Handbook of Ceramic Hard Materials

Ralf Riedel (Editor)

Handbook of Ceramic Hard Materials

@WILEY*VCH

Related titles from WILEY-VCH

M. Swain (Ed.) Structure and Properties of Ceramics ISBN 3-527-26824-3 R. J. Brook (Ed.) Processing of Ceramics Part I: ISBN 3-527-26830-8 Part 11: ISBN 3-527-29356-6 J. Bill, F. Wakai, F. Aldinger Precursor-Derived Ceramics ISBN 3-527-29814-2

Ralf Riedel (Editor)

Handbook of Ceramic Hard Materials

@WILEY-VCH Weinheim . New York . Chichester . Brisbane . Singapore . Toronto

Editor: Prof. Dr. Ralf Riedel Fachgebiet Disperse Feststoffe Fachbereich Materialwissenschaft Technische Universitit Darmstadt PetersenstraRe 23 64287 Darmstadt Germany

This book was carefully produced. Nevertheless, authors, editor and publisher do not warrant the information contained therein to be free of errorb. Readers are advised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate.

Library of Congress Card No. Applied for A catalogue record for this book is available from the British Library Deutschc Bibliothek Cataloguing-in-Publication Data: A catalogue record for this publication is available from Die Deutschc Bibliothek ISBN 3-527-29912-6

C WILEY-VCH Verlag GmbH, D-69469 Weinheiin (Federal Republic of Germany), 2000 Printed on acid-free and chlorine-free paper All rights reserved (including those of translation in other languages). N o part of this book may be reproduced in any form by photoprinting, microfilm, or any other means nor transmitted or translated into machine language without written permission from the publishers. Registered names, trademarks, etc. used in this book, even when not specifically maked as such, are not to be considered unprotected by law. Composition: Alden Bookset, Oxford Printing: betz-druck, Darmstadt Bookbinding: Buchbinderei Osswald, NeustadtiWstr. Printed in the Federal Republic of Germany ~

~

This book is dedicate to

Ute, Vincent, Lorenz and Marlene

Preface

With increasing demand for improved efficiency of engines, plants and production processes, ceramics have gained great importance as structural engineering materials in recent years. Within the group of the so called advanced materials, carbon in form of diamond or diamond-like structures, carbides, nitrides and borides have reached an outstanding position due to their excellent hardness and thermo-chemical and thermo-mechanical properties. The distinct covalent bonding of the aforementioned structures positively influences their hardness and their tribological behavior. Moreover, a series of oxides such as stishovite, a high pressure modification of silica, or boron sub-oxides have been recently discovered to exhibit high hardness apart from the well known alumina. There is presently much effort in basic science and applied research to work on novel ceramic hard materials denoted as super- or ultra-hard materials that can compete with the hardness of conventional diamond. Aim and scope of the research in this field is to develop hard materials with superior mechanical and chemical properties and with similar hardness. Moreover, calculations of properties of hypothetical carbon nitrides like C3N4 indicated that there might be compounds exhibiting even higher hardness values than that of diamond. The low-temperature synthesis of diamond and cubic boron nitride on the one hand as well as the successful research on new carbon nitrides on the other hand have caused an enormous impact around the world on both the basic science and the technological development of these novel ultra-hard materials. With the present book we wish to review comprehensively and concisely the state of the art concerning the structure, synthesis, processing, properties and applications of ceramic hard materials in general. In particular, the synthesis, modeling and properties of novel hard materials like binary carbon nitrides, ternary boron carbonitrides and others are also addressed. It is the aim of this reference book not only to reflect the state of the art and to give a sound review of the literature, but to delineate the underlying concepts and bearing of this interdisciplinary field. With the present edition we wish to show that the field of hard materials research and development has to be recognized into the wider context of chemistry, physics as well as materials science and engineering. The book is organized in two volumes and three parts, covering the structure and properties of ceramic hard materials (Volume 1, Part I), synthesis and processing (Volume 1, Part 11) as well as the typical fields of applications (Volume 2, Part 111). Volume 1 starts with an introduction into novel ultra hard ceramics including diamond and diamond-like carbon, carbon nitrides and silicon nitrides as well as boron containing carbides, nitrides and carbonitrides. Here we wish to recognize the great fundamental and technological challenge of developing new superhard

VIII

Preface

materials which can compete with the hardest counterparts such as diamond and cubic boron nitride. In dealing with properties, the first Chapter in Part I is then devoted to the structure of crystalline and amorphous ceramic hard materials. The structural features are responsible in particular for the intrinsic materials properties such as melting point and hardness. It has been found that in many cases the hardness of a crystalline substance correlates with its melting point. Therefore, detailed knowledge of the 3dimensional arrangement of the atoms is required to understand the materials behavior under certain conditions. More details of the individual crystal structures with respect to a 3dimensional view can be found on our hard materials homepage under the web address ‘www.hardmaterials.de’. Phase transitions and materials synthesis under high pressure in laser heated diamond cells is the topic of the continuing Chapter. The materials behavior under high pressure and temperature is of fundamental interest for the synthesis of hard materials since many of the ultra-hard substances like diamond, cubic boron nitride or stishovite are formed naturally or synthetically under these harsh conditions. The next three Chapters are concerned with the mechanical behavior and corrosion of ceramic hard materials and their relation to microstructure. This correlation is an important feature since hardness is not only governed by the intrinsic atomic structure of the respective material but also to a great extend by its polycrystalline nature. Therefore, the grain morphology and grain boundary chemistry play a decisive role in the materials response under environmental or mechanical load. In the following Chapter transition metal carbides, nitrides and carbonitrides are discussed with a focus on their structure and bonding, thermodynamic behavior as well as on their physical and mechanical properties. Part I is then completed by two Chapters which deal with the theoretical design of novel sp2-bonded carbon allotropes and novel superhard materials based on carbon and silicon nitrides. These Chapters tribute to the fact that with proceeding computerization the number of calculated novel solid structures that led to the prediction of new materials with hardness comparable to or exceeding that of diamond has increased enormously in recent years. Part I1 continues with the synthesis and processing of ceramic hard materials. Since the conventional powder technological synthesis and processing of ceramics has been treated in a large number of published review articles here we concentrate on novel synthetic routes that provide ceramic hard materials. Consequently, six Chapters report on i) directed metal oxidation, ii) self-propagating high temperature synthesis, iii) hydrothermal synthesis of diamond, chemical vapor deposition of diamond (iv) and cubic boron nitride (v) films and finally vi) the polymer to ceramic transformation. All these processes are particularly suitable for the formation of refractories with high hardness. Part I1 is then closed by a Chapter on nano structured superhard materials. In the course of this work high hardness is achieved by microstructural control rather than by the synthesis of a distinct crystal structure. In Volume 2 ceramic hard materials are highlighted in the light of their applications. Chapter 1 of Part I11 concisely reviews the history of diamond and diamondlike super abrasive tools while Chapter 2 and 3 are concerned with the application of chemical vapor deposited diamond and diamond-like carbon films. These sections

Prejace

IX

include the synthesis of optical grade CVD diamond windows and discuss their physical and mechanical properties. The most important and wide-spread ceramic hard materials are based on alumina. Chapter 4 reports on the processing developments to increase the hardness of alumina based ceramics for grinding and cutting applications. Silicon carbide and silicon nitride materials are the most technologically important non-oxide compounds and have gained great significance in the field of cutting ceramics and are treated in Chapters 5 and 6. Boron-based ceramics are a further group of either established or candidate materials with extreme hardness. Therefore, Chapter 7 deals with boron carbide or transition metal borides like titanium diboride and their distinct properties and applications. In Chapter 8, classical hard metals comprised of tungsten carbide as the hard phase and cobalt as the binder phase are discussed. Volume 2 is finally completed by a data base (Chapter 9) containing approximately 130 hard materials including carbides, nitrides, borides, silicides and oxides. The data base references the crystal structure, physical properties like melting point and density, mechanical properties (Youngs modulus, micro hardness) and oxidation resistance of the respective compounds. Future developments of novel hard materials such as the recently discovered intermetallic phase A1MgBl4will be updated on our internet homepage ‘www.hardmaterials.de’. In closing these introductory remarks, I would like to emphasize that the special chance to place a summary of the outstanding expertise on the field of present hard materials research and development would not have been possible without the great enthusiasm and commitment of all the colleagues who contributed in the writing of this two volume set. I am grateful for their enormous efforts in compiling a fascinating series of articles imparting depth insight into the individual fields of modern hard materials research. Finally, I wish to thank the Wiley-VCH Editors Peter Gregory and Jorn Ritterbusch for encouraging me in the preparation of this book and for their continuous support throughout the editorial process. Ralf Riedel March 2000 Darmstadt

Foreword

One of the clearest hierarchies in materials science and engineering is provided by the property of hardness. There are, of course, many properties where remarkable differences exist between groups of materials. An example is provided by electrical conductivity where a ratio of 10l8can be readily found; with electrical conductivity, however, the different materials do not come into direct competitive opposition. In the case of hardness, the very value of this property lies in the ability of one material to demonstrate a higher place in the hierarchy than another; the one material is used in effect to overpower the other. The existence of this hierarchy, which has been long recognised in the traditional measurement scale for the property, has direct relation to applications. In any use of materials it is important to be able to shape them to be fit for purpose; where the shaping process involves some type of machining, as it most commonly does, then the property of hardness becomes the unambiguous figure of merit. It is for these reasons that there has been long standing and productive interest in hard materials, in their design, in their fabrication, in their use, and in the underlying science and engineering. It is thorougLlv in keeping with this tradition of research relevant to application that the present book brings together a set of authoritative reviews of the progress which has been made. The organisation of the book is a direct reflection of the logic which has been used in developing hard materials. One of the great attractions of the subject has been the close link that exists between hardness on the one hand and the bonding and structure of the material on the other. The link between these two has proved to be one of the best foundations on which to base materials development. The link is a central theme in the first part of the book where fine examples are given of the rich contribution which has been made and which continues to be made by fundamental studies of bonding and structure to materials performance. It has long been recognised that the very aspect of their extreme resistance to deformation would make it a particular challenge to manufacture hard materials in reliable and cost-effective ways. It is here that the materials community has shown itself to be imaginative and forward looking in seeking innovative fabrication routes. These are well presented in the second part of the book where specific attention is given to the paths which can be used to assemble materials of precisely defined form without sacrifice of their characteristic mechanical resilience. The most striking aspect of hard materials, however, is the direct link to applications. This link has brought an unusual degree of purpose to materials development which has enjoyed the benefits of being conducted in full recognition of the target to be reached. It has also meant that the progress made in research can be rapidly evaluated since the testing procedures relate so directly to the end use. The third

XI1

Foreword

part of the book accordingly gives close accounts of the performance of the different classes of hard materials in the applications context. The contributors to this text are to be congratulated on bringing their many disciplines to bear on this central theme. Materials science is well known to undergo fashions as materials are developed and discarded and indeed as sectors of application grow and decline. The one requirement which will remain is that the forming and shaping of materials will always be necessary whatever the eventual sector of application. We can accordingly be confident that the long history of hardness studies, not least in the last two hundred years from the carbon tool steels, to high speed steels, to stellite, to tungsten carbide, to cermets, to ceramics, and now to diamond, boron, nitride and other special systems, will be continued with informed imagination and with creative innovation. The present book is a splendid platform on which to base such future development. Richard Brook January 2000 Oxford, UK

Contents

List of Contributors XXVII List of Symbols XXXIII List of Abbreviations XXXIX Introduction: Novel Ultrahard Materials A . Zerr and R. Riedel Introduction XLV Hard Materials XLVI Hardness XLVII Carbon-based Hard Materials L Diamond LII Diamond-like and Amorphous Carbon LV Novel Hypothetical Three-dimensional Carbon Phases Fullerenes LIX Carbon Nitride (C3N4) LIX Boron-based Hard Materials LXIV Boron Nitrides LXIV Boron-rich Boron Nitrides LXVII Nitrogen-rich Boron Nitride LXVIII Boron Carbonitrides (B,C,N,) LXVIII Boron Suboxides LXXI . Silicon-based Materials LXXI Concluding Remarks LXXII Acknowledgement LXXIII References LXXIII Part I

Structures and Properties

1

Structural Chemistry of Hard Materials W . Jeitschko, R. Pottgen, and R.-D. Hoffmann

1.1

Introduction 3 Diamond and Diamond-Related Structures 5 The Crystal Structure of Diamond 5 The Isoelectronic Compounds c-BN and S i c 6 Crystal Chemistry of Borides and Boron Carbides 8 The Structures of Transition Metal Carbides 12 Silicides and Silicide Carbides of Transition Metals 20

1.2

1.2.1 1.2.2 1.3 1.4

1.5

LVI

XIV

Contents

1.6 1.6.1 1.6.2 1.6.3 1.7 1.7.1 1.7.2 1.8

Nitrides 23 Nitrides of Main Group Elements 24 Transition Metal Nitrides 25 Perspectives: Nitridosilicates 29 Oxide Ceramics 30 Hard Ceramics of Main Group Elements 30 Transition Metal Oxides 32 Amorphous Hard Materials 36 References 37

2

Phase Transitions and Material Synthesis using the C02-Laser Heating Technique in a Diamond Cell A . Zerr, G. Serghiou, and R . Boehler

2.1 2.2 2.2.1 2.2.2 2.2.3

Introduction 41 Technique of C02-Laser Heating in a Diamond Anvil Cell 42 Sample Assemblage in a Diamond Anvil Cell 42 Pressure Conditions in the Sample Volume 43 Experimental Set-up for C02-Laser Heating in a Diamond Anvil Cell 44 Temperature Determination 45 Temperature Stabilization 45 Radial Temperature Gradients 48 Raman and Fluorescence Spectroscopic Analysis of Samples in a Diamond Anvil Cell 48 Determination of Melting Temperatures at High Pressures 49 Melting of Cubic BN at 10 GPa 49 Melting Temperatures of Materials Relevant to the Earth’s Lower Mantle 51 Phase Diagrams, Decomposition Reactions, and Stability of Solids at High Pressures and Temperatures 54 Coesite-Stishovite Phase Boundary 55 High Pressure and Temperature Phase Diagram and Decomposition Reactions in a Ternary System 56 Stability of a Perovskite Oxide with Respect to its Component Oxides 59 C02-laser Heating Experiments on Organic Compounds 60 Conclusion 62 Acknowledgments 62 References 62

2.2.4 2.2.5 2.2.6 2.2.7 2.3 2.3.1 2.3.2 2.4 2.4.1 2.4.2 2.4.3 2.5 2.6

3

Mechanical Properties and their Relation to Microstructure D. Sherman and D . Brandon

3.1 3.1.1

Introduction 66 Applications and Engineering Requirements

66

Contents

XV

3.1.2 3.1.3 3.1.4 3.2 3.2.1 3.2.2 3.2.3 3.2.4 3.3 3.3.1 3.3.2 3.3.3 3.3.4 3.3.5 3.3.6 3.4 3.5 3.5.1 3.5.2 3.5.3 3.6

Bulk Components 68 Coatings 70 Engineering Requirements 70 Principal Mechanical Properties 7 1 Elastic Modulus 71 Strength 72 Fracture Toughness 74 Hardness 79 Mechanical Testing of Hard Materials 81 Elastic Modulus 81 Fracture Strength 8 1 Fracture Toughness 83 Hardness 84 Indentation Toughness 86 Erosion, Wear and Scratch Tests 89 Microstructural Parameters and Mechanical Properties 9 1 Failure Mechanisms 94 Creep Behavior 94 Mechanical Fatigue 95 Ballistic Properties 97 Conclusions 98 References 99

4

Nanostructured Superhard Materials S. Veptek

4.1 4.2 4.2.1 4.2.2 4.3

Introduction 104 Concept for the Design of Superhard Materials 109 Nanocrystalline Materials 110 Heterostructures 114 Preparation and Properties of Superhard Nanocrystalline Composites 116 Preparation 116 Properties of the ncM,N/aSi3N4 Composites 119 Other Superhard Nanocomposites and the General Validity of the Design Principle 124 Discussion of the Possible Origin of the Hardness and Stability of the Nanostructure 128 Conclusions 133 Acknowledgments 134 References 134

4.3.1 4.3.2 4.3.3 4.4 4.4

5

Corrosion of Hard Materials K . G. Nickel and Y. G. Gogotsi

5.1

Introduction

140

XVI 5.2 5.3 5.3.1 5.3.2 5.4 5.4.1 5.4.2 5.4.3 5.4.4 5.4.5 5.5 5.5.1 5.5.2 5.6 5.6.1 5.6.2 5.6.3 5.6.4 5.6.5

Contents

Corrosive Media 140 Corrosion Modes 141 Active and Passive Corrosion 141 Homogeneity and Location of Attack: Internal, External and Localized Corrosion 141 Corrosion Kinetics 142 Physical Boundary Conditions 142 Active Corrosion Kinetics 143 Basic Passive Corrosion Kinetics 145 Kinetic Breaks 147 Complex Kinetics 148 Corrosion Measurement 150 Experimental Methods 150 Corrosion Data 151 Materials 154 Diamond and Diamond-like Carbons 154 Carbides 155 Nitrides 166 Carbonitrides 173 Titanium Diboride 176 References 177

6

Interrelations Between the Influences of Indentation Size, Surface State, Grain Size, Grain-Boundary Deformation, and Temperature on the Hardness of Ceramics A . Krell

6.1 6.2

Introduction 183 The Assessment of Residual Porosity and Flaw Populations: A Prerequisite for any Hardness Investigation 184 Theoretical Considerations 185 The Role of the Lattice and of Grain Boundaries in the Inelastic Deformation at an Indentation Site in Sintered Hard Materials 185 Quantitative Understanding the Load Effect on the Hardness: Theoretical Considerations Compared with Single Crystal Data 188 Influences of the Grain Size and the State of the Surface 191 The Grain Size Influence on the Load Effect of the Hardness: Modeling Experimental Results 191 The Effect of the Grain Size and the Surface State in Ceramics when Recorded by Different Measuring Approaches 193 Comparing the Grain Size Effect and the Indentation Size Effect: The Role of Grain Boundaries at Room Temperature 195 The Effects of Temperature on the Hardness of Ceramics 198 Summary 199 References 20 1

6.3 6.3.1 6.3.2 6.4 6.4.1 6.4.2 6.5 6.6 6.7

Contents

XVII

7

Transition Metal Carbides, Nitrides, and Carbonitrides W. Lengauer

7.1 7.2 7.2.1 7.2.2 7.3 7.4 7.4.1 7.4.2 7.5 7.5.1 7.5.2 7.5.3 7.5.4 7.5.5 7.6

Introduction 202 General Features of Structure and Bonding 205 General Structural Features 205 General Features of Bonding 206 Preparation 207 Characterization 2 10 Chemical Analysis 2 10 Physical Microanalysis 21 1 Thermodynamics 2 12 Stability of Carbides 212 Nitrogen Partial Pressure of Nitrides 212 Phase Equilibria of Important Carbide Systems 213 Transition Metal-Nitrogen Systems and Structure of Phases 216 Carbonitride Systems 221 Properties of Important Transition Metal Carbides, Nitrides, and Carbonitrides 224 Melting Points 224 Color 224 Thermal and Electrical Conductivities 225 Thermal Expansion 228 Diffusivities 229 Elastic Properties 23 1 Microhardness 234 Industrial Applications 238 Cemented Carbides and Carbonitrides 238 Deposited Layers 241 Diffusion Layers 246 Acknowledgments 248 References 248

7.6.1 7.6.2 7.6.3 7.6.4 7.6.5 7.6.6 7.6.7 7.7 7.7.1 7.7.2 7.7.3

8

New Superhard Materials: Carbon and Silicon Nitrides J . E. Lowther

8.1 8.2 8.2. I 8.2.2 8.2.3 8.2.4 8.3 8.3.1 8.3.2 8.3.3

Introduction 253 Modeling Procedures 254 Semi-empirical Approaches 254 Tight-binding Schemes 255 Ab initio Pseudopotential Approach 256 Transition Pressures and Relative Stability 256 Carbon Nitride 257 Crystalline Structures 258 Graphitic Structures 259 Amorphous Structures 261

XVIII

Contents

8.3.4 8.4 8.4.1 8.4.2 8.4.3 8.5

Relative Stability 263 Silicon Carbon Nitride 264 j3SiC2N4 265 Near-cubic Forms of SiC2N4 266 Relative Stability 268 Conclusions 268 Acknowledgements 269 References 269

9

Effective Doping in Novel sp2 Bonded Carbon Allotropes G. Jungnickel, P. K. Sitch, T. Frauenheim, C. R. Cousins, C. D. Latham, B. R. Eggen, and M . I. Heggie

9.1 9.2 9.3 9.4 9.5 9.6

Introduction 271 Lattice Description 274 Computational Methods 276 Static Properties 278 Electronic Properties 279 Conclusions 282 Acknowledgments 283 References 283

Part I1

Synthesis and Processing

1

Directed Metal Oxidation V. Jayaram and D. Brandon

1.1 1.2 1.2.I 1.3 I .4 1.5 1.5.1 1.5.2 1.5.3 1.5.4 1.5.5 1.5.6 1.6 1.6.1 1.6.2 1.6.3 1.6.4 1.6.5 1.6.6

Historical Background 289 Oxidation and Oxide Formation 290 Initial Oxidation 291 Related Ceramic Processing Routes 293 Directed Metal Oxidation Incubation 295 Directed Metal Oxidation Growth 300 Introduction 300 Directed Metal Oxidation Composites from Al-Mg Alloys 300 Directed Metal Oxidation Growth from other Aluminum Alloys 304 Microstructural Scale 305 Growth into Particulate Preforms 307 Growth into Fibrous Preforms 309 Mechanical Properties 310 Elastic Modulus 310 Strength and Toughness 3 11 Thermal Shock 313 High Temperature Strength 313 Wear Properties 314 Mechanical Properties of Fiber-reinforced DMO Composites 3 14

Contents

XIX

1.7 1.8 1.9 1.9.1 1.9.2 1.9.3

Corrosion of Directed Metal Oxidation Composites 3 16 Other Properties 316 Applications 3 I6 Wear Resistant Components 317 Ceramic Composite Armor 3 17 Thermal Barriers and Heat Sinks 318 References 3 18

2

Self-propagating High-Temperature Synthesis of Hard Materials Z . A . Munir and U. Anselmi-Tumburini

2.1 2.2 2.3 2.3 2.4

Introduction 322 Mechanistic Characterization of the Process 327 Effect of Experimental Parameters 33 1 Synthesis of Dense Materials 342 Synthesis by Field-Activated Self-propagating High-temperature Synthesis 348 Selected Recent Examples of Synthesis of Hard Materials 356 Acknowledgment 368 References 368

2.6

3

Hydrothermal Synthesis of Diamond K. G. Nickel, T. Kruft, and Y. G. Gogotsi

3.1 3.2 3.3 3.3.1 3.3.2 3.4

Introduction 374 Evidence from Nature 376 Hydrothermal Synthesis 377 C-H-0 System 377 Hydrothermal Treatment of S i c 382 Outlook 387 Acknowledgments 387 References 387

4

Chemical Vapor Deposition of Diamond Films C.-P. Klages

4.1 4.2 4.2.1 4.2.2 4.2.3

Introduction 390 Preparation Methods for Diamond Films 391 Hot-filament Chemical Vapor Deposition 392 Microwave-plasma-based Methods 397 Preparation of Special Forms: Textured and Heteroepitaxial Films 400 Thermochemistry and Mechanism of Chemical Vapor Deposition Diamond Growth 407 Transformation of Graphite to Diamond at Low Pressures 407

4.3 4.3.1

xx 4.3.2 4.4 4.4.1 4.4.2 4.4.3 4.4.4 4.5

Contents

Reactive Species in Diamond Chemical Vapor Deposition, the Role ofCH3 408 Properties and Applications of Chemical Vapor Deposited Diamond 410 Diamond Coated Cutting Tools 41 1 Thermal Conductivity of Chemical Vapor Deposited Diamond: Thermal Management Applications 412 Electrical Properties and Electronic Applications 413 Electrochemical Use of Chemical Vapor Deposited Diamond 415 Summary 417 References 4 17

5

Vapor Phase Deposition of Cubic Boron Nitride Films K. Bewilogua and F. Richter

5.1 5.2 5.2.1 5.2.2 5.2.3 5.3 5.4 5.4.1 5.4.2

Introduction 420 Empirical Results 421 Deposition Methods 421 Morphology and Structure of cBN Films 423 Film Adhesion 427 Models of cBN Formation 427 Sputter Deposition of cBN Films 429 Sputter Deposition with Conducting Targets 430 Deposition by d.c. Magnetron Sputter with a Hot Boron Target 43 1 Discrimination between Nucleation and Growth Phase 433 Detection of hBNxBN Transition 433 RF Magnetron Sputtering 435 Properties of cBN Films 440 Mechanical and Tribological Properties 440 Optical Properties 440 Electrical Properties 441 Other Properties 441 Summary and Outlook 442 References 442

5.5 5.5.1 5.5.2 5.6 5.6.1 5.6.2 5.6.3 5.6.4 5.7

6

Polymer to Ceramic Transformation: Processing of Ceramic Bodies and Thin Films G. D. Soraru and P. Colombo

6.1 6.2 6.3

Introduction 446 Processing of Monolithic Components 450 Preparation and Characterization of SiAlOC Ceramic Bodies by Pyrolysis in Inert Atmosphere 452 Experimental Procedure 452

6.3.1

Contents

6.4 6.4.1 6.4.2 6.4.3 6.4.4 6.5 6.6 6.6.1 6.7 6.8 6.9 6.10 6.10.1 6.10.2 6.1 1

XXI

Results 453 Characterization of the Pre-ceramic Precursors 453 Characterization of the Pre-ceramic Components 454 Characterization of the Ceramic Components 455 Mechanical Characterization at High Temperature 457 Discussion 458 Preparation and Characterization of SiAlON Ceramics by Pyrolysis in Reactive Atmosphere 460 Experimental 460 Results and Discussion 460 Processing of Thin Ceramic Films 463 Experimental 463 Results and Discussion 464 Conventional Conversion Process: Annealing in Controlled Atmosphere 464 Nonconventional Conversion Process: Ion Irradiation 467 Conclusions 472 Acknowledgments 473 References 473

Part 111

Materials and Applications

1

Diamond Materials and their Applications Edited by R. J . Caveney

1.1 1.1.1 1.1.2 1.1.3 1.1.4 1.1.5 1.1.6 1.1.7 1.2 1.2.1 1.2.2 1.2.3 1.2.4 1.2.5 1.2.6 1.2.7 1.3 1.3.1 1.3.2 1.3.3

Superabrasive tools: A Brief Introduction 479 Introduction 479 Early History 479 Synthetic Diamond 48 1 Cubic Boron Nitride 482 Polycrystalline Diamond and Cubic Boron Nitride 482 Chemical Vapor Deposited Diamond 484 Outline of Chapter 485 The Crystallization of Diamond 485 The Carbon Phase Diagram 485 Diamond Crystallization at High Pressure 487 High Pressure Apparatus 490 The Synthesis of Particulate Diamond Abrasives 49 1 Growth of Large Synthetic Diamonds 496 Novel Diamond Synthesis Routes 504 Cubic Boron Nitride Crystallization 5 10 Polycrystalline Diamond and Cubic Boron Nitride 5 12 Natural Polycrystalline Diamond 5 12 Synthetic Polycrystalline Diamond 512 Mechanisms involved in Polycrystalline Diamond Manufacturing Process 513

XXII

Contents

1.3.4 1.4 1.4.1 1.4.2 1.4.3 1.4.4 1.4.5 1.4.6 1.4.7 1.5 1.5.1 1.5.2 1.5.3 1.5.4 1.5.5

Polycrystalline Cubic Boron Nitride 5 18 New Ultrahard Materials 521 Introduction 521 Hardness 521 C3N4 523 Boron Rich Nitride 526 Boron Carbonitrides 526 Boron Suboxides 526 Stishovite 526 Industrial Applications of Diamond and cBN 527 Introduction 527 Abrasive Application 528 Machining of Stone and Concrete 540 Applications of Polycrystalline Ultra-hard Materials Applications of Single Crystal Diamond 559 Acknowledgments 566 References 566

2

Applications of Diamond Synthesized by Chemical Vapor Deposition R. S. Sussmann

2.1 2.2 2.2.1 2.2.2 2.2.3 2.2.4 2.2.5 2.2.6 2.3 2.3.1

Introduction 573 Properties of Chemical Vapor Deposited Diamond 574 Material Grades 574 Optical Properties 576 Strength of Chemical Vapor Deposited Diamond 580 The Young Modulus 581 Thermal Conductivity 582 Dielectric Properties 583 Optical Applications 583 Chemical Vapor Deposited Diamond for Passive Infrared Windows in Aggressive Environments 584 Windows for High-power Infrared Lasers 589 Windows for High Power Gyrotron Tubes 597 Window Requirements 598 The Development of Chemical Vapor Deposited Diamond Gyrotron Windows 599 Thermal Management of Laser Diode Arrays 606 Laser Diode Arrays: General Issues 607 Modelling of Submount Heat Resistance 607 Flatness of Submount 610 Thermal Stress 610 Cutting Tools, Dressers and Wear Parts 61 1 Cutting Tools Trends 61 1 Cutting Tool Application of Chemical Vapor Deposited Diamond 612

2.3.2 2.4 2.4.1 2.4.2 2.5 2.5.1 2.5.2 2.5.3 2.5.4 2.6 2.6.1 2.6.2

548

Contents

XXIII

2.6.3 2.6.4

Chemical Vapor Deposited Diamond Dressers 616 Chemical Vapor Deposited Diamond Wear Parts 617 References 6 19

3

Diamond-like Carbon Films C.-P. Klages and K . Bewilogua

3.1 3.2 3.2.1 3.2.2. 3.2.3 3.3 3.3.1 3.3.2 3.4. 3.4.1 3.4.2 3.5. 3.5.1 3.5.2 3.5.3 3.5.4

Introduction 623 Preparation Methods for Diamond-like Carbon Films 623 Hydrogenated Amorphous Carbon (a-C: H) 623 Hydrogen Free Amorphous Carbon (ta-C) 627 Metal-containing Amorphous Hydrocarbon 629 Microstructure and Bonding of Diamond-like Carbon 630 Amorphous Carbon and Hydrogenated Amorphous Carbon Metal-containing Amorphous Carbon Films 634 Physical Properties of DLC Films 637 Electrical and Optical Properties 637 Mechanical Properties 639 Applications of DLC Films 640 Adhesion of DLC Films 640 Tribology of DLC Coatings 642 Tribological Applications 644 Other Applications 644 References 645

4

Ceramics Based on Alumina: Increasing the Hardness for Tool Applications A . Krell

4.1 4.2

Recent Trends in the Application of Ceramic Tool Materials 648 Technological Essentials for Producing Hard and Strong Tool Ceramics 650 Typical Defects in Ceramics Tool Materials: The State of The Art 651 Recent Trends in Ceramic Technologies Related to Tool Ceramics 653 Tool Materials with Undefined Cutting Edge: Sintered Grinding Materials 658 Technical Demands for Grinding Materials 660 661 Advanced Commercial Products: Sol/gel-derived Corundum Sintered Alumina Grits Produced by Powder Processing Approaches 665 New Trend for Cutting Hard Workpieces: Submicrometer Cutting Ceramics for Tools with Defined Cutting Edge 666 Demands for Cutting Materials Used for Turning Hard Workpieces 667

4.2.1 4.2.2 4.3 4.3.1 4.3.2 4.3.3 4.4 4.4.1

630

XXIV

Contents

4.4.2 4.4.3 4.4.4

Carbide Reinforced Composite Ceramics Based on A1203 669 Single Phase Sintered Corundum 670 Comparative Cutting Studies with Submicrometer Ceramics: A1203 and Composites Reinforced with Ti(C,N) and Ti(C,O) 670 Summary 680 References 68 1

4.5

5

Silicon Carbide Based Hard Materials K. A . Schwetz

5.1 5.1.1 5.1.2 5.2 5.3 5.3.1 5.3.2 5.3.3 5.4 5.4.1 5.4.2 5.4.3 5.5 5.6 5.7

lntroduction 683 History 683 Natural Occurrence [7] 684 Structure and Phase Relations of S i c 685 Production of S i c 688 The Acheson/ESK Process 688 Other Production Methods 69 1 Dense S i c Shapes 699 Properties of Silicon Carbide 719 Physical Properties 719 Chemical Properties 720 Tribological Properties 723 Quality Control 734 Toxicology and Occupational Health 736 Uses of Silicon Carbide 736 Acknowledgments 740 References 740

6

Silicon Nitride Based Hard Materials M . Herrmann, H. Klemm, Chr. Schubert

6.1 6.2 6.3 6.4 6.4.1 6.4.2 6.5 6.5.1 6.5.2 6.5.3 6.5.4 6.6

Introduction 749 Crystal Structure and Properties of the Si3N4Modifications 753 Densification 755 Microstructural Development 758 Microstructural development of P-Si3N4materials 758 Microstructural development of a’-SiALON materials 768 Properties of Si3N4Materials 771 Mechanical properties at room temperatures 771 High-temperature properties of silicon nitride materials 777 Wear resistance of Si3N4materials 782 Corrosion resistance of Si3N4 786 Conclusions/Further potential of silicon nitride materials 792 Acknowledgements 795 References 795

Contents

xxv

7

Boride-Based Hard Materials R. Telle, L. S . Sigl, and K. Takagi

7.1 7.2 7.2.1 7.2.2 7.3 7.3.1 7.3.2 7.4 7.4.1 7.4.2 7.4.3 7.4.4 7.4.5 7.5 7.5.1 7.5.2 7.5.3 7.6

Introduction 802 Chemical Bonding and Crystal Chemistry of Borides 803 Chemical Bonding of Borides 803 The Crystal Structure of Borides 804 Phase Systems 812 Binary Phase Diagrams of Technically Important Systems 813 Ternary and Higher Order Systems 818 Boron Carbide Ceramics 837 Preparation of Boron Carbide 837 Sintering of Boron Carbide 839 Properties of Boron Carbide 851 Chemical Properties and Oxidation of Boron Carbide 855 Boron Carbide-Based Composites 857 Transition Metal Boride Ceramics 874 Preparation of Transition Metal Borides 875 Densification of Transition Metal Borides 876 Properties of Transition Metal Borides Ceramics 878 Multiphase Hard Materials Based on Carbide-Nitride-Boride-Silicide Composites 888 Boride-Zirconia Composites 888 Cemented Borides 895 Boron Carbide-Based Cermets 895 Titanium Diboride-Based Cermets 897 Cemented Ternary Borides 919 Potentials and Applications 927 Future Prospects and Fields of Application 933 References 936

7.7 7.8 7.8.1 7.8.2 7.8.3 7.8.4 7.9

8

The Hardness of Tungsten Carbidecobalt Hardmetal 946 S . Luyckx 946

8.1 8.2 8.2.1 8.2.2 8.3 8.3.1 8.3.2 8.3.3 8.3.4 8.3.5 8.4

Introduction 946 The Hardness of the Two Component Phases 947 The Hardness of Tungsten Carbide 947 The Hardness of Cobalt 948 Factors Affecting the Hardness of WC-Co Hardmetal 950 Cobalt Content and Tungsten Carbide Grain Size 950 Grain Size Distribution and Cobalt Mean Free Path 952 Binder Composition and Carbon Content 952 Porosity 953 Effect of Temperature 953 Relationships between Hardness and Other Hardmetal Properties 960

XXVI

Con ten /s

8.4.1 8.4.2 8.5

Relationship between Hardness and Toughness 962 Relationship between Hardness and Abrasive Wear Resistance Conclusions 963 Acknowledgments 963 References 964

9

Data Collection of Properties of Hard Materials G. Berg, C. Fviedrich, E. Broszeit, and C. Bevger

9.1 9.2 9.3

Introduction 965 Profile of Properties 965 Organization and Contents of the Data Collection Acknowledgement 967 Refercnces 99 1

Index 997

966

962

List of Contributors

U. Anselmi-Tamburini Dipartimento di Chimica Fisica Universita di Pavia 27100 Pavia Italy

R. Bohler Max-Planck-Institute for Chemistry Saarstrasse 23 D-55020 Mainz Germany

M. W. Bailey De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 21 35 South Africa

D. Brandon Department of Materials Engineering Technion - Israel Institute of Technology Haifa 32000 Israel

G. Berg Fachgebiet und Institut fur Werkstofiunde der TU Darmstadt und Staatliche Materialpru fungsanhalt Grafenstrasse 2 D-64283 Darmstadt Germany C. Berger Fachgebiet und Institut fur Werkstofiunde der TU Darmstadt und Staatliche Materialpriifungsanhalt Grafenstrasse 2 D-64283 Darmstadt Germany K. Bewilogua Fraunhofer Institut fur Schicht und Oberflachentechnik (IST) Bienroder Weg 54 E D-38 108 Braunschweig Germany

J. R. Brandon De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2 135 South Africa E. Broszeit Fachgebiet und Institut fur Werkstofiunde der TU Darmstadt und Staatliche Materialprufungsanhalt Grafenstrasse 2 D-64283 Darmstadt Germany

R. C. Burns De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 21 35 South Africa

XXVIII


R. J. Caveney De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2135 South Africa

G. J. Davies De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2135 South Africa

S. E. Coe De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2135 South Africa

B. R. Eggen School of Chemistry, Physics and Environmental Sciences University of Sussex Falmer Brighton BNl 9QJ UK

J. L. Collins De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2135 South Africa

D. Fister HC Starck Gmbh Kraftwerkweg 3 D-79725 Laufenburg Germany

P. Colombo Universita di Bologna Dipartimento di Chimica Applicata e Scienza dei Materiali viale Risorgimento 2 1-40I36 Bologna Italy M. W. Cook De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 21 35 South Africa C. R. Cousins Department of Physics University of Exeter Stocker Road Exeter EX4 4QL UK

T. Frauenheim Fachbereich Physik Universitat/Gesamthochschule Paderborn D- 33095 Paderborn Germany C. Friedrich Fachgebiet und Institut fur Werkstoffkunde der TU Darmstadt und Staatliche Materialprufungsanhalt Grafenstrasse 2 D-64283 Darmstadt Germany Y. G. Gogotsi Institut fur Angewandte Mineralogie Universitat Tubingen Wilhelmstrasse 56 D-72074 Tubingen Germany


J. 0. Hansen De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2 135 South Africa M. Hoffmann Fakultat fur Maschinenbau Institut fur Werkstoffkunde I1 Universitat Karlsruhe Kaiserstrasse 12 Postfach 6980 D-76 128 Karlsruhe Germany R. D. Hoffmann Westfalische Wilhelms-Universitat Miinster Anorganisch-Chemisches Institut Wilhelm-Klemm-Strasse 8 D-48 149 Miinster Germany

M. I. Heggie School of Chemistry, Physics and Environmental Sciences University of Sussex Falmer Brighton BN1 9QJ UK V. Jayaram Department of Metallurgy Indian Institute of Science Bangalore India W. Jeitschko Westfalische Wilhelms-Universitat Miinster Anorganisch-Chemisches Institut Wilhelm-Klemm-Strasse 8 D-48149 Miinster Germany

XXIX

G. Jungnickel Fachbereich Physik Universitat/Gesamthochschule Paderborn D- 33095 Paderborn Germany C. P. Klages Fraunhofer Institut fur Schicht und Oberflachentechnik (IST) Bienroder Weg 54 E D-38108 Braunschweig Germany T. Kraft lnstitut fur Angewandte Mineralogie Universitat Tubingen Wilhelmstrasse 56 D-72074 Tubingen Germany A. Krell Fraunhofer Institute for Ceramic Technologies and Sintered Materials Winterbergstrasse 28 D-0 1277 Dresden Germany C. D. Latham Department of Physics University of Exeter Stocker Road Exeter EX4 4QL UK

W. Lengauer Institute for Chemical Technology of Inorganic Materials Vienna University of Technology Getreidemarkt 9/161 A-1060 Vienna Austria

XXX


J. E. Lowther Department of Physics University of Witwatersrand Johannesburg South Africa S. Luyckx School of Process and Materials Engineering University of the Witwatersrand Johannesburg 2050 South Africa

R. Pottgen Westfalische Wilhelms-Universitat Munster Anorganisch-Chemisches Institut Wilhelm-Klemm-Strasse 8 D-48149 Munster Germany

Z. A. Munir

F. Richter Technische Universitat ChemnitzZwickau Institut fur Physik D-09107 Chemnitz Germany

Facility for Advanced Combustion Synthesis Department of Chemical Engineering and Materials Science University of California Davis CA 95616 USA

R. Riedel Fachbereich Materialwissenschaft Technical University of Darmstadt Petersenstrasse 23 D-64287 Darmstadt Germany

K. G. Nickel Universitat Tubingen Applied Mineralogy Wilhelmstrasse 56 D-72074 Tubingen Germany

K. A. Schwetz Advanced Ceramics Lab Elektroschmelzwerk Gmbh Max-Schaidhauf-Strasse 25 D-87437 Kempten Germany

S. Ozbayraktar De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2135 South Africa

P. K. Sen De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2135 South Africa

C. S. J. Pickles De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2135 South Africa

G. Serghiou Max-Planck-Institute for Chemistry Saarstrasse 23 Mainz Germany


D. Sherman Department of Materials Engineering Technion - Israel Institute of Technology Haifa 32000 Israel M. Sibanda De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2135 South Africa

I. Sigalas De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2 135 South Africa P. K. Sitch Fachbereich Physik Universitat/Gesamthochschule Paderborn D- 33095 Paderbron Germany G. D. Soraru Universita di Trento Dipartimiento di Ingegneria dei Materiali Via Mesiano 77 1-38050 Trento Italy R. S . Sussmann De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2135 South Africa

XXXI

K. Takagi Toyo Kohan Co. Ltd. Tokyo Japan R. Telle Institut fur Gesteinshuttenkunde RWTH Aaachen MauerstraBe 5 D-52056 Aachen Germany S. Veprek Institute for Chemistry of Inorganic Materials Technical University Munich Lichtenbergstrasse 4 D-85747 Garching b. Munich Germany C. J. H. Wort De Beers Industrial Diamond Division Pty Ltd Diamond Research Lab PO Box 1770 Southdale 2135 South Africa A. Zerr Fachgebiet Disperse Feststoffe Technical University of Darmstadt Petersenstrasse 23 D-64287 Darmstadt Germany

List of Symbols

a!

a a!

a a?P, x

P P Y Yi

r(n) "/S

r s

tan 6 Ac AG!98 ASint AT AX & &

i rl rl

0 8

20 6 6 61

62

x x x x x x x

A

absorption coefficient atomic attraction constant growth parameter power absorption coefficient polytypes or phases atomic repulsion constant geometrical factor rake angle secondary ion yield orientational surface energy surface energy width of X-ray reflection microplastic deformability dielectric loss factor concentratation difference Gibbs free energy interfacial entropy temperature change change of size or mass elastic strain emissivity strain rate degree of conversion to nitride fraction of reaction completed angle constant relating tensile strength and hardness X-ray scattering angle entering angle thermal conductivity thermal conductivity of rectants thermal conductivity of products empirical parameter relating bulk modulus and inclination angle layer thickness, mean free path polarity of bond thermal conductivity wavelength X-ray wavelength mean free path

degrees J

J K or "C

degrees J cm-2 s-I bond length degrees m

m m

K-'

XXXIV P P

v 7r

P P P Pm U U

ff UO

ffb

ffij ffS

UY 7 7-

4 4, w, 74 4 X W

(4

7

0 a a a a0

acr ai

a, b a, 6 , c

A A A ABAB, ABCABC A,B,H,HA b B B (B4C)

List of Symbols

C/Ti ratio coefficient of friction Poisson ratio complementary energy J density kg m-3 dislocation density resistivity theoretical density of product conductivity electrical conductivity stress Pa median failure stress Pa fracture strength in bending, modulus of rupture local stress field Stefan-Boltzmann constant J cm-* sC1 KC4 yield stress GPa annealing time help time S angle between crack and tensile stress ternary phases azimuthal angle between polarization vector and substrate direction constraint factor electron affinity wear coefficient average diffusion distance in time r vacancy in crystal structure indent size, half length of diagonal crack length depth of cut equilibrium bond distance critical flaw size depth of lateral crack on erosion heat capacity coefficients crystal unit cell parameters lattice constants area contact area material constant

m m mm m m m

nm m2 m2

stacking sequences stacking positions Burgers vector bulk modulus GPa designation of a dissolved species in, e. g., a liquid designation of a non-stociometric compound (solid solution)

List of Symbols

Ci CP d d d d d d d d

dP

D D D D

DO DH, DC

e E E E EO E" EC EF

4

Ei Eirnax EP

f f f f

h F

XXXV

radius of radial crack m m s-' velocity of light in vacuo interfacial concentration m radius of lateral crack on erosion proportionality constant specific heat per unit volume concentration of impurities heat capacity J g-' K-' bond length A degree of dilution diameter of Brinell impression m m diameter of Vickers impression grain size m height of beam m layer thickness rad spacing of powder diffraction rings pore diameter diameter of Brinell indenter m diameter of median crack m diffusivity size of particles diffusion factor diffusion coefficients at high and low temperature unit electron charge activation energy J binding energy J Young's modulus theoretical Young's modulus E/(l-v2) J composite potential energy Fermi energy J band gap J ion energy J maximum ion energy J potential energy mm mine' feed rate Hz frequency volume fraction Hz Weibull safety factor volume fraction of Z N force statistical failure probability ion flux J external work (linear elasticity) vapor phase shear modulus

XXXVI

i, a

I IBB

4 IRB ji

J k k k0 hog kL

k* kr K K KIC KICO J

1 1 1 1

L Lm,, m

M M M A4 Mi

n n

List of Symbols

strain energy release rate toughness, fracture energy, work of fracture convective heat transfer coefficient indentation depth Planck constant hardness enthalpy enthalpy of formation at 298 K Brine11 hardness Meyer hardness Knoop hardness plastic hardness Vickers hardness average applied compressive stress in hardness GPa fluxes of impinging ions and deposited atoms rank order of test result intensity of black body radiation peak height intensity of real body radiation current density mass flux Boltzmann constant constant for layer growth reaction constant logarithmic rate constant linear rate constant parabolic rate constant reaction coefficient proportionality constant stress intensity factor fracture toughness diffusion path length length of sample long diameter of Knoop impression span of beam defined load maximum load Weibull modulus mode of deformation (I, 11, or 111) metal Mohs hardness weight molecular mass of impurity number order of reaction

N m-l J cm-2 s - ~K-'

GPa GPa GPa GPa test

m2s-l or kg2m-4s-'

Paris exponent for fatigue refractive index stress exponent number of tests average coordination number no. of constraints for coordination number r average coordination number grain size exponent momentum porosity fixed load gas pressure porosity pressure ambient and high pressure confining pressure in powder pressure of nitrogen partial pressure of O2 Porod scattering vector scattering vector activation energy heat of reaction resonance factor heat of transport radius distance from crack tip interatomic distance radius of curvature equilibrium interatomic distance gas constant film growth rate average roughness radius of plastic zone solid phase elastic recovery stoichiometric ratio entropy thickness of window time ChI4? delay time time for wave propagation temperature ternary phase ambient temperature temperature limits

N

Yo GPa Pa Pa Pa

J J g-' J

m m

m

m

Yo Jmol-' K-' S S S

K or "C K or "C K or "C

XXXVIII

UP

V

List of Symbols

adiabatic combustion temperature critical temperature and pressure eutectic temperature absolute melting temperature substrate temperature displacement stoichiometry factor wave velocity internal energy substrate bias voltage elastic strain energy plasma potential surface free energy average velocity machining velocity volume wavenumber longitudinal velocity of sound volume of pores volume volume fraction symbol for vacancy in a chemical formula applied substrate bias voltage volume lost in erosion impact molar volume of metal orientational growth rate molar volumes of product and reactants sample thickness RF power carbon-to-metal ratio coordinate layer thickness liquid phase lattice directions Miller indices

K or "C K or "C K or "C m

J J

v

J m min-' m3 cm-' m s-' m3 V m3

m W

m

List of Abbreviations

3PB 4PB ACC AES AFM APW AR ASEA ASTM b.c.c. b.c.t. BB BET 3C, 4H, 6R CAD CALPHAD cBN CCD CED CMC COOP CN CVD CVI

cw

d.c. DAC DCC DF-TB DH DIN dlC DLC DMO DOS DTA ECH ECR

three-point bend four-point bend amorphous covalent ceramics Auger electron spectroscopy atomic force microscope augmented plane wave antireflection Swedish company American Society for Testing and Materials body-centered cubic body-centered tetragonal black body Brunauer-Emmett-Teller method for determining porosity polytype notations of Sic (C: cubic, H: hexagonal, R: rhombohedral) cathodic arc deposition calculation of phase diagrams model cubic boron nitride charge coupled device cutting edge displacement ceramic matrix composite crystal orbital overlap population coordination number chemical vapor deposition chemical vapor infiltrated continuous wave direct current diamond anvil technique direct coagulation casting density-functional tight-binding method methy ldiethoxysilane Deutsche Industrie Norm diamond-like carbon diamond-like carbon directed metal oxidation density of states differential thermal analysis electron cyclotron heating electron cyclotron resonance

XL


EDAX EELS EP EPMA EPR ERD ERDA ESCA ESK EXAFS f.c.c. FEPA FTIR FWHM FZK GA-XRD

gc

GEC GFRP GGA h hBN HF-CVD HIP HK HOMO HOPG HPHT HPL HPMS HR HR-TEM HSS HV IBAD ICSD ICDD IED IR ISE IS0 ITER JAERI JFM JIS KFM

energy-dispersive analysis of X-rays electron energy loss spectroscopy electroplated electron probe microanalysis electron paramagnetic resonance elastic recoil detection elastic recoil detection analysis electron spectroscopy for chemical analysis Elektroschmelzwerk Kempten extended X-ray absorption fine structure face-centered cubic Federation Europeen des Fabricants de Produits Abrasifs Fourier transform infrared full width at half maximum Forschungzentrum Karlsruhe glancing angle XRD glassy carbon General Electric Company, USA glass-fiber reinforced plastic generalized gradient approximation hexagon a1 hexagonal boron nitride hot filament CVD hot isostatic pressing Knoop hardness highest occupied molecular orbital highly ordered pyrolytic graphite high-pressure high-temperature high-pressure laminate high-pressure microwave source Rockwell hardness high-resolution TEM high-speed steel Vickers hardness ion-beam assisted deposition inorganic crystal structure database international center for diffraction data ion energy distribution infrared indentation size effect International Standards Organization international thermonuclear experimental reactor Japan atomic energy research institute Johnson figure of merit Japanese Standards Keyes figure of merit


LAS LDA LDA LIDT LPI LPSSiC LPSSS LRO LSF LSI LWIR MAK MAS-NMR Me-DLC MMC MOR MOSFET MS MSIB MTES MTF MW MWP-CVD NASA ncTi02 NDE Nd-YAG NEA NICALON NIRIM NMR NRA ORNL PA-CVD PAlC PCS p.p.m. PBC PC pcBN PcD PCS PCT PTC PTES PTFE

XLI

lithium aluminosilicate local density approximation laser diode array laser induced damage threshold liquid polymer infiltration liquid-phase sintered S i c low-pressure solid-state source long-range order line spread function liquid silicon infiltration longer wavelength infrared Maximal zulassige Arbeitsplatz Konzentration magic angle spinning NMR metal-DLC hybrid metal matrix composite modulus of rupture metal-oxide silicon field effect transistor mass spectroscopy mass-selected ion beam methy ltriethoxy silane modulation transfer function microwave microwave plasma CVD National Aeronautics and Space Administration (USA) nanocrystalline titania nondestructive evaluation neodymium-yttrium-aluminum-garnet laser negative electron affinity branded Si-C-0 composite fiber from Nippon Carbon National Institute for Research in Inorganic Materials (Japan) nuclear magnetic resonance nuclear reaction analysis Oak Ridge National Laboratory plasma assisted CVD polyaluminocarbosilane polycarbosilane parts per million periodic bond chain potential cycling polycrystalline boron nitride polycrystalline diamond . polycarbosilane Patent Cooperation Treaty polytitanocarbosilane phen y ltriethoxy silane polytetrafluoroethylene

XLII PP PVD r RBAO RBM rBN RBS RBSN RF RSF RSSC S.C. s.c.cm. SAD SAXS

scs

SEM SERR SHS SIALON SiCAlON Si-DLC SIF SIMS SNMS SP SRO STM ta-C taC TD TCNE TEM TGA TH TRS TZP UHP UPS

uv VAMAS vc VEC VLS

vs

wBN


polymer pyrolisis physical vapor deposition rhombohedral reaction bonded aluminum oxide reaction bonded mullite rhombohedral boron nitride Rutherford back-scattering reaction bonded silicon nitride radio frequency reduced spatial frequency reaction sintered silicon carbide simple cubic standard cubic centimeters small angle diffraction small-angle X-ray scattering Textron process Sic fibers with C core and C surface scanning electron microscope strain energy release rate self-propagating high-temperature synthesis Si-A1-0-N (silicon aluminum oxynitride) fiber SiC-AlN-Al20C composite fiber Si-DLC hybrid stress intensity factor secondary ion mass spectrometry secondary neutron mass spectrometry sintered powder short-range order scanning tunnel microscopy hydrogen-free amorphous carbon tetrahedral amorphous carbon theoretical density tetracyanoethylene transmission electron microscope thermogravimetric analysis triethoxysilane transverse rupture strength (= MOR) tetragonal zirconia polycrystals ultahigh purity ultraviolet photoelectron spectroscopy ultaviolet Versailles Agreement on Materials and Standards vapor phase formation and condensation process valence electron concentration vapor-liquid-solid process vapor-solid reaction wurtzitic boron nitride


XANES XPS XRD YAG YLF

XLIII

X-ray absorption near edge structure X-ray photoelectron spectroscopy X-ray diffraction yttrium aluminium garnet, yttrium aluminate, Y2A15012 yttrium-lithium-fluorite

Introduction: Novel Ultrahard Materials A. Zerr and *R. Riedel

Introduction The synthesis of new materials with hardness comparable to or even harder than diamond is of considerable fundamental and technological interest and is a great challenge to chemists, physicists, and materials scientists. Most of the known ultrahard materials, including diamond and cubic boron nitride, were first synthesized in the 1950s and industrially manufactured using high pressure-high temperature processes [I-31. Extensive research in this domain continues and recently a few new materials have been synthesized or rediscovered as superhard ones (Si02-stishovite [4], cubic Si3N4 [5]). New vapor deposition methods (CVD, PVD, laser ablation etc.) which allow the deposition of diamond, cubic boron nitride and other hard materials films at low temperature and low pressure (i.e. often under metastable conditions) on a variety of substrates have been developed since the early 1980s [&lo]. Diamond and cBN (cubic boron nitride) combine excellent mechanical, chemical, and physical properties. However, owing to its instability at high temperatures, diamond cannot be used, for example, as a cutting tool for steel. Moreover, with increasing temperatures diamond and cBN weaken due to the onset of the transformation to the graphite structure so that above 1100°C in a nonoxidizing atmosphere boron carbide B4C(with a hardness of about 30 GPa) has been identified as the hardest material [I I]. For this reason and because of the need to replace expensive diamond in many other applications, new hard materials with comparable or even superior properties are required. Theoretical work on the carbon nitride C3N4predicted that this compound could have a hardness comparable to or even greater than that of diamond [12,13]. The synthesis of polycrystalline C3N4films was first reported in 1992 [I41 and in 1993 [ 151and has been under extensive investigationsince then, resulting in hundreds of publications on this subject. However, in a few recent publications there has been doubt the thermodynamic stability of theoretically predicted ultrahard C3N4 phases [ 161 and whether their hardness would be comparable to that of diamond [ 171. In consequence, research on the low-temperature synthesis of diamond and cubic boron nitride [18],the search for other possible candidate compounds as well as microstructure design of known materials and their composites [ 19-21] (heterostructures, whiskers, nanocomposites) remain the subject of experimental and theoretical efforts. Microstructure design is addressed in this book by D. Sherman and D. Brandon in Part I, by S. Vepfek in Part I1 and by A. Krell in Part 111. This introduction deals with some of the latest experimental and theoretical developments in the field of novel boron- and carbon-based ultrahard materials as well as with new observations on a class of silicon-based compounds which previously were not classified as ultrahard. * This is a revised version of the paper published earlier [ 1521.

XLVI

Introduction: Novel Ultrahurd Materials

Hard Materials Generally, hard materials are solids with high hardness in the range 8-10 on the Mohs scale of hardness, given by the sequence of minerals which can be scratched by the next (Table 1). Usually, common hard materials are subdivided into compounds with metallic (like TIN or WC), ionic (Al2O3>,or covalent bounding (diamond, Si3N4)[22]. The definition of ultrahard materials is that their hardness values are comparable to that of diamond. Another definition often used in the literature is that the hardness of such materials exceeds 40GPa. Among all known single phase compounds diamond, cBN and probably boron carbides (Bl3C2-Bl2C3)satisfy the latter definition. Accordingly one can expect that novel ultrahard compounds will be found in the isothermal ternary phase diagram BC-N given schematically in Fig. 1. In particular, the carbon nitrides (e.g. C3N4) and the boron carbonitrides (B,C,NZ) have been discussed as substitute materials Table 1. Hardness ranking of minerals and some prominent synthetic ceramic materials according to F. Mohs. In the case of the synthetic materials microhardness values are given in units of the Knoop scale. The microhardness variations result from variations in the grain size, the load of indentation, the phase composition and the used densification techniques. Modified after [ 1521. Minerals/ synthetic materials

Formula

Mohs hardness

Talcum Hexagonal boron nitride* Gypsum Calcite Fluorite Apatite Feldspar Quartz Topaz p-Silicon nitride*

Mg,[(OH)z/Si4O,oI hBN

1

CaS04 2 H 2 0 CaCO, CaF2 Ca5[(F3OH)/(PO&I K[AlSi,O,] Si02 A1z[F2/Si041 BSi3N4

2 3 4 5 6 7 8

Corundumt Titanium nitride* Silicon carbide* %-Silicon nitride Titanium carbide* Boron carbide' Titanium diboride* Boron suboxides Stishovite* Cubic boron nitride' Diamond

t

9

g

3

}

Sic %Si3N4 TIC B4C TiBz B,O Si02 cBN C

Microhardness,' Knoop 100 Wal 0.15-0.30

17 21 26 26-35 28

}

30

10

30-59 33 45 75-100

' The microhardness values are taken from published sources [7] and [102]. The microhardness varations are taken mostly from previous work [25]. * Synthetic material.

Synthetic material or natural mineral.

Microhardness variations for polycrystals' [GPa]

14-26 14-29 20-38 2648 13-32 20-38 19-35

Hard Materials

XLVII

Figure 1. Schematic of the isothermal ternary B C - N phase diagram at temperatures below the decomposition of the stoichiometric compounds and the regimes of composition of some solid solutions discussed. For explanation of the compounds see text. First published in [I 521 and reproduced with permission.

for diamond. One can also expect that metastable dense high pressure phases of the compounds based on silicon or oxygen can belong to ultrahard materials. One of these is the high pressure-high temperature stishovite phase of Si02 with a hardness exceeding 33 GPa [4], which is almost an order of magnitude above that of quartz, the ambient pressure phase of Si02 (Table 1). The second one is cubic Si3N4with the spinel structure, whose hardness may be comparable to that of Si02-stishovite [5].

Hardness Hardness is one of the quantitative parameters that describe resistance of a material towards plastic (irreversible) deformation. Plastic deformations begin when the shear component of the stress applied to a material exceeds some value called the yield stress. There are many ways to create a plastic deformation and consequently many ways to define and to measure resistance of a material towards such deformations. Hardness can be determined in several ways: -

Scratching methods (Mohs, Martens), Grinding methods (Rosiwal). Indenting methods (Vickers, Knoop, Brinell, Rockwell, Shore).

For example, the Vickers hardness, Hv, is defined as the applied load P divided by the surface area of the impression, while the Knoop hardness H K is derived from the load P divided by the projected area of the impression: 1854.4P Hv = d2 ’ 14 229 P HK=-

12

XLVIII

Introduction: Novel Ultrahard Materials

where d is the diagonal of the square-based diamond pyramid of the Vickers indenter and I is the long diagonal of the rhombus-based diamond pyramid of the Knoop indenter. In the case of applied test loads below 1.96N the determined hardness is defined as microhardness and depends on the applied amount of load [23]. Generally, the unit of the hardness measured is given in kgmm-* or in GPa (1000 kgmmP2 = 9.81 GPa). The Vickers or Knoop hardness can be also expressed in terms of Mohs hardness, M , by the following expression [23]:

H v = 3.2M3.

(3)

However, this expression cannot be applied for superhard materials like diamond. Each testing method yields different hardness values for one material. Thus, in order to compare hardness values of different materials, the specific test method and the test conditions have to be described carefully. Moreover, the hardness of single crystals depends for many compounds on the crystallographic plane tested and for the Knoop indentor additionally on the orientation of the indentor’s long axis relative to the crystallographic axis of the examined crystal. For example, in experiments on single-crystalaSi3N4the Vickers hardness measured on different crystallographic planes varied by more than 34% and the Knoop hardness measured on the same crystallographic plane by 7% [24]. Hardness of polycrystalline materials is strongly influenced by: residual stresses, toughening phases, microstructural textures, the grain size, the applied load and the porosity as well as by the structure and composition of grain boundaries. As a consequence, the hardness of polycrystals may significantly differ from that of the single crystal counterpart. For example, due to the above reasons the measured microhardness of polycrystalline TIC varies between 1200 and 3250Hv or by more than 270% and that of polycrystalline Si3N4between 1500 and 4800 (Hv)or by more than 300% [25]. In Table 1, the microhardness values and their variations for some prominent synthetic ceramic materials are listed ranked according to the Mohs scale. The above topics are discussed in more detail in Part I by D. Sherman and D. Brandon, and by A. Krell. There are few theoretical attempts to describe hardness quantitatively in terms of the elastic bulk or shear moduli of an ideal solid. The elastic moduli and the corresponding hardness of a compound or element are strongly related to its molar volume, to its chemical bonding, and to its crystal structure. Basically, it can be stated that the higher the hardness the lower the molar volume and the more covalent the bonding of the material. Liu and Cohen suggested in their paper on hypothetical PC3N4 [12] that ‘on the microscopic level, for ideal systems, hardness is determined by the bulk modulus’. To estimate the bulk modulus they used an empirical model developed earlier [26], where the bulk modulus scales as a homopolar energy gap divided by the volume of the bond charge. The resulting relation gives theodependence of the bulk modulus B (in GPa) as a function of the bond length d (in A) and the empirical parameter A: 1971 - 220X (4) B= d3.5 ’

Hurd Materials

XLlX

Table 2. Calculated equilibrium volumes, bulk moduli, and cohesive energies of some postulated ultrahard materials. The calculated and/or experimental values of diamond, lonsdaleite, cubic BN, and !3Si3N4are given for comparison. The experimental volume and density values are derived from crystal structure data. CH.6 = all sp2-bonded hexagonal carbon phase 26,73,74]; Cb,c,t,.4= all sp*-bonded body-centered tetragonal carbon [26,72,76]; BNh.c.t.-4= all sp -bonded body centered

1

tetragonal BN [119]. First published in [152] and reproduced with permission. Material

Crystal symmetry

Vo

nm3/ atom]

Density [g/cm3]

B, [GPa]

Ecoh

444 (calc) 435 (exp) 440 (calc) 367 (calc) 369 (exp) 265 (calc) 256 (exp) 427 (calc) 372 (calc) 362 (calc) 268 (calc)

8.17/atom (calc) 7.37/atom (exp) 8.14/atom (calc) 13.2/BN pair (exp)

Cdiarnond

cubic

5.67 (exp)

3.51 (exp)

Clonsdalelte cBN

hexagonal cubic

6.47 (exp) 5.90 (exp)

3.52 (exp) 3.45 (exp)

hexagonal

10.42 (exp)

3.19-3.20 (exp) 3.49 3.17 2.97 2.82

hexagonal hexagonal tetragonal tetragonal

6.25 6.29 6.72 7.37

Refs

[eVI

74.3/unit cell (calc) 81.5/unit cell 9.06/atom 8.47/atom 13.36/BN pair

X is 0 for homopolar solids of group IV elements, and 1/2, 1, and 2 for heteropolar solids of group 111-IV, 111-V, and 111-VI elements, respectively. From this, it is evident that increasing ionicity results in a decrease in B. This expression has been further improved to account for non-octet compounds such as Si3N4 by using the average coordination number N, # 4 [26]:

B=-

(N,.) (1971 - 220X) 4 d3.5

In Table 2, some experimental and calculated structural properties such as the equilibrium volume, bulk modulus, and cohesive energies of candidate materials with potentially ultrahigh hardness are summarized. The equilibrium volume corresponds to the minimum of the total crystal energy as a function of volume [26]. The cohesive energy of a crystal represents the difference between the energy of the isolated constituent atoms or molecules and the energy of the crystal at absolute zero temperature [27]. It is evident from Table 2 that the higher the cohesive energy and the lower the equilibrium volume the higher the bulk modulus of the material. On the other hand materials deform plastically only when subjected to shear stress. According to Frenkel analysis, strength (yield stress) of an ideal crystalline solid is proportional to its elastic shear modulus [28,29]. The strength of a real crystal is controlled by lattice defects, such as dislocations or point defects, and is significantly smaller then that of an ideal crystal. Nevertheless, the shear stress needed for dislocation motion (Peierls stress) or multiplication (Frank-Read source) and thus for plastic deformation is also proportional to the elastic shear modulus of a deformed material. Recently Teter argued that in many hardness tests one measures plastic deformation which is closely linked to deformation of a shear character [17]. He compared Vickers hardness data to the bulk and shear

L


Figure 2. Scattering of the Vickers hardness for a common set of hard materials when compared with bulk modulus (gray region) and shear modulus (black region). Following this comparison Teter suggested that ‘the shear modulus is a significantly better qualitative predictor of hardness than the bulk modulus’ [ 171.

moduli from over thousand measurements and found that the shear modulus of polycrystalline aggregates is a significantly better qualitative predictor of hardness than the bulk modulus (Fig. 2). Experimental bulk moduli can be obtained from the measurement of lattice parameters and volumes as a function of pressure [30]. The single crystal elastic moduli can be measured using the Brillouin spectroscopy, inelastic neutron scattering, ultrasonic measurements or the Schaefer-Bergmann method [311. Once the single crystal moduli are known one can derive the bulk B and shear G moduli of a polycrystalline material [32].

Carbon-based Hard Materials The most common allotropes of elemental carbon are graphite and diamond. Graphite crystallizes in a sheet structure with hexagonal symmetry and sp2 hybridization of the trigonally coordinated C-atoms whereas diamond exhibits a cubic lattice related to the zinc blende structure with sp3 hybridization of the tetrahedrally coordinated C-atoms. Less common carbon polymorphs are lonsdaleite with the hexagonal wurtzitic structure 1331, and the rhombohedral polymorph called pdiamond (3R polytype of wurzitic structure) [34,35]. The rhombohedral polymorph, which was reported as birefringent lamellae in natural diamonds, has not yet been obtained in significant amounts needed for detailed studies [35]. Graphite,

Carbon-based Hard Materials

LI

Figure 3. The different carbon modifications: (a) hexagonal graphite; (b) cubic diamond; and (c) hexagonal lonsdaleite. First published in [ 1521 and reproduced with permission.

diamond, and lonsdaleite are, however, well investigated and their structures are shown in Fig. 3. In lonsdaleite, comprised of the same tetrahedral configuration as in diamond, the planes of six-membered carbon rings are in the chair and boat conformation and are stacked in an ABAB sequence while in diamond, these planes exhibit the chair conformation exclusively and are arranged in an ABCABC sequence. The recent discovery of the c 6 0 molecule shown in Fig. 4, which is the most prominent representative of the fullerene family, led to a new type of carbon allotrope. In fullerenes, the carbon atoms form spherical clusters comprised of six- and five-membered rings of the general composition C20+2m (m = 0 , 2 , 3 . ..) where m is the number of hexagons. The c 6 0 fullerene has been predicted by Kroto and Smalley since 1985 [36-38] and was isolated first in macroscopic quantities by Kratschmer et al. in 1990 [39-41]. At present, the spherical carbon clusters and their derivatives are discussed with respect to their application in material science [42,43]. Furthermore, a variety of new polymeric carbon networks have been postulated to exhibit interesting materials properties [41,44]. Recently, carbon nitrides such as the hypothetical compound C3N4 have been also considered for the synthesis of ultrahard materials [ 14,151.

LII


Figure 4. Molecular model of the fullerene C60.The gray carbon atoms illustrate one of the 12 pentagons present in C6,,. First published in [152] and reproduced with permission.

Diamond Besides the aesthetic appearance of diamond in the form of gemstones, diamond is the hardest material known, and is, therefore, on the top of the Mohs scale and has the lowest molar volume (3.4cm3) of any material. Owing to this unique property, diamond is used for many technological applications such as abrasion, cutting, and polishing. Diamond also has the highest thermal conductivity (2000 W m-' K-') of any material at room temperature, four times as high as the value of Cu or Ag, is a good electrical insulator, it has a small dielectric constant and exhibits a high electron/hole mobility. Boron doped diamond (Typ IIb) exhibiting a p-type conductivity has been known for a long time. Production of diamond with a reasonable n-type conductivity was one of the most difficult tasks in diamond synthesis. Koizumi and coworkers recently obtained diamond films exhibiting n-type conductivity by using microwave enhanced plasma CVD with phosphine PH3 as a donor source for homoepitaxial deposition of n-type diamond 1451. Therefore, diamond will be a most important candidate material for future applications in electronic devices. In addition, diamond has low coefficients of friction and thermal expansion, high chemical and corrosive resistance towards most acids and oxidizing substances, it is transparent to visible and infrared light, withstands ionizing radiation and can, therefore, also be used as window or lens material or as a protective coating for this application. Conventionally, diamond is synthesized at high pressure and high temperature according to the method developed by the General Electric Company in the 1950s


LIII

[l, 461. Synthetic diamond is now commercialized and is utilized for cutting, grinding, and polishing. Recent developments in the high-pressure synthesis technique allow growth of diamond single crystals up to 25 carats in weight [47]. The high-pressure synthesis of diamond is described in more detail by M. W. Bailey et al. in Part 111, while hydrothermal synthesis is discussed in Part I1 by K. G. Nickel et al. Since the mid-1980s the vapor deposition synthesis of diamond films has attracted increasing scientific and industrial interest. At present, vapor-grown diamond products are commercially viable as thin-film-coated cutting tools and freestanding thick-film cutting tools, as substrates for thermal management application and radiation detectors, as optical windows and for production of high-range audio-speaker diaphragms. Commercialization of the vapor-grown diamond became possible due to the cost reduction below $5/carat as a result of the dramatic increase in growth rate and of improvements in energy-use efficiency [481. Advanced diamond films are synthesized using the gas-phase decomposition of volatile carbon sources such as methane (CH4), acetone (H3C-CO-CH3), carbon monoxide (CO), acetylene (HC-CH), or adamantane (CI0Hl6)and is conducted under temperature and pressure conditions where graphite is the stable polymorph (Fig. 5). The most important methods for producing CVD-diamond under metastable conditions are: -

Microwave Plasma Assisted Chemical Vapor Deposition (PACVD) [49,50], Heated Filament Assisted Chemical Vapor Deposition (HFCVD) [5 13,

Figure 5. Pressure and temperature conditions of the diamond synthesis: (a) shock wave production of diamond; (b) high temperature, high pressure regime for the synthesis of diamond; (c) catalytic region for diamond formation; (d) chemical vapor deposited diamond; and (e) transformation of CG0into diamond. The most recent review of the P, T phase diagram of carbon can be found elsewhere [151].

LIV -


Direct Current (DC) Plasma Jet Deposition [52,53],and Oxygen-Acetylene Torch [54].

The chemical-vapor deposition of diamond films and their applications are reviewed by C.-P. Klages and by R. S. Sussmann et al. in Parts I1 and 111, respectively. To date the most effective CVD method (with the greatest mass deposition rate) is based on the hydrogen/hydrocarbon gas mixtures. In this method diamond is formed for kinetic reasons according to the simplified reaction:

H2

Cdiamond f 2H2. In a typical process, the hydrocarbon precursor containing more than 95% H2-gasis passed through a plasma (700-1000°C) or over a heating filament ( T M 2000") at less than atmospheric pressure. The reaction product, solid carbon, is then deposited on a substrate heated at 800-1000°C and contains both graphite and diamond. In the presence of H2-gas, the formation of unsaturated carbon nuclei and hence the growth of graphite is suppressed by atomic hydrogen formed under these conditions and is due to the reaction of the solid carbon with hydrogen radicals (H') giving volatile hydrocarbons. The reaction rate of Cgraphite with H' is about 20 times as high as the rate of diamond. Therefore, the formation of diamond is promoted. Growth rates of up to 0.9 mm h-' [55] or 20 carats per hour [48] have been achieved on a variety of substrates making the CVD-diamond a highly interesting material for technological applications. Diamond films have also been deposited from hydrogen-free gas mixtures such as C60/Ar using microwave apparatus. The deposition rates are, however, significantly lower. Additionally, use of hydrogen-poor plasmas results in nanocrystalline (3-10nm) diamond films in contrast to micrometer sized crystals from the hydrogen-rich plasmas [56]. There are continuous theoretical attempts to describe the mechanism of CVDdiamond synthesis including mechanisms of surface reactions, diamond nucleation, and film growth. To achieve this aim various phenomenological or first-principles models, molecular dynamics and Monte Carlo simulations have been used [57,58]. While the dominant substrate for low pressure growth of diamond films is single crystal silicon, several different materials such as Ta, Mo, W, Cu, Au, Ni, Sic, SO2, and Si3N4have been coated with polycrystalline diamond or diamond-like deposits. The nucleation rates and the adhesion of the generated films vary with the type of substrate material, which is related to the formation of intermediate carbide layers such as Sic or WC. Identification of vapor grown diamond is accomplished by X-ray diffraction or electron diffraction using a transmission electron microscope (TEM) and by Raman spectroscopy. Diamond coatings show the characteristic first-order Raman peak for diamond at 1332cm-' [59]. In many cases, an additional peak located at ~ 1 5 5cm-' 0 occurs in the Raman spectrum which is attributed to a highly disordered carbon phase (diamond-likecarbon) [60]. This diamond-like carbon contains sp3 and sp2 hybridized C-atoms which do not coincide with graphite. Recently a new simple method for the synthesis of diamond, called metallic reduction-pyrolysis-catalysis, was reported. In this method diamond powder was CH4


LV

obtained as a product of reaction of carbon tetrachloride with metallic sodium in an autoclave at 700°C, where CC14 served as a carbon source [18]: CC14

700°C + 4NaCl- Catalyst

C

+ 4NaCl.

As a metal catalyst an Ni-Mn-Co alloy (70 :25 :5 wt%) was used. About 2% of the carbon yield was well crystallized diamond and the residue was most probably in an amorphous form, as was evident from X-ray powder diffraction and Raman spectroscopic measurements.

Diamond-like and Amorphous Carbon In the course of the research into the synthesis of diamond under metastable conditions, a new class of materials, diamond-like carbon and hydrocarbon phases, have been discovered. The diamond-like hydrocarbons (aC :H) are generated by the R F self-bias method, a technique derived from R F sputtering, developed by L. Holland [61,62]. The molecular ions, C,H;, derived from the particular hydrocarbon used in the plasma, disintegrate upon colliding with the substrate surface resulting in the formation of diamond-like hydrocarbon films [63]. The main structural feature of diamond-like hydrocarbons is the presence of both sp3- and sp2-carbon. Solid-state NMR-investigations revealed that the material contains sp3-carbon atoms of the form -C-H or H-C-H [6]. No quaternary carbon atoms could be detected while methyl groups, -CH3, were found to be present in small amounts. In addition, resonant Raman spectroscopy showed small 7r-bonded clusters of sp2 sites [64]. It was found that conducting the R F self-bias experiment with hydrocarbons in the presence of H2 results in production of small diamond clusters. This experimental result indicates that aC : H could be considered as an intermediate compound formed during the transformation of hydrocarbons to diamond in the plasma processes [65]. The mechanical properties of diamond-like hydrocarbon films strongly depend on the hydrogen content. Increasing the ratio of sp3 sites results in decreased hardness values and lower wear resistance [66]. This phenomenon is attributed to the fact that hydrogen is monovalent and cannot contribute to formation of a covalently bonded and highly cross-linked carbon network as found in the diamond structure. However, microhardness values in the range 30-50 GPa have been measured in aC :H films, values significantly higher than the hardness of Sic (20-38 GPa). In contrast, diamond-like carbon (aC) is free of hydrogen or contains only little hydrogen and does not represent microcrystalline diamond [67,68]. The preparation of aC has been accomplished by magnetron sputtering [68]. Again, the structure and properties of the aC phase is related to the ratio of the sp3/sp2 sites. However, hydrogen is not responsible for the stabilization of the sp3-hybridizedcarbon atoms. Here, the mechanical constraints resulting from random covalent networks, which can be calculated using the constraint-counting method developed by J. C. Phillips [69], can be reduced by the generation of medium-range or long-range order by clustering or crystallization.

LVI


Another class of novel carbon material is amorphic carbon which can be obtained by laser ablation [70]. This amorphous phase contains sp3 -bonded nodules of pure carbon embedded in a matrix of other types of carbon phases. The mean particle size of the nodules is 20 nm, the density of the films is between 1.85 and 2.89 g crnp3.The internal stresses derived by distortions of the bonding angles in random networks are counterbalanced by the high surface to volume ratios of the nanoparticles. Substrate discs 30mm in diameter were coated with uniform layers of amorphous carbon by laser ablation of a graphite feedstock. Maximum growth rates of about 0.5 pm h-' and film thicknesses of up to 5 ym have been achieved on different substrate materials such as Si, Ge, ZnS, Cu, stainless steel, quartz, glass, and plastics. One outstanding property of laser ablation in comparison with the CVD process for the preparation of carbon films is that the substrate temperature does not exceed 35°C during the whole deposition procedure. In contrast to the CVD diamond, which develops polycrystalline columnar structures, the laser-ablated amorphous carbon is deposited in the form of self-seeding nodules. Diamond-like carbon films are discussed in more detail in Part I11 by C.-P. Klages and K. Bewilogua.

Novel Hypothetical Three-dimensional Carbon Phases A dense carbon phase with a calculated density of 4.1 g cmP3was predicted by N. N. Matyusenko and V. E. Strel'nitzkii in 1979 [71]. Due to this high density value, ultrahigh hardness of this carbon material is expected. In addition, several different hypothetical three-dimensional polymeric carbon networks with interesting materials properties have been proposed. The most relevant ones with respect to the potential of high hardness are the following carbon networks. (i) In 1983, R. Hoffmann published a metallic allotrope of carbon in which layers of infinite polyene chains are connected by bonds parallel to the c-axis. Each layer is rotated by 90" about the c-axis (Fig. 6a). The unit cell of the crystal structure is primitive body centered tetragonal and contains four atoms (b.c.t.-4 structure). In this network, carbon is present in the form of trigonal sp2 atoms. However, in contrast to graphite where the carbon atoms are arranged in a two-dimensional sheet structure, the trigonal carbon atoms of the b.c.t.4 structure form a three-dimensional network [72]. (ii) A similar carbon allotrope was proposed by M. A. Tamor and K. C. Hass in 1990. In this structure, the chains of the trigonal sp2 carbon atoms are rotated by 60" rather than 90" between the layers. The three-dimensional all sp2 phase of carbon has a hexagonal Bravais lattice with six atoms in the primitive unit cell and is known as the H-6 structure (Fig. 6b). However, theoretical calculations indicate that H-6 carbon is unstable with respect to the transformation to diamond. This instability is probably due to the short distances between carbon chains [73,74]. (iii) The 'super diamond' structure shown in Fig. 7 can be derived from tetraethynylmethane C(C2H)4. In principle, the polymeric network of the super diamond could be obtained by oxidative coupling of the methane derivative [41,75].

Curbon-based Hard Materials

LVII

Figure 6. (a) Model of a section of the hypothetical polymeric network of the body centered tetragonal structure (b.c.t.4) of carbon and BN suggested by R. Hoffmann et al. [72]. (b) Model of a section of the hypothetical hexagonal structure (H-6) of carbon postulated by M. Tamor and K. Hass [73,74]. In both structures, each carbon atom is considered to be trigonally coordinated and sp2-bonded, exclusively. First published in [ 1521 and reproduced with permission.

(iv) The carbon skeleton of allene, 2HC=C=CH2, could also be utilized for the formation of a three-dimensional carbon phase [41]. In this case, the oxidative coupling of allene could provide a carbon structure with orthogonally arranged chains of sp2 carbon atoms (Fig. 8). A similar carbon phase could be obtained by the coupling of tetraethynylallene, (HCGC)~C=C=C(C=CH)~[75].Here, the carbon chains are connected by butadiene units instead of single bonds. Both the b.c.t.-4 and the H-6 phase of carbon are discussed in terms of intermediate structures formed during the chemical vapor deposition of diamond. Furthermore, these types of structures could play an important role in diamond-like phases. The calculations of the structural properties of the b.c.t.-4 and H-6 carbon phase using

LVIII


Figure 7. Hypothetical super-diamond structure suggested earlier [41,75]. First published in [ I 521 and reproduced with permission.

the first-principles pseudopotential total-energy method revealed high bulk moduli B of 350 GPa and 372 GPa, respectively [26,76]. The extraordinary high values of B are in the range of cBN, the second hardest material known. Presently, the synthesis of new carbon phases by the coupling of unsaturated molecular organic compounds such as allenes or alkynes is being intensively investigated in several laboratories.

Figure 8. Hypothetical allene structure of carbon postulated earlier [41,75]. First published in [I521 and reproduced with permission.


LIX

Fullerenes It has been shown that fullerenes or their derivatives can exhibit very interesting chemical, electrical, magnetic, and mechanical properties. Besides, for example, the superconductivity, which has been experimentally verified for the alkali metal fullerides M 3 C 6 0 with M = K or Rb [77], fullerenes can serve as a starting material for diamond synthesis [78,79] and may exhibit high hardness themselves under high pressure conditions [go]. In 1992 M. Regueiro et al. reported on the transformation of c 6 0 molecules into polycrystalline diamond by nonhydrostatic compression at room temperature [78]:

& c60

P > 20 GPa/20"C

' Cdiarnond.

In contrast c 6 0 withstands hydrostatic pressure up to 20 GPa [81]. However, the football molecules seem to be unstable towards uniaxial or shear stresses whereas they are stable under isotropic stress where the spherical molecules are homogeneously deformed. In a dense arrangement of c 6 0 spheroids, 48 of the 60 carbon atoms have a quasi-tetrahedral coordination which is required in the diamond structure. Only small structural rearrangements are then necessary for the transformation into diamond [78]. R. S. Ruoff and A. L. Ruoff proposed that c 6 0 is stiffer than diamond [go]. This result has been estimated from the calculated bulk modulus of individual c 6 0 molecules. The calculations revealed a bulk modulus B = 843 GPa which is nearly twice the experimental value of 441 GPa determined for diamond. Since solid c 6 0 forms a van der Waals crystal with f.c.c. lattice and the distance between the single c 6 0 molecules is about l.Onm, B would be relatively small under normal conditions. However, when the individual carbon spheres are compressed until they touch each other, the bulk modulus of the crystal would become in the range of that of the molecule. Taking into account a volume filling factor of 74% valid for the f.c.c. structure, B is calculated to be 624 GPa using the volume fraction rule [80]. Therefore, diamond might be outranked by the hardness of c 6 0 crystals. This result is of great fundamental interest, despite the fact that the extraordinary hardness value will be only achieved at high pressures (20 GPa).

Carbon Nitride (C3N4) Theoretical predictions that covalent solids comprised of carbon and nitrogen are excellent candidates for extraordinary hardness [82,83] lead to a strong interest in the scientific community which resulted in hundreds of theoretical and experimental studies on this system. In this chapter we only depict some results on C-N compounds without an attempt to be exhaustive. The first of such predicted structures, namely PC3N4the structural analogue of the well-known hexagonal PSi3N4, was suggested almost 10 years ago [12]. A number of other structures of C3N4 were proposed later: graphite-like (P-6m2) and cubic defect-zincblende (P-43m) [84], the hexagonal analogue of aSi3N4

LX


(P31c), pseudocubic (P-42m), and cubic (I-43d) [13]. Bulk moduli (and hardness) of most of these compounds were predicted to be comparable with and the last one even to exceed that of diamond. The hardness of the hypothetical carbon nitride, pC3N4, has been comprehensively discussed by A. Liu and M. Cohen [12,83]. In the case of &N4, the bulk modulus is estimated to be higher than that of diamond (483 GPa) by semi-empirical methods using Eq. (4) with a bond length d = 0.147 nm and an ionicity factor of X = 0.5. From ab initio calculations of the total structural energy of C3N4 using the first principles pseudopotential total energy method, a somewhat lower value of 427GPa for B is derived. This finding can be attributed to the structural difference between this material and the completely tetrahedrally coordinated zinc blende structure from which the scaling law (Eq. (4)) was developed [12,83]. In contrast, the compound pC3N4 is considered to have the pSi3N4 structure and is only a partially tetrahedrally coordinated structure. The pSi3N4structure is related to the well known hexagonal structure of the beryllium silicate Be2Si04.Therefore, the unit cell of PC3N4 is hexagonal and contains two formula units (Fig. 9). The carbon atoms are sp3 hybridized and tetrahedrally surrounded by N atoms, whereas nitrogen has a planar trigonal coordination (sp3 hybridization). The overall structure can be described by a network of CN4 tetrahedra and NC3 sites. Teter and Hemley suggested the cubic high pressure phase of Zn2Si04denoted as willemite-I1 as another possible structure of a superhard C3N4[13]. This suggestion was based on the fact that the structure of the zero-pressure phase of Zn2Si04 (willemite-I) is isomorphous with the structure of Be2Si04. Using first principles pseudopotential total energy techniques they calculated energetics, structure, and some physical properties of the cubic willemite-I1 structure as well as of four other candidate structures of C3N4. They found that aC3N4 (the analogue of aSi3N4) is energetically favorable at zero-pressure and the cubic willemite-I1 structure might be a high pressure phase of C3N4. The predicted transition pressure from a-to the cubic willemite-I1phase was 68 GPa. The calculated bulk modulus of this cubic C3N4was 496 GPa, which is significantly higher than the measured bulk modulus of diamond (442 GPa) or predicted moduli of a- and pC3N4 (425 GPa and 450 GPa, respectively). This result suggested that the hardness of cubic C3N4with willemite-I1 structure may exceed that of diamond [13]. However, in his later work Teter revised this conclusion to the contrary according to the almost linear relation he has found between shear moduli of polycrystalline compounds and their hardness [17]. The calculated shear moduli of the polycrystalline p- or cubic C3N4 are only 60% of the shear modulus of diamond and smaller than that of cubic BN. Accordingly the hardness of these carbon nitrides (once synthesized) will probably not exceed that of cubic BN [17]. Further discussion of possible structures of C3N4can be found in Part I in the chapter of J. E. Lowther. Synthesis of carbon nitrides has been attempted for example by the pyrolysis of organic nitrogen-containing compounds such as melamine-formaldehyde resin and subsequent shock compression of the obtained residues [85]. However, no tetrahedral carbon nitride could be detected in the reaction products. In a similar approach, L. Maya et al. obtained a carbon nitride with trigonally coordinated carbon by the pyrolysis of organic compounds at 700°C and 225 MPa pressure


LXI

a)

Figure 9. Theoretically suggested crystal structures of carbon nitride C3N4(filled circles represent the carbon atoms): (a) Hexagonal structure corresponding to that of PSi3N4or phenacite, Be2Si04. (b) Cubic structure corresponding to that of willemite-11, the high-pressure phase of Zn2Si04[ 131. First published in [ 1521 and reproduced with permission.

[86]. Plasma decomposition of CH4 and N2 provided amorphous solids containing C, N, and H, exclusively [87]. High pressure pyrolysis of tetracyanoethylene, (CN)2C=C(CN)2, gave C4.66Nwith graphitic structure [88]. Similarly, the gas phase decomposition of pyridine in the presence of chlorine at 800°C provided the carbon nitride C5N with a hexagonal sheet structure [89]. Martin-Gil et al. reported on the synthesis of cubic carbon nitride with the zinc blende structure by pyrolysis of N,N-diethyl- 1,4-phenylene-diarnmoniumsulphate at 800°C in an N2 atmosphere using Se02 as a catalyst [90]. The material obtained contained both an amorphous matrix and polycrystalline regions with crystallites of 5-50nm in size. The sample as whole had a composition CN, with 0.2 < x < 0.5. The composition of the embedded crystallites was estimated using EELS (electron energy loss spectroscopy) data to be C,N,,, with x and y between 3 and 4. The

LXII

Introducfion: Novel Ultrahard Matericils

cubic zinc blende structure of the crystallites was derived from TEM measurements and confirmed by extended energy loss fine structure analysis [90]. Besides the preparation of polycrystalline C- and N-containing thin films by using direct current magnetron sputtering with nitrogen as the sputtering gas [ 141, the synthesis of C3N4 was experimentally verified in the work of C. Niu et a/. [15]. In this paper, the preparation of the carbon nitride was realized by pulsed laser ablation of graphite targets in the presence of a nitrogen source. The C-N films deposited on Si (001) and on polycrystalline Ni were characterized by Rutherford backscattering (RBS). photoelectron spectroscopy (XPS, ESCA), and electron diffraction measurements in the TEM mode. The analysis of the diffraction data (convergent beam electron diffraction) clearly showed that the reflections were in accord with that expected for the PC3N4 structure. Furthermore, the spectroscopic analyses revealed the presence of C and N in the deposited films. The bonding in these solids is covalent. The analysis of the chemical composition of the films determined by RBS exhibited nitrogen contents of up to 45% which is the highest amount measured in any binary C-N solids to that time. The nitrogen content did not decrease after thermal annealing in flowing N2 up to 800°C indicating the relative stability of the C-N material. Preliminary investigations of the hardness showed that the films could not be scratched by hard metals [15]. Thin carbon nitride films containing small crystals were deposited by Yu e t a / . on Si wafers in the (100) orientation using R F diode sputtering of graphite target with pure N2 [91]. The films were about 1 pm thick with a nominal composition of Co6N0.300.1 and contained polycrystalline grains of 0.5-1 pm in size with higher nitrogen content. Many of the lattice spacings obtained from the ring patterns measured with TEM matched those of the theoretically predicted PC3N4 [91]. Analogous results were reported by Fernandez et al. who used a dual ion beam sputtering system to deposit C-N films with a nitrogen content of about 44% on KCI substrates [92]. Preparation of crystalline carbon nitrides was reported to be possible by use of chemical vapor deposition techniques. Films containing a-,PC3N4 and an unknown C-N phase were obtained on nickel substrates from the gas mixture of nitrogen and methane in a hot-filament CVD apparatus [93]. The composition of the grown films close to the stoichiometric value N / C = I .33 was measured using energy-dispersive X-ray analysis and Auger electron spectroscopy. The lattice constants of synthesized SI- and (jC3N4agreed with the theoretically predicted values. The authors have not observed bands of diamond or graphite in Raman spectra of the synthesized films. Presence of amorphous C-N compounds was ruled out after analysis of the films’ morphology [93]. Plasma enhanced CDV process was applied by He and Chang for synthesis of PC3N4 on a silicon wafer as a substrate [94]. As a precursor they used a mixture of methane, hydrogen, and nitrogen. The nitrogen concentration was varied between 0 and almost 75%. The strongest evidence for synthesis of BC3N4 at the highest N2 content was obtained from the TEM measurements. The lattice spacings calculated from the powder diffraction rings agreed with the theoretically predicted dspacings of PC3N4. However, contamination of the deposited carbon nitride by iron carbide or nitride most probably in amorphous form and possible presence of amorphous CH, double bounds C=C or C=N was also reported [94].


LXIII

According to the theoretical calculations, the synthesis of dense C3N4structures should be possible by applying high pressures and temperatures [12]. The simplest possible starting materials for synthesis of carbon nitrides are elementary carbon and nitrogen. Nguyen and Jeanloz used different forms of carbon (graphite, amorphous carbon, and c 6 0 fullerene) and nitrogen for their high pressure synthesis experiments where the samples were heated by an Nd :YAG laser [95]. After heating of the c 6 0 N2 mixture at 30 GPa to 200CL2500 K they obtained a crystalline material at ambient conditions, the X-ray powder diffraction pattern of which did not match any of the known carbon phases, the carbon nitrides prepared by deposition techniques, or the theoretically predicted PC3N4.The measured d-spacings can be better described by a cubic cell with lattice parameter a = 0.77 nm. Similar results were obtained with the use of other carbon precursors. Presence of more than one phases was not excluded. The chemical composition of the synthesized material was not determined [95]. Molecular precursors containing carbon and nitrogen were used for synthesis experiments in the Nd :YLF laser heated diamond cell. Tetracyanoethylene (TCNE) with the stoichiometry C6N4 was investigated to 42 GPa [96,97]. EDX and EELS measurements showed increase of the nitrogen content in the treated sample with pressure: After heating at 20GPa the nitrogen content was 24% and by 42GPa the nitrogen content reached 38%, which is the composition of the starting material. However, at all pressures crystalline carbon was formed in addition to the amorphous sp2-bonded carbon nitride. It was also found that the amount of carbon increases with the heating time [96,97], indicating need of higher pressure for synthesis of single phase carbon nitrides. Such experiment was performed on a mixture of TCNE (C6N4) and cyanuric triazide (C3N12) with a nominal composition C3N4 at 119GPa and 2500°C [97]. X-ray powder diffraction spectra showed six new lines, which were indexed to a tetragonal cell with a = 0.7815 nm and c = 0.6219 nm. The stability of another carbon nitride precursor material paracyanogen (pCN) at high pressures and temperatures has been investigated in detail [98]: The decomposition temperature of the precursor was found to increase with pressure from 550°C at 3GPa to 750°C at 20GPa. This work demonstrated that there is a well defined line for decomposition of pCN and the decomposition temperatures are moderate. It was likely that the carbon in pCN retained sp2 hybridization to the highest pressure of the work. Additionally, above the decomposition temperature the authors obtained crystalline graphite even in a region of the carbon phase diagram where diamond was stable. Thus the pCN precursor, with sp2 carbon bonding, favors the formation of the carbon phase with sp2-bonded atoms instead of thermodynamically favored sp3-bonding [98]. Successful synthesis of a bulk crystalline carbon nitrogen was reported by use of an organic precursor containing carbon, nitrogen, and hydrogen in [99]. After heating this precursor to 400°C at 6 GPa the author obtained a three phase-agglomerate consisting of poorly ordered graphite, a-carbyne and a hexagonal phase with lattice spacings close to that of the predicted PC3N4. The most significant problems in almost all experiments on synthesis of carbon nitrides are as follows:

+

LXIV


(i) The samples contain many phases often with different chemical composition. (ii) Crystalline carbon nitrides have been obtained in small amounts and mostly Xray or electron powder diffraction data were presented. Such data do not always allow unambiguous determination of the structure especially when other phases were not identified. (iii) Small sample amounts and presence of other phases is a serious hindrance for accurate determination of chemical composition of the obtained crystalline carbon nitrides. Improvements in the synthesis techniques and analytic equipment may help to solve these problems. It also has to be mentioned here that the stability of some of the predicted carbon nitrides was questioned in a number of studies based on chemical and thermodynamical grounds. A recent discussion of this subject and further references can be found elsewhere [ 171.

Boron-based Hard Materials Carbon-based materials such as diamond or diamond-like carbon have several disadvantages. Among them, the low oxidation resistance of diamond is one of the most important drawbacks for its technological applications. Boron-based hard materials of the binary or ternary systems B-C, B-N, B-0, or B-C-N might improve this property significantly. For example, diamond can be used up to about 600°C in air environment, whereas cubic boron nitride (cBN) withstands oxidation up to 1100°C [loo, 1011. Elemental boron itself is a hard material. The high microhardness of boron of 27 GPa [lo21 is related to the low molar volume of 5 cm3 and the covalent bonding of the rhombohedral crystal structure comprised of B12-icosahedra which are linked together by multiple-center bonds. Besides diamond and cBN, the well known boron carbide B4C is among the hardest materials and has been comprehensively reviewed by F. Thkvenot [1031. In the present chapter, the latest developments concerning the binary and ternary systems B-N, boron carbide nitrides (B-C-N), and boron suboxides are discussed. Other hard materials based on boron are described by R. Telle et al. in Part 111.

Boron Nitrides Boron nitride is isoelectronic with elemental carbon and is, therefore, found in similar crystal structures: -

the hexagonal (hBN) [lo41 the cubic zinc blende (cBN) [2,3] the wurtzitic (wBN) [lo51 the rhombohedral (rBN) [lo61 structure

In contrast to rhombohedral carbon, rBN has been synthesized in significant amounts by several methods and is well investigated [107,108]. In contrast to

Boron-based Hard Materials

LXV

Figure 10. P,Tphase diagram of boron nitride: (a) catalytic conversion of hBN into cBN; (b) metastable region for the vapor phase formation of cBN. An alternative phase diagram of BN suggested previously [I 1I] is shown by dashed lines.

graphite, the hexagons of hBN are arranged on top of each other, while the structures of cBN and wBN completely correspond to diamond and lonsdaleite, respectively. The zinc blende polymorph of BN is also known as sphalerite or PBN. Cubic boron nitride is the second hardest material and is produced conventionally at high pressures and temperatures, analogous to those methods developed for the fabrication of synthetic diamond. It has been commercially available since 1969 [loo]. Hot-pressed monolithic and polycrystalline BN, comprised of both the cubic and the wurtzitic polymorphs, is already used in cutting tool devices [log]. The generally accepted P,T phase diagram of BN shown in Fig. 10 is based on the work of Wentorf and colleagues, in which melting temperatures of hBN and the phase boundary h B N - + c B N at high pressures were measured. This phase boundary extrapolated to room temperature by analogy with the carbon phase diagram intersects the pressure axis [110], thus implying that hBN is the most stable modification at ambient conditions. In contrast, it was suggested in some recent publications [l 1 I] that cBN is stable at ambient pressure at temperatures up to 1300°C. This suggestion is based on measurements of the thermodynamic properties of all modifications of BN, where the difference of enthalpies of formation of cubic and hexagonal BN was found to be negative [l l l]:

AHFBN(298K)- AHfhBN(298K)= -16kJmol-'. Using these thermodynamic data an alternative phase diagram of BN (also shown in Fig. 10) was developed. The technological applications of cBN are similar to those of diamond. However, because of improved oxidation resistance, cBN allows high-speed cutting or

LXVI


grinding of steel-based materials which cannot be machined with diamond. The discovery of diamond synthesis within the metastable regime of the carbon phase diagram depicted in Fig. 5 resulted in an increased effort for the deposition of cBN films under similar conditions. The main processes developed are given below. (i) Deposition of cBN by electron cyclotron resonance plasma using NH3 and BF3 as the reactant gases. The films were grown on single crystal Si (100) wafers at a substrate temperature of 675°C and were characterized by infrared and Auger spectroscopy. The film thickness was 100400 nm, the average growth rate was 10 nm min-' [I 121. (ii) Formation of cBN by activated reactive evaporation of H3B03in the presence of NH,. Cubic BN could be deposited on different substrate materials such as steel, silicon, glass, or NaCl crystal with a rate of 150 nmmin-' at 450°C substrate temperature [113]. (iii) Synthesis of cBN by R F plasma CVD under low pressure conditions and negative self-bias. B2H6 and N2 were applied as the reactants. The cBN films were obtained on single crystal Si wafers heated at 600°C with a deposition rate of 6 nm min-' . Cubic BN was verified by transmission electron diffraction (TED) and microscopy (TEM). The TEM investigations revealed crystallite sizes of 10-20 nm [ 1141. (iv) Plasma enhanced CVD from BC13 + NH3 + H2 Ar under irradiation of an ArF excimer laser at 193 nm. The cBN films deposited were heteroepitaxially grown on Si (100) surfaces at 500-900°C substrate temperatures. Molecular orbital calculations of the cBN growth mechanism indicate an important role of atomic chlorine formed during the deposition [I 151. (v) Finally, cBN was synthesized in the form of films by the dual ion beam deposition technique. In this process, boron atoms are sputtered from a boron target by an Ar' ion beam and subsequently deposited on a substrate (Si wafer) simultaneously bombarded by N i ions. The deposition rate is 2 .2 4 .6 nm min-' and film thicknesses of up to 300nm could be achieved [116].

+

However, besides the formation of cBN, hexagonal hBN is also found in the deposited films prepared by all of the different methods mentioned above. Vapordeposition of cBN films is reviewed in Part I1 by K. Bewilogua and F. Richter. Additionally, it is also worth mentioning that cBN has been obtained by the pressure pyrolysis of borazine, B3N3H6, in a hydrothermal apparatus at 25100 MPa and 250-700°C followed by a high temperature treatment at 6.5 GPa pressure in a girdle type apparatus [117]. The decomposition of magnesium boron nitride (Mg3BN3) in the presence of water and under high pressure ( P > 4GPa) and high temperature (1400°C) conditions according to the reaction equation Mg3BN3

+ 3H20 * 3 Mg 0 + cBN + 2NH3

also provided cBN [117]. Magnesium boron nitride is a compound known in the pseudobinary system BN-Mg3N2 and is synthesized from a mixture of hBN and Mg3N2 at 4GPa and 800°C. The morphology of the cBN crystals formed is strongly dependent on the temperature and pressure applied. In the case of the

Boron-based Hard Materials

LXVII

decomposition at 1400°C and 4 GPa pressure, tetrahedral crystals 10 mm in diameter were isolated [ 1 181. By analogy to the body-centered tetragonal carbon, J. Corkill et al. postulated also a b.c.t.-4 phase of BN with alternating B and N atoms (Fig. 6a) which has been calculated by the pseudopotential total energy method [119]. While the calculations for b.c.t.-4 carbon showed a reduced stability with respect to diamond due to the occupation of antibonding states at the Fermi level, the structure of b.c.t.4 BN has a small density of states at the Fermi level. Consequently, b.c.t.-4 BN is estimated to have higher stabilization compared to b.c.t.-4 carbon. The calculated bulk modulus B (268GPa) of b.c.t.-4 BN is significantly lower than that of the cBN structure (367GPa) owing to the relative openness of the b.c.t.-4 phase.

Boron-rich Boron Nitrides Crystalline boron-rich boron nitrides of the compositions B25N up to B53N as well as amorphous phases of the composition B3N and B5N have also been synthesized by CVD methods [120,121]. The formation of a rhombohedral structure of the boron-rich nitride, B4N, is described in [122]. The films deposited at 1200°C on hBN substrates by a tungsten hot filament assisted vapor-phase reaction x>l were characterized by Raman and photoelectron spectroscopy as well as by X-ray diffraction. The spectroscopic measurements indicate the presence of a B4N phase with a crystal structure analogous to that of rhombohedral B4C, a compound which also belongs to ultrahard materials. In this crystal structure, B l l N icosahedra are linked by N-B-N chains as shown in Fig. 11 [122]. Due to this

Figure 11. Model of the unit cell of the B4N structure. The atoms marked in black represent the nitrogen atoms within the N-B-N-chain. First published in [152] and reproduced with permission.

LXVIII


structural feature, the new BN phase can be expected to exhibit interesting hardness properties.

Nitrogen-rich Boron Nitride Existence of a nitrogen-rich boron nitride denoted as h’BN was suggested by Yo0 et al., who observed appearance of a low-density hexagonal compound after laser heating of boron in excess of nitrogen in a diamond cell [123]. This compound was synthesized both in the stability fields of conventional hBN (2 GPa, 1300 K) and of cubic BN (15 GPa, 1800 K). h’BN is highly transparent and can be recovered as white polycrystals, which do not convert to other forms of boron nitride at ambient conditions [ 1231. Unfortunately, chemical composition of the compound obtained was not determined quantitatively.

Boron Carbonitrides (B,C,N,) In addition to diamond and the compounds found in the binary phase diagrams of C-N, B-C, and B-N (Fig. I), materials of the extended ternary system containing boron, carbon and nitrogen can be also considered as potential candidates with advanced hardness properties. These compounds are of broad interest because hexagonal (BN),C, can be considered for applications as high temperature semiconductors, especially if they can be altered by doping or structural changes [ 1241. Moreover, it is expected that the cubic form of ternary compounds or solid solution with the general formula (BN),C, might exhibit hardness approaching that of diamond, but will remain strong for steel cutting and withstand oxidation at higher temperatures than diamond. Low density B,C,N,, with hexagonal or turbostratic structure can be synthesized in three different ways: -

nitriding of solid -phase precursors at high temperatures using CVD technique by pyrolysis of inorganic polymers containing boron, carbon, and nitrogen.

Boron carbonitrides of the general composition B,C,N, were first synthesized by Kosolapova using the reaction of lampblack and amorphous boron in N2 atmosphere at 1800-2000°C [ 1251. Most recently, B,CyN, with hexagonal graphite-like sheet structure was obtained by CVD methods providing thin films at lower temperatures (1000-1400°C). Typically, BC13/HC=CH/NH3 or BC13/NH3/CH4/H2gas mixtures are used for the deposition of B-C-N films [124,126-1301. Boron carbonitride powders could be synthesized in high amounts by the pyrolysis of amine-borane derived inorganic polymers [ 131, 1321. The molecular amine-boranes, pyridine-borane (C5H5N BH3) and piperazine-borane (C4H10N2)BH3, form highly cross-linked polymers by annealing at moderate temperatures (100-200°C) with or without solvents [1331. Subsequently, the polymeric materials can be thermally decomposed

-

-

LXIX

Buron-bused Hard Materials r! I

H

\

c-c,

A H C N - i - H

I

\

/

110 OC

H

Figure 12. Schematic representation of the decomposition of pyridine-borane into turbostratic boron carbonitride, BC4N, with graphite like sheet structure. In the graphitic structure on the right, only the sp2 o-bonds are shown. First published in [I521 and reproduced with permission.

giving boron carbonitrides, B,C,N2, with different carbon contents. The following simplified equations represent the overall pyrolysis reactions providing BC4N and BC2N (Fig. l), respectively: CN-B.3

n

NH N-BH3 U

1050"C/Ar

1050"C/Ar

BC4N

BC2N

+ CH4 + 2H2,

+ 2CH4 + NH3 + H2.

Powder diffraction as well as photoelectron spectroscopy (XPS), electron energy loss spectroscopy (EELS), and solid state NMR investigations of the synthesized BC2N and BC4N revealed a graphite-like sheet structure with carbon substituted for boron and nitrogen and with little sheet-to-sheet registry (turbostratic boron carbonitride). The pyrolytic conversion of pyridine-borane to turbostratic B-C-N via the formation of an intermediate polymeric compound is shown schematically in Fig. 12. These boron carbonitride materials can be considered as solid solutions of C and BN and are thermally stable up to 1800°C in an Ar atmosphere. At higher temperatures, phase separation providing crystalline hBN, B4C, and elemental C is observed (see Fig. 1) [ 1331. Microstructural investigation of the synthesized BC4N by means of TEM showed also the presence of small amounts of isolated crystals with cubic symmetry as determined by electron diffraction data [ 132,1341. The diffraction data of the cubic crystals do not correspond to those of diamond or cBN. Therefore, a cubic boron carbonitride phase is assumed for those single crystals embedded in a turbostratic BC4N matrix. The most difficult issue in the synthesis of hexagonal B,C,NZ compounds is proving of a single phase content. Indeed there is no direct method of distinguishing between hexagonal B. C ,Nz, hBN and ..J graphite, if all are turbostratically distorted. Andreev and Lundstrom pointed out that in some of the above described work the available data are insufficient to make unambiguous conclusions about single phase content [1351. They prepared hexagonal B,C,N, materials by nitration of boron carbide B4C powder at 2100°C

LXX


at different partial nitrogen pressures and found in all cases at least two coexisting phases. For example, samples prepared at partial nitrogen pressures between 30 and 100% contained a carbon-poor three-dimensionally ordered phase and a carbonrich turbostratic phase [135]. Synthesis of ceramic materials from polymers is discussed in Part I1 by G. D. Soraru and P. Colombo. The high-pressure and high-temperature formation of crystalline boron carbonitride with a cubic lattice was reported in 1977 by various Russian authors using hBN + C, hBN + B4C, or cBN + C as the reacting species [136,137]. High microhardness values in the range between 60 and 80 GPa have been measured for these cubic phase E C - N materials. High-pressure synthesis of the cubic B,C,N, phases from the starting graphite-like (BN),C1-,x (with x = 0.15, 0.26 and 0.6) prepared by CVD-method was performed at 14GPa and above 3300K [124]. Although detailed chemical analysis was not performed, it was assumed that the obtained material was a solid solution of cubic BN and diamond. X-ray powder diffraction data suggested that this cubic material exhibits ideal mixing according to the Vegard’s law and has the same composition as the starting hexagonal material. However, the author also observed reflexes allowed only for cBN, which he explained by ordering of materials. Additionthe B-N pairs in the lattice of the synthesized cubic (BN),C1 -, ally, the author always observed the presence of graphite-like phases in a heat-treated charge and small amounts of B4C in a few samples [124]. Ideal mixing in the cubic (BN),C, compounds synthesized in this way was also reported 15 years later by Kagi et al. [138]. In contrast, measurement of molar volume and of the bulk modulus of cubic (BN),C1 --x phases synthesized in the YAG-laser heated diamond cell at pressures of 30-50 GPa and temperatures of 2000-2500 K showed strong deviation from ideal mixing [139]. According to these studies the molar volumes of cubic (BN),yCl at ambient conditions are larger than that of the ideal solid solution and cubic (BN),C1_, phases with x < 0.5 have a bulk modulus smaller than that of endmember cBN [139]. If the first observation holds at pressures and temperatures of synthesis then a mechanical mixture of diamond and cBN will appear at equilibrium conditions. This is because high pressures favor formation of denser materials [ 1401. It follows that this cubic (BN),C1 --x should be a metastable intermediate product of to the mixture of diamond and cBN. transformation from hexagonal (BN),C, -, The second observation suggests that the hardness of cubic (BN),C, phases with x < 0.5 should be smaller than that of cBN, if the systematic approach of Liu and Cohen [12] is applied. The conclusion of nonideal mixing made earlier [139] can be called in question because hexagonal materials did not transform completely into the cubic phases and quantitative chemical analysis of the synthesized cubic phases was not performed. Nevertheless, metastability of cubic (BN),CI is supported by the following results of Japanese authors. Sasaki et al. observed crystallization of hexagonal BC2N in diamond and cBN in the presence of a Co catalyst at relatively low pressures and temperatures (5.5 GPa and 1400-1600 K) [141]. Nakano found that upon phase separation of metastable cubic BC2N to did not change diamond and cBN the composition of residual cubic (BN),C1 -, (x = 0.33). Accordingly, no other stable phases of type (BN),C, except diamond and cBN exist at the temperature and pressure conditions of the work (7.7 GPa, 2300-2400°C) [142].

Silicon-based Materials

LXXI

Because formation of cubic boron carbonitride is of great fundamental interest with respect to superhard materials much additional effort is needed to succeed in the preferential synthesis of the cubic B-C-N phase. As follows from the above results the most promising way would be synthesis under nonequilibrium conditions such as flash-heating at static pressures or shock-wave compression [ 1401. Successful synthesis of cubic BC2,5Nsolid solution in 18% yield by shock-compression of hexagonal BC2,5Nhas been reported [143]. The material obtained was a single cubic BC2.5N phase with a diamond-like structure and crystals between 5 and 20 nm in size.

Boron Suboxides In 1988, A. R. Badzian reported that boron suboxides exhibit exceptional high hardness values [144]. The suboxides with compositions between B 2 0 and B220 were synthesized by the solid state reaction of elemental boron with B2O3 at 1600-2000°C. Subsequent melting of the reaction product under Ar provides boron containing 4-5 at.% oxygen corresponding to B I 8 0and BZ20,respectively. The crystal structure of the boron rich boron suboxides is related to that of arhombohedra1 boron. The microhardness of boron suboxides was determined to be between 31 GPa and 59 GPa, which is rather close to the hardness of cubic BN [144, 1451. Application of high pressure and high temperature allows synthesis of boron suboxides with the high degree of crystallinity needed for studying the properties of these compounds. For example, only fine powders with some secondary amorphous products were obtained in the synthesis of boron suboxide with a nominal composition B 6 0 at ambient pressure. Additionally, the synthesized compounds were nonstoichiometric with compositions varying from B600,72to B600.86.Hubert et al. showed that by reduction of B2O3 by boron at 1700°C and 5.5 GPa a highly crystalline material consisting of multiply twinned particles with icosahedral symmetry can be obtained. The size of these icosahedral particles was up to 20-30pm. The measured molar oxygen content was 0.96 f 0.08 (B600.96) thus approximating the ideal value of 1 [146,147]. The a-rhombohedra1 structure of elemental boron has an electron deficiency that can be reduced by the incorporation of interstitial atoms in structural voids. Filling these voids with atoms can result in cross-linking of the boron icosahedra which is realized for example in the structures of B4C (B12C3), B12P2, B4N, or boron rich boron suboxides (see Fig. 11) and can contribute to a significant increase of the rigidity of the solid state structure.

Silicon-based Materials This section deals with high pressure phases of two silicon based compounds Si02 and Si3N4,which previously did not belong to the family of superhard materials.

LXXII


While the a- and p-phases of silicon nitride were known for a long time as hard materials, the common phases of Si02 (e.g. quartz) lie much further down the hardness scale (Table 1). The high pressure phase of Si02 with the rutile structure (stishovite) was first reported in 1961 [148]. Si02-stishovite is stable at pressures above 11 GPa and persists metastably at ambient conditions. Its density (4.287gcmP3) is almost 62% higher than that of the ambient pressure quartz phase (2.648 g ~ m - [149]. ~ ) The bulk modulus of Si02-stishovite (298 GPa) exceeds that of B4C (200 GPa) which hardness is comparable with that of cubic BN 141. Such a strong increase in density and bulk modulus is due to an increase of the coordination number of Si-atoms from four to six. Recently LCger et al. measured the Knoop hardness of polycrystalline single phase Si02-stishovite to be between 31 and 35GPa, which exceeds the literature values of 30GPa for B4C and B 6 0 [4]. However, one has to bear in mind that the measured hardness values depend strongly on the state of the sample and on the specific test method used. Thus to make reliable comparisons hardnesses of the materials have to be measured using the same method and taking into account the above mentioned peculiarities. A change to six-fold coordination of Si-atoms occurs upon transformation of hexagonal c1- and pSi3N4 to the high pressure cubic phase with spinel structure observed in the laser heated diamond cell at 15GPa [5]. The technique of C02laser heating in a diamond cell is described in Part I by Zerr et al. The density of the cubic Si3N4is 23% above that of a- or p-phases. According to first-principles calculations, the elastic bulk and shear moduli of cSi3N4exceed that of Si02-stishovite. Consequently, the hardness of the cubic silicon nitride can be comparable to that of Si02-stishovite [5]. Detailed discussion of silicon nitride based materials is given in Part 111 by M. Herrmann. These results on the high pressure phases of two silicon based materials show a strong potential of the high pressure techniques in synthesis of new materials with promising mechanical properties. Furthermore, there are many unexplored paths for future work, even if some of these ways seem on first glance to be completely worked out, as for the case of Si02-stishovite.

Concluding Remarks The experimental and theoretical research activities in the field of materials with hardness comparable to or higher than diamond have been stimulated in recent years by the discovery and postulation of novel carbon allotropes and novel carbon- and boronbased materials. Comparison of the experimental and theoretical results obtained in the past shows that first principles quantum mechanical and semi-empirical calculations are very useful for the screening of the structural and electronic properties of new materials. The fascinating variety of new possible carbon allotropes which have been calculated to be stable relative to diamond or can be stabilized for kinetic reasons reflects the high potential and the topicality of modern carbon chemistry. Novel carbon allotropes and carbon and silicon nitrides are subjects of discussion in Part I in the chapter written by G. Jungnickel et al. and J. E. Lowther.

However, the experimental evidence for the synthesis and structure of dense superhard C3N4 phases reported in the literature is not unambiguous. Thus, future research should be focused on the synthesis of sufficient amounts of these new superhard materials and on their structural characterization. The structures of all C3N4-polymorphsreported so far still remain to be verified by single crystal X-ray or electron diffraction. Furthermore, the difficulties in the determination of reliable hardness values of these new materials not only at room temperature but also at elevated temperatures have to be overcome. Diamond will not lose its reputation as the hardest material until the newly developed phases and compounds have been characterized in detail and can be produced in reasonable amounts.

Acknowledgement The authors thank J. Nicolich for preparation of crystal structure plots. This work was supported by the Deutsche Forschungsgemeinschaft under the contract number Ri 510/14-1.

References 1. 2. 3. 4. 5. 6.

7. 8. 9.

10. 11.

F. P. Bundy, H. T. Hall, H. M. Strong, and R. J. Wentorf, Nature, 1955, 176, 51. R. H. Wentorf, J . Chem. Phys. 1957, 26, 956. R. H. Wentorf, J . Chem. Phys. 1961, 34, 809. J. M. Leger, J. Haines, M. Schmidt, J. P. Petitet, A. S. Pereira, and J. A. H. da Jornada, Nature, 1996, 383, 401. A. Zerr, G. Miehe, G. Serghiou, M. Schwarz, E. Kroke, R. Riedel, H. FueR, P. Kroll, and R. Boehler, Nature, 1999, 400, 340. J. C. Angus and C. C. Hayman, Science, 1988, 241, 913. K. E. Spear, J . Am. Ceram. SOC.1989, 72, 171. W. A. Yarbrough and R. Messier, Science, 1990, 247, 688. P. K. Bachmann, Adv. Muter. 1990, 2, 195. P. K. Bachmann, Keramische Zeitschrift, 1992, 2, 93. R. Telle, in Materials Science and Technology: A Comprehensive Treatment, R. W. Cahn, P. Haasen, and E. J. Kramer (Series Eds), Vol. 11, N. V. Swain (Ed.), VCH, Weinheim, 1994,

p. 173. 12. A. Y. Liu and M. L. Cohen, Science, 1989, 245, 841. 13. D. M. Teter and R. J. Hemley, Nature, 1996, 271, 53. 14. J. Hoggin, Chem. Eng. News, May 25, 1992, 20. 15. C. Niu, Y. Z. Lu, and C. M. Lieber, Science, 1993, 261, 334. 16. A. J. Stevens, T. Koga, C. B. Agee, M. J. Aziz, and C. M. Lieber, J. Am. Chem. SOC.1996,118, 10900. 17. D. M. Teter, M R S Bull. 1998, 23, 22. 18. Y. Li, Y . Qian, H. Liao, Y. Ding, L. Yang, C. Xu, F. Li, and G. Zhou, Science, 1998,281,246. 19. U. Helmersson, S. Todorova, S. A. Barnett, J.-E. Sundgren, L. C. Markert, and J. E. Greene, J. Appl. Phys. 1987, 62, 481. 20. I-W. Chen and A. Rosenflanz, Nature, 1997,389,701.

LXXIV


21. S. VepTek, P. Nesladek, A. Niederhofer, F. Glatz, M. Jilek, and M. Sima, Surf. Coat. Technol. 1998, 108-109, 138. 22. H. Holleck, J. Vac. Sci. Technol. 1986, A4, 2661. 23. K.KuBmaul, in Lexikon Werkstoftechnik, H. Grafen (Ed.), VDI, Dusseldorf, 1991, p. 409. 24. K. Niihara, and T. Hirai, J. Muter. Sci. Lett. 1978, 13, 2276. 25. R. W. Rice, C. Cm. Wu, and F. Borchelt, J . Am. Ceram. Soc. 1994,77, 2539. 26. M. L. Cohen, J. Hard Muter. 1991, 2, 13. 27. B. Farid and R. W. Godby, Phys. Rev. B, 1991,43, 14248. 28. R. W. Hertzberg, Deformation and Fracture Mechanics of Engineering Materials, 3rd edn, Wiley, New York, 1989. 29. A. Kelly and N. H. Macmillan, Strong Solids, 3rd edn, Clarendon Press, Oxford, 1986. 30. E. Knittle, R. M. Wentzcovitch, R. Jeanloz, and M. L. Cohen, Nature, 1989, 337, 349. 3 1. H. B. Huntington, in Solid State Physics: Advances in Research and Applications, F. Seitz and D. Turnbull (Eds), Vol. 7, Academic Press, New York, 1958, p. 213. 32. 0. L. Anderson, in Physical Acoustics: Principles and Methods, W. P. Mason (Ed.), Vol. 111, Part B, Academic Press, New York, 1965, p. 43. 33. R. E. Hanneman, H. M. Strong, and F. P. Bundy, Science, 1967, 155, 995. 34. H. Strunz, Mineralogische Tabellen, 7th edn, Akademische Verlagsgesellschaft Geest & Portig K.-G., Leipzig, 1978. 35. A. Neuhaus, Proc. Intern. Mineralog. Assoc. 1960, 1, 33. 36. H. W. Kroto, J. R. Heath, S. C. O'Brien, R. F. Curl, and R. E. Smalley, Nature, 1985,318, 162. 37. H. W. Kroto, Science, 1988,242, 1139. 38. R. F. Curl and R. E. Smalley, Science, 1988, 242, 1017. 39. W. Kratschmer, L. D. Lamb, K. Fostiropoulos, and D. R. Huffman, Nature, 1990,347, 354. 40. H. W. Kroto, Angew. Chem. 1992, 104, 113. 41. F. Diederich and Y. Rubin, Angew. Chem. 1992, 104, 1123. 42. R. F. Curl and R. E. Smalley, Spektrum der Wissenschaft, 1991, 88. 43. W. Kratschmer and K. Fostiropoulos, Physik in unserer Zeit, 1992, 3, 105. 44. H. R. Karfunkel and T. Dressler, J . Am. Chem. Soc. 1992, 114, 2285. 45. S. Koizumi, M. Kamo, Y. Sato, H. Ozaki, and T. Inuzuka, Appl. Phys. Lett. 1997,71, 1065. 46. P. W. Bridgman, Sci. Am. 1955, 193, 42. 47. R. C. Burns, S. Kessler, M. Subanda, C. M. Welbourn, and D. L. Welch, in Proceedings of the 3rd NIRIM International Symposium on Advanced Materials ( I S A M '96), M. Akaishi et al. (Eds), NIRIM, Tsukuba, 1996, p. 105. 48. J. E. Butler and H. Windischmann, M R S Bull. 1998, 23, 22. 49. M. Kamo, Y. Sato, K. S. Matsumoto, and N. Setaka, J . Cryst. Growth, 1983, 62, 642. 50. P. K. Bachmann, W. Drawl, D. Knight, R. Weimer, and R. F. Messier, in Diamond and Diamond-Like Materials Synthesis, G. H. Johnson, A. R. Badzian, and M. W. Geis (Eds), M R S Symp. Proc., Vol. EA-15, Materials Research Society 1988, 99. 51. S. Matsumoto, Y. Sato, M. Kamo, and N. Setaka, Jpn. J . Appl. Phys. 1982, 21, L183. 52. S. Matsumoto, J. Muter. Sci. Lett. 1985, 4, 600. 53. K. Kurihara, K. Sasaki, M. Kawarada, and N. Koshino, Appl. Phys. Lett. 1988, 6,437. 54. Y. Hirose and N. Kondo, in 35th Spring Meeting of the Japanese Appl. Phys. Society, 1988, p. 434. 55. N. Ohtake, H. Tokura, Y. Kuriyama, Y. Mashimo, and M. Yoshikawa, in Proc. 1st Int. Symp. Diamond and Diamond-Like Films, J. P. Dismukes et a/. (Eds), The Electrochemical Society, Pennington, NJ, Proceedings Vol. 89-12, 1989, p. 50. 56. D. M. Gruen, M R S Bull. 1998, 23, 32. 57. D. W. Brenner and J. H. Harrison, Am. Ceram. Soc. Bull. 1992,71, 1821. 58. A. M. Stoneham, I. J. Ford, and P. R. Chalker, MRS Bull. 1998, 23, 28. 59. S. A. Solin and A. K. Ramdas, Phys. Rev. B, 1970, 1, 1687. 60. C. Wild, N. Herres, J. Wagner, P. Koidl, and T. R. Anthony, in Proc. 1st Int. Symp. on Diamond and Diamond-Like Films, J. P. Dismukes et al. (Eds), The Electrochemical Society, Pennington, NJ, Proceedings Vol. 89-12, 1989, p. 283. 61. L. Holland, J . Vac. Sci. Technol. 1977, 14, 5. 62. L. Holland, Thin Solid Films, 1979, 58, 107.

References

LXXV

0. S. Selwyn and E. Kay, Plasma Chem. Plasma. Process, 1985, 5, 1983. M. Ramsteiner and J. Wagner, Appl. Phys. Lett. 1987, 51, 1335. K. Kobayashi, N. Mutsukara, and Y. Machi, J . Appl. Phys. 1986, 59, 910. F. Jansen, M. Machonkin, S. Kaplan, and S. Hark, J. Vuc. Sci. Technol. 1985, A3, 605. J. W. Rabelais and S. Kasi, Science, 1988, 239, 623. N. Savvides, J . Appl. Phys. 1985, 58, 518. J. C. Phillips, Phys. Rev. Lett. 1979, 42, 153. C. B. Collins, F. Davanloo, T. J. Lee, J. H. You, and H. Park, Am. Cerum. Soc. Bull. 1992,71, 1535. 71. N. N. Matyusenko and V. E. Strel'nitzkii, J . Exp. Theor. Phys. Lett. 1979, 30, 199. 72. R. Hoffmann, T. Hughbanks, M. Kertesz, and P. H. Bird, J . Am. Chem. Soc. 1983,105,4831. 73. M. A. Tamor and K. C. Hass, J . Mater. Res. 1990, 5, 2273. 74. A. Y. Liu, M. L. Cohen. K. C. Hass, and M. A. Tamor, Phys. Rev. B, 1991,43, 6742. 75. A. H. Alberts and H. Wynberg, J. Chem. Soc., Chem. Commun. 1988, 748. 76. A. Y. Liu and M. L. Cohen, Phys. Rev. B, 1992,454579. 77. A. F. Hebard, M. J. Rosseinsky, R. C. Haddon, D. W. Murphy, S. T. Glarum, T. T. M. Palstra, A. P. Ramirez, and A. R. Kortan, Nature, 1991, 350, 600. 78. M. N. Regueiro, P. Monceau, and J.-L. Hodeau. Nature, 1992, 355, 237. 79. M. N. Regueiro, Adv. Mafer. 1992, 4, 438. 80. R. S. Ruoff and A. L. Ruoff, Nature, 1991, 350, 663. 81. S. J. Duclos, K. Brister, R. C. Haddon, A. R. Kortan, and F. A. Thiel, Nature, 1991, 351, 380. 82. M. L. Cohen, Phys. Rev. B, 1985, 32, 7988. 83. A. Y. Liu and M. L. Cohen, Phys. Rev. B, 1990,41, 10727. 84. A. Y. Liu and R. M. Wetzcovitsch, Phys. Rev. B, 1994, 50, 10362. 85. M. R. Wixom, J . Am. Cerum. Soc. 1990, 73, 1973. 86. L. Maya, D. R. Cole, and E. W. Hagaman, J . Am. Ceram. Soc. 1991, 74, 1686. 87. H. X. Han and B. J. Feldman, Solid State Commun. 1988. 65, 921. 88. T. Sekine, H. Kanda, Y. Bando, M. Yokoyama, and K. Hojou, J . Muter. Sci. Lett. 1990, 9, 1376. 89. J. Kouvetakis, R. B. Kaner, M. L. Sattler, and N. Bartlett, J . Chem. Soc. Commun. 1986, 1758. 90. J. Martin-Gil, F. J. Martin-Gil, M. Sarikaya, M. Qian, M. Jose-Yacaman, and A. Rubio, J. Appl. Phys. 1997,81, 2555. 91. K. M. Yu, M. L. Cohen, E. E. Haller, W. L. Hansen, A. Y. Liu, and I. C. Wu, Phys. Rev. B, 1994,49, 5034. 92. A. Fernandez, P. Prieto, C. Quiros, J. M. Sanz, J.-M. Martin, and B. Vacher, Appl. Phys. Lett. 1996, 69, 764. 93. Y. Chen, L. Guo, and E. G. Wang, Phil. Mag. Lett. 1997, 75, 155. 94. J. L. He and W. L. Chang, Surf. Coat. Technol. 1998,99, 184. 95. J. H. Nguyen and R. Jeanloz, Muter. Sci.Eng. A, 1996, 209, 23. 96. D. C. Nesting and J. V. Badding, Chem. Muter. 1996, 8, 1535. 97. D. C. Nesting, J. Kouvetakis, and J. V. Badding, in Proceedings ofthe 5th NIRIM International Symposium on Advanced Materials (ISAM '98), M. Kamo et al. (Eds), NIRIM, Tsukuba, Japan, 1998, 1. 98. A. J. Stevens, T. Koga, C. B. Agee, M. J. Aziz, and C. M. Lieber, J . Am. Chem. Soc. 1996,118, 10900. 99. V. P. Dymont, E. M. Nekrashevich, and I. M. Starchenko, JETP Lett. 1998,68,498. 100. L. Vel, G. Demazeau, and J. Etourneau, Mater. Sci. Eng. 1991, B10, 149. 101. B. P. Singh, Muter. Res. Bull. 1986, 21, 85. 102. F. Binder, Radex-Rundschau, 1975, 4, 531. 103. F. Thevenot, J . Eur. Cerum. Soc. 1990, 6, 205. 104. A. Lipp, K. A. Schwetz, and K. Hunold, J . Eur. Serum. Soc. 1989, 5, 3. 105. R. R. Wills, Int. J . High Technol. Serum. 1985, 1, 139. 106. T. Ishii, T. Sato, Y. Sekikawa, and J. M. Iwata, J . Cryst. Growth, 1981, 52, 285. 107. T. Matsuda, N. Uno, H. Nakae, and T. Hirai, J . Muter. Sci. 1986, 21, 649. 108. V. L. Solozhenko, V. A. Mukhanov, and N. V. Novikov, Dokludy Chem. 1990,312, 125. 63. 64. 65. 66. 67. 68. 69. 70.

LXXVI


A. Nagel, Ceram. Tec AG, Plochingen, 1993, private communications. F. R. Corrigan and F. P. Bundy, J . Chem. Phys. 1975, 63, 3812. V. L. Solozhenko, High Press. Res. 1995, 13, 199. S. Y. Shapoval, V. T. Petrashov, 0.A. Popov, A. 0.Westner, M. D. Yoder, and C. K. C. Lok, New Diamond Sci. Tech. 1991, 1093. 113. K. L. Chopra, V. Agarwal, and D. Vankar, Thin Solid Films, 1985, 126, 307. 114. A. Chayahara, H. kokoyama, T. Imura, and Y. Osaka, Jap. J . Appl. Phys. 1987, 26, L1435. 115. S. Komatsu, Y. Moriyoshi, M. Kasamatsu, and K. Yamada, J . Phys. D: Appl. Phys. 1991,24, 1687. 116. W. L. Lin, Z. Xia, and Y. L. Liu, Muter. Sci. Eng. 1990, B7, 107. 117. S. Hirano, A. Fujii, T. Yogo, and S. Naka, J. Am. Ceram. Soc. 1990,73, 2238. 118. S. Nakano, H. Ikawa, and 0. Fukunaga, J. Am. Ceram. Soc. 1992,75,240. 119. J. L. Corkill, A. Y. Liu, and M. L. Cohen, Phys. Rev. 1992, B45, 12746. 120. K. Ploog, P. Rauh, W. Stoeger, and H. Schmidt, J. Cryst. Growth, 1972, 13/14, 350. 121. K. Ploog, H. Schmidt, E. Amberger, G. Will, and K. H. Kossobutzki, J. Less-Common Met. 1972, 29, 161. 122. H. Saitoh, K. Yoshida, and W. A. Yarbrough, J. Muter. Res. 1993, 8, 8. 123. C. S. Yoo, J. Akella, and H. Cynn, Phys. Rev. B, 1997, 56, 140. 124. A. R. Badzian, Mat. Res. Bull. 1981, 16, 1385. 125. T. Y. Kosolapova, G. N. Marenko, T. I. Serebryakova, 0. Prilutskii, T. Khorpyakov, and 0. I. Chernysheva, Sov. Powder Metall. Met. Ceram. 1971, 10, 22. 126. A. R. Badzian, T. Niemyski, and E. Olkusnik, in Proc. Third Int. Con$ Chemical Vapor Deposition, F. A. Glaski (Ed.), Salt Lake City, Utah, 1972, p. 747. 127. J. Kouvetakis, R. B. Kaner, M. L. Sattler, and N. Bartlett, J . Chem. Soc., Chem. Commun. 1986, 1758. 128. R. B. Kaner, J. Kouvetakis, C. E. Warble, M. L. Sattler, and N. Bartlett, Muter. Res. Bull. 1987, 22, 399. 129. T. M. Besmann, J. Am. Ceram. SOC.1990,73, 2498. 130. F. Saugnac, F. Teyssandier, and A. Marchand, J . Am. Ceram. Soc. 1992,75, 161. 131. R. Riedel, J. Bill, and G. Passing, Adv. Muter. 1991, 3, 551. 132. J. Bill, R. Riedel, and G. Passing, Z. Anorg. Allg. Chem. 1992, 610, 83. 133. J. Bill, M. FrieB, and R. Riedel, Eur. J. Solid State Inorg. Chem. 1992, 29, 714. 134. J. Bill, P h B . Thesis, University of Stuttgart 1993. 135. Yu. G. Andreev and T. Lundstrom, J. Alloys Comp. 1994,210, 311. 136. N. N. Sirota, M. M. Zhuk, A. M. Mazurenko, and A. I. Olekhnovich, Vestsi Akad. Nauk BSSR, Ser. Fiz.-Muter. Nauk, 1977, 2, 111, or CA 1977, 87, 59009j. 137. A. K. Butylenko, G. V. Samsonov, I. 1. Timofeeva, and G. Makarenko, Pis’ma Zh. Tekh. Fiz. 1977,3, 186, or CA 1977,86, 175812. 138. H. Kagi, I. Tsuchida, Y. Masuda, M. Okuda, K. Katsura, and M. Wakatsuki, in Proceedings of the Joint XV AIRAPT and XXXIII EHPRG International Conference, Warsaw, Poland, W. A. Trzeciakowski (Ed.), World Scientific Publishing, Singapore, 1995, p. 258. 139. E. Knittle, R. B. Kaner, R. Jeanloz, and M. L. Cohen, Phys. Rev. B, 1995, 51, 12149. 140. V. L. Solozhenko, Eur. J. Solid State Inorg. Chem. 1997, 34, 797. 141. T. Sasaki, M. Akaishi, S. Yamaoka, Y . Fujiki, and T. Oikawa, Chem. Mat. 1993, 5, 695. 142. S. Nakano, in Proceedings of the 3rd NIRIM International Symposium on Advanced Materials (ZSAM P6), M. Akaishi et al. (Eds), NIRIM, Tsukuba, Japan, 1996, p. 287. 143. T. Komatsu, M. Nomura, Y. Kakudate, and S. Fujiwara, J . Muter. Chem. 1996, 6, 1799. 144. A. R. Badzian, Appl. Phys. Lett. 1988, 53, 2495. 145. I. A. Bairamashvili, G. I. Kalandadze, A. M. Eristavi, J. Sh. Jobava, V. V. Chotulidi, and Yu. I. Saloev, J. Less-Common Met. 1979, 67, 455. 146. H. Hubert, L. A. J. Garvie, B. Devouard, P. R. Buseck, W. T. Petuskey, and P. F. McMillan, Chem. Mater. 1998, 10, 1530. 147. H. Hubert, B. Devouard, L. A. J. Garvie, M. O’Keeffe, P. R. Buseck, W. T. Petuskey, and P. F. McMillan, Nature, 1998, 391, 376. 148. S. M. Stishov and V. Popova, Geokhimiya, 1961, 10, 837. 109. 110. 111. 112.

149. J. R. Smyth and T. C. McCormic, in Mineral Physics and Crystallography: A Handbook of Physical Constants, T. J. Ahrens (Ed.), American Geophysical Union, Washington, DC, 1995, p. 1. 150. S. Fahy and S. G. Louie, Phys. Rev. B, 1987, 36, 3373. 151. F. P. Bundy, W. A. Bassett, M. S. Weathers, R. J. Hemley, H. K. Mao, and A. F. Goncharov, Carbon, 1996, 34, 141. 152. R. Riedel, Novel ultrahard materials, Adv. Muter. 1994, 6, 549.

Part I

Structures and Properties

Handbook of Ceramic Hard Materials Edited by Ralf Riedel Copyright Q WILEY-VCH Verlag GmbH, D-69469 Weinheim (Federal Republic of Germany), 2000

1 Structural Chemistry of Hard Materials W. Jeitschko, R. Pottgen, and R.-D. Hoffmann

1.1 Introduction Most hard materials are formed from high-melting elements in the middle of the periods of the transition metals and the main group elements. The stability of these elements and compounds arises from a maximum of filled bonding bands and a minimum of filled antibonding bands. The hardness of many elements and materials is frequently not well known, but usually hardness and high melting points correlate, and melting points are better known. Figure 1 shows the melting points of the elements and some well-known high-melting and hard materials. At this point it is worthwhile to note that refractory carbides like HfC or TaC have significantly higher melting points than the frequently used crucible ceramic alumina, A1203. Most well-known hard materials have surprisingly simple crystal structures. Here we give an overview of the crystal chemistry of hard materials. Different aspects of the various crystal structures are discussed, such as close-packed structures with filled octahedral voids like titanium carbide, the tetrahedral arrangements in Sic or the three-dimensional inorganic networks in a- and P-Si3N4. Also we briefly mention the synthesis and some applications of various hard materials. The basic crystallographic data of the most important hard materials are listed in Table 1. Further access to crystallographic data can be gained through various compilations. The older literature has critically been evaluated in the Strukturberichte [ 13, the Structure Reports [2] and in the Landolt-Bornstein [3]. The non-oxidic materials are well covered in several editions of Pearson’s Handbook [MI.The various structure types of these materials have been critically compiled by Parthe and coworkers [7]. Access to the more recent literature is best achieved through 4000

-*

4000

-

TaC

. =- ZrC

3000

Figure 1. Melting points of most elements and some selected high-melting inorganic materials.

4

1 Structural Chemistry of Hard Materials

Table 1 Crystallographic data of hard materials

Compound

Pearson symbol

Space group

Z

cF8 hP4 cF8 hP12 cF8 hR45 B13C2 hR42 B6° TiB2 hP3 FeB oP8 CrB OC8 R-MOB tI16 cP7 YB6 wc hP2 hP6 w2c TIC cF8 Fe3C oP16 oP20 cr3c2 cF116 cr23c6 oP40 cr7c3 W3Fe3C cF112 MoSi2 t16 CrSi2 hP9 TiSi, oF24 Ti3SiC2 hP12 hP18 Mo5Si3C a-Si3N4 hP28 p-Si3N4 hP14 AIN hP4 oP12 5-Fe2N &-Fe3N hP8 y'-Fe4N cP5 Ti2N tP6 Ti,AIN cP5 Ti2AIN hP8 Zr3A1N oc20 a-A1203 hR30 a-Be0 hP4 cF8 MgO cF56 MgA1204 Ti02, rutile tP6 Ti07. anatase t112 TiO;; brookite oP24 y-Zr02 cF12 P-Zr02 tP6 u-Zr02 mP12

Fdjm P63/mmc F43m P63mc F43m Rjm R h P6/mmm Pnma Cmcm 141lamd PmSm ~6m2 P3ml Fmjm Pnma Pnma Fmh Pnma Fdjm I4/mmm P6222 Fddd P63/mmc P63/mcm P31c p63 l m P63mc Pbcn P6322 Pm3m P42lmnm PmSm P63/mmc Cmcm R% P63-mc Fm3m Fdh P42/mnm 14, lamd Pbca Fmh P42/nmc p21 1.

8 4 4 6 4 3 6 1 4 4 8 1 1 2 4 4 4 4 4 16 2 3 8 2 2 4 2 2 4 2

CeOz Tho2

Fm3m Fmh

4 4

c-c h-C c-Sic 6H-Sic C-BN

cF12 cF12

1

2 1

2 4 6 2 4 8 2 4 8 4 2 4

a bm)

b (pm)

356.7 252.2 434.9 308.1 361.5 563.3 538.6 303.1 294.6 549.5 785.8 296.9 310.5 412.8 290.6 299 432.9 673.0 507.9 282.9 553.3 1066.0 701.0 452.6 1108.7 320.2 442.1 480.0 826.7 306.8 728.6 775.2 759.5 311.0 554.1 443.7 469.8 379.0 494.5 411.1 299.4 1149.8 336.9 476.0 269.8 421.1 807.5 459.3 378.4 544.7 918.4 506.5 364 520.8 514.5 p = 99..23" 552.6 560.4

C

V

bm)

(nm3)

41 1.9 1511.7 1216.4 1231.9 322.9 405.3 293.3 1697.0 283.7 473 451.4 1147.2 1214.2 785.2 635.1 855.1 1766.9 504.6 561.9 290.2 498.0 484.3 437.9 303.4 1361 898.3 1299.3 427.7 295.8 951.5 514.5 527 531.1

0.0454 0.0262 0.0823 0.1243 0.0472 0.3343 0.3095 0.0257 0.0656 0.0684 0.1636 0.0703 0.0208 0.0366 0.08 1 1 0.1543 0.1796 1.2113 0.3852 1.3628 0.0805 0.1075 0.3393 0.1440 0.2320 0.2925 0.1450 0.0417 0.1191 0.0837 0.0544 0.0742 0.0695 0.1057 0.3480 0.2549 0.0270 0.0747 0.5265 0.0624 0.1362 0.2574 0.1300 0.0698 0.1404 0.1687 0.1760

Ref.

1.2 Diamond and Diamond-Related Structures

5

the ICSD data base [8]. The X-ray powder patterns of most solid state materials are stored in the ICDD files [9]. Knowing the crystal structure they can also be generated by the program LAZY PULVERIX [lo].

1.2 Diamond and Diamond-Related Structures 1.2.1 The Crystal Structure of Diamond Carbon in the structural form of diamond is the only element used industrially as a hard material. Each year about ten tons of natural diamond and about twenty tons of synthetic diamond (produced via high temperature high pressure synthesis) are marketed as hard materials. While pure diamond is transparent, a yellow tint results from the replacement of some carbon atoms by nitrogen, and a blue, yellow, or even green tint through substitution of carbon by boron atoms. Polycrystalline diamond with impurities, used as an abrasive, is often black. The structure of cubic diamond (a = 0.3567nm) is presented in Fig. 2. Each carbon atom is tetrahedrally coordinated by four other carbon atoms with the C-C single bond distance of 0.1545nm [ll]. These tetrahedra are connected via shared corners. Hexagonal puckered nets can be visualized (perpendicular to the three-fold axes) where the condensed c6 rings are all in chair conformation. These layers are stacked in the sequence ABC, ABC. Starting from well-crystallized graphite with a large degree of AB, AB stacked planar layers, a hexagonal modification of diamond (a = 0.2522 nm, c = 0.41 19 nm) can be obtained under static pressure at high temperature [12]. This modification has an AB, AB stacking sequence of the chair-like layers (Fig. 2). It is frequently called hexagonal diamond or lonsdaleite. Small particles of lonsdaleite have been discovered in the Canyon Diablo meteorite crater [12]. The C-C distance in lonsdaleite is equal to that in diamond [l 11. One quarter of the c 6 rings are in chair and three quarters are in boat conformation. The lonsdaleite structure is possibly idealized. An imperfect ‘lonsdaleite’ could be formed by stacking

C

6

Figure 2. Crystal structures of cubic diamond (left-hand side, space group F d h ) and hexagonal lonsdaleite (space group P6Jmmc). The tetrahedral environments of the carbon atoms and the stacking sequences ABC, ABC and AB, AB, respectively, are indicated.

6

1 Structural Chemistry qf Hard Materiuls

Figure 3. Crystal structure of c-BN. Two unit cells of the cubic structure are shown at the right-hand side. In the middle the structure is projected along one of the threefold axes (cube diagonal). At the left-hand side the tetrahedral environments around one kind of atoms is emphasized. Boron and nitrogen atoms occupy equivalent positions, hence the structure may be thought to be composed of NB4/4 as well as BN4/4tetrahedra.

faults of the ABC stacking variant resulting in a significant degree of AB stacking. Different stacking faults of such chair-like layers are frequently also observed in thin films of diamond prepared by chemical vapor deposition (CVD) techniques. The initial X-ray investigations of several samples of synthetic hexagonal diamond indicated mixtures of both modifications. One problem of such X-ray powder investigations is the overlap of several broad reflections from both modifications.

1.2.2 The Isoelectronic Compounds c-BN and S i c Cubic boron nitride (c-BN) [6,13] is almost as hard as diamond and it is sometimes called inorganic diamond. c-BN can be synthesized from hexagonal graphite-like boron nitride (h-BN) under high pressure (50-90 kbar) and high temperature (1500-2200°C) with the help of small amounts of a catalyst such as Li3N [14]. cBN crystallizes with a zinc blende type structure. The boron and nitrogen atoms are ordered on the carbon positions of the diamond structure, resulting in a three-dimensional network of corner-sharing BN414 and NB4/4 tetrahedra (Fig. 3) with B-N distances of 0.157nm. Due to the ordering of boron and nitrogen atoms, the space group symmetry is reduced: c-BN crystallizes in the noncentrosymmetric space group F43m which is a translationengleiche subgroup of index 2 of Fdjm. The density of cubic boron nitride (3.45gcmp3) is slightly lower than that of diamond (3.51 g ~ m - ~ At ) . higher temperatures c-BN is less oxidizable than diamond. A wurtzite type modification of boron nitride corresponding to the hexagonal lonsdaleite modification of diamond is also known. It is formed by shock compression at higher pressure than necessary for c-BN [15]. An inexpensive diamond substitute is silicon carbide S i c which is produced in megaton amounts per year. S i c can be synthesized from Si02 and carbon at high temperatures: Si02 3C -+ Sic + 2CO. Between 2200 and 2400°C the commercially sold 6H-modification [ 161 is obtained, while cubic 3C-Sic [ 171 (zinc blende type) forms at lower temperatures (<2O0O0C) [18]. Pure S i c is colorless

+

1.2 Diamond and Diamond-Related Structures

c-Sic

7

GH-SIC

Figure 4. Crystal structures of 3C- and 6H-Sic. The tetrahedral environments of the silicon atoms and the stacking sequences are indicated.

and transparent, while technical Sic (carborundum) is black with a bluish tint, due to a small excess of silicon. S i c is a very non-corrosive material with a remarkable chemical and thermal stability. The cubic modification of 3C-Sic (space group F43rn) is isotypic with c-BN. In this modification the chair-like layers are stacked in ABC sequence (Fig. 4) as in cubic diamond with a Si-C distance of 0.189nm in all polytypes. This is somewhat less than the average distance of 0.195 nm, between the C-C (0.154nm) and Si-Si (0.235 nm) single bond distances in the elements [ 1 11. For Sic a huge number of different stacking sequences have been observed. In 1972 Parthe listed some 77 different polytypic forms [19]. With the exception of the cubic 3C-Sic modification all these polytypes have rhombohedral, trigonal, or hexagonal symmetry. The c dimensions of their respective hexagonal cells are all multiples of 0.252 nm. They range from 0.505 nm to 149.5 nm, all with the same a dimension of 0.308 nm. Growth by screw dislocations has been discussed as an explanation for the long-range order of the many polytypes [20,21]. As an example the structure of 6H-Sic (space group P6gnc) is presented in Fig. 4. Here, the stacking sequence is ABCACB. Various symbols have been introduced to describe the many possible stacking sequences of (not only) the modifications of Sic. As an example, the different notations are presented in the following scheme for a structure with fourteen layers: A B C B A B C B A B C A C B, A B C B A B C B A B C A C B

14H

A B C notation

Ramsdell

8


The simplest one is the ABC notation, where the letters A, B, C are used for the three possible positions of the close packed layers, starting with the letter A for the first layer. For more complex stacking sequences the Jagodzinski-Wyckoff notation has been introduced. This notation is only applicable to structures which allow not more than three stacking positions and where two subsequent layers cannot have the same stacking position. B layers of a stacking sequence ABA, where the two neighboring layers (A) are displaced by the same amount in the same direction are designated with the letter h (hexagonal); and B layers of a sequence ABC where the two neighboring layers (A and C) are displaced by opposite amounts are assigned the letter c (cubic). These sequences can be represented in a further abbreviated form by summing up equal subsequences counting from one h to the next one, so the sequence ABCBABCBABCACB can be represented by ( h ~ ) ~ ( h c cUsing ) ~ . the Zhdanov notation this can be further abbreviated by stating the number of layers and how often they are repeated. Thus ( h ~ ) ~ ( h cis c )represented ~ by 2432in the Zhdanov notation. A more general notation was introduced by Ramsdell. This notation lists the number of slabs in one unit cell and the respective symmetry with a capital letter (C for cubic, H for hexagonal, T for trigonal, and R for rhombohedral). Since there is sometimes more than one way to arrange a certain number of layers in a trigonal or hexagonal cell this notation may be ambiguous.

1.3 Crystal Chemistry of Borides and Boron Carbides The technically important boron carbide, usually designated with the formula B4C, is synthesized from boron oxide and coke at about 2500°C [22]: 2B203 7C + B4C 6CO. The high costs for elemental boron make the direct synthesis unprofitable. The product has a large homogeneity range [23] extending from carbon-rich B8C2 to boron-rich B13C2; the latter may be considered as the ideal composition. This material is almost as hard as diamond and cubic BN, but less expensive. At temperatures above 1200°C its hardness even exceeds that of diamond [24]. It has a surprisingly low density of only 2.51 gcmP3. B13C2itself shows an extremely high chemical inertness. It is only attacked in alkaline media in the presence of oxygen. Boron carbide then forms borates and carbonates, such as B4C 40H- 4 0 2 + B40:C0:2H20. Boron carbide is widely used as an abrasive. In this respect it is better than silicon carbide; however, it is more expensive. The unit cell of the ideal B13C2structure (space group R j m ) [25] is shown in Fig. 5. B12icosahedra occupy the corners of the rhombohedral cell. They are linked directly via B-B bonds and in addition by linear C-B-C units. Hence, the compound may be represented by the formula (B12)CBC.The C-B distance of 0.143 nm indicates some double bond character. The homogeneity range of this boron carbide results from a substitution of some boron atoms of the chains and/or of the icosahedra by carbon. Also, some related

+

+

+

+

+

+

1.3 Crystal Chemistry of Borides and Boron Carbides

9

Figure 5. The crystal structure of B13C2(‘B4C’)in the hexagonal setting. All boron atoms indicated by filled circles form (empty) icosahedra. Other boron atoms (open circles) are located in the centers of linear C-B-C units. At the left-hand side the stacking (ABC) of this rhombohedra1 structure is indicated.

aluminum containing compounds, A1B40C4 [26] and A12,1BS1C8[27] are known, which again contain BI2 icosahedra as the most important structural units. Such compounds are reported to contain C-B-C, C-B-B, C-Al-B, or C-Al-C chains between the icosahedra. A detailed list of these structures has been given by Wells [28]. The structure of so-called ‘P-AlBl2’ was redetermined recently, resulting in the composition A13B48C2[29]. This structure may be described as an orthorhombically distorted variant of tetragonal boron. Besides aluminum, also silicon, titanium, vanadium, copper, and iron can be incorporated in this structure tYPe. A considerable hardness was also reported for the boron suboxide B 6 0 [24]. The structure of B 6 0 (B6P type, space group Rjm) was refined from X-ray powder data [30]. Recently twinned crystals of B 6 0 with icosahedral morphology were grown at high temperature and high pressure 1311. Again in this compound BI2icosahedra are the most important structural unit. Within one layer, the B12 icosahedra are connected via oxygen atoms. In this way each icosahedron has six oxygen neighbors and each oxygen atom is connected to three icosahedra (Fig. 6). Subsequent layers are stacked in ABC sequence and connected via boron-boron bonds. Of the transition metal borides, the titanium boride TiB2 is the technically most important compound. In the industrial process, TiB2 is synthesized by carbothermal reduction of boron oxide in the presence of titanium dioxide [22,32]: Ti02 + B 2 0 3 + 5C -+ TiB2 + 5CO. TiB2 crystallizes with A1B2 type structure [46,331 (space group P6/mmm, Fig. 7), where the boron atoms fill trigonal prisms formed by the titanium atoms. Each boron atom has three boron neighbors in a trigonal planar arrangement, forming a two-dimensional honeycomb network with B-B distances of 0.175 nm. TiB2 has excellent thermal and chemical stability up to 1700°C. It has a relatively low density of 4.52gcmP3 and a high melting point of 3225°C. ZrB2, HfB2, VB2, NbB2, CrB2, MoB2, and WB2 are isotypic with TiB2 and A1B2; however, most of

10

I Structural Chemistry of Hard Materials

Figure 6. Crystal structure of B,O. On the right-hand side of the figure one layer of oxygen-bridged (empty) B12icosahedra is outlined. The hexagonal cell of this rhombohedra1 structure and the stacking of subsequent layers is presented at the left-hand side.

these diborides have a higher density and their production is more expensive. Nevertheless, some of them are used for special applications [22,32]. Apart from the diborides, also some monoborides are technically important. Most of them are used as hard components in composite materials. One notable compound is FeB. It has a remarkable adhesiveness on iron-based materials and thus increases their hardness. Such hard coatings are produced via CVD techniques. As an intermediate product elementary boron is obtained through the reaction 2BC13 3H2 -, 2B 6HC1 between 500 and 900°C, which subsequently reacts with iron to form the FeB coating [32,34]. The transition metal monoborides crystallize with three different structure types. The orthorhombic structures of CrB [35] and FeB [36] are presented in Fig. 8. The boron atoms have a trigonal-prismatic metal environment and two near boron neighbors, thus forming infinite boron zig-zag chains. The B-B distances are comparable to the shorter B-B distances in the modifications of elemental boron. The Cr-B and Fe-B distances within the trigonal prisms range from 0.212 to 0.223 nm. Each boron-centered trigonal prism shares two rectangular faces with neighboring prisms forming infinite prism columns, which extend along the z direction in CrB and along y in FeB. The difference between the two structures is

+

+

Figure 7. The AlB, type structure of TiB,. A projection along the hexagonal axis is shown on the right-hand side and a perspective view is presented on the left. The two-dimensional boron network is emphasized in both drawings.

1.3 Crystal Chemistry of Borides and Boron Carbides

11

X C - 4V

CrB

FeB

Figure 8. The crystal structures of CrB (space group Cmcm) and FeB (Pnrna). Each boron atom in these orthorhombic structures is located in a trigonal prism of metal atoms. Each prism shares two rectangular faces with adjacent prisms. Thus, the boron atoms form infinite zig-zag chains.

the way in which the infinite columns are connected with each other (Fig. 8). The monoborides VB, NbB, TaB, WB, NIB, and the high-temperature beta modification of MOB also crystallize with CrB type structure, while TiB, MnB, and COB are isotypic with FeB [6]. The structure of (low-temperature) a-MOB[37] is closely related to that of CrB. It is also composed of infinite columns of boron-centered molybdenum prisms (Fig. 9). However, in contrast to the CrB type, the boron zig-zag chains are now rotated by 90" with respect to each other (as a consequence of the four-fold screw axis). The Mo-B distances within the trigonal prismatic units range between 0.223 and 0.265 nm. The infinite boron zig-zag chains have B-B distances of 0.174 nm. WB is isotypic with a-MOB [37]. The rare earth hexaborides, well known for their superior electron emitting properties [38], are also quite hard materials. The hardness of YB6 is reported to be comparable to that of TiB2 [32]; however, such borides are quite expensive. YB6 is isotypic with CaB6 and LaB6 [3840], where the boron atoms form B6

Mo B

Figure 9. Crystal structure of tetragonal wMoB (space group I4,lumd).

12

1 Structural Chemistry of’ Hard Materials

X

J-Y

Figure 10. Crystal structure of cubic YB6 (space group Pm3m). The three-dimensional network of B, octahedra and the B24 coordination polyhedra around the yttrium atoms are emphasized.

octahedra which are bonded to six other octahedra forming a three-dimensional cubic network (Fig. 10). The yttrium atoms occupy the large cavities formed by this network. They have 24 boron atoms in their coordination shell. Alternatively, the YB6 structure can be described as a CsC1-like packing of yttrium atoms and B6 groups.

1.4 The Structures of Transition Metal Carbides Cemented carbides form one of the most important groups of hard materials [41,42]. The carbides WC, Tic, and TaC are the technically most important ones. They are produced in amounts of several thousand tons per year. VC, NbC, ZrC, HfC, Mo2C, and the chromium carbides are somewhat less important. The extreme hardness and high melting points of many transition metal carbides were already recognized in the 1890s by Moissan [43]. The transition metal carbides, especially WC, are used as the hard components in cutting tools, wire drawing dies, drills, and other wear resistant materials. For this purpose they are embedded in a binder matrix of a lower melting metal of the iron group, mainly cobalt, which supplies the necessary tenacity. The composition of these composite materials is around 92% WC, 8% Co. Usually these cemented carbides are prepared by powder metallurgical techniques, starting from the elemental components. Alternatively one can reduce the respective transition metal oxides with pure graphite at about 2O0O0C, for example Ti02 + 3C + T i c 2CO. Very non-abrasive coatings of TIC can be obtained by chemical vapor deposition, e.g. Tic& + CH4 + T i c + 4HC1 [22]. WC is the technically most important transition metal carbide. It adopts a very simple structure (Fig. 11, space group P6m2) [44,45], which may be described as a defect-A1B2 structure in which every other boron position is unoccupied. This results in a symmetry reduction. In space group P6m2 (a translationengleiche subgroup of index 2 (t2) of P6/mmm) the two-fold 2d position of P6/mmm is reduced to two one-fold positions. This allows the introduction of ordered defects. The

+

I .4 The Structures of Transition Metal Carbides

13

W OC

Figure 11. Crystal structure of WC. A projection along the hexagonal c axis is shown on the lefthand side of the drawing and a view of the unit cell is presented on the right-hand side. The trigonal W6C and WC6 prisms are outlined.

carbon atoms in WC are isolated (no C-C bonds) within the trigonal tungsten prisms. This is a significant difference when compared with the various boride structures discussed above. The filled prisms are stacked one upon each other by sharing common triangular faces. Thus, filled and unfilled columns alternate. The W-C and shortest W-W distances are 0.220 and 0.284 nm, respectively. Besides the trigonal prismatic coordination of the carbon atoms, octahedral coordination is observed, especially for the carbon atoms in the carbides of the early transition metals [46]. Important examples are the NaCl type carbides TIC, ZrC, HfC, VC, NbC, and TaC [42], where the metal atoms have a cubic closepacked arrangement, and the tungsten carbide W2C, where the tungsten atoms form a hexagonal close packed structure [47,48]. It is well known that close packed atoms form octahedral voids. The number of these voids equals that of the number of close packed atoms. In the NaCl type structure of TIC all octahedral voids formed by the cubic close packed titanium atoms are filled by carbon atoms, and the Ti& octahedra share only edges. In contrast, in W2C the tungsten atoms form a hexagonal close packed arrangement and the carbon atoms fill only half of the octahedral voids. The w6c octahedra are linked not only via common edges, but also by common faces. The Parthe-Yvon rule states that carbon atoms of transition metal carbides occupying octahedral voids of face sharing transition metal octahedra, never occupy both of these adjacent octahedra [47]. This is demonstrated in Fig. 12 using the Jagodzinski-Wyckoff notation for the close packed metal atoms. For every close packed transition metal ( T )layer with an h type packing only one of the adjacent face sharing octahedral voids can be filled. For the carbide with the composition T2C this unoccupied void is marked with an asterisk in Fig. 12. Examples for the carbide structures shown in that figure are W2C with (anti-)Cd12 type structure Mo3C2 (hcc), (Ta2V)C2 (hhc), V4C3, and TIC [47, 491. Perspective views of the TIC (space group Fm3m) and W2C ( P h l ) structures are shown in Fig. 13 with an emphasis on the octahedral carbon coordination. The Ti6C octahedra in TIC share all edges, while edge-sharing only within the ab plane is observed for the W6C octahedra in W2C. It is interesting to compare the structures of WC and W2C. The main difference between these two tungsten carbides is the carbon coordination: trigonal prisms of tungsten atoms in WC and octahedra in W2C with W-C distances of 0.220 and

14

1 Structural Chemistry of Hard Muteriuls

I

I

h'

h' 4

*

'h

T2C h

T3C2

T3C2

T4C3

TC

hcc

hhc

hhcc

C

Figure 12. Stacking sequences of the hexagonal metal atom layers in several carbides with close packed transition metals [47]. The structures are represented by cuts along the (1 10) planes of the respective hexagonal cells. The transition metal (7') and carbon atoms are drawn as large open and small filled circles, respectively. A 'forbidden' carbon position, discussed in the text, is marked with an asterisk in the structure of T2C.

0.209 nm, respectively. The difference in the W-C distances may be rationalized by geometrical considerations: in a hard sphere model (equal W-W distances) trigonal prismatic voids are larger than octahedral voids. However, there are also differences in the coordination numbers: the tungsten atom in WC has six carbon neighbors, in W2C it has only three. In addition to the W-C bonding both compounds have strong W-W bonds. Each tungsten atom has eight (WC) or twelve (W2C) tungsten neighbors with W-W distances ranging between 0.284 and 0.300 nm, comparable to the W-W distances in the cubic body centered structure of the element

Tic

w2c

Figure 13. Crystal structure of cubic T i c (NaC1 type) and hexagonal W2C (Cd12 type). Some carbon-filled Ti6 and W6 octahedra are emphasized.

1.4 The Structures of Transition Metal Carbides

15

(8 x 0.274 nm and 6 x 0.316 nm). W,C has a carbon content of only 3.2 weight-%. It has a larger microhardness than WC, but it is also more brittle [42]. The transition metal carbides with the composition T2C are sometimes called hemicarbides. In addition to tungsten, such carbides are also formed by molybdenum, vanadium, niobium, and tantalum. A large number of somewhat differing crystal structures have been reported for these, all with very similar X-ray patterns, corresponding to the hexagonal close packed arrangement of the metal atoms. The differences arise through slight distortions and/or larger unit cells of lower symmetry, which result through differing order of occupied and empty carbon positions [50]. Usually the Parthk-Yvon rule (of two face sharing metal octahedra only one can be occupied by carbon atoms) is obeyed. The different structures may be stable in differing ranges of temperature and they may also be stabilized by interstitial impurities like nitrogen or oxygen. Besides the systematic carbon defects due to the hexagonal stacking sequences, carbon defects also occur by nonoccupancy of ‘allowed’ octahedral sites. Thus, the compositions TC, T3C2, T4C3,and T2C discussed above are the upper limits of the homogeneity ranges. Cubic titanium carbide for example has a homogeneity range from TIC to about TiCo.5.At high temperature the defects are randomly distributed, while an ordering of the defects is to be expected for thermodynamic reasons at low temperature. As examples, partial phase diagrams of the systems titanium-carbon and vanadium
16


-

I

9

3200

-s

3000

3200

Y

3

4 Liquid

2800

2800

2600

2600

2400 2200

2400 2200

2000

2000

1800

1800

1600

1600

1400

1200

1400 1200

1000

1000

800 600

800 600 0 10 20 30 40 50 60 70 Ti

atomic % carbon

-

0.70 0.75 0.80 0.85 0.90 0.95 C/V Ratio

42 43 44 45 46 47 48 atomic % carbon

Figure 14. Partial phase diagrams of the systems titanium-carbon [51] and vanadium-carbon [52]. In the latter diagram, regions with short-range (SRO) and long-range order (LRO) are indicated.

interactions. For these calculations usually full occupancy of all atomic sites of the NaCl type structure in assumed, a supposition which may not be entirely correct. In NbO with a defect-NaC1 type structure 25% of the niobium and oxygen positions are vacant, and the vacancies are ordered [75]. Carbon plays an important role in steels, both in the form of solid solution and as component of carbide phases. The cubic face centered modification of iron (y-Fe) dissolves as much as 8 at-% (2 weight-%) carbon, which randomly occupies the octahedral voids of the cubic close packed iron atoms. This cubic phase is called austenite. On quenching, this phase distorts by a displacive (‘martensitic’) phase transition to the corresponding tetragonal structure (martensite). The degree of the distortion is proportional to the carbon content. For a carbon content of zero, the distortion extrapolates to zero, that is, the lattice constants of the (tetragonal) martensite extrapolate to those of pure y-Fe [76,77]. Cementite Fe3C [78] plays an important role in steels. This iron carbide is significantly harder than pure iron. Its orthorhombic crystal structure is presented in Fig. 15. The best structure refinement of a cementite type structure seems to have been done for Fe2,7Mno.3C[79]. The carbon atoms occupy trigonal prismatic voids of the iron atoms with Fe-C distances of 0.201 and 0.202nm. Three additional iron neighbors at 0.237(2 x) and 0.28 1 nm complete the coordination shell of the carbon atoms, resulting in tricapped trigonal prisms. The trigonal prisms are condensed via common triangular faces. Columns of such trigonal


17

Figure 15. Crystal structure of cementite, Fe3C (space group Pnma). The carbon filled trigonal Fe6 prisms and one empty Fe6 octahedron are emphasized.

prisms are connected via Fe-Fe bonds forming empty Fe6 octahedra. The Fe-Fe distances between the two crystallographically different iron atoms range from 0.250 to 0.268 nm. The shorter ones compare well with the Fe-Fe distance of 0.248 nm in b.c.c. iron [l 11. The various chromium carbides are relatively hard and brittle. They significantly increase the hardness and pyrophoric stability of carbon rich hard materials. These compounds are known as Stelliten. The hardness of alloyed steels [9] results from several chromium-iron double carbides of compositions (Fe, Cr)3C2,(Cr, Fe)23C6r and (Fe, Cr)7C3.These mixed carbides crystallize all in the structures of the respective pure chromium carbides with a mixed occupancy of the chromium positions by chromium and iron atoms. Cr3C2 [80] is the technically most important chromium carbide. It is used as a cemented carbide with nickel as binder [41]. Its orthorhombic structure is shown in Fig. 16. The striking structural motifs of Cr3C2 are carbon centered trigonal prisms of the chromium atoms. These trigonal Cr6C prisms are connected by sharing common faces as well as common edges. The two crystallographically different carbon atoms in Cr3C2have Cr-C distances from the center to the corner of the

Figure 16. Crystal structure and coordination polyhedra of the chromium atoms of Cr3C2 (space group Pnma). The carbon filled trigonal Cr6 prisms are also emphasized.

18


Crl

CR

Cr3

C r4

Figure 17. Projection of the crystal structure of cubic Cr23C6(space group Fm3m). Some carbon filled square antiprisms are emphasized at the left-hand side. The coordination polyhedra of the chromium atoms are also shown.

prisms ranging from 0.202 to 0.213 nm. In addition, the C1 atoms have two further chromium neighbors outside the rectangular faces of the prisms at 0.229 and 0.262nm, and the C2 atoms have a seventh chromium neighbor at 0.226nm. The shortest Cr-Cr distances in Cr3C2 of 0.250nm are exactly as long as in b.c.c. chromium [l 11. Again, the stability of this peculiar carbide structure arises from strongly bonding metalkarbon and metal-metal interactions. The cubic structure of the chromium carbide with the highest metal content, Cr23C6[81] is shown in Fig. 17. Here, the carbon atoms occupy square antiprisms of the chromium atoms with Cr-C distances of 0.21 1 and 0.213 nm. The square antiprisms are condensed via common edges and common corners. The Cr-Cr distances for the four crystallographically different chromium atoms range from 0.242 to 0.293 nm. Each chromium atom has at least four chromium neighbors at short distances. The shortest Cr-Cr contacts are significantly shorter than in b.c.c. chromium (0.250 nm) [ 1 11, indicating strongly bonding Cr-Cr interactions. In the structure of Cr13C6 the carbon filled square antiprisms are built by the Cr3 and Cr4 atoms. Crl and Cr2 have exclusively chromium neighbors in their coordination shell. The Crl atoms have twelve chromium neighbors in a cuboctahedral environment, while the 16 chromium atoms around the Cr2 atoms form a FrankKasper [82,83] polyhedron. The structure of Cr7C3[84] is presented in Fig. 18. Structural elements are carbon filled trigonal prisms of chromium atoms which are condensed via common edges and corners. The Cr-C distances within the trigonal prisms range from 0.197 to 0.21 1 nm, essentially the same range as in Cr3C2.All chromium atoms in Cr7C3 have at least three carbon neighbors and the Cr-Cr distances range from 0.249 to 0.276nm. The predominant features of chemical bonding in Cr7C3 are similar to CrZ3C6and Cr3C2,that is, strong Cr-C and Cr-Cr interactions. Besides the chromium-iron carbides, the so-called q-carbides, like W3Fe3C [85,86] and W3C03C [87] play an important role in steels [42]. They are not separately produced. The q-carbides (and nitrides) comprise a wide variety of different compositions T3T$X,T 6 T i x , and T4TiX, where T = Mo, W; T' = Fe,


19

Figure 18. Crystal structure of orthorhombic Cr7C3(space group Pnma). The carbon filled trigonal prisms are emphasized.

Co, Ni; X = C (and sometimes also nitrogen or even oxygen, e.g. Ti4Fe200.4[SS]). In the cubic structure of the compounds T3T&Y and T6TLX the metal atoms occupy the same atomic positions, but the structures differ in their X positions, which are either sixteen- or eightfold [86,89]. In both cases, however, the C(N) atoms occupy octahedral voids formed solely by the molybdenum or tungsten atoms. As an example, the structure of W3Fe3C is presented in Fig. 19. At first sight it seems very complex; however, a description by two independent, interpenetrating three-dimensional networks makes this structure type comprehensible. One network consists of corner-shared stelh quadranguh [90] of iron atoms (right-hand part, Fig. 19), and the second network is composed of corner-sharing w6/2c octahedra. The latter network is different from that known for the Re03 structure. The two

Figure 19. Crystal structure of W3Fe3C.In the middle of the drawing all atoms of this cubic structure (space group Fd3rn) are shown. The carbon-filled octahedra of tungsten atoms are emphasized on the left-hand side, only the condensed Fed tetrahedra are drawn on the right-hand side of the figure.

20


crystallographically different iron atoms have six iron neighbors each, at Fe-Fe distances ranging from 0.238 to 0.250nm. In view of the Fe-Fe distance of 0.248 nm in b.c.c. iron [l 13, these contacts in W3Fe3C certainly indicate strong Fe-Fe bonding. In addition, there is W-W and W-Fe bonding. As it turns out, in most ternary phases the carbon atoms are surrounded by those metal atoms forming the more stable (i.e. higher melting) carbide. The W-C distances of 0.206nm within the W6& octahedra of W3Fe3C, are essentially the same as in the W6/3Coctahedra of W2C(0.209 nm) [48]. The remarkable segregation of the W3Fe3Cstructure into an iron and a W6/2Csublattice parallels the stability of the binary carbides; that is, high-melting WC and metastable Fe3C.

1.5 Silicides and Silicide Carbides of Transition Metals The melting points and the hardness of transition metal silicides are lower than those of the respective carbides [22]. Nevertheless, some of these silicides find applications as hard materials, since they exhibit a remarkable resistance against acids and they are also stable in air up to high temperatures because they form dense coatings of silica and silicates. A large disadvantage of the silicides, however, is their extreme brittleness. Today mainly MoSi, and the isotypic tungsten compound WSi, have technical importance. They are used mostly for protective coatings for molybdenum and tungsten wires and sheets. For this purpose MoSi2 can be prepared by CVD techniques: Mo 2SiC14 4H2 + MoSi, 8HC1. Their use as components of cermets is less important. TiSi2 has gained some importance for the construction of turbines [91]. Three closely related structure types were found for the transition metal disilicides: TiSi2 [92], CrSi2 [93], and MoSi2 [94]. The common structural motif of the three structures is a two-dimensional close-packed layer of composition 7Si2 as outlined in Fig. 20. Within these layers each transition metal atom has six silicon neighbors. The structures of TiSi2, CrSi2, and MoSi, then result from different stackings of these 7Si2 layers. The stacking, however, is not close-packed. The layers are stacked in such a way that the transition metal atoms of one layer are situated (in projection) between two silicon atoms of an adjacent layer. Thus, the transition metal atoms fall above the points B, C, or D, indicated as small dots in Fig. 20. In this way each transition metal atom obtains ten silicon and four transition metal neighbors. Each silicon atom has five transition metal and five silicon neighbors. The stacking sequences are AB, AB for MoSi,, ABC, ABC for CrSi2, and ABCD, ABCD for TiSi2. Again, it is important to note that these structures should not be confused with the close-packed layers described earlier, in which a stacking AB, AB or ABC, ABC is hexagonal or cubic, respectively, for dense packed layers. With the disilicides these correspond to tetragonal (AB, AB) and hexagonal (ABC, ABC) structures. For a more familiar view of these structures we now turn to the unit cells. The tetragonal crystal structure of MoSi2 (space group I4/mmm) is presented in Fig. 21. It may also be considered as derived from the body-centered structure of

+

+

+

I S Silicides and Silicide Carbides of Transition Metals

21

MoSi,, AB

CrSi,, ABC

TiSi,, ABCD

Figure 20. The crystal structures of MoSi2, CrSi,, and TiSi2. All of these structures contain hexagonal close-packed layers TS4; one of these is shown in the upper left-hand corner. The positions of the transition metal atoms of adjacent layers above and below are marked with small black dots in this drawing. The three structures have the stacking sequences AB (MoSi,), ABC (CrSi2), and ABCD (TiSi,), respectively. The hexagons connecting the silicon atoms around a transition metal atom serve as guides for the eye.

tungsten by an ordered arrangement of the molybdenum and silicon atoms on the tungsten sites. This leads to a tripling of the tungsten subcell. Due to the different sizes of molybdenum and silicon the cubes are strongly compressed ( c / o = 2.45 instead of 3.0 for the ideal arrangement). The molybdenum atoms have ten silicon

MoSi,

CrSi,

TSi,

Figure 21. Crystal structures of tetragonal MoSi2, hexagonal CrSi,, and orthorhombic TiSi,. The C N 10 polyhedra of silicon atoms around the transition metal atoms and parts of the twodimensionally close packed 7Si2 layers are outlined for each structure.

22


neighbors in the form of a bicapped square prism. These polyhedra are condensed via common edges and faces. The unit cells of the orthorhombic TiSi2 (space group Fddd) and hexagonal CrSi2 (space group P6222) structures are also presented in Fig. 21. For both structures, the 7Sil0 polyhedra and parts of the T-centered Si6 rings are outlined. The Si-Si distances range between 0.247 and 0.279 nm. These bonding Si-Si distances are all significantly longer than the Si-Si bond length of 0.235 nm in elemental silicon [I 11. This has to do with the fact, that the coordination number (CN) of the silicon atoms in the transition metal disilicides is considerably higher (CN 10) than in the diamond structure of elemental silicon (CN 4). Besides Ti5Si3C1- with partially filled Mn5Si3type structure (see below), Ti3SiC2 is the only other phase in the ternary system Ti-Si-C [95-971. Ti3SiC2crystallizes in the form of thin platelets which can be plastically deformed [98]. It combines refractoriness (m.p. M 3200°C) and stability against aggressive media with thermal shock resistance. A high potential as a component of ceramics and composite materials was discussed [99,100]. The electronic structure of Ti3SiC2 was investigated by self-consistent full-potential linear muffin-tin orbital calculations [ 1011. The structure of Ti3SiC2is shown in Fig. 22. The titanium and silicon atoms are close packed with the stacking sequence ( h h h ~ )Thus, ~ . Ti3SiC2adopts a structure typical for a metal-rich carbide. The titanium octahedra are filled by carbon atoms, and these Ti6C octahedra are condensed via common edges, forming double layers which are separated by the silicon atoms. No siliconxarbon bonds are observed. In compounds of similar composition Si-C bonds (0.193 nm) have been found only in U3Si2C2[102].

Figure 22. Crystal structure of Ti3SiC2 (space group P6,lmmc). The close-packed titanium layers containing the carbon atoms are shown on the left-hand side. The edge-sharing Ti& octahedra are emphasized. On the right-hand side a (110) cut through the hexagonal cell is shown. The titanium and silicon atoms form close-packed layers with the stacking sequence hhhc where the silicon atoms correspond to the second h.

1.6 Nitrides

Mol

23

Si

Mo2 C Figure 23. Crystal structure of Mo5Si3C(space group P6,lmcm). On the left-hand side of the drawing a projection of the structure along the hexagonal z axis is presented. The carbon atoms (not shown) are situated inside the emphasized Mo6 octahedra. These share faces, thus forming infinite chains, which extend along the hexagonal axis. The near-neighbor coordinations of all atoms are presented at the right-hand side of the figure.

The molybdenum silicide carbide Mo5Si3C [I031 belongs to a large family of compounds with structures derived from the well known Mn5Si3 type structure [104], also called D8* or Nowotny phases. The carbon atoms of Mo5Si3C occupy octahedral voids formed by the molybdenum atoms. Today more than 300 compounds [5] with unfilled, filled, or partially filled Mn5Si3 structure are known. As is usually the case for compounds with a high content of metal atoms, in the structure of Mo5Si3C all atoms have high coordination numbers. The Mo2 atoms (Fig. 23) form strings, which extend along the c axis with rather short Mo2-Mo2 distances of 0.252nm. In addition, the Mo2 atoms have 6 Si and 6 Mol neighbors. The silicon atoms are coordinated by 9 Mo and 2 Si atoms. The Mol atoms form chains of facesharing octahedra, which are surrounded by the Si and Mo2 atoms. These octahedra are filled with carbon atoms. According to the Parthe-Yvon rule [47] the carbon positions should not be fully occupied and indeed the structure determination of this compound showed considerable deviations from the ideal composition corresponding to the formula Mo4.8Si3C0.6[ 1031. The structural chemistry of compounds with filled Mn5Si3 structure is complex, because the octahedral voids cannot only be filled by small impurity atoms like carbon, nitrogen, or oxygen, but also by transition metal and main group elements, such as Hf5Sn3Cu [lo51 and Ti5Ga4[106]. This gives many possibilities for deviations from the ideal composition.

1.6 Nitrides In elemental form, carbon has many applications in both modifications as diamond and as graphite. In contrast, the neighboring element nitrogen is a gas. Nevertheless,

24


a-Si3N4

P-Si3N4

Figure 24. Crystal structures of trigonal a-Si3N4 (space group P31c) and hexagonal P-Si3N4 (P63/m), emphasizing the corner-sharing SiN4 tetrahedra.

both elements form hard materials, which are structurally closely related, and both form such refractory compounds with main-group elements (e.g. Sic, Si3N4)and with transition metals (e.g. WC, TiN).

1.6.1 Nitrides of Main Group Elements Silicon nitride, Si3N4,is the technically most important nitride of the main group elements. It is dimorphic. Usually both forms are obtained together during the preparation, for example by reaction of the elements at 1200°C. The low temperature a-modification is transformed to the P-modification by heating above 1650°C. However, the inverse reaction has not been observed. At low preparation temperatures the a-modification predominates. The proportions of a-Si3N4 and p-Si3N4in mixtures can reliably be determined by X-ray powder diffraction. A high amount of a-Si3N4is desired, since it significantly increases the sintering activity [ 107,1081. The crystal structures of a-Si3N4 [lo91 and p-Si3N4 [7,110] are presented in Fig. 24. Both modifications have a complex three-dimensional network of cornersharing SiN4 tetrahedra with an average Si-N bond length of 0.174nm. The structure of the low-temperature modification has two crystallographically different silicon atoms, while only one silicon position occurs in the high-temperature modification. In both modifications the nitrogen atoms have more or less trigonal-planar carbon coordination. The high stability of silicon nitride is due to a complex interplay of predominantly covalent bonding and a high degree of condensation of the corner-sharing SiN4 tetrahedra (three tetrahedra share one common corner). The high corrosion stability of Si3N4results from the formation of a thin surface coating of SO2. In mixtures with aluminum oxide, Si3N4 is used as Sialon mixed ceramic Si3N4-A1203. Today, Si3N4 is the most frequently utilized nitridic material. It is used for the construction of bearings, pistons, and turbines.

1.6 Nitrides

25

Another main group element nitride with considerable hardness is aluminum nitride, AlN. It is important as a ceramic insulator with high thermal conductivity. It has the hexagonal wurtzite type structure (h-ZnS) [l 111. In contrast to the highpressure compound c-BN, it is prepared by various methods at ambient pressure [22, 1121. Many attempts have been made in recent years to prepare carbon nitrides with the ideal composition C3N4 in analogy to Si3N4 [113-1171. So far only thin offstoichiometric films with considerable nitrogen deficiency have been obtained. Ab initio calculations have predicted that this nitride could be isomorphic with (a-Si3N4 and/or b-Si3N4. A cubic modification of C3N4 with defect-zinc blende type structure has also been discussed, and it was predicted that such a compound could be even harder than diamond. This, however, is improbable for the simple structure-chemical reason, that a compound containing only carbon and nitrogen has too many electrons. In all of the structures proposed for C3N4 the nitrogen atoms are three-coordinate with a nonbonding (lone) pair of electrons as the fourth ‘ligand’, thus not utilizing one potential coordination place. In contrast, the diamond structure has the ideal number of four electrons per atom, and all of these can be used for bonding interactions.

1.6.2 Transition Metal Nitrides Transition metal nitrides are used as hard surface coatings. They also play a role in steels. The most important compounds are the NaCl type mononitrides of titanium, zirconium, hafnium, niobium, and tantalum. They are isotypic with the corresponding carbides (Fig. 13). Titanium nitride can be obtained by reduction of titanium dioxide: 2Ti02 N2 4C + 2TiN 4CO. For industrial applications several composite ceramics like A1203/TiN/TiC and Si3N4/TiN have been developed. Surface coatings of titanium nitride are produced via CVD techniques by the gas phase reaction 2TiC14 H2 2NH3 + 2TiN + 8HC1. Such TIN coatings have yellow color, while mixed nitrides like TiN/AlN are blue. The color of such coatings can be varied by a substitution of some nitrogen by carbon or oxygen atoms. It also depends on the concentration of unoccupied nitrogen sites x of TiNl --x.Thin films of TIN are used as hard, abrasion-proof surface coatings with almost golden color, also for decorative purposes, like for watches. An overview of the industrially important nitrides was given by Benesovsky [118]. The mononitrides and nitride carbides have, like the carbides discussed earlier, large homogeneity ranges, for example the composition of the NaCl type titanium nitride TiN, p x extends from stoichiometric TiNl.o to TiNo,5[119]. At low temperatures the vacant nitrogen sites can be expected to be ordered, and indeed a tetragonal superstructure with a doubled c axis has been reported for this phase for the composition (Ti2N) [120]. The degree of the nitrogen deficit strongly influences the hardness and the melting point of these compounds. Also, the nitrogen atoms in TIN can to some degree be replaced by carbon and/or oxygen atoms resulting in Ti(C, N, 0) materials. Chemical bonding in these nitrides is similar to that of the corresponding carbides TIC and HfC [71-741.

+ +

+ +

+

26


X

A

B

A

B

A

B

Figure 25. Crystal structure of c-Fe2N. The iron atoms form hexagonal close-packed layers which extend along the yz-plane of this orthorhombic structure (space group Pbcn). In the x direction these planes have the stacking sequence AB, AB. Hence, the iron atoms form a distorted hexagonal closepacked arrangement where the nitrogen atoms fill half of the octahedral voids.

Nitrogen also plays a role in steel, not only as a component of solid solutions in carbides, but especially in the form of hard surface coatings. The crystal structures of the respective hard materials are not all well known. In the present section we discuss some representative structures that were determined reliably. The nomenclature of the iron nitrides derives from the phase diagram of the binary system iron-nitrogen. Thus, the a- and y-phases are solid solutions of nitrogen in the aand y-modifications of iron. The structure of E,-Fe2N [I211 was established recently from a combination of neutron diffraction and high-resolution synchrotron powder data. c-Fe2N crystallizes with anti-a-PbO, type structure as outlined in Fig. 25. The structure consists of edge- and corner-sharing Fe6N octahedra with Fe-N distances ranging from 0.189 to 0.201 nm. The iron atoms form a hexagonally close-packed array and the nitrogen atoms fill half of the octahedral voids in an ordered manner. The structures of E-Fe3N and y'-Fe4N are also based on close-packed arrangements of iron atoms [122]. In &-Fe3Nthe iron atoms are hexagonal close-packed and nitrogen atoms occupy only corner-sharing octahedra. The ordered arrangement of the nitrogen atoms results in a & times larger a axis. In this structure the deviations of the iron atoms from the atomic positions in an ideal hexagonal close-packed structure are small. The iron atoms in y'-Fe4N have a face-centered cubic arrangement, and the nitrogen atoms occupy octahedral voids (Fig. 26). This results in a three-dimensionally infinite arrangement of all-corner-sharing Fe6N octahedra with Fe-N distances of 0.190nm, slightly shorter than the average Fe-N distance of 0.195nm in E,-Fe2N. Since the nitrogen atoms occupy only one of the four octahedral voids formed by the face-centered cubic iron atoms, the space group symmetry is lowered from Fmjm to Pm%. In addition to the congruently melting (m.p. x 3300°C) NaCl type phase TiNl - -x (x = 0.5 to 1.O) the phase diagram of the system titanium-nitrogen contains at least three other nitrides, all of which are stable at temperatures of less than 1300°C[119]. Of these the nitride Ti2N crystallizing with (anti-)rutile type structure (Fig. 27) is known for some time [123]. In this structure the titanium atoms form octahedra which are filled by nitrogen atoms. The Ti6N octahedra are linked via common

1.6 Nitrides

Fe4N1-x

27

li3AIN

Figure 26. The crystal structures of y'-Fe4N and Ti,AIN (space group P m h ) . In the binary nitride the nitrogen atoms occupy one quarter of the octahedral voids formed by the cubic close-packed iron atoms. In the ternary compound the nitrogen atoms fill those octahedral voids which are formed solely by the titanium atoms.

edges, thus forming chains which extend along the c axis of the tetragonal cell (Fig. 27). These chains are further condensed by sharing corners, and in this way a three-dimensional network is formed. For the other two titanium nitrides the structure determinations resulted in the compositions Ti3N2- and Ti4N3- x. These nitrides are stable at only very limited temperature ranges between 1050 and 1300°C [119,124,125]. Considering composites and thin-film coatings containing TIN and AlN, one has to be aware of the ternary compounds Ti3A1N [126], Ti2AlN [127], and Ti3A12N2 [126]. Of these the latter is thermodynamically stable only between 1200 and 1300°C. In all three structures the metal atoms are close packed. While the structure of Ti3A12N2has a large cell with a ten-layer stacking sequence, the metal positions in Ti3AlN and Ti2AlN correspond to cubic (ABC) and hexagonal (AB) close packed structures (Fig. 28). In both of these the nitrogen atoms are situated in octahedral voids formed solely by the titanium atoms. The cubic structure of Ti3A1N is known as perovskite (CaTi03 type) structure and Ti2AlN is an H phase (Cr2A1C type). There are many representatives of these carbides T3MX and T2MX where T are early transition metals, mainly from groups 14 and 15, M are main-group elements, especially Al, Ga, In, Sn, Pb, and X = C, N [5, 128-1311. Layers of zirconium and aluminum atoms can also be discerned in the structure of Zr3AlN [132]. In this complex nitride the nitrogen atoms are again situated in

Figure 27. Crystal structure of Ti2N with tetragonal (unti-)rutile type structure (space group P42/mnm).The edge- and corner-sharing Ti6N octahedra are emphasized.

28


I

Y41lOfl

YL

X

Ti,AIN

Ti,AIN

Figure 28. Crystal structure of Ti3AlN (CaTiO,, perovskite type) and Ti2AlN (Cr2AlC, H phase type structure). In both structures the metal atoms are close packed. In the cubic perovskite type structure the positions of the close-packed metal atoms correspond to those of the Cu3Au structure (space group Pm3m) with the stacking sequence ABC, ABC. In the hexagonal Cr2A1Ctype structure of Ti2AIN the stacking sequence of the close-packed metal atoms is AB, AB.

octahedral voids formed solely by the transition metal atoms. These Zr6N octahedra share edges and corners, thus forming two-dimensionally infinite sheets with aluminum layers in between them (Fig. 29). It is striking that the carbon and nitrogen atoms in complex carbides and nitrides, containing transition metal and main-group elements, are usually situated in octahedral voids formed by the

i “ I

“ I

” ‘ I

“ I

“ I

W

Figure 29. Crystal structure of Zr,AlN (space group Cmcm). The layers of edge- and corner-sharing Zr6N octahedra are emphasized.

1.6 Nitrides

29

transition metals. Other examples already discussed above are the structures of Ti3SiC2 [98] and Mo5Si3C [103]. Another example is the filled P-manganese type structure of Nb3A12N and Mo3A12C [133]. Also, ,as already mentioned, in the qW3Fe3C type carbides and nitrides the carbon atoms are situated in octahedral voids formed by those transition metal atoms which form the more stable (highmelting) binary carbides. It can be assumed, that these octahedral Tgc and T6N groups are very stable and that they possibly exist already in the melts.

1.6.3 Perspectives: Nitridosilicates Nitridosilicates of the alkaline earth and rare earth metals have been investigated intensively in the past five years. A recent review by Schnick and Huppertz [134] summarizes the rich crystal chemistry and some potential applications for this class of compounds. Such nitridosilicates can be synthesized via two different routes. One possibility is the direct reaction of binary silicides under flowing nitrogen at high temperature like 3CeSi2 + 1 1/2N2 + Ce3Si6NI1[135]. A more sophisticated method is the reaction of silicium diimide with an alkaline earth element like Ba 7Si(NH)24 BaSi7Nlo+ 2N2 + 7H2. The latter reaction is carried out in tungsten crucibles under flowing nitrogen in a high-frequency furnace [136] at temperatures up to 1600°C. The nitridosilicates are highly stable compounds. They might find applications as hard materials possibly in combination with Si3N4. In their structures the silicon and nitrogen atoms form highly condensed networks of corner-sharing SiN4tetrahedra. The alkaline earth and rare earth atoms are embedded within the cages of these tetrahedral networks. The structures of Sm3Si6N11type Ce3Si6NII [135] and Ca2Si5N8[136] are shown as examples in Fig. 30. The Si-N distances in the various nitridosilicates range from 0.164 to 0.180 nm. The different connectivities of the SiN4 tetrahedra in this large family of structures are reviewed in reference 134.

+

e N

Ca2Si5N8

Ce,S,iN,

Figure 30. Crystal structures of monoclinic Ca2Si,N, (space group Cc) and tetragonal Ce3Si6NI (Sm3Si6NII type, space group P4bm). The three-dimensional networks of comer-sharing SiN4 tetrahedra are emphasized.

30


1.7 Oxide Ceramics 1.7.1 Hard Ceramics of Main Group Elements One technically very important hard ceramic is corundum, a-A1203.Due to its outstanding resistance to abrasion, wear, and corrosion, its high thermal stability and high electrical resistance, its compatibility with living tissue, and last but not least its availability and relatively moderate price, it has many applications, for example as polishing powder, material for crucibles, cutting tools, spark plugs, substrate for electronic circuits, and artificial hip-joints [18,22, 137, 1381. The gemstones ruby and sapphire are essentially a-A1203. The red color of ruby is due to a small amount of Cr203 in solid solution. Blue sapphire contains minor amounts of iron and titanium. Such crystals are grown also industrially by the Verneuil technique. Large amounts of aluminum oxide are produced for the production of aluminum. By heating A1(OH)3/A10(OH) mixtures to 400°C one obtains the relatively soft y-modification of A1203,while hard a-A1203is produced by heating above 800°C. In the crystal structure of a-A1203[I391 the oxygen atoms form a hexagonal close packed array in which the aluminum atoms occupy two-thirds of the octahedral voids. In Fig. 31 the AIOs octahedra are emphasized. These octahedra share faces, edges, and corners. Each oxygen atom has four aluminum neighbors in approximately tetrahedral arrangement (2 x 0.185 nm, 2 x 0.197 nm). Beryllium oxide is a high melting compound with an exceptionally high thermal conductivity, very low electrical conductivity, and a remarkable chemical stability. The breaking strength of Be0 is comparable to that of a-A1203, however, the industrial production of Be0 is more expensive. The high toxicity of beryllium compounds, for example abrasive dust of BeO, prevents a broad use of t h s material.

Figure 31. The rhombohedra1 crystal structure of C L - A ~(space ~ O ~group R3c). Pairs of face-sharing octahedra extend along the threefold axis. Octahedra of adjacent pairs share corners and edges. Two-dimensionally infinite layers of condensed octahedra may be discerned. Such packages are indicated by braces. At the right-hand side such a layer is projected along the [110] direction.

1.7 Oxide Ceramics

31

Figure 32. The wurtzite type structure of cL-BeO, space group P63nrc. The three-dimensional network of condensed Be04-tetrahedra is outlined. The stacking of these tetrahedra has the sequence AB, AB. The corresponding stacking ABC, ABC is known to occur for the zinc blende (sphalerite) structure (Fig. 3 ) .

Be0 is a dimorphic oxide. The low-temperature a-form (<2O0O0C) crystallizes with the hexagonal wurtzite type structure [140]. Each beryllium atom has a slightly distorted tetrahedral oxygen coordination (1 x 0.162 nm and 3 x 0,164 nm). These tetrahedra share all corners with other tetrahedra, thus forming a three-dimensional network of condensed Be04-tetrahedra. It can easily be seen (Fig. 32) that the z axis of this structure is polar: the Be04 tetrahedra all point in the direction of the c axis. The high-temperature p-form of Be0 is stable above 2000°C. It cannot be quenched. In its tetragonal structure [I411 the arrangement of the oxygen atoms is the same as in rutile, TiOz. Magnesia, MgO, is industrially produced by thermal decomposition of magnesium carbonate. It has a higher corrosion stability than a-A1203. Depending on the thermal treatment two forms of MgO are obtained: sintered magnesia through an annealing process between 1700 and 2000°C or molten magnesia prepared in an arc-melting furnace between 2800 and 3000°C [18]. Sintered magnesia is used as crucible material in metallurgical furnaces. Melted magnesia is very expensive and used only for special applications as an insulating material at high temperature. MgO crystallizes with NaCl type structure [5, 1421 (space group Fmjm) which is shown in Fig. 13 for titanium carbide. Each magnesium atom has octahedral oxygen coordination with Mg-0 distances of 0.210 nm. These MgOh octahedra are condensed via all edges. Another relatively hard corrosion resistant material is the complex oxide MgA1204 which is well known as spinel. Compared with a-A1203 spinel has a lower thermal shock stability. The crystal structure of spinel [143] can be derived from an almost ideal cubic close packed arrangement of oxygen atoms. The magnesium cations fill one eighth of the tetrahedral voids and the aluminum cations occupy one half of the octahedral voids. For electrostatic reasons the smaller A13+ cations fill the larger octahedral voids. Alternatively, one can describe the spinel structure with small building blocks (Fig. 33). Half of the magnesium atoms build an f.c.c. cell which consists of eight smaller cubes. The latter are alternatingly occupied by Mg04 tetrahedra and distorted A1404 cubes, denoted as A and B in Fig. 33.

32


4Mg

+

4Mg04

+

4A1404 =

8MgA1204

Figure 33. The crystal structure of cubic MgA1204(spinel, space group F d h ) . For details see text.

1.7.2 Transition Metal Oxides Of the transition metal oxides, the highest melting and most stable ones are those of the titanium group: Ti02, ZrOz, Hf02. Chemically related are the fluorite (CaF2) type compounds Ce02 and Tho2. In this context it is worthwhile to remember that for good reasons up to 1940 thorium was positioned directly under hafnium in the Periodic Table. These oxides have many applications, for example Ti02 is used as a white pigment, calcium- or yttrium-containing Zr02 as solid electrolyte, or Zr02 and Thoz as high-melting materials. They are also quite hard, although they are not used primarily as hard materials [18,221. Titanium oxide TiOz occurs as a mineral in three well-known modifications: rutile, anatase, and brookite. Today rutile is by far the most important white pigment (others are anatase, ZnS, ZnO, BaS04). As a pigment rutile, when compared to anatase, has superior quality. It has the highest density of the three forms of Ti02 and it has also certain applications as an inexpensive hard ceramic. To our knowledge the p-T diagram of TiO, is not firmly established. However, it is known that under normal conditions rutile is the most stable form. On heating both anatase and brookite irreversibly transform to rutile. For comparison, we briefly discuss the structures of all three modifications of Ti02. In all of these the titanium atoms have somewhat distorted octahedral oxygen coordination and the oxygen atoms have a trigonal, more or less planar titanium environment. The most simple structure of Ti02 is the rutile structure (Fig. 34). All of the Ti06 octahedra share two opposite edges with adjacent Ti06 octahedra, in this way forming chains, which extend along the z axis of this tetragonal structure [144]. In the other directions the Ti06 octahedra share corners, thus forming an infinite threedimensional framework. Alternatively to the description by condensed Ti06 octahedra, the rutile structure can also be derived from a distorted hexagonal closepacked arrangement of oxygen atoms in which one half of the octahedral voids are filled with titanium atoms in an ordered manner. In contrast to rutile, anatase [145] and brookite [146] are derived from distorted cubic close-packed oxygen arrangements in which again half the octahedral voids are filled by titanium atoms. The voids are filled in such a way, that the octahedra

I . 7 Oxide Ceramics 21-

+y

33

Brookite

Figure 34. The crystal structures of orthorhombic brookite (space group Pbca), tetragonal anatase (14,lamd), and tetragonal rutile (P4*/mnrn).The edge- and corner-sharing Ti06 octahedra and the more or less distorted trigonal-planar environments of the oxygen atoms are emphasized.

share three common edges with other octahedra in brookite and even four common edges in anatase. According to the Pauling rules for ionic crystals, the stability of such an atomic arrangement decreases as the number of common edges increases, so the repulsion of the titanium atoms increases. The Ti-0 distances in anatase are 0.193nm (4x) and 0.198nm (2x), nearly identical with rutile (4 x 0.195nm, 2 x 0.198nm). In brookite on the other hand, they vary from 0.187nm to 0.204nm. As a consequence, one would expect the average Ti-0 distance of 0.196 nm in brookite to be significantly greater than the average Ti-0 distances in rutile (0.1959nm) and anatase (0.1949nm), which is not the case, probably because the structure refinement of brookite, carried out from film data, was not that accurate. In this context it is worth mentioning that the stishovite form of SiOz [147] with rutile type structure is a very hard material. Recent investigations on polycrystalline

34


cubic Zr02

tetragonal Zr02

X--lY

L--

- - - _ _ _ _ _ I

. I

monoclinic Zr02 Figure 35. The crystal structures of the cubic, tetragonal, and monoclinic forms of Zr02. The different zirconium-oxygen polyhedra are outlined. The cubic fluorite type (space group, Fm3m) high-temperature y-modification transforms on cooling by a displacive phase transition to tetragonal P-ZrOz (P4,/nmc), which on further cooling transforms, again by a displacive phase /c). transition, to monoclinic a-Zr02 (P2,

stishovite have shown that this material is harder than alumina and almost equals the hardness of c-BN [148].It is thus the hardest known oxide. The high-temperature (y-) forms of Zr02 and HfOz [149] as well as CeOz [150], Thoz, and UOz [I511 crystallize with the cubic fluorite type structure. The metal atoms form a close-packed f.c.c. arrangement and the oxygen atoms fill all tetrahedral voids. This way each metal atom obtains a cubic environment of eight oxygen atoms (Fig. 35). Thoz and UOz are used as nuclear fuels. Tho2 has the highest melting point of all oxides (>3200"C) and an extremely low oxygen vapor pressure. It is therefore used

1.7 Oxide Ceramics

35

as a corrosion-resistant ceramic whenever the less expensive oxides MgO, A1203,or Zr02 are not suitable, for example as crucibles for high-melting metals [18, 1381. Of the rare earth oxides, Ce02 is the most frequently used material. It is utilized for polishing and abrasive purposes and it is obtained in large quantities during the separation process of the early rare earth elements [152]. Zirconium dioxide has many applications mostly in its cubic, high-temperature y-modification, which for pure y-Zr02 is only stable above 2300°C. However, it can be stabilized by adding about 10% MgO, CaO, or Y 2 0 3 .In these yellow ceramics the metal atoms M2+ or M3+ substitute for Zr4+. As a consequence some oxygen positions are not occupied in the solid solutions Zrl - y M ~ + 0 2 - , 0 , and Zr2- 2 x M ~ ~ 0 4 p 1 0where x , 0 indicates the void positions. In this form stabilized zirconia is used in the glass and steel industry as container material. It also has applications as solid electrolyte, like in fuel cells and as an oxygen sensor. The electric charges are carried by the oxygen ions, which move from occupied to unoccupied sites. Below 2300°C cubic y-Zr02 transforms to the tetragonal p-form which is thermodynamically stable down to about 1 100°C. On further cooling, P-ZrOz transforms to the monoclinic a-form, which also occurs as a mineral, called baddeleyite. Both transformations are of the displacive type (as opposed to a reconstructive phase transition), where the atoms move only by a few picometers. As a consequence the y- and P-modifications cannot be quenched. Since the crystal systems change from cubic to tetragonal to monoclinic, both transformations are ferroelastic transformations, where the domain walls of the resulting twin domains are planar. Even though the transformations can be classified as displacive, the corresponding lattice constants change considerably from cubic a = 0.5065 nm [149] to tetragonal ah = 0.515nm, c = 0.527nm [153] to monoclinic a = 0.5145nm, b = 0.5208nm, c = 0.531 1 nm, P = 99.23' [154]. As a consequence, the transformations can be expected to have large hystereses, especially as compact polycrystalline ceramics. For unstabilized pure Zr02 the alp-transformation occurs at about 1200°C on heating and at about 1000°C on cooling [155]. In comparing the corresponding cell volumes for Z = 4 formula units for the (pseudo-) cubic cells of a-Zr02 (0.1404 nm3), P-Zr02 (0.1397 nm3), and y-Zr02 (0.1300 nm3) one notices the anomaly that the high-temperature cubic modification has a cell volume, which is about 7% smaller than those of the other two modifications. In part this may be due to an oxygen deficiency of the black high-temperature y-modification of the binary Zr02- ,r. There is also an increase in the volume for the P/a (tetragonal to monoclinic) transformation. For stabilized tetragonal zirconia this volume increase amounts to about 4%. In annealed ceramics of stabilized zirconia the tetragonal and monoclinic modifications coexist in the two-phase field with different compositions. Both of these may undergo phase transitions not only on cooling or heating, respectively, but also because of the change in volume under stress. When a crack arrives at an embedded grain of tetragonal zirconia the grain may transform to the more voluminous monoclinic modification, thus dissipating the stress to the whole surface of the embedded grain. This mechanism is called 'transformation toughening' [156]. Furthermore, since the y/P- as well as the B/a-transformations are ferroelastic, stress can be dissipated by movement of domain walls. From a crystallographic point of view this is analogous to the change

36


in form as it is known for the low-temperature modification of alloys with shapememory. In the crystal structure of tetragonal P-Zr02 the zirconium atoms have again (as in the cubic y-modification) eight oxygen neighbors [153]. This distorted cubic coordination can best be visualized as two interpenetrating distorted tetrahedra, one compressed with Zr-0 distances of 0.2065 nm, the other stretched with Zr-0 distances of 0.2455nm. These may be compared to the eight oxygen atoms of a zirconium atom in y-Zr02 at 0.2193 nm. In contrast to the y- and P-modifications, in the monoclinic baddeleyite type structure of a-Zr02 [154] the zirconium atoms have only seven oxygen neighbors at Zr-0 distances ranging from 0.205 to 0.228nm. Of the two kinds of oxygen atoms, one has three zirconium neighbors in approximately planar configuration (Zr-0 distances of 0.205, 0.206, and 0.216 nm), the other has a distorted tetrahedral environment (Fig. 35) with Zr-0 distances between 0.215 and 0.229 nm.

1.8 Amorphous Hard Materials A recent strategy is the synthesis of hard materials by pyrolysis (sometimes in the presence of N2, NH3, HzO, H2, and/or Ar) from molecular precursors. Such hard materials are frequently composed of inorganic networks which are amorphous to X-rays. Examples are ceramics with the approximate compositions BSi3N40, BSi3C4N,B3Si3N7,and BSiN3C [157-1591. These ceramics stay amorphous to Xrays up to temperatures between 1500 and 1900°C. At higher temperatures they decompose to composites containing BN, P-Sic, a-Si3N4, and P-Si3N4. For the characterization of the reaction intermediates and the resulting amorphous solids various thermal analysis and spectroscopic techniques are useful: differential thermal analysis (DTA), thermal gravimetric analysis (TGA), mass spectroscopy (MS), Fourier transform infrared (FT-IR) spectroscopy, l 1 B and 29Simagic angle spinning nuclear magnetic resonance (MAS-NMR), and X-ray photoelectron spectroscopy (XPS). These reveal that the atoms in the amorphous ceramics have the coordination numbers known from the corresponding binary phases: four for boron and silicon, three and four for carbon, three for nitrogen, and two for oxygen. In characterizing a solid as 'amorphous' to X-rays one should be aware that this does not rule out a high degree of order. The Scherrer formula D = k . X/Bcos 8 can be used to determine the particle size of a crystallite by X-ray diffraction. In this formula D is the diameter of a particle, k a constant close to k = 1, X the wavelength of the X-rays, 0 the Bragg angle of the diffracted radiation, and B the line broadening (in radians). The latter is defined as the additional line width (apart from the line width due to the instrumental set-up) at one half of the maximum intensity of the considered reflection [16&162]. As a hypothetical example we may consider a cube of 4 x 4 x 4 unit cells of diamond (8 atoms per cell, a = 0.357nm) at a diffraction angle of 8 = 21.9" (i.e. the 111 reflection with Cu ka: radiation). Using the Scherrer formula, this perfectly ordered arrangement of 8 x (4)3= 512 carbon

References

37

atoms results in a diffraction peak, which at half intensity has a width of B = 0.116 rad = 6.7". With this width the 111 peak (at 21.9") practically overlaps with the 200 peak (at 25.6" and a half-width of 6.9'). Thus a powder, made-up of such hypothetical crystallites with 5 12 perfectly ordered carbon atoms each, looks amorphous to X-rays!

References 1. P. P. Ewald and C. Herrmann, Strukturberichte Vol. 1-7, in Z . Kristallogr. Akademische

Verlagsgesellschaft mbH, Leipzig, 1913-1939. 2. A. J. C. Wilson, W. B. Pearson, J. Trotter, and G. Ferguson (Eds), Structure Reports Vol. 858, International Union for Crystallography, N. V. A. Oosthoek's Uitgevers MIJ, Utrecht, 1956-1993. 3. Landolt-Bornstein, Numerical Data and Functional Relationships in Science and Technology, K.-H. Hellwege (Ed.), Group III. Crystal and Solid State Physics, Vol. 6, Structure Data of Elements and Intermetallic Phases, P. Eckerlin, H. Kandler, and A. Stegher (Eds), Springer, Berlin, 1971. 4. W. B. Pearson, The Crystal Chemistry and Physics of Metals and Alloys, Wiley-Interscience, New York, 1972. 5. P. Villars and L. D. Calvert, Pearson's Handbook of Crystallographic Data for Intermetallic Phases, ASM International, Materials Park, OH, 1991. 6. P. Villars, Pearson's Handbook, Desk Edition, Crystallographic Data for Intermetallic Phases, Vol. 1 and 2, ASM International, Materials Park, OH, 1997. 7. E. ParthC, L. Gelato, B. Chabot, M. Penzo, K. Cenzual, and R. E. Gladyshevskii, TYPIXStandardized Data and Crystal Chemical Characterization of Inorganic Structure Types, in Gmelin Handbook of Inorganic and Organometallic Chemistry, 8th edn., Springer, 1993. 8. Fachinformationszentrum FIZ Karlsruhe, Inorganic Crystal Structure Data Base - ICSD, Karlsruhe, 1998. 9. International Centre for Diffraction Data - ICDD, Powder Diffraction File, Newtown Square, Pennsylvania, 1997. 10. K. Yvon, W. Jeitschko, and E. Parthe, J. Appl. Crystallogr. 1977, 10, 73. 11. J. Donohue, The Structures ofthe Elements, Wiley, New York, 1974. 12. F. P. Bundy and J. S. Kasper, J. Chem. Phys. 1967,46, 3437. 13. V. L. Solozhenko, V. V. Chernyshev, G. V. Fetisov, V. B. Rybakov, and I. A. Petrusha, J . Phys. Chem. Solids, 1990,8, 1011. 14. L. F. Vereshchagin, E. V. Zubova, L. N. Burenkova, and N. I. Revin, Dokl. Akad. Nauk SSR, 1968, 13, 25. 15. T. Soma, A. Sawaoka, and S. Saito, Muter. Res. Bull. 1974, 9, 755. 16. A. H. Comes de Mesquita, Acta Crystallogr. 1967, 23, 610. 17. Z. Li and R. C. Bradt, J . Muter. Sci. 1986, 21, 4366. 18. W. Buchner, R. Schliebs, G. Winter, and K. H. Buchel, Industrielle Anorganische Chemie, Verlag Chemie, Weinheim, 1984. 19. E. Parthe, Crystallochimie des Structures Tetraedriques, Gordon & Breach, Paris, 1972. 20. P. T. B. Shaffer, Acta Crystallogr. Sect. B, 1969, 25, 477. 21. M. Dubey, U. S. Ram, and G. Singh, Acta Crystallogr. Sect. B, 1973, 29, 1548. 22. H. Briehl, Chemie der Werkstofle, B. G. Teubner, Stuttgart, 1995. 23. H. L. Yakel, Acta Crystallogr. Sect. B, 1975, 31, 1797. 24. R. Telle, Chemie in unserer Zeit, 1988, 22, 93. 25. A. Kirfel, A. Gupta, and G. Will, Acta Crystallogr. Sect. B, 1979,35, 1052. 26. H. Neidhard, R. Mattes, and H. J. Becher, Acta Crystallogr. Sect. B, 1970, 26, 315. 27. A. J. Perrotta, W. D. Townes, and J. A. Potenza, Acta Crystallogr. Sect. B, 1969, 25, 1223. 28. A. F. Wells, Structural Inorganic Chemistry, 5th edn., Clarendon Press, Oxford, 1990.

38


29. H. Hillebrecht and F. D. Meyer, Z . Kristallogr. 1997, Suppl. 12, 11. 30. H. Bolmgren, T. Lundstrom, and S. Okada, AZP Conference Proceedings (American Institute of Physics) 1991, 231, 197. 31. H. Hubert, B. Devouard, L. A. J. Garvie, M. O’Keeffe, P. R. Buseck, W. T. Petuskey, and P. F. McMillan, Nature, 1998, 391, 376. 32. F. Benesovsky, Borcarbid und Boride, in Ullmanns Enzyklopudie der technischen Chemie, Band 8, VCH, Weinheim, 1974, pp. 657462. 33. M. Stiimke and G. Petzow, Z . Metallkd. 1975, 66, 292. 34. R. Fichte, M. Pistor, G . Knippschild, A. Meffert, and J. Rasch, Bor und Ferrobor, in Ullmanns Enzyklopadie der technischen Chemie, Band 8, VCH, Weinheim, 1974, pp. 644-648. 35. R. Kiessling, Acta Chem. Scand. 1949, 3, 595. 36. T. Bjurstrom, Arkiv. Kemi Mineral. Geol. 1933, 11A, I . 37. R. Kiessling, Acta Chem. Scand. 1947, 1, 893. 38. J. Etourneau, J. Less-Common Met. 1985, 110,267. 39. P Blum and F. Bertaut, Acta Crystallogr. 1954, 7,81. 40. M. M. Koruskova, V. N. Gurin, T. Lundstrom, and L.-E. Tergenius. J . Less-Common Met. 1986, 117, 73. 41. F. Benesovsky and W. Schedler, Hartmetalle, in Ullmanns Enzyklopadie der technischen Chemie, Band 12, VCH, Weinheim, 1974, pp. 5 15-52 1. 42. F. Benesovsky, Carbide, in Ullmanns Enzyklopadie der technischen Chemie, Band 9, VCH, Weinheim, 1974, pp. 122-136. 43. See A. W. Frad, Advances in inorganic Chemistry and Radiochemistry, H. J. Emeleus and A. G. Sharpe (Eds), Academic Press, New York, 1968, 11, 153. 44. E. Parthe and V. Sadagopan, Monatsh. Chem. 1962,93, 263. 45. E. Rudy, F. Benesovsky, and E. Rudy, Monatsh. Chem. 1962,93, 693. 46. W. Jeitschko, Structural Chemistry of Transition Metal-Metalloid Compounds, in M T P international Review of Science, inorg. Chem., Series 2, Vol. 5, Butterworths, London, 1974, pp. 219-281. 47. E. Parthe and K. Yvon, Acta Crystallogr. Sect. B, 1970, 26, 153. 48. Yu. B. Kuz’ma, V. I. Lakh, V. Ya. Markiv, B. I. Stadnyk. and E. I. Gladyshevkii, Poroshkovaya Metallurgiya, 1963, 2, 286. 49. K. Yvon and E. Parthe, Acta Crystallogr. Sect. B, 1970, 26, 149. 50. T. Epicier, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.), Kluwer, Dordrecht, 1990, pp. 215-248 and 297-327. 51. M. Hansen and K. Anderko, Constitution of Binary Alloys,Genium Publishing, Schenectady, NY, 1985. 52. J. Billingham, P. S. Bell, and M. H. Lewis, Acta Crystallogr. Sect. A , 1972, 28, 602. 53. C. H. De Novion, R. Lorenzelli, and P. Costa, C. R. h a d . Sci. Paris, 1966, 263, 775. 54. C. Froidevaux and D. Rassier, J . Phys. Chem. Solids, 1967, 28, 1197. 55. D. Kordes, Phys. Stat. Sol. 1968, 26, K103. 56. A. W. Henfrey and B. E. F. Fender, Acta Crystallogr. Sect. B, 1970, 26, 1882. 57. J. D. Venables, D. Kahn, and R. G. Lye, Phil. Mag. 1968, 18, 177. 58. J. Billingham, P. S. Bell, and M. H. Lewis, Phil. Mag. 1972, 25, 661. 59. M. P. Arbuzov, V. G. Fak, and B. V. Khaenko, Sov. Phys. Crystallogr. 1970, 15, 164. 60. M. Sauvage, E. Partht, and W. B. Yelon, Acta Crystallogr. Sect. A , 1974,30, 597. 61. J. B. Venables and M. H. Meyerhoff, NBS Special Publication 364, Proceedings of 5th Mater. Res. Symposium, 1972, p. 583. 62. H. Goretzki, Phys. Stat. Sol. 1967, 20, K141. 63. J.-L. Chermant, P. Delavignette, and A. Deschanvres, J . Less-Common. Met. 1970, 21, 89. 64. P. S. Bell and M. H. Lewis, Phil. Mag. 1971, 24, 1247. 65. B. V. Khaenko and V. V. Kukol’, Sov. Phys. Crystallogr. 1989, 34,905. 66. V. V. Kukol’, B. V. Khaenko, and 0. A. Gnitetskii, Crystallogr. Rep. 1995,40, 66. 67. Ch. H. de Novion and V. Maurice, J. Phys. (Paris), Colloq. 1977, 211. 68. Ch. H. de Novion, B. Benneu, T. Priem, N. Lorenzelli, and A. Finel, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.), Kluwer, Dordrecht, 1990. 69. B. V. Khaenko, 0. A. Gnitetskii, and M. V. Karpets, Crystallogr. Rep. 1999, 44,281.

References

39

70. G. Hagg, Z . Phys. Chem. B, 1931, 12, 33. 71. A. Neckel, K. Schwarz, R. Eibler, P. Weinberger, and P. Rastl, Ber. Bunsenges. Phys. Chem. 1975,79, 1053. 72. A. Neckel, P. Rastl, R. Eibler, P. Weinberger, and K. Schwarz, J. Phys. C: Solid State Phys. 1976, 9, 579. 73. P. Blaha, K. Schwarz, F. Kubel, and K. Yvon, J . Solid State Chem. 1987, 70, 199. 74. S. D. Wijeyesekera and R. Hoffmann, Organometallics, 1984, 3, 949. 75. A. L. Bowman, T. C. Wallace, J. L. Yarnell, and R. G. Wenzel, Acta Crystallogr. 1966,21,843. 76. C. S. Barrett and T. B. Massalski, Structure of Metals, McGraw-Hill, New York, 1966. 77. C. S . Roberts, J. Metals, 1953, 5 (Trans. AIME 197), 203. 78. D. Meinhardt and 0. Krisement, Archivfiir das Eisenhiittenwesen, 1962, 33, 493. 79. E. J. Fasiska and G. A. Jeffrey, Acta Crystallogr. 1965, 19, 463 80. S. Rundqvist and G. Runnsjo, Acta Chem. Scand. 1969, 23, 1191. 81. H. L. Yakel, Acta Crystallogr. Sect. B, 1987, 43, 230. 82. F. C. Frank and J. S. Kasper, Acta Crystallogr. 1958, 11, 184. 83. F. C. Frank and J. S. Kasper, Acta Crystallogr. 1959, 12, 483. 84. M. A. Rouault, P. Herpin, and M. R. Fruchart, Ann. Chim. Paris, 1970, 5, 461. 85. A. Westgren, Jernkontorets Annaler, 1933, 111, 525. 86. Z . Bojarski and J. Leciejewiez, Archiwunz Hutnictwa, 1967, 12, 255. 87. P. Ettmayer and R. Suchentrunk, Monatsh. Chem. 1970, 101, 1098. 88. B. Rupp and P. Fischer, J. Less-Common Met. 1988, 144, 275. 89. J. M. Newsam, A. J. Jacobson, L, E. McCandlish, and R. S . Polizzotti, J . Solid State Chem. 1988, 75, 296. 90. B. G. Hyde and S . Anderson, Inorganic Crystal Structures, Wiley, New York, 1989. 91. U. Westermann, E. Lugscheider, and J. Wonka, Metal1 1993, 47, 741. 92. W. Jeitschko, Acta Crystallogr. Sect. B, 1977, 33, 2347. 93. B. Boren, Ark. Kem. Miner. Geol. A , 1933, lla(10), 1. 94. H. Nowotny, R. Kieffer, and H. Schachner, Monatsh. Chem. 1952,83, 1243. 95. J. J. Nickl, K. K. Schweitzer, and P. Luxenberg, J . Less-Common Met. 1972, 26, 335. 96. W. J. J. Wakelkamp, F. J. J. van Loo, and R. Metselaar, J . Eur. Ceram. SOC.1991, 8, 135. 97. Yong Du and J. C. Schuster, Ber. Bunsenges. 1998, 102, 1185. 98. W. Jeitschko and H. Nowotny, Monatsh. Chem. 1967,98, 329. 99. T. El-Raghy and M. W. Barsoum, J . Appl. Phys. 1998,83, 112. 100. S. K. Lee, B. R. Lawn, and M. W. Barsoum, J . Am. Ceram. SOC.1998,81,225. 101. N. I. Medvedeva and A. L. Ivanovskii, Russ. J . Inorg. Chem. 1998, 43, 398. 102. R. Pottgen, D. Kaczorowski, and W. Jeitschko, J. Muter. Chem. 1993, 3, 253. 103. E. Partht., W. Jeitschko, and V. Sadagopan, Acta Crystallogr. 1965, 19, 1031. 104. B. Aronsson, Acta Chem. Scand. 1960, 14, 1414. 105. W. Rieger, H. Nowotny, and F. Benesovsky, Monatsh. Chem. 1965,96, 232. 106. M. Potzschke and K. Schubert, Z . Metallkd. 1962, 53, 474. 107. W. Schnick, Angew. Chem. 1993, 105, 846. 108. H. Lange, G. Wotting, and G. Winter, Angeih,. Chem. 1991, 103, 1606. 109. M. Billy, J. C . Labbe, A. Selvaraj, and G . Roult, Mater. Res. Bull. 1983, 18, 921. 110. R. Griin, Acta Crystallogr. Sect. B, 1979, 35, 800. 111. H. Schulz and H. H. Thiemann, Solid State Commun. 1977, 23, 815. 112. A. W. Weimer (Ed.), Carbide, Nitride and Boride Materials, Chapman & Hall, London, 1997. 113. C. Niu, Y. Z . Lu, and C. M Lieber, Science, 1993, 261, 334. 114. T. Hughbanks and Y. Tian, Solid State Commun. 1995,96, 321. 115. A. K. Sharma and J. Narayan, Int. Mater. Rev. 1997,42, 137. 116. J. V. Badding, Adv. Muter. 1997, 9, 877. 117. D. M. Teter, Mater. Res. SOC.Bull. 1998, 22. 118. F. Benesovsky, Nitride, in Ullmanns Enzyklopadie der technischen Chemie, Band 17, VCH, Weinheim, 1974, pp. 315-321. 119. W. Lengauer, J . Alloys Compd. 1992, 186,293. 120. A. N. Christensen, A. Alamo, and J. P. Landesman, Acta Crystallogr. Sect. C 1985, 41, 1009. 121. D. Rechenbach and H. Jacobs, J. Alloys Compd. 1996,235, 15.

40


122. H. Jacobs, D. Rechenbach, and U. Zachwieja. J. Alloys Compd. 1995, 227, 10. 123. B. Holmberg, Acta Chem. Scand. 1962, 16, 1255. 124. W. Lengauer and P. Ettmayer, J. Less-Common Met. 1986, 120, 153. 125. W. Lengauer, J. Less-Common Met. 1986, 125, 127. 126. J. C. Schuster and J. Bauer, J . Solid State Chem. 1984, 53, 260. 127. W. Jeitschko, H. Nowotny, and F. Benesovsky, Monatsh. Chem. 1963,94, 1198. 128. H. H. Stadelmaier in Developments in the Structural Chemistry of Alloy Phases, B. C. Giessen (Ed.), Plenum Press, New York, 1969. 129. L. E. Toth, Transition Metal Carbides and Nitrides, Academic Press, New York, 1971. 130. H. Nowotny, Angew. Chem. Intern. Edit. 1972, 11, 906. 131. Th. M. Gesing, K. H. Wachtmann, and W. Jeitschko, Z. Naturforsch. 1997, 52b, 176. 132. J. C. Schuster, Z . Kristallogr. 1986, 175, 211. 133. W. Jeitschko, H. Nowotny, and F. Benesovsky, Monatsh. Chem. 1964,95, 1212. 134. W. Schnick and H. Huppertz, Chem. Ew. J. 1997, 3, 679. 135. M. Woike and W. Jeitschko, Inorg. Chem. 1995, 34, 5105. 136. T. Schlieper and W. Schnick, Z . Anorg. Allg. Chem. 1995,621, 1037. 137. H. W. Hennicke, U. Hofmann, W. Lehnhauser, and K.-J. Leers, Keramik, in Ullmanns Enzyklopadie der technischen Chemie, Vol. 13, VCH, Weinheim, 1974, pp. 711-735. 138. K. Heldt and A. Reckziegel, Oxidkeramik, in Ullmanns Enzyklopadie der technischen Chemie, Vol. 17, VCH, Weinheim, 1974, pp. 515-529. 139. J. Lewis, D. Schwarzenbach, and H. D. Flack, Acta Crystallogr. Sect A , 1982, 38, 733. 140. T. M. Sabine and S. Hogg, Acta Crystallogr. Sect. B, 1969, 25, 2254. 141. D. K. Smith, C. F. Cline, and S . B. Austerman, Acta Crystallogr. 1965, 18, 393. 142. E. Schiebold, Z. Kristallogr. 1921, 56, 430. 143. N. G. Zorina and S . S . Kvitka, Kristallografiya, 1969, 13, 599. 144. R. Restori, D. Schwarzenbach, and J. R. Schneider, Acta Crystallogr. Sect. B, 1987,43,251. 145. M. Horn, C. F. Schwerdtfeger, and E. P. Meagher, Z . Kristallogr. 1972, 136, 273. 146. W. H. Baur, Acta Crystallogr. 1961, 14, 214. 147. M. Sugiyama, S. Endo, and K. Koto, Mineral. J. Jpn. 1987, 13, 455. 148. J. M. Leger, J. Haines, M. Schmidt, J. P. Petitet, A. S. Pereira, and J. A. H. de Jomada, Nature, 1996,383,401. 149. L. Passerini, Gazz. Chim. Ital. 1930, 60, 762. 150. G. Brauer and H. Gradinger, Z. Anorg. Allg. Chem. 1954, 277, 89. 151. K. Clausen, W. Hayes, J. E. Macdonald, P. Schnabel, M. T. Hutchings, and J. K. Kjems, High Temp. High Press. 1983, 15, 383. 152. I. McGill, J. Helgorsky, A. LevCque, Th. Petzel, and K. Reinhardt, Rare Earth Elements, in Ullmann’s Encyclopedia of Industrial Chemistry, 5th edn., Vol. A22, VCH, Weinheim, 1993, pp. 607-649. 153. G. Teufer, Acta Crystallogr. 1962, 15, 1187. 154. D. K. Smith and H. W. Newkirk, Acta Crystallogr. 1965, 18, 983. 155. G. Routschka and K.-E. Granitzki, in Ullmann’s Encyclopedia of Industrial Chemistry, 5th edn., Vol. A23, VCH, Weinheim, 1993, pp. 1 4 8 . 156. L. Michalowsky (Ed.), Neue keramische Werkstofle, Deutscher Verlag fur Grundstoffindustrie, Leipzig, 1994. 157. R. Riedel, Advanced ceramics from inorganic polymers, in Materials Science and Technology, R. W. Cahn, P. Haasen, and E. J. Kramer (Eds.), Vol. 17B, Chapter 11, VCH, Weinheim, 1996. 158. R. Riedel, J. Bill, and A. Kienzle, Appl. Organomet. Chem. 1996, 10, 241. 159. H.-P. Baldus and M. Jansen, Angew. Chem. 1997, 109, 338. 160. B. D. Cullity, Elements of X-Ray Diffraction, Addison-Wesley, Reading, Massachusetts, 1959. 161. L. V. Aziroff, Elements of X-Ray Crystallography, McGraw-Hill, New York, 1968. 162. B. E. Warren, X-Ray DifSraction, Addison-Wesley, Reading, Massachusetts, 1969.


2 Phase Transitions and Material Synthesis using the C02-Laser Heating Technique in a Diamond Cell A. Zerr, G. Serghiou, and R. Boehler

2.1 Introduction At room temperature, pressures in excess of 500GPa can be attained using the diamond anvil cell (DAC) technique [l]. This technique, described in a number of comprehensive reviews [24], has found broad application in the high pressure sciences, because diamond serves as an optical window from the far infrared to the near ultraviolet wavelength regime and is transparent to X-rays [4,5]. Development of the convenient ruby pressure scale, where the red shift of the R1fluorescence line with pressure is used in situ [6,7], led to a much wider use of this experimental technique. A few high-pressure devices are used to achieve high temperatures at high static pressures: Large-volume high-pressure devices (with characteristic sample dimensions of 2-3mm) can be used at temperatures up to 3000K, and pressures of about 25GPa (white field in Fig. 1). In an externally heated DAC temperature is limited to about 1800 K. Above this temperature the diamond anvil transforms to graphite, even in an inert atmosphere [2]. Moreover, these devices are limited to pressures of about 30GPa due to weakening of the diamond anvils and of other parts of the DAC upon heating. Additionally, pressure measurements are complicated by the absence of accurate high-pressure-high-temperature equations of state for any solid that could be used as a pressure gauge. Another possibility involving use of an internally heated DAC, where the conducting sample is heated by passing of an electric current through it [8,9], is constrained by its range of applicability. The laser-heated DAC allows us to access much higher static pressures and temperatures than the techniques listed above (gray field in Fig. 1). The technique of laser heating in a DAC is based on three main features: optical transparency of diamond anviIs; the samples can be heated via the optical absorption of intense laser radiation, and the temperature can be determined from the thermal radiation spectrum of the heated sample using the Planck formula [lo]. Laser radiation for heating of a sample in a DAC was first implemented by Ming and Bassett [ll], who used a pulsed ruby laser, and a continuous-wave Nd-YAG (yttrium-aluminum-garnet) laser to heat samples in a DAC above 3300 K, and up to 2300K, respectively. Today two types of continuous wave infrared (IR) lasers are extensively used in laser heating experiments: Solid state lasers (Nd-doped YAG, or YLF (yttrium-lithium-fluorite) crystals with the most intense line at

42

2 Phase Transitions and Material Synthesis using the COI-Laser Heating Technique

s!?

SUN

3

b

20

60

Pressure (GPa)

Figure 1. Pressure-temperature regime accessed to date using the C02-laser heating technique is shown by the light gray field. The P-T values available for multianvil high pressure devices are given by the white field. The temperature on the surface of the sun is shown for comparison.

1.06 pm), and C 0 2 gas lasers (most intense line at 10.6 pm). The Nd-YAG lasers have been used for heating semiconductors [12], metals, and insulators containing transition metals such as Fe, Ni, etc. [l 11. The absorption mechanism involves coupling of the laser radiation with electronic excitations within the sample. The C02laser heating technique, first employed by Boehler and Chopelas [ 131, has been used in experiments on numerous nonconducting inorganic (oxides, silicates, nitrides, etc.) and organic materials (e.g. hydrocarbons C,H2, + 2 ) . Heating of these materials with C02-laser radiation is possible due to the resonant absorption of the laser radiation by lattice vibrations. The C02-laser heating technique is particularly suitable for high-pressure-high-temperature experiments on hard materials (i.e. cubic BN, Sic, Si3N4),since most of them are insulators. In section two of this paper we describe the technical aspects of C02-laserheating in a DAC. The third section focuses on the methods for measuring melting temperatures at variable pressures, the fourth section on the determination of high pressure and temperature phase diagrams, and in the fifth section some experiments focusing on the synthesis of diamond and cubic BN from organic precursors will be described.

2.2 Technique of C02-Laser Heating in a Diamond Anvil Cell 2.2.1 Sample Assemblage in a Diamond Anvil Cell Figure 2 shows schematically the sample assemblage in a DAC. The sample container, which is made by drilling a hole in a pre-indented metal gasket, has

2.2 Technique of C02-Laser Heating in a Diamond Anvil Cell

43

typical dimensions of about 150 pm in diameter and 50 pm in height. The sample, which can be a single crystal, glass, or pressed powder pellet with typical dimensions of about 70 x 70 x 15 pm, is separated from the highly thermally conducting diamond anvils using small discs of the sample material or of an alkali halide (for example NaC1, KBr). The sample is embedded in a pressure transmitting medium, which provides thermal insulation of the sample from the diamond, and quasihydrostatic pressure conditions in the sample volume (see below). Depending on the chemical composition of the sample, different pressure media can be used: Noble gases provide chemically inert conditions. Alkali halides, which are less hydrostatic and less chemically inert, are better thermal isolators and do not require use of gas loading devices. Nitrogen is used in specific experiments to provide chemical stability of nitrides (for example BN [14]) at high temperatures or as a reactant source for synthesis of nitrogen containing materials at high pressures and temperatures [ 12,151.

2.2.2 Pressure Conditions in the Sample Volume For pressure measurement we used the ruby fluorescence scale, calibrated under quasi-hydrostatic load conditions to 80 GPa with the known equations of state of two metals [7]. The pressure was measured at room temperature using micronsized ruby crystals, distributed on the edge of the sample volume (Fig. 2) to avoid chemical reaction with the sample material during heating. The ruby fluorescence was excited by Ar-laser radiation (Fig. 2). At room temperature the pressure medium is solid, resulting in pressure gradients in the pressure chamber by uniaxial compression. Use of soft pressure media (e.g. argon) reduces these pressure gradients to about 10-15% of the average value. It was found in C02-laser heating experiments to 20 and 43 GPa [16,17] that

COz-laser

Ar-laser

Figure 2. Schematic drawing of a sample assemblage in the diamond cell. The sample is heated with a C 0 2 laser. Nitrogen is used as a pressure medium in experiments on nitrides to preclude decomposition. Argon is used as a pressure medium when chemically inert conditions are required. Fluorescence of ruby crystals is excited using an Ar laser.

44

2 Phase Transitions and Material Synthesis using the C02-LaserHeating Technique

after heating of the sample the pressure gradients are reduced to 3% of the average value. During laser heating the pressure in the sample chamber increases owing to thermal expansion of the sample and of the surrounding pressure medium. To reduce this effect in our C02-laser heating experiments, the sample to pressure medium volume ratio was kept below 1 :10. We found that upon heating silicates up to 2600K at around 20 GPa the thermal pressure does not exceed 0.3 GPa, which is less than 2% of the average pressure value [13].

2.2.3 Experimental Set-up for C02-Laser Heating in a Diamond Anvil Cell An experimental set-up for C02-laser heating in a DAC is shown schematically in Fig. 3 . The incident angle of radiation of the C 0 2 laser (Melles Griot, 150W, >90% of the intensity is in the single transverse electromagnetic mode TEM,,,) is about 25" (see also Fig. 2). The laser beam is focused by a ZnSe-lens onto the sample in a DAC. By changing the distance between the ZnSe-lens and the DAC one can vary the size of the laser-heated spot. The heated area, which exhibits a bell-shaped temperature distribution (see below), is imaged by a reflecting objective on the plane of the entrance pin-hole of the monochromator. The magnification of about 20, and a pinhole diameter of about 50 pm, result in a spatial resolution for the sample temperature measurement of about 2.5pm in diameter. The thermal

l4

I

I

L

13 Figure 3. Schematic drawing of the experimental set-up for the C02-laser heating experiments in a DAC: 1 = C 0 2 laser, 2 = Si mirror, 3 = ZnSe lens, 4 =diamond cell, 5 =reflecting objective, 6 = monochromator, 7 = CCD detector, 8 = PC, 9 and 12 = beam splitter, 10 = optical diode, 11 =regulating unit, 13 = ocular, 14 =power supply of the C 0 2 laser. The dashed and dotted line shows the path of the C02-laser beam, the dashed lines show the visible light path, and the solid ones depict electrical connections.

2.2 Technique of C02-Laser Heating in a Diamond Anvil Cell

45

radiation spectrum from this area is measured with a charge coupled device (CCD) detector. To determine the temperature, the Planck formula is fitted to the measured spectrum (see below). Approximately 50% of the incandescent radiation from the sample is directed to an optical diode by a beam-splitter. The diode signal, after amplification and conversion in the regulating unit, is used to stabilize the laser power (see below). The second beam-splitter directs the sample radiation to the ocular for visual observation.

2.2.4 Temperature Determination The heated sample emits thermal radiation, which is used for temperature determination. The spectrum collected was measured in the wavelength range 5 15-820 nm corresponding to the range of maximal quantum efficiency of our CCD detector. To determine the temperature we fitted the Planck formula with a wavelength independent emissivity to the measured spectrum. The Planck formula [ 101 contains the temperature and the wavelength dependence of the thermal radiation intensity I B B ( T , A) of the black body (BB):

where h is the Planck constant, k the Boltzmann constant, c the velocity of light in vacuum, T the temperature of the BB, and X the wavelength. The temperature and wavelength dependent emissivity E ( T ,A) is introduced for description of a thermal radiation spectrum of a real body ZRB(T,X). Emissivity is simply the ratio of the intensities of the real body to the BB at the same X and T [5]: In our experiments we assumed that the materials investigated are graybodies; that is that their emissivities, which are smaller than unity, are wavelength independent E ( T ,A) = e( T ) .The wavelength dependence of emissivity is only known for a few materials in vacuum or at atmospheric pressure [18,19]. If we assume that E of the investigated materials has the same wavelength dependence as tungsten [ 191 then the temperature correction would be about -100K at 3000K and about -350K at 5000K. Figures 4a and 4b show the thermal radiation spectra (thick solid lines) of S i c heated to about 6700K at 12GPa and of MgSi03 heated to 3000 K at 100 GPa, respectively. The temperature values correspond to the fits (thin solid line) of the Planck formula with a wavelength-independent emissivity to the measured spectra.

2.2.5 Temperature Stabilization In C02-laser heating experiments in a DAC it is important to eliminate sample temperature fluctuations due to fluctuations of the laser intensity. The peak-topeak amplitude of the COz-laser power fluctuations in our experiments exceeded

46

2 Phase Transitions and Material Synthesis using the COr-Laser Heating Technique

SIC P=12 GPa T=6700 K

.E fn C

9 c

-

I

0

550

a

.

1

’

600

I

650

’

I

700

’

I

.

I

750

800

750

800

wavelength (nm)

E Y, C

d c

-

500

b

550

600

650

700

wavelength (nm)

Figure 4. (a) Thermal emission spectrum of S i c heated to 6700 f200 K at 12 GPa (thick line). The thin line shows the fit of the Planck formula to the spectrum with a wavelength independent emissivity. (b) Thermal emission spectrum of MgSiO, heated at 3000 K and 100GPa (thick line). The thin line is the fit of the Planck formula.

5% of a chosen average value, when feedback stabilization was not employed. This resulted in temperature fluctuations with a peak-to-peak amplitude exceeding 30% of the average sample temperature. Figure 5a shows temperature fluctuations in a sample heated for 60s with an unstabilized C 0 2 laser. This nonlinear relation between laser power fluctuations and temperature fluctuations is most probably due to the temperature dependent absorption of the sample material: For example, experimental measurements of the absorption coefficient k(X) of MgO between 1 and 9 pm to about 1300K at atmospheric pressure showed that k(X) increases nonlinearly with temperature as well as with wavelength [20].

2.2 Technique ofC02-Las e r Heating in a Diamond Anvil Cell

3000 -. 34.5 GPa

T=3000 K ~

2800 -

-

2400-

-..

L

2200 -

-9

,

..'

1

~

.

s

m

~

,,--

:

I

% .;-

T=2140 K

1 -

2ooo 1800). , 0 10

-%\

,L

*.

.

...

..-. .

2600 - c

,

47

. , 20

.

I

30

.

40

.

,

50

I

.

.

60

70

80

, . 90

I

100

time (x0.6 s)

3010+50 K

3000

(Mg,Fe)SiO, 29 GPa

28001

2600

2000

0

5

10

15

20

25

30

35

40

45

50

time (min) Figure 5. (a) Temperature fluctuations in the (Mg,Fe)Si03 sample heated in a DAC with an unstabilized COz laser: 100 temperature measurements within a time interval of about 60 s from the center of the heated area (squares). During this time interval the temperature fluctuated between 2140 and 3000 K. (b) Temperature fluctuations in the (Mg, Fe)SiO, sample heated in a DAC with laser stabilization to average temperatures of 2090 K (diamonds), 2420 K (circles) and 3010 K (squares) were monitored during 35, 20, and 17 min, respectively. The laser stabilization suppressed the peak-to-peak fluctuations to below 120 K, or to below 5% of the average temperature. After [21].

To suppress sample temperature fluctuations a feedback stabilization circuit is employed (see above): Figure 5b shows the sample temperature fluctuations measured on a (Mg, Fe)Si03 sample heated at 29 GPa for 17-35 min when the feedback stabilization system was activated. In these long-term heating runs the peak-to-peak amplitude of the sample temperature fluctuations were kept below 5% of the average temperature.

48

2 Phase Transitions and Material Synthesis using the COz-Laser Heating Technique I

2800 2600

1

.

I

.

I

.

I

,

I

.

I

(Mg ,Fe)Si 0,

,

l

P=29 GPa

h

2400-

!? 3

2

.l-i

2200-

8

E

I-

0 0

2ooo: 1800 -

1600!, -30

.

I

-20

.

,

-10

.

0

10

.

, 20

.

I

!

30

rad. distance (pm) Figure 6. Radial temperature distribution in the (Mg, Fe)Si03 sample heated with a defocused COl laser beam at 29 GPa. Temperatures were measured (symbols) in 4 pm steps with a spatial resolution of 2.5 pm. In the central region of the heated area (with about 30 Fm in diameter) the temperature gradients are below 10Kpm-'. After [21].

2.2.6 Radial Temperature Gradients As mentioned above, the laser beam can be defocused in order to increase the heated area (Figs 2 and 3). Owing to the high power of the C02-laser, and high absorption of its radiation by the sample, the temperature gradients in the central portion of the heated area can be reduced to below lOKpm-': Fig. 6 shows the radial temperature gradients measured in an (Mg, Fe)Si03 sample heated at 29 GPa to peak temperatures of about 2000 K and 2700 K, respectively.

2.2.7 Raman and Fluorescence Spectroscopic Analysis of Samples in a Diamond Anvil Cell Samples were characterized using Raman and fluorescence spectroscopy at room temperature and high pressure up to 100 GPa. These techniques allow us to characterize rapidly micron-sized samples which can be irregularly shaped and oriented. In particular, with the use of available databases one can obtain detailed structural information. Spectroscopic signals are sensitive to structural changes at the 0.52.0nm level [22] and are less dependent on atomic number than X-ray signals. Therefore, these techniques are very effective in monitoring structural changes in crystals with small X-ray scattering cross-sections such as MgSi03 or Si02 and in disordered solids [23,24]. Additionally, by using the shift of the vibrational modes of the solids upon compression one obtains the pressure dependence of physical properties such as the thermal expansion and the sound velocity [25,26]. In our

2.3 Determination of’ Melting Temperatures at High Pressures

49

experiments the samples were excited with an argon ion laser (A = 457.9 nm) with powers varying from a few milliwatts to about 100mW. The spatial resolution of the measurements was typically 10-15pm. Spectra (see below) were taken from the center of the laser-heated spot at room temperature using CCD, or photon counting detection systems.

2.3 Determination of Melting Temperatures at High Pressures Measurement of the melting temperatures at static pressures above 25 GPa or at temperatures above 3000 K (at high pressures) first became possible after development of the laser heating technique in a DAC. In our C02-laserheating experiments we have used two melting criteria: The first involves a strong increase in the absorption of the laser radiation in the melt, resulting in a strong temperature increase. In these experiments the melting temperatures were taken as the last temperatures of the solid before the temperature increase due to melting. In the molten area of the sample convective motion was observed, accompanied by strong temperature fluctuations of about f 6 0 0 K. Characteristic textural features due to melting were found on the recovered samples (see below). An increase in the absorption of the laser radiation in the melt was usually observed for materials which melted congruently. Changes in the absorption of laser radiation were also observed in solid-solid phase transitions, for example when Si02 quartz transforms to coesite at high pressure. In these transitions changes in the absorption are usually smaller and the temperature fluctuations remain within the limits provided by the stabilization system (below f 6 0 K). A different criterion was employed when no changes in absorption of the laser radiation in the melt were detected. This method takes advantage of the observed change in the surface topography on the quenched sample, which is due to migration of the melted sample material towards the laser beam (see below). Such migration is possible when the melting temperature of the pressure medium is below the melting temperature (or solidus) of the sample material. This effect was observed both in congruently melting solids, where strong increases in the absorption of the laser radiation in the melt are observed, and in partial melting of multicomponent materials, where changes in the absorption of the laser radiation in the melt are undetectable .

2.3.1 Melting of Cubic BN at 10GPa An increase in the absorption of the laser radiation in the melt, as well as textural changes on the sample surface, were clearly observed in melting experiments on cubic BN at 10GPa described below. Based on previous estimates of the triple point between hexagonal BN, cubic BN, and liquid [27], melting of cubic BN at

50

2 Phase Transitions and Material Synthesis using the C02-Laser Heating Technique

Laser power (arb. units) Figure 7. Sample temperature of cubic boron nitride versus C02 laser power at 10GPa: 100 temperature measurements were taken within a time interval of about 5 s while the laser power was steadily increased. The two discontinuous temperature increases at 3600 K are due to melting.

10GPa requires temperatures in excess of 3500K. In our experiment we used nitrogen as a pressure medium to preclude decomposition of boron nitride upon heating [14]. Long-term heating of the sample at the temperatures required for melting of cubic BN (35004000K) led to diamond anvil failure. Therefore we avoided long heating durations: The laser power was increased rapidly while temperatures were measured continuously every 0.05 s. Sample temperatures versus laser power (solid line) are shown in Fig. 7. This plot shows a typical power-temperature dependence with two temperature jumps at about 3600 K. Sudden temperature increases are due to an increase in the absorption of the laser radiation in the melt. The subsequent temperature drops are most probably due to changes in the surface topography which were observed on the quenched sample (Fig. 8): The melted portion of the sample migrated toward the laser beam in the melted nitrogen pressure medium [28], which decreased the thickness of the insulating pressure medium between diamond and sample, leading to higher heat conduction to the diamond anvil. As a result the sample temperature decreased below the melting temperature and the second melting event could be observed after further increase of the laser beam intensity. The melting temperature was 3600 f 100 K, taken as the last temperature of the solid before the temperature jumps. Raman spectroscopic analysis of the melted sample area showed the lines of

2.3 Determination of Melting Temperatures at High Pressures

co2-laser

51

\

Figure 8. Schematic of the cubic BN sample melted in a DAC at 10 GPa. Temperatures recorded for this sample are shown in Fig. 7. The outgrowth in the center of the heated area is due to melting.

cubic BN only, no traces of the low pressure hexagonal BN phase were observed. These results are consistent with the previous estimation of Corrigan and Bundy ~71.

2.3.2 Melting Temperatures of Materials Relevant to the Earth's Lower Mantle (Mg, Fe)O, (Mg, Fe)Si03-, and CaSi03-perovskite are considered to make up about 97 weight-% of the Earth's lower mantle, which extends from about 670 to about 2900km depth and makes up about 50% of the Earth's mass. Their melting curves are important for modeling of chemical differentiation, rheology of the lower mantle [29], and for interpretation of seismic wave attenuation [30]. In our melting experiments on these compounds the experimental set-up and the sample assemblage was the same as described above with the exception that argon was used as a pressure medium. Using C02-laser heating in a DAC we measured, for the first time, the melting temperatures of MgO to about 32 GPa [31]. These measurements were not possible previously because of the high initial melting temperature of 3063 K at 1 atm [32]. A Nd-YAG laser is unsuitable for heating of MgO, because of the low absorption of that radiation (A = 1.06pm) by MgO [5,20]. The results for MgO are shown in Fig. 9. The initial slope of the MgO melting curve of about 36KGPa-' is significantly smaller than theoretical estimates, where values between 100 K GPa-' and 300 K GPa-' were reported [37-391. Using the theoretical [40,41] and empirical 1421 relations between melting temperature and pressure we extrapolated the high pressure data back to 1 atm and obtained 3050K, which is very close to our measurement at 1 bar of 3040 f l00K and to the data in the literature [32]. A fit

52

2 Phase Transitions and Material Synthesis using the C02-Laser Heating Technique I

'

I

'

I

'

I

5000 4500 -

~

I

.

I

.

I

,

CaSi0,-pv

A,..

1500-.'

.

(Mg,Fe)SiO,-pv

,*-

1

multi-component

-

multi-anvil -

I

-

I

~

I

-

I

-

of the theoretical melting relation [40,41] to the data revealed that the Griineisen parameter of MgO at 1 atm is 1.3, in good agreement with the literature value of 1.4 [43]. Melting temperatures of orthorhombic (Mg, Fe)Si03-perovskite and of cubic CaSi03-perovskite were measured using the COJaser heating technique to 62 GPa [34] and 43 GPa [33], respectively (Fig. 9). Melting temperatures of both perovskites (which are considered to be the most abundant lower mantle phases) have similar strong pressure dependencies: The initial melting slope for both systems is about 60 K GPa-', which is almost twice that of MgO. At the highest pressures of these experiments the obtained melting temperatures exceeded 5000 K for (Mg, Fe)Si03-perovskite and 4000 K for CaSi03-perovskite. Precise measurement of such high melting temperatures is only possible if the radial and axial temperature gradients are small, and if chemically inert pressure media (for example noble gases) are used. These requirements were not fulfilled in earlier DAC experiments on (Mg, Fe)Si03-perovskite [44-46]. In those experiments samples were heated with YAG-laser radiation, which is only weakly absorbed by the sample, and no thermally insulating pressure medium was used. In previous melting experiments on CaSi03-perovskite [47], rhenium plates in contact with the sample were used to absorb the Nd-YAG laser radiation, thus indirectly heating the sample. We have

~

~

2.3 Determination of Melting Temperatures at High Pressures

53

Figure 10. (a) Back-scattering electron topographic image of the glass sample with the mantle relevant composition heated just above the solidus to 2550 K at 20.6 GPa. Within a circular patch of about 60 pm in diameter the starting material converted to a crystalline high pressure assemblage. Cracks within this area became evident after pressure release. Another type of texture of about lOpm in diameter in the center (arrow) is due to migration of molten sample material towards the laser beam, as evident from quantitative measurements with an atomic force microscope in the area indicated by dashed lines. (b) Three-dimensional surface topography of the 34 x 34pm area, shown in Fig. 10(a) by dashed lines, measured with an atomic force microscope. Partial melting of the sample within an area of about 10 pm in diameter caused height variations of up to 0.77 pm. In contrast, height variations in the surrounding area are smaller than 0.2 pm. Reprinted with permission from [35]. Copyright [1998] American Association for the Advancement of Science.

54

2 Phase Trunsitions and Material Synthesis using the C02-Laser Heating Technique

found, however, that at high temperatures CaSi03 reacts with rhenium [33], which probably depresses the melting temperatures by up to 650 K [47]. The solidus temperature of a multicomponent system (for example, the Earth‘s mantle material) cannot be predicted even if the melting temperatures of all endmember components of the system are known. Previous solidus temperatures of mantle relevant materials were measured to about 25 GPa using multianvil high pressure devices [36,48,49]. The solidus of the (Mg, Fe)O-(Mg, Fe)Si03 assemblage, which can be considered as a simplified model for the Earth’s mantle material, was measured to about 33 GPa in a DAC using a Nd-YLF laser [33]. Additionally, melting of (Mg, FehSi04, which breaks down to the same phase assemblage, was estimated in shock-wave experiments to be 4300 K at about 130 GPa [50]. For the measurement of the solidus temperature of a multicomponent material with a realistic Earth’s mantle composition we used [35] a different melting criterion involving changes of the surface texture due to melting. Textural changes on the sample surface due to melting were verified using a scanning electron microscope (SEM) and an atomic force microscope (AFM). Figure 10a shows a back-scattered electron topographic image of the sample which was partially molten at 2550K and 20.6 GPa. Characteristic melting features (about 10 pm in diameter) can be seen in the center of the heated area. In the molten area the height variations are almost four times larger than in the surrounding subsolidus material, as evident from quantitative topographic measurements with an AFM (Fig. lob). At pressures between 20 and 30 GPa our data (Fig. 9) agree well with previous multianvil data and with the solidus of the (Mg, Fe)-(Mg, Fe)Si03 assemblage measured using an YLFlaser. Our results indicated a strong depression in the melting temperatures of the multicomponent system as compared with the end-member phases: At 60 GPa the solidus is about 900 K below the melting temperature of MgO. Our data [35] extrapolate to the melting point estimation from the shock wave experiment at 130GPa [50]. This supports the assumption that the solidus of the Earth’s mantle material is dominated by the melting behavior of the (Mg, Fe)@(Mg, Fe)Si03 assemblage.

2.4 Phase Diagrams, Decomposition Reactions, and Stability of Solids at High Pressures and Temperatures The present C02-laserheating technique in a diamond anvil cell allows us to access a large P-T range (see Fig. 1) for examining structural properties, establishing the stability fields of solids, and for synthesizing new materials. Additionally, the present method of laser heating in a diamond cell circumvents several difficulties associated with prior large volume presses at high temperatures: Hydrostatic pressures eliminate effects of shear on the transition pressures; no thermocouple corrections as a function of pressure are required since the temperatures are measured using the Planck radiation function; the temperatures are measured

2.4 Phase Diagrams, Decomposition Reactions, and Stability of Solids

55

from the same parts of the samples that serve for phase identification, thus, avoiding problems associated with temperature gradient; the samples are characterized in situ, avoiding reverse transitions upon pressure release; higher temperatures decrease possible kinetic problems on phase transitions. We illustrate the use of the laser heated diamond anvil cell for synthesizing high pressure phases and establishing phase boundaries of solids with the following three studies.

2.4.1 Coesite-Stishovite Phase Boundary We examine first the high pressure and temperature phase boundary between the tetrahedrally coordinated (coesite) and the octahedrally coordinated (stishovite) SiOz polymorphs. The phase diagram of Si02 is the focus of intense interest, because: it has been shown recently that stishovite is the hardest known quenchable oxide and is a member of an emerging new family of superhard materials based on the A 0 2 formula (A = Si, Os, Ru, Mn, Sn, Ge, Pb) [51,52]; the coesite-stishovite transition is used for pressure calibration in multianvil devices. However, there I

'

I

'

I

'

I

.

1

.

1

wesite (tetrahedral coordination) after heating at 2600 K

10.0 GPa

stishovite (octahedral coordination) after heating at 2600 K

10.9 GPa


12.0 GPa


13.0 GPa

30

500

600

700

800

90

rel. wavenumbers (cm -I) Figure 11. Raman spectra of laser-heated SOz taken at room temperature which show that stishovite converts to coesite at 10.0GPa and 2600K. After [17].

56


"0°1

Z (1993)

:

Se(1995)

.

-I

I-

2000 1500

:

. , . .,

1

:

,'Z(1993) ,>-

3

3-..z'Z(1994)

R

1000-

-

.

8

C = Coesite

1'0 15 Pressure GPa

1'4

Figure 12. Measurements of the coesite-stishovite transformation [ 171: Se. The phase boundary (dotted line) is an extension of Yagi and Akimoto's results [54]: Y through our data points. Previous large volume press data from [56]: A, [57]: S, [55,58]: Z are also presented in the figure (see text). After [ 171.

are strong discrepancies (of order 2 GPa) between existing estimates of this boundary [17,53-551. The experiments were carried out in a C02-laserheated diamond cell using argon as pressure medium. The starting materials were thin discs of Si02 glass. Samples were typically heated for about 1-2min close to their melting temperatures to avoid kinetic effects on the transition, and then temperature quenched by switching off the CO, laser. Phase identification of the samples was carried out using Raman spectroscopy. The phase boundary was determined with three different experiments [17]. In the first experiment Si02 glass was converted to stishovite and the phase boundary was approached in 1 GPa increments until coesite was formed (Fig. 11). Analogously, in the second experiment, coesite was heated at increasing pressures until stishovite was formed. In the third experiment coesite or stishovite was synthesized directly from the starting glass material in the vicinity of the phase boundary. These results are summarized in Fig. 12 together with previous work on the phase boundary at lower temperatures. We obtained a phase boundary that is represented by P(GPA) = 7.8 + 0.0010 T(K). Our estimated slope is less than one half that obtained by Zhang [ 5 5 , 5 8 ] using multianvil high pressure devices.

2.4.2 High Pressure and Temperature Phase Diagram and Decomposition Reactions in a Ternary System In our second study we examined the high P-T behavior of Mg3Al2Si3OI2garnet, a major constituent of the Earth's upper mantle, and transition zone. Garnets are the

2.4 Phase Diugrams, Decomposition Reactions, and Stability of Solids

57

primary host minerals for A1203at these depths [59], containing structural elements (polyhedra) of low density solids (tetrahedra), as well as structural elements of high density phases, such as end-member MgSi03 ilmenite (octahedra), or perovskite (octahedra and dodecahedra) [60]. It is known that end-member MgSi03 ilmenite is a low-temperature, high-pressure phase, and transforms to the garnet structure above 2100 K [61]. Both phases transform to MgSi03-perovskite above about 23GPa [61]. In order to investigate if the addition of aluminum changes the transition pressure to the perovskite structure, and whether magnesian perovskite can accommodate the available A1203 in the perovskite structure, we performed C02-laser heating experiments on Mg3A12Si3012garnet to 45 GPa, and temperatures close to the melting point. This doubles the experimental P-T range of previous investigations [62,63]. To identify the phases we used Raman as well as fluorescence spectroscopy. The latter allowed us to identify small amounts of A1203, which is hard to detect using Raman spectroscopy. To do this, we used Mg3Al2Si3OI2 : Cr3+ garnet as a starting material because a free A1203: Cr3+ phase has a strong fluorescence signal. Moreover, the spectrum of A1203:Cr3+ can be distinguished easily from that of Mg3Al2Si3OI2 :Cr3+. Typical Raman spectra of pyrope, ilmenite, and perovskite are shown in Fig. 13. Figure 14a shows fluorescence spectra of Mg3A12Si3012 : Cr3+ after heating in the perovskite stability field.

f i I after heating at 2600 K

30 GPa

after heating at 3000 K

Ilrn

h

c v)

26.4 GPa

C

3

.E

I

Ilrn

after healing at 2730 K

-

Ilrn

v)

C 0, C

a

22.1 GPa

IT

after heating at 2500 K P P

&LA

P

PP

I .

1

400

.

21.2 GPa 1

.

I

I

.

1

600 800 I000 re\.wavenumbers (cm-')

B

7

'

1; I0

Figure 13. Room temperature Raman spectra after heating pyrope glass between 21.2 and 30 GPa showing pyrope (P), ilmenite (Ilrn), and perovskite (Pv) lines. The peak labeled with an asterisk is an artifact. After [ 161.

-


I

-

I

-

1

-

45.3GPa

R,/R,ruby

43.3 GPa

0 GPa

710

I

I

I

706 708

R,fluorescence from ruby chip: R,fluorescence from sample -

0

b

I

I

1

I

10

20

30

40

50

Pressure (GPa)

Figure 14. (a) Room temperature fluorescence spectra after heating Mg3Al2Si3OI2 : Cr3+ pyrope crystals at about 2500 K at various pressures. R1 and R2 are the primary electronic peaks due to Cr3+-A1203, the peaks labeled with N are N-lines (Nebenlinien) and those labeled with S are vibrational sidebands of Al2O3 [64]; the feature labeled with an asterisk is residual fluorescence from unconverted pyrope. (b) Comparison of the pressure dependencies of the R I fluorescence lines of a ruby chip (located at the edge of the gasket) and A1203:Cr3+ produced after heating Mg3A12Si3012 :Cr". Agreement of the R I peak positions of the ruby chips and the Cr3+-A1203 in the center of the sample shows that no pressure gradients are present in the sample chamber after heating. After [16].

2.4 Phase Diagrams, Decomposition Reactions. and Stability of Solids

59

P(GPa) Figure 15. Mg3A12Si3012-pyrope-ilmenitcperovskitephase relations. The diagram also includes data points from earlier multianvil work by [63]: K and [62]: I. After [16].

Pyrope transforms to the ilmenite structure above 21.5 GPa, to perovskite plus ilmenite above 24GPa, and to perovskite above 29GPa (Fig. 15). The stability field of the ilmenite structure is expanded to higher temperatures with respect to end-member MgSi03-ilmenite. The transition pressures to aluminous perovskite is about 2GPa higher with respect to pure MgSi03. Furthermore, we found that some A1203 exists as a free phase together with perovskite up to 43 GPa, because we detected A1203:Cr3+fluorescence up to this pressure (Fig. 14). Our results show that the perovskite structure can accommodate at least 25mole% A1203 above 43 GPa.

2.4.3 Stability of a Perovskite Oxide with Respect to its Component Oxides In the third study we investigated whether MgSi03-perovskite, considered to be the most abundant mineral in the Earth, decomposes to its component oxides above 65 GPa, as suggested by recent studies [65]. We performed for the first time C02laser heating experiments on the stability of MgSi03 perovskite under hydrostatic conditions between 73 and l00GPa and up to 3000K. In addition to using MgSi03 glasses and crystals, we also employed a new approach using stoichiometric mixtures of Si02 and MgO as starting materials. MgSi03-perovskite formed as the single phase to the highest pressures and temperatures of our measurements (lOOGPa, 3000K) [66].

60


Thus, at the present pressure-temperature conditions, which span most of the Earth's lower mantle, silicate perovskite is denser than an assemblage of its component oxides, MgO and SO2.

2.5 C02-laser Heating Experiments on Organic Compounds The C02-laser heated DAC can be used for in situ investigation of the high P-T behavior of organic compounds containing carbon, boron, and nitrogen, as well as the synthesis of new materials. Graphite and diamond, hexagonal and cubic BN, molecular hydrogen, and nitrogen can be easily identified in the diamond cell, or in the recovered reaction products by Raman spectroscopy. It was shown [67] that diamond can be synthesized from a number of organic compounds (like polyethylene or paraffin) at high pressures and temperatures. Shock-wave experiments on methane [68] suggest that this alkane decomposes into molecular hydrogen and diamond at pressures above 20 GPa and temperatures of about 2500 K. In contrast, recent first-principles molecular dynamic simulations [69] suggest that below 100 GPa methane polymerizes into a mixture of methane, ethane, and butane. In order to examine if the decomposition products depend on the chain length of the starting hydrocarbon material, we performed C02-laserheatand nonadecane ing experiments in a DAC on methane CH4, octadecane C18H38, C19H4,, [70]. In all these experiments no pressure medium was used, resulting in axial temperature gradients of several hundred K pm-' [21], thus the temperature values given below are only approximate. After heating at around 20 GPa, we observed formation of diamond for all three hydrocarbons, as evident from Raman-spectroscopic analysis of the recovered samples at 1 bar. We note that other decomposition products escaped from the sample volume after opening the diamond cell. We found that diamond formed from longer chain alkanes (e.g. octadecane) at lower temperatures than from short ones (e.g. methane). Figure 16A shows an SEM picture of the recovered diamond aggregate synthesized from octadecane at 21 GPa and at temperatures exceeding 3000K. On the surface of this diamond aggregate we found a few spheres with diameters of less than 1 pm, which indicated melting of diamond (see Fig. 16B). The above results suggest that alkanes can be used as a carbon source for diamond synthesis. We also performed C02-laser heating experiments on borane-dimethylamine BH3 NH(CH3)*to check if organic materials containing boron and nitrogen can be used for synthesis of ternary BNC, compounds with diamond-like structure. Borane-dimethylamine was heated at 23 GPa to about 3000 K. Raman spectra from the heated sample area revealed the presence of cubic BN, and of liquid H2 at pressures below the solidification point of hydrogen (5.5 GPa at room temperature [71]) (Fig. 17). A Raman spectrum of the recovered white agglomerate showed a weak line of diamond in addition to the Raman bands of cubic BN. Thus, borane-dimethylamine transformed to cubic BN, diamond and hydrogen

2.5 COJaser Heating Experiments on Organic Compounds

61

Figure 16. (A) SEM-image of the diamond agglomerate recovered from a diamond cell after heating of octadecane to above 3000K at 21 GPa. The area bounded by the dashed lines is enlarged in (B), showing two spheres of molten diamond (indicated by arrows) on the agglomerate surface.

rn

I

400

'

I

600

.

,

800

.

I

1000

.

,

1200

.

1400

4200

4400

rel. wavenumber (cm-') Figure 17. Raman spectra of liquid H2 and cubic BN at 0.6 GPa in a DAC after heating of boranedimethylamine to about 3000 K at 23 GPa.

62


after heating at 23 GPa. No evidence for the presence of ternary BNC, in the heated sample volume was found.

2.6 Conclusion The studies described above show that the C02-laser heated diamond anvil cell can be used for investigating nonconducting materials under controlled conditions in a very broad pressure and temperature range. Samples can be analyzed in situ using Raman spectroscopy, fluorescence spectroscopy, and X-ray diffraction. For structural, chemical, and textural analysis of the recovered samples techniques such as atomic force microscopy, scanning- and transmission electron microscopy can be employed. All of the above techniques can be used for determining melting curves, developing high P-high T phase diagrams, evaluating the stability of solids, and monitoring of chemical reactions, which are essential for controllable synthesis and characterization of new materials.

Acknowledgments We thank 0. Tschauner for fruitful discussions and the Editor of the book for encouraging us to undertake this work. A.Z. gratefully acknowledges the financial support of the Deutsche Forschungsgemeinschaft.

References 1. A. L. Ruoff, H. Xia, and Q. Xia, The effect of a tapered aperture on x-ray diffraction from sample with a pressure gradient: Studies on three samples with a maximum pressure of 560GPa, Rev. Sci. Znstrum. 1992, 63, 43424348. 2 . M. Eremets, High Pressure Experimental Methods, Oxford University Press, Oxford, 1996. 3. R. M. Hazen and L. W. Finger, Comparative Crystal Chemistry, John Wiley & Sons Ltd., Chichester, 1982. 4. A. Jayardman, Diamond anvil cell and high-pressure physical investigations, Rev. Mod Phys. 1983,55, 65-108. 5 . W. L. Wolfe and G. J. Zissis, The Infrared Handbook, Office of Naval Research, Department of the Navy, Washington, DC, 1985. 6. R. A. Forman, G. J. Piermarini, J. D. Barnett, and S. Block, Pressure measurement made by the utilization of ruby sharp-line luminescence, Science, 1972, 176, 284-285. 7. H. K. Mao, J. Xu, and P. M. Bell, Calibration of the ruby pressure gauge to 800 kbar under quasi-hydrostatic conditions, J . Geophys. Res. 1986, 91, 46734676. 8. R. Boehler, M. Nicol, and M. L. Johnson, Internally heated diamond anvil cell: Phase diagram and P-V-T of iron, in High Pressure Research in Mineral Physics, Geophys. Monogr. Ser., M. H. Manghnani and Y. Syono (Eds), AGU, Washington, DC, 1987, pp. 173-176.

Reterences

63

9. L. G. Liu and W. A. Bassett, The melting of iron up to 200 kbar, J . Geophys. Res. 1975, 80, 3777. 10. M. Planck, Entropie und Temperatur strahlender Warme, Ann. P h j x 1900, 1, 719-737. 11. L. Ming and W. A. Bassett, Laser heating in the diamond anvil press up to 2000°C sustained and 3000°C pulsed at pressures up to 260 kilobars, Rev. Sci. Instrum. 1974, 45, 11 15-1 118. 12. C.-S. Yoo, J. Akella, and M. Nicol, Chemistry at high pressures and temperatures: In-situ synthesis and characterization of P-Si3N4 by DAC X-ray/laser heating studies, in The 3rd NIRIM International Symposium on Advanced Materials (ISAM'96), M. Akaishi et al. (Eds), NIRIM, Tsukuba, Japan, 1996, pp. 175-179. 13. R. Boehler and A. Chopelas, A new approach to laser heating in high pressure mineral physics, Geophys. Res. Lett. 1991, 18, 1147-1 150. 14. M. Eremets, K. Takemura, H. Yusa, D. Goldberg, Y. Bando, and K. Kurashima, Melting of cubic and hexagonal BN, in The 3rd NIRIM International Symposium on Advanced Materials (ISAM'96). M. Akaishi et al. (Eds), NIRIM, Tsukuba, Japan, 1996, pp. 169-174. 15. J. H. Nguyen and R. Jeanloz, Initial description of a new carbon-nitride phase synthesized at high pressures and temperatures, Muter. Sci.Eng. 1996, A209, 23-25. 16. G. Serghiou, A. Zerr, A. Chopelas, and R. Boehler, The transition of pyrope to perovskite, Phj:F. Chem. Min. 1998, 25, 193-196. 17. G. Serghiou. A. Zerr, L. Chudinovskikh, and R. Boehler, The coesite-stishovite transition in a laser-heated diamond cell, Geophys. Res. Lett. 1995, 22, 441444. 18. F. Cabannes, Facteurs de reflexion et d'emission des metaux, J. Physique, 1967, 28, 235-248. 19. J. C. De Vos, A new determination of the emissivity of tungsten ribbon, Physica, 1954, 20,69& 714. 20. U. P. Oppenheim and A. Goldman, Infrared spectral transmittance of MgO and BaF, crystals between 27" and IOOO'C, J . Optic. Soc. Am. 1964,54, 127-128. 21. A. Zerr, Schmelztemperaturen von (Mg, Fe)S03-Perowskit und von MgO bei hohen Drucken, PhD thesis, University of Mainz, Mainz, 1995. 22. S. R. Elliot. Physics of Amorphous Materials, Longman, London, 1990. 23. R. J. Hemley, P. M. Bell, and H. K. Mao, Laser techniques in high-pressure geophysics, Science, 1987, 237, 605-612. 24. G. Serghiou, A. Zerr, and A. Chopelas, Phase transitions and disordering of CaSi03 upon compression, Rev. High Press. Sci. Techno/. 1998, 7, 359-361. 25. A. Chopelas, The fluorescence sideband method for obtaining acoustic velocities at high compressions: application to MgO and MgA1204, Phys. Chem. Min.1996, 23, 25-37. 26. A. Chopelas and R. Boehler, Thermal expansivity in the lower mantle, Geophys. Res. Lett. 1992, 19, 1983-1986. 27. F. R. Corrigan and F. P. Bundy, Direct transition among allotropic forms of boron nitride at high pressures and temperatures, J . Chem. Phys. 1975, 63, 3812-3820. 28. D. A. Young, C . 4 . Zha, R. Boehler, J. Yen, M. Nicol, AS. Zinn, D. Schiferl, S. Kinkead, R. C. Hanson, and D. A. Pinnick, Diatomic melting curves to very high pressure, Phys. Rev. B, 1987, 35, 5353-5356. 29. P. E. van Keken, D. A. Yuen, and A. P. van den Berg, Implications for mantle dynamics from the high melting temp,eratures of perovskite, Science, 1994, 264, 1437-1439. 30. %-I. Karato, Importance of anelasticity in the interpretation of seismic tomography, Geophys. Res. Lett. 1993, 20, 1623-1626. 31. A. Zerr and R. Boehler, Constraints on the melting temperature of the lower mantle from high pressure experiments on MgO and magnesiowustite, Nature, 1994,371,506-508. 32. B. Riley, The determination of melting points at temperatures above 2000" Celcius, Rev. Hautes Temper. Refract. 1966, 3, 207-2 11. 33. A. Zerr, G. Serghiou, and R. Boehler, Melting of CaSi03-perovskiteto 430 kbar and first in-situ measurements of lower mantle eutectic temperatures, Geophys. Res. Lett. 1997, 24, 909-912. 34. A. Zerr and R. Boehler, Melting of (Mg, Fe)Si03-perovskite to 625 kbar: indication of a high melting temperature in the lower mantle, Science, 1993, 262, 553-555. 35. A. Zerr, A. Diegeler, and R. Boehler, Solidus of Earth's deep mantle, Science, 1998,281,243-246. 36. J. Zhang and C. Herzberg, Melting experiments on anhydrous peridotite KLB-1 from 5.0 to 22.5 GPa, J . Geophys. Res. 1994,99, 17729-17742.

64

2 Phase Transitions and Material Synthesis using the CO,-Laser Heating Technique

37. Z. Gong, R. E. Cohen, and L. L. Boyer, Molecular dynamics simulations of melting of MgO at high pressures, in Annu. Rep. Director Geophys. Lab. Carnegie Instn. Wash., Washington, DC, 1991, pp. 129-134. 38. I. Jackson, Phase relations in the system LiF-MgF2 at elevated pressures: Implications for the proposed mixed-oxide zone of the Earth’s mantle, Phys. Earth Planet. Inter. 1977, 14, 86-94. 39. E. Ohtani, Melting temperature distribution and fractionation in the lower mantle, Phys. Earth Planet. Inter. 1983, 33, 12-25. 40. J. J. Gilvarry, The Lindemann and Gruneisen laws, Phys. Rev. 1956, 102, 308-316. 41. F. A. Lindemann, Uber die Berechnung molekularer Eigenfrequenzen, Phys. Z . 1910,11,609612. 42. E. A. Kraut and G. C. Kennedy, New melting law at high pressures, Phys. Rev. Lett. 1966,16, 608-609. 43. R. Boehler, Adiabats of quartz, coesite, olivine, and magnesium oxide to 50 kbar and 1000K, and the adiabatic gradient in the Earth’s mantle, J . Geophys. Res. 1982,87, 5501-5506. 44. D. Heinz and R. Jeanloz, Measurement of the melting curve of Mgo 9Feo Si03 at lower mantle conditions and geophysical implications, J. Geophys. Res. 1987,92, 11437-1 1444. 45. E. Knittle and R. Jeanloz, Melting curve of (Mg, Fe)Si03 perovskite to 96 GPa: evidence for a structural transition in lower mantle melts, Geophys. Res. Lett. 1989, 16,421424. 46. J. S. Sweeney and D. L. Heinz, Melting of iron-magnesium-silicate perovskite, Geophys. Res. Lett. 1993, 20, 855-858. 47. G. Shen and P. Lazor, Measurement of melting temperatures of some minerals under lower mantle pressures, J . Geophys. Res. 1995, 100, 17699-17713. 48. E. lto and E. Takahashi, Melting of peridotite at uppermost lower-mantle conditions, Nature, 1987,328, 514-517. 49. E. Takahashi, Melting of a dry peridotite KLB-1 up to I4GPa: Implications on the origin of peridotitic upper mantle, J. Geophys. Res. 1986, 91, 9367-9382. 50. K. G. Holland and T. J. Ahrens, Melting of (Mg, Fe)2Si04at the core-mantle boundary of the Earth, Science, 1997, 275, 1623-1625. 51. J. Haines, J. M. Leger, and 0. Schulte, Pa3 modified fluorite-type structures in metal dioxides at high pressure, Science, 1996, 271, 629-631. 52. J. M. Leger, J. Haines, M. Schmidt, J. P. Petitet, A. S. Pereira, and J. A. H. Jornada, Discovery of hardest known oxide, Nature, 1996, 383, 401. 53. V. Swamy, S. K. Saxena, B. Sundman, and J. Zhang, A thermodynamic assessment of silica phase diagram, J. Geophys. Res. 1994,99, 11787-1 1794. 54. T. Yagi and S. Akimoto, Direct determination of coesite-stishovite transition by in-situ X-ray measurements, Tectonophysics, 1976,35, 259-270. 55. J. Zhang, B. Li, W. Utsumi, and R. Liebermann, Reversal of the coesite-stishovite phase transformation, in AGU Spring Meeting, American Geophysical Union, Baltimore, 1994, p. 346. 56. S. Akimoto and Y. Syono, Coesite-stishovite transition J . Geophys. Res. 1969,74, 1653-1659. 57. K. Suito, Phase relations of pure Mg2Si04up to 200 kilobars, in High Pressure Research: Application in Geophysics, M. H. Manghnani and S. Akimoto (Eds), Academic Press, Inc., New York, 1977, pp. 255-266. 58. J. Zhang, R. C. Liebermann, T. Gasparik, and C. T. Herzberg, Melting and subsolidus relations of Si02 at 9-14GPa, J. Geophys. Res., 1993,98, 19785-19793. 59. A. E. Ringwood, Composition and Petrology of the Earth’s Mantle, McGraw-Hill, New York, 1975. 60. P. McMillan, M. Akaogi, 1. Ohtani, Q. Williams, R. Nieman, and R. Sato, Cation disorder in garnets along the Mg3A12Si3012-Mg4Si,0,2join: an infrared, Raman and NMR study, Phys. Chem. Min. 1989, 16,428435. 61. H. Yusa, M. Akaogi, and 1. Eiji, Calorimetric study of MgSi03 garnet and pyroxene: heat capacities, transition enthalpies, and equilibrium phase relations in MgSi03 at high pressures and temperatures, J . Geophys. Res. 1993, 98, 64534460. 62. T. Irifune, T. Koizomi, and J. Ando, An experimental study of the garnet-perovskite transformation in the system MgSiO3-Mg3AI2Si3Ol2,Phys. Farth Planet. Inter. 1996, 15, 90-106. 63. M. Kanzaki, Ultra high-pressure phase relations in the system MgSiO3-Mg3AI2Si3Ol2,Phys. Earth Planet. Inter. 1987, 49, 168-175.

References

65

64. L. Zhang and A. Chopelas, Sound velocity of A1203 to 616kbar, Phys. Earth Planet. Inter. 1994,87, 77-83. 65. S. K. Saxena, L. S. Dubrovinsky, P. Lazor, Y. Cerenius, P. Haggkvist, M. Hanfland, and H. Jingzhu, Stability of perovskite (MgSi03)in the Earth’s mantle, Science, 1996,274, 1357-1359. 66. G. Serghiou, A. Zerr, and R. Boehler, (MgFe)Si03-perovskite stability under lower mantle conditions, Science, 1998,280, 2093-2095. 67. R. H. Wentorf, The behavior of some carbonaceous materials at very high pressures and high temperatures, J. Phys. Chem. 1965,69, 3063-3069. 68. M. Ross and F. H. Ree, Repulsive forces of simple molecules and mixtures at high density and temperature, J Chem. Phys. 1980, 73, 614M152. 69. F. Ancilotto, G. L. Chiarotti, S. Scandolo, and E. Tosatt, Dissociation of methane into hydrocarbons at extreme (planetary) pressure and temperature, Science, 1997, 275, 1288-1290. 70. A. Zerr, G. Serghiou, and R. Boehler, C02-laser heating in a diamond cell to 80GPa and 5000 K: phase transitions and synthesis under controlled conditions, in The 5th NIRIM International Symposium on Advanced Materials (ISAM’98), M. Kamo et al. (Eds), NIRIM, Tsukuba, Japan, 1998, pp. 5-8. 71. S. K. Sharma, H. K. Mao, and P. M. Bell, Raman measurements of hydrogen in the pressure range 0.2-630 kbar at room temperature, Phys. Rev. Lett. 1980, 44,886-888.


3 Mechanical Properties and their Relation to Microstructure D. Sherman and D. Brandon

3.1 Introduction In this contribution we review current knowledge of the mechanical properties of hard materials and the relation between these properties and their microstructure. However, a quantitative relationship is only meaningful to the extent that test methods are clearly defined and understood, while the test data are only useful within the context of the engineering requirements which determine material selection for a specific application. It is also important to define what is meant by a ‘hard material’. We therefore discuss briefly some typical applications of hard materials and the engineering requirements that must be satisfied in these applications. In Section 3.2 we define the mechanical properties which need to be evaluated, while in Section 3.3 we review the test method which have been developed to ensure accurate and reproducible measurements. We then examine the microstructure-property relations and the size effects that dominate the mechanical strength of brittle materials, as well as some important failure mechanisms.

3.1.1 Applications and Engineering Requirements Structural components are required to transmit a load, and their mechanical strength and stiffness are the primary concerns of the design engineer [1,2]. Hard materials, on the other hand, are selected, as the term implies, for their hardness and we should consider why ‘hardness’ is not a synonym for ‘strength’. A good definition for hardness is ‘the resistance to penetration’ [3], emphasizing that this property relates to the surface response to a mechanical load, and is not necessarily a ‘bulk’ property. Hard materials may be used for structural components, but they are also used as coatings [4,5], to improve the resistance of a structural component to superficial mechanical damage. Thin, hard coatings may be submicron, but are more commonly a few microns in thickness. Thick coatings may approach or exceed a millimeter in thickness. All hard coatings are dependent for their engineering performance on their adhesion to the substrate, which is commonly achieved by combining a high-strength chemical bond with control of the substrate roughness, in order to optimize mechanical interlocking [6]. Coatings are susceptible to the presence of residual stress, either associated with the mismatch in physical properties across the coating-substrate interface, or due to the deposition mechanism. Compressive residual stress in the coating

3.1 Introduction

67

may be beneficial, inhibiting subcritical crack growth and delaying the onset of damage, but residual tensile stresses are always detrimental. Mismatch in the thermal expansion coefficient and the elastic moduli of the coating and the substrate are common sources of residual stress. It is important to distinguish between elastic and plastic penetration of an indentor. An elastomer has a low resistance to elastic deformation, but releasing the load will generally leave no mark on the rubbery elastomer. By contrast, a refractory metal may have a poor resistance to plastic penetration, but show relatively little elastic relaxation when the indentor load is relaxed [3]. Throughout this review we assume the elastic response to be small (but not negligible) compared to the plastic response, so that the high hardness of a ‘hard material’ is also accompanied by high stiffness, corresponding to a large tensile modulus. The mode of loading and the rate at which load is applied at the surface is important [3]. There is a clear distinction between a sliding bearing, in which contacts between the components are predominantly subjected to shear parallel to the contact surface, and a roller (or ball) bearing in which the mechanical contacts are made and broken in tension (perpendicular to the contact surface). Most mechanical contacts can be approximated to ‘point’ contacts, since the ‘perfectly flat’ surface has yet to be invented. More exactly, the ‘true’ contact area depends on the load applied, the rigidity of the components, and their yield strength. A high hardness implies a minimum contact area at a given normal load. It follows that contact damage is, in the first instance, localized and restricted to the contact regions. Mechanical wear occurs by removal of material from points of contact, followed by transfer of the load to new points of contact. Areas worn down by mechanical wear will establish renewed contact once neighboring areas are similarly eroded, leaving the original contact areas once more proud of the nominal contact surface. Since surface damage is localized, events taking place at the surface often occur rapidly. True contact areas are often on the scale of the surface finish, less than a micron for a well-polished surface. Sliding velocities of the order of meters s-’ then correspond to contact times of less than a microsecond, so that the localized damage events are commonly adiabatic, and accompanied by intense ‘frictional’ heating [ 7 , 8 ] . Hence the importance of cooling the workpiece in all grinding, polishing, and mechanical working operations. Susceptibility to brittle failure is the penalty accompanying the desirable properties of hard materials [9]. Premature brittle failure can be prevented, either by suitable engineering design or by ensuring that brittle crack propagation is inhibited. In successful engineering design tensile loading is minimized, since brittle materials are commonly strong in compression. This may be achieved either by pre-loading in compression (assembly of the component in constraint) or by ensuring that the operating loads are compressive. Fracture toughness is the common parameter used to measure the resistance to brittle crack propagation [9-111, but many hard materials exhibit R-curve behavior, in which the critical stress intensity factor increases with crack length [9,12]. The mode I fracture toughness determines the resistance to macroscopic crack propagation, but the short-crack resistance or ‘damage tolerance’ [131 often determines the performance in service, when damage is commonly localized at the surface.

68

3 Mechanical Properties and their Relation to Microstructure

Clearly, the primary engineering requirements for bearing components and cutting tools include high stiffness and hardness, maintained to high temperatures [14]. These are also the primary considerations in the selection of grit for grinding wheels and polishing compounds. Erosion-resistant surface coating systems are often based on hard and rigid materials, which may also employ elastomeric materials to resist plastic but not elastic penetration. Armor systems are also designed to resist penetration, at both sub- and supersonic velocities, and also employ hard materials [15,161. Finally, two other characteristics of hard materials are important: their density and their chemical stability. The importance of these properties depends on the application. Oxides are chemically more stable than nitrides, which are in turn more stable than borides or carbides. The susceptibility to oxidation of boron carbide prevents its application at high temperatures, but for ballistic protection, where hardness, rigidity and low density take precedence, boron carbide is unparalleled. Tungsten carbide, on the other hand, is the material of first choice for cutting tools, because of its high hardness and stiffness, and high-temperature oxidation resistance, but is far too dense for application in which weight or inertial forces are important. Its use for ballistic protection is therefore out of the question.

3.1.2 Bulk Components Hard materials available for the production of bulk components fall into two major groups: ‘hard metals’, used primarily in the manufacture of cutting tools and related applications, and structural ceramics, primarily the oxides, carbides, borides, or nitrides of the low atomic number cations. Relatively few materials are of engineering importance, and we will list these, explaining briefly why similar compositions are less useful.

3.1.2.1 Tungsten Carbide-Cobalt, Tungsten Carbide-Nickel, Vanadium Carbide The bonding of tungsten carbide particles by cobalt (and, to a lesser extent, nickel) has dominated the cutting tool industry for over 70 years [17,18]. There has been a steady improvement in the compositions available, including additional carbide particulates, control of the cobalt-nickel binder composition, and a reduction in the tungsten carbide particle size. The commonest cutting tool formulations contain 6-12 weight-% binder. Production of these ‘hard metal’ components depends on active wetting of partially soluble carbide particles by the molten binder phase. The formation of deleterious secondary phases during liquid phase sintering and cooling to room temperature must be avoided. The success of hard metals in cutting tool applications is a result of their unique combination of hardness and toughness, tailored by controlling the ratio of binder to carbide phase, together with their unique microstructure. Both the ductile binder

3.1 Introduction

69

and the hard carbide phases are continuous. Since the ductile binder is continuous, no crack can propagate without expending energy in ductile failure of the binder. On the other hand, the three-dimensional continuity of the carbide skeleton maximizes rigidity and stiffness of the composite for any given carbide content, so toughness is obtained with a minimum loss of hardness. Bulk cutting tools based on alternative materials, most notably alumina and silicon nitride, have been developed [ 19,201, but, although niche applications have been found, no substitute for tungsten carbide has yet been discovered. The closest to a ‘breakthrough’ has been the development of polycrystalline diamond (which lacks oxidation resistance) [21]. More recently, cubic boron nitride (cBN) has found a large and growing market in high speed machining thanks to a combination of hardness and oxidation resistance [22,23]. 3.1.2.2 Ceramics: Oxides, Carbides, Borides, Nitrides While ‘hard metals’ may have a fracture toughness of up to 20MPam1/2, most engineering ceramics have a toughness well below 10 MPa m1I2 [ l ,3,9]. The sole exceptions are the composite microstructures, associated with ‘directed metal oxidation’ [24] and some other processing routes, and the transformationtoughened zirconia compositions. Most of these materials lack the stiffness and hardness required for wear-resistant components. The strength and rigidity of the chemical bond in a ceramic depend on the valency and the radius of the cation [25]. The bond strength (but not the chemical stability) generally decreases in going from carbides to nitrides to oxides, while the highest elastic moduli are expected for low atomic number cations. Any choice is a compromise in which the primary requirements for hardness and stiffness (optimizing wear resistance) are modified by secondary requirements for chemical stability, inertial response and selected thermal properties (expansion coefficient, thermal conductivity or thermal diffusivity). Many ceramics have limited engineering applications because they fail to meet one or other of the secondary requirements. For example, both magnesium oxide and aluminum carbide are unstable in the presence of moisture. In other cases an otherwise important engineering ceramic is unsuitable for a specific application. Thus alumina has poor toughness and a relatively large thermal expansion coefficient. Thermal stresses generated by adiabatic heating at the surface of an alumina wear component can cause surface microdamage, leading to the formation of hard wear particles which may cause abrasive damage of components far removed from the site of the alumina bearing surface. Silicon nitride bearings, on the other hand, have found a rapidly growing market, thanks to painstaking development of high performance grades of silicon nitride, with high failure strength and fracture toughness at temperatures of up to 1200°C. Silicon nitride has a lower elastic modulus than alumina, a significantly smaller coefficient of thermal expansion, and on oxidation forms soft, often protective, silica films. This combination of properties gives better wear resistance than can be achieved in the aluminas. By contrast, alumina is the material of first choice for the ball joint in a hip prosthesis, mounted against a high density polyethylene cup. In

70

3 Mechanical Properties and their Rehtion to Microstrircture

this application lack of toxicity is a primary concern, while the wear path is limited, so that wear is not a major concern.

3.1.3 Coatings Hard coatings can solve the problem of maximizing the hardness of the surface, to optimize wear resistance, without sacrificing the toughness of the bulk component and risking brittle fracture [26]. A successful hard coating must be compatible with the substrate, exhibiting good adhesion and minimizing residual stress in the coating. A wide spectrum of coating technologies is available, encompassing electrolytic coatings, anodizing, diffusion coatings, plasma spraying, chemical vapor deposition, laser deposition, and more. The various coating technologies have one thing in common: the extraordinary difficulty in determining their mechanical properties. The problem can be reduced to three separate factors: The mismatch in elastic properties between the coating and the substrate, which leads to interfacial shear stresses at any discontinuity in the coating; the dependence of the mechanical properties on thickness, both because of thickness-dependent microstructural changes and owing to a size dependence of the properties; difficulties in standardizing and interpreting mechanical tests on coatings. These considerations are independent of other problems: surface roughness and adhesion, which are usually avoided by assuming, or requiring, that the adhesive strength of the coating-substrate interface exceeds the cohesive strength of the coating, and residual stresses in the coating, typically biaxial, which also affect the critical load for microcracking, peeling, and (eventually) spalling (flaking).

3.1.4 Engineering Requirements It is the ability of a material to satisfy the engineering requirements for a given component which is important, not the mechanical properties as such. Determining the mechanical properties is, in many respects, just a way of grading the desirability of candidate materials for a specific task. Specification of the value of specific properties (bend strength, elastic modulus, or fracture toughness) is generally insufficient (by itself) to determine fitness for purpose. The engineering requirements for specific applications (wear components, cutting tools, mechanical bearings, or armor) vary dramatically. A wear component must exhibit high wear resistance, and good performance is associated with a high elastic modulus, a high damage tolerance, and good chemical stability (absence of attack by acid residues in a lubricating oil, for example). A good candidate for composite armor is expected to have a high Vso(the velocity at which a given projectile has a 50% chance of penetration). This is clearly a function of the armor thickness and the rigidity of the mounting, and it is usual to add that the weight per unit area should not exceed a given value (placing a premium on a low density material such as boron carbide).

3.2 Principal Mechanical Properties

71

3.2 Principal Mechanical Properties 3.2.1 Elastic Modulus The elastic moduli are an intrinsic mechanical property of a material dominated by the strength of the chemical bond between atoms. The potential energy associated with the bond between a pair of atoms is determined by a balance of the forces of attraction and repulsion

where a and ,B are constants related to the attraction and repulsion components, r is the distance between the atoms, and n > m. All four constants can be treated as material properties. The net force between the atom pair is just

and is zero for r = ro, the equilibrium spacing in the absence of external forces. An external driving force displaces the atoms by a distance u from the equilibrium position, where u = r - ro, and the relationship between the external driving force and the displacement is F=-=dEp dr

u.

(3)

U=O

The second derivative, which determines the curvature, of the potential energy well near y o , is constant for a given material and temperature, leading to Hooke's law: F = Ku. The constant K is proportional to the elastic, or Young's modulus K

o(

*I

dU2 "=o

= f ( a ,p, n, m).

(4)

K is therefore the proportionality constant between stress and strain in elastic deformation. The modulus of elasticity is effectively independent of microstructure and minor alloying additions, but does depend on temperature, reflecting changes in equilibrium spacing associated with thermal expansion. Ionic or covalent bonds deform less than metallic bonds under a given external force, and so have a higher modulus of elasticity. The high elastic modulus of hard materials reflects the strong chemical bonds between the atoms. For comparison, the elastic moduli of steel, alumina, and WC are 210, 400, and over 500GPa, respectively [ 1,2,9]. Many hard materials of practical importance are polyphase materials, and it is important to formulate their effective elastic modulus. Two bounds exist for calculating the composite modulus, based on elastic deformation of the individual phases. The upper, Voigt bound, assumes compatible displacements or strains.

72


The elastic modulus, Ec, is then given by the ‘rule of mixtures’:

Ec =

c

EiL,

n

whereh is the volume fraction of component i. The Reuss model gives a lower bound for the elastic modulus, and assumes compatible loads or stresses: 1

J;:

E,=FE+ Most polyphase materials have elastic moduli which lie between these two bounds, and more sophisticated models have been developed to narrow the gap. The Hashin-Shtrikman model, in particular [27], has given better predictions for the elastic modulus of a composite, while, more recently, Aboudi has used a cell model [28] and finite element analysis has also been used to derive the relationship between the volume fraction and effective modulus [29-321. These methods also estimate the displacement and stress field in the composite.

3.2.2 Strength The mechanical strength of hard materials is critical for load-bearing, structural applications. These brittle materials only deform plastically at high temperatures, or under severe hydrostatic constraint, since the Peierls stress for dislocation movement is high. Failure is usually by unstable crack propagation under a tensile stress that exceeds the tensile strength of the material. In terms of fracture mechanics, brittle failure occurs when the Mode I stress intensity factor KI reaches the fracture toughness of the material, K I c (see below). An experimental uniaxial stress-strain relationship, determined in tension, would provide the necessary design information for ceramics and hard materials: Young’s modulus, Poisson’s ratio and the tensile fracture strength. However, tensile data for brittle materials are often unreliable, due to parasitic bending stresses associated with dimensional inaccuracy in machining tensile ‘dog-bone’ specimens and misalignment of the specimen grips. The test procedure most commonly used to validate the strength of ceramics and hard materials is a flexure test, in either three-point (3PB) or four-point bending (4PB). The specimens for bend tests are simple to machine and require less material, while misalignment and parasitic bending can be avoided. The strength determined in bending is variously termed the flexural strength, the transverse rupture strength, TRS, or the modulus of rupture, MOR, and corresponds to the maximum tensile stress developed in the surface ligaments of the sample at the point of fracture. Several issues need to be addressed when a bend test is used to evaluate the strength: The sources of errors that are responsible for statistical scatter in the results; the statistical nature of the strength of a brittle material; the relationship between the measured flexural strength and the strength derived from a valid uniaxial tensile test, together with the effect of sample size on the strength; and


Loadpointfriction

73

u

.

-

Twisted specimen orfixture

n

n

Reducedmomentofinertia

Figure 1. Source of errors in bending tests. Reproduced with permission of . . . . . . . . . . (Ref. 33).

the relationship between the properties measured on a laboratory sample or test piece with those required of an engineering component under operating conditions. A major cause of experimental scatter in determining the strength of brittle materials is the deviation from ideal geometry [33]: nonprismatic specimens, a nonparallel loading path, deviations from pure articulation, eccentricity in point-loading, inaccurate measurements of load span and specimen dimensions, or inadequate span :height ratio; all resulting in nonuniform bending and stress concentrations, Fig. 1. 'Real' scatter of material properties is typically associated with processing defects and microstructural inhomogeneity in the material, especially residual porosity, impurity inclusions, and exaggerated grain growth. Brittle materials fail by unstable crack propagation from a pre-existing flaw. The size, orientation, and shapes of the flaws in a given flaw population determine the variability of the failure stress; the statistical failure probability, F, first formulated by Weibull [34] and described by the Weibull function: F = I -exp[-jv[g]"du],

(7)

where uoand m denote the median failure stress [35,36] and the Weibull modulus, u, is the volume of the test piece. The material parameter uomeasures the amplitude of the failure strength, while the parameter rn measures the strength variability. If the

74


failure probability function for N specimens is known, then the failure probability of the ( N 1)th specimen can be predicted. The number of tests N required to ensure predictive reliability is typically 20. If a 3PB test piece is loaded, the volume element subjected to the maximum tensile stress is very small. This volume is much larger for a 4PB specimen, but a unidirectional tensile specimen will have the largest volume under maximum stress. It follows that the median strength of a 3PB specimen may exceed that of a 4PB specimen, which in turn should have a higher average strength than a tensile specimen. The measured strength ratio between specimens of identical size subjected to 3PB and to pure tension is predicted to be [34,37]:

+

which yields 1.73 for m = 10, a typical value for a structural ceramic. The predicted strength ratio between a 4PB specimen (inner/outer span = 0.5) and a same-size tensile specimen is:

ffT

(9)

which gives a strength ratio of 1.45 for same value of m = 10. Clearly, the loading geometry must be considered when comparing the strength of ceramic components. Actually, it was Leonard0 da Vinci (1452-1519) who first noticed that the strength of steel wire increased for decreasing wire length. Indeed, assuming uniform stresses in test pieces having different volume but the same failure probability, Weibull statistics predicts a strength ratio:

which yields the value of 2 for V2 = 1000V1and m = 10, a typical situation for some small test samples and the corresponding large engineering component. The elastic modulus, the fracture toughness, the density, and the microstructure, by contrast, are not usually sensitive to the size of the engineering component, and it is principally the strength which is size-sensitive, and hence not a true material parameter. This has been demonstrated for a commercial glass ceramic [38], and for alumina [39]. It has been suggested that the microstructure and the defect populations themselves may vary throughout the volume of a component, causing variations in strength unpredictable by Weibull analysis [40,41]. An attempt to verify the minimal specimen size required for strength determination was carried out lately [42].

3.2.3 Fracture Toughness Griffith [43,44], who was the first to relate the strength of brittle materials to material properties and crack length, and initiated the discipline of fracture


75

S

Figure 2. The Griffith cracked body A, with free surfaces S, and crack length 2a.

mechanics. Griffith calculated the balance between mechanical energy and internal energy in a solid body containing a crack: UE us, (11) where FLand UE are the external work and elastic strain energy stored in the body under stress, respectively and Us is the surface free energy, Fig. 2. For linear elastic bodies, FL = 2uE. UE was calculated from Inglis' [45] expressions for the stress field in a body containing an elliptical hole. An instability occurs when the change in the mechanical energy exceeds the surface energy required for a unit increase in crack area: dU -= 0. dA This critical point marks the onset of crack propagation. Griffith calculated the critical stress far from the crack location for a plate of unit width, and showed that for a homogeneous, isotropic material, a relationship exists between the critical stress, uc and half the crack length, a:

u = -FL

76


J.

I Figure 3. The three basic modes of fracture, distinguished by the displacement driving force: I, the opening mode; 11, the shear mode; and 111, the out of plane shear mode.

where ”is is the surface energy, E* = E for loading under plane stress conditions, and E* = E/(1 - v2) for loading under plane strain, E and v are the Young modulus and Poisson ratio respectively, assuming homogeneous, isotropic materials. Eq. (13) does not involve the local stress field in the vicinity of the crack tip, and that the only energy characteristic of the material is the surface free energy. Three possible deformation ‘modes’can be defined, Fig. 3. Mode I, is a tensile mode characterized by an identical displacement normal to the crack plane for any two adjacent points on the free surface of the crack. Mode I1 is an in-plane shear mode, in which the two adjacent points have the same (but with opposite sign) displacement in the plane of the crack and parallel to the direction of crack propagation. Mode I11 corresponds to an out-of-plane shear mode; with shear displacements in the plane of the crack but perpendicular to the direction of crack propagation. Crack propagation seldom occurs under a pure tensile stress, and shear stresses are commonly present in the early stages of crack propagation, so that most crack propagation takes place, to some extent, under mixed-mode loading conditions. Irwin [461and Orowan [47] generalized the Griffith energy balance, using the term ‘strain energy release rate’ (SERR), 9,to characterize a general relationship between crack length, sample geometry and the mode of loading. In mode I deformation, is defined by: assuming unit thickness of the cracked body,

or

3.2 Principal Mechanical Properties P

77

P

P,6

(4

(b) 6 = constant

(c) P = constant

Figure 4. Strain energy release during constant displacement and constant load.

where U is the strain energy stored in the cracked body, 7r the complementary energy, a the projected crack length, and da is crack increment, as shown in Fig. 4a. For an imposed displacement (displacement control) the strain energy is reduced for increasing crack length, Fig. 4b, and the crack advances stably, while for a fixed applied load (load control), Fig. 4c, the strain energy increases for increasing crack length, and the crack advances unstably, if brittle materials is considered. Increasing the external applied driving force will increase the strain energy release rate until the strain energy release rate reaches a critical value gIC, which defines a failure criterion for a brittle, flaw-containing component. The material property qIc,is variously referred to as the toughness, the fracture energy, or the work of fracture. gICis determined by the rate at which the stored strain energy is relaxed when a crack advances. The critical ‘strain energy release rate’ per unit area of crack advance has units of N m-’ or J mP2.The lower bound for the work of fracture must , two new surfaces are created by the advancing be twice the surface energy, T ~ since crack. The actual value of the fracture energy will always be much higher than that predicted from the equilibrium surface energy, since fracture is irreversible (and approximately adiabatic) and additional energy dissipating mechanisms are active: localized friction between microfacets, mechanical interlocking of grains which bridge the crack, and some possible plasticity at the crack root. An alternative stress analysis approach, is based on the local stress field near a crack tip, Fig. 5 . The solution of the boundary value problem for a semi-infinite, linear-elastic cracked body was found by Williams [48], and yields the Williams’ expansion of the stress field in a cracked body. The first term determines the local stress field in the vicinity of the crack tip:

where r is the distance from the crack tip, 6’ is the angle (Fig. 5), and M denotes the mode of deformation. The parameter Kis the stress intensity factor (SIF), which is a function of geometry and loading only, and not a material property. The nature of the near-stress field is that of square root singularity in distance from the crack tip [9,10]. K is the amplitude of the stress field singularity near the crack tip. As the external load increases, the stress intensity approaches the fracture toughness, KIc , and crack propagation is initiated. This alternative failure criterion

78


/”

X

crack tip Figure 5. Stresses at the vicinity of a crack tip.

becomes, for mode I: KI = KIc. The fracture toughness of a material K I c characterizes the maximum stress field developed in the vicinity of a crack, and has the dimensions of MPa m1/2. Fracture toughness is a material property and measures the resistance to crack propagation in the absence of generalized plastic yielding [49]. For an ideally brittle material, the fracture toughness is independent of crack length, and crack initiation under static loading is followed by crack propagation. In many hard materials the fracture toughness increases with crack length, corresponding to an increased resistance to crack propagation as the crack advances. In such cases crack initiation is followed by a period of stable crack growth, with increasing stress intensity factor. This is ultimately replaced by unstable crack propagation as the toughness approaches the steady-state value of KIc characteristic of long cracks. This behavior is summarized by the crack propagation resistance or R-curve (resistance curve) [9,12]. The steady-state value of the toughness is significantly influenced by test piece geometry: the initial ratio of crack length to specimen width and the length of the uncracked ligament in the cross section. Most static failures in brittle materials approximate to Mode I loading conditions, while cracks that initiate in other in-plane loading modes, tend to propagate in Mode I [49]. The Mode I crack is the most efficient relaxation path for the strain energy stored in an elastic body. Plastic deformation can be ignored, since plasticity is both limited and highly localized in hard materials. The fracture toughness of hard materials generally ranges between 2 and 10 Mpa m1l2 [ 1,2,9], although higher (and often invalid) values are sometimes reported, dependent on the method of testing. This is an order of magnitude smaller than the toughness of most engineering alloys [l, 2,101. It is the limited fracture toughness of hard materials which limits their structural applications. The fracture toughness determines the critical amplitude of the stress singularity at the crack tip. For an ideal brittle continuum, provided the crack length, a, is


79

known, there is an exact relationship between the fracture toughness and the remote stress:

where p is a function of the geometry of the component or test piece. Geometrical functions for P have been published in the literature, most calculated by finite element analysis [50-541. Irwin [46] derived the relationship between the strain energy release rate and the stress intensity factor from the stress field at the crack tip. For a pure mode I deformation:

An analogous relationship holds for the fracture toughness and the fracture energy:

Excellent texts on fracture mechanics are available [9-111.Related topics are failure criteria for the onset of cracking in laminated systems [55],dynamic fracture mechanics [56],and nonlinear fracture mechanics [9,1I].

3.2.4 Hardness Hardness is commonly defined as the resistance of a material to static penetration by a harder material [3].The qualifier, ‘static’, excludes dynamic effects associated with erosion or ballistic penetration. Hardness can be quantified by determining the depth of penetration of a hard indentor, usually diamond, loaded perpendicular to a planar surface of the material under test [57].The penetration depth is related to the contact area over which the load is transferred between the indentor and the sample. The ‘hardness’ is equal to the load divided by the area of plastic contact. There is some ambiguity in ignoring recoverable, elastic penetration, but, apart from elastomeric materials, the effects of elastic recovery during unloading are usually small. The fixed load, P , applied to the indentor and the known indentor geometry are used to calculate the hardness from the contact area, A , between the indentor and the sample: H

P

--, -

A

(20)

where H , is an average applied compressive stress, generally measured in GPa. In polycrystalline solids, plastic deformation beneath the indentor is associated with dislocation multiplication and glide [58].In metallic solids the critical flow stress or yield stress of the material, oY,governs the hardness [59].There is therefore a

80


direct relationship between the hardness and the plastic flow stress, often expressed as:

H, = Cgy, (21) where C M 3 for a perfectly elastic, plastic solid [60]. The large proportionality factor reflects the elastic constraint exerted on the plastic zone beneath the indentor by the surrounding elastic material. No such constraint exists in the unidirectional, tensile or compressive tests used to determine the uniaxial flow stress. When the constitutive relationship describing the mechanical response is more complicated, or for a brittle material (when the plastic response depends on the hydrostatic component of the constraint), the proportionality constant may be appreciably higher [61]. Amorphous, glassy materials posses a clearly defined hardness, related to the Young modulus [62]. The same elastic constraint inhibits cracking in the vicinity of the indentor, ensuring that some plastic flow occurs beneath the area of elastic contact and precedes crack nucleation and propagation, even in very brittle materials. From the engineering perspective, a hardness test is an ideal method for monitoring the mechanical properties of hard materials, since minimal sample preparation is required and the test can be performed on actual components using simple apparatus operated at low loads. The hardness test may also be considered ‘nondestructive’, since components can often be put into service after testing. The mechanical performance of hard materials is often ranked by their indentation hardness, which in itself accounts for both the popularity and the technological success of this simple, cost effective test. The measured hardness of any material depends on parameters associated with the test method and indentor geometry (Brinell, Vickers, Knoop, etc.), so that hardness is not an intrinsic bulk property, comparable to elastic modulus, yield strength or fracture toughness. In general, the measured hardness varies with the applied load and the indentor shape and dimensions, but also with the microstructure and prior history of the material, as well as with the loading rate, the environment and the test temperature [3,63-671. In crystalline materials the resistance to dislocation glide determines the hardness, and glide is activated by the shear component of the compressive stress generated beneath the indentor. The maximum strain beneath an indentor (elastic and plastic) may be some two orders of magnitude greater that the tensile strain to failure of a brittle solid [68]. Such conditions are seldom encountered by hard materials operating in engineering systems. The strain rate in an indentation test is also high when compared to a static tensile test, since the velocity of the indentor is much greater than the usual crosshead speed in tensile testing. A high strain rate increases the flow stress ((~y),the effect increasing with indentation load and rate. By contrast, at high homologous temperatures thermally activated creep is possible, reducing hardness with increasing temperature and decreasing loading rate [69]. Indentation testing dominates the determination of hardness, but some alternative tests are available: The scratch test ranks materials in order of their ability to scratch materials lower on the scale [70], and assigns a ‘Mho’ hardness (1822) to each material, with diamond having the highest value of 10 on the Mho scale.

3.3 Mechanical Testing of Hard Materials

81

Erosion testing is of considerable importance, especially for components subjected to particulate bombardment. Dust and sand particles, but also rain, are all sources of damage in supersonic aircraft and jet engines. Erosion is also common in pumping systems required to transfer slurries and other solid-phase dispersions. The rebound test (the ‘bounce’ of a steel ball dropped onto the surface) is of importance in testing the resilience of elastomeric components, while damping losses, assessed by measuring the decay of high frequency vibrations, are also used, primarily in acoustic and vibrational damping systems.

3.3 Mechanical Testing of Hard Materials 3.3.1 Elastic Modulus It is possible to determine the tensile (Young) modulus by direct measurement of stress and strain in unidirectional tensile or compressive loading. The average stresses are monitored by a load cell, while the average strains are read from a strain-gauge or extensometer. The elastic modulus is given by Hooke’s Law, E = c/E.This method requires both a suitable test specimen and a uniaxial testing machine that will guarantee unidirectional loading: Accurately machined ‘dogbone’ configurations are mandatory for tensile specimens, precision cylinders for compressive tests. A measurement of the velocity of sound in the material, using an ultrasonic transducer, is less prone to error. If the density p and longitudinal sound velocity vL are known, then the elastic modulus is given by: It is also possible to determine the lowest resonant frequency of a rectangular beam statically loaded by a mass M in three-point bending. The tensile modulus is then given by: 16 rM13f2 3 d4 ’ where I is the span of the beam, d the height and f the resonant frequency. E=-

(23)

3.3.2 Fracture Strength Since the strength of brittle materials is stochastically distributed, a large number of tests is required to characterize the mechanical properties accurately. The two most important properties are the median strength and Weibull Modulus [34,37], and the minimum number of (valid) test results needed to meet most engineering requirements is 20. When the results of N tests of the fracture stress c are ranked in ascending order, then the best statistical estimate of the failure probability, F, for each result is given

82

3 Mechanical Properties and their Relution to Microstructure

Inert strength, u,(MPa) Figure 6. Typical ln(l/(l - F)) against In0 plots for three and four point bending bars. Reproduced with permission of . . . . . . . . . . (Ref. 9).

by (i + 0.3)/(N - 0.4), i being the rank order of the ith test result. The Weibull modulus m is estimated from the slope of the graph of In(ln[l/(l - F ) ] ) vs. lnc, as shown for various geometries and processing routes, Fig. 6 . A detailed discussion of Weibull strength analysis is given in [37]. The determination of m should always be made by graphical plotting of the test results, since alternative failure mechanisms can give rise to different median strengths and Weibull moduli. These may be associated with the probability of failure initiation at a surface as opposed to a bulk defect. Surface flaws commonly have a wider size distribution, giving a lower Weibull modulus, while bulk flaws, with a narrower distribution, correspond to a higher value of m [71-741. In consequence, the slope of the Weibull plot changes, increasing for the higher strength values, dominated by the size and distribution of the bulk defects. A full understanding of the In In[I / ( 1 - F ) ] vs. In c curve requires fractographic analysis of the failed samples and correlation of the observed failure origins (surface cracks, bulk porosity, hard aggregates, inclusions) with the measured strength. Various internationally recognized standards exist for the determination of flexure strength (3PB or 4PB):ASTM C1161 and MIL STD 1942 (USA), DIN 51-1 10-1 (Germany), B41-104 (France), JIS R 1601 (Japan), GB 8489 (China). The specimens are prismatic rectangular bars. Commonly specified outer spans are 20, 40, or 80 mm, with corresponding 4PB inner spans of 10, 20, and 40 mm, respectively. The specified cross section (height x width, all dimensions in mm) varies from 1.5 x 2 (American) and 2 x 4 (French) for small-span specimens, to 3 x 4 for the medium span, and 6 x 8 for the large span. Other test samples and loading geometries are sometimes required for specific components or applications. The biaxial strength of thin plate is commonly


83

evaluated using a ring-on-ring test or an alternative mounting suitable for biaxial flexure (a hardened steel ball with an articulated tripod support, for example).

3.3.3 Fracture Toughness A bewildering number of specimen geometries and experimental methods has been proposed (and adopted) to determine the fracture toughness of brittle materials, and values for fracture toughness of hard materials are routinely determined, published and compared. However, at the time of writing, no universally agreed standard for fracture toughness exists most internationally accepted methods specify a precracked specimen loaded to fracture under three point bending (for example ASTM E-399) [75]. The fracture toughness is then calculated using an appropriate function from the maximum recorded load and the specimen geometry. The generation of a reproducible, sharp, and accurately positioned precrack in a hard material is a technical problem which partially accounts for the variability of reported fracture toughness results. While an individual laboratory may obtain a reproducible result, different laboratories, employing nominally the same test method for the same material, may find average values for fracture toughness that lie outside the scatterband determined by their colleagues elsewhere [7&78]. Any fracture toughness test must approximate the square root singularity of the stress field calculated for an atomically sharp crack. Deviations from the square root singularity will reduce the local stress intensity and increase the apparent value of fracture toughness. Unfortunately, since engineering applications for hard materials usually specify a minimum toughness, it is not uncommon to welcome misleadingly high values of toughness. Any test procedure (or test laboratory!) consistently giving unusually high values for fracture toughness should be viewed with suspicion, and it is the lowest measured value of the fracture toughness which should be used to estimate maximum allowed loads and specify the requirements for defect detection in the nondestructive evaluation of a component. Unstable crack propagation often follows immediately on crack initiation in brittle materials, and it is difficult to generate a straight, sharp precrack. It is therefore common practice to cut a thin (typically about 100pm) notch, using a diamond cutting disk [79], to act as a crack initiator. The presence of a square root stress singularity at the notch tip is unlikely, although microcracking at the root of the notch may approximate the square root singularity [SO], so the fracture toughness determined from notched specimens is often an overestimate, while the uncertainty in initial crack length, associated with the size of the microcracked damage zone, introduces an additional error. A precrack can be generated from microindentation cracks placed near the center of the test beam prior to loading [81]. The indentation crack tip front is elliptical, and the residual stress field associated with localized plastic flow beneath the indentor has a significant effect on the square root stress singularity. A more successful method for preparing a precracked beam is by applying a compressive load to a notched or indented specimen held in a stiff fixture. Controlled crack growth from the original defect gives the required sharp and straight crack front, and the

84


fracture toughness results obtained from such precracks have proved to be highly reproducible [82-841. Controlled precracking of a thin, notched specimen can be achieved by gluing the sample to the side face of a ductile metal beam loaded in 3-point bending. Crack arrest in the thin, brittle specimen occurs at the point where the stress in the thick supporting beam becomes compressive [85]. Precracking of ceramic specimens by cyclic fatigue of notched specimens at a slowly decreasing stress intensity factor has also been reported [86]. Most applications of these test methods for precracked-beams loaded by 3PB evaluate the plane-strain fracture toughness using the methodology summarized in the ASTM standard E-399 [75]. The precise measurement of crack length is also difficult. Partial crack closure after precracking, upon removal of the load, may preclude accurate crack length measurement. The use of die penetrant as a visual crack marker is helpful but inaccurate. Optical inspection at low magnifications is usually adequate, providing the specimen finish is good. The fracture toughness of materials generally quoted in the literature is the static or quasi-static fracture toughness, and is a poor guide to dynamic behavior, slow crack growth, or mechanical fatigue strength. Negotiations for an international fracture toughness standard are well advanced within the framework of ASTM and VAMAS (Versailles Agreement on Materials and Standards) [87]. The test geometry which is expected to become an ASTM standard is the chevron-notched three-point bend beam. The chevron notch is cut with a diamond saw, and the crack grows at low loads from the tip of the notch. Stable crack growth continues until the crack front reaches the full cross-section of the beam, when growth becomes unstable at the point of maximum load, the critical load for fracture toughness evaluation. The fracture toughness determined from a precracked beam may give values which differ significantly from those determined from a chevron notched specimen. This difference is usually materialdependent and it is not possible to predict which test will give the lower (more dependable) value. Alternative geometries and test methods to determine the fracture toughness of ceramics and hard materials have been developed over the last decades: Herzian indentation [88], the double torsion test [89], and compact tension tests [90-921.

3.3.4 Hardness Microhardness tests in metals are performed under very low load [3], 0.01-2N, which nevertheless activates dislocation glide. The higher values of microhardness characteristic of hard materials require higher loads, typically 0.05-0.5 N, which do not usually initiate microcracking (indentation cracking). At still higher loads, 2-100 N, microcrack initiation and propagation is expected in brittle materials, resulting in higher values for the measured hardness, since microcracking increases the compliance, and hence the elastic component of penetration, which is relaxed (apart from frictional effects) when the load is removed. In addition to plastic deformation, other viscous flow mechanisms may operate under high loads, such as densification in both glasses and ceramics containing residual porosity [93]. It


85

Figure 7. A common load-hardness relationship. Reproduced with permission of . . . . . . . . . . (3).

follows that hardness tests should be performed over a range of load to ensure that a load-independent value of this property can be defined [66],Fig. 7. In hard materials a high-angle, pyramidal indentor, generates high residual tensile stresses on load removal which may exceed the tensile strength, resulting in microcrack initiation and propagation. Valid hardness measurements require that the indentation be either very much less than the grain size, in order to characterize the microhardness of individual grains, or else appreciably greater than the grain size, to ensure that it is the hardness of the polycrystal that is determined. Typical indent sizes, using standard microhardness testing systems, are in the range 3-50pm, and up to 200pm in the low load regime. A good optical microscope has a resolution of the order of 0.5pm and should be capable of achieving 10% accuracy providing several measurements are made to monitor the reproducibility. The methods which have been developed to determine hardness vary in indentor geometry and definition of the contact area. In Brinell’s method, a hard sphere of diameter D is indented into the softer sample, and the Brine11 hardness is calculated from the diameter of the circular indent which is left in the surface after removing the load:

where d is the diameter of the impression. In the Rockwell method the depth of penetration of the indentor, usually but not always a diamond cone, is used to calculate the Rockwell hardness number on one of several scales (Rockwell A, B, C , or D), depending on the applied load and the geometry of the indentor. The Meyer hardness number is derived from the projected area of the residual

86


impression:

P

HM =4-. ?rd2

In the Knoop method an elongated diamond indentor with a specified ratio of the two diagonals is used. Measurements taken from the long diagonal of the indent can be used to assess hardness anisotropy by rotating the indentor about the loading axis between measurements. The Berkovitz microhardness indentor employs a triangular pyramid which avoids variations in hardness associated with anisotropy in the plane of the surface and ensures that the indentor tip is truly a point (inaccuracy in cutting a square pyramid results in a chisel edge at the apex). In the Vickers method a square diamond pyramid with a standard apex angle of 136" is used and the contact area is determined from the mean length of the two diagonals of the indent, 2a, which is then translated into a Vickers hardness number:

P

Hv = 0.464-.

a2

(26)

It is the Vickers test which is most commonly used to specify the hardness of hard materials, with results often quoted in kgmm-2, but more correctly in SI units, GPa. The repeated measurement of the size of indents, and the interpretation of indent geometry for the purposes of calculation, may be tedious, and operator bias is almost unavoidable. The edge of the impression is not always well defined, and misleading edge effects may be associated with anisotropic plasticity or plastic recovery. Faceted and elongated grains, or other microstructural features, together with the limitations of contrast and resolution in the optical microscope, complicate the interpretation, while the shape of the indent may differ in different materials: socalled 'pin-cushion' or 'barreled' indents, associated with different constitutive relations and frictional shear on the faces of the indentor in contact with the plastic zone [3]. Mismeasurement of indent size is a major source of scatter in the experimental data and the relative errors in the results of different operators.

3.3.5 Indentation Toughness Palmqvist [94,95] has used the Vickers diamond pyramid indentor to initiate and propagate cracks in hard materials, and Evans, Lawn and their coworkers [96-1001 have related indentation cracking to fracture toughness by assuming a simplified stress field. Since then, considerable effort has been made to characterize fracture toughness using Vickers indentation cracking, Fig. 8. The primary advantage seen for such a test is the limited volume of material needed and the inexpensive test apparatus. Furthermore, such a test can be performed on an actual component, with no need to machine a special test piece. There are several problems: the small volume element of material under load, the variety of possible cracking mechanisms, analysis of the stress field needed to define a calibration function, variations in the shape of the crack front, and accurate measurement of crack length. These problems reduce the significance of indentation toughness results, although this remains a


81

Figure 8. Typical plastic deformation and cracking after indentation toughness test. Reproduced with permission of . . . . . . . . . . (3).

useful test for monitoring damage tolerance and the resistance to short crack propagation. The measured values should not be interpreted as ‘fracture toughness’. Four microcrack morphologies may develop in a brittle solid after formation of the plastic zone beneath the indent: median cracks, Fig. 9a, radial cracks (a halfpenny shaped elliptical crack Fig. 9b, much larger than the median crack), lateral cracks (in a plane normal to the median and radial cracks), Fig. 9c, and Palmqvist

i

I _ 2c‘

I

\-

I I

!

a

*I I

b

Median mqvlst radial

vent

I C

d

Figure 9. Cracks formed during an indentation toughness test: (a) median cracks; (b) radial cracks; (c) lateral cracks; and (d) Palmqvist cracks. Reproduced with permission of . . . . . . . . . . (3).

88


radial cracks (originating near the edge of plastic zone beneath the indent), Fig. 9d. In addition, Herzian cone cracks may develop, concentric with but outside the plastic indent left by a hard sphere or cylindrical punch. The median crack is a single, penny-shaped crack nucleated beneath the apex of the plastic zone created by the indentor. The diameter of the median crack is comparable with the indent size, and a median crack is not visible in a polycrystalline, opaque material. The driving force for nucleation of the median crack is the elastic tensile stress developed normal to the indentation direction at the elastic-plastic boundary when the external load is relaxed. Nucleation of a median crack depends on the presence of a suitable flaw. Once nucleated, the median crack will propagate spontaneously to a stable flaw size. The critical flaw size for growth is:

where 0 is a constant which relates the tensile strength to the hardness. After nucleation, the median crack propagates away from the elastic-plastic zone boundary. Stable propagation occurs on increasing the external load, and the ratio of the diameter of the median crack, D,to the indent size is given by:

D

-m*. a

When the diameter of the median crack exceeds the indent diameter, it can ‘pop’ out to the surface of the specimen to become a half-penny shaped, radial crack, which is visible on the surface and whose length can be measured. A polished surface may be necessary to identify accurately the position of the crack tip, while annealing may be required to eliminate compressive surface stresses due to polishing. The exact shape of the radial crack is unclear, but in a transparent specimen the ‘popin’ of such cracks may observed on unloading the indentor. Multiple radial cracks may occur, especially for indentors with a small pyramid angle. High frictional forces may prevent sliding at the indentor contact surface and cause large shear deformation adjacent to the indentor facets. Upon unloading, the compressive stress normal to the surface reverses sign to become tensile, due to the previous irreversible plastic deformation beneath the indentor. If a suitable flaw is present, a lateral vent crack may be generated. This shallow crack can ‘pop-in’ parallel to the surface and terminate subsequently at the surface to remove a flake of material. The lateral crack is larger the plastic zone, and may be larger that the radial cracks. Finally, Palmqvist cracks are short, shallow cracks, initiated at the end of the indent diagonals at relatively low loads, Fig. 9d. The plane of a Palmqvist crack coincides with the plane of the radial cracks, and there is some evidence that the Palmqvist cracks may, in some ceramics, initiate before the radial cracks, at the same time as the median crack. Given the variety of cracking mechanisms, both the stress field under load and the residual stress field are clearly very complex [loll. Difficulties of stress analysis have led to some inconsistency in the formulae derived to calculate indentation toughness. Rather than review the large number of published equations [102-1071, a

3.3 Mechanical Testing of’ Hard Materials

89

basic parametric formulation may be given in the form of a nondimensional fracture toughness [66]:

where Hv is the Vickers hardness, a, c, and R,, are the half-length of the diagonal of the indent, the radius of the radial crack, and the radius of the plastic zone, respectively, q5 the constraint factor, v and /-L are Poisson’s ratio and the coefficient of friction between the indentor and the sample. Alternative’formulations exist! The indent size, the surface crack morphology, and the extent of subsurface damage are all influenced by the crystal structure, chemical composition, and microstructural morphology of the material.

3.3.6 Erosion, Wear and Scratch Tests 3.3.6.1 Erosion

Solid particle erosion is a major cause of damage in pipes, gas turbines, and compressor components [108]. Erosion may be defined as the resistance of a material to surface damage due to impinging particles of high kinetic energy [109]. Erosion is a dynamic phenomenon, and the damage may be mechanical (loss of bend strength due to surface microcracking) or physical (loss of transparency due to roughening of a surface). Rain erosion, in which the impinging ‘particles’ are rain drops arriving at sonic or supersonic velocities, may be as damaging as solid particle erosion, and the loss of visibility due to surface roughening of a pilot’s cockpit canopy can be as dangerous as any loss of strength. Erosion damage is commonly measured as the change in a given property as a function of the number of impacts or the total momentum transferred. The property of interest may be the weight loss of an eroded sample, its residual strength, or some other measurable property (residual transmittance, for example) [1lo]. In ductile materials the impinging particles gouge out fragments by micromachining of the surface, and the maximum weight loss is observed at an impingement angle of about 20” [lll]. Erosion of hard materials is primarily associated with the generation of lateral cracks beneath the point of impact, and the rate of damage is a maximum at normal impact. The erosion rate depends on the fracture toughness or damage tolerance of the target material and the hardness of the impinging particles [112]. For an ideally brittle material, the volume of material lost in a single impact, Vi, can be estimated from the radius of the lateral crack, ci, and its depth, ai [9]:

where w is a ‘wear coefficient’. If additional energy absorbing mechanisms exist, and the material exhibits R-curve behavior, then the value of the fracture toughness should be that appropriate to short cracks of length ci. The above relation suggests

90


that the influence of fracture toughness should exceed the effect of hardness in improving erosion resistance. 3.3.6.2 Wear Assessment

A good example of extreme wear is the attack that occurs at the top face of a cutting tool when a fully-plastic, machined chip is sheared adiabatically across the tool face. Adhesive and abrasive wear are major mechanisms [113], as is chemical wear, especially at elevated temperatures. Adhesive wear occurs when two surfaces in contact bond during shear, and is associated with large compressive stresses generated at points of contact formed during sliding under pressure [114]. If the local compressive stresses are sufficient, subsurface microcracks propagate and fragment near-surface material to form wear particles. This occurs when the toughness of the bulk material is lower than that of the interface under shear. For a sufficiently low interfacial toughness, the debond 'crack' will propagate at the interface without wear. The distribution of defect-like, near-surface voids, impurities, or surface microcracks determines the extent of the wear zone and the size of the wear particles. Abrasive wear takes place when two hard materials are brought into contact without bonding, when the primary damage is associated with mechanical interlocking (shear of asperities) and the generation of Herzian cone-cracks surrounding the contact points [115, 1161. 3.3.6.3 Scratch Testing

In a scratch test, a diamond or WC, conical or pyramidal indentor is sheared across a specimen under a controlled normal load while monitoring the tangential force [117]. The adhesive bond and the tangential plowing force determine the total work performed by 'frictional' forces. The coefficient of friction is determined from the ratio of the tangential to the normal load. The results of a scratch test depend on the indentor material and geometry, the normal load, the indentor velocity, the surface roughness, the lubricant (if any), and the homogeneity and isotropy of the sample [118]. Crack damage in a scratch test approximates that of a Herzian crack, and results from the tensile stress generated near the trailing edge of the contact. Cracks are formed when the external load exceeds a critical value. The trace of the scratch contains a set of cracks whose density is a function of the load, the surface conditions and the indentor velocity. Tensile stresses that generate cracks are enhanced by frictional forces between the indentor and the sample in the contact area [119]. The presence of a lubricant increases the threshold load for cracking [120]. A chemically active liquid, such as water, may increase the friction load and decrease the threshold for cracking. The fracture surface energy of glass may be reduced by the presence of scratches, which decrease the threshold load for cracking. The shape of the indentor is important. No damage was observed in soda-lime glass scratched by spherical indentor at loads up to 50 g. Damage increased with load for a 90" square pyramidal indentor, but only initiated at 15g load when the pyramidal indentor angle was 136". The width and the depth of the scratch decreased with increasing indentor velocity.

3.4 Microstructural Parameters and Mechanical Properties

91

Scratch tests have been used to evaluate the adhesion of thin films to a ceramic substrate.

3.4 Microstructural Parameters and Mechanical Properties Since the failure of hard, brittle materials is dominated by fracture originating from sources of stress concentration, their performance in service is dominated by the presence of processing defects as much as by the microstructure [121]. It is to distinguish between the two, even though ambiguity remains when the scale of a ‘processing defect’ approaches that of the microstructure. Processing defects resolvable by available methods of nondestructive evaluation (NDE) and quality control present no difficulties. These include porosity, microcracking, and those‘inclusions whose density and elastic properties enable detection. Ultrasonic testing and X-ray radiography are capable of a resolution well below 100 pm, providing the component cross-section is not too large. Both techniques can be automated and have tomographic capability, that is the defects can be located in three dimensions with reasonable accuracy, not just in projection [122-1241. Below the resolution limit of available NDE procedures, but for dimensions which exceed the scale of the microstructure, it is still useful to talk of ‘processing defects’, even though they may only be detectable by fractographic analysis, which traces the origin of failure to an abnormally large grain, a hard aggregate, a brittle inclusion, or other defect. The ‘undetectable’ processing defects are a function of processing history, and constitute the major obstacle to the improvement of performance through improvements in the quality of raw materials for hard materials. Reducing powder particle size below 0.1 pm is of little help if the resultant finer grain size is accompanied by particle aggregation and residual porosity. Similarly, the price of improved powder purity is only worth paying if contamination by milling media wear can be avoided during processing. It is helpful to regard all ‘microstructural defects’ as features of the microstructure, and avoid confusing them with processing defects, even though their presence is clearly a result of processing history. Subgranular residual porosity and pockets of glassy phase at grain boundary junctions, intergranular and transgranular microcracks, and small second phase particles associated with impurities or precipitation, are all microstructural defects [125, 1261. The effort to relate mechanical properties to the microstructure of engineering materials has a long and respectable history, but with many misconceptions. Before the analysis of crystal structure by X-ray diffraction became generally accepted, it was thought that the mechanical fatigue of steel was due to ‘crystallization’,while ductility was attributed to a ‘noncrystalline’microstructure, and t h s despite the clear metallographic evidence for a granular structure. It is useful to distinguish three aspects of microstructure in hard materials, all of which are related to mechanical properties. The first is the physical nature of the features present, including the composition and crystal structure of the solid

92


phases, and the presence of porosity, microcracking, and impurity phases [121,127]. The second aspect is the microstructural scale of the features, including their size, size distribution, volume fraction, and distribution in space (microstructural homogeneity). The third factor is the orientation of microstructural features, both their morphological anisotropy (alignment of platelets, whiskers, and fibers), and their crystallographic anisotropy (preferred orientation of crystalline phases with respect to the component geometry, commonly termed ‘texture’). As in metals, the scale of the microstructure is related to the hardness (and, through the hardness, to the plastic flow stress), with the Petch law applying well into the submicron range of grain size. Increasing the hardness for a given composition is readily achieved by reducing grain size, typically by reducing powder particle size for the raw material together with the sintering temperature used to densify (sinter) the component. Unfortunately, processing to full density becomes more difficult as the powder particle size decreases, while any residual porosity reduces the hardness. Grain growth during sintering is limited by appropriate dopant (trace) additions. Fine zirconia particles in alumina pin grain boundaries [ 1281, while MgO additions inhibit grain boundary migration without affecting densification [1291. Commercial aluminas with submicron grain sizes and near-theoretical densities are available with a hardness of the order of 20 GPa (compare the tensile modulus of 380 GPa). Fracture toughness has been improved by providing microstructural toughening mechanisms, discussed more fully below. Direct melt oxidation (DMO) leads to alumina products which contain a network of some 15vol-% residual metal [130,131]. The metal fibers intersect a propagating crack front and dissipate energy by deforming plastically, reducing the stress intensity factor at the crack tip. Platelets or whiskers in the microstructure, present as additions to the powder mix or, preferably, grown in situ as a result of trace additions, also bridge the propagating crack front, supporting some of the load and generating frictional forces as they detach from the matrix. Tetragonal zirconia grains or particles transform to the lower density monoclinic phase in the stress field at a crack tip, generating crack closure forces which lead to ‘transformation toughening’. The high temperature strength and creep properties of hard materials are sensitive to the presence of glassy phases at grain boundaries [132], which promote viscous sliding at the boundaries, lead to enhanced stress concentration at the boundary junctions and nucleate cavities. The cavities grow under the hydrostatic tensile component of applied load, coalescing to form ductile microcracks which initiate the third stage of creep failure (see below), leading to stress rupture. In some grades of silicon nitride a glassy oxynitride phase responsible for the excellent sinterability of the material, can be crystallized after sintering, greatly improving the creep resistance. Since creep failure is commonly initiated by cavitation, which results in a volume increase, creep in hard materials depends on the sign of the uniaxial load. Creep rates in compression are generally lower than those in tension [ 1331. The viscous flow of the glassy phase in the grain boundary channels, a percolation process, commonly determines the overall strain history. The availability of nanopowders has led to the development of submicrometer, fine-grained ceramics, which exhibit improved mechanical properties. Many

3.4 Microstructural Parameters and Mechanical Properties

93

glass-free oxide ceramics with a submicron grain size are superplastic under conditions of moderate temperature (= 0.5 T,) and low strain rate to lop2s-’) [134-1 361. The exact mechanism of superplastic deformation is controversial, but it is clear that the grain boundaries play an important role, and that diffusion is a dominant contributor to plastic deformation. Tensile strains to failure of 1001000% are routinely observed [ 134-1 391, and should enable complex shapes to be formed from dense ceramic preforms. The strain rates required for ceramic superplasticity are too slow for most forming operations. Through trial and error, it has been found that grain sizes of less than 1 pm are required for superplasticity, and the finer the grain size, the larger the superplastic strain to failure of the ceramic [140,141]. A second advantage offered by a fine grain size is an increase in the strain rates observed during superplastic deformation. The effect has been rationalized using the following phenomenological relation:

where C is the strain rate, A a material constant, k is Boltzmann’s constant, b the Burgers’ vector, d the grain size, 0 an applied stress, E the Young modulus, Do a pre-exponential factor for diffusion, Q the activation energy for superplastic flow (usually Qsuperplastic M Qdiffusion), and T is the temperature. The constant n is the stress exponent, while p is the grain size exponent. The value of p is typically between 2 and 3, and a 10-fold reduction in grain size should lead to a factor of 100 to 1000-fold increase in strain rate at constant applied stress. Conversely, the same grain size reduction should lead to a decrease in flow stress or lower deformation temperature, at the same strain rate. The latter phenomenon may explain the room temperature plastic deformation in nanostructured Ti02, reported by Karch, Birringer, and Gleiter [142], who were the first to report significant room-temperature ductility (M100% plastic deformation) in nanocrystalline ncTi02. By contrast, a compression test on a TiOz single crystal resulted in brittle fracture. Recent experiments [ 1431 have shown that nanocrystalline oxides may be plastically deformed under relatively low tensile stresses and at temperatures below half the melting point. The Hall-Petch relation relates the flow stress to the grain size: 0

= 00

+ k(d)-”2,

(32)

where 0 is the flow stress, go is a lattice friction stress, k a (positive) material constant, and d is the grain size. This equation implies that any decrease in grain size should increase the flow strength. The relation between hardness and grain size is expected to be given by a similar equation, although this may not be true of nanocrystalline materials, for which abnormal Hall-Petch behavior has been reported for both nanocrystalline metals [144,145] and ceramics [I431 (a decrease in grain size decreasing the hardness and increasing the ductility). Grain-boundary sliding, grain boundary diffusion, and the generation of lattice dislocations from boundary junctions have all been suggested as alternative mechanisms of plastic

94


shear, replacing dislocation glide, in nanocrystalline materials [ 146-1491. Plastic deformation in crystalline ceramics can occur by dislocation glide, climb or diffusion, usually at grain boundaries. The high Peierl’s stress and limited dislocation density in ceramics usually restrict plastic deformation mechanisms to diffusion-assisted processes. The ballistic response of hard materials is also microstructure-dependent, primarily through the hardness, although the density and elastic modulus determine the mechanical impedance, and hence the shock-wave behavior. Light-weight, stiff ceramics are effective armor modules. Boron carbide provides the best ballistic performance, although the much cheaper alumina dominates the market. Since the time scale of ballistic impact is so short, the failure processes are adiabatic, and the presence of a glassy phase may be almost as deleterious as residual porosity for ballistic performance. Adiabatic heating above the glass transition temperature allows the glassy phase to flow, leading to cavitation failure under impact, not dissimilar to that observed in creep!

3.5 Failure Mechanisms 3.5.1 Creep Behavior Hard materials and ceramics are used in load bearing applications at high temperatures and for extended times. Under these conditions the primary failure mode is often by creep and stress rupture. ‘Creep’ is time-dependent, irreversible plastic deformation under constant load or stress, terminating in time and temperature dependent failure, termed ‘stress rupture’. Ceramics, like metals, begin to creep at homologous temperatures of 0.4-0.5 T,. The melting temperature T, of a structural ceramic is usually well above 2O0O0C, and creep initiates at temperatures well beyond the operating temperatures of heat resistant superalloys. The resistance of ceramics to environmental attack is substantially better than the superalloys, and certainly far superior to the refractory metals (molybdenum and tungsten, for example). Two engineering parameters are commonly used to rank the behavior of materials under creep conditions (constant temperature and applied uniaxial stress): the minimum strain rate and the time to rupture. The first parameter is related to the useful life of components susceptible to shape-change in service (in heat engines, for example), while the second estimates the time-dependent failure probability. The deformation and damage mechanisms in creep of ceramics and hard materials are similar to those in metals [150,151]. Under normal loading conditions (in the absence of severe elastic constraint) ceramics fracture at room temperature before any significant plastic flow. Dislocation glide in ionically bonded ceramics is complicated by the presence of both anions and cations, which create electrostatic (Coulombic) barriers to shear. As in metals, three creep regimes have been identified. The initial high strain-rate, observed on applying the load, decreases rapidly

3.5 Failure Mechanisms

95

as mobile slip dislocations are exhausted and boundary shear leads to grain interlocking at the triple junctions. This primary, transient creep terminates at the minimum creep rate, which may then persist for most of the creep life, corresponding to the second, steady state, creep regime. When an extended period of steady-state creep is present, this rate is denoted i. and can often be described by the relation:

i = A 6 exp(-QIRT), (33) where A and n are material parameters, and Q is an ‘activation energy’ for creep [152]. The constant creep rate in the steady-state regime is usually associated with strong work-hardening, arising from the blocking of secondary slip dislocations by dislocation barriers on the primary slip system. The stress exponents close to 5 are due to dislocation climb and fully ductile behavior, and a stress exponent close to 3 is due to dislocation climb [151, 1531. The mechanisms responsible for fracture in structural ceramics at elevated temperatures have been reviewed [ 1541. Sensitivity to flaws or microstructural inhomogeneities which nucleate microcracks are among the failure mechanisms. The flaws which control failure under creep conditions are different from those responsible for fast fracture at room temperature. A common feature is the development of cracks through gradual damage accumulation, depend on the microstructure. The role of cracks in the deformation and fracture behavior of polycrystalline structural ceramics have been reviewed [1551.

3.5.2 Mechanical Fatigue Mechanical fatigue under cyclic loading is a common cause of failure in engineering systems. Fatigue cracks usually initiate at sources of stress concentration: re-entrant angles in component design, hard inclusions and aggregates, residual porosity, or shrinkage cracks. However, the parameter controlling crack advance in metals and alloys is the stress intensity factor at the fatigue crack tip after nucleation, and observed values of the SIF for fatigue crack propagation are significantly below the fracture toughness ( K I c )of the material. In these ductile materials dislocation mobility at the crack tip during loading and unloading in the fatigue stress cycle determines the rate of crack growth. The semiempirical Paris’ law [lo] describes this rate of crack advance as a function of the number of load cycles, expressed in terms of the SIF amplitude: da -= C ( A K ) ” , (34) dN where C and IZ are material parameters. The predictive capability of this formula permits the inclusion of allowed crack length or maximum working life in the design of a component subjected to cyclic loading in service. The Paris exponent n is in the range 2 4 for most metals and alloys, but is often much larger in ceramics. This implies that crack advance in ceramics after fatigue crack nucleation is much more rapid and occurs over a narrow range of

96

3 Mechanical Properties and their Relation to Microstructure FRONTAL-WAKE

BRIDGED-INTERFACE

.--.

Figure 10. Schematic presentation of toughening mechanisms. Frontal-wake mechanisms: (a) dislocation glide, (b) microcracking, (c) phase transformation, (d) ductile second phase. Bridging mechanisms: (e) grains, (f) fibers, (g) whiskers, (h) ductile second phase. Reproduced with permission of . . .. . . . . . . (9).

AK before terminating in catastrophic failure at K = KIc. For example, crack growth in Sic was found to depend on K,,, rather than AK, and was very dependent on the stress ratio R [S6]. The damage mechanisms associated with crack growth under cyclic loading were essentially identical to those in static loading [156-1581. Fatigue failure in brittle solids is often assumed to be absent, since no significant plastic deformation occurs at the loaded crack tip. However, when toughening and energy absorbing mechanisms exist [9,12], progressive changes in the stress field at the crack tip occur during cyclic loading. Frontal, or crack-wake toughening mechanisms include localized phase transformations (typical of zirconia) [ 1591, and constrained plastic deformation of a ductile second phase (for example, metal particles in a ceramic matrix), Fig. 1Oa-d. Extrinsic or crack-bridging mechanisms are associated with oriented fibers whiskers [ 1601, grain interlocking, and the presence of uncracked ligaments [161-1631, or bridging of the advancing crack by ductile particles [164], Fig. 1O+h. These all reduce the SIF by shielding the crack tip, and those toughening mechanisms that improve the mechanical performance under static loading progressively increase the internal stress at the crack tip as a

3.5 Failure Mechanisms

97

result of the debris and damage generated during cyclic loading, leading to unstable crack propagation.

3.5.3 Ballistic Properties An effective armor module prevents penetration of a projectile at minimum weight and cost per unit area. Steel armor has been preferred for over a century. Aluminum and fibrous-composite armors have been developed for armored vehicles and personnel protection over the past 40 years. A thick armor plate may be required to defeat a projectile. To minimize the weight of armor, damage to the armor must be minimized and damage to the projectile maximized. Cracking and erosion of the projectile in the first microseconds of impact will reduce the effective mass and velocity of the projectile, increase the contact area over which energy is dissipated, reduce the impact stress, and increase the volume of armor material participating in energy dissipation. Destroying the projectile is thus a first priority, and has led to the development of bilayer ceramic/metal or ceramic/fibrous composite armor. The hardness of the ceramic component fractures and erodes the projectile, but with only a limited absorption of kinetic energy due to the low fracture toughness of the ceramic. The metallic or composite backing then absorbs the kinetic energy of the projectile and ceramic fragments generated in the primary impact. The backing also reduces the bending stresses in the module, preventing premature failure of the ceramic. The efficiency of a ceramic/metal or ceramic/composite assembly in defeating a given projectile is significantly better than that of a metallic armor of the same weight/unit area. Monolithic ceramics have been evaluated extensively for their ballistic performance [15,16,165]. The failure modes observed during projectile penetration include: deflection of the projectile, erosion of projectile and target [1661, shockwave propagation and reflection [ 166,1671, and crack initiation and coalescence [168,169]. These mechanisms are affected by the rigidity and inertia of the armor module [167], the constraints introduced by the mounting system, and the design of the backing layer [167]. Each mechanisms may have a role in determining the minimum weight of armor required to defeat a given threat. The damage resulting from projectile impact on a hard, brittle target has been studied for over 70 years. Basic studies have used spherical projectiles [170-1721, planar impact configurations, or Hopkinson bar tests [ 1731. The results are generally consistent, and may be summarized in terms of three basic phenomena: cracking, crushing, and erosion. Damage to alumina tiles impacted by armor piercing rounds has been analyzed in some detail [ 174-1761. Quasi-static, drop-weight tests have been correlated with armor piercing tests, and similar damage mechanisms identified. The damage mechanisms, both with and without lateral constraint, have been described for fully supported alumina tiles and for tiles supported by a thin backing plate. Two types of damage were identified:short-range damage due to quasi-static stresses generated by the stress front propagating into the ceramic, and long-range, dynamic damage associated with multiply reflected shock waves. The quasi-static damage is characterized by a wide size

98


distribution of large fragments, while the dynamic damage is characterized by much smaller fragments and granules, down to an eroded ‘dust’ of alumina powder. The projectile is fractured and eroded during the initial stages of impact, reducing projectile mass and velocity, and the amplitude of the stress front decreases rapidly as it advances in a semi-hemisphere from the contact zone. The quasi-static damage is in the form of radial tensile cracks resulting from biaxial bending and initiates at the bottom surface of the tile, opposite the point of impact, propagating through the thickness as a semi-elliptical crack. A Herzian cone crack is subsequently initiated at the edge of the tileprojectile contact zone and propagates toward the bottom surface [177-1821. At high kinetic energies, crushing occurs within the cone envelope beneath the impact, with microcrack coalescence under the high compressive stress. Compressive cracks have been termed ‘wing’ cracks by Ashby and coworkers and identified as a primary source of damage in brittle materials subjected to compression [177,180]. Dynamic damage in the form of spa11 cracks is generated when an elastic shock wave is reflected from the tile-backing interface [176]. A high impedance, steel support plate transmits the shock-waves through the ceramic-support interface, reducing the damage in the impedance-matched ceramic. Delaying the onset of failure in a ballistic ceramic tile (by only microseconds) can promote erosion of the projectile. To achieve this, a delay in the onset of quasi-static damage is required, primarily by reducing the bending moment. Limiting dynamic damage is especially important for multiple impact performance, and is probably best addressed by optimizing the lateral dimensions of the tile in order to limit long-range damage without sacrificing overall protection.

3.6 Conclusions The design of components for applications requiring high stiffness, hardness and wear resistance under extreme conditions of load and temperature leaves little choice for materials selection outside the realm of ‘hard materials’. The spectrum of materials options available to the design engineer is by no means as broad and varied as could be wished, but a combination of advances in materials and process development, together with improved methods of inspection and testing, has led to a significant reduction in the uncertainties and risks associated with the estimation of mechanical failure probabilities and component for brittle materials in service. Coating technology remains the least understood area in this domain, and the improvements in sophistication of production technology (the development of diamond coating, for example) have not necessarily been accompanied by improved understanding of the factors that determine the performance of hard coating in service. The optimization of coating thickness, the development of graded (‘functionally graded’) coatings, the effect of substrate surface preparation prior to coating, and the compatibility of the elastic and thermal properties of substrate and coating, are all complex topics. They need to be addressed.

Ref&wms

99

References I . M. F. Ashby and D. R. H. Jones, Engineering Materials I and 2, Int. Series on Materials Science and Technology, Vol. 39, Pergamon Press, Oxford, 1988. 2. M. F. Ashby, Materids Selection in Mechanical Engineering, Pergamon Press, Oxford, 1992. 3. I. J. McColm, Ceramic Hardness, Plenum Press, New York, 1990. 4. F. S. Petit, in Coating f o r High Temperature Applicutions, E. Lang (Ed.), Applied Science Publications, Essex, 1983, pp. 1-32. 5. C. Puret and P. Pichoir, in Coating,/or Hight Tempertrture Applications, E. Lang (Ed.), Applied Science Publications, Essex, 1983, pp. 33-78. 6. D. Satas, Coating Technology Handbook, Dekker, 1991. 7. Y. Kimura, K. Okada. and Y. Enomoto, Wear, 1989, 133, 147-161. 8. 0. 0. Ajayi and K. C. Ludema, Wear, 1988, 124, 237-257. 9. B. R. Lawn, Fracture of’ Brittle Solids, 2nd edn., Cambridge Solid State Science Series, Cambridge University Press, 1993. 10. D. Broek, Elementnry Engineering Fracture Mechanics, 4th edn., Kluwer Academic Press, Dordrecht, 1986. 11. M. F. Kanninen and C. H. Popelar, Advanced Fructure Mechanics, Oxford Engineering Science Series 15, Oxford University Press, Oxford, 1985. 12. A. G. Evans, J . Am. Ceram. Soc. 1990, 73, 187-206. 13. J. Gallagher, Damage Tolerance Design Handbook, Revised edn., Metals and Ceramics Information Center, Wright-Patterson Air Force Base, Columbus, OH, 1983. 14. T. Nishioka et al., Sumitomo Elec. Tech Rev. 1990. 29, 262-269. 15. R. C. Laible, in Ballistic Materials and Penetrution Mechanics, R. C. Laible (Ed.), Elsevier Scientific Publishers, Amsterdam, 1980, pp. 133-143. 16. D. G. Brandon, Ceramic Armor, in Concise Encj~clopcJdiaof Advanced Ceramic Materials, R. J. Brook (Ed.), Pergamon Press, Oxford, 1991, pp. 22-24. 17. D. W. Lazenby and P. Phillips, Cutting,for Construction, Architectural Press, London, 1978. 18. H. J. Swinheart (Ed.), Cutting Tools Materials Selections, Am. Soc. of Tools and Manufactury Eng., Dearborn, Michigan, 1986. 19. T. Tatsuno and M. Miyamoto, Kobelco Technol. Rev. 1994, 117, 11-15. 20. G. Lutz, Int. Ceram. Rev. 1993, 42. 367-371. 21. R. T. Coelho, S. Yamada, D. K . Aspinwall, and M. L. H. Wise, Int. J . Machine Tools Manuf: 1995,35, 761-774. 22. K. Przyklenk, Indust. Diamond Rev. 1993, 53. 192-195. 23. T. Ikeda, Y. Kawate, and Y. Hiral, Kobelco Technol. Rev. 1989, 6, 1 4 . 24. J. Bruhn, S. Schicker, D. E. Garcia, R. Jenssen. F. Wagner, and N. Claussen, Ke.v Eng. Muter. 1997, 121, 73-80. 25. W. D. Kingery, H. K. Bowen, and D. R. Uhlmann, Introduction to Ceramics, 2nd edn., A Wiley-Interscience Publication, John Wiley & Sons, Inc., New York, 1976. 26. S. Chattergee and S. Chandrashekhar, Key Engin. Muter. 1996, 114, 67-104. 27. Z. Hashin and S. Shtrikman. J. Mech. Phys. Solids, 1963, 11, 163. 28. M. Paley and J. Aboudi, Comp. Sci. Technol. 1991, 41,411429. 29. T Christman, A Needelman, and S. Suresh, Actu Metall. 1989, 37, 3029-3050. 30. G. Bao, J. W. Hutchinson, and R. M. McMeeking, Acta Metall. 1991, 39, 1871-1882. 31. A. Maewal, Comput.Meth. Appl. Mech. Eng. 1986, 57, 191-205. 32. C. T. Sun and W. C. Liao, J . Cotnp. Mater. 1990, 24, 977-993. 33. G . D. Quinn and R. Morrell, J . A m . Ceram. Soc. 1991, 74,2037-2066. 34. W. Weibull, Sived. Ins. Eng. Res. Proc., 151, 1-45, 35. E. J. Gumbel, Statistics of’E.utremes, Columbia University Press, New York, 1958, p. 136. 36. L. G. Johnson, Theory and Technique of’ Variation Research, Elsevier, Amsterdam, 1964, p. 137. 37. R. W. Davidge, Mechanical Behavior of Ceramics, Cambridge University Press, Cambridge, 1979. 38. D. Lewis, A m . Ceram. Soc. Bull. 1982, 61, 1208-1214. 39. M. J. Hanney and R. Morrell, Proc. Br. Ceram. Soc. 1982, 13, 277-290.

100 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50.

51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84.


D. Lewis 111 and S. M . Oyler, J. Am. Ceram. SOC.1976, 59, 289-307. J. Seidel, N. Claussen, and J. Rodel, J. Europ. Ceram. SOC. 1995, 15, 395404. T. Lube, M. Manner, Key Eng. Muter. 1997,132,488-491. A. A. Griffith, Phil. Trans. Roy. SOC.Lond. 1920, A221, 163. A. A. Griffith, Proc. First Znt. Cong. Appl. Mech., C. B. Biezeno, J. M. Burgers, and J. Waltman Jr. (Eds), 1924, Delft, p. 35. C. E. Inglis, Proc. Roy. SOC.Lond. 1913, A127,290. G. R. Irwin, in Handbuch der Physik, Vol. 6, Springer-Verlag, Berlin, 1958, p. 551. E. Orowan, Weld. Res. Supp. 1955, 34, 157s. M. L. Williams, J. Appl. Mech. 1958, 24, 109. T. T. Shih and C. Opoku, Eng. Frac. Mech. 1979, 12,479498. D. P. Rooke and D. J. Cartwright, Compendium of Stress Intensity Factors, HMSO, London, 1976. H. Tada, P. Paris, and G. Irwin, The Stress Analysis of Cracks Handbook, Del Research Corporation, Hellertown, PA, USA, 1987. Y. Murakami, Stress Intensity Factors Handbook, Vol. 1 and 2, Pergamon Press, Oxford, 1987. D. P. Rooke, F. I. Baratta, and D. J. Cartwright, Eng. Fract. Mech. 1981, 14, 397426. G. C. Sih, Handbook of Stress Intensity Factors, Lehigh University, 1973. J. W. Hutchinson and Z . Suo, in Advances in Applied Mechanics, J. W. Hutchinson and T. Y. Wu (Eds), Academic Press, New York, 1991. L. B. Freund, Dynamic Fracture Mechanics, Cambridge University Press, Cambridge, 1990. E. C. Combe and D. C. Smith, Br. Dent. J. 1964, 117, 237. A. H. Cottrell, Dislocations and Plastic Flow in Crystals, Oxford University Press, Oxford, 1965. D. Tabor, The Hardness of Metals, Clarendon Press, Oxford, 1951. R. Hill, The Mathematical Theory of Plasticity, Clarendon Press, Oxford, 1950. S. R. Witek, G. A. Miller, and M. P. Harmer, J. Am. Ceram. SOC.1989, 72, 469473. L. A. Davis, c.-P. Chou, L. E. Tanner, and R. Ray, Scripta Metall. 1976, 10, 937. H. Li and R. C. Bradt, J. Non-Cryst. Sol. 1992, 146, 197. P. M. Sargent and T. F. Page, Proc. Br. Ceram. SOC.1987,26,209. P. F. Becher, J. Am. Ceram. SOC.1974, 57, 107. J. B. Quinn and G. D. Quinn, J. Muter. Sci. 1997, 32, 4331. R. W. Rice, C. C. Wu, and F. Borchelt, J. Am. Ceram. SOC.1994.77. 2539-2553. K. Zeng, E. Soderlung, A. E. Giannakopoulus, and D. J. Rowcliffe,’Acta Muter. 1996, 44, 1127-1141. J. Lankford and D. L. Davidson, 3rd International Conference on Mechanical Behavior of Materials, Pergamon, Oxford, 1980. J. Valli, J. Vac. Sci. Technol. A , 1986, 4, 3007-3014. S. Herada, N. Noda, T. Endo, and M. Nagano, J . SOC.Muter. Sci. Jpn. 1988, 37, 904-907. Y. Le Petitcorps, M. Leheye, R. Pailer, and R. Naslain, Cornp. Sci. Technol. 1988, 32, 31-55. B. Thomas, D. West, W. E. West, and K. Wyckoff, Powder Metall. Znt. 1988, 20, 39-44. R. W. Davidge, Muter. Sci. Technol. 1986, 2, 902-909. American Society for Testing and Materials, Standard Test for Plane Strain Fracture Toughness of Metallic Materials, ASTM E-399, Philadelphia, Pennsylvania, 1987. I. Hadley and M. G. Dawes, Fatigue Fract. Eng. Muter. Struct. 1996, 19, 963-973. H. Awaji, T. Yamada, and H. Okuda, J. Jpn. Ceram. SOC.,Znt. Edn. 1991, 4, 403408. G. D. Quinn, J. J. Kubler, and R. J. Gettings, VAMASTechnical Report No. 17, NIST, Gaithersburg, Maryland, 1994. T. Nishida, Y. Hanaki, and G. Pezzotti, J. Am. Ceram. SOC.1994,77, 606608. D. Munz, in Fracture Toughness of Ceramics, Vol.6, R. C. Bradt, A. G. Evans, D. P. H. Hasselman, and F. F. Lang (Eds), Plenum Press, New York, 1983. B. R. Lawn and E. R. Fuller, J. Muter. Sci. 1975, 10,2016-2024. R. Warren and B. Johnson, Powder Metall. 1984, 27, 25-29. T. Sadahiro and S. Takatsu, in Modern Developments in Powder Metallurgy, Vol. 14, H. H. Hauser, H. W. Antes, and G. D. Smith (Eds), Metal Powder Industries Federation, Princeton, NJ, 1981, pp. 61-72. T. Nose and T. Fujii, J. Am. Ceram. SOC.1988,71, 328-333.

References

101

85. D. Sherman, J . Am. Ceram. SOC.1997,80, 1904-1906. 86. C. J. Gilbert, R. H. Dauskardt, and R. 0. Ritchie, J. Am. Ceram. Soc. 1995,78,2291-2300. 87. J. A. Salem, G. D. Quinn, I. Bar-On, and M. G. Jenkins, A S T M Provisional Standard P S 07097, ASTM Committee C-28, 1997. 88. B. R. Lawn, J. Am. Ceram. SOC.1998,81, 1977-1994. 89. R. B. Tait, P. R. Fry, and G. G . Garrett, Exp. Mech. 1987, 27, 14-22. 90. M. Manoharan, S. v. Kamat, Znt. J. Fract. 1995, 73, R41LR45. 91. C. J. Gilbert, J. J. Cao, L. C. De Jonghe, and R. 0. Ritchie, J. Am. Ceram. Soc. 1997,80,22532261. 92. B. R. Murphy, W. W. Predebon, and B. J. Pletka, J. Muter. Sci. Lett. 1994, 13, 13461348. 93. V. S. Krilov, V. I. Kovalenko V. G. Marinin, I. T. Ostapenko, and V. P. Podtyakn, Sov. J. Superhard Mater. 1983, 5, 21. 94. S . Palmqvist, Jernkontorets Annalen, 1957, 141, 300. 95. S . Palmqvist, Jernkontorets Annalen, 1963, 167, 208. 96. B. R. Lawn and M. V. Swain, J . Mater. Sci. 1975, 10, 113. 97. B. R. Lawn and M. V. Swain, J. Muter. Sci. 1975, 10, 1049. 98. B. R. Lawn and A. G. Evans, J. Muter. Sci. 1979,12, 2195. 99. D. B. Marshall, B. R. Lawn, and A. G. Evans, J . Am. Ceram. SOC.1982,65, 561. 100. D. B. Marshall and B. R. Lawn, in Microindentation Techniques f o r Materials Science and Engineering, P. Blau and R. Lawn (Eds), ASTM STP 889, Philadelphia, Pennsylvania, p. 26. 101. J. R. Kelly, M. E. Cohen, and J. A. Tesk, J. Am. Ceram. SOC.1993,76,2665-2668. 102. S . G. Cook, J. E. King, and J. A. Little, Muter. Sci. Techno/.,995, 11, 1093-1098. 103. S. Srinivasan, R. 0. Scattergood, Li Zhuang, A. Grosh, A. S. Kobayashi, and R. C. Bradt, J. Am. Ceram. Soc. 1991,74,887-890. 104. C. B. Ponton and R. D. Rawlings. Mater. Sci. Technol. 1989, 5, 961-976. 105. R. Kurth, B. F. Lawlor, Y. J. Stockmann, and R. W. Steibrech, Fract. Mech. Ceram. 1996,11, 151-162. 106. R. F. Cook and L. M. Braun, J. Muter. Sci. 1994, 29, 2192-2204. 107. S. I. Athabai, J. Bodel, T. Dabs, and B. R. Lawn, J. Mater. Sci. 1991, 26, 2313-2321. 108. J. Foster, Key Eng. Mater. 1996, 122, 247-278. 109. E. Ness and R. Zibbel, Wear, 1996, 196, 12e125. 110. J. E. Ritter, Key Eng. Muter. 1992, 171, 93-106. 111. Y. I. Oka, H. Ohnogy, T. Hosokawa, and M. Matsumura, Wear, 1997,203,573-579. 112. J. E. Ritter, S. R. Choi, K. Jakus, P. J. Whalen, and R. G. Rateick Jr., J. Muter. Sci. 1991,26, 5543-5546. 113. G. Barbezat, A. R. Nicoll, and A. Sickinger, Wear, 1993, 162-164, 529-537. 114. Chen Yin Qian et al., Wear, 1990, 137, 175-186. 115. K. Kato, Wear, 1992, 153, 277-295. 116. A. L. Ryzhikov V. M. Gropyanov, and L. G. Tarakanchikov, Sov. J. Frict., Wear, 1991,12, 143- 146. 117. J. Echigoya, Strength of Materials, Proceedings of the 10th International Conference, Japan Institute of Metals, Sendai, Japan, 1994, pp. 853-856. 118. J. Echigoya, T. Hyakubu, Z. T. Liu, K. Sasaki, and H. Suto, Surf. Coat. Techno/. 1994, 67, 11 1-1 17. 119. D. A. ONeil and S. F. Wayne, J. Eng. Mater. Tecnol. 1994, 11, 471478. 120. V. Hajek, K. Rusnak, J. Vlcek, L. Martinu, H. M. Hawthorne, Wear, 1997,213, 80-89. 121. R. Raj and R. K. Boria, Tailoring Multiphase and Composite Ceramics, Proceeding of the 21st University Conference on Ceramic Science, Plenum, New York, 1986 pp. 27-39. 122. P. Basetty, T. Jensen, and J. N. Gray, Review of Progress in Quantitative Nondestructive Evaluation, Plenum, New York, 1997, pp. 2167-2174. 123. J. A. Medding and C. Duke Jr., Review of Progress in Quantitative Nondestructive Evaluation, Plenum, New York, 1997, pp. 1503-1508. 124. D. J. Roth, D. B. Stang, S. M. Swickard, M. R. DeDuire, and L. E. Dolhert, Muter. Eva/. 1991,49, 883-888. 125. J. Lamon, Rev. Metall. 1995, 92, 265284.

102


126. G. Petzow, R. Telle, and R. Danzer, Muter. Churact. 1991, 26, 289-302. 127. A. Zimmermann, M. Hoffman, B. D. Flinn, R. K. Bordia, T. J. Chuang, E. D. Fuller, and J. Rodel, J . Am. Ceram. SOC.1998,81, 2449-2457. 128. L. A. Xue, K. Meyer, and I. Wei Chen, J. Am. Ceram. SOC.1992,75,822-829. 129. Yet Ming Chiang and W. D. Kingery, J. Am. Ceram. SOC.1989,72,271-277; 1990,73, 11531158. 130. V. Jayaram, R. Manna, M. G. Kshetrapal, J. Sarker, and S. K. Biswas, J . Am. Ceram. SOC. 1996,79,710-772. 131. B. S. S. Daniel and V. S. R. Murthy, Muter. Des. 1995, 16, 155-161. 132. R. W. Evans, D. P. Thompson, and B. Wilshire, Br. Ceram. Trans. 1993,92, 28-30. 133. W. E. Luecke, S. M. Wiederhorn, B. J. Hockey, R. F. Krause Jr., and G. G. Long, J. Am. Ceram. SOC.1995,78,2085-2096. 134. Y. Maehara and T. G. Langdon, J . Muter. Sci. 1990, 25, 2275-2286. 135. I.-W. Chen and L. A. Xue, J . Am. Cerum. SOC.1990,73,2585-2609. 136. T. G. Nieh, J. Wadsworth, and F. Wakai, Int. Muter. Rev. 1991,36, 146-161. 137. F. Wakai, S. Sakaguchi, and Y . Matsuno, Adv. Cerum. Muter. 1986, 1, 259-263. 138. T. G. Nieh, C. M. McNally, and J. Wadsworth, J. Muter. 1989, 41, 31-35. 139. K. Kajihara, Y. Yoshizawa, and T. Sakuma, Scripta Metall. Muter. 1993, 28, 559. 140. T. G. Nieh and J. Wadsworth, J. Muter. Res. 1990, 5, 2613-2615. 141. F. Wakai and T. Nagano, J. Muter. Sci. 1991, 26, 241-2447. 142. J. Karch, R. Birringer, and H. Gleiter, Nature, 1987, 330, 556558. 143. Z. Cui and H. Hahn, NanoStructured Muter. 1991, 1, 419-425. 144. A. H. Chokshi, A. Rosen, J. Karch, and H. Gleiter, Scripa Metall. 1989, 23, 1679-1684. 145. A. M. El-Sherik, U. Erb, G. Palumbo, and K. T. Aust, Scripta Metall. 1992, 27, 1185-1188. 146. H. J. Hofler and R. S. Averback, Scriptu Metall. 1990, 24, 2401-2406. 147. R. 0. Scattergood and C. C. Koch, Scripta Metall. Muter. 1992, 27, 1195-1200. 148. C. S. Pande, R. A. Masumura, and R. W. Armstrong, NanoStructuredMater. 1993,2,323-331. 149. V. G. Gryaznov, M. Yu. Gutkin, A. E. Romanov, and L. I. Trusov, J . Muter. Sci. 1993, 28, 43594365. 150. W. R. Cannon and T. G. Langdon, J . Muter. Sci. 1988,23, 1-20. 151. E. Yasuda, K. S. Ramesh, S . Kimura, Report of the Research Laboratory of Engineering Materials, Tokyo Institute of Technology, 1985, 10, 115-134. 152. A. Hynes and R. Doremus, Crit. Rev. Solid State Muter. Sci. 1996,21, 129-187. 153. A. H. Chokshi and T. G. Langdon, Muter. Sci. Technol. 1991,7, 577-584. 154. D. S . Wilkinson, Fruct. Mech. Ceram. 1992, 10, 349-366. 155. D. P. H. Hasselman, K. V. Donaldsod, and A. Venkateswaran, Fract. Mech. Ceram. 1992,10, 493-507. 156. G. Grathwohl, Muterialwiss. Werkstofftech. 1988, 19, 113-124. 157. E. D. Whitney, Powder Metall. Znt. 1983, 15, 201-205. 158. A. G. Evans, Znt. J . Fract. 1980, 16,485498. 159. D. B. Marshal, M. C. Shaw, R. H. Dauskard, R. 0 Ritchie, M. J. Readey, and A. H. Heuer, J . Am. Ceram. SOC.1990,73,2659-2666. 160. R. H. Dauskardt, B. J. Dalgleish, D. Yao, R. 0. Ritchie, andP. F. Becher, J. Muter. Sci. 1993, 28, 3258-3266. 161. C. J. Gilbert, J. J. Cao, W. J. MoberlyChan, L. C. DeJonghe, and R. 0. Ritchie, Actu Muter. 1996,44, 3199-3214. 162. C. J. Gilbert, R. N. Petrany, R. 0. Richie, R. H. Dauskardt, and R. W. Steinbrech, J . Muter. Sci. 1995, 30, 643. 163. C. J. Gilbert and R. 0. Ritchie, Eng. Fract. Mech. 1998, 60, 303-313. 164. Jian-Ku-Shang and R. 0. Ritchie, Metall. Trans. A , 1989, 20, 897-908. 165. M. L. Wilkins, C. F. Cline, and C. A. Hondol, Tech. Rept. UCRL-50694, Lawrence Radiation Laboratory, 1969. 166. M. L. Wilkins, in Ballistic Materials and Penetration Mechanics, R. C . Laible (Ed.), Elsevier, Amsterdam, 1980, pp. 225-252. 167. H. Kolsky, in Ballistic Materials and Penetration Mechanics, R. C. Laible (Ed.), Elsevier, Amsterdam, 1980, pp. 185-252.

References

168. 169. 170. 171. 172. 173. 174. 175. 176. 177. 178. 179. 180.

103

Y. J. Lee, Int. J . Frac. 1992, 57, 2546. C. C. Ma and S. K. Chen, Int. J . Frac. 1992,58, 345-359. A. S. Kobayashi, Muter. Sci. Eng. 1991, A143, 1 11-1 17. Z. Rosenberg, Y. Yeshurun, and G. Brandon, J . Physique, 1985, 8, 331-341. D. A. Shockey, A. H. Marchand, S. R. Skaggs, G. E. Cort, M. W. Burkett, and R. Parker, Int. J . Impact Eng. 1990,9, 263-275. J. E. Field, S. M. Walley, N. K. Bourne, and J. M. Huntly, Journalde Physique IV, 1994,4, C8/ 2-22. D. Sherman and D. G. Brandon, J. Muter. Res. 1997, 12, 1335-1343. D. Sherman and T. Ben-Shushan, Int. J. Impact Eng. 1998,21,245-265. D. Sherman, Int. J . Impact Eng., accepted, 1999. B. R. Lawn, J. Am. Ceram. SOC.1998, 81, 1977-1994. C. Kocer and R. E. Collins, J. Am. Cerurn. Soc. 1998,81, 1736-1742. R. Mouginot and D. Maugis, J. Muter. Sci. 1985, 20, 4354. M. F. Ashby and S. D. Hallam, Acta Metall. 1986, 34, 497.


Nanostructured Superhard Materials S. Vepfek

4.1 Introduction Hardness of a material is a measure of its ability to resist deformation [l, 21. The upper limit of the hardness is determined by the rigidity of the crystal structure upon isostatic pressure. This ‘theoretical’ hardness is proportional to the reciprocal value of the bulk modulus, B. From the Hooke’s law B = da/dE (CTis the applied stress and E the resulting elastic strain) and from the relationship between the stress (a force) and the first derivative of the binding energy, E, with bond distance, a, at the equilibrium position ao, CT = (dEb/da)o, one obtains [3] B=

(2) 0

.

Thus, the theoretical hardness increases with the increasing curvature of the interatomic potential curve at the equilibrium bond distance, ao, that is, with increasing bond energy and decreasing bond length. A high bond energy means a high electron density between the atoms as found for a nonpolar, covalent bond between atoms of small radii of the first period. In addition, a high coordination number is required in order to maximize the value of B. For these reasons carbon in its metastable, fourfold coordinated sp3-hybridization, diamond polymorph, is the hardest material known, followed by cubic boron nitride, cBN [2]. Based on such general considerations C. M. Sung predicted in 1984 that C3N4 should have a hardness comparable to diamond [2]. Later on, Cohen [4] and coworkers [5-71 derived Equation (2), which reproduced very well the bulk moduli of many materials and confirmed the prediction of C. M. Sung regarding the theoretical hardness of C3N4,

Here, ( N c ) is the average cpordination number and X is the polarity of the bond, B is in GPa and a. is given in Angstroms [6]. For nonpolar, covalent bonds in diamond X = 0, whereas in other compounds, such as cBN, Si3N4and C3N4 X > 0 which decreases the value of the elastic modulus. The expected high theoretical hardness of C3N4 is based on the small bond distance and relatively small polarity A. However, like diamond and cBN, C3N4is a thermodynamically unstable compound whose preparation requires a high degree of dissociation of nitrogen in order to achieve the desired stoichiometry, a medium-energy ion bombardment to promote the carbon into the metastable fourfold coordination and a relatively high temperature of about >800°C in order to evaporate the paracyanogen,

4.1 Introduction

105

(CN),, which is always formed during the deposition of the stoichiometric C3N4(see [S] for further details). However, the Vickers hardness reported for stoichiometric amorphous C3N4 films reached only about 30 GPa, far below that of diamond (70-100 GPa [2,5,6]). Using similar synthetic principles, other authors succeeded in the deposition of crystalline, stoichiometric C3N4 films [9-131 but they could not measure the hardness because of an insufficient density of the polycrystalline films [lo]. A large number of papers report on the ‘synthesis of carbon nitride’ but those films are highly substoichiometric with a large nitrogen deficiency even below that of paracyanogen and have a lower hardness. Nevertheless, CN, may find some ‘niche’ applications such as protective, low friction coatings on magnetic storage discs [14-16]. In contrast to the C3N4 and substoichiometric CN, thin films of turbostratic graphite with a relative low concentration of nitrogen, CN0,25p0,35 and three dimensional cross-linking deposited at a relatively high temperature of 400°C have a hardness of 40-50GPa [17-191. Also the hardness of diamond crystals can be increased by nitrogen impurities 120,211, thus reaching the upper end of the hardness scale in Fig. 1. These are two examples of the well known rule that the microstructure determines the practically achievable strength (and hardness) of engineering materials, which is orders of magnitude smaller than the theoretical strength. The latter is calculated for the slip of crystal planes as c,(theor) G/2n, where cT,(theor) is the theoretical critical fracture stress and G the shear modulus [3,22]. The reason is that the deformation and fracture of materials occurs due to the multiplication and movement of dislocations in crystalline materials and growth of microcracks in glasses and ceramics, all of which require a much smaller stress than the theoretical value a,(theor) mentioned above [3,22241. Therefore, the practical strength and hardness of materials is determined by the

6000{

(nc-M,N/a-Si3N, 160

4000

0

Figure 1. Vickers hardness of selected materials in comparison with that of the heterostructures and nanocrystalline composites. Notice that the hardness of diamond depends on the quality and purity of the crystal (see text).

106

4 Nanostructured Superhard Materials

I

9000

8000

-

I

I

.

I

r

n

I

.

I

I

-

0

7000 N -

E 6000E

0 O

0

.

0

-

3 500025 (I) (I)

E

4000-

-

0 0

3000-

nJ 2000r

1000

0

-

0

I

0

20

'

I

40

'

I

60

'

I

80

'

I

.

100

Figure 2. (a) An example of the indentation curve measured on a =5pm thick layer of ncTiN/aSi3N4.L,,, and h,,, are the maximum applied load and indentation depth, hplastic is the depth of the remanent, plastic deformation. The arrows -+ and t denote the loading and unloading part. (b) Measured values of the hardness against the maximum applied load. (c) Indentation depth at a constant applied load of 50mN against time which shows the absence of creep.

4.1 Introduction

'ma,

,

I

,

I

,

I

I

.

I

.

I

.

0.26

0.24 sl

sf$0.22-

-

-0

.g 0.20 C

0

H, = 7529 [N/mm2] El(1-v2) = 520.28 [GPa]

.....................

107

c-

-

c

E

9,

TI

-

-

E 0.18

c

0

0.16z0

time [sec] (C)

microstructure which hinders the multiplication and movement of dislocations and growth of the microcracks. Of course one always should use a material with a possibly high cohesive energy and optimize the design of its microstructure depending on the conditions of its applications and on the nature of the load imposed on that material (tensile, compressive, shear, static, dynamic, . . .). The same applies for the hardness. Besides the 'theoretical' hardness as the resistance of a material against isostatic compression, depending on the reciprocal value of the bulk modulus mentioned above (see Equation (2)), three categories of hardness are distinguished in practical measurements: Scratch hardness (e.g. Mohs scale used by mineralogists), static indentation hardness (e.g. Brinell, Vickers, Knoop) and rebound or dynamic hardness. In the latter measurement the indentor is usually allowed to fall under gravity from a given height which, together with the indentor mass, defines the impact energy [ 11. Obviously, the theoretical hardness is of little use in considering a material for an application. Instead one has to consider the practically measured type of hardness depending on the particular applications. Here we shall limit ourselves to the static indentation measurements using the Vickers method for the simple reason that this, and the Knoop technique, are the only ones which can be simply applied to the measurement of the superhard materials. The Vickers indentor is a regular pyramid made of diamond with an angle of 136" between the opposite faces. The choice of this angle is based on an analogy with the Brinell test because both methods yield similar values of hardness, however only for relatively soft materials with H 5 1000kg mm-2 (see [ 11for further details). The diamond pyramid is pressed into the material to be tested under a defined load L (kg) and, after unloading, the average value of the two diagonals d

108


(mm) of the resultant plastic deformation is measured with a microscope. The Vickers hardness is proportional to the ratio of the applied load and the area of the plastic deformation H = const L / d 2 and given in units of kgmm-2 or GPa [ 1,2,25-281. (For historical reasons, the values of Vickers hardness are usually given in kgmmP2, the conversion factors to other units used are 1000kgmmP2 = 9.807 GPa = 9807NmmP2.) Figure 1 shows the Vickers hardness of some materials including the heterostructures and nanocomposites to be discussed here. The Knoop hardness is measured in a similar way, but the diamond pyramid has two different angles of 172’30’ and 130’, thus yielding an elongated plastic deformation. The hardness is calculated from the longer diagonal whereas the shorter one yields information on the elastic properties (see [l, 24,291 for further information). Modern indentometers for the measurement of the hardness of thin films use a computer controlled step-wise increase of the load up to a chosen maximum L,,, followed by a step-wise unloading. Instead of measuring the diagonal of the plastic deformation, the indentation depth, h, is measured electronically and the indentation curve evaluated (see Fig. 2a). The ‘plastic hardness’ is calculated from the maximum load, L,,,, and the depth of the plastic indentation, hplastic,as

The ‘universal hardness’, H,, is obtained from the same formula if h,,, is inserted instead of hplastic. The universal hardness includes both the elastic and plastic deformation. The linear part of the unloading curve corresponds to the elastic recovery when the diamond pyramid is in a constant area contact with the material. Therefore it represents Hooke’s law and allows one to calculate the corresponding ‘elastic modulus’ E/(1 - v 2 )which is a complicated function of the bulk, shear, and tensile moduli (v is the Poisson ratio). The details of the apparatus, the measuring procedure and possible errors are given in the relevant papers to which we refer here [25-281. If done correctly, the plastic hardness measured by the indentation agrees within about 10-1 5% reasonably well with that from the classical Vickers method at least in the range H 5 1500 kgmmp2 [25]. The so called ‘nanoindentation’, which is frequently used to measure < 1 pm thin films, is subject to a number of possible errors [27,28]. For example, when measuring soft materials, such as aluminum or pure iron with a small load and indentation depth, the dislocations are pinned in the surface contaminant layer (oxides, carbides) and, consequently, unrealistically high values of hardness are found. The plastic deformation may also need a certain time to reach equilibrium under the given load because of a finite velocity of dislocation movement. This can be seen as creep (increase of the indentation depth) when a constant load is applied for 10-30 s. Therefore, in order to obtain reliable values of the hardness, one should always use at least >4pm thick films, make sure that the maximum indentation depth, h,,,, does not exceed 10% of the film thickness, check for possible dynamic response of the material by sustaining the indentor at the maximum load L,,,

4.2 Concept for the Design of Superhard Materials

109

for at least 2&30 s and measure the hardness (always at a different position on the layer) as a function of the maximum applied load. Figure 2b shows the typical dependence of the measured plastic hardness on L,,, and the typical scattering of individual measurements. In the case of superhard materials, one typically finds a smaller hardness and large scattering of the individual measurements when using small loads of SlOmN because the indentation depth of 50.1 pm is comparable with the surface roughness. Figure 2c shows the time dependence of A,, under the constantly applied maximum load L,,,. One notices that there is no creep seen, at least on the time scale available with the given instrument. We emphasize these problems here to make sure that in the measurements to be reported hereafter, all these precautions have been taken in order to make the results as reliable as possible. However, one should keep in mind that the Vickers technique, which was originally developed for the conventional alloys and materials with hardness not exceeding 2000-2500 kg mm-*, is not suitable for the measurements of materials whose hardness approaches that of diamond. Therefore we consider values of hardness of 7000-10 000 kg mmp2 (and even higher) frequently found in our measurements only as an indication that these materials approach the hardness of diamond, but the Vickers hardness scale is probably not linear in that range. As it will be shown later, there is no dislocation activity in the superhard nanocomposites and, therefore, no ‘plastic deformation’ in the classical sense. Strictly speaking one should use the term ‘remaining’, ‘permanent’ or ‘pseudoplastic’ deformation because it is a brittle fracture as found, for example when indenting into a glass. Because the current literature does not differentiate between these terms we adopt the term ‘plastic deformation’ keeping, however, in mind that it is in fact a brittle fracture whose microphysics and micromechanics are not yet understood. Last but not least, one should not forget that for practical applications the hardness at room and at a high temperature is only one of many properties which determine the applicability of a material. Further mechanical properties, such as fracture toughness under static and dynamic load, coefficient of friction, corrosion resistance, and reactivity with the material to be machined at room as well as at elevated temperature are decisive for most applications.

4.2 Concept for the Design of Superhard Materials Most researchers agree on the definition accordin to which ‘superhard’ materials are those with Vickers hardness 24000 kg mm-’ (about 40 GPa), and that the ‘theoretical hardness’ (see Equation (2) and related discussion) is only of an academic interest but hardly of any use. There are various possible approaches to achieve such a goal. A strong material must have a high cohesive energy, be free of flaws (e.g. voids), and its microstructure must be designed in such a way as to efficiently hinder the multiplication and movement of dislocations and growth of microcracks. The fracture stress, af,,can be increased in various ways

110


known from the metallurgy, which are briefly summarized in Equation (4) (see e.g. ~31):

Here, the first term describes the Peierls-Nabarro stress which determines the dislocation movement (slip within the crystallographic plane) in a crystal, the second term corresponds to the solution hardening (C is the concentration of the impurities), the third term is the Hall-Petch relationship [3,30,3I], which describes grain boundary strengthening (d is the crystallite size, see below), the fourth term describes the strain hardening due to dislocation pile up ( a is a material constant, G the shear modulus, b the Burgers vector, and p the dislocation density). The last term describes the overall dependence of the crack growth on the size of microcracks (voids) of the size a. ( E is the Young modulus, ys the surface energy). Any of these processes allows the engineer to increase the strength of a material by a different amount. For example, the hardness of TIN films deposited in glow discharge plasmas under energetic ion bombardment which causes strengthening due to the formation of dislocations or due to a high biaxial compressive stress is typically higher than that of a single crystal.

4.2.1 Nanocrystalline Materials The decrease of the grain size d strengthens metallic alloys (Hall-Petch relationship, third term)

as well as ceramics (last term in Equation (4), Griffith theory [3]). There are several different mechanisms and theories which describe Hall-Petch strengthening [3,24,30-331: Pile-up models, work hardening and grain-grain boundary composite models which yield different relationships for a. and kgb [32]. The work hardening models and the core-mantle composite model of Koks 34 yield also a more comI . can originate plicated dependence of ofr on d . In ceramics, the d-'I dependence from the fact that the size a. of the possible flaws such as voids and microcracks, which are formed during the processing of the material, also decreases with decreasing grain size. For these reasons, the Hall-Petch relationship (5) should be considered as a semiempirical formula which is valid down to a certain minimum grain size, for example of d > 450b = l00nm for the pile-up models and d > 60b x 15 nm for a dislocation source model [32]. There are many examples which show that the strength and hardness of an alloy can be increased by a factor of 4-7 if the grain size is decreased into the nanometer range. One such example is the rapidly solidified A19&r3CelCo, (x = 1-2%) alloy containing as a main component a nanosized quasicrystalline icosahedral phase [35]. A similar strengthening is found in many metals such as TiAl[36], Cu [37], and other

1

4.2 Concept ,for the Design of Superhard Materials

111

nanocrystalline metals and intermetallics [38,39]. However, when the dislocations activity decreases with grain size decreasing below about 20-50 nm (which causes the strengthening) grain boundary sliding becomes active. This results in a softening of such materials when the crystallite size decreases below 6-10 nm [36-391. A simple phenomenological model describes the softening in terms of an increasing volume fraction of the grain boundary material, fgb, with crystallite size, d, decreasing below 4-6nm [40]: 1 H ( d ) = (1 -fgblHc +fgbHgb with f g b (6) d’ Because the hardness of the grain boundary material, H g b , is smaller than that of the crystallites, H,, the overall hardness of the material decreases with d decreasing below 10nm. The amount of such decrease depends on the nature of the grain boundary material. The decrease of the hardness in the grain size range 56-20 nm, sometimes called a ‘negative Hall-Petch slope’ was first reported by Chokshi et al. [41] and, afterwards, subjected to many studies with controversial conclusions [38, 39,42-501. In a recent paper Sanders er al. [50] have shown, that in the grain size range between about 10 and 100 mn the Hall-Petch slope depends on the sample preparation. If nanocrystalline Cu and Pd samples, prepared by inert gas condensation, were compacted at elevated temperature, the slope followed nearly that extrapolated from coarse-grain samples. However, even in this work one can see that the slope of the experimental points in the lower grain size range of 20-8 nm decreases and levels off (see Figs 6 and 7 in [50]). The authors emphasize that even such a compaction is probably not sufficient to remove all flaws from the grain boundaries which lead to softening. In a more recent paper, Musil and Regent [51] have found such a softening even for dense nanocrystalline NiCr thin films prepared by argon ion sputtering. These results, which are shown in Fig. 3, clearly demonstrate that the softening in the

y1600-

.

I’

E

m

251400cn u)

E g1200m r

.g 1000-

-cnmn

800;, d’ r

€

I I

I

I

crystallite size [nm] Figure 3. Hardness of ncNiCr (80/20 weight-%) alloy prepared by argon ion sputtering against crystallite size (from [51] with permission).

1 12


range d < 5-10 nm is most probably an intrinsic property of simple nanocrystalline materials which have inherent flaws in the grain boundaries. Although many details are still not understood there is little doubt that grain boundary sliding is the reason for the softening in this range of the crystallite size [38,39]. Various mechanism of grain boundary creep and sliding were discussed and are described by deformation mechanism maps in terms of the temperature and stress [3,52-571. The theories have been critically reviewed [32]. More recent computer simulations have shown that the grain boundary sliding in nanocrystalline materials is a universal phenomenon which can occur without any significant activation at room temperature [58,59]. Based on our understanding of these mechanisms one can expect that an appropriate design of a nanocrystalline/amorphous composite, which would avoid the grain boundary sliding, may cause the hardness of such material to continue increasing even for grain sizes below 6-10 nm and at elevated temperature. The absence of grain boundary sliding in a nanocrystalline material at room temperature does not exclude its appearance at elevated temperature which may occur under the conditions of their particular application, such as the leading edge of cutting tools which may reach a temperature of 700-1000°C. An idealized microstructure of such a composite is shown in Fig. 4. In reality, one might also evoke an interwoven bi-phase system as observed by Milligan et al. for ncAu/aSi (see Fig. 6 in [60]) or a bicontinuous two-phase system. However, as will be discussed below, we have not observed any such microstructure in our detailed high resolution transmission electron microscopic study but only isolated nanocrystals imbedded within the amorphous matrix [61,62]. In the interwoven bi-phase system ncAu/aSi, Milligan et al. have observed a significant improvement of the toughness of the composite as compared to a pure ncAu [60]. If such a composite consists of strong materials, such as hard nitrides with a hardness of about 2000 kg mmP2, even a relatively modest increase of the hardness by a factor of four would result in a hardness comparable to that of diamond. In order to

a-Si,N,

Ol Figure 4. Idealized structure of the nanocrystalline-amorphous composite.

4.2 Concept ,for the Design OjSuperhard Materials

113

form such a bi-phase system, both materials must be immiscible (i.e. display a thermodynamically driven segregation during the deposition) and the cohesive energy at the interface between both phases must be high. Last but not least, the amorphous phase must possess a high structural flexibility in order to accommodate the coherency strain without forming dangling bonds and other flaws. In order to ensure that such a composite will show no grain boundary sliding even at a high temperature, both materials should be refractory in order to ensure that the operational temperature of the tools will be less than half the melting or decomposition temperature, Tm,decomp 2 2500 K. It is important to note that, although the strengthening caused by the decreasing dislocation activity in the range of crystallite size of d 5 lOOnm finally reaches a saturation due to the absence of dislocations in ultrasmall nanocrystals of d 5 10 nm [38,39], the strength and hardness of such material can still increase upon a further decrease of the crystallite size when another mechanism of strengthening becomes dominant. In the absence of dislocations and grain boundary sliding, the nanocomposites show a brittle behavior which means that the fracture strength (and hardness) are proportional to the elastic modulus of the material. This behavior has been confirmed in our nanocrystalline superhard composites [63-651. The fracture stress of such material is then determined by the critical stress for the growth of microcracks (see the last term in Eq. (4)). The size of the 'microcrack', ao, that is, of a possible flaw and voids, is in a well compacted material always smaller than the size of the nanocrystals, uo 5 d . Moreover, even if the critical stress for the crack growth

should be exceeded by the applied (e.g. uniaxial) stress, the crack could hardly grow to a size larger than d because it would have to bend to pass around the nanocrystals. If p is the angle between the plane of such a bent crack and the direction the component ( T = ~ oappl sin p < rappl is acting in the of the tensile stress gappl, direction normal to the crack plane. Only this component can cause the crack to grow. For randomly oriented nanocrystals embedded in a 51 nm thin matrix, as found in our nanocomposites with the smallest crystallite size and highest hardness, ~ oappl. Furthermore, upon the cracks have to bend so strongly that cp + 0 and ( T << the bending also crack branching ('meandering') occurs in a three-dimensional isotropic nanocomposite which further decreases the ability of such nanocracks to propagate [3, 1701. These are the reasons why the hardness of our composites increases when the crystallite size decreases below 5-7 nm [64,66,67]. The obvious first (but not the only) choice of such an ncja system which meets the above discussed criteria for achieving the superhardness is straight forward: Many transition metals M, such as Ti, Zr, Hf, V, Nb, Ta, Cr form stable (refractory) and hard nitrides which crystallize upon deposition even at relatively low temperature of >lOO"C [68-701. Silicon nitride, on the other hand, grows amorphous even at 1100°C and it possesses the desirable structural flexibility due to the fourfold coordination of silicon combined with the threefold coordination of nitrogen.

1 14


TiN-Si3N4 are immiscible up to a relatively high temperature of 1000°C [71-731. A simple thermodynamic estimate [63] shows that similar applies to the majority of the transition metal nitrides mentioned above [74]. An even better choice of the amorphous phase should be boron nitride because the boron atoms can be three- and fourfold coordinated and the B-N bond is stronger than the Si-N one [6,7,74]. Because silicides and borides are refractory materials, also the cohesive energy at the ncM,N/aSi3N4 and ncM,N/(a and nc)BN interface should be large. We shall see later that this concept indeed works and superhard materials can be obtained in the M,N/aSi3N4 and ncM,N/(a and nc)BN systems. Because silicon dissolves in many metals at high temperatures, the choice of boron nitride, aluminum nitride and others offers advantages with respect to many applications such as machining. In the next section we shall briefly summarize another approach towards the design of nanostructured superhard materials which has been successfully developed in recent years.

4.2.2 Heterostructures In a theoretical paper published in 1970 [75] Koehler suggested a concept for the design of strong solids which are now called ‘heterostructures’ or ‘superlattices’. Originally, he suggested depositing epitaxial multilayers of two different metals M(l) and M(2) having as different elastic constants as possible, EM(2) > EM(,), but a similar thermal expansion and strong bonds. The thickness of the layers should be so small that no dislocation source could operate within the layers. The dislocations then would form within a layer M(l) with the smaller elastic modulus. If, under applied stress, the dislocation would move towards the M( 1)/M(2) interface, the elastic strain induced in the second layer M(2) with the higher elastic modulus would cause a repelling force which will hinder the dislocation movement across the interface. Thus, the strength of such multilayers should be much larger than that expected from the rule of mixtures. His prediction was further developed and experimentally confirmed by Lehoczky who deposited AI-Cu and Al-Ag heterostructures and measured their mechanical properties [76,77]. According to the rule of mixture, the applied stress, ca,which causes elastic strain, E , is distributed between the layers proportionally to their volume fractions VMcX)and elastic moduli EM(.) ga = (VM(l)Eh4(l) + VM(2)EM(2))&. (7) Lehoczky has shown that the tensile stress-strain characteristics measured on multilayers consisting of two different metals displayed a much higher Young modulus and tensile strength, which both increased with decreasing thickness of the double layer (‘layer period’). For layers thickness of 570 nm the yield stress of Al/Cu multilayers was 4.2 times larger and the tensile fracture stress was 2.4 to 3.4 times larger than the values given by the rule of mixture. This work was followed by a number of researchers who confirmed the experimental results on various metallic multilayer systems (e.g. Cu/Ni [78-801 as well as on hard epitaxial and polycrystalline superlattices of nitrides, such as TiNjVN

4.2 Concept f o r the Design of Superhard Materials

1 I5

[81], TiN/NbN [82,83], TiNV,Nbl -xN [84-861, Ti/TiN and WjWN [87], oxides [88,89] and superlattices consisting of transition metal nitride and CN, [90]. In all these cases, the hardness of the heterostructures increases with decreasing lattice period, reaches a maximum of about 4000-5000kgmm-2 at 5-7nm and, afterwards, decreases again. The theoretical work on the heterostructures was summarized in several excellent review papers [88,89,91]. Therefore, only a brief summary will be given here. For a large lattice period, where the dislocation multiplication source can still operate, the increase of the hardness and tensile strength (most researchers measured the hardness because it is simpler than the measurement of tensile strength as done by Lehoczky [76,77]) with decreasing layer thickness is due to the increase of the critical stress needed to multiplicate dislocations, a, = Gb/lpp (Ipp is the distance between the dislocation pinning sites [3]). Usually one finds strengthening dependence similar to the Hall-Petch relationship, Eq. (9,but with a somewhat different dependence on the layer period X-" (instead of the crystallite size, d , in Eq. ( 5 ) ) with n = for layers with different slip systems and n = 1 for layers with a similar slip systems [91]. In thin layers where no dislocation source can operate and the Koehler's and Lehoczky's model applies, the maximum value of the critical shear stress a,(X) which can cause a movement of a dislocation from the weaker layer 1 into the stronger layer 2 is given by [75-771

Here G M ( I ) are the shear moduli of the two materials, XI is the thickness of the softer layer, b is the Burgers vector of the dislocation, and cp is the angle between the dislocation slip plane in the layer M(1) and the interface M(l)/M(2). This theory predicts that the strength (and hardness) depends mainly on the relative difference between the shear moduli ( G M ( 2 ) - G M ( ~ ) ) / ( G M ( ~ ) G M M ( I ) )and the angle cp. For small period but still X1 >> 46 the enhancement reaches an asymptotic value of

+

a,(max) =

GM(2)

-

GM(2)

+ GM(l)

sincp 8.ir

GM(l) GM(l)

(9)

The decrease of the hardness experimentally observed at small values of X < 57nm is due to two effects: For XI smaller than about 15-20 times the Burgers vector, b, the repelling mirror forces induced by the strain in the layer of stronger materials on both sides of the layer M(l) begin to cancel out (these forces have opposite sign with respect to each other) and, therefore, the enhancement of the strength decreases. The other effect is a roughness of the interface due to the interdiffusion during the preparation. These effects were discussed in a number of papers ([82,92-951). In particular, Chu and Barnett [93] have shown that an interface roughness of about l n m will cause a strong decrease of the hardness and probably dominates the experimentally observed softening for X < 5-7 nm. Recent experimental data on superlattices with a high immiscibility which form sharper interfaces support this conclusion [94]. In a remark added in proofs, Koehler mentioned that the ideas described in his paper should also be valid if one of the layers is amorphous. In spite of that note,

116


the researchers studied only the epitaxial and, later on, polycrystalline heterostructures. Only recently, several papers appeared in which one of the layers consist of amorphous CN,, the other of a transition metal nitride such as TIN [88,96,97] or ZrN 1901. However, with decreasing layer thickness the layered structure vanished and a nanocrystalline composite structure appeared [90]. Such films also show a high hardness exceeding 4000 kg mm-2. In summary, significant progress has been achieved in the preparation and understanding of the hardness and strength enhancement in heterostructures and it is best described in the recent theoretical papers by Chu and Barnett [93] and Anderson and Li [94]. The application of heterostructures for cutting tools has been well developed [98] and cutting tools with superior properties as compared to single layer coatings are available. However, the relatively complicated preparation technique, which requires a periodic change of the source for the subsequent deposition of the multilayers, makes the investigation into the novel superhard nanocrystalline composites very attractive because they can be deposited in a simple plasma chemical vapor deposition (CVD), plasma vapor deposition (PVD) or a combined PVD and CVD process. Therefore, the remaining part of this chapter will concentrate on these nanocomposites.

4.3 Preparation and Properties of Superhard Nanocrystalline Composites 4.3.1 Preparation As already discussed, the formation of a nanocrystalline/amorphous (or an ncjnc) composite schematically illustrated in Fig. 4 and its thermal stability require a high immiscibility of the components. This is fulfilled, for example, in systems consisting of a stable transition metal nitride and silicon nitride. A simple estimate shows that in such a system, the immiscibility is assured if the activity of nitrogen is sufficiently high at the given temperature to shift the equilibrium of reaction (10) far to the left hand side [63,73,74] M,N

4mi-3

+ (m/3)Si3N4= Me,Si, + ___ 6 N2.

For most of the stable nitrides, such as TIN and others, this applies at a nitrogen pressure above about 1 mbar and temperature below 600°C. A similar estimate shows that the segregation should be much stronger in the TiN/BN and many other M,N/BN systems as compared to the TiN/Si3N4 one. It should also occur in the TiN/AlN and many other systems. We are primarily interested in superhard coatings for cutting tools, which have to be deposited on appropriate substrates, such as cobalt cemented carbide or steel. Thus, the preparation technique should assure a high nitrogen activity and a sufficient activation energy for the thermodynamically driven segregation of the

4.3 Preparation and Properties of Superhard Nanocrystalline Composites

117

phases to occur at such a low temperature which is compatible with these materials. The problem is mainly the steel, because the maximum tolerable temperature is about 550°C or even lower for high speed steel. Therefore, an activated chemical or physical vapor deposition technique which allows one to coat simultaneously a large number of nonplanar substrates (i.e. a large total area) at a sufficiently high deposition rate of 2 3 pmh-' (about 1 nms-') such as plasma CVD and PVD appear to be most appropriate. Because plasma CVD usually operates at a somewhat higher pressure of about 1-5 mbar whereas a lower pressure of
r

118


feedthreugh

,/ T'C*

R=30kQ

HV-supply

Pirani

%\themostate

'

XCI*

Figure 5. Schematics of the apparatus for the deposition of ncTiN/aSi3N4 coatings in an abnormal direct current glow discharge (see text).

TiN prepared from TiC14 usually suffers from a relatively large chlorine content which, for thermal CVD films deposited at 1000°C, can approach almost 1 at.%, and may be much higher in films prepared by means of plasma CVD at lower temperatures. Following our earlier studies about the shifts of the chemical equilibrium in intense glow discharge plasmas [log, 1091we have shown that the chlorine content of TIN deposited at 500°C can be significantly reduced by increasing the discharge current density at the cathode to 2 1 mA cm-2 [ 1101. Using this principle and a large excess of nitrogen and hydrogen, ncTiN/aSi3N4 films deposited at 500-550°C contain 50.5 at.% of chlorine, which is fairly tolerable. Another question is how to achieve the desirable stoichiometry, that is, a sufficiently high activity of nitrogen, a(N2),in order to achieve the segregation in reaction (10). The phase diagram of the ternary Ti-Si-N system at 1273K shows that this is achieved at loga(N2) 2 -6 which corresponds to a nitrogen pressure of lop6atm [71,73]. At a lower value of a(N2) the TiN, Si phases coexist and for loga(Nz) 5 -7.5TiNx TiSi2 appear. The phase diagram becomes more complicated at an even lower nitrogen activity, but this is not of any interest here. The important conclusions are: First, for a nitrogen pressure of 2 1 mbar as used in our experiments, fully stoichiometric TIN + Si3N4 phases should be formed at 550°C if the activation energy for N2 reaction with Ti and Si (i.e. a sufficient

+

+

4.3 Preparation and Properties of Superhard Nunocrystalline Composites

119

degree of dissociation) is provided by an intense discharge; second, decomposition of Si3N4is expected upon a long term annealing at T 2 1100°C under vacuum with a partial pressure of nitrogen of mbar [68,71,73,74]. This will be of importance later in this discussion. Thus, any nonstoichiometry observed experimentally during the plasma CVD must be due to either an insufficient activation of N2 or a preferential sputtering of nitrogen from the films during the deposition. It is therefore not surprising that in most cases to be reported here the films are fully stoichiometric and have shown gold-like color typical of TiN as long as the Si3N4 content was below about 15mol.% 1671. A small substoichiometry which we observed at a relatively high current density of 22.5 mA cm-2 might be due to preferential sputtering of nitrogen during the deposition. When the Si3N4content is increased significantly above the percolation threshold the color becomes darker (for Si3N4content of 230 mol.%) and finally gray (>40mol.%) [67]. The small fraction of elemental silicon observed by Li Shizhi et al. [lo11 and J. Patscheider et al. [ l l l ] was probably caused by insufficient nitrogen activity during the deposition. More recently, stoichiometric and superhard films of ncTiN/aSi3N4 and of ncTiN/ncBN coatings were prepared by means of a chlorine free, combined plasma PVD (vacuum arc evaporation of Ti) and CVD (silane and borazine used as a source of silicon and boron, respectively) technique by the SHM company [99] which is compatible with large scale industrial production. Also reactive sputtering was demonstrated to yield similar very hard ncTiN/aSi3N4 [l 11,1121, ncTiN,/CN,x [113], ncNbN/CN, [I 141, and TiC/aC [115] nanocomposites. Voevodin and Zabinski reported the deposition of ncTiC/aC nanocomposites by means of laser ablation [116,117]. Although the hardness of this material reached only about 32 GPa, the coatings have shown a remarkably high toughness. Because of the preliminary stage of this work it is too early to discuss details of these preparation techniques here. It should be just stated that this more recent development shows that it is possible to prepare the superhard nanocomposites also by techniques which are compatible with industrial large scale processing.

4.3.2 Properties of the ncM,N/aSi3N4 Composites Figure 6 shows the dependence of the average TiN crystallite size in the ncTiN/ aSi3N4 composites prepared in an H F discharge [63] on the silicon content as determined by energy dispersive analysis of X-rays (EDX). The squares correspond to the results obtained from the X-ray diffraction using the Warren-Averbach analysis [l IS] which in the range of the crystallite size of L1Onm yields the same values as the Scherrer equation if the integral width of the Bragg reflection is used [119].This is because the line broadening due to the random stress (50.5 GPa in our films) is small compared to that due to the finite crystallite size. Only Bragg reflections from the f.c.c. lattice of TIN were found in the XRD [63,64,66, 1011 and electron diffraction study [61,62] investigations of all ncTiN/aSi3N4 composites. The solid circles in Fig. 6 show the data obtained from direct lattice image in high resolution transmission electron microscopy (HR-TEM) [62]. Both series of data show a

120

4 Nanostructured Superhard Materials 1

I

very good agreement. The somewhat larger value of the crystallite size determined from the HR-TEM is probably due to the fact that the experimentalist easily overlooks small crystallites in the HR-TEM micrographs, in particular if these are in the shadow of the larger ones. XPS studies have shown that the silicon is bonded as Si3N4[63,101] and is always amorphous [61,62]. The available EDX does not allow us to measure the nitrogen content. Thus, the content of silicon nitride [Si3N4]corresponds to (7/3)-times the measured content of silicon [Si]. This scale is used in Fig. 7 which shows a similar dependence of the average TiN crystallite size on the Si3N4 content for films deposited in an abnormal d.c. glow discharge (see apparatus in Fig. 5) where the surface of the growing film is exposed to energetic ion bombardment. The measured hardness of the films is shown as well. One notices that the TIN crystallite size initially decreases with increasing content of Si3N4,passes a minimum, and increases afterwards again. The hardness shows exactly the opposite behavior reaching a maximum at the Si3N4content of about 20mol.% when the TiN crystallite size approached its minimum. Such behavior has also been found for other systems studied so far, ncW2N/aSi3N4[64] and ncVN/aSi3N4 [66,67] and it appears to be a universal phenomenon in these systems. For the understanding of the development and stability of the nanostructure, which will be discussed in the following section, it is important to realize that all other deposition parameters, such as temperature, discharge current density, partial pressure of all gases including TiC14 but the SiH4 (which was the only variable parameter) are kept constant in this series of experiments. Thus, all these ncM,N/ aSi3N4 systems choose the minimum crystallite size at the Si3N4content of about 20 mol.% because this minimizes their total Gibbs free enthalpy.


[ 01

0

10 20 30 a - Si,N, content [mol %]

121

2000

I

Figure 7. Dependence of the average TIN crystallite size and of the hardness on the Si3N4content in composites deposited in a d.c. discharge.

One might speculate whether the initial decrease of the crystallite size is a result of a decrease of the TiN (or more generally M,N) growth rate because of a decrease of the supply of the Ti gaseous source when the SiH4 flow is increased. However, this can be almost certainly ruled out because the nucleation rate remains high and almost constant since the large supersaturation changes only by less than I % in these experiments when the small flow rate of SiH4 is varied between 0 and 5 1 sccm as compared to 20-100 sccm. of the total gas flow. This is supported by the fact that the deposition rate remains fairly constant even if the Si3N4 content is varied from 0 to almost 40mol.% (see Fig. 3a in [63]). Moreover, even if this speculative explanation should hold, it can never explain the increase of the crystallite size with increasing the Si3N4content above about 20 mol.%, which is close to the percolation threshold in three dimensional f.c.c. system [120]. The crystallite size of TIN should continuously decrease with increasing content of Si3N4if the development of the microstructure were kinetically controlled. Therefore, we conclude that the minimum crystallite size together with the concomitant maximum hardness of the ncTiN/aSi3N4 composites is a result of a (relative?) thermodynamic stability of such a nanostructure at the percolation threshold (see remark added in proof and ref. [170]). We recall that at this percolation threshold the HR-TEM and XRD data show that the nanostructure consists of isolated TiN nanocrystals with a nearly spherical shape and only very few nanocrystals touching each other [61,62], that is, there is no indication of bicontinous or an interwoven bi-phase systems. Such a nanostructure with the minimum crystallite size has a maximum specific area of the interface. If, as experimentally observed, the system adjusts the minimum crystallite size at the percolation threshold resulting in the maximum specific area of the interface, this interface must possess an unusual stability. As such a behavior was also found with the other systems (ncW2N/aSi3N4

122


and ncVN/aSi,N4) it must be a result of some universal phenomenon. The possible nature of this phenomenon will be discussed later. The absence of dislocation activity in the nanocrystalline material, together with the absence of grain boundary sliding, as discussed above, should result in brittle behavior, so the measured hardness should be proportional to the elastic modulus. Both the hardness (Figs 6 and 7) and modulus (not shown here) of the composites were varied via the variation of the Si3N4content and the resulting hardness was plotted against the elastic modulus. For all three systems a linear relationship was found regardless of whether the deposition was done in the H F or d.c. discharge [63,64,66,67,102]. These results lend further support to the universality of this concept for the design of superhard nanocomposites. The films also show a fairly high elastic recovery of 280% upon indentation (see Fig. 2a and [64]), which is the result of the absence of dislocation activity and grain boundary sliding. Such a property is of great importance for applications as protective coatings on machining tools, because a local stress which arises when the cutting edge hits a grain of a hard precipitate in an alloy is distributed over a larger volume of the coatings thus decreasing the local load and increasing the material's resistance against fracture. A semiquantitative figure-of-merit is the ratio H" : Ey where the exponents x and y depend on the geometry of the tool and the machining conditions as well as on the theoretical model used for such an analysis [121]. Some researchers prefer to use the ratio H 3 :E 2 . This ratio approaches the value of 0.9 in our nanocomposites and is (compared with other hard materials) relatively high. In many applications the coating reaches a high temperature. For example, the temperature of the leading edge of cutting tools can be more than 800°C. With increasing cutting speed and decreasing consumption of coolant, which both increase the economical performance of the machining, this temperature increases. Therefore it is important to investigate the resistance of the coatings against oxidation in air and the stability of the mechanical properties at high temperatures. In our earlier study we found that the ncTiN/aSi3N4coatings are resistant to oxidation in air up to about 8OO"C, which is significantly higher than that for T i c (500°C) and TiN (600°C) and fairly comparable with Til -,Al,N [63,66,67]. The absence of creep and grain boundary sliding at room temperature does not exclude its occurrence at elevated temperature (see the deformation mechanisms maps [3,52-541). Because the equipment available for the measurements of hardness cannot be used at such high temperatures, and other measurements of the mechanical properties require much more time and labor, we have studied so far only the stability of the nanostructure and hardness upon isochronal annealing of the films in vacuum for 0.5 h at a given temperature. After each annealing of the sample the X-ray diffraction pattern and the hardness were measured, and the crystallite size was evaluated. Because such experiments are very time consuming, only preliminary data obtained from eight samples with average crystallite size chosen between about 3.5 and 9nm could be systematically studied. For films deposited on a steel substrate the maximum temperature is limited to about 900°C because at 1000°C the films peel off. In this series of measurements with three different samples with an average crystallite size of about 3.5 nm, both the crystallite size and the hardness of about


F

4-1 -(u3

1-

O0O0

4000

0

700

600

800

123

(I) v)

900

annealing temperature ["C] (a)

r 6000

61

4

-....

E

0

o-,

o

+

Tdep

044 I 0 600

'

o

o

0

-5000

E

-4000

-3000

o-,

I

'

I

1000

'

I

Y

; a,

-2000 0 09.07.97, [Si] 10.3%, [Cl] 0.9% '-1000 H = 2650 kg/rnrn2, d = 3.4prn silicon - substrate 800

$

5 m

0

1200

annealing temperature ["C] (b)

Figure 8. (a) Dependence on the crystallite size and hardness of ncTiN/aSi3N4films deposited on a stainless steel substrate on the annealing temperature (0.5 h for each step). Note that after annealing at 1000°C for 0.5 h the sample peeled off. (b) The crystallite size and the hardness of ncTiN/aSi3N4 films deposited on an Si wafer against the annealing temperature (0.5 h each step), see text.

5000 kgmm-* remained unchanged up to the highest temperature before the sample peeled off. One example of such measurements is shown in Fig. 8a. For reasons which are not fully understood yet [122] the hardness of the composites deposited on Si-wafers is always somewhat lower. With this in mind we show in Fig. 8b that for such samples both the crystallite size and the hardness do not change upon isochronal annealing for 0.5h for each step up to 1100°C and only a small coarsening is observed at 1150°C. Because Si3N4 decomposes under vacuum at this temperature (saturation pressure of about lop6atm [74]), the absence of coarsening (Ostwald ripening) is surprising. The lack of any change of the hardness is just a consequence of the stability of the nanostructure. Nevertheless, this clearly

124


indicates that the grain boundaries and the ncTiN/aSi3N4interface possess a surprising stability. In view of the fact that the maximum annealing temperature of 1100°Cwhere no coarsening occurred is about 0.7 Tdecat 1 atm of Si3N4,this finding merits more detailed discussion (see below). In contrast to these results, ncTiN/aSi3N4composites with a larger crystallite size of 2 6 nm show recrystallization accompanied by a decrease of the hardness already at 800°C. This suggests that the stabilizing effect of the ncTiN/aSi3N4 interface increases with decreasing crystallite size.

4.3.3 Other Superhard Nanocomposites and the General Validity of the Design Principle In this section we shall discuss further nanocrystalline composites in which the superhardness has been observed with emphasis on their practical applicability for industrial applications and on the general nature of the design principle. Let us start with the latter question. The fact that the same development of the nanostructure and hardness exceeding 5000 kgrnmp2 with increasing content of Si3N4as shown in Figs 6 and 7 was found also in the ncW2N/aSi3N4 system strongly supports the prediction that superhardness will be achieved in any other thermodynamically stable and hard transition metal nitride in a combination with Si3N4.The reason is easily seen by considering the thermodynamically driven segregation described by Eq. (10). WN is thermodynamically unstable at >5OO0C and 30 bar of N2, and W2N decomposes under vacuum above about 600°C [68]. Therefore, the occurrence of the segregation and the superhardness in the ncW2N/aSi3N4system shows that it will occur also with any other M,N, because for all thermodynamically stable metal nitrides with a strongly negative Gibbs free enthalpy of formation the equilibrium of reaction (10) will be shifted farther to the left hand side. This principle can be further generalized for any other nitride system which meets these conditions, and it will also apply to other systems, such as oxides, borides, carbides etc. showing the segregation. This is of great importance because this large variety of the possible materials and their combinations allows one to optimize the properties of the coatings, such as hot hardness, toughness, friction coefficient, chemical stability in air and against the material to be machined, as required for the given applications. This is illustrated by the following example. An apparent disadvantage of silicon nitride with respect to the applications of the superhard nanocomposites for machining tools, is the solubility of silicon in many materials including ferrous and aluminum alloys, superalloys [24], and others at high temperatures. In view of the surprisingly high stability of the ncTiN/aSi3N4 composites one cannot exclude the possibility that ncM,N/aSi3N4 will show sufficient stability against silicon dissolution in such applications. Nevertheless, both the question of the applicability of the superhard composites as well as of the universality of their design principle call for the investigation into further ncM,N/(a and nc)A,Bb systems.

4.3 Preparation and Properties of Superhard Nunocrystalline Composites

125

The choice of boron nitride instead of Si3N4 is one of the most attractive possibilities for several reasons: BN (in particular cBN) does not react with ferrous-, aluminum-, and many other alloys and is fairly stable against oxidation in air at elevated temperatures. Moreover, the equilibrium in the system described by reaction (1 1) is shifted much more to the left hand side than in the reaction (10). M,N

+ xBN

M,B,

1

x+l +2 N2.

(11)

This is due to the much higher stability of BN (AGf(1300K) = -68 kJ/mole and atom) at high temperatures as compared with Si3N4 (AGf(1300K) = -44.9kJ/ mole and atom) [74]. The boron nitride matrix possesses also a high structural flexibility because of the coordinational flexibility of boron (threefold in hBN, fourfold in cBN, and mixed in aBN). Furthermore, because TiB2 and many other transition metal borides are refractory materials, also the ncM,N/(a and nc)BN interface should have a high cohesive energy which is required for blocking the grain boundary sliding and achieving the superhardness. These general considerations were recently confirmed by experiments in which we have demonstrated that the hardness in the ncTiN/(a and nc)BN coatings reached 7000 kgmm-2. This is important particularly in comparison with the earlier results of other researchers to be briefly discussed here. Using a conventional plasma PVD (e.g. sputtering) at a low pressure where such conditions are not fulfilled leads to the formation of homogeneous phases (metastable solutions), such as Til_,A1,N (see e.g. [123-1251) or Ti-B-N [126-1281. Other authors have found phase segregation with the formation of binary phases TIN and TiB2 in the Ti-B-N coatings with relatively little nitrogen [129-1361 corresponding to ‘region 4’ of the equilibrium phase diagram [73,131]. These apparently contradictory results of different research groups can be probably attributed to different plasma conditions at the surface of the substrate during the deposition, that is, the above mentioned activation as well as to different film composition (mainly nitrogen content). Although these data are not quite clear from the published papers one can find there data supporting this hypothesis. Gissler, Mitterer, and coworkers found that the deposited Ti-B-N films with a relatively low nitrogen content formed two segregated binary phases TIN (or TiNl - J + TiB2 as expected from the equilibrium phase diagrams [129,131-1361. The hardness of the films depended on the composition showing a maximum for a composition of about Til B0.6p1N0,5 where the coatings had a very fine, almost isotropic nanocrystalline microstructure [ 1311331. The value of the maximum hardness depended on the substrate temperature of 3OW0O0C, ion energy of 126eV and and applied negative bias. For Tsubstrate the ratio of the ion to atom flux to the substrate of 0.7, the hardness reached about 50 GPa or slightly more for a nitrogen content of 5 15 at.% [134-1361. In general, the hardness as a function of the composition did not follow the rule-of-mixture but displayed a maximum at the given composition where the microstructure of the films was very uniform (no columnar growth) and the crystallite size was 3-5 nm. If the films were deposited at a nominal temperature of 20°C and post-annealed at 600-800°C, the hardness increased from 25-30 GPa to about 40 GPa after 30 min [132]. Similar values of superhardness were obtained if the films were deposited at about room temperature as ultrathin TiN/BN multilayers and post-annealed at

126


400°C for 20 h [131]. However, in spite of the significantly higher hardness as compared with TiN, the performance of these films in cutting tests was slightly worse than TIN and (TiA1)N coatings [ 131,132,1351. With increasing nitrogen content, when the hexagonal BN (hBN) phase was formed, the hardness of the films strongly decreased to 15-20 GPa for an overall composition of TiB2N2[135]. This decrease was attributed to the formation of the soft hBN. Andrievski has also found a strong hardness enhancement up to 52GPa in TiBo.7No.5 thin films with a grain size of about 3 nm [137,138]. The microstructural information regarding the possible segregation is unfortunately not available in his papers. However, the development of the morphology of the films as a function of the TiB2:TiN ratio (see Fig. 1 in [137]) is very similar to that found in our ncTiN/ aSi3N4and ncW2N/aSi3N4films (see Figs. 5 and 6 in [64]) thus suggesting that the microstructure in his TiB2-TiN films corresponds to a similar two phase system. In view of these results and the above discussed thermodynamic and kinetic considerations we have chosen deposition conditions which promote both, the phase segregation and the formation of cubic cBN which alone is ultrahard. This means using a high plasma density, high deposition temperature and high ion bombardment. The plasma CVD using halides, which can easily provide these conditions, has the advantage of a large flexibility for fundamental research but is not the best choice for large scale industrial processing (see the foregoing section). Therefore a chlorine free, combined plasma CVD and PVD technique was developed for the deposition of superhard ncTiN/BN coatings whose hardness reaches 70 GPa [99] (see also remark added in proof and ref. [171]). The list of possible superhard nitride nanocomposites can be further extended including further immiscible systems. For example, the above mentioned metastable solid solution Til -,Al,N coatings prepared by plasma PVD show a high stability and corrosion oxidation at elevated temperature (see [123-125,1391 and references therein). However, if deposited under conditions of a high activity of nitrogen, these coatings will undergo segregation into the TIN and AlN phases which, under an appropriate volume ratio, should also form a nanocrystalline superhard composite. The TIN may act as template for the pseudomorphic growth of the metastable cubic A1N [88,89,14&142] if the thickness of the TIN is sufficiently small. Similar effects have been found also in the TiN/NbN, TiN/CrN, and TiN/CN, heterostructures [88,89] and will surely be found in many others. Because CrN has similar tribomechanical properties as TiN but a significantly higher oxidation resistance (see e.g. [143,144] and references therein), nanocomposites based on this material in combination with BN, AlN, Si3N4 or other immiscible nitrides have a large technological potential. A very interesting template effect of TIN resulting in the development of a superhard nanocomposite microstructure was reported by Sproul, Y.-W. Chung and their co-workers for the TiN/CN, [96,97,145] and ZrN/CN, [90] systems. The researchers were depositing heterostructures by unbalanced magnetron sputtering from two opposing targets (Ti and graphite), between which the substrate was mounted on a turntable. The substrate temperature was not controlled, but the authors give 200°C as the upper limit. The thickness of the heterostructures was varied by changing the rotation speed of the turntable. When the apparent ‘layer

4.3 Preparution and Properties

of Superhard Nanocrystalline Composites

127

thickness’ (calculated from the deposition rate) decreased to 2 4 nm, the hardness suddenly increased to almost 5000 kg mmP2 and the small angle X-ray diffraction peaks from the layered structure vanished. The XRD showed a (1 11) texture of the films and the HR-TEM micrographs revealed that the material consisted of ncM,N/CN,x nanocomposite (M = Ti and Zr) with CN, pseudomorphic growth on the M,,N template. It is interesting to compare these TiN/CN, heterostructures and composites prepared by reactive sputtering at a relatively low pressure of about 0.01 mbar and temperature of 1200°C with the results obtained by plasma CVD at significantly higher pressure of 1-5mbar and temperature of 560°C [67]. Because carbon can substitute nitrogen in the TIN crystal lattice, TiC,N1-, forms a homogeneous solid over the whole region of 0 < x < 1. The higher deposition temperature and a high activation energy provided by the intense glow discharge plasma during plasma CVD resulted in the formation of a well mixed titanium carbonitride TiC,Nl --I. The hardness of the films plotted against the composition parameter, x, did not show any extremum, but followed monotonically the ruleof-mixture increasing from the hardness of TIN of about 2000 kg mm-2 to that of TIC of about 4000 kgmm-2 [67]. Knotek and coworkers [ 1151 reported on the preparation of ‘superstoichiometric carbides’ MCI+x, which, in fact, consist of nanocrystals of stoichiometric carbides imbedded in a aC:H matrix. With increasing excess, x,of the carbon, the hardness of the films increases, reaches a maximum of about 4000 kgmmP2 and, afterwards, decreases again upon a further increase of x in a similar manner as in our nanocomposites (Fig. 7). Similar results were recently reported by Voevodin and Zabinski who used laser ablation for the deposition of ncTiC/aC films [I 16,1171. The maximum hardness of about 32GPa was found for the total carbon content of about 65at.% [116]. Although this is below the range of the ‘superhard’ materials, the very high toughness of 70 N and low coefficient of friction of 0.15 [ 1171 are impressive. The hardness of stoichiometric amorphous boron carbide of 25000 kg mm-2 [ 1461 exceeds significantly that of the crystalline B13C3 (often incorrectly written as ‘B4C’ [147,148]) of about <3500kgmm-2. The fundamental unit of the crystalline BI3C3and of elemental boron is the B12boron icosahedron with a five-fold symmetry, which is incompatible with a closed packed three-dimensional crystalline lattice. Therefore, the icosahedra are connected via bridging carbon (in BI3C3) or boron (in cB) atoms [148]. The amorphous boron carbide consists also of the boron icosahedra which are, however, randomly oriented and bridged. Thus, aB,C (x M 4) can be considered as a composite consisting of randomly oriented ‘quasicrystals’ of the B-icosahedra which are imbedded in a carbon (or carbon rich) amorphous matrix. Figure 9 shows the dependence of the hardness of aB,C on the composition of the films [148,149]. The significant limitation regarding the applicability of these carbon containing superhard nanostructured materials is the high solubility of carbon in ferrous alloys (however, not in Al-alloys) and the low thermal stability of transition metal and boron carbides against oxidation in air. The stability of BI3C3is limited by the low melting point of the oxide of about 460”C, which allows a fast diffusion of oxygen and resulting oxidation of the film. In the case of composites containing

128


0

0

[C]:[B] in the gas inlet 1 2 3

4

6

x in B,C(s)

Figure 9. Hardness of amorphous boron carbide deposited from diborane and methane by plasma CVD against the composition (upper scale) and the C : B ratio in the gas inlet.

aC and CN, one has to keep in mind that these materials decompose at a temperature of 30O-50O0C, which limits their possible applications. These examples show that superhard nanocomposites of various composition can be prepared and their properties tailored according to the desired applications. Many of these nanocomposites may have limited applications due either to their chemical and structural instability at elevated temperature or their solubility (or chemical reactivity) with the machined material. However, the stable nanocomposites, such as ncTiN/aSi3N4,ncTiN/BN or more generally ncM,N/A,Nb will surely find interesting applications. Preliminary results from the dry, hard- and high-speed turning which are currently being done at our institute in collaboration with the SHM company are very promising (for more recent results see [171]).

4.4 Discussion of the Possible Origin of the Hardness and Stability of the Nanostructure As already discussed the practical hardness (and strength in general) of a material is determined by its microstructure. An appropriate design of the microstructure can easily result in an increase of the hardness by a factor of 4-7. The experimentally found superhardness of the ncM,N/aSi3N4 and ncTiN/BN nanocomposites of 5000-7000 kgmm-2 corresponds to an increase by a factor of 2.54, rather in the lower range. Also the hardness of 10000-13 000 kgmm-2, which we often measure on our composites, corresponds to an increase by a factor of 5-6.5. For the reasons

4.4 Discussion of the Possible Origin of the Hardness and Stability

of

the Nanostructure

129

outlined above, we do not consider these values of hardness as quantitatively unambiguous. It is, however, difficult to dispute the measured value of the elastic modulus of 1500 GPa (which was calculated from the linear part of the unloading curve and corrected for the elastic deformation of the diamond indentor; see Fig. 4 in [102]) as irrelevant because the linear part of the load-indentation-depth dependence is equivalent to the linear part of any stress-strain characteristic, that is, to the Hooke's law which defines the elastic modulus. Thus, the superhardness of the nanocomposites is due to the nanostructure, which efficiently eliminates the dislocation activity and grain boundary sliding. The question of the origin of the superhardness is thus reduced to the question of the formation and stability of the nanostructure and the high cohesive energy of the interface which avoids grain boundary sliding. The comparison of the ncTiN/CN, composites prepared by reactive sputtering with the TiC,Ni, + I carbonitride films deposited by plasma CVD clearly shows that, in this case, the nanocomposites are stabilized kinetically and they do not form under conditions of intense plasma with a sufficient availability of activation energy. On the other hand, the absence of Ostwald ripening in our nanocomposites with the smallest crystallite size and largest value of the hardness (see Fig. 8) strongly suggests that these materials may be thermodynamically stabilized. Such a hypothesis is surely provoking because, to best of our knowledge, there are no known grain boundaries with a negative Gibbs free energy of formation. The structural disorder within the grain boundaries and the incommensurability of the grain boundaries consisting of materials with a different crystal structure results in an excess lattice energy due to strain [150] and a positive Gibbs free energy of the grain boundaries, that is, lower thermodynamic stability of the nanocrystalline metals [151, 1521. As a result, all these systems show coarsening (Ostwald ripening) upon annealing to a temperature where the diffusion becomes fast enough, which is typically T 2 OST,,,, (Tm,"is the melting or vaporization temperature). Let us therefore discuss the possible explanation of the absence of the Ostwald ripening in our superhard nanocomposites as described above. The Ostwald ripening is a thermodynamically driven process due to a difference in chemical potential, &A,

= 2vM 7 0 (rl

-

r 2 ) / r 1 r2

(12)

( V , is the molar volume and "yo is the specific interfacial Gibbs free energy) between particles of different radii rl > r2. The smaller the particle, the higher is its Gibbs free energy of formation [I 53,1541. Upon the Ostwald ripening the system decreases its total Gibbs free energy due to the growth of larger particles at the expense of the smaller ones. In the quantitative treatment of the rate of this process two limiting cases can be distinguished: Either the transport (diffusion between the particles) or the reaction at the phase boundary is rate limiting. Let us first assume that the observed structural stability of the nanostructure, the absence of Ostwald ripening, is due to a slow diffusion [153,154]. In other terms, we assume, in contradiction to the experimentally found percolation behavior of these systems (see Figs. 6, 7, Fig. 10 in [155], and the discussion above), that the nanostructure is thermodynamically unstable and it forms due to the lack of diffusion

130


during the deposition of the films. Consequently, the experimental result that Ostwald ripening does not occur during the 30min annealing at 1100°C means that the mean diffusion length of Ti in Si3N4must be much smaller than the thickness of the aSi3N4tissue for the films at the percolation threshold, with the smallest crystallite size and highest hardness. From the Einstein formula [153,154]

(x), = J207 (13) we can estimate the upper limit of the diffusion coefficient D(1000°C)at the annealing temperature of 1100°C assuming (in accord with the absence of the Ostwald ripening) that during the annealing time 7 = 30 min the average diffusion distance (x), is much smaller than the average thickness of the aSi3N4tissue, about 0.5 nm. Thus, we obtain as an upper limit for the diffusion coefficient of Ti in aSi3N4 at 1100°C a value of D(1100"C) 5 10-22-10-20cm2s-1. Such a value is far too small for any system reported in the literature. As a rule, the lower limit of the coefficient for bulk diffusion in crystals at a temperature T = 0.57', is of the order of >10-15-10-'4cm2 s-' and for the grain boundary diffusion [23,156]. A value of much larger (lO-'cm s the order of ci2 was reported for the selfdiffusion of nitrogen in TIN (N15 exchange technique) at 0.5Tdecomp in [157] and 2 x lo-'' cm2 s-* for the interdiffusion between TIN and Si couple at 1100°C [158]. Because these values are many orders of magnitude larger than the above calculated upper limit of the diffusion coefficient D(1100°C) 5 lop2' cm2s-' we conclude that the experimentally observed absence of the Ostwald ripening due to a slow diffusion can be with certainty ruled out. Consequently, the reaction at the nc/a-interface must be rate controlling. This means that the titanium atoms do not dissolve in the amorphous Si3N4 tissue. Because many metals, such as Ni, Au, Ag, Al, Ga, and others diffuse easily through Si3N4 at elevated temperatures, and in view of the large number of vacancies in aSi3N4 at 1100°C (which is close to 0.77'") the lack of dissolution of Ti must be due to the stability of the ncTiN/aSi3N4 interface. This is in accord with the observation of the percolation threshold discussed above. Moreover, the increase of the TiN-crystallite size beyond the percolation threshold means that the thermodynamic driving force, which is causing the maximum specific area of the ncTiN/ aSi3N4 interface at the percolation, decreases with increasing thickness of the Si3N4tissue in that range. The results of annealing experiments with samples deposited either with a smaller or a higher Si3N4 content, that is, having a larger crystallite size of 6-7nm, show that such films coarsen already at a temperature of about 800°C. This result is particularly important for the sample with a higher Si3N4content of about 40mol.%, because the diffusion length of Ti is larger than in the stable films (Si3N4content of about 17-23 mol.%). If the systematic study, which is presently being done at our institute, confirms these results, there will be hardly any doubt left that the ncja interface of the material with the minimum crystallite size and maximum hardness is thermodynamically stabilized. With this limitation in mind let us hypothesize what could be the physical origin of such stabilization. The strong increase of the hardness occurs at a TiN-crystallite size of 545 nm and thickness of the aSi3N4-tissue of <1 nm, that is, in the range where surface

i"

4.4 Discussion of the Possible Origin of’the Hardness and Stability of the Nunostructure

0

.

0

1

4

4

4

100

120

131

.-K 0.008-

z

0.006-

0.0040.00220

40

60

80

crystallite size [A] Figure 10. Dependence of the measured tensile strain on the crystallite size for ncSi and ncTiN/ aSi3N4,see text.

energy strongly influences the mechanical and thermal properties (e.g. lattice dilatation or contraction [159,160] and a decrease of the melting point [160]) and the quantum localization phenomena dominate the optoelectronic properties of solids [ 155,1601. In thin films of nanocrystalline materials lattice dilatation has been found to increase with decreasing crystallite size for nanocrystalline silicon [159] as well as for ncTiN [63] and ncW2N [64]. These data are reproduced in Fig. 10. The lattice dilatation in films which do not have any significant biaxial stress has been explained due to the density defect in the grain boundaries and the anharmonicity of the interatomic potential [161]. (Our ncTiN/aSi3N4 films have a very small compressive stress of 0.2-0.5 GPa [ 1621. Energetic ion bombardment during the deposition can increase this stress to 5GPa [163].) Because the elastic lattice energy increases with the second power of the strain [22] it represents a destabilizing contribution to the total free energy of the system regardless whether it is compressive or tensile. Thus, there has to be some stabilizing contribution. Considering the Gibbs free enthalpy of the nanocomposite given by Eq. (14), the only possibility seems to be to assume that the stabilizing effect is associated with a high cohesive energy Eintof the ncM,N/aSi3N4 interface A G j ( d ) = A G $ ( C O ) + + N ~ T E E* ~N~r d 2 ( E $ +A&,,).

(14) Here, A G j ( d ) and AG;(co) are the Gibbs free energy of the nanocrystalline and coarse grained systems, respectively, the second term describes the destabilizing effect of the lattice strain (see Fig. 10) and the last term the stabilizing effect of the grain boundaries. N is the number of nanocrystals per mole, E$ > 0 is the is the entropy term of increase of the cohesive energy (dependent on d ) and the interface. The decrease of the surface energy of the ionic transition metal nitride by ‘wetting’ it with aSi3N4 may account for the initial decrease of the crystallite size with

132


increasing silicon nitride content, but it cannot explain the increase of the crystallite size at high Si3N4 content above the percolation threshold. Therefore, Eint must reach a maximum stabilizing effect at the minimum crystallite size and the thickness of the Si3N4interface at the percolation into separate ncTiN and aSi3N4 phases. With a further increase of the Si3N4 content this stabilizing effect must decrease and vanish, that is, E i t 5 E i y ) for d > d(min) M 3-3.5nm. Thus, for larger crystallite size of d > 5-8nm the destabilizing effect of the grain boundaries dominates and the thermodynamically driven Ostwald ripening (Eq. (12)) occurs. We can not offer any complete theoretical explanation of this phenomenon now, but only a hypothesis which may explain the experimentally observed results. The possible explanation is associated with the changes of the band structure due to quantum confinement phenomena in the nanocrystals and resultant increase of the cohesive energy at the interface. Indeed, a strong increase of the band gap was theoretically calculated [164,1651and experimentally observed in silicon [1661 as well as in compound semiconductors [160]. In metallic nanoclusters deposited on a carbon substrate, the conduction band narrows and the Fermi level shows a downshift [160,167]. This means, that the bonds which are formed at the nc/a interface by a combination of the states near the Fermi level of the nanocrystalline transition metal and the upper edge of the valence band of the 5 1 nm thin Si3N4 should also show downshift with decreasing crystallite size as long as the aSi3N4 tissue is thin. In such a case the system gains stability by increasing the specific area of that interface. However, when the fraction of Si3N4 increases above the value corresponding to the percolation threshold, its thickness increases and the band gap decreases towards the bulk value. Thus, the stabilizing effect at the nc/a-interface is lost and there is no driving force which would maximize its area. Consequently, the crystallite size increases again. The observed minimum of the crystallite size at the percolation threshold corresponds to a minimum of the total free energy of the system, that is, to a balance between this stabilizing effect at the interface and the destabilizing increase of the elastic lattice energy associated with the strain. This hypothesis, of course, needs further theoretical development and experimental verification. One may imagine other effects which could explain the one or the other observation. For example, the stability of the nanostructure might be explained by a strong immiscibility of the TIN and Si3N4phases in a similar way as it is believed to apply for heterostructures prepared of materials which show high immiscibility [89,95,168,169]. However, such an explanation is in contradiction with the phase diagram of the TiN/Si3N4 at high temperature and low pressure of nitrogen. Furthermore, it cannot explain why the ncTiN/aSi3N4 films with a large crystallite size coarse at much lower temperature and why, during the deposition, the crystallite size passes a minimum at the Si3N4 content corresponding to the percolation threshold. For similar reasons also a strain induced decrease of the diffusion in the ncTiN/aSi3N4 is unlikely to play any important role. Of course, one should keep in mind that the high structural stability of the nanostructure at high temperature is not necessary condition for the occurrence of the superhardness. Any phenomenon which avoids the grain boundary sliding will

4.4 Conclusions

133

result in the superhardness of nanostructured materials. Recently, we have found the superhardness also in ncTiN/aSi3N4 films with crystallite sizes of about 610 nm, when they were prepared under different plasma conditions (lower current density and lower voltage at the substrates during the deposition). These films, although superhard, recrystallize at about 800°C. We mention these observations here just to emphasize that there may be also other phenomena which avoid the grain boundary sliding, at least at room temperature. One example is the template effect which was observed, for example, in the ncTiN/ncAlN and ncTiN/ncCN, systems mentioned above. Any such effect can easily result in an enhancement of the strength and hardness by a factor of 4-8 and thus in superhardness of nanocomposites (ncja as well as ncjnc) with a crystallite size of the order of lOnm or less. The above discussed explanation of the high stability against coarsening at elevated temperatures applies probably only to ncja nanocomposites, where the percolation behavior as seen in Figs 6 and 7 is found. The high thermal stability of such nanocomposites may lead to a high hot hardness and toughness whereas that of superhard nanocomposites with d = 6-10 nm or more may deteriorate due to the thermally activated grain boundary sliding. Future research will, hopefully, help to clarify these questions.

4.4 Conclusions The practical strength and hardness of materials can be increased by an appropriate control of the microstructure, which hinders the multiplication and movement of dislocations and growth of microcracks. One such possibility is the decrease of the grain size which leads to a decrease of the dislocation activity and of the size of microcracks and an increase of the strength and hardness by a factor of 3-8. However, a new deformation mechanism, grain boundary sliding, becomes important in nanocrystalline materials when the grain size decreases below about 10-20 nm. Grain boundary sliding results in a softening of nanocrystalline metals typically for a grain size less than 6-10 nm. Grain boundary sliding can be avoided by an appropriate design of nanocrystalline composites, such as ncTiN/aSi3N4, ncVN/aSi3N4, ncW2N/aSi3N4,ncTiN/BN, and others (Si3N4and BN content of 17-23 mole.%), which have a small crystallite size of 3-9 nm and a high cohesive energy of the ncja interface. These nanocomposites show a high elastic recovery of 280% and Vickers microhardness exceeding 5000 kg mm-*. Presently available data indicate that these nanostructures are thermodynamically stabilized and do not show Ostwald ripening up to a temperature of 1100°C. Films with either a larger or smaller content of Si3N4 than that corresponding to the percolation threshold of about 20 mole.% have a larger crystallite size, smaller hardness and show coarsening (Ostwald ripening) by 800°C. A hypothesis based on a strengthening of the interface due to quantum confinement phenomena is suggested to explain these observations. For many practical applications, such as tribological coatings of machine tools

134


for ferrous metals, aluminum alloys, superalloys, and others, nanocomposites which do not contain silicon (which is soluble in such alloys), new ncM,N/a or ncA,Bb nanocomposites were suggested and their preparation investigated. Among others, ncTiN/a and ncBN nanocomposites, which were prepared by a combined plasma PVD and CVD technique without any halides, showed a very high hardness of 6000-7000 kg mm-2. Some theoretical aspects were discussed to show the generality of the principle for the design of superhard nanocomposites. The present review was complemented by a brief summary of the recent development in the field of heterostructures which represent another approach to the design of superhard nanostructured materials.

Acknowledgments I should like to thank the coworkers from my institute for their enthusiastic collaborations, my colleagues and friends Profs. Li Shizhi, H. Warlimont, V. Vitek, and many others for illuminating discussions, and my wife Dr. Maritza Vepiek-Heijman for her valuable comments on the manuscript. I am very much indebted to Drs. M. Jilek, M. Sima, and P. Holubar from the SHM Company for a very fruitful collaboration towards the industrialization of our superhard nanocomposite coatings. Partial financial support of our work by the German Science Foundation (DFG) and the Bavarian Ministry for Education and Science is thankfully acknowledged.

References 1. D. Tabor, The Hardness of Metals, Clarendon Press, Oxford, 1951. 2. Ch.-M. Sung and M. Sung, Muter. Chem. Phys. 1996,43, 1-18. 3. R. W. Hertzberg, Deformation and Fracture Mechanics of Engineering Materials, 3rd edn., Wiley, New York, 1989. 4. M. L. Cohen, Phys. Rev. B, 1984, 32, 7988-7991. 5. A. Y. Liu and M. L. Cohen, Science, 1989,245, 841-842. 6. M. L. Cohen, Solid State Commun. 1994, 92, 45-52. 7. M. L. Cohen, Mater. Sci. Eng. A , 1996, 209, 1 4 . 8. S. Vepfek, J. Weidemann, and F. Glatz, J . Vac. Sci. Technol. A , 1995, 13, 2914-2919. 9. Y. Zhang, Z. Zhou, and H. Li, Appl. Phys. Lett. 1996,68, 634-636. 10. Y. Zhang, private communication (1997). 11. Y . Chen, L. Guo, and E. G. Wang, Mod. Phys. Lett. B, 1996, 13, 615-617. 12. Y. Chen, L. Guo, and E. G. Wang, J . Phys.: Condens. Muter. 1996,8, L685SL690. 13. Y. Chen, L. Guo, and E. G. Wang, Philos. Mag. Lett. 1997, 75, 155-162. 14. E. C. Cutiongco, D. Li, Y.-W. Chung, and C. S. Bhatia, J . Tribol. 1996, 118, 543-548. 15. A. Khurshudov, K. Kato, and S. Daisuke, J . Vac. Sci. Technol. A , 1996, 14, 2935-2939. 16. X. Yun, R. C. Hsiao, and D. B. Bogy, IEEE Trans. Magn. 1997, 33, 938-943.

References

135

17. H. Sjostrom, S. Stafstrom, M. Boman, and J.-E. Sundgren, Phys. Rev. Lett. 1995, 75, 13361339; see also errata 1996, 76, 2205. 18. H. Sjostrom, L. Hultman, J.-E. Sundgren, S. V. Hainsworth, T. F. Page, and G. S. A. M. Theunissen, J . Vac. Sci. Technol. A , 1996, 14, 56-62. 19. W. T. Zheng, H. Sjostrom, I. Ivanov, K. Z. Xing, E. Broitman, W. R. Salaneck, J. E. Greene, and J.-E. Sundgren, J. Vac. Sci. Technol. A , 1996, 14, 2695-2701. 20. G. B. Bokii, N. F. Kirova, and V. I. Nepsha, Sov. Phys. Doklady, 1979, 24, 83-84. 21. A. F. Goncharov, Sov. Phys. Uspechy, 1987, 30, 525-534. 22. Ch. Kittel, Introduction in Solid State Physics, Wiley, New York, 1971. 23. E. Hornbogen and H. Warlimont, Metallkunde, Springer, Berlin, 1996. 24. A. Kelly and N. H. MacMillan, Strong Solids, 3rd edn., Clarendon Press, Oxford 1986. 25. H.-H. Behnke, Harterei Technische Mitteilungen, 1993,48, 3-10. 26. M. F. Doerner and W. D. Nix, J . Muter. Res. 1986, 1, 601-609. 27. T. F. Page and S. V. Hainsworth, Surf. Coat. Technol. 1993,61, 201-208. 28. W. C. Oliver and G. M. Pharr, J . Muter. Res. 1992, 7, 613-617 and 15641583. 29. K. G . Schmitt-Thomas, Metallkundejiir das Maschinenwesen, Vol. I & 11, Springer, Berlin, 1990. 30. E. 0. Hall, Proc. Phys. SOC.B, 1951, 64, 747-753. 31. N. J. Petch, J . Iron Steel Inst. 1953, 174, 25-28. 32. A. Lasalmonie and J. L. Strudel, J . Muter. Sci. 1986, 21, 1837-1852. 33. N. Hansen, Metall. Trans. A , 1985, 16A, 2167-2190. 34. U. F. Koks, Metall. Trans. 1970, 1, 1121-1143. 35. A. Inoue, H. M. Kimura, K. Sasamori, and T. Matsumoto, Muter. Trans. Jpn. Inst. Metals, 1994,35, 85-94. 36. H. Hahn and K. A. Padmanabhan, Philos. Mag. B, 1997,76, 559-571. 37. R. W. Siegel, Spektrum der Wissenschaft, 1997, 3, 62-67. 38. R. W. Siegel and G. E. Fougere, Muter. Res. Soc. Symp. Proc. 1995, 362, 219-229. 39. R. W. Siegel and G. E. Fougere, NanoStructured Muter. 1995, 6 , 205-216. 40. J. E. Carsley, J. Ning, W. W. Milligan, S. A. Hackney, and E. C. Aifantis, NanoStructured Muter. 1995, 5, 441448. 41. A. H. Chokshi, A. Rosen, J. Karch, and H. Gleiter, Scripta Metall. 1989, 23, 1679-1684. 42. G. W. Nieman, J. R. Weertman, and R. W. Siegel, Scripta Metall. 1989, 23, 2013-2018. 43. J. S. C. Jang and C. C. Koch, Scripta Metall. Muter. 1990, 24, 1599-1604. 44. K. Lu, W. D. Wie, and J. T. Wang, Scripta Metall. Muter. 1990, 24, 2319-2323. 45. G. W. Nieman, J. R. Weertman, and R. W. Siegel, Scripta Metall. Muter. 1990, 24, 145-150. 46. G. W. Nieman, J. R. Weertman, and R. W. Siegel, J . Muter. Res. 1991, 6, 1012-1027. 47. G. E. Fougere, J. R. Weertman, R. W. Siegel, and S. Kim, Scripta Metall. Muter. 1992, 26, 1879-1 883. 48. R. Z. Valiev, F. Chmelik, F. Bordeaux, G. Kapelski, and B. Baudelet, Scripta Metall. Muter. 1992,27, 855-860. 49. A. M. El-Sherik, U. Erb, G. Palumbo, and K. T. Aust, Scripta Metall. Muter. 1992,27, 11851188. 50. (a) P. G. Sanders, C. J. Youngdahl, and J. R. Weertman, Muter. Sci. Eng. 1997, A234, 77-82; (b) P. G. Sanders, J. A. Eastman, and J. R. Weertman, Acta Muter. 1977, 45, 40194025; (c) C. J. Youngdahl, P. G. Sanders, J. A. Eastman, and J. R. Weertman, Scripta Muter. 1997,37, 809-813. 51. J. Musil and F. Regent, J . Vac. Sci. Technol. 1998, A16, 3301-3305. 52. M. F. Ashby, Acta Metal. 1972, 20, 887-897. 53. H. Luthy, R. A. White, and 0. D. Sherby, Mafer. Sci. Eng. 1979, 39, 211-216. 54. F. A. Mohamed and T. G. Langdon, Metall. Trans. 1974,5,2339-2345. 55. J. Crampon and B. Escaig, J . Am. Ceramid Soc. 1980, 63, 686686. 56. R. C. Gifkins, Metall. Trans. 1976,7A, 1225-1232. 57. J. W. Edington, Metall. Trans. 1982, 13A, 703-715. 58. S . Yip, Nature, 1998, 391, 532-533. 59. J. Schiotz, F. D. Di Tolla, and K. W. Jacobsen, Nature, 1998, 391, 561-563. 60. W. W. Milligan, S. A. Hackney, M. Ke, and E. C . Aifantis, NanoStructured Muter. 1993, 2, 267-276.

136

4 Nunostructured Superhard Materials

61. S. Vepiek, S. Christiansen, M. Albrecht, and H. P. Strunk, Muter. Res. SOC.Symp. Proc. 1997, 457,407412. 62. S. Christiansen, M. Albrecht, H. P. Strunk, and S. Vepfek, J. Vac. Sci. Technol. B, 1998, B16, 19-22. 63. S. Vepfek and S. Reiprich, Thin Solid Films, 1996, 268, 64-71. 64. S. Vepiek, M. Haussmann, and S. Reiprich, J. Vuc. Sci. Technol. 1996, A14, 4 6 5 1 . 65. S. Vepfek, Proc. of the 10th Znt. Conf. on Thin Films, Salamanca, Spain, September 1996, Thin Solid Films, 1998, 317,449-454. 66. S. Vepiek, M. Haussmann, and Li Shizhi, Electrochem. SOC.Proc. 1996,96-5,619-628. 67. S. Vepfek, M. Haussmann, S. Reiprich, Li Shizhi, and J. Dian, Surf. Coat. Technol. 1996,86/ 87, 394401. 68. G. W. Samsonow, Nitridi, Naukova Dumka, Kiev, 1969 (in Russian). 69. H. Holleck, J. Vuc. Sci. Technol. A , 1986, 4, 2661-2669. 70. H. Fischmeister and H. Jehn, Hurtstofschichten zur Verschleissminderung, DGM-Informationsgesellschaft, Stuttgart, 1987. 71. S. Sambasivan and W. T. Petuskey, J. Muter. Res. 1994, 9, 2362-2369. 72. T. Hirai and S. Hayashi, J. Muter. Sci. 1982, 17, 1320-1328; 1982, 17, 3336-3340; 1983, 18, 240 1-2406 73. P. Rogl and J. C. Schuster (Eds.), Phase Diagrams of Ternary Boron Nitride and Silicon Nitride Systems, ASM International, Materials Park, OH, 1992. 74. M. W. Chase, C. A. Davies, J. R. Downey, D. J. Fruip, R. A. McDonald, and A. N. Syverud, JANAF Thermochemical Tables, 3rd edn., Phys. Chem. Re$ Data, 1985, 14, No. 1. 75. J. S. Koehler, Phys. Rev. 1970, B2, 547-551. 76. S. L. Lehoczky, J. Appl. Phys. 1978,49, 5479-5485. 77. S. L. Lehoczky, Phys. Rev. Lett. 1978, 41, 1814-1818. 78. C. A. 0. Henning, F. W. Boswell, and J. M. Cobert, Acta Metall. 1975, 23, 193-197. 79. R. F. Bunshah, R. Nimmagadda, H. J. Doerr, B. A. Movchan, N. I. Grechanuk, and E. V. Dabizha, Thin Solid Films, 1980, 72, 261-275. 80. S. Menezes and D. P. Anderson, J. Electrochem. SOC.1990, 137, 440-444. 81. U. Helmersson, S. Todorova, S. A. Barnett, J.-E. Sundgren, L. C. Markert, and J. E. Greene, J. Appl. Phys. 1987, 62, 481-484. 82. M. Shinn, L. Hultman, and S. A. Barnett, J. Muter. Res. 1992, 7, 901-911. 83. X. Chu, M. S. Wong, W. D. Sproul, S. L. Rohde, and S. A. Barnett, J. Vac. Sci. Technol. 1992, A10, 1604-1609. 84. K. M. Hubbard, T. R. Jervis, P. B. Mirkarimi, and S. C. Barnett, J. Appl. Phys. 1992,72,44664478. 85. P. B. Mirkarimi, L. Hultman, and S. A. Barnett, Appl. Phys. Lett. 1990, 57, 2654-2656. 86. P. B. Mirkarimi, S. A. Barnett, K. M. Hubbard, T. R. Jervis, and L. Hultman, J .Muter. Res. 1994,9, 1456-1467. 87. K. K. Shih and D. B. Dove, Appl. Phys. Lett. 1992,61, 654-656. 88. W. D. Sproul, Surf. Coat. Technol. 1996,86/87, 170-178. 89. W. D. Sproul, Science, 1996, 273, 889-892. 90. M. L. Wu, X. W. Lin, V. P. Dravid, Y. W. Chung, M. S. Wong, and W. D. Sproul, J. Vac. Sci. Technol. 1997, A15, 946950. 91. S. A. Barnett, in Physics of Thin Films, Vol. 17, Mechanics and Dielectric Properties, M. H. Francombe and J. L. Vossen (Eds), Academic Press, Boston, 1993, p. 2. 92. J. E. Krzanowski, Scripta Metall. Muter. 1991,25, 1465-1470. 93. X. Chu and S. A. Barnett, J. Appl. Phys. 1995, 77,44034411. 94. S. A. Barnett, Invited Lecture presented at the TSM Annual Meeting, San Antonio, Texas, February 1998 unpublished (see also [168, 1691). 95. P. M. Anderson and C. Li, Nunostructured Muter. 1995,5, 349-362. 96. D. Li, X. Chu, S. C. Cheng, X. W. Lin, V. P. Dravid, Y. W. Chung, M. S. Wong, and W. D. Sproul, Appl. Phys. Lett. 1995,67,203-205. 97. D. Li, X. W. Lin, S. C. Cheng, V. P. Dravid, Y. W. Chung, M. S. Wong, and W. D. Sproul, Appl. Phys. Lett. 1996, 68, 1211-1213: 98. T. L. Selinder, M. E. Sjostrand, A. Ostlund, M. Larsson, S. Hogmark, and P. Hedenqvist,

References

131

Proc. 14th Int. Plansee Seminar, G. Kneringer, P. Rodhammer, and P. Wilharditz (Eds), Plansee AG, Reutte, Austria, 1997, p. 44. 99. S. Vepiek, P. Nesladek, A. Niederhofer, F. Glatz, M. Jilek, and M. Sima, 25th Int. Conf. on Metallurgical Coatings and Thin Films, San Diego, April 1998, Surf. Coat. Technol. 1998, 108-109, 138-147. 100. S. Vepiek, S. Reiprich, and Li Shizhi, Appl. Phys. Lett. 1995, 66, 2540-2642. 101. Li Shizhi, Shi Yulong, and Peng Hongrui, Plasma Chem. Plasma Proc. 1992, 12, 287-297. 102. S. Vepiek, A. Niederhofer, P. Nesladek, and F. Glatz, Electrochem. SOC.Proc. 1997, 97-25, 3 17-324. 103. A. von Engel, Ionized Gases, 2nd edn., Clarendon Press, Oxford, 1965. 104. S . C. Brown, Introduction to Electrical Discharges in Gases, Wiley, New York, 1966. 105. G. Francis, Ionization Phenomena in Gases, Butterworth, London, 1960. 106. S. Vepfek, J. Th. Elmer, Ch. Eckmann, and M. Jurcik-Rajman, J. Electrochem. SOC.1987,136, 2398-2405. 107. S. Vepiek, Ch. Eckmann, and J. Th. Elmer, Plasma Chem. Plasma Proc. 1988,8,445466. 108. S. Vepi-ek, C. Brendel, and H. Schafer, J. Cryst. Growth, 1971, 9, 266-272. 109. S. Vepiek, in Current Contents in Materials Science, E. Kaldis (Ed.), North-Holland, Amsterdam, 1979, pp. 151-236. 110. J. Patscheider, Li Shizhi, and S. Vepiek, Plasma Chem. Plasma Proc. 1996, 16, 341-363. 11 1. M. Diserens, J. Patscheider, and F. Levy, 25th Int. Conf. on Metallurgical Coatings and Thin Films, San Diego, April 1998, Surf. Coatings Technol. 1998, 108-109, 241-246. 112. F. Vaz, L. Rebouta, S. Ramos, M. F. da Silva and J. C. Soares, 25th Int. Conf. on Metallurgical Coatings and Thin Films,San Diego, April 1998, Surf. Coat. Technol. 1998, 108-109,236240. 113. H. Jensen, J. Sobota, and G. Sorensen, Surf. Coat. Technol. 1997, 94/95, 174178. 114. H. Jensen, J. Sobota, and G. Sorensen, 25th Int. Conf. on Metallurgical Coatings and Thin Films, San Diego, April, 1998, Surf. Coat. Technol. unpublished. 115. 0 . Knotek, private communication, 1997. 116. A. A. Voevodin and J. S. Zabinski, J . Muter. Sci. 1998, 33, 319-327. 117. A. A. Voevodin and J. S . Zabinski, Diamond Rel. Mater. 1998, 7, 463467. 118. H. P. Klug and L. E. Alexander, X-Ray DzfSraction Procedures, Wiley, New York, 1974. 119. M. Riickschloss, PhD Thesis, Technical University, Munich, 1993. 120. R. Zallen, The Physics of Amorphous Solids, Wiley, New York, 1983. See chapter 4, in particular Figure 4.5for a two dimensional network and Table 4.4. Notice the following important facts: ( I ) The structural percolation corresponds to the situation when the percolating phase (here Si3N4)forms a continuous connection (‘cluster’) through the matrix phase (here TIN). It occurs at a critical fraction of the percolating phase (here aSi3N4)Pc which depends on the nature (site or bond percolation), structure see e.g. the f.c.c., b.c.c., and diamond lattices in Table 4.4) and dimensionality of the network (here TIN). Pc = 1 for one dimensional chain, 0.354.7 for two and 0.124.4 for three dimensional network. For the f.c.c lattice of TiN Pc N 0.2 for the site percolation as observed in the present work. (2) Because of different near-threshold scaling laws, the percolation phenomena of electrical and optical properties, such as conductivity and optical reflection, commence only at a fraction of the percolating phase which is significantly larger than P, (see e.g. Figure 4.5 and Table 4.3 in [120] and the change of the reflectivity of the ncTiN/aSi3N4 films reported in [67]). (3) Even above the percolation threshold (P(Si3N4)> Pc) there still may be some local connectivity within the matrix (nc-TiN). In our particular case it means, that some of the TIN nanocrystals can be in contact with each other. 121. T. E. Fischer, private communication, 1998. 122. So far, only films deposited on Si show a lower hardness. Films deposited on hard metal (Co cemented tungsten carbide) or on such a substrate precoated with TiN or Ti, -.Al,N show the same or somewhat larger hardness as those deposited on steel. For both steel and silicon substrates the crystallite size and the Si3N4content are equal. The possible reasons of this effect are under investigation. 123. 0. Knotek, M. Bohmer, and T. Leyendecker, J. Vac. Sci. Technol. 1986, A4, 2695-2700. 124. T. Leyendecker, 0. Lemmer, S. Esser, and J. Ebberink, Surf. Coat. Techno/. 1991, 48, 175178.

138


125. Y. Tanaka, T. M. Giir, M. Kelly, S. B. Hagstrom, T. Ikeda, K. Wakihira, and H. Satoh, J . Vac. Sci. Technol. 1992, A10, 1749-1756. 126. 0.Knotek, R. Breidenbach, F. Jungblut, and F. Loffler, Surf. Coat. Technol. 1990,43144. 107115. 127. 0. Knotek and F. Loffler, J . Hard Muter. 1992,3, 29-38. 128. 0.Knotek, A. Schrey, J. W. Schultze, and B. Siemensmeyer, WerkstofSKorros. 1992,43,511519. 129. C. Mitterer, M. Rauter, and P. Rohdammer, Surf. Coat. Technol. 1990,41, 351-363. 130. M. Tamura and H. Kubo, Surf. Coat. Technol. 1992,54155,255-260. 131. W. Gissler, Surf. Coat. Technol. 1994, 68/69, 556563. 132. P. Hammer, A. Steiner, R. Villa, M. Baker, P. N. Gibson, J. Haupt, and W. Gissler, Suyf. Coat. Technol. 1994, 68/69, 194-198. 133. T. P. Mollart, M. Baker, J. Haupt, A. Steiner, P. Hammer, and W. Gissler, Surf. Coat. Technol. 1995, 74/75, 491496. 134. P. Losbichler, C. Mitterer, P. N. Gibson, W. Gissler, F. Hofer, and P. Warbichler, Surf. Coat. Technol. 1997,94195, 297-302. 135. W. Gissler, M. A. Baker, J. Haupt, P. N. Gibson, R. Gilmore, and T. P. Mollart, Diamond Films Technol. 1997, 7, 165-170. 136. C. Mitterer, P. Losbichler, F. Hofer. P. Warbichler, P. N. Gibson, and W. Gisler, Vacuum, 1998,50, 313-318. 137. R. A. Andrievski, J . Solid State Chem. 1997, 133, 249-253. 138. R. A. Andrievski, J. Muter. Sci. 1997, 32, 4463-4484. 139. J. G. Han, J. S. Yoon, H. J. Kim, and K. Song, Surf. Coat. Technol. 1996,86-87, 82-87. 140. I. Petrov, F. Adibi, J. E. Greene, L. Hultman, and J.-E. Sundgren, Appl. Phys. Lett. 1993,63, 3638. 141. F. Adibi, 1. Petrov, J. E. Greene, L. Hultman and J.-E. Sundgren, J. Appl. Phys. 1993, 73, 858G8589. 142. (a) M. Setoyama, A. Nakayama, T. Yoshioka, T. Nomura, A. Shibata, M. Chudou, and H. Arimoto, Surnitomo Electric Industries, 1995, 146, 91-95; (b) M. Setoyama, A. Nakayama, M. Tanaka, N. Kitagawa, and T. Nomura, Surf. Coat. Technol. 1996,86-87,225-230. 143. G. Berg, Ch. Friedrich, E. Broszeit, and Ch. Berger, Surf. Coat. Technol. 1996,86-87,184-191; 1996,3, 1597-1602. 144. J. Vetter, R. Knaup, H. Dwuletzki, E. Schneider, and S. Vogel, Surf. Coat. Technol. 1996,8687, 739-147. 145. Y.-W. Chung, Surf. Rev. Lett. 1996,3, 1597-1602. 146. S. Vepfek and M. Jurcik-Rajman, Proc. of 7th Int. Symp.on Plasma Chem., C. J. Timmermans (Ed.), Eindhoven University of Technology, Eindhoven, 1985, pp. 90-94. 147. G. Will and K. H. Kossobutzki, J. Less Common Metals, 1976, 47, 4348. 148. N. N. Greenwood and A. Earnshaw, Chemisty of the Elements, Pergamon Press, Oxford, 1984. German translation, VCH-Verlag, Weinheim, 1990. 149. S. Vepfek, Plasma Chem. Plasma Proc. 1992, 12, 219-235. 150. H. P. Aubauer and H. Warlimont, Z . Metallkunde, 1974, 65, 297-303. 151. F. Gartner, R. Bormann, R. Birringer, and A. Tschope, Scripta Muter. 1996, 35, 805810. 152. I recommend reference 11511for more details because it shows that some of the earlier papers predicting a higher stability of the nanocrystalline metals were incorrect. 153. H. Schmalzried, Chemical Kinetics of Solids, VCH, Weinheim, 1995. 154. H. Schmalzried, Solid State Reactions, Verlag Chemie, Weinheim, 1981. 155. S. Vepfek, Thin Solid Films,1997, 297, 145-153. 156. D. Gupta and P. S. Ho (Eds), Diffusion Phenomena in Thin Films and Microelectronic Materials, Noyes, Park Ridge, NJ, 1988, p. 43. 157. Hj. Matzke, in Difusion in Metals, A. L. Laskan, J. L. Bocquet, G. Brebec, and C. Monty (Eds), Kluwer, Dordrecht, 1990, pp. 429455. 158. W. Wakelkamp, PhD Thesis, Technical University, Eindhoven, 1991. 159. S . Vepfek, Z. Iqbal, and F.-A. Sarott, Philos. Mag. B, 1982,54, 137-145. 160. A. P. Alivisatos, Science, 1996, 271, 933-937.

References

139

161. S. Vepiek, F.-A. Sarott, and Z. Iqbal, Phys. Rev. B, 1987, 36, 3344-3350. 162. P. Nesladek and S. Vepiek, unpublished results, 1998. See also [170]. 163. S . Vepfek, in Proc. of Muter. Res. Soc. Europe, Strasbourg, 1984, P. Pinnard and S. Kalbitzer (Eds), Les Cditions de physique, Les Ullis, France, 1984, pp. 425444. 164. C. Delerue, M. Lannoo, and G. Allan, J . Lumin. 1993, 57, 249-256. 165. B. Delley and E. F. Steigmeier, Phys. Rev. B, 1993, 47, 1397-1400. 166. S. Vepiek and Th. Wirschem, in Handbook of Optical Properties Vol. ZZ: Optics of Small Particles, Znterfaces and Surfaces, R. E. Hummel and P. Wissmann (Eds), CRC Press, Boca Raton, 1997, pp. 129-143. 167. S. B. DiCenzo and G. K. Wertheim, in Clusters of’ Atoms and Molecules ZZ, H. Haberland (Ed.), Springer Verlag, Berlin, 1994, pp. 361-383. 168. S. Barnett and A. Madan, Physics World, 1998, 11, 4548. 169. A. Madan, Yun-yu Wang, S. A. Barnett, C. Engstrom, H. J. Ljungcrantz, L. Hultman, and M. Grimsditch, J. Appl. Phys. 1998, 84, 776-779. 170. A. Niederhofer, P. Neslidek, H. Mannling, S. Vepfek and M. Filek, Surf. Coatings Technol. (1999) in press (Proc. Int. Conf. Metallurgical Coatings and Thin Films, San Diego, April 1999). 171. P. Holubai, M. Jilek and M. Sima, Suyf. Coatings Technol. (1999) in press (Proc. Int. Conf. Metallurgical Coatings and Thin Films, San Diego, April 1999). 172. A. Voevodin, ibid.

Remarks added in proofs. Since the submission of the manuscript in 1998 many new publications appeared about the industrialization of the (TiA1)N/aSi3N4nanocomposites [ 1711, thermal stability [ 1701 of the ncTiN/aSi3N4 and nc(TiAC)N/aSi3N4, on the ultra-tough nanocomposites ncTiC/aC and ncTiC/aC + MoS2 [ 1721 and others. Microhardness of 80 to 2 l00GPa which is in the range of diamond, was achieved in ultrahard nanocomposites consisting of nc-TiN/a-Si3N4/a-2nc-TiSiz [ 1701 when the surface of the TIN nanocrystals is covered with about one monolayer of ‘Si3N4’[170].


5 Corrosion of Hard Materials K. G. Nickel and Y. G. Gogotsi

5.1 Introduction Corrosion may be defined as the physical and chemical alteration of a material due to its interaction with the environment of interest. It must be emphasized that corrosion resistance is not a material property but a system property and real environments have a high variability in both chemical and physical boundary conditions. Corrosive alteration leads to a change in the desired original properties and so corrosion usually, but not exclusively, has adverse effects and causes a decrease in strength, toughness, hardness, and increases wear. There are many reasons for the degradation of properties by corrosion. Decrease of the cross-section carrying the load, pitting, stress induced by phase changes, and grain separation are some of the common features. Some hard materials, namely Sic and Si3N4,are designed to serve at high temperatures, where corrosion resistance is an essential feature. This is the reason why a large proportion of the ceramic literature on corrosion, including recent textbooks [ 1-51, deals with these materials. In order to react with the environment, the material must not be in equilibrium with it. The hard materials covered in this chapter are non-oxides and hence this condition is fulfilled for all of them in almost every application environment, including ambient air. The applicability of hard materials thus rests on the kinetics of the processes. Before discussing individual materials we will introduce important terms and concepts in corrosion like attack modes, media classes, location of attack, kinetics, and their parameters. We feel that this is necessary for understanding of the discussion of the materials, because there is no generally accepted standard procedure that allows us to compare the corrosion behavior of materials as easily as other properties like mechanical strength or hardness.

5.2 Corrosive Media The compositional variability of environments is so vast that only very few attempts have been made to classify corrosion behavior from the point-of-view of the discrete environment at hand. A noteworthy exception is the Dechema handbook [6],in which valuable information may be found if the corrosive agent is simple and known (e.g. acetates, chlorine, steam). However, its main thrust is on the corrosion of metals.

5.3 Corrosion Modes

141

The more common classification scheme is to divide the corrosive media into their state of aggregation, that is to subdivide into corrosion by solids, liquids and gases. While solid state corrosion is rarely dealt with, we have vast amount on literature on hot gas corrosion. The case of corrosion by liquids is commonly further subdivided into more specific cases, such as aqueous corrosion (e.g. acids and water), corrosion by glasses, metal melts and salt melts. The last case is for historic reasons known in the form of a rather misleading expression: hot corrosion. A special case, which spans from the liquid into the gaseous state is given by the corrosion in hot water systems: hydrothermal corrosion. At first glance this may not make too much sense, but the experience is that the different names mentioned above correspond to main forms of the attack mode and often refer to different application temperature levels.

5.3 Corrosion Modes 5.3.1 Active and Passive Corrosion A fundamental distinction between corrosion modes is the division into active and passive corrosion. The active mode is characterized by the loss of material to the environment and results in the decrease of size and weight of the specimen. The loss may be in form of gaseous or dissolved species. The passive mode denotes a process, where the material reacts with the environment to yield a new condensed phase, often in form of a layer or scale on the surface. In most cases it is associated with a weight and/or component size gain. In the strict sense, the passive mode implies that the newly formed phase is retarding the process, that is corrosion is slowed down with time (protective scale), but there are cases in which the scale is nonprotective: a scale with cracks, low viscosity, or foamy texture may not hinder the access of the corrosive agent to the substrate.

5.3.2 Homogeneity and Location of Attack: Internal, External and Localized Corrosion If there is a perfectly homogeneous surface attack and a removal of material or a scale growth is taking place solely and uniformly on the surface of the material, the process is denoted external corrosion. This is most likely if we have a perfect single-phase material. As soon as grain boundaries in general or secondary phases appear as discrete particles or grainboundary phases the process is likely to become more complex, because each phase of the assemblage will react differently to the environment at hand. An easily envisaged form of the problem is the attack of oxygen on a S i c containing discrete free carbon particles. At high temperature the carbon reacts in an active mode (formation of CO,T), while the Sic forms condensed Si02.Inhomogeneous or

142

5 Corrosion of Hard Materials

local corrosion with bubble and pit formation is the result, at least in the early stages of this oxidation [7]. If the secondary phase is a grain-boundary phase, which allows a relatively fast transport of the corroding agent into the material via the grain boundaries, we will have an attack of the main material and/or the grain-boundary phase not only on the surface, but in a zone stretching out into the bulk of the specimen. This is known as internal corrosion. Despite its localized character on a microscopic scale, intergranular corrosion may be quite homogenous on the macroscopic scale. Additionally, the process of internal corrosion is also capable of slowing down with time, so it may offer some protective quality. It must be emphasized that those processes do not exclude each other. Internal corrosion may occur while a scale is formed on the outside. We will discuss the problem in the context of the individual materials.

5.4 Corrosion Kinetics 5.4.1 Physical Boundary Conditions In modeling the kinetics of corrosive processes and quoting numbers of kinetic constants it is usually inherently implied that we have steady-state conditions with a constant activity of the corrosive agent in the environment and an infinite reservoir of the material to be corroded. The first condition is met in many technical applications, where the corrosive medium is flowing, the second one may be true or not. But even the first condition is violated for some corrosion conditions, such as when a substance is deposited on the surface of the material by a physical process and can hence be treated as an attack by a substance which was implanted at a given amount without replenishing. Exact corrosion kinetics must be modeled by solving the second law of Fick for the geometry of the case at hand. However, in some cases a net effect may be calculated from simple thermodynamics, as for closed system conditions in active corrosion [8]. For the case of diffusion through scales it has been demonstrated that quasi-steady-state modeling is often a good approximation for an exact solution, at least for conditions t D / x 2 > 2 [9] (where t = time, D = diffusivity, x = layer thickness). Some basic solutions for situations with instant singular corrosion can also be found in the literature [lo]. But even under steady-state conditions there is a profound influence of physical boundary conditions on corrosion behavior. The most widely known example of this is the boundary between active and passive oxidation of silica-formers. The classic modeling has been done by Wagner (1 11 for silicon. The line of argument is as follows. Active oxidation of silicon occurs by the formation of gaseous SiO and passive oxidation by the formation of silica: Si

+ 402H SiOt

(1)

5.4 Corrosion Kinetics

143

and Si + O2 H Si02. Calculations of the oxygen partial pressure P(0,) for the boundary condition both by simply taking the values from the JANAF-tables [12] or by using a modern thermochemical program [ 131 yield similar results with an equilibrium P(0,) as low as bar at 1000°C. The prediction from the equilibrium would thus be, that at conditions with P ( 0 2 ) > lop2*bar we should find the formation of a silica layer on a bare Si surface. Experimental observation shows that this only takes place at P(02)-levels more then 20 orders of magnitude higher. Wagner’s solution was to transform the problem into a physically controlled one. Silica is formed at conditions, which allow its condensation from the gas phase in equilibrium with the substrate:

4Si +

Si02

* SiO.

(3)

If reaction (1) is fast, a concentration gradient of O2 inwards and SiO outwards must exist. Then the interdiffusion coefficients and the effective thickness of the boundary layer will control whether the critical P(Si0) of the condensation reaction (3) is reached or not. From tables and estimations of those physical constants we can relate back to the oxygen pressure of the bulk gas necessary to induce the critical P(Si0) on the surface. The calculation yielded reasonable results for the activepassive boundary of Si in streaming atmospheres with low oxygen contents and accordingly the theory was later applied to other silica-formers [ 14,151. This success story demonstrates that equilibrium calculations disregarding physical boundary conditions can be misleading, and that thermochemical calculation is capable of giving good results if those boundaries are taken into account.

5.4.2 Active Corrosion Kinetics Under steady-state conditions the loss of material to the environment by decomposition into gaseous species or dissolution is usually a reaction or a diffusional problem. For reaction control we have the flux J as J = krci,

(4)

where J = mass flux, k , = reaction coefficient, ci = concentration of agent at interface, and for first-order reactions a linear law follows under steady-state conditions. The basic shape of the kinetics for diffusion control can be seen by looking at Fick’s first law J=

-D-dc

dl ’

where I = diffusion path length. With constant conditions, the gradient from the surface to the bulk of the corrosive medium becomes constant, 1 stays constant and accordingly the flux, J , becomes constant with time. Therefore the change in

144


10

h

- 5

f

Y

c

.-0

In In

a 0

0

2 v

-8 -5

0

1

2

3

4

5

6

7

8

log (time [s]) Figure 1. Predicted behavior of a model case for active corrosion under different external conditions.

size or mass (Ax) is also linear with time t:

AX = kit. (6) This expected linearity is usually experimentally confirmed. A theoretical prediction of the size of the h e a r rate constant kl in E q . (6) for a given temperature and composition of medium and material is possible, when thermodynamic data are available to calculate the reaction, and boundary parameters such as the geometry of the sample, viscosity, stream velocity, Schmidt and Reynold’s numbers are known or can be estimated [16]. Other approaches to active corrosion prediction utilizing thermochemical calculations [8,17,18] require the experimental determination of effective parameters. They show both the importance of physical boundary conditions and the extremely low level of partial pressures at which active corrosion is potentially dangerous. In Fig. 1 a model case applying such simplified kinetic equations [17] is illustrated. Thermodynamic calculations for Sic in H2 yield partial pressures of about lo-’ bar for SiH4 and CH4 even at moderate temperatures of 1 100°C, so that the reaction Sic + 4H2 w SiH4 + CH4

(7)

is used as the dominant reaction. Geometric boundary conditions (a tube of 1 m length with an inner diameter of 30 cm) allow us to calculate a volume flow through such a pipe from a given stream velocity. The calculated material recession after a given time ranges from less than a monolayer in the closed system to more then 70mm per year for ideal, and hence gas velocity independent, evaporation. The model for intermediate cases indicates that both mild and conservative applicability limits for high speed civil transport systems [18] are not met even at low gas velocities.


145

It is interesting to note that different approaches [17,18] result in applicability limits, which specify partial pressure limits of
5.4.3 Basic Passive Corrosion Kinetics Passive corrosion with scale formation on the substrate is often more complicated. The simplest form is that of reaction control. If a reaction is slow compared to the delivery of the agent to the reaction site, the concentration of the agent at the interface is as high as the external concentration of the corrosive medium. For a given temperature this reaction rate constant, k , dictates then the linear rate constant k~ in Eq. (6) and linear kinetics prevail. Only this time Ax is positive and denotes a layer growth or a mass gain. Likewise Eq. (6) is correct, if the chemical reaction rate constant is high (= fast reaction) and the growing scale is not hindering the agent from access to the substrate. This situation is most easily envisaged in a scale with plenty of shortcircuits like cracks or open porosity. Here the concentration gradient develops from the substrate into the bulk of the environment. Accordingly the analysis is identical to that for active corrosion kinetics above, which gave the linear kinetics. Another simple case is a homogeneously growing single-phase layer, which allows only a slow diffusional transport of corrosive medium to the substrate. Here the steady state means for Eq. (5) that a layer, and hence the diffusion path length 1, is growing with time while the absolute concentration difference Ac stays constant. At the substrate-oxide interface, the equilibrium with the material is achieved or the concentration of the agent can be approximated to zero, while the concentration at the surface of the scale is constant at the solubility limit of the scale material. The law appropriate for this situation as derived from Fick's first law is called the parabolic law due to diusional control and introduces the parabolic rate constant k,, which has units of m2 s-' or kg2 m-4 s-' , depending on the definition of Ax as scale growth or mass gain:

(ax)2 = k,t

(8)

or

ax=

fi.

(9)

The derivation of the parabolic law above is only a good approximation for a situation with a sufficiently thick pre-existing scale. It has been shown in a classic paper by Deal and Grove [19] on the oxidation of silicon, that the correct parabolic law should recognize the chemical reaction rate and the gas transport coefficient along with the diffusion coefficient. Their analysis demonstrated that the parabolic rate law is rather

(ax)2 + axA = B ( t +).

(10)

146


with a linear ‘short-term’-solution for t << T and a parabolic ‘long-term’ solution identical to Eq. (8) for t >> T ( A , B, T = constants). It should be noted that all those constants vary with temperature and that the so called ‘help time’, T , can be of appreciable length: For silicon at 1700°C we have a time of at least several tens of hours, during which the kinetics are basically linear! A nice review of the complexity of the ‘simple’ parabolic oxidation of Si, particularly in its initial stages, has been given by Irene [20]. A third basic law for passive corrosion comes from the analysis of the situation, where the reaction product formed on the material is completely blocking the further access of the corrosive agent. This law was derived from the study of metals at moderate temperatures. It is easily envisaged: If every particle formed protects the substrate completely, we have a situation, in which the surface area available for reactions is diminished rapidly. The mathematical form following is the logarithmic law

with the logarithmic rate constant klog.The same equation is valid for a more complex situation, pertinent to ceramic systems. If the diffusion through a scale is rate controlling and the scale crystallizes a phase, which is either completely blocking for the agent or has a diffusion coefficient much smaller than the original scale, the effective cross section for the agent transport is likewise diminished. For a constant crystallization rate we will end up with Eq. (1 1). However, natural nucleation and crystal growth is complex, dependent on many variables and usually not constant with time. The energy requirements are often so strongly temperature dependent that we see either none or full crystallization. Thus in pure systems we will often have kinetic breaks with time or temperature rather

-

h

u)

._

c

=I

P 2 c

300

e

2.200 ln Q) c

u)

Y

.-0

Jt>. m -

100

0 0

50

100

150

200

Time (arbitrary units) Figure 2. Basic passive laws.

250

3 I0


147

then a nice curve following Eq. (11) exactly. This may be different for systems in which phase formation or crystallization is linked to chemical changes of the scale, induced by diffusion of components from the interior. The shapes of curves with behavior according to the basic laws is illustrated in Fig. 2. It is obvious from Fig. 2 that data from a single point (here: t = 28 and x = 100) would not allow prediction of the behavior at any other time unless the law is specified.

5.4.4 Kinetic Breaks The basic laws of Section 5.4.3 are for constant growth under constant conditions. However, a scale growing according to the parabolic law may change to a different behavior only after some time has elapsed. There are a number of reasons for such kinetic breaks, shown in Fig. 3 . Again the most easily envisaged case is a crystallizing scale. Crystallization is often accompanied by a volume expansion, which may induce stresses in the scale and lead to spalling. The renewed fast access of the medium to the substrate accelerates reaction rates. If spalling is connected to a critical thickness, we find pseudo-linear growth built from individual parabolic segments (Fig. 3). If the first break is not repaired by the following growth, for example if the scale is not spalling but permanently cracks, we find break-away behavior, where parabolic behavior changes to linear. We may also find breaks which slow down corrosion, such as if a scale crystallizes only after some critical thickness or time is reached without disintegration. A parabolic-logarithmic behavior may follow. Note that this break is not apparent to the naked eye unless very long times of observation are involved.

Figure 3. Common kinetic break types.

148


*Oo0 n

=!

I

1500-

cu

Y

m

P

Y

1000-

.-0 5

5

-

500-

0

20

40

60

80

100

10

Time (a.u.) Figure 4. Growth change due to developing diffusion coefficients, D.

5.4.5 Complex Kinetics The real corrosion behavior may be far more complex then the ideal laws suggest, because of changes in properties of the scale with growth and the operation of differing types of corrosion at the same time. Changes in the properties may be microstuctural changes like the development of bubbles, pits, or internal porosity, where effective surfaces or effective diffusion path lengths are changed. Other common changes are those of the effective diffusion coefficient. If a scale is not built from a phase of constant composition like the SiOz formed on Si, but contains elements from external impurities or internal secondary phases, which are concentrated or diluted with time, the diffusion coefficient will change with these changes. Accordingly the overall behavior must deviate from parabolic behavior. Such changes are illustrated in Fig. 4. Note that the increase of D with time makes the shape more linear with time, while the decrease in D brings it to a curve more like the logarithmic law. The development of scales on highly porous media has been reported to show a typical break of growth speed with increasing temperature. The model case shown in Fig. 5 involves indeed no true kinetic break, the parabolic rate constant increases regularly with temperature and is identical for short and long times. The accepted reason for the shape of the 'high-T' curve in Fig. 5 is the pore closure with the associated breakdown of the effective surface. A typical factor is of the order of 300 if a sample with a geometric surface of 1 cm2 has a specific surface of 0.1 m2/g-' in the porous state. A relatively simple analytical treatment for complex growth patterns has been suggested by Nickel [21]. It works on the assumption that the basic laws of


149

30

Time (a.u.) Figure 5. Typical development of layer growth with increasing temperature in porous media with pore closure due to passive corrosion.

growth can operate simultaneously. Growth according to

Ax = k,t +

+ klogIog(t)

is evaluated by stepwise multiple linear regression. The stepwise procedure was adopted to give fit constants with more physical meaning. The fit allows quantification of individual contributions of the basic laws and is illustrated in Fig. 6. Other attempts to model complex kinetics have been put forward by Ka11 et al. [22]. Both complex modeling methods do not incorporate true kinetic breaks and thus need improvement in order to be used as a general growth law.

0

500

I000

1500

Time (a.u.) Figure 6. Model for complex growths kinetics after Eq. (1 1) [21] with individual contributions.

150


6 - 4

? m

Y

% 2

c m c 0 0

E

.-0) a 3 -2 -4 0

20

40

60

80

Time (a.u.) Figure 7. Weight change curves obtained from the simultaneous action of parabolic weight gain and two different linear weight loss rates (‘para-linear behavior’).

An important special case of complex kinetics is the simultaneous action of basic passive laws together with active corrosion. A typical case for this is when a scale grows but is consumed at the same time by evaporation. The resulting shape of curves is shown in Fig. 7 and has been described as ‘para-linear’ [23] behavior. It may be analyzed with Eq. (12) omitting the logarithmic term. It should be noted that small variations in the linear loss rate constant give macroscopically different behavior: the curve labeled ‘data 1’ in Fig. 7 might wrongly be interpreted as ‘hardly any change after a short period and ‘data 2’ as ‘active corrosion after an incubation period’.

5.5 Corrosion Measurement 5.5.1 Experimental Methods The most common type of corrosion experiment is the after-treatment test, that is investigations on property changes are done after the end of the test by any of a variety of means (weight, dimensions, penetration, phase change, retained strength etc.). The limitations are discussed in the section on corrosion data (5.5.2). For the corrosion of ceramics by ordinary liquid media, the testing is usually done by immersion tests: the sample is placed in a heated retort or autoclave inside a stirred excess of corrosive medium. Common test conditions are at the boiling point of the medium (e.g. 10% H2S04, 30% NaOH) for a week. The corrosion by melts is classically studied by the methods of the refractory industry like drop-, crucible-, and finger tests [24]. While those methods allow an

5.5 Corrosion Measurement

151

easy ranking they are hardly capable of producing quantitative data outside the narrow conditions set by the standardized experimental set-up [25]. For hot corrosion testing burner rigs can be applied [26].The agent is injected into the flame of the burner and the sample is held or agitated at a distance, which allows deposition of the corrosive agent from the gas phase at a constant rate. In the ordinary furnace test for gas corrosion the samples are placed in a furnace for a given time in the atmosphere of interest. The advantage of the furnace test is that many different samples can be tested in one set of conditions, the drawback that it gives only one data point in time per experiment. The pitfalls are apparatus related: Temperature and atmosphere constancy have to be guaranteed, crosscontamination must be avoided, and interaction with furnace linings and crucible material minimized. The danger of those points is often underestimated; there are reports of different results from a mere change in the furnace size [27]. Contamination by furnace linings is a source of error for all corrosion experiments. There is ample evidence that the commonly used linings of A1203-ceramics do contaminate samples with A1 and alkalis via gas phase transport, even if highpurity A1,03 is used. The contamination is strongly enhanced in wet atmospheres [281. In situ measuring of corrosion is mostly done by thermogravimetric measurement, that is the monitoring of a weight change during a isothermal period of time. The advantage here is that the true shape of changes is recorded, which allows proper kinetic modeling. The limitations of the method come firstly from the apparatus: Resolution, accuracy, precision, and the often unchecked longterm stability are major points of concern. The pitfalls of contamination by furnace interior and crucibles, and local atmosphere changes are identical to those mentioned above. For coarse estimations of danger levels in corrosion nonisothermal thermogravimetry or differential thermal analysis (DTA) can be used.

5.5.2 Corrosion Data From the descriptions above the reader will accept that there are plenty of problems of defining data and parameters for the characterization and comparison of corrosion resistance. Standardization is under discussion [29], but is not generally accepted. Only for very specific corrosion cases are the first norms and recommendations available [30-341. This is the reason why even excellent data collections on advanced ceramics [35], become very fuzzy when corrosion is concerned. We will discuss here briefly the different data types. For a general case we first look for the changes in dimensions (Fig. 8). There are three principal changes: a thickness of a scale growing on the substrate, a change in the overall size of the component, and penetration depths. The most interesting parameter from the point-of-view of an engineer would be the penetration depth, because it would allow evaluation of the effective cross section of the component after corrosive attack and hence prediction of the remaining strength.

152


layer thickness

corroded size

Figure 8. Principal dimensional changes during corrosive attack.

For homogenous surface attack by active corrosion the case simplifies to the recession as there is no scale growing and the remaining cross section is of virgin material. The parameter of interest is then the recession rate, which is identical to kl of Eq. (6) for linear kinetics. In corrosion science the SI-units ms-' or the more vivid unit mmyear-' of engineering practice are useful. The relation to mass loss is obtained by multiplying by the density of the lost material to yield kgmp2s-' or equivalent units. It should be emphasized again that this parameter is useless unless the physical boundary conditions are specified. The quotation of partial pressures of outgasing species is often more meaningful. A problem arises when the attack is localized, because the attack is then different for different phases and the outer surface is no longer identical to the virgin material. The evaluation of penetration depth and rate may be performed in some cases using dye tests. If penetration goes with formation of porosity, dyes like methylene-blue or phosphorescing agents can be applied to show the depth of corrosion in cross sections. More complex situations, such as with chemical modification of the grainboundary phase, may be measured in optical thin sections, but they need usually more careful evaluation by chemical profiling with microanalytical techniques. Note that the kinetics of active corrosion may then deviate strongly from linear kinetics because leaching by liquid media or evaporation out of channels often involves diffusional problems. A k1 is then not the parameter for ranking. A better choice would be the reference to critical effects, such as the 'time needed for 0.1 mm penetration at 0.7 T,' [36]. Regrettably, such parameters are extremely scarce in the literature. As the experimental determination and the investigations afterwards are highly time-consuming and expensive, an often followed method of characterization is by the effect of corrosion on mechanical properties, usually strength or wear properties [37,38], in addition to thermogravimetric data.

5.5 Corrosion Measurement

153

Changes (a.u.)

0.0 \

\

\

\

I

Moving to the more complex cases with scale growth, Fig. 9 shows the range of kinetic curves one may find. Figure 9 shows a parametric study for a single case of para-linear behavior. Not even the sign of the curves is constant. We must therefore discuss the individual contributions. For linear scale growth the discussion above is valid. For processes of retarded scale growth with time we describe the layer growth by k, (units of m2 s-l or equivalents) for simple parabolic behavior (Eq. (8)). The calculation of the appropriate mass change unit is not as straightforward as in the linear case. This is because the mass change recorded by weighing or thermobalance is measuring the difference in masses and not the mass of the layer. The latter may be inferred if the chemical reaction and the properties of the scale material are known, introducing a stoichiometric factor u. As an example the reaction Sic

+ 1 4 0 2 = SiO2 + CO

(13)

means that I mole of the condensed product Si02 (= 1Msio2= 60.09g with a density psio2 of 2 . 2 g ~ m -in ~ the amorphous state) is produced from 1 mole of S i c (= 40.09 g). Thus a recorded mass change of Ax = nMproduct- nkfmaterial (= 19.99g) in Eq. (12) represents the production of a full mole of S O 2 . In general it follows that the factor u is defined from the reaction equation as (nM)condensed

U =

(nM)condensed

product

product

- (nM)material

'

(14)

which is 3 for reaction (13). Because of the form of Eq. (8) the two k, are hence related by

[ y1

Pproduct 2

[$1

k,(layer-thickness) -

3

= kp(mass-change) U2

154


Neglecting the factor u has lead to occasional inconsistencies in the literature. The discussion of it seems trivial but a close look at Eq. (14) shows that for other cases, for example SiBCN3

+ 2.25 O2 H S O 2 + 4B203+ CO + 1.5N2,

(16)

u can become very large with the result that substantial scale production may occur almost invisible to thermogravimetry. The layer thickness derived from any of the kinetic equations may be recalculated to give a solution for homogeneous recession:

From this discussion one might conclude that layer thickness measurements are the better choice for monitoring passive corrosion. While this is true from the point of resolution (ellipsometry allows the determination of scales down to the nm scale) it is difficult for high-temperature in situ recording and it implies dense transparent layers, which may not be the case. Thus a safe characterization will ask for both quantities to be measured independently. Inconsistencies between measured and calculated values from Eq. (15) will then immediately point to problems concerning scale structure and/or internal corrosion. Additional recording of mechanical property changes has the advantage of including information on potentially dangerous new flaw populations derived by corrosion. The overlay of several parameters (flaw geometry, cross-section changes, stress field changes) will not allow us to use mechanical testing as the only criterion. Thus, at the present time we prefer to compare and rank materials in terms of time needed for penetration to a certain depth at the temperature of interest to make any type of kinetic equation accessible. The fixed ‘critical condition’ is taken to be 0.1 mm, following Ashby and Jones [36], the time unit is in years to give a measure for engineering practice. An engineer may then be able to decide that a material is not useful for long-term applications if it drops to log t, < 0 at the temperature of interest where tc is the critical time in years for a penetration of 0.1 mm. In the following sections on the materials this is calculated but the uncertainties have to be kept in mind.

5.6 Materials 5.6.1 Diamond and Diamond-like Carbons Diamond is known to resist all acids and most other chemical substances in the low-temperature range to 100°C. However, transformation of diamond into other carbon phases may occur under local contact stress at room temperature [39,401

5.6 Materiuls

155

Diamond will be etched by caustic alkalis and some oxidizing salts, for example by NaClO and KOCl at 380°C [41], and K and Na nitrates above 400°C. Diamond reacts with metals, which form carbides (e.g. W, Ta, Ti, and Zr) or which dissolve carbon (e.g. Fe, Co, Mn, Ni, and Cr) [42]. Detailed knowledge of the temperatures of reactions with metals is reviewed in the context of experiments with metals for the catalysis of diamond synthesis [43]. Diamond corrosion is usually a problem of oxidation and graphitization. Graphitization in inert atmospheres was observed at about 1500°C and becomes very fast at 2100°, where sizable diamonds are converted to graphite within minutes [44]. An extrapolation of the data gives a T M 1450°C for t, = 0. It has been reported that graphitization is prevented in pure H2 up to 2200°C [45]. Surface graphitization associated with concurrent oxidation, which may be found at T > 600°C, is not thought to be a true conversion but an autocatalytic surface reaction involving CO, species. The oxidation of diamond is clearly an active corrosion process. At least up to 700°C diamond has a fast reacting { 11 1)-plane, an intermediate { 1 lo} and a slow { 100)-plane, which indicates reaction control. At higher temperatures and/or lower oxygen pressures gas diffusion becomes rate determining in analogy with graphite [46],and this is indicated by a more even attack [47]. Hence corrosion rates are faster or start at lower temperatures for fine powders compared to films and the corrosion in air is faster than in low-oxygen environments [48]. At 720°C diamond will burn in an oxygen jet. It is often stated in the literature that oxidation starts at 600°C. The theoretical onset of oxidation should be well below 600"C, because from the etching rates of Evans [44] for the { 111)-plane in 1 bar oxygen a t, well below zero ( ~ 0 . 0 2 )is calculated. Linear extrapolation would give T M 340°C for tc = 0. However, in analogy with graphite there may be an accumulation of C-0 complexes at the surface, which is faster then their removal between 300°C and 500°C [49]. Surface graphitization at T > 600°C points in this direction. Also in analogy with other carbon forms, impurities may act as catalysts for the oxidation, which can decrease oxidation temperatures by more then 100°C and increase rates by orders of magnitude [50]. To a first approximation the temperature sensitivity of the oxidation rate at T < 800°C is less dependent on the structure of diamond (films) then on crystallographic orientation, the data for the diamond faces bracket those of thin layers [47]. As described in more detail in the contribution on hydrothermal synthesis of diamond in this book, hydrothermal reactions start for fine powders at 7OO0C, but a significant etching and dissolution of large diamond grains is observed only at T 2 800°C [51].

5.6.2 Carbides The carbides considered as hard materials include B4C, WC, and Sic. In oxidizing environments, in particular, their behavior is so different that they can not be compared directly.

156


5.6.2.1 Silicon Carbide Compilations of corrosion data on Sic-based materials have been given in some hand- and textbooks [2,35,52,53], many papers on it are included in the proceedings of specific meetings [3,5,54]. 5.6.2.1.1 Corrosion by Liquid Media There is only limited information about the behavior of pure Sic in low-temperature water, acids, and bases. A relatively pure sintered S i c is applied in many pumping systems for the chemical industry [%].This provides evidence that it is hardly affected by chemical attack. Significant attacks can occur via a grain-boundary phase in other Sic types, and this is usually accompanied by a decrease in strength and Weibull modulus. Thus it is clear that liquid-phase sintered S i c will show less corrosion resistance with a behavior depending on the chemistry of the additives. For Y/Al-sintered S i c significant corrosion was reported for a treatment in boiling 20% HC1 and 50% NaOH; concentrated acids (HZSO4, H3P04)and bases (KOH) had much less effect [56]. Sic with free Si (Si-Sic) is affected in particular by hot strong bases (10M NaOH) and hot diluted acids (H2SO4 at pH = 3) [57]. 5.6.2.1.2 Hydrothermal Corrosion In contrast to the good behavior of S i c at lower temperatures a strong attack can be obtained in supercritical water. At pressures of 100MPa there is some indication from powder experiments that initial reactions start at 300°C [58, 591, strong corrosion is observed at T > 500°C [60,61]. The lower temperature onset is supported by tribological studies [62]. The main mechanism is the formation of silica from Sic along with CH4, COXf C. The silica is then dissolved in HzO. The dissolution rate of silica will play a vital role in the kinetics of the process. Basically the attack should have active corrosion character (Eq. (6)). Chemical modification of the water to increase the pH will enhance the dissolution of silica [63] and hence further increase corrosion of Sic. 5.6.2.1.3 Corrosion by Metals Sic can be attacked by a number of metals, because often silicides and in some cases other carbides are stable phases and eutectic temperatures are low. From [52] we have evidence of reaction with Bi (600"C), Li (815"C), Mg (8OO0C), A1 (750"C), Ti (575"C), Ta (12OO0C), Nb (1300"C), Cr (lOOO"C), Mo (1200"C), W (1500"C), Ni (500"C), Co (1150"C), Fe (lOOO"C), Cu (950"C), Pd (600"C), and Au (950°C). Resistance to attack was noted for Na (350"C), Ca (1 180"C), Cd (500"C), Sn (600"C), Pb (815"C), and Ag (962°C). These data have to be treated with some caution, because many results are obtained from treatment in vacuum without a protective scale present. The kinetics of the reaction have been described as parabolic for a number of cases. However, they depend also on physical conditions like the wetting behavior of the liquid metal or a formed melt [64]. Thus, it is obvious that Sic with free Si will react more readily than Sic. Temperatures of possible reactions may be evaluated with the aid of phase diagrams for the appropriate Si-metal system.

5.6 Materials

157

From own experience we know that Pt can be used as crucible material in oxidation studies of Sic. However, in vacuum the formation of melt and silicides at the eutectic of the system (= 1400°C) is so fast that individual Pt grains drill deep holes into the material within seconds. The use of Sic as a hard material in wear application, can thus be endangered by a reaction with the metal if the silica scale is eroded by the process and contact temperatures are high enough. 5.6.2.1.4 Corrosion by Salt Melts (Hot Corrosion) Data and reviews of the mechanisms of the hot corrosion behavior of Sic have been presented in a number of papers by Jacobson and coworkers [65-681. A representative member of the hot corrosion agents is Na2S04. Its generation is explained in combustion environments by 2 NaCl

+ SO2 + 0.5 O2 + H 2 0 H Na2S04+ 2 HCl.

(18) Since the action in the liquid state is regarded as most effective, the first condition for the window of hot corrosion is given by the melting and dewpoint of Na2S04. The melting point of 884°C does not vary much, the dewpoint depends on pressure, S-content of the fuel and NaC1-concentration in the atmosphere. For conditions with NaCl > 1 p.p.m. and combustion pressures of 100 bar it is calculated to be 1100-1200°C from Eq. (18). However, the effective corrosion action is generally believed not to be by the sulfate itself. It stems from the activity of N a 2 0 arising from the decomposition Na2S04

* N a 2 0 + SO3

(19)

because only the N a 2 0 is attacking the protective scale of Si02 by

-

x Si02 + N a 2 0 H N a 2 0 (SiO2).,.

(20)

This is the reason why other compounds releasing alkali oxide on decomposition, such as Na2C03,have similar hot corrosion effects. It should be noted that complex salts, in particular those containing vanadium, can have an even stronger hot corrosion effect, as is well known from metals and oxides [69,70]. From Eq. (19) it can be calculated that only for conditions with a low P(SO3)will we have a dissociation sufficient to drive reaction (20) to the right side. Consequently the window of hot corrosion has a third axis, the P ( S 0 3 )of the combustion, and it has been computed [71] that the window should only open for low-S fuels (<0.5%, a condition met by many modern fuels) and at comparatively high temperatures (FZ>IOOO"C). In reality the attack already starts at the melting point of Na2S04. This is attributed to internal controlling factors. Many Sic-based materials contain free carbon, which promotes the dissociation of Na2S04 and makes the silicate melt more basic. The dramatic attack of Na2S04 on S i c with free carbon has been confirmed experimentally [72]. As is described in more detail for Si3N4,recent studies suggest that free carbon is an enhancing factor but not a necessary condition [73,74]. S i c may act as the reducing agent to promote sulfate dissociation either directly or via an oxygen gradient in the silica scale consumed by reaction (20).

158


It is very difficult to quantify hot corrosion in laboratory experiments with preloaded samples, because the salt melts propagate rapidly both by spreading and gas phase transport, resulting in uneven coverage of the sample and changes with time. A much better constancy of conditions is achieved in burner-rig methods [75-781. However, even here a quantification is difficult as the usually highly porous scales containing sulfate, silicate and silica are liable to spalling. Furthermore, the attack is often in the form of pitting, and so there is a change in exposed surface and the attack is nonuniform. The statistical nature of pit distribution is probably not due to a distribution of secondary phases but the result of the action of surface tension effects, because two immiscible liquids (sulfate and silicate) with differing surface stresses are present [74]. Therefore, while weight changes of chemically cleaned corroded samples are good indications for the strength of attack, they remain unsatisfactory and a widely used criterion is by the influence of corrosion on mechanical properties. Even for cases with good control of chemical changes, a prediction of the effect on strength can not follow directly, because the fractures occur usually from the newly formed pits (Fig. 10 [78]). None the less, as seen in Fig. 10, the relation between pit size and failure is evident for most Sic materials. A model for the prediction of properties from corrosion conditions would thus have to give an expression for the formation and growth of pits with time, temperature, and external properties. The author is not aware of a fully consistent model to yield this result. Thus we are left with individual studies. A study on commercial Sic [78] yielded the typical result that a reduction in strength of 20-30% is obtained after 4 0 h exposure at 1000°C.

500

2

z d

5E

(ac)"', pm 200 100

1

50

400

--0.75K,c (a~)"~

300

;200

c

3 c

0 Type A 0 Type B

A Type C

0

m

t

1

100

0 Type D

0.1 0.2 (Pit dimensions)-'", (ac)-'", pm-"'

0.3

Figure 10. Relation of fracture stress and pit dimension from a study of four different Sic materials 1781.

5.6 Materials

159

Figure 10 also shows that the relation with pit size is not true for all Sic-materials: Type D shows no correlation and has a much stronger decrease in strength (= -50%). It was a material with free silicon. It follows that Si-Sic has less hot corrosion resistance and that the attack and its consequences are due to the distribution of the preferentially corroded Si. To improve the hot corrosion resistance current developments aim for a protection by mullite-based coatings [79,80]. 5.6.2.1.5 Corrosion by Hot Gases Halogens do not form condensed species with Si

at the temperatures of interest. Thus, if they are pure (no oxygen or water present) they will automatically induce active corrosion behavior and calculated pressures of species like SiF4 or SiCI4over Sic will be as high as the concentration of F and C1 in the gas. In the presence of enough oxygen to form silica, these pressures will be significantly reduced for the case of attack by C12 and accordingly lower corrosion rates are measured [81]. However, pressures of SiC14 over Si02 can exceed lop6bar at temperatures above 700°C in high-C12 environments. Calculated pressures for an attack by H2 are likewise significant even at low temperatures [8]. The calculated main low-T species is CH4, but the kinetics are so unfavorable for its formation that in reality the beginning of significant active corrosion is at temperatures above 13OO0C,as is known from etching studies [82]. This is only true for pure Sic: grain boundaries and secondary phases in sintered S i c are attacked at temperatures as low as 1000°C [83]. For the survival of S i c it is therefore of great importance whether silica is formed or not. The boundary for the active-to-passive oxidation is described in Fig. 1 1. It shows the literature data and a Wagner-type theoretical model [15] based on the

H i m 76 Gulbranran 66

Vaughn

so

*

-8

I

1,200

1

1,400

I

1,600

1

1,800

2,000

Temperature ("C) Figure 11. Transition temperatures and oxygen pressures for active and passive modes of oxidation of Sic after [15].

160


influence of carbon activity and the presence of condensed SiO. Even though the model might be challenged in view of new thermodynamic data, arguing against condensed SiO, the general position of the boundary between active (I) and passive oxidation is known from the experiments. The region denoted active (11) in Fig. 11 is for a special situation, when the pressures of the produced gas species (CO) exceed 1 bar, which can lead to bubble formation and spalling, and hence quasi-active behavior, even though nominally silica is produced. Experimental evidence for this was obtained at temperatures of about 1760°C [84]. It should be noted that the oxygen pressures for the transition in Fig. 1 I should not be confused with those obtained by COX-equilibria.Despite low oxygen partial pressures of, for example CO in equilibrium with carbon, we have passive oxidation of S i c up to x 1400°Cin CO, because CO is a reactive and oxidizing gas species for S i c [52,85,86]! Consequently active corrosion in COX-environmentsis measured only at very high temperatures [87]. In passive oxidizing environments S i c forms silica (Eq. (13)). For pure S i c a parabolic behavior is evident [88]. For SSiC there is evidence that oxygen does not penetrate deeply into the interior [89], leading to a relatively homogeneous surface oxidation. Therefore SSiC has often parabolic oxidation rate constants nearly as good as chemical vapor deposition (CVD)-Sic and powder. From a data collection [53] plus some more recent data [88] the critical time t, for 0.1 mm recession is computed and shown in Fig. 12. We have omitted data from experimental changes, which were deliberately made with high additive contents to study their influence on oxidation. It is evident from Fig. 12 that basically all cases meet a long-term stability criterion up to very high temperatures. Some cautionary remarks have to be added: At very high temperatures the kinetics may not be adequately described by the parabolic law; at T > 1700°C the problem of active (11) oxidation will become important; SSiC can show bubble and pit formation at much lower temperatures, which is usually attributed to the additives B and/or A1 [90]. This bubble-and-pit formation is sometimes only present after a ‘incubation period’ of several lOOh at T > 1600°C. Most oxidation studies run only for times < 100 h. Therefore long-term studies may be necessary for true qualification of Sic. Industrial utility of Sic is not guaranteed by a sufficient resistance to oxidation by O2 or air. The influence of other components of the atmosphere or transferred impurities has to be evaluated. Often this is difficult because strong variations in temperature and gas composition are encountered. None the less it has been shown that SSiC can resist flue gases of aluminum remelting furnaces for up to 1000h without much change in strength [91], although this is certainly not true for all plants of this type [92]. Similarly nitrogen-based gases of the system N2-H2-C0 did not decrease the strength of SSiC, while endothermic gases had an adverse effect (-20-50%) at 1200-1300°C [93,94] in long-term experiments. Typically B-doped Sic behaved better then Al-doped or those with free Si. A good resistance of Sic-based materials to SO2 or H2S has also been reported [95].

5.6 Materials

161

A

I

" 4

* * *

*

* ssic A powder&SC A

CVD

1000

800

900

1200

1100

1300

1400 1600 1800 1500 1700

Temperature ("C) Figure 12. Calculated critical times (years) for 0.1 mm recession of S i c materials, using the data compilation of [53].

The influence of water on the corrosion of Sic seems to be threefold: It enhances the oxidation rate slightly; it transports impurities (e.g. Al and alkalis) which can strongly enhance oxidation; and it is able to evaporize the silica scale inducing para-linear behavior under appropriate flow conditions [23]. This may impair some applications. 5.6.2.2 Boron Carbide Boron carbide is among the hardest materials yielding only to diamond and boron nitride. It is also one of the most corrosion-resistant compounds at room or moderate temperatures. When considering the corrosion resistance of boron carbide materials, it is important to remember that they are rarely stoichiometric, with the carbon content varying from 9.88 to 23.4% [96] Many of them contain free carbon or sintering aids. Thus their behavior depends on the chemical composition. 5.6.2.2.1 Corrosion by Liquid Media Pure boron carbide is insoluble in HCl, H2S04, and HN03, even when boiling. Resistance of hot-pressed boron carbide to sulfuric acid solution has been investigated [97]. The material was produced from a powder containing 77.8-79% boron, 21.6-20.4% carbon, 0.25-0.7Y0

162


boron oxide, and 0.25-0.7% free boron. Dense and additive-free B4C specimens did not interact with 10% sulfuric acid either at room temperature or on boiling. A small mass loss (
+

+

+

+

+

+

5.6.2.2.2 Hydrothermal Corrosion and Corrosion in Water Vapor The interactions of boron carbide with water vapor starts at 250°C. The following reactions are possible:

+

B4C + 8 H20 + 2 B 2 0 3 C02

+ 8 H2,

(21)

+ H20 + 2 HBO2,

(22)

+ 3 H20 + 2 H3B03.

(23)

B2O3 and B203

Gaseous boric acid removes a boron oxide film. The rates of formation and removal of the B2O3 film are equal at 550-600°C in air with a dew point of 25-70°C and at 650°C with a dew point of 88°C. At higher temperatures, B203 is formed at a higher rate than it is removed by the interaction with water vapor. Therefore, at low temperatures boron carbide is oxidized with water vapor more rapidly than with dry air, at high temperatures the situation is quite the opposite [2]. Hydrothermal corrosion of B4C has been studied theoretically and experimentally [loo]. Reaction (21) as well as reactions leading to the formation of CH4 and CO are possible. Reactions that lead to the formation of carbon oxides dominate only at low pressures and carbide:water ratios. Under all other conditions, the formation of methane has been predicted. HB02, H3B03and H3B306gas molecules, which were predicted by thermodynamic simulation under various temperatures and pressures, are very stable thermodynamically. They may condense to some type of boric acid on cooling. Unlike Sic, boron carbide did not show any evidence of free carbon formation under hydrothermal conditions. Neither thermodynamic simulation, nor experiments show the formation of elemental carbon under hydrothermal conditions. Even the traces of carbon that were present in the as-received B4C powder disappeared after a hydrothermal treatment. Tests of a hot-pressed B4C-lO%SiAlON composite demonstrated high corrosion resistance up to 500°C. Even at 500"C, almost no changes of the sample surface occurred under the experimental conditions.

5.6 Materials

163

5.6.2.2.3 Corrosion by Melts Hot-pressed boron carbide is attacked by fused alkalis. The use of boron-carbide-based materials at high temperatures is limited at present, which probably results from the fact that they have been investigated primarily at room or moderate temperature [ 1011. 5.6.2.2.4 Corrosion by Hot Gases Boron carbide is mainly used at relatively low temperatures [96], therefore its oxidation has not been studied as thoroughly as that of the Si3N4and Sic ceramics. There are data on the oxidation of B4C powders [lo21 and several papers on oxidation of boron carbide materials produced by different methods [103-1071. In [ 1081 different reactions in the system B4C-02 were calculated thermodynamically at oxygen pressures of 105 and 1.3x lop3kPa. It was shown that the oxidation of B4C may result both in condensed (B203,B, C) and gaseous (CO, COz, B202, BO, BOz) products. At high oxygen partial pressures the reaction B4C

+ 4 0 2 ++ 2 B 2 0 3+ C02

(24) should dominate. This reaction was confirmed by the results of experimental investigations on powders and sintered specimens [ 1091. Oxidation of boron carbide powders starts at about 600°C. Studies on hot-pressed B4C containing iron, aluminum, magnesium, calcium, titanium, silicon, and other impurities at a level of up to 4% and oxidized in air [lo41 demonstrated that upon nonisothermal heating of specimens exothermal effects due to oxidation can be registered above 600°C. A detectable mass gain starts above 700°C. At 1000°C the oxidation process intensifies and above 1200°C the material starts oxidizing considerably, with a high thermal effect. At temperatures above 12OO0C,the specimen weight decreases due to the B203vaporization, the rate of which increases significantly at this temperature [ 1091. The X-ray diffraction (XRD) analysis of oxidized specimens did not reveal any solid oxidation products other than B203.This is in agreement with the thermodynamic calculations [98] and the experimental investigations showing that reaction (24) dominates B4C oxidation. The kinetic curves obtained under isothermal conditions (Fig. 13) follow the paralinear law (Fig. 7) with competing B4C oxidation and B 2 0 3 vaporization. 3,

.

1

2

N

6

1

F o

@

E -10

-20

<

-300

1

2

3

4

5

t, h

Figure 13. Oxidation kinetics of hot-pressed B4C at: (1) 800°C; (2) 900°C; (3) 1000°C;(4) 1100°C; and (5) 1250°C; reproduced from [2].

164


Thus, the oxidation kinetics due to the boron oxide vaporization at temperatures above 1200°C result in weight loss of specimens (Fig. 13), which becomes linear in time after the first 30min of oxidation. On oxidation at temperatures below 1100°C a liquid boron oxide layer (T, of amorphous B2O3 is 450°C [llO]) uniformly covers the specimen surface. Under the conditions of active B203 vaporization the diffusion of impurities apparently does not exert great influence on the oxidation process, unlike in the case of oxidation of silica formers. One may suppose that up to 1200°C the oxidation rate of B4C is controlled by the oxygen diffusion through the B203 layer. Differencesin the thermal expansion coefficients of B203 and B4C lead to cracking of the oxide layer after cooling from high temperatures. The XRD analysis revealed B203 and H3B03 in the surface layer of oxidized specimens. However, H3B03 is not formed during the oxidation of materials, but during the exposure of specimens to air at room temperature for several days due to the hydration of boron oxide according to reaction (23). The oxidized specimens kept in air were covered with a white film due to the hydration of boron oxide. It is important to note that the specimens of ceramics oxidized at 1400°C had a graphite layer of up to 30 pm thickness under the B203film [107]. If the oxidation process was accompanied by friction, the graphitized layer was formed at much lower temperatures. The formation of free carbon was also observed upon oxidation of B4C-Sic composites [l 111. Active oxidation of B4C at a low partial oxygen pressure leads to the formation of gaseous boron oxides and removal of them from the surface of the specimen. Similar to other materials, porosity increases the reactive surface of boron carbide specimens and weight gain on oxidation [1061. All the above primarily refers to materials containing relatively small amounts of impurities. However, boron carbide ceramics usually contain 7-20% Al or A1203to increase hardness and 6-1 3% silicon to improve mechanical characteristics [98]. Aluminum- and silicon-doped materials often exhibit structural nonuniformity. They have areas of pure boron carbide, and silicon- and aluminum-doped areas [107]. In contrast to the oxidation of the additive-free material, weight gain instead of its loss is observed above 1200"C, that is the rate of oxide formation is higher than the rate of its vaporization. Electron microscopy investigations of specimen surfaces of Si- and Al-doped materials heated to 1430°C demonstrated glassy oxide layers containing boron, oxygen, silicon, and aluminum. Thus, the oxide layer is a borosilicate glass. Crystallization of aluminum borates forming needle-like crystals also has been found in our work. The dissolution of silica and alumina in boron oxide retards the rate of its vaporization. In general, addition of silicon and aluminum to boron carbide facilitates the formation of borosilicate glass on oxidation and decreases the rate of boron oxide vaporization, thus increasing the oxidation resistance of the material [ll 11 [2]. However, to form a continuous protective layer on the specimen surface, it is necessary to provide a uniform distribution of additives in the material excluding the existence of pure boron carbide areas.

5.6 Materials

165

B4C reacts with C 0 2 to yield B203 and CO or free carbon [96]. Boron carbide neither interacts with sulfur and phosphorus vapors, nor with nitrogen up to 1200°C. BN can be formed upon reaction with nitrogen at higher temperatures, or when ammonia is added. With chlorine it reacts above 1000°C to form BC13 and graphite. Bromine and iodine do not react with B4C [98]. 5.6.2.1 Tungsten Carbide

Hard materials on the basis of WC have usually a metallic binder phase of Co and/or Cr, Fe, Ni, and additions of other carbides (Tic, TaC, NbC), which determine the corrosion resistance. Accordingly, these composites are attacked by many acids and bases. The authors refer the reader to previous work [112,113] for more details. Tungsten carbide has a low resistance to oxidation. It is the result of low protective properties of tungsten oxides. Dissolution of oxygen in the carbide structure starts at low temperatures, and W 0 2 and W 0 3 form upon oxidation at 500°C and higher. The oxidation rate increases drastically at 800°C due to vaporization of W 0 3 [108]. According to thermodynamic simulation [loo], at a carbide:water ratio of 1: 10, tungsten carbide should be completely oxidized producing H2W04or W 0 3 at lower temperatures and W 0 2 at higher temperatures. Carbon is oxidized to CO and C 0 2 in the case of a low W C : H 2 0 molar ratio. Formation of free carbon upon hydrothermal corrosion of WC was predicted to be possible at a high WC:water ratio upon hydrothermal corrosion of WC. The stability of WC increases with increasing pressure. Raman spectra obtained from the hydrothermally treated WC powders demonstrate the presence of graphite. Hydrothermal treatment of a WC-G%Ni commercial cemented carbide demonstrated a rather low corrosion rates below 500°C. Only a very thin layer of corrosion products covered the surface of the samples after a treatment at 350°C. At 500"C, the corrosion rate increased and all WC grains from the surface layer of the sample were dissolved in the supercritical fluid. This led to a slight decrease of the sample weight. Ni does not interact or interacts very slowly with water under these conditions. Therefore, a network of Ni grain-boundary layers remained on the surface. Ni grain-boundary phase serves probably as a diffusion barrier and protects the material from further corrosion. At 700°C, the corrosion rate increases and the surface of the sample was covered with a thick layer of corrosion products. No changes in the phase composition of the samples were found by XRD after the treatment at 350°C. WC, Ni, and traces of W 0 3 were found after hydrothermal treatments in the temperature range 400-500°C. W 0 3 dissolves in water and a weight loss of the samples was observed. At 700"C, oxidation of Ni occurs and nickel oxide interacts with W 0 3 (or H2W04) leading to the formation of NiW04. Also, hydrothermal corrosion of WC-Zr02 composites has been studied, but at lower temperatures when less significant interaction of WC with water occurred [114, 1151.

166


5.6.3 Nitrides 5.6.3.1 Silicon Nitride In many cases the references quoted for Sic can be used to access corrosion data on Si3N4-basedmaterials [2,3, 5, 35, 53, 54, 116, 1171. Si3N4ceramics are much more complicated in their behavior because sintered types come with a significant amount of grain-boundary phase, which determines the overall behavior. 5.6.3.1.1 Corrosion by Liquid Media The higher susceptibility of Si3N4ceramics to corrosion by liquid media is already apparent from reports on hydroabrasive wear, which is much higher then the wear of S i c [55]. It is not the attack of the matrix Si3N4,which controls aqueous corrosion but that of the grain-boundary phase. Often this grain-boundary phase is a oxidic silicate glass with or without small nitrogen contents. To predict the corrosion resistance in water, acids, and bases the reader is thus referred to the literature on glass and glass corrosion [118,119]. Rules of thumb would imply that a glass with more network-modifiers is less corrosion resistant to corrosion, materials with less glass content are better and that a greater crystallinity of the grain-boundary phase is of advantage. The kinetics of the leaching processes are often linear for H F and strong bases, while selective acid etching can be parabolic. Leaching is always associated with a loss in strength and tends to correlate linearly with degree of dissolution from the glass matrix as has been shown for H F and HCl with Y - and Al-leaching [120]. At 70°C a treatment for 240 h with 1 M HCl can reduce the bending strength by z 50%. 5.6.3.1.2 Hydrothermal Corrosion The onset of reactions of Si3N4 powder with water at 100 MPa according to

+

Si3N4+ 6 H 2 0 H 3 Si02 4 NH3

(25)

starts at temperatures below 200°C [61], probably at about 130°C [121]. At 500°C the powder is completely oxidized within a few hours. Reaction (25) would suggest a passivating behavior but the solubility of silica is favored even at low temperatures in the alkaline water present due to the dissolution of NH3. Hence basic reaction kinetics are linear, it is a form of active corrosion. The grain-boundary phase of Si3N4-ceramicsmay be even stronger affected by hydrothermal corrosion. This leads to pit formation at conditions of 300°C and 8.6MPa [122]. The shape and depths of pit formation depend on the additive type. It is interesting to note that a hot isostatically pressed (HIP) Si3N4 ‘without sintering additives’ was attacked more strongly then those with additives, because the grain boundaries with silica from the oxygen content of the powder were leached and resulted in grain dislodgment [ 1231. It should be noted that hydrothermal corrosion should not be confused with corrosion by wet atmospheres and that the behavior changes at higher temperatures when N2 + H2 is the product gas rather then NH3 [60].

5.6 Materials

167

5.6.3.1.3 Corrosion by Metals Reactions of Si3N4 with metals and alloys are a complex issue, because the atmosphere both for oxygen and nitrogen partial pressures plays a role and silicides, nitrides, or ternary compounds can form. A good compilation of the partly conflicting observations has been provided [1161. For the interaction of the related SiAlONs and Fe-based alloys see published work [124]. Predictions for reactions can be made on the basis of Ellingham-type diagrams. In noble gas atmospheres or under vacuum, Si3N4reacts readily with most transition elements below 1000°C and with many other metals at varying temperatures. Whether this is true for a given Si3N4 ceramic under application conditions depends on many parameters like a silica layer protection, diffusion coefficients, wetting behavior etc. The reader is here referred to the reference mentioned above [116]. 5.6.3.1.4 Corrosion by Salt Melts Hot corrosion of Si3N4 has been extensively studied and reviewed by researchers at NASA labs [67,68,77, 125, 1261. The basic attack is that on the protective silica layer, which makes the behavior similar to that of Sic and basic mechanisms can be taken from the chapter on S i c above. Recent investigations [74] gave more insight into the details of the kinetics, which is shown in Fig. 14. The surface tension differences between sulfate and silicate melts result in a fractal pattern of hillocks of silicate melt beneath the sulfate, hereby exposing unaltered silica with a strong oxygen potential gradient to the sulfate. On consumption of this silica the decomposition of the sulfate is enhanced up to a gas bubble formation. The exposed areas are the sites for stronger corrosion, which can ultimately produce pits. However, the process does slow down, when the sulfate is used up or a critical thickness of silicate layer is reached. There is a strong indication that the grain-boundary phases modify the development. Wide shallow pits and/or exposure of individual Si3N4-grainsare observed [127,128], and Na advances into the grain-boundary phase. The resulting degradation of mechanical properties is of similar order as for S i c (20-50%) [26,127,129]. However, it seems that the degradation is not linear with time but levels out with time [129]. Furthermore, in pitted samples there is an increase in the Weibull modulus, with the result that the failure stress at low failure probabilities is decreasing much less than the average strength [ 1301. Taking into account the higher level of initial strength of Si3N4-basedceramics the remaining strength after hot corrosion is usually higher compared to SIC. 5.6.3.1.5 Corrosion by Hot Gases Active corrosion has usually an even more pronounced effect on strength, compared to Sic, as the phase between the grains may vanish leading to grain dislodgment and inhomogeneous internal corrosion [ 1311. All gases which produce high partial pressures of Si-bearing species (namely halogens) are therefore dangerous, particular in low-oxygen environments [ 1321341. In mixed and oxidizing atmospheres Si3N4 has a superior resistance to C1containing environments relative to Sic materials [ 135-1 371. A small water content in H2S-H2 mixtures is also capable of producing a silica scale, which slows down

168


of the Na-sulfate

with amounts of

immiscible at 9OOOC

1 -

Ess

n Na,SO, (watersoluble)

Na,O * @i02

3J

SiO,

(not watersoluble)

Figure 14. Schematic model for the hot corrosion kinetics of silica-formers after [74].

5.6 Materials

169

corrosion [ 1341. As for Sic the plain influence of a water vapor content in oxidizing environments is probably moderate [ 1381 unless it transports impurities. The active(1)-to-passive transition has been discussed with some controversy but most models [14,139,140] and experimental data [141-1431 point to very similar conditions as outlined in Fig. 11 for Sic. The passive oxidation of pure Si3N4has been intensively studied. Thermodynamic data suggest that a two-step oxidation

+

(26)

+ N2

(27)

2 Si3N4+ 1; O2 H 3 Si2N20 N2 and

+

Si2N20 14 O2 + 2 SiOz

and accordingly a duplex layer in a oxygen potential gradient, developing in a scale, is favored. However, the discussion of the correct kinetics, the presence or absence of a sublayer of Si2N20,the activation energy, and other issues are still not settled [14,144-1511. Extremely low oxidation rates up to high temperatures are, however, clear. This contrasts strongly with the behavior of sintered Si3N4. In Fig. 15 we have calculated tc for pure and sintered qualities, assuming parabolic growth and using the data compiled earlier [53].

0

... ::'"""": .: , 1 -0

-1

0 .

~

0

0

-21

,

I

,

I

.

.

0

,

,

HPISSN

-3

800

1200

1000

900

1100

1400

1300

,,, 1800

1600 1500

1700

Temperature ("C) Figure 15. Calculated critical times (years) t, for 0.1 mm homogeneous recession by parabolic silica scale growth on Si3N4-basedmaterials using data from the collection of [53].

170


A number of things can be learned from Fig. 15. First, sintered Si3N4 never reaches the level of oxidation resistance of pure Si3N4. Typically a temperature difference of 250°C exists to have comparable behavior for even the best qualities. Second, at low to moderate temperatures pure Si3N4oxidizes much slower then S i c (Fig. 12), but sintered Si3N4is comparable to Sic. Third, there is a very wide scattering of results for sintered Si3N4.And last, but not least, at about 1400°C there is a rapid breakdown in log t, to values below and much below zero, indicating very rapid oxidation of sintered Si3N4. The reasons for these features are manifold and again related to the presence of grain-boundary phases. We often have a very complex passive oxidation behavior with leaching of grain-boundary constituents to the scale, crystallization, internal oxidation, bubble formation, scale cracking, and other features. All these phenomena are different for different systems and vary with the amount and exact composition of the additives, For detailed information the reader is referred to the textbooks and proceedings already mentioned and some reviews, which cite papers from a vast amount of literature [140,152, 1531. We have to point out some facts, which should make the reader cautious in interpreting Fig. 14 too strictly. At low temperatures some Si3N4 types (sintered with pure Y or rare earth additives) contained secondary phases, which oxidize rapidly and with large volume increases to yield the oxide phase. Since at low temperatures ( 1300"C, we commonly have deviations from the parabolic law and all the forms described in Sections 4.3.2 and 4.4.3 can be observed. Since the extrapolation to t, is done on the basis of a parabolic approximation it will not give very reliable t,. Perhaps the most disturbing characteristic of sintered Si3N4 oxidation is the appearance of internal oxidation accompanied by outer scale growth. Leaching of grain boundary constituents into the scale (Fig. 16) is observed for Y, Al, Mg, REE, and others to varying degrees. The leaching is connected to internal oxidation and is for some cases the rate limiting step [156]. In the leached zone, which extends into the sample with the same order of magnitude as the outer scale thickness, we often find increased porosity and hence decreased hardness [155]. At least this zone must be incorporated into the value for penetration but can not be gained from measuring layer thickness or mass change. There is furthermore recent evidence from transmission electron microscopy (TEM) studies for the dissolution of matrix grains in the grain-boundary phase during oxidation at high temperatures, which widens the grain boundary films as a function of depth into the material [157,158]. This type of penetration will be even more difficult to detect and quantify. Therefore, for most Si3N4 ceramics it will be necessary to have mechanical property measurements along with corrosion to evaluate their applicability. None the less there are rules of thumb, which may help in materials selection. The most important one is that the level at which oxidative corrosion becomes the prime problem, is connected to the eutectic of the system sintering additive(s) + Si02.

5.6 Materials

171

Figure 16. EDX map for Y and SEM picture of SSN oxidized at 1400°Cfor 990 min [ 1551, showing a leaching zone beneath the scale.

This may well apply also to the pure Si3N4 qualities, because here the eutectic Si3N4-Si02 in the vicinity of 1600°C may be the ultimate temperature for longterm use [1591. The challenge for sintered Si3N4to be applicable at T > 1400°C has found first success in studies which show that, for example, additions of MoSi, and surface treatments are capable of enhancing the oxidation resistance [ 160, 1611. 5.6.3.2 Boron Nitride Hard materials based on dense crystalline modifications of boron nitride find wide application in cutting tools [162]. Since these materials are produced at high pressures, the size of samples is usually small. Corrosion has been studied primarily in view of abrasive and tool applications of BN and the data are limited. However, there exists much information on corrosion of hexagonal graphite-like BN. Similar to the graphite and diamond phases of carbon, the reaction products are the same for all modifications of BN. However, hard modifications of BN typically show a higher corrosion resistance. Thus the available information on graphitic BN can be used for evaluation of the lowest corrosion resistance limit of hard BN. Although materials based on the high-pressure phases possess very high hardness and strength, they are unstable at high temperatures, which imposes restrictions on their application. When they are heated in air, along with oxidation, the wurzite and sphalerite modifications can be transformed into the graphitic one [ 1621. 5.6.3.2.1 Corrosion by Liquid Media The resistance of pure boron nitride to attack by diluted acids is lower than that to attack by concentrated ones and decreases in

172


the order H N 0 3 -+ HC1 + H2S04. It interacts slowly with concentrated sulfuric acid to form ammonium sulfate and boric acid [I631 and the corrosion rate of dense specimens is extremely low. Boron nitride dissolves on interaction with concentrated hydrofluoric acid (94% H F solution) by the reaction BN

+ 4HF H NH4BF4

(28)

and on heating with ammonium fluoride and concentrated sulfuric acid by the reaction BN 3 NH4F 2 H2S04 % 2 (NH4)2S04 BF3. (29)

+

+

+

Boron nitride specimens are resistant to attack by hydrochloric, sulfuric, and phosphoric acid solutions with additions of oxidizing agents (KMn04, K~Cr207, KC104), but decompose most rapidly in a 5% H2S04solution with KC104 addition [ 1631. The corrosion resistance of powders and bulk specimens in alkaline solutions is rather high and almost independent of the concentration. 5.6.3.2.3 Corrosion by Melts Boron nitride fused with potassium carbonate decomposes by the reaction

BN + K2C03 H KB02 + KOCN.

(30)

Nitride excess leads to a certain amount of KCN. If carbon is present, the reaction proceeds as follows: 4 BN

+ 3 K2CO3 + 2 C H K2B407 + C02 + 4 KCN.

(31)

Boron nitride reacts with sodium formate [163]. It must be emphasized that boron nitride has an exceptionally good resistance to most molten salts and glasses, probably due to its resistance to wetting. It is inert to a number of corrosive molten salts, such as alkali halides, lithium borate and cryolites, and has a good resistance to molten glasses. However, it does react with molten alkali carbonates and hydroxides. 5.6.3.2.2 Corrosion by Hot Gases Thermodynamic calculations [1081 show that the oxidation of boron nitride at atmospheric and low pressures over the whole temperature range to 2000°C is dominated by

4BN+302~22203+2N2. (32) The oxidation of boron nitride as well as boron carbide is distinguished by an oxide layer remaining in a liquid state and vaporizing considerably over the whole temperature range. However, the study of the mechanisms of oxidation of boron nitride materials is complicated by the existence of several BN modifications with considerable differences in their structure and properties. The oxidation of a stable hexagonal a-modification of BN with a graphite lattice has been studied most thoroughly [104]. The studies on Hexanite-R ceramics based on the wurzite modification of boron nitride [164] demonstrated that the oxidation kinetic curves (Fig. 17) were very similar to that for B4C (Fig. 13). A rather high oxidation resistance is also exhibited

5.6 Materials

E -1

a

-2

3O

173

I

I

1 t, h

2

Figure 17. Oxidation kinetics of wurzite BN Hexanite-R at: ( I ) 900°C; (2) 1000°C; (3) 1IOo"C, and (4) 1200°C; reproduced from [2].

by a cubic boron nitride-based material having sphalerite structure (Elbor-R) and materials that are a mixture of wurzite and sphalerite BN [165]. Graphitic boron nitride is hydrolyzed with humid air to form ammonia and boric acid: BN

+ 3 H 2 0 @ H3B03+ NH3.

(33)

Boron nitride is resistant to hydrogen and sulfur dioxide; it reacts with fluorine at room temperature by the reaction 2 BN + 3 F2 @ 2 BF3 + N2.

(34)

The absence of interaction with other halogens and some halides can be used for removal of boron, boron carbide and other impurities from boron nitride [98].

5.6.4 Carbonitrides 5.6.4.1 Silico-carbonitrides The few investigations available for the new precursor-derived silico-carbonitrides do not permit their corrosion behavior to be assessed in detail. Most studies of interest here deal with thermal stability only. The issue of thermal stability is important for these phases, because precursorderived ceramics are amorphous and have stoichiometries where detrimental structural and chemical changes may occur on crystallization. The highest temperatures recorded so far come from phases in the system Si-B-C-N (2000°C [166]). For the carbonitrides, the stoichiometry of the amorphous product is Si3+.TN4Cs+p [167] and thus in terms of possible crystallization products there is always C in addition to Si3N4 and Sic. In precursor-derived Si-C-N-ceramics,

174


micro/nanocrystalline Si3N4 and Sic appear at M 1400-1500°C [168]. Thermal stability experiments show that in Si-GN-ceramics the reaction

+

Si3N4 3 C @ 3 Sic

+ 2N2

(35)

takes place at about the thermodynamically predicted temperature (M 1405°C at 1 bar N2). The effectivity of this reaction is well known from the sintering of Si3N4-SiC composites [ 1691. Its suppression by oxidation [I 701 indicates an internal pressure build-up of several bar. In ceramics from polyborosilazanes, on the other hand, ordering effects are demonstrated only at 1700°C and strong crystallization at 1800°C [171]. The crystallization of Si3N4 at T 2 1750°C in this system was also confirmed by Baldus et al. [ 1721 who reported even higher crystallization temperatures, sometimes exceeding 1900°C. In boron-containing systems the initial Si3N4-crystallization without decomposition may indicate that the carbon is not present in the form of a reactive free carbon: boron may keep carbon in some other (amorphous?) form to suppress or to retard the kinetics of reaction (35). It is a very old experience that the crystallization of amorphous substances is influenced by numerous parameters [ 1731. Therefore we consider it very dangerous to take the observed crystallization behavior as a fixed value. Despite a lack of data on active oxidation behavior some thermal stability tests give good hints. It relates to the oxygen impurities of the precursor material. The mass loss observed and clearly attributed to oxygen removal from silicocarboboronitrides [166] starts from T M 1300°C. Thus in an environment with low P ( 0 2 ) , at which no protective layer is formed, we will have a substantial loss by active oxidation. In passive oxidation some very promising data for the new ceramics have been given [170,174, 1751. It is indicated that silicocarbonitrides behave as good as pure Si3N4 [170] and that the boron-containing types are even more resistant [175]. A t, value calculated from the k, given by [174] would be over 300 years at 1550"C! We are not aware of hot corrosion testing of precursor-derived ceramics. A prediction rests upon the argument that for those which contain boron a top scale compositions in the Na-B-Si-0-system should form, that is a classical glass forming system with low melting points and viscosities [176]. Apart from the hot corrosion degradation by itself it is completely unknown whether this will induce crystallization of the amorphous precursor-derived ceramics. 5.6.4.2 Titanium Carbide, Nitride, and Carbonitrides Carbides and nitrides of transition metals are usually hard and some of them are used as hard materials. Tungsten carbide, which was briefly described above, is a major component of cutting tools and wear-resistant materials. Titanium carbide, TIC, and nitride, TiN, are two other major hard materials that are widely used in tools. They also find application as hard coatings (in particular, TiN [177]) and components of ceramic matrix composites [178]. Both TIC and TiN show a very

5.6 Materials

175

wide stoichiometry range and almost never exist in the stoichiometric composition. Due to their structural identity, they also form solid solutions in the Ti-C-N system. Oxygen can be dissolved in the material until saturation of carbon or nitrogen vacancies [179] with no changes in the f.c.c. (NaC1) crystal structure. In all cases, whether we consider TIN, Tic, and TiC,N, materials, the oxidation behavior is similar and these rutile-formers will be analyzed together. The peculiarities of the defect structure of TIC and TiN allow for oxygen dissolution on the first stage of oxidation. Titanium oxides appear later, when the material structure is saturated with oxygen. Anatase was observed at lower temperatures and/or short oxidation times ( 5 min). With increasing time and temperature, rutile is formed as the only oxidation product [108]. Oxidation of TIN powders [180], thin films [177], and dense bodies [I081 was extensively investigated. It was shown that oxidation of TiN, following a logarithmic law, starts at room temperature. Above 500°C the oxidation rate increases and the kinetics follow a parabolic law. Oxidation of a TiN powder prepared by plasma vapor phase synthesis and having the specific surface area of 46 m2 g-' is noticeable above e2250"C and results in the formation of Ti02 as anatase and brookite at lower temperatures, and as rutile at higher temperatures [181]. The scales formed at oxidation of sintered TIN in air can be divided into two major types with the following layer systems: At low temperatures ( 5800°C) TiN-dense Ti02-air; at high temperatures (1000°C) TiN-porous Ti02-dense Ti02-air. The latter can develop into a multilayered scale during the long-term oxidation above 1000°C [181]. Growth of such rutile scales was based on the simultaneous diffusion of titanium and oxygen ions through the oxide layer in opposite directions. We can suppose that the fine grain inner layer grows by inward diffusion of oxygen, while the coarse grain outer layer grows by outward cation diffusion. Oxidation of TIC is similar to that of TIN. Many carbides of transition metals form carbon at the carbide-oxide interface. In the case of TIC, no carbon was formed [182]. Oxidation of titanium and carbon at 1000°C occurs with the same rate [183]. Both TiN and TIC are widely used in ceramic matrix composites for improving electrical conductivity and mechanical properties [ 178, 184,1851. The formation of a rutile scale was observed on the surfaces. The growth of such a scale should be related to the diffusion of titanium to the surface of the composite and its oxidation according to the reaction: 2TiN + 2 0 2 = 2Ti02

+ N2.

(36) The protective properties of rutile are fairly low compared to those of alumina or silica. Thus, the presence of TiN or TIC limits the high-temperature applications of the above composites. At small amounts and particle size of TIN in silicon nitride ceramics, a continuous silicate film can be formed, covering TIN particles and protecting them from further oxidation [ 1781. Thermodynamic analysis of the chemical equilibria in water-carbide systems [loo] indicated that carbon, corresponding oxides, CH4, C 0 2 , CO, and H2 are the main products of hydrothermal corrosion of metal carbides in the temperature

176


range 300-1000°C. There exists a possibility for selective etching of T i c and carbon layers formation. A low H 2 0 to carbide molar ratio results in the formation of carbon layers. Lower pressure is beneficial here. A high H 2 0 to carbide molar ratio means that under equilibrium conditions carbon can be oxidized to CO and C02. Experimental studies showed that no protective oxide layer was formed upon hydrothermal corrosion of carbides of transition metals. Raman spectra obtained from the hydrothermally treated TIC powders demonstrated the presence of graphite. Anatase crystals were also observed on the surface of TIC. With growing temperature, anatase was partly transformed to rutile [loo]. This is in agreement with other published data on corrosion of TIC and TiN [186].

5.6.5 Titanium Diboride TiB2 is widely used as a high-temperature material, particularly in coatings [187] and as a component of composites [ 1881. According to thermodynamic calculations for the Ti-B-0 system under oxygen partial pressures close to the ambient one and sufficiently high partial pressures of B203, Ti02 is a stable reaction product. Lower titanium oxides are stable only at reduced oxygen pressure, and TiB only at extremely low values of PBl0,. Thus, the main products of TiB2 oxidation in air or oxygen are Ti02 and B2O3. As the oxygen partial pressure is reduced, Ti305 becomes more stable. It was experimentally detected at the scale-boride interface at an oxygen partial pressure of 14.66 kPa [189]. TiB2 oxidation above 700°C was approximated by the parabolic rate law (9) [190]. Deviations from the parabolic oxidation start at temperatures of 950-1 100°C when vaporization of B2O3 becomes noticeable. A cubic law was suggested [191,192], but we assume that a paralinear law (Fig. 7) should better describe the kinetics, similar to the case of BN and B4C. However, unlike for boron carbide and nitride, the kinetics becomes parabolic again at higher temperatures (>12OO0C). This can be explained by the fact that when B203 is completely evaporated the process is controlled by the diffusion through the titanium dioxide film. Impurities and metal binders were found to improve the oxidation resistance of TiB2 [193]. This was explained by the fact that they accelerate the sintering of a rutile scale, thus increasing its protective properties. The oxidation behavior of an equimolar TiB2-A1N composite has been studied in the temperature range 700-1300°C in flowing oxygen [194]. Throughout this temperature range, the oxidation rate is governed by diffusional mechanisms, but three temperature domains can be identified. In the range 7O0-85O0C, only the oxidation of titanium diboride occurs. Calculations assuming a cubic diffusional kinetic law result in the apparent activation energy of 180 kJ mol-' . In the temperature range 900-1000°C, the oxidation of TiB2 still dominates over oxidation of AlN, but the formation of aluminum borates A14B209and A118B4033occurs and leads to a more complex behavior. An apparent activation energy of 80kJmol-' was determined for this stage. Above 1050"C, aluminum nitride oxidation becomes

References

177

more significant and the oxidation kinetics can be described by a parabolic law with an apparent activation energy of 472 kJ mol-' .

References 1. R. J. Fordham, D. J. Baxter, and T. Graziani (Eds), Corrosion of advanced ceramics. Key Eng. Muter. 1995, 113, 256. 2. Y . G. Gogotsi and V. A. Lavrenko, Corrosion o f High-Performance Ceramics, Springer

Verlag, Berlin, 1992, p. 181. 3. K. G. Nickel (Ed.), Corrosion of Advanced Ceramics Measurement and Modelling. NATO AS1 Series E. Vol. 267, Kluwer, Dordrecht, 1994, p. 467. 4. R. Telle and P. Quirmbach (Eds), Korrosion und VerschleiJ von keramischen Werkstoffen, Deutsche Keramische Gesellschaft, Koln, 1994, p. 359. 5. R. E. Tressler and M. McNallan (Eds), Corrosion and corrosive degradation of ceramics, Ceram. Trans. 1990, 10, 493. 6. D. Behrens, Dechema Corrosion Handbook, Vol. I , VCH, Weinheim, 1987. 7. D. M. Mieskowski, T. E. Mitchell, and A. H. Heuer, J . Am. Ceram. SOC.Comm. 1984, 67, 17-18. 8. K. G. Nickel, H. L. Lukas, and G. Petzow, in The SGTE Casebook. Thermodynamics at Work, K. Hack (Ed.), Institute of Metals, London, 1996, pp. 163-175. 9. J . R. Frade, in Corrosion of Advanced Ceramics - Measurement and Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 47-58. 10. G. H. Frischat, Glastechn. Ber. 1979, 52, 143-154. 11. C. Wagner, J . Appl. Phys. 1958, 29, 1295-1297. 12. M. W. Chase et al., J . Phys. Chem. Re$ Data, 1985, 14 (Suppl. I). 13. K. Hack, Metall. Trans. 1990, 21B, 1013-1023. 14. S. C. Singhal, Ceram. Int. 1976, 2, 123-129. 15. K. G. Nickel, J. Europ. Cerum. Soc. 1992, 9, 3-8. 16. H. E. Kim and D. W. Readey, in Silicon Carbide Transactions, Ohio State University, Westerville, OH. 1989, pp. 301-312. 17. K. G. Nickel et a/., Powder Metall. Int. 1989, 21, 29-34. 18. N. S. Jacobson, High- Temperature Durability Considerations f o r HSCT Combustor, Report: 3162, NASA, Cleveland, OH, 1992. 19. B. E. Deal and A. S. Grove, J . Appl. Phys. 1965,36, 3770-3778. 20. E. A. Irene, CRC Crit. Rev. Solid State Muter. Sci. 1988, 14, 175-223. 21. K. G. Nickel, in Corrosion of Advanced Ceramics - Measurement and Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 59-72. 22. P. 0. Kall, M. Nygren, and J . Persson, in Corrosion oj'Advanced Ceramics - Measurement and Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 73-84. 23. E. J. Opila and R. E. Hann, J . Am. Ceram. Soc. 1997,80, 192-205. 24. R. Schober, G. Konig, and D. Grimm, in Technische Keramische Werkstofe,J. Kriegesmann (Ed.), Deutsche Keramische Gesellschaft, Koln, 1993, Chapter 6.1.6.2 (16 pp.). 25. M. Cable, in Corrosion of Advanced Ceramics - Measurement and Modelling, K . G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 285-296. 26. D. J. Baxter, in Corrosion of Advanced Ceramics Measurement and Modelling, K . G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 249-260. 27. M. Maeda, K. Nakamura, and A. Tsuge, J. Muter. Sci. Leu. 1989,8, 195-197. 28. E. Opila, J . Am. Ceram. Soc. 1995, 78, 1107-1 110. 29. R. Telle, K. G. Nickel, and P. Quirmbach, in Korrosion und Verschleg von keramischen Werkstoffen, R. Telle and P. Quirmbach (Eds), Deutsche Keramische Gesellschaft, Koln, 1994, pp. 347-352. 30. G. C. Padgett, in Proc. of 1st ECERS, Maastricht, 1989. -

-

178

5 Corrosion of Hard Material.)

31. J. S. Association, in Japanese Industrial Standard JIS R 1609-1990, Japanese Standard Association, Japan, 1990. 32. J. J. Kubler and R. Morrell, A S M E , 1993, 93-GT-158, 1-5. 33. K. G. Nickel et al., Ceram. Forum Int. 1994, 71, 624631. 34. K. G. Nickel and Y. G. Gogotsi, Key Eng. Muter. 1996, 113, 15-30. 35. R. Morell, Handbook of Properties of Technical & Engineering Ceramics, HMSO, London, 1987. 36. M . F. Ashby and D. R. H. Jones, Engineering Materials I , Pergamon, Oxford, 1980, p. 279. 37. H. Hoppert, H. Rasch, and E. Litzow, in Korrosion und VerschleiJ von keramischen Werkstofen, R. Telle and P. Quirmbach (Eds), Deutsche Keramische Gesellschaft, Koln, 1994, pp. 98-106. 38. T. Graas et al., Ceram. Forum Int. 1997, 74, 723-726. 39. A. Kailer, Y. G. Gogotsi, and K. G. Nickel, in Proc. EuroMat ’97,Maastricht, 1997, pp. 171-174. 40. Y. G. Gogotsi, A. Kailer, and K. G. Nickel, Muter. Res. Innovat. 1997, 1, 3-9. 41. Y. L. Orlov, The Mineralogy of the Diamond, Wiley, New York, 1977. 42. J. E. Field, in Diamond and Diamond-like Films and Coatings, R. E. Clawing et al. (Eds), Plenum, New York, 1991, pp. 17-35. 43. R. J. Wedlake, in The Properties qf Diamond, J. E. Field (Ed.), Academic Press, New York, 1979, pp. 501-535. 44. T. Evans, in The Properties of Diamond, J. E. Field (Ed.), Academic Press, New York, 1979, pp. 403424. 45. S. Kumar et al., Diamond Rel. Muter. 1996, 5, 12461248. 46. E. A. Gulbransen and J. S. A,, in Heterogeneous Kinetics and Elevated Temperatures, G. R. Belton (Ed.), Plenum, New York, 1970, pp. 181-208. 47. K. Tankala and T. Debroy, in Proc. New Diamond Science and Technology, R. Messier et al. (Eds), MRS, Pittsburgh, 1991, pp. 827-831. 48. W. Zhu et al., in Proc. New Diamond Science and Technology, R. Messier et al. (Eds), MRS, Pittsburgh, 1991, pp. 821-826. 49. F. Schroder (Ed.), Chemisches Verhalten von Graphit. Gmelins Handbuch der Anorganischen Chemie, Vol. Kohlenstoff, Teil B, Lieferung 3, Verlag Chemie, Weinheim, 1968, pp. 789-809. 50. F. Marcuccilli, Etude exphrimentale et thhoretique de la structure et de la rhactivite‘ de suies issues de moteur diesel, Dissertation, Universiti de Haute-Alsace, 1993, p. 212. 51. Y. G. Gogotsi and K. G. Nickel, Ceram. Eng. Sci. Proc. 1997, 18, 747-754. 52. V. Haase, in Gmelins Handbuch der Anorganischen Chemie, F. Schroder (Ed.), Springer, Berlin, 1986, pp. 325-395. 53. K. G. Nickel and P. Quirmbach, in Technische Keramische Werkstofie,J. Kriegesmann (Ed.), Deutscher Wirtschaftsdienst, Koln, 1991, Chapter 5.4.1.1, pp. 1-76. 54. J. R. Blachere and F. S. Pettit, High Temperature Corrosion of Ceramics, Noyes Data Corporation, Park Ridge, NJ, 1989, pp. 188. 55. W. Prechtl, Keram. Z. 1993, 45, 197-201. 56. E. Gehrke et al., Mechanical, Thermal and Chemical Properties of’a Sic-material, Report VI l / 94, IWM, Freiburg, 1994. 57. K. Wendland et al., in Korrosion und Verschleij von keramischen Werkstoffen, R. Telle and P. Quirmbach (Eds), Deutsche Keramische Gesellschaft, Koln, 1994, pp. 107-1 13. 58. Y. G. Gogotsi and M. Yoshimura, M R S Bull. 1994, 10, 39 45 . 59. Y. G. Gogotsi et al., Diamond Rel. Muter. 1996, 5, 151-162. 60. M. Yoshimura, J. Kase, and S. Somiya, in Proc. 2nd Int. Symp. Ceramic Materials and Componentsf o r Engines, W. Bunk and H. Hausner (Eds), Germany, 1986, pp. 529-536. 61. M. Yoshimura et al., in Corrosion and Corrosive Degradation ojCeramics, R. E. Tressler and M. McNallan (Eds), American Ceramic Society, 1990, pp. 337-354. 62. S. Kitaoka et al., J. Am. Ceram. Soc. 1994, 77, 1851-1856. 63. R. 0. Fournier and J. J. Rowe, Am. Min. 1977,62, 1052-1056. 64. R. Sangiorgi, in Corrosion of Advanced Ceramics, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 261-284. 65. N. S. Jacobson, J. Am. Ceram. Soc. 1986, 69, 7482. 66. N. S. Jacobson, J. L. Smialek, and D. S. Fox, in Handbook qf’ Ceramics and Composites, N . P. Cheremisinoff (Ed.), Dekker, New York, 1990, pp. 99-136.

References

179

67. N.S. Jacobson, J. Am. Ceram. SOC.1993,76, 3-28. 68. N. S.Jacobson, J. L. Smialek, and D. S. Fox, in Corrosion of Advanced Ceramics Measurement and Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 205222. 69. H.J. Grabke, in Corrosion of Advanced Ceramics - Measurement and Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 223-234. 70. F. S. Pettit, G. H. Meier, and J. R. Blachere, in Corrosion of Advanced Ceramics - Measurement and Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 235-248. 71. N.S . Jacobson, Oxid. Metals, 1989, 31, 91-103. 72. N. S.Jacobson, J. Am. Ceram. SOC.1985,68,432-439. 73. C. Berthold and K. G. Nickel, Key Eng. Muter. 1997, 132-136, 1588-1591. 74. C. Berthold and K. G. Nickel, J . Europ. Ceram. SOC.1998, in press. 75. J. Schlichting, Werkstof Korros. 1975, 26, 753-758. 76. N. S. Jacobson, C. A. Stearns, and J. L. Smialek, Adv. Ceram. Muter. 1986, 1, 154-161. 77. D. S. Fox, N. S. Jacobson, and J. L. Smialek, Hot Corrosion of Ceramic Engine Materials, Report DOE/NASA/50111-2, DOE/NASA, Cleveland, 1988. 78. D. S. Fox and J. L. Smialek, J . Am. Ceram. Soc. 1990, 73, 303-31 1. 79. K. N. Lee e f al., 1995. 80. K. N. Lee and R. A. Miller, J . Am. Ceram. SOC.1996, 79, 62G-626. 81. S. Y.Ip, M. J. McNallan, and D. S. Park, J . Am. Ceram. SOC.1992, 75, 1942-1948. 82. G. S. Fischman, A. Zangvil, and S. D. Brown, Adv. Ceram. Muter. 1988, 3, 66-70. 83. G. W. Hallum and T. P. Herbell, Adv. Ceram. Muter. 1988, 3, 171-175. 84. G. A. Schneider, K. G. Nickel, and G. Petzow, in Physics and Chemistry of Carbides, Nitrides and Borides, NATO ASI Series E: 185, R. Freer (Ed.), Kluwer, Dordrecht, 1990, pp. 387401. 85. E. Fitzer and R. Ebi, in Silicon Carbide, K. L. Marshall, T. W. Faust, and C. E. Ryan (Eds), South Carolina University Press, Columbia, SC, 1974, pp. 320-328. 86. K. Schwab, Korrosion poroser Sic-Keramiken unter Rujjilterbedingungen. Dissertation, University Tiibingen Mineralogy, 1996, p. 170. 87. T. Narushima et al., J . Am. Ceram. SOC.1993, 76, 2521-2524. 88. L. U. J. T. Ogbuji and E. J. Opila, J . Electrochem. SOC.1995, 142, 925-930. 89. J. A. Costello and R.E. Tressler, Ceram. Int. 1985, 11, 3944. 90. J. W. Fergus and W. L. Worell, in Corrosion and Corrosive Degradation of Ceramics, R. E. Tressler and M. McNallan (Eds), Martinus Nijhoff, Boston, 1990, pp. 43-51. 91. R. M. Persoons, J. Luyten, and A. D. Stalios, in Corrosion of Advanced Ceramics, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 297-306. 92. M. McNallan, M. Van Roode, and J. R. Price, in Corrosion and Corrosive Degradation of Ceramics, R. E. Tressler and M. McNallan (Eds), Martinus Nijhoff, Boston, 1990, pp. 445467. 93. D. P. Butt, R. E. Tressler, and K. E. Spear, J. Am. Ceram. Soc. 1992, 75, 3257-3267. 94. D. P. Butt, R. E. Tressler, and K. E. Spear, J. Am. Ceram. Soc. 1992, 75, 3268-3277. 95. H.Nickel and R. Forthmann, in Corrosion of Advanced Ceramics, K. G . Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 99-1 16. 96. P. S. Kisly et al., Naukova Dumka, Kiev, 1988. 97. G. I. Postogvard, Y. G. Parkhomenko, and I. T. Ostapenko, Poroshkovaya Metall. 1983,8, pp. 28-30. 98. T. Y.Kosolapova (Ed.), Nonmetallic Refractory Compounds, Metallurgiya, Moscow, 1985. 99. A. M.Pivak et at., in New Technological Processes in Powder Metallurgy, I. D. Radomyselsky (Ed.), Institute for Problems of Materials Science, Kiev, 1986, pp. 31-35. 100. N. S. Jacobson, Y. G. Gogotsi, and M. Yoshimura, J . Muter. Chem. 1995. 101. F. Thevenot, in Proc. 1st ECERS, G. De With, R. A. Terpstra, and R. Metselaar (Eds), Elsevier, Maastricht, 1989, pp. 2.1-2.25. 102. Y. L. Krutsky, G. V. Galevsky, and A. A. Kornilov, Poroshkovaya Metall. 1983,2, pp. 47-50. 103. V. A. Lavrenko et at., Oxid. Metals, 1976, 10, 85-95. 104. V.A. Lavrenko, Y. G. Gogotsi, and I. N. Frantsevich, Dokl. Akad. Nauk SSSR, 1984,27,114117. 105. V. A. Lavrenko and Y. G. Gogotsi, Oxid. Metals, 1988, 29, 193-202. 106. L. N. Yefimenko, E. V. Lifshits, and I. T. Ostapenko, Poroshkovaya Metall. 1987, 4,56-60. -

180


107. Y. G. Gogotsi and V. V. Kovylyaev, Poroshkovaya Metall. 1989, 4, 71-75. 108. R. F. Voitovich, Oxidation of Carbides and Nitrides (in Russian), Kiev, 1981. 109. A. P. Pomytkin, Kinetics of High Tempetrature Oxidation of Titanium, Niobium, Chromium, Boron and Silicon Carbides in Oxygen, Thesis, Academy of Science of USSR, Institute for Problems of Materials Science, Kiev, 1976. 110. R. C. Weast (Ed.), Handbook of Chemistry and Physics, CRC Press, Boca Raton, 1987. 111. T. Narushima et al., Key Eng. Muter. 1996, 113, 99-104. 112. I. M. Low and X. S. Li (Eds), Key Eng. Muter. 1996, 114, 284. 113. I. M. Low (Ed.), Key Eng. Muter. 1998, 138, 696. 114. T. Yoshio et al., in Proc. Fall Meeting of the Ceramic Society of Japan, 1993, p. 206. 115. T. Nishida et al., in Proc. 10th Japan-Korea Seminar on Ceramics, 1993, pp. 385-388. 116. R. C. Sangster and G. Bar, Silicon, Suppl. B 5dl,8th edn., F. Schroder (Ed.), Vol. 15, SpringerVerlag, Berlin, 1995, p. 255. 117. R. C. Sangster, P. Kampf, and U. Nohl, Silicon, Suppl. B 5d1, 8th edn, F . Schroder (Ed.), Vol. 15, Springer-Verlag, Berlin, 1995, p. 304. 118. D. E. Clark and B. K. Zoitos, Corrosion of Glass, Ceramics and Ceramic Superconductors, Noyes Publications, Park Ridge, NJ, 1992, pp. 672. 119. H . Scholze, Glas, 3rd edn, Springer-Verlag, Berlin, 1988, pp. 342. 120. M. Shimada and T. Sato, in Corrosion and Corrosive Degradation of Ceramics, R. E. Tressler and M. McNallan (Eds), American Ceramic Society, 1990, pp. 355-365. 121. K. G. Nickel, U. Daumling, and K. WeiBkopf, Key Eng. Mater. 1994,89-91,295-300. 122. T. Yoshio and K. Oda, in Corrosion and Corrosive Degradation of Ceramics, R. E. Tressler and M. McNallan (Eds), American Ceramic Society, 1990, pp. 367-386. 123. K. Oda et al., J . Am. Ceram. SOC.1993,76, 1365-1368. 124. J. Vleugels and 0. Van der Biest, Key Eng. Muter. 1998, 138-140, 127-176. 125. D. S . Fox and N. S. Jacobson, J. Am. Ceram. SOC.1988,71, 128-138. 126. N. S . Jacobson and D. S. Fox, J. Am. Ceram. SOC.1988,71, 139-148. 127. D. S. Fox, N. S . Jacobson, and J. L. Smialek, in Corrosion and Corrosive Degradation of Ceramics, R. E. Tressler and M. McNallan (Eds), Martinus Nijhoff, Boston, 1990, pp. 227-249. 128. C. Berthold, Degradation von Si,N, als Ventilwerkstoffdurch Oxidation und Salzschmelzkorrosion bei Ein&tztemperaturen, Dissertation, University Tubingen, Applied Mineralogy, 1997, D. 142. 129. T. E. Strangman and D. S. Fox, in Proc. Erosion, Corrosion and Foreign Object Damage Eflects in Gas Turbines, Rotterdam, 1994, pp. AGARD36/1-36/11. 130. C. Berthold and K. G. Nickel, in Proc. Werkstoff-Woche '96, F. Aldinger and H. Mughrabi. (Eds), DGM, Frankfurt, 1997, pp. 577-582. 131. D. R. Clarke, J . Am. Ceram. SOC.1983,66, 92-95. 132. H. E. Kim and A. J. Moorhead, J. Am. Ceram. SOC.1990, 73, 3007-3014. 133. F. A. Oliveira, R. A. H. Edwards, and R. J. Fordham, Werkstoff.Korros. 1990,41,73&742. 134. F. A. Costa Oliveira, High Temperature Gaseous Corrosion of Silicon Nitride Ceramics, Dissertation, TU Delft, JRC Petten, 1992, pp. 271. 135. J. E. Marra et al., J. Am. Ceram. SOC.1988,71, 1067-1073. 136. M. McNallan et al., in Corrosion and Corrosive Degradation of Ceramics, R. E. Tressler and M. McNallan (Eds), Martinus Nijhoff, Boston, 1990, pp. 309-333. 137. P. Hsu, C. Park, and M. J. McNallan, J. Am. Ceram. SOC.1993,76, 1621-1623. 138. M. Maeda, K. Nakamura, and M. Yamada, J. Muter. Sci. 1990,25,379&3794. 139. A. H. Heuer and V. L. K. Lou, J. Am. Ceram. SOC.1990,73,2785-3128. 140. K. G. Nickel, Z. Fu, and P. Quirmbach, J . Eng. Gas Turb. Power, 1993, 115, 76-82. 141. J. E. Sheehan, J . Am. Ceram. SOC.Comm. 1982,65, 11 1-1 13. 142. W. L. Vaughn and H.G. Maahs, J . Am. Ceram. SOC.1990,73, 1540-1543. 143. T. Narushima et al., J. Am. Ceram. SOC.1994,77, 2369-2375. 144. H. Du, R. E. Tressler, and K. E. Spear, J. Electrochem. SOC.1989, 136, 3210-3214. 145. L. Ogbuji and J. Smialek, J. Electrochem. SOC.1991, 138, 51-53. 146. L. U. J. T. Ogbuji, J. Am. Ceram. SOC.1992, 75, 2995-3000. 147. L. U. J. T. Ogbuji, J. Am. Ceram. SOC.1995,78, 1272-1278. 148. L. U. J. T. Ogbuji, J . Am. Ceram. SOC.1995, 78, 1279-1284.

References

181

149. L. Ogbuji, in Proc. Microscopy of Oxidation 3, 1996, Cambridge, UK 150. K. L. Luthra, J . Am. Ceram. SOC.1991,74, 1095-1103. 151. K. L. Luthra, in Corrosion ofAdvanced Ceramics - Measurement and Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 23-34. 152. M. Billy, in Progress in Nitrogen Ceramics, F. L. Riley (Ed.), Martinus Nijhoff, Boston, 1983, pp. 403419. 153. K. G. Nickel, Ceram. Int. 1997, 23, 127-133. 154. J. K. Patel and D. P. Thompson, Br. Ceram. Trans. J. 1988,87, 70-73. 155. Z. Fu-Rudolph, Oxidationsverhalten von additivhaltigen Si3N4-Werkstoffen,Dissertation, Max Planck Institut fur Metallforschung, PML, 1992, p. 159. 156. D. R. Clarke, in Progress in Nitrogen Ceramics, F. L. Riley (Ed.), Martinus Nijhoff, Boston, 1983, pp. 421426. 157. M. Backhaus-Ricoult and Y. Gogotsi, Key Eng. Muter. 1996, 113, 81-90. 158. C. O'Meara and J. Sjoberg, J . Am. Ceram. SOC.1997, 80, 1491-1500. 159. E. Butchereit et al., in Proc. Werkstoflwoche '96, F. Aldinger and H. Mughrabi (Eds), DGM Frankfurt, 1997, pp. 569-575. 160. H. Klemm et al., J . Am. Ceram. SOC.1996, 79,2429-2435. 161. H. Klemm, M. Herrmann, and C. Schubert, Ceram. Eng. Sci. Proc. 1997, 18, 615-623. 162. A. V. Kurdyumov and A. N. Pilyankevich, Phase Transformations in Carbon and Boron Nitride, Naukova Dumka, Kiev, 1979. 163. G. V. Samsonov, 0. P. Kulik, and V. S. Polishchuk, Processing and Methods of Analysis of Nitrides, Academy of Science of USSR, Kiev, 1978. 164. V .A. Lavrenko, T. G. Protsenko, and A. V. Bochko, Dokl. Akad. NaukSSSR,1975,224,877879. 165. A. V. Bochko et al., Sverkhtverdye Materialy, 1986, 1, 16-18. 166. T . Riedel et al., Nature, 1996, 382, 796-798. 167. T. Riedel, in Processing of Ceramics, Part 2, R. Brook (Ed.), VCH, Weinheim, 1996, Chapter 11, pp. 1-50. 168. Friess, M. et al., Key Eng. Muter. 1994, 89-91, 95-100. 169. K. G. Nickel et al., Adv. Ceram. Muter. 1988, 3, 557-562. 170. T. Riedel et al., Nature, 1995, 374, 526-528. 171. Funayama. 0. et al., J . Am. Ceram. SOC.1993,76,717-725. 172. H.-P. Baldus, M. Jansen, and 0. Wagner, Key Eng. Muter. 1994, 89-91, pp. 75-80. 173. W . Vogel, Struktur und Kristallisation der Glaser, VEB Deutscher Verlag fur Grundstoffindustrie, Leipzig, 1965. 174. H. P. Baldus and G. Passing, Muter. Res. SOC.Symp. Proc. 1994, 346, 617422. 175. M. Jansen et al., in Proc. Werkstoff-Woche'96, F. Aldinger and H. Mughrabi (Eds), DGM, Frankfurt, 1997, pp. 669-674. 176. G. W. Morey, J . SOC.Glass Technol. 1951, 35, 27G283. 177. H. G. Tompkins, J . Appl. Phys. 1991,70,3876. 178. Y .G. Gogotsi, J . Muter. Sci. 1994, 29, 2541-2556. 179. E. M. Fryt, in Microscopy ofOxidation, S . B. Newcomb and J. A. Little (Eds), Institute of Materials, London, 1997, pp. 739-747. 180. Y. Sakka, S. Ohno, and M . Uda, J . Am. Ceram. SOC.1992,75,244248. 181. Y .G. Gogotsi, F. Porz, and G . Dransfield, Oxid. Metals, 1993, 39, 69-91. 182. S. Shimada and M. Kozeki, J. Muter. Sci. 1992, 27, 1869-1875. 183. M. Munster, Z. Elektrochem. 1959, 3, pp. 806-819. 184. A. Tampieri and A. Bellosi, J . Am. Ceram. SOC.1992,75, 1688-1690. 185. J. Mukerji and S. K. Biswas, J . Am. Ceram. SOC.1990, 73, 142-145. 186. M. Yoshimura, M. Hayakawa, and S . Somiya, Nihon Ceramikkusu, 1989,97, 1339-1347. 187. J . Desmaison, in Corrosion of Advanced Ceramics - Measurement and Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 309-327. 188. A. Bellosi, in Corrosion of Advanced Ceramics - Measurement and Modelling, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 131-142. 189. V. V. Svistunov, Investigation of Preparing and Oxidation Processes of Transition Metals Melted Borides, Thesis, Sverdlovsk, 1975.

182 190. 191. 192. 193. 194.

5 Corrosion of Hard Materials A. Lebugle and G. Montel, Rev. Int. Hautes Temp. Refract. 1974, 11, 231. R. J. Irving and I. G. Worsley, J . Less-Common Metals, 1968, 16, 103. A. Munster and G. Schlamp, 2. P h p . Chem. 1960,25, 116. V. B. Voitovich, V. A. Lavrenko, and A. M. Adejev, Oxid. Metals, 1994, 42, 145-161. V. Schneider et al., in Corrosion of Advanced Ceramics, R. J. Fordham, D. J. Baxter, and T. Graziani (Eds), TransTech, 1996, pp. 49-58.


6 Interrelations Between the Influences of Indentation Size, Surface State, Grain Size, Grain-Boundary Deformation, and Temperature on the Hardness of Ceramics A. Krell

6.1 Introduction This contribution describes the most important influences on the hardness of materials from the viewpoints of the measuring approach and of probable chances for further progress in the development of new materials. Therefore, the discussion will be focused on questions that arise when precise measurement has to be achieved in the investigation of hard materials. Which testing load should be applied? Is there a real size effect to be expected from basic properties of the material (or is it an artifact of the measurement?)? - In investigations of new hard materials, the size of available samples is often small, and many people prefer microhardness approaches. Can grain size effects be recorded independent of the applied load? - How does the hardness depend on the surface preparation? Do ground ceramics (which are important in technical use) exhibit the same hardness as polished samples (often preferred for tests in development procedures)? - In ceramics, the dependence of the strength on flaws is a well-known phenomenon associated with stress concentrations that cannot relax by plastic deformation in these brittle materials. Does hardness testing and the development of sintered materials with an increased hardness also have to consider flaw populations? - Is the intensively investigated activity of lattice processes inside the grains (dislocations, twins) the one significant contribution to the inelastic deformation at an indentation site at room temperature? What is the role of grain boundaries? - Which kind of temperature dependence is to be expected in different temperature ranges? Is there a different impact of the microstructure on the hardness at, for example, 400°C or at 1400°C? Which elements of the microstructure are most important for obtaining a high hardness in the different ranges of the temperature? -

It is not intended here to present new evidence for special micromechanisms. Instead, the available knowledge about such processes is used to answer the questions outlined above.

184

6 Interrelations Between Influences

6.2 The Assessment of Residual Porosity and Flaw Populations: A Prerequisite for any Hardness Investigation Whereas the influence of the porosity on the hardness is not an object of general controversy, it is often neglected both in characterizing and in materials development investigations. In fact, the grain size effect on the hardness of a material gives little information when samples with relative densities of 96 and 99% are compared. Figure 1 shows that the grain size effect may be not only obscured but offset by porosity phenomena: a small porosity of 1-2% can affect the hardness more than a decrease in the grain size from 2 to 0.5ym. Therefore, it is imperative to use specimens with relative densities 299% if data are required to characterize the material independent of porosity influences. With the large influence of even small amounts of porosity shown in Fig. 1 and with a surprisingly large contribution from grain boundaries to the inelastic deformation at an indentation site (demonstrated below), it is not surprising to find a strong influence of many kinds of defects on the hardness. Individual pores which

Figure 1. The Vickers hardness (testing load 1 kg) of fine-grained sintered alumina ceramics prepared by a sol/gel approach starting with boehmite (y-seeding).

6.3 Theoretical Considerations

185

are so small that they are invisible in the optical equipment of hardness testers, clusters of pores, microcracks, and other defects located within the subregion of the material affected by the indentation, all promote the penetration of an indenting body in one way or another. Therefore, the development of new hard materials requires the preparation of phases with small microplastic deformability of the crystal lattice, and reduced grain sizes in the sintered polycrystals, and highly perfect microstructures. For example, sintered alumina ceramics which have a submicrometer microstructure and associate this small grain size with a macrohardness of 20-24 GPa (compared with 17-19 GPa in conventionally sintered corundum) were obtained only when new processing approaches became available, which could reduce the flaw population [ 11.

6.3 Theoretical Considerations 6.3.1 The Role of the Lattice and of Grain Boundaries in the Inelastic Deformation at an Indentation Site in Sintered Hard Materials On a microscale, the permanent plastic deformation at an indentation site in brittle ceramics is associated with similar deformation structures as in metals. The dislocation density is only 106-107 cm-* in virgin pressureless sintered alumina (A1203),an extremely low value which should make microplastic deformation on indentation very difficult. However, transmission electron microscopy (TEM) investigations in sa phire (single crystal A1203) revealed a very high dislocation density of 101'-10''cm-2 and even more after indentation at only 50g [2]. This high degree of plasticity under the tip of the indenter in the brittle material is possible because the highly localized stress field exceeds the high yield stress of the ceramic (Table 1: 4 8 G P a for single and polycrystalline A1203 at room temperature [3-5]) while macroscopic fracture is avoided due to the extreme spatial concentration of the stress field. It is important to understand that small finite crack systems generated beyond some critical stress increase the elastic compliance at the site; contrary to incorrect interpretations they do, however, not significantly change the area of the material exposed to the (constant) load of the indenter. Hence, the size of the permanent indent and the conventionally measured hardness remain unaffected. It is generally assumed that the higher hardness of most polycrystals compared with the single crystalline state, and the increase of the hardness at reduced grain sizes, has to be attributed to hindrance of the movement and the growth of dislocations and twins by grain boundaries. By qualitative observations, the generation of dislocations and twins at the indentation site has been investigated repeatedly, but no attempt has been made to describe theoretically the relationship between deformation structures in the lattice, its hardness, and the grain size dependence

186


of these processes. Without t h s knowledge, it is not surprising that even basic questions like the dependence of the hardness on the testing load (i.e. on the size of the indent) are still a subject of controversy. Considering the very complex dislocation and twin structures at an indentation site in ceramics, it is not possible to describe precisely this situation theoretically, and probably any model has to be a phenomenological one. Such a model describing the indentation size effect (ISE) independent of special micromechanisms, is introduced in Section 6.3.2 and may be used, for example, for extrapolation purposes when very small indents have to be discussed (e.g. on investigating surface properties). Even with small loads of 1&100mN (1log), the length of Vickers diagonals in A1203 is 1-3pm, and the size of the associated plastic zone is rather large (x4-12 pm) compared with today’s most interesting Jine-grained microstructures. At further reduced loads, unfortunately, even advanced indentation techniques often do not give reliable data, and a theoretically supported extrapolation towards smaller sizes can be useful. Surprisingly, even less experimental and theoretical attention has been paid to another mechanism which should contribute to the inelastic deformation additionally to the role of dislocations: the activity of grain boundaries. Table 1 compares the critical stresses for microplastic deformation of the crystal lattice and for grain boundary fracture. There is no doubt that the compressive yield stress and the grain boundary strength are similar at least within an order of magnitude. Therefore, stresses at an indentation site which exceed the yield stress of 4-8 GPa should also exceed the grain boundary strength of 2-8 GPa. The key idea for investigating the question about the amount of grain boundary contribution to the generation of a permanent indent at room temperature is outlined in Fig. 2: it is assumed that the grain size effect in the hardness of polycrystals and the influence of the testing load (the indentation size) are analogous features. Considering a finite volume within which dislocations or twins can grow, move, or multiply, a limitation of this space should reduce the microplastic deformability of this segment of material (increasing its hardness). It seems unimportant whether this limitation is due to close spacing of grain boundaries (lower part of Fig. 2) or a small spatial extension of the plastically deformed zune (limitation of the stress field) at the tip of the indenter (demonstrated in the upper part of Fig. 2 for a single crystal where the ISE does not interfere with other mechanisms that might affect the mobility of the dislocations). Table 1. Comparison of critical stresses for microplastic deformation of the crystal lattice and for grain boundary fracture in AI2O3at room temperature [6]. Compressive yield stress estimations from

Dynamic shock wave investigations Sapphire [4] Sintered A 2 0 3[3]

4 GPa 6 GPa

Grain boundary strength, ngb, for sintered alumina (grain size D = 15 pm) with grain boundary fracture toughness [7] FZ 1.5 MPam’’’, size of defects (e.g. pores at triple junctions) =

Indentation analysis

I

mSapphire [5] Sintered A 2 0 3 [5]

4.5 GPa 7.5 GPa

a

U f7gb = 2-8

GPa


I

I

I---

1Id.nr :tri

grain boundary contrit,,,,.

II

187

I? 51 7e

6

necr

% ,

to deformat

grain size effect

--_

--

at constant load ( d a m zone size)

Grain Size in Sintered Polycrystals

L

I

Figure 2. The limitation of the free space available for dislocation or twin activity by the spatial size of the plastic zone in a single crystal (upper curve/upper x-axis), and by the grain size of polycrystalline microstructures (lower curve/lower x-axis) [6].

Under this condition, the size effect (load effect) as measured in single crystals simulates the grain size influence of the hardness in polycrystalline materials. In the mean time, this simulation of the grain size influence in the hardness of polycrystals by the indentation size effect (load influence) in a single crystal applies under a very interesting condition: it excludes, in the single crystal, all possible contributions from grain boundaries. Therefore, the dzflerence between the plastic zone size effect (load effect) measured in a single crystal (e.g. in sapphire) and the real grain size effect observed in the sintered polycrystals (e.g. sintered corundum) gives a measure that characterizes the amount of grain boundary contributions to the deformation that governs the hardness at room temperature (experimental data c.p. ##6.5, Fig. 8). To be able to compare grain size and indentation size influences, the size of the plastically deformed zone must be known. This information comes from TEM investigations where an approximately constant ratio 2RP1/2aM 4-5 between the plastic zone size 2RP1and the length of the Vickers diagonals 2a can be derived for single as well as polycrystalline ceramics and for hard materials with fundamentally different bonding like (ionic) alumina [2] and (covalent) S i c [8]. Together with the measurements of the ISE in sapphire and of the grain size dependence of the hardness in available sintered microstructures with grain sizes 20.2 l m , the theoretical extrapolation will give a forecast of the probable hardness of nanostructured alumina ceramics. The model also explains the observed influence of

188


the grain size on the measured load effect and demonstrates why reliable information about the grain size dependence of the hardness requires the use of larger testing loads.

6.3.2 Quantitative Understanding the Load Effect on the Hardness: Theoretical Considerations Compared with Single Crystal Data Considering the very complex dislocation and twin structures at an indentation site in ceramics, it is neither possible to describe precisely this situation nor would it be straightforward to ask for a model that depends on special configurations. Hence, the consideration has to be a phenomenological one, and it cannot be intended to justify one or another special deformation mechanism. Instead, the limited scope of the question to be answered here is whether the dislocation idea will generally give a size effect which agrees with the experimental observation, or if some ideas have to be rejected because of obvious disagreements. For example, people are often upset when sometimes a strong size (load) effect is observed - which disappears in other microstructures of the same material (or simply on measuring on surfaces prepared in another way). Therefore, it seems interesting to ask whether dislocation activity (or an associated deformability that changes with increasing load) can introduce an indentation size effect that possibly depends on the microstructure. With the observed high dislocation and twin densities at the indentation site [2], an idea proposed by Upit and Varchenya [9] is adopted here: microscopic forces (or energies) required to generate and to move dislocations depend on the size of (e.g.) dislocation half loops (and, therefore, on the space available for their formation), and cause the observed size effect of the hardness. A definition of the size of the plastically deformed zone 2RpIbeneath indentations in ceramics was given by Perrott [lo] together with an analysis of the local stress field assuming a ratio 2Rp1/2aM 4 (2a = length of Vickers diagonals). It is important to understand that, neglecting grain boundary contributions in polycrystals, the microplastic deformability at the indentation site is governed by the local compressive yield stress of each of the grains. This yield stress of individual grains in a polycrystalline microstructure should be similar to the yield stress of a single crystalline volume when, at small testing loads, the plastic zone size 2RP1in the single crystal equals the size of the grains in the considered polycrystal. The typical 'as sintered' structure of grains in sintered alumina consists of aplurality of grains without any lattice defects (elements of plastic deformation); only in a few grains can single dislocations, dislocation loops, and dislocation networks be observed [l l-121. Therefore, as has already been mentioned, at the beginning of an indentation process the average dislocation density is 106-107 cm-2 or even less, and it will be rather hard to initiate microplastic deformation. However, at a small load of only 0.5 N Hockey [2] observed regions with a dislocation density at the upper limit of resolution (10'1-10'2 cm-*), and it is probable that the deformation process has arrived at a state where strain hardening (if existing at all) has already been passed through.


189

Then, on increasing the load further, fundamental principles of solid state physics suggest that the further extension of the plastically deformed zone in a single crystal should occur much more easily by growth or by multiplication of pre-existing elements of lattice deformation (dislocations, twins) than by the more difficult generation of new dislocations in the virgin material around. Hence, on increasing the load (the size of the indent), the microplastic deformability increases (the hardness decreases) because the more easy growth of pre-existing dislocations becomes more important and the role of the (difficult) generation of new dislocations decreases. To simplify the mathematical procedure, a characteristic area A e (2Rp1)2is introduced. At some load PI, the microplastic deformability S can be defined as an inverse local compressive yield stress:

S = (dP/dA)-'

(la) the smaller the increase in the load required to produce some incremental growth dA = A - A l of the plastic zone, the larger is the deformability. With different dislocation structures in the virgin ceramic and generated during indentation as outlined above, the deformability may change with proceeding penetration. Without hypothesizing any mechanism or specific character of this change, two different symbols Si and 6,distinguish the following situations. For the preceding penetration at loads P < P I , the average deformability Si is simply expressed by

Note that Si characterizes an initial deformability only when PI is very small, and that 6i gives dzflerent averages for increasing values of P I . With Eq. (la), the latter dA = A , , and the expression for Si integrates to integral equals

Jt'

Si = AI/P1. (1b) On the other hand, with Eq. (la) the deformability 6,for the further extension of the plastic zone at PI is 6,

PI, A , )

(dP/dA)ibp,,A=Al, (lc) where the considerations discussed above suggest 6, 2 Si. In the hardness definition H = a P / ( 2 ~ 2 )( P ~ = testing load, a = 1.8544 for the Vickers pyramid), 2a can be replaced by the size of the plastically deformed zone 2Rp1as far as the ratio n = 2Rp1/(2a)is approximately constant. With the characteristic area A = (2Rp1)2,the hardness becomes =

H = o'P/A, Pa) where the parameter a' = n 2 0 depends on the value of n = RpI/a. In the associated differential equation

dH -dP H P

dA A

190


dA/A can be substituted by means of Eq. (lb) by an expression that depends on the load P only: dA--dA A

- Se(P1)dP

A1

=

A1

+

(;)p=p,(E) =

(Se/Si> d P P

(2c)

For finite load ranges P I -+ PI AP, and Eq. (2b) can be integrated if a slowly varying ratio Se/Si is regarded as a small perturbation within this load interval. The result is H(P)=

(g)

H [(6c/6i)-ll ,

(3)

with H , = H(P,) and a scaling parameter P, that generally can be chosen arbitrarily. The choice of a large P,, however, offers the advantage that H , acquires (P 4 00). the character of an asymptotic (constant) level H , = Hasympt It is only here that the knowledge about the processes of microplastic deformation becomes physically important for the model: if the deformability 6,for the further extension of the plastic zone is larger than the average deformability Si for the preceding penetration at any PI, the ratio Se/Si > 1 attributes a specific character to the load influence in Eq. 3, and the hardness decreases with increasing load in a way that agrees qualitatively with the measured data obtained for a Czochralski grown sapphire crystal in the load range 3-100 N (Fig. 3 , interrupted line). To obtain reliable data which are really characteristic for the single crystalline material (independent of special features associated with one or another orientation),

1.o

I

I

2.0

4.0

I

I

I

7.0 10.0 20.0 indenter load (N)

I

40.0

I 1 70.0 100.0

Figure 3. Conventionally me2sured Vickers hardness on sapphire ((1510) surface, Vickers diagonals parallel with the traces of (1012) and (1102) p 1 a n e s ) m T h e interrupted line is the fit according to Eq. (3) with a constant (average) ratio (Se/Si) resulting in H , = 15.40GPa and (Se/S,) = 1.119. An only slightly different fit results with the varying (S,/Si) of Eq. (3a) ( H , = 16.20GPa, ( & / c ? ~=) ~1.115).

6.4 Influences of the Gruin Size and the State of the Surface

191

investigations were performed on (1210) sapphire surfaces which offer the opportunity of comparing hardness values associated with most difficult and most easy indentation fracture. One series of measurements was performed with the diagonals of the Vickers pyramid parallel with the traces of crystallographic (lOT0) and (0001) planes, and another one with the diagonals nearly parallel with two (1072) planes (the basis (0001) shows the highest resistance to cracking, whereas { 1072) planes exhibit a minimum fracture toughness [14,15]). When the diagonals are aligned with (1070) and (OOOl), the cracks only start with this orientation but change to the (1072) system a few micrometers beyond the boundaries of the indent [13]. No significant differences in the hardness were observed for the two orientations. With a constant ratio Se/Si > 1, Eq. (3) gives a hardness that decreases infinitely for very large testing loads (interrupted line in Fig. 3). To understand the asymptotic character of the function H ( P ) . the (slow) variation of (Se/Si) with P needs to be taken into consideration: for very large values of the load, the deformability 6, which characterizes the further extension of the plastic zone approaches the average deformability Si for the preceding penetration (due to A l >> AA at P I + 00). Then, the exponent in the denominator approaches zero, the denominator becomes unity, and H ( P ) approaches a constant value H,. It has been shown [16] that the exponent of the denominator in Eq. (3) may be described, for example, by

with = (he/Si)p+o. Since (S,/Si) as given by Eq. (3a) is a very slowly varying function, fits according to (3a) or using an average (constant) value show only small differences in practically interesting load ranges (Fig. 3), and it is obvious that use and the form of Eq. (3a) has little practical meaning for the derived load effect over the experimentally investigated range. The expressions (3) and (3a) give a load influence that compares well with measurements in single crystalline A120, (Fig. 3). However, it should be emphasized again that with the phenomenological character of the model this agreement is not an evidence for some specific micromechanism, it just means that there is no general discrepancy between the dislocation idea and the observed indentation size effect.

6.4 Influences of the Grain Size and the State of the Surface 6.4.1 The Grain Size Influence on the Load Effect of the Hardness: Modeling Experimental Results Previous conventional and depth sensing Vickers hardness measurements performed on sapphire and with sintered traditional A1203 ceramics (grain sizes 2-10 pm,

192


30

Sintered Al 203 with different grain sizes

Sapphire

10

:

1.o

I

3.0

I 6.0

I

I

10.0 indenter load (N)

30.0

I

60.0 100.0

Figure 4. Load dependence of the Vickers hardness of sapphire (same data as in Fig. 3) and of finegrained sintered alumina ceramics (relative density 99%, purity > 99.7% A120,) [16].

>

99.8-99.9% A1203)[ 131 were supplemented by measurements with advanced more fine grained (0.4-1.5 pm) and highly pure (99.99%) microstructures [17]. Figure 4 gives the results of conventional load-dependent measurements of the Vickers hardness in different fine grained alumina ceramics. Generally, the hardness of the polycrystals is (of course) higher than the hardness of sapphire, but it is also obvious that the increasing hardness of more fine-grained microstructures is significant at larger indentation sizes only, associated with a decreasing load (size) influence. With Eq. (3/3a), this decreasing influence of the load in more fine-grained microstructures is expressed by smaller values of the ratio (Se/Si)o. Table 2 displays fitting parameters for the experimental data of Fig. 4: for smaller grain sizes, increasing asymptotic hardness values H , are associated with decreasing parameters ( C ~ ~ / SIn~ submicrometer )~. alumina microstructures, the increase of the microplastic deformability (the decrease of the hardness) becomes smaller and smaller and approaches zero already at small indent sizes of 10-20pm (see Fig. 4), and 6, which characterizes the extension of the plastic zone approaches the ‘initial’ deformability Si at rather small loads of about 1 N. Table 2. Parameters of a least squares fit (Eqs (3) and (3a)) for the load effect in the hardness of alumina ceramics with different grain sizes (Fig. 4). Average grain size (pm)

(bel4)O

H , = VH (P= 98.1 N) (GPa)

0.2 0.5 3 Single crystal (sapphire)

1.04 1.04 1.09 1.115

20.5 19.7 18.0 16.2

6.4 Influences of the Grain Size and the State

of

the Surface

193

The measurements in Fig. 4 demonstrate that any polycrystal exhibits a hardness like the single crystal if the load is small enough (if only a couple of grains are affected). A difference appears only when, on increasing the load, the indentation size becomes much larger than the grains: whereas in sapphire the deformability increases with the growing plastic zone, in the polycrystals this size effect is partly offset by the hindrance of dislocation activity due to the close spacing of the grain boundaries. This deviation from the single crystalline behavior starts earlier, the smaller the grain size is (limiting at higher testing loads the increasing deformability, i.e. the decrease of the hardness).

6.4.2 The Effect of the Grain Size and the Surface State in Ceramics when Recorded by Different Measuring Approaches Figure 4 clearly demonstrates the reason of the declining grain size effect at decreasing testing loads (smaller indentation size). A more conventional presentation is given in Fig. 5. A similar grain size influence is recorded by Vickers indents at 10 kg (HVlO) and 1 kg (HV1; HV1 > HVlO due to the ISE), and close results are obtained by HV1 and by a Knoop approach at 100 g (HKO. 1) because the longer of the two Knoop diagonals exhibits a similar length at a load of 100 g to the Vickers diagonal size at HVl. With further decreasing sizes of the plastic zones in the

0

0

a

22

-

0-

s9!

m

4

g20m m C al

T?

2 18HKO.l

2

-

t - - - ~

al

*---

Y 0

5

.

HV1 HVlO

polished surfaces high-purity Al (no doping additives) 14 0.1

0.2

0.4

0.6 0.8 1.0

2.0

4.0

D (w)

Figure 5. Investigations of the grain size effect in alumina ceramics by using different approaches [6]. Each data point indicates the average value of 6-15 indents.

194


ground

24-1

22 -

polished

i' 4200

2 $18C

: r

16-

14

I

0.1

0.2

I

I

0.4

I

I 1 1 1 1

I

0.6 0.8 1.0 D (PC.Lm)

2.0

I

)

4.0

Figure 6. Influence of the surface preparation on the hardness of sintered alumina ceramics with different grain sizes. In the submicrometer range, grinding results in a higher hardness than measured on polished surfaces.

Vickers microhardness (400 g, HV0.4), however, the grain size effect becomes smaller, and a similar hardness is derived for the single crystal (sapphire) as for an even extremely fine-grained 0.2 pm polycrystal. Figures 6 and 7 show that the hardness of ground surfaces may be higher than measured on polished surfaces (unfortunately, no detailed investigation of this effect is known in ceramics). The reason for this seems to be similar to the grain size influence. Whereas no significant difference exists in alumina ceramics with grain sizes >1 pm, an increasing advantage of the ground state is observed for fine-grained materials with HV10 > 20 GPa (submicrometer A1203, A1203Ti(C, 0,N) composites). TEM has revealed extended dislocation activity in both ground and in polished alumina surfaces [18], but the depth of the affected surface layer was very different: the maximum depth of damage was about 20pm in ground (40pm diamond) but less than 2pm in polished (0.25pm diamond) surfaces. Therefore, the most probable reason for the different hardness in ground and in polished ceramics is that a high density of grinding-induced dislocation and twin networks in the machined surfaces reduces the microplastic deformability (increase the hardness) at an indentation site, in a similar way to small grain sizes. Within the framework of the understanding outlined above it is clear that after grinding, a subsequent further microplastic deformation (on indentation) must be the more difficult the less mobile the grinding-induced dislocations are, that is the closer the spacing of grain boundaries, the smaller the grain size is. As a

6.5 Comparing the Grain S i x E j e c t and the Indentation Size Effect

T

14

I

'

-

raw mixture synthesized via Al-titanate

I

'

l

l

1

I

I

1%

I

consequence, the influence of surface preparation on the hardness is stronger in hard, very fine-grained microstructures (Figs 6 and 7).

6.5 Comparing the Grain Size Effect and the Indentation Size Effect: The Role of Grain Boundaries at Room Temperature Figure 8 compares the Vickers hardness HVlO (P= 98.1 N) of alumina ceramics with the experimentally observed size effect in sapphire described by Eq. (3) (including the load-dependent exponent given by Eq. (3a)). The curves for n = 1 and n = 4 represent extrapolations of the experimentally measured size effect on the hardness of single crystalline alumina. As additional information, the curve of the indentation size effect of sapphire in Fig. 8 was complemented by data measured and analyzed by Oliver and Pharr [I91 on a (0001) plane. It is very remarkable that neither the different crystals and orientations, nor the different (depth sensing) measurements (Vickers-Berkovich) and the very different data analyses affect the character of the size effect observed in sapphire: it can be described by one curve according to Eq. (3)/(3a) for available data in an extremely wide range of testing loads between 1 mN and 100N. With the two parameters H , and (6e/Si)o = (6e/6i)p+o (Table 2), it is possible to analyse the experimentally recorded dependence of the load effect in sintered

196

6 Interrelations Between Influences For indentation size effect of sapphire: 2a (pm) for 1.o

0.1 I

I

I t

4

I

I

I

l

l

n=4

n=l

n=l ,2R,, (pm) for n=4

10.0 I

I

I

l

l

1000.0

100.0 I

I

I

l

l

I

I

I

--- - - = - - - - -

' I

I

I

0.1

I

I

I

I 1

1.o

l

L

(curves derived from measured load effect in sapphire)

+-

HVlO - grain size effect measured in AI,O,

10

l

I

I

I

l

l

I

I

I

10.0

l

l

100.0

I

I

I

I [

I

1000.0

For HV10: Grain size D (prn)

Figure 8. Grain size effects in the hardness of alumina ceramics (HV-10, triangles), and indentation size effect (load influence) in the hardness of sapphire (squares). For HVlO values, the lower x-axis gives the average grain size (data for sapphire indicated at 1OOOpm). For the curves describing the size effect in sapphire, the upper x-axis represents the length of the Vickers diagonal (n = 1 ) or the plastic zone size (n = 4) [6,16]. See Fig. 2 (p. 166) for theoretical background.

alumina ceramics on the grain size. Note that the deformabilities 6, and Si are phenomenological parameters that describe the microscopic deformability without assuming specific mechanisms. For the use of these parameters within the framework of the model introduced in Section 6.3.2 it is, therefore, unimportant whether inelastic microdeformations are generated by lattice processes only (dislocations, twins), or if additional grain-boundary deformation takes place. If grain-boundary processes are active, they will increase the deformation and reduce the hardness. Therefore, it is not a surprise to find in Fig. 8 a quite different grain size dependence for polycrystalline alumina ceramics (HV 10) compared with the (also experimentally measured!) size effect in the single crystal (sapphire, represented by the curve derived under the assumption of a plastic zone size 2R,1 = 4 x 2a). The degree of this difference, however, is much larger than might be expected: even microstructures with coarse grain sizes of the order of 10ym exhibit a much higher microplastic deformation (lower hardness) than isolated sapphire subregions of the same size. It should be emphasized that this conclusion is independent of theoretical speculations: all data points in Fig. 8 are measured values. Only the extrapolation

6.5 Comparing the Grain Size Effectand the Indentation Size Effect

191

of the indentation size effect in sapphire towards smaller sizes depends on the model, but this extrapolation was performed in a way that agrees well with published nanoindentation results. Comparing in Fig. 8 microstructures with average grain sizes of about 10 and 0.2 pm, respectively, the dzfeerence between the hardness of these plycrystalline microstructures on the one hand, and of plastically deformed single crystalline volumes that exhibit the same size as the grains in the polycrystals on the other hand, is almost doubled when fine-grained materials are compared with coarser sizes, and it will further increase for nanoceramics. Therefore, with the data in Fig. 8 it seems rather improbable to find a sharp increase of the hardness in future nanoscale corundum microstructures with grain sizes of 50nm or less. Additionally to efforts intended to reduce the grain size, grain-boundary engineering is, therefore, a second approach to the development of hard polycrystals, which becomes the more important the more fine-grained the material is. For example, the ratio between the contributions of lattice processes and of grain boundaries to the inelastic deformation at an indentation site may be quite different if different types of bonding (ionic, covalent) are considered. In fact, first investigations indicate an improved resistance against inelastic grain boundary microdeformation in submicrometer microstructures with a covalent bonding character if these materials are produced without ionic or metallic binder phases. Figure 9 compares highly-pure sintered alumina (>99.9% A1203) and binderfree (!) ceramics based on WC and (Ti, W)C [20]. Whereas all these materials exhibit a similar hardness when microstructures with average grain sizes 1 4 p m are

30

\v WC\\

25 -

\? \

(Ti,W)C

*

\

a m

s?

pure AI,O,

0

-

I 20> (I)

\T\

0

*'

*** *.

\

\ . - m

(I)

0

acl

*=

\

E

\

c

\

+\

* \ \ \

15-

T \

I

0.1

\

blished surfaces]

0.2

I

I

0.4

l

~

l

~

0.6 0.8 1.0

l

l

I

2.0

1

\

1

4.0

1

198


considered, a difference appears in the submicrometer range. With grain sizes <1 pm, the slope of the curve decreases for alumina, indicating an increasing contribution of grain-boundary deformation. The carbides, however, show a steady increase of the hardness towards smaller grain sizes.

6.6 The Effects of Temperature on the Hardness of Ceramics This section considers the direct influences of the temperature on the hardness, excluding secondary consequences when phases become thermodynamically unstable. Figure 10 compares the temperature effect in the hardness of two single crystalline materials with a different bonding character (A1203,Sic) with the behavior of polycrystalline alumina ceramics. The authors did not discuss the unusual result of a constant hardness of Sic between room temperature and about 400°C. It is, however, a general observation that in single crystals the influence of the temperature on the hardness decreases at high temperatures >lOOO°C, whereas a much larger effect is observed at room temperature and associated with a further increase of the hardness below 0°C due to the thermally activated character of dislocation movement (contrary to the strength, which is almost constant up to about 600-1000°C in most ceramics). Below 1000°C, Fig. 10 displays a similar temperature effect on the hardness of single and polycrystalline A1203. The one difference is that at room temperature

0

I

I

I

I

200

400

600

I

1000 Temperature (OC)

800

I

I

I

1200

1400

1600

Figure 10. Temperature dependence of the hardness in highly ionic (A1203)and highly covalent (Sic) single crystals compared with polycrystalline (sintered) ceramics [21].

6.7 Sunitnary

199

Figure 11. Intergranular damage at the edge of a 4 kg indent in a 2.5 pm alumina ceramic at 1600 C P11.

the 9 pm material is less hard than the fine-grained ceramic due to the usual grain size effect and less hard than the sapphire crystal (probably associated with some defects). In the range 500-1 000°C the disadvantage of the coarser microstructure compared with the 2.5 pm ceramic is offset because it is less susceptible to inelastic grain boundary deformation due to its lower frequency of grain boundaries per unit volume. Altogether, however, the influence of the microstructure on the effect of the temperature is small at temperatures < 1000°C. The situation changes when at some critical temperature > 1000°C weakening grain boundaries cause a global instability of the microstructure at the indentation site (Fig. 11). This temperature range is of special interest for ceramic tools, because both cutting ceramics and grinding materials have to resist temperatures of more than 1000°C at processing zones (cutting edges and tips). The consequence of the general instability of the grain boundaries at high temperatures is the decrease of the hardness to values which are much smaller than observed for the single crystal. It is obvious that the critical temperature for this drop of the hardness depends strongly on the grain boundary structure (amorphous interfaces, segregating species) and on the grain size.

6.7 Summary The measured size effect on the hardness of single and polycrystalline alumina can be fitted assuming an increasing deformability, S,, when the extension of the plastic zone proceeds by growth and multiplication of pre-existing elements of plasticity instead of the more difficult generation of new dislocations, dislocation loops, or twins. With increasing testing load (size of the indent), there is a decreasing hardness due to the involvement an increasing number of pre-existing dislocations but, on the

200


other hand, with increasing size of the plastic zone the value of 6, approaches the average deformability Si of the preceding stages of the penetration process, and the degree of the load effect decreases. As a consequence, the hardness approaches an asymptotic value H , at larger testing loads. Smaller grain sizes restrict the microplastic deformation to smaller volumes, and the process approaches H , at a smaller size of the indent. Therefore, the influence of the load (size) on the hardness decreases in more fine-grained microstructures. The grain size effect is larger when larger testing loads are applied, it disappears at small loads. Hence, microhardness data should not be used to study grain size effects in the hardness. Comparing the hardness of polished and of ground surfaces, there is a tendency for higher values in the latter. Whereas no significant difference exists in alumina ceramics with grain sizes >1 pm, an increasing advantage of the ground state is observed for fine-grained materials with HVlO > 20 GPa (submicrometer A1203, A1203-Ti(C,0,N) composites). The most probable reason is that grinding-induced dislocation and twin networks in the machined surfaces reduce the microplastic deformability (increase the hardness) at an indentation site in a similar way to the effect of small grain sizes. A remarkably large discrepancy between the hardness of plastically deformed single crystalline volumes and of polycrystalline microstructures with grains of the same size, indicates a significant contribution of grain-boundary deformation at the indentation site in polycrystalline alumina, even at room temperature and even for coarse microstructures. The increase of grain-boundary contributions to the deformation at small grain sizes makes a sharp increase of the hardness in nanoceramics like alumina highly improbable. Additionally to efforts that reduce the grain size, grain boundary engineering is, therefore, a second approach to the development of hard polycrystals, which becomes more important the more fine-grained the material is. First investigations indicate an improved resistance to inelastic grain-boundary microdeformation in submicrometer microstructures with a covalent bonding character if these materials are produced without ionic or metallic binder phases. Under this condition, an increase of the hardness with decreasing grain sizes may still exist at grain sizes of lOOnm or less. Different mechanisms control the temperature effect in the hardness below 1000°C and at higher temperatures when weakening grain boundaries induce a loss of the global stability of the microstructure. At lower temperature, thermally activated processes result in a similar influence of the temperature in polycrystalline ceramics to that in single crystals, and the influence of the microstructure on the temperature effect is small. A steady decrease of the hardness with increasing temperature is observed between room temperature and about 1000°C, with a decreasing slope the higher the temperature is. The influence of the temperature again increases when, beyond some critical temperature, weakening grain boundaries induce a drop of the hardness to much lower values than observed for the single crystal at same testing load and temperature. The influence of the microstructure (e.g. of amorphous grain-boundary phases and of the grain size) on this critical temperature is technically important because many ceramic tool materials

Rejerences

20 1

run under conditions where the local temperature at the processing edge or tip exceeds 1000°C.

References 1. A. Krell and P. Blank, A1203-Sinteringmaterial, process for the production of said material and

use thereof, European Patent EP-756 586 BI, Int. CL6 C04B 35/11 I , published 29 October 1997. 2. B. J. Hockey, Plastic deformation of aluminum oxide by indentation and abrasion, J . Am. Ceram. SOC.1971,54, 223-231. 3. D. E. Munson and R. J. Lawrence, Dynamic deformation of polycrystalline alumina, J . Appl. Phys. 1979,50,6272-6282. 4. R. A. Graham and W. P. Brooks, Shock-wave compression of sapphire from 15 to 420kbar, J. Phys. Chem. Sol. 1971,32,2311-2330. 5. A. Krell, Gezielte Mikrostrukturentwicklung in Sinterkorund, PhD Thesis, Academy of Sciences/Central Institute for Solid State Physics and Materials Science, Dresden, 1982. 6. A. Krell, Effects of load, grain size, and grain boundaries on the hardness of alumina, Proc. 22nd Cocoa Beach Conference (2&24 January 1998), American Ceramic Society, Westerville, OH, 1998, pp. 159-168. 7. A. Krell and P. Blank, Inherent reinforcement of ceramic microstructures by grain boundary engineering, J. Europ. Ceram. Soc. 1992, 9, 309-322. 8. J. Lankford and D. L. Davidson, The crack-initiation threshold in ceramic materials subjected to elastic/plastic indentation, J . Muter. Sci. 1979, 14, 1662-1668. 9. G. P. Upit and S. A. Varchenya, The size effect in the hardness of single crystals, in The Science of Hardness Testing and its Research Applications, J. H. Westbrook and H. Conrad (Eds), The American Society for Metals, Metals Park, OH, 1973, pp. 135-146. 10. C. M. Perrott, Elastic-plastic indentation: Hardness and fracture, Wear, 1977,45,293-309. I I . A. Krell, J. Woltersdorf, E. Pippel, and D. Schulze, On grain boundary strength in sintered A120,, Philos. Mag. A , 1985, 51, 765-776. 12. A. Krell, Improved hardness and hierarchic influences on wear in submicron sintered alumina, Muter. Sci.Eng. A , 1996, 209, 156163. 13. A. Krell, Vickers hardness and microfracture of single and polycrystalline AI2O3,Cryst. Sci. Technol. 1980, 15, 16667-16664. 14. S . M. Wiederhorn, Fracture of sapphire, J . Am. Ceram. SOC.1969, 52, 485-491. 1976, 59, 59-65. 15. P. F. Becher, Fracture-strength anisotropy of sapphire, J . Am. Ceram. SOC. 16. A. Krell, A new look at grain size and load effects in the hardness of ceramics, Muter. Sci. Eng. A245, 1998, 277-284. 17. A. Krell and P. Blank, Grain size dependence of hardness in dense submicrometer alumina, J. Am. Ceram. SOC.1995,78, 1118-1120. 18. B. J. Hockey, Observations by transmission electron microscopy on the subsurface damage produced in aluminum oxide by mechanical polishing and grinding, Proc. Br. Ceram. SOC. 1972,20,95-115. 19. W. C. Oliver and G. M. Pharr, An improved technique for determining hardness and elastic modulus using load and displacement sensing indentation experiments, J . Muter. Res. 1992, 7, 15641583. 20. V. Richter, M. v. Ruthendorf, and J. Drobniewski, Perspektiven feinkorniger Hartmetalle, in Pulvermetallurgie in Wissenschaft und Praxis, Vol. 13, Proc. Hagen Symposium Powdermetallurgy, 13-14 November 1997, R. Ruthardt (Ed.), Werkstoff-Informationsgesellschaft, Frankfurt, 1997, pp. 29-45. 21. A. Krell and 0. V. Bakun, High-temperature hardness of A1203-base ceramics, Acta Metall. 1986,34, 1315-1319.


7 Transition Metal Carbides, Nitrides, and Carbonitrides W. Lengauer

7.1 Introduction Transition metal carbides, nitrides, and carbonitrides belong to a class of materials that show an unusual combination of outstanding properties, amongst which are exceptional hardness, high melting point (e.g. TaC and HfC have among the highest known melting points of all materials), metallic lustre combined with vivid colors, and simple metallic structures. In addition, they show high electrical and thermal conductivities. This combination of properties has attracted considerable attention and has resulted in numerous technical applications [ 11 and theoretical investigations. Many binary transition metal carbides and nitrides especially the face centered cubic 6 phases of the IVB group, exist over broad ranges of composition with an upper limit of the nonmetal/metal ratio near 1. Practically all solid state properties show a gradual change with this ratio. Other carbides with a different structure such as tungsten carbide and chromium carbides have a very narrow homogeneity region. Transition metal carbides and nitrides often form solid solutions resulting in carbonitrides in which the solid-state properties also change with the [C]/([C] + [Nl) ratio. This behavior offers a large field for tailoring carbonitrides. The carbides, and to a minor extent also nitrides and carbonitrides, are easily wetted by liquid metals, which has found application in hardmetals and cermets. Since the basic invention of preparing composite materials by liquid phase sintering of carbides with iron group metals, where a very successful combination of hardness with toughness was achieved, these materials are applied as components in hardmetals for a variety of cutting and drilling operations. In steels, carbide and carbonitride precipitates cause hardening known as dispersion hardening. The chemical stability of transition metal carbides, nitrides, and carbonitrides is high, they are not readily attacked by dilute acids (except oxidizing acids and hydrofluoric acid) or alkaline solutions. The thermal stability is connected with their free energy of formation, the stability decreasing with increasing group number. While the group IV transition metal nitrides TIN, ZrN, and HfN and carbonitrides Ti(C, N), Zr(C, N), and Hf(C, N) can be melted without decomposition, the nitrides and carbonitrides of the other groups decompose before melting points are reached. As a result nitrogen is released. Tables 1 and 2 give a short overview of the properties important for technical applications of the most well characterized and important binary transition metal carbides and nitrides. Some specific properties which cannot easily be listed in a table, as well as the properties of transition metals carbonitrides) especially those of the most commonly used group IVB carbonitrides) are described in detail later.

Pearson symbol

cF8 cF8 cF8 cF8 cF8 cF8 0P20

hP3

hP2

Compound

TiC ZrC HfC VCO.88 NbC TaC Cr3C2

Mo~C

WC

0.4328 0.4700 0.4638 0.4164 0.4470 0.4455 a=0.28290 h = 0.55329 c = 1.14719 u = 0.30233 c = 0.47344 a = 0.29058 c = 0.28365

Lattice parameter(s) (nm)

28 25 20 26 18 16 27 17 23

4.93 6.46 12.3 5.36 7.78 14.48 6.68 9.18 15.72

(gcm-')

Microhardness (GPa)

Density

2776

2520

3067 3420 3930 2650 3610 3985 1810

Melting point ("C)

707

530

450 350 420 430 340 290 380

Young's modulus (GPa)

19

15

28.9 24.6 25.1 26.8 27.0 22.1 14

Heat conductivity (Wm-' K-

3.9

7.8

8.5 7.5 6.1 7.2 6.6 6.3 10.3

Linear thermal expansion coefficient (IO-~)K-'

17

51

75 67 69 20 15 75

100

Electrical resistivity (FQ cm)

Table 1. Bulk properties at room temperature of some important transition metal carbides (disordered state). For f.c.c. carbides (Pearson symbol cF8) the composition is near 50 at-% C, except for VC0.88.The melting points do not apply to this composition.

6'

k

u

cF8

VN NbN TaN TaN

CrN

* d = decomposes

HfN

TiN ZrN

cF8 cF8 cF8 cF8 cF8 cF8 hP3

Phase Pearson symbol

0.4242 0.4570 0.4516 0.4138 0.4392 0.4238 a = 0.2936 c = 0.2885 0.4148

gray brown

6.14

13

17 15 18 5.7 11 32 11 d

3050 3000 3330 2350 d d d

MicroMelting hardness point* ( g ~ m - ~ (GPa) ) (“C)

Density

golden yellow 5.39 green yellow 7.32 dark yellow 13.83 brown 6.04 light yellow 8.16 gray yellow 15.9 dark gray 14.3

Lattice Color parameter( s) (nm)

450

~

~

420 460 380 380 360 ~

11.7

9.5

~

3.8

11 11 11

29

-

~

9.9 7.8 8.5 10.8 10.2 8.0

640

128

-

27 24 27 65 60

-

-

5.8 10.47 6.92 8.9 17.2 8.9

Young’s Heat Linear thermal Electrical T, modulus conductivity expansion coefficient resistivity (GPa) (Wm-’ K- ) (10-6K-’) (PQcm) (K)

Table 2. Bulk properties of important transition metal nitrides. Values for room temperature and for stoichiometric composition.

h,

2.

2.

-

?I

3

$

. 3 ?

2

5

$.

?I

5k

3

z.g.

3 a

u

0 P

7.2 General Features of Structure and Bonding

205

7.2 General Features of Structure and Bonding 7.2.1 General Structural Features Transition metal carbides, nitrides, and carbonitrides belong to the family of interstitial alloys or compounds. The common features of this class of materials are the very simple metallic structures with small carbon and/or nitrogen atoms in the interstitial voids of the often densely-packed host lattice. They generally have face-centered cubic (f.c.c.) or hexagonal close packed (h.c.p.) metal lattices with nonmetal atoms more-or-less randomly distributed on the interstitial sites. The metal atoms are arranged in such a way that they form close packed arrangements of metal layers with ABAB.. . (hexagonal, h) or ABCABC.. . (cubic, c) stacking sequence (Fig. 1) or with a mixture of these. The nonmetal atoms in these phases occupy the octahedral interstitial sites. According to a crystallochemical rule [2,3] the phases of pure h type can have a maximum nonmetal/ metal ratio of 1/2 and the c type phases have a maximum carbon content of 1. The phenomenon of vacancy ordering is often observed especially in the transition metal carbides [4,5]. In the carbides of the group IVB and VB transition metals the common structural element is the T6X octahedron. With increasing radius ratio r\./rT the trigonal prismatic group T6X is favored over the octahedral group. Therefore the transition metal carbides of groups VIB-VIIIB feature trigonal prismatic T6X elements in different crystallographic arrangements. Deviations from the simple close packed structure are more frequently encountered with nitrides where T5X quadratic pyramids are observed as another structural element. This is the case particularly in group VB compounds. These prevail even in the complex nitrides, such as the Z-phases NbCrN, TaCrN, NbMoN, and TaMoN, with the filled-up BCuTi-type structure. With this quadratic pyramid, which has lower symmetry than an octahedron, such nitrides are more complex. Because of the introduction of structural elements other than T6N octrahedral, the f.c.c. structure of 6 nitrides is destabilized increasingly with increasing group number, and (within a group) with increasing periodic number. The nitrides of group IVB,

E h

c

Figure 1. Hexagonal (h) and cubic (c) stacking sequence of metal atoms of close-packed h.c.p. and f.c.c. transition metal carbides and nitrides. Open large circles designate metal atoms, filled small circles designate interstitial nonmetal. In h sequences only 50% of the interstices can be occupied. Nitrogen and carbon atoms are randomly distributed among the nonmetal sublattice.

206

7 Transition Metal Curbides, Nitrides, and Carbonitrides

FTiNl --x, GZrN1--x, and 6HfN1px, crystallize in the sodium chloride structure and no low-temperature modifications of these stoichiometric nitrides are known. In the VB group, also all three metals form stoichiometric f.c.c. nitrides, but GVNl.oo transforms at 205K into a tetragonal modification [6], FNbNl,ooat around 13101320°C into hexagonal qNbN [7] and GTaNI.ooat around 1800°C into hexagonal ETaN [8], the transformation temperatures depending on the nitrogen equilibrium pressure. In the VIB group these f.c.c. nitrides are increasingly less stable. GCrN1--x requires a relatively high nitrogen pressure of preparation and transforms at below ca. 280 K into a tetragonal antiferromagnetic compound [9] and yMoNl --x occurs only at up to 38 at-% N using flowing ammonia, a value that could be increased if the nitriding potential is additionally increased. This stability feature continues further to the Mn-N system, where the f.c.c. phase has the stoichiometry T4N and to the Fe-N system. In the latter only the f.c.c. solid solution of nitrogen in yFe is observed [lo] (the Fe4N phase has f.c.c. structure with respect to the metal positions only).

7.2.2 General Features of Bonding A mixture of metallic, covalent and ionic components prevails in the bonding of transition metal carbides, nitrides, and carbonitrides. The metallic character is shown by the high electrical conductivities of these compounds. The bonding mechanism has been described extensively by a variety of approaches for calculating the density of states (DOS) and hence the electron density in f.c.c. transition metal carbides, nitrides, and oxides [l 11. In the DOS of these compounds there is a minimum at a valence electron concentration (VEC) of 8, which corresponds to the stoichiometric composition of the group IVB carbides TIC, ZrC, and HfC. Transition metal carbides have a lower DOS at the Fermi level than the corresponding transition metal nitrides, hence the electrical properties such as electrical and thermal conductivity and the superconducting transition temperature, T,, are lower than those of the nitrides. Although transition metal nitrides and carbides show sometimes complete mutual solubility, differences due to the different bonding character of nitrogen as compared to that of carbon can be encountered. Figure 2 compares the decomposed and the total DOS of TiN with T i c [12]. At low energies the s band is characterized by a large contribution of nonmetal 2s state. At intermediate energies the p band follows and there is a significant contribution of d states to this band. At higher energies the d band composed of d states follows with the Fermi energy, EF, located well within this band. The DOS at EF is higher for TiN than for Tic, corresponding well with the observed superconducting properties (TIN: T, = 5.4 K, TIC: T, < 1.2 K) and room-temperature electrical conductivities. The covalent contribution to bonding in these compounds can be drawn in the molecule orbital (MO) scheme [13], where atomic orbitals of atoms are combined to form bonding and antibonding states. It could be shown that eg bonding symmetry in T i c changes to t bonding 26 symmetry in TIN. Thus the change in bonding strength from a strong Ti-C interaction in TIC towards a more pronounced Ti-Ti interaction in TiN is to be expected.

7.3 Preparation

-0.6

-0.4

u

0.4 E (Ryd)

207

I ?

Figure 2. Density of states (decomposed into the different contributions and total) of Tic and TIN [12].

The ionic contribution to the binding mechanism in these compounds can be estimated from the charge transfer which occurs from the metal atom to the nonmetal atom. It is about half an electron, which contributes to the electrostatic interaction of metal and nonmetal.

7.3 Preparation Transition metal carbides can be prepared by direct combination of the elements, preferably as powder mixtures, at elevated temperatures: Ti + C --+ GTiC. If heat is generated by the reaction such as with Ti and C, the reaction is exothermic enough to keep up the reaction temperature. The process can be carried out in form of a self-sustaining high-temperature synthesis (SHS). The reaction front, once ignited, travels through the powder bed producing the nitride and dissipating the heat over a certain period of time. Also nitrides can be produced by SHS techniques.

208

7 Transition Metal Carbides, Nitrides, and Carbonitrides

Likewise carbon containing gases such as methane or benzene can be utilized: Ti + CH4 -+ TIC + 2H2. The preparation of TIC, VC, NbC, TaC, Mo2C, WC, and chromium carbides are important technical processes for the production of carbide powders for hardmetals. Generally, the carbides are prepared by the reduction of oxides with carbon, while Mo2C and WC are made by reacting the metal with carbon such as graphite or carbon black. Solid solution of transition metal carbides and carbonitrides, which also play an important role in the hardmetal industry, can be advantageously manufactured by the simultaneous reduction of oxide or oxide-carbide mixtures in the presence of carbon

+

Ti02 WC + 2 C

--f

+

(Ti,W)C 2CO.

The generally volatile chlorides of the transition metals can be employed to obtain transition metal carbides via the reaction scheme

+

TiC14 CH4 -+ T i c + 4 HC1. This reaction proceeds in the gas phase at temperatures above about 600°C and is usually applied either in chemical vapor deposition of carbide layers on solid substrates or to produce carbide powders with very fine (submicron) grain sizes. In the latter case the nucleation of the carbide must proceed in the homogeneous gas phase. The oxides of the transition metals can be reduced by carbon according to the following overall reaction

+

w O 3 4 c -+ wc

+ 3 co.

Coarse chromium carbide powders can be produced by reducing Cr203 with aluminum powder in the presence of carbon:

3 Cr203+ 6A1+ 4 C -+ 3 A1203+ 2 Cr3C2. Bulk transition metal nitrides can be prepared by reacting the metal or the metal hydride in powdered or compact form with molecular or atomic nitrogen or flowing ammonia. This can be expressed by following examples: Ti + i N 2 -+ TiN,

+ ZrN + H2, MO+ NH3 -+ MoN + +H2.

ZrH2 4N2

-+

The pressure of molecular nitrogen at a given temperature that is necessary to obtain a specific nitride phase is defined by the nitrogen potential (partial free energy of nitrogen) of the nitride. The formation of nitrides from oxides is performed by the aid of carbon as a reducing agent, according to the overall reaction equation Ti02

+2C +iN2

-+

TiN + 2CO.

7.3 Preparation

209

This reaction usually proceeds over several intermediate products and the resulting nitride can contain oxygen and carbon. For the preparation of large crystals (1 cm3) zone-annealing techniques were developed [14],which are similar to those zone melting already known from material purification processes. The starting materials are either solid-state sintered nitride powders or metal powders pressed into rods. The applied temperatures must be very high (around 3000°C) in order to overcome the low diffusivities of nitrogen in nitrides, but they must not exceed the melting temperatures of the compounds in order to avoid incongruent melting and decomposition. Carbonitride powders for industrial applications are prepared by mixing carbide and nitride powders intimately together and heating the mixtures to high temperatures, TIC + TIN + 2 Ti(Co,5No.5), whereas for the case of the Ti-C-N system any proportion leads to a single phase product Ti(C,N1 -J provided that sufficient time or a sufficient high temperature is provided for equilibration. For this reaction the nitrogen equilibrium pressure is important either for preparation of Ti(C,N) or if this compound is heated at high temperatures such as in sintering procedures. Figure 3 gives experimental as

1800 1400

-300 otm -3 0 otm -I

.

otm/

-5k

- our thwmodvnomic estimation our experimental determination +

-7

---X

0

.x’

0I

ZELIKMAN 8 GOROVITS

PORTNOI 8 LEVINSKII KiEFFER 8 o l

0 . 2 0 3 0 4 0.5 0.6 0.7 0.8 0.9 x

Figure 3. Nitrogen equilibrium pressure of Ti(C, N) as a function of composition from [16] with literature data. The composition is x in Tic, -,yN,, which is 1 - [C]/([C] [Nl).

+

210


well as calculated results [15, 161 for the nitrogen equilibrium pressure as a function of the carbon/nitrogen ratio in Ti(C,N, -~r). A great variety of methods exists for the preparation of nitride and carbonitride layers. These can be divided into the groups CVD, PVD [17], and diffusion layer and thermal surface modification techniques. (i) Chemical vapor deposition (CVD): a volatile transition metal compound such as a chloride is brought to reaction with ammonia, nitrogen/hydrogen, or ammonia/hydrogen/nitrogen mixtures. For carbonitrides a carbon-containing gas such as CH4 or other hydrocarbon is used [18]. Also single metal-organic precursors can be used which make the construction of the furnaces more simple. An example is Bis(benzene)vanadium, V(C6H6). which is solid at room temperature, and can be used for deposition of V(C, N ) layers [ 191. The nitride or carbonitride that forms as a reaction product in the gas phase nucleates on solids, like hardmetals, and forms solid, strongly adhering layers at temperatures around 1000°C. Thus hardened steels cannot be coated unless a hardness decrease is accepted. (ii) Physical vapor deposition (PVD): The transition metal is evaporated or sputtered by glow discharge or arc heating, and reacted with low-pressure. Numerous modifications of PVD methods have been developed [20-231. The nitride layers that are deposited on the substrate surfaces often have high lattice defect concentrations [24] and hence exhibit significant deviations from generally accepted bulk values (lattice parameter, hardness, composition) [25]. PVD processes applied to coat hardmetals generally result in a smoother interface than the CVD process because of the lower temperatures. (iii) Diffusional techniques: Carbide, nitride, and carbonitride formation, which is used for surface hardening, can be achieved by subjecting the material surface to the action of various salt baths, ammonia, methane, molecular or atomic nitrogen, or nitrogen ions [26]. Reactive plasma techniques and high-pressure techniques were developed to keep the workpiece temperature low and to avoid thermally induced changes in the workpiece. Laser irradiation together with an N2 jet directed on the heated spot is employed for limited local heating of the sample [27].

7.4 Characterization 7.4.1 Chemical Analysis Carbon analysis of transition metal carbides is nearly exclusively performed by combustion of the carbide in oxygen and measurement of the reaction products C 0 2 or CO. A common technique is vacuum or carrier-gas hot extraction where CO is formed and measured with an IR spectrometer or with a thermal conductivity

7.4 Characterization

21 1

detector. The sample is heated to more than 2000°C in the presence of a bath metal such as platinum in a stream of a carrier gas such as He. Due to the high temperatures involved, the method is suitable for powder samples with larger grain size or even for small solid pieces. The accuracy of the method is on the order of 1-2 rel-% C. In another technique measurement of the change in electrical conductivity of an alkaline solution resulting from the C 0 2 absorption is performed. An oxidizing aid such as V 2 0 5 or copper oxide is used to facilitate the oxidation. Several procedures have been applied for the analysis of nitrogen in nitrides and carbonitrides. The Kjeldahl analysis involves the dissolution of nitrides in acids whereby the nitrogen is converted into ammonium ions. The dissolution process is time consuming since some nitrides, such as GTiN,-,, are chemically very stable against acids and must be dissolved under rigorous conditions by the action of hydrofluoric acid in PTFE containers at 100-200°C. Also vacuum or carrier-gas hot extraction such as for the carbides can be applied. If carbon is present the CO is absorbed after IR measurement and the remaining nitrogen is measured usually in a thermoconductivity cell. Temperature-controlled furnaces can yield a fractional nitrogen analysis which resolves nitrogen differently bonded or different nitride phases. In another combustion analysis the sample is flash-heated at temperatures of up to 1800°C.The evolved N2 is either measured volumetrically in an azotometer or via a gas chromatograph [28,29] equipped with a thermoconductivity detector. N2, COz, and H 2 0 can also be determined simultaneously and hence the C and the H contents. The accuracy of the method is better than 1 rel-% N for finely powdered samples.

7.4.2 Physical Microanalysis Among the microphysical methods for determining nitrogen in nitrides are: electron probe microanalysis (EPMA), Auger electron spectroscopy (AES), ultraviolet and X-Ray spectroscopy (UPS, XPS), secondary ion or neutron mass spectrometry (SIMS, SNMS) and Rutherford backscattering (RBS). EPMA has the highest reproducibility and accuracy, provided that the lateral resolution of not less than a few microns is sufficient. The main problem in EPMA are the rather low energies of the CK, and NK, radiations which requires windowless or ultra-thin window detector systems and results in a low background-to-peak ratio together with low net intensities [30,31]. High-accuracy EPMA requires external or internal standardization [32] together with a wavelength-dispersive spectrometer equipped with high-quality crystals. Recent development of analyzing crystals have yielded multilayer crystals of the type WjSi, NijC and Mo/B4C, which show excellent peak-to-background properties. The background signal obtained with these crystal is more smooth, the sensitivity for high-order intensities is low which facilitates background measurements and avoids line coincidences.

212


A problem of EPMA of titanium nitrides and carbonitrides is the close overlap of the Ti L1 line with the N K, line which cannot be separated by diffraction methods. Therefore, one procedure to overcome this problem is the measurement of the Ti K, line and to subtract the appropriate contribution from the Ti L1 N K, peak for determination of the net N K, peak. If this method is made iteratively the accuracy of the method is within 1 at-% N.

+

7.5 Thermodynamics 7.5.1 Stability of Carbides The thermodynamic stability of the carbides is different between heavier and lighter metals. Stoichiometric composition of the carbides can more easily be attained with the heavier elements of the IVB and VB groups the than with the carbides of the metals of the first long period. Vanadium carbide shows a maximum carbon content of VCo.89.Also chromium does not form a carbide with 50 at-% C, instead a carbide of stoichiometry Cr3Cz.While the monocarbides of the IVB and VB groups can be melted congruently the carbides of the VIB group metals decompose peritectically before their melting points are reached. The monocarbide MoC decomposes in a peritectoid manner already at about 1230°C into Mo2C C.

+

7.5.2 Nitrogen Partial Pressure of Nitrides The nitrogen equilibrium pressure of the transition metal nitrides is a function of temperature and increases substantially with increasing group number. This is reflected in the Richardson diagram [l]. The lower (the more negative) the free energy of formation, the lower the equilibrium nitrogen pressure of the nitride is. Group IVB nitrides can easily be obtained by nitridation with molecular nitrogen under low pressures even at high temperatures, whereas for group VB nitrides the nitrogen pressure must exceed normal pressure at temperatures above 1400"C, and even higher pressures or more energetic nitriding conditions must prevail for the preparation of group VIB nitrides. The nitrogen pressure determines the composition of the nitride if equilibrium is attained. Since nitrogen pressure determines the phase stability, nitride phases with a nonmetal/metal ratio greater than 1 can be obtained if high-pressure nitrogen or high nitrogen potentials, such as those prevailing in flowing ammonia or in atomic or ionic nitrogen atmospheres, are applied. Either an extension of the mononitride phase field to compositions [Nl/[Me]> 1 occur, such as in the N b N , Hf-N, and Ta-N systems, and/or distinct nitride phases, such as in the Zr-N, Nb-N, and Ta-N systems are obtained. Ion implantation techniques can lead to nitrides with substantially more than 50 at-% N, but these phases are probably not in thermal equilibrium.

7.5 Thermodynamics

2 13

7.5.3 Phase Equilibria of Important Carbide Systems A lot of effort has been made to model phase equilibria by the CALPHAD (CALculation of PHAse Diagrams) method 1331. These methods, however, need experimental results, the least information being that a phase occurs or not. Thus a variety of calculated phase diagrams were published which are no longer relevant because other phases have been found. In the following presentation phase diagrams or portions mainly of such experimental investigation are presented and discussed if they have appeared recently. Ordered phases [4,5] are generally not discussed, unless they are interesting for industrial applications.

7.5.3.1 The Group IVB Carbide Systems The group I V B transition metal carbon systems are very similar to each other concerning the carbide phases: the f.c.c. monocarbides T i c l --x, ZrCl --x, and HfCl --x show extensive homogeneity ranges and have high congruent melting temperatures. Carbon does not stabilize the h.c.p. low-temperature modifications of the IVB metals (in contrast to nitrogen) and the solubility of C in the CL phases is much smaller when compared to the IVB transition metal nitrides (see below).

7.5.3.2 The Group VB Carbide Systems The VB transition metakarbon systems (phase diagrams in Fig. 4) are again very similar to each other and contain structurally identically subcarbide phases PMe2C and
c

7.5.3.3 The Group VIB Transition Carbide Systems The crystal structures of the phases in these systems are characterized by destabilization of the octahedral structure elements and the appearance of trigonal prismatic structural elements. F.c.c. monocarbides MeCl --x in the Mo-C and W-C system exist only at high temperatures and with substoichiometric compositions. The stoichiometric monocarbides WC and MoC are hexagonal. The Cr-C system is characterized by the absence of a monocarbide phase. Three carbide phases Cr23C6,Cr7C3,and Cr3C2with crystal structures that deviate from the simple concept of a densely packed metal atom host lattice and interstitial carbon atoms. A recent investigation on the homogeneity ranges have confirmed the accepted values, except for the Cr23C6phase which has found to be lower in carbon. Cr3C2has trigonal prismatic structural elements in an orthorhombic lattice.

214


2000

/

T/OC

"b

1800

I

L

2200

T/ K

2000

1600

1400

40

1200

0

10

20

30

at% C

1800

1600

50

2000

2200

T/OC

T/ K

1800

2000 1600

1800 1400 0 (b)

Nb

10

20

30

at% c

40

50

Figure 4. Recently reinvestigated portions of phase diagrams of the group (a) V-C, (b) N b C , (c) Ta-C [8].

VB

carbide systems

Each carbon atom has eight metal neighbors in the form of a quadratic antiprism. In that respect the structure differs basically from all other carbides.

7.5.3.4 Ternary Transition Metal Carbides There are several families of ternary carbide phases with transition metals and main group elements several of them being of structural interest such as the H-phases, the filled p manganese type carbides and the K carbides. Of technical importance are however only phases which are used as hard constituents in cemented carbides

7.5 Thermodynamics

2 15

2300

T/OC

2500

2200

T/ K 2400

21 00

2300

2000

2200

1900

2100

1800

2000

1700

1'0

I

Ta

20

30

40

50

at% C

(C)

such as (Ti, W)C and the q phases [35] such as W3C03Cof which the occurrence in decarburized compacts (e.g. near interfaces of CVD coated cutting tools) is unwanted causing embrittlement. F.c.c. transition metal carbides often form complete solid solutions with each other particularly at high temperatures. They undergo phase separation at lower temperatures [36,37]. Miscibility gaps have been observed in the quasibinary systems Tic-ZrC, Tic-HfC, VC-NbC, and VC-TaC. At high temperatures also 33

T

31

m

a (3

29

i

Q)

c

P 27

r"0 L

0

2 25 1

23

60

0 70

0.80

0 90

Composition [ C ] / M Figure 5. Microhardness of ordered (B) and disordered (A) VC, --I as a function of composition 1341.

216


complete miscibility is observed in the systems where the f.c.c. phase is a hightemperature phase at least for one binary constituent such as in the group IVB carbide (f.c.c. WC and group VB carbide) f.c.c. MoC quasibinary systems. Even at low temperatures the solid solubility of MoC and WC in the f.c.c. carbides of the IVB and VB groups is substantial and is exploited for use in cutting tools.

7.5.4 Transition Metal-Nitrogen Systems and Structure of Phases Several of the transition metal-nitrogen systems were established only recently and there is still some lack of knowledge for systems for the investigation of which a high nitrogen partial pressure is necessary. As for the carbides systems only recent results will be presented here (see also [lo]).

7.5.4.1 The Group IVB Nitride Systems The industrially very important Ti-N system features a high solubility of nitrogen in aTi, three subnitride phases, qTi3N2-x,(Ti4N3-x and ETi2N, and the f.c.c. phase GTiN, --x with the NaC1-type structure [38]. qTi3N2-x and cTi4N3-x have been found to be isostructural with the hafnium subnitride phases and are characterized by metal atom layers comprising cubic and hexagonal stacking sequences. Figure 6 shows the phase diagram of the Ti-N system. The system was recently modeled by the CALPHAD method [39] where good agreement with the experimental results was obtained. The Zr-N system is characterized by the presence of only one nitride phase, FZrN, --x, up to a composition of [N]/[Zr]= 1. It should be mentioned that none

3500

I

I

T/OC

3500

T/K

3000

3000 2500

2500 2000 2000 1500 1500 1000

1000 500 0

Ti

10

20

30

at% N

Figure 6. Phase diagram of the Ti-N system [38].

40

50

7.5 Thermodynamics

2 17

2500

T/OC

2500

T/ K

2000

2000 1500 1500 1000 1000

500 500

0 0

v

10

20

30

40

50

at% N

Figure 7. Phase diagram of the V-N system [lo] with reinvestigated homogeneity ranges in the temperature range 1100-1700°C (thick lines) [46].

of the subnitride phases which exist in the neighboring systems Ti-N and Hf-N occur here. The Zr3N4phase [40] is obviously not of technical importance. Besides the f.c.c. 6HfNl the Hf-N system contains two subnitride phases [41], qHf3Nz-, and
In the vanadium-nitrogen system the phases BV2N and 6VN1--x have h.c.p. and f.c.c. structure, respectively [6,4245]. A recent investigation of the homogeneity ranges by diffusion couples [46] has yielded a revised version of the phase diagram mainly with respect to the subnitride phase of which the homogeneity was found to reach up to 33.3 at-% N, the most probable value also from a crystallographic point of view (Fig. 7). The Nb-N system has already been subjected to many investigations [47,48]. Near the composition NbN at least three phases have been observed. For qNbN two different structures have been proposed with respect to the nitrogen positions. It contains trigonal prismatic and octahedral T6N structure elements and features a metal layer sequence AABB. qNbN undergoes a phase transition at about 1320°C (depending on the nitrogen pressure) into f.c.c. GNbNI --x with an NaC1-type structure. Another phase, 6'NbNo.95,crystallizes in the anti-NiAs-type structure which is closely related to qNbN but is most probably not stable and occurs only transiently within the transformation of 6NbNl --x into qNbN. After recent in situ investigations by high-temperature XRD [49] and differential scanning calorimetry [7] as well as diffusion couple investigations [7,50] the phase reactions in the Nb-N system seem to be clarified (Fig. 8). The GNbN1-, 4 yNb4N3*xtransition occurs

218


2000 j

j

1400/

P-Nb2N

lzyyl

1000

0

Nb

10

20

30

40

50

60

at% N

Figure 8. Phase diagram of the Nb-N system with reinvestigated homogeneity ranges and phase reactions (thick lines and data points from electron-probe micoanalysis and thick broken line for the yNb4N3+tx+ GNbNI --x transition [7]). The thin solid lines are previous data.

in the temperature range 112O-122O0C, depending on the composition. The transition is of quasi-continuous character without a two-phase field between yNb4Ns*, and GNbNI-,. By reacting Nb powder with nitrogen under high presphase can be extended beyond [N]/[Nb]= 1 sure the composition of the GNbNI-, and the phase Nb5N6can be prepared. The lower stability of the tantalum nitride phases as compared to the nitrides of the IVB metals is reflected in the substantially higher equilibrium pressures [51]. In order to prepare the high-temperature f.c.c. phase GTaN, - ,the nitrogen pressure must exceed 10 bar. It is the hardest known transition metal nitride phase. The pTa2N phase belongs to the &Fe2Ntype at low temperatures and undergoes an order-disorder transition to an L’3 type between 1600 and 2000°C. TaN is expected to melt at about 2800-3000°C and nitrogen equilibrium pressures of 1-3 kbar. The BTaN phase with a WC-type structure is often observed in the reaction products between Ta powder and nitrogen or ammonia at temperatures below 1100°C and is believed to be a high-pressure phase which is metastable under normal pressures [52]. ETaN has a structure not encountered in carbides, featuring an arrangement of quadratic Ta5Npyramids. There are three hyperstoichiometric phases in the Ta-N system with decreasing metallic character. The one with the highest nitrogen content, the red Ta3N5,is nonmetallic and has been synthesized by the reaction of TaC15 with NH3. In industrial applications these nitrides occur in the deposition of tantalum nitrides for electronic devices (see 7.7.2). Recent investigations by diffusion couples have yielded a revised version of a phase diagram of the Ta-N system [8], which has also been modeled recently [53] (Fig. 9a and 9b).

7.5 Thermodynamics

2 19

2000 -

T/OC

2200

T/ K

1800 -

2000

-

1800

1600

1400

0

10

20

Ta

30

40

50

at% N

(a)

3500

*

3000

2500

s!3

c

E 2000

a

0.

+E,

1500

E-TaN

1000

500 0

(b)

t

, 1

5 0.6 0.7 0.8 0:l 0:2 0:3 0:4 C Mole-Fraction N

Figure 9. (a) phase diagram of the Ta-N system from experimental investigations [8]. (b) Phase diagram modeled by the CALPHAD method [53].

7.5.4.3 The Cr-N and Mo-N Systems

The Cr-N system is characterized by the nitride phases pCr2N and GCrNI --x.The former has an arrangement of the metal atoms is h.c.p. and the nitrogen atoms are distributed in an ordered arrangement in the interstitial voids corresponding to the &Fe2N-typestructure. CrN, --x crystallizes in the NaC1-type structure. It has been found only recently [54]that this phase has a large homogeneity range at high temperature and has only a line compound character at temperatures below 1150°C. The phase diagram of this system (recently investigated on the basis of diffusion couples) is given in Fig. 10. The substantial increase of the

220


Figure 10. Phase diagram of the Cr-N system [lo] with recent results on the homogeneity ranges of Cr,N and CrN, - X . The homogeneity range of the latter broadens substantially upon increasing temperature. The nitrogen-rich phase boundary of CrN,-, is for 30 bar N2 and extends most probably at least up to 50 at-% N at higher pressures.

homogeneity range of CrNl --x is interesting and it is not completely clear whether the line compound occurring below about 1200°Chas the same structure as CrNl - x at higher temperature. Cr2N is readily dissolved by dilute acids, whereas CrN appears to withstand acid attack for prolonged periods. CrN is paramagnetic at room temperature and undergoes a paramagnetic-antiferromagnetic first-order transition at temperatures of 276-286 K, depending on the composition [9]. In the Mo-N system three nitride phases have been reported 1551. An f.c.c. phase, originally named yMo2N because it has a composition around 33 at-% N, should instead be written as yMoNl - x , to be consistent with the notation for other f.c.c. nitrides. yMoNlPx has been reported to also exist at nitrogen contents above 33 at-% N. It can be assumed that it is a phase with a very broad homogeneity range. This is also indicated by the results obtained from high-pressure experiments. yMoNl - x undergoes an ordering transition between 400 and 850"C, depending on the composition, into the so-called pMo2N. Another phase, GMoN, with a structure similar to that of WC, has been prepared by various methods, including nitridation of Mo powder with flowing ammonia, or reaction of MoC15 with NH,. It has been structurally characterized to have several types of ordering arrangements of nitrogen in the centers of trigonal prismatic structure elements [56]. Molybdenum nitrides are studied for their interesting properties in catalytic reactions such as for the desulfurization and denitridation of oil products which is of increased importance for use in cars where sulfur can destroy the activity of the catalyzer. These powders can be produced with a high surface area by reaction of MOO, or MoS2 with NH3. Also layers of molybdenum nitrides, prepared by various techniques such as reactive sputtering or ion beam deposition,

7.5 Thermodynamics

22 1

were studied for application of wear-resistant coatings as well as for diffusion barriers [57,58]. 7.5.4.4 Ternary Transition Metal Nitrides One of the most important ternary nitride system is the Ti-Al-N system, however, the only technically important phase (Ti, A1)N is not thermodynamically stable. Applications are described in 7.7.2. The most important transition metal nitride, TIN, has complete miscibility with ZrN, HfN, VN, and NbN at high temperatures but miscibility gaps occur upon lowering the temperature [36]. The two-phase mixtures which form upon demixing have increased hardness and are thus interesting for commercial applications. The Ti-Nb-N system is interesting because of the superconducting properties of the (Ti, Nb)N phase having a slightly higher superconducting transition temperature, T, [59] than the NbN phase (once the world record holder of T,). Thus, Ti/Nb alloys (used for commercial superconducting applications dealing with high magnetic fields such as nuclear magnetic resonance spectrometers in medicine and for accelerator tubes in high-energy physics because of their excellent processing properties) can be modified by nitrogen implantation or in-diffusion to form a (Ti,Nb)N layer [60]. The TIN and NbN phases are completely miscible, however, recently an indication for a spinodal-like decomposition in (Ti, Nb)N was found to occur in the composition region around Tio.75Nbo.~sN-Tio.6~Nbo.37N at high temperatures [61].

7.5.5 Carbonitride Systems 7.5.5.1 The Group IVB Carbonitride Systems Because of the evident similarities between transition metal carbides and transition metal nitrides, carbon can be replaced completely by nitrogen in the IVB and VB group carbides. without changing the structure of the binary phases. Within the IVB-VIB group, only one ternary phase Cr2(C,N)2 has been reported [62]. Solid solubility in the VIB group carbonitride systems Cr-C-N and Mo-C-N is not complete because in the boundary systems the crystal structures of the carbide and nitride phases are different. The most important ternary carbonitride system is the Ti-C-N system, followed by the carbonitride systems of the other group IVB elements Zr-C-N and Hf-C-N. The phase diagrams of the Ti-C-N and Zr-C-N systems are presented in Fig. 11 [63]. It can be seen that the binary f.c.c. carbides and nitrides show a complete quasibinary solid solution. The carbon and nitrogen atoms occupy all the octahedral interstitial lattice sites randomly so that also the voids are randomly distributed on these interstitial sites. Because of the wide range of homogeneity a large variation of composition is possible for the preparation of carbonitrides powders and layers. Most of these powders and layers contain 50 at-% C + N. The excellent properties of Ti(C, N) have been exploited for cermets and hardmetals for a long time; and also Zr(C, N) was recently introduced in industrial coating processes (see 7.7.2).

222


Ti

TIN

Tic at% N

(a)

Zr

ZrC

10

(b)

20

30

40

ZrN

at%N

Figure 11. Phase diagrams of (a) the Ti-C-N and (b) Zr-C-N systems for T = 1120°C [63].

7.5.5.2 The Ti-Mo-C-N

and Ti-W-C-N Systems

The quaternary systems Ti-Mo-C-N [64] and Ti-W-C-N [65] are very important because the quaternary phases (Ti, Mo)(C, N) and (Ti, W)(C, N) are used as a

7.5 Thermodynamics I

I

I

I

I

0.1

0.2

I

I

0.3

\

I

I

223

I

I

X

LL

a 0.2 0.1

fcc

01

Tic

0

I

0.3 0.4 Site Fraction Mo

I

0.5

I I 0.6

(a)

TIN

G+graphite/G+WC+graphite

WN

wc Figure 12. Calculated phase diagrams of (a) the Ti-Mo-C-N [64] system for 1600°C with experimental data of [67] and of (b) the Ti-W-C-N [65] system for 1423°C with experimental data of [68]. Both for 50 at-% nonmetal in the representation of a reciprocal salt pair.

hard constituent in cermets [66]. These systems are technically most important for 50 at-% carbon + nitrogen and were described in terms of a section TiN-TicMoN-MoC and TiN-Tic-WN-WC. Figure 12a, b presents the calculated version of these phase diagrams for a nonmetal/metal ratio = 1 together with experimental data. A homogeneous f.c.c. phase (Ti, Mo)(C, N) separates into two f.c.c. phases, of

224


which one is richer in Ti and N than the other (Fig. 12a). The two-phase field f.c.c. + f.c.c.2 becomes smaller at higher temperatures. A more complicated situation exists for the Ti-W-C-N system because of the presence of a hexagonal WC phase. Figure 12b shows the calculated phase diagram of this system, also for a nonmetal/metal ratio of 1 [65] together with experimental data [68]. Also two f.c.c. phases form and the two phase field is even larger than in the corresponding system with Mo because of the higher nitrogen equilibrium pressure. Upon increasing carbon concentration graphite and WC form together with the two f.c.c. phases. The phase equilibria in the Ti-Mo-C-N and Ti-W-C-N systems have influence on the microstructures of cermets (see 7.7.1), at least partially responsible for a favorable core-rim type microstructure of the hard particles [69].

7.6 Properties of Important Transition Metal Carbides, Nitrides, and Carbonitrides 7.6.1 Melting Points Amongst the transition metal carbides, one of the highest known melting temperatures of any material has been measured: 6TaCo,89melts congruently at 3985°C and 6HfCo,94congruently at 3930°C (see Table 1 for the melting temperatures of other carbide phases). Melting points of various nitrides [70] have been measured under nitrogen pressures up to 300 bar. The nitrogen pressure has to be chosen in such a way as to be equal to the nitrogen equilibrium pressure of the nitride at its melting point. The melting point temperatures of the various nitrides can reach more than 3000°C (e.g. for TIN), see Table 2.

7.6.2 Color Transition metal carbides generally show a gray color, the exceptions are TaC and NbC, which have dark-brown/bronze color. Many mononitrides exhibit an intense color of bright metallic appearance which depends on the nonmetal/metal ratio. Table2 lists the colors of the mononitrides near the composition indicated by their formula. These colors can be observed not only for bulk samples but also play an important role when thin films [71] are produced for ornamental purposes (see 7.7.2). Very interesting is the color change of the stoichiometric group IVB carbonitrides ratio (Fig. 13). The higher number of the period as a function of the [C]/([C] [“I) the more the vivid violet color extends into the ternary region starting from the yellow colored nitrides.

+

7.6 Properties of Important Transition Metal Carbides, Nitrides, and Curbonitrides

225

+

Figure 13. Colors of the group IVB carbonitrides as a function of the [C]/([C] [N])ratio (which is x in MeC,NI -.J. Note that the colored region extends wider into the carbon-rich region the higher is the period number [73].

7.6.3 Thermal and Electrical Conductivities The thermal conductivity of transition metal carbides increases with increasing temperature, an unexpected phenomenon that has been investigated extensively on titanium and zirconium carbides [72]. Previous studies have reported a linear increase of the thermal conductivity with temperature, but more recent investigations have detected a nonlinear relationship [73]. Carbon deficiency increases the low temperature thermal conductivity of &TicI--x but decreases the hightemperature thermal conductivity. Transition metal carbides have generally a much lower thermal conductivity than isotypic transition metal nitrides. The thermal conductivity of WC is much higher than that of the f.c.c. transition metal carbides which can probably be explained by the stronger covalent bonding in WC and hence heat transport by phonons. Figure 14a illustrates the thermal conductivities for the group IVB nitrides, carbides and carbonitrides. The substitution of nitrogen by carbon in group IVB transition metal nitrides significantly reduces the thermal conductivity (compare 7.2.2). A variety of other ternary and quaternary carbides and carbonitrides is especially interesting for use in metal cutting tools in the form of layers, and powder particles have been investigated recently for their specific heat, temperature and thermal conductivity [66]. The behavior of the heat conductivity of most of these compounds is similar to that of the group IVB carbonitrides for (near) stoichiometric composition showing a positive curvature of heat conductivity as a function of temperature. Interestingly, however, Nb- and Mo-containing ternary titanium carbides and nitrides and quaternary titanium carbonitrides show a linear behavior in the same temperature interval. The similarity between f.c.c. transition metal carbides and nitrides is also valid for room-temperature electrical resistivity which decreases with increasing carbon or nitrogen content. Generally the transition metal carbides have higher electrical resistivities than the transition metal nitrides (Tables 1 and 2). This is also reflected in the

226


50 45

F

5

40

-0,79

iij 25

-

a,

0,4

L

20

-..1

15 0

,

1

200

,

1

400

,

(

600

,

1

800

0,oi

,

1

1200

1000

1400

temperature [“C]

50 45

_----------

0.61 0,43 0,25

20

0 _ ,1.1. I

15 0

200

I

I

400

I

I

600

,

,

800

,

,

1000

,

,

1200

,

1400

temperature [“C] (b) Figure 14. Thermal conductivity of group IVB carbonitrides with a nonmetal content close to 50 at-% as a function of temperature [73]: (a) Ti(C, N), (b) Zr(C, N), (c) Hf(C, N), The numbers denote the [C]/([C]+ [N]) ratio.

resistivity of group IVB carbonitrides (Fig. 15) which shows a general increase of resisratio. Only at around [C]/([C] [N]) = 0.5 the tivity with increasing [C]/([C] + [q) resistivity increase is much less than at high and low carbon contents, respectively. Some transition metal carbides are superconducting. The highest T, value has been reported for NbC (1 1.1 K) and TaC (10.3 K) and hexagonal WC (10.OK).

+

227

7.6 Properties of Important Transition Metal Carbides, Nitrides, and Carbonitrides

50

z-

45

i.$

.c

40 35

U

$

30

8

g 25 c

20 15 0

200

400

600

800

1000

1200

1400

temperature ["C] (C)

Tic, VC, HfC, and ZrC are not superconducting above 1 K. The electrical conductivities of transition metal carbides and nitrides are greatly influenced by the nonmetal/metal ratio; as this ratio approaches unity the electrical conductivity reaches a maximum. This is shown for FTiNl --x in Fig. 16 [74]. A similar behavior can be observed for other carbides and nitrides. The electrical conductivities of these compounds decrease linearly with increasing temperature [75]. Superconductivity is common among all f.c.c. transition metal nitrides of the IVB and VB groups as well as yMoN, -.x and GMoN. GNbN, -.x shows the highest

0.05 0.06

7 E 0.04 u

F

b

m

0.03

0

0.02

O.O10

. R 0

0.2

0.4

0.6

0.8

1

COMPOSITION, [C]I([Cl+[NI)

+

Figure 15. Electrical conductivity of group IVB carbonitrides as a function of the [C]/([C] [N]) ratio [74]: Ti(C,N), Zr(C, N), 0 Hf(C, N).

+

228


0,4

0,50

0,60

0,70

0,80

0,90

1 ,oo

composition [N]/p] Figure 16. Electrical conductivity of TiN, --x as a function of composition [74].

superconducting transition temperature of 17 K, which was the highest known of any material in the early superconductivity research. It is noteworthy that at around 15 K, hexagonal GMoN also has one of the highest transition temperatures of these materials [56] and f.c.c. yMoNl,oo,although never prepared with a stoichiometric composition, was predicted theoretically to have a T, of up to 30K. All transition metal nitrides are type I1 (hard) superconductors. For several f.c.c. nitrides, such as GTiN1--x,GZrN1--x, and 6HfNl --x (data reported for 6HfN1--x are, however, contradictory) T, increases linearly with increasing nitrogen content up to [Nl/[T]= 1 [76]. Table 1 gives the highest T, for each compound in absolute temperature.

7.6.4 Thermal Expansion Thermal expansion is a very important property of transition metal carbides, nitrides and carbonitrides because they are practically never used in pure form but mostly in contact with other materials (metals). Upon thermal load different thermal expansion of the carbide phase and the substrate may cause degradation of the composite. Generally their thermal expansion is higher than that of the corresponding metals. The thermal expansion of WC has recently been compiled [77]: the volume expansion increases steeply up to 15 x lop6 K-' at 1000 K. Figure 17 gives the data calculated also for higher than ambient pressure. The thermal expansion of various transition metal nitrides has been determined for both low temperatures [78] down to 8 K and for high temperatures [79] up to around 1500K. 6VN1--x exhibits a very low or even negative thermal expansion

7.6 Properties of Important Transition Metul Carbides, Nitrides, and Carbonitrides

229

25

20

.-'Y

h

15

% v

v

d

lo 5

0

100

1100

2100

3100

4100

T (K) Figure 17. Thermal expansion data of WC for ambient and higher pressures [77]. Reprinted by permission of The American Ceramic Society.

at temperatures below than 70 K. The high-temperature thermal expansion is given in Fig. 18 [79]. Interestingly, considering its low-temperature thermal expansion, 6VN1 has the highest high-temperature expansion coefficient up to 1 I x lop6K-I among these compounds. For the group IVB carbonitrides also data for well characterized f.c.c. phases exist [80]. The average linear thermal expansion (which is the linearized thermal expansion as a function of temperature) for the three carbonitrides is: = (9.9 GTi(C,N1_,y): LY,,

GZr(C,NI

-.y):

GHf(C,NI

1.4 x [C]/([C]+ [N]))x 10-6K-1

+ [N])) x lop6K-I = (8.5 - 2.4 x [C]/([C]+ [N])) x loP6 K-',

a,, = (7.8 Q,,

-

- 0.3 x

[C]/([C]

where [C] and [N] are the carbon and nitrogen contents, respectively. These expressions are valid for 50at-% [C] [N]. It can be realized that Zr(C,N) has the smallest thermal expansion coefficient at nitrogen-rich compositions, and also the smallest increase as a function of increasing carbon content. For the (nonaveraged) linear thermal expansion fitted by polynomial expressions see [go].

+

7.6.5 Diffusivities In transition metal carbides, nitrides, and carbonitrides, nitrogen diffusion occurs via a vacancy mechanism in the nonmetal sublattice. Because the vacancy concentration

230


12.0 -

1

-

11.0

-

h

id

L 10.0 v

8

e .-

-

9.0 -

0

2 -3

8.0-

.-

-

L

0

-

C

I! 1.0

f 8

U

6.0

-

5.0 ,

1

300

1

1

500

1

1

700

1

1

900

1

1

1100

1

1

1300

,

1

I500

1

1

1700

Figure 18. High-temperature thermal expansion of various transition metal nitrides [79].

is a function of composition within the homogeneity range of the f.c.c. phases, the nitrogen diffusivity is often composition-dependent as well. This has only been recently started to be investigated in transition metal carbides and nitrides for which a detailed analysis of the nitrogen diffusion profile is a necessary pre-requisite. Many diffusivity studies were conducted by investigating nitride or carbide layer growth as a function of time, which gives the average nitrogen diffusivity in the various phases [81]. A parabolic relationship, d = kt’/*, holds for this layer growth, where d is the layer thickness, k is a constant, and t is the diffusion time. The parabolic behavior indicates that the nitride layer growth is indeed diffusioncontrolled, provided that the sample is semi-infinite. Recent efforts for investigation this layer growth has shown that the thickness of the sample can have crucial influence on the layer thickness [82,83]. Methods were developed using wedgeshaped diffusion couples in order to exploit this behavior for diffusivity measurements [84]. A wedge shaped diffusion couple is presented in Fig. 19, which shows the thickness enhancement of diffusion layers upon decreasing sample thickness.


23 1

Figure 19. Diffusion layers of hafnium nitrides formed in a wedge-type sample. Note the thickness enhancement upon decreasing sample thickness [82].

A quantitative description of the dependency of diffusivity on the composition in f.c.c. carbides and nitrides can be made by the expression

D =f3dexp(-E/(kBT) exp(a(c+ - c))

[cm2/s],

wherefis 1/12 for f.c.c. compounds, v is the jump frequency of the nonmetal atoms NM, d is the mean distance between lattice planes, E is the activation energy, kB is the Boltzmann constant, and a is a parameter for the steepness of concentration dependency. The concentration c is in mol NM x cm-3 and c+ is the maximum concentration (in [NM]/[M] units this value is 1). Hence the diffusivity term is split into the activation energy term and a concentration-dependent term. The product f v 2 d is often expressed through a single factor, Do, the so-called preexponential factor and is in fact an extrapolation to infinite temperature, so that the accuracy is poor, especially if only a small temperature interval is investigated. A detailed study of concentration-dependent diffusivities was conducted for GTiCI--y and GZrCI--x [85], GNbCI-,x and GNbNI--x [86], GTaC1--x and GTaN1--x [87], and GCrN, --x [54]. These compounds show a large concentration-dependent diffusivity the higher the more nonmetal deficient the compound is and is a result of the vacancy diffusion mechanism. Further concentration-independent diffusion coefficients were measured for vanadium nitrides [46], hafnium nitrides [88], and chromium carbides, GTiNI--x [89] and GZrNI--x [90]. Although the metal diffusivity in transition metal nitrides has not yet been investigated, the activation energy of that process is much higher than for nitrogen diffusion and can be estimated to be of the order of about 8eV. Tables 3a and 3b summarize some nonmetal diffusivity data of transition metal carbides and nitrides.

7.6.6 Elastic Properties Transition metal carbides, nitrides and carbonitrides exhibit high Young’s moduli, appreciably higher than that of the transition metals. Many of the carbide and

232


Table 3. (a) Concentration dependency of the nonmetal diffusivity of transition metal carbides and nitrides (for compounds with a = 0 no significant concentration dependency is present). If two values are listed the first value corresponds to the nonmetal rich and the second to the nonmetal poor composition. ~

&Tici GZrCi GNbCI GTaCi - x GNbNi PTa2N GCrN,

-

0.058 0.167 0.065 0.004 0.43 1.40 0.018

3.12 3.59 3.53 3.28 3.13 3.3 2.92

112 144 98 99 20 0 184

Table 3. (b) Concentration independent non-metal diffusivities in transition metal carbides and nitrides.

2.9 20 1.28 0.718 0.0121 0.02 0.14 0.54 12.7 13.6 3.4 120 3.51

3.95 6.5 5.0 3.2 2.9 2.5 2.70 2.14 2.73 2.93 2.92 3.6 4.0 2.68

nitride phases were investigated for their elastic properties, mainly the Young’s modulus E, and the Poisson’s ratio v. Both, bulk samples as well as thin film samples were measured and generally polycrystalline material was studied [911. Among these studies, many of the data suffer from lacking or inappropriate characterization of composition and microstructure. The porosity of bulk samples as well as the substrate of thin films can influence the results substantially. Correction procedures for accounting for such influences are sometimes inappropriate [92]. Depending on the correction function, 5% porosity can cause a drop in E of about 50 GPa. For T i c most values for the Young’s modulus group around 450-460 GPa for a composition around 50 at-% carbon, the Poisson ratio is 0.17-0.19. The Young’s modulus of the important hexagonal phase WC is very high at room temperature [77] and drops by about 20% upon reaching 2000 K (Fig. 20). Although the measured Young’s modulus of TIN1-, show, quite a scatter because of the above-mentioned problems, it can be stated that E increases with


L

a v1 3 3 W 0

0

Kauk &Kockelmanna

0

Kreimer

0

Lee 8 Gilmre (polyuys.)

0 0

Koster L Rauscher ”

+

X

0

v) ET)

C

+

c

3 0

600

550

0

233

28 24

Bukatov et al. Brown et al.

n

25

Neshpor 8 Sarnsonov

A

C

X

1000

2000

3000

4000

Figure 20. Young’s modulus of WC as a function of temperature of various authors compiled by [77]. Reprinted by permission of The American Ceramic Society.

increasing nitrogen content. Many of the thin film samples show a higher Young’s modulus than bulk samples (see Fig. 21). Poisson’s ratios of 0.18-0.321 were reported, the most probable value is likely to be around 0.22. Similar statements can be made for ZrNl-, and H f N - , with values reaching up to 500 and 459 GPa, respectively (both values are for different kind of bulk samples: diffusion annealed or hot-pressed) and with Y = 0.25 for both compounds. Young’s modulus data on group VB transition metal nitrides are scarce but were measured for single crystals of 6VN and 6NbN [93], yielding the maximum values in the (100) direction of 487 and 490 GPa, respectively. For CrN, values up to 520 GPa were reported. For Ti(C,N) an increase of E with carbon content was measured on several samples (see citations in [92]), the porosity of which was, however, rather larger and different in different samples. A well characterized sample Ti(C0.20N0.80) with a porosity of only 3.5% [63] yielded the highest value of 451 GPa (and v = 0.212), which probably comes very near to the value for an ideal zero-porosity samples. An investigation of the E and G behavior as a function of temperature was recently conducted on polycrystalline samples of TIN [94], where a linear relationship was found for both quantities (see Fig. 22). For an extensive review on the elastic properties ( E and Y) of transition metal nitrides, carbonitrides (and carbides) as well as on the porosity correction routines the reader is referred to the tables given in [92]. Upon application and development of high-temperature investigation techniques, such as Brillouin scattering [95,96] a

234


a (3

I

I

800

I

I

0

I

-

600 -

Y

Y,

3 3

0

400 0

tn c3)

c

13

0

a

>-

n " 0.0

0.2

0.6

0.4

0

hot-pressed

0 sintered A not stated

0

zoneannealedfilm

0

film

A

0.8

#

1 .o

diffusion annealed

single crystal

Figure 21. Room temperature Young's modulus of TiN, - as a function of composition and for different sample types [92].

gain in knowledge on the high-temperature properties of these materials (even in single crystalline form) can be expected in the near future.

7.6.7 Microhardness Sometimes large differences are encountered between the microhardness of bulk material and thin films. This is most probably due to the different dislocation density and the grain size of both types of materials. Also the difference in testing (micro- or nano-indentation for films, heavier load up to several N for bulk samples) most probably yield differences. In addition, the composition (as for any of the properties discussed here) influences the microhardness. Because of these peculiarities the values are difficult to compare. Many of the transition metal carbides such as T i c and ordered VCI --x are very hard compounds. In GTiC1- the microhardness increases with increasing carbon content [97], a phenomenon that is probably closely related to the valence electron concentration (VEC) with a maximum stability at VEC = 8 at the composition TIC.

235


435

I

I

I

180

I

430

-

425

- 170

F

420

-

9

41 5

- 160 53

41 0

- 155

175

165 0 rn

U

-

405 400 395

I

I

I

I

Q 0)

150 f 145 140

Figure 22. Young’s modulus ( E ) and shear modulus (G) of TIN as a function of temperature for a composition of 50 at-% N [94].

Hardness data for transition metal carbides near the composition indicated by the formula are contained in Table 1. Single crystal investigations and orientation-dependent measurements on polycrystalline material of transition metal carbides show that the microhardness is orientation dependent [98] (e.g. for WC and NbC [99]). For T i c the Knoop hardness in the (1 10) plane is 27 GPa whereas that in (100) plane is 31 GPa, for example. Upon ordering the microhardness can change significantly (compare Fig. 5). Generally the hardness values of annealed crystals are lower than as-grown crystals due to their lower dislocation density. Polycrystalline material has higher values by about 1-2GPa than single crystals due to the grain boundary influence on hardness (Hall-Petch relation). Single-crystal and polycrystalline transition metal carbides have also been investigated with respect to creep, plasticity and slip systems. The f.c.c. carbides show slip upon mechanical load within the (111) plane in the (110) direction [98,100]. The ductile-to-brittle transformation temperature of TIC is about 800°C and is dependent on the grain size. The yield stress in TIC obeys a Hall-Petch type of relation, that is, the yield stress is inversely proportional to the square root of the grain size. TIC and ZrC show plastic deformation at surprisingly low temperatures around 1000°C. Figures 23a-e show the hardness vs. composition for a variety of transition metal nitrides [43,74,79,101] and carbonitrides [88]. The hardness decreases with increasing nitrogen content for the group VB nitrides [43,79,101]. This behavior can probably be explained by an electronically induced lattice softening as a result of the generation of antibonding states. It is known from band structure

236

7 Transition Metal Carbides, Nitrides, and Carbonitrides 26 25 24 23

19 18

17 16

0,50

0,60

(a)

0,80 composition [N]/[Ti] 0,70

1,oo

0,90

35

30

5 0 0,70 (b)

0,80

0,90

1,oo

composition "]/[Me]

Figure 23. Microhardness of nitrides and carbonitrides as a function of composition. (a) TiN [74], (b) VN [43], NbN and TaN [loll, (c) Cr-N [54], (d) Ti(C, N) and Ti(C, B, (e) Zr(C, N) and Hf(C, N) [73].

+,

+


-m1 4 1

a 12

E

I

p-Cr,N

0

I0

20

30

50

40

Composition [at% N]

(C)

ln

3 C

gc

24 22

g 20

.-0

E 18

16 14 0

0,4

02

0,6

03

1

composition [C]/([C]+[N])

(d)

28 26 24

2 ln

22

E

20

2

18

.-

16

rn

P

E

14 12 0

(e)

I

1

1

I

02

0,4

0,6

0,8

composition [C]/([C]+[N])

1

231

238


calculations that in f.c.c. transition metal compounds with more than eight valence electrons the antibonding states are occupied. An increasing valence electron concentration due to increasing nitrogen content would then lead to a decrease in hardness when the number of valence electrons exceeds eight. Indeed in GTiN, --x [74] a maximum in microhardness was found for the compound STiN0.67 corresponding to 7.3 valence electrons. For 6HfNl --x, however, a smooth increase in microhardness with increasing nitrogen content was observed. This cannot be explained in the same manner, and is probably a result of the difference in electronic states due to 5f electrons. The microhardness of chromium nitrides [54] is shown in Fig. 23c as obtained from diffusion couples. Upon use of diffusion couples, which exhibit thick diffusion layers, bulk values of the microhardnesses can be measured as a function of composition, provided that the nitrogen diffusion profile is measured. The microhardness of the IVB group carbonitrides (Fig. 23d, e) increases with increasing [C]/([C]+ [Nl) ratio. For Zr(C, N) and Hf(C, N) the increase is steep on the nitrogen-rich side and flattens on the carbon-rich side, so that a significant positive deviation from linearity can be observed. Contrary, for Ti(C, N) this deviation is only very small for 50 at-% nonmetal but significant for a substoichiometric composition of 45 at-% nonmetal.

7.7 Industrial Applications There is a widespread field of applications of transition metal carbides, nitrides and carbonitrides. The most important applications are: -

powders for hardmetals, cermets and spray coatings, deposited layers made by various techniques such as PVD and CVD, as well as diffusion layers for surface-modified components made by high-temperature and plasma techniques.

The layers are prepared for enhancing the abrasive resistance of the components, for ornamental purposes and layers in electronic devices, most of them for preventing the interaction of Si with other metals. The three application fields are discussed in more detail later. It should be noted that a variety of potential or just beginning applications mainly of nitride films are reported in the literature and discussed below.

7.7.1 Cemented Carbides and Carbonitrides The most widely used transition metal carbide is tungsten carbide, hexagonal WC, which is employed as the hard constituent in WC-Co hardmetals. Such hardmetals are sintered composite materials with 80-90% of hard particles such as WC embedded in a ductile binder phase such as Co. For these applicationsWC combines a number of

7.7 Industrial Applications

P

239

Atomic fraction of tungsten

Figure 24. Phase diagram of the W-C-Co system for 1423°C[102]. The top C corner is not shown.

favorable properties such as high hardness and wear resistance, good electrical and heat conductivities and is readily wetted by the binder metals Co and Ni. The ternary system W-C-Co is shown in Fig. 24 at 1423°C [102]. WC coexists with the Co phase and forms a pseudobinary lower melting eutectic between WC and Co. This eutectic facilitates full densification during liquid phase sintering. The carbon content must be kept close to the theoretical value because a decrease in total carbon content will lead to the formation of -q carbides (W, Co)& which are brittle and harmful to the performance of these materials. In modern hardmetals, also f.c.c. carbides such as TaC, NbC and T i c are admixed which increase the high-temperature performance. Figure 25 shows a microstructure of a modern WC/Co-based hardmetal containing about 10weight-% of f.c.c. carbides Tic, TaC, and NbC. Shortly after the invention of WC-Co hardmetals, T i c based hardmetals have been proposed as cutting tools but have not found extended use because of their brittleness. Only in the last two decades Tic-based hardmetals have found wider use when titanium nitride, GTiN, had been introduced as a further constituent. GTiN is used as one of the hard components in sintered carbonitride compositions containing TIC, Mo2C, WC, TaC, NbC, and other compounds [69,103]. Carbonitrides for such materials can be produced either by blending the various powders before liquid phase sintering with Ni and/or Co as a binder phase, or directly within the liquid phase sintering cycle. The modern titanium carbonitride cermets [ 103,1041 are characterized microstructurally by a structured hard phase featuring core-rim structures embedded in a tough Co-Ni binder. These carbonitrides are often called cermets or Ti(C, N) hardmetals, a microstructure of which is given in Fig. 26. The hard particles a roundly shaped as compared to the faceted WC grains (Fig. 25). In the microstructure of Fig. 26, which is a SEM image, corerim type grains with a dark core and a gray rim as well as grains with a white core and a gray rim can be seen.

240

7 Transition Metal Carbides, Nitrides, trnd Carbonitrides

Figure 25. Microstructure (light-optical microscope) of a WC/Co-based hardmetal (WIDIA TTR) containing hexagonal WC and cubic carbides (right: cubic phase, left hexagonal phase enhanced by specific etching).

Figure 26. Microstructure (SEM image) of a cemented carbonitride (WIDIA TTI 25) showing round hard particles with core-rim type structure.


241

The boundary between hardmetals and cermets is not strict because many of these compacts resemble microstructure features of both type of materials [ 1061 faceted WC crystals together with round-shaped titanium carbonitride-based hard particles. Generally, these titanium carbonitride hardmetals are comparable with respect to properties and microstructure to WC-based hardmetals. The powders of these materials are liquid phase sintered with Ni or Ni-Co binder metal alloys. The core-and-rim structure of the hard phase usually exhibit a molybdenum- and carbon-rich (Ti, Mo)C rim and a titanium- and nitrogen-rich Ti(C, N) but can also be inverted (compare Fig. 26). The metallurgy of the phase reactions is (because of the complexity of the multicomponent system) not yet fully understood [69]. According to the theoretical background of the phase reactions within the system Ti-Mo-C-N [67] the quaternary compound (Ti, Mo)(C, N) with a random distribution of the metals Ti and Mo on the metal sites and the nonmetals C and N on the nonmetal sites undergoes a decomposition with the formation of isotypic f.c.c. phases but with a preferential arrangement of N around the Ti atoms and Mo around the C atoms. This decomposition is of a spinodal type, which is a decomposition induced by local composition fluctuations without necessary activation energy rather than a classical nucleation and growth process. The question of whether the spinodal decomposition is a necessary prerequisite of core-and-rim formation is still unsettled because core-and-rim type structures may develop via reactions of the hard constituents with the liquid binder metal as well. The core-rim structure has turned out to be most probably stable [64, 1051. It has been claimed to be favorable because the molybdenum-containing rim is much better wetted by the binder alloy than the core. Modern Ti(C, N) hardmetals have been successfully applied for finishing operations of high-strength steel grades and ductile cast irons at high cutting speeds but with moderate cross section of the chips. The main advantages of Ti(C, N) hardmetals are the high chemical resistance and the low frictional welding tendency giving longer life of the cutting edge and good workpiece surface. Hardmetal-like microstructures are also produced by spray coating processes in which material is deposited by high velocity. Sprayed parts are used in a variety of processes where sliding wear occurs (aerospace industry, stone industry), advantage being that the parts can be refurbished. Various methods have been developed, such as detonation gun spraying and high-velocity oxygen fuel spraying. With these techniques layers of WC-Co, Cr3C2-Ni/Cr, and TiC/Ni mixtures are deposited with a thickness up to the mm range [107].

7.7.2 Deposited Layers Transition metal carbide, nitride and carbonitride layers of the IVB group made by CVD (chemical vapor deposition) and PVD (physical vapor deposition) are produced in order to increase the service life of hardmetal tools [18,21,108] for cutting and milling operations. About 90% of cutting tools are coated. The surface of such an insert must resist interaction with the workpiece (e.g. steel) and environment (oxidation), the adherence to the sintered compact must be excellent and the material in itself must have high hardness, strength and fracture

242


Figure 27. Milling and turning inserts with various coatings. The top coating is TiN (golden yellow), Zr(C, N) (violet) and A1203 (black). From WIDIA, Germany.

toughness. A number of binary, ternary, and quaternary compounds were studied and are actually in use, a large number of them contain titanium. Figure 27 shows various milling an turning inserts, which are multilayers coated with TiN, Ti(C, N), Zr(C, N) together with A1203. After the early established monolayers such as of TIN, multilayers composed of TIN, TIC, Ti(C, N), and A1203were developed. Frequently used layer systems are TiC/Ti(C, N)/TiN and TiN/Ti(C, N)/TiN, which have better wear and thermal cracking resistance than TiN monolayers [109]. Although Ti(C, N) is suitable for interaction with the workpiece the top layer is often TIN because of its appealing golden yellow color. Figure 28 shows a microstructure of such a TiN/Ti(C,N)/ TiN coating produced by WIDIA for a hardmetal grade used to turn steels. A1203is often used to provide a ceramic material in contact with the workpiece or to apply several multilayers of the type A1203/TiN.The thicknesses of the sublayers are on the order of about 1 pm and the total layer thickness is about 10-12 pm. Such a structure is shown in Fig. 29 for the WIDIA grade TN150. Most of these layers are deposited by CVD. Also a combination of PVD and CVD (‘duplex’) techniques was studied and has yielded better performing layers than other titanium nitride and carbonitride coatings [110]. The wear mechanism of multilayer coatings is assumed to occur layer by layer and the cracks are deflected by the interfaces between the sublayers. Besides cutting test the performance of coatings can be tested (at least compared) by scratch tests, in which a critical load is obtained at which the coatings are removed. The critical loads are typically 50-IOON and are influenced by the type of the deposition process. Also the obtained hardness is dependent on the preparation method; magnetron sputtered Ti(C, N) layers reach up to about 37 GPa [108]. For ornamental layers bright golden yellow TIN and yellow-green ZrN, as well as


243

Figure 28. Microstructure of a coated WC/Co-based cutting tool for turning steel (WIDIA TN250) where the layer is composed of three sublayers.

(Ti,Al)N gold + dark blue, depending on the Ti/AI ratio, Ti(C,N), Zr(C,N) (colors: see Fig. 13), CrN (metallic to brown) are used [l 1I, 1121 for example on watch cases, lighters, frames for glasses, and bathroom furnishings where the combination of color with increased scratch and corrosion resistance is exploited. The color of these materials stems from the absorption behavior, which is the result of the response of the material to the incident light, in turn a function of

Figure 29. Microstructure of a multilayer-coated steel-turning hardmetal (WIDIA TN150), with various TiN/A1203 sublayers a TIN and a Ti(C, N) layer.

244


the electron density of states. The CIE system is widely applied for defining colors. This is a plot of in the coordinates L* (lightness), a* (red-green value), and b* (yellow-blue). The color can be also dependent on the surface roughness, the layer thickness and the structural disordering. For example, in TiN the color appearance is compositionally dominated whereas in ZrN it is microstructure dominated. Thus, a certain variability exists for establishing colors apart from compositional point of view and related materials can behave differently. However, this behavior brings some difficulties with respect to the reproducibility of colors which in turn is influenced by the color perception of the human eye [113]. The deposition of TIN and ZrN on glass for windows represents a combination of ornamental application with the economic constraint of energy saving in buildings. ZrN layers, especially, have been found to exhibit favorable properties for transparent heat mirrors, i.e. windows which transmit visible light but reflect heat radiation [114]. Chromium nitride layers (fabricated by, e.g. cathodic arc plasma deposition) are interesting because of their corrosion properties as well as because of their excellent adhesion properties and fine-grained structure. They are applied for die-casting moulds where excellent edge properties are necessary [115,116]; some of these layer can have a multiphase character composed of Cr(N), Cr2N, and CrN1-, [117]. Sputter deposited ternary chromium nitrides such as Cr,Me, -,N with Me =Ti, Nb, Mo, and W additions and with grain sizes of up to 25 nm have been found [118] to show either a hardness minimum (Me = Mo, Ti) or a maximum of up to 27 GPa (Me = W, Nb). The Ti-A1-N system contains three ternary nitride phases but the most important phase for industrial applications is the thermodynamically unstable f.c.c. phase (Ti, A1)N [119,120], which can be prepared by magnetron sputtering and has found application for hardmetal tools [121-1231. (Ti,Al)N was found to exhibit a better oxidation resistance than TiN, depending on the Ti/Al ratio which can vary over a broad range. At high A1 contents a hexagonal phase forms probably related to or identical with the wurtzite-type AIN. Drill bits coated with (Ti, A1)N show a better performance than coated with TIN. The best results (about twofold lifetime) were obtained with (Tio,5Alo.5)N. Film hardnesses are on the order of 20-26 GPa, depending on the composition. Layers composed of (Ti, Zr)N show a higher hardness (up to 33.5 GPa) but the performance is not as good as of (Ti,Al)N [121]. In (Ti,Zr)N layers a spinodal decomposition around the composition (Tio.sZro.7)Nwas found [ 1241 which was also described for the phase equilibria of bulk samples [36,125]. Zr(C,N) layers together with a variety of other layers were successfully introduced for cutting tools for turning steel. A microstructure of such a multilayer is shown in Fig. 30 where Zr(C, N) represents a thick top layer. The favorable performance of such Zr(C, N) containing layers stems for the thermal expansion behavior of Zr(C,N), which is the lowest amongst the group IV carbonitrides, and which reduces tensile stress upon thermal load. Such a behavior enhances the performance of the coatings because of increased crack resistance and adhesion. While the first and second generation of coatings contained monolayers of TiN or T i c and multilayers composed of TiN, Ti(C, N), TIC, TiB2, and (Ti, A1)N (together with A1203)there is a general trend in research towards multicomponent/multiphase


245

+ Zr(C,N) + AI,O,

+ Ti(C,N) 6

TIN

f-

Substrate

10 pm

H Figure 30. Microstructure of a Zr(C, N) top-layer coated steel turning hardmetal (WIDIA TN7015).

coatings. Recently, layers composed of (Ti, - ,- y - zA1,CryY,)N, with y = 0.03 and z = 0.02, to were found to have an even higher oxidation resistance than (Ti, A1)N layers, with an onset of rapid oxidation about 80°C higher than of the latter [126]. The Knoop hardness is dependent on the composition and reaches values around 27 GPa. A further example of such thin films are the superlattices, which are in principle composed of two different nitrides, such as TiN/NbN and TiN/VN, which are successively deposited by PVD techniques such magnetron sputtering with a thickness in the nm range [127,128]. Hardnesses of up to about 50 GPa were measured on such layers depending on the thickness of the individual sublayers, called wavelength, which reach a maximum at around 5 nm. The most recent development which has, in addition, already be applied in industrial scale, is the preparation of superhard nano-crystalline layers by a combined plasma-CVD/PVD method. Two-phase layers of the type ncTiN/cBN and ncTiN/ aSi3N4were prepared, where nc denotes nano-crystalline, c cubic, and a amorphous [ 129-1 3 13. Hardnesses of up to 70 GPa were measured. The grain size of the crystalline phases of these materials is of the order of a few nm, whereas the maximum hardnesses are observed at the lower end of grain sizes studied (about 2 nm). The hardness increase of such multilayers or nanocrystalline coatings as compared to conventional bulk nitrides has not yet completely clarified. Three different phenomena were discussed [ 1321, which are known to increase the hardness of a material. The well-known Hall-Petch type grain size hardening ( H = Ho k / d ’ ” , where H = hardness, Ho = single-crystal hardness, k = constant, d = grain size), is probably only responsible for a minor extent, the coherency strain and the dislocation line energy effect. From the experimental finding that also a combination of layers practically without a lattice parameter mismatch (e.g. TiN/Nbo,4Vo,6N; mismatch: <0.2% [127]) also show a hardness increase, coherency strain effects (in which strain is caused

+

246


due to the mismatch) can also be ruled out at least for a part of the superlattices. The presence of dislocation line energy effects in which the shear modulus is important to move a dislocation (line energy= Gb2/2,where G is the shear modulus and b is the Burgers vector) can also contribute to the hardness increase but the observed increase is much higher than calculated from this theory (there is, however, some uncertainty in these calculation due to the lack of data for G ) . While a Hall-Petch type hardening is probably not the most prevailing mechanism in the above cited multilayers, it has been made responsible in nanocrystalline layers composed of two phases rather different in nature (crystalline, amorphous or different crystal structure). Although extremely fine-sized powder can show a negative Hall-Petch effect (nanosoftening), films composed of two such different phases having strong interphase boundaries suppress such effect resulting in a drastic increase in hardness. Films of the Ti-B-N system [133,134] consisting of a thermodynamically unstable f.c.c. phase Ti(B, N), where the B/N ratio can be varied, were also found to show increased hardness of up to 50 GPa. These films can be made single crystalline, the hardness of which is dependent on the ion bombardment intensity. Thus a dislocation density mechanism seems to be responsible for the hardening effect. Nitrides in the form of thin layers have an important application in electronic devices such as VLSI (very large scale integrated circuits) as Ohmic interface materials to prevent the formation of intermediate phases between Si and A1 or, in the most recent developments, where A1 is replaced by Cu because of the better electromigration properties upon continuing miniaturization, between Si and Cu. This application uses the chemical inertness, their high electrical conductivities and the ease with which nitrides can be deposited as layers. Mainly TiN deposited by PVD techniques is used in these circuits but also tungsten and molybdenum nitrides [58]. Also Tio.8A10,2N was studied and has shown a better performance than TIN [135]. For the new Si/Cu interconnection, TaN has proved to be of superior performance [136]. Films of tantalum nitrides are also used or of potential use for a variety of electronic applications ranging from thin-film resistors as well as for thermal printing heads and for wear and corrosion resistant coatings [137]. These films can be deposited by CVD with and without plasma [I 381 and PVD methods [136,137]. For CVD methods organic precursors [139,140], TaC1, [141], or TaBr5 [138] can be used. If a low resistivity of the tantalum nitride films is wanted, then the f.c.c. TaN,_, is desired without presence of the insulation Ta3NS phase, which is difficult if TaCl, NH3 as starting substances are used. TaNl --x films (deposited by both by PVD and CVD) have a resistivity of 130-250pRcm [142]. Niobium nitride films show good performance for vacuum microelectronic devices for displays because of a high work function and resistance to sputtering [143].

+

7.7.3 Diffusion Layers Diffusion layers of transition metal nitrides are (similar to case hardened steels, which should not be discussed here) prepared by heating metals and alloys in nitrogen or ammonia, occasionally diluted with an inert gas. Mainly titanium and


247

its alloys (e.g. TiA16V4) are of interest in order to circumvent the strong cladding behavior of these materials [26]. Also plasma nitriding techniques are used for the preparation of surface layers on the order of several microns thickness. The microstructure of these diffusion layers depend strongly of the alloy composition, the temperature and the nitrogen pressure [ 1441. A technique which is related to this surface refining method, is the surface laser heating of alloys. In this method a strong laser is used to melt titanium alloys at the surface with simultaneous supply of nitrogen via a jet or by operating in nitrogen atmosphere [27,145]. By such a method nitrides (though high-melting) form by surface melting together with an intermediate diffusion layer form in the solid state. Nitride layer thicknesses of the order of millimeters can be prepared. These layers are, for example, applied in large turbine blades in order to prepare a surface withstanding the intensive cavitation of water droplets reaching supersonic velocity. One of the most interesting developments in the field of hardmetals and cermets for cutting tools is the creation of a graded surface which makes use of the favorable properties of functional-gradient materials by dissipating the thermal and mechanical load onto a larger part of the cutting insert in order to increase the lifetime or to withstand extreme conditions of cutting [146]. Figure 31 shows surface-near microstructures of such functional-gradient cermets composed of (Ti, W)(C, N) with additions of TaC and NbC and a Co binder metal. Upon sintering, the hexagonal phase WC forms (which can also be present in the starting formulation). Depending on the composition and the sintering cycle a smooth variation of the

Figure 31. Microstructures of functional-gradient cemented carbonitrides based on the system (Ti, W)(C, N F C o which are surface modified by applying a reactive gas phase during sintering in a one step fabrication process [146].

248


phase content as a function of the distance from the surface can be seen. This morphology is caused by the action of reactive gases such as nitrogen and carbon monoxide introduced at specific points of the sintering cycle. The action of reactive gases can be such modified that different gradients can be fabricated and even layers of different chemical and phase composition can be prepared together with gradients. This functional-gradient hardmetal has a Ti(C, N) top layer, an intermediate WC-Co layer and a gradient in which the (Ti, W)(C, N) phase is enriched at the outer side and continuously reaches bulk composition. In view of the rather high production costs of coated materials (the coating process is about 15-20% of the total production costs of a cutting insert) such modified cemented carbonitrides could be lower the production costs significantly because the diffusion layers and gradients are established during the sintering cycle and (provided similar performance can be obtained) an additional coating cycle is then unnecessary.

Acknowledgments The author would like to thank Dr. K. Dreyer, WIDIA GesmbH, Essen (Germany) for support with figures and microstructures of tools parts. The help of Mr. W. Prohaska with the preparation of the phase diagram drawings is gratefully acknowledged.

References 1. P. Ettmayer and W. Lengauer, Nitrides in Ullmann’s Encyclopedia of Industrial Chemistry, Verlag Chemie, Weinheim, 1991, A17, p. 341. 2. E. Parthe and K. Yvon, Acta Crystallogr. B, 1970, 26, 153. 3. K. Yvon and E. Parthe, Acta Crystallogr. B, 1970, 26, 149. 4. T. Epicier, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.), Kluwer, Dordrecht, 1990, p. 215. 5 . C. H. de Novion, B. Beuneu, T. Priem, N. Lorenzelli, and A. Finel, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.), Kluwer, Dordrecht, 1990, p. 329. 6. F. Kubel, W. Lengauer, K. Yvon, A. Junod, and K. Knorr, Phys. Rev. B, 1988,38, 12908. 7. W. Lengauer, M. Bohn, B. Wollein, and K. Lisak, Acta Muter. 1999, in press. 8. H. Wiesenberger, W. Lengauer, and P. Ettmayer, Acta Muter. 1998,46,651. 9. J. D. Browne, P. R. Lidell, R. Street, and T. Mills, Phys. Stat. Solid, 1970, Al, 715. 10. T. B. Massalski (Ed.), Binary Alloy Phase Diagrams, 2, ASM International, Metals Park, OH, 1990. 11. K. Schwarz, CRC Crit. Rev. Solid State Muter. Sci. 1987, 13, 211. 12. A. Neckel, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.), Kluwer, Dordrecht, 1990, p. 485. 13. P. Blaha and K. Schwarz, Int. J. Quantum Chem. 1983,23, 1535. 14. A. N. Christensen and S. Fregerslev, Acta Chem. Scand. A , 1978, A31, 861. 15. R. Kieffer, H. Nowotny, P. Ettmayer, and M. Freudhofmeier, Monatsh. Chem. 1970,101,65. 16. H. Pastor, Muter. Sci. Eng. 1988, A105/106, 401. 17. J.-E. Sundgren and L. Hultman in: Materials and Processes f o r Surface and Interface Engineering, Y .Pauleau (Ed.), Kluwer, Dordrecht, 1995, p. 453.

References

18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59.

249

W. Schintlmeister and 0. Pacher, J . Vac. Sci. Technol. 1975, 12/4, 743. S. Abisset and F. Maury, Surf. Coat. Technol. 1998, 108-109, 200. W. D. Sproul, J . Vac. Sci. Technol. A , 1994, 12, 1594. J. Smolik and K. Zdunek, Vacuum, 1998, 51/3, 441. V. Craciun, D. Craciun, C. Ghica, L. Trupina, C. Flueraru, and N. Nastase, Appl. Surf: Sci. 1999, 138-139, 593. K. Obata, K. Sugioka, H. Takai, and K. Midorikawa, Applied Surf. Sci. 1999, 138-139, 335. A. J. Perry and J. R. Treglio, J . Vac. Sci. Technol. A , 1995, 13/3, 1067. S. Xiao, C. P. Lungu, and 0. Takai, Thin Solid Films, 1998, 334, 173. J. Magnan, G. C. Weatherly, and M.-C. Cheynet, Metall. Muter. Trans., A , 1999, 30A, 19. C. Hu, H. Xin, L. M. Watson, and T. N. Baker, Acta Muter. 1997, 45/10, 431 1. W. Lengauer, Talanta, 1991, 38, 659. R. Tiubler, S. Binder, M. Groschner, W. Lengauer, and P. Ettmayer, Mikrochim. Acta I, 1992, 107, 337. G. F. Bastin and H. J. M. Heijligers, J . Microsc. Spectrosc. Electron. 1986, 11, 215. G. F. Bastin and H. J. M. Heijligers, Quantitative Electron Probe Microanalysis of Nitrogen, University of Technology, Eindhoven, 1988. W. Lengauer, J. Bauer, M. Bohn, H. Wiesenberger, and P. Ettmayer, Mikrochim. Acta, 1997, 126, 279. K. Frisk, A Thermochemical Evaluation of the Cr-N, Fe-N, M e N and Cr-Mo-N Systems, Report TRITA-MAC-0393, Materials Research Center, Royal Institute of Technology, Stockholm, 1989. V. N. Lipatnikov, W. Lengauer, P. Ettmayer, E. Keil, G. Groboth, and E. Kny, J . Alloys Comp. 1997,261, 192. R. Ward in M T P International Review of Science, Inorganic Chemistry Series One, D. W. A. Sharp (Ed.), Butterworths, London 1972, p. 5. H. Holleck, Binare und ternare Carbid- und Nitridsysteme der Ubergangsmetalle, Borntrager, Berlin 1984. H. Holleck, J . Vac. Sci. Technol. A , 1986, A6, 2661. W. Lengauer, Acta Metall. Muter. 1991, 39, 2985. K. Zheng and R. Schmid-Fetzer, Z . Metallk. 1996, 87, 540. R. Juza, A. Rabenau, and I. Nitschke, Z. Anorg. Allgem. Chem. 1964,332, 1. E. Rudy, Metall. Trans. 1970, 1, 1249. 0. N. Carlsson, J. F. Smith, and R. H. Nafziger, Metall. Trans. A , 1986, A17, 1647. W. Lengauer and P. Ettmayer, J . Less-Common Met. 1985, 109, 351. W. Lengauer, J . Phys. Chem. Solids, 1988,49, 59. N. Kieda, N. Mizutani, and M. Kato, J . Less-Common. Met. 1988, 144, 293. C. Teichmann, W. Lengauer, P. Ettmayer, J. Bauer, and M. Bohn, Metall. Muter. Trans A , 1997,28A, 837. G. Brauer and W. Kern, Z. Anorg. Allgem. Chem. 1983,507, 127. G. Brauer and W. Kern, Z. Anorg. Allgem. Chem. 1984,512, 7. R. Berger, W. Lengauer, and P. Ettmayer, J . Alloys Comp. 1997, 259, L9. M. Joguet, W. Lengauer, M. Bohn, and J. Bauer, J . Alloys Comp. 1998,269,233. J. Gatterer, G. Dufek, P. Ettmayer, and R. Kieffer, Monatsh. Chem. 1975, 106, 1137. G. Brauer, E. Mohr, A. Neuhaus, and A. Skokan, Monatsh. Chem. 1972, 103,794. K. Frisk, J. Alloys. Comp. 1996, 278, 216. W. Mayr, W. Lengauer, D. Rafaja, J. Bauer, and M. Bohn, J . Phase Equil. 1999, 20, 35. P. Ettmayer and A. Vendl, in Science of Hard Materials, R. K. Viswanadham, D. J. Rowcliffe, and J. Gurland (Eds.), Plenum, New York, 1983, p. 47. A. Bezinge, K. Yvon, J. Muller, W. Lengauer, and P. Ettmayer, Solidstate Commun. 1987,63, 141. S. L. Roberson, D. Finello, and R. F. Davis, Muter. Sci. Eng. A , 1998, 248, 198. V. P. Anitha, A. Bhattacharya, N. G. Patil, and S. Major, Thin Solid Films, 1993, 236, 306. W. Mayr, W. Lengauer, V. Buscaglia, J. Bauer, M. Bohn, and M. Fialin, J. Alloys Comp. 1997, 262-263, 521.

250


60. V. Buscaglia. - A. Martinelli, R. Musenich, W. Mayr, and W. Lengauer, J . Alloys Comp. 1999, 283, 24. R. Musenich, P. Ettmayer, and M. Bohn, ECASZA 99, 61. B. Wollein, V. Buscaglia, - W. Lengauer, Sevilla, Spain, 1999. 62. P. Ettmayer, G. Vinek, and H. Rassaerts, Monatsh. Chem. 1966, 97,1248. 63. S. Binder, W. Lengauer, P. Ettmayer, J. Bauer, J. Debuigne, and M. Bohn, J . Alloys Comp. 1995, 217, 128. 64. H.-J. Chung, J.-H. Shim, and D. N. Lee, J. Alloys Comp. 1999, 282, 142. 65. S. Jonsson, Z . Metallkd. 1996,87, 788. 66. L. Chen. G. Groboth, E. Kny, W. Lengauer, and P. Ettmayer, Proc. I l l hPlansee Seminar '97 1997,2,498, G. Kneringer et al. (Eds.), Reutte (A). 67. E. Rudy, J. Less-Comm. Met. 1973, 33, 43. 68. A. Doi, T. Nomura, M.-A. Tobioka, K. Takahashi, and A. Hara, High Temp.-High Press, 1986, 18,443. 69. P. Ettmayer, Ann. Rev. Muter. Sci. 1989, 19, 145. 70. P. Ettmayer, R. Kieffer, and F. Hattinger, Metall. (Berlin), 1974, 28, 1151. 71. A. J. Perry, M. Georgson, and W. D. Sproul, Thin Solid Films, 1988, 157, 255. 72. W. S . Williams, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.), Kluwer, Dordrecht 1990, p. 625. 73. W. Lengauer, S. Binder, K. Aigner, P. Ettmayer, A. Guillou, J. Debuigne, and G. Groboth, J . Alloys Comp. 1995,217, 137. 74. W. Lengauer, J . Alloys Comp. 1992, 186,293. 75. R. E. Taylor and J. Morreale, J . Am. Ceram. Soc. 1964,47, 69. 76. W. Lengauer, Surf. Znterf. Anal. 1990, 15, 377. 77. R. R. Reeber and K. Wang, J . Am. Ceram. SOC.1999,82, 129. 78. W. Lengauer, P. Ettmayer, and K. Yvon, J. Less-Common Met. 1991, 168, L7. 79. W. Lengauer and P. Ettmayer, Monatsh. Chem. 1986, 117, 275. 80. K. Aigner, W. Lengauer, D. Rafaja, and P. Ettmayer, J . Alloys Comp. 1994, 215, 121. 81. W. Lengauer, H. Wiesenberger, M. Joguet, D. Rafaja, and P. Ettmayer, in Transition Metal Carbides and Nitrides: Preparation, Properties and Reactivity, S . T. Oyama (Ed.), Blackie, London, 1996, p. 9 1. 82. W. Lengauer, D. Rafaja, R. Taubler, C. Kral, and P. Ettmayer, Acta Metall. Muter. 1993,41, 3505. 83. W. Lengauer, J . Alloys Comp. 1995, 229, 80. 84. D. Rafaja and W. Lengauer, Model. Simul. Muter. Sci. Eng. 1998, 6, 141. 85. T. C. Wallace Sr., D. P. Butt, in Transition Metal Carbides and Nitrides: Preparation, Properties and Reactivity, s. T. Oyama (Ed.), Blackie, London, 1996, p. 53. 86. D. Rafaja, H. Wiesenberger, W. Lengauer, and M. Joguet, Metall. Muter. Trans A, 1998,29A, 439. 87. D. Rafaja and W. Lengauer, Acta Muter. 1998, 46, 3477. 88. W. Lengauer, D. Rafaja, G. Zehetner, and P. Ettmayer, Acta Muter. 1996, 44, 3331. 89. F. W. Wood and 0. G. Paasche, Thin Solid Films, 1970,40, 131. 90. H. Matzke, in The Physics and Chemistry of Carbides, Nitrides and Borides, R. Freer (Ed.), Kluwer, Dordrecht 1990, p. 357. 91. E. Tiirok, A. J. Perry, L. Chollet, and W. D. Sproul, Thin Solid Films, 1987, 153, 37. 92. C. Kral, W. Lengauer, D. Rafaja, and P. Ettmayer, J . Alloys Comp. 1998, 265,215. 93. H. S. Kim, G. Bugli, and G. Djega-Mariadassou, J. Solid. St. Chem. 1999, 142, 100. 94. F. Li, K. Fatrdla, and W. Lengauer, to be published 1999. 95. P. R. Stoddart, J. D. Comins, and A. G. Every, Physica B, 1996, 219-220, 717. 96. A. G. Every, W. Pang, J. D. Comins, and P. R. Stoddart, Ultrasonics 1998, 36, 223. 97. J.-L. Chermant, P. Delavignette, and A. Deschanvres, J. Less-Common Met. 1970, 21, 89. 98. F. W. Vahldiek and S. A. Mersol, J . Less-Common Met. 1977, 55, 265. 99. Y . Kumashiro, E. Sakuma, Y . Kimura, H. Ihara, and S. Misawa, J . Cryst. Growth, 1981, 52, 597. 100. D. W. Lee and J. S . Haggerty, J. Am. Ceram. SOC.1969,52, 641. 101. C. Politis and G. Rejman, Kernforschungszentrum Karlsruhe, KfK Report Ext6178-I, 1978.

References

25 1

102. L. Akesson, Thesis, Royal Inst. Technol. Stockholm, Sweden 1982. 103. K. Dreyer, P. Drohmann, H. Holzhauer, M. Miiller, H. van den Berg, and W. Weith, Proc. PM World Congress, Granada, 1998. 104. P. Wally, P. Ettmayer, and W. Lengauer, J . Alloys Comp. 1995, 228, 96. 105. P. Ettmayer, H. Kolaska, W. Lengauer, and K. Dreyer, Int. J . Refract. Met. Hard. Mat. 1995, 13, 343. 106. T. Viatte, S. Bolognini, D. Mari, and W. Benoit, Proc. PM World Congress, Granada 1998. 107. L.-M. Berger, W. Hermel, P. Vuoristo, T. Mantyla, W. Lengauer, and P. Ettmayer, in Thermal Spray: Practical Solutions for Engineer. Problems, C. C. Berndt (Ed.), ASM International, Materials Park, OH, 1996, p. 89. 108. J. Deng, M. Braun, and I. Gudowska, J . Vac. Sci. Technol. A , 1994, 12/3, 733. 109. Y. L. Su and W. H. Kao, Wear, 1998, 223, 1 19. 110. T. Bell, Met. Muter. 1991, 7, 478. 11 1. U. Kopacz and R. Riedl, Z. Metallkd. 1992, 83, 492. 112. G. Reiners. U. Beck, and H. A. Jehn, Thin Solid Films, 1994, 253, 33. 113. G. Reiners, H. Hantsche, H. A. Jehn, U. Kopacs, and A. Rack, S u r - Cout. Technol. 1992,54/ 55, 273. 114. K. E. Anderssson, M. Veszelei, and A. Roos, Solar Energ. Muter. Solur Cells, 1994, 32, 199. 115. 0. Knotek, F. Loffler, and B. Bosserhoff, SUI$ Coat. Technol. 1993,62, 630. 116. S. S. Kim, J. G. Han, and S. Y. Lee, Thin Solid Films, 1998, 334, 133. 117. K. Volz, M. Kiuchi, and W. Ensinger, Surf. Coat. Technol. 1998, 108-109, 303. 118. P. Hones, R. SanjinCs, and F. Levy, Thin Solid Films, 1998, 332, 240. 119. W.-D. Miinz, J. Vuc. Sci. Technol. 1986, A4, 2717. 120. 0. Knotek, W. D. Miinz, and T. Leydendecker, J . Vac. Sci. Technol. 1987, AS, 2173. 121. H. Randhawa, P. C. Johnson, and R. Cunningham, J. Vac. Sci. Technol. A 1988, 6/3, 2136. 122. C. Jarms, H.-R. Stock, and P. Mayr, Surf. Coat. Technol. 1998, 108-109, 206. 123. B.-J. Kim, Y.-C. Kim, J.-W. Nah, and J.-J. Lee, J . Vuc. Sci. Technol. A , 1999, 17/1, 133. 124. 0. Knotek and A. Barimani, Thin Solid Films, 1989, 174, 51. 125. R. Kieffer, H. Nowotny, P. Ettmayer, and G. Dufek, Metall. (Berlin) 1972, 26, 701. 126. L. A. Donohue, I. J. Smith, W.-D. Miinz, I. Petrov, and J. E. Greene, Surf: Cout. Technol., 1997,9495, 226. 127. M. Shinn, L. Hultman, and S. A. Barnett, J . Muter. Res. 1992, 7/4, 901. 128. M. Larsson, P. Hollman, P. Hedenqvist, S. Hogmark, U. Wahlstrom, and L. Hultman, Surf. Cout. Technol. 1996,8647, 351. 129. S. Veprek and S. Reiprich, Thin Solid Films, 1995, 268, 64. 130. S. Veprek, M. Haussmann, S. Reiprich, L. Shizhi, and J. Dian, Surf. Coat. Technol. 1996,8687, 394. 131. S. Veprek, P. Neslidek, A. Niederhofer, F. Glatz, M. Jilek, and M. &ma, Surf Coat Technol. 1998, 108-109, 138. 132. X. Chu and S. A. Barnett, J. Appl. Phys. 1995,77/9, 4403. 133. P. Losbichler, C. Mitterer, P. N. Gibson, W. Gissler, F. Hofer, and P. Warbichler, Sure Coat. Technol. 1997, 94-95, 297. 134. C. Mitterer, P. Losbichler, F. Hofer, P. Warbichler, P. N. Gibson, and W. Gissler, Vacuum, 1998, SO/-, 313. 135. I. Petrov, E. Mojab, F. Adibi, J. E. Greene, L. Hultman, and J.-E. Sungren, J . Vuc. Sci. Techno/. 1993, A l l , 11. 136. M. H. Tsai, S. C. Sun, C. E. Tsai, S. H. Chudng, and H. T. Chiu, J . Appl. Phys. 199679,6932. 137. K. Radhakrishnan, N. G. Ing, and R. Gopalakrishnan, Muter. Sci. Eng. B, 1999, 57, 224. 138. X. Chen, H. L. Frisch, A. E. Kaloyeros, B. Arkles, and J. Sullivan, J . Vac. Sci. Technol. 1999, B17, 182. 139. R. Fix, R. Gordon, and D. Hoffman, Chem. Muter. 1993,5, 614. 140. H. Chiu and W. Chang, J. Muter. Sci. Lett. 1992, 11, 92. 141. K. Hieber, Thin Solid Films, 1974, 24, 157. 142. X. Sun, E. Kolawa, J. Chen, J. Reid, and M.-A. Nicolet, Thin Solid Films, 1993, 236, 347.

252


143. Y. Gotoh, M. Nagao, T. Ura, H. Tsuji, and J. Ishikawa, Nucl. Znstr. Meth. Phys. Res. B, 1999, B148, 925. 144. J. K. Gregory, W. Lengauer, and W. Mayr, in Surface Performance of Titanium, J. K. Gregory, H. J. Rack, and D. Eylon (Eds.), TMS, Warrendale, PA, 1996, p. 65. 145. H. Xin, S. Mridha, and T. N. Baker, J . Muter. Sci. 1996, 31, 22. 146. W. Lengauer, L. Chen, J. Garcia, V. Ucakar, K. Dreyer, D. Kassel, and H. W. Daub, Proc. PM2tec 99, Znt. Congress on Powder Metallurgy and Particulate Materials, Vancouver 1999.


8 New Superhard Materials: Carbon and Silicon Nitrides J. E. Lowther

8.1 Introduction Novel procedures continuously being introduced into the synthesis of new materials have lead to the speculation that it could be possibie to produce a material with hardness exceeding that of diamond, the hardest material presently known to man. As the chemistry of diamond is based upon understanding the structural bonding of carbon, a modification of known carbon structures through application of high pressure or temperature would seem a promising route. Indeed there has been much speculation about the hardness of carbon found in fullerene complexes, yet production of ‘post-diamond’ phases has not met with any success, at least through known synthesis procedures. Even so the commercial synthesis of diamond (and more recently cubic boron-nitride) has lead to much speculation in the field and with it enormous excitement as there is potential for new superhard materials. Properties of diamond, its hardness, unique electrical and thermal properties, arise from a three-dimensional structure of very strong covalent bonds. Similar properties are also associated with cubic boron-nitride. That other similar highly covalent three-dimensional structures may exhibit similar properties to that of diamond is highly likely. A possible material that first stimulated interest in such structures was PSi3N4 [I], a hard material with quite interesting properties. A possible replacement of one (or several) of the Si atoms with C would tentatively suggest the synthesis of a material that is harder than PSi3N4 and even compete with diamond. There is therefore a rich possibility of structures in the tertiary Si,C,,N, system that afford unique potential for synthesis. But problems are always encountered when dealing with nitrogen, the N2 dimer has a very strong covalent bond and does not react with carbon readily when it competes with either carbon or silicon in forming stable Si,C,,Nz structures. Graphite is very likely more stable than diamond, and in a tertiary material like Si,C,,N, the formation of a SirC, structures is also possible. Quite clearly sophisticated techniques are needed for a successful synthesis [2,3]. Given the cost and complexity of a commercial synthesis procedure, even when identified, an unknown synthesis route requires considerable investment. For this reason computer modeling procedures have been introduced that attempt to predict new structures and examine phase transitions connecting them. Such procedures are continuously being applied [4] and, bearing in mind that fundamentals of the modeling procedures are also being debated, indications in the last few years have shown that reliable calculations of material properties can be

254

8 New Superhard Materials: Carbon and Silicon Nitrides

achieved at little expense. Liu and Cohen [5] applied computer modeling to investigate a hypothetical phase of PC3N4 and suggested that such a material would have a large binding energy and may be metastable under ambient conditions. More importantly, the calculations implied that this material would have a hardness comparable with (or even greater than) diamond. The computational technique used had already been applied to diamond with considerable success [6] and its reliability simulated enormous interest and effort in investigating other materials. This chapter examines some potential structures that have been proposed for different phases of carbon-nitride and also the tertiary material silicoti-carbonnitride. First we briefly examine some basic modeling procedures that have been employed to date when investigating these materials and then discuss properties of the predicted structures.

8.2 Modeling Procedures Several approaches are essentially used in the computational modeling of solids. These are classified as being either semi-empirical or ab initio. The term semiempirical is appropriate to those systems where significant knowledge exists about the system under specific conditions of temperature or pressure. Subsequently the system is modeled under other conditions. Semi-empirical modeling is really an interpolation from one system to other, probably unknown or experimentally inaccessible, conditions. On the contrary the term ab initio is appropriate for systems where little or no knowledge exists about the system. The ab initio procedure is applicable for predictive modeling of new phases of materials and is far more reliable than semi-empirical methods. Molecular dynamics procedures can be implemented at both levels of the formalism. Such techniques endeavor to examine the trajectories of atoms under different temperatures and pressures. Presently there are many excellent algorithms readily available to achieve this task when the atomic interactions can be described in analytic form. Unfortunately for covalent solids, which is likely to be the character of superhard materials, evaluation of the atomic interactions is complex and computationally very demanding.

8.2.1 Semi-empirical Approaches This category of approaches assumes that interactions in the solid can be represented by two or three-body potentials or force-fields between atoms. The interaction is usually obtained by fitting models to results on molecular systems. This is a convenient approach for ionic materials where long range Coulomb forces are strong and for a typical material like SiOz at least eleven parameters are needed to characterize the system. Such a choice of potential, but in the form of V(r) = D{exp[-a(r - req)]- 1}2 ( k / 2 ) ( 8- Oe,)2 with empirical parameters

+

8.2 Modeling Procedures

255

D,a, re,, k, and Be, fitted to small molecular systems was applied to structures of carbon-nitride by Guo and Goodard [7]. However, these results were in conflict with many other calculations for carbon-nitrides highlighting that use of such a simple interatomic interaction has danger when the potential is transferred to materials with a far higher density. In the case of highly covalent materials like diamond and (most likely) related superhard materials the potential is likely to be far more complicated than used by Guo and Goodard. In order to study fracture and related degradation processes in silicon nitride a similar effective potential has been obtained [8] representing an average of the interactions. This allows for simulations of up to lo6 effective atoms. However, the drawback of such an approach is that it apparently disregards the possibility of atomic segregation. One of the most promising analytic potentials that has been introduced in this regard is the Tersoff potential [9] that has the form V ( r ) = f c ( r ) ( f R ( r )+ / f A ( v ) ) where f ( r ) and b are complicated functions containing angular as well as a radial (v) dependences. For a typical binary solid some 24 empirical parameters are now needed to adequately describe the potential, with considerable success being reported for C and Si structures. Suitable parameters for the Tersoff potential have now been obtained for cubic BN [lo] and recently for PSi3N4 [l 11 but earlier results [12] had indicated that parameters in the Tersoff potential may not be unique. Further work is now being undertaken to try and find a transferable Tersoff potential for the Si,C,,N, system, but in all likelihood such a potential may have a limited range of transferability amongst closely related structures. Quite clearly the semi-empirical procedure is very dependent upon the choice of parameters, although it has been possible to obtain sensible parameters from fixed systems. The shortcoming with the approach is that the parameters may not be transferable to other (especially unknown) situations. At the same time once a reliable potential is obtained there are many excellent molecular dynamics algorithms available to compute the subsequent motion of the system quite efficiently and with low computer cost.

8.2.2 Tight-binding Schemes Although various tight-binding schemes have proven successful in studies of carbon and silicon surfaces, the problem of obtaining a successful interaction with N has not been satisfactorily solved. And even if such approaches such as the virtual crystal approximation could be employed it is doubtful whether such a potential could be truly transferable, a similar problem as with the semi-empirical approaches. Recently an approach using density functional concepts has been introduced [ 131 and applied to high density structures of carbon-nitride [ 141 with various stoichiometries. The tight-binding interactions are only evaluated once for the system and stored numerically making the method rather similar to conventional tight-binding methods yet avoiding an interpolative scheme for their evaluation at different distances. Even though the method still retains some reliance on molecular

256


rather than solid state concepts for computing electronic interactions it has proven quite useful in examining various stoichiometric forms of carbon-nitride.

8.2.3 Ab initio Pseudopotential Approach The pseudopotential approach employs a quantum mechanical description for the electronic interactions as specified by either through density functional or Hartree Fock techniques [15]. In the case of materials modeling the density functional approach appears to be extremely accurate in predicting ground state structures of systems [16] and it is this approach that presently appears to be gaining ground in the modeling of superhard materials. Even within the density functional formalism two approaches, the local density approximation (LDA) [17] or the generalized gradient approximation (GGA) [ 181, are presently receiving attention, although in the case of carbon based systems the LDA approach seems more promising for structural properties. Once the electronic interaction is described, forces acting on the atoms can be calculated and in turn their positions established using well developed techniques [19]. Overall the procedure does not require any input apart from elementary chemical details of the interacting atoms and the crystalline structure of the system. The operational shortcoming with this approach is that it is computationally expensive requiring very long computer runs as well a being memory intensive in that large arrays are necessary to describe the electronic degrees of freedom. Without doubt however this procedure is now regarded as state-of-the-art and is most reliable for predictive modeling. The quantum mechanical basis of the approach very often consist of simple plane waves which together with fast Fourier transform techniques and a suitable pseudopotential give accurate results. One of the more successful pseudopotentials is a soft pseudopotential that attempts to limit the number of planes waves needed in the basis, with the pseudopotential generated from the Troullier-Martins [20] method.

8.2.4 Transition Pressures and Relative Stability The equation of state of the material is usually represented by the Morgeneau [21] or Birch [22] form and the material modeled through a unit cell structure in which the individual atoms have fractional (Wykoff) parameters optimized as the cell changes in volume. From the fitted equation of state we then find the pressure on the system, P = - d E / d V . However this is not the full story. Ideally there are two important factors specifying a phase transformation between two phases. First two (or more) phases can coexist when the separate phases have the same energy. Accordingly an investigation of the full thermodynamic energy of the system against pressure would reveal the coexistence of distinct phases. But energy is also needed when compressing one phase to another: the energy required to surmount the energy barrier between the phases. The latter value needs to know specific details of phases

8.3 Carbon Nitride

257

involved. Thus there are two ways of deducing the pressure for a phase transformation. Either from the slope at the common energy of E / V ,or by the intersection of E/P.The former is specific in that it refers to distinct phases whereas the latter makes no assumption of the path followed. For example, if we look at the transformation from graphite into diamond we calculate from E/V that a pressure of P = 80 GPa is needed to transform graphite into diamond by overcoming the energy barrier. The commercial synthesis process involves pressures far less than this, although high pressure treatment of graphite (or CG0type material) leading to diamond is consistent with application of such large pressures. However, the calculated relative energies of graphite and diamond are very close to each other. This means that the pressure for the transformation found from the E/P intersection is calculated to be very small, and this would be the situation if the graphite to diamond transformation itself was continuous. Of course the transformation process of graphite into diamond itself is not continuous but rather involves changes from graphite into other (rhombohedral) graphitic forms through to hexagonal diamond and then finally diamond. Chemical influences (such as metal flux) as well as temperature factors are important in the synthesis. Collectively these latter features are influencing the relative energetics of the various phases. Thus we must conclude that whereas the calculated pressure from E/P has some guiding significance, the pressure obtained from E / V identifies a chosen synthesis route. The pressure calculated in this way will be an upper value needed for the transformation between the two phases. Accordingly, an in depth modeling of the synthesis procedure needs an investigation of the stability of the intermediate phases if they are known. Yet the important physical properties of the system are essentially contained in relative values of the energies of the two phases and how the energy changes under compression.

8.3 Carbon Nitride The difficulty of producing large enough samples of C3N4 for a precise structural characterization is proving problematic. It has been observed to exist in apparent amorphous or disordered phases [23-251 and there are several reports of crystalline phases [26,27] being dispersed in the amorphous matrix. Laser ablation [28,29] appears to be a promising technique for the synthesis of crystalline carbon nitride and vapor deposition has recently proven to be promising [30,31] as has ionbeam deposition [32,33]. Graphite is an important starting point for the synthesis of diamond. It exists in several forms and as such through applied pressure leads to other structures such as hexagonal diamond and then finally diamond. Graphite structures for carbon nitride have been considered theoretically as possible forms of the material [6,34,35]. These all consist of C3N4layers in a hexagonal planar lattice in which a C atom is missing. This form is not as hard as diamond but the location of the

258

8 New Superhard Materials: Carbon and Silicon NitridcJ

Table 1. Structural details of some predicted superhard crystalline phases of C3N4. Phase

Z

Unit cell (nm)

Wykoff parameters

a = 0.641-0.644, C: (0.178, 0.772, 0.250) c = 0.24040.2467 N: (0.331, 0.033, 0.250) N: (0.333, 0.667, 0.250) Cubic-C3N4 (Pz3rn) 1 a = 0.342-0.344 C: (0.500, 0.500, 0.000) (zinc-blende) N: (0.254, 0.254, 0.254) Cubic C3N4 (1534 4 a = 0.540 C: (0.875, 0.000, 0.250) N: (0.281, 0.281, 0.254) a G N 4 (P31c) 4 a =0.6466 C: (0.773, 0.178, 0.250) c = 0.4710 C: (0.227, 0.821, 0.750) N: (0.333, 0.667, 0.750) N: (0.033, 0.331, 0.250) N: (0.033, 0.331, 0.250) N: (0.967, 0.670, 0.750) PC3N4 (P6,lrn)

2

Bulk modulus (GPa) 425 [37], 437 [5], 451 [35] 425 [37], 430 [26], 448 [35], 480 [26], 496 [35], 425 [35]

missing C atom in the C3N4layer suggests there could be several ways of stacking in much the same way as, for example, graphitic BN [36]. A rhombohedra1 [37] form of the material in graphitic form is predicted to be stable at a material density lower than the 0-phase and more recently other graphitic forms with still lower density have been suggested as being possible precursors to the formation of the superhard phase [34,35]. High energy amorphous structures have also been investigated [14,38] and these have shown competition between the C-C, C-N, and N-N bonds in forming such a structure. Possible crystalline structures with stoichiometries different from C3N4[6,39] have also been examined.

8.3.1 Crystalline Structures To date, several forms of C3N4have been suggested for crystalline structures with superhard forms being BC3N4[37,40,41] (hexagonal structure and space group P3), cubic or defect zinc-blende C3N4 [37] (P43rn) and another cubic form [35] (1434 suggested to be the hardest as measured by the calculated bulk modulus. In each of these structures, carbon atoms are in four-fold coordination by nitrogens, the nitrogens in three-fold coordination by carbons. All structures are predicted to be stable with unit cells given in Table 1. The harder structures are shown in Fig. 1, as can be seen each C atom is surrounded in four-fold coordination by N atoms. The C-N bond length is small: 0.145nm in 0C3N4and 0.148nm in cubic (zinc-blende) C3N4. Some idea of the nature of the bonding in phases of C3N4can be obtained from a mapping of the electronic charge density. In Fig. 2 we show such a plot for the 0 structure, as expected it is quite directional, yet there are quite unique features about N associated with the N(2p) orbits directed away from the C-N bond. It is possible that such a feature could lead to sites for passivation (i.e. hydrogen association) leading to sources of brittleness in such a material.

8.3 Carbon Nitride

259

Figure :1 Some superhard crystalline phases of C3N4:(a) P-phase (P63/m),(b) cubic (P43m),(c) cubic (1434 structures. Black spheres represent C atoms.

8.3.2 Graphitic Structures Graphitic phases consist of hexagonal planes of bonded C-N atoms with a missing C on one of the conventional planar hexagonal lattice sites. Various sites are possible for the ordering of the graphitic planes depending on the relative location of the missing C atom from one plane to another [42]. First is the case where empty C sites are aligned relative to each other along the hexagonal c-axis. This structure emphasizes C-C and N-N bonding between the adjacent layers so that all bonds along the c axis are saturated though there is a rather large region where bonding is probably weak. Such a structure was the first type of graphitic structure considered for C3N4and named p-C3N4 by Ortega and Sankey [34]. We shall refer to the structure as p-C3N4(even-phase). If we consider that the planes may move relative to one another, another structure can arise with mixed C-N bonding between the

260


Figure 2. Charge density of P(P6,Irn) C3N4plotted perpendicular to the c-axis.

layers. This structure we shall name p-C3N4(odd-phase). Probably there are many possibilities between. There is also the possibility that one of the planes may have no missing C atoms with the adjacent plane having two missing C atoms: a C2N4-C4N4 structure. The three possibilities are shown in Fig. 3. The other type of graphitic structure considered recently by Teter and Helmley [35] also contains 14 atoms in the unit cell but is somewhat different to the p-C3N4 structures as here two possible C-N bond lengths are in the hexagonal plane. This structure is shown in Fig. 4 and emphasizes C-N bonding between layers. As the stability of the graphitic phases probably relates to the intralayer structure than the interlayer spacing we investigate the layer charge density for the phases. In Fig. 5 we show the charge density of the P63/m phase; rather surprisingly most of the charge on N is directed away from the C-N bond toward the vacant site, as is the case for some superhard structures.

8.3 Carbon Nitride

26 1

Figure 3. p-structures of hexagonal graphite C3N4: (a) odd-phase structure, (b) the even-phase structure, and (c) the CZN4-C4N4planar structure. Carbon atoms are shown in black.

The low energy P63/m graphite phase appears to change to one of the higher energy p-C3N4 graphite phases following compression well before the superhard (3-phase is reached. However this p-C3N4 graphite phase seems slightly softer than the P63/m phase and as such requires less pressure to surmount the energy barrier that leads to the p-phase. If a graphitic phase of C3N4 can be established it would therefore appear quite fruitful to use pressure (especially uniaxial) techniques to change between the graphitic phases and possibly attain a superhard phase. Of the graphite structures stacking involving C4N4-C2N4 sheets was unlikely but each of the p-graphite phases were found quite stable and with relative energies lying very close to each other. Forces between adjacent planes in the graphitic structures are very small and so graphitic carbon nitride is extremely soft along the c-axis. This weak interaction probably leads to dispersed layers of graphitic sheets in the material as recently observed using high resolution spectroscopy of carbon nitride thin films [43].

8.3.3 Amorphous Structures To model disorder, a 56-atom unit cell as described in Fig. l(b) was used [38] and heated to l0000K using a very approximate Tersoff potential [12]. We chose

Figure 4. P6m2 structure of hexagonal graphite C3N4.

262


Figure 5. Charge density of the graphitic C3N4.

standard Tersoff parameters for C and for N those that give best results specifically for the cubic form of C3N4,bearing in mind (as pointed out by Kroll [12]) that the Tersoff potential has limitations in describing N-N bonds. The subsequent computational procedure was quite time consuming for convergence, this indicating that the initial starting point was indeed a highly metastable configuration. Relaxation was monitored throughout and we noted importance of the various bonding configurations, in particular the build up of C-N bonds and competition with C-C and NN bonds. Although the number of atoms in the simulation was small we did observe that N-N bonds were often not connected into the complete network as much as CC bonds were. Instead the N atoms tended to group into relatively loosely bonded structures often isolated from the main continuous network. This can be seen in

8.3 Carbon Nitride

263

N

Figure 6. Calculated structure of amorphous C3N4.

Fig. 6 which shows the distribution of atoms in the amorphous 56-atom cell at a volume very near the stable volume for such a structure. ~, This equilibrium volume corresponds to a material density of p = 2.9 g ~ m - still relatively high but far lower than the crystalline hard material p = 3.8 gcmP3. Specifically at p = 2.9 gcmP3 there are a larger number of C-N bonds than either C-C or N-N bonds and this is typical of most volumes considered up to a material density of 4.5 g ~ m - Thus ~ . the modeling has shown that C-N bond formation is preferred over others.

8.3.4 Relative Stability The equilibrium results for the various crystalline, graphitic and amorphous C3N4 systems are given in Table 2 and the relative energies in Fig. 7. Only modest pressure is needed to overcome the energy barrier separating the graphitic phases and induce a transition between them. Estimating this pressure from the slope at the common energy/volume intersection gives a hydrostatic Table 2. Structure, bulk modulus of some phases of C3N4. Phase

Shortest C-N distances (nm)

P

0.145 0.148 0.132-O.145 0.137 0.137

Cubic (defect zinc-blende) P63/rn graphite p-graphite (odd-phase) p-graphite (even-phase) Amorphous

Interplanar distance (nm)

0.329 0.337 0.372

B (GPa)

Relative energy (eV/atom)

432 428 253 205 198 128

0.000 $0.200 -0.002 +0.029 +0.038 +0.564

264


t

1

-220.4

-221.0'. ,

'

40

I '

'

50

60

vo~ume(a.u.~) Figure 7. Energy-volume relationships for various phases of C3N4.

transition pressure necessary to transfer between the various graphite phases to be less than 5 GPa: a uniaxial compression is far less than this. Hydrostatic pressures that lead to the formation of the superhard P-phase calculated in the same way for graphite phases however are much higher. For example again from the slope at the common energy intersection we calculate that a pressure of about 80GPa is needed to go from the P63/m phase to the superhard P-phase and a little less at about 55 GPa to go from the odd-phase p-C3N4 phase and only slightly more for other graphitic phases. Such transition pressures are well above the value of 12 GPa suggested recently by Teter and Helmley [35] to go from P63/m graphite to the superhard j3-phase, although the way in which this value was estimated is not clear. Very likely this value was based upon a upon common energy/pressure arguments that neglect the importance of intermediate changes in volume of the structures.

8.4 Silicon Carbon Nitride As discussed earlier the p-form of Si3N4was the prototype material leading to the postulate that C3N4may be harder than diamond [5,40]. Recently tertiary ceramics in the form Si-C-N are being investigated [4447] for potentially superhard properties. Amorphous forms of Si-C-N materials are well documented [48] but often such materials segregate into Sic or Si3N4when heated; a stable crystalline form of Si-CN that does not behave in such a manner is being sought. Crystalline S i x - N compounds that have so far been produced have had their structure contested depending upon the nature of the locality about Si. Using extended X-ray-absorption fine structure studies (EXAFS) either a local tetrahedral [49] or random [50] structure about Si has been suggested. In both these cases Si atoms are linked by a C-N network. More recent EXAFS measurements [47] coupled with computer modeling favored the local tetrahedral Si-N-C arrangement. The crystalline structure of this Si-C-N material was found to have a structure very near to that of ~ t S i ~ N ~ ( Pwith 3 ~ ca) unit cell of dimension a = 0.6904nm and c = 0.5260nm.

8.4 Silicon Carbon Nitride

265

Figure 8. Beta phase of SiC2N4.

The importance of the local tetrahedral structure has also been noted in the analysis of a lower density form of SiC2N4 compound recently produced by Riedel and coworkers [45]. Here X-ray diffraction showed the possibility of an almost cubic (Pn3m) phase with a unit cell of measured dimension a = 0.61885 nm. In such a material the local Si related tetrahedral structure is realized as SiN4 units that are connected through N=C=N chains. The flexibility of the N=C=N bond has been considered an essential feature for more extended SiC2N4structures [45,51]. Following earlier theoretical results on fJC3N4,one of the hardest forms of SiC2N4is expected to be fJSiC2N4.This is obtained by replacing two C atoms in the 14-atom unit cell of fJC3N4with Si atoms maintaining lowest C-N bond lengths [52]. Possibly this form can be attained through a synthesis starting from one of structures discussed above.

8.4.1 PSiC2N4 In Fig. 8 we show the structure of fJSiC2N4as viewed along the c-axis of the hexagonal unit cell with details of the structure being presented in Table 3 . We see that the bulk modulus of SiC2N4is about $ that of C3N4. In Fig. 9 we show the calculated charge density in one of the hexagonal planes. There are two important features to be noted on comparing this with a similar charge distribution for the fJC3N4shown in Fig. 1. First very little charge is located about the Si atom, suggesting that Si bonding be insignificant compared with Table 3. Calculated structure and Bulk Modulus of some phases of SiC2N4. Phase

Unit cell (nm)

B (GPa)

B'

PSiC2N4( P 2 / m) a (Pn3m) b c (~4n2)

a = 0.678, b = 0.692; c = 0.258 a = 0.664 a = 0.665 a = 0.632

330 106 95 32

4.01 0.000 3.90 -0.048 2.68 -0.029 4.89 -0.052

Relative energy (eVjatom)

266


Figure 9. Charge density of pSiC2N4 plotted perpendicular to the c-axis.

the C=N bond. Second is the direction of the C=N bonding; if we look at the N atom we see that the main charge is not directed along the C=N bond itself but with N(2p) orbitals being directed perpendicular to the bond. Surprisingly the C-N bond is very similar in both silicon-carbon-nitride and carbon nitride again suggesting the importance of the C=N bond.

8.4.2 Near-cubic Forms of SiC2N4 Recently a low density phase of SiC2N4has been investigated [45]with a possible cubic structure (Pn3m) shown in Fig. 10. In such a structure, Si surrounded tetrahedrally by four N atoms is the

8.4 Silicon Carbon Nitride

267

Si N C N Si

a

N b

Si

Si

C N

N N C

Si

C Si

Figure 10. Cubic phase of SiC2N4.Large open circles are Si, small black circles C and small open circles N. The nature of the Si-N=C=N-Si bond is shown for each case.

essential building block and a linear Si-N=C=N-Si bridging unit. The measured unit cell constant is 0.618nm, and Riedel et al. noted that this size of cell would have a short interatomic spacing between the C and N atoms of 0.11 9 nm. They therefore suggested that the measured cell constant may correspond to an effective value relating to a random distribution of N atoms about the Si-C-Si unit. The linearity of the N=C=N bond can also be maintained in a similar symmetry in the manner suggested in Fig. 10b and more recently Kroll et af. [51] have suggested another structure for the cubic phase involving N=C=N bond with overall P4n2 symmetry shown in Fig. 1Oc. The nonlinear C=N=C bond is expected to lead to a softening of the structure [53]. As yet the several cubic structures have not been related to the potentially superhard BSiC2N4 phase. We have calculated the structure of the three new possible phases and the results are given in Table 3.

268


60

120 volurne(a.u.3)

Figure 11. Energy-volume behavior of different phases of SiC2N4:(a), (b), and (c) identify the different phases shown in Figure 10. The pressure shown is obtained from the intersection point of the c-phase and the P-phase.

8.4.3 Relative Stability In Fig. 11 we show the energy/volume relations for each model of the cubic phase and for the PSiC2N4phase. The energy of each of the cubic phases lies below that of the P-phase. This is not unlike the situation for graphite and diamond and so a cubic to P-phase synthesis may be possible. Estimating the transition pressure needed to change from one phase to another is from the slope at the energy intersection of the common phases; this is the pressure necessary to overcome the energy barrier between the two phases. In the softer of the three cubic phases the pressure is quite large, about 62 GPa.

8.5 Conclusions There is understandably some pessimism regarding a commercial synthesis of superhard carbon nitride especially in view of the intense effort that has been put into its synthesis by many laboratories in the last few years with no firm result to date. Yet the wisdom of the computational arguments upon which the various structures of the material have been predicted seems firm and they have been shown to be capable of giving us valuable insight in many other areas of materials science. Experiences gained regarding the commercial synthesis of diamond and cubic BN could probably lead ways that carbon nitride can be synthesized, yet other more novel routes also cannot be excluded.

Acknowledgements The National Research Foundation (South Africa) is thanked for its support.

References

269

References 1. 0. Borgen and H. M. Seip, Actu. Chem. Scand. 1961, 15, 1789. 2. C. Niu, Y. Z. Lu, and C. M. Lieber, Science, 1993, 261, 334. 3. K. M. Yu, M. L. Cohen, E. E. Haller, W. L. Hansen, A. Y. Liu, and I. C. Wu, Phys. Rev. B, 1994,49, 5034. 4. M. L. Cohen and J. R. Chebkowsky, Electronic Structure and Optical Properties of Semiconductors, Springer Verlag, Germany, 1988. 5. A. Y. Liu and M. L. Cohen, Phys. Rev. B, 1990,41, 10727. 6. Y. Miyamoto, M. L. Cohen, and S. G. Louie, Solid State Commun. 1997, 102, 605. 7. Y. Guo and W. A. Goddard 111, Chem. Phys. Lett. 1995, 237, 72. 8. R. K. Kalia, A. Nakano, A. Omeltchenko, K. Tsurta, and P. Vashishta, Phys. Rev. Lett. 1997, 78, 2144. 9. J. Tersoff, Phys. Rev. Lett. 1988, 61, 632. 10. W. Sckkal, B. Bouhafs, H. Aourag, and M. Certier, J . Phys.: Condens. Mutter, 1988,10,4975. 11. F. de Brito Mota, J. F. Justo, and A. Fazzio, Phys. Rev. B, 1998, 58, 8323. 12. P. M. Kroll, PhD Thesis, University of Paderborn, 1996. 13. D. Porezag, Th. Frauenheim, Th. Kohler, G. Seifert, and R. Kaschner, Phys. Rev. B, 1995,51, 12947. 14. F. Weich, J. Widany, and Th. Frauenheim, Phys. Rev. Lett. 1997, 78, 3326. 15. R. M. Dreizler and E. K. U. Gross, Density Functional Theory, Springer Verlag, Berlin, 1990. 16. M. C. Payne, M. P. Teter, D. C. Allan, T. A. Arias, and J. D. Joanopoulos, Rev. Mod. Phys. 1992, 64, 1045. 17. D. M. Ceperley and B. J. Adler, Phys. Rev. B, 1980, 45, 566. 18. J. P. Perdew, Phys. Rev. B, 1986, 33, 8822. 19. M. Bockstede, A. Kley, J. Neugebauer, and M. Schemer, Comp. Phys. Comm. 1997,107, 187. 20. N. Troullier and J. L. Martins, Phys. Rev. B, 1991, 43, 1993. 21. F. D. Murnaghan, Proc. Nat. Acud. Sci. USA, 1994,3, 24. 22. F. Birch, J . Geophys. Res. 1952, 57, 227. 23. A. R. Merchant, D. G. McCulloch, D. R. McKenzie, Y. Yin, L. Hall, and E. G. Gerstner, J. Appl. Phys. 1996, 79, 6914 24. M. Y. Ching, D. Li, X. Lin, V. P. Dravid, Y. Chung, M. Wong, and W. D. Sproul, J. Vuc.Sci. Technol. 1993, 11, 521. 25. D. M. Bhusari, C. K. Chen, K. H. Chen, T. J. Chuang, L. C. Chen, and M. C. Lin, J. Muter. Res. 1997, 12, 322. 26. J. Martin-Gill, F. Martin-Gill, M. Sarikaya, M. Qian, M. Jose-Yacaman, and A. Rubio, J . Appl. Phys. 1997, 81,2555. 27. K. Yamamoto, Y. Koga, K. Yase, S. Fijiara, and M. Kubota, Jpn. J . Appl. Phys. 1997, 36, L230. 28. Z. J. Zhang, S. Fan, and C. M. Lieber, Appl. Rev. Lett. 1995,66, 3582. 29. M. Tabbal, P. Merel, S. Moisa, M. Chaker, A. Ricard, and M. Moisan, Appl. Rev. Lett. 1996, 69, 1698. 30. Y. Chen, L. P. Guo, and E. G. Wang, Philos. Mag. Lett. 1997, 75, 155. 31. Kehui Wu, E. G. Wang, Jian Qing, and Guichang, Xu, J . Appl. Phys. 1998, 83, 1702. 32. C. Ronning, H. Feldermann, R. Merk, H. Hofsass, P. Reinke, and J. U. Thiele, Phys. Rev. B, 1998,58, 2207. 33. H. W. Lu, X. R. Zou, and J. Y. Feng, J . Phys. D: Appl. Phys. 1998,31,363. 34. J. Ortega and 0. F. Sankey, Phys. Rev. B, 1995, 51, 2624. 35. D. M. Teter and R. J. Hemley, Science, 1996, 271, 53. 36. J. Furthmuller, J. Hafner, and G. Kresse, Phys. Rev. B, 1994, 50, 15606. 37. A. M Liu and R. M. Wentzcovitch, Phys. Rev. B, 1994, 50, 10362. 38. J. E. Lowther, Phys. Rev. B, 1998, 57, 5724. 39. C. Cote and M. L. Cohen, Phys. Rev. B, 1997, 55, 5684. 40. A. Y. Liu and M. L. Cohen, Science, 1989, 245, 841. 41. J. V. Badding, Adv. Muter. 1997, 11, 877.

270 42. 43. 44. 45. 46. 47. 48. 49.

50. 51. 52. 53.


J. E. Lowther, Phys. Rev. B, 1999, 59,11683 . Sjostrom, S. Stafstrom, M. Boman, and J. E. Sundren, Phys. Rev. Lett. 1995, 75, 1336. R. Riedel, H. J. Kleebe, H. Schonfelder, and F. Aldinger, Nature, 1995, 374, 526. R. Riedel, A. Greiner, G. Miche, W. Dressler, H. Fuess, J. Bill, and F. Aldinger, Angew. Chem. Znternat. Edn, 1997 36, 603. A. Bendeddouche, R. Berjoan, E. Beche, T. Merie-Mejean, S. Schamm, V. Serin, G. Taillades, A. Pradel, and R. Hillel, J . Appl. Phys. 1997, 81, 6147. Y. K. Chang, H. H. Hseich, W. F. Pong, M. H. Tsdi, T. E. Dunn, F. Z. Chien, P. K. Tseng, K. L. Tsang, W. K. Su, L. C. Chen, S. L. Wei, K. H. Chen, D. M. Bhusari, andY. F. Chen, Phys. Rev. B, 1998, 58,9018. M. Driss-Khodja, A. Georghiu, G. Dufour, H. Roulet, C. Senemaud, and M. Cauchetier, Phys. Rev. Lett. 1985, 54, 1985. A. Gheorghiu, S. Senemaud, H. Roulet, G. Dufour, T. Moreno, S. Bodeur, C. Reynaud, M. Cauchetier, and M. Luce, J . Appl. Phys. 1992,71,4118. F. Tenegal, A. M. Flank, and N. Herlin, Phys. Rev. B, 1996, 54, 12029. P. Kroll, R. Riedel, and R. Hoffman, Phys. Rev. B, 1999,60, 3126. Cheng-Zhang Wang, En-Ge Wang, and Qingyun Dai, J. Appl. Phys. 1998, 83, 1975. E. Kroke, personal communication, 1998.


9 Effective Doping in Novel sp2 Bonded Carbon Allotropes G. Jungnickel, P. K. Sitch, T. Frauenheim, C. R. Cousins, C. D. Latham, B. R. Eggen, and M. I. Heggie

9.1 Introduction It has been speculated that there exists a multitude of metastable forms of carbon. This arises from the preparation of macroscopic quantities of fullerenes [l, 21, the observation of carbon nanotubes [3] and onions [4]. and the successes in carbon network chemistry [5,6]. This is based on the well-known fact that carbon atoms can frequently occur in different local environments. In particular, carbon can form hybrid orbitals and is found to be either sp3 bonded and four-fold coordinated as in diamond, sp2 bonded and, hence, threefold coordinated as in graphite, or sp bonded and two-fold coordinated as in acetylene. Whereas in diamond one finds o bonds as in silicon, in the latter two forms the carbon atoms can lower their energy by creating additional 7c bonds to their neighbors, that is they form double and triple bonds. In contrast to diamond, materials based on these under-coordinated types of carbon tend to form layered or chain-like structures which are usually less dense than graphite. Diamond forms very strong and short bonds and is, therefore, a promising material for applications in harsh environments. The intraplane bonding in graphite is very strong but the weaker interplanar forces prevent graphite from being well protected against mechanical or chemical attack. Carbon atoms can develop intermediate hybridization states dependent on their local neighborhood as in the various amorphous phases that have been prepared during the past decades. Semiconductor devices based upon carbon would be very attractive when taking advantage of the short and stiff carbon-carbon bonds in combination with the lightness of the carbon atoms since this would result in a material with a thermal conductivity almost as good as in diamond, which is stable against thermal degradation. A large band gap and a high saturation velocity for electrons as in diamond [7,8] would make it a superior material for superfast, high-power, high-temperature, and wide band gap applications. Unfortunately, diamond itself, though intensively studied, has been found to be less suitable for device applications. Although it can be doped by boron to make it a p-type semiconductor it is almost impossible to obtain the n-type counterpart for a p-n junction. Diamond cannot be doped by the most likely candidate, nitrogen, as a substitutional impurity. Instead nitrogen moves off its lattice site causing a dangling bond on one of its carbon neighbors which in turn results in a deep level [9-121. Other impurities such as phosphorus [16-181 or lithium [19] have been tried for ntype doping in diamond but their size reduces the solubility in diamond to near

272

9 Effective Doping in Novel sp2 Bonded Carbon Allotropes

zero [9]. The most promising alternative, P, results in an electrical activity much less than would be expected from its concentration so that the doping efficiency is rather low [17]. Moreover, it has been found that the phosphorus-vacancy complex [20] is more stable than substitutional P. This complex traps injected electrons and deactivates the impurities. In amorphous carbon (a-C) substitutional impurities can relax off their sites as in diamond or cause substantial rearrangements in the carbon host. Injected holes or electrons migrate in a-C to occupied high energy n: or low energy unoccupied n* levels, respectively. Associated n-bonded clusters relax to screen the additional charge [12-141 and the conductivity can only occur through variable range hopping between such clusters. In the light of these findings, in a recent paper [ 151 the doping problem in carbon has been attacked from a different point of view. Instead of searching for other atom types or complexes that effectively would act as shallow donors metastable carbon modifications which accept both boron and nitrogen as substitutional impurities have been proposed to exist. Among the hypothetical forms of carbon predicted prior to this work are stacked layers made of double and acetylene triple bonds [21] yielding structures like ‘graphyne’ and related polymeric networks. These planar configurations were predicted to have quite large electronic gaps but also found to have a lower number of atoms per unit layer area than graphite. Equivalent building blocks were used to create polyines in quasi-one-dimensional frameworks [22] or ‘superdiamond’ and related structures [5] in three-dimensionally connected networks. Another class of networks has been proposed by combining sp2 atoms in a nonlayered arrangement. The idea was to find a material based upon the thermodynamically most stable short-range order but with properties close to diamond. In order to make use of the short bonding between two graphitic carbon atoms to obtain the highest bulk moduli, Hoffman et al. [23] predicted the particular structure which they named BCT4 to be metastable. Tamor and Hass proposed that a very similar structure called H6 possibly exists [24]. Though the latter is topologically directly related to diamond it has been found to be rather unstable unlike the metallic phase BCT4 [25,26] and converts without a pronounced barrier back to diamond [27]. The most remarkable feature of these structures, as well as another energetically unfavorable structure [28] (BCC4 in the following), is that they do not contain any six-membered rings as is typical for graphite. Other purely three-fold coordinated networks have been proposed inspired by tesselations of infinite triply periodic minimal surfaces [28-321. All of these structures are composed of five-, six-, and seven-membered rings which are the major components in amorphous structures too. They establish the whole family of negatively curved graphitic sheets. The density of these structures is usually lower than that of graphite but they have been predicted to be more stable than fullerene c 6 0 [32,33]. Electronically the proposed structures behave like either metals or insulators with gaps smaller than that of f.c.c. c 6 0 [34]. An exception with respect to this is polybenzene [28] which has been found to be a low energy form of carbon that opens an indirect band gap of 2.96 eV [34]. This exceptionally large gap has been attributed to the very special topology which consists of

9.1 Introduction

273

benzene-like six-membered rings interconnected by a three-dimensional framework of single bonds [28]. By combining three- and four-connected atoms, Bucknum and Hoffmann generated a carbon network based upon the 1,4-cyclohexadiene motif. This structure has a density between that of graphite and diamond. Owing to the geometric constraints that force unconjugated double bonds face-to-face onto each other, the system shows a significant indirect band overlap across the Fermi level and is predicted to be thermodynamically quite unstable [35]. A whole family of similar networks which mix three-fold and four-fold coordinated carbon atoms has been proposed by Merz et al. [36]. Yin [37], Biswas et al. [38], Fahy and Louie [39], Mailhiot and McMahan [40], Boercker [41], Crain et al. [42], Furthmuller et al. [43], and Teter [44] all investigated a dense carbon phase containing sp3 atoms only as in diamond, which is known from silicon and is called BC8. This and other interesting metastable structures with four-fold coordination such as ST12 [45], R8 [42], lonsdaleite, and H3 [44] have been investigated to establish theoretically the stability limit for diamond. Another class of four-fold coordinated low-energy forms of carbon has been predicted by Nesper et al. [46], Adams et al. [47], and Benedek et al. [48], and is due to a carbon decoration of zeolite structures. In particular, structures known as clathrates provide ideal hosts for four-fold coordinated carbon atoms. In contrast, the compact icosahedral clusters with tetrahedral bonding proposed by Zeger and Kaxiras [49,50], though nice geometric arrangements, are quite high in energy and thermodynamically rather unstable if the outer dangling bonds are not saturated with hydrogen. Finally, metallic allotropes with higher coordination numbers have been studied [39,51] and found to be well removed in enthalphy from four-fold coordinated structures below a pressure of 10 Mbar. Here, we report on the eometric optimization and doping properties of the recently proposed purely sp bonded carbon phases [15] and a propotype zeolite structure (sp3 bonded) called melanophlogite using the density-functional based tight-binding method (DF-TB) [52,53]. The sp2 bonded structures are based upon the concept of connecting ‘ideal’ conformations of eight-membered rings. Hence, the structures fill the gap between carbon phases based upon benzene rings (graphite, polybenzene, some polyines) and the models based on an interconnection of tenfold rings (BCT4, H6, BCC4). The carbon crystals using eight-membered rings as the main structural motif have a similar stability to BCT4 and fullerene C60,and belong to local minima in the energy landscape for carbon [ 151. They are expected to be thermodynamically very stable, since a transition to diamond or graphite would require a large number of bonds to be broken and rearranged. The sp3-bonded zeolite type structures are obtained from natural silica frameworks by substituting the silicon atoms with carbon and removal of oxygen. We demonstrate that considering n-type doping such structures [4643] behave very much like diamond. A mid-gap state occurs after a substantial relaxation of the impurity which is localized on a nearby carbon atom. In contrast, the topologies of the purely sp2-bonded structures favor the incorporation of boron or nitrogen on substitutional lattice sites and do not permit a

5

274


reorganization of the carbon neighborhood. The relaxation is weak and ensures that injected holes or electrons remain associated with the substituent. They provide half-filled electronic levels close to either the valence or the conduction band. We compare the situation with doping effects in diamond, amorphous carbon, polybenzene, and the zeolites. The recent synthesis of crystals made of the DGhC36 fullerenes [54] and of a material called Quasam [55] (which has been declared to be the lightest hard substance yet discovered) supports the belief that there is a real chance of preparing novel forms of carbon such as those described in this paper. Quasam is believed to be made up of graphite-like planes joined together by four-fold coordinated carbon atoms and has been produced by fragmenting an organic compound.

9.2 Lattice Description The hypothetical carbon phases that we predict to be metastable are presented in their relaxed state in Fig. l(a, b). Figures l(c, d) show polybenzene and the melanophlogite structure. By exploiting the special feature of boat and chair conformations of eightmembered rings which contain sp2-sp2 bonds, potentially double bonded extremely hard three-dimensional networks can be constructed. Similar to ethylene these networks give rise to large n-n* splittings of the electronic levels at the Fermi energy. As a result a wide bandgap material with superior properties for electronic applications can be formed. For this, the connection of the octahedrons must be such as to minimize the x-orbital overlap of adjacent double bonds. A structure made from chair conformations of eight-membered rings is shown in Fig. 1(a). There exist two very similar uniform three-connected networks, the space groups of which have been identified by Wells [56] to be hexagonal P6222 and rhombohedral RTm. Carbon models can be made for these nets with equal bond length and angles as in graphite. The two topologies differ only slightly in the form of the screw axis incorporated. Hence, the properties of crystalline structures obtained by decoration with carbon atoms are expected to be rather similar. The screw axis in both cases is such that there is little n-orbital overlap between sp2 neighbors along the axis, whereas the n-orbital overlap parallel to bonds interconnecting the screw axis is maximized. Since both structures are topologically almost identical we will concentrate on discussing the rhombohedral form only. During relaxation an ideal crystal with graphite-like bond properties undergoes a transition to a less symmetric monoclinic form of space group P112/m. Following the naming conventions for BCT4 and H6, we will nevertheless call the structure R6. Structure l(b) which we will call BCT8 in its ideal configuration has bodycentered tetragonal symmetry (I4lmd) with the generating atom at the position i, 0). For the idealized structure with equal bond lengths and angles the lattice constants are given by a = b = 4d and c = J8d, where d is the common bond

(a,

9.2 Lattice Description

275

Figure 1. Perspective views of the wide band gap models showing the three-dimensionally connected networks.

length. This structure results from connecting boat-shaped eight-membered rings while avoiding the construction of smaller rings. Dropping the latter requirement and allowing six-membered rings to occur gives rise to a simple cubic lattice which is known as the polybenzene structure [28] and shown for completeness in Fig. l(c). The properties of this simple cubic structure denoted here as SC24 are expected to be different since the n-bonded subunits are isolated six-membered rings within the rigid single bond environment. Isolated double bonds do not exist in this structure. Topologically, both models belong to the class of Archimedian three-connected nets [56]. Each lattice point belongs to two eight-membered rings and to one of another kind. In case (b) the other ring type is a ten-membered ring whereas in model (c) it is the six-membered ring as mentioned above. It is worth noting that structure BCT8 is topologically related to R6 in the sense that it contains screw axes that are interconnected by bonds potentially double

276

9 Eflective Doping in Novel sp2 Bonded Carbon Allotropes

bonded. Here, however, the projection of the axis onto a plane perpendicular to it is a square, whereas in the former case it is a triangle. SC24 does not contain such a feature. BCT8 and R6 both contain a rigid multichannel structure which may be promising for applications where anisotropic transport at a microscopic level is of interest. Such properties have been predicted for hypothetical crystals made of carbon nanotubes [57]. However, synthesizing such crystals requires one to prepare, purify, and align carbon tubes of equal size [58]. Finally, we investigate the doping behaviour of a prototype zeolite structure named melanophlogite, which has been detected to be one of the lowest energy forms of this kind [46]. The space group of this structure is P m n . The simple cubic system contains 46 atoms in its primitive cell and we will refer to it as SC46. The model used throughout these studies is shown in Fig. l(d).

9.3 Computational Methods To prove that the topologies described above form metastable configurations if decorated with carbon we explored the energy landscape with respect to variations of the lattice shape. For this, we started from the idealized structures with graphite bond length and angles and let them relax to find the ground state with respect to the energy functional as defined in the density-functional based tight-binding method (DF-TB) [52,53]. Within this technique the total energy of a crystal is approximated by a sum over all occupied electron levels obtained by solving the many-body Schrodinger equation and a classical interaction that describes the ion-ion core repulsion and the double counting corrections. To solve the electronic problem efficiently we expand the wave function in a minimal basis of localized contracted orbitals and transform the Schrodinger equation into a general eigenvalue problem. For this we make use of the two-center approximation to evaluate the Hamiltonian and overlap matrix elements. The eigenvalues are then obtained by diagonalization and the Brillouin zone integration is done employing a Gamma-point sampling through the large conventional unit cell approach. The DF-TB method has been shown to coincide with more sophisticated self-consistent field methods considering the energetics and ground state geometries of small carbon clusters [52], fullerenes and fullerene oligomers [59,60], diamond surfaces [61,62] and amorphous carbon phases [63]. B and N impurities in amorphous material [ 121 and in sublayers of diamond surfaces [64,65] have been successfully described by DF-TB. The true local minima in the energy landscape were found by updating the atomic coordinates for a sufficiently large number of lattice vector combinations using an efficient conjugate gradient algorithm. It should be noted that the lattice vectors have been varied independently in order to find energetic minima without applying symmetry constraints. The lattice shape could change during the optimization process. Finally, the vibrational density of states within the quasi-harmonic

9.3 Computational Methods

277

Table 1. DF-TB ground state properties of hypothetical model structures compared with the results for natural carbon allotropes and recent calculations with more sophisticated methods. Structurea

Lattice parameters [nml

Density Atomic Bulk volume modulus [ g ~ m - ~[nm’/atom] ] [GPa]

Total energy E-EDlamond

[eV atom-’]

Bond lengthsb [nm]

Bond anglesb [degl

DF-TB results Diamond {(66)-a} Graphite

a = 0.3569

3.51

0.00568

49 1

a = 0.2458 c = 0.6700

2.27

0.00877

-

0.00 -0.01

c60

BCT8 { (82.10)-b} R6 {(8,3)-b) SC24 {(6.8*)} SC46 BCC4 {(10,3)-a} BCT4 {(10,3)-b) H6 {(10,3)-~}

0.1545 ( l / l ) 109.5 ( l / l ) 0.1419 ( l / l ) 120.0 ( l / l )

0.41

0.1397 (1/3) 108.0 (1/3) 0.1448 (2/3) 120.0 (2/3) 0.1342 (1/3) 0.1496 (2/3) 0.1343 (1/3) 0.1498 (2/3)

a = 0.5873 c = 0.3937 a = 0.4318 b = 0.3589 c = 0.7101 a = 0.6070

2.35

0.00849

297

0.52

2.26

0.00882

316

0.54

2.14

0.00932

324

0.22

a = 0.6671

3.09

0.00645

395

0.12

a = 0.41 16

2.29

0.00872

338

1.44

a = 0.2502 c = 0.8698 a = 0.2607 c = 0.6396

2.90

0.0068 1

449

0.49

3.18

0.00627

519

0.75

115.6 (1/3) 122.2 (2/3) 117.2 (1/3) 121.4 (2/3)

0.1399 (2/3) 120.0 ( l / l ) 0.1494 (1/3) 0.1550f 109.3 f 5.3 0.0028 0.1455 (I/]) 120.0 (I/]) 0.1425 (2/3) 0.1492 (1/3) 0.1462 (2/3) 0.1465 (1/3)

118.6 (2/3) 122.8 (1/3) 117.1 (2/3) 125.8 (1/3)

Reference section SC24 [28]

a = 0.6033

2.19

0.0091 1

-

0.21

SC46 [47]

a = 0.6680

2.95

0.00675

-

0.14

BCC4 [28] a = 0.4065 BCT4 [26Ic a = 0.2512 c = 0.8599 a = 0.2618 H6 [25jC c = 0.6327

2.37 2.94

0.00842 0.00679

-

362

1.26 0.52

3.18

0.00626

372

0.82

a

0.1396 (2/3) 120.0 ( l / l ) 0.1474 (1/3) 0.1437 ( l / l ) 0.1423 (2/3) 0.1480 (1/3) 0.1445 (2/3) 0.1468 (1/3)

120.0 (l /l ) 118.1 (2/3) 123.8 (1/3) 116.9 (2/3) 126.2 (1/3)

Wells’s [57] symbols for 3D-nets given in curly brackets. Ratios in brackets denote percentage of bonds or angles. Except for bulk moduli extended basis set results ([75] differences from [25,26] are small). ~

approximation has been calculated by diagonalization of the dynamical matrix for the ground state lattice to guarantee that there are no negative nor zero phonon modes. The static properties of the relaxed ground state models of BCT8 and R6 are summarized in Table 1 together with structural details of the lattices SC24, SC46, BCC4, BCT4, and H6 for comparison, all handled as described above. The results for the latter models agree very well with the recent more sophisticated

278

9 EfSeective Doping in Novel sp2 Bonded Carbon Allotropes

calculations, see Table 1, and demonstrate the accuracy of the tight-binding method used here.

9.4 Static Properties As has been expected from the local geometries, the lattices with screw axes inside respond to the carbon interaction by a strong n-orbital overlap along bonds connecting these helical subunits. Hence, the associated bonds are strengthened and shortened. Double bonds with the typical length of about 0.134nm are established. The bonds within the helices, however, have a pronounced single-bond character since the dihedral angles between the atoms prevent a strong n-orbital overlap. Therefore, these bonds are longer and have typical o-bond lengths close to 0.150 nm. For comparison, the ideal bond length for the diamond lattice is calculated to be 0.155nm with the DF-TB method. These quite different bond lengths introduce stress into the idealized crystals. Accompanied bond angles can no longer be equal and differ significantly from 120". However, on average the graphite value is still preserved which has been found for H6 and BCT4, too. The strain energy is balanced by the energy gain due to the n-bonded atom pairs. As a result, R6 and BCT8 are about 0.5eVlatom less stable than graphite. Due to the strong overlap between the n-orbitals along the shorter bonds the new crystals are expected to become insulators with a clean n-n* band gap. The energy of the metallic phase BCT4 is found to be about the same. Here, the nbonded system which lowers the energy consists of extended zig-zag chains within the crystal. Isolated double bonds do not exist in this system. Instead, graphitelike bonds parallel to the chains are established with a length of about 0.142nm. Note also that the unique bond length in graphite is calculated within DF-TB to be 0.142 nm. The polybenzene structure SC24 is more stable than the R6 and BCT8 by about 0.2eV/atom because all the bond angles in this structure remain at there ideal values during relaxation and the n-interaction further stabilizes the six-membered ring structures. While the former is also true for BCC4 the overall n-orbital overlap weakens the whole structure and makes it less stable than graphite by almost 1.5 eV/ atom. For comparison, the total energy of the lowest energy forms of amorphous carbon have been determined by the DF-TB method to be about 0.7-0.8 eV/atom above graphite. C60 is less stable than graphte by about 0.42eV/atom with the DF-TB technique which compares well to the experimental value of 0.44 eV/ atom [66]. Diamond is slightly less stable than graphite by 0.01 eV/atom which is close to the experimental value of 0.025 eV/atom [67]. Therefore, the structures BCT8, R6, SC24, and BCT4 may well exist, since their total energies are well below amorphous carbon material and comparable to the natural allotrope fullerene C60.In contrast, BCC4 is unlikely to be synthesized by the same arguments. The bulk moduli of BCT8, R6, SC24, and BCC4 are found to be about the same, i.e. almost 2/3 that of diamond. Hence, the structures are expected to be very hard.

9.5 Electronic Properties

279

H6 and BCT4 both have a bulk modulus which, within the tight-binding model is comparable in magnitude to that of diamond. In accord with an empirical TB scheme [24] H6 seems to be even harder than diamond while SCF plane wave calculations [25,26] predict that the almost equal bulk moduli of BCT4 and H6 are 17% below the diamond value. The bulk moduli of the models have been determined by calculating the elastic compliances after applying suitable strains to the crystals and inversion of the volume compressibility [68]. A four-fold coordinated structure obtained by decoration of melanophlogite with carbon has been determined to be one of the two lowest energy forms of zeolite type hypothetical carbon allotropes [46]. Within a converged SCF plane wave calculation the total energy of this structure was determined to exceed that of diamond by only 0.09 eV/atom. The band gap was found to be almost 4 eV and only 6% lower than that of diamond utilizing the local density approximation. We find the structure to be 0.12 eV/atom less stable than diamond with a band gap which is about 4.4 eV and 68% that of diamond.

9.5 Electronic Properties As mentioned above, BCT8 and R6 contain localized n-bonds within a o-bonded frame. These n bonds are well relaxed so that there is maximum overlap between the parallel n-orbitals. Hence, the structures develop pronounced gaps due to the n-n* splitting of states as is shown in Table 2 and the densities of states (DOS), see Fig. 2. In order to estimate the gap width and the activation energies more accurately than is possible within a minimal basis (sp-basis) approach used to relax the structures, we have recalculated the electronic properties of the models using an extended basis set that includes d-type polarization functions (spdbasis). This set was particularly constructed to improve the diamond and graphite conduction band structure above the Fermi level. The electronic DOS obtained for SC24, SC46, diamond, and graphite with the spd-basis is shown for comparison in Fig. 2. Table 2. Electronic properties of the hypothetical structures compared with diamond. sp indicates minimal, spd extended basis set, see text. B-C and N - C mark the HOMO participation of the atomic orbitals at the impurity and the nearest carbon neighbors. Structure

Diamond BCT8 R6 SC24 SC46

Gap width [eV]

B acceptor level [ev] N donor level [eV]

A 0 contribution to HOMO

SP

SPd

SP

SPd

SP

spd

B-C

N-C

7.75 2.99 3.37 3.65 8.22

6.49 2.72 2.79 3.45 4.43

0.09 0.17 0.37 0.03 0.20

0.09 0.20 0.24 0.01 0.34

4.61 0.32 0. I9 0.04 4.10

3.28 0.09 0.03 0.02 2.44

0.03 0.07 0.48 0.01 0.46

0.50 0.25 0.17 0.04 0.64

280

9 Effective Doping in Novel sp2 Bonded Carbon Allotropes 1.0 0.8 0.6 0.4

0.2 0.0

0.8 0.6

8n

0.4

0.2 0.0

"1"

0.8 0.6 0.4 0.2

-25

-20

-15

-10

-5

0

5

-20

-15

-10

-5

0

0.0

Figure 2. The (broadened) electronic densities of states for BCTB, R6, SC24, and SC46 show the wide band gap character of these materials. Diamond and graphite DOS are plotted for comparison.

The most remarkable feature of the structures presented is that they can take up boron and nitrogen at substitutional lattice sites without a strong relaxation. In recent DF-TB studies of the doping effect in diamond [12] we could reproduce the experimental findings that boron yields a shallow acceptor level and nitrogen a deep donor level which is impractical for electronic applications. Whereas boron relaxes only a little, nitrogen moves substantially off the substitutional lattice site. A lone pair electron and a dangling state at one of the carbon neighbors is created. The carbon defect lies deep in the gap at 3 eV below the conduction band

9.5 Electronic Properties

28 1

and pins the Fermi level. A Boron atom in contrast stays close to its lattice site. The associated acceptor level is about 0.1 eV above the valence band and pins the Fermi level. Both values are close to the experimental activation energies of about 2.00 eV for the single nitrogen donor and 0.37 eV for the boron acceptor [8]. Testing zeolite structures as possible hosts for boron and nitrogen is more complicated because the impurity can be placed on different lattice sites. We have checked the three unique places in the melanophlogite prototype structure and present the results for the best bonding situation found for either boron or nitrogen. We only observed shallow acceptor (activation energy = 0.32 eV) and donor levels (activation energy = 0.17eV) in the pure sp3 environment for SC46 when the local symmetries forced all four carbon neighbors of boron or nitrogen to relax away by exactly the same distance such that the B-N or C-N bonds had the same length. This caused degenerate levels at the Fermi energy to appear. Jahn-Teller distortions removed these degeneracies and yielded lower energy structures. Generally, all the ground states found were characterized by a off lattice site relaxation of the impurity. A bond to one of the carbon neighbors was stretched and became weaker than the others. In the nitrogen case this bond finally was broken due to the electrostatic interaction between the nitrogen lone pair orbital and the carbon dangling bond leaving an occupied deep carbon state. Therefore, it seems to be fairly general that n-type doping by nitrogen is inhibited in a carbon sp3 environment. Significant lattice relaxations to screen the injected charge plus the creation of too deep a carbon level seem to be always favored over establishing a real donor state. This is in sharp contrast to the hypothetical carbon crystals BCT8 and R6. In these structures the nitrogen atom is naturally incorporated at a lattice site in a favorable local geometry. It must take part in the two single and the isolated double bond associated with each site. The lattice relaxation is extremely weak and N stays at the lattice site as does a substitutional boron atom. The highest occupied molecular orbital (HOMO) is occupied by just one electron and acts as a donor and acceptor, respectively. The activation energies determined within DF-TB are summarized in Table 2. The HOMO is primarily located at the substitutional atom and spread over its neighbors as can be deduced from the participation analysis, see Table 2. Note, that the numbers given are the largest contributions found. The remaining parts are almost equally distributed over all the other atomic orbitals in the system. In the case of diamond and the zeolites the donor level participation ratio is very large indicating that the highest molecular orbital is strongly localized in the neighborhood of the impurity. Amorphous carbon materials consist of a mixture of sp3- and sp2-bonded carbon atoms and the latter form more or less extended n-bonded subclusters in a rigid single-bond environment [63]. Considering this structure a similar doping effect as discussed above might be expected. However, the size distribution of xclusters [69] and the associated strain in the systems yield broad bond length distribution functions [70]. As a result, the n- and rc*-bandsare broadened and the tails of the bands overlap with potential acceptor and donor levels. This causes electrons to be either trapped from the highest energy x-levels or injected into lowest energy x*bands, repectively. It is accompanied with relaxation processes of the carbon

282


network in order to saturate carbon defect states which are always present in amorphous carbon. Hence, doping is very inefficient in full contrast to the situation in the proposed crystalline allotropes where the injected hole or electron cannot migrate to distant n-states. Finally, we carefully checked the doping results obtained within the DF-TB scheme against a parameter-free self-consistent-field method employing normconserving pseudopotentials to model atom cores and a localized real-space basis set of Gaussian orbitals to expand the molecular wave functions and the charge density [7 11. This computational scheme invokes the local spin density approximation for exchange and correlation terms within the density functional theory to solve the many-body Schrodinger equation. The technique has been successfully applied to study diamond growth [72] reactions, graphitization tendencies at diamond surfaces [73] and the auto-catalysis of Stone-Wales transformations in fullerene C60[74]. However, the computational efforts within this code forced us to represent the new crystals by relatively small clusters. We cut out representative pieces of the structures including two screw axes and three double bonds in a row interconnecting them. The cut was done to ensure that only (J dangling bonds occured on the outside of the clusters which were saturated by hydrogen. The hydrogen atoms then were kept fixed in order to simulate the crystal boundary. The carbon atoms and the substitutional impurity could fully relax within this hydrogen cage by a conjugate gradient algorithm. The cluster representing BCT8 was found to have an electronic gap of 3.9 eV and the one representing R6 had a gap of 2.7 eV. The energy differences of the acceptor and donor levels to the respective band edges calculated within this technique are 0.63eV and 0.67eV in the case of BCT8 and 1.37eV and 0.38eV in the case of R6. The activation energies tend to be a little higher than determined within DFTB. However, this is not surprising taking into consideration that a hydrocarbon molecule with an artificially strained boundary is used to model the extended crystal. In the molecules, the central double bond is elongated compared to the outer two already without any impurity. Nevertheless, the calculations prove that the relaxation of the impurities is very small so that they do not remarkably move off the substitutional lattice site. Except for the boron acceptor in R6 we find shallow levels close to states representing the valence or conduction band edges, respectively. The differences compared to the periodic DF-TB calculation are enhanced owing to charge transfer effects to the surface of the molecule in particular into constraint bonds.

9.6 Conclusions We could show that it is possible to get n-type doping effects in carbon based materials using nitrogen as a dopant. The structures where this works, however, are purely sp2 bonded low energy forms of carbon. The semiconducting allotropes are characterized by unique screw axes of single bonded sp2 carbon atoms held

References

283

together by isolated well-relaxed double bonds. This particular local geometry opens wide band gaps and provides an ideal host situation for nitrogen. In contrast to usual sp3 bonded environments, nitrogen atoms stay at their lattice sites during relaxation and take part in the isolated double bonds, one of which is associated with each lattice point. The carbon crystals are metastable forms with a density close to graphite and large bulk moduli. These mechanical properties together with pronounced electronic gaps make the BCTS and R6 crystals attractive for electronic purposes, as protective coatings, and for optical applications. The purely sp2 bonded phases have a distinct multichannel architecture which may cause unusual anisotropic transport effects at a microscopic level.

Acknowledgments We gratefully acknowledge support from the Deutsche Forschungsgemeinschaft and the EPSRC.

References 1. H. W. Kroto, J . R. Heath, S. C. O’Brien, R. F. Curl, and R. E. Smalley, Nature, 1985,318, 162.

2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18.

19. 20. 21. 22.

W. Kratschmer, L. D. Lamb, K. Fostiropoulos, and D. R. Huffman, Nature, 1990, 347, 6291. S. Iijima, Nature, 1991, 354, 56. D. Ugarte, Europhys. Lett. 1993, 22, 45. F. Diederich and Y. Rubin, Angew. Chem.: Internat. Edn. 1992, 104, 1123. Z. Xu and J. S. Moore, Angew. Chem.: Internat. Edn. 1993, 105, 1394. A. T. Collins, Muter. Sci. Eng. B, 1992, 11, 257. 0.Madelung (Ed.), Semiconductors: Group IV Elements and III- V Compounds, Springer, Berlin, 1991. S. A. Kajihara, A. Antonelli, J. Bernholc, and R. Car, Phys. Rev. Lett. 1991, 66, 2010. C. Cunha, S. Canuto, and A. Fazzio, Phys. Rev. B, 1993,48, 17806. S. J. Breuer and P. R. Briddon, Phys. Rev. B, 1996, 53, 7819. P. K. Sitch, Th. Kohler, G. Jungnickel, D. Porezag, and Th. Frauenheim, Solid State Commun. 1996, 100, 549. P. Stumm, D. A. Drabold, and P. A. Fedders, J . Appl. Phys. 1997, 81, 773. S. Poykko, M. Kaukonen, M. J. Puska, and R. M. Nieminen, Comp. Muter. Sci. 1998,10,351. G. Jungnickel, P. K. Sitch, Th. Frauenheim, B. R. Eggen, M. I. Heggie, C . D. Latham, and C. S. G. Cousins, Phys. Rev. B, 1998,57, R661. J. F. Prins, Diamond Rel. Muter. 1995, 4, 580. H. Spicka, M. Griesser, H. Hutter, M. Grasserbauer, S. Bohr, R. Haubner, and B. Lux, Diamond Rel. Muter. 1996, 5, 383. S. Koizumi, M. Kamo, Y. Sato, S. Mitd, A. Sawabe, A. Reznik, C. Uzan-Saguy, and R. Kalish, Diamond Rel. Muter. 1998, 7, 540. S. Prawer, Diamond Re/. Muter. 1995, 4, 684. R. Jones, J. E. Lowther, and J. Goss, Appl. Phys. Lett. 1996, 69, 2489. R. H. Baughman, H. Eckhardt, and M. Kertesz, J . Chem. Phys. 1987, 87, 6687. U. H. F. Bunz, Angew. Chem.: Internat. Edn. 1994, 106, 1127.

284 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65.

9 Eflective Doping in Novel sp2 Bonded Carbon Allotropes R. Hoffmann, T. Hughbanks, M. Kertesz, and P. H. Bird, J . Am. Chem. SOC.1983,105,4831. M. A. Tamor, and K. C. Hass, J . Muter. Res. 1990, 5 , 2273. A. Y. Liu, M. L. Cohen, K. C. Hass, and M. A. Tamor, Phys. Rev. B, 1991,43, 6742. A. Y. Liu, and M. L. Cohen, Phys. Rev. B, 1992,45, 4579. M. 1. Heggie, and C. D. Latham, Euromat 91: Proc. Second Europ. Con$ Adv. Materials and Processes, T. W. Clyne, P. J. Withers (Eds), 1992, 3, 239. M. O’Keeffe, G. B. Adams, and 0. F. Sankey, Phys. Rev. Lett. 1992,68,2325. H. Terrones, and A. L. Mackay, Carbon, 1992, 30, 1251. H. Terrones, J. Fayos, and J. L. Aragon, Acta Metall. Muter. 1994, 42, 2687. D. Vanderbilt, and J. Tersoff, Phys. Rev. Lett. 1992, 68, 511. S. J. Townsend, T. J. Lenosky, D. A. Muller, C. S. Nichols, and V. Elser, Phys. Rev. Lett. 1992, 69, 921. R. Phillips, D. A. Drabold, T. Lenosky, G. B. Adams, and 0.F. Sankey, Phys. Rev. B, 1992,46, 1941. M.-Z. Huang, W. Y. Ching, and T. Lenosky, Phys. Rev. B, 1993, 47, 1593. M. J. Bucknum, and R. Hoffmann, J . Am. Chem. Soc. 1994, 116, 11456. K. M. Merz, Jr., R. Hoffmann, and A. T. Balaban, J . Am. Chem. Soc. 1987, 109, 6742. M. T. Yin, Phys. Rev. B, 1984,30, 1773. R. Biswas, R. M. Martin, R. J. Needs, and 0. H. Nielsen, Phys. Rev. B, 1984, 30, 3210. S. Fahy, and S. G. Louie, Phys. Rev. B, 1987, 36, 3373. C. Mailhiot, and A. K. McMahan, Phys. Rev. B, 1991, 44, 11578. D. B. Boercker, Phys. Rev. B, 1991,44, 11592. J. Crain, S. J. Clark, G. J. Ackland, M. C. Payne, V. Milman, P. D. Hatton, and B. J. Reid, Phys. Rev. B, 1994,49, 5329. J. Furthmiiller, J. Hafner, and G. Kresse, Phys. Rev. B, 1994, 50, 15606. D. M. Teter, private communication. R. Biswas, R. M. Martin, R. J. Need, and 0. H. Nielsen, Phys. Rev. B, 1987, 35, 9559. R. Nesper, K. Vogel, and P. E. Blochl, Angew. Chem.: Znternat. Edn. 1993, 105, 786. G. B. Adams, M. O’Keeffe, A. A. Demkov, 0. F. Sankey, and Y.-M. Huang, Phys. Rev. B, 1994, 49, 8048. G. Benedek, E. Galvani, S. Sanguinetti, and S. Serra, Chem. Phys. Lett. 1995, 244, 339. L. Zeger and E. Kaxiras, Comp. Muter. Sci. 1993, 1, 403. L. Zeger and E. Kaxiras, Phys. Rev. Lett. 1993,70, 2920. M. T. Yin and M. L. Cohen, Phys. Rev. Lett. 1983,50, 2006. D. Porezag, Th. Frauenheim, Th. Kohler, G. Seifert, and R. Kaschner, Phys. Rev. B, 1995,51, 12947. G. Seifert, D. Porezag, and Th. Frauenheim, Znt. J . Quantum Chem. 1996,98, 185. C. Piskoti, J. Yarger, and A. Zettl, Nature, 1998, 393, 771. S. Hill, New Scientist, 1997, 153, 14. A. F. Wells, Three-dimensional Net, and Polyhedra, Wiley & Sons, New York, 1977. J. Tersoff and R. S. Ruoff, Phys. Rev. Lett. 1994, 73, 676. P. M. Ajayan, Cond. Muter. News, 1995, 4, 9. G. Jungnickel, D. Porezag, Th. Kohler, Th. Frauenheim, and M. R. Pederson, Fullerene, and Fullerene Nanostructures, H. Kuzmany, J. Fink, M. Mehring, and S. Roth (Eds), World Scientific, Singapore, 1996. D. Porezag, G. Jungnickel, Th. Frauenheim, G. Seifert, A. Ayuela, and M. R. Pederson, Appl. Phys. A - Mat. Sci. & Processing, 1997, 64, 321. Th. Kohler, M. Sternberg, D. Porezag, and Th. Frauenheim, Phys. Stat. Sol. A , 1996, 154, 69. G. Jungnickel, D. Porezag, Th. Frauenheim, M. I. Heggie, W. R. L. Lambrecht, B. Segall, and J. C. Angus, Phys. Stat. Sol. A , 1996, 154, 109. Th. Frauenheim, G. Jungnickel, Th. Kohler, and U. Stephan, J . Non-Cryst. Sol. 1995,182, 186, and refs. therein. P. K. Sitch, G. Jungnickel, M. Kaukonen, D. Porezag, Th. Frauenheim, M. R. Pederson, and K. A. Jackson, J. Appl. Phys. 1998, 83,4642. M. Kaukonen, P. K. Sitch, G. Jungnickel, R. M. Nieminen, S. Poykko, D. Porezag, and Th. Frauenheim, Phys. Rev. B, 1998,57, 9965.

References

285

66. H.-D. Beckhaus, S. Verevkin, C. Ruchardt, F. Diederich, C. Thilgen, H.-U. ter Meer, H. Mohn, and W. Miiller, Angew. Chem.: Internat. Edn. 1994, 33, 996. 67. M. T. Yin and M. L. Cohen, Phys. Rev. B, 1984, 29, 6996. 68. J. F. Nye, Physical Properties ojCrystu/s, Clarendon, Oxford, 1985. 69. U. Stephan, Th. Frauenheim, P. Blaudeck, and G. Jungnickel, Phys. Rev. B, 1994, 50, 1489. 70. G. Jungnickel, Th. Kohler, M. Haase, S. Deutschmann, U. Stephan, P. Blaudeck, and Th. Frauenheim, J . Non-Cryst. Sol. 1995, 183, 161. 71. R. Jone, and A. Sayyash, J. Phys. C: Solid State Phys. 1986, 19, L653. 72. C. D. Latham, M. I. Heggie, R. Jone, and P. R. Briddon, Diamond Rel. Muter. 1994, 3, 1370. 73. C. D. Latham and M. I. Hegaie. Diamond Re/. Muter. 1995. 4, 528. 74. B. R. Eggen, M. I. Heggie, EyJungnickel, C. D. Latham, R. Jone, and P. R. Briddon, Science, 1996. 272. 87. 75. A. Y. Liu; private communication.

Part I1

Synthesis and Processing


1 Directed Metal Oxidation V. Jayaram and D. Brandon

1.1 Historical Background ‘Directed metal oxidation’ (DMO) is a term coined by Newkirk and his coworkers [ 1,2] to describe the formation of a ceramic product by oxidation of a liquid metal in a growth direction normal to the original free surface of the liquid metal. In the processes which have been patented by these workers, the liquid metal is commonly an aluminum alloy and the oxidant air, while growth is usually into a porous ceramic preform. The end product is a polyphase ceramic composite which may be tailored to have rather unusual properties, most notably good thermal conductivity, high wear resistance, and good fracture toughness. These properties are a direct consequence of the microstructure of the product, which usually contains some residual metal (enhancing both the thermal conductivity and the fracture toughness). Although the commercial development and exploitation of DMO is relatively recent, the phenomenon has been known for over 70 years, and can be traced back to the search for suitable refractories which accompanied the dramatic growth of the aluminum industry following the end of the First World War. The refractory linings of crucibles and furnaces used in the new industry were sometimes observed to be attacked by the molten metal, the products of the attack penetrating into the refractory brick to form a hard dense mass which cracked under the thermal and growth stresses, allowing the molten metal to escape. The removal and replacement of the damaged refractory could only be accomplished using road-drills and jack-hammers, so that, even at that time, the unusual mechanical properties of the DMO material were noted, if not appreciated [3]. Residual sodium in the refractory was quickly identified as the primary element catalyzing the attack, but with the development of magnesium-containing aluminum alloys it was recognized that sodium-free refractory linings might also be attacked in the presence of magnesium, originating either in the refractory or in the composition of the alloy melt [4,5]. Work on the development of refractories for the aluminum industry continues to this day, with specific instances of refractory penetration recurring as new alloys are developed, most recently for the Al-Li alloys. In all cases it has been noted that the metallic forms of the elements catalyzing the attack have unusually low vapor pressures: Na, Mg, Li, and to a lesser extent Zn. The process patented by Newkirk and his associates (more than 60 patents were filed in the period 1985-90) covered both the composition of the aluminum alloy (now to be used as a raw material for the manufacture of the DMO product), the use of an ‘initiator’ (to promote nucleation of the growth process), control of the growth rate and microstructure (primarily through the composition, structure,

290

1 Directed Metal Oxidation

and geometry of the preform), and the termination of the reaction (by a calcium containing ‘poisoning’ agent, used to coat the external surface of the preform), as well as the actual DMO process conditions (temperature and oxygen partial pressure). Subsequent work extended the DMO concept to the gaseous nitridation of liquid aluminum to form AlN, as well as to a number of boride and carbide products (especially titanium and zirconium) formed by preform infiltration [2,6]. At the time of writing (1998) it is not at all clear what the commercial future of these processes will be. It should be noted that there is a wide range of ‘ceramic’processing technologies under development, all of which are based on chemical reactions occurring in combinations of solid, liquid, and gaseous phases. Concrete is an unconventional example, not usually classed as a ceramic. More conventionally, reaction bonded silicon nitride (RBSN - gaseous nitridation of a silicon preform), reaction sintered silicon carbide (the REFEL process - molten silicon infiltrated into a carboncontaining silicon carbide preform), reaction bonded aluminum oxide (RBAO gaseous oxidation of an aluminum-containing alumina preform), and boron carbide reaction bonded by infiltrating a molten aluminum alloy, all rely on a chemical reaction to bond the ceramic product into a shaped component. The DMO process produces a near net-shape product without the need for any externally applied pressure, and is (in principle) capable of being used for the economic manufacture of very large ceramic components. It is also capable of yielding a ceramic composite product with some very desirable engineering properties. Despite the high expectations attached to the process and the extensive investment, especially by the American government but also by private industry, it is still not clear which areas of application are going to prove economically viable in the long term.

1.2 Oxidation and Oxide Formation The classical approach to the oxidation of metals has been as a one-dimensional mass transport problem, in which the transport is across a growing, planar film of the product and is driven by the free energy difference between the anodic and cathodic processes occurring at the metal-oxide and oxide-atmosphere interfaces, respectively. The anodic reaction at the metal-oxide interface, in its simplest form, is just: M = Ma+ ne, and represents the transfer of metal atoms across the interface to form cations with the accompanying release of free electrons. The corresponding cathodic reaction may take many forms, of which the simplest (in the presence of 2e = 02-,in which a partial pressure dry oxygen and no other reactants) is: of oxygen in the atmosphere supplies oxygen molecules which are ionized at the oxide-interface to form anions. Mass transport may limit the one-dimensional process in a number of ways. Transport of electrons (or holes) across the growing film is generally rate limiting

+

i02+

1.2 Oxidation and Oxide Formation

29 1

in the passivation reaction occurring during the formation of stoichiometric, nonconducting oxides (aluminum or titanium, for example). Electron and hole transport often depends on quantum tunneling, leading to a logarithmic growth law and the formation of the uniform, passivating film. In semiconducting, nonstoichiometric oxides it is the diffusion of cations or anions through the growing oxide layer that limits the growth. Ionic diffusion generally depends on the presence of point defects: cation or anion vacancies or interstitials, and is a thermally activated solid-state diffusion process. At constant temperature, diffusion limited growth generally follows a parabolic growth law, and the oxide layer offers only limited protection to continued oxidation of the metal. In two well-recognized cases the oxidation reaction proceeds at a constant rate, unimpeded by the reaction product. Thus, molybdenum at high temperatures forms a volatile oxide which leaves the metal unprotected against continued attack, while copper at elevated temperatures forms an oxide layer in residual tension which cracks and peels to expose fresh metal to attack. In such cases the rate of reaction is only limited by the supply of reactants to the two interfaces, the sites of the anodic and cathodic reactions. In general, the supply of metal atoms for the anodic reaction is not a problem, and it is the supply of oxygen (the partial pressure of oxygen in the gas phase) which is the rate-limiting step. However, neither the anodic nor the cathodic reactions need be quite as simple as represented here, and the rate-limiting reaction step in the formation of a steadily growing oxide may not be obvious. In the DMO process, as we shall see, the metal is supplied to the reaction front by capillary flow through a network of interconnected channels and growth occurs entirely at the oxide-atmosphere interface, outwards from the oxide product. DMO is therefore one oxidation process in which the supply of metal for the anodic reaction may indeed be rate controlling. However, the processes occurring at the oxide-atmosphere interface have been found to be rather complex, and involve both the formation of a metastable oxide film and a wetting reaction between the oxide product and the molten metal emerging from the capillary channels. Clearly, neither the multiple cracking and fragmentation of an oxide product, to allow free access of oxidant to the metal surface, nor the percolation of liquid metal through a network of capillary channels in the product, as occurs in DMO, are particularly suited to process modeling based on one-dimensional mass transport.

1.2.1 Initial Oxidation Some of the voluminous literature on the oxidation and corrosion of aluminum and its alloys has a direct bearing on DMO. Pure aluminum is normally covered by an amorphous native oxide film which is partially converted to y-alumina at the interface between the parent metal and the amorphous oxide when heated to ~ ~ 4 5 0 in °C dry air [7-91. This unusual behavior is explained by growth of the amorphous phase, through outward cation migration, while thickening of the y-alumina is by epitaxial growth on the parent metal, controlled by inward oxygen diffusion. Termination

292

I Directed Metal Oxidation

of the oxidation process at elevated temperatures has been attributed [9] to the build-up of biaxial residual stresses which inhibit further diffusion. The presence of moisture initially increases the rate of oxidation due to the development of hydrogen blisters. Only when the metal is in the liquid state (700-1 100°C) does a-alumina form directly [lo-121. The influence of magnesium additions has been summarized by Lea and Ball [13] and Field [14]. The first authors [13] showed that y-alumina formation is accelerated at the Al-amorphous alumina interface, as expected since magnesium cations are incorporated in the spinel structure. Subsequently, magnesium diffuses rapidly to the surface along grain boundaries to form MgO, with concurrent crystallization of the amorphous phase to y-alumina. Finally, the MgO layer becomes continuous so that the rapidly diffusing magnesium no longer has direct access to external oxygen and reacts instead with y-alumina to form MgA1204, releasing aluminum metal particles which are embedded in the spinel. By contrast, Field [14] detected no spinel but rather direct oxidation of magnesium to primary MgO below a cracked, amorphous surface oxide layer. This was followed by secondary precipitation of MgO within the amorphous alumina, which was ascribed to reduction of alumina by magnesium. The formation of MgO on alloys containing as little as 0.5% Mg [15] at =40O0C has been confirmed, while spinel and reduced, aluminum metal-rich particles are frequently observed in the oxide film [16]. Additions of copper and silicon inhibit crystallization [ 171, possibly by keeping magnesium in solution. Strong segregation of magnesium at grain boundaries [18-201 and surfaces [20] was cited as the probable cause for formation of MgO in dilute alloys [18,19,21]. In situ microscopy [22] has revealed MgO nucleating at grain boundaries in Al-Mg-Zn, suggesting that segregation could stabilize MgO and rapid grain boundary diffusion could maintain a flux of magnesium for further growth. Experiments on liquid alloys of A1-Mg [23] in the range 650-900°C have demonstrated that a duplex layer of MgO and spinel is formed initially, the former by direct oxidation at the free surface and the latter by reaction of MgO with the liquid alloy. The proportion of MgO in the duplex layer peaked (after 2000 s at 800°C) and was gradually replaced by MgA1204,with reducing magnesium content in the alloy. Measurements of the strength of the oxide film have implied [24] that additions of Ca, Li, and Na initially increase the oxidation rate, while Be, in quantities as small as 0.1 YO,considerably reduces the rate of oxidation. Zinc additions to A1-Mg alloys [25] were found to greatly increase the oxidation rate at very low concentrations (0.2%). In one elegant study of diffusion couples between Al-Mg and alumina/spinel/magnesia [26] it was found that reduction of spinel to alumina by aluminum was slow, while an A1-3.6Mg alloy readily reacted with MgO and alumina to form spinel. In particular, silicon contamination was found to promote alloy wetting of the oxide and to render the interfacial oxide less protective. The relevance of these results for the DMO process can be summarized by three points: The importance of MgO as an ubiquitous surface layer that promotes rapid oxidations; the importance of limiting the amount of magnesium in the alloy so that the oxide layer, through which the melt migrates, is not too thick; and the effects of Si, Zn, and Li in promoting the DMO process.

1.3 Related Ceramic Processing Routes

293

1.3 Related Ceramic Processing Routes The production of ceramic components has traditionally been by powderprocessing, in which the conversion of the porous ‘green’ body into the final, sintered product has been accompanied by a large volume change, typically of the order of 40%. DMO is just one possible production route for ceramic products which promises substantial savings by eliminating this sintering shrinkage and ensuring that the initial product approximates the final, specified geometry. That is, DMO is a ‘near-net-shape’ process. The casting of concrete and the blowing of glass into a mold are also ‘near-net-shape’ processes, but in the case of structural ceramics the mechanical properties of engineering interest are well beyond the values achievable in a cement reaction or in a glass. In RBSN, a silicon metal powder preform reacts with nitrogen gas to form silicon nitride. The reaction occurs by percolation of the gas phase into the open porosity of the silicon preform, while the large specific volume of the silicon nitride product substantially reduces the volume fraction of residual porosity (down to a lower limit of approximately YO). The mechanical properties of RBSN are inferior to those of sintered silicon nitride powder products, but the reduced costs of machining the near-net-shape product is a very attractive advantage. RBAO involves a similar production route, but in this case the aluminum metal powder is a minor component in a powder mix that contains a substantial fraction of alumina. Oxidation of the aluminum metal by gas percolation yields a metastable oxide phase, which is replaced at higher temperatures by the stable corundum phase. Again, the large specific volume of the oxide product reduces the volume fraction of porosity, but in this case the thermal cycle actually ‘sinters’ the product, and for the optimized powder formulation, expansion during oxidation of the metal at a moderate temperature is followed by the sintering contraction at a higher temperature to yield a dense alumina product whose final shape and volume is still very close to that of the ‘green’ powder preform. A similar reaction route has been developed for reaction bonded mullite (RBM, aluminum silicate). In this case the powder formulation also includes a source of silicon (silicon metal, silicon carbide, or silica), in addition to aluminum metal and alumina, and the final product contains mullite as the major phase. The RBM process is frequently modified by additions of zircon (ZrSi04), which provide an additional source of silicon and result in a dispersion of fine zirconia particles in the product which restricts grain growth to ensure a fine uniform microstructure. It is again possible to reduce residual porosity while retaining the near-net-shape characteristics, but at the time of writing this process is still under development and some way from commercial exploitation. A rather different approach has been developed for reaction sintered silicon carbide (RSSC), first developed in the former Soviet Union. In this process a powder preform of mixed graphite and silicon carbide is immersed in a liquid bath of molten silicon. The silicon wets and infiltrates the preform, reacting with the finely-divided graphitic component (carbon black). In the best case, all graphite is reacted and the residual silicon content is no more than a few percent. The product of the reaction, silicon carbide, firmly bonds the silicon carbide preform powder

294


with effectively zero residual porosity. This process (patented as the REFEL process in the western world) has been somewhat neglected in recent years, and the heattreatments developed to tailor high performance, high toughness microstructures in Sic have yet to be applied to RSSC. Infiltration of alumina preforms by aluminum alloy melts has also been achieved, both by applying external pressure (in the absence of adequate wetting and capillary absorption), and by ensuring the presence of wetting aids in the melt (most notably magnesium, as in the DMO process). This infiltration processing route has not yet been exploited commercially. Reactive wetting by an aluminum alloy melt is also the basis of the process developed for boron carbide and marketed by the Dow Corporation. In this case a B4C powder preform is reacted with an aluminum melt at elevated temperatures. The molten aluminum is sucked into the pore channels in the preform by capillary forces and reacts to form a range of mixed aluminum borocarbides. The final ceramic product is essentially pore-free, but contains a network of residual aluminum alloy. Since the ceramic phases are continuous, the product is hard, but the residual aluminum phase is also interconnected (while the volume fraction of residual alloy is to some extent controllable) so that crack propagation is significantly inhibited. As we shall see below, both the microstructure and the mechanical behavior of DMO products are similar to those obtained by reactive infiltration. The combination of infiltration and reaction that characterizes DMO has been exploited to make a number of composites. As long ago as 1953, it was shown that silica containing refractories were reduced by molten aluminum to form alumina and silicon [4]. Subsequently [27], the displacement reaction was extended to the formation of composites of alumina with residual Al-Si. More recently, the A1-Si02 displacement reaction has been used in the infiltration of dense preforms of silica [28] and mullite [29,30] by molten aluminum. Extension of the reactive infiltration process to porous silica-containing preforms [31,321 has resulted in the fabrication of metal-matrix composites in which the silica was replaced by a mixture of about 65% alumina and 35% metal, while the pores were infiltrated by molten alloy. In contrast to DMO, the displacement reaction appears to proceed at a critical temperature of X I 100-1200°C and without the need for a volatile solute element or oxygen. Borosilicate glass has also been used as an initiator to enable the infiltration of Al-Si alloys into alumina preforms [33]. Closer in concept to the DMO process is the infiltration of aluminum alloys in nitrogen to yield A1-A1N composites. Low temperatures (< 1000°C) and high magnesium or strontium content promote the spontaneous infiltration of liquid metal with a small concurrent nitridation to yield dispersions of A1N [34]. Similar alloys may be infiltrated at higher temperatures, resulting in nitride contents that increase with temperature to yield A1N-matrix composites [35-381. This process can result in particulate loadings of up to 75%. The mechanical properties of these aluminum nitride composites have been extensively characterized [39]. A further extension of the DMO process has been in the fabrication of silicon nitride and titanium nitride matrix composites by nitridation of the appropriate molten alloys [6,40], but it is unclear whether these processes will have commercial applications. Other examples of reactive infiltration include aluminum into Si3N4to

295

1.4 Directed Metal Oxidation Incubation

yield an A1N-Si3N4-Al-Si composite [41] and zirconium into B4C to yield ZrB2ZrC-Zr [4244].

1.4 Directed Metal Oxidation Incubation Incubation is the first stage in the directed metal oxidation of a liquid metal, preceding the actual growth process, and corresponds to a well-defined initiation process. In the case of Al2O3-A1 composites, initiation is frequently delayed for a considerable period, up to several hours, and depends on the alloy composition, the heating rate, and the final temperature [45-511. Gross variations in this incubation period lead to lack of reproducibility in the thickness of composite obtained in a given time. It has sometimes been concluded that melt oxidation was absent when the duration of the experiment was shorter than required for the transition from incubation to growth (e.g. [52]). Incubation can be avoided by seeding the melt surface [45], but the development of surface oxide during the incubation period provides an understanding of the transition from low temperature behavior to the high temperature growth regime and sheds some light on the phenomenon of oscillatory growth rates, which commonly accompany formation of the aluminum/alumina composite. While incubation is found in nearly all DMO alloy systems, the most detailed work has been on A1-Mg and Al-Mg-Si alloys. A typical isothermal ternary section and PO2-composition diagram are shown for A1-Mg-0 in Figs 1 and 2. The alloy composition corners for three-phase equilibria (alloy alumina spinel and alloy spinel magnesia) shift to higher magnesium content with increasing temperature, while at a fixed temperature, magnesium depletion of the melt reduces the stability of MgO to form first MgA1204 and then A1203.The PO2 dependence shows that decreasing magnesium levels in the melt increase the equilibrium oxygen activity of the melt in equilibrium with spinel. When heated in air, the oxide layer formed on the melt at low temperatures is MgO. As the temperature increases, the stable oxide in equilibrium with the melt changes from magnesia to spinel [51]. For a typical magnesium content of 3%, the critical temperature for this transition is about 750°C. Thermogravimetric analysis (TGA) at a constant heating rate reveals a characteristic abrupt rise in the oxidation rate when spinel is formed at the interface between the alloy and the surface film of MgO. The temperature of the peak in oxidation rate increases with heating rate, confirming the importance of kinetic factors. Two mechanisms have been suggested for the formation of spinel: Direct oxidation of the alloy melt through cracks in the MgO film [48,49]; or an interfacial displacement reaction between aluminum and MgO, followed by solid state diffusion of magnesium through the oxide to repeat the oxidation to MgO at the surface [47,5 I]. Evidence for the latter mechanism includes the smaller molar volume of oxygen ions in spinel when compared to that in magnesia, so that for a given quantity of

+

+

+

+

296

1 Directed Metal Oxidution

T=l100"C

A'2°3

\ \ \

iquid)

~ ~ ~ 0 . at.'IoMg 4 4 6 (0.40wt.%)

Figure 1. Ternary isothermal section for the A1-Mg-0 system at 1100°C [47].

T = 1100°C

.23 I

I

0.2

Al

I

I 0.4

I

I

0.6

I

I

I

1-36

0.8

Mg

Figure 2. Stability fields for the alloy and corresponding oxides as a function of oxygen partial pressure at 1100°C in the system A1-Mg-0 [47].

I .4 Directed Metal Oxidation Incubation

297

I I ‘ I

Composite Growth (Oscillatory) I

Time Figure 3. Schematic of weight gain (TG) and corresponding rate of weight gain (DTG) curves for oxidation of a binary A1-Mg alloy in the temperature range 1050-1250°C[51].

oxygen, the volume of spinel formed should be less than the equivalent volume of periclase, suggesting that the remaining volume could persist as liquid metal-filled channels in the spinel which maintain contact with the MgO. It has also been shown that mechanical disruption of the surface [48] can induce preferential formation of spinel, although this does not appear to be a prerequisite for the DMO process [5 11. A classical incubation period, characterized by extremely low rates of weight gain, is then observed as the temperature increases (Fig. 3). The rate of spinel formation drops as the spinel zone develops a duplex structure characterized by a dense, metal-free layer at the interface with MgO but containing microchannels of alloy near the melt [48,51]. It has been established that the true external surface is composed of MgO, by using low voltage X-ray spectroscopy in the scanning electron microscope [51], and this has been attributed to de-mixing of cations in spinel in the presence of an oxygen potential gradient [53]. These microstructural changes have been attributed to the diffusion-limited supply of magnesium from the melt reservoir which prevents the continued formation of MgO. Direct oxidation of the alloy melt then leads to sealing of the metal channels by spinel, preventing contact between the melt and the surface MgO. It is worth emphasizing the change in the oxidation maxima, and the subsequent drop in oxidation rate between the ‘low’ and ‘high’ temperature regimes. In the low temperature regime, below 950°C for A1-3 Mg, magnesium is completely depleted from the alloy reservoir and the oxidation product shifts from MgO to spinel, though never to alumina [23,51]. In the high temperature regime, the incubation

298


period is characterized by bulk magnesium levels that are only marginally reduced, together with a duplex surface oxide as described above. The temperature for the onset of the spinel ‘peak’ increases with both magnesium content and heating rate [46,51], as would be expected from both the phase equilibria and the activation energy for spinel formation. The end of the incubation period [49,5 11 is marked by a reversal of the phenomena that led to the low incubation rate: Alloy channels that were isolated from the surface, reconnect and advance through the spinel layer. The mechanism for regeneration of the liquid metal channels seems to be associated with an increase in melt oxygen activity in equilibrium with spinel when the magnesium concentration decreases. Hence magnesium depletion at the surface, due to the rapid formation of MgO, should promote spinel dissolution and promote capillary wicking of molten alloy through the spinel to the free surface. This hypothesis is supported by the reduced incubation time for ternary alloys that contain silicon, since silicon is known to reduce the activity of magnesium in the melt and should promote spinel dissolution [54]. Cracking of the surface oxide film could also lead to direct exposure of the melt to the gas phase [48], but this does not appear to be a prerequisite for terminating the incubation stage. A simple thermodynamic criterion for the molten alloy to repenetrate the channels is that the liquid-spinel interfacial energy should be less than one half the grain-boundary energy, and this criterion may only be satisfied above some minimum temperature [ 1,491. This could explain why alloy repenetration does not begin at a temperature of 900°C for an A1-3Mg alloy, although this temperature is past the peak rate of spinel formation and well into the incubation regime. When the alloy in the channels penetrates to the spinel-magnesia interface near the free surface, their composition should be in equilibrium with spinel, leading to the formation of spinel nodules [49,51,55]. Spinel-alloy composite growth should then continue until limited by further magnesium solute depletion to below the concentration defined by the alloy + spinel alumina ternary equilibrium. At this juncture, the spinel nodules should be replaced by alumina nodules, and the alloy composition in the advancing composite should be given by that determined by the spinel-alumina buffer layer at the original melt surface. The extent of spinel growth which occurs before the A1203-A1 composite begins to form increases with the magnesium content of the alloy and, for a fixed magnesium composition, decreases as the temperature increases, as predicted from the ternary Al-Mg-0 equilibrium. For example, an A1-8 Mg alloy has been observed to grow a spinelbased composite to a thickness of hundreds of microns at 1100°C until the alloy magnesium concentration reaches =0.5%, at which point alumina is the stable oxide. In contrast, an A1-3 Mg alloy was depleted to a similar critical magnesium concentration after ~ 1 0 pm 0 of spinel growth [51]. The oxide at the external surface of the composite is predominantly a thin (0.1 pm) layer of MgO with an intermittent, underlying spinel layer: The microstructure is discussed below. It could be argued that the entire process of incubation and reinitiation is caused by excess magnesium, and that a starting composition near the three-phase equilibrium should trigger immediate growth of alumina. However, controlled growth of a DMO composite has never been reported from a bulk

+

1.4 Directed Metal Oxidation Incubation

299

binary Al-Mg alloy of the predicted composition, possibly due to additional depletion of solute in the temperature gradient of the furnace during the initial heating period. In binary Al-Zn alloys, alumina composite growth has been initiated, without the formation of a surface spinel layer, by immersing a magnesium spinel seed in the melt [56]. Partial dissolution of the spinel provides the required concentration of magnesium in the alloy. Similar growth could not be achieved in the absence of zinc. Even in the case of Al-Zn it appears that incubation is not entirely avoided, although the incubation time may now reflect the time taken for the seed spinel to dissolve. Binary Al-Zn alloys containing no magnesium also have an incubation period for DMO [57], even though thermodynamic criteria do not predict any oxide other than alumina. In this case the delay in the onset of growth could be associated with delayed rupture of the surface oxide by zinc vapor. The duration of incubation is determined partly by the heating procedure. Anomalously long incubation times are encountered at intermediate temperatures (>70 000 s at 1 100°C) when the alloy is heated in argon [46]. Subsequent introduction of oxygen at the final processing temperature leads to reaction with magnesium vapor to form a thick layer of MgO, which then has to be reduced and penetrated by the underlying molten alloy. Because the magnesium vapor pressure increases rapidly with temperature in the range 950-1200°C, the advantage of rapid reduction could be outweighed by the greater thickness of MgO formed [46]. Consistent elimination of the incubation stage can be achieved by seeding the melt surface with silica [45]. While bulk silicon appears to reduce the incubation time marginally, the effect of a surface layer of S i 0 2 is much more pronounced, probably serving as an additional source of oxygen during the spinel growth peak. The large molar volume contraction which accompanies the reduction of silica and the formation of spinel helps to form alloy channels that are enriched in silicon and are associated with residual seed particles [51]. If the seed particle is large enough, the magnesium concentration will be depleted to the point at which alumina is stabilized, even before the particle is penetrated by melt channels. If the seed particle is too small, spinel nodules will grow from the silica. In either case, the size of the silica grains determines the time for the persistence of a silicon-rich alloy in the channels, which in turn determines the transient growth rate of spinel and the subsequent reduction of magnesium activity. The silica seed particles are converted initially to the binary spinel-alloy, and finally to the alumina-alloy composite, at a rate which is determined by the A1-SiO2 displacement reaction. The passivating effect of spinel and the ensuing solid state growth of MgA1204-Mg0 are then bypassed. Other seed materials have been shown to eliminate the incubation stage, specifically MgO [50] and NaOH [58]. In the case of seeding by MgO, an MgO-coated Al-alloy gave immediate DMO alumina growth, although the extent of composite formation was limited by the quantity of seed material. NaOH seeding gave the same immediate onset of DMO growth on pure A1 with limited composite growth. In both cases, it would appear that the seed particles act as sources of oxygen and facilitate the formation of a surface aluminate by a displacement reaction which is interspersed with alloy channels. Both the MgO and the NaOH seed particles also act as a source of the volatile constituent that is essential for growth of the DMO composite.

300


1.5 Directed Metal Oxidation Growth 1.5.1 Introduction The formation and subsequent breakdown of the spinel layer which occurs during the incubation stage is followed by the growth of A1203-Al composite. This is a complex phenomenon whose details are still not fully understood. Incremental advance of the growing composite involves at least three steps. (i) Transfer of oxygen from the atmosphere to the condensed phases. (ii) Reaction between A1 and 0 to form A1203. (iii) Flow of molten alloy from the reservoir toward the surface. In addition to these basic processes, the infiltration of preforms by the oxidation of a complex alloy in an oxidizing atmosphere (air) must involve diffusion of the nonoxidizing solute elements (and nitrogen) away from the interface, as well as wetting of the preform by the molten alloy and adaptation of the growth process to the tortuosity of the preform pore space. Early attempts to model DMO [l, 45,591 assumed a steady state process controlled by ionic transport across a surface oxide layer. Subsequent evidence demonstrated that the surface chemistry was more complicated and that growth was cyclic, far from equilibrium, and selflimiting. Most of the systems that have been studied were based on A1-Mg, but DMO composites have also been grown from Al-Zn, Al-Na, and Al-Li alloys, implying that the microstructures produced and the underlying growth mechanisms should be similar. The growth of an A1203-A1 composite from a binary Al-Mg alloy into air will serve as a model system that avoids the complications of multiple solute elements but retains the essential features of oxidative growth.

1.5.2 Directed Metal Oxidation Composites from AI-Mg Alloys An Al,03-A1 composite growing outward from the original molten Al-Mg alloy surface into the atmosphere contains two fully interconnected, continuous phases, the ceramic oxide and the molten metal. The ceramic volume fraction is normally in the range 7&85% and the temperature of formation is in the range 10001300°C [ 1,601. Higher growth temperatures lead to porosity and premature termination of the growth process. The contiguity of the metallic phase is a condition for the supply of molten metal from the alloy reservoir to the advancing growth interface, while the presence of an interconnected ceramic skeleton is inferred from the crystalline texture in the alumina, which consists of columnar ‘grains’, several tens of microns wide, percolated throughout the thickness of the composite and with occasional low angle boundaries [47,60,61]. Elastic moduli measurements, discussed later [60], confirm the high degree of connectivity of the stiffer oxide phase.

1.5 Directed Metal Oxidation Growth

301

The mean intercept lengths of the ceramic and alloy channels, measured perpendicular to the growth direction, are about 5 pm and 2 pm, respectively. These intercept lengths are not related to the grain sizes, and both phases are monocrystalline over distances hundreds of times larger than the measured intercept lengths. In effect, the liquid metal in the composite develops a grain size upon solidification which is similar to that seen in a slowly cooled, cast aluminum alloy. A typical growth history is shown schematically in Fig. 3. Following the incubation period (when present) the growth rate rises to a maximum. During this period, the oxide nodules nucleated during incubation (initially spinel and subsequently alumina) grow radially from their point of nucleation, coalesce and cover the surface of the melt. Within each nodule the orientation of the alumina crystal lattice is approximately constant, with the c-axis aligned along the growth direction [61]. The alumina grain size perpendicular to the growth direction is therefore determined by the nodule spacing. As the nodules coalesce to form a planar growth front the growth rate begins to drop monotonically with time and (equivalently) composite thickness [46,62]. Occasionally, at high temperatures or with small samples, growth acceleration has been observed and correlated, through differential thermal analysis, with adiabatic heating due to the exothermic nature of the oxidation reaction [46]. Frequently, oscillations in the growth rate are superimposed upon the average, decreasing rate [47,63] and have been associated with the passivation of old nodules and nucleation of new growth centers. Microstructural variations that accompany growth are of two kinds. Firstly, the scale of the microstructure is finer near the oxidation front and coarser at the original initiating surface [64,65], corresponding to the nucleation of new nodules during the growth process. Secondly, oscillations in the growth rate are often reflected in microstructural banding in which the residual metal content of the layers differs slightly. At higher growth temperatures, above 1300°C for A1-3 Mg, growth may cease abruptly after 1-2 mm of composite has formed. The incubation period is terminated by re-penetration of molten alloy through the passivating spinel layer and the nucleation of spinel nodules. Selective oxidation of the volatile magnesium leads to depletion of the solute at the surface in order to reach the three-phase equilibrium: (Al, Mg) alumina spinel. Alumina then becomes the preferred oxide product. The top (1-2pm) of the surface at the growth front has a complex chemistry whose details are still a matter for debate. In general terms, the surface consists of a layer of MgO separated from the growing composite by a thin wetting layer of molten alloy. This model borrows schematically (Fig. 4) from the microstructure found in liquid-solid reactive diffusion couples, such as Al-MgO or A1-Si02, in which the alloy wets its own oxidation product and penetrates the solid-solid interface, which in the present case is A1203-Mg0. Such microstructural observations were the basis for the steady state growth hypothesis [45,48,59],which assumed anionic oxygen transport through the MgO layer, controlled by either electronic or cationic vacancy defects, followed by dissolution of MgO into the molten alloy, diffusion of dissolved oxygen through the thin intervening alloy layer and, finally, an oxidation reaction to form alumina at the interface with the preformed, columnar alumina crystals. As indicated in

+

+

302


Figure 4. Schematic of surface microstructure during DMO growth. A thin alloy film is formed between a surface MgO layer and the growing alumina skeleton and is continuous with underlying alloy channels. The reactions at A, B, and C are described in the text.

Fig. 4, the following reactions have been proposed. $02(gas'-+ 3 0 z + 3 V k g + 6 h )

(1)

+ 6 h' + 3 0; -+3 0 (liquid) 3 0 (liquid) + 2 A1 (liquid) +. A1203

(2)

3 Vh,

(3)

where Vhg indicates a doubly charged cation vacancy, h is a positive hole and 0; refers to an oxygen ion occupying its own site. The above mechanism was consistent with several experimental observations. MgO was known to be preferentially formed on the surface of Al-Mg alloys under many circumstances, as described earlier, either because of cation demixing in spinel, to form periclase, or due to rapid diffusion of Mg in the solid state and subsequent surface segregation. The 'single crystal' oxidation product clearly precluded any mechanism that involved repeated nucleation of corundum (alumina) grains, which in any event is known to be difficult at temperatures below 1200°C. Finally, the presence of a film of molten alloy at a distance of about 1 pm from the surface accounted for the relative absence of a growth rate dependence on composite thickness (a rather slow decrease is actually observed). The steady-state microstructure described above is a simplification. The MgO layer is far from uniform, both with respect to position and time [50,56,66], and thicknesses: <0.1 pm has occasionally been observed [66]. In some samples the interface between aluminum and MgO was found to contain spinel [47], while the solidified metallic phase frequently contained nanocrystals of MgO [66]. Even less consistent with the steady state hypothesis was the apparent 'equilibrium' between aluminum and MgO in the presence of a magnesium concentration that was necessarily, and experimentally, close to the three-phase equilibrium with spinel and alumina [50]. Cation de-mixing of spinel, as a source of MgO (periclase), seems unlikely when spinel is the minor phase or apparently absent. An alternative hypothesis assumes that oxidation of magnesium in the vapor phase is responsible for the presence of MgO, an assumption that is supported by the demonstration of


303

DMO composite growth from aluminum alloys that contained no Mg but other, alternative volatile elements (Na, Li, Zn). In each case DMO growth occurs at temperatures for which the vapor pressure of the volatile solute is an appreciable fraction of the partial pressure of oxygen in the gas phase. The mechanism of oxygen transfer across the surface oxide layer is also problematic. Observed growth rates require that diffusion occur through the surface oxide at rates far greater than those measured for the fastest diffusing species (cation vacancies) in polycrystalline MgO [66].In addition, the decrease in growth rate with increasing thickness implies that viscous flow of liquid through the capillary channels of the composite might also be rate determining. The results of interrupted experiments [63] suggest that microstructural banding might be related to the formation of alumina by alternate dissolution of both spinel and magnesia into the molten alloy. The volume fraction of the alumina phase in these experiments also corresponded approximately to that expected from a comparison of the molar volume of oxygen in the alumina-spinel and alumina-magnesia phase pairs. This hypothesis also predicts correctly the volume fraction of spinel formed from MgO during the spinel growth peak in the incubation stage, as well as that of alumina formed from a silica nucleant. Alternating bands of magnesia and spinel at the growth interface with the molten alloy could be associated with periodic fluctuations in the magnesium concentration of the melt. Calculation of the partial pressure of oxygen (PO,) in equilibrium with different alloy-oxide combinations suggests that a low magnesium content in the melt maximizes the driving force for the formation of alumina by dissolution of MgO, while the resulting enrichment of magnesium then favors epitaxial spinel formation on the MgO crystals [63].Subsequent growth of alumina then occurs by dissolution of spinel, leading to a decrease in the magnesium concentration by diffusion into the bulk, and the cycle is then repeated. Isolation of molten metal from the MgO decreases the rate of magnesium oxidation and hence the rate of weight gain during this period of spinel dissolution. Subsequent re-penetration of spinel by the molten alloy once more initiates direct dissolution of MgO. The repeated formation of MgO required by the above sequence is most likely due to the oxidation of magnesium in the vapor phase, although cation de-mixing may also be possible if the spinel layer is sufficiently thick. In summary, composite formation by DMO of Al-Mg alloys is considered to be made possible by the following sequence of events. 0) A buffer layer of spinel at the original melt surface ensures a supply of molten alloy, containing =0.2-0.5% of magnesium, to the free surface, close to the three-phase equilibrium composition of the alloy with spinel and alumina. (ii) The high vapor pressure of magnesium leads to the formation of MgO at the surface of the composite. (iii) A displacement reaction with aluminum leads to the formation of alumina, not by direct nucleation but rather by transport of dissolved oxygen to the AlA1203 interface. This reaction is periodically mediated by spinel, due to local variations in the magnesium activity that are not immediately compensated by bulk diffusion from the melt reservoir.

304


The high-temperature limit for oxidative growth has been attributed to excessive loss of magnesium vaporizing from the surface [50,62,65]. Unless the magnesium can be replenished by diffusion through the liquid, a passivating layer of alumina or spinel will form. In principle, providing the bulk reservoir contains sufficient magnesium, oxidative growth is expected to resume eventually, the times involved being comparable to the incubation period. Related experiments [65] have shown that sufficient magnesium may be available, either in the alloy channels of a previously grown composite or in the spinel layer at the interface with the melt, to continue DMO growth when placed in contact with a pure aluminum melt. This demonstrates the very low levels of magnesium actually necessary for composite formation, but it is likely that, in the absence of a magnesium reservoir, growth will eventually cease due to magnesium losses by spinel precipitation or magnesium evaporation.

1.5.3 Directed Metal Oxidation Growth from other Aluminum Alloys Several other aluminum alloy systems which have added to an understanding of the DMO growth process are based on additions of magnesium, silicon, zinc, lithium, and sodium. Magnesium need not be an actual alloy constituent but may be added by doping the surface of a pure aluminum melt with MgO powder [50]. This combination eliminated the incubation period and permitted composite growth to a thickness determined by the amount of the original dopant. The elemental magnesium released by reduction of MgO remained in the vicinity of the surface alloy channels long enough to allow oxidative growth to occur well beyond the thickness of the oxide dopant layer. The MgO particles were therefore assumed to migrate with the oxidation front. A suitable magnesium dopant concentration can also be generated by immersing spinel in the alloy reservoir [56] or by doping the preform with MgO-Mg [67]. Additional binary aluminum alloys which have been oxidized by the DMO process include Al-Zn [57], aluminum doped with NaOH [58], aluminum doped with sodium from p-alumina [68], A1-Li and Al-Na [66], and aluminum doped with lithium derived from lithium carbonate [69]. In all cases the growth of the composite was promoted by the presence of an oxide which contained a potentially volatile cation (ZnO, LiAlsOs, NaA102 etc.) reducable to aluminum through a displacement reaction. In the case of sodium, the cessation of growth after only limited composite formation was attributed to evaporative loss of the solute. A similar observation with zinc as the dopant was probably due to localized surface depletion, reducing the activity of zinc below that needed to retain a surface layer of ZnO. Al-Li alloys seem to be capable of sustained composite growth but have not been studied extensively, presumably due to the cost and difficulty of lithium doping. Al-Zn alloys with minor amounts of magnesium [57] invariably lead to more sustained growth compared to the binary alloys, and it has been argued that trace levels of magnesium ensure that, in the event of temporary zinc depletion, the surface remains covered with MgO-MgA1204 rather

1 .SDirected Metal Oxidation Growth

305

than alumina. During growth in this system, however, the principal surface oxide appears to be ZnO [56,57], although it should be noted that the two cations magnesium and zinc readily substitute for each other in the periclase lattice. When magnesium-rich alloys are used, the surface oxide has been found to revert to MgO, as found in binary Al-Mg [65]. Additions of silicon improve the uniformity of growth in Al-Mg [48] and Al-Zn-Mg alloys [56], even though binary Al-Si alloys cannot initiate composite formation. For example, binary Al-Mg alloys may nucleate only a few nodules per square centimeter, which grow irregularly outwards from the isolated nucleation sites and never develop a planar growth front. Silicon additions greatly increase the surface density of nodules and lead to a very much more uniform and smooth growth front. Improved wetting by reduction of the alloy-oxide interfacial energy has been suggested as the primary reason for more regular growth, although prolonged incubation and microstructural banding, due to periodic magnesium depletion, are also observed in Al-Mg-Si alloys. It has already been suggested that oscillations in the growth rate are probably linked to the periodic surface depletion of magnesium. The observation that DMO composites grow initially with a uniform microstructure and a smoothly varying growth rate, before the onset of the growth instability [62,63], is consistent with the hypothesis that the instability occurs when the distance from the growth surface to the magnesium reservoir in the alloy melt is sufficient for solute diffusion through the liquid channels to become rate limiting. Furthermore, uniform microstructures are commonly found in ternary alloys which contain additions such as silicon, palladium, or indium [70], and for these solutes thermodynamic calculations have shown that the stability of alumina is extended to higher magnesium concentrations. This improved alumina stability should make spinel formation less likely and so ensure a single alumina growth mechanism via MgO dissolution into molten aluminum. However, an important feature of oscillatory growth that still has no satisfactory explanation is the lateral uniformity of alloy composition across the growth surface that should be necessary to achieve coordinated cyclic behavior over the whole of the growth front.

1.5.4 Microstructural Scale The volume fractions of the alumina and alloy phases present in a DMO composite may be related to the relative molar volumes of oxygen in alumina, spinel, and periclase, but the factors that determine the spacing of the metal alloy channels and the ceramic ‘grain’ size (the nodule spacing) are much less apparent. Additions of transition metals (Ni, Cu, Ti, or Zr) and some other elements, such as zinc, refine the microstructure of composites grown from an Al-Mg-Si alloy [71-731. Isothermal aging of composite samples coarsens the scale of the microstructure, as would be expected for a classical precipitation or a spinodal system. The coarsening mechanism has been attributed to capillarity-driven dissolution and reprecipitation of alumina, with the liquid alloy providing the medium for oxygen transport [64]. A similar phenomenon could also account for some coarsening of the microstructure created early in the growth process, when compared to the microstructure formed in

306


the later stages. Nevertheless, there is still no clear explanation relating the microstructural length scale to the alloy composition and the DMO growth temperature. It is possible to produce composites with an alumina content in excess of 85 vol-% at high temperatures. However, this appears to be achieved either at the expense of a reduced growth rate or with premature termination of the growth process before complete conversion of the alloy to composite [60,74]. Excessive evaporative loss of magnesium and sealing of alloy channels by alumina are the probable combined causes of this high temperature behavior, which prevails above about 1300°C (typically 1300°C for Al-Cu-Mg and 1200°C for Al-Ni-Mg [74], 1300°C for Al3 Mg [62], and 1400°C for Al-Zn-Mg [65] and Al-Si-Mg [46]). A further complication in complex alloys is the enrichment of nonoxidizing elements, such as silicon, nickel, or copper, which fail to diffuse away from the interface and instead precipitate refractory intermetallic compounds [74]. A controversial area in the growth of DMO A1203-A1 are the temperature and oxygen partial pressure dependencies of the oxidation rate. One difficulty in reconciling experimental results obtained by different authors and assigning an unambiguous growth mechanism has been the absence of a clear regime of constant growth rate [65]. The available evidence suggests that the rate of weight gain starts to fall after little more than 1-3 mm of composite growth. This decrease in growth rate approximates a parabolic time dependence in some instances [46,75] and has been attributed variously to enrichment of nonoxidizing elements near the surface, depletion of magnesium in the melt, or percolation restrictions on the supply of metal through the capillary channels [62,71]. Absence of data on melt viscosity, alloy channel tortuosity, alloy-alumina interfacial energies, and melt-oxide contact angles precludes a meaningful analysis of viscous flow of the melt through the percolation channels and its possible control of the growth rate. Some thermogravimetric studies have been conducted over a sufficient range of temperature and time to permit the determination of average growth rates and an analysis of activation energies which are valid for a limited range of temperature and composite thickness. Variations in alumina content across the composite and uneven growth across the surface make it difficult to define a linear interface mobility from gravimetric data. Measured activation energies range from 90 kJ mol-' for a complex Al-Si-Zn-Mg-Cu-Fe alloy [45], to 213 kJ mol-' for Al-1 Mg [76], 360 kJ mol-' for A1-5 Mg [77], and 400 kJ mol-' for Al-Mg-Si [71], with other values typically around 270-300 kJ mol-' [46]. Attempts to relate these values to specific mechanisms, such as ionic diffusion in MgO [59], have largely been abandoned because of the wide energy spread and, in some cases, the absence of well-defined Arrhenius behavior [50].Similar inconsistencies have been observed in the oxygen partial pressure dependence of DMO growth. Power law dependencies of 1/4 and 1/6 for the oxygen partial pressure were reported for Al-Mg-Si [71], but growth from Al-Si-Zn-Mg [45] was found to be insensitive to PO2. In contrast, an exponent of unity was proposed for Al5Mg with 2 and 5% Si, while growth from an A1-5Mg-10% alloy was again found to be insensitive to PO2 [78]. It would appear that the growth of the composite is inherently cyclic in the short term, involving the formation and dissolution of oxides such as MgO and ZnO, and


307

is self-limiting in the long term, due to surface depletion of the volatile element, enrichment of the melt in the channels by nonoxidizing elements, or to an insufficient flux of molten alloy through the capillary channels.

1.5.5 Growth into Particulate Preforms Commercial applications of melt oxidation generally require infiltration of a DMO Al2O3-A1 composite into a shaped preform, either a particulate aggregate or a fiber lay-up. The role of a particulate aggregate in modifying the DMO process is complex. The aggregate poses a physical hindrance to the growth, forcing changes in the growth direction and restricting the reaction between oxygen and the vaporizing solute. When the particle size decreases to approach the alloy channel width the growth constraints become pronounced, since the natural microstructural scale characteristic of DMO growth in free space can no longer be maintained. Chemical effects are also unavoidable, since the interfacial energy and contact angle between the molten alloy and a preform particle replace those with the growing alumina body to dominate the wetting by the melt of the percolation channels. In addition, the depassivating MgO/ZnO phase may now form on the preform particles, while heterogeneous nucleation of new alumina grains becomes possible. More substantial effects are to be expected if the particles react with the liquid alloy (partial dissolution in the melt or the formation of additional phases). Of the two most commonly used preform materials, one, alumina, is chemically inert during DMO while the other, silicon carbide, is reactive. We first examine the growth of a DMO composite into an alumina particulate, followed by that into S i c preforms, and finally the DMO infiltration of a suitably coated fibrous preform. Coatings of calcium sulfate with calcium silicate may be used to ensure that oxidative growth ceases at the external boundary of the preform, thereby ensuring near-net-shape fabrication. The mechanism by which these barriers ‘poison’ the oxidation without impeding the flow of oxygen is not clear at this time. Oxidative growth of Al-Mg-Si into alumina preforms was first reported by Nagelberg et al. [71,72]. It was subsequently observed that the crystallographic texture of the matrix alumina was reduced in the infiltrated preform when compared to that found after growth in free space [72]. In addition, wetting of the alumina preform by the alloy was found several particle diameters ahead of the fully infiltrated thickness [71]. The presence of an MgO-rich surface together with the underlying alloy film appeared similar to that found during growth into free space, and it was concluded that the growth mechanism remained substantially unchanged. However, in high partial pressures of oxygen, the growth rate during infiltration (normalized by the true surface area) was 2-3 times higher than that into free space, but also dropped significantly as the composite thickened. This behavior was attributed both to the wetting of alumina preform particles ahead of the growth interface, and to the difficulty of transporting unoxidized silicon by diffusion from the growth surface back into the melt through the tortuous alloy channels. At low partial pressures of oxygen the growth rate was found to increase as the composite growth interface approached the external surface of the preform.

308


Here, it was argued that the oxygen supply through the pores of the alumina preform was the rate-limiting step, and that growth became progressively easier as the alloy percolation channels approached the free surface. The oxidation rate was also found to decrease with increase in the silicon content of the melt. In the above work, the presence of alumina preform particles appeared to favor infiltration but no residual metal was found between the preform alumina particles and the matrix alumina and, surprisingly, neither was any epitaxy identified [72]. If the particles aid infiltration, then the rate of infiltration would be expected to increase with the particle surface area, and this was indeed the case for coarse alumina particles DMO infiltrated using Al-Zn, Al-Zn-Mg, and Al-Mg alloys [54,75,79,80]. A maximum in the growth rate was observed for particle sizes of about 20-50pm [57] and 5-8pm [80]. The decrease in growth rate at smaller particle sizes was attributed either to the difficulty of developing the surface oxide layer (MgO/ZnO) or to a diminished supply of liquid metal owing to the restriction in viscous flow due to the reduced size of the pore channels. Interestingly, in contrast to earlier work on Al-Mg based alloys, composites grown from Al-Zn-Mg alloys, in which copious nucleation of ZnO was observed on alumina particles ahead of the main infiltration front, also exhibited rapid wetting of the Zn0-A1203 interface, as well as epitaxy between the matrix alumina and the particulate alumina preform. In one paper the microstructure of the preforminfiltrated matrix has been reported to be refined when compared to that obtained in the absence of an alumina preform, but the magnitude of the effect was not specified [72]. The importance of wetting of the alumina preform by the liquid alloy and of the preform-ZnO interface was demonstrated by the absence of infiltration of Al-Zn-Mg into Zr02 [57]. Quantitative assessment of the role of pore size and particle packing fraction (the ‘green’ density of the preform) in the growth process remains elusive, largely owing to the absence of steady-state growth over a sufficiently wide range of preform parameters. Watari et al. [75] have reported a growth rate that was proportional to the square root of the pore size. However, the rate of infiltration into the alumina preform observed in this study was nonlinear and the data only appear to obey the given pore size relationship for composites of 1-2 mm in thickness and at a single temperature. Silicon carbide has been a popular choice of particulate for preforms in the DMO growth of A1203-A1 composites, promising both hardness and chemical stability in an oxidizing atmosphere. A distinctive feature of DMO growth into S i c preforms is the reactivity of Sic with both the alloy melt and with oxygen. Sufficient silicon must be present in the alloy to prevent formation of aluminum carbide during growth, since aluminum carbide is hygroscopic and leads to timedependent disintegration of the composite under ambient conditions [8 11. Superficial oxidation of S i c leads to reactive infiltration of liquid aluminum into the silica-coated Sic. Initial published work on the A1203-SiC-A1 formulations showed that growth rates increased with decreasing particle size 1821, and attributed the increase to wetting of the silica layer by the melt. It was subsequently pointed out [83] that oxidation of Sic, either during pre-sintering of the preform or during infiltration, would be expected to increase the amount of silica available

1.5 Directed Metul Oxidation Growth

309

with decreasing Sic particle size in the preform. In effect, the volume fraction of Sic in the composite is reduced and the silica product acts as an additional source of oxygen that promotes reactive infiltration. This effect was found to increase the residual alloy content of the DMO composite up to 30 vol-%, using a 5pm particle size S i c particulate preform, as compared with 10-15 vol-% of residual alloy with 50-300 pm S i c particles. The reactive infiltration of a porous silica preform to yield a metal-rich A1203-A1 DMO composite appears to be similar [31]. In this case, the individual silica particles are completely reacted and replaced by Al2O3-Al, while the pore spaces are filled with the resulting silicon-rich alloy. Similarly, it was shown that additions of 1-20% Si02 and Sn02 to Sic preforms could increase the final alloy content of the resulting composite [84]. In summary, optimizing the S i c preform particle size can promote the production of DMO composites with a higher alloy content. However, this is often at the expense of significant residual porosity in the final product, particularly at higher temperatures (> 1000°C). Residual porosity may be avoided by coating the Sic particles with reactive alumina (by the precipitation of aluminum hydroxide followed by calcination). Composites made from preforms with these alumina-coated S i c particles contain less porosity and residual metal, and have higher hardness [85]. When excessive reactive wetting of the preform leads to metal migration over particle surfaces ahead of the main reaction front, residual porosity becomes a ubiquitous feature of DMO infiltration. The surface of a partially-infiltrated region then becomes sealed before the underlying pores are filled with the advancing composite. In the case of oxidation in air, a similar phenomenon can lead to the entrapment of nitrogen-containing pores. The crystallinity of alumina formed by DMO infiltration into S i c particulate preforms follows the expected behavior of heterogeneous nucleation on S i c particles and large monocrystalline regions with [00011 parallel to the growth direction within the larger voids [83,86]. Manor et al. [82] found that the infiltration rate increased as the Sic particle size decreased. However, no systematic trend was found in another work [83], most probably due to widely different silicon concentrations in the percolation channels resulting from reduction of the silica layer during infiltration.

1.5.6 Growth into Fibrous Preforms Silicon carbide fiber preforms have been successfully DMO infiltrated after being coated in order to prevent attack by the melt and produce weak interfaces with the DMO matrix in order to maximize pull-out during fracture. To achieve the desired properties, a double layer coating was used in which the outer layer of S i c provided an oxidation resistant surface on which the alumina could nucleate, while the inner coating, probably BN, ensured a low shear strength for the interface between the composite matrix and the reinforcing fiber [87]. Alumina fibers have also been successfully infiltrated [88,89]. Double layers of BNjSiC and triple layers of Zr02/A1203/Si02have been used to control the reactivity and shear

310

I Directed Metal Oxidution

strength of the fiber interface with the matrix. In the latter coating system, the role of alumina is to reduce the thermal mismatch with the fiber, while the middle zirconia layer both promotes fiber pull-out and protects the fiber against attack by the melt. The outer silica layer promotes wetting by the molten metal. The alloy used in the last example was A1-5 Mg, but a quaternary alloy, with silicon and zinc additions and less magnesium, would probably have minimized degradation of the fiber and reduced the incubation time, possibly obviating the need for the silica coating. An attempt to infiltrate aluminosilicate (mullite-based) fibers resulted in extensive attack associated with the reduction of silicon from the oxide and its dissolution in the melt [90].

1.6 Mechanical Properties Three distinctive features of the microstructure of DMO composites determine their mechanical properties: the particulate or fiber loading in the preform, the presence of a continuous ductile (and low-melting point) alloy phase, and the geometry of the anisotropic ceramic skeleton. Not surprisingly, the transverse rupture strength (TRS or MOR = modulus of rupture), which is sensitive to the initial flaw distribution, varies widely with the dimensions of the reinforcement and the residual porosity. The tribological properties, such as the resistance to adhesive wear, are more forgiving. The thermal shock resistance can also vary widely, since this depends on the thermal expansion coefficient, the thermal conductivity and the toughness, all of which can be tailored by controlling the residual metal content, the nature and volume fraction of the particulate preform, and the porosity. High temperature creep of DMO composites is to some extent dependent on the content of the residual alloy, which melts above about 600°C. A composite with most of the alloy phase replaced by pores is creep resistant, similar to the behavior of chemical vapor infiltrated (CVI) Sic-Sic composites, in which 1@15% of fine, uniformly distributed porosity can be tolerated (at least, in the absence of a low-viscosity grain-boundary phase). Published data on the mechanical behavior of DMO composites is somewhat limited. The following section summarizes the different mechanical properties and, where possible, compares the properties of the base composite with those of a particulate-reinforced material. The fiber reinforced composites are considered separately.

1.6.1 Elastic Modulus The stiffness of a DMO composite of A1203-Al with 22% alloy and 4% porosity (231 GPa) was modeled successfully [60] by assuming the metal and the ceramic skeleton to deform equally in series and in parallel, that is, by taking a ReussVoigt average. The additional effect of isolated pores could be included by using empirical expressions derived from data on porous alumina. The elastic modulus

1.6 Mechanical Properties

3 11

of a composite fabricated at a high temperature, yielding a higher ceramic content with significant amounts of isolated alloy pockets (13% alloy, 5% porosity), was also successfully modeled by using Eshelby's treatment for elastic spheres dispersed in an elastic matrix. This latter composite had a Young's modulus of 304 GPa, illustrating the large increase in stiffness possible through a reduction in volume fraction and connectivity of the metal phase. In both the above DMO materials the modulus was measured by standing wave resonance, with the test bar aligned perpendicular to the growth direction, approximating to a [uwtO]-alumina orientation. The modeling of the results assumed isotropic, bulk polycrystalline behavior for all the phases present, ignoring textural anisotropy. Since elastic anisotropy of the elastic constants for alumina is not very great, these assumptions are reasonable. The major difficulty in measuring elastic moduli parallel to the growth direction stems from the substantial infiltration times required to produce the necessary length for the through-thickness resonating bar. Pickard et al. [91] measured a modulus of 220 GPa for a DMO composite grown from A1-3 Mg that contained 20% residual alloy and 10% porosity. They also reported values up to 260 GPa for composites grown from complex Al-Si-Mg-Fe-Cu alloys, and, since there was no observable change in the proportions of alloy and ceramic, the increase in modulus probably reflects the presence of a significant volume fraction of elastically stiff, intermetallic phases. The above values of the Young's modulus lead to values of the specific modulus (modulus per unit weight) of the DMO composite in the range 65-84MPacm3 g-', which may be compared with 100 (polycrystalline alumina), 140 (dense Sic), and 25 (metals, silica). DMO composites with particulate reinforcements of alumina and silicon carbide were found to possess moduli of 302 and 324 GPa, respectively [92]. In the former case, the total alumina content, including the reinforcement, was 8 1YOwhile in the latter, the S i c volume fraction was 46% and the matrix alumina 37%. The alloy content and porosity were in the vicinity of 15 and 2 vol-YO,respectively. A similar value of 313 GPa was reported for a S i c reinforced DMO composite whose composition was not disclosed [93].

1.6.2 Strength and Toughness Most of the available data on strength and toughness have been reported for transverse rupture strengths measured in three- or four-point bending, both at room and elevated temperature. It is difficult to interpret strength data without a prior understanding of 'toughness'. Accordingly, we first examine the limited measurements of the fracture toughness of these materials. Double torsion tests of DMO A1203-Al grown from A1-3Mg have shown that K I C drops from a maximum of 9.5 MPam'/2 (74% alumina, 22% alloy) to a low of 3 MPam'/2 (73% alumina, 1% alloy) [60], with an intermediate value of 6MParn'I2 for a composite with 72% alumina and 13% alloy. Thus, the toughness, as might be expected, appears to increase monotonically with alloy content. Flinn et al. [94], working with Lanxide DMO composites, determined the contribution to the fracture energy from the plasticity of ductile alloy ligaments. Their result, derived from the measured yield

312


strength of the alloy and fractographic analysis of the composite, gave an estimated fracture energy contribution of 150-200 J mP2.This value corresponds to a toughness increment of about 7MPam'/2 (assuming a Young's modulus of 250GPa) over that for polycrystalline alumina, which is consistent with the maximum value reported by Aghajanian and coworkers. The alloy in this case, though not specified, was reported to contain Cu, Fe, and Si, implying a higher fraction of intermetallics than in the Al-Mg alloy studied by Aghajanian and coworkers. Pickard et al. [91] found R-curve behavior in four-point bending of A1203-Al composite samples made from Al-Mg and Al-M Si Zn, with an initial toughness of 3 MPam'I2 rising to a plateau of 5 MPa m 12. However, a complex commercial alloy (A-380) yielded DMO composites which contained intermetallic phases and also gave a constant toughness of 3MPam'I2. The toughness appears to be increased by additions of Sic [91]; in particular, the increment in the plateau of the crack resistance curve increased with S i c particle size. It was suggested that particle bridging and frictional sliding at the interface could contribute significantly to energy dissipation. A model developed to explain the increase in KIc accounted satisfactorily for the toughness of composites containing 67 and 165 pm Sic particles, but underestimated the value for 35 pm Sic. It was concluded that particulate contributions to toughening exceeded those due to plastic dissipation, at least for the composites derived by DMO from this alloy (A-380), which contained a high fraction of brittle intermetallics and displayed a flat R-curve. In situ measurements of crack opening displacement in a A1203-SiC-A1 DMO composite [95] confirmed that the intrinsic (crack tip) toughness was about 2 MPa similar to many other alumina matrix composites) and that the observed increases of up to 7.5 MPa ml/' could be attributed entirely to crack wake effects, including frictional sliding, S i c particle bridging and plastic deformation of alloy bridges, rather than to changes in the crack-tip profile. Anderson [96] measured KI in four-point bending for both a base composite and a Sic-reinforced composite and obtained 7.8 and 5.6 MPa m1/2,respectively. The decrease observed after growth into the particulate preform is inconsistent with the results of Pickard et al. but it is unlikely that the alloys used for the two composites were the same. In additional results [97] for the fracture toughness, KIc, determined from a single edge-notched beam test, values of 3.6MParn'l2 (for a 100pm Sic particulate preform) up to 6.2MPam'12 (13 pm Sic) were reported. This trend is the opposite of that reported by Pickard et al. [91]. Values of the elastic modulus reported in this later work ranged from 261 to 328 GPa. Most mechanical properties data report only rupture strength, and, since such data are sensitive to the size and distribution of preexisting bulk defects and surface finish, they are not easily interpreted without some help from microstructural analysis. Flexural strengths in the range 250-300MPa were reported for two alumina-aluminum composites, the higher values for composites that contained more metal and less porosity, and which also had a higher fracture toughness [60]. Values around 150-300 MPa have been published for alumina and silicon carbide-reinforced compositions, but the particle sizes were not disclosed [92]. Pickard et al. [91] studied Sic reinforced materials and found that the finest (35 pm) Sic particle size gave the highest strength (400 MPa) but the minimum

P-


3 13

toughness. The coarsest preform particles, 165 pm Sic, yielded strengths only of the order of 150 MPa. In this work microcracking at the Sic particles due to thermal expansion mismatch with the alumina matrix was held responsible for the strength degradation. Raman peak shift measurements have shown [98] that residual stresses in the silicon carbide particulate preform are a minimum for large particles, so it is possible that unfilled pore space in the preform (whose dimensions should scale with the particle size) could be responsible for the strength loss.

1.6.3 Thermal Shock The residual strength behavior of Al2O3-A1 DMO composites after quenching from elevated temperatures is similar to that of other ceramic systems, with an abrupt and monotonically increasing loss of strength above a critical temperature drop, A T , which increases in the range 200-300°C as the volume fraction of residual metal increases [60]. In contrast, a composite in which most of the metal had been replaced by open porosity, showed a low, monotonically decreasing strength with increasing A T ,but with no abrupt transition above a critical A T . Such behavior is typical of porous ceramics. Experiments on Sic reinforced DMO composites [97] showed similar effects, with the thermal shock strength degradation increasing as the particle size decreased, so much so that, although fine particulate material was initially stronger, the coarser microstructures retained a higher residual strength after thermal shock. These results were attributed to the beneficial role of an array of preexisting flaws at large particles, which could extend and prevent the growth of a single large crack. The critical value of A T was found to be independent of particle size in the range 13-100 pm.

1.6.4 High Temperature Strength A Sic particulate reinforced (48%) alumina-aluminum DMO composite [99] has been found to give lower creep rates at 1200-1300°C, when compared to silicon carbide whisker-reinforced (10 and 50%) alumina. Creep cavitation damage was attributed to alumina grain-boundary sliding, and stress exponents for the stress dependence of the minimum creep rate were between 2 and 2.6. Molten metal pockets were associated with cavitation at higher temperatures, but could also blunt local stress concentrations, similar to the effect of viscous, glassy grainboundary phases at triple junctions in a conventional ceramic. Clearly, the presence of the coarse-grained alumina skeleton in the DMO composite microstructure more than compensates for any lack of load bearing capacity associated with the 10-15% of liquid alloy which is present. Hot hardness values [92] and elevated temperature transverse rupture strengths [60] show the expected strength reduction in the vicinity of the melting point of the alloy phase (500-700°C). Nevertheless, significant high temperature hardness remains ( 3 4 GPa at 900"C), at least in the particulatereinforced DMO composites.

1.6.5 Wear Properties In terms of their potential engineering applications, the wear performance of DMO composites is probably the most important class of mechanical behavior. Product brochures published by the manufacturers have highlighted the superior performance of silicon carbide reinforced composites when compared to monolithic alumina, cemented carbides, cast irons or stellites, particularly in abrasive slurry wear tests [loo]. Sliding block tests and slurry pump tests [93] have also demonstrated lower wear rates for silicon carbide reinforced DMO composites in comparison with ceramics such as silicon nitride-bonded silicon carbide and alumina. In block-onring tests [loll it was found that the total wear of a A1203-SiC-AI DMO composite was comparable to that recorded for Si3N4,S i c and SiAlON at high loads and tested against steel, and superior to both alumina and cast iron. In addition, the wear of the composite itself proved to be the lowest of all the materials studied, while the measured friction coefficients were similar to, or lower than, those determined for the other ceramics. Pin-on-disc wear tests of DMO composites against steel [102,103] established that low wear rates and friction coefficients under high speed dry sliding conditions were due to the formation of a lubricating tribofilm which was rich in magnetite and which reduced both the contact stresses (by spreading the load) and the friction coefficient. It was postulated [102-1041 that the aluminum-rich metal phase in the composite could be crucial in forming a lubricating surface layer by softening or melting at the high surface temperatures generated adiabatically. At higher speeds and applied loads, cyclically varying wear rates and friction were encountered, associated with the formation and removal of the tribofilm. While wear rates were comparable to (or lower than) those of silicon nitride and zirconia toughened alumina on steel, it was found that a higher metal content in a fine particulate DMO composite reduced the wear rate, despite the lower hardness and increased porosity [102]. A similar blunting role of aluminum has been suggested during sliding of composites against abrasive S i c grinding discs [ 1051.

1.6.6 Mechanical Properties of Fiber-reinforced DMO Composites Alumina-aluminum DMO compositions have been grown into both S i c and alumina fibers. The fibers have generally been coated, either to maximize fiber pull-out, or to protect the fiber from the melt and the atmosphere, or both. D M O composites uniaxially reinforced with 15% Nicalon fibers [lo61 have been reported to have mean bend strengths of 720 MPa with a fracture toughness of 27 MPam'/2 and a measured tensile strength of 460MPa at room temperature. Fareed et al. [107,108] have published two data sets for the flexural strength of D M O composites reinforced with 35% Nicalon cloth (satin weave), from which the residual aluminum had been leached. In one case [107], the flexural strength dropped from 480MPa at room temperature to 400 MPa at 1000°C and 350 MPa at 1200°C. This impressive retention of strength in the presence of 20% open porosity in the capillary channels was accompanied by a fracture toughness that increased from 21 MPa rn'/*at room


3 15

temperature to 23 MPam'12 at 1000°C, before dropping to 18 MPam112at 1200°C. In their second report [108], the flexural strength increased slightly, from 460 MPa at room temperature to 488 MPa at 1200°C before dropping to 400 MPa at 1300°Cand 340 MPa at 1400°C, while the fracture toughness dropped monotonically from 27.8 MPa m'12 at room temperature to 15.6 MPa m112at 1400°C. The fiber coatings used in this work were not disclosed, but other reports on processing of DMO fiber reinforced composites, described earlier, arc consistent with a duplex BN-SIC system, the BN to introduce a low shear strength interface and the S i c to protect the coating and fiber from oxidation. Cyclic fatigue tensile and shear tests at room temperature, which were accompanied by microstructural examination [ 1091, have revealed that the general behavior of Nicalon-reinforced DMO composites is similar to other fiber-reinforced brittle ceramic, glass, and glass-ceramic systems, such as Nicalon reinforced lithium aluminosilicate (LAS) glass-ceramic. Residual stresses, introduced during fabrication, commonly generate extensive microcracking in the matrix of these brittle materials, so that the matrix contributes little to the stiffness of the composite. In the case of the DMO fiber-reinforced composites, the fatigue behavior in uniaxial stress-strain tests was dominated by progressive matrix cracking (at a stress amplitude of about 70 MPa), which was followed by interface debonding and fiber sliding. The bend strength of notched samples remain high (70-80% of the unnotched values) due to the well-distributed damage and global load sharing. Tensile strengths under uniaxial loading for these biaxially reinforced composites were found to be about 250MPa [108,109]. Substantial fiber pullout was observed, with fiber fracture strengths, as deduced from the fracture mirrors, slightly reduced compared to that of the virgin fiber before infiltration [109]. Strength retention of 70-100% was observed at temperatures up to 1100°C after thermal cycling to 1200°C in an inert atmosphere [108]. Tests in air at a constant stress level, just above that required for matrix microcracking, gave a stress-rupture life of over 500h at 1100°C and over lOOOh at 1200°C. No creep was detected at 12OO0C, even though unreinforced Nicalon fibers are known to degrade and creep under such conditions. The apparent improvement in the high temperature properties of the fiber was ascribed to the absence of oxidative degradation due to the protection of the Al2O3-A1 matrix. Some limited strength measurements have been made on alumina fiber (Nextel 610)-reinforced DMO Al2O3-Al, using duplex BN-Sic fiber coatings [88]. The four-point load flexural strength was found to drop from 400 MPa at room temperature to 300 MPa at 700°C and 160 MPa at 1000°C. An exposure to air at 1000°C for 100 h decreased the strength by 20%. The drop in high temperature strength was associated with a change from transgranular to intergranular fracture of the fibers, possibly due to the precipitation of impurity phases at the grain boundaries. The toughness of this oxide fiber-reinforced DMO composite fell from 17 MPa m'I2 at room temperature to 9MPam'/2 at 1000°C. In general, the high temperature mechanical properties of the oxide fiber-reinforced composites appear to be inferior to those of the Nicalon (Sic)-reinforced material. Some other grades of fiber have also been used to reinforce DMO composites, including Sic-SCS-6, but data are very limited.

316


1.7 Corrosion of Directed Metal Oxidation Composites Published data on the resistance of DMO composites to oxidation, corrosion, and other environmental factors is scanty. Standard Sic particulate-reinforced compositions have been rated for use at room temperature under conditions ranging from slightly acidic to basic [ 1001. Concentrated hydrochloric acid is known to cause progressive dissolution of the residual aluminum alloy. Resistance to acid attack can be achieved [loll by active diffusion-infiltration of molten nickel to replace residual aluminum and silicon with nickel-rich intermetallics. The corrosion of Sic particulate-reinforced composites in chloride solutions [ 1 101 suggests that the alloy phase is more susceptible to pitting than monolithic A1-6061, and that corrosion is accompanied by hydrogen evolution under acidic as well as basic conditions. The possibility of using SIC-DMO composites in heat exchangers for the glass industry prompted a study of high temperature corrosion in the presence of sodium silicate at 1000-1300°C [l 111. The composite DMO material appears potentially superior to monolithic Sic and alumina, on the basis of corrosion resistance, in the case of Sic, and thermal shock resistance, in the case of alumina. Under static conditions, the principal mechanism of degradation was the continued oxidation of residual alloy in the capillary channels to form additional composite at the surface. By contrast, when a silicate was sprayed on to the surface under dynamic conditions at high temperature, the weight gain due to oxidation of the alloy was accompanied by dissolution of the protective silica film formed on the silicon carbide filler particles into the molten silicate (as observed in the hot corrosion of monolithic Sic). A proprietary alumina-rich surface layer appeared to reduce the attack. Similar hot corrosion tests with coal slags [ 1 121 demonstrated that a Sic particulate-reinforced DMO could possess a lifetime of 4000-1 5 000 h at 1260°C, as estimated from the time taken to degrade a 1 mm thick alumina-rich surface layer of the composite.

1.8 Other Properties Some limited measurements have been made of the electrical conductivity [113] and dielectric constant [114] of unreinforced DMO composites. These studies confirm the much higher degree of connectivity of the alloy phase along the infiltration direction as compared to the transverse directions. Anisotropy in the dielectric constant of acid-leached composites may be ascribed to the crystallographictexture of the alumina.

1.9 Applications The engineering advantages of DMO composite ceramics determine the possible areas of application, although actual penetration into these markets depends on

I .9 Applications

3 17

economic factors, in particular the cost of replacing existing or competing materials. It is helpful to list the primary engineering features of these materials and the advantages they convey. (i) Since no external pressure is applied and directed metal oxidation is a near-netshape process, very large components can be produced which approach theoretical density and are essentially free of residual stress. (ii) The hardness of the dense ceramic and the resistance to crack propagation imparted by the network of metal-filled capillary channels give these materials excellent wear resistance. (iii) The presence of the continuous metal phase enhances the thermal conductivity, allowing components to act as a heat sink and avoid excessive temperature gradients. (iv) The combination of high stiffness and high hardness, together with the relatively low specific gravity and reasonable production cost, make the DMO material an attractive candidate for ballistic protection.

No data are available on the sales of DMO products and the following assessment is based almost entirely on unverified information taken from the commercial literature and the manufacturer’s published reports.

1.9.1 Wear Resistant Components Sic particulate-reinforced DMO materials are being used for the manufacture of large protective chute liners in high-wear applications for the mining industry and are being evaluated for applications as wear parts for pumps, valves, nozzles, ducts, plates, seals, and piston engines [ 1 151. The standard DMO composites are suitable for neutral and basic operating conditions, while modified compositions, possibly with nickel infiltration replacing the residual aluminum alloy in the capillary channels, are available for use under acid conditions (pH as low as 1) [loo]. Results of tests in slurry pumps indicate that a Sic particulate-reinforced DMO composite is superior to traditional steels, cermets, and alumina in this application [93]. Extensive performance tests of Sic and A1,03 particulate-reinforced composites for valve seat inserts in piston engines have given promising results which were ascribed to the higher hardness and thermal conductivity of the DMO material when compared to metallic inserts [92]. An outer layer of a particulate-reinforced DMO composite has also been used to improve the wear resistance of a metal-matrix composite camshaft [116]. Other applications which have been explored include slide gates to control the flow of molten steel (replacing magnesia or magnesia/carbon composites), and tooling for the plastics injection molding industry (replacing various grades of tool steel) [117].

1.9.2 Ceramic Composite Armor The published information on ceramic composite armor is neither extensive nor very informative, although it is clear that incorporating a hard, stiff and light-weight

318


ceramic plate into a composite armor system can significantly reduce the weight of armor which is required to defeat a given threat. Two examples are the boron carbide armor used to protect the pilot’s seat in combat helicopters and the alumina plate inserts used to improve the personal protection provided by a polymer reinforced armored vest. The accepted criterion for judging the performance of an armor system is the V5,test, which is the minimum weight per unit area of armor required to defeat a given projectile under standard conditions (a 50% chance of penetrating the armor module at a specified projectile velocity V ) . Sic particulate-reinforced DMO tiles are known to have attractive ballistic properties against light and medium threats (up to and including medium caliber, high velocity rounds and fragments). The DMO materials are candidates for vehicular add-on armor [ 1171 and are reported to have been used to provide added protection to armored personnel carriers by the US Army in the Persian Gulf war against Iraq. There is no indication that DMO armor has ever been considered seriously as a candidate ballistic material to defeat the heavy threat faced by a main-line battle tank. In terms of cost, DMO armor should be competitive with steel because of the reduced weight [118]; however, it is by no means clear that the DMO product is superior to other grades of alumina-based ceramic armor tile, either in terms of cost or performance.

1.9.3 Thermal Barriers and Heat Sinks Porous aluminum titanate DMO has both a low thermal expansion coefficient (nearly zero at least up to 600°C) and a low thermal conductivity, suggesting possible applications in thermal barriers and abradable seals for turbine rotor shrouds [I 161 and automobile exhaust port liners and gas desulfurization nozzles [117]. Other possible thermal applications of silicon carbide reinforced DMO include heat exchangers [117] and flaps and seals in the afterburners of jet engines [119]. In addition, Sic particulate-reinforced DMO could serve as a lossy insert for high powered microwave tubes, withstanding temperatures up to 1000°C in ultrahigh vacuum. It has even been suggested that very large DMO composite cylinders might be suitable for applications in deep-sea submersibles [ 1201.

References 1. M. S. Newkirk, A. W. Urquhart, H. R. Zwicker, and E. Breval, J . Muter. Res. 1986,1,81-89. 2. M. S. Newkirk, H. D. Lesher, D. R. White, C. R. Kennedy, A. W. Urquhart, and T. D. Claar, Ceram. Eng. Sci. Proc. 1987, 8, 879-885. 3. E. T. Thews, Met. Ind. (Lond.) 1927, 31, 597. 4. K. J. Brondyke, J . Am. Ceram. SOC.1953, 36, 171-174. 5. J. G. Lindsay, W. T. Bakker, and E. W. Dewing, J . Am. Cerum. Soc. 1964, 47, 90-94. 6. A. W. Urquhart, Muter. Sci. Eng. A , 1991, A144, 75-82.

References

7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50.

3 19

A. F. Beck, M. A. Heine, E. J. Caule, and M. J. Pryor, Corros. Sci. 1967, 7, 1-22. R. K. Hart and J. K. Maurin, Surf. Sci. 1970, 20, 285-303. G. M. Scamans and E. P. Butler, Metall. Trans. A , 1975,6A, 2055-2063. W. Thiele, Aluminium, 1962, 38, 707-715 and 780-786. S. Balicki, Pruce. Inst. Hutnic. 1958, 10, 208-213. M. V. Mal’tsev, Izv. Akad. Nuuk. SSSR Ser. Fiz. (English Translation) 1956, 20, 747-750. C. Lea and J. Ball, Appl. Surf. Sci. 1984, 17, 344362. D. J. Field, G. M. Scamans, and E. P. Butler, Metall. Trans. A , 1987, 18A, 463472. G. D. Preston and L. L. Bircumshaw, Philos. Mug. 1935,20, 70G720. W. W. Smeltzer, J . Electrochem. SOC.1958, 105, 67-71. M. J. Dignam, J . Electrochem. SOC.1962, 109, 184191 and 192-198. J. M. Chen. T. S. Sun, R. K. Viswanadham, and J. A. S. Green, Metall. Trans. A , 1977,8A, 1935-1940. T. S. Sun, J. M. Chen, R. K. Viswanadham, and J. A. S. Green, Appl. Phys. Lett. 1977,31, 580-582. C. Lea and C. Molinari, J. Mater. Sci. 1984, 19, 23362352. T. Malis and M. C. Chaturvedi, J. Muter. Sci. 1982, 17, 1479-1486. G. M. Scamans, R. Alani, and P. R. Swann, Corros. Sci. 1976, 16,443459. I. Haginoya and T. Fukusaka, Trans. Jpn. Inst. Met. 1983,24, 613-619. W. Kahl and E. Fromm, Metall. Trans. B, 1985, 16B, 47-51. M. Drouzy and M. Richard, Fonderie, 1974, 29, 121-128. D. A. Weirauch Jr., J. Muter. Res. 1988, 3, 729-739. B. F. Devereaux, Trans. TMS-AIME,1968,242, 795. M. C. Breslin, J. Ringnalda, L. Xu, M. Fuller, J. Seeger, G. S. Daehn, T. Otani, and H. Fraser, Mater. Sci. Eng. 1995, A195 113. R. E. Loehman, K. Ewsuk, and A. P. Tomsia, J . Am. Ceram Soc. 1996,79,27-32. W. G. Fahrenholtz, K. G. Ewsuk, D. T. Ellerby, and R. E. Loehman, J. Am. Ceram. Soc. 1996, 79, 2497-2499. M. Hanabe and P. B. Aswath, Actn Mater. 1997, 45, 40674076. M. Hanabe and P. B. Aswath, J . Muter. Res. 1996, 11, 1562. W-P. Tai, T. Watari, and T. Torikai, Am. Cerum. Soc. Bull. 1997, 76, 8 6 8 9 . M. K. Aghajanian, M. A. Rocazella, J. T. Burke, and S. D. Keck, J . Muter. Sci. 1991,26,447454. D. K. Creber, S. D. Poste, M. K. Aghajanian, and T. D. Claar, Cerum. Eng. Sci. Proc. 1988,9, 975-982. M. K. Aghajanian, J. P. Biel, R. G. Smith, and C. R. Kennedy, J. Muter. Sci. Lett 1994, 13, 293-296. H. Scholz, R. Gunther, J. Roedel, and P. Greil, J . Muter. Sci. 1993, 28, 939-942. V. Jayaram, B. S. S. Daniel, N. Nagendra, and H. R. Muralidhar, Truns. Muter. Res. Soc. Jpn 1993, 14A, 621-624. M. K. Aghajanian, R. A. Langensiepen, M . A. Rocazella, J. T. Leighton, and C. A. Anderson, J. Muter. Sci. 1993, 28, 6683-6690. W. B. Johnson, Cerum. Eng. Sci. Proc. 1992, 13, 573-580. M. K. Aghajanian, J. P. Biel, and R. G. Smith, J . Am. Cerum. Soc. 1994,77, 1917-1920. T. D. Claar, W. B. Johnson, C. A. Anderson, and G. H. Schiroky, Cerarn. Eng. Sci. Proc. 1989, 10, 599-609. E. Breval and W. B. Johnson, J . Am. Ceram. Soc. 1992,75, 2139-2145. W. B. Johnson, A. S. Nagelberg, and E. Breval, J . Am. Ceram. Soc. 1991, 74, 2093-2101. A. S. Nagelberg, Solid State Ionics, 1988, 32/33, 783-788. K. C. Vlach, 0. Salas, H. Ni, V. Jayaram, C. G. Levi, and R. Mehrabian, J . Muter. Res. 1991, 6, 1982-1995. 0.Salas, H. Ni, V. Jayaram, K. C. Vlach, C. G. Levi, and R. Mehrabian, J. Muter. Res. 1991, 6, 19641981. A. S. Nagelberg, J . Muter. Res. 1992, 7, 265-268. S. Antolin, A. S. Nagelberg and D. K. Creber, J . Am. Cerum. SOC.1992, 75, 447-454. P. Xiao and B. Derby, J . A m . Cerum. SOC.1994,77, 1761-1770.

3 20


51. 0. Salas, V. Jayaram, K. C. Vlach, C. G. Levi, and R. Mehrablan, J . Am. Ceram. SOC.1995, 78, 609-62 I. 52. C. D. Qin, M. 0.Lai, B. Cotterell, and S. Shi, Muter. Res. SOC.Symp. Proc. 1995,365, 133-138. 53. H. Schmalzried and W. Laqua, Oxid. Metals, 1981, 15, 339-353. 54. K. C. Vlach, C. G. Levi, and R. Mehrabian, in Processing and Manufacturing of Advanced Materials for High Temperature Applications, V. A. Ravi and T. S. Srivatsan (Eds), The Minerals, Metals and Materials Society, Warrendalc, USA, 1992, pp. 157-171. 55. V. S. R. Murthy and B. S. Rao, J . Muter. Sci. 1995, 30, 3091-3097. 56. M. Sindel, N. A. Travitzky, and N. Claussen, J . Am. Ceram. SOC.1990, 73, 2615-2618. 57. M. Hanabe, V. Jayaram, and T. A. Bhaskaran, Acta Muter. 1996,44, 819-829. 58. P. Xiao and B. Derby, J. Am. Ceram. Soc. 1994,77, 1771-1776. 59. A. S. Nagelberg, S. Antolin, and A. W. Urquhart, J. Am. Ceram. Soc. 1992, 75, 455462. 60. M. K. Aghdjanian, N. H. Macmillan, C. R. Kennedy, S. J. Luszcz, and R. Roy, J . Muter. Sci. 1989,24, 658-670. 61. E. Breval, M. K. Aghajanian, and S. J. Luszcz, J . Am. Ceram. SOC.1990, 73, 2610-2614. 62. 0.Salas, A Fundamental Study of the Formation of A1,03/Al Composites from the Oxidation of Liquid A1 Alloys, PhD thesis, University of California at Santa Barbara, USA, 1992. 63. 0. Salas, V. Jayaram, K. C. Vlach, C. G. Levi, and R. Mehrabian, in Processing and Manufacturing of Advanced Materials for High Temperature Applications, V. A. Ravi and T. S. Srivatsan (Eds), The Minerals, Metals and Materials Society, Warrendale, USA, 1992, pp. 143-156. 64. P. Padmapriya, T. A. Abinandanan, and V. Jayaram, Scripta Muter. 1997,37, 1883-1888. 65. P. Xiao and B. Derby, J. Europ. Ceram. SOC.1997, 12, 185-195. 66. V. Jayaram, J . Muter. Sci. 1996,31,45914598. 67. F. J. A. H. Guillard, R. J. Hand, and W. E. Lee, Br. Cerum. Trans. 1994, 93, 129-136. 68. X. Gu and R. J. Hand, J . Muter. Sci. 1995,30,6005-6012. 69. X. Gu and R. J. Hand, J . Europ. Ceram. SOC.1996, 16, 929-935. 70. M. Pinkas, A. Venkert, and E. Manor, Scripta Muter. 1997,36, 377-383. 71. A. S. Nagelberg, Muter. Res. Soc. Symp. Proc. 1989, 155, 275-282. 72. E. Breval and A. S. Nagelberg, Mater. Res. Soc. Symp. Proc. 1989, 132,93-98. 73. A. Guha and V. S. R. Murthy, Bull. Muter. Sci. 1996, 19, 1117-1124. 74. S. C. Khatri, M. J. Koczak, T. Chou, and Y. Kagawa, Ceram. Eng. Sci. Proc. 1992,13,494-502. 75. T. Watari, K. Mori, T. Torikai, and 0. Matsuda, J . Am. Ceram. Soc. 1994,77,2599-2602. 76. S. C. Koh, D. K. Kim, and C. H. Kim, J . Muter. Synth. Proc. 1995,3, 105-110. 77. H. Venugopalan, K. Tankala, and T. Debroy, Muter. Sci. Eng. 1996, A210, 6 4 7 5 . 78. S. C. Khatri, M. J. Koczak, T. Chou, and Y. Kagawa, Ceram. Eng. Sci. Proc. 1992,13,485493. 79. D. D. Upadhyay, R. Bhat, S. Ramanathan, and S. K. Roy, J . Alloys Compd. 1994, 205, 274279. 80. H. Venugopalan and T. Debroy, Muter. Sci. Eng. 1997, A232, 39-46. 81. S. Kumar, V. Jayaram, T. V. Mani, and K. G. K. Warrier, J . Muter. Sci. 1997,32,47194722. 82. E. Manor, H. Ni, C. G. Levi, and R. Mehrabian, J . Am. Ceram. Soc. 1993,76, 1777-1787. 83. S. P. Dhandapani, V. Jayaram, and M. K. Surappa, Acta Metall. Muter. 1994, 42, 649-656. 84. S. Lee and D. K. Kim, Ceram. Eng. Sci. Proc. 1990,11,795-805. 85. V. Jayaram, S. Kumar, T. V. Mani, M. S. M. Saifullah, J. Sarkar, and K. G. Warrier, in Inorganic Matrix Composites, M. K. Surappa (Ed.), The Minerals, Metals and Materials Society, Warrendale, USA, 1996, pp. 193-205. 86. V. S. R. Murthy and A. Deepak, Br. Ceram. Trans. 1996, 95, 173-176. 87. A. S. Fareed, G. H. Schiroky, and C. R. Kennedy, Ceram. Eng. Sci. Proc. 1993,14,794-801. 88. A. S. Fareed and G. H. Schiroky, Ceram. Eng. Sci. Proc. 1994, 15, 344352. 89. X. Gu and E. G. Butler, Ceram. Eng. Sci. Proc. 1997, 18, 371-378. 90. I. S. Kim and H. W. Hennicke, Znterceram. 1992,41, 75-76. 91. S. M. Pickard, E. Manor, H. Ni, A. G. Evans, and R. Mehrabian, Acta Metall. Mater. 1992, 40, 177-184. 92. D. J. Landini, H. D. Lesher, and J. Gesing, in Proc. Third International Symp. Ceramic Materials und Components for Engines, V. J. Tennery (Ed.), American Ceramic Society, Westerville, USA, 1988, pp. 1536-1551.

Rejerences

32 1

93. J. Weinstein and B. Rossing, in High Performance Composites for the 1990s, S. K. Das, C. P. Ballard, and F. Marikar (Eds.), The Minerals, Metals and Materials Society, Warrendale, USA, 1991, pp. 339-357. 94. B. D. Flinn, M. Ruehle, and A. G. Evans, Acta Metall. 1989, 37, 3001-3006. 95. J. Roedel, M. Sindel, M. Dransmann, R. W. Steinbrech, and N. Claussen, J . Europ. Ceram. SOC.1994, 14, 153-161. 96. C. A. Andersson and M. K. Aghajanian, Ceram. Eng. Sci. Proc. 1988, 9, 621-626. 97. I-S. Kim and I-G. Kim, J . Muter. Sci.Lett. 1997, 16, 772-775. 98. R. Arvind Singh, A. K. Sood, V. Jayaram, and S. K. Biswas, Scripta Muter. 1998,38,617-622. 99. H-T. Lin and K. Breder, J . Am. Ceram. Soc. 1996,79, 2218-2220. 100. Reported in Product Brochures, ALANX Products, 101 Lake Drive, Newark, Delaware 19702, USA. 101. R. K. Dwivedi, Cerum. Eng. Sci. Proc. 1991, 12, 2203-2221. 102. V. Jayaram, R. Manna, M. G. Kshetrapal, J. Sarkar, and S. K. Biswas, J. Am. Ceram. Soc. 1996,79, 770-772. 103. A. Ravikiran, V. Jayaram, and S. K. Biswas, J . Am. Ceram. SOC.1997, 80, 219-224. 104. J. Zhang, D. G. Zhu, and L. Yang, Wear, 1998,215, 34-39. 105. R. Arvind Singh, V. Jayaram, and S. K . Biswas, J . Muter. Res. 1999, 14, 6 6 6 7 . 106. C. A. Anderson, P. Barron-Antolin, A. S. Fareed, and G. H. Schiroky, in Proc. Int. Con$ Whisker and Fiber Toughened Ceramics, American Society for Metals, 1988, pp. 209-21 5. 107. A. S. Fareed, B. Sonurpalak, C. T. Lee, A. J. Fortini, and G. H. Schiroky, Ceram. Eng. Sci. Proc. 1990, 11, 782-794. 108. A. Fareed, B. Sonurpalak, P. A. Craig, and J. E. Garnier, Ceram. Eng. Sci. Proc. 1992, 13, 804818. 109. F. Heredia, A. G. Evans, and C. A. Andersson, J . Am. Ceram. SOC.1995,78,2790-2800. 110. D. G. Kolman and D. P. Butt, J . Electrochem. SOC.1997, 144, 3785-3791. 111. W. A. Kem, M. J. McNallan, and R. E. Tressler, J . Am. Ceram. Soc. 1996, 79, 1881-1891. 112. T. M. Strobel, J. P. Hurley, K. Breder, and J. E. Holowczak, Ceram. Eng. Sci.Proc. 1994, 15, 579-586. 113. H. Venugopalan, K . Tankala, and T. Debroy, J . Am. Ceram. Soc. 1994,77, 3045-3047. 114. S. L. Swartz, D. R. White, and L. E. Cross, J . Muter. Sci. 1992, 27, 3932-3938. 115. Reported in Futuretech, Technical Insights Inc., Englewood, USA, 1990, 101, 1-19. 116. B. W. Sorenson, G. H. Schiroky, and A. W. Urquhart, Turbomachinery, 1990,31,20-26. 117. S. Ashley, Mech. Eng. 1991, 113, 44-49. 1 18. Reported in Structural CeramicslCerumic Matrix Composites, Techmonitoring, SRI International, 1995, 44-45. 119. L. P. Zawada and S. S. Lee, Ceram. Eng. Sci. Proc. 1995, 16, 337-339. 120. Reported in: Society and Industrial News, Am. Ceram. Soc. Bull. 1995, 73, 28.


2 Self-Propagating High-Temperature Synthesis of Hard Materials Z. A. Munir and U. Anselmi-Tamburini

2.1 Introduction Of the many advantages of self-propagating high-temperature synthesis (SHS), most can be realized in the synthesis of hard (primarily refractory) materials. Among the most important and generally applicable of these advantages are: the energy efficiency provided by this process, its tendency to produce relatively pure products, and its suitability for the simultaneous synthesis and densification of monolithic and composite materials. The synthesis and densification of composites can be accomplished in one step through the use of this method. In recent years, the utilization of the SHS process for the preparation of hard and other types of materials has been extensive with a variety of experimental approaches and modifications, including chemical, thermal, mechanical, and electromagnetic activations. The basis of the SHS method is the propensity of highly exothermic reactions to become self-sustaining after they have been initiated by a brief energy input. A requirement for such a phenomenon is that a portion of the heat generated by the chemical reaction is conducted forward to raise the temperature of the next reactants layer to a temperature at which the reaction rate becomes significant, a temperature commonly referred to as the 'ignition' temperature. Part of the heat generated in this layer will be conducted forward and thus the process is repeated and the reaction moves along the reactants within a usually narrow zone, typically referred to as the combustion zone or combustion wave. The process is demonstrated by the sequence of photographs in Fig 1 . As will be seen in more detail below, the SHS process is governed by the chemical and physical parameters of the reacting system. And under real experimental conditions, the process is also controlled by the ambient conditions. To understand the SHS process and its governing parameters, we consider the general Fourier heat transfer relationship accounting for a heat source,

+

pC,(dT/dt) = fi(d2T/dx2) p Q ( d q / d t ) - h / w ( T - To) - u s / w ( T 4- T t ) , ( 1 )

where p is the density of the product (g ~ m - ~Cp ) , is its heat capacity (J g-' K-I), K. is the thermal conductivity (JcmP2s-' K-*), Q is the heat of the reaction (Jg-I), q is the fraction of the reaction completed, h is the convective heat transfer coefficient (J cm-2 sC1 K-' , us is the Stefan-Boltzmann constant (J cmP2s-' KP4), T is temperature (K), To is the ambient temperature, w is the sample thickness (cm), t is time (s), and x (cm) is the coordinate along which the combustion wave is advancing. The term on the left-hand side of Eq. (1) is the algebraic total rate of heat accumulated in the combustion zone and is the sum of the net rate of conducted

2.1 Introduction

323

Figure 1. Photographic sequence of self-propagating combustion synthesis: reaction between Ti powder compact and gaseous nitrogen.

heat, the rate of chemical heat evolution, the rate of heat loss due to convection, and the rate of heat loss due to radiation. The reaction rate is assumed to have an Arrhenius dependence on temperature and has the form

dq/dt

= ko(l - q)"exp(-E/RT),

(2) where E is the activation energy for the process, n is the order of the reaction, R is the gas constant, and ko is a constant. Under ideal (adiabatic) conditions, the last two terms on the right-hand side of Eq. (1) are neglected and the calculated maximum temperature in the combustion zone is the adiabatic combustion temperature, Tad.For most of the important refractory materials (including many hard materials) Tad is high, exceeding 4000 K in some cases. The high value of this parameter is a thermodynamic consequence of the high exothermic heat of formation of these materials. Table 1 lists the values for Tad and the enthalpies of formation for selected hard materials, including carbides, nitrides, oxides, and borides. With few exceptions, the adiabatic temperatures of the listed materials exceed the empirically established minimum limit for SHS, about 2000 K [I]. Generally the exceptions are compounds of tungsten (W2C,WC, WB, W2B),molybdenum (Mo2C,MoN, MOB),and chromium (Cr3C2).It should be emphasized that the temperatures in Table 1 are based on adiabatic conditions and thus are upper limits which are seldom achieved in practice. In some cases, notably in the case of nitrides, the adiabatic temperature is so high as to exceed the dissociation temperature of the nitride phase. Examining Table 1 shows a general trend between the enthalpy of formation of the material and its adiabatic temperature. However, a more accurate relationship is and the adiabatic between the ratio of the enthalpy to the heat capacity (H/Cp)29X

324

2 Self-propagating High-Temperature Synthesis of Hard Materials

Table 1. Adiabatic combustion temperatures and enthalpies of formation of selected materials. Material

Enthalpy, - H (kJ)

Combustion temperature, Tad(K)

TIC HfC ZrC NbC

184.5 208.2 196.7 138.9 100.8 144.1 83.3 53.1 40.2 26.4 323.0 373.6 371.5 236.4 217.2 252.3 40.8 323.8 329.3 328.0 25 1.O 66.1 66.9 138.5 209.2 1675.3 1097.5 944.7

3 290 3 802 3913 2821 2 228 2 902 957 1015 1164 603 4 459 5 622 4941 3 508 3 758 3 436 1239 3 193 3 979 4 183 2715 1617 1158 2 490 2 729 11 312 12 153 9 519

vc

TaC Cr3C2

Mo~C

wc w2c

TIN HfN ZrN NbN VN TaN MoN TiB, HfB2 ZrB2 NbB2 WB W2B VB TaB2 A1203

Zr02 Ti02

temperature (Tad),as has been shown previously [2]. Such a relationship is shown in Fig. 2 for the materials listed in Table 1 for which reliable thermodynamic data are available. With two notable exceptions, the data fit a straight line and a more useful limit for SHS can now be stated as a (H/Cp)298 RZ 3000 K. In SHS reactions, the maximum temperature is ideally achieved within the narrow combustion zone, indicating that the reaction is complete within the combustion wave. However, for kinetic reasons, some reactions continue beyond the zone and consequently the maximum temperature is reached after the passage of the wave. Schematic representations of these two cases are shown in Figs 3(a) and (b), respectively. The figures show profiles of temperature and the conversion parameter. In the latter case, diffusion of one or both reactants through a product layer can provide a kinetic barrier for the process and hence its lack of completion within the combustion zone. Such a behavior can be seen in systems where the particle size of the reactants is relatively large, as was reported earlier for the synthesis of titanium carbide [3]. Since in this case the conversion is incomplete in the wave, the temperature of the wave is relatively low and hence its rate of propagation is low. In general the nature of the product is not affected by a change in velocity as long as the propagation is in the steady-state mode [4-6]. When the propagation is in the

2.1 Introduction

2 0

1 5

325

E

1 0

5

F

TiCh

0 0

2

6

4

8

1 0

1 2

1 4

Tadx1 0- 3( K ) Figure 2. The relationship between (AH,,,/C,)

and the adiabatic temperature for hard materials.

non-steady-state mode (e.g. spin or pulsating wave propagation), the reaction may be incomplete or the products are not the same as those obtained under steady-state propagation [6]. An example of spin propagation is shown in Fig. 4 for the synthesis of nickel silicides [7].

Propagation Wave

I I&

XrdL Wave propagation

~

T,

t-------

I

Vf

k 4b k i .

After-bum

4

*

Figure 3. Schematic representation of the temperature, T , degree of conversion, 7,and rate of heat release, 4, during wave propagation for: (a) a reaction completed within the wave, (b) a reaction which extends beyond the wave.

326

2 Self-propagating High-Temperuture Synthesis of Hard Materials

Figure 4. Evidence of spin combustion during the synthesis of Ni silicides [7].

2500

1

0 0

7.5

15

22.5 30 37.5 time (s)

45 52.5

60

2500

Figure 5. Temperature profiles during the combustion synthesis of: (a) Ti + C + 25 weight-% Ni, (b) Ti N2 (gas) [12].

+

2.2 Mechanistic Characterization of the Process

327

2.2 Mechanistic Characterization of the Process Determinations of the mechanism of self-sustaining synthesis reactions have been attempted through two experimental approaches. The first, and most widely used, method examines the temperature dependence of the wave velocity [8] and the second utilizes temperature profiles associated with wave propagation [9-111. Although in principle both methods should provide the same value for the activation energy, in reality results obtained by the two methods can be significantly different [ 11. The difference is partly attributable to the spatial distribution of the conversion parameter. Referring to Fig. 3(b), which depicts a case where the reaction is not complete within the wave, the temperature in the wave is not the maximum, but it is the value that controls the rate of wave propagation. Thus activation energies obtained for such cases are not valid for the process. Examples of synthesis processes in which the reactions are complete within the zone and those in which the reactions continue beyond the wave are shown in Figs 5(a) and (b),

2500 2000

G 1500 0

G

I000 500 O L J I

0

1

I

I

I

15

30 t

I

1

60

45

6)

2500 2000

-Y G

1500

-

~

1000

-

500

-

O r .

I

I

I

Figure 6. Temperature profile of a 52% dense Zr + 1.5B + 15 weight YOTiB2: (a) complete profile, (b) enlarged portion of the initial segment of the profile [lo].

328

2 Self-Propagating High-Temperature Synthesis of Hard Materials

n v

1000

-

s b

500 0 3.00 3.04 3.08 3.12 3.16 3.20

150000

50000 0

-50000 3.00 3.04 3.08 3.12 3.16 3.20

PI

-. -I.E+7

"

!

-1.5E+7 3.00 3.04 3.08 3.12 3.16 3.20

t 6) (C)

+

+

Figure 7. Thermal profile characterizations of the wave in the reaction of Zr 1.5B 15 weight-% TiB2: (a) smoothed temperature profile, (b) first derivative of the profile, (c) second derivative of the profile [lo].

respectively. The former is the temperature profile for the synthesis of Tic-Ni composite and the latter is for the synthesis of TIN [12]. In the latter case, both visual observations and thermogravimetric analyses confirmed that the reaction between Ti and nitrogen continues well after the passage of the wave [8,13]. As indicated above, determination of the kinetic parameters of the SHS process can be made through a mathematical analysis of temperature profiles [14]. Integration of an adiabatic form of Eq. (1) along with appropriate substitutions gives the following expression for the conversion parameter [9], [Cpu(T- TO)- 61( d T / d x ) I / [ ( ~ ~ l ) ( d T / d x+ ) QWI,

(3) where u is the wave velocity, K~ and K~ are the thermal conductivity of the reactants and products, respectively, To is the ambient temperature, and the other symbols are as defined above. Equation (3), which gives the spatial distribution V(X) =

2.2 Mechanistic Characterization of the Process

329

U L ":E 20 10 0

3.00 3.04 3.08 3.12 3.16 3.20

(a)

t (s)

0.6

0.2 3.00 0 -0.2

3.05

3.10

3.15

3.20

+

+

Figure 8. Thermal profile characterization of the reaction Zr 1.5B 15 weight-% TiB,: (a) the time dependence of the reaction rate, aq/dt, (b) the time dependence of the conversion factor, q [lo].

of the conversion parameter, can be differentiated with respect to time to give the reaction rate, &/at, which is related to x by d v / d t = u ( a q / a x ) .Utilization of this (the Boddington-Laye) method of kinetic analysis has been made in investigations on the combustion synthesis of zirconium boride, molybdenum silicide, and titanium silicide [lo, 111. Figure 6(a) shows a typical profile during the reaction of Zr 1.5B 15weight-% TiB2 and Fig. 6(b) shows an enlarged section of the initial portion of the profile. A smoothed version of the latter as well as its first and second derivatives are shown in Figs 7(a) to (c), respectively. From Fig. 7(b), the maximum heating rate at the leading edge of the wave is calculated as 1.6 x lo5K s-'. Using Eq. (3), the reaction rate and the temporal distribution of the conversion parameter can be calculated, as shown in Figs 8(a) and (b), respectively. It can be seen from Fig. 7(b) that the conversion is complete in about 8 x lO-*s and with the knowledge of the wave velocity, the width of the combustion zone is calculated as about 0.4 mm. Using the calculated reaction rates and their dependence on temperature, activation energies for the formation of zirconium boride can be calculated, as shown in Fig. 9. The lines represent data for different levels of conversion. Those representing 7 values of 0.6, 0.7, and 0.8 have similar slopes. The lack of conformity of the lines for q of 0.5 and 0.9 is related to the uncertainties in the process of obtaining derivatives at these ends of the conversion profile, as explained elsewhere [lo]. The average value for the activation energy for the combustion synthesis reaction Zr + 1.5B + XTiB2 (with X between 0.15 and 0.22) was calculated as 144.5 x 15 kJ molF' . The activation energy was also calculated from measurements of the

+

+

330

4

g3 5 5 2

I

2 Self Propagating High-Temperature Synthesis of Hard Materials

-

-

*

F \ ,

-

-

t 0.9 0 1 3

I

I

I

5

4

I

6

1 0 ' (IK) ~

Figure 9. Arrhenius plot of the temperature dependence of the reaction rate for Zr + 1.SB+XTiB, for 7 values from 0.5 to 0.9.

temperature dependence of the wave velocity, shown in Fig. 10. The resulting value, 148.6 x 12kJmol-', is in good agreement with that determined from the profile analysis, indicating that the reaction was complete within the relatively narrow reaction zone. Similar analyses were made for the synthesis of MoSi2 and Ti2Si3 [l 11. In these cases, particularly for MoSi2, the reaction is not complete within a narrow zone as can be seen from Fig. 11. Here the width of the zone is calculated as approximately 1.3mm. The wide reaction zone for molybdenum silicide is, in part, the consequence of a relatively low enthalpy of formation as manifested by an adiabatic temperature near the empirically established lower limit. This is the reason why some composites of MoSi2 cannot be directly synthesized by SHS without some form of activation.

-7.8 E = 134 W/mol (Qconstant)

1

-8.0 r. F -3 -8.2 e C

d

-8.4 0

-8.6 3.4

E = 148 Wimol (Qvarying) I

I

I

I

3.5

3.6

3.7

3.8

-II 3.9

1 0 ' (~1 1 ~ )

Figure 10. Arrhenius plot of the temperature dependence of the wave velocity in the reaction Zr = 1.5B+XTiB2 [lo]: 0, from velocity measurements; 0 , from temperature profiles.

2.3 Efect of Experimental Parameters

33 1

0.9 0.8 0.7

0.6

0.5 Er

0.4

0.3 0.2 0.1 0.0 -0.1

2.4

2.5

2.6

2.7 t

2.8

2.9

6)

Figure 11. The time dependence of the conversion profile, 7, for the synthesis of MoSi,

2.3 Effect of Experimental Parameters Typical SHS experiments for materials synthesis involve the use of powders. Powders are mixed in the desired stoichiometry, pressed into a compact, and then ignited at one end to produce the self-sustaining combustion wave. In some cases, a diluent is added. This is done either because its presence in the product is desired, or it is added to control (i.e. to lower) the temperature of the reaction. In the latter case, the product of synthesis is added as a diluent. The SHS procedure, therefore, includes the materials and experimental parameters of powder particle size, degree of dispersion, relative density of the compact, the stoichiometric ratio, the degree and nature of dilution, and the initial temperature of the reactants. The initial temperature, To, has a direct influence on the adiabatic temperature and hence the feasibility of establishing self-sustaining combustion. A schematic representation of the general effect of To on Tad is depicted in Fig. 12(a) and the specific effect in the case of T i c is shown in Fig. 12(b) [15]. In the latter case, Tad coincides with the melting point of TIC (3210 K) with the fraction of melted product = 0.33 for To 7299 K. Figure 12(a) shows that increasing To from T , to T, increases the adiabatic temperature from Tal to Ta2. The relationship between Toand Tad can be derived for a general case in which Tad < T,, where T,, is the melting point of the product as TO

(a H298

+

1298

+ bT)d T =

/

Tad

(up TO

+ b p )dT,

(4)

where H298is the enthalpy of formation of the product at 298 K, A u and Ab are the algebraic sums of the Cp coefficients of the product and reactants (assuming a Cp = a + bT form), ap and b, are the coefficients for the heat capacity equation of the product. In the case of TIC, shown in Fig. 12(b), the adiabatic temperature (for To = 298 K) is at the melting point of the product and increasing To initially

332


3340

-

3320

-

3300 -

8 2

3280 3260 3240 3220

-

3200 -

A

I 400

(b)

800 1200 1600 2000 2400 2800

To(K)

Figure 12. (a) Schematic representation of the influence of the initial temperature on the adiabatic temperature: His the enthalpy and T is temperature. (b) The dependence of the adiabatic temperature on the initial temperature for TIC [15].

has no effect on Taduntil all the product is in the molten state. Further increase in To will result in an increase in the adiabatic temperature. The preheating of reactants before ignition is a method employed in the synthesis of materials whose adiabatic temperatures are relatively low. This is to be distinguished from the practice of heating the reactants (e.g. in a furnace) until the entire sample reacts. This latter form of combustion synthesis has been erroneously referred to as SHS, but, in fact, it has no discernible combustion wave and it is more appropriately referred to as the simultaneous (or volume) combustion form of combustion synthesis. For this form of combustion synthesis, the rate of heating to the ignition temperature plays a major role in the thermal behavior of the process and the nature of the resulting products [16]. Lower heating rates result in the formation of precombustion phases through interfacial diffusion with the consequence of a reduced driving force in the main (combustion) reaction [17,18].

2.3 Effect of Experimental Parameters

1

9

o

o

r

I

' '

I

I

'

'

'

I

'

'

h

'

---

333

Y

Y

e3

1800

-

c

E

1700

i

-

E

I-

.c

1600

-

m n

m *e I500 -

U

-0.2

0

0.2

0.4

0.6

0.8

1

1.2

Stoichiometric increment (x)

Figure 13. The effect of the stoichiometry of the reactants Mo + (l+x)Si and Si adiabatic temperature.

+ (1 + x)C on the

In principle, the roles of stoichiometry and dilution are similar for product phases with a narrow compositional stability range. For example, in the synthesis of MoSi2 and Sic, changing x in Mo (1 x)Si and Si (1 x)C has the same effect as adding a diluent, as seen in Fig. 13, which shows the effect of stoichiometry on the adiabatic temperature. This, however, is not the case when the product phase is thermodynamically stable over a range of composition, as is the case of many hard compounds (e.g. TIC, ZrC, NbC, TIN, ZrN, and others). For such materials, the enthalpy of formation is a function of composition and thus changing the stoichiometry of the reactants has the effect of changing the heat of the reaction as well as the stoichiometry of the product. Using a modeling approach in which the product is presumed to have a fixed nominal stoichiometry [19], the influence of the mixture ratio of the nonmetal to the metal on the combustion wave velocity was calculated and compared to experimental results, as shown in Figs 14(a) and (b) for synthesis in the Ti-B and Zr-B systems [20-251. In these cases, increasing the value of p from about 0.3 to 1.0 has the effect of reducing the degree of dilution by the excess metal in the starting mixture. An experimental investigation on the effect of stoichiometry on the combustion synthesis of TIC was recently carried out by Makino et al. [26]. Figure 15(a) shows the dependence of the existence and nature of combustion waves on the C/Ti ratio (p) for samples (1Omm diameter and 35mm length) with initial relative densities ranging from about 55 to 95% and a carbon particle (diameter) size of 20pm. Steady-state combustion was observed in the range 0.47 5 p 5 1.4 and a pulsating mode of combustion was observed for 1.4 5 p 5 1.9. For p > 1.9 and p < 0.47, the samples either were ignited but the wave became extinguished, or the samples could not be ignited at all. As the figure shows, the relative density had little influence on these boundaries, The observed lower limit for steady-state propagation, p M 0.47, coincides with the lower limit of phase stability in the Ticl p.y or about 32 at% C. On the other hand,

+ +

+ +

334


Ti - B

1

0

(b)

Zr- B

1 .o

0.5 P

Figure 14. Modeling results of the influence of the mixture ratio of nonmetal to the metal, p, on the combustion wave velocity in the: (a) Ti-B system, (b) Zr-B system [19]. Tois the initial temperature: data from [20-251.

the corresponding upper limit, p = 1.4, falls beyond the stoichiometric limit of TIC phase and thus the product must contain excess carbon. The dependence of the velocity on the stoichiometric ratio is shown in Fig. 15(b). The rate of propagation increases to a maximum corresponding to the limit of steady-state propagation, that is, the highest velocity (reaction rate) is obtained under a condition of excess carbon. It has been suggested that excess carbon enhances the kinetics of the reaction between liquid titanium and solid carbon [27]. The effect of dilution on the synthesis reaction of hard (refractory) materials has been examined experimentally and through modeling. Figure 16 shows the effect of adding TIC as a diluent on the adiabatic temperature and the fraction melted of the product [ 151. The adiabatic temperature remains constant with the addition of TIC up to 20 mol%, at this point the product is entirely solid and further addition of the diluent decreases the adiabatic temperature as shown in the figure. The decrease in the combustion wave temperature is expected to result in a decrease in the wave velocity. Experimental confirmation of this is provided in Fig. 17 for the case of

2.3 Effect of’E.xperimcnta1 Parameters

I

335

Specimen diameter=lO mm

0 Steady

OS

3 -f .-

0 Pulsating

t

A Extinction X No ignition

I 0

I

I

1

2

3

15 P*I

A 0.55-0.65 00.65-0.75 0 0.75-0.85 0 OM-0.95

fl

0

10 -

0

-

0 .

0

a&

.E

mE

5

m *

.*

-

Open: steady

Filled: pulsating 0

2 Mixture ratio, p

3

1

(b)

Figure 15. The effect of the mixture ratio C/Ti (= p) on: (a) the relative density and nature of the combustion mode, (b) the velocity of the wave for various relative densities of the reactants [26].

1 60

3300

50

3200

h

s. 3100 3000 i-”

2700

k

40 h

20

0

0

10

20

30 40 50 TIC (mol%)

60

70

Figure 16. The influence of dilution (by TIC) on the adiabatic temperature and fraction of product melted (v) in the synthesis of TIC [15].

336

2 Self-propagating High-Temperature Synthesis of Hard Materials d,. W)

p = 1.0 h

Borovinskaya Present

E

.-

0 &el= 0.56

0

.-

0

v

m 0.0

Degree of dilution

0.5

Figure 17. The effect of dilution on the wave velocity in the synthesis of TIC [26]. Experimental results from [27].

TIC 1261 and compared to results of a previous investigation [27]. It is interesting to note that the use of smaller carbon particles (0.1 pm) in the earlier study 1271resulted in steady-state wave propagation in the same range as for the case of larger carbon particles (25 pm). However, the use of smaller particles extended the dilution range, albeit with pulsating wave propagation. Non-steady-state (pulsating and spin) propagation results in laminated products that are macroscopically inhomogeneous [7]. As indicated above, the addition of a diluent reduces the combustion temperature and the velocity of the wave. When a sufficient amount of diluent is added, the wave becomes unstable and ultimately ceases to advance. The concept of a wave extinction limit was utilized to construct SHS diagrams [4,5]. These diagrams were based on the Zeldovich criterion for wave extinction [28],

where vmaxis the maximum (adiabatic) velocity and wmin is the minimum velocity before the wave becomes extinguished. Equation (5) was used with the relationship between velocity and temperature, w2 = ( K RT2/QE)exp(-E/RT),

(6)

where Q is the heat of the reaction, E is the activation energy of the process, T is temperature, R is the gas constant, and K is a constant. The resultant relationship, which provides boundaries for the existence of SHS, is derived as

where T,, and Tminare the maximum (adiabatic) temperature and the minimum, that is, the temperature of the wave just before it becomes extinguished, and d is the degree of dilution [4,5]. Using this relationship, SHS diagrams for TiB,, H B 2 ,ZrB2, and TIC have been constructed. An example is shown in Figs 18(a) and (b) for TiB2 and Tic, respectively. Details of the experimental results and the theoretical derivations of this model as well as others are given in the literature [4,5].

2.3 Ejfect of Experimental Parameters

331

1400

TiB,

0.

400 200

0

20

(E=398KJ/Mol)

1200

3 g 1000

-

60

40

-/ /

(b)

\

(E=398KJ/MoI)

-

400 200

100

This Work

800 -

600

80

'

0

(E=l38KJ/Mol) I

I

I

I

0.2

0.4

0.6

0.8

1.0

d

Figure 18. Theoretical limits of SHS (SHS diagram) and experimental results for: (a) TiB2, (b) Tic [4,51.

The effect of particle size of one or more of the reactants on the SHS process has been examined in experimental and modeling studies. In general, an increase in particle size results in an increase in the diffusion length and hence a decrease in the reaction rate, with the rate expected to be inversely proportional to particle size. Figure 19 shows experimental observations on the relationship between the particle size of titanium and the wave velocity in the synthesis of TIC and TiB2 [21]. This trend is also predicted by modeling studies of Makino and Law, showing the anticipated inverse dependence of the velocity on particle size [29]. Their prediction of this dependence is compared to results of experimental observations in Fig. 20 for the synthesis of TIC [26]. The wave velocity decreases by a factor of about four as the carbon particle size increased from 1 to 25pm. However, the particle size of Ti had no apparent influence on the velocity within the range of 10-75 pm. The fact that titanium is molten in the combustion zone is believed to provide the explanation for the lack of a dependence. The experimental results of the trend of velocity with carbon particle size provide an unexpected observation at the small particle size end. When small (<5 pm) carbon particles were used, the

338

2 Self-Propagating High-Temperature Synthesis of Hard Muterials I

24 1

i

. : h

v

12

>

6

n ”

0

0.1

1.5

0.5 1.0 r, (mm)

Figure 19. Dependence of the combustion rate of Ti + 2B and Ti i-C mixtures on the Ti particle size for several initial temperatures (To):0 , To = 20°C; 0, To = 200°C; 0, TO= 400°C [21].

wave propagated in a pulsating mode, albeit at a much higher average velocity. The significance of this, as was pointed out earlier, is that the products of non-steadystate combustion are inhomogeneous. In another investigation, the size of the Ti particles was shown to have little effect on the density of the TIC product, but it had an effect on the resulting microstructure [30]. Another important experimental parameter in combustion synthesis is the relative density of the reactants. Figure 21 shows the dependence of the wave velocity on density in the combustion synthesis of TiNi [31]. These and similar experimental p = 1.0

30

P..,

d&mm) 0 Deevi et al. 1.6

0.56 1.0 0.52-0.60

h

2”E

-

B

25

”

25

”

20

-

Open: steady Filled pulsating

0 d)

I

0

I

I

I

20 30 40 50 Diameter of C particles, d, (pm) 10

Figure 20. The effect of the particle size of carbon on the combustion wave velocity in the synthesis of TIC [26].

2.3 Effect

of

E.uperimentul Purameters

339

1.5

1.o

0.5 40

20

0

60

Porosity ( % ) Figure 21. The effect of porosity on the wave velocity in the synthesis of TiNi [31].

results show a relatively common pattern in which the wave velocity and combustion temperature have maximum values at an optimum relative density [213. The relative density has three roles in the SHS reactions. It influences: the energy density (e.g. J ~ m - of ~ )the reaction; the kinetics of the reaction through changes in the particle-particle interfacial contact; and the wave propagation dynamics through changes in the effective thermal conductivity of the reactants. At low relative densities, the velocity and temperature are expected to be low because of all the three factors listed above. On the other hand, at high relative densities the effective thermal conductivity is so high that heat is conducted at such a high rate to distant regions ahead of the wave making it impossible to reach the ignition temperature in the layer immediately ahead of the wave. Although the existence of a maximum is not always observed experimentally, the concept of wave extinction at high densities is always verified. Highly dense reactant powders cannot be ignited. The role of particle size in this behavior is shown by study on the synthesis of TIC, Fig. 22 [26]. Here the particle size of carbon is changed while holding the size of the Ti particles constant at 25 pm. With small carbon particles (1 pm), a maximum in 40

3

4(w)p , ( w

h

f

30

0 Shkiro and Borovinskaya 0. I A Present 1 0 10 v " 25

AA A

8

v

I

.0

OA

20 25 25 25

0

0

20 > M

'gm

10

n "

0.4

0.5

0.6 0.7 0.8 Relative density

0.9

1.0

Figure 22. The effect of relative density of the reactants on the wave velocity in the synthesis of TIC P61.

340

2 Self-Propagating High-Temperature Synthesis of' Hard Materials

the wave velocity is observed at a relative density of 55%. However, regardless of density, wave propagation was in the pulsating mode when these powders were used. In contrast, no maximum is observed when relatively large particles (25 pm) were used and in this case wave propagation was in the steady-state mode over the entire range up to the extinction limit. The results further show that this limit decreases as the particle size increases. For example, the limit is about 85% relative density for samples with carbon particles of 1 pm diameter and it is about 75% relative density for samples with carbon particles of 25 pm diameter. The role of relative density in gas-solid synthesis is significant for a different reason. This case relates to the synthesis of nitrides for the category of materials of interest to this chapter. Porosity in this case is essential to maximize the interfacial contact area between the metal and nitrogen gas and to provide a conduit for the replenishment of the gas to maintain the local supply at the interface. In principle, the latter requirement can be negated if the total pores of the solid reactant contain, at the outset, the needed number of nitrogen gas molecules for complete conversion to the nitride phase. Permeation of the gas through the pores would not play a significant role in such a case. An analysis has been made to provide the following relationship between the degree of conversion to the nitride and the pressure of nitrogen within the pores [32],

v = {(1/S)f'n[P/(1

VrnlIRT, (8) where 7 is the degree of conversion to the nitride, S is the stoichiometric ratio of the nitride (i.e. number of moles of N2 per mole of metal in the reaction), Pn is the pressure of the nitrogen gas, V, is the molar volume of the metal, p is the initial porosity of the sample, R is the gas constant, and T is temperature. Applying Eq. (8) to the synthesis of selected refractory nitrides results in the curves of Fig. 23. The figure shows, for example, that complete conversion of Ti to TIN with an initial porosity of 0.7 would require a pressure of nitrogen of about 2 x lo3 atm. The corresponding value for the case of BN is about 6 x lo3 atm. This does not imply that complete conversion would be accomplished with such

0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

Figure 23. The dependence of the degree of completion of conversion to TiN on the pressure of nitrogen gas in preexisting pores: solid lines, 1000 K; dashed lines, 1500 K [32].

2.3 Effectof Experimental Parameters

10'

'

I

I

I

I

I

I

4.0

6.0

8.0

10.0

12.0

14.0

341

V,"(cm')

Figure 24. A generalized relationship between the pressure of nitrogen and the molar volume of elements with useful nitrides at complete conversions. T = 1500 K, porosity = 0.5. Nitride phases: A = M3N4,B = MN, C = M,N,, D = M,N [32].

a pressure, since Eq. (8) does not account for the kinetics of the process. It only provides an indication of the significance of permeation in the synthesis of the nitrides. This can be better demonstrated by Fig. 24 in which the pressure for complete conversion is plotted as a function of the molar volume of the metal and shown for homologous series of nitrides. It can be seen from this figure, for example, that the nitriding of Hf and Zr (to form HfN and ZrN) is less affected by permeation than B (to form BN). Experimentally, the role of porosity (relative density) in nitride formation by SHS is complex and depends on phase transformation (as it affects permeation) and the diffusion of nitrogen atoms through the metal or the nitride. The influence of the former relates to the melting of the metal. If the combustion temperature exceeds the melting point of the metal, then permeation is affected by the concomitant change in the continuous porosity. Such a behavior is shown in Fig. 25(a) where the degree of conversion to TIN is plotted against the relative density of the starting Ti powder for synthesis in one atmosphere nitrogen pressure [33]. Maximum conversion occurs at a relative density of about 55% and the decrease with subsequent increase in density is due to a decrease in the local availability of nitrogen resulting from the decrease in permeability. The decrease on the other side, that is, as the density decreases from 55%, is due to the melting of Ti and its influence on the open porosity in the Ti reactant. For this end of the curve, the surface area of the Ti in contact with gaseous nitrogen is high, leading to a higher reaction temperature and more extensive melting of the titanium [34]. In systems where the combustion temperature does not exceed the melting point of the metal, the dependence of conversion on density does not show a maximum but, as predicted, decreases with an increase in the relative density, as shown in Fig. 25(b) for the synthesis of NbN [35]. As can be seen from Figs 25(a) and (b), complete conversion to the nitride phases is not accomplished under the experimental conditions of one atmosphere of nitrogen. In the case of TiN,

342

2 Self-Propuguting High-Tetnpeniture Synthesis of Hurt1 Muterid\

+

8

a 40 45

I

I

I

I

I

I

50

55

60

65

I0

15

Relative density of Ti pellets (%) (a)

z

2

48 44 40

0 *

3 36 .v1

3 -

0 -

0

0

0

-

32 28

0

0

-

0

0

1

,,LLL 2056

57

58

59

60

61

62

Relative Density (%)

Figure 25. The effect of relative density of the metal compact on the combustion wave velocity: (a) for the synthesis of TIN, (b) for the synthesis of NbN [33,35].

complete conversion can be achieved with the appropriate control of the melting of Ti by the addition of a diluent.

2.3 Synthesis of Dense Materials When the temperature of the SHS process exceeds the melting points of all product components, the product can be highly dense without the use of external pressure. Such products are referred to as ‘cast SHS materials’ and are primarily the result of thermite reactions [36]. However, in other than these extremely exothermic reactions the products tend to be highly porous. While this is probably the most important disadvantage of the S H S process, it can actually be an advantage in some special cases, as in the synthesis of electrode materials [37].A direct relationship between the initial and the final densities is observed, see Fig. 26 for the case of T i c [30]. However, attempts to use highly dense reactants, taken as a step towards ameliorating

2.3 Synthesis of Dense Materials

343

45

5

h

,x 40 .A .

5

5

+ *

35

rt;

30 45

-

I

I

I

50 55 60 65 Uncombusted reactants density (YO)

Figure 26. The relationship between the relative densities of the reactants and product in the synthesis of TIC [30].

this problem, cannot overcome the primary intrinsic source of porosity (see below). Thus the need to densify the product through simultaneous or subsequent application of pressure is of major concern in SHS synthesis. The sources of porosity in SHS products have been analyzed and classified as intrinsic and extrinsic 1381. The extrinsic sources of porosity are the initial porosity of the reactants and the porosity generated by the expulsion of volatile impurities during synthesis. The contribution of the latter is difficult to calculate exactly, but an approximate relationship has been derived which shows the dependence of the level of porosity on the amount of impurity as

+ CipmRTIMiPc,

(9) where wp is the volume of pores per unit volume of product, Ci is the concentration of the impurity in the reactants, pm is the theoretical density of the product material, Mi is the atomic or molecular mass of the impurity, P, is the pressure to overcome the confining forces in the powder, R is the gas constant, and T is the absolute temperature. Results of such an analysis for the synthesis of TIC are plotted in Fig. 27 for selected (assumed) impurities. The calculations were made assuming a combustion temperature of 3000 K (the adiabatic temperature of TIC is 3210 K) and the confining pressure, P, (the pressure needed to push the compacted particles apart), was assumed to be 200 atm. The figure shows that at an impurity level of, for example, lOOOppm the volume percentage porosity can range from about 4.5% for Ba as an impurity to about 26% for Mg. Intrinsic sources of porosity are the molar volume change (due to the formation of the product) and the porosity generated by the thermal migration. Figure 28 shows the relationship between the molar volume of the reactants and the product for selected refractory materials. The lines represent a value of x in the relationship V , = xVR, where Vp and VR are the molar volumes of the product and reactants, respectively. The figure shows that the formation of TaC, B4C, for example results in a porosity of about lo%, while the formation of Sic, MoSi2, results in a porosity of about 30%. The other intrinsic source porosity is thermal migration (the Soret up

344

2 SelflPropagating High-Temperature Synthesis of' Hard Muteriuls

100

Impurity:

.-

VI

a

.s 1.o Tic : T = 3000K 2. 0.1 1.o

1o2

10

10'

10'

Impurity level, ppm

Figure 27. The contribution of impurity volatilization to pore formation in the synthesis of TIC [38].

effect). As a consequence of a temperature gradient, a nonuniform vacancy concentration can arise, even in pure materials [39]. The coalescence of the vacancies can give rise to porosity. The extent and location (relative to the temperature gradient) of the excess vacancies from this effect depends on the magnitude and sign of the heat of transport [40]. Calculations have been made to determine the size of the pore diameter, dp, as a function of the temperature gradient and the result is expressed as

(WT&- D C / ~31 P3

dp = { (6e2/n)W J R ) ( d ~ / d x )

(10) where 1 is the length of the sample across which the temperature gradient exists, t is the time interval over which the gradient exists, Q, is the heat of transport, DH and Dc are the diffusion coefficients at the high (TH)and low (Tc) temperature regions of the sample, respectively, and dT/dx is the temperature gradient. Assuming that the temperature gradient occurs across one powder particle, the ratio of pore 34

v, = xv,

x= I .o

30 26

-

10

14

18

22

26

30

34

38

v, ( c d Figure 28. The relationship between reactants and product molar volume for selected compounds 1381.

2.3 Synthesis of' Dense Materirrls

r-50

30

345

rl

I 40 50

i

10

log (dT/dX) (Wcm)

log (dl'idX) (Wcm)

4 6 8 log (dT/dX) (Wcm)

Figure 29. The effect of temperature gradient on the pore/grain size ratio for: (a) Ti + Ni; (b) Zr + 2B; (c) Ta + C [40].

diameter to particle diameter was calculated and the results are shown in Fig. 29 for the three selected examples of TIN, ZrBz, and TIC [40]. For each case, curves are shown as a function of combustion wave velocity, Slower wave propagation and higher temperature gradients give rise to higher porosity from thermal migration. Fi ure 29 shows that under realistic conditions of temperature gradients (lo510f Kcm - ') [lo] and wave velocities (l-lOcms-'), the ratio d p / l ranges from 2 to 10%. However, it should be noted that while thermal migration can give rise to pore formation, experimentally it is difficult to isolate its contribution. All of the sources discussed above, in some combination, contribute to the overall porosity of SHS products, which can be as high as 50%. With few exceptions, the general desire is to prepare dense materials by the SHS process and thus several simultaneous or sequential steps are taken to densify these materials. In the former, the synthesis and densification is achieved in one step, and in the latter the combustion process is followed, within a short time, by a densification process. The idea is to take advantage of the plasticity of the hot product. Examples of the sequential densification include the application of uniaxial pressure [4 1-44], hot-isostatic pressing (HIP) [45,46], pseudo-HIPing [47], dynamic (explosive) compaction [48-5 11, high speed forging [52-541, and pressureless and microwave sintering [55-571. Although the relatively rapid nature of the SHS process was expected to give rise to products with high defect concentrations, experimental observations on the sintering of SHS produced materials have not substantiated this expectation [58,59]. In a related investigation, it was found that shock compressed Ti + C powders could be reaction-sintered to give a highly refined microstructure (grain size of 6 pm) [60]. The simplest form of densification, uniaxial pressing, has been used to densify a variety of materials such as TIC + NiAl, TIC + Ni3Al [41], T i c + Ni, TIC + Fe [42], and others [43,44]. Two parameters play an important role in this type of postreaction densification: the magnitude of the applied pressure and the time of its application relative to the completion of the combustion process. Figure 30 shows the dependence of the residual porosity in Tic-Ni/Al cermets on the applied pressure. As the figure shows, the relative density increases from about 50 to 99% as the pressure is increased from ambient (0.1 MPa) to about 28 MPa [41]. The timing of the application of the pressure relative to the synthesis reaction has also

346

2 SelJ Propagating High-Temperature Synthesis of Hard Materials

Pressure (MPa)

Figure 30. The effect of the applied pressure on the residual porosity in the synthesis of: 0, TiCNiAI, and W TiC-Ni3A1[41].

been investigated [42,61]. Fu et al. examined the various parameters related to pressure application [42]. Figure 31 shows a schematic diagram of the pressure-time parameters. The time for wave propagation, tw, is followed by a delay time, td, before the pressure is changed from the ambient, Po, to the desired (high) pressure, Ph. The time duration through which the pressure is maintained at P h is designated as t,, as shown in the figure. The effect of the delay time on the (relative) density of the composites 75 weight-% TiB2-25 weight-% Fe and 75 weight-% TIC-25 weight-% Ni is shown in Fig. 32. The results show that for these two cases, the highest densities are obtained with little or no delay time (td 5 1 s). In both cases, the time at pressure, t,, was 10s but the values for P h were 95 and 80MPa for the Fe and Ni-containing cermets, respectively. The effect of the time of application of the pressure is shown in Fig. 33. For both cermets, the densification approaches an asymptotic value after a hold of about one minute. Furthermore, and in qualitative agreement with the work of Dunmead et al. [41], the final density depended on the level of the applied pressure, Fig. 34. The difference in behavior of the two

Time

Figure 31. Schematic representation of the sequence of pressure application during SHS synthesis and densification [42].

2.3 Synthesis of' Dense Materials 1 00

341

75TiB1-25Fe t,= I0 scc

98' h

5

. c

Q

92 88

75TiC-25Ni

84

t,=l 0 sec

P = 80 MPa

80

2

6

4

8

10

td(set)

Figure 32. Effect of the delay time in pressure application on the relative density of TiB2-Fe and Tic-Ni composites [42].

material systems is believed to be a consequence of wettability differences. For example, the addition of Mo improves the wettability in the 75 weight-% TiB2-25 weight-% Fe system such that when small amounts of Mo were added, a density of 98.5% could be reached. Thus, in addition to the intrinsic mechanical behavior of the materials, the parameters of temperature and wettability play major roles in the densification process. Poor wettability of the ceramic by the molten metal can result in a nonuniform microstructure, and in the case of the pseudo-HIPing, it can give rise to the expulsion of the metal phase from the composite to the surrounding pressure medium (e.g. silica) [62]. The difficulty in assessing the roles of these parameters independently makes it necessary to rely on specific experimental data for any given systems, but with the above consideration taken as a general guideline. 100

99

5 h

6

98 0

1

4

Q

-

0

75TiBJ25Fe 97 96

t,= 1 sec P = YSMPa

i I 0

20

40

60

80

100 120

t, (set)

Figure 33. Effect of the time of pressure application on the relative density of Tic-Ni and TiB2-Fe composites [42].

348

2 Self Propagating High-Temperature Synthesis of Hard Materials

I00 99

tp= 10 sec

98

g

97

," 96

75TiBJ25Fe

0.

t"= 1 sec

95

93 94

tp= 10 sec

I

U

40

60

L 80

100

120

Pressure (MPa)

Figure 34. The effect of the applied pressure on the final density of Tic-Ni and TiB,-Fe composites ~421.

Simultaneous synthesis and densification has been attempted using different experimental approaches. In the case of nitride cermet synthesis, the application of a nitrogen pressure of 4.0MPa resulted in TiN/Ti-Ni cermets with relative densities in excess of 97% without the application of external pressure [63]. Another approach for the simultaneous synthesis and densification is through shock wave application [64]. The pseudo-HIPing technique has also been used to simultaneously synthesize and densify T i c tiles [65]. For such large samples (about 10 x 10 x 2.5cm), the formation of cracks during cooling is of major concern and whose minimization requires the use of a less conducting pressure transmitting medium to effect a slower cooling rate. Recently, a process was developed to simultaneously synthesize and densify materials using a modified field-activated combustion synthesis [66].

2.4 Synthesis by Field-Activated Self-propagating High-temperature Synthesis The occurrence of SHS reactions is predicated upon the ability of a reaction front to produce heat at a magnitude and a rate which can raise the temperature of the adjacent layer to the 'ignition' point, and thus sustain the process in the form of a moving reaction front or combustion wave. However, for thermodynamic and kinetic reasons this process may not be possible, as is the case for the synthesis of many important materials. Referring to Eq. (l), a thermodynamic limitation is operative when the reaction enthalpy, Q , is low and this is the general basis for the use of the empirical limit of the adiabatic temperature discussed in an earlier

2.4 Synthesis hj. Fielcl-Activuted Sdflpropaguting High-trmperature Sjnthe.si.r

349

part of this chapter. Kinetic limitations of SHS reactions are dominant when the rate of the reaction, &1/8r, is too low or when the thermal conductivity, K , is not optimum. Too low a K value leads to inadequate heat build up in the reactant layer just ahead of the front and hence the absence of ignition conditions. Similar end results are obtained when, for too high a K , the heat is diffused to regions far ahead of the front. Because of a low reaction enthalpy (thermodynamic limitation), materials such as B4C, Sic, WC, Cr3C2,Mo2C, and others cannot be synthesized by a normal SHS process. Kinetic limitations have been given as reasons for the inability to synthesize such materials as TaC. AlN-SIC, and others. Many composites of hard materials where the formation of one of the involved phases is energetically lethargic cannot be synthesized by SHS because of these limitations. These include such materials as MoSi,-SiC, .xB4C-TiB2 (for x > 0.5), and others. The common approach to activate these reactions is to preheat the reactants in order to either raise the adiabatic temperature for a subsequent SHS reaction or to cause a reaction in a simultaneous combustion mode, as discussed above. This is done through furnace heating or by a ‘chemical oven’ [67]. The latter involves the placement of the reactants within a powder mixture of other more exothermic reactants. However, in both of these methods diffusional reactions can occur before the main combustion reaction and can result in the formation of extraneous (undesirable) phases in the product [68,69]. To overcome the limitation of the SHS process and thus facilitate the synthesis of a large number of materials systems, a new process was recently developed in which the activation is effected by the imposition of an electric field [70]. A schematic representation of this method of activation is shown in Fig. 35. Two electrodes are placed in contact with opposite faces of the compact containing the reactants and a voltage is applied across them. In most cases, the application of a field (within limits) does not result in the initiation of a combustion wave. Such a wave, however, can be initiated and propagated after the addition of energy from an ignition source, in the usual way. In all cases investigated, the initiation of a combustion wave is not accomplished below a minimum (threshold) value of voltage (or voltage per sample thickness,

Product

Reactants

Figure 35. Schematic diagram of the field-activated combustion synthesis

350

2 Self Propagating High-Temperature Synthesis of' Hard Materials

0

5

10

15

20

25

E (V/cm) Figure 36. The effect of the applied field on the combustion wave velocity in the synthesis of p-Sic.

field). Figure 36 shows that in the case of the synthesis of Sic, the threshold is about 6Vcm-' [71]. The figure further shows that there are three regions in the dependence of wave velocity on field strength. As indicated, below the threshold no wave propagates, and above the threshold the wave velocity increases nearly linearly with field. At relatively high field values, above about 20Vcm-', the reaction between Si and C does not require an ignition source but takes place in a simultaneous mode. In this case the Joule heating by the current is sufficient to ignite the reaction. As will be discussed in more detail, the role of the applied field is believed to be primarily thermal in nature, providing additional (Joule) energy. In most cases, this energy is localized to the narrow reaction zone, as will be shown later. Real-time measurements of the changes in voltage and current during wave propagation provide the results shown in Fig. 37 for the synthesis of Sic [71]. Both the measured parameters of current and voltage and the calculated parameter of resistance remain reasonably constant during wave propagation, as indicated by the interval between the start (S) and end (E) of the process. In view of the fact that the electrical conductivities of the reactants and product are significantly different, these observations are taken as indicative of a current that is confined to the combustion wave. Modeling and other experimental studies have provided support for this conclusion, as will be discussed below. The modeling study was based on a modification of the applicable Fourier relationship, Eq. (1). The modification includes the addition of another heat source from the Joule heating by the current. The rate of heat generation from this source is uE2, where u is the electrical conductivity and E is the applied field. The modeling was initially done on the synthesis of Sic, using a two-dimensional approach and accounting for heat loss by radiation, to bring the results closer to

’7.4 Syirtliesis h!, Field- Activuted Self+vpagating High-temperature Synthesis 20

,

35 1

1

15 d ‘c)

>

-

10

5

0

0

5

15

10

time (s) Figure 37. Variations of real-time voltage and current and calculated resistance during wave propagation in the synthesis of P-SiC [71]: S and E signify the start and end of wave propagation, respectively.

real experimental conditions [72]. The modified relationship is

+

+

+

pC,(dT/dt) = K [ ( ~ ’ T / ~ . Y ’( a) 2 T / a y 2 ) ] Q(i3v/at) oE2

+ ( o s / W ) ( T-4 T:), (11)

where o,is the Stefan-Boltzmann constant, W is the thickness of the sample, and To is the ambient temperature. In this modeling, the parameters C,, 6,and o are taken as functions of composition and temperature and the reaction is assumed to obey second-order kinetics. The modeling analysis provided results which were qualitatively consistent with the experimental observations. Of these the propagation mode and the localization of the current are the most important results. Figure 38 shows conversion (v) profiles calculated at constant time intervals during wave propagation for the modeled synthesis of Sic. The equal spacing between the profiles is indicative of a steady-state wave propagation, in agreement with the experimental observations. The corresponding current density profiles are shown on Fig. 39. Two important aspects of this figure are noteworthy: the current is

0

x (cm) Figure 38. Calculated conversion profiles for the synthesis of S i c with applied voltage 20V [72].

352


(E=14.2 V/cm)

0.0

0.5

1.0

1.5

2.0

x (4 Figure 39. Current density profiles for the synthesis of S i c [72].

confined to a very narrow zone; and the confinement zone is located in the leading edge of the combustion wave. The confinement of the current to the reaction zone is an obvious advantage when the energy efficiency of this form of activation is considered. However, this situation is not universally true but is dependent on the electrical conductivity of the product relative to the reactants. The implication of this is that in systems where the product is more conducting, the current is not confined to the reaction zone but is, at least partially, carried by the product. Both experimental and modeling studies have provided support for this expectation [6]. The application of a field has been shown to have the following effects in combustion synthesis: it makes possible the synthesis by SHS of materials which, previously, could not be prepared by this method [7&74]; influences the nature and the relative amounts of product phases [75,76]; makes possible the synthesis in relatively dense reactants [77]; and influences the microstructure of the product [78]. A few examples of these will be discussed here and more details on the others are provided in the referenced papers. An example of the benefit of the field is demonstrated by the synthesis of B4C-TiB2 composites. With a goal of producing hard and light materials for armor application, the desire is to increase the relative content of the carbide phase in this composite. With normal SHS, composites with a B4C/TiB2molar ratio, y , of 50.5 can be synthesized without preheating the reactants [79]. While composites with higher values y can be synthesized by preheating the reactants, the product contained other extraneous phases (e.g. TIC). With application of a field, composites with y values up to 5 could be synthesized [80]. Figure 40 shows the effect of the field on the wave velocity in the synthesis of B4C-TiB2 composites with 1 5 y 5 5. As in other field-activated combustion synthesis, a minimum (threshold) field value was necessary to propagate the combustion wave. The threshold value depended on the value of y , as seen in Fig. 41. Higher threshold values were required for higher molar ratio 01) values. The products in all cases contained the two desired phases only. Since the imposition of the field is a requirement for the synthesis of composites

2.4 Synthesis by Field-Activated Selfpropagating High-temperature Synthesis

3 53

H B,C/TiB,=I 0 B,CmiB,=2 0

0

0

20 30 40 Voltage ( V )

10

B,C/TiB,=3 B,C/TiB,=5

50

60

Figure 40. Effect of field on the wave velocity for the synthesis of yB4C-TIC composites with 1 5 y 5 5 [SO].

with y > 0.5, the removal of the field during wave propagation affords the opportunity to investigate the mechanism of formation of the two phases from elemental reactants. Figure 42 shows the X-ray diffraction (XRD) patterns obtained from a B4C-TiB2 composite with y = 1 in which the wave was frozen by the removal of the field [80]. The upper pattern was obtained from the region behind the combustion zone, and as can be seen it contains only the peaks for B4C and TiB2, the product of synthesis. The middle pattern is for the combustion zone itself and shows the presence of TiB2 and carbon (no peak for boron is expected since it was used as an amorphous powder). Finally, the lower pattern is for the area ahead of the wave and shows the presence of the reactant elements only. These observations indicate that the formation of TiB2 is the first step in the composite formation, an expected conclusion in view of the large difference in the thermodynamic driving force for the formation of the two compounds. It should be recalled that B4C cannot be formed without field activation. 2 2

F

2 0

Y

-o m

1 8

C

>

1 6

8 0

1

2

3

4

B,C/Ti B, Molar Ratio

5

6

Figure 41. Effect of the molar ratio B,C/TiC on the threshold voltage in the combustion synthesis of B,C-Tic composites.

2 SelfiPrapagating High-Ternperuture Synthesis of Hurd Materials

354

reacted region 300

100

TiR,

1

'fi

unreacted region

0

25

30

35

40

45 50 2 theta

55

60

65

Figure 42. X-ray diffraction patterns of frozen reaction zone with analyses for the reacted region, combustion zone, and unreacted region [80].

Another example of the role of the field in synthesis is that related to the preparation of tungsten silicides [75]. The two stable silicides, WSi2 and W5Si3, have relatively low adiabatic combustion temperatures (1459 and 954 K, respectively) and thus cannot be synthesized by SHS. However, both can be synthesized by field activation. For WSi2, a combustion wave can be initiated above a threshold value of about 15Vcm-' and the product is the desired phase. On the other hand, the case of W5Si3 is more complex. Although a combustion wave can be initiated above a threshold field of about 14Vcm-', the product does not contain the desired phase unless the field is above about 19 Vcm-I. Microstructural observations on the products for the case of W5Si3showed that under the higher field the product contained a liquid phase and it was proposed that the formation of the desired phase was aided by the higher diffusion rates in the liquid phase. The formation of composites and solid solutions in the binary A1N-SIC offers another important example of the role of the field in the synthesis of materials. The two compounds form extensive solid solutions with the hexagonal 2H structure above about 1960°C [81]. The spinodal decomposition of these, at lower temperatures, gives rise to the formation of composites modulated at the nanoscale [82]. The common method to synthesize the solid solutions is to heat mixed powders of AlN and Sic (aor p) at 2100°C for up to 16h. As will be seen here, using field-activated synthesis for the reaction S3N4 4A1+ 3C, the formation of solid solutions can be effected in less than a minute [76,83]. Although this reaction is highly exothermic (with an adiabatic temperature of 2504 K), it does not sustain a self-propagating combustion wave without the application of a field above a threshold value of 8.0 V cm-' . At this field, however, the reaction is not complete, with the product containing SiC-rich and A1N-rich 2H solid solutions and Si and C . The reaction becomes complete at a field of 12.5Vcm-' but the product is a composite of two solid solutions. When the applied field is 25.0 V cm-I, the product is a single phase solid solution. Careful analysis of the product phases showed that their compositions changed with the strength of the applied field, as shown in Fig. 43. As the field increased, the compositions of the two phases moved closer to each other

+

2.4 Synthesis hy Field-Activated SelJlpropagating High-temperature Synthesis

3 55

0.8 0.6

zd

0.4

0 WDS-AIN

o WDS-Sic

0.2

n

XRD-AIN x XRD-Sic A

I

12.5

I

20

25

30

Field (V/cm) Figure 43. Effect of the applied field on the phase composition of combustion-synthesized AIN-SIC composites and solid solutions [76].

Figure 44. X-ray dot maps of A1 and Si in AIN-SIC solid solutions prepared under fields of 25 (left) and 30Vcm-' (right) [76].

356

2 SeIf-Propagating High-Temperature Synthesis of Hard Materials

and at 25 V cm-' , the compositions are the same, indicating the formation of the solid solution. Increasing the field to 30 V cm-', had no effect on the XRD generated composition, but a more significant effect on the elemental distribution, as seen in Fig. 44. Although the XRD results indicate that the product with 25 V cm-' is single phase, the elements (A1 and Si) are not uniformly distributed in the sample. However, a nearly complete uniform phase is obtained when the field was 30 V cm-' . These observations are further evidence to the role of the field in the synthesis of materials.

2.6 Selected Recent Examples of Synthesis of Hard Materials During the past few years, the use of the SHS method to obtain technologically important hard materials has been primarily focused on the synthesis of composites, mainly TiC- and TiB2-based materials. Only in a few cases has the synthesis of monolithic hard compounds been considered. In these cases the emphasis is on the development of new techniques (such as the field-assisted combustion synthesis described in the previous section) or on the understanding of the reaction mechanism [5,7,8,84,85]. While the primary goal is to synthesize dense materials, the production of porous materials, to be used mainly as a source of powders, has also been investigated. In this case the powders can be the final desired product (e.g. for use as abrasives) or can be subsequently densified. The emphasis on the synthesis of dense materials is motivated by the desire to develop new processes to replace the traditional techniques for the production of, for example, cutting tools. The traditional methods for the fabrication of these materials typically involve powder metallurgical techniques which are both labor and energy-intensive. The extremely high temperatures, typically attained in the SHS process, makes it feasible, at least in principle, to synthesize and densify pure carbides, through the simultaneous application of pressure to the combustion products when they are still above the ductile-brittle transition temperature, which for T i c is around 1700°C. Early attempts to utilize this approach were only partially successful, as they were plagued by the macrocracking of the final products, probably due to the presence of residual thermal stresses [86]. An example of such an approach is the work of Wang et al. [87], who used explosive compaction on combustion synthesized TIC and TiB2. The dynamic compaction followed the SHS process with a time-delay of 5-20 s, with the best results being obtained with a delay time of 5 s. The compacts produced had relative densities higher than 99% for TiB2 and 96.8% for TIC. The microhardness, elastic modulus, and compressive strength of TiB2 were comparable to those reported for hot-pressed materials. For the case of Tic, the microhardness and elastic modulus (NO0 and 46 700 kg mm-2, respectively) were equivalent to those obtained from hot-pressed samples, but the compressive strength (103 kg mm-') was lower, presumably because of extensive cracking in the SHS-prepared samples. Samples of TiBz were less prone to cracking, probably because the melting point of TiBz (2900°C) is exceeded during the synthesis while

2.6 Selected Recent Examples of Synthesis of Hard Materials

357

that of TIC (3140°C) is not likely to be reached during real (i.e. nonadiabatic) conditions. Moreover, it was observed that the grain boundary strength of the TIC and TiB2 prepared by the SHS-dynamic compaction method were relatively weak. This conclusion was based on observations of grain boundary failure during hardness measurements and was explained by a formation of microcracks along grain boundaries during the rapid cooling. Support for this conclusion was provided by significant improvements in mechanical properties after annealing for several hours at 1500 and 2000°C. As pointed out above, the existence of cracks is related to the rate of cooling subsequent to synthesis and densification. The use of a pressure medium in the 'pseudo-HIPing' approach is believed to ameliorate this problem by decreasing the rate of cooling of the densified sample [65]. In order to improve the mechanical properties of such monolithic hard materials, a second phase is usually added. In many cases the second phase is a metal and thus the formation of a cermet is accomplished. The addition of a ductile phase acts to reduce the residual stress and to increase the fracture toughness of the material. The characteristics of the combustion processes aimed at the formation of metalceramic composites based on T i c and TiB2 have been investigated in several recent studies. Capaldi and Wood [88] investigated the SHS processes of synthesizing TIC cermets with the metallic component being iron, nickel, copper, and manganese. Notable differences in the process characteristics were observed, depending on the nature of the metal additive (binder). Figure 45 shows the dependence of wave propagation velocity on combustion temperature for all four metals. These results imply differences in the activation energy of the synthesis process, see Eq. (6). Contrary to what may be attributed to the role of the metal as an inert component, these results suggest that the metal plays a more direct role in the synthesis. Thus differences between the metals in terms of influencing the kinetics of the process may arise from possible differences, for example, in the formation of intermediate intermetallic phases, or differences in the kinetics of a dissolution-precipitation process responsible for the formation of the ceramic component [88]. The possibility of formation of low melting phases, represented

9

5 -2

3

u

-2

-8.0 -8.2 -8.4 -8.6 -8.8 -9.0 -9.2 -9.4 -9.6 -9.8

- 10.0 3.5~10' 4.0~10' 4 . 5 ~ 1 0 5.0~10' ~ 5.5~10.~ I/T, (I/K)

Figure 45. Apparent activation energy determination for Tic, Fe-Tic, Ni-Tic, and Cu-Tic [88].

358


by eutectics between Ti and the binding metal, can alter the kinetics of the dissolution-precipitation process involved in the combustion synthesis of carbides. Another difference in the synthesis of these cermets is represented by the variation in the ‘ignition temperature’. In some cases, for example in the system Ti-C-Fe, the ignition temperature is well below the melting point of the pure Ti metal, an occurrence which has been subject to much debate in the early SHS literature. The prevalent point of view has been that ignition occurs after the melting on the metal. In the work of Capaldi and Wood [88], however, when iron was used as the binder, titanium and iron react in the solid-state producing FeTiz which is a eutectic compound with a melting point of 1085°C. In the case of the use of Mn, the T i c that formed was carbon-deficient relative to that formed with other metals, probably because of the formation of manganese carbide. The synthesis of cermets in dense form have been reported by several authors. La Salvia et al. reported the results of high-speed forging of TiC-Ni [54] and TiC-NiMo [89] cermets. In both cases the high speed load was applied 5-10 s after the end of the combustion process, when the samples’ temperature was still above the ductile-to-brittle transition. In the first case cermets containing 5 , 25, and 30 mol% of the metal were prepared. The microstructure was generally characterized by spheroidal grains of T i c surrounded by a discontinuous Ni layer probably due to poor wetting of melted Ni on Tic. A substantial fraction of TIC particles exhibited bonding and necking, forming a continuous carbide skeleton. The samples with the lower metal content exhibited a certain porosity due to partial vaporization of Ni due to the high reaction temperature. The hardness and compressive strength of these samples were in the range of values of conventionally processed cermets. This indicates that the bonding between Ni and T i c is excellent. The addition up to 10 weight-% of Mo [89] was based on the suggestion that its presence improves the wettability of the metallic phase on Tic. In traditional processing this improvement in wettability decreases drastically the tendency of T i c grain growth, reducing the possibility of carbide-carbide grain contacts. In the combustion-synthesized/dynamically-compacted materials this effect was not so evident. Figure 46 shows that the influence of Mo content on the size of the TIC particles is not significant. On the other hand, evidence for incomplete solubility of Mo within the matrix is observed. However, a decrease in the number of microstructural flaws such as voids, interphase debonding, and binder microcracks, attributed to the improved wetting of the Ni-alloy binder on the carbide phase, is observed with an increase in Mo content. As a result, the compressive strength, bend strength, fracture toughness, and Young’s moduli, were slightly improved, while the Vickers microhardness seemed to be independent of the Mo content. Another example of synthesis of Tic-based compact cermets is reported by Xing et al. [90]. These authors describe the synthesis of dense NiA1-20 vol-% T i c composites through a reaction of a mixture of Ni, Al, Ti, and C powders in a hot press. The reaction was performed using the volume combustion mode: heating the green mixture until a reaction is observed in the entire sample. The samples were heated up to 1500°C under the influence of an applied pressure. The products were 98.9% dense but the T i c particles (0.2-1 pm) were not uniformly dispersed in the matrix. The combustion process had similar ignition characteristics to the


359

h

E

, I4.5

I

al

.*v1

-.U 0

z

.g

4.0

-

3.5

-

3.0

I I

i

0

2

4

6

8

10

12

Mo (wt%) Figure 46. Influence of the amount of Mo on carbide particle size for combustion synthesized/ impacted forged Tic-Ni based materials [SS].

system Ni-A1, with an ignition temperature of 588°C. This temperature is below the lowest eutectic temperature in the Ni-A1 system, suggesting that the combustion process commences through a solid state reaction. A different approach to the synthesis of dense hard materials is reported by Degnan and Wood [91]. They describe a process in which Fe-(W,Ti)C cermet powders, produced by the SHS method, are used as additives to molten iron for the production of a particulate-reinforced material with improved wear properties. The presence of the additive provided an enhancement of wetting between the metal matrix and the added particles. The product was a fairly uniform material which showed a considerably improved wear resistance over iron-based alloys. Sharivker et al. [92] followed a somewhat similar route by using SHS-produced TIC or Tio.97M00.03C as additives to produce metal-based hard materials. An interesting exception to the synthesis of dense materials through external pressure application is the work of Ohyanagi et al. [93] in which the formation of a surface dense TIC by a pressureless process is reported. Such a material was obtained using reactants with high green densities (6549%). The formation of a fully dense surface layer appears to be the result of the solidification of a supercooled solid solution of carbon in molten Ti. The presence of a nonequilibrium solution is suggested by the existence of an exothermal peak on the cooling curve which became evident in samples with green densities greater than 76%. The temperature corresponding to the maximum of this peak decreased when the cooling rate was increased. The surface became completely dense with a texture typical of solids obtained from solidification of supercooled liquid. The presence of the liquid only at the surface was explained by a combination of capillary and gas expulsion effects [93]. In the case of borides, the production of pure dense materials is more difficult because TiB2 does not deform plastically even at very high temperatures due to its intrinsically high concentration of Peierls barriers to dislocation movement. Recent investigations on borides have been totally devoted to the synthesis of composites either metal-ceramic or ceramicseramic. Woodger et al. [94,95]

360


2000, h

5 1900

:1800

*

$

g2

1700

c

.gs D 3 u

1600 1500 1400-

I

20

I

50 60 70 Percent relative green density

30

40

80

Figure 47. Influence of green density on the combustion temperature for the synthesis of Al-HfB2 composites for samples with 60 vol-% Al [94].

investigated the formation of dense materials through liquid metal infiltration obtained by adding large amounts of metal to the reaction mixture. The infiltration was done with pure A1 or with Ni3Ti intermetallic compound. In the case of HfB2+ Al, up to 70 vol-% A1 was added to the reacting mixture. An unusual effect of the green density of the reactants on the combustion temperature and propagation rate was observed, Figs 47 and 48. Initially, as the relative density increased, both the reaction temperature and propagation rate decreased, but with still higher densities, these parameters increased dramatically. Such a behavior is attributed to an assumed enhancement of liquid spreading at the reaction front for reactants with high green densities. However, the presence of a large amount of liquid aluminum does not lead to the formation of highly dense products. The relative density of the products is primarily related to the density of the green compacts, and it decreases with the amount of the metal. Furthermore, sedimentation

*

h

$

v

0.35

.-3

-

0

.>”

s

0.30

0.25

-

--

20

30 40 50 60 70 Percent relative green density

80

Figure 48. Influence of green density on the wave velocity in the synthesis of Al-HfB2 composites for samples with 60 vol-% Al [94].

2.6 Selected Recent Examples of’Synthesis of Hard Materials 1600 1

361

T

92 1550 f 1500 2 1450 2

-

C

2

. 3

1400 -

P

3E

1350

-

I

35

40

,

I

I

I

50 5 5 60 65 70 Percent relative green density 45

75

Figure 49. The influence of green density on the combustion temperature for the synthesis of Ni,TiTiB2 composites. Samples containing 70 vol-% Ni,Ti [95].

effects, due to the large density difference between A1 and HfB2, were observed together with the presence of small amounts of Hf-A1 intermetallic phases. In the case of TiB2,cermets with up to 90 vol-% Ni3Ti have been synthesized from elemental reactant powders. An unusual dependence of the combustion temperature on relative green density was also observed, Fig. 49. The bimodal nature of the combustion temperature curve was explained by a two-mechanism process. The first peak in combustion temperature is likely to be related to the highly exothermic TiB2 reaction in the presence of a small amount of diluent metals. The second peak can be associated with the improved reaction kinetics resulting from the enhanced capillary spreading of the molten phase. The presence of the molten phase makes possible the synthesis of samples with very high relative densities. As shown in Fig. 50 a rapid increase in relative density with metal content is observed up to 85% relative density, then this is followed by a marked decrease, probably related to the decrease in the reaction temperature as a consequence of increased dilution of the reacting mixture. The synthesis of a similar cermet has been reported by Yang and Petric [96] and Yi and Petric [97] who used elemental mixtures to synthesize Ti3A1(Nb)-TiB cermets. In this case the combustion process was reported to be initiated by the reaction between Ti and Al, which subsequently ignites the combustion between Ti and B. A further example of production of boride-based fully-dense materials is recently reported by Olevsky et al. 199). The product, a TiB,-TiN-Ni cermet, a material with a unique combination of high temperature hardness and stability and with a reasonable ductility/fracture toughness. Other composite materials based on TIC and TiB2 have been produced recently through the SHS process. Attempting to develop a process for the production of ceramic armor, Lis et al. [55] considered the combustion synthesis of TiB2-Tic composites. Such composites have a lower hardness than either TiB2 and T i c single phases, but a fracture toughness and wear resistance which are significantly higher than the corresponding values of the single phase materials. In order to

362

40

2 SeEf-Propagating High-Temperature Synthesis of Hard Materials

'

'

50

I

I

I

I

60

70

80

90

Volume percent Ni,Ti Figure 50. The effect of Ni3Ti content on final density of Ni,Ti-TiB, composite. All samples were 60% relative green density [95].

reduce the cost of the raw materials, these composites were obtained through the use of Ti and B4Cpowders as reactants with the anticipated product containing a TiB2/ T i c molar ratio of 2. Under such conditions, the adiabatic temperature is high enough to reach the lowest eutectic temperature (T, = 2520°C) making it possible to attain high product density. The combustion process was found to be strongly dependent on the reactants' particle size. A stable combustion process could be obtained only for B4C and Ti powders with particle sizes below 10 and 49pm, respectively. The synthesized composite was then densified by pressureless sintering at temperatures between 2000 and 2100°C or by hot-pressing at temperatures between 1800 and 2000°C using a ram pressure of 200 atm. The sintering time for the combustion synthesized samples was considerably shorter than for mixtures of commercial T i c and TiB2 powders, and the microhardness of the product (Hv = 29.5 GPa) was higher than values reported in the literature. The addition of alumina to TiB2 or TIC has become an attractive approach in recent years. The addition of alumina to titanium carbide or boride improves the oxidation resistance and results in composites which possess superior mechanical strength and fracture toughness, without a significant drop in their overall microhardness. Moreover, these composites can be obtained through SHS using, as reactant, the relatively inexpensive titanium oxide instead of the metal. In a recent example, a ceramic-ceramic composite to be used as a grinding material has been prepared by Vershinnikov et al. [98] using the reaction 3TiO2 + 3Bi2O3+ lOAl = 3TiB2 5A1203.Kecskes et al. [50] showed that the dynamic consolidation technique can be used to produce dense TiB2-A1203 composites through SHS using the previous thermite reaction. Sample densities of 90% can be obtained by varying the reaction ignition-consolidation delay time. The TiB2 grain size was found to be related to that of the precursor Ti02. A detailed study on the influence of several experimental parameters on product microstructure was reported by Bowen and Derby [loo] for TiC-A1203 composite always obtained through a reaction between TiOz and A1 plus C. The combustion kinetics and the product microstructure

+


O

0

20 30 40 Aluminium particle size (pm)

10

363

h

50

Figure 51. Combustion wave velocity against A1 particle size for the combustion synthesis of TICA1203 composites [100].

result is strongly dependent on the reactant particle size. The wave propagation rate, for example, was strongly dependent on A1 particle size, but less significantly on Ti02 particle size. In both cases a decrease of propagation rate is observed when the particle size is increased, as seen in Figs 5 1 and 52. In contrast, when the carbon particle size was increased from 0.02 to 7.2 pm, the wave propagation increased but then decreased for larger particle size, Fig. 53. This behavior was explained by an assumed barrier role played by the extremely fine carbon, decreasing the titanium-aluminum particle contact, a necessary first step in the reaction, Fig. 54. The effect of diluents on the reaction products was also considered in the cited work [loo]. When A1203 was added as a diluent it tended to induce cracking in the final product, probably because the combustion temperature decreased below the melting point of alumina. On the other hand, the addition of A1 as a diluent produced an unexpected increase in wave velocity with increasing diluent content, an observation that can be explained

2 '

0

5 10 15 Titania particle size (Fm)

20

Figure 52. Combustion wave velocity against TiOz particle size for the combustion synthesis of TiC-A1203 composites [ 1001.

364


0

10

20

30

40

50

Carbon particle size (pm) Figure 53. Combustion wave velocity against C particle size for the combustion synthesis of TiCiA120, composites [loo].

by an enhanced diffusion process or heat transfer in the presence of a higher liquid content. The synthesis of TiC-AI2O3-Al composites by SHS has also been investigated by Yi et al. [loll and Kim et al. [102]. In both investigations the products were densified at the end of the combustion reaction using a hot-press system in one case [loll and a weight-drop apparatus in the other [102]. In a few cases the SHS process was used as source of heat in order to achieve rapid densification. An interesting result in this regard is the work of Li et al. [lo31 and Miyamoto et al. [46], in which fully-dense, multifunctionally-graded A1203-TiCNi-TiC-A1203 materials were produced using a HIP compaction method. The

Figure 54. Schematic diagram showing how small carbon black particles act as a barrier to TiO2-A1 particles contact [loo].

2.6 Selected Recent Examples of Synthesis of’ Hard Materials I

2

3

4

5

6

365

7

vent

N,

Figure 55. Experimental set-up for SHS Cr7C3coating preparation: 1 =pressure gauge, 2 = reactor, 3 = heater, 4 = container, 5 =igniter, 6 =reactant charge, 7 = substrate [106].

samples, represented by a layered compact, were surrounded by a large amount of Si powder and were ignited in high pressure nitrogen. The silicon burns in a nitrogen pressure of above 3 MPa generating a high reaction heat and a temperature of over 2300 IS.The green body was densified within a few minutes. Because of the graded composition, the products had a stress distribution of compression/tension/compression which resulted in a desired compressive stress in the outer ceramic layers. Thus far we have considered only the synthesis of powders or bulk densified materials. However, the SHS approach has been successfully applied to the production of thick protective layers. A typical example of such a process is the development of technologies for the application of ceramic linings on hollow cylindrical bodies by means of thermite reactions under the influence of a centrifugal force [104]. A related technology, usually referred to as SHS cast-coating, has been developed in the former USSR [105]. Several types of coatings have been produced in this way, including carbides, borides, and silicides. A recent example of this approach is reported by Lin and Xue [106]. In this work a layer of Cr7C3-Cr cermet was produced on a flat substrate of carbon steel via the SHS cast route. The synthesis is based on the thermite reaction 7Cr03 14A1+ 3C = Cr7C3 7A1203. The experimental setup is shown in Fig. 55. A layer of the reactants mixture, a few millimeters thick, is compressed over the substrate, and a layer of an igniter powder is placed on top of it. The entire assembly is then heated under an atmosphere of nitrogen gas until a volume combustion is obtained. The two products of the thermite reaction are obtained in the molten phase. Since the density of the chromium compound is higher than that of alumina, a thick layer of Cr-containing phases is formed in contact with the substrate. After cooling, the loose alumina layer on top is easily removed. The thickness and composition of the green layers, the gas pressure, and reactants preheating temperature were found to strongly influence the characteristics of the final layer. Figure 56 shows that a strongly adherent layer could be obtained only when using either high preheating temperatures or high gas pressure, with a complex structure of the green compact, represented by three layers with different proportions of the reactants. The best results were obtained with a triple reacting layer, a 300°C preheating temperature and a 5-10MPa gas pressure, or with a 200°C preheating temperature and a 15 MPa pressure. The surface coating was primarily composed of Cr7C3and Cr with traces of Cr23C6.Transverse sections

+

+

366


(A)

(B)

EI13 Non of any product is formed on the substrate U Only some droplets are formed on the substrate IT3 Only some islands are formed on the substrate About 60% area is covered 111 Covered completely, but not smootly Covered completely and smootly Figure 56. Effect of preheating temperature and ambient pressure on the formation of SHS Cr7C3 coating on Cr: (A) single layer of reactants; (B) triple layer of reactants [106].

show that the composition of the layer is not homogeneous: the Cr content increases and the C content decreases with distance from the top of the coating. Concomitantly, the mechanical properties change with distance, with the maximum hardness being measured on the top of the surface. The adhesion of the coating to the substrate was found to be exceptionally good, even after several thermal cycles. Considerable interest has been generated during the past few years in the synthesis of diamond containing materials through SHS. In this regard the advantage of SHS in avoiding the drawback of the traditional methods of production of diamond-containing tools has been highlighted. These drawbacks relate to the necessity to use long reaction times at very high pressures. In the SHS process, reaction times can be reduced (from several hours in the traditional methods) to a few seconds. Furthermore, the suitability of the SHS process for the production of functionally-graded materials is an additional attractive feature. However, there is one possible drawback in the use of the SHS process: the extremely high reaction temperatures. Diamond transforms into graphite above about 14001500°C under an inert gas atmosphere or in vacuum. As a result, it is of concern that the mechanical characteristics of diamond might degrade during a typical SHS thermal cycle. Theoretical analysis of the combustion process in diamondcontaining mixtures have been proposed by Padyukov et al. [107]. The extent of diamond grain preservation was shown to depend more on the time the diamonds stayed in the combustion zone than on the absolute value of the combustion temperature. Surface graphitization, on the other hand, increases with the combustion temperature. Three different approaches have been proposed in order to reduce the effect of the combustion temperature on the diamond degradation [108]. In the first approach two-layer samples are used, as shown in Fig. 57. One contains the pure reacting mixture, for example Ti + B or Ni + Al, while the second contains the same mixture diluted with diamond powder. The first layer, without diamond acting as an inert diluent, has a more rapid combustion rate and acts to increase the rate of combustion in the second layer. This results in a reduction of the dwell time of the diamond in the high-temperature region. Moreover, such a procedure


367

Figure 57. Schematic of the bilayer pellet used for the SHS synthesis of diamond containing materials [108].

makes possible the incorporation of higher amounts of diamond in the mixture while still maintaining a self-sustaining process. Another approach is based on the feasibility of producing a protective reducing atmosphere during the combustion synthesis process. This can be accomplished by the addition to the reactants of a hydride with a dissociation temperature well below the reaction temperature. Titanium hydride (TiH2) is an example of such an addition. For the system Ti B diamond an optimum concentration of 5 weight-% of the hydride was shown to preserve the grain strength of the diamond, Fig. 58 [108]. A third modification aims at lowering the reaction temperature by the use of less exothermic reactions. The use of an intermetallic formation process, particularly in the Ni-A1 system, is an example of this. The materials used for the incorporation of diamond fall in two categories, corresponding to two different methods for fixing the diamond to the matrix [47]. In one case the diamond is retained only mechanically, while in the second case a sort of chemical fixing is obtained through formation of a covalent bond between the diamond and some component of the matrix. Metal carbides are generally used in order to form covalent bonds with diamonds. A metal is also sometimes

+ +

z

a"

15 14

I

0

10

I

I

20 30 Mass % ,TIH,

I

40

I 50

Figure 58. Dependence of diamond grain strength in combustion products of Ti + B + 10 weight-% diamond system on concentration of TiH2 additive [108].

368


included in the matrix in order to improve the mechanical properties of the product. Intermetallics (Ni-A1 or Ni-Al-Cu-Ti), TiB2, and TiB2/Si represent examples of a matrix of the first kind, while Tic-Si, Tic-Ti-A1, and Tic-TiN-Ni-Mo-A1 represent a matrix of the second kind. The densification of the products has been obtained using the typical approaches already used for other SHS processes such as hot-pressing, hot-isostatic pressing, pseudo-HIP, explosive consolidation, and high-velocity forging. Ohyanagi et al. [47] obtained good results with a diamond content up to 70 mol% (34 weight-%) using dynamic pseudo-isostatic compaction. Similar results are reported by Padyukov and Levashov [109]. SHS reactions have also been used in the metal brazing of diamonds on cutting tools [110]. In this case an SHS process is used to supply the heat necessary for brazing a single particle of diamond on a copper-tungsten plate through a silver, copper, and titanium alloy. A pellet of Ti + C was placed under the tungstencopper plate and then ignited. The speed of the thermal cycle ensured that no damage to the diamond occurs. It has been suggested that a supercooled liquid forms as a result of the rapid rate of cooling under such experimental conditions. The presence of the liquid brazing alloy even at relatively low temperatures ensures the process of stress relief.

Acknowledgment This work was supported by a grant from the (US) Army Research Office, ARO.

References 1. Z. A. Munir and U. Anselmi-Tamburini, Self-propagating exothermic reactions: The synthesis of high temperature materials by combustion, Mater. Sci. Rep. 1989, 3, 277-365. 2. Z. A. Munir, Synthesis of high temperature materials by self-propagating combustion methods, Ceramic Bull. 1988, 67, 342-349. 3. V. M. Shkiro, V. N. Doronin, and I. P. Borovinskaya, Study of the concentration structure of the titanium-carbon combustion wave, Combust. Explos. Shock Waves, 1981, 16,270-376. 4. Z. A. Munir and N. Sata, SSH diagrams: theoretical analysis and experimental observations, Int. J. SHS, 1992, 1, 355-365. 5. Z. A. Munir and W. Lai, The SHS diagram of Tic, Combust. Sci. Technol. 1992,88,201-209. 6. A. Feng and Z. A. Munir, The effect of product conductivity on field-activated combustion synthesis, J . Am. Ceram. SOC.1997,80, 1222-1230. 7. S. Zhang and Z. A. Munir, Spin combustion in the system nickel-silicon, J. Mater. Sci. 1992, 27,4789-5794. 8. M. E. Grami and Z. A. Munir, The effect of nitrogen pressure and diluent content on the combustion synthesis of TiN, J. Am. Ceram. SOC.1990,73, 2222-2227. 9. S. D. Dunmead, Z. A. Munir, and J. B. Holt, Temperature profile analysis in combustion synthesis: I, theory and background, J . Am. Ceram. SOC.1992, 75, 175-179. 10. S. D. Dunmead, Z. A. Munir, and J. B. Holt, Temperature profile analysis in combustion synthesis: 11, experimental observations, J. Am. Ceram. SOC.1992, 75, 180-188.

References

369

1 1. L. L. Wang and Z. A. Munir, Kinetic analysis of the combustion synthesis of molybdenum and titanium silicides, Metall. Trans. 1995, 26B, 591-601. 12. S. D. Dunmead, Z. A. Munir, and J. B. Holt, Gas-solid reactions under a self-propagating mode, Solid State Zonics, 1989, 32/33, 47&480. 13. S. Deevi and Z. A. Munir, The mechanism of synthesis of titanium nitride by self-sustaining reactions, J. Mater. Res. 1990, 5, 2177-2183. 14. T. Boddington, P. G. Laye, J. Tipping, and D. Whalley, Kinetic analysis of temperature profiles for pyrotechnic systems, Combust. Flame, 1986, 24, 359-368. 15. J. B. Holt and Z. A. Munir, Combustion synthesis of titanium carbide: theory and experiment, J. Mater. Sci. 1986, 21, 251-259. 16. K. A. Philpot and Z . A. Munir, The influence of pre-ignition processes on the combustion synthesis of nickel aluminides, in Materials Processing by Self-Propagating High- Temperature Synthesis, K. A. Gabriel, S. G. Wax, and J. W. McCauley (Eds), 1987, p. 105. 17. L. L. Wang, Z. A. Munir, and J. B. Holt, Synthesis of Al20,-B4C composites via a thermitebased combustion reaction, J. Mater. Synth. Proc. 1994, 2, 227-237. 18. L. L. Wang, Z. A. Munir, and J. B. Holt, Formation of MgO-B4C composites via a thermitebased combustion reaction, J. Am. Ceram. SOC.1995, 78, 756-764. 19. A. Makino and C. K. Law, SHS combustion characteristics of several ceramics and intermetallic compounds, J. Am. Ceram. Soc. 1994, 77, 778-786. 20. A. P. Hardt and R. W. Holsinger, Propagation of gasless reactions in solids - 11. Experimental study of exothermic intermetallic reaction rates, Combust. Flame, 1973, 21, 91-97. 21. A. I. Kirdyashkin, Y. M. Maksimov, and A. G. Merzhanov, Effects of capillary flow on combustion in a gas-free system, Combust. Explos. Shock Waves, 1981, 17, 591-595. 22. A. A. Zenin, A. G. Merzhanov, and G. A. Nersisyan, Thermal wave structure in SHS processes, Combust. Explos. Shock Waves, 1981, 17, 63-71. 23. I. P. Borovinskaya, A. G . Merzhanov, N. P. Novikov, and A. K. Filonenko, Gasless combustion of mixtures of transition metals with boron, Combust. Explos. Shock Waves, 1974,10,2-10. 24. A. G. Merzhanov and I. P. Borovinskaya, A new class of combustion processes, Combust. Sci. Technol. 1975, 10, 195-201. 25. N. P. Novikov, I. P. Borovinskaya, and A. G. Merzhanov, Dependence of the composition of the products and the combustion rates in metal-boron systems on the ratio of the reagents, Combust. Explos. Shock Waves, 1974, 10, 177-178. 26. A. Makino, N. Araki, and T. Kuwabara, Flammability limits, dilution limits, and effect of particle size on burning velocity in combustion synthesis of TIC, J S M E Znt. J. B: Fluids Therm. Eng. 1994, 37, 576582. 27. V. M. Shkiro and I. P. Borovinskya, Study of characteristics of the combustion of mixtures of titanium with carbon, in Combustion Processes in Chemical Technologyand Metallurgy, A. G. Merzhanov (Ed.), Akad. Nauk SSSR, OIKhF, Chernogolovka, 1975, p. 253. 28. Y. B. Zeldovich, Theory for the limit of the quiet flame propagation, J. Exp. Theor. Phys. 1941, 11, 159-169. 29. A. Makino and C. K. Law, Heterogeneous flame propagation in the self-propagating high-temperature synthesis (SHS) process: theory and experimental comparisons, Twentyfourth Symposium (International) on Combustion, The Combustion Institute, 1992, pp. 1883-1 89 1. 30. D. C. Halverson, K. H. Ewald, and Z. A. Munir, Influence of reactant characteristics on the microstructure of combustion-synthesized titanium carbide, J. Mater. Sci. 1993, 28, 45834594. 31. A. D. Bratchikov, A. G. Merzhanov, V. I. Itin, V. N. Khachin, E. F. Dudarev, V. E. Gyunter, V. M. Maslov, and D. B. Chernov, Self-propagating high-temperature synthesis of titanium nickelide, Sov. Powder Metall. Met. Ceram. 1980, 19, 5-8. 32. Z. A. Munir and J. B. Holt, The combustion synthesis of refractory nitrides, Part I: theoretical analysis, J. Mater. Sci. 1987, 22, 710-714. 33. M. E. Grami and Z. A. Munir, The effect of porosity on the combustion synthesis of titanium nitride, J. A m . Ceram. Soc. 1990, 73, 1235-1239. 34. Z. A. Munir, The mechanism of synthesis of high temperature materials by combustion processes, High Temp. Sci. 1990, 27, 279-293.

310


35. S. Zhang and Z. A. Munir, The combustion synthesis of refractory nitrides: Part 11, the synthesis of niobium nitride, J. Muter. Sci. 1991, 26, 3380-3385. 36. V. A. Gorshkov, G. N. Komratov, and V. I. Yukhvid, Production of cast higher chromium carbide using self-propagating high-temperature synthesis, Sov. Powder Metall. Met. Ceram. 1992.31.952-955. 37. U. Anselmi-Tamburini, F. Maglia, G. Spinolo, and Z. A. Munir, NijYSZ cermets from combustion synthesis: effect of process parameter on product microstructure, J . Am. Ceram. SOC.1998,81, 1765-1772. 38. Z. A. Munir, Analysis of the origin of porosity in combustion synthesized materials, J. Muter. Synth. Proc. 1993, 1, 387-394. 39. J. H. Matlock and J. P. Stark, Thermal diffusion ofvacancies in aluminum, Acta Metall. 1971, 14,923-929. 40. Z. A. Munir and L. L. Wang, The contribution of thermal migration to pore formation during SHS reactions, in Proceedings of the First US-Japan Workshop on Combustion Synthesis, Y. Kaieda and J. B. Holt (Eds), National Research Institute for Metals, Tokyo, Japan, 1112 January 1990, pp. 123-137. 41. S. D. Dunmead, Z. A. Munir, J. B. Holt, and D. D. Kingman, Simultaneous synthesis and densification of TiC/Ni4-A1 composites, J. Mater. Sci. 1990, 26, 241c-2416. 42. Z. Y. Fu, W. M. Wang, R. Z. Yuan, and Z. A. Munir, Fabrication of cermets by SHS-QP method, Int. J. SHS, 1993, 2, 307-313. 43. S. Adachi, T. Wada, T. Mihara, Y. Miyamoto, M. Koizumi, and 0. Yamada, Fabrication of titanium carbide cermets by high-pressure self-combustion sintering of titanium powder and carbon fiber, J. Am. Ceram. SOC.1989,72, 805-809. 44. P. D. Zavitsanos, J. J. Gebhardt, and A. Gatti, The use of self-propagating high-temperature synthesis of high density titanium diboride, in Combustion and Plasma Synthesis of HighTemperature Materials, Z. A. Munir and J. B. Holt (Eds), VCH Publishers, New York, 1990, pp. 170-178. 45. Y. Miyamoto, SHS/HIP compaction using inorganic fuels, Int. J. SHS, 1993, 1,479489. 46. Y. Miyamoto, Z. Li, Y. S. Kang, and K. Tanihata, Development of ceramic/metal heterogeneous composites by SHS/HIP compaction, Int. J. SHS, 1996, 5, 69-75. 47. M. Ohyanagi, T. Yoshikawa, M. Koizumi, S. Hosomi, E. A. Levashov, and I. P. Borovinskaya, Fabrication of diamond-dispersed cermets by SHSidynamic pseudo isostatic compaction (DPIC), Znt. J. SHS, 1995, 4, 387-394. 48. H. A. Grebe, A. Advani, N. N. Thadhani, and T. Kottke, Combustion synthesis and subsequent explosive densification of titanium carbide ceramics, Metall. Trans. 1992, 23A, 2365-2372. 49. L. J. Kecskes, T. Kottke, and A. Niiler, Microstructural properties of combustion-synthesized and dynamically consolidated titanium boride and titanium carbide, J. Am. Ceram. SOC.1990, 73, 1274-1282. 50. L. J. Kecskes, A. Niiler, T. Kottke, VK. V. Logan, and G. R. Villalobos, Dynamic consolidation of combustion-synthesized alumina-titanium diboride composita ceramics, J. Am. Ceram. SOC.1996,79, 2687-2695. 51. B. H. Rabin, G. E. Kroth, and R. L. Williams, Fabrication of TiC-AI2O3 composites by combustion synthesis and subsequent dynamic consolidification, Int. J. SHS, 1992, 1, 336341. 52. J. C. La Salvia, L. W. Meyer, and M. A. Meyers, Densification of reaction-synthesized titanium carbide by high-velocity forging, J. Am. Ceram. SOC.1992, 75, 592-602. 53. D. A. Hoke, M. A. Meyers, L. W. Meyer, and G. T. Gray, Reaction synthesis/dynamic compaction of titanium diboride, Metall. Trans. 1992, 23A, 77-78. 54. J. C. La Salvia, M. A. Meyer, and D. K. Kim, Combustion synthesis/dynamic consolidation of Tic-Ni cermets, J. Muter. Synth. Proc. 1995, 2, 255-274. 55. J. Lis, S. Majorowski, V. Hlavacek, and J. A. Puszynski, Combustion synthesis and densification of TiB2-TiC composite powders, Int. J. SHS, 1995, 4, 275-285. 56. H. Zhang, S. Yin, J. Zhong, and S. Guo, TiC-TiB2-AI2O3 ceramics made by combustion synthesis and pressureless sintering, Int. J. SHS, 1996, 5, 185-190. 57. G. Xu, H. Zhuang, J. Cai, and S. Xu, Microwave sintering of SHS-produced AIN powders, J. Muter. Synth. Proc. 1996, 4, 307-314.

References

371

58. B. Manley, J. B. Holt, and Z. A. Munir, Sintering of combustion-synthesized titanium carbide, in Sintering and Heterogeneous Catalysis, G . C. Kuczynski, A. E. Miller, and G. A. Sargent (Eds), Muter. Sci. Res. 1984, 16, 303-316. 59. M. Ouabdesselam and Z. A. Munir. The sintering of combustion-synthesized titanium boride, J. Muter. Sci. 1987, 22, 1799-1807. 60. J. Lee, N. N. Thadhani, and H. A. Grebe, Reaction sintering of shock-compressed Ti + C powder mixtures, Metall. Muter. Trans. 1996, 27A, 1749-1 759. 61. A. G. Merzhanov, A. S. Rogachev, A. S. Mukasyan, and B. M. Khusid, Macrokinetics of structural transformation during the gasless combustion of a titanium and a carbon powder mixture, Combust. Explos. Shock Waves, 1990, 26, 92. 62. J. Martinez and Z. A. Munir, Combustion synthesis of Zr0,-Cu composites during the application of a pseudo-hot isostatic pressure, unpublished work. 63. M. Shibuya, 0. Odawara, M. Ohyanagi, and M. Koizumi, Simultaneous synthesis and densification of TiN/Ti-Ni composites by SHS nitridation, Int. J . SHS, 1996, 5, 77-83. 64. N. N. Thadhani, R. A. Graham, T. Royal, E. Dunbar, M. U. Anderson, and G. T. Holman, Shock-induced chemical reactions in titanium-silicon powder mixtures of different morphologies: time-resolved pressure measurements and materials analysis, J. Appl. Phys. 1997, 82, 1113-1128. 65. R. V. Raman, S. V. Rele, S. Poland, J. Lasalvia, M. A. Meyers, and A. R. Niiler, The one step synthesis of dense titanium carbide tiles, J . Muter. 1995, 47, 23-55. 66. Z. A. Munir, I. J. Shon, and K. Yamazaki, Simultaneous synthesis and densification by fieldactivated combustion, US Patent, approved, 6 January 1998. 67. U. Anselmi-Tamburini and Z. A. Munir, The propagation of solid-state combustion wave in Ni-A1 foils, J . Appl. Phys. 1989, 66, 5039-5045. 68. S. C. Deevi, Diffusional reactions in the synthesis of MoSi2, Muter. Sci. Eng. 1992, A149,241251. 69. D. C. Halverson, B. Lum, and Z. A. Munir, The combustion synthesis of boride composites, Proceedings of the Symposium on High-Temperature Materials Chemistry - IV , Z. A. Munir, D. Cubicciotti, and H. Tagawa (Eds), PV 88-5, pp. 613-622, The Electrochemical Society, Pennington, NJ, 1988. 70. Z. A. Munir, W. Lai, and K. H. Ewald, Field-assisted combustion synthesis, US Patent No. 5,380,409; 10 January 1995. 71. A. Feng and Z. A. Munir, The effect of an electric field on self-sustaining combustion synthesis: Part 11. Field-activated synthesis of Sic, Metall. Muter. Trans. 1995, 26B, 587593. 72. A. Feng and Z. A. Munir, The effect of an electric field on self-sustaining combustion synthesis: Part I. Modeling studies, Metall. Muter. Trans. 1995, 26B, 581-586. 73. Z. A. Munir, Electrically-stimulated SHS, I n t . J . SHS, 1997, 6, 165-185. 74. Z. A. Munir, Field effects in self-propagating solid-state synthesis reactions, Solid State Ionics, 1997, 101, 991-1001. 75. S. Gedevanishvili and Z. A. Munir, The influence of an electric field on the mechanism of combustion synthesis of tungsten silicides, J . Muter. Res. 1995, 10, 2642-2647. 76. H. Xue and Z. A. Munir, Synthesis of AIN-SIC composites and solid solutions by fieldactivated self-propagating combustion, J . Europ. Ceram. Soc. 1997, 17, 1787-1792. 77. H. Xue, K. Vandersall, E. Heian, N. N. Thadhani, and Z. A. Munir, Initiation of selfpropagating combustion waves in dense Mo + 2Si reactants through field activation, to be published. 78. H. Xue and Z . A. Munir, The effect of field on the microstructure of MoSi, synthesized by field-activated combustion, to be published. 79. D. C. Halverson, Z. A. Munir, and B. Lum, Combustion synthesis of ceramic preforms for molten-metal infiltration, in Combustion and Plasma Synthesis ofHigh Temperature Materials, Z. A. Munir and J. B. Holt (Eds), VCH Publishers, New York, NY,1990, p. 262. 80. H. Xue and Z. A. Munir, Extending the compositional limit of combustion-synthesized BIGTiB2 composites by field activation, Metall. Trans. 1995, 27B, 475480. 81. A. Zangvil and R. Ruh, Phase relationships in the silicon carbide-aluminum nitride system, J . Am. Ceram. Soc. 1988,71, 884890.

372


82. M. Miura, T. Yoyo, and S. Hirano, Phase separation and toughening of SIC-AIN solidsolution ceramics, J . Muter. Sci. 1993, 28, 418883. H. Xue and Z. A. Munir, The synthesis of composites and solid solutions of SIC-AIN by fieldactivated combustion, Scripta Muter. 1996, 35, 979-982. 84. C. He and G. C. Stangle, The mechanism and kinetics of the niobium-carbon reaction under self-propagating synthesis-like conditions, J . Muter. Res. 1995, 10, 2829-2841. 85. H. Xue and Z. A. Munir, Field-activated combustion synthesis of TaC, Int. J . SHS, 1996, 5, 229-239. 86. J. H. Lee, N. N. Thadhani, and H. A. Grebe, Reaction sintering of shock-compressed Ti + C powder mixtures, Metall. Muter. Trans. 1996, 21A, 1749-1 759. 87. L. Wang, M. R. Wixom, and L. T. Thompson, Structural and mechanical properties of TiB, and TIC prepared by self-propagating high-temperature synthesis/dynamic compaction, J. Muter. Sci. 1994, 29, 524-543. 88. M. J. Capaldi and J. V. Wood, Production and characterization of Tic-containing materials by self-propagating high-temperature synthesis, J. Muter. Synth. Proc. 1996, 4, 245-253. 89. J. C. La Salvia and M. A. Meyers, Microstructure, properties, and mechanisms of Tic-Mo-Ni cermets produced by SHS, Int. J . SHS, 1995, 4, 43-57. 90. Z. P. Xing, J. T. Guo, G. Y. An, and Z. Q. Hu, Hot pressing aided exothermic synthesis and densification of NiAl and NiAI-TIC composites, Int. J . SHS, 1996, 5, 51-56. 91. C. C. Degnan and J. V. Wood, Processing and wear properties of a (W,TiC) particulate reinforced ferrous-based metal matrix composite, A. Poursartip (Ed.), Proc. 10th Int. Conf. Compos. Muter. Vancouver, Canada, 1995, pp. 77 1-778. 92. S. Y. Sharivker, I. P. Golovinskaya, I. N. Vorob’eva, and V. M. Bunin, Wear of tungstenless hard alloys based on SHS-titanium carbide, Powder Metall. Metal Ceram. 1996, 35, 55-57. 93. M. Ohyanagi, M. Kanno, and M. Koizumi, T i c combustion synthesis and fabrication of a body with a densified surface layer by the pressureless method, J . Muter. Synth. Proc. 1993, 1, 31 1-322. 94. T. C. Woodger, J. J. Moore, H. C. Yi, and J. Y. Guigne, Combustion synthesis of A1-HfB, metal matrix composites, T. S. Srivatsan and J. J. Moore (Eds), Proceedings of Processing and Fabrication of Advanced Materials V ,TMS, 1996, pp. 223-233. 95. T. C. Woodger, J. J. Moore, H. C. Yi, and J. Y. Guigne, Combustion synthesis of Ni3Ti-TiBz intermetallic matrix composites, T. S. Srivatsan and J. J. Moore (Eds), Proceedings of Processing and Fabrication of Advanced Materials V ,TMS, 1996, pp. 399408. 96. W. Y. Yang and A. Petric, Microstructure of the Ti3A1(Nb)/TiB composite produced by combustion synthesis, Metall. Muter. Trans. 1995, 26A, 3037-3043. 97. H. C. Yi and A. Petric, Combustion synthesis of Ti-Al-Nb matrix composites reinforced by titanium borides, J. Muter. Synth. Proc. 1994, 2, 161-167. 98. Y. I. Vershinnikov, N. S. Pesotskaya, S. S. Mamyan, and I. P. Borovinskaya, SHS of composite ceramic powder TiB,-AI,03 and its abrasive characteristics, Muter. Powder Technol. 1993, PTM93, 749-753. 99. F. Olevsky, P. Mogilevsky, E. Y. Gutmanas, and I. Gotman, Synthesis of in situ TiB2/TiN ceramic matrix composites from dense BN-Ti and BN-Ti-Ni powder blends, Metall. Muter. Trans. 1996,27A, 2071-2079. 100. C. R. Bowen and B. Derby, The formation of TiC/Alz03 microstructures by a selfpropagating high-temperature synthesis reaction, J. Muter. Sci. 1996, 31, 3791-3803. 101. H. C. Yi, H. J. Feng, J. J. Moore, A. Petric, and J. Y. Guigne, Combustion synthesis and mechanical properties of TiC/Al2O3-AI composites, Int. J . SHS, 1996, 5, 39-50. 102. C. W. Kim, C. S. Kim, S. C. Koh, and D. K. Kim, Combustion synthesis and densification of the TiC-AIzO3-AI composites, Int. J . SHS, 1996, 5, 159-168. 103. Z. Li, K. Tanihata, and Y. Miyamoto, Fabrication of multifunctional gradient materials in the system of A1~03/TiC/Ni/TiC/Alz03 by SHS/HIP, Advanced Materials ’93, N. Mizutani (Ed.), Elsevier Science, Amsterdam, 1994, pp. 677-680. 104. 0. Odawara, Long ceramic-lined pipes produced by a centrifugal thermal process, J. Am. Ceram. SOC.1990,73, 629-633.

References

373

105. A. G. Merzhanov, Self-propagating high-temperature synthesis: twenty years of search and findings, in Combustion and Plasma Synthesis of High Temperature Materials, Z . A. Munir and J. B. Holt (Eds), VCH Publishers, New York, 1990, pp. 1-53. 106. L. Lin and H. Xue, Investigation of Cr,C,/Cr hard surfacing via SHS cast route, Znt. J . SHS, 1995, 4, 171-182. 107. K. L. Padyukov, E. A. Levashov, and I. P. Borovinskaya, Regularities of the behaviour of synthetic diamond in the combustion wave of self-propagating high-temperature synthesis, Eng. Phys. J . 1992,63, 577-583. 108. E. A. Levashov, I. P. Borovinskaya, A. Rogachov, M. Koizumi, M. Ohyanagi, and S. Hosomi, SHS: A new method for production of diamond-containing ceramics, Znt. J . SHS, 1993, 2, 189-201. 109. K. L. Padyukov and E. A. Levashov, Self-propagating high-temperature synthesis: a new method for the production of diamond-containing materials, Diamond Rel. Muter. 1993, 2, 207-21 0. 110. M. Ohyanagi, Y. Taketani, Y. Nakamura, E. Kamijou, and M. Koizumi, Instantaneous bonding of diamond and metal by brazing, Am. Ceram. SOC.Bull. 1993, 72, 8 6 8 9 .


3 Hydrothermal Synthesis of Diamond K. G. Nickel, T. Kraft, and Y. G. Gogotsi

3.1 Introduction Cubic diamond and hexagonal graphite are the thermodynamically stable condensed phases in the P-Tdiagram of carbon with diamond and graphite dominating at very high and low pressures respectively. Direct transformation between the forms is kinetically impeded due to the necessity of a complete reconstruction of strongly bonded covalent networks. Natural diamonds are supposed to have grown in the depth of the mantle 121 under high-pressure high-temperature (HPHT) conditions above the BermanSimon line [ 11, shown in Fig. 1. Although possible mechanisms of natural diamond formation are still debated, C-H-0 fluids seem to play an important role [3]. Reports on man-made diamond obtained by HPHT synthesis were first published in 1955 by General Electric [4]. Usually, metals able to dissolve carbon under HPHT conditions are used as catalysts and increase growth rates. Diamond crystals of several millimeters in size can be obtained in this way, but usually small grains for abrasives are produced. Direct conversion of graphite to diamond without catalyst in HPHT apparatus is possible, but uneconomical for industrial application. Direct transformation can be done by the detonation method and produces nanosized powders of diamond and diamond-like carbon [5]. Diamond films and coatings can be obtained via various chemical vapor deposition (CVD) methods at and below atmospheric pressure [6]. This highly metastable and kinetically driven diamond growth process requires the presence of atomic hydrogen, which selectively etches graphitic and amorphous carbon while diamond is unaffected. While HPHT, shock wave, and CVD synthesis of diamond have been commercialized, these methods have serious limitations. None of them allows economical growth of large single crystals and the diamond quality does not satisfy the requirements for electronic or gem applications. Since the quality and size of the best natural diamonds have not yet been reproduced in the lab, the search for a better method of diamond synthesis continues. Recent work demonstrates that diamond can be formed metastably under a variety of conditions. Related to the CVD process is the low-pressure solid-state source (LPSSS) process, where preformed mixtures of carbon and diamond seeds [7] or carbon and metal [8] are treated in a H2 plasma, thus forming compact diamond ceramics. Diamond particles have been found in combustion products [9], synthesized from fullerenes by pressurization at room temperature [ 101 or sputtering at high temperatures [ 111, obtained by decomposition of polymers with an appropriate structure [ 121, by heating carbon ‘onions’ with electron beam in a TEM [ 131, by chlorination of S i c [14], or by deposition from supercritical fluids [15-171.

3.1 Introduction

0

6

375

0

Temperature [“C] Figure 1. P-Tdiagram of the low-pressure, low-temperature labile equilibriums of carbon solution: 1 = graphite-diamond equilibrium line, 2 = glassy carbon-diamond transition line, 3 =range of pneumatolytic hydrothermal processes, 4 = oxidative corrosion of diamond, 5 = anticipated area of diamond hydrosynthesis, 6 and 7 =diamond synthesis from glassy carbon precursors, 8 = lowpressure, low-temperature hydrothermal homoepitaxy of diamond. Reproduced from [ 151 with permission from A. Szymanski.

The presence of small diamond particles has been confirmed independently in all cases using several methods of analysis. These recent findings became primarily possible due to the development of new analytical techniques that allow the detection of diamond in small quantities and with particle sizes down to a few nanometers. The formation of diamond particles in the range of the thermodynamic stability of graphite can be explained either by kinetic factors, or by a higher stability of small diamond particles compared to graphite, which has been controversially discussed [ 181. While some of the metastable routes to diamond mentioned above will probably remain a scientific curiosity, hydrothermal synthesis has attracted more attention due to its commercial potential. Synthesis from a supercritical water-based fluid is an excellent method for preparing powders, coatings, and single crystals of many different materials [ 191, which offers many advantages over other preparation methods because it is inexpensive, environmentally benign, reduces free energies for various equilibria, enables the synthesis of phases that may not be stable under other conditions, and permits production of composites from incompatible phases that can not be obtained by other means. It is the only commercial method of growing large quartz single crystals and synthetic emeralds. Thus, if applied to diamond, it may potentially yield diamond crystals, which might be competitive with the Cullinan

376

3 Hydrothermal Synthesis of Diamond

diamond. Rustum Roy calls the hydrothermal synthesis ‘The fourth revolution in diamond synthesis’ [20]. Although we think that it is too early to consider it to be a ‘revolution’, it is certainly worthy of attention and further evaluation as a potential method of diamond synthesis. The goal of the ongoing research is to make diamonds from a liquid phase which is several orders of magnitude denser than vapor, in the hope of increasing growth kinetics compared to CVD synthesis. In this chapter, we explain why hydrothermal synthesis deserves special consideration and present a review of the current state of the art in hydrothermal synthesis of diamond as a potential future method of making diamond or hard diamond-like sp3-carbon.

3.2 Evidence from Nature DeVries was among the first to suggest the possibility of the hydrothermal synthesis of diamond [3]. A hot fluid in the C-H-0 system, not quite saturated with carbon, should precipitate carbon as it cools down [21]. Depending on pressure and temperature, diamond, graphite [22], or poorly crystallized carbonaceous materials like anthraxolite [3] can form. Any supply of carbon rich C-H-0 fluids at a depth greater than about 150km in the lithosphere has a high probability of becoming diamondiferous as indicated by some investigations on diamond inclusions and diamond host rocks [23-251. C-H-0 fluids should be the base of the inorganic solutions from which graphite and diamond precipitate and grow in nature [3]. Little evidence for metastable growth of diamond in earth systems was found. Diamond should form only at temperatures and pressures above about 1100°C and 4.5 GPa, although a theoretical minimum set for diamond formation was calculated to be 2GPa and 1000°C [2]. However, it has been suggested [26] that natural epitaxial coarsening of diamonds could proceed in colloidal pneumatolytic-hydrothermal environments in nearsuperficial earth zones, because equilibrium data for material trapped as inclusions in diamond indicated closing temperatures of 600-800°C and pressures below 10 MPa [15] and fit to early theories of diamond formation [27]. Based on the petrologic circumstances mentioned in [26], an area of probable diamond hydrogenesis and hydrosynthesis was sketched (Fig. 1). Theoretical calculations and experimental studies [28] suggested that smaller, colloidal diamond particles of about 10nm may be stable at lower pressures and higher temperatures than macroscopic diamond. A so called ‘crystallite size effect’ [29] might account for metastable diamond nucleation and growth under hydrothermal conditions. Thus, the HPHT graphite-diamond equilibrium should only be applied to crystallites >100nm in diameter [28]. Based on observations of turbid rims of fibrous diamond and layers of the lonsdaleite polymorph on some natural diamonds [30] as well as of embryocrystals and phantoms in central parts of natural diamonds [311, a mechanism for homoepitactic diamond coarsening and crystal growth in nature was presented [32,33]. Assuming

3.3 Hydrothermal Synthesis

377

that diamond grows from a nucleus in the postmagmatic fluid, the growth rate should be controlled by the diffusion process of sp3-structural elements to the seed crystal surface resulting in thin colloidal cluster layers which may in turn form lonsdaleite or fibrous diamond. Due to coalescence, slow transformation to thick polycrystalline layers or monocrystals with cubic structure might occur. Coexistence, intergrowth and thus presumable paragenesis of crystals of diamond, quartz and a-and 0-Sic have been found in a diamondiferous kimberlite [34] where crystallization of quartz instead of coesite indicates maximum pressures of <2.0-2.8 GPa at temperatures of 1000-1200°C. This might be important in the view of diamond hydrosynthesis from silicon carbides in the metastable range.

3.3 Hydrotherma1 Synthesis Investigations of hydrothermal processes and development of hydrothermal equipment have their origin in the attempts to understand and mimic nature which produces thousands of well-crystallized minerals at very modest temperatures [19]. The hydrothermal processes in this paper deal with water or aqueous fluids in the supercritical range of water, above its critical point ( T , = 374"C, P, = 22.14MPa). This is opposed to 'just boiling in water', and supercritical processes dealing merely with hydrocarbons or halogenated hydrocarbons [35]. Most interesting applications of supercritical fluids occur at 1 < TITc < 1.1 and 1 < PIP, < 2 [36]. At these conditions, the fluid exists as a single phase with sufficient density to give high dissolving power and high diffusivity, facilitating mass transport. Because of the high compressibility of fluids near their critical point, physical and chemical properties change drastically and density and dissolving power can be tuned through changes in temperature and pressure. Supercritical water can act as an acidic or basic catalyst, and its reactivity can often be enhanced by autocatalysis from water-soluble reaction products [37]. The discussion of hydrothermal diamond synthesis is divided into two sections, dealing with synthesis from C-H-0 liquids and synthesis based on decomposition of silicon carbide, respectively. Both start with thermodynamic calculations in order to demonstrate the theoretical possibility of carbon formation before the experimental findings are summarized. Naturally, equilibrium calculations do not consider kinetic limitations.

3.3.1 C-H-0 System In many instances diamond synthesis in the C-H-0 system is related to diamond formation in nature as discussed in part 2. It was shown that precipitation of diamond can occur in the C02/CH4 system [38]. No reagent external to the C-H-0 system should be required for carbon precipitation while high water fugacities may exist at the site of diamond precipitation.

378


400 500 600 700 800 900 1.000 Temperature I 'C

Figure 2. Calculated temperature dependence of the major species in the carbon-water system for a C : H 2 0 molar ratio of 1 : 1 at 100 MPa.

3.3.1.1 Thermodynamic Analysis

Since experiments related to carbonaceous liquids summarized and discussed below were carried out in the presence of diamond seed, the calculations presented here [39] have been restricted to the water-diamond system, excluding graphite from the calculations. While CO and H2 are the main equilibrium species at low pressures (P< 1 MPa), at higher pressures (P> 2MPa), a 1 : 1 mixture of CO and H2 converts to a fourspecies assemblage involving C02, CH4, H20, and diamond, suggesting two independent reactions:

CO+H2 =iCH4+iCO;?,

(1)

co + H2 = Cdiamond+ H20.

(2) Combined, the overall reaction governing the system to a first approximation is

4

2CO + 2H2 = CH4 + C02 + Cdiamond + H20.

(3) The individual contributions of reactions (1) and (2) in reaction (3) at higher pressures (100MPa) are strongly temperature dependent (Fig. 2). It is clear that the equilibrium amounts run in pairs CH4 + C02, Cdiamond H20, and CO H2. The deviation at higher temperatures stems from the influence of the Boudouard equilibrium

+

+

+

2C0 = Cdiamond c 0 2 . (4) Clearly, at higher temperatures diamond will be oxidized to yield CO. Reaction ( 3 ) shows the dependence of the equilibrium composition on the C: H 2 0 ratio. In order to maintain the equilibrium, an increase in the amount of H 2 0yields less CH4, C02, and diamond and more CO and H2. The amount of diamond decreases with decreasing C :H 2 0 ratio. Accordingly, the temperatures for the


379

elimination of diamond from the system decrease with decreasing C : H 2 0 ratio, from 950°C at a C: H 2 0 ratio of 1 : 1.2, to 700°C at a C: H 2 0 ratio of 1 : 1.4 [39]. Repetition of the calculations for graphitic carbon gives a similar result with shifted boundaries (we need a lower C: H 2 0 ratio of approximately 1:2 to eliminate C at 700°C). Thermodynamic equilibrium calculations carried out for very high pressures and temperatures [40] are consistent with our calculations. Summarizing, the hydrothermal behavior of diamond is predicted to depend crucially on temperature and the C : H 2 0 ratio with a strong interaction of both variables while it shows a weak pressure dependence in the range of supercritical water.

3.3.1.2 Interaction between Diamond and Supercritical Water Knowledge of the hydrothermal behavior of different allotropes and forms of carbon is essential for optimization of growth parameters of diamond. Noticeable interaction between nanocrystalline diamond powder and supercritical water starts at about 700°C but even long-term high-temperature treatment (94 h, 850"C, 200MPa) did not cause complete dissolution [39]. The sharpness of the diamond Raman band increases with increasing reaction temperature and time, while the band position moves from the downshifted value of 1323cm-' for untreated nanocrystalline diamond powder to the typical diamond value of 1332 cm-* . This was explained by ordering and healing of the diamond structure. Electron paramagnetic resonance (EPR) measurements indicated that the broken carbon bond concentration decreased by a factor of 2 after 100 h at 750°C and 300 MPa. Hydrothermal treatment of mixtures of graphite and nanocrystalline diamond showed, that nanocrystalline diamond reacts with supercritical water faster than graphite [39]. Microcrystalline diamond powders showed a significant dissolution with weight loss and gas formation after a week at 750°C and 500 MPa [39]. Healing of diamond as found in the case of nanocrystalline diamond can be explained by redeposition of carbon from the hydrothermal fluid, which also explains the increasing formation of graphitic carbon with increasing temperature observed in these experiments. It was reported that hydrothermal treatment in the presence of a catalyst at 900°C and 1 GPa enhanced the graphitization of amorphous carbon [41]. Formation of etching pits has been observed on the surface of diamond single crystals after 15h at 800°C and l00MPa [39]. Long-term treatment at 800°C caused significant weight loss. Graphitization of the diamond surface was not observed. No weight change of the diamond single crystals occurred in the presence of nanocrystalline diamond powder or graphite. Deposition of carbon films on the walls of the gold capsules indicated the saturation of the fluid with carbon, and thus preservation of diamond. The interaction of artificial diamond single crystals with a high pressure water fluid at 900 and 1400°C under 5.2GPa yielded nondiamond carbon films [42]. Scanning electron microscopy (SEM) examinations revealed the formation of deep and shallow flat-bottomed trigons on diamond-( 11 1) faces at reaction temperatures above 1000°C.

380


Summarizing, the hydrothermal behavior of diamond depends strongly on the grain size. Noticeable dissolution of monocrystalline, microcrystalline and nanocrystalline diamond start at 800, 750, and 700°C, respectively. In agreement with thermodynamic calculations, complete dissolution of diamond was never observed at high C : H 2 0 ratios. Precipitation of graphitic carbon and healing of defective diamond crystals have been indicated. While some data suggest the formation of nanocrystalline diamond, no epitaxial diamond growth was observed.

3.3.1.3 Hydrothermal High-pressure High-temperature Synthesis Epitaxial growth of diamond on diamond has been demonstrated in the carbonwater system at 7.7GPa starting at 2000°C and yielding a considerable weight increase at 2200°C [40]. The newly grown diamond coats were tightly bonded to the seed crystals. The colorless and transparent crystals became black and opaque, indicating inclusions of graphite. Rounded corners and edges were sharpened, and rough and irregular growth steps and numerous pits were formed. No evidence of direct conversion of graphite to diamond was found. According to the thermodynamic calculations, it was concluded that graphite dissolved in the water rich fluid, forming COz and CH4. Subsequent carbon precipitation from the carbon-saturated fluid resulted predominantly in homoepitaxial diamond growth. 3.3.1.4 Synthesis outside the Range of Thermodynamic Stability

The range of thermodynamic stability is 30 MPa < P < 1 GPa. Hydrothermal treatment of glassy carbon and diamond seeds at 80&900"C and 140MPa in Au capsules resulted in a substantial increase of the diamond (1 1 1) reflection in X-ray diffraction (XRD) patterns [35,43]. Hexagonal crystal shapes were etched out of the substantially reacted glassy carbon but possibly new-grown diamond crystals were too small to be seen by SEM. Less diamond growth was found at higher temperatures. No diamond growth was observed using Pt capsules suggesting that hydrogen presence is essential for hydrothermal growth of diamond. Aggregates of small diamond crystals were obtained after experiments, when mixtures of glassy carbon, quarter micron diamond seed and different metal powders (Ni, Pt, Fe) were hydrothermally treated at 800°C and 140MPa in gold tubes [17]. In typical runs, the original black powder charge turned gray and contained some large aggregates with irregular morphology. The aggregates, which were up to tens of microns in size, contained no metal and were built of diamond particles of 0.1-1 pm size. Some particles several microns in size were found. No diamond growth was observed in the absence of diamond seeds. Formation of new diamond was deduced from strong differences in the luminescence spectra of diamond seed and hydrothermally treated material, intensity increase of the diamond Raman band of an order of magnitude after hydrothermal treatment, and different downshifts of the Raman band with increasing excitation power. We found similar agglomerates after hydrothermal treatment of nanocrystalline diamond at 740°C and 300 MPa in pure water (Fig. 3a). Hydrothermal treatment


38 1

Figure 3. SEM micrographs of: (a) aggregate after 12 h hydrothermal treatment of nanocrystalline diamond at 740°C and 300 MPa, (b) plate-like crystals on the surface of a diamond seed crystal after 150 h hydrothermal treatment of graphite and diamond single crystals at 800°C and 300 MPa, (c) liquid-phase hydrothermal homoepitaxy of diamond on diamond seed at 170MPa and 400°C in a specially prepared water solution. Reproduced from [I51 with permission from A. Szymanski.

382


of nanocrystalline diamond can result in changes in its Raman-spectra without change in grain size [39] and it is not clear whether new diamond is formed. Hydrothermal growth of diamond on diamond single crystal substrates was reported when mixtures of glassy carbon with quarter micron diamond seed or with Ni and a diamond single crystal were treated with highly concentrated NaOH at 800°C and 140MPa [44]. Aggregates of about 10pm of transparent crystals were found in the presence of fine-grained diamond, while hexagonal, octahedral and cubic shaped transparent crystals of about 2 pm were observed in the presence of Ni. We observed formation of very similar two-dimensional crystals on diamond single crystals without addition of Ni or NaOH (Fig. 3b). The crystals formed could be diamond or a new crystalline transparent material without a Raman band, however, their identification was difficult. Successful hydrothermal diamond synthesis was carried out in autoclaves filled with a ‘specially prepared carbon enriched water solution’, the composition of which was not disclosed [15,45,46]. The carbon precursor should be fine-grained diamond, vitreous carbon or emulsion of crude oil and water [29]. The presence of free radical catalysts was mentioned and paragenetic crystallization of quartz needles and diamond indicate the presence of silicon [45]. The synthesis was described as a sol/gel colloidal process working in the range 200-600°C and 100200 MPa. Healing and joining of diamond crystals was reported. After 21 days at 400°C and 170 MPa, thin colorless films of polycrystalline diamond were obtained on (1 11) surfaces of seed crystals (Fig. 3c). With a reported size of 1 5 4 0 pm, these are the largest diamond crystals from hydrothermal experiments. Summarizing: Strong experimental evidence exists for diamond growth on diamond seed outside the range of its thermodynamic stability. While nanocrystalline diamond might grow under a variety of conditions, possibly due to the ‘crystallite size effect’ [18,28], growth of microcrystalline or possibly larger diamond requires special conditions which have only been attained very roughly so far.

3.3.2 Hydrothermal Treatment of Sic The first observation of hydrothermal carbon formation on S i c [47,48] led to the suggestion of the following basic decomposition reactions of S i c [49,50]:

+

Sic + 2H20 = Si02 CH4 S i c + 3H20 = Si02 CO 3H2 Sic + 4H20 = Si02 + C 0 2 + 4H2 S i c + 2H20 = Si02 + C + 2H2

+

+

(5)

(6)

(7) (8)

3.3.2.1 Thermodynamic Calculations of SiC-H20 Interactions The calculations presented here are in qualitative agreement with other published calculations [5 1,521. We concentrate on the activity of the condensed phases SO2, C, and Sic, presented as a function of the amount of water added (Fig. 4).


383

p-Sic -

I

01 0

I

1

2 3 amount of water I mol

4

5

Figure 4. Calculated equilibrium activity of Si02, S i c and C for compositions of 1 mol S i c plus variable amounts of H 2 0 at 500°C and 200 MPa.

This computation is typical for the whole supercritical T-P range. It shows that Si02 is stable in the whole range. Sic becomes unstable at a H 2 0 :Sic molar ratio of 2 : 1, which is a critical point, whereas carbon remains stable until a slightly larger H 2 0 :Sic molar ratio than 2: 1 is reached. This indicates the possibility of direct carbon formation in the course of hydrothermal leaching of S i c by selective oxidation of Si according to reaction (8). Reaction (8) is restricted to low H20:SiC molar ratios whereas reaction (6) and (7) are impeded at H20:SiC molar ratio 5 2 : 1. The equilibrium amount of carbon increases with increasing temperature and decreasing pressure. 3.3.2.2 Experimental Results for the SiC-H20 System Hydrothermal formation of carbon on Sic was first observed in the course of corrosion studies of amorphous Si-Ti-C-0 (Tyranno) fibers performed at 300-800°C at about 100 MPa [47,48,53]. Increasing downshift of the carbon D band in the Raman spectra with increasing reaction temperature from 1355-1360 cm-' in microcrystalline graphite to 1350cm-', 1336 cm-I, and 1330 f 3 cm-' after hydrothermal treatments at 300°C, 500"C, and 600-800°C respectively indicated the possible formation of sp3-bonded carbon [50]. The investigations were extended to a great variety of different forms of Sic. Carbon formation was observed on powders, platelets, and single crystals of a-Sic, powders and whiskers of p-Sic, B-doped SSiC and polycrystalline CVDSic samples [16,48-501. The formation of carbon depends crucially on the H 2 0 :Sic ratio, where three cases can be distinguished [50,52]. (i) Low H 2 0 :Sic ratio: Carbon and Si02 precipitate on the Sic surface. (ii) Very high H 2 0 :Sic ratio: Neither carbon nor Si02 precipitates, only in some cases carbon films of nanometer thickness where found. (iii) Medium H 2 0 :Sic ratio: Carbon film formation accompanied by dissolution of SO2.

384


I

I

1100 1200 1300 1400 1500 1600 Wavenumber / cm-I

Figure 5. Raman spectra of carbon on a-Sic crystals, X = 632.8 nm, after 4 h hydrothermal treatment at 600°C and 200 MPa: (a) homogeneous carbon film, (b) diamond particle at seed crystal edge, (c) example of an inhomogeneous carbon film after 4 h hydrothermal treatment at 700°C and 200 MPa.

Depending on the Sic material used, carbon formation starts at temperatures between 300°C and 600°C. In all cases IR-absorptions corresponding to hydrogenated sp3 carbon were present at about 2850, 2920, 2960 (C-H stretch) [16]. TEM investigations and analysis of the Raman spectra consistently showed formation of fine graphitic carbon with a grain size of a few nanometers. Carbon yield and grain size increase with increasing temperature. The carbon yield reaches a maximum at about 700°C and decrease with further temperature increase due to subsequent oxidation of the carbon films. Downshift of the Raman D-band of carbon and splitting in two or more bands in the range 1100-1400 cm-' were often observed in the whole temperature range under investigation and indicated disorder as well as the presence of sp3 carbon (Fig. 5), although the respective contribution of each of the processes into observed shift has not been established. Hydrothermal treatment of p-Sic whiskers for 24h at 600°C yielded Raman bands at 1327 and 1352cm-' which were assigned to nanosized diamond and disordered carbon, respectively [53]. Generally, downshift of the Raman D-band increases with increasing reaction temperature. Carbon layers grown on a-Sic crystal surfaces, visible in the optical microscope, revealed only the formation of small bulges, which may rather be attributed to domain borders than to delaminating carbon on SEM micrographs. Homogeneous films of amorphous carbon with high sp3-contribution (Fig. 5a) formed at 600°C. Very few transparent crystals of micrometer size, exhibiting the typical diamond Raman band at about 1330cm-' (Fig. 5b), were found at the edges of S i c crystals. Higher temperatures led to inhomogeneous carbon films with often high sp3 content


385

(Fig. 5c) on the Sic-crystals as well as on the gold capsules which indicates also carbon precipitation from the fluid. Carbon formation via hydrothermal treatment of P-Sic was intensively studied at 400-800°C and 100-500 MPa [54]. Carbon formed at all temperatures, but reasonable amounts in reasonable time were found only at and above 550°C. Above 600°C carbon deposits on the capsule walls indicated carbon precipitation from the fluid. In effort to reveal the presence of diamond after hydrothermal treatment, the residues were leached in 20% H F solution to remove Si02 and in hot HC104 to remove graphitic carbon because large amounts of silica and nondiamond carbon obscured the picture. XRD investigations showed graphite reflections and in some cases a very weak reflection at 0.206nm, attributed to diamond. In the Raman spectra, shifts of the graphite D band, and additional bands and broad features in the range 11001340 cm-', confirmed the presence of amorphous and nanocrystalline carbon and indicate the possible presence of disordered nanocrystalline sp3-carbon. Small particles with a size below 0.3pm were found, and sometimes intergrown with several larger P-SIC particles after wet oxidation, and exhibited predominantly hexagonal (0001) and face-centered cubic (1 11) and (100) faces that were attributed rather to diamond than to graphite. The morphology of the particles was very similar to that of gas phase diamond powders. Reflections corresponding to lattice spacing of 0.206 and 0.192 nm in small angle diffraction (SAD) patterns suggested the presence of cubic diamond as well as of lonsdaleite or other diamond polytypes. Hydrothermal experiments with P-Sic powders and watery solutions of 10% Na2C03, 10% NaHC03 and 50% NaHCOO yielded results which were very close to each other, because sodium hydrogen carbonate and sodium formiate can be converted to disodium carbonate at high temperatures according to

+

2NaHC03 = Na2C03 H 2 0

(9)

and

+ + H20.

2NaHCOO = Na2C03+ C H 2 0 = Na2C03 C

(10)

Carbon formation was observed at and above 400°C and became pronounced at 600°C. Raman spectra strongly suggest the formation of larger quantities of sp3bonded carbon than in the experiments with distilled water. Small amounts of well crystallized diamond were found after hydrothermal treatment of p-Sic powder at 700-750°C in the presence of diamond seed [55]. After removing silica and nondiamond carbon, small (<3 pm) carbon particles of predominantly octahedral shape, thus being probably diamond, were found. They were attached to the surfaces of the single crystal seed but could be removed by intense ultrasonic treatment. Tetrahedral hillocks of <1 pm size as well as tetrahedral etch pits appeared on the seed diamond surfaces. The hillocks always showed a faceted structure and sometimes common orientation, as expected for homoepitaxial diamond growth. The microcrystalline diamond seed did not show any increase in size after hydrothermal treatment, but particles of <50 nm size, which typically show a regular

386


crystalline shape, were found. They were identified as new-grown nanocrystalline diamond by Raman spectra, which show a downshift of the diamond band to 1323-1324 cm-' and an increased band halfwidth, full width at half maximum (FWHM) of 10-12 cm-' [56] compared with 1332 cm-' and 5 cm-' respectively, found for the microcrystalline diamond seed used. In another attempt to obtain diamond from Sic, no evidence for diamond growth was found by means of XRD and SEM after hydrothermal hydrolysis of p-Sic powder at 140MPa and 800°C [57]. Formation of quartz, a second acicular phase, graphite and maybe nanosized diamond crystals was reported. TEM analysis of whiskers of the second acicular phase revealed lattice spacing of 0.9 nm, a d-value of 0.56nm and a composition of Si01,6Co,2. A very sharp band at 1330cm-I was found in a Raman spectrum and assigned to the silicon oxycarbide, Si01.6Co.2, rather than to diamond. However, we would like to comment, that the lattice spacing, the d-value, and the acicularity of the so called 'oxycarbide' whiskers are close to the c-axis (0.861 nm), the (101) reflection (0.564nm) and the space group (P41212)of the silica modification keatite [58], which has no Raman band at 1330 cm-' . We often found keatite after hydrothermal experiments with p-Sic powder and CVD-SIC fibers [Sl]. The small amounts of carbon found can easily be introduced during hydrothermal treatment or TEM investigation. Thus, there is no alternative assignment to the band at 1330 cm-' except of diamond. Summarizing, strong evidence for the formation of nanosized diamond in the course of hydrothermal treatment of Sic was found. Seeding might enhance the size of newly formed diamond crystals. Above all, the hydrothermal method seems to be a versatile way for carbon coatings on Sic with emphasis on formation of amorphous sp3-bonded carbon. Depending on the substrate, reaction conditions and composition of the fluid, it should be possible to vary the coating structure from graphitic to diamond-like carbon. 3.3.2.3 Comparison of CVD- and Hydrothermal Diamond Synthesis from the Sic-H20 System

Formation of diamond-like sp3-carbon in the course of hydrothermal decomposition of Sic may be explained in the same way as diamond formation at low pressures by CVD. Due to its crystal structure, cubic p-Sic is a good substrate for heteroepitaxial diamond growth. Tetrahedral coordinated carbon in the Sic structure may become a good precursor for the further growth of sp3-bonded carbon (diamond) when Si is extracted. Hydrothermal treatment of S i c results in the formation of hydrogen, which, dissociated into atoms, plays a crucial role in CVD diamond synthesis by etching preferentially sp2-bonded carbon, thus facilitating formation of sp3-carbon and nucleation of diamond. Even though the equilibrium concentration of atomic hydrogen is very low, it can not be ruled out that hydrogen initially forms as atomic or highly reactive species, thus affecting the carbon in the neighborhood of its formation. These facts indicate some similarities between hydrothermal diamond synthesis from carbides and diamond synthesis via the vapor phase.

References

387

3.4 Outlook Taking into account the fact that hydrothermal formation of carbon takes place at relatively low temperatures (200-800°C) and pressures (<100 MPa), this method seems to be very attractive. It has obvious advantages over the HPHT synthesis due to less extreme temperature-pressure conditions and the possibility of much larger pressure vessels, which would make the hydrothermal method attractive for industrial use. However, the temperature of synthesis should be decreased preferably to 200-250°C to make commercialization and up-scaling easier. A better understanding of the mechanisms of diamond growth under hydrothermal conditions should make it possible to optimize the conditions for the growth of diamond coatings or large single crystals. High growth rates compared to CVD synthesis may be achievable. Current understanding of the diamond synthesis under hydrothermal conditions and quantity of diamond produced are comparable with those of CVD synthesis of diamond at the end of the 1950s or the beginning of the 1960s. Thus, we are at the beginning of the road and further research is strongly recommended.

Acknowledgments We thank Professor Dr. Andrzej M. Szymanski, Faculty of Chemistry of the Warsaw University of Technology for supply of materials for publication, and the German Research Society for financial support (grant DFG Ni 299/4).

References I . R. Berman and S. F. Simon, Z . Elektrochem. 1955, 59, 333-338. 2. R. C. DeVries, R. Roy, S. Somiya, and S. Yamada, Trans. Mater. Res. Soc. Jpn. 1994, 19B, 64 1-663. 3. R. C. DeVries, in Advanced Ceramics IIZ, S. Somiya (Ed.), Elsevier, Amsterdam, 1990, pp. 181205. 4. F. P. Bundy, H. T. Hall, H. M. Strong, and R. H. Wentorf, Nature, 1955, 176, 51-55. 5. K. Yamada and A. B. Sawaoka, J . Am. Ceram. Soc. 1994, 77, 1104-1106. 6. P. K. Bachmann, D. Leers, and H. Lydtin, Diamond Rel. Muter. 1991, 1, 1-12. 7. R. Roy, H. S. Dewan and P. Ravindranathan, Mater. Res. Bull. 1993, 28, 861-866. 8. R. Roy, H. S. Dewan, and P. Ravindranathan, J. Mater. Chem. 1993,3,685-686. 9. B. V. Derjaguin and D. V. Fedoseev, in New Diamond Science and Technology,M R S Znt. Conf. Proc. 1991, pp. 3-10. 10. M. N. Regueiro, P. Monceau, and J.-L. Hodeau, Nature, 1992,355, 237-239. 11. A. R. Krauss, D. M. Gruen, D. Zhou, T. G. McCauley, L. C. Qin, T. Corrigan, 0. Auciello, S. Pimenov, V. Konov, and A. Karabutov, in M R S Fall Meeting Abstracts, Paper W10.5, MRS, 1997, p. 499. 12. G. T. Visscher, D. C. Nesting, J. V. Badding, and P. A. Bianconi, Science, 1993,260, 1496-1499.

388


13. F. Banhart, Phys. Bl. 1997, 53, 33-35. 14. Y. G. Gogotsi, I.-D. Jeon, and M. J. McNallan, J . Muter. Chem. 1997, 7, 1841-1848. 15. A. M. Szymanski, E. Abgarowicz, A. Bakon, A. Niedbalska, R. Salacinski, and J. Sentek, Diamond Rel. Muter. 1995, 4, 234-235. 16. Y. G. Gogotsi, P. Kofstad, M. Yoshimura, and K. G. Nickel, Diamond Rel. Muter. 1996, 5, 151-162. 17. X.-Z. Zhao, R. Roy, K. A. Cherian, and A. Badzian, Nature, 1997, 385, 513-515. 18. P. Badziag, W. S. Verwoerd, W. P. Ellis, and N. R. Greiner, Nature, 1990, 343, 244-245. 19. R. Roy, in Workshop on Solvothermal and Hydrothermal Reactions, Takamatsu, Japan, 1996, pp. 11-120. 20. R. Roy, in M R S Fall Meeting Abstracts, Paper W10.4, MRS, 1997, p. 499. 21. B. R. Frost, Amer. J. Sci. 1979, 279, 1033-1059. 22. D. Rumble I11 and T. C. Hoering, Geochim. Cosmochim. Acta, 1986,50, 1239-1247. 23. 0. Navon, I. D. Hutcheon, G. R. Rossman, and G. J. Wasserburg, Nature, 1988,335, 784789. 24. 0. Navon, Nature, 1991, 353, 746-748. 25. G. D. J. Guthrie, D. R. Veblen, 0. Navon, and G. R. Rossman, Earth Planet. Sci. Lett. 1991, 105, 1-12. 26. A. M. Szymanski, Arch. Miner. 1994, 50, 13-30. 27. S. J. Thugutt, Rev. Sci. 1923, 61, 97-102. 28. A. Niedbalska, R. Salacinski, J. Sentek, and A. M. Szymanski, in Proceedings of the Fourth International Conference on Advances in New Diamond Science and Technology, S. Saito, N. Fujimori, 0. Fukunaga and M. Kamo (Eds), Scientific Publishing Division MYU, Tokyo, Japan, 1994, pp. 415418. 29. A . M . Szymanski, J. Miner. SOC.Korea, 1997, 10, 18-32. 30. C. Frondel and U. B. Marvin, Nature, 1967, 214, 587-589. 31. V. Mironov and B. Antonyuk, Arch. Miner. 1994, 50, 3-12. 32. A. M. Szymanski, in Paper presented at the European Conference on Diamond Materials, Barcelona, 1995. 33. A. M. Szymanski, Szklo Ceram. 1996,47, 26-28. 34. I. Leung, W. Guo, I. Friedman, and J. Gleason, Nature, 1990, 346, 352-354. 35. R. Roy, D. Ravichandran, P. Ravindranathan, and A. Badzian, J . Muter. Res. 1996,11, 11641168. 36. C. A. Eckert, B. L. Knutson, and P. G. Debenedetti, Nature, 1996,383, 313-318. 37. M. Siskin and A. R. Katritzky, Science, 1991, 254, 231-237. 38. P. Deines and J. W. Harris, Geochim. Cosmochim. Acta 1995,59, 3173-3188. 39. Y. G. Gogotsi, T. Kraft, K. G. Nickel, and M. E. Zvanut, Diamond Rel. Mater. 1998, submitted. 40. S. Yamaoka, M. Akaishi, H. Kanda, and T. Osawa, J. Cryst. Growth, 1992, 125, 375-377. 41. %-I. Hirano, K. Nakamura, and S. Somiya, in Hydrothermal Reactions for Materials Science and Engineering, S. Somiya (Ed.), Elsevier, London, 1989, pp. 331-336. 42. N. A. Bendeliani, T. D. Varfolomeeva, A. N. Glushko, N. A. Nikolaev, V. N. Slesarev, and S. V. Volkov, Kristallograjiya, 1995, 40, 345-346. 43. R. Roy, D. Ravichandran, and P. Ravindranathan, in Proceedings of the Applied Diamond Conference 1995, Applications of Diamond Films and Related Materials, Third International Conference, A. Feldman, Y. Tzeng, W. A. Yarbrough, M. Yoshikawa and M. Murakawa (Eds), NIST, Gaithersburg, MD, 1995, pp. 355-359. 44. D. Ravichandran and R. Roy, Muter. Res. Bull. 1996,31, 1075-1082. 45. A. M. Szymanski, J. Chem. Vap. Dep. 1996,4, 278-285. 46. E. Abgarowicz, A. Bakon, S. Maciejowska, R. Salacinski, A. Niedbalska, and A. M. Szymanski, in Proceedings of the Joint XV AIRAPT & XXXIII EHPRG International Conference Warsaw, W. A. Trzeciakowski (Ed.), World Scientific Publishing Co., Singapore, 1995, pp 247-249. 47. Y. G. Gogotsi and M. Yoshimura, J. Am. Ceram. SOC.1995,78, 1439-1450. 48. Y. G. Gogotsi and M. Yoshimura, Nature, 1994,367, 628-630. 49. Y. G. Gogotsi and M. Yoshimura, M R S Bull. 1994, XIX, 3 9 4 5 . 50. Y. G. Gogotsi, in Korrosion und VerschleiJ von keramischen Werkstoffen, R. Telle and P. Quirmbach (Eds), DKG, Aachen, 1994, pp. 114-122.

References

389

51. T. Kraft, K. G. Nickel, and Y. G. Gogotsi, J. Muter. Sci. 1998, submitted. 52. N. S. Jacobson, Y. G. Gogotsi, and M. Yoshimura, J. Muter. Chem. 1995, 5, 595-601. 53. Y. G. Gogotsi, M. Yoshimura, M. Kakihana, Y . Kanno, and M. Shibuya, Ceram. Trans. 1995, 51, 243-241. 54. Y. G. Gogotsi, K. G. Nickel, D. Bahloul-Hourlier, T. Merle-Mejean, G. E. Khomenko, and K. P. Skjerlie, J . Muter. Chem. 1996, 6 , 595-604. 55. Y. G. Gogotsi, K. G. Nickel, and P. Kofstad, J. Muter. Chem. 1995, 5, 2313-2314. 56. V. D. Andreev, T. A. Nachalnaya, and E. V. Gabrusenok, Sverktverd. Muter. 1993, 2, 11-18. 51. R.Roy, D. Ravichandran, A. Badzian, and E. Breval, Diamond Rel. Muter. 1996,5, 913-916. 58. J. Shropshire, P. P. Keat and, P. A. Vaughan, Z. Kristallogr. 1959, 112, 409413.


4 Chemical Vapor Deposition of Diamond Films C.-P. Klages

4.1 Introduction The development of the scientific knowledge of diamond, of its properties and composition, and the various attempts to synthesize this much sought-after material form an exciting chapter in the general history of science. Several review articles covering historical aspects of diamond in general or of its synthesis under conditions of metastability in particular are available. A recent short history of diamond synthesis, in which older sources back to the year 1711 are quoted, was written by Angus [l]. DeVries et al. [2], in an article containing numerous references to earlier publications, laid their emphasis on the contributions of Russian researchers. The modern era of diamond growth can be traced back to the days between 26 November 1952 and 7 January 1953, when Eversole at the laboratories of Union Carbide Corporation in the USA achieved the first overgrowth of diamond from low-pressure gas phases on the surface of diamond powder particles. Thus the growth of diamond under pressure-temperature conditions at which graphite is the thermodynamically stable phase of carbon, obviously predates the first successful high-pressure, high-temperature syntheses of diamond at ASEA in Sweden in February 1953 and at the General Electric Company in the USA in December 1954 [l]. Eversole was the first to synthesize diamond by any method, the procedure used by him, however, was far from economic viability, owing to a very low yield and the necessity of repeated removal of simultaneously deposited graphite by a treatment in an atmosphere of pure hydrogen. Additionally, there was much doubt in the scientific community that diamond could be grown at all under conditions of thermodynamic metastability. Therefore the work of Eversole did not attract enough attention to be a starting point of much more intensified efforts. The results of Eversole were confirmed about 15 years later independently by the group of Angus at Case Western Reserve University in the USA and by researchers of the Physical Chemistry Institute of Moscow. The important role of hydrogen in the low-pressure synthesis of diamond was recognized in the work of Angus. Molecular hydrogen suppresses the spontaneous nucleation and growth of graphite during the formation of diamond, but it does not completely prevent the formation of the stable carbon allotrope, which has to be removed from the deposit in a separate etching step. Therefore it was a very important breakthrough in the efforts to grow diamond by a chemical vapor deposition process, when in the seventies the Soviet group found out (stimulated by results of Case Western Reserve University group on the action of atomic hydrogen as an efficient remover of nondiamond carbon phases), that in the presence of a superequilibrium concentration of atomic hydrogen diamond could be deposited continuously, with much higher growth rates, largely free from

4.2 Preparation Methodsfor Diamond Films

391

graphite and not only epitaxially on diamond crystallites but also on nondiamond substrates. This Russian achievement can most probably be dated back to the early 1970s. Nevertheless, it took another decade, until (thanks to a series of papers from the Japanese NIRIM (National Institute for Research in Inorganic Materials), revealing detailed experimental information) the possibility and the potential of diamond growth from the gas phase was recognized by the general scientific community. The Japanese work, connected with the names of Kamo, Matsumoto, Sato, and Setaka, had started in 1974, based on results in the Russian literature. Among various diamond CVD processes, developed over about 15 years, the hot-filament CVD process (HF-CVD) and the microwave-plasma CVD method (MWP-CVD), for the first time described in detail by the Japanese group, have continued to be the most used diamond deposition processes.

4.2 Preparation Methods for Diamond Films At pressures below the diamond-graphite equilibrium line of the carbon phase diagram, graphite will eventually be the only product of any gas-phase pyrolysis process of a carbon-bearing atmosphere running very close to chemical equilibrium and leading to the deposition of carbon. All continuously working diamond deposition methods, on the other hand, have as a common feature the presence of a pronounced gas-phase nonequilibrium in close vicinity to the substrate, maintained by some sort of gas phase activation. As an empirical fact, the gas phase must contain, in addition to a carbon carrier, a powerful carbon etchant like oxygen, fluorine or (as indicated above) atomic hydrogen. The nonequilibrium situation close to the growing diamond film can be maintained either by thermal means (external heating like in the hot-filament process or by a chemical reaction as in the combustion (oxyacetylene) CVD method) or by the generation of a plasma, sustained by direct current (d.c.), radio frequency (RF) or microwave (MW) electrical fields. A thorough comparison of diamond CVD processes, including a discussion of their advantages and drawbacks, was given by Bachmann and Lydtin [3]. It was pointed out by these authors that typical linear growth rates, which can be achieved by these methods, correlate positively with the estimated gas phase temperatures, reaching approximately 1 mm h-l in plasma jets running at pressures between several 100mbar and 1 bar at gas temperatures exceeding 4000K, see Fig. 1. Growth rates less than 1 pm h-' are obtained in thermal decomposition processes as investigated mainly in the 1950s and 1960s and in nonisothermal plasmas at pressures typically below 1 mbar, in which heavy-particle temperatures are at least an order of magnitude smaller than the electron temperatures (ECR in Fig. 1 stands for electron cyclotron resonance, a special form of a microwave plasma supported by a magnetic field at pressures in the pbar region, see below). An intermediate position is occupied by low power density MW plasmas and hot-filament activated

392

4 Chemical Vupor Deposition of Diamond Films 1000

DC arcjet

100 *

V

10. 10

oxy-acetylene flames

r I pmlh low-power density MW plasmas hot filaments

1

0.1

non-isothermal DC, RF or ECR plasmas thermal decomposition

0.01 1000

2000

3000

4000

5000

6000

71 DO

gas phase temperature I K

Figure 1. Correlation of typical gas temperatures and growth rates r (in pm h-') of diamond CVD processes (after [3]). See text for the explanation of abbreviations.

processes, running typically at pressures beyond 10 mbar and allowing growth rates of several pm h-' . In the following, the HF-CVD method will be presented in some detail because of its simplicity and scientific as well as technical importance (thanks to its up-scaling potential). Several results of studies on HF-CVD processes are of fundamental importance for an understanding of diamond deposition in general. The presentation of plasma-based methods will focus on microwave-activated processes because they are probably the most often used diamond CVD growth methods for virtually impurity-free material.

4.2.1 Hot-filament Chemical Vapor Deposition The essential feature in the HF-CVD process which turns the pyrolysis of a hydrocarbon from a graphite-forming process to a diamond CVD method, is a hot refractory metal wire as an atomic hydrogen source in the close vicinity of the substrate. The elementary principle was described by Matsumoto et al. in 1982 [4,5], see Fig. 2. The researchers at NIRIM used coiled tungsten filaments, mounted roughly 1 cm from the substrate surface and heated to about 2000°C by passing an electrical current. The temperature of the substrate was usually 700-1000"C, as measured directly at the substrate holder. From a mixture of hydrogen and methane (about 1 vol-%) as a feed gas (10-1 00 cm3min-I), diamond was formed at pressures between roughly 10 and l00mbar. Dependent on the substrate conditions, the deposits consisted of cubo-octahedral or multiply twinned isolated particles or closed films of diamond. A substrate scratching with diamond powder is not expli-

4.2 Preparation Methods,for Diamond Films

393

To pump

Figure 2. Schematic view of the HF-CVD deposition chamber as used by Matsumoto et al. (adapted from [4,5]).

citly mentioned in these papers; it is however probable that this or a similar kind of pretreatment was used in order to achieve a dense nucleation and early coalescence of the crystallites to closed films. Figure 3 shows the surface of a closed diamond film with well-shaped facets grown by HF-CVD in the author’s laboratory. The typical appearance of isolated, non-twinned particles, is shown schematically in Fig. 5. There can be no reasonable doubt that the diamond phase of carbon was actually

Figure 3. Scanning electron micrograph of a diamond film grown by HF-CVD at Fraunhofer-IST (by courtesy of L. Schafer).

394

4 Chemical Vapor Deposition of Diamond Films

formed by the HF-CVD method as run at NIRIM: Scanning electron micrographs (SEM) clearly showed the cubo-octahedral morphology typical for a material with a cubic crystal lattice, electron diffraction data were consistent with ASTM values and the Raman peak at 1334.5cm-' was very close to the position reported for diamond single crystals. SEM photographs, X-ray or electron diffraction, and especially Raman spectroscopy are still the most conclusive methods for identifying the diamond phase and these methods have therefore been used routinely in many subsequent publications on diamond CVD. Linear growth rates around 1 pm h-' are obtained under the conditions quoted above. A growth velocity of this order of magnitude can still be considered as typical for HF-CVD processes employing hydrocarbon-hydrogen mixtures forming good quality diamond. Numerous investigations of the influence of deposition parameters on diamond growth rates and film qualities have been reported in the literature. Several important factors can be summarized from various investigations leading to basically identical results. (i) There is generally hardly any nucleation of HF-CVD (or MWP-CVD) diamond on a smooth, thoroughly cleaned substrate. Therefore, a suitable substrate pretreatment is essential in order to increase the nucleation density, leading to a faster formation of a closed film and better utilization of growth species. This can be done, for example, by ultrasonic irradiation in an aqueous diamond powder suspension or by lapping with diamond paste. Diamond residues are most probably responsible for this effect. (ii) A maximum growth rate is achieved at substrate temperatures between 850 and 950°C: the exact figure depends on the pressure used. (iii) The growth rate increases with the filament temperature and higher growth rates are achieved at lower filament-substrate distances, thanks to an increased flux of H atoms to the substrate 'driving' the formation of diamond. (iv) Increasing diamond growth rates can also be achieved with methane concentrations growing from 0.1 to 2 vol-% in hydrogen, however the film quality will deteriorate due to increasing amounts of sp2-bonded carbon. (v) As a function of gas pressure, the growth rate has a maximum at pressures typically between 10 and 100 mbar, dependent on the methane concentration, and the temperatures of filament and substrate. The positive effect of increasing carbon concentration in the feed gas on the linear growth rates, accompanied by a deterioration of the film quality under otherwise unchanged conditions, is a very general effect in all diamond CVD processes. Therefore, a compromise between growth rate and film quality will have to be made, depending on economic considerations and the film properties required for a specific application. In the early HF-CVD studies tungsten was exclusively used as a filament material. Among other refractory metals, only tantalum and rhenium have since then gained greater importance. An advantage of tantalum over tungsten is its much smaller carburization rate: W as well as Ta are thermodynamically unstable with respect to carbides Me2C or MeC under typical diamond forming conditions; however the carbide-forming reaction which severely deforms and embrittles the filaments,

4.2 Preparation Methods for Diamond Films

395

is generally significantly slower for a tantalum wire. Under comparable conditions, Ta keeps its mechanical strength longer than a W filament of the same thickness, thanks to a core of uncarburized material [6]. In contrast to Ta and W, Re does not form a carbide, albeit it dissolves a certain amount of carbon. Therefore rhenium filaments, if used under suitable conditions, can retain their ductility and do not show significant dimensional changes [7]. However, the much higher price of Re has to be taken into account, if deposition of diamond on a larger than laboratory scale is considered. The role of the hot filament in diamond HF-CVD is clearly to dissociate molecular hydrogen in order to meet the requirement for a continuous growth of the diamond phase. The question, if there are any differences between filament materials (for the same geometries and surface temperatures), as far as the hydrogen dissociation capabilities and achievable growth rates are concerned, is closely related to chemical surface reactions involving carbon and hydrogen. If the concentrations of carbon- and hydrogen-bearing species on and near to the filament surface are close to equilibrium concentrations at the filament temperature, no significant effects of the filament material on gas phase composition near the substrate will be expected. From the experimental results reported in the literature, no clear conclusions can be drawn with respect to this question: In a comparison of tantalum and tungsten filaments Singh et al. [8] did not find significant differences in diamond growth rates, and the authors concluded that the dissociation of the gas mixture is independent of the filament material. In contrast, distinct differences between rhenium, tantalum, and tungsten filaments in HF-CVD of diamond were found by Okoli et al. [6]. In order to obtain well faceted good quality diamond films different methane concentrations were needed for the three materials at a filament temperature of 2200°C. According to results of Sommer and Smith [7], material and temperature dependent formation of carbon contamination surface layers, reducing the filament activity, has to be taken into account in interpreting comparisons of this kind. Gas phase diagnostic investigations also show differences in the hydrogen dissociation capabilities of different filament materials: In methane-containing gas phases the quantitative determination of H concentrations by two-photon laser-induced fluorescence yielded values below the equilibrium concentrations calculated for the filament temperatures. Among the materials investigated (W, Ta, Ir) the highest concentrations of atomic hydrogen were generated by W filaments [9]. The deviation of molecular hydrogen dissociation from thermal equilibrium, indicating material specific surface effects, is also indicated by higher degrees of dissociation in pure hydrogen gas phases, where no carburization or carbon impurities affect the filament surface. Aside from the filament stability and its hydrogen dissociation capability the incorporation of evaporated filament material into the growing diamond films is one aspect which has to be taken into account in HF-CVD: Quantitative determinations of filament-related impurities in diamond films have been carried out for rhenium, tantalum and tungsten filaments. For filament temperatures between 1900°C and 2350°C the amount of incorporated Re increases by three orders of magnitude to a maximum of about 0.1 at-% [lo]. Using Rutherford backscattering

396


(RBS) a tungsten concentration around 0.1 at-% was measured by Gheerhaert et al. in films grown at filament temperatures of about 2250°C [Ill. For tantalum filaments at 2300°C maximum Ta concentrations of 0.03 at-% were detected using secondary mass spectrometry (SIMS) [121. For carbide-forming filament materials the degree of filament material incorporation strongly depends on the carburization state of the filaments. This is caused by the fact that at the relevant temperatures the carbides have lower vapor pressures than the corresponding uncarburized materials [ 121. Therefore, in several studies relatively high amounts of filament impurities were found at the diamond-substrate interface when the deposition process was started with uncarburized filaments. The mostly used gas atmospheres for hot-filament deposition of diamond generally consist of mixtures of 0.1 to a few vol-% methane and hydrogen. In contrast to plasma-activated deposition methods, additions of oxygen-containing source gases in HF-CVD of diamond are possible only to a limited extent, due to the tendency of oxidation and volatilization of the filament material. Nevertheless, additions of oxygen, water vapor, or oxygen-bearing organic compounds to the HF-CVD gas feed can have several advantages if the quantitative gas phase composition is carefully chosen. This was first demonstrated by Hirose and Terasawa, who used a range of oxygen- and nitrogen-containing organic compounds [13]. Diamond films with good crystallinity and high phase purity (according to Raman and electron diffraction characterization) were obtained at growth rates about an order of magnitude higher than typical for oxygen-free gas feeds. Compared with methane-hydrogen mixtures, much higher gas pressures, up to nearly one atmosphere, and gas phase concentrations of the carbon-bearing species were possible in these experiments. These results were later essentially confirmed by other authors, achieving growth rates up to 10 pm h-' by using ethanol and acetone (1-8 vol-%) in hydrogen at pressures up to 13 kPa [14]. Kawato and Kondo carried out a comparison of CH4-H2 and CH4-H2-02 mixtures and showed that an increased growth rate and extended pressure and concentration ranges (40 kPa, 4 vol-%) are also obtainable with additions of molecular oxygen [15]. Numerous investigations have been performed in order to reveal the reasons for the beneficial effects of oxygen-containing gases observed in diamond HF-CVD (and MWP-CVD) [15-201. Results of these studies can only be summarized here without further comments. The effects of oxygen have generally been attributed to -

-

promotion of hydrogen dissociation and methane decomposition, a reduction of the acetylene concentration, leading to less nondiamond carbon phases, formation of active growth sites by improved hydrogen abstraction, and an efficient removal of non-diamond carbon phases by hydroxy radicals.

As shown by Bachmann, there are more general limits to the C,H,O-concentrations which can be used in diamond HF-CVD and similarly in most other diamond CVD processes. A compilation of diamond-forming gas phase compositions from a large number of publications shows that (within a ternary C,H,O-concentration diagram) those concentrations leading to diamond growth are restricted to an essentially

4.2 Preparution Methods for Diamond Films

397

wedge-shaped area extending from the H-corner along the CO line of the triangle [21]. Smaller carbon concentrations do not lead to the formation of deposits at all, due to the dominance of etching processes over carbon deposition. At concentrations higher than those situated in the diamond wedge sp2-coordinated carbon phases tend to dominate. As a key to the growth of fiber-textured diamond films, additions of molecular nitrogen or nitrogen-containing compounds have in the recent years gained increased interest, see below. Homoepitaxial growth experiments of Cao et al. [22] showed, that nitrogen strongly enhances the growth rate of { 100) facets. This effect can be attributed to nitrogen compounds breaking the dimers of the (2x1) reconstructed {loo} surface, thus favoring the (100) growth direction.

4.2.2 Microwave-plasma-based Methods As indicated above already, diamond growth rates beyond a few 100 nm h-' generally require gas phase temperatures above 2000K. Aside from hot filaments and combustion flames, various kinds of plasmas are in use today to heat the gas phase. Depending on the ratio of average energies of electrons and heavy particles (neutrals and ions), electrical gas discharges can be divided into nonisothermal plasmas, characterized by a large difference between the average energies (or temperatures) of electrons and neutrals (or ions), and isothermal plasmas, in which electrons and heavier particles have virtually the same temperature. In the pressure region below about 1 mbar, for example in sputtering, many low-pressure plasmaCVD or glow-discharge polymerization processes, average electron energies of a few electron volts, corresponding to several l0000K, are quite typical while the gas phase temperatures usually do not exceed a few hundred degrees Celsius. The success of plasma CVD methods in diamond growth is generally based on the use of higher pressures, mostly beyond 10 mbar and up to 1 bar. Under these conditions, a local thermal equilibrium prevails between the electrons, gaining their energy from an electrical field, and heavier particles which are heated by frequent collisions with the hot electrons. As an example, the use of microwave plasmas will be discussed in the following sections. 4.2.2.1 Isothermal Microwave Plasma Chemical Vapor Deposition Aside from hot-filament chemical vapor deposition, microwave plasma CVD (MWP-CVD) of diamond has become the most popular method for growing diamond films, since the pioneering paper of Kamo et al. appeared in 1982 [23], in which the experimental setup and deposition conditions were described in detail. The tubular apparatus, as used by the NIRIM researchers and constructed in many other laboratories subsequently, typically consists of a silica reactor tube running through two opposing bores in a rectangular microwave guide, connected to a 2.45 GHz magnetron source. Using a stub tuner between the magnetron and the silica tube and a plunger at the end of the waveguide, a position of maximum MW

398


electrical field strength can be generated within a section of the tube, just beyond a substrate table. Owing to the geometric limitations of suitable waveguides at 2.45GHz, quartz tubes are limited in diameter to around 5cm. Therefore the coatable area on a substrate is usually between 1 and 10cm2. The experimental conditions leading to successful growth of diamond are very similar to those in HF-CVD: Pressures are typically between 10 and loombar, the feed gas in the first NIRIM publication consisted of 1 vol-YOmethane in hydrogen and the substrate was heated to 800-1000°C in a plasma generated at several 100W to 1 kW. Under these conditions, growth rates between 0.5 and 3 pm h-’ can be achieved, the lower value being more realistic for well faceted, high phase purity films. In early MWP-CVD experiments oxygen-free gas feeds, generally 0.1-1 vol-% methane in hydrogen, were commonly in use. However, since it was shown by Hirose and Terasawa and others, that higher growth rates and better film qualities at larger pressures and carbon concentrations could be achieved using oxygen compounds in HF-CVD [13], it became popular to use oxygen-containing compounds like 02,H 2 0 , CO, C02, alcohols, ethers, and others in microwave plasma CVD of diamond. Independent of the gas composition actually used, hydrogen, methane, and acetylene are the prevailing components of the in situ and exhaust gas of typical MWP-CVD processes in C,H gas phases. Oxygen-containing components are predominantly converted to carbon monoxide. Oxygen additions suppress the acetylene formation and lead eventually to a practically complete disappearance of hydrocarbons in favor of CO. The spectrum of stable products is fairly similar to that of HF-CVD processes, pointing to a basic similarity of both types of processes [24]. In fact the main role of the plasma in MWP-CVD (analogous to the filament in HF-CVD) seems to be the dissociation of molecular hydrogen into atoms, which initiate a reaction chain leading to the methyl radical as the most important diamond growth species, and help to gasify unwanted sp2-bonded carbon deposits (see below). While the tubular reactor design was used in most laboratories working on MWPCVD of diamond in the mid-80s, commercial reactors appeared on the market after a few years. Major drawbacks of the NIRIM-type reactors are the limited substrate size and difficulties in preventing the plasma from interactions with the tube material, leading to etching and redeposition of silicon contaminations in the films, localized heating, deformation, or even puncturing of the tube. Also, MW power independent adjustment of the substrate temperature by external heating is difficult. The development of MWP-CVD sources at ASTeX, Applied Science and Technologies, at Woburn, MA, USA was outlined by P. K. Bachmann [25], who also contributed to the design of the so called ‘bell jar’ reactor which was the first microwave plasma CVD apparatus permitting growth of diamond on substrates up to 7.5cm in diameter and became the first commercially available diamond CVD reactor at all. This reactor contains a dome-shaped silica bell jar of lOcm diameter in which a plasma is generated at a location of maximum electrical field strength well apart from the wall material. Today the ASTeX high-pressure microwave source reactor (HPMS) is in use in many laboratories. A schematic cross-section of this apparatus is shown in Fig. 4 [25]. Microwave power is transmitted by a coupling device from a rectangular wave-

4.2 Preparution Metkods,for Diamond Films

399

MW

=i

inlet gas

,-

substrate

t o pump

+----

Figure 4. Schematic cross-section through a modern 2.45 GHz ASTeX HPMS (‘High-pressure Microwave Source’) MWP-CVD reactor. Microwave radiation is coupled through a silica window separating the atmospheric pressure region within the waveguide and around the antenna from the plasma region under reduced pressure. The reactor is equipped with a R F inductionheatable graphite stage and can be used for substrates up to lOcm diameter. Modified by a water-cooled graphite stage, reactors of this type can be used for up to 5 kW MW power. Inspection windows are not shown here (adapted from [25]).

guide through a silica MW window into a cylindrical double-walled water-cooled steel shell equipped with quartz windows for process inspection and diagnostic investigations. The substrate is located on a graphite substrate stage independently heatable from below by a R F induction coil. The system was introduced with a 1.5 kW MW magnetron and power supply and can now be obtained in a modified version equipped with a 5 kW magnetron and a water-coolable substrate stage. Growth rates up to 15 pm h-’ can be obtained at 5 kW using suitable process parameters. However, process conditions leading to very phase-pure material as usually needed for optical, electrical, or thermal purposes, always lead to much smaller deposition rates. Under suitably optimized deposition conditions, reactors of this type are very stable and can be left running unattended for days or even several weeks if very thick diamond films are needed. In order to coat substrates with diameters in excess of 4 inches, reactors working at lower frequency have been developed in recent years. Using a 75 kW source operating at 91 5 MHz, an apparatus was built by ASTeX permitting a coatable area of 8 inches diameter and a total mass deposition rate which is a factor of 10-100 larger than for the 5 kW HPMS source [25].

400

4 Chemical Vapor Deposition o j Diamond Films

4.2.2.2 Non-isothermal Plasma Deposition

A different approach was pursued by Kawarada, Suzuki and others [26,27], trying to increase the size of the microwave plasma by operating it at pressures well below 1 mbar. In the case of MW fields, it becomes increasingly difficult to sustain a plasma under decreasing pressure owing to insufficient energy absorption by the electrons from the oscillating electrical field in the absence of frequent collisions with heavy particles. The number of ionizing collisions will be decreased and eventually the loss rate of electrons by recombination and diffusion to the chamber walls will become larger than the generation rate; the plasma will be extinguished. In order to operate a plasma sustained by microwave radiation down to pressures of lop4mbar, electron cyclotron resonance (ECR) sources have gained a certain importance in the field of thin film deposition and etching. The Lorentz force imposed by a magnetic field of 875 G field strength normal to the oscillating electrical field of 2.45 GHz will force the electrons onto closed circular loops (cyclotron motion), allowing a much higher average electron energy. However, in case of diamond CVD deposition, use of an ECR source at pressures below 0.1 mbar, which is about the maximum pressure for the ECR mechanism to come into play, yields only fairly small growth rates and a poor crystallinity and phase purity of the deposit. In a wide range of experiments conducted in the author’s laboratory at pressures below mbar, no carbon deposition could be detected at all. This is just one example of the failure of conventional nonisothermal plasmas, which are frequently used for thin film deposition and etching, to grow diamond films with acceptable rate and quality.

4.2.3 Preparation of Special Forms: Textured and Heteroepitaxial Films Throughout the first decade of the modern diamond CVD era, counted from the first NIRIM papers in the early 80s, it was common practise to treat a substrate to be coated by scratching with diamond paste in order to enhance film nucleation. Diamond films of a few micrometers thickness, grown on such a substrate, consist of randomly oriented crystallites because neither the nucleation nor the growth process introduce any significant ordering of crystallite orientations in the film. The development of methods allowing the growth of diamond films with an increased orientational order (i.e. a reduced degree of orientational freedom) of crystallites was not only of academic interest; for many application of diamond films it would be of considerable importance to control the orientation of crystallites during film growth and thereby improve the film morphology or physical characteristics like the optical or (thermal and electrical) transport properties. The ultimate goal would doubtlessly be the growth of large-area films approaching the quality of single crystals. Two methods will be presented here allowing the preparation of films in which the crystallites have lost two and three degrees of freedom. By an appropriate control of the growth process, diamond films with fiber texture can be fabricated, in which

4.2 Preparation Methods for Diamond Films

40 1

special crystallographic axes of individual crystals are restricted to make a given angle with the film normal, which in a perfectly textured film is zero (loss of two degrees of freedom). Three degrees of orientational freedom are lost in epitaxial and heteroepitaxial films, in which the regularity of a substrate lattice imposes a complete, or nearly complete, orientation on the (nuclei of the) growing film. 4.2.3.1 Textured Diamond Films The formation of a texture, that is, a preferred orientation of crystallites, during deposition of diamond is a nice example of a mechanism which can be observed during mass crystallization of many materials. The solidification of a melt in a container, for example, usually starts simultaneously at many nucleation sites on the container walls. During the competitive struggle among the growing crystallites only those with fastest growth directions close to the substrate normal survive. Crystals with unfavorable orientations will die out during growth, leaving a continuously decreasing number of still-growing crystals, accompanied by an increase of their frontal area. This process of ‘geometric selection’ was analyzed theoretically by Kolmogorov in 1945, leading to his law n(h) 0; l/h”* for the crystallite population density n at an ingot thickness h [28]. In 1967 the principle was re-named by Van der Drift as an ‘evolutionary selection’ and discussed quite generally for various situations playing a role in vapor-deposition of thin films (crystal morphology, angle of incidence of growth species, surface diffusion) [29]. For an isotropic supply of growth species, freely growing crystals and crystals in a film develop in the same way: The direction of fastest growth in the isolated crystal is the direction from the center to the remotest point of the crystal. In a film, the crystals which are the ‘fittest’ for survival during the evolutionary selection, are those which have this direction most parallel to the average growth direction. Texture formation in growing diamond films is closely related to the morphology of individual diamond crystallites formed under identical conditions. In discussing crystal morphologies, on must differentiate between the equilibrium shape of a crystal, which is determined by the thermodynamics of equilibrium between, for example, vapor and solid phase, taking into account the orientation dependence r(n) of surface energy of the solid phase, and the growth shape, determined by the kinematics of faces during growth, allowing for the orientation dependence of growth rates V(n). Sharp cusp-like minima in polar diagrams of r(n) or V(n) correspond to flat faces on the equilibrium and growth shape. Dominant faces are those with the smallest distances from the origin, that is, those with smallest surface energy and growth rates. Frequently growth shapes are quite similar to equilibrium shapes, as lowest growth rates are expected for faces parallel to a maximum number of atomic chains connected by strong bonds, so-called F-faces (flat faces) containing at least two different ‘periodic bond chains’ (PBCs [30]), characterized by the lowest surface energies. In the absence of defects like screw dislocations, growth on these faces has to overcome the barrier for the formation of two-dimensional nuclei of new layers. Higher surface energies and growth rates are generally found for S-faces (Stepped) and K-faces (kinked), characterized by the presence of only one type of parallel

402


PBCs or the lack of parallel PBCs. Growth on these faces involves the addition of growth units to pre-existing steps or kinks, without an activation barrier. These interrelations between surface energy and growth rates (in spite of often being able to describe general features of crystal morphology) do not necessarily hold always and exactly: Growth rates of different faces of crystals are often observed to depend differently on growth conditions like temperature, supersaturation and kind of growth species, impurities acting as inhibitors or catalysts and so on. Diamond CVD is a nice example illustrating the influence of growth conditions on crystal shape and film texture. A PBC analysis of the diamond lattice leads one to expect ( 11 1) faces (containing at least two different types of periodic bond chains) as the only well-defined flat faces of diamond crystallites. (001) faces should be kinked (K-faces) and appear (if at all) as curved surfaces owing to the lack of pronounced cusp-like growth rate or surface energy minima in (001) directions. This expectation is, however, in conflict with experimental evidence, showing that (001) faces of diamond and other crystals with a diamond lattice are generally slow-growing well-defined flat faces of similar or even larger morphological importance than ( 111) faces. Giling and Van Enckevort explained this prominence of (001) faces in terms of a surface reconstruction whereby surface atoms are able to lower their energy by dimerization with neighboring atoms, thus reducing the number of dangling bonds at the surface. The surface reconstruction process leads to F-face-like behavior of (001) faces, that is, growth by two-dimensional nucleation of new layers [311. A cubic crystal bounded by (001) and (111) faces generally forms a cubooctahedron which degenerates into a cube or an octahedron as the limiting shape for vanishing portions of ( 111) and (001) faces. The relative share of these two types of faces depends on the ratio of growth rates Vool/Vlll. In Fig. 5 crystal shapes are shown for a number of ‘growth parameters’ a, defined by 31/2VooI/ V11 [32]. The arrows indicate the largest diameter, which is the direction of fastest growth of an individual crystal. In order to realize a cubic diamond crystallite, the growth rate ratio has to become Vool/V111I 3-‘/2 = 0.58; for a pure octahedron a ratio 2 31/2= 1.73 is required. As shown by Clawing et al. [33], all forms of cubo-octahedrons, including the limiting forms of cubes and octahedrons, can be produced as single isolated crystallites by a suitable selection of growth conditions, that is, growth rate ratios 0.58 V0ol/Vl11 2 1.73 are accessible. Thick films grown under the extreme condition of a growth rate ratio near 0.58 (a = 1) exhibited a (1 11) texture with 1

1.5

2

2.5

3

Figure 5. Morphologies of cubo-octahedral crystals, exhibiting (001) and { 11I } facets, grown at different growth parameters, a. Arrows indicate the directions of fastest growth (Reprinted from Diamond and Related Materials, 3, C. Wild et al., Oriented CVD diamond films: twin formation, structure and morphology, pp. 373-38 1, Copyright 1994, with permission from Elsevier Science).

4.2 Preparation Merhodsfor Diamond Films

403

Figure 6. SEM micrographs of a (001)-textured diamond film with {OOl) morphology (by courtesy of M. Fryda).

(001) facets as to be expected from the considerations made above. Similarly, (001)textured films with { 11 1) facets were obtained in the opposite extreme (a! = 3), where pure octahedra were grown as single crystallites. Upon a reduction of a! to about 1.9, corresponding to single crystallites appearing as octahedra with truncated corners, a material dominated by (001) facets, with (001) about 10-15" from the substrate normal, was formed. Figure 6 shows the surface of a film of this kind, grown at the author's institute. Growth of diamond films is frequently accompanied, however, by the formation of twins stabilizing a pronounced (011) texture. An analysis of the formation mechanism of this type of texture was presented by Wild et al. According to the principle of evolutionary selection, a (01 1) texture should only occur at cy = 1.5, with protruding corners formed by {OOl} and { 11 1) faces. However, roof-shaped structures bounded only by { 11 1) faces were observed in thick (01 1)-textured films [34]. Taking into account the effect of twinning on (111) faces the authors presented arguments for the destabilization of growth of (001) tips in favor of a preferential (01 1) growth direction, forming the observed roof-shaped appearance. As shown in [34] and [35], high values of a are favored by high methane concentrations and low deposition temperatures. These findings are in agreement with results of direct growth rate measurements performed on single cubo-octahedral crystallites and homoepitaxial films, showing a stronger methane-concentration dependence of the [OOl] growth rate, however a larger activation energy (temperature dependence) for [11 11 growth [36]. Knowledge of the growth parameter dependence of a permits a large scope in tailoring film texture and morphology independently. Thus, using a two step procedure, the growth of smooth, perfectly (001)-textured films with (001) morphology is made possible [37]. The concentration of molecular nitrogen in the source gas was recognized as a very important third parameter determining the growth morphology and texture: Small amounts (10-100 p.p.m.) of nitrogen were found to improve the crystalline perfection and favor the formation of smooth (001 }-faceted (001)-textured films considerably [22]. An analysis of HF-CVD homoepitaxial film growth rates

404


showed, that the addition of 100 p.p.m. nitrogen accelerated the [ l l l ] growth by a factor of 2.2, however, a 4.4-fold increase was found for the [OOl] growth rate. Therefore, (Y is multiplied by a factor of 2. A probable explanation of this phenomenon was given in terms of the surface reconstruction by dimerization on (001) faces, leading to the F-face type behavior (see above): The adsorption of nitrogen-containing species could partially break this surface reconstruction, leading to the formation of kink sites and an enhancement of the growth rate on this face. An alternative mechanism leading to the growth of textured diamond films is related to the bombardment with low-energy ions [38]. By using a carefully controlled substrate bias-voltage during film growth, films with (001) texture and (001) morphology could be obtained on diamond (film) substrates of any orientation. Seemingly the ion flux leads to a renucleation and a preferential growth of only those crystallites which are oriented with one [OOl] direction parallel to the ion flux direction, possibly due to their higher stability with respect to a flow of etching H+ ions. 4.2.3.2 Heteroepitaxy of Diamond Films 4.2.3.2.1 Bias-induced Heteroepitaxy on Silicon Substrates In the early nineties it was discovered independently by research groups in the USA and in Germany, that heteroepitaxial nucleation of diamond can be achieved on 3C-Sic and silicon, by applying a negative substrate bias voltage to the substrate for a limited period of time preceding conventional microwave plasma-assisted diamond CVD (MWPCVD) 139,401. The work at the author’s institute was started in 1991, inspired by literature on the ‘beneficial’ effects of ion beam assistance on heteroepitaxial nucleation and growth. The use of an ion beam during MW-PCVD is precluded due to the low mean free path length of ions at the pressures typically used. The energy of positively charged species arriving at the substrate can, however, easily be increased beyond the l-lOeV region, which is typical for self-bias effects, by connecting the substrate holder to an external d.c. voltage source. The bias-induction of diamond nucleation had already been reported by Yugo et al. in 1991 [41], and thoroughly studied also by Stoner et al. [42]. However, they found no evidence of the formation of heteroepitaxially oriented nuclei, due to the selection of experimental parameters in their studies. Using a commercial MW-PCVD reactor equipped with a substrate bias, nucleation and growth of epitaxially orientated diamond films can be achieved on silicon substrates by applying a three-step procedure consisting typically of: -

-

-

plasma etching of the substrate in pure H2 at 850°C and an applied substrate bias voltage, V,, of -150V; a nucleation step in CH4/H2 (partial pressure ratio 1:lOO) at 850°C and VB = -150V, 5-10min; and film growth in CH4/H2(0.5: 100) at 800°C.

A scanning electron micrograph of a heteroepitaxial film grown on (001) silicon is shown in Fig. 7. The epitaxial orientation of most of the grains relative to the substrate is obvious: The grains exhibit nearly square (001) facets bounded by

4.2 Preparation Metho& for Diamond Films

405

Figure 7. SEM micrograph of a heteroepitaxial diamond film obtained on (001) silicon by biasinduced nucleation, film thickness =6.5 pm. Vertical and horizontal edges of the photograph are orientated parallel to the Si [I 101 and [LlO] directions, respectively (by courtesy of X. Jiang).

edges oriented along [110] and [I101 with (OO1)diamondIl(OO1)silicon and [l 10]diamondll[l10]silicon. The overall four-fold symmetry of a heteroepitaxial film on silicon is also evident from polar plots of Raman intensities taken with parallel polarization of incident and scattered light as a function of the azimuthal angle, 4, between the polarization vector and the [110] substrate direction. X-ray pole figures also reveal the symmetry of heteroepitaxial films on silicon and confirm the orientation relations given above for lattice planes of silicon and diamond. In addition, this technique is able to give a measure of the crystallographic quality of the films: Due to a residual misalignment between individual diamond crystallites and the substrate (a tilt misorientation between the [OOI] directions of individual grains and the substrate as well as a twist misorientation around the [OOI] substrate direction) the reflections corresponding, for example, to { 11l} lattice planes in [001]-oriented films are broadened. The full width at half maximum (FWHM) of pole figures taken on early samples was about 12" in both the x and 4 variable [43]. The tilt component of the misalignment can also be measured by quantitative image analyses of atomic force microscopy (AFM) data: Tilt angles of diamond (001) facets with respect to the substrate plane exhibit a Gaussian distribution around o", independent of the azimuthal direction of the measurement [MI. The latter result indicates that the perfect orientation represents an energy minimum in this epitaxial system (albeit possibly a rather shallow one): Interfacial tilting and azimuthal rotational misorientations are not necessarily a consequence of the large lattice constant misfit between diamond and silicon. Therefore, the crystallographic quality of heteroepitaxial diamond films on silicon can be considerably improved by adjusting the process conditions (substrate cleaning, base vacuum, nucleation parameters, especially voltage and time). In fact, considerable progress

406

4 Chemical Vapor Deposition of’Diamond Films

in this direction has already been made; tilt angle as well as pole figure FWHM values of less than 5” were observed in films of 10-20pm thickness; in thicker films (100 pm), about 2” have been reached. 4.2.3.2.2 The Mechanism of Bias-induced Nucleation The mechanism of biasinduced nucleation is presently still a matter of many investigations. Owing to the difficulty of surface analyses under the high-pressure conditions typical of diamond MW-PCVD, experimental clues to the kind of decisive steps are rare. Only two fairly different hypotheses out of many discussed in the literature shall therefore be mentioned here. A subplantation process was discussed by Gerber et al. [45,46], due to the pronounced voltage dependence which they found for the bias-induced nucleation of diamond on silicon, exhibiting a sharp maximum at -225V for 2% methane. Measurements of the ion energy distribution (IED) by a retarding field probe in the substrate table showed that these voltages lead to ion energies of 70-80eV, according to the authors the optimum region for the formation of dense sp3bonded carbon by a subplantation mechanism. The subplantation model was originally proposed by Lifshitz et al. [47] in order to explain what was claimed (but seemingly not confirmed later) to be an epitaxial growth of diamond on (111) Si at room temperature from a mass-selected 120eV C+ beam [48]. Subplantation according to this model involves: -

-

-

shallow implantation of hyperthermal ions (energies of the order of 100 eV) onto subsurface interstitial sites; internal growth of a dense, rigid phase and simultaneously substrate surface sputtering and dilution; and possible oriented growth by a ‘mold’ effect, that is, endotaxy with the surrounding crystalline environment.

On the other hand, a more classical reaction scheme was proposed by Jiang et al., based on AFM and SEM observations of the temporal development of diamond crystallites. They considered the formation, destruction and capture of ‘active sites’, ‘germs’, and ‘nuclei’ and attributed the role of ion-bombardment to an enhancement of adatom diffusion [49]. 4.2.3.2.3 Heteroepitaxy on Other Substrates Many other substrate materials have in recent years been used to grow epitaxially oriented diamond crystallites or continuous films. An exhaustive review with 50 references, covering growth on the nonmetals c-BN, Si, Sic, BeO, graphite and metals like Cu, Ni, Pt, Ir appeared recently [50]. In spite of a good lattice match (misfit 1.3%) and a ready nucleation of diamond on its boron-terminated {lll}, face, c-BN has hardly any value as a substrate for technical purposes because it can only be grown by high-pressure high-temperature synthesis with sizes of, at most, 0.5 mm diameter. Cubic silicon carbide is an important substrate as far as potential applications are concerned, because 3C-Sic films of high quality can be grown on Si either by chemical conversion in carbon-containing gas phases or by CVD processes. Using bias-induced nucleation, very smooth heteroepitaxial diamond films have recently been grown

4.3 Thermochemistry and Mechanism of Chemical Vapor Deposition Diamond Growth

407

on 3C-Sic films on silicon, exceeding probably the quality of other films grown on Si or S i c before [51]. The ultimate goal of diamond heteroepitaxy is doubtlessly the achievement of deposition of very high quality films on silicon, owing to its cheap.ness and ready availability of large-diameter wafers.

4.3 Thermochemistry and Mechanism of Chemical Vapor Deposition Diamond Growth 4.3.1 Transformation of Graphite to Diamond at Low Pressures In the early era of metastable diamond growth, the decisive role of atomic hydrogen for the achievement of continuous diamond CVD growth was usually explained by its etching effect on graphite; it was implied that there was simultaneously a precipitation of diamond from the gas phase and a gasification of graphite, if there was any. In at least two publications, it was explicitly demonstrated that in the presence of atomic hydrogen diamond can nucleate and grow, while graphite is simultaneously etched under formation of hydrocarbons. At first glance, this seems to be a thermodynamic paradox because in these reactions graphite is virtually converted to diamond, which should be impossible at subatmospheric pressures owing to an overall increase of the Gibbs free energy: C,

--+ Cd,

AG;98 = +2.9 kJmole-'.

(1)

It can be shown, however, that the formation of diamond from graphite under the conditions of typical diamond CVD processes is not in conflict with classical thermodynamics if the CVD process is looked at in some more detail [52 and refs. therein]. An inevitable first highly endothermic step in the CVD reaction is the dissociation of molecular hydrogen by the application of energy, for example by means of a heated filament, a flame or an electrical gas discharge, reaction (2). The gasification of graphite, reaction (3) on the other hand, is strongly favored thermodynamically: H2 +2H C,

+ 4H +CH4

AG$,OO = +75 kJ mole-',

(2)

AGy500

(3)

=

-470 kJ mole-'.

Methane reacts with H forming a series of CH, radicals (x = 3.0)

+ H +CH3 + H2 CH3 + H CH2 + H2 CH2 + H +CHI + H2

AGy5oo = -42 kJ mole- 1 ,

AGySoo= -38 kJmole- 1 ,

(6)

CHI+H+C+H2

AGy5oo

(7)

CH4

--

AGySoo=

=

+17 kJ mole-', -84 kJ mole-',

(4) (5)

408

4 Chernicd Vapor Deposition of Diurnond Films

and carbon atoms are finally added to diamond ( c d is a carbon atom in the diamond lattice):

c + Cd

--f

2cd

AGy5oo = -480 kJ mole-'.

(8)

Summarizing reactions (3) to (8), one obtains a thermodynamically strongly favored net reaction (9) for the graphite-to-diamond transformation, driven by the input of atomic hydrogen: C,

+ 8H -+

Cd

+ 4H2

AGy5oo = - 1097 kJ mole-',

(9)

4.3.2 Reactive Species in Diamond Chemical Vapor Deposition, the Role of CH3 For the purely thermodynamic argument given above, it is not necessary that the direct incorporation of atomic carbon into the diamond lattice is actually taking place in any diamond CVD mechanism. It could have been formulated alternatively using the incorporation of a methyl radical or an acetylene molecule into the diamond lattice as a central step in the reaction sequence. Many theoretical and experimental investigations have focused on the question: Which gas phase species must actually be present at the diamond surface in order to keep the crystal lattice growing? Several molecular models, involving mostly an attachment of methyl radicals or acetylene molecules to the diamond surface, have been proposed in order to evaluate possible growth mechanisms. A review of these theoretical investigations, which are not completely conclusive as to which species are actually most responsible for CVD diamond growth, was recently given by Goodwin and Butler [53]. Several gas-phase diagnostic studies have been performed especially on the hotfilament CVD process in an attempt to analyze the spatial distribution of stable and reactive species in the region between filament and substrate and to obtain information about the decisive steps in the diamond formation mechanism. One of the first quantitative investigations of this kind was performed by Celii et al. using tunable infrared-laser detection of methane, acetylene, ethylene and methyl [54,55]1.Aside from a considerable conversion of methane (starting concentration 8 x 10 ~ m - ~to) acetylene (2 x 1014cmP3) near the hot filament (tungsten, 22OO0C), methyl radicals were found with a concentration of 2 x 10l2~ m - ~ . An indirect access to the concentrations of methyl and hydrogen atoms was used by Harris and Weiner [56]. The authors used a quartz probe 3-5mm from the hot (2600 K) tungsten filament and determined the concentrations of ethylene and ethane formed by 'self-scavenging' reactions (10, 1 1) of methyl, forming the main loss channels of CH3 at relatively low pressure within the probe (<10 mbar):

+ + CH3 + CH3 --+ C2H6.

+

CH3 CH3 -+C2H5 H +C2H4 H2

(or +C2H6),

(10)

(11)

4.3 Tl~ertiioi.hmii.~tr~ atid Mecliunisrn of Clirmicul Vripor Deposition Diurnoticl Growth

409

The mole fraction of methyl in the growth chamber was then calculated by doubling the sum of the ethane and ethylene mole fractions. From the methyl mole fractions H atom concentrations in the chamber were estimated by assuming a partial equilibrium for the reaction (12): CH4+H*CH3+H2.

(12)

The methyl mole fraction increased from 2 x lop4 to 7 x lop4, when the methane concentration was raised from 0.5 to 3.2 vol-YO. The H fraction amounted to about 0.01 at 2vol-% methane. The most direct experimental approach to the species involved in diamond growth, molecular beam mass spectrometry of stable and reactive species extracted through a 300 pm orifice in the substrate plane, was used by Hsu to investigate the HF-CVD as well as the MWP-CVD process [57,58]. The most interesting results of these measurements are as follows. (i) In HF-CVD, acetylene is the predominant carbonaceous species near the substrate, if a multi-filament arrangement is used, while methane prevails under a single filament. (ii) Methyl and acetylene mole fractions saturate at lo-’ and 7 x lo-’, if the methane concentration is increased beyond 2%, accompanied by a decrease in the H atom fraction from 2 x lop’ at methane fractions below 1% to about 1 Op4 at 7 % methane (attributable to filament poisoning). (iii) Owing to a rapid removal of H atoms by surface recombination, reaction (12) is not in partial equilibrium, as assumed [56], but shifted to the methane side. Based on these relatively early investigations, no decision could be made as to which of the various species detected was actually responsible for diamond growth, although in many of these studies only CH3 and C2H2 were left as possible candidates. Very sophisticated diagnostic techniques involving nonlinear laser spectroscopy have, in recent years. been applied to study growth atmospheres in diamond CVD processes. However, the most conclusive information about which species are responsible for bringing carbon to the growing diamond film came from a number of suitably designed growth experiments. Several of these experiments point to CH3 as the principal species in commonly used CVD environments, although minor contributions of other species like acetylene cannot completely be excluded. Isotopic labeling experiments using 13CH4/’2C2Hz feed gases in the hot-filament process were performed by Chu et ul. and D’Evelyn et al., first for polycrystalline film growth [59,60], later for homoepitaxial growth on diamond {OOl}, {Oll}, and { 111) surfaces [61], and reanalyzed in order to account for the nonlinearity of the first-order Raman shift frequency versus I3C mole fraction relation, which was used to determine the isotopic composition of the diamond films grown [62]. Irrespective of the crystallographic orientation, corrected ”C mole fractions in the films were identical, within experimental errors, to the corresponding mole fractions in methane sampled directly above the growth surface. Owing to the fast reaction (12), the isotopic composition of methyl and methane can be assumed to be virtually equal. Isotopic scrambling between methane and acetylene, on the other

410

4 Chiwiccrl Vripor. Deposition of Dicltnoncl Films

hand, was suppressed by injecting the gas sufficiently close to the diamond surface. Similar experiments were performed by MWP-CVD, also demonstrating that the methyl radical is about an order of magnitude more efficient in diamond formation than acetylene [63]. Prevention of the scrambling between C, and C2 species is also the essential idea behind the flow tube studies of Martin and Hill [64], experiments with forced convection of Schiifer et ul. [65],and experiments with supersonic jets performed by Lee et ul. [66],all demonstrating the importance of methyl radicals as diamond precursors. Martin and Hill used a microwave plasma in order to generate a stream of hydrogen atoms in a flow-tube system. After injection of a small amount of methane into the flowing gas the formation of diamond could be achieved in a furnace at 8SO’C a few centimeters downstream, corresponding to a reaction time of the order of 1 ms. This time is sufficiently long for reaction (12) to come to equilibrium, but too short for major Cl-species formation. Diamond formation was also observed after acetylene injections, however the rates were about an order of magnitude smaller than for methane. The investigations of Schiifer et ul. were stimulated by observations made during HF-CVD experiments using arrays of four parallel uncoiled Ta wires, 1 1 .S mm from each other. Under conditions of free convection of the gas phase a strong decrease of the diamond growth rate from the edges of a wafer towards its center was noted. The authors concluded that the reason for this observation was a virtually complete conversion of carbon species to acetylene, which does not act as a growth species. In order to test these assumptions, they investigated the effects of vertical gas jets (hydrogen with 0.5% methane and acetylene) blown through the filament array with a velocity of about 1000 cm s-’ and striking the substrate vertically. With methane, maximum growth rates of several pm h-’ were found at the stagiiation point of the jet while in the case of acetylene a growth rate depression was observed at this point. A practically complete prevention of the interconversion between CH, and C2H, was achieved in the experiments of Lee rt al., using supersonic free jets of methyl radicals, generated by the pyrolysis of azomethane, and atomic hydrogen. Gas phase chemistry can thus be virtually eliminated. Good quality diamond crystallites could be grown on submicron diamond seeds, exposed to these jets at 650-8SO’C. Diamond deposition ceased, when either the azomethane pyrolysis or the glow discharge generating atomic hydrogen was turned of. Graphitic carbon was the main product, when an acetylene jet was used instead of a methyl jet.

4.4 Properties and Applications of Chemical Vapor Deposited Diamond The world-wide interest in CVD diamond is in a considerable part due to its unsurpassed mechanical hardness and wear resistance, promising the ultimate wear protection for tools and machine parts to be achieved by direct coating from the gas phase. However, the exceptional position of diamond under natural

4.4 Properties and Applications of Chemical Vapor Deposited Diamond

411

and man-made materials is not only based on its mechanical properties but also its optical transparency over a broad wavelength range, its unsurpassed thermal conductivity at ambient temperatures, several outstanding electrical properties (band gap, breakdown field strength, charge carrier mobilities, saturated carrier velocities) and chemical inertness. Many prospective applications make use of one or more of these physical properties. Several handbooks have meanwhile appeared which give very exhaustive accounts of the properties and applications of natural or man-made diamond single crystals [67] and diamond films [68,69]. The subsequent sections will therefore be restricted to a concise introduction of only the most important applications (which often are still in a state of infancy or even only prospective) and the properties which are relevant for these applications.

4.4.1 Diamond Coated Cutting Tools Before the deposition by CVD methods became possible, diamond was in use as a cutting material in three different forms: As a single crystalline natural material since early times; as a synthetic single crystal and as compacted and sintered polycrystalline diamond chips (pcD) after the high-pressure, high-temperature synthesis was invented in the mid-1950s. CVD allows two further forms of diamond to be used for the machining of materials: a free standing sheet of several 100 pm thickness, grown on a temporary substrate and used similarly to the pcD chip (and in many cases outperforming the corresponding pcD tool); and a diamond film of up to a few 10 pm thickness deposited directly from the gas phase onto a suitably shaped cutting tool [69]. The technical potential of diamond CVD is fully exploited only by the last method, because it allows the simultaneous conformal coating of many, possibly different, cutting tools with complex three-dimensional shapes (drills, mills, cutting tool inserts with multiple cutting edges equipped with chip breakers) in a single run. In spite of a tremendous technological potential and (at a first glance) lack of sophistication, diamond coating of cutting tools has not yet found the widespread use which was expected 15 years ago. Owing to the catalytic effect of iron on the conversion of diamond to graphite at elevated temperatures [67], diamond tools cannot be used for cutting steels. Nevertheless, there is a growing application potential for diamond tools thanks to the growing use of materials like hypereutectic AlSi alloys, A1 alloys containing alumina or silicon carbide, and fiber reinforced plastics, which are difficult to machine. The introduction of diamond-coated tools, especially those based on cemented carbide, has mainly been slowed down by the difficulty of obtaining sufficient adhesion between the hard metal substrate and diamond film reliably and reproducibly. Therefore, a major part of recent work on diamond tool coatings consisted of the development of suitable substrate pretreatment and process schemes aiming at a tight mechanical anchoring of the diamond phase, removal of Co (which has a very detrimental effect on the film adhesion because it tends to convert diamond into graphite) from the upper substrate regions and/or formation of an interlayer between WC/Co and diamond. Although a consensus as to an optimum procedure cannot be expected, due to different tool materials, deposition conditions and work

412


piece materials, significant progress has been made in tool lifetime and reliability. Therefore, diamond-coated tools are now just beginning to play a role in the machining of nonferrous and nonmetallic materials [70].

4.4.2 Thermal Conductivity of Chemical Vapor Deposited Diamond: Thermal Management Applications Pure diamond is the material with the highest thermal conductivity, K., at room temperature. A compilation made by Vandersande [68] compares the best results of thermal conductivity measurements performed by several groups on the very rare and pure (less than 100 p.p.m. N) type IIa of natural diamond. At 80-100 K, K. has a maximum value of about 125 W cm-' K-'; at room temperature about 25 W cm-' K-', more than six times the value of copper (4 W cm-' K-I), is obtained. Within the classical kinetic theory of heat conductivity, treating the phonons, carriers of heat in insulating solids, as a gas of particles with mean free path A average velocity v and specific heat C per unit volume, the expression (13) can be deduced from a few simple arguments: K.

= f(CvA).

(13)

Using v w 1O6cmC' (of the order of the velocity of sound in diamond) and the literature value of C (1.82 J K-' cmP3), a mean free path length of 0.4 pm can be calculated, roughly two orders of magnitude larger than for NaCl or silica single crystals. Theoretically, the large value of A can be traced back to the strong bonds between light atoms forming the diamond lattice. This consideration also demonstrates that a diamond crystal has to be perfect over dimensions of at least a few micrometers in order to prevent phonon scattering by point, line, or planar defects like impurities, dislocations, or stacking faults, reducing K. below the best single crystal values. The thermal conductivity is therefore critically dependent on crystalline quality and purity of a diamond crystal or film. Early thermal-conductivity measurements on CVD diamond were performed on relatively thin films (less than 30 pm) with average lateral grain sizes of only a few micrometers, grown by MWP-CVD [71] or HF-CVD [72]. For both growth methods the in-plane thermal conductivity K~~ decreased from about 10 to less than 2 Wcm-' K-', if the methane concentration in the source gas was increased to several percent, accompanied by a deterioration of the film quality as to be judged from Raman spectra and (in the case of the HF-CVD samples) an increase of the H-content from 0.1 to 1.0%. Significantly larger thermal conductivities were achieved, when films of several l00pm thickness were grown by MWP-CVD or HF-CVD [73,74]. Owing to grain dimensions attaining 50-100 ym in the top regions of films of 350-600 pm thickness, grain boundary scattering was strongly reduced and room temperature thermal conductivities of 17 Wcm-' K-'could be achieved. The average thermal conductivity of a thick diamond film measured in-plane or vertically to the film plane can be considered as an integral over increments from local contributions

4.4 Properties and Applications of' Chemical Vapor Deposited Diamond

4 13

of differential slices parallel to the film plane. Graebner and coworkers analyzed thick films with respect to the local values of in-plane and vertical thermal conductivities and discovered, that both quantities increased with the distance from the substrate surface up to values of 21 and 23.5 W cm-' K-' in the top region of a 350 pm thick film [75,77]. The latter value is already close to the best single crystal data. The high thermal conductivity of diamond forms the basis for the application of CVD diamond in heat sinks which is expected to play an increasing role in the near future: Semiconductor-based laser diodes or IMPATT microwave diodes generate a large amount of heat which has to be removed efficiently in order to prevent an early failure of the device due to excessive heating. The efficiency of water-cooled heat sinks is limited, due to the relatively small area across which the heat flux is entering into the sink. Therefore heat spreaders are used in order to enlarge the area, over which the heat is conducted to the heat sink. These spreaders must not present a large thermal resistance themselves. Single-crystalline diamonds synthesized by the high-pressure, high-temperature method are mainly used for this purpose today [67]; however, the price of these crystals increases very rapidly with size. Therefore CVD films can gain importance where larger heat sinks with dimensions reaching 1 cm or more are necessary [69]. Thermal management of computers could possibly open a much larger future market for thick diamond films of maximum thermal conductivity. Modern multichip modules involve power density levels which are difficult to cope with using conventional substrate materials. The physical properties of diamond, aside from thermal conductivity, the highly insulating nature, and a very low coefficient of thermal expansion, make it an optimal choice for managing the thermal problems encountered in present days two-dimensional packaging and in prospective threedimensional arrangements of integrated circuits [69].

4.4.3 Electrical Properties and Electronic Applications The literature on the electrical properties of CVD diamond and on the performance of first laboratory samples of electronic devices made from it, has grown to such an extent that an appropriate review would need at least an order of magnitude more space than available here. Important publications are referred to in Chapters 8 and 29 of [69], a good introduction was recently given by Prins, who had been the first to demonstrate transistor action in natural type IIb semiconducting diamond in 1982 [78]. Table 1, adapted from his article, compares the most relevant properties of diamond to those of other semiconductors. JFM and KFM are the Johnson and Keyes figures of merit, in the table normalized to the corresponding figures for Si. They are defined as JFM = (27t-'ESus and KFM

= K.(cv,/~~cE)''~,

(14)

414


Table 1. Comparison of physical properties of diamond with those of other semiconductors (after [781)

Band gap, E, (eV) Thermal conductivity, K (W cm-' K-') Carrier mobility, p (cm2V-' s-l) Holes Electrons Saturated electron velocity, us (10' cms-:) Breakdown field strength, E, ( lo5 V cm- ) Dielectric constant, E JFM (relative to Si) KFM (relative to Si)

Si

GaAs

3C-Sic

Diamond

1.1 1.5

1.4 0.46

2.3 4 5

5.47 25

600 1400 1.o 3.0 11.8 1 1

400 8500 1-2 4.0 12.5 2.6 0.5

70 800 2.5 10-50 9.1 34 6.3

1700 2150 2.7 50-200 5.7 90 35

where c in KFM is the speed of light, the other symbols are defined in the table. JFM was designed to measure the upper limits of a material's performance in a transistor handling simultaneously high frequency and high power. The superior position of diamond is mainly due to its very large breakdown field strength. In the derivation of KFM the thermal limits on the speed of a semiconductor device are considered; it is mainly the unsurpassed thermal conductivity of diamond which gives it the highest KFM value. Apart from these advantages of diamond as a semiconductor material for high-power, high-frequency applications, its very large band gap, making undoped pure diamond very highly insulating at room temperature and up to very high temperatures, is the prerequisite for prospective applications in active devices working at elevated temperatures up to 500°C or more. So far, only p-type doping with boron (incorporated from a suitable volatile source during CVD growth) has become a standard practice. In homoepitaxial films, mobilities reaching 1600 cm2V-' s-' have been reported, indicating a good crystalline quality similar to natural IIb-type diamond single crystals. Doping of the n-type by phosphorus during CVD growth is still in its infancy. Correspondingly, the list of 24 publications on active diamond devices in [69] mainly refers to unipolar field-effect transistors, mostly prepared from in situ p-doped homoepitaxial films or B ion implanted layers on single crystals. In several cases transistor function and operation at elevated temperatures up to 500°C were demonstrated. In order to develop an economically viable diamond device technology, however, diamond single crystals are hardly suitable substrates. Owing to grain-boundary effects in randomly oriented polycrystalline films grown on cheap silicon substrates, device performance in this material is inferior by orders of magnitude. Therefore, the growth of heteroepitaxial films on Si or Sic substrates might gain importance for future electronic applications. Even if the films are still polycrystalline, the detrimental effect of grain boundaries on physical properties is reduced owing to much smaller intergrain angles. Electrical properties (in-plane hole mobility, through-plane leakage currents) as well as possibly the thermal conductivity are therefore superior to those of randomly oriented material [79]. For a metal-oxide

4.4 Properties and Applications of Chemical Vapor Deposited Diamond

4 15

E"AC

----xt-

clean

5.5 eV

H-terminated (1 11) diamond surface

Figure 8. Alignments of valence and conduction bands of clean and H-terminated (1 11) surfaces of p-doped diamond, EF = Fermi level. The figures have been aligned at the vacuum level (adapted from [SO]).

silicon field-effect transistor (MOSFET) made from heteroepitaxial material, transistor operation significantly improved over that of a random material has already been demonstrated up to 400°C [69]. Apart from transistors, several other solid state devices have been discussed [78], like junctions, photon and electron beam switches and various kinds of sensors. One property of diamond which has stimulated considerable interest in the recent years is the negative electron affinity (NEA) of suitably prepared surfaces [78, SO]. The electron affinity, x,of a material is defined as the difference between the energy of a free electron in vacuum and the bottom of the conduction band Eva,- E,. In Fig. 8 the electronic bands of p-doped clean and H-terminated (1 1 1) diamond surfaces near the surface are depicted, based on the results of UV-photoemission measurements. For the H-terminated surface, the electron affinity becomes negative; once an electron is injected into the conduction band from a suitable contact or by UV excitation, it will easily leave the crystal and be emitted into vacuum. This effect, which is also observed on monohydride terminated (1 00) surfaces, is not unique to diamond but was also observed in a few other semiconductors with high band gaps [80]. Apart from a scientific interest, the NEA of diamond makes it an attractive candidate for the replacement of thermionic emitters as electron beam sources and as a miniature electron emitter for field emission displays.

4.4.4 Electrochemical Use of Chemical Vapor Deposited Diamond The chemical inertness of diamond, even in hot acidic or caustic oxidizing media [8 I], and the possibility of making conductive CVD diamond films with resistivities down to less than 0.01Rcm [82] on a wide variety of substrates and over large areas, has in the recent years stimulated an increased interest in applications for analytical and preparative electrochemistry. In a comparative study, Swain

4 16


investigated the response of electrodes consisting of heavily boron-doped diamond films, highly ordered pyrolytic graphite (HOPG) and glassy carbon (gC), towards repeated potential cycling (PC) for 2 h in a solution of 1 M HN03 0.1 M NaF at 50°C. The electrode properties were investigated before and after PC by Raman spectroscopy, optical and scanning electron microscopy, and electrochemical techniques. The potential ranges extended to the onset potentials for the evolution of hydrogen and oxygen, for each electrode. For diamond, as a result of the potential cycling the anodic onset potential shifted from 1.2 to 1.7V (vs. Ag/AgCl), at the cathodic potential of -0.6V, a current reduction was observed. Simultaneously, the total charge in a cyclovoltammogram taken after PC was reduced by a factor of two. These changes were discussed in terms of a surface cleaning by removal of nondiamond phases. In contrast to diamond, a much larger degree of microstructural damage and surface oxydation was observed for HOPG and gC electrodes [83]. The width of the ‘potential window’ with very low background currents in aqueous electrolytes depends critically on the diamond phase purity. Comparing HF-CVD samples grown from atmospheres with 1 and 4% methane, and doped with about lo2’ boron atoms per cm3, a potential range extending from -1.25V to +2.3V (vs. SHE) without significant water decomposition could be detected for the high quality sample grown from the carbon-poor atmosphere. Samples grown from 4% methane, on the other hand, showed a range of only -0.6 to 1.7 V, owing to their content of nondiamond phases. The inert nature of the Hterminated diamond surface can be made responsible for the high overpotentials observed for hydrogen and oxygen [84] as well as halogen evolution [85]. The evolution of halogen is promoted by the absorption of halogen atoms on the electrode surface (16); however, owing to a small affinity of the hydrogenated diamond surface towards absorption of atomic species, the subsequent reaction (17) is kinetically strongly hindered:

+

X-

+

Xads X-

-

Xads

+X2

+ e- ,

+ e-.

(16) (17)

It is inferred that generally electrochemical reactions involving surface-absorbed species will proceed comparably slowly on pure diamond surfaces. On the other hand, outer sphere electron transfer reactions not requiring a strong surface interaction of the species involved, will in principle not be affected by the inert nature of the diamond surface, as to be demonstrated, for example, by the reversible behavior of the Ru(III)/Ru(II) redox couple [85]. The actual technical potential of diamond electrodes for analytical and preparative electrochemical applications can so far not be fully assessed. From various prospective applications mentioned in the literature, making use of the exceptional stability, wide potential window and low background currents of diamond electrodes, only one example shall be mentioned here, the reduction of nitrate to ammonium according to the net Eq. (18) NO,

+ 7H20 + 8e- +N H 4 0 H + 90H-,

Eo = -0.15 V vs. SHE.

(IS)

References

4 17

Thanks to the special properties of diamond electrode surfaces, this reaction can be run with high efficiency and stability [86]. A potential application is the removal of nitrates from the effluents of combustion processes or from drinking water.

4.5 Summary Only a small selection of possible applications of CVD diamond could be touched on here. Many of the expectations which were placed on the applications of CVD diamond in the early years of the ‘new diamond era’ have so far not come true, due to reasons like still too high deposition costs or insufficient film qualities for certain applications, major advances in competitive materials and technologies, or just the lack of a market. Nevertheless, thanks to the tremendous progress which has been made in the recent years and will undoubtly be made in the future, CVD diamond can be expected to play an ever increasing role as the ultimate material for a variety of technical purposes.

References 1. J. C. Angus, in The Physics of Diamond, A. Paoletti and A. Tucciarone (Eds), 10s Press, Amsterdam, Oxford, Tokyo, Washington DC, 1997, p. 1. 2. R. C . DeVries, A. Badzian, and R. Roy, M R S Bulletin, 1996, 65-75. 3. P. K. Bachmann and H. Lydtin, in Diumondand Diamond-like Films and Coatings, R. E. Clausing, L. L. Horton, J. C. Angus, and P. Koidl (Eds), Plenum, New York, 1991, p. 829. 4. S. Matsumoto, Y. Sato, M. Kamo, and N. Setaka, Jpn. J. Appl. Phys. 1982, 21, L183-Ll85. 5. S. Matsumoto, Y. Sato, M. Tsutsumi, and N. Setaka, J. Muter. Sci. 1982, 17, 3106-31 12. 6. S. Okoli, R. Haubner, and B. Lux, J. PhJjsique ZV 1991, C2, 923-930. 7. M. Sommer and F. W. Smith, J . Muter. Res. 1990, 5, 2433-2440. 8. B. Singh, Y. Arie, A. W. Levine, and 0. R. Mesker, Appl. Phys. Lett. 1988, 52, 4511152. 9. L. Schafer, U. Bringmann, C.-P. Klages, U. Meier, and K. Kohse-Hoinghaus, in Diamond and Diamond-Like Films and Coatings, R. E. Clausing et al. (Eds), Plenum Press, New York, 1991, p. 643.. 10. F. Jansen, M. A. Machonkin, and D. E. Kuhman, J. Vac. Sci. Technol. 1990, AS, 3785-3790. 11. E. Gheeraert, A. Deneuville, M. Brunel, and J. C. Oberlin, Diamond Rel. Muter. 1992,1,504-507. 12. M. Griesser, G. Stingeder, M. Grasserbauer, H. Baumann, F. Link, P. Wurzinger, H. Lux, R. Haubner, and B. Lux, Diamond Rel. Muter. 1994, 3, 638-644. 13. Y. Hirose and Y. Terasawa, Jpn. J. Appl. Phys. 1986, 25, L519-L521. 14. M. Murakawa, S. Takeuchi, H. Miyazawa, andY. Hirose, Surf: Coat. Technol. 1988,36,303-310. 15. T. Kawato and K. Kondo, Jpn. J. Appl. Phys. 1987,26, 1429-1432. 16. Y. Saito, K. Sato, H. Tanaka, K. Fujita, and S. Matsuda, J. Muter. Sci. 1988, 23, 842-846. 17. C . Chen, Y. C. Huang, S. Hosomi, and I. Yoshida, Muter. Res. Bull. 1989, 24, 87-94. 18. J. A. Mucha, D. L. Flamm, and D. E. Ibbotson, J. Appl. Phys. 1989, 65, 3448-3452. 19. S. J. Harris and A. M. Weiner, Appl. Phys. Lett. 1989, 55, 2179-2181. 20. D. N. Belton and S. J. Schmieg, J . Appl. Phys. 1991,69, 3032-3036. 21. P. K. Bachmann, D. Leers, and H. Lydtin, Diamond Rel. Muter. 1991, 1, 1-12. 22. G. Z. Cao, J. J. Schermer, W. J. P. van Enckevort, W. A. L. M. Elst, and L. J. Giling, J. Appl. Phys. 1996, 79, 1357-1364.

418


23. M. Kamo, Y. Sato, S. Matsumoto, and N. Setaka, J. Cryst. Growth, 1983, 62, 642-644. 24. C.-P. Klages, Appl. Phys. A 1993, 56, 513-526. 25. P. K. Bachmann, in Handbook of Industrial Diamond and Diamond Films, M. A. Prelas, G. Popovici, and L. K. Bigelow (Eds), Marcel Dekker, New York, Basel, Hong Kong, 1998, Chapter 21. 26. H. Kawarada, K. S. Mar, and A. Hiraki, Jpn. J . Appl. Phys. 1987, 26, L1032-LIO34. 27. J. Suzuki, Jpn. J . Appl. Phys. 1989, 28, L281LL283. 28. A. A. Chernov, Modern Crystallography ZIZ, Springer Verlag, Berlin, Heidelberg, New York, 1984, p. 283. 29. A. Van der Drift, Philips Res. Repts. 1967, 22, 267-288. 30. P. Hartman, in Crystal Growth; An Introduction, P. Hartmann (Ed.), North-Holland, Amsterdam, 1973, p. 367. 31. L. J. Giling and W. J. P. Van Enckevort, Surface Science, 1985, 161, 567-583. 32. C. Wild, P. Koidl, W. Miiller-Sebert, H. Walcher, R. Kohl, N. Herres, R. Locher, R. Samlenski, and R. Brenn, Diamond Rel. Mater. 1993, 2, 158-168. 33. R. E. Clausing, L. Heatherley, L. L. Horton, E. D. Specht, G. M. Begun, and Z. L. Wang, Diamond Rel. Muter. 1992, 1, 411415. 34. Ch. Wild, N. Herres, and P. Koidl, J. Appl. Phys. 1990, 68, 973-978. 35. Ch. Wild, R. Kohl, N. Herres, W. Muller-Sebert, and P. Koidl, Diamond Rel. Muter. 1994, 3, 373-381. 36. H. Maeda, K. Ohtsubo, M. Irie, N. Ohya, K. Kusakabe, and S. Morooka, J . Muter. Res. 1995, 10, 3115-3123. 37. R. Locher, C. Wild, N. Herres, D. Behr, and P. Koidl, Appl. Phys. Lett. 1994,65, 34-36. 38. X. Jiang, W. J. Zhang, M. Paul, and C.-P. Klages, Appl. Phys. Lett. 1996, 68, 1927-1929. 39. B. R. Stoner and J.T. Glass, Appl. Phys. Lett, 1992, 60, 698-700. 40. X. Jiang and C.-P. Klages, lecture at Diamond 92, Heidelberg 1992, published in: Diamond Rel. Muter. 1993,2, 1112-1 113. 41. S. Yugo, T. Kanai, T. Kimura, and T. Muto, Appl. Phys. Lett. 1991,58, 10361038. 42. B. R. Stoner, G.-H. M. Ma, and S. D. Wolter, J.T. Glass, Phvs. Rev. B 1992.45. 11067-11084. 43. X. Jiang, C.-P. Klages, M. Rosler, R. Zachai, M. Hartweg, and H.-J. Fusser, Appl. Phys. 1993, A51,483489. 44. K. Schiffmann and X. Jiang, Appl. Phys. 1994, A59, 17. 45. J. Gerber, S. Sattel, K. Jung, H. Ehrhardt, and J. Robertson, Diamond Rel. Muter. 1995, 4, 559-562. 46. S. Sattel, J. Gerber, and H. Ehrhardt, Phys. Stat. Sol. ( A ) 1996, 154, 141-153. 47. Y. Lifshitz, S. R. Kasi, and J. W. Rabalais, Phys. Rev. Lett. 1989, 62, 1290-1293. 48. J. L. Robertson, S. C. Moss, Y. Lifshitz, S. R. Kasi, J. W. Rabalais, G. D. Lempert, and E. Rapoport, Science, 1989, 243, 1047-1050. 49. X. Jiang, K. Schiffmann, and C.-P. Klages, Phys. Rev. B 1994,50,8402-8410. 50. T. Suzuki and A. Argoitia, Phys. Stat. Sol. ( A ) 1996, 154, 239-254. 51. H. Kawarada, T. Suesada, and Nagasawa, Appl. Phys. Lett. 1995, 66, 583-585. 52. J. C. Angus, in The Physics of Diamond, A. Paoletti and A. Tucciarone (Eds), 10s Press, Amsterdam, Oxford, Tokyo, Washington DC, 1997, p. 9. 53. D. G. Goodwin and J. E. Butler, in Handbook of Industrial Diamond and Diamond Films, M. A. Prelas, G. Popovici, and L. K. Bigelow (Eds), Marcel Dekker, New York, Basel, Hong Kong, 1998, Chapter 11. 54. F. G. Celii, P. E. Pehrsson, H.-T. Wang, and J. E. Butler, Appl. Phys. Lett. 1988,52,2043-2045. 5 5 . J. E. Butler and F. G. Celii, in Proc. 1st Znt’l Symp. on Diamond and Diamondlike Films, J. P. Dismukes et al. (Eds), The Electrochemical Society, Pennington, NJ, 1989, p. 317. 56. S. J. Harris and A. M. Weiner, J . Appl. Phys. 1990,67,6520-6526. 57. W. L. HSU,Appl. P h y ~Lett. . 1991,59, 142771429, 58. W. L. Hsu, in Proc. 2ndZnt’l Symp. on DiamondMaterials, A. J. Purdes et al. (Eds), The Electrochemical Society, Pennington, NJ, USA, 1991, p. 217. 59. C. J. Chu, M. P. D’Evelyn, R. H. Hauge, and J. L. Margrave, Muter. Res. SOC.Symp. Proc. 1990, 162, 85-90. 60. C. J. Chu, M. P. D’Evelyn, R. H. Hauge, and J. L. Margrave, J . Muter. Res. 1990,5,2405-2413.

References

61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79.

80. 81. 82.

83. 84. 85. 86.

419

C. J. Chu, M . P. D’Evelyn, R. H. Hauge, and J. L. Margrave, J . Appl. Phys. 1991,70, 1695-1705. M. P. D’Evelyn, C. J. Chu, R. H. Hauge, and J. L. Margrave, J . Appl. Phys. 1992,71, 1528-1530. C. E. Johnson, W. A. Weimer, and F. M. Cerio, J. Muter. Res. 1992, 7, 1427--1431. L. R. Martin and M. W. Hill, J. Muter. Sci. Lett. 1990, 9, 621-623. L. Schafer, M. Sattler, and C.-P. Klages, in Applications of Diamond and Related Materials, Y. Tzeng e f al. (Eds), Elsevier, Amsterdam, 1991, p. 453. S. S. Lee, D. W. Minsek, D. J. Vestyck, and P. Chen, Science 1994, 263, 1596-1598. J. E. Field (Ed.), The Properties qf Diamond, Academic Press, London, 1997. G. Davis (Ed.), Properties and Growth qf Diamond, EMIS Datareview Series, INSPEC, London, UK, 1994. M. A. Prelas, G. Popovici, and L. K. Bigelow (Eds.) Handbook of Industrial Diamond and Diamond Films,Marcel Dekker, New York, Basel, Hong Kong, 1998. C.-P. Klages, M. Fryda, T. Matthee, L. Schafer, and H. Dimigen, in Proceedings ofthe 14th International Plansee Seminar, G. Kneringer, P. Rodhammer und P. Wilhartitz (Eds), Plansee AG, Reutte, Austria, 1997, Vol. 3, 1-15. A. Ono, T. Baba. H. Funamoto, and A. Nishikawa, Jpn. J . Appl. Phys. 1986, 25, L808-LSlO. D. T. Morelli, C. P. Beetz, and T. A. Perry, J . Appl. Phys. 1988, 64,3063-3066. D. T. Morelli, T. M. Hartnett, and C. J. Robinson, Appl. Phys. Letf. 1991, 59, 2112-2114. J. E. Graebner and J. A. Herb, Diamond Films Technol. 1992, 1, 155-163. J. E. Graebner, S. Jin, G. W. Kammlott, J. A. Herb, and C. F. Gardinier, Appl. Phys. Lett. 1992, 30, 1576-1 578. J. E. Graebner, S. Jin. G. W. Kammlott, B. Bacon, L. Seibles, and W. Banholzer, J. Appl. Phys. 1992,71, 5353-5356. J. E. Graebner, S. Jin, G. W. Kammlott, J. A. Herb, and C. F. Gardinier, Appl. Phys. Lett. 1992, 359,401403. J. F. Prins, in The Physics of Diamond, A. Paoletti, and A. Tucciarone (Eds), 10s Press, Amsterdam, Oxford, Tokyo, Washington DC, 1997, p. 41 1. C.-P. Klages in The Physics of Diamond, A. Paoletti, and A. Tucciarone (Eds), 10s Press, Amsterdam, Oxford, Tokyo, Washington DC, 1997, p. 73. R. J. Nemanich, P. K. Baumann, M. C. Benjamin, S. P. Bozeman, and B. L. Ward, in The Physics of Diamond, A. Paoletti and A. Tucciarone (Eds), 10s Press, Amsterdam, Oxford, Tokyo, Washington DC, 1997, p. 537. P. M. Natishan and A. Morrish, Materials Letters, 1989, 8. 269-272. M . Werner, R. Job, A. Zaitzev, W. R. Fahrner, W. Seifert, C. Johnston, and P. R. Chalker, Phys. Stat. Sol. (‘A), 1996, 154, 385-393. G. M . Swain, J. Electrochem. Soc. 1994, 141, 3382-3393. H. B. Martin, A. Argoitma, U. Landau, A. B. Anderson, and J. C. Angus, J. Electrochem. Soc. 1996, 143, L133-LI36. N. Vinokur, B. Miller, Y. Avyigal, and R. Kalish, J . Electrochem. Soc. 1996, 143, L238-L240. R. Tenne, K. Patel. K. Hashimoto. and A. Fujishima, J . Elecfroanal. Chem. 1993,347,409415.


5 Vapor Phase Deposition of Cubic Boron Nitride Films K. Bewilogua and F. Richter

5.1 Introduction Connected with the advancement of synthetic diamond as a technical material, cubic boron nitride (cBN) has attracted increasing interest during the past decades. Boron nitride exists in high density diamond-like and in low density graphite-like modifications. High density cubic boron nitride has a zinc-blende structure with eight atoms in the elementary cell. Each B atom and each N atom is surrounded by four N or B atoms, respectively. Like in the diamond structure the neighboring atoms are tetrahedrally arranged with fourfold coordination. Diamond and cBN have similar lattice constants. However, diamond has a pure covalent bonding and cBN a covalent-ionic mixed bonding. Hexagonal boron nitride (hBN) and graphite have similar structures, both with a threefold coordination of nearest neighbors. Beside hBN and cBN two other BN structures are known, the rhombohedral (rBN) and the metastable high-density wurtzite-type BN (wBN). Apparently the wBN phase only can be synthesized by shock compression. On the one hand hBN and rBN and on the other hand cBN and wBN have similar structures and the same short range order but different stacking sequences of atomic planes (see [l]). In Table 1 the structural data as well as some properties of cBN, diamond, hBN, and graphite are summarized. Cubic BN is the second hardest material after diamond. It does not react with iron in tribological contact [2] and can therefore (in contrast to diamond) be used to coat tools for machining of ferrous materials. Because of its wide band gap (Eg= 6.4eV [ 3 ] ) , cBN transmits electromagnetic radiation in a broad range including the near ultraviolet as well as the visible and infrared region. Moreover, cBN can be doped both n- and p-type [4] and has good thermal conductivity. Therefore, additional potential applications are as protective coatings for optical devices or in electronics, either as a high temperature semiconductor or as heat dissipating material. Analogous to graphite in the carbon case, boron nitride exhibits a low-density relatively soft hexagonal phase hBN. Formation of boron nitride by chemical reaction at atmospheric pressure yields the hexagonal phase. However, by using the high pressure, high temperature (HPHT) synthesis established by Wentorf [5] it is possible to transform hBN into cBN. In this method, ( p , T ) values in the range 1200-2000°C and 4.5-7.5 GPa, respectively [6], were applied corresponding to conditions where cBN is thermodynamically stable [7] and formed comparatively readily. Today, small crystallites of cBN having a typical size in the several 100 mm range [S] are produced in considerable quantities with this method and are processed

42 1

5.2 Empirical Results Table 1. Structure and properties of cBN, diamond, hBN, and graphite. cBN

Diamond

hBN

Graphite

Crystal structure Lattice constants (nm)

zinc blende 0.362

diamond 0.356

Bond length (nm) Interplanar spacing (nm) Density (gcm-') Hardness (GPa) Young's modulus (GPa)

0.157 3.48 5&70 700-900

3.52 9&100 1050

Band structure

Indirect semiconductor 6.1 2.12

Indirect semiconductor 5.5 2.42

hexagonal a 0.250 c 0.666 0.144 0.333 2.28 approx. 10 a 87 c 34 Direct semiconductor 5.2 1.72

hexagonal a 0.246 c 0.671 0.142 0.335 2.27 <10 a 19 c 5 Metallic

10'O

1013

Optical band gap (eV) Refractive index (A = 589.3 nm) Resistivity (0 cm) Dopability Thermal conductivity (W m-I K - I ) Oxidation at air ("C) Reactivity with iron at elevated temperatures

0.154 ~

a c

10'' lo1*

-

a c

lo-' 1

p and n <750

P (n ?) 2000

1

6.4-10.8

>loo0 No

>500 Yes

>I000 No

-

Yes

mainly to sintered polycrystalline boron nitride (pcBN) pieces as cutting edges for tools. However, by growing thin BN films from the vapor phase it is possible to get cubic BN as a thin film at moderately increased temperatures as well as subatmospheric pressure. Starting in the SOs, mainly by Japanese researchers [9], great efforts have been made in this field up to now. The aim of this chapter is to present an overview of some important recent developments in cBN thin film deposition. The physical mechanisms which are causal for cBN formation in ion-assisted thin film deposition will be discussed in some detail.

5.2 Empirical Results 5.2.1 Deposition Methods A great variety of deposition methods has been successfully used up to now to deposit cBN-containing thin films, among them both physical (PVD) and chemical vapor deposition (CVD) techniques. All these methods have in common that the growing film is exposed to an intense ion bombardment which plays a crucial role in cBN thin film formation. Among the PVD methods initially such techniques

422

5 V q m Phuse Deposition

of' Cubic Boron Nitride Films

did predominate which use electron beam evaporation of boron. The necessary ion impact to the growing film was performed either by special activation and ionization of the nitrogen containing gas atmosphere together with a suitable substrate bias to accelerate the ions [10-12] or by means of ion beam sources [13,14]. The use of ion beam sources to realize ion beam impact to the growing film has also been combined with boron vaporization by ion beam sputtering [15,16] or laser ablation [17] of either a boron or BN target. Mass-selected ion beam deposition (MSIB) [18] and cathodic arc deposition (CAD) [19] have also been successfully used to deposit cBN thin films. With the last mentioned two methods the film is formed essentially by deposition of ions. Because of its potential for application, R F sputtering [20,21] and R F magnetron sputtering [22,23] has attracted increasing attention. Deposition by d.c. magnetron sputtering (which is even more interesting from an application point of view) has also been successfully applied to cBN thin films, either using a B4C target [24] (section 5.4.1) or a boron target which is operated at elevated temperature where boron gets conductive [25] (section 5.4.2). Different variants of CVD techniques were applied, for example R F glow discharge using the gases diborane B2H6, N2, Ar, He, and H2 [26] as well as electron cyclotron resonance (ECR) plasma CVD processes with B2H6and N2 [27] or with the vapor of N-trimethylborazine [28]. The authors reported that cBN could only be generated if the negative substrate bias was sufficiently high. BN films deposited by such CVD methods seem to have in general somewhat lower cBN contents than those prepared by PVD techniques. Probably this is due to the incorporation of hydrogen into the films [29]. Zhang et ul. [30,31] deposited films on nickel in a hot filament assisted R F glow discharge with the reactant gases being B2H6, NH3, and H2. Obviously the substrates were not biased. Scanning electron microscopic images revealed crystallites with dimensions of about 2 pm. As an essential reason for the rather large crystallites, a catalytic effect of Ni was assumed. Unfortunately, for the evidence of the cubic phase only X-ray diffraction patterns, but no infrared spectra, were published. Therefore considerable doubt that these crystallites consist of cBN remains. Konyashin et al. [32] deposited BN films using plasma assisted CVD with gas mixtures of argon, hydrogen, and borane-ammonia (BH3-NH3) without substrate bias at temperatures up 1000°C. Regrettably no data on hydrogen content and B and N concentrations in the films were given. From IR spectra and X-ray diffraction patterns of films deposited with high hydrogen concentrations in the process gas, the existence of single phase cBN was concluded. The authors assumed a deposition mechanism which includes a selective etching of sp2 bonded BN and the formation of metastable BNH, species. Reinke et al. [33] analyzed numerous deposition methods with respect to film structure (cBN or hBN) on the one hand and ion bombardment on the other hand. To describe the ion bombardment quantitatively they used the ion energy and the ratio of fluxes of impinging ions (i) and deposited boron atoms (a). Figure 1 shows the results for the ratio i/u against the ion energy. The total number of deposited atoms can be estimated taking into account the deposition rate and the mass densities of the BN phases. As can be seen from that presentation,

5.2 Empirical Results

423

0 0 0

0

0

oo

0

0 0 .

0

0 0

I

10

100

0

.

1000

Ion energy [eV] Figure 1. Data compilation according to Reinke et al. [33] showing the phase of BN thin films in dependence on ion energy, Ei, and the ratio of bombarding ions and deposited atoms, i/a, for numerous deposition methods. Filled and open symbols represent cBN and hBN films, respectively.

cBN has been grown in a broad range of ion energies Ei,beginning with quite low energies connected with a large i/u (these are mostly CVD methods) up to higher E; together with small i/u. The limiting case is represented by mass selected ion deposition where the film grows solely by deposition of Bf and Nf ions. In this case i/u is one and cBN was formed at ion energies down to 125 eV [34]. In their successful attempts to grow cBN thin films, most authors have applied moderately increased substrate temperatures, T,, mostly in the range between 200 and 500°C. However, the growth of cBN thin film is also possible at temperatures as high as 1000°C [35]. The low temperature limit for cBN film deposition is at about 150'C, as has been found by Mirkarimi [36] for ion-assisted pulsed laser deposition and more recently by Hofsass et ul. [34] for MSIB deposition.

5.2.2 Morphology and Structure of cBN Films Up to now, most work on cBN deposition has been done using single crystal silicon substrates. Those films grow in a typical phase sequence which has been shown for the first time by Kester et al. [37] in 1993 and since then confirmed by many groups: At first an amorphous layer at the interface to the substrate is formed which is usually referred to as aBN. This layer probably contains also some silicon from the substrate and is due to ion-induced intermixing. After that a turbostratic boron nitride (tBN) layer is formed. Turbostratic boron nitride is a BN form consisting of nearly parallel hBN-like layers which, however, do not have defined three-dimensional orientation. The distance of these layers is a few percent enlarged in comparison to the hBN crystal (see e.g. [ 11). Under deposition conditions which

424

5 Vapor Phase Deposition of Cubic Boron Nitride Films

Figure 2. Cross-section TEM micrograph of a cBN film on single crystal silicon with the typical interlayers of amorphous material (probably a mixture of B, N, and Si), turbostratic BN.

finally yield cBN growth (i.e. under proper substrate temperature and ion bombardment as discussed above) the tBN exhibits a preferred orientation such that the caxis of the tBN clusters are predominantly perpendicular to the substrate surface normal. After a certain amount of tBN deposition the film structure changes into cBN. In Fig. 2 a cross-section transmission electron micrograph of this typical phase sequence is shown. The thickness of the amorphous and tBN interlayers depends on the deposition conditions. Especially the tBN thickness may be very large, however, under optimum conditions the transition to cBN growth takes place very soon and the thickness of the hexagonal BN layer is only about 1 nm or even not detectable in cross-section TEM. The cBN films are polycrystalline with columnar grains about 10-20nm in diameter extended perpendicular to the surface. The cBN crystallites have a preferential orientation. Some authors report on a (1 10) texture [38,39], others [40] found a so-called double-fiber texture. Both textures are characterized by an alignment of { 11l} planes perpendicular to the surface. The difference is that in the case of the (110) texture the (111) planes have a preferred azimuthal orientation (i.e. the (1 10) direction is parallel to the substrate normal), whereas in the case of the double-fiber texture the orientation about the in-plane [l 1 11 axis is random. However, both textures do support the suggestion that the structural transition from hBN to cBN growth is characterized by three (1 11) spaces of cBN (dl11 = 0.209 nm) matching two (0002) spaces of tBN

5.2 Empirical Results

-Yr

425

c-BN (1 1I), 2.09 A

Figure 3. High-resolution cross-section electron micrograph of the tBN/cBN 2 : 3-transition in a magnetron sputtered BN film.

(do,,, M 0.34 nm), thus forming a semicoherent interface. In the meantime this model has been confirmed using cross-section TEM by us and others (see Fig. 3). A necessary condition for cBN phase formation in thin films is that boron and nitrogen are incorporated in a nearly 1: 1-ratio. Based on highly accurate composition measurement using neutron depth profiling, Hackenberger et al. [41] investigated the relationship between film stoichiometry and the phases present in the film. From these measurements, together with an analysis of data from the literature, they found evidence that film stoichiometry is one of the factors that stabilize the cubic phase. Boron nitride films containing both the hBN and cBN phase cause strong infrared absorption bands at about 1050cm-' owing to the presence of cBN (transversal optical mode) [42] as well as at about 800 and 1400cm-' corresponding to B-N bending (out-of-plane vibration) and stretching (in-plane vibrations) modes, respectively [43]. Actually, the IR absorption features depend on the chemical short-range order of the lattice, and their appearance is only a necessary (not a sufficient) condition for the existence of the respective phases. There exist some more BN phases with either sp2 or sp3 short-range order which show essentially the same IR absorption as hBN and cBN, respectively. However, if used in combination with more direct methods such as electron diffraction, IR absorption is a fast and reliable method for determination of the phase content of BN thin films and has found widespread application. In Fig. 4 the IR absorption spectrum of a BN film is given as an example which shows a distinct absorption near 1050cm-' and only weak absorption at about

426

5 Vupor Phusr Depositiori of Cubic Boron Nitride Films

1 .o

0.9 T-

2

Y

t 600

8~1 8

800

,

-

1000

1200

-7, ' 1 7 --1400 1600 1800 2000 7-

Wavenumber [cm'] Figure 4. Infrared absorption spectrum of a 130 nm thick BN film on silicon consisting mainly of nanocrystallline cBN

800 and 1400 cm-I, respectively. This film consists practically of phase pure cBN. The slight sp2-like absorption is mainly due to the non-cubic (amorphous and turbostratic) interlayer mentioned above. Probably the not perfectly sp3 bound grain boundary regions of the nanocrystalline cBN film contribute to the sp2 signal, too. There were some attempts to reveal quantitative data on phase contents in mixed BN films from IR spectra. Jager et al. [44] determined the cBN content from the absorption coefficients a h (at a wave number near v = 1400cm-' for hBN) and a, (near v = 1050cm-' for cBN). The ratio of the volume fractions,f,, and,fvhof both phases was calculated as fvc/fih = 0.6ac/ah. However, in practice it is much more simple to consider the peak heights 1;(i = h, c) in the spectra than to determine the absorption coefficients ai. Schiitze et al. [24] derived the rough approximation Z; z Q; (i = h,c) which leads to,jvc/fvh= 0.61,/Zh. Considering the mass fractions,f,, it follows that,fi,,,/f;,,h = ( p , B N / p h B N ) *jvc/fvh and,fm,/fn,h z Z,/Z, and the cBN content simply can be calculated as Zc/(Zc + Zh). Using this relation one gets for the film with the spectrum shown in Fig. 4 a cBN content of SO-90% which is also in good agreement with the impression gained from cross-section TEM images.

5.3Morlel.~of c B N Forrnution

427

5.2.3 Film Adhesion An important and typical experimental result is the strong biaxial compressive stress which is always found in cBN thin film. The stress usually has values of 4-10 GPa which is high enough to cause serious adhesion problems as soon as the film thickness exceed a few 100 nm. In fact, bad film adhesion is the most important obstacle in making cBN ready for application. Beside high stress. the ambient conditions also influence the adhesion of cBN films. Several research groups observed that cBN films stored in vacuum, dry air, or nitrogen did not delaminate while those kept under normal humid environmental conditions delaminated much faster from the substrates. In detail this effect was discussed by Cardinale rt al. [45]. They suggested a generation of oxide compounds by an aqueous oxidation of BN. This oxidation is accompanied by a volume expansion which destroys the film. Probably water vapor comes into the films via defects which can be assumed to be located in the regions between the columnar grains. Therefore, the way to overcome the adhesion problems should be to prepare more homogeneous films with higher density. Most cBN deposition experiments reported so far were carried with silicon substrates. Only a few papers concerned ‘non-silicon’ substrates. Mirkarimi rt al. [46] used different metallic substrates and found a decrease of the cBN content with decreasing metal hardness. Schutze rt al. [47] found that it is not possible to deposit cBN films on steel substrates under conditions which are suitable for silicon substrates. On steel cBN could only be prepared after deposition of a high-density hBN buffer layer. Concerning the effect of the substrate material on the cBN film growth there are still a lot of open questions.

5.3 Models of cBN Formation McKenzie et al. [48] considered the above mentioned strong biaxial stress in cBN and mentioned that the respective stress tensor may be broken down into two components: a tensor describing a hydrostatic pressure and a tensor describing a pure shear stress. The hydrostatic pressure has a magnitude of two thirds of the biaxial compressive stress. Because of these stress values in thin BN films, the hydrostatic pressure may have a magnitude of several GPa. This is approximately the same magnitude as the pressure values applied to transform hBN into cBN using the HPHT method. The authors argue that provided the pressure is high enough, the formation of cBN is in principle possible, even at low temperature, due to the fact that the layer is formed from individual atoms. As a possible mechanism, they discuss the localized melting of small regions of the layer due to energy transfer from the ions (‘thermal spikes’) followed by the solidification of the material as cBN under the influence of the high stress. The authors back up their model by experimental results. For instance they have varied the conditions for boron nitride thin film deposition which yielded different

428

5 Vupor Plitrsr Deposition o f ’ Cubic BOIYJ~I Nitride Filtm

100 a, 0

0 0

20

c

I

m 8

0

2

4 6 8 Compressive Stress [GPa]

10

12

Figure 5. Cubic phase content in RN thin film deposited by IBAD against the compressive film stress (adapted from [48]).

stress level in the films. An analysis of the respective cBN content of the films (Fig. 5) showed that a certain threshold of compressive stress exists above which cBN formation in the film takes place. They also reported on a typical dependence of the film stress on the deposition time (zero stress + increase up to maximum + decrease to saturation value) which at least qualitatively agrees with the above mentioned sequence aBN + hBN cBN. An important conclusion from this model is that the compressive stress not only exists in conjunction with the cBN phase in the layer, but is the cause of the cBN formation. An analogous model has already been applied by the same authors to explain the formation of tetrahedral amorphous carbon (taC) [49]. Several authors conducted experiments under well-controlled conditions and varied both the energy and mass of the ions in order to discriminate between the specific role of ion energy and momentum. Kester and Messier [ 141 found a distinct increase of the cBN content of the films at a certain threshold value of the transferred momentum per arriving atom. For the momentum transferred from a single impinging ion they used an expression given in [50] for a single, binary, elastic, head-on collision betwcen the ion and an atom of the film. These results underline the relevance of momentum transfer processes in cBN formation. However, because of the quite restricting suppositions as well as an error existing in the formula used, their quantitative accuracy is limited. Mirkarimi rt al. [17] reported on similar BN deposition experiments performed by pulsed laser evaporation of an hBN target together with an ion beam bombardment of the growing film performed by a special ion source. They varied the ion energy (Eion= 500-1200 eV) and used ions of different mass (mion = ???Ar, m K r , nzxe).The authors found that the cBN content of the films scaled best with ( M Z ~ ~ ~ E ~ ~ ~ ) (i.e. with the total momentum of impinging ions per deposited atom, ptoi/ii).Despite the quite different deposition parameters in the particular experimental series all dependences of the cBN percentage on those parameters could be traced back to practically one and the same curve when ptoi/a as parameter was used. As an ---f

5.4 Spirfter-Deposirion 100

of

cBN Films

429

,

80 c

E 6

--

60

s

2 4 0

41

20 0 0

100

P,,,/a

200 (eV x amu)’

300

400

Figure 6. Dependence of the cBN percentage in BN thin films on p t o t / ufor different ion masses (adapted from [ 171).

example, in Fig. 6 the cBN content dependence on ptot/u for different ion masses is shown. To get practically phase-pure cBN films (i.e. Zc/(Zc Zh)2 0.8),p,,,/a has to exceed a critical value of about 20(1-250 (eV amu)’I2 (amu: atomic mass units). In addition, based on their ion impact approach, Mirkarimi et (11. [ 171 established a quantitative model of defect generation during ion-assisted film growth. They found that the maximum height of the defect production distribution exhibits dependence as cBN formation. Moreover, as the authors the same (w~lol,Eion)”2 emphasize, this is the very same dependence as that found by Windischmann [51] for the formation of intrinsic compressive stress in thin films deposited under ion bombardment. Therefore, we conclude that the ion impact and stress models approaching the subject from different directions using different ‘pictures’, however, have a great deal in common. The ion impact during film growth causes subplantation [52] of particles into the subsurface region of the film thus modifying its physical state: The film gets more defect-rich, densified and compressively stressed. These are all different characteristic of the same physical phenomenon. From that point of view it is not surprising that critical values of the compressive stress or ptot/u can be found which mark the onset of cBN growth. However. those critical values represent ‘projections’ of the cBN formation mechanism onto the p l o , / a and film stress scales. respectively. They d o not deliver information on the physical mechanism of cBN formation itself.

+

5.4 Sputter Deposition of cBN Films In this section cBN film preparation by sputter deposition will be discussed in detail. Such techniques have a high potential for later scaling up of deposition devices and for practical utilization. such as in the field of tool coatings.

430

5 Vapor Phase Deposition of' Cubic Boron Nitride Films

5.4.1 Sputter Deposition with Conducting Targets Most of the sputter deposition experiments known so far were performed using either hexagonal boron nitride (hBN) or boron (B) targets. Due to the low electrical conductivity of hBN and B, the sputter deposition of cBN films requires radio frequency (RF) power at the target electrodes, both in diode [20,21] or magnetron [23] arrangements. With respect to industrial demands like high rate and large area deposition processes, d.c. magnetron sputtering seems to be very promising. However, for the application of this technique, conducting target materials have to be used. Possible candidates are low-resistivity boron rich compounds, like the boride ZrBI2.In the work reported here, as target materials boron carbide (B4C, p < lo2 Rcm at room temperature) was chosen [24,53]. The development of a d.c. magnetron sputter process for the B4C target was subdivided in different phases. First, an R F diode sputter process was optimized for an hBN target [21] and transferred to a B4C target [53], followed by d.c. magnetron sputtering experiments [24,54]. Boron carbide was preferred because it could be expected that the only 'impurity' carbon, at least partially, can be removed by a reaction with nitrogen, such as towards stable gas molecules like C2N2 [55]. For the d.c. magnetron a mixed material target consisting of boron nitride and titanium diboride (BN/TiB2, both with 50 weight-%, p 5 R cm) was also tested [56]. As mentioned above, Hackenberger et al. [41] concluded that stoichiometric or nearly stoichiometric BN films contained the highest amount of the cubic phase. In sputter deposited films both from the hBN target and the B4C target, atomic ratios B : N z 1 in the films were achieved [21,24,54]. The process data for the sputter deposition of cBN films by R F (13.56MHz) diode sputtering with hBN and B4C targets as well as by d.c. magnetron sputtering with B4C targets have been reported in detail elsewhere [21,24,54]. In both configurations the ionization was enhanced by additional magnetic coils positioned around the substrate holder. The substrate electrodes were operated either with a d.c. potential, with pulsed d.c. [57] or with R F power. Some important parameters of the deposition experiments with conducting targets will be summarized as follows. R F diode sputtering: target diameter: 150 mm (hBN and B4C) - target-substrate distance: 70 mm target power: 750-1000 W - substrate potential UBs)-50 to -500 V (d.c. or d.c. self bias/RF) - substrate temperature: up to 350°C - nitrogen flow (N2/Ar + N2): 0-100% - deposition rate: 5 0.5 pm h-' -

~

d.c. magnetron sputtering: target dimensions: 254 mm x 127mm (B4C and BN/TiB2) - target-substrate distance: 80 mm -

5.4 Sputter Deposition of'cBN Filnts -

43 1

target power: 750-1 500 W negative substrate potential ( UHS):up to -550 V (d.c. or d.c. self mode) substrate temperature: up to 500°C nitrogen flow: up to 50% for B4C, up to 100 YOfor BN/TiB2 deposition rate: 5 0.5 pm h

In the d.c. magnetron mode at higher nitrogen flows for the boron carbide target the danger of poisoning and arcing increased. On the other hand, the BN/TiB2 target could be operated with a high stability even at high target power and high nitrogen flows. With the R F as well as the d.c. magnetron sputter technique a films consisting of almost single phase cBN could be deposited. For both techniques the process parameters (target power, substrate voltage, ion current density) necessary to generate cBN (cBN parameters) were determined [24,54]. To achieve B : N ratios near 1 and high ( ~ 6 0 % cBN ) contents, for the hBN target only a few percent nitrogen flow in the sputter gas and for the boron carbide target N2 flows > l o % were needed [24,58]. Cross-section TEM investigations of these sputter deposited films revealed the typical phase evolution well known for cBN films. Electron diffraction patterns indicated that in our films the c-axis ([0002]) of turbostratic BN is parallel to ( 1 1 1 ) direction of cBN and both are perpendicular to the substrate normal [58]. The maximum thicknesses reached so far with the both sputter techniques were about 500 nm on Si and about 200 nm on steel substrates.

5.4.2 Deposition by d.c. Magnetron Sputter with a Hot Boron Target In order to prevent any incorporation of undesired target constituents, we developed a special d.c. magnetron sputtering technique utilizing a pure boron target. In this technique the target is operated at elevated temperature. Since boron has semiconducting properties it becomes increasingly conductive if the temperature is raised. Owing to the special design of the magnetron source the thermal contact between the water-cooled copper support and the clamped target plane was relatively bad. Therefore, by R F sputtering under appropriate conditions, the temperature of the boron target could be increased quickly to about 800°C. This led to an increase in conductivity of six orders of magnitude compared to that at room temperature [59], which was high enough to ignite a stable d.c. magnetron sputter discharge with the boron cathode. During d.c. discharge the steady state temperature of the target is about 600°C [60]. The experiments have been performed using a circular R F magnetron source (diameter l00mm). The substrate holder (diameter 150mm) was mounted at a distance of l00mm from the target surface and was separately R F powered yielding a certain negative d.c. potential of the substrate. This d.c. self-bias will be referred to in the following as the substrate bias voltage, UBS.The substrate temperature and total pressure were held constant during deposition at about 350°C

432

5 Vapor Phase Deposition of Cubic Boron Nitride Films 1.0

1000

-*0.8 n

4

+

3u v

intensity ratio

-

- 800

-

5?

0.6 -

-

1 3 0

-1500

3r

-400

2

0

8

3 0.4-

2

0

-

.M

E

cn

8

U

0.2-

Y Y

- 200

c 3 0.0

*-*'

l

0

-50

~

l

-100

'

-150

l

'

-200

l

-250

~

-300

l

'

-350

l

I

-400

~'

l0 -450

'

~

Substrate bias voltage [V] Figure 7. cBN content together with the deposition rate of BN films deposited at different substrate bias voltages by d.c. magnetron sputtering of a boron target.

and 0.2Pa. The sputter gas was argon plus 10% nitrogen and polished silicon wafers have been used as substrates. For d.c. sputtering of the boron target a power of 200 W was utilized. Varying the substrate bias voltage leads to the evolution of BN layers containing various phase compositions (Fig. 7): From noncubic modifications obtained without external substrate bias and at UBs = -lOOV to films with the highest cBN content at UBs = -350 V. A further increase of the substrate bias to UBs = -400 V causes growth conditions resulting in BN films with a lower cBN/hBN ratio. The growth rate decreases if - UBs is increased mainly due to resputtering. In contrast to that, for R F magnetron sputtering with either an hBN or a boron target performed in the same vacuum chamber maximum cBN content was obtained at UBs = -15OV [25]. However, it has been revealed by Langmuir probe measurements [61] that this reduction of substrate bias voltage for optimum cBN formation is connected with an about twice as high plasma density in front the substrate. Therefore, qualitatively the decreased plasma density has to be compensated by a higher ion energy. A quantitative evaluation of the three magnetron sputter processes with respect to the film-forming particle fluxes (for the analysis procedure see section 5.5.2) revealed that within the error of measurement the d.c. process with boron target had the same criticalptotlabetween 180 and 265 (eV amu)'I2 as the two R F processes [62]. This is in good agreement with the values found for pulsed laser deposition by Mirkarimi et al. [17].

'

5.5 Discrimination between Nucleation and Growth Phase

433

5.5 Discrimination between Nucleation and Growth Phase In nearly all investigations published until now certain experimental conditions were maintained during the whole growth process and then assessed with respect to the cBN percentage in the films grown. This is also true for all experiments discussed in section 5.2 of this chapter. To the knowledge of the authors there are only few exceptions: McKenzie et al. [63] used a helicon activated reactive evaporation process and found that for good cBN formation an R F substrate bias of 100 W was necessary. However, after growing 40 nm of the film the cBN growth could be maintained even when the substrate bias was reduced to 50 W. They concluded that the cBN growth, once nucleated, can be continued at a lower bias level. However, they did not give quantitative data in terms of bias voltage, ion energy or ptot/a.Schutze et al. [57] investigated the reactive sputtering from a B4C target. They monitored the tBN/cBN transition by measuring the substrate current and changed the parameters after cBN nucleation to ‘softer’ conditions. Hahn et al. [64,65] were the first who systematically varied and quantitatively characterized the deposition conditions in order to investigate nucleation and growth of cBN separately. In an R F magnetron sputtering process they started with conditions for optimum cBN formation in order to nucleate the cubic phase. After that, however, they continued the growth under reduced ion impact. It was found that the cBN growth can be maintained even when the ion impact is drastically reduced in comparison with the conditions which are necessary to start cBN formation. Experiments concerning the discrimination between nucleation and growth will be discussed in the following sections in more detail.

5.5.1 Detection of hBN-cBN Transition To the knowledge of the authors there are few experimental experiences concerning the discrimination of nucleation and growth of cBN. Schiitze et al. [57] developed a simple method to control the cBN nucleation in situ by monitoring the substrate current. This is possible, for example, if in the R F sputter device the substrate electrode is operated with d.c. voltage. Fig. 8a shows the principle. Keeping all process parameters constant, an increase in the substrate current was observed if the deposition had been performed in the cBN mode, but it was not observed if the deposition was performed in an hBN mode. This can be explained as follows. The measured substrate current Zsub = Zi ZsE = Zi( 1 r i ) is the sum of the ion current Zistriking the substrate surface and the current of secondary electrons, ZsE, generated at the substrate surface by the incoming ions (ri = secondary electron yield). With the assumption that the 7icBN > YihBN, one can correlate the increase of the substrate current with the change from hBN growth (range A) to cBN growth (B) in the films. The idea of different secondary electron yields was proved by measurements of the ion current component Zi using a Faraday cup arrangement mounted into the substrate holder. While the substrate current

+

+

434

5 Vapor Phase Deposition of Cubic Boron Nitride Films 80

b

A1 20 0 0.o001

Deposition time [a. u.]

0.001

0.01

0.1

1

AUI

Figure 8. (a) Scheme of the observed substrate current increase during BN film deposition where f means the current during hBN growth, (b) cBN content against relative current increase A f /f , the parameters of the deposition experiments were: R F powered boron carbide target (750W), 10% nitrogen in Ar/N2 and UBs = -200 V.

increased (typical values of AZ/Isub z 0.1) the measured ion current Z,stayed nearly constant. Furthermore a clear correlation between AI and the cBN content in the films was found (Fig. 8b). With increasing AZ the cBN content in the films increased, too. An additional hint to higher secondary electron yields of cBN compared to hBN was the brightness difference in scanning electron microscope images taken in the secondary electron mode under identical conditions for both samples [55]. The method described allows control of the cBN deposition process. For example in the cBN nucleation range (left part of interval B in Fig. 8a) the deposition parameters can be modified to ‘softer’ conditions. One possibility is to change the sputter gas composition from high Ar contents, which are related to high stress and cause cBN nucleation [48], to pure nitrogen. Figure 9 shows a SIMS profile of a film grown under changed conditions. The deposition was started under typical ‘cBN conditions’ with 10% N2 in the sputter gas and after nucleation the gas composition was changed to 100% NZ.The diagram reveals that the higher nitrogen

’

60

r_l

50

s40

Y

,g*

+g

u

30

20 10

0 0.00

0.05

0.10

0.15

0.20

Depth [PI Figure 9. SIMS depth profile of boron, nitrogen, and carbon in a cBN film sputter deposited under different conditions on a silicon substrate.

5.5 Discriniination between Nucleation and Growth Phase

435

Table 2. Deposition conditions for separate investigations of nucleation and growth phase Step

R F power (W) N2 (YO)

At (min)

uBS

(v)

Aiil

(nm)

~

1

1000

3

-15

-

-

2

500

3

-15

-400

2 -100

-1

-150

-10

1&30

-200 -175 -150 -125

3

1000

3

4

1000

100

-100

-100-150

-80 -60 -40 +15

For all steps Ts = 350"C, p = 0.2 Pa.

concentration leads to a considerable decrease in the carbon content to about 5 at-% This behavior is to be expected, taking into account element concentration measurements of cBN films deposited with different nitrogen contents in the sputter gas. An interpretation of the observed carbon reduction was given in section 5.4.1. The total incorporated momentum per deposited atom, p t o t/ a , was estimated to be about 20% lower in the pure nitrogen deposition compared to the Ar/N mixture. There are still other methods which are suitable for controlling the phase evaluation in cBN films. Barth et al. [66] measured in situ infrared reflectance spectra during the deposition time and revealed different deposition processes on silicon and steel substrates, respectively. McKenzie et al. [48] concluded from in situ stress measurements that there is a stress threshold above which the cBN growth is beginning. However, in both cases the process conditions were not modified during film growth.

5.5.2 RF Magnetron Sputtering The deposition experiments using RF magnetron sputtering of an hBN target [64,65]were carried out using the deposition chamber described in section 5.4.2 following the experimental design given in Table 2. The whole deposition process consisted of four steps. During step 1 the hBN target was sputter-cleaned in an Ar/N2 mixture. Then, the shutter between target and substrate was opened, and a relatively high substrate bias voltage UBs = -4OOV was applied. Owing to the corresponding high ion energy, strong resputtering prevents any net film growth. Nevertheless, an intense intermixing of boron and nitrogen with the silicon substrate takes place yielding an intermixed region and hence an improved film adhesion. After that, step 3 was carried out with UBs reduced to - 150 V which had been found before to be the optimum value for cBN growth for that process (see section 5.4.2). Under these conditions a thin cBN film (about 5nm thick) was already

436


formed after a deposition time of 1 min. In step 4 the deposition was continued in a pure nitrogen atmosphere. The whole multistep process has been repeated several times and the substrate bias voltage during step 4 was varied in each case within the range from -2OOV and the floating potential of the substrate holder (about +10 V). The total deposition time was chosen so that all films got comparable thickness values of 100-150nm. The film structure has been characterized by infrared absorption, electron diffraction, and cross-section transmission electron microscopy. It was found that by changing the gas atmosphere to pure nitrogen during step 4, but maintaining the substrate bias voltage at -150V, the IR peak intensity ratio Zc/(Zc + Zh) was about 0.84.9, which is the same value as in the case of cBN growth under optimum conditions in an argon-nitrogen mixture (97% Ar/3 YON2). By transmission electron microscopy and electron diffraction it was confirmed that the BN films were composed of small (10-50nm) grains of the cubic phase. The film composition was measured using elastic recoil detection analysis (ERDA) and it was found that the films were (within the error of measurement) stoichiometric BN (B/N = 0.95 f 0.05). Practically 100% cubic film structure was obtained within a wide range of substrate bias voltage during step 4, ranging from -200 V to values as low as -6OV (Fig. 10). The deposition rate was about 200nmh-' for UBs = -200 V and increased gradually with decreasing bias voltage due to reduced resputtering. The increase of the deposition rate between -40 and -60 V (Fig. lo), however, is mainly due to the different mass density of the cubic and hexagonal BN phase. From that it was concluded that for voltage values below - 100 V, resputtering could be neglected and the rate of mass deposition was practically constant.

0.8

-

- 0.8

0.6

-

-0.6

0.4

-

-0.4

0.2

-

n

cf

+

u*

--. W

U"

.-0

E

-

I

3

8

0.2

I

-*-

U

0.0

3. v

: E-

."0

8

0

3

a

0

intensity ratio

- 0.0

0

I

I

I

I

I

I

0

-40

-80

-120

-160

-200

I

Substrate bias voltage [V] Figure 10. Cubic BN phase content together with the deposition rate of BN films deposited at different substrate bias voltages during step 4.

5.5 Discrimination between Nucleation m d Growth Phase

437

a

k' 600

I

800

'

'

I

1000

I

1200

'

I

1400

'

I

1600

"

1800

'

200

Wavenumber [cm'] Figure 11. FT-IR transmission spectra demonstrating the phase inhomogeneity of BN films deposited at U,, = -60 V. The IR beam with a diameter of about 1 mm was moved in 1 mm steps radially from the outer region towards the center of the substrate (spectra a-g). Within the lateral resolution limit of the method a sharp transition from noncubic (outer part) to cubic BN (inner part) is found.

For UBs= -60V and -8OV, only a circular central part of the film (having a diameter of about 3cm) was cubic BN. Outside this central part, the film was purely sp2 bonded (hBN). Considering the lateral resolution of the method, the boundary between the cBN and hBN region was very sharp. This can be seen from Fig. 11: the infrared spectra shown in this figure were measured in distances of 1 mm using a beam diameter of 1 mm. Within a distance of only about 3 mm the absorption features change significantly between the typical pictures of the cubic and hexagonal BN phase. The charge carrier density in the magnetron plasma was measured by Langmuir probe measurements [61]. As an example, in Fig. 12 the distribution of the charge carrier density is presented which had been measured lOmm in front of the substrate. Because the charge carrier density at the plasma sheath is directly related to the ion current towards the substrate (see below), the inhomogeneity visible in Fig. 12 will be reproduced in the ion flux towards the substrate, that is, the ion

438

5 Vapor Phase Deposition qf Cubic Boron Nitride Filnis

1

-60

'

1

-40

'

1

-20

'

'

1

.

20

1

40

'

0

Radial probe position [mm] Figure 12. Radial distribution of the charge carrier density lOmm from the substrate surface (hBN target, R F target power IOOOW, substrate bias voltage UBS= -15OV.p = 0.2Pa, 97% Ar. 3% N?).

flux will be about 15% higher in the center of the substrate holder than at r = 2cm. This is obviously a relevant fact for the BN phase formation and will be discussed later. In order to describe the growth conditions in a quantitative manner the particle fluxes towards the substrate have been estimated. The total flux of incorporated atoms (both boron and nitrogen), Fa, was estimated from film stoichiometry, growth rate, and mass density. The flux of ions cannot be directly measured for several reasons, above all because of the high resistivity of the film which necessitates an RF substrate bias. Therefore an estimation based on the Bohm sheath criterion was made [25]. The Bohm criterion [67] yields an estimation of the ion flux, F,, towards the substrate when ion density, ni, electron temperature, T,, and ion mass, mi, are known. These quantities, however, can be determined by Langmuir probe measurements (ni, T,) or estimated from gas composition (mi). Finally, the kinetic ion energy, Ei, was roughly estimated (neglecting the plasma potential) to be Ei = elUBsl. Using the values of Fa, Fi, and Ei, the total incorporated momentum per deposited atom, ptot/a, for the experimental series was calculated (Fig. 13). Under optimum conditions for cBN growth with the given method used during step 3 (97% Ar/3% N2, Uss = -150V) ptot/a is about 200(eVamu)'/2, which is in good agreement with Mirkarimis results. Changing to ure nitrogen as the sputtering gas, ptot/a reduces to approximately 150 (eV amu)' *. With an additional reduction of UBs it was finally found that the continuation of cBN growth withptot/avalues as low as about 50 (eV amu)'/2 was possible. These results lead to the conclusion that the continuation of cBN growth after a certain initial cBN layer has been formed is

P '

5.5 Discrimination between Nucleation and Growth Phase

439

800 H

h-BN

-

C-BN

600 -

N

n

s

*2

. .

ArM2=9713

400-

2cd --v u

a” 200 -

0

. I

H

. I

.

a 1

I

I

I

governed by an epitaxial mechanism. It has been found that the upper few monolayers of a growing cBN film are sp2 bonded [68,69]. The film forming particles penetrate this layer and reach somewhat deeper regions of the film where they finally become incorporated into the already existing cBN. The certain minimum energy and/or momentum of the ions which was found to be still necessary to maintain the cBN growth is obviously due to the fact that the film-forming particles have to penetrate the sp2 bonded layer. In the meantime, Litvinov and Clarke [70] have confirmed the results of Hahn et al. [64,65]. They deposited boron nitride films in an UHV chamber by ion beam sputtering of a boron target together with a nitrogen ion bombardment of the growing film. The substrate temperature was 1000°C. The authors found that after cBN nucleation at UBs = -96 V the substrate bias voltage can be significantly reduced without interruption of the cBN growth. At UBs = -56V the crystalline quality of the layers is even better that at -96 V. The reduction of the bias voltage below approximately -50 V, however, leads to the formation of predominantly hexagonal phase material. Interestingly, a similar nucleation effect as found for the ion impact seems to exist in the case of the substrate temperature. As has been found by several authors [34,36] a minimum substrate temperature of 150°C is necessary in order to nucleate cubic BN in ion-assisted deposition. However, very recently Feldermann et al. [71] have found for mass selected ion beam deposition that switching off the substrate heating after cBN nucleation and allowing the substrate to cool down to room temperature, does not interrupt cBN growth.

440


5.6 Properties of cBN Films The aim of this section is to give an impression on properties of thin cBN films reported so far. It was found that the film properties are very similar to those known for bulk material cBN.

5.6.1 Mechanical and Tribological Properties Only few reports concerning mechanical and tribological data are available so far. Therefore these properties are a main topic of this chapter. Because of the relatively small thicknesses (<1 pm) of cBN films which can be reached with good adhesion, it is difficult to determine reliable data of hardness, Young’s modulus, and wear. Knoop hardness values of about 50 GPa for cBN and 25 GPa for amorphous BN (aBN) films were reported by Murakawa et al. [72]. From nanometer indentation measurements hardness ( H ) and Young’s modulus ( E ) were derived for different sputter deposited cBN films (thickness some 100 nm) [54,55,73]. The hardness values ranged between 60 and 80GPa which is close to the bulk hardness of cBN (up to 90GPa [l]). On the other hand the Young’s moduli, E, in the range 300600GPa, are lower than the cBN bulk values (800-900GPa) [l]. Mirkarimi et al. [73] explain this discrepancy by substrate effects which decrease with increasing film thickness. Microwear tests performed using atomic force microscopy (AFM) were reported by Miyake et al. [74] and by Schutze [55]. The result was that the wear caused by a diamond tip is much greater for hBN than for cBN. The microwear rates of sputter deposited cBN films were found to be nearly the same as that of hard diamond like carbon films [55]. Friction coefficients, p, of cBN films were measured under different conditions (counterpart, humidity, number of cycles, sliding speeds) at normal loads up to 10N. Data available from the literature are summarized in Table 3. In all cases considered hBN films had clearly higher friction coefficients than cBN. Firsts attempts to apply cBN coatings for wear protection on drills were reported already in 1990 by Murakawa and Watanabe [77]. Owing to the still unsolved problem of depositing films of several micrometers thickness, no essential progress in tool applications of cBN films has been made so far.

5.6.2 Optical Properties Several data on refractive index, n, and extinction coefficient, k, of BN films for the visible, UV, and near IR spectral ranges were published. In all cases the refractive index of cBN films varies between about 2.0 and 2.1, slightly decreasing with increasing wavelength (A) [78,55]. These n values agree well with data known for bulk cBN [79,42]. In the range X > 400 nm the extinction seems to be negligible.

441

5.6 Properties of cBN Films

Table 3. Data on friction behavior of cBN films. Counterpart

Relative humidity

Friction coefficient,

("/.I

P

Steel Diamond

45 45

0.2

Steel

60

0.35

Steel

40

0.1-0.4

Alumina

40

0.0 5-0.5

Remarks

Decreasing with increasing load Extreme decrease under vacuum (<0.01) Strong fluctuations, p decreases under dry conditions Strong fluctuations

Reference

[751 [751 [76] [55] [55]

For films sputter deposited from a boron carbide target, k markedly increased at X < 400nm [55]. This indicates an effect of the incorporated carbon (5-10 at-%, see section 5.5.1) in the films. The cathodoluminescence behavior of hBN and cBN films was investigated by Taylor et al. [80] with respect to possible applications of the large band gap material BN in the deep blue and UV range.

5.6.3 Electrical Properties Undoped BN films exhibit low electrical conductivities. Ronning et al. [81] determined for ion beam deposited films resistivities of 109-10'0Rcm and the Z-U curves were best approximated assuming a Frenkel-Poole conduction mechanism. These resistivities are near to the expected bulk value (up to 10" R cm [82]). Furthermore the dielectric constant was estimated as e = 8-10. The resistivities of films sputter deposited with a boron carbide target were lower (107-109 R cm). Comparing the results of lateral and sandwich measuring arrangements, no essential differences were found [55]. Obviously the conductivity of BN films can be decreased by doping, for example with magnesium to prepare a p-type semiconductor [83].

5.6.4 Other Properties Boron nitride is a wide band gap semiconductor. For BN a negative electron affinity (NEA) can be expected [84]. Materials with this property can be used, for example, in cold cathode emitters. NEA occurs if the vacuum energy level lies below the minimum of the conduction band. Beside the electronic structure of the bulk material the surface conditions of the considered material are quite important [85]. For sputter deposited cBN films no hints of NEA were found, but UPS spectra of the same cBN films, additionally coated with a monolayer of Cs, provided indications of NEA [55,86].

442

5 Vapor Phuse Deposition of Cubic Boron Nitride Films

With the Faraday cup arrangement in the substrate holder used for the experiments on cBN nucleation which were discussed in section 5.5.1 additionally secondary electron yields, yi, both of hBN and cBN films could be determined. Here yi is the yield for ion-induced electron emission. For both phases in the ion energy range between 100 and 600 eV for an argon-nitrogen gas mixture, "ii near to 1 were estimated. According to the observed substrate current increase (section 5.5.1) the values of cBN were calculated to be about 20% higher than those of hBN [86]. The plasma etching behavior of BN films is interesting, for example with respect to deposition process modifications. However, there are still many open questions concerning the effect of additional gas components in the process gases of cBN deposition processes. For ion etching in pure Ar the ratio of the sputter yields of hBN and cBN was revealed to be YhBN/YcBN5 1 [55,64]. On the other hand, Harris et al. [87] observed, for an etching gas consisting of 1% CH4 and 99% HZ, a considerably higher etch rate for the hBN than for the cBN phase. The surface tension of sputter deposited cBN films was determined by contact angle measurements with different well known liquids [88]. The value of 40 mN m-l is nearly the same as that of DLC (diamond like carbon) [89].

5.7 Summary and Outlook There are much experimental data on deposition of cBN films. Both CVD and PVD methods are suitable for preparing films with high contents of the cubic phase material. Essential parameters determining the film growth are ion impact onto the growing film as well as the substrate temperature. The cBN content in the films obviously scales best with the total momentum of impinging ions per deposited atom. Sputter deposition techniques in the R F as well as in the d.c. mode were successfully used to prepare nearly single phase cBN films. At present the insufficient adhesion, caused by high compressive stress and sensitivity to humidity, is the most essential problem to be solved. A promising way to overcome this difficulty could be discrimination between the nucleation and growth phases. Furthermore a better understanding of the correlation between growth conditions, film structure, and stress is needed. If a sufficient adhesion of cBN films on different substrates can be achieved for a thickness range of 1-2pm or more, cBN coatings will have a high potential in several fields of application, especially in the field of tools for ferrous materials.

References 1. J. H. Edgar, in Properties ojGroup III Nitrides, J. H. Edgar (Ed.), INSPEC, the Institution of Electrical Engineers, London, 1994, pp. 3-21, 2. A. Inspektor, C. E. Bauer, and E. J. Oles, Surf. Coat. Technol. 1994,68/69, 359-368.

References

443

3. W. R. L. Lambrecti and B. Segall, in Propertics qf Group III Nifrides, J. H. Edgar (Ed.), INSPEC, the Institution of Electrical Engineers, London, 1994, pp. 129-1 34. 4. R. H. WentorfJr., J . Chem. Phys. 1962,36, 1990-1991. 5. R. H. Wentorf Jr., J . Chem. Phys. 1957, 26, 956. 6. R. H. Wentorf Jr., J . Chem. Phys. 1961, 34, 809-812. 7. V. L. Soloshenko, Diamond Rel. Muter. 1994,4, 1 4 , has delivered strong evidence that cBN is the thermodynamically stable phase at ambient conditions, too. The formation of cBN at those conditions, however, is obviously kinetically inhibited. 8. R. J. Caveney, Muter. Sci. Eng. 1992, B11, 197-205. 9. For a review see: J. J. Pouch and S. A. Alterowitz (Eds), Synthesis and Properties of Boron Nitride, TransTech Publications, Aedermannsdorf, Switzerland, 1990. 10. K. Inagawa, K. Watanabe, K. Saitoh, and Y. Yuchi, Surf: Coat. Technol. 1989, 39, 253-264. 11. M. Murakawa and S. Watanabe, Surf: Coat. Technol. 1990,43/44, 128-136. 12. T. Ikeda, Y. Kawate, and Y. Hirai, J . Vac. Sci. Technol. A , 1990, 8, 3168-3174. 13. T. Wada and N. Yamashida, J . Vuc. Sci. Technol. A , 1992, 10, 515-520. 14. D. J. Kester and R. Messier, J . Appl. Phys. 1992, 72, 504513. 15. N. Tdnabe, T. Hayashi, and M. Iwaki, Diamond Rel. Muter. 1992, 1, 883-890. 16. N. Tanabe, M. Iwaki, Diamond Rel. Muter. 1993, 2, 512-516. 17. P. B. Mirkarimi, K. F. McCarty, D. L. Medlin, W. G. Wolfer, T. A. Friedmann, E. J. Klaus, G. F. Cardinale, and D. G. Howitt, J . Muter. Res. 1994, 9, 2925-2938. 18. H. Hofsass and C. Ronning, in Proc. Int. Conf. on Beam Processing of Advanced Materials, ASM International, Materials Park, OH, 1996, pp. 29-56. 19. G. Krannich, F. Richter, J. Hahn, R. Pintaske, V. B. Fillipov, and Y. Pddero, Diamond Rel. Muter. 1997, 6, 1005-1009. 20. M. Mieno and T. Yoshida, Jup. J . Appl. Phys. 1990,29, L1175-Ll177. 21. K. Bewilogua, J. Buth, H. Hiibsch, and M. Grischke, DiamondRel. Muter. 1993,2, 1206-1210. 22. V. Yu. Kulikowsky. L. R. Shaginydn, V. M. Vereschaka, and N. G. Hatynenko, Diamond Re/. Muter. 1995, 4, 113-1 19. 23. D. G. Rickerby, P. N. Gibson, W. Gissler, and J. Haupt, Thin Solid Films, 1992, 209, 155-160. 24. A. Schiitze, K. Bewilogua, H. Liithje, S. Kouptsidis, and S. Jager, Surr Coat. Technol. 1995,7475, 711-122. 25. J. Hahn, M. Friedrich, R. Pintaske, M. Schaller, N. Kahl, D. R. T. Zahn, and F. Richter, Diamond Rel. Muter. 1996, 5, 1103-1 112. 26. W. Dworschak, K. Jung, and H. Ehrhardt, Diamond Rel. Muter. 1994,3, 337-340. 27. M. Okamoto, Y. Utsumi, and Y. Osaka, Jpn. J . Appl. Phys. 1992,31, 3455-3460. 28. A. Weber, U. Bringmann, R. Nikulski, and C.-P. Klages, S u r t Coat. Technol. 1993,60,493497. 29. R. Freudenstein, S. Reinke, and W. Kulisch, Diamond Rel. Muter. 1997, 6, 584588. 30. F. Zhang, Y. Guo, Z. Song, and G. Chen, Appl. Phys. Lett. 1994,65,971-913. 31. Z. Song, F. Zhang, Y. Guo, and G. Chen, Appl. Phys. Lett. 1994,65, 2669-2671. 32. I. Konyashin, J. Loeffler, J. Bill, and F. Aldinger, Thin Solid Films, 1997, 308-309, 101-106. 33. S. Reinke, M. Kuhr, W. Kulisch, and R. Kassing, Diamond Rel. Muter. 1995, 4, 272-283. 34. H. Hofsass, H. Feldermann, M. Sebastian, and C. Ronning, Phys. Rev. B, 1997, 55, 1323013233. 35. P. B. Mirkarimi, D. L. Medlin, K. F. McCarty, D. C. Dibble, W. M. Clift, J. A. Knapp, and J. C. Barbour, J . Appl. Phys. 1997,82, 1617-1625. 36. P. B. Mirkarimi, D. L. Medlin, K. F. McCarty, and J. C. Barbour, Appl. Phys. Lett. 1995, 66, 28 13-28 15. 37. D. J. Kester, K. S. Ailey, R. F. Davis, and K. L. More, J . Muter. Res. 1993, 8, 1213-1216. 38. D. R. McKenzie, W. D. McFall, S. Reisch, B. W. James, I. S. Falconer, R. W. Boswell, H. Persing, A. J. Perry, and A. Durandet, Surf: Coat. Technol. 1996,78,255-262. 39. A. K. Ballal, L. Salamanca-Riba, C. A. Taylor 11, and G. L. Doll, Thin Solid Films, 1993,224, 46-5 1. 40. D. L. Medlin, T. A. Friedmann, P. B. Mirkarimi, and K. F. McCarty, J . Appl. Phys. 1996,79, 3567-351 1. 41. L. B. Hackenberger, L. J. Pilione, R. Messier, and G. P. Lamaze, J . Vac. Sci. Tecl?nol.A , 1994, 12, 1569-1575.

444


42. P. J. Gielisse, S. S. Mitra, J. N. Plendl, R. D. Griffis, L. C. Mansur, R. Marshall, and E. A. Pascoe, Phys. Rev. 1967,155, 1039-1046. 43. R. Geick, C. H. Perry, and G. Rupprecht, Phys. Rev. 1966, 146, 543-547. 44. S. Jager, K. Bewilogua, and C.-P. Klages, Thin Solid Films, 1994, 245, 50-54. 45. G. F. Cardinale, P. B. Mirkarimi, K. F. McCarty, E. J. Klaus, D. L. Medlin, W. M. Clift, and D. G. Howitt, Thin Solid Films, 1994, 253, 130-135. 46. P. B. Mirkarimi, K. F. McCarty, G. F. Cardinale, D. L. Medlin, D. K. Ottesen, and H. A. Johnsen, J. Vac. Sci. TechnolA, 1996,14,251-255. 47. A. Schiitze, K. Bewilogua, H. Liithje, S. Kouptsidis, and M. Gaertner, Surf. Coat. Technol. 1997, 97, 33-38. 48. D. R. McKenzie, W. D. McFall, W. G. Sainty, C. A. Davis, and R. E. Collins, Diamond Rel. Muter. 1993, 2, 970-976. 49. D. R. McKenzie, D. Muller and B. A. Pailthorpe, Phys. Rev. Lett. 1991, 67, 773-776. 50. J. D. Targove and H. A. McLeod, Appl. Opt. 1988,27,3779-3781. 51. H. Windischmann, J . Appl. Phys. 1987, 62, 1800-1807. 52. J. Robertson, Diamond Rel. Muter. 1996, 5, 519-524. 53. H. Liithje, K. Bewilogua, S. Daaud, M. Johansson, and L. Hultman, Thin Solid Films, 1995, 257,4045. 54. S. Kouptsidis, H. Liithje, K. Bewilogua, A. Schutze, P. Zhang, Diamond Rel. Muter. 1998, 7, 2 6 3 1. 55. A. Schutze, Dr. rer. nat. Thesis, TU Braunschweig/Germany, published in: Fraunhofer-Institut fur Schicht- und Oberflachentechnik - Berichte aus Forschung und Entwicklung Nr. 2, Fraunhofer IRB Verlag, 1997. 56. S. Kouptsidis, Dr. rer. nat. Thesis, TU Braunschweig/Germany/Fraunhofer-Institutfur Schicht- und Oberflachentechnik-Berichte aus Forschung und Entwicklung Nr. 4, Fraunhofer IRB Verlag, 1998. 57. A. Schiitze,K. Bewilogua, H. Liithje, and S. Kouptsidis, DiamondRel. Muter. 1996,5, 1130-1135. 58. K. Bewilogua, A. Schiitze, S. Kouptsidis, and H. Liithje, Proc. Third Int. Conf. on Applications of Diamond Films and RelatedMaterials, A. Feldman, Y. Tzeng, W. A. Yarbrough, M. Yoshikawa, and M. Murakawa (Eds), NIST Special Publications, Washington DC, 1995,831-838. 59. L. Gmelin, Handbuch der anorganischen Chemie, Suppl. Vol. 2, Springer Verlag, Berlin, Heidelberg, New York, 1981, p. 82. 60. V. LinD, Diploma Thesis, Institut fur Physik, Technische Universitat Chemnitz, Germany, 1998. 61. R. Pintaske, Th. Welzel, N. Kahl, M. Schaller, J. Hahn, and F. Richter, Surf. Coat. Technol. 1997, 90, 275-284. 62. F. Richter, R. Pintaske, J. Hahn, and Th. Welzel, in Hard Coatings Based on Borides, Carbides and Nitrides: Synthesis Characterization and Applications, A. Kumar, Y.-W. Chang, and R. W. J. Chia (Eds), The Minerals, Metals and Materials Society, Warrendale, PA, 1998, pp. 153-168. 63. D. R. McKenzie, W. D. McFall, H. Smith, B. Higgins, R. W. Boswell, A. Durandet, B. W. James, and 1. S. Falconer, Nucl. Instr. Meth. B, 1995, 106, 90-95. 64. J. Hahn, Dr. rer. nat. Thesis, Technische Universitat Chemnitz, Germany, 1997. 65. J. Hahn, F. Richter, R. Pintaske, M. Roder, E. Schneider, and Th. Welzel, Surf. Coat. Technol. 1997,92, 129-134. 66. K.-L. Barth, W. Sigle, D. Stockle, J. Ulmer, and A. Lunk, Thin Solid Films, 1997, 301, 65-70. 67. B. Chapman, Glow Discharge Processes, Wiley, New York, 1980, Chapter 3. 68. G. Sent, D. Bouchier, S. Ilias, M. A. Djouadi, V. Stambuli, P. Moller, G. Hug, and S. Reinke, in Proc. 4th Intern. Symp. on Diamond Materials, K. V. Ravi, J. L. Davidson, R. H. Hauge, J. P. Dismukes, K. E. Spear, and B. V. Spitsyn (Eds), Reno, NV, 1995, The Electrochemical Society, Pennington, NJ, 1995, pp. 377-382. 69. T. A. Friedmann, P. B. Mirkarimi, D. L. Medlin, K. F. McCarty, E. J. Klaus, D. Boehme, H. A. Johnsen, M. J. Mills, and D. K. Ottesen, J . Appl. Phys. 1994,76, 3088-3101. 70. D. Litvinov and R. Clarke. Aool. Phvs. Lett. 1997. 71. 1969-1971. 71. H. Feldermann, R. Merk, 'H: Hofsass, C. Ronning, T. Zheleva, Appl. Phys. Lett., 1999, 74, 1552-1554. 72. M. Murakawa, S. Watanabe, and S. Miyake, Diamond Films Technol. 1991, 1, 55-66.

References

445

73. P. B. Mirkarimi, D. L. Medlin, K. F. McCarty, D. C. Dibble, W. M. Clift, J. A. Knapp, and J. C. Barbour, J. Appl. Phys. 1997,82, 1617-1625. 74. S. Miyake, S. Watanabe, M. Murakawa, R. Kaneko, and T. Miyamoto, Thin SolidFilms, 1992, 212, 262-266. 75. S. Watanabe, S. Miyake, and M. Murakawa, Surf: Coat. Technol. 1995, 76-77, 600-603. 76. Z. Yu, K. Inagawa and Z. Jin, Surf. Coat. Technol. 1994,70, 147-150. 77. M. Murakawa, S. Watanabe, Surf. Coat. Technol. 1990,43144, 145-153. 78. W. D. McFall, D. R. McKenzie, and R. P. Netterfield, Surf. Coat. Technol. 1996, 81, 72-78. 79. G. L. Doll, in Properties of Group III Nitrides, J. H. Edgar (Ed.), INSPEC, the Institution of Electrical Engineers, London, 1994, pp. 169-1 74. 80. C. A. Taylor 11, S. W. Brown, V. Subramaniam, S. Kidner, S. C. Rand, and R. Clarke, Appl. Phys. Lett. 1994,65, 1251-1253. 81. C. Ronning, E. Dreher, H. Feldermann, M. Gross, M. Sebastian, and H. Hofsass, Diamond Re/. Muter. 1997,6, 1129-1 134. 82. V. V. Lopatin, in Properties of Group III Nitrides, J. H. Edgar (Ed.), INSPEC, the Institution of Electrical Engineers, London, 1994, pp. 89-100. 83. M. Lu, A. Bousetta, A. Bensaoula, K. Waters, and J. A. Schultz, Appl. Phys. Lett. 1996, 68, 622-624. 84. M. J. Powers, M. C. Benjamin, L. M. Porter, R. J. Nemanich, R. F. Davis, J. J. Cuomo, G. L. Doll and S. J. Harris, Appl. Phys. Lett. 1995, 67, 3912-3914. 85. R. J. Nemanich, P. K. Baumann, M. C. Benjamin, S. W. King, J. van der Weide, and R. F. Davis, Diamond Rel. Muter. 1996, 5, 790-796. 86. K. Bewilogua, A. Schiitze, H. Walter, S Kouptsidis, Proceedings of the Second Symposium on III-V Nitride Materials and Processes (Eds. C. R. Abernathy, W. D. Brown, D. N. Buckley, J. P. Dismukes, M. Kamp, T. D. Moustakas, S. J. Pearton, F. Ren), The Electrochemical Society, Inc., Pennington, NJ, 1998, Proceedings volume 97-34, 134141. 87. S. J. Harris, A. Weiner, G. L. Doll, and W.-J. Meng, J . Muter. Res. 1997, 12, 412415. 88. K. Trojan, M. Grischke, and H. Dimigen, Phys. Stat. Sol. ( a ) , 1994, 145, 575-585. 89. M. Grischke, K. Bewilogua, K. Trojan, and H. Dimigen, Surf. Coat. Technol. 1995,7475,739745.


6 Polymer to Ceramic Transformation: Processing of Ceramic Bodies and Thin Films G. D. Soraru and P. Colombo

6.1 Introduction Polymer pyrolysis to form advanced ceramics allows the production of highly covalent refractory components (fibers, films, membranes, foams, joints, monolithic bodies, ceramic matrix composites) that are difficult to fabricate via the traditional powder processing route [14]. Yajima was the first to demonstrate the feasibility of producing high-strength SiC-based fibers from pyrolysis of polycarbosilane [5]. In this process, a thermoplastic pre-ceramic polymer is first shaped into the desired form, cross-linked into a pre-ceramic network and finally converted into a ceramic material by a pyrolysis process in a controlled atmosphere (Fig. 1). A common feature of the polymer route is the formation of intermediates called amorphous covalent ceramics (ACC) [6]. These are formed after removal of the organic components and before crystallization that occurs at higher temperatures. Many experimental techniques have been applied to characterize the polymer-toceramic transformations. They range from differential thermal analysis/thermogravimetrical analysis/mass spectometry (DTA/TGA/MS) [7,8] to spectroscopies which are especially helpful for the low temperature range when the ceramic residue is still amorphous such as: magic angle spinning nuclear magnetic resonance (MAS NMR) [9,10], XPS [11, 121, and X-ray absorption near edge structureextended X-ray absorption fine structure (XANES-EXAFS) [I 3,141, to X-ray diffraction (XRD) and transmission electron microscopy (TEM) to characterize the crystallization process [15,16]. Among the various systems already synthesized, S i c obtained from polycarbosilane is certainly the most widely studied and the proposed structure of the amorphous silicon carbide phase obtained after pyrolysis in Ar at 800°C is shown in Fig. 2 [17]. One of the most attractive and challenging features of this process is the possibility of designing the composition, the microstructure, and the properties of the final ceramic material by controlling each processing step. The processing steps are: ~

-

the polymer composition and/or its architecture, the cross-linking step, the pyrolytic conversion process, and the final high temperature annealing to induce densification and/or crystallization.

Indeed, many Si-containing ceramic systems have been already prepared from polymers of different composition such as: polysiloxanes (-R, R2Si-O-), to give SiOC [ 18,191, polysilanes (-R1 R2-Si-), [20,2 I], and polycarbosilanes (-R2Si-CH2-),

6.1 Introduction

f

Pb'O'ecular recursor

447

I

1

Inert or Reactive Atmosphere

Fibers

A m o r p hou s Covalent C e ra m ic

Films Foams

Joints lnfiltralion

for CMC

Bulks

N a n o - c' r y s t a I Iin e Ceramic

Figure 1. Flow chart for the preparation of ceramic components viu polymer pyrolysis.

[22-241 to give S i c and polysilazane (-R2Si-NH-), to give SiCN [25] and SiBCN ceramics [26]. Moreover, the various polymeric precursors can be easily modified by chemical reactions to obtain new systems. A good example is the preparation of polytitanocarbosilane (PTC) from polycarbosilane and titanium alkoxide [27]. This precursor transforms into SiC/TiC during pyrolysis. The fibers produced from this modified precursors seem to exhibit better mechanical properties than those of S i c [28]. The same reaction used to produce PTC gives the opportunity of modifying the polycarbosilane with different elements just by changing the nature of the metal alkoxide and accordingly a large variety of new metal organic polymers, precursors for S-M-C4 (M = Al, Zr, Ta) systems have been obtained [29-311. Another approach to control the composition and the microstructure of the final product is to use ceramic or metallic powders as filling agents or reagents in the starting polymers [32,33]. The selectivity toward defined ceramic phases can also be controlled by varying the polymer architecture. In this case the pyrolysis in inert atmosphere of-[MeHSiNH],- leads to SiCN ceramics at temperatures >8OO0C whereas pyrolysis of isostructural -[H2SiNMeIxgives Si3N4and carbon [34] . The precursor molecular structure can also lead to dif-

448

6 Polymer to Ceramic Transformation: Processing of Ceramic Bodies and Thin Films

,

. I

-

\

I

Si,

’S .i .

I

-Si-C,

/

CH+~,

“i:

si-Cqc/ ’ \

I

Figure 2. Proposed structure of the amorphous silicon carbide phase obtained from polycarbosilane.

ferent microstructures of the cross-linked pre-ceramic networks that in turn have a significant effect on both the pyrolysis chemistry and the composition and microstructure of the ceramic products. Indeed, a study performed on polycarbosilane/siloxane precursors having similar chemical composition but different molecular structures pointed out the formation of different preceramic microstructures that survive up to high temperatures (1000°C) leading to SiOC glasses of different microporosity and thermal stability [35]. The curing step can play a vital role in defining the structure and mechanical properties of polymer derived ceramics. Indeed, the tensile strength of Sic-based Nicalon fibers drops quickly at high temperatures due to the oxygen introduced during the thermal cross-linking stage. It is believed that the oxygen gives rise, upon pyrolysis, to a SiOC phase that decomposes above 1300°C with evolution of SiO and CO [36]. This problem has been overcome by using a radiation (electron beam or y-ray irradiation) curing process in helium gas which reduces the oxygen content down to 0.4weight % and improves the fiber resistance up to 1750°C [37]. A different approach to controlling the composition and structure of the ceramic products is through the control of the transformation process itself. The heating rate and the maximum pyrolysis temperature can have a strong influence on the ceramic yield and the ceramic composition [38]. In Fig. 3 thermogravimetry curves obtained on a commercial polycarbosilane at different heating rates are reported. It can be noticed that the ceramic yield is strongly related to the pyrolysis rate, and it increases by using low heating rates. The pyrolysis atmosphere can be varied from inert (Ar, He, N2) to reactive (NH,, 0,)and accordingly silicon nitride and carbonitride-based materials, such as Si-N, Si-C-N, Si-0-N, and SiM-0-N (M = Al, Zr, Ti) have been produced from various polysiloxanes, polycarbosilane and polysilanes precursors [3942]. More recently synthetic silicates have been fabricated by thermooxidation of polysiloxanes filled with different oxide or metallic powders [43]. As far as the final heat treatment step is concerned, it is well known that the maximum pyrolysis temperature governs the densification of the amorphous ceramic phase and its crystallization process [44]. The densification of polycarbosilane-derived amorphous silicon carbide was studied [ 17,451 in the temperature

6.1 Introduction

449

0

-1 0

-$ u1

-

-20

Em

.-

0)

-30

200

600

400

800

1000

1200

Temperature ("C)

Figure 3. Thermogravimetry curves recorded on a commercial polycarbosilane (Dow Corning X96348) in Ar flow (100cm3 min-') at different heating rates.

range 900-1000°C (Fig. 4). An activation energy of 82 kcal mole-' was obtained from the initial slope of the densification curves (Fig. 5). The observed increase of density during the isothermal firing treatments has been primarily ascribed to reduction of the free volume of the amorphous phase. Two main processes may account for this effect: the progress of the condensation reactions between residual CH groups in the structure with the elimination of H2 and CH4 and the formation of new Si-C-Si bridges with a consequent increase of the cross-linking of the network; and a rearrangement of the open amorphous covalent structure toward a more compact configuration with no change in chemical composition. For both of these

0

2

4

6

8

10

12

Time (h)

Figure 4. Evolution of the density, p, of polycarbosilane-derived amorphous silicon carbide with the 900°C; 1000°C; 950°C. firing temperature.

+

450

6 Polymer to Ceramic Trunsftirrnution: Processing of' Ceramic Bodies and Thin Films

3.9

4.0

4.1

4.2

4.3

IO~IRT

Figure 5. Arrhenius plot for the densification rate of the polycarbosilane-derived amorphous silicon carbide.

mechanisms, a key step should be the cleavage of chemical bonds: C-H (99 kcalmole-') and Si-C (76 kcal mole-'). The obtained value of activation energy suggests that, at least in the initial stages, the densification process occurs via the cleavage of chemical bonds present in the material. Up to now, polymer pyrolysis has been investigated especially to develop ceramic fibers [46,47] and ceramic matrix for ceramic matrix composites [48-501. More recently studies have been undertaken to exploit this method to develop ceramic thin films [51-531, foams [54], joints [55], and bulk materials [56]. Moreover, nonconventional heating systems such as laser [57],microwave heating [53], or even athermal conversion processes such as ion bombardment are just now starting to be applied to the polymer route and the preliminary results are very promising [58-601. In this chapter we focus on the polymer processing of bulk ceramics obtained by pyrolysis of partially cross-linked preceramic bodies and of thin ceramic films (obtained either by traditional pyrolysis or by the innovative ion irradiation process).

6.2 Processing of Monolithic Components The use of polymer pyrolysis for the fabrication of bulk ceramics has been hindered mainly by the significant volume shrinkage (250%) accompanying the polymer-toceramic conversion, often resulting in the formation of cracks which destroys the structural integrity of the ceramic component. In order to overcome these difficulties, many hybrid processes have been studied for the fabrication of bulk ceramic components via polymer pyrolysis. Greil and coworkers proposed to reduce the shrinkage by the use of active fillers that can react during pyrolysis leading to a volume expansion [33], pre-ceramic polymers

6.2 Processing of Monolithic Components

45 1

have been used as active binders in the traditional powders processing [61-631; and polymer-derived amorphous ceramic powders have been sintered to give dense, monolithic ceramics [64-671. However, in all these hybrid processes, the potential advantages of the polymer route are lost, at least to some extent. Recently, Riedel and coworkers [56,68-721, demonstrated the possibility of producing a dense, crack-free, monolithic ceramic in the Si-C-N system by direct pyrolysis of a polymeric green compact. In this method, a pre-ceramic polymer is first converted into a pre-ceramic network with a controlled cross-linking process. The powders obtained are then sieved and pressed to get a polymeric pre-ceramic component, that can be pyrolyzed without cracking, into the final dense ceramic body. These innovative ceramic components display negligible creep rates up to very high temperatures [73,74]. The key point of this process is the partial crosslinking of the polymer precursor that leads to an infusible network preventing the component from melting and bloating during the first stages of the pyrolysis process. Moreover, the open porosity of the green body guarantees the efficient removal of the gaseous products during pyrolysis preventing the development of cracks. Thus, the significant volume shrinkage usually associated with the polymer-toceramic transformation no longer prevents the fabrication of bulk, crack-free ceramic components. However, in order to optimize the microstructure of the ceramic product, the influence of many polymer properties and processing parameters still must be assessed. As an example, the cross-linking degree and the weight loss during pyrolysis of the polymeric precursor, the powder size and powder size distribution, the amount of porosity and the pore size distribution of the green compact, the pyrolysis rate and atmosphere are just some of the parameters that could strongly influence the microstructure and the related physical, chemical, and mechanical properties of the final ceramic product. Here we report on the fabrication and characterization of dense ceramic materials in the SiAlOC system from pre-ceramic green bodies obtained by pressing partially crosslinked polyaluminocarbosilane (PAlC) powders with different aluminum loads. The synthesis of PAlC is carried out reacting a commercial polycarbosilane (PCS) with an aluminum alkoxide, following a published procedure [27]. PAlC has been already investigated in our laboratory both with the aim of getting insights into the chemical reaction between the polycarbosilane chains and the aluminum alkoxide and to clarify the structure of the resulting amorphous Sic-based ceramic [29,40,47]. The objective of the present work is to investigate the role of some polymer properties (cross-linking degree, weight loss during pyrolysis, etc.) on the microstructure and mechanical properties of the ceramic materials obtained by pyrolysis in inert atmosphere at 1000°C. Indeed, the forementioned parameters can be easily modified by varying the amount of aluminum alkoxide reacted with the polycarbosilane. Accordingly, several PAlC compositions have been prepared with a nominal Al/ Si ratio between 0 and 0.25. The corresponding ceramic materials have been characterized mainly by measuring their density, porosity and mechanical properties through four point bending tests both at room temperature and at high temperature up to 14OO0C,in air. The results will be discussed with the aim of disclosing the key parameters that allow the processing of high strength and high modulus ceramic components.

452


Finally, we will show that this innovative processing route can be extended to the production of bulk ceramic components by pyrolysis in reactive atmosphere, such as for example, production of SiAlON ceramics by nitridation of PAlC in ammonia flow [75].

6.3 Preparation and Characterization of SiAlOC Ceramic Bodies by Pyrolysis in Inert Atmosphere 6.3.1 Experimental Procedure Polyaluminocarbosilanes (PAlCs), with a nominal AljSi ratio of 0.05, 0.1, 0.15, and 0.25 were prepared reacting a commercial polycarbosilane (Dow Corning X9-6348) with an aluminum sec-butoxide modified with ethylacetoacetate, A1(Obus)*etac (Alfa product, 89349). Details of the synthesis are reported elsewhere [47]. A PAlC composition with Al/Si= 1 was also prepared and used only for the MAS NMR study. PAlCs were subsequently heat-treated for 1 h at 300°C in Ar flow in a silica tube furnace in order to ensure that all the starting pre-ceramic polymers were treated at the same maximum temperature. In some cases, more than one batch of PAlC with the same AljSi ratio was prepared; however, every batch was studied individually. For comparison purposes, partially cross-linked, pure polycarbosilane was also prepared. In this case, however, the PCS was cross-linked at 400°C for 1h since polymers processed at lower temperatures showed melting and bloating during pyrolysis. The pre-ceramic polymers were ball-milled in an agate mortar and the powders with a size lower than 63 pm were used to prepare green compacts by cold uniaxial pressing into a steel mold 60 mm in length and 8 mm wide. Green samples (approximately 60 x 8 x 5 mm3) were accordingly obtained with a pressure of 143 MPa applied for 30min (see Fig. 6). After complete release of the molding pressure and extraction of the sample from the mold, the length of the sample was measured and compared with the length of the mold in order to evaluate the relative size increase, usually designated as ‘springback’ and defined as: S=----IC - Id x 100, ld

where Id is the mold length and I, is the corresponding compact length. This parameter is a function of many factors, such as the powder properties, the compacting pressure and time, and the die rigidity. However, in our experiments the forming process was performed always with the same conditions. Thus, the springback is directly related to the ability of the pre-ceramic polymers to be permanently deformed under load and gives a measurement of the cross-linking degree of the pre-ceramic network. Indeed, highly cross-linked polymers behave as purely elastic materials and display a higher value of springback compared to polymers with a lower cross-linking degree that show a permanent deformation under pressure, ultimately leading to a lower S value. The green components were also characterized

6.4 Results

453

Figure 6. Green and corresponding ceramic components obtained from PAlC with AljSi = 0.05 ratio.

measuring their open porosity (P)by mercury infiltration (Carlo Erba equipment) and their bulk density, Pb. The structure of the partially cross-linked polymers was investigated by recording 29Si CP MAS NMR spectra on a MSL 400 spectrometer operating at 79.6MHz. Details of the experimental set up are reported elsewhere [47]. Monolithic, dense and crack-free SiAlOC ceramic components were obtained (see Fig. 6) after a pyrolysis process (1 "C min-' , Ar flow, 100 cm3min-') at 1000°C for 1 h. The weight and the length of the samples were recorded before and after the heat treatment in order to measure the associated weight loss and the linear shrinkage. The microstructure of the ceramic samples was investigated measuring the open porosity, P,and the bulk density, &,. Modulus of rupture, MOR. was obtained on polished samples by four point bend tests (4 x 20 mm) with 1 mm min-' deformation rate using Instron equipment. The elastic modulus, E , was measured at room temperature, by the acoustic resonance method using a self-assembled apparatus and, at high temperature, from the stress-strain curve obtained in the four point bend tests.

6.4 Results 6.4.1 Characterization of the Pre-ceramic Precursors The cross-linking mechanism of polycarbosilane with various metallic alkoxides has been already published [45]. The Si-H bonds of the PCS react with the M-OR groups (M=Ti, Al, Zr, Ta) of the metallic alkoxide to give Si-0 bonds that probably form siloxane bridges between the polycarbosilane chains. The 29Si CP MAS NMR spectra recorded on the starting PCS and on two PAlCs with an AljSi ratio of 0.25 and 1 are presented in Fig. 7. The spectra show two peaks at 0p.p.m. (Sic4 sites) and at - 17.6p.p.m. (SiC3H sites), characteristic of the polycarbosilane backbone and a third peak, present only in the Al-containing compositions, at 10 p.p.m. that is assigned to SiC30 sites. By increasing the A1 content, a decrease of

454


AVSi = 0.25

50

0 -50 Chemical Shift (ppm)

-100

Figure 7. 29Si CP MAS NMR spectra of the partially cross-linked pre-ceramic polymers.

the component at -17.6p.p.m. can be noticed and an increase of the peak at 10p.p.m., in agreement with the proposed mechanism. Thus, these results suggest an increase in the number of the siloxane bridges, that is, of the cross-linking degree, with the amount of aluminum alkoxide.

6.4.2 Characterization of the Pre-ceramic Components Bulk density of all the pre-ceramic bodies was ~1 g ~ m - ~Figure . 8 shows the evolution of open porosity and elastic recovery as a function of the Al/Si ratio. For all the PAlC precursors P falls in a narrow range between 9% and 12%. A higher value ( P = 17%) has been observed for the green bodies obtained from pure PCS. These low values of open porosity are due to the plastic deformation during pressing of the partially cross-linked precursors. Moreover, mercury infiltration experiments always resulted in a bimodal distribution of the open porosity, in agreement with the published literature [56]. Accordingly, the finer pore fraction below 10 nm can be assigned to flat pores between two particles whereas the coarser one, with a maximum in the range 8&600nm, is due to pores at triple point junctions. The size of the pores at the triple point junctions is ~ 6 0 0 n mfor the pure PCS precursor and increases with the A1 load, that is with the cross-linking degree, from a minimum value of ~ 8 nm, 0 observed for the AljSi = 0.05 composition to 300 nm for the Al/Si = 0.25 samples. The elastic recovery (Fig. 9) shows an evident relation with the aluminum content: the lowest S values being observed for the samples with AljSi = 0.05 ( S M 0.4%) and

6.4 Results

455

18 A

8 W

14

s

.d c,

v1

g 10

0 pc

6 t.

1

0

..

#

.

.

.

I

0.1

I

. . I

.

.

0.2

I

.

0.3

Al/Si (Atomic ratio) Figure 8. Evolution of the porosity of the pre-ceramic samples as a function of the AljSi ratio.

the highest ones for the samples with Al/Si=0.25 (1% 5 S 5 1.5%). The S value measured for the green compact derived from pure PCS is 1.3%, close to the values measured on PAlCs with the maximum aluminum load. These results suggest that, by increasing the amount of aluminum alkoxide used for the preparation of the modified polycarbosilanes, the cross-linking degree advances, in agreement with the NMR results, and correspondingly increases the elastic recovery of the pre-ceramic components. The high S value measured for the PCS green bodies indicates that the thermal cross-linking process used in this case leads to a polymer with a high crosslinking decree which exhibits a highly elastic behavior during pressing.

6.4.3 Characterization of the Ceramic Components All the pyrolyzed samples were XRD amorphous. Weight loss ( ~ 2 weight-%) 8 and linear shinkage ( ~ 2 9 % )associated to the pyrolysis treatment were found to be

€

i

1 2

3 4

0

0.1

0.2

Al/Si (Atomic ratio)

T

0.3

Figure 9. Evolution of the elastic recovery of the pre-ceramic samples as a function of the AljSi ratio.

456

.cd


200

5

W

d 100

0

E

0.1

0

0.2

0.3

Al/Si (Atomic ratio) Figure 10. Evolution of the flexural strength, modulus of rupture (MOR), of the ceramic samples pyrolized at 1000"C, 1h, with the AljSi ratio.

independent of the A1 content. Values of density, porosity, flexural strength, and elastic modulus measured on the various ceramic components are reported in Table 1. The open porosity of the PAlCs-derived ceramic samples shows a weak dependence from the amount of aluminum and falls in the range ~ 8 - 1 3 % .It increases up to ~ 2 0 %for the PCS-derived component. Such low values of open porosity have been already reported in the literature for similar polymer-derived ceramics [56] and indicate that this processing route is suitable for the fabrication of dense (up to 92% of theoretical density) covalent ceramics at low temperature. The porosity values measured for the ceramic products are comparable with the starting pre-ceramic samples. Indeed, results of the mercury infiltration experiments indicate that the pyrolysis process leads to the removal of the finer pore fraction (below 10 nm) leaving almost unchanged the amount and the size of the coarser pores. The evolution of the MOR and E with the Al/Si ratio is reported in Figs 10 and 11. Surprisingly, contrary to the behavior of all the other studied parameters, these Table 1. Results of the microstructural and room temperature mechanical characterization performed on the ceramic samples pyrolyzed at 1000°C for 1 h. AljSi (Atom ratio)

Batch number

Bulk density Pb (gcm-3)

Porosity P (Yo)

PCS 0.05 0.05 0.10 0.10 0.10 0.10 0.15 0.25 0.25

I I

1.82 f 0.01 2.15 f 0.03 2.19 f 0.01 2.14 f 0.04 2.24 f 0.01 2.12 z t 0.06 2.13 f 0.05 2.15 f 0.05 2.11 f 0.05 2.12 f 0.05

I1 I

I1 111

IV I I I1

Flexural strength (MOR) (MPa)

Elastic modulus (GPa)

19.0 0.3 7.8 f 0.5 8.0 f 0.5 10.3 f 0.5 7.5 f 0.4 12.4 f 0.5

40 f 1 181 f 15 214 f 10 190 f 38 177 f 16 -

-

-

10.6 f 1.3 11.9 f 1.5

101 f 7 63 f 2

55 f 3 117 f 1 136 f 1 123 f 9 177 f 7 112 f 5 122 f 3 78 f 8 56f 1 90 f 1

*

~

-

457

6.4 Results

160

2 120

9

batch 2

A

batch 3

k

u

W

W 80

40

0.1

0

0.2

0.3

Al/Si (Atomic ratio) Figure 11. Evolution of the elastic modulus, E , of the ceramic samples pyrolized at lOOO"C, 1 h, with the Al/Si ratio.

properties show a strong dependence from the composition of the starting polymers with up to 100% of variation. The highest values of MOR (up to ~ 2 5 MPa) 0 and of E (up to ~ 1 4 0 G P ahave ) been recorded for the samples with AljSi ratio of 0.05, then the increase of the A1 load leads to a decrease of both mechanical parameters. PCS-derived samples display very low values of MOR and E similar to the samples with the highest AljSi ratio.

6.4.4 Mechanical Characterization at High Temperature High temperature mechanical characterization was performed on the PAlC compositions with AljSi = 0.05 and 0.1. The elastic modulus has been measured in air at various temperatures between 800 and 1400°C by four point bend tests (40 x 20 mm) with a 0.2 mm min-' deformation rate. E has been calculated, through the standard equation valid for rectangular bars, by measuring the displacement with a LVDT. All the ceramic samples obtained by pyrolysis at 1000°C have been pre-annealed at 1400°C for 1 h in argon atmosphere, before the high temperature tests. This treatment lead to the crystallization of microcrystalline PSiC with a minor amount of clSiC. Aluminum atoms are present both as a solid solution of A120C in clSiC and in the residual amorphous phase. The results of the elastic modulus measurements are reported in Table 2. The evolution of the elastic modulus with the temperature for the Al/Si = 0.05 composition is reported in Fig. 12. It can be noticed that the elastic modulus is constant up to Table 2. Results of the elastic modulus, E , measurements performed at high temperature, in the range 800-1400°C. ~~~~~~

~

T ("C)

20

AljSi = 0.05 AljSi =0.1

129 f 6 118 f 8 126 f 5 107 f 1

800

900

1000

1100

-

118f7 104f 7

115f6 104f6 9 4 f 4 96 f 12 85 f 13 72 f 7

108 f 8

1200

1300

1400 81 f 2 -

458

6 Polymer to Ceramic Transformation: Processing of Ceramic Bodies and Thin Films 200

1

'

'

'

'

1

'

'

'

'

'

"

1

'

1

'

'

'

'

'

'

'

'

'

'

~

'

'

-

1 5 0n

I

i

Q

0

T

T +

-

* r

100-

m

W

-

-

50-

0

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1100°C. Above this temperature the slight decrease of E is probably related to the viscous deformation of the amorphous silicon oxycarbide phase that shows a glass transition temperature of ~ ~ 1 3 0 0 [76]. °C

6.5 Discussion In this work, crack-free, dense (up to 92% theoretical density), amorphous SiAlOC ceramic components with good mechanical properties have been obtained at low temperature ( T = 1000°C) by direct pyrolysis of pre-ceramic green bodies. Elastic modulus and flexural strength show a clear dependence from the A1 content and the maximum values were recorded for the composition low in Al. These behaviors cannot be ascribed to a different weight loss or linear shrinkage during pyrolysis due to the constancy of these parameter with the polymer composition. Thus, in order to explain this experimental data we tried to correlate the elastic modulus with the amount of porosity measured by mercury infiltration. The dependence of E on the porosity shows a high level of scatter (see Fig. 13). As an example, ceramic components with a porosity level of about 10% display elastic modulus values from z60GPa to =120GPa, with a 100% variation. These findings suggest that other microstructure parameters, apart from the amount of open porosity must play a crucial role in governing the Young modulus of the studied polymer-derived ceramics. Recently some studies have been published that take into account the role of the contact area on which mechanical stresses are really acting between the particles [77]. According to these models, samples with the same P but with different E values must have a different amount of contact area between the ceramic grains. In our samples, the interface between the grains (derived by the decomposition of the initial pre-ceramic particles) in the ceramic component can be high only if the

6.5 Discussion

459

t

5

10

15

20

25

Porosity, P (YO) Figure 13. Elastic modulus of the samples pyrolyzed at 1000°Cfor 1 h plotted against the porosity of the ceramic components.

contact area between the pre-ceramic powders in the pressed green body is also high. In fact, during the pyrolysis, the development of the ceramic bond between the individual particles takes place with the formation of primary chemical bonds via condensation reactions such as:

+ H-Si+ H3C-Si-

-Si-CH3 ESi-CH,

+S - C H 2 - S i s +=Si-CH2-Si-

+ H2 + SH4.

These reactions can occur only if the reacting groups are close enough for the formation of -Si-CH2-SiE bridges, that is, if they belong to the interface between two pre-ceramic grains. The extension of the contact surface between the partially cross-linked polymeric grains should be high for the compositions that can be easily plastically deformed under pressure, with a low cross-linking degree. In order to test this hypothesis, the elastic modulus has been plotted against the elastic recovery, that is a parameter directly related to the cross-linking degree of the preceramic polymers (see Fig. 14). In this case a very good relation between E and S can be observed suggesting that, actually, the cross-linking degree is a vital parameter for tailoring the mechanical properties of polymer derived ceramics. In particular, the cross-linking degree must be high enough to prevent bloating but also low

0

0.5

1

1.5

2

Elastic Recovery, S (YO) Figure 14. Elastic modulus of the samples pyrolyzed at 1000°C for 1 h plotted against the elastic recovery of the green components.

460


enough to allow viscous deformation during pressing with the formation of extended contact area between the polymeric powders, which in turns gives rise to ceramic components with high mechanical properties.

6.6 Preparation and Characterization of SiAlON Ceramics by Pyrolysis in Reactive Atmosphere 6.6.1 Experimental The green pre-ceramic samples were nitrided in NH3 flow (100 cm3min-') in a silica tube furnace at different temperatures up to 1000°C for 1 h with a heating rate of 1"Cmin-I. The ceramics obtained at 1000°C were subsequently fired in N2 flow in a graphite furnace at various temperatures up to 1500°C. At the end of the pyrolysis treatments the weight loss and the specific surface area (BET method) was recorded. The skeleton density was measured with a helium picnometer while the porosity was obtained by mercury infiltration tests. In order to follow the sintering process, dilatometric tests and differential thermal analysis experiments were performed in flowing nitrogen at 5°C min-' up to 1550°C on samples already pyrolized under ammonia at 1000°C. Structural characterization of the final PSiAlON ceramic was accomplished by X-ray diffraction analysis on powdered samples. Microstructural investigation of the ceramic phase was performed by SEM on polished or fracture surfaces. Mechanical characterization, namely hardness (Hv), elastic moduls (E), and fracture toughness (&) was carried out by means of Vickers and Knoop indentation tests performed at 40-60 and 10 N, respectively.

6.7 Results and Discussion The nitridation under ammonia of the yellowish PAlC compacts leads to the formation at 1000°C of white, crack-free amorphous SiAlON pellets with a linear shrinkage of 20%. After an additional pyrolysis step under flowing nitrogen at 1500°C the amorphous samples convert into microcrystalline PSiAlON monolithic components with a further linear shrinkage of 20%. A weight loss of 25% has been observed from room temperature up to 400°C (Fig. 15). It is due to the decomposition of the organic component of the modified aluminum alkoxide, since PCS chains are stable up to this temperature. This process leads to a corresponding increase of the specific surface area up to 350 m2 g-' (see Fig. 16), without any appreciable variation in the apparent density. In the temperature range 400-600°C the nitriding of PCS occurs with a further weight loss of up to ~535%and an additional increase of surface area to 500 m2 g-I. In this case, however, the formation of a three-dimensional silicon nitride network

6.7 Results and Discussion

46 1

0 - I \

I

/I

-.'

-10

2. (P

z

-20

0

400

800

T

1200

-

0

fn fn

Adensity

/.

*

-30

-E

-40 1600

("C)

Figure 15. Processing of SiAlON ceramics by nitridation of PAlC precursors: evolution of the density and weight loss with the firing temperature.

leads to an increase in density to 1.4 gcmP3,indicating that a sintering mechanism is active in this temperature range. Above 600°C the weight of the samples does not change up to the maximum temperature reached in this study (1 500°C). However, its apparent density remains constant (1.4 g cmp3)up to 12OO0C,then increases dramatically above 13OO0C,approaching a value of 2.5 g cmp3at 1500°C.The evolution of the surface area with the pyrolysis temperature is similar to the density, that is, it remains constant (500m2g-') up to 1200°C and then decreases rapidly above 1300°C, being close to zero at 1500°C. The dilatometric and DTA tests performed 600

NH3 <-

I -> N2

. ol

N

E

Y

m

E!

m

0 .c 0

al

n a

0

400

800

1200

1600

T ("C)

Figure 16. Processing of SiAlON ceramics by nitridation of PAlC precursors: evolution of the specific surface area with the firing temperature.

462

6 Polymer to Ceramic Transformation: Processing of Ceramic Bodies and Thin Films 0.25

4

. 0

-I

-I

-0.05 LIL"

-0.15

1 I

1

T ("C) Figure 17. Processing of SiAlON ceramics by nitridation of PAlC precursors: differential thermal analysis (DTA) and dilatometry of a monolithic amorphous sample obtained after pyrolysis in ammonia at 1000°C.

on samples previously nitridated at 1000°Care reported in Fig. 17. The dilatometric curve clearly shows that the shrinkage of the amorphous AlON compacts starts around 1300°C and it is complete before 1500°C. DTA analysis indicates that, in the same temperature interval, a strong exothermic effect, related to the crystallization of PSiAlON phase, is active. Indeed, the crystallization of the sample has been with a verified by XRD analysis and the phase was identified as PSi2.5A10,500,5N3.5, mean particle size of 40 nm. Thus the observed sintering process has to be related with the amorphous to crystalline transformation that occurs via the formation of a liquid phase. Monolithic PSiAlON ceramics obtained at 1500°C are 85% dense, according to the porosity measurements. The results of the mechanical characterization performed on samples pyrolyzed at 1300°C and 1500°C are reported in Table 3. This work his clearly shown, for the first time, that the direct pyrolysis of partially cross-linked pre-ceramic bodies can be applied for the processing of monolithic ceramics even when a reactive pyrolysis process is used. Table 3. Measured values of hardness (H,), elastic modulus ( E ) and fracture toughness (KIc) for amorphous (1300°C) and microcrystalline (1500°C) DSiAlON pellets. Temperature ("C)

H, (GPa)

E (GPa)

KIc (MPa m112)

1300 1500

2.0 f 0.3 11.0 f 1.4

28 !i 12 198 f 28

1.2 f 0.5 4.7 f 0.8

6.9 Experimental

463

6.8 Processing of Thin Ceramic Films Processing is always a key step in the development of a material or a device. Parameters such as the heat treatment temperature, the heating atmosphere, and the dwelling time greatly influence the microstructure and composition of the material. Moreover, thin films are more likely to be affected by the processing conditions than a monolithic body, so more care should be taken to avoid the presence of undesired impurities, such as oxygen. This work reports the development of a polymeric/sol-gel route for the deposition of silicon carbide and silicon oxycarbide thin films for applications such as heat-, corrosion-, and wear-resistant coatings, coatings on fibers for controlling the interaction with the matrix in ceramic matrix composites, or films in electronic and optoelectronic devices. This method, in which the pre-ceramic films are converted to a ceramic coating either by a conventional high temperature annealing or by ion irradiation, is alternative to conventional methods such as chemical or physical vapor deposition (CVD, PVD), molecular beam epitaxy, sputtering, plasma spray, or laser ablation, which are not always practical or cost efficient. Using a conventional high temperature annealing (500-12OO0C), important characteristics like the degree of crystallization, the crystallite dimension, the presence of hydrogen and the CjSi ratio can all be varied in a fairly large range of values. However, low-temperature melting substrates cannot be employed, and if the precursors contain carbon, it is difficult to obtain pure ceramics due to the retention of a noticeable amount of hydrogen (because of the strength of the single C-H bonds) and of free carbon. The use of ion implantation allows for the conversion to ceramics through the breaking or three-dimensional cross-linking of bonds in the precursor films, accompanied by the radiolysis of hydrogen atoms and alkyl groups. Several advantages might result from the use of this room temperature processing technique with respect to conventional annealing. In particular ceramic coatings can be obtained on non-refractory substrates or as components of integrated circuits for electronics without degrading metallic overlayers. Other interesting potentialities are the introduction of dopants and the patterning of the coating by using masks or by focusing the ion beam.

6.9 Experimental S i c and SiOC films were produced by spinning a solution containing the precursor on (100) silicon waters or silica slides. Detailed experimental and analysis procedure is reported elsewhere [59,60]. The pre-ceramic polymers used were: for S i c films, polycarbosilane (PCS, Dow Corning), or allylhydridopolycarbosilane (HPCS, Starfire Systems, Watervliet, NY). For SiOC films, methylhydroxylsiloxane or phenylmethylhydroxylsiloxane (SR350 and SR355, respectively, General Electric Silicone Products, Waterford, NY). In addition, SiOC films were produced also from different alkoxides: methyltriethoxysilane or phenyltriethoxysilane (MTES

464


and PTES, respectively, Aldrich, Milan), or from a mixture of methyldiethoxysilane and triethoxysilane (DH and TH, respectively, ABCR, Karlsruhe), with a DH/TH molar ratio of 1/2. The precursors were chosen in order to achieve different CjSi ratios in the final ceramic product. The conversion from precursor to ceramic was obtained either by conventional heating in a tube furnace (heating at temperatures in the range 200-1200°C in different inert atmospheres: regular grade Ar (99.99%), ultra high purity Ar (99.999%), vacuum (1 x 10-6mbar)), or by irradiating the films with different doses of ions (C, Au, He). Part of the irradiated specimens were subsequently heat treated at 1000°C for 1 h in vacuum. After conversion, the average thickness for the films was in the range 200-500 nm. The composition and microstructure of the films was investigated using several techniques, such as Fourier transformed infrared spectroscopy (FTIR), secondary ion mass spectroscopy (SIMS), Rutherford back-scattering spectrometry (RBS), elastic recoil detection analysis (ERDA), nuclear reactions analysis (NRA), glancing angle X-ray diffraction (GA-XRD), Raman spectroscopy, transmission electron microscopy (TEM), scanning electron microscopy (SEM), and atomic force microscopy (AFM). Film thickness was measured by ellipsometry or using a tactile profilometer. Some optical and mechanical properties of the films were also evaluated, using a UVvis spectrophotometer and a nano-indenter [51, 52, 59, 60, 78-80]. In this paper, only the results concerning SiOC films obtained from SR350 and S i c films obtained from PCS will be reported.

6.10 Results and Discussion 6.10.1 Conventional Conversion Process: Annealing in Controlled Atmosphere When considering the thermal conversion, it was found that the processing gas, the heating temperature, and the dwelling time have a strong influence on the structure and the composition of the produced ceramic thin films. In Fig. 18 are reported the FTIR spectra collected for SR350 and PCS films heat treated in argon and vacuum. The peaks at about 1090cmpl (with the shoulder at 1190cm-') and at 460 cm-' are assigned to the Si-0 stretching and rocking modes, respectively, while the peak at about 790cm-' corresponds to the S i x stretching vibration. Pyrolysis in regular grade Ar results in coatings whose structure resembles that of silica, and compositional analysis confirmed that little carbon was retained in the SR350 and PCS films. It has to be emphasized that, in the case of PCS, oxygen was not present in the as deposited coatings, so it can be concluded that oxygen was introduced during the heat treatment. When UHP Ar gas was used, the compositional analysis showed that carbon was retained in both the SR350 and PCS coatings (with a very high CjSi ratio in PCS films), and that oxygen contamination was almost completely suppressed in the latter. X-ray diffraction analysis shows that

6.10 Results and Discussion I

"

I

,

'

1

'

'

465

-

SR350

1600

1200

.

.

l

.

.

.

800

Wavenumbers / cm-'

400 1600

1200

800

400

Wavenumbers / cm-l

Figure 18. Fourier transform transmission spectra for SR350 and PCS films heat treated in: (a) regular grade Ar, 1000°C; (b) ultra high purity Ar, 1000°C;(c) vacuum, 1000°C; (d) vacuum, 1200°C.

the structure of the PCS films remained amorphous, possibly due to the excess carbon inhibiting crystallization or growth of Sic nanocrystals. Performing the pyrolysis of PCS films at 1000°Cin vacuum, a structure characteristic of nanocrystalline silicon carbide was obtained, but the compositional data show that some oxygen contamination was present. Increasing the processing temperature to 1200°C results in the increase of crystallization of PSiC (see later). The SR350 films heat treated at 1200°C in vacuum exhibits a composition and structure that can best be described as an oxycarbide glass, that is an amorphous silicate material in which some of the Si-0 bonds are replaced by Si-C bonds. In fact, the FTIR spectrum shows the persistence of Si-C bonds after heating, even if this finding is less evident when heating at temperatures 11000°C. It is thus clear that the processing atmosphere strongly affects the resulting composition and microstructure of both SiOC and S i c coatings, and that vacuum is the only environment which allows the synthesis of both silicon carbide and silicon oxycarbide thin films [51]. In Fig. 19 is shown the influence of the processing temperature on the compositional and microstructural development of PCS films, heat treated in vacuum for 1h. The peak at about 2100cm-' (see Fig. 19a) is attributable to the Si-H stretching; the peaks at 1400cmp' are due to the CH3 asymmetric in Si-CH3 whereas the peak at 1250cm-' is due to the Si-CH, symmetric deformation. The peaks at 1355 and 1020cm-' are respectively assigned to the CH2 deformation and wagging of Si-CH,-Si, and the peaks at 820 and 790cm-' are respectively assigned to the SiCH3 rocking and S i x stretching. The spectra show that, for heat treatments above 600°C, the absorption bands that are associated with organic species are eliminated, indicating the conversion to a ceramic material of the pre-ceramic polymer.

466


2000

1600

1200

Wavenumbers / cm"

800

10

20

30

40

50

60

70

80

29 / Degrees

Figure 19. (a) Fourier transform transmission spectra and (b) glancing angle X-ray diffraction (XRD) patterns for PCS films heat treated for 1 h in vacuum at different temperatures.

The narrowing and intensification of the Si-C absorption band located at 800 cm-' indicates an increase of the ordering of the structure, which corresponds to a crystallization of the films. Indeed, glancing angle X-ray diffraction analysis (see Fig. 19b), reveals that above 1000°Cthe PCS films develop into nanocrystalline ceramic coatings. The compositional analysis showed that, with increasing processing temperature, a strong decrease in the hydrogen concentration occurs [8 11. In the intermediate pyrolysis temperature, just after the polymer-to-ceramic conversion, there is a rather large oxygen contamination, which is 19.7 (at 800°C) and 15.6at-% (at 9OOOC). This is believed to be due to the presence of reactive sites in the films at this stage of the transformation, which are generated by the cross-linking of the PCS polymer and the creation of Si radicals. With increasing processing temperature, the oxygen contamination strongly decreases, and films heated at 1100°C have a C:Si ratio of about 1, while the ones heated at 1200°C appear to be silicon-rich. Raman analysis allowed to follow the evolution of the free carbon (turbostratic C) contained in the Sic films: at elevated heating temperatures (> lOOO"C), the free C phase is drastically reduced, and disappears completely in films pyrolyzed at 1200"C, presumably being eliminated as CO and/or CH4. The effect of heat treatment at 1200°Cvs. 1000°Cis primarily to enhance the evolution of hydrogen and the more effective crystallization of PSiC at the higher temperature. Combining compositional and microstructural data we can thus infer that the presence of free carbon has a hindering effect on the development of PSiC crystallites in the silicon carbide films [52]. With increasing processing temperature the PCS-derived films approach the composition and microstructure of stoichiometric polycrystalline 0 silicon carbide. Concurrent with the compositional and microstructural development, a shrinkage of the films occurs, which is of about 75%


1600

1200

800

Wavenumbers / cm-'

400

467

26 28 30 32 34 36 38 40

28 / Degrees

Figure 20. (a) Fourier transform transmission spectra and (b) glancing angle X-ray diffraction (XRD) patterns for PCS films heat treated for different times in vacuum at 1000°C.

at 1200°C yielding a Sic coating 200 nm thick. No major defects (cracks, pits) are visible in the films. The effect of heating time on PCS films, pyrolyzed at 1000°C in vacuum, is reported in Fig. 20. We can see that the crystallinity of the S i c coatings clearly increases with increasing the isothermal treatment time. Accordingly, the compositional analysis revealed that for heat treatments shorter than 6 h (at 1000°C), the films are carbon-rich, whereas for longer dwelling times they become silicon-rich, with a significant reduction in hydrogen and a lesser reduction in carbon and oxygen. At the same time, Raman spectroscopy shows a strong reduction of the free turbostratic carbon phase [52]. In Fig. 21 is reported the variation of the average dimension of the PSiC crystallites in S i c films from polycarbosilane, as a function of temperature and dwelling time (at 1000°C). The grain size seems to be enhanced with respect to PCS-derived Nicalon fibers and PCS pyrolysis residues that have been treated at similar temperatures. This is most probably due to differences in oxygen contamination among the different PCS-derived materials, and/or to their different processing histories (especially the kind of annealing atmosphere). Again, concurrent with the transformations happening in the material during the isothermal treatment, the film thickness decreases as a function of the dwelling time.

6.10.2 Nonconventional Conversion Process: Ion Irradiation In the experimental conditions adopted in these studies, the irradiating ions penetrate through the entire thickness of the film, and stop into the substrate.


800

900

1000 1100 1200

Pyrolysis Temperature /"C

Dwelling Time at 1000°C / h

Figure 21. Average PSiC crystallite dimension for PCS-derlved thin films as a function of: (a) processing temperature and (b) dwelling time (heating in vacuum at lO0OT).

During this travel, the ions transfer energy to the atoms, which constitute the coating material, by nuclear collisions and electronic excitations. As a result, a change in composition, structure and density occurs, which depends on the product of the fluence (which is proportional to the number of irradiating ions) by the linear density of energy deposited by each impinging ion. In particular, besides merely promoting densification, ion irradiation has a much higher efficiency than UV or electrons for breaking all bonds and creating three-dimensional cross-links in polymers (organic or inorganic), and especially for radiolyzing hydrogen and alkyl groups. One has to bear in mind that the number of irradiated ions amounts to just a few monolayers, and thus the properties of the substrates are not strongly affected by this processing method (Si is amorphized within the ion range). In Fig. 22 are reported the FTIR spectra for SR350 and PCS films, submitted to irradiation with different doses of 3 MeV Au ions, or annealed, or irradiated and then annealed (1 h at 1000°C in vacuum). We can easily see that with increasing ion dose the bands relative to the organic moieties (see previous discussion) show a progressive smearing out, and are replaced by broad bands of conjugated bonds 0 Similar behavior and threshold when the deposited energy reaches values of ~ 4 eV. of collapse into a three-dimensional network are generally found for organic polymers, and in our case indicate the actual occurrence of the polymer-to-ceramic transformation. With respect to spectra of annealed films, the bands of conjugated bonds are significantly broader and shifted toward lower wavenumbers (of =2Ocm-' for the main Si-C and Si-0-Si modes) in spectra of films converted by irradiation. An oxygen sub-stoichiometry in silica films or a compressive state of stress can induce such a shift. In our case, this is more probably due to a higher degree of bond distortion because of the combination of free radicals at random,


469

Irradiated + Annealed Annealed 5x10''

= 4 5x10''

As deposited

As deposited

2000

2000

1600

1200

Wavenumbers I cm-'

800

1600

1200

800

Wavenumbers I crn"

Figure 22. Fourier transform spectra of SR350 and PCS films irradiated with different doses of 3 MeV Au ions, or annealed (1 h lOOO"C), or irradiated with 5 x lOI5 Au' at cm-2 and then annealed (1 h lO0OT).

and to some atomic disorder in Au-irradiated films. Associated with the structural transformation in the specimens, there is a compositional change, which depends on the dose and the type of irradiating ion (see Table 4). At high doses, the irradiation process allows to obtain ceramic films whose composition is very similar to that of the starting precursor, apart from a large decrease in hydrogen content. No major contamination (oxygen) is introduced, and carbon is retained in the structure, whereas conventional annealing is shown to strongly alter the Si:C ratio (and to introduce some oxygen in the PCS films) [58,59,60]. Under irradiation, the films also are subjected, without cracking, to a compaction that can be as high as 65%. After irradiation, all the films are X-ray amorphous, while PSiC nanocrystallites develop in PCS films, even after they have been irradiated, during the annealing treatment. SiOC and Sic films obtained by ion irradiation contain some carbon clusters (free C). Ion irradiation, besides provoking structural and compositional chances in the films, affects the hybridization state of the carbon clusters, as can be seen in Fig. 23, which reports some Raman spectra collected on irradiated, annealed, irradiated and then annealed SR350 and PCS samples. The presence of a single Raman peak in the case of both SR350 and PCS irradiated films, indicates that the carbon clusters possess a high degree of sp3 hybridization, which is typical of diamond-like films. The hybridization state of the carbon clusters becomes a little more graphite-like in films irradiated and then annealed (according to the degree of splitting in two components of the Raman carbon peak), and is absolutely graphitic (sp2 hybridization, turbostratic carbon) in precipitates formed during annealing of as-deposited films [58].

470


Table 4. Composition, density, and hardness of the studied films. Film

Composition (Si,C,O,H,)

Treatment

Density Hardness (g ~ m - ~ ) (GPa)

J

X

PCS

As-deposited He 1 x 10l6 c 1 x 10I6 Au 1 x 10" Annealed Irradiated Au

1.90 f 0.1 1.90 1.85 1.75 0.80 + annealed 1.65

0.08 f 0.02 0.92 0.06 0.06 0.20 0.05

4.50 f 0.2 1.10 1.oo 2.14 0.60 2.15 0.62 2.01 0.25 2.45 0.04 2.41

0.28 17.1 29.0 19.4 19.0 34.6

SR350

As-deposited He 5 x lot6 c 1 x 10'6 Au 1 x 10'' Annealed Irradiated Au

1.00 4~0.05 1.oo 1.oo 1.oo 0.60 + annealed 0.98

1.55 f 0.05 1.40 ISO 1S O 1.28 1.45

2.9 f 0.1 0.29 0.37 0.55 0.52 0.02

0.23 21.0 18.5 15.5 7.0 18.5

1.25 2.05 2.16 2.06 2.08 2.40

The carbon clusters afford some interesting optical properties to the SiOC and Sic films [79], but are also responsible for the high hardness measured on irradiated films (see Table 4). This is at least two times larger than that of conventionally annealed specimens, mainly because of the previously mentioned diamond-like nature of the C precipitates in the amorphous SiOC or Sic ceramic matrix of irradiated specimens. The effect depends on the type of irradiating ion, but also on the nature of the side groups of the polymeric chain (CH3 vs. C6H5) [59,60]. The annealing of irradiated films does not seem to affect much their hardness,

1"'""""""l

I

I

1000

1200

1400

*

'

'

I

'

"

I

'

1600

Wavenumbers/ cm-l

18001000

.

1200

'

'

PCS

Irradiated

Irradiated

.

,

I

.

1400

.

.

I

.

1600

.

.

1800

WavenumbersI cm-'

Figure 23. Raman spectra of SR350 and PCS films irradiated with 5 x 10153 MeVAu' atcm-', or annealed (1 h 1000°C), or irradiated with 5 x 10"Au+ atcm-2 and then annealed (1 h lOOWC).


47 1

Figure 24. Profilometer trace and AFM image of a -250 pm-wide strip realized by ion irradiation (1 x loL5500 keV C+ at cm-’) in a PCS film.

despite the partial change in hybridization state of the C clusters, probably because this effect is compensated by the further hydrogen removal and compaction. Another interesting consequence of the ion irradiation of polycarbosilanes and polysiloxanes is that the resulting films are thermochemically very stable towards high temperature annealing, as shown by compositional analysis (see Table 4). Indeed, carbon atoms having already lost their hydrogen neighbors can no longer evolve (as CO or C,H,) and leave Si atoms with dangling bonds able to react with O2 or H 2 0 of the atmosphere, as instead happens during annealing of nonirradiated films [79]. Thus, the postirradiation annealing treatment results mainly in the nearly complete elimination of hydrogen from the ceramic films. Finally, ion irradiation promotes a good adherence of the coatings to the substrate, due to some interfacial mixing. The capability of patterning the films using ion irradiation, which can be quite important when fabricating actual devices, is illustrated in Figs 24 and 25. Figure 24 shows the profilometer trace and the AFM image of a strip realized in a PCS film by ion irradiation (1 x lOI5 500 keVC+ at cm-*) using a 250 pm mask, while Fig. 25

Figure 25. Profilometer trace and AFM image of a -250 pm-wide strip realized by ion irradiation 500 keV C+ at cm-2) in a PCS film, after dissolving the non-irradiated part of the film. (1 x

472


shows the same specimen after the nonirradiated part of the film was removed by dissolving it in monochlorobenzene. It is evident the ~ 4 8 % compaction of the film due to the irradiation, as well as the rather good lateral resolution of the strip profile. The high edges present in the strip after dissolving the non-irradiated part of the film (Fig. 25) are probably due to an irradiation-induced cross-linking of the PCS polymer at the interface between the irradiated and non-irradiated areas, which makes it insoluble, but is not accompanied by any compaction.

6.11 Conclusions Polymer pyrolysis appears to be a very promising processing route to advanced covalent ceramics. This process allows the formation of new amorphous covalent ceramic phases in the general system Si-M-C-N-0 (with M = B, Ti, Al, Zr . . .) which show exceptional oxidation and creep resistance at high temperature up to 1500°C. Another advantage of this process is its high flexibility. Indeed, the composition, structure (amorphous or crystalline), microstructure, and properties of the ceramic material can be controlled and adjusted by varying many different parameters such as: the composition and architecture of the pre-ceramic polymer, the amount and nature of the filler, the cross-linking step, and the pyrolysis parameters (atmosphere, heating rate, final temperature). Also, nonconventional heating systems such as laser or microwave heating or even athermal conversion processes like ion bombardment can be efficiently used. Polymer pyrolysis is also very flexible regarding the type of component that can be obtained: fibers, films, membranes, foams, ceramic/ceramic joints, ceramic matrix composites, and monolithic bodies can all be fabricated. The present study has been focused on the formation of monolithic component and thin ceramic films. 0 Dense Sic-based monolithic components with MOR values up to ~ 2 5 MPa and E values up to ~ 1 4 0 G P ahave been fabricated from partially cross-linked polyaluminocarbosilanes, PAlC, by pyrolysis at 1000°Cin a controlled atmosphere. The cross-linking degree of the pre-ceramic polymers was controlled by varying the amount of aluminum alkoxide reacted with the PCS. Among the many investigated parameters (weight loss and linear shrinkage during pyrolysis, porosity of the preceramic and ceramic components) only the cross-linking degree was shown to really play a key role in the development of ceramic materials with elevated mechanical properties. This result has been related to the ability of the pre-ceramic powders to be permanently deformed under pressure leading to an extended contact surface both between the green particles and between the ceramic grains. Defect-free Sic and SiOC ceramic thin films were obtained by heating in inert atmosphere or by ion irradiation. The composition and microstructure of the coatings can be varied by changing processing conditions (heating atmosphere, maximum heating temperature, dwelling time, type and dose of irradiating ions). The segregation of C clusters during ion irradiation affords interesting optical (luminescence) and mechanical (high hardness) properties to the films.

References

473

Acknowledgments Gian Domenico Soraru would like to thank Prof. J. D. Mackenzie, who introduced him to the field of polymer-derived ceramics, Dr. Florence Babonneau for the MAS NMR study performed in the framework of a very effective, 10-year long collaboration, and the contribution of his students and colleagues: Dr. Alberto Ravagni, Ing. Massimo Mercadini, Ing. Fabiano Vulcan, Ing. Alberto Chiari, Ing. Marcello Beber, Ing. Paolo Bosetti, Dr. Vincenzo Sglavo, Dr. Antonella Glisenti. MURST is also acknowledged for the 40% financial support. Paolo Colombo would like to thank Prof. C. G. Pantano for the constant support and encouragement in pursuing research in the field of polymer-derived ceramics. He also gratefully acknowledges the fundamental contribution of Dr. J. C. Pivin, who performed the ion irradiation experiments as well as many characterizations on the specimens.

References 1. J. Lipowitz, Cerum. Bull., 1991, 70, 1888-1894. 2. Y. D. Blum, R. M. Platz, and E. J. Crawford, J . Am. Cerurn. Soc. 1990,73, 170-172. 3. R. Riedel, Advanced ceramics from inorganic polymers, in Materials Science and Technology, Vol. 176, Processing of Ceramics, R. J. Brook (Ed.), VCH, Weinheim, 1996, pp. 1-50. 4. J. Bill and F. Aldinger, Precursor-derived covalent ceramics, in Precursor-Derived Ceramics, J. Bill, F. Wakai, and F. Aldinger (Eds.), Wiley-VCH, Weinheim, 1999, pp. 33-51. 5. S. Yajima, J. Hayashi, and M. Omori, Chem. Lett. 1975, 931-934. 6. G. D. Sorard, F. Babonneau, and J. D. Mackenzie, Structural concepts on new amorphous covalent solids, J . Non-Cryst. Solids, 1988, 106, 25G261. 7. V. Belot, R. J. P. Corriu, D. Leclercq, P. H. Mutin, and A. Vioux, Thermal redistribution reactions in crosslinked polysiloxanes, J . Polym. Sci.: Part A: Polym. Chem. 1992,30,613-623. 8. R. Campostrini, G. D’Andrea, G. Carturan, R. Ceccato, and G. D. Sorard, Pyrolysis study of methyl-substituted Si-H containing gels as precursors for oxycarbide glasses, by combined thermogravimetric, gas chromatographic and mass spectrometric analysis, J . Muter. Chem. 1996,6, 585-594. 9. F. Babonneau, J. Livage, G. D. Sorard, G. Carturan, and J. D. Mackenzie, SiC/TiC ceramics via polymer route: a structural investigation. New J . Chem. 1990, 14, 539-544. 10. F. Babonneau, G. D. Sorard, and J. D. Mackenzie, 29Si MAS-NMR investigation of the conversion process of a polytitanocarbosilane into SiC/TiC ceramics, J. Muter. Sci. 1990, 25, 36643670, 11. G. D. Sorard, A. Glisenti, G. Granozzi, F. Babonneau, and J. D. Mackenzie, The pyrolysis process of a polytitanocarbosilane into SiC/TiC ceramics: an XPS study, J . Muter. Res. 1990, 5, 1958-1962. 12. G. Granozzi, A. Glisenti, and G. D. Sorard, XPS characterization of mixed carbides obtained from polymer precursors, Muter. Rex Soc. Symp. Proc., 1990, 180, 81 1-814. 13. C. Laffon, A. M. Flank, P. Lagarde, M. Laridjani, R. Hagege, P. Orly, J. Cotteret, J. Dixmier, J. L. Miquel, H. Hommel, and A. P. Legrand, Study of Nicalon-based ceramic fibres and powders by EXAFS spectrometry, X-ray diffractometry and some additional methods, J . Muter. Sci. 1989, 24, 1503-1512. 14. F. Babonneau and G. D. Sorard, Spectroscopic characterization of the nitridation process of polymeric precursors to Si-M-N-0 systems (M = Ti, Zr, Al), Muter. Res. SOC.Symp. Proc. 1993, 287, 239-244.

414


15. B. A. Bender, R. W. Rice, and J. R. Spann, Transmission electron microscopic characterization of ceramics formed by pyrolysis of organometallic polymer precursors, J . Am. Ceram. SOC. 1987,70, C58-C60. 16. H.-J. Kleebe, D. Suttor, H. Miiller, and G. Ziegler, Decomposition-crystallization of polymerderived Si-C-N ceramics, J. Am. Ceram. Soc. 1998,81, 2971-2977. 17. G. D. Sorarh, F. Babonneau, and J. D. Mackenzie, Structural evolution from polycarbosilane to S i c ceramics, J. Muter. Sci. 1990, 25, 38863893. 18. G. M. Renlund, S. Prochazka, and R. H. Doremus, Silicon oxycarbide glasses: Part I. Preparation and chemistry, J . Muter. Res. 1991, 6, 2716-2722. 19. G. M. Renlund, S. Prochazka, and R. H. Doremus, Silicon oxycarbide glasses: Part 11. Structure and properties, J. Muter. Res. 1991, 6, 2723-34. 20. R. West in Ultrastructure Processing of Ceramics, Glasses and Composites, L. L. Hench and D. R. Ulrich (Eds), Wiley, New York, 1984, pp. 235-244. 21. R. H. Baney, J. H. Gaul, and T. K. Hilty, Methylchloropolysilanes and Derivatives Prepared from the Redistribution of Methylchlorodisilanes, Organometallics, 1983, 2, 859-863. 22. S. Yajima, J. Hayashi, M. Omori, and K. Okamura, Development of Silicon Carbide Fiber with High Tensile Strength, Nature, 1976, 261, 683. 23. C. L. Schilling, J. P. Wesson, and T. C. Williams, Polycarbosilane Precursors for Silicon Carbide, Am. Ceram. Soc. Bull. 1983, 62, 912-915. 24. L. V. Interrante, C. W. Whitmarsh, W. Sherwood, H.-J. Wu, R. Lewis, and G. E. Maciel, High yield polycarbosilane precursors to stoichiometric Sic. Synthesis, pyrolysis and application, Mater. Res. SOC. Symp. Proc. 1994,346, 593-603. 25. D. Seiferth and G. Wiseman, High-Yield Synthesis of Si,N,/SiC Ceramic Materials by Pyrolysis of a Novel Polyorganosilazane, J. Am. Ceram. SOC.1984, 67, C-132. 26. R. Riedel, A. Kienzle, W. DreBler, L. Ruwisch, J. Bill, and F. Aldinger, A silicon carbonitride ceramic stable to 2,0Oo”C, Nature, 1996, 382, 796-798. 27. S. Yajima., T. Iwai, T. Yamamura, K. Okamura, and Y. Hasegawa, Synthesis of Polytitanocarbosilane and Its Conversion into Inorganic Compounds, J . Muter. Sci. 1981, 16, 1349-1355. 28. K. Okamura, M. Sato, T. Matsuzawa, and Y. Hasegawa, Formation of S i c Fibers and Related Ceramic Fibers from Polycarbosilane, in Ultrastructure Processing of Advanced Ceramics, J. D. Mackenzie and D. R. Ulrich (Eds), Wiley-Interscience, New York, 1988, pp. 501-518. 29. F. Babonneau, G. D. Soraru, K. J. Thorne, and J. D. Mackenzie, Chemical characterization of Si-AI-C-0 precursor and it pyrolysis”, J . Am. Ceram. Soc. 1991, 74, 1725-1728. 30. F. Babonneau and G. D. Sorarh, Synthesis and Characterization of Si-Zr-C-0 ceramics from polymer precursor, J. Eur. Ceram. Soc. 1991, 8, 29-34. 31. K. Thorne, E. Liimatta, and J. D. Mackenzie, Synthesis of SiC/TaC ceramics from tantalum alkoxide modified polycarbosilane, J. Muter. Res. 1991, 6, 2199-2207. 32. D. Seyferth, N. Bryson, D. P. Workman, and C. Sobon, Preceramic polymers as “reagents” in the preparation of ceramics, J. Am. Ceram. Soc., 1991,74,2687-2689. 33. P. Greil, Active-Filler-Controlled Pyrolysis of Preceramic Polymers, J . Am. Ceram. Soc. 1995, 78, 835-848. 34. R. M. Laine, F. Babonneau, K. Y. Blohowiak, R. A. Kennish, J. A. Rahn, and G. J. Exharos, The evolutionary process during pyrolytic transformation of poly(n-methylsilazane) from a preceramic polymer into an amorphous silicon nitride/carbon composite, J . Am. Ceram. Soc. 1995, 78, 137-145. 35. G. D. Sorarh, Q. Liu, L. V. Interrante, and T. Apple, The role of precursor molecular structure on the microstructure and high temperature stability of silicon oxycarbide glasses derived from methylen-bridged polycarbosilanes, Chem. Muter. 1998, 10, 4047-4054. 36. S. Yajima, K. Okamura, T. Matsuzawa, Y. Hasegawa, and T. Shishido, Anomalous Characteristics of the Microcrystalline State of S i c Fibers, Nature, 1979, 271, 706. 37. K. Okamura, M. Sato, T. Seguchi, and S. Kawanishi, Preparation of high-temperature strength S i c fiber, in Controlled Interphases in Composite Materials, H. Ishida (Ed.), Elsevier, Amsterdam, 1990, pp. 209-218. 38. D. Bahloul, M. Pereira, P. Goursat, N. S. Choong Kwet, and R. J. P. Corriu, Preparation of silicon carbonitrides from an organosilicon polymer; I, Thermal decomposition of the crosslinked polysilazane, J. Am. Ceram. Soc. 1993, 76, 1156-1 162.

References

475

39. K. Okamura, M. Sato, and Y. Hasegawa, Silicon nitride fibers and silicon oxynitride fibers by the nitridation of polycarbosilane, Cerum. Int. 1987, 13, 55-61. 40. G . D. Soraru, A. Ravagni, R. Campostrini, and F. Babonneau, Synthesis and characterization of p-sialon ceramics from organosilicon polymers, J . Am. Ceram. Soc. 1991, 74, 2220-2223. 41. G. D. Soraru, A. Ravagni, R. Dal Maschio, G. Carturan, and F. Babonneau, Polymer-derived Si3N4-Zr02 nanocomposite powders, J. Muter. Res. 1992, 7, 1266-1270. 42. G. Granozzi, A. Glisenti, R. Bertoncello, and G. D. Soraru, XPS study of the nitridation process of a polytitanocarbosilane into Si-Ti-N-0 ceramics, Muter. Res. Soc. Symp. Proc. 1992,271, 899-904. 43. D. Suttor, H.-J. Kleebe, and G. Ziegler, Formation of mullite from filled siloxanes, J . Am. Ceram. Soc. 1997,80,2541-2548. 44. M. Monthioux and 0. Delverdier, Thermal behavior of (organosilicon) polymer-derived ceramics. V: main facts and trends, J . Eur. Ceram. Soc. 1996, 16, 721-737. 45. G. D. Soraru, F. Babonneau, and J. D. Mackenzie, Mixed carbides via polymer route, Muter. Res. Soc. Symp. Proc. 1990, 180, 815-818. 46. M. Narisawa and K. Okamura, Polymer pyrolysis, in Chemical Processing ofcerumics, B. I. Lee and E. J. A. Pope (Eds), Marcel Dekker, New York, 1994, pp. 375-394. 47. G . D. Soraru, M. Mercadini, R. Dal Maschio, F. Taulelle, and F. Babonneau, Si-AI-0-N fibers from polymer precursor: Synthesis, structural and mechanical characterization, J . Am. Ceram. Soc. 1993, 76, 2595-2600. 48. L. V. Interrante, C. W. Whitmarsh, C-Y. Yang, W. Sherwood, W. Schmidt, P. S. Marchetti, and G. E. Maciel, Processing of Si-based ceramics and ceramic composites using hydridopolycarbosilane (HPCS), in Silicon-BusedStructural Ceramics, Ceramic Transactions, Vol. 42, 1994, pp. 57-69. 49. S. T. Gonczy, E. P. Butler, N. R. Khasgiwale, L. Tsakalakos, W. R. Cannon, and S. C. Danforth, Blackglas-Nicalon composites with CVD boron nitride fiber interface coatings, Ceram. Eng. Sci. Proc. 1995, 16, 433441. 50. P. Carri and G. D. Soraru, Sol-gel processing of continuous fiber reinforced composites by the liquid infiltration and pyrolysis (LIP) method, in Innovative Processing and Synthesis of Ceramics, Glasses and Composites, N. P. Bansal, K. V. Logan, and J. P. Singh (Eds), Ceramic Transactions, Vol. 85, 1997, pp. 405416. The American Ceramic Society, Westerville, Ohio, USA. 51. P. Colombo, T. E. Paulson, and C. G. Pantano, Atmosphere effects in the processing of silicon carbide and silicon oxycarbide thin films and coatings, J . Sol-Gel. Sci. Technol. 1994,2,601-604. 52. P. Colombo, T. E. Paulson, and C. G. Pantano, Synthesis of S i c thin films with polycarbosilane (PCS), J . Am. Ceram. Soc. 1997, 80, 2333-2340. 53. P. Colombo, G. A. Danko, and R. Silberglitt, Production of SIC and SiOC from preceramic polymers: comparison of microwave and conventional heating, in Ceramics ~- Getting into the 2000s, Part C (Section F), Techna Srl, Faenza, 1998. pp. 353-360. 54. P. Colombo and M. Modesti, Silicon oxycarbide ceramic foams from a preceramic polymer, J . Am. Ceram. Soc. 1999,82, 573-578. 55. P. Colombo, V. Sglavo, E. Pippel, and J. Woltersdorf, Joining of Reaction-bonded silicon carbide using a preceramic polymer, J. Muter. Sci. 1998, 33, 2409-2416. 56. R. Riedel, G. Passing, H. Schonfelder, and R. J. Brook, Synthesis of Dense Silicon-based Ceramic at Low Temperatures, Nature, 1992, 355, 714717. 57. K. Jakubenas and H. L. Marcus, Silicon carbide laser pyrolysis of polycarbosilane, J . Am. Ceram. Soc. 1995. 78, 2263-2266. 58. J. C. Pivin and P. Colombo, Conversion of inorganic-organic polymers to ceramics by ion implantation, Nucl. Instr. Meth. Phys. Res. B, 1996, 120, 262-265. 59. J. C. Pivin and P. Colombo, Ceramic coatings by ion irradiation of polycarbosilanes and polysiloxanes. Part I: Conversion mechanism, J . Muter. Sci. 1997, 32, 61634173. 60. J. C. Pivin and P. Colombo, Ceramic coatings by ion irradiation of polycarbosilanes and polysiloxanes. Part 11: Hardness and thermochemical stability, J . Muter. Sci. 1997, 32, 61756182. 61. S. Yajima, T. Shishido, and K. Okamura, SIC bodies sintered with three-dimensional crosslinked polycarbosilane, Ceram. Bull. 1977, 56, 1060-1063.

476

6 Polymer tn Ceramic Transformation: Processing of Ceramic Bodies and Thin Film&

62. K. B. Schwdrtz, D. J. Rowcliffe, and Y. Blum, Microstructural development in Si3N4/Polysilazane bodies during heating, Adv. Ceram. Muter. 1988, 3, 320-323. 63. R. Riedel, K. Strecker, and G. Petzow, In Situ Polysilane-Derived Silicon Carbide Particulates, J . Am. Ceram. SOC.,1989 72, 2071-2077. 64. H. Kodama and T. Miyoshi, Preparation of Fine-Grained Sic Ceramics from Pyrolyzed Polycarbosilane, Adv. Ceram. Muter. 1988, 3, 177-179. 65. H. Tanaka and Y. Kurachi, Sunthesis of P-SiC Powder from Organic Precursor and is Sinterability, Ceram. Int. 1988, 14, 109-1 15. 66. R. Riedel, M. Seher, and G. Becker, Sintering of Amorphous Polymer-Derived Si, N and C Containing Composite Powders, J . Europ. Ceram. SOC.,1989, 5, 113-122. 67. V. M. Sglavo, R. Dal Maschio, G. D. Sorard, and A. Bellosi, Fabrication and characterization of polymer-derived Si2N20-Zr02 ceramics, J . Muter. Sci. 1993, 28, 6437-6441. 68. G. Passing, R. Riedel, H. Schonfelder, and R. J. Brook, Polymer-Derived Ceramic Components, in Euro-Ceramics II, Vol. 2, G. Ziegler and H. Hausner (Eds), Deutsche Keramische Gesellschaft, Koln, 1992, pp. 601-606 69. R. Riedel, H. J. Kleebe, H. Schonfelder, and F. Aldinger, A Covalent Micro/Nano-Composite Resistant to High-Temperature Oxidation, Nature, 1995, 374, 526528. 70. R. Riedel, M. Seher, J. Mayer, and D. V. Szabo, Polymer-Derived Si-Based Bulk Ceramics, Part I: Preparation, Processing and Properties, J . Europ. Ceram. Soc., 1995, 15, 703-715. 71. M. Mayer, D. V. Szabo, M. Ruhle, M. Seher, and R. Riedel, Polymer-Derived Si-Based Bulk Ceramics, Part I: Microstructural Characterization by Electron Spectroscopy Imaging, J . Europ. Ceram. SOC.1995, 15, 717-727. 72. J. Bill, R. Riedel, and F. Aldinger, Hybrid Processing of Silicon-Based Polymers to Composite Ceramics, in Euro-Ceramics IV, Vol. 4, A. Bellosi (Ed.), Gruppo Editoriale Faenza Editrice, 1995, pp. 125-132. 73. L. An, R. Riedel, C. Konetschny, H.-J. Kleebe, and R. Raj, Newtonian viscosity of amorphous silicon carbonitride at high temperatures, J . Am. Ceram. SOC.,1998, 81, 1349-1352. 74. R. Riedel, L. M. Ruswisch, L. An, and R. Raj, Amorphous siliconboron carbonitrlde ceramic with very high viscosity at temperatures above 1500"C,J . Am. Ceram. Snc. 1998,81,3341-3344. 75. G. D. Sorard, V. M. Sglavo. F. Vulcan, and F. Babonneau. Fabrication and characterization of p-SiAlON components from polymeric precursors, Muter. Res. SOC.Symp. Proc. 1993, 287, 245-2 50. 76. T. Rouxel, G. Massouras, and G. D. Sorard, High temperature behavior of an SiOC oxycarbide glass: Elasticity and viscosity, J. Sol-Gel Sci. Technol. 1999, 14, 83-94. 77. R. Cytermann, A New Way to Investigate the Dependence of Elastic Moduli on the Microstructure of Porous Materials, Pow. Mettall. Int. 1987, 19, 27-30. 78. J. C. Pivin, P. Colombo, and M. Tonidandel, Ion irradiation of preceramic polymers thin films, J . Am. Ceram. SOC.1996,79, 1967-1970. 79. J. C. Pivin, P. Colombo, M. Sendova-Vassileva, J. Salomon, G. Sagon, and A. Quaranta, Ioninduced conversion of polysiloxanes and polycarbosilanes into ceramics: Mechanisms and properties, Nucl. Instr. Meth. Phys. Res. B, 1998, 141, 652-662. 80. J. C. Pivin and M. Sendova-Vassileva, Visible photoluminescence of ion irradiated polysiloxane films, Solid State Commun. 1998, 106, 133-136. 81. T. E. Paulson, Polymer processing of silicon (0xy)carbide thin films and microcomposite coatings, M.Sc. Thesis, The Pennsylvania State University, University Park, PA, 1994.

Part I11

Materials and Applications


Diamond Materials and their Applications Edited by I. Sigalas and R. J. Caveney

1.1 Superabrasive tools: A Brief Introduction M. W. Bailey

1.1.1 Introduction Tooling systems containing diamond and cubic boron nitride materials are used today across the breadth of modern industry, which could not function effectively without them. Their range of application covers petroleum exploration, mining, stone and concrete sawing, the grinding of tungsten carbides, glass, and ceramics, the machining of a wide range of engineering materials including aluminum alloys, modern composite materials, hard ferrous materials, and also in such unlikely fields as the cutting of frozen foods. Most applications take advantage of the extremely high hardness and wear resistance of these so-called ‘ultrahard’ or ‘superabrasive’ materials. However, more recently, the outstanding thermal and optical properties of diamond are being more widely exploited, giving rise to further (and new) areas of application.

1.1.2 Early History Diamond, as a material, has been recognized as having unique properties for several thousands of years and references of it being used as an industrial tool can be traced back to at least 300 BC. At this time, whole stones appeared to have been used for engraving and, by 150 BC, India, where the early diamond deposits were discovered, had established an export trade with China for such diamond-tipped engraving tools for use in cutting very hard jade stone (Fig. 1). No major change in the use of diamond as an industrial tool occurred for some considerable time. Reference was made by Pliny in his ‘The History of the World,’ published in AD 77, to splinters of diamond being used as engraving tools. Other industrial applications developed very slowly until the 18th and 19th centuries. In AD 1751, Diderot describes in his ‘Encyclopaedia’ a method of drilling rock using fragments of diamond held in the lower end of a diamond rod. This rod, a precursor to modern rock drills, was raised and dropped by an operator who also rotated it with his hand (Fig. 2). By the mid 18OOs, such rock drills were machine powered and a report in 1852 claimed that a hole, 5 cm in diameter, had been drilled in granite to a depth of 37 cm in 87 min. With modern diamond technology, this can now be achieved in a matter of seconds.

480

1 Diamond Materials and their Applications

Figure 1. Diamond trade routes from the 1st century BC to the 3rd century AD.

Other developments were taking place during this period. A diamond tipped lathe tool was used by J. Ramsden in 1773, the beginning of precision turning. The use of diamond as a wire drawing die was patented in England in 1819. By the end of the 19th century, large saw blades with diamond set around their

Figure 2. The first diamond-tipped ‘percussion’ drill (Diderot).

1. I Superabrasive tools: A Brief Introduction

48 1

periphery were in use in France for cutting stone. The cutting of marble and limestone used during the construction of large buildings in Paris in the early 1900s utilized diamond-containing saw blades up to 2-3 m in diameter. In 1842, the first mention was made of a diamond tool utilizing relatively small pieces of diamond when Pritchard reported grinding and polishing microscope lenses using a tool consisting of diamond grit hammered into the surface of a cast iron tool. The first patent for a diamond wheel was granted to A. L. Caverdon in France, in 1878, when he produced a wheel containing fine diamond particles held onto a metal wheel form possibly by jamming them into indentations made in the surface of the wheel. This method was described by Carl Zeiss in 1906. Most modern diamond tools, including the majority of saw blades and grinding wheels, consist of small particles of diamond held in a matrix or ‘bond’ which wears down during use to expose new diamond particles, which carry out the cutting. This is in contrast to the very early tools which used relatively large single stones or slivers from single stones. A patent for such a bond system was granted to the Western Electric Company in 1927. During the early 1930s, in order to grind the then-new material, cemented tungsten carbide, which had been invented by Krupp in Germany in 1928, grinding wheels containing diamond held in a resin bond were developed and several patents were granted between 1932 and 1933.

1.1.3 Synthetic Diamond The commercial availability of synthetic diamond has been a major contributor to the dramatic advances which have occurred during the past 50 years in both diamond tool technology and also in the range of applications in which they are used. In the period between 1953 and 1958, three independent teams of scientists from ASEA in Sweden, General Electric in the United States and De Beers in South Africa were successful in producing synthetic diamond by converting graphite to diamond using high-pressure, high-temperature technology. Prior to this, the limited supply of natural diamond impeded development of new diamond tools. The commercial availability of synthetic diamond offered two advantages. First, there was potentially unlimited availability of industrial diamond compared to the limited volumes of suitable natural material and, second, it offered the opportunity of engineering material to have specific properties suited to particular industrial applications. Of the diamond used for industrial purposes today, some 90% is synthetic, and 10% natural. The major use of diamond grit is in the machining (sawing and drilling) of stone and concrete (Fig. 3). Finer sizes of diamond are used widely for the grinding, sawing, and polishing of glass, ceramics, tungsten carbides, and a very wide range of other mainly nonferrous engineering and industrial materials. Specialist applications, for example fine high precision turning (Fig. 4) of nonferrous and precious materials, drawing of fine wire, and the dressing of conventional abrasive wheels have, for several decades, used natural diamond. During the 198Os, the technology to produce, economically, large synthetic single diamond crystals

Figure 3. Diamond drilling a 100 mm hole for a new sea wall drainage system

was developed. Whole synthetic crystals or cut pieces of the order of several millimeters in size are used for these high precision industrial applications.

1.1.4 Cubic Boron Nitride Following the successful commercial synthesis of diamond in the 1950s, the second hardest material known, cubic boron nitride, cBN, was introduced to the market in the 1960s and is complementary to diamond. The iron, and its alloying elements, in ferrous materials has a tendency to react chemically with diamond under machining conditions and this can reduce the efficiency of the tool. cBN, however, although not as hard as diamond, does not react chemically with iron and is therefore particularly well suited to machining hard ferrous materials.

1.1.5 Polycrystalline Diamond and Cubic Boron Nitride The next notable material development was the introduction to the market in the early 1970s of polycrystalline diamond (pcD). This made available to industry

1.1 Superabrasive tools: A Brief Introduction

483

Figure 4. Natural diamond cutting tool machining a contact lens.

relatively large pieces of diamond, albeit polycrystalline rather than monocrystalline, at an economical cost. Such materials can be produced with varying mechanical properties and are used in a wide variety of cutting applications and also as a wear resistant surface. As a cutting tool material, it is used predominantly to machine nonferrous abrasive materials including drilling of rock in oil and gas exploration, and in the engineering and manufacturing industries for the machining of aluminum alloys, reinforced plastics, new wear resistant lightweight composite materials, wood, and wood composites (Figs 5 and 6). pcD is also used in noncutting applications and one of its original and important applications is as a die for drawing wire. More recently, the use of pcD as a wear resistant component in thrust bearings, work rests, and high precision gauging equipment has become more widespread. The development of polycrystalline cBN (pcBN) in the mid-1970s made it economically possible to machine fully hardened ferrous workpieces on a lathe or milling machine when previously the only method of machining such workpieces was by grinding, normally with silicon carbide or aluminum oxide (conventional) abrasives. Huge increases in productivity can be achieved by changing to turning (or milling) with pcBN. In addition, environmental improvements can also be achieved, since in many cases pcBN does not need coolant.

484


Figure 5. A pcD-tipped cutter machines oak panels for kitchen furniture.

1.1.6 Chemical Vapor Deposited Diamond The development in the late 1980s of an entirely new method of diamond synthesis using chemical vapor deposition (CVD), as an alternative to ultra-high pressure and temperature systems, has caused a considerable stir in the industry.

Figure 6. pcD machining a cast metal-matrix composite automotive brake motor in A1 20% Sic,.

1.2 The Crystullizution qf Diamond

485

With this technique, it is possible to deposit layers of diamond onto a substrate from a hot carbon-containing plasma, enabling relatively large areas of diamond to be produced (at the time of writing, the order of 10000mm2). Traditional diamond tools utilize the extreme hardness of diamond (and cBN) but, with this new technology, which is still developing, the exceptional optical and thermal properties can be applied. Diamond has a very high thermal conductivity (up to six times that of copper at room temperature) and is transparent to visible and infrared radiation, opening the door to a new varied range of industrial applications where these properties can be exploited in areas such as thermal management of electronics and windows for high technology infrared equipment, in addition to its use as an alternative material for use in the more traditional diamond tool areas.

1.1.7 Outline of Chapter Section 1.2 of this Chapter reviews the crystallization of diamond and cubic boron nitride using high pressure, high temperature techniques. Chapter 4 of Part I1 deals with the subject of chemical vapor deposition of diamond at low pressure. The high pressure sintering of diamond and cubic boron nitride to form superhard composites is described in Section 1.3, a review of attempts to produce other hard materials is given in Section 1.4. Section 1.5 summarizes the many and varied applications of diamond and cBN materials. The applications of diamond grown using the CVD technique are discussed in Chapter 2 of Part 111.

1.2 The Crystallization of Diamond G. J. Davies

1.2.1 The Carbon Phase Diagram It was established that diamond is an allotrope of carbon by Tennant in 1797 [1]. This led to many attempts to crystailize diamond using various carbonaceous starting materials, but it was not until about a century and a half later that successful synthesis was proven, as referred to in Section 1.1.3. The first clear success was by a Swedish group at the ASEA Company in February 1953 [2]. This was followed by the General Electric Company in December 1954 [3]. The early attempts and the subsequent successes are well reviewed [4-61. Natural diamond has been used by man since at least biblical times, not only as a gem but also, due to its extreme properties relative to other materials, as a n abrasive and even as a medicine in crushed form, ‘a panacea for all ills’. It is not intended here to enter the debate concerning the crystallization of natural diamond. Some general

486


h

h (3 W

cc

I3 v) v)

W

cc a

0

1000

2000

3000

4000

5000

6000

TEMPERATURE (K)

Figure 7. The phase diagram of carbon adapted from [9]. A: region of solvent/catalyst-based synthesis of diamond from graphite using: A, : commercially used transition metal solvent/catalyst [ 161 A2: nonconventional metal solvents: Cu, Zn, Ge [17-201 A,: nonmetallic solvents, mainly metal salts, silicates [22-281 B: the synthesis of diamond from phenolic resin and cobalt [30] C: region of direct conversion of hexagonal graphite to either hexagonal or cubic diamond, where the dotted line C1-2,represents the solid-solid transformation line of graphite or hexagonal diamond to cubic diamond through fast pressure/temperature cycles [9] C,: conversion of hydrocarbons to cubic diamond [33] C4: transformation of C60fullerenes into polycrystalline diamond [35] D: low pressure metastable growth of diamond [12, 13, 36-38] N: regions of natural diamond genesis [7, 8, 21, 291.

natural diamond genesis information is to be found elsewhere [7, 81. The conditions under which natural diamonds are formed will, however, be touched upon when the phase transformation diagram for carbon is discussed later in this section on crystallization. To gain a balanced general view of the many ways in which man has crystallized diamond, the thermodynamically stable and metastable phases of elemental carbon and reaction dynamics between them, over obtainable pressures and temperatures, should be considered. The most up to date phase and transformation diagrams are to be found in a review by Bundy et al. [9]. Figure 7 is an adaptation of the pressuretemperature phase transition diagram taken from this review article, together with further information gleaned from the extensive diamond- and carbon-related scientific literature. The dominant thermodynamically stable forms of carbon are graphite, diamond, liquid, and vapor. The phase diagram presented by Bundy [9] and adapted in Fig. 7

1.2 The Crystallization of‘ Diamond

487

has some new notable features over many of the earlier diagrams. The slope of the diamond melting line is positive, and molten carbon is said to be metallic with little evidence for transformations from electrically conducting to nonconducting forms. For consideration of diamond crystallization, one of the most important features is the diamond-graphite equilibrium line which stretches from 1.7GPa at OK to 12 GPa at 5000 K, the diamond-graphite-liquid carbon triple point. The activation energies required to initiate transformations from one form of carbon to another are very largely due to the high cohesive energy of carbon in its crystalline structures. Davies and Evans [lo] showed that the activation energy for the graphitization of the { 1 lo} surface of pure diamond in vacuo is 728 f 50 kJ mol-’, which is about twice the bond energy of the C-C bond in aliphatic hydrocarbons (3.7eV per CJ C-C bond). This is also about the same energy required for the vaporization of graphite. These very high activation energies result in very large areas of pressure and temperature over which a metastable form of carbon can subsist. Probably the most important example of this is the indefinite existence of diamond at room temperature and pressure even though, at these conditions, it is very ‘deep’within the graphite stability field. To overcome the activation energy for conversion to graphite in vacuo, diamond must be heated to a temperature-of about 2000 K before graphitization becomes rapid [ 101. In addition to the allotropes of carbon, there are polytypes of carbon, that is structural variations based upon lattice plane stacking difference. Important examples are hexagonal diamond, which occurs in methods of very rapid and hot direct transformation into diamond, and rhombohedra1 graphite, produced by cold working normal hexagonal graphite crystals. When the nanoscale groupings of pure carbon atoms are considered, other solid forms with very distinctive properties are known and are probably best described as ‘metastable’ over the whole range of pressures and temperatures in Fig. 7. These are amorphous forms such as carbon black and glassy carbon, condensate of very long linear carbon molecules collectively called carbynes and, of course, the now very extensive variations of fullerenes (C60, C70, etc.) and the derived nanotubes [ 1 I]. Importantly for diamond crystallization, all these forms of carbon, including graphite, very stubbornly persist at pressures far into the diamond stable region. The practical synthesis of diamond could thus be considered as the search for means of finding practical convenient kinetic pathways to avoid the high activation barriers between the multifarious nondiamond carbon structures and diamond. The two very general approaches are to use solvents for carbon and to precipitate diamond primarily in the diamond stable part of the phase diagram or to ‘quench’ carbonaceous species from very high energy states such as vapors, liquid, or plasmas. Some of the more recently reported means could well be considered to be combinations of these [12, 131. Some of the most important diamond crystallization techniques in the recent literature will now be itemized using Fig. 7 as a guide.

1.2.2 Diamond Crystallization at High Pressure The shaded region in Fig. 7 labeled with pressure from about 4.5 GPa to 9 GPa and temperature from about 1475K to 2475 K, gives the conditions under which the

488


bulk of the reported solution grown, diamond stable region, crystallization of diamond has taken place. The graphite-diamond equilibrium expression in this region, following Berman and Simon [14] is P

AG:=AH~-7.AS~,o[

AVdP=O

It has stood the test of time and experimental verification [ 151. Al is where most of the commercial diamond abrasive products and high pressure polycrystalline diamond products are manufactured using alloys of transition metals, mainly iron-nickel, cobalt-iron and to a lesser extent nickel-manganese for the grits and larger individual crystals, and mainly cobalt for the polycrystalline products [16]. In this reference, the general transition metal solvent work up to the end of the 1980s is covered. In principle many of the other transition metals and alloys can be used but a large number of them require very difficult to attain and control temperatures and pressures dependent upon melting points, carbide stabilities, and carbon solubilities. The position A2, somewhat higher in pressure and temperature than A , , covers the region where the higher melting and or lower carbon solubility metals and alloys can be used. Diamond synthesis from graphite has been demonstrated using copper, zinc, and germanium [17, 181. Magnesium was shown to be a workable solvent/catalyst [19] at about 7.7 GPa and 2275 K. Later, by the addition of 5&60% copper to magnesium, diamond synthesis conditions were reduced to 6 GPa and 1825 K [20]. The A3 position in Fig. 7 is the general condition under which inorganic compounds have been used to aid in the conversion of graphite to diamond. The more recent work in this area has mainly been carried out at the National Institute for Research in Inorganic Materials, Tsukuba, Japan. This work was probably stimulated by the possibility that volatiles such as C 0 2 , H 2 0 , CH4, and O2 may play an important role in natural diamond formation [21]. Carbonates [22], hydroxides [23] and sulphates [23, 241 with alkali and alkali earth cations have proven to work at 7.7GPa and temperatures in the range 1825-2475K. The diamonds formed were aggregates with particles up to 20 pm in size with normal faceting. In order to clarify the possible role of C 0 2 , silver carbonate has been investigated compared to silver metal [25]. This compound aided diamond formation above about 2075K and 7.7GPa, whereas pure silver did not act as a solvent/catalyst under these conditions. Further, volatile-rich silicate melts of kimberlite composition (kimberlite is the igneous rock usually associated with natural diamond in diamond bearing pipes) and Si02-H20 combinations aided graphite conversion to fine diamond at 7.7 GPa and 20750-2475 K [26, 271. The need for the presence of volatiles when using silicates to aid diamond formation has been recently confirmed [28]. In this report, anhydrous magnesium silicate melts corresponding to the composition of the minerals forsterite (Mg2Si04)and enstatite (MgSiO,) were compared with hydrated versions of these compositions. Fine diamonds of euhedral shape were formed only in the hydrated magnesium silicate graphite systems above about 2075 K at 7.7 GPa. No diamond was detected in the anhydrous systems up to

1.2 The Crystallization of Diamond

489

2175 K at this pressure even after treatments lasting 8 h or more. These findings are consistent with the current theories of natural diamond genesis. The two regions both marked N in Fig. 7 are intended to illustrate the conditions under which natural diamond formed in the earth’s crust, probably over geological time scales (about 3.3 billion years). There are two environments linked to natural diamond formation, the peridotitic (depicted by the lower pressure N region on Fig. 7) and the eclogitic (at higher pressures and about 200 K hotter, also marked N). The pressure given by the N positions are estimated from the accepted correspondence of 0. 1 GPa pressure to 3 km depth into the earth crust. The lower pressure limit of N is given by the boundary zone between the upper region lithosphere and the mobile asthenosphere at about 150 km in the earth’s mantle. It is believed that natural diamonds, crystallized from C 0 2 or CH4 volatiles below the lithosphere and after existing for great periods of time at these depths, are brought to the surface by subsequent eruptions of magma. For more detailed discussions of the geology, mineral inclusions in natural diamond and general matters related to diamond genesis, the reader is referred to Harris [7], Kesson [8], Haggerty [29], and Meyer [21]. The position marked B at 4 GPa and temperatures between 1775 and 2075 K, gives the condition under which well faceted diamonds of about 200 pm diameter have been crystallized using specially treated phenolic resins as the source of carbon and molten cobalt as solvent [30]. The notable point of this particular crystallization is that it occurred well into the graphite stable region of Fig. 7 and is thus an example of the ‘metastable growth of diamond’. An explanation for this may be found in consideration of the relative solubilities of phenolic resins and diamond in molten cobalt allowing dissolution of one metastable form and precipitation of another, namely diamond. This is an example where rules governing the transitions between metastable states, such as the Ostwald and Ostwald-Volmer rules, can be applied [3I]. The curved dashed line CI-Cz in Fig. 7 depicts the proposed boundary between the formation of the hexagonal (below the line) or cubic (above the line) polytypes of diamond, after explosive shock compression or fast heating, mainly using laser heated diamond anvil cells, of graphite structures in the diamond stable region, described by Bundy et al. [9]. This reference is a detailed review article and is a major source of reference for work concerned with the direct conversion of graphite and other carbon structures to diamond. The diamonds that can be retrieved from these various methods, either of the hexagonal or cubic form, are always small and crystallographically defective. This is because of the very short times at high temperature and pressure, and the need to satisfy fast thermal quench requirements. Despite this, diamond powders have been commercially produced by the Du Pont Company of the USA and others using explosive shock of metal containing small crystals of graphite as a carbon source [32]. The direct conversion of hydrocarbons to diamond has also been demonstrated. C 3 is the region where organic compounds such as anthracene, camphene, fluorene, pyrene, sucrose, polyethylene, adamantane, and paraffin wax have been converted to very fine diamonds under high pressure (about lpm in size) (Wentorf [33]). Aliphatic compounds seem to work but compounds with aromatic rings and/or

490


large amounts of nitrogen only formed graphite of excellent crystallinity under high pressure pyrolysis and did not form diamond. There is a growing body of literature describing the crystallization of diamond from fullerenes as a starting material. One of the latest [34] claims that c 6 0 can be transformed to diamond at 5GPa and 1675K using a solvent/catalyst without forming graphite first. The generation of ‘bulk’ polycrystalline diamond at room temperature has been reported by the crushing of c 6 0 at about 20 GPa [35], (position C4 in Fig. 7). The conditions used in the polycrystalline diamond (cobalt-matrix) anvil cell were deliberately nonhydrostatic with respect to pressure. It was suggested, from the high efficiency of conversion and fast kinetics involved, that, if cheap bulk C60became available at some time in the future, a feasible method for the commercial production of polycrystalline diamond might result. The metastable crystallization of diamond in the appropriate pressure range of 0-0.5 GPa and temperature range 675-1300K has now been accomplished using a ‘galaxy’ of techniques in region D in Fig. 7. These include chemical vapor deposition from carbonaceous gases, mainly methane, hydrogen mixtures using many varied means of generating plasma [36, 371 (see also Chapter 4 of Part 11), the use of a combination of lasers and carbon dioxide as source (the so called QQC deposition process) [13], hydrothermal synthesis in the C-H-0 and C-H-Ohalogen systems [38], low pressure solid source processes (the so called LPSSS methods) [12], and many more. The literature in this general area of metastable diamond growth has now become extensive, as discussed in Section 1.2.6. From a commercial point of view, the production of diamond abrasives is, at this time, dominated by the high-pressure, high-temperature, transition metal solvent/catalyst techniques where the crystallization takes place in the diamond stable region.

1.2.3 High Pressure Apparatus Early attempts at generating high pressure used opposed anvils, frequently referred as ‘Bridgman anviis’ after the father of high pressure, Professor P.W. Bridgman [39]. A constraint of this system was the limited volume of material to which the high pressure could be applied. This problem was addressed by forming recesses in the two opposing anvils, to form the so-called) ‘toroidal’ device used extensively in Russia [40]. However, most commercial growth of diamond using metallic solvent/catalysts has been carried out in the belt/girdle and multi-anvil devices described below. 1.2.3.1 Belt/Girdle Devices The pressure limit of the piston and cylinder apparatus [39] was extended to allow commercial production of synthetic diamonds by designing the cylinder to accommodate pistons shaped as truncated cones. If the die is reinforced by prestressing with concentric rings with interference fits, the device can sustain a pressure of about 6GPa through use of an appropriate gasket (talc, pyrophyllite, and mixtures thereof) at the die/anvil interface.

I .2 The Crystallization of Diamond

49 1

The gasket allows an even distribution of stress in the die taper and along the tapered anvil cone. In this way, both the die and anvil are supported by the resulting compressive stresses. The essential difference between the belt and girdle devices is the shape of the die bore/taper and the matching anvils. For the belt apparatus, the die bore is curved continuously from center to outer surface whereas, for the girdle, the die bore is straight and proceeds out on both sides in a linear progression.

1.2.3.2 Multi-anvil Devices The first multi-anvil press was invented by Von Platen [41]. It comprised six anvils in a cubic arrangement. The benefit of a muiti-anvil device is the possibility of multistaging, which allows the yield strength of a compressed component to be increased. Russian workers [42] have made extensive use of a multi-staged split sphere device. Pressure is applied to two sets of anvils. The outer arrangement contains eight anvils which form an octahedral-shaped cavity, and an inner set of six anvils is placed to form a cubic-shaped central cavity that contains the high pressure capsule in which the diamonds are grown.

1.2.4 The Synthesis of Particulate Diamond Abrasives Crystal growers in most of the world’s research and commercial organizations [ 16, 43,441 have efficiently harnessed the solvent/catalyst growth of diamond to produce both abrasive grits and large single crystals. Scientists have developed mechanisms to explain the formation of diamond. These have been described by many workers and could involve the formation of a C+ ion or the formation of an intermediate carbide which decomposes at synthesis conditions. However, the more accepted explanation of growth is based on supersaturation. For example, if the well known Ni-C equilibrium phase diagram, Fig. 8 [45], at 5.7 GPa is studied, an explanation of the formation of diamond under isothermal (thin film growth) or temperature gradient conditions can be obtained. Both diamond growth techniques are possible since the solubility of diamond in the molten solvent/catalyst, nickel, is less than the solubility of graphite. This section will focus upon the crystallization principles used to produce commercially synthetic diamond ‘grit’ abrasives. The size of particles discussed here are within an envelope of about 850-200 pm for grits used mainly in stone and concrete sawing (saw sizes) and from about 200-50 pm used for grinding applications (wheel sizes). The solvent/catalyst crystallization method generally uses graphitic carbons as the source material at constant high temperature and high pressure. The general pressure and temperature conditions used are all within the diamond stable region, at position A, in Fig. 7. The lowest conditions are about 5 GP a and 1500 K, and the highest about 5.8 GPa and 1775 K. These conditions are determined by the transition metal alloy systems together with the crystal nucleation and growth conditions all necessary to crystallize the diamond structures required for specific

492

1 Diamond Materials and their Applications 1700

,

I

I

1600

1300 1240 0

1.0 4.7

2.0 9.0

3.0 13.15

4.0 Wt.% C 16.7 At.% C

99

100

Figure 8. Ni-C phase diagram at 5.7 GPa [45].

product types. The diamond industry has spent many years and significant amounts of money developing and refining the large scale high pressure equipment appropriate to generate and maintain economically the required crystallization conditions. The high pressure equipment used by the two largest manufacturing companies, namely De Beers Industrial Diamond Division and the General Electric Corporation are proprietary, highly evolved designs based upon the belt and/or girdle apparatus described in Section 1.2.3.1. The Chinese producers mainly use cubic multi-anvil high pressure designs, and the Russians and Ukrainians use various modifications of the toroidal equipment. Possibly the first question a crystal grower must ask when investigating or developing a crystal growth method is, ‘how is an appropriate driving force for nucleation and growth generated and controlled’? The central part of Fig. 8 is considered as representative of a general transition metal/carbon phase diagram at a pertinent pressure (5.7 GPa) and temperature range. The driving force for crystallization of diamond using ‘ideal’ graphite as source carbon would be given by the difference in solubility between graphite and diamond (the supersaturation) at the chosen conditions. For example, in Fig. 8 at 1450°C and 5.7 GPa this would be the solubility difference AB, (A’B’). Note that this supersaturation for diamond crystallization (let it be Ac) would decrease as temperature is increased and would essentially be zero at the diamond/graphite equilibrium temperature, E(eq.)in Fig. 8. The slopes of the diamond/metal liquidus line and graphite/metal liquidus line at constant temperature will change with pressure. The difference between these solubility lines, Ac, will thus also change with pressure and similarly become zero at the diamond/graphite equilibrium condition. Thus the driving force for crystallization, Ac, is a function of the ‘distance’ from the diamond graphite equilibrium line in pressure/temperature space, Fig. 9. The supersaturation, Ac, can of course be related to the nucleation rate and growth rate of the diamond in the classical way as shown in Fig. 10. Note the extremely strong dependence of nucleation rate upon supersaturation and that there is generally a critical level of supersaturation, Accrit,which must be


493

DIAMOND STABLE

a GRAPHITE STABLE

TEMPERATURE

Figure 9. Pressure-temperature diagram for diamond synthesis showing typical conditions ( P I ,T I ) . The supersaturation Ac is a function of P , T I giving rise to different crystal morphology at different conditions.

Figure 10. The classical relationships of the rates of nucleation and growth to supersaturation, Ac, for crystallization from a solution.

494


exceeded before the probability for nucleation becomes large enough for an observable nucleation rate. The growth rate is an increasing function of supersaturation. The relative growth rates of different crystallographic growth zones is also a function of Ac, hence the morphology shift with pressure, temperature condition as depicted in Fig. 9. The relationships between solubility, supersaturation, pressure, and temperature depend upon the details of the microscopic mechanisms operative in the context of the detailed active impurity chemistry and the specific geometric relationship set up between the source carbon, molten metal solution, and nucleation site, and or growing diamond. For an introduction into these extensive subjects the reader is referred to Burns and Davies [16] where the issues such as nitrogen and boron active impurity chemistries, diamond surface reconstruction, morphology and solvent/catalyst are discussed. An important conclusion to be drawn from the above fundamental explanation is that the structure of the source carbon as it pertains to solubility in the chosen molten metal is very important. Will and Graf [46] compare six graphite types using an iron-nickel solvent and show a clear dependence of rate of diamond formation and diamond size upon the starting crystallinity of the graphites. Moreover these authors include consideration of the influence of gaseous impurities in the graphites and the cluster theories of Sunagawa [47] concerning the carbon species in solution in the metal. As the source carbon became more crystalline, the pressure needed for diamond nucleation increased by up to 0.5GPa. This could be interpreted as the need to increase the pressure in order to exceed the critical supersaturation for spontaneous nucleation, Accrit,owing to the much lower solubility of the recrystallized and thus more crystalline graphite. Another striking example of the importance of the structure of source carbon leading to crystallization consequences via the supersaturation driving force is described in Wang et al. [48]. Here, the pressure needed for spontaneous nucleation was shown to increase with time as the graphitic source material recrystallized in the high pressure reaction chamber. In addition to the explained (at least in general terms) dependence of crystallization on the structure of the source carbon, the character of the transport of solute from dissolving source to growing diamond crystals needs to be considered. When a diamond synthesis reaction chamber from a process for making grits is broken open after completion of the crystallization cycle, it is observed that each grown crystal is separated by a metal alloy film from the undissolved carbon source material [16]. The metal film, which of course was molten during the growth of the diamonds, is usually about 50-200 pm in thickness. Figure 11 is a schematic drawing of this growth geometry. The carbon source material is dissolved at the molten metal interface, D,and the resultant carbon species in solution transported by diffusion to the growing diamond surface. The dimensions of these molten metal films are such that it is unlikely that convection plays any significant role in carbon transport for this type of crystallization. The growth of diamond grits of good morphology occurs through lateral propagation of steps over the surface. The detailed kinetics of surface adsorption, surface diffusion and incorporation of carbon species into the step structures needs to be studied to understand the

1.2 The Crystallizatioiz of Diamond

495

Figure 11. Schematic growth environment of a synthetic diamond particle.

microscopic mechanisms of growth. The different chemical and reconstructed structures of the lowest energy crystal facets of diamond give rise to different growth rate responses to the flux of carbon species resulting in a growth zone history and particular final morphology for each crystal. The morphology of diamond grits may be related to the pressure and temperature conditions of growth through the relative growth rate response of the different crystallographic surfaces to the flux set up by the supersaturation, Ac, produced by these pressure and temperature conditions, as in Fig. 10 [ 161. Thus a morphology map may be drawn on such diagrams showing the regions of pressure and temperature of common morphology [ 161. The difficulty of investigating the detailed growth mechanisms is extreme due to the ‘inaccessibility’ of the growing crystal in practical high pressure, high temperature equipment. The crystal defect types and distributions which characterize any real crystal, diamond abrasive grits being no exception, may also be discussed in terms of the driving force for crystallization, and the impurities in solution and/or suspension in the solvent and the geometry/transport considerations of the growth environment. Examples of this are the oriented metal solvent inclusions described by Wakatsuki et al. [49]. These inclusions occur as an array, along certain growth zone boundaries leading to certain edges of the crystals. The most important crystal defects for diamond abrasives are those which may affect their behavior in applications and also those which give rise to their appearance, such as color and clarity. A discussion of the characterization of synthetic diamond abrasives in terms of strain, impurity content, both as inclusions and lattice defects, and crush strengths is given by McCormick et al. [50].These authors consider the macroscopic structure such as shape, external morphology, cracks, and large inclusions together with the microscopic structure such as small inclusions, dislocations, and some nitrogen related lattice defects. Using strain sensitive techniques such as Raman spectroscopy, photoluminescence, and birefringence, together with information from extensive literature, they arrive at hierarchies of the macroscopic and microscopic structures

496

I Diamond Materials and their Applications

and properties that affect the strengths of the diamond particles. Not surprisingly, the most important macroscopic structures of the diamond particles turn out to be size, shape, crack, and inclusion content. It was also demonstrated that particular nitrogen related lattice defects such as the H3 and N-V center affect the strength of the crystals. For a comprehensive discussion of the observed impurity lattice defect centres such as H3 and N-V, the reader is referred to Davies [51]. An important observation in [50] was that some of the microscopic defects may well strengthen the diamond by impeding dislocation motion and possibly acting as crack arresters. The very complex nature of the macroscopic and microscopic structures as they affect strength and the behavior of diamond abrasive/particles still requires extensive work to elucidate fully. However, from a crystallization point of view, gaining control over the crystallization behavior is the key to the production'of optimal diamond abrasives. This, of course, may be achieved by choice and manipulation of the pressure/temperature conditions, source carbon structure and solvent/catalyst metal type, leading to control over nucleation and growth rates.

1.2.5 Growth of Large Synthetic Diamonds R. C. Burns and M. Sibanda Large diamonds are best grown using the temperature gradient technique, where diamond is used as the source of carbon. In this technique, as described by Wentorf [52], the driving force for reconstitution, under diamond stable conditions, is provided by the higher solubility of diamond in a hot zone of the solvent/catalyst and the consequent crystallization of diamond in a cooler zone at constant pressure (Fig. 9). The design of the high pressure cell also influences the rate at which this carbon transport mechanism proceeds. A schematic diagram of a high pressure cell used to grow reconstituted diamond is shown in Fig. 12. A useful axial temperature gradient is established by manipulating the distance between the diamond source and seed pad, which (at constant cell pressure) determines the growth rate and hence, quality and size of the diamonds grown on the seeds in a fixed time. The small, well formed diamond seeds, which act as a template for new growth, are mechanically attached to the seed pad. For a reaction cell pressure of about 5.5 GPa, the temperature at the diamond source is about 1450°C. The solvent/ catalyst bath is agitated by convection and soon saturates with carbon. The solvent/catalyst in the vicinity of the seeds, at a temperature of 1420°C, is therefore supersaturated with respect to diamond, providing the driving force for diamond growth on the seeds, which are selected to be about 0.5 mm in size [53, 541. The nucleated diamond is restrained by the seed pad, resulting in truncated cubooctahedral growth, whereas the orientation of the grown diamonds depends on the orientation of the seed. The habit of the reconstituted diamond depends largely on the solvent/catalyst, internal pressure and temperature of the reaction cell.


DIAMOND POWDE

t

CARBON HEATER SOLVENT-CATALYS

497

AT

GROWING CRYSTAL SEED CRYSTAL

T2

f

Figure 12. Schematic diagram of a high pressure cell used in the reconstitution method.

Diamonds of a cubic habit tend to be formed at a lower growth temperature than those of an octahedral habit [47] (refer to Fig. 9). The chosen spacing between diamond seeds is determined by the intended size of the reconstituted diamond. This spacing also affects the growth rate. In practice it is found, for a fixed temperature gradient, that the smaller the spacing between seeds, the slower the growth rate. Figure 13 shows a selection of large synthetic diamonds (0.5 to 1.8ct in weight) grown by De Beers Diamond Research Laboratory from seeds using the temperature gradient technique. Crystal perfection is determined by a reasonably low growth rate which is achieved by optimizing, in combination, the space between seeds, the axial temperature gradient and the solvent/catalyst under diamond synthesis conditions.

Figure 13. Diamonds grown by De Beers Diamond Research Laboratory from seeds using the temperature gradient technique.

498


1.2.5.1 Solvent/Catalysts for the Growth of Large Synthetics 1.2.5.1.1 Type Ib Diamonds Effective solvent/catalysts for high pressure, high temperature diamond synthesis (thin-film and temperature gradient techniques) are derived from metals and alloys of Group VIII of the periodic table and metals such as chromium, manganese, tantalum and niobium [55-571. The effectiveness of transition metals and their alloys depends on a number of factors which include: their melting points, the degree to which they influence the supersaturation of the molten solution and also the stability of possible metal carbides at synthesis conditions [58]. Most commercial solvent/catalyst alloy systems make use of a combination of metals [16, 59-61] (e.g. iron-nickel-chromium, iron-nickel, iron
Co-Ti [62] Fe-A1 [65] Fe-Co [62] Fe-AI-B (0.02 wt YOB) [65] Ni [43] Ni -Fe Fe-Co-Ti [65]

Solvent

-

-

-

-

-

3.5 4

2.3 1.8 5 1.3

~

Growth rate (mg h-')

300 500 1000 160

Growth time (h)

3.4 4.6 25 5.1

~~

Max. size ( 4 near colorless near colorless yellow blue yellow yellow near colorless

Color

50-100 50--400
-

<0.01-2 0.4-1 1 100-300

Nitrogen content (p.p.m.1

( l l l ) , (100) > {113), {115), {llO} { l l l ) > (113) > (100) > (110) { l l l } , {loo} >> {IlO), j113) { l l l ) , {110), (100) {113), {115} { I l l } , (110) >> {110), (113) (1111, {loo) {Ill}, (100) > {113}, {115}, {llO)

Morphology

Table 1. Summary of large synthetic diamonds grown for research purposes by some of the leading producers of synthetic diamonds

W

P W

R

;

b 8'

=:

-

$CI

2 0

x

500


Co content in Fe-Co solvent (&A)

Figure 14. Nitrogen concentration in diamond versus the cobalt content of iron + cobalt solvent. The diamonds were grown with 1.5 weight-% titanium addition using (a) high-purity graphite and (b) synthetic diamond powder [67].

a more efficient nitrogen-getter under conventional synthesis conditions. At higher temperatures, however, aluminum becomes a more efficient nitrogen-getter than titanium as shown by diamond synthesis experiments at 1550°C where the use of titanium yielded yellowish and heavily included crystals [62]. Boron is often added in the form of a stable metal boride to gettered reaction volumes, to generate blue semiconducting diamonds (type IIb) [62].

1.2.5.2 Purity of Large Synthetic Diamonds Synthetic diamonds invariably contain

, \

,

,

',

Al \

Ti

--_ Added amount of Al or Ti to the solvent-catalyst(wt%)

Figure 15. Nitrogen concentration of diamond grown with A1 or Ti added as a nitrogen getter [67].

1.2 The Crystullization

of' Diamond

501

impurities. These impurities fall into two classes. There are impurities on an atomic scale (point defects) and these include nitrogen, boron, and metallic elements such as nickel. The second class of inclusions are present in the form of metallic or nonmetallic particles, both microscopic and macroscopic. The occurrence of these inclusions is controlled by crystal growth kinetics.

1.2.5.2.1 Inclusions Solvent/catalyst metal inclusions can occur in large synthetic diamonds. The number, size, and shape of these inclusions depends to a large extent on the growth rate and the synthesis temperatures and pressures. Extensive research in this area shows that the uptake of metallic inclusions occurs predominantly during the early stages of growth when the carbon flux density is very large, particularly when a small seed crystal is used [68]. One suggestion is the use of a two-stage growth method [63, 691 in which a small recess is made on the seed pad and the seed is located at the bottom, such that it does not protrude above the recess. The net result is that carbon supply to the seed is limited as the seed grows within the recess during the initial stages of growth. Once the seed has grown out of the recess, it has an enlarged surface which can accept more carbon without compromising quality. The uptake of inclusions during the early stages of a growth can, therefore, be minimized by paying special attention to the depth and diameter of the seed recess. The use of larger seed crystals tends to increase the growth rate. The coupled effect of growth rate and seed crystal size [70] on the uptake of inclusions is summarized in Fig. 16 where it is evident that the size of the seed crystal is an important parameter in the growth of high purity diamonds. In general, the population and size of inclusions increases with an increase in growth rate. There are many different types of inclusions that occur and most of these have been comprehensively dealt with elsewhere [44]. One of these types of inclusions results from hoppered growth when there is a high supersaturation [71], as shown in Fig. 17. Low growth rates and stable growth conditions are therefore important for the growth of high purity large synthetic diamonds. The choice of solvent/catalyst alloy which incorporates a getter is important in controlling the quality of grown diamond. In the growth of near-colorless diamonds, the tendency for the getter elements to form stable carbides can be a problem. When titanium is used, a very stable carbide is formed whilst, with aluminum, the A14C3carbide formed is readily decomposed at diamond synthesis conditions. Inclusions therefore occur more readily with titanium than with aluminum gettered systems. At high levels of titanium or aluminum, inclusion content can be a problem. When titanium is used, many fine particles of TIC are formed [72, 731 and included in the growing diamond crystal in addition to the solvent/catalyst metals (Fig. 18). It has been suggested [73] that the addition of copper to the reaction volume has the effect of suppressing the formation of titanium carbide. Copper is thought to form a cation in the molten solvent/catalyst bath which can decompose the carbide. 1.2.5.2.2 Point Defects Nitrogen and boron are ubiquitous atomic-scale impurities in synthetic diamonds. The free energy of formation of nitrides and borides in the solvent/catalyst melt will determine the extent to which these impurities are

502


'

Crystals containing inclusions

Seed crystal (100) face area (mmz)

Figure 16. Dependence of critical growth rate of diamond crystals on { 100) faces of seed crystals, on varying seed size [70].

available for incorporation into ie growing crystal. As mentioned in Section 1.2.5.1.1, unless nitrogen-getters are used during synthesis, nitrogen is responsible for the highest concentration of impurity-related point defects in synthetic diamond. This impurity is incorporated principally as isolated substitutional atoms which give rise to a strong optical absorption below 500 nm, leading to a yellow or yellow-brown color. Such diamonds are classified as

Figure 17. Cross section of a cobalt-titanium grown crystal with inclusions resulting from hoppered growth.

1.2 The Crystullizution

of

Diamond

503

Figure 18. Micrograph of cross-section of a carbon, solvent/catalyst mixture Co-Fe-Ti showing large Tic crystallites and some Tic particles that have been trapped at the diamond-solvent/catalyst interface.

type Ib. Synthetic diamonds tend to take up impurities at different concentrations for each of the types of growth sector (regions of the crystal with a common crystallographic growth plane). Under standard synthesis conditions, nitrogen concentrations are highest in { 11 1) growth sectors, typically 100-300 p.p.m. for iron-cobalt grown crystals [62]. Generally, the average nitrogen concentration in {loo} sectors is about half that for { 11 l} sectors [73, 741. However, at low growth temperatures, cube growth predominates and the nitrogen concentration in {loo} sectors exceeds that in { 11 l} sectors [75]. The reduced solubility of the solvent/catalyst for nitrogen at low temperatures has led to nitrogen concentrations up to about 1000p.p.m. for the ironxobalt system. Nitrogen concentrations are substantially less in the minor growth sectors ({I lo}, (1 13) and occasionally { 115)) compared with { 11 l} and { IOO}, with the { 110) sectors containing about 1 p.p.m. If nitrogen getters are used to reduce the amount of nitrogen available to the growing diamond, as described in Section 1.2.5.I .2, the residual nitrogen concentration tends to be approximately proportionally similar to the ungettered case. Blue, semiconducting type IIb diamonds are synthesized by adding both boron, and a nitrogen getter to the reaction volume. The total boron acceptor concentration is highest for { 11 l} sectors, somewhat less for { 1 lo} sectors and substantially less for the remaining types of growth sector [74-761. The presence of uncompensated boron acceptors gives rise to a blue coloration. The strength of the blue color, or indeed any residual yellow color, depends upon the difference between the boron acceptor and nitrogen donor concentrations for any growth sector. When nickel or cobalt is used as the solvent/catalyst, these elements are incorporated in the diamond lattice as optically active point-defects, but only in (111) growth sectors [77, 781. Frequently, when large, high-quality synthetic diamonds are discussed, the question of their potential use as gemstones is raised. At the time of writing, none of the major manufacturers of synthetic diamonds markets material for gem use. How-

504


ever, small numbers of cut and polished synthetic diamonds, mostly produced in the former Soviet Union using the so-called BARS equipment, have appeared in the gem trade in recent years. The majority have been yellow in color but a few near-colorless synthetic diamonds have also appeared. As uncut crystals, it is easy to distinguish synthetic diamonds from natural diamonds by their distinctive morphology and the presence of a seed. But cut and polished synthetic diamonds are also readily identifiable as such by standard gemmological techniques such as microscopy of inclusions and of features on polished faces (due to sectordependent differential hardness) and zonation of color or fluorescence (due to sector-dependent differential uptake of optically active impurities) [79-841. Nearcolorless and boron-doped, blue in color, synthetic diamonds tend to exhibit strong, long-lived phosphorescence. To enable large numbers of diamonds to be screened rapidly for the presence of synthetic diamonds, instruments have been developed. The first of these instruments detects an optical absorption line at 415 nm present in the majority of natural diamonds. Those diamonds not exhibiting this feature may be examined using a second instrument which, by exciting near-surface fluorescence with very short wavelength ultraviolet light, produces a fluorescence pattern from which the growth structure of the diamond may be inferred. These patterns are very different for synthetic as compared to natural diamond [85].

1.2.6 Novel Diamond Synthesis Routes J. 0. Hansen 1.2.6.1 Introduction: Thermodynamically Stable and Metasfable Processes The two major innovations in diamond technology in this century must be the reproducible high pressure process with metal solvent/catalyst, and the low pressure sustained growth from methane-hydrogen plasma reviewed by Kamo et al. [86] which we call CVD. There may be some argument that the work of Angus et al. [87] or the patents of Eversole [88] predate and anticipate metastable diamond synthesis by CVD, but the authors did not achieve sustained growth. They did demonstrate that the decomposition of methane or carbon monoxide can produce active carbon atoms or hydrocarbon fragments, which in a suitable temperature range and in the presence of diamond surfaces to act as nuclei, will permit overgrowth of a new diamond phase. The appearance of graphite nuclei leads to a cessation in diamond growth which suggests that the success of the ‘metastable’ growth, without plasma methods, depends upon an energy barrier to nucleation of graphite. In the now mature CVD technology, graphite nucleation is inhibited by the continuous bombardment of the growing surface by atomic hydrogen, which has a greater tendency to etch graphite than diamond [89]. Even so, it appears that certain ‘tricks of the trade’ are required to initiate nucleation of diamond on a nondiamond substrate because of a substantial energetic barrier. The graphitediamond equilibrium line in Fig. 7 is based on a thermodynamic ana-


505

lysis by Berman and Simon [ 141, which has been reviewed recently [90] in the light of new data, particularly the direct electrochemical determination of the free energy difference for the diamond to graphite transformation [91]. From the equilibrium expression of Berman and Simon, Eq. (1) in Section 1.2.2, it may be possible to include additional terms which offer an explanation for some cases of novel non-CVD synthesis, or to expand the existing terms such as AH', the enthalpy change at zero pressure. 1.2.6.2 Increase in Enthalpy: Synthesis from Nongraphitic Carbons

Onodera et ul. [30] and Brannon and McCollum [92] have published evidence of diamond growth from nongraphitic carbons at pressures below the phase-equilibrium line at 2 4 GPa and 1300-1900°C. They suggest as explanation an increased chemical potential (free energy) of the nongraphitic phase with respect to diamond, or the catalytic action of the hydrogen associated with the cokes, and not present in graphite, by analogy with diamond CVD. The second explanation seems unlikely, since hydrogen has been shown to be a deleterious impurity in metal solvent/catalyst synthesis from graphites [93, 941. The enthalpy of nongraphitic carbon (furan coke) has been measured at low temperature [95,96] and found to be higher than both graphite and diamond, by between 0.4 and 2.8 kJ mol-* . This would have the effect of eliminating or reversing the sign of the first term in Eq. (1) and so decreasing the pressure requirement for diamond synthesis, perhaps as far as point B in Fig. 7. As Vereschagin et al. [96] point out, there is an even greater driving force for graphite crystallization. Even in synthesis from synthetic graphites of high graphitization index (dOo2= 0.337 nm), there is a competitive process of recrystallization of graphite as coarse single crystal flakes [56]. See also Fig. 19. The increased enthalpy of the nongraphitic carbon follows from a higher internal energy because of lower total bond energy or equivalently, a higher concentration of defects by comparison with single-crystal graphite. A similar argument has been used by Bar-Yam and Moustakas [97] for the case of CVD synthesis. High vacancy concentrations in graphite and diamond can cause diamond to become the stable phase because the formation energy of vacancies in diamond is lower than that in graphite. The possible success of a synthesis route using nongraphitic carbon would seem to depend upon the suppression of graphite nucleation and purity factors, since growth of high quality diamond requires low levels of impurities. This is more difficult to achieve in a carbon which has not been graphitized. 1.2.6.3 Interfacial Energy: Growth in the Colloid Size Range

Ideal diamond has an exceptionally high surface energy because of the high density of strong bonds which must be destroyed in order to generate unit area of surface. It has been estimated at 5000-9000 mJ mP2 depending upon the crystal face [98]. Graphite has a lower average surface energy but there is extreme anisotropy, with the basal plane having an energy of 150mJm-* and the prism planes over 4000mJm-* [99]. This explains the tendency for graphite to crystallise as large thin flakes, so reducing the contribution of the edge surfaces to the total surface

506

I Diamond Muteriuls and their Applications

Figure 19. Micrograph of a section of a diamond synthesis capsule in polarized light at 200x magnification. Recrystallization of coarse graphite flakes of high crystallinity is a process which competes with diamond growth even in the diamond stable region. The tabular shape suggests that surface energy is a significant driving force for recrystallisation even in this size range.

(interfacial) energy. In practice, these solid-vacuum interfacial energies are not likely to be seen, since the diamond surface and the graphite prism-face surfaces will reconstruct or react with adatoms such as hydrogen and oxygen to reduce their surface energy. Under atmospheric conditions, diamond is terminated by a range of hydride and oxide groups and is a low energy surface [loo]. During high pressure solvent/catalyst synthesis, the interfacial energies of metal melt-diamond and metal melt-graphite have values lying between the metal surface tension, 1800mJ mP2,and the solid-vacuum surface energy. Some values have been estimated from contact angle data [loll although the complexities of contact angle work in the diamond stable region are formidable. By including a surface energy (interfacial energy) term in the equilibrium expression it is possible to make some interesting predictions.

1

AG', = A H ! - T L I S ; , ~ A VTd P

+ A(yA) = 0

where y is the interfacial (surface) energy of any phase and A is its molar area, which is not an intensive variable like pressure or temperature but varies with both the size of the particles and the density of the phase, diamond having a lower molar area for the same particle size than graphite. For instance, using this approach, Nuth [ 1021 has predicted that diamonds (in vacuo) are more stable than graphite in the size range 1-5nm, a result which was supported by Badziag et al. [lo31 using a different approach. There is experimental support for this in the finding that nanometre diamonds are abundant in primitive meteorites and possibly in the interstellar dust clouds [ 1041. Similarly, growth of graphite in the diamond-stable region during high pressure, high temperature synthesis, and the corollary, an overpressure for nucleation of


507

diamond, are explainable using the additional interfacial energy term without the need for kinetic arguments [105, 1061. Once the growing particles are in the size range of microns, the molar surface area becomes too small for this term to have any effect and the classical phase equilibrium line is applicable. There is another way of looking at the same effect, since a surface energy per unit area is equivalent to a line force per unit length, or surface tension, hence there is an excess pressure in a spherical droplet given by the Laplace equation:

A P = -2Y Y

(3)

where y is the surface energy of any phase and r is the radius of curvature. For diamond with sigma at 5 J mP2and Y at 1 nm, the excess pressure in a nanometre sized sphere is 5 GPa, which would be sufficient to stabilize the diamond phase to at least 1600K. This could explain processes such as laser pyrolysis of hydrocarbons [ 1071, where small quantities of nanometre sized diamond have been detected in the carbonaceous products. A recent result suggests that it is even possible to generate and stabilize a diamond core in the centre of a graphitic ‘onion’ a giant fullerene, by electron irradiation and annealing of the outer graphite layers. In this process, ‘the sputtering-induced loss of atoms and the closure of the shells around vacancies lead to extreme surface tension, which is finally responsible for the spherical shape of the onions’ [108]. This publication is a good example of the development of analytical techniques such as high resolution electron microscopy lattice imaging [1081 and micrometrearea Raman spectroscopy (microRaman) [ 1091 which permit the researcher to determine that diamond is present in crystallites of submicrometre, perhaps even nanometre size. Early scientists like Moissan [I 101 and Hannay [I 113 performed experiments in the pressure and temperature ranges now being cited as feasible for hydrothermal synthesis of diamond [38, 112, 1 131. Had they been equipped to use the analytical techniques now available, they may have been proven correct in their claims, but in a particle size regime much finer than they were able to inspect.

1.2.6.4 Non-equilibrium Thermodynamics: Superequilibrium Atomic Hydrogen The process of CVD of diamond may be explained by kinetic arguments [89], or alternatively by the nonequilibrium thermodynamic coupling theorem [ 1 141. In the latter case, it is the presence of a superequilibrium concentration of atomic hydrogen and the coupling of the process of recombination of hydrogen atoms with the processes of diamond growth which allow the diamond deposition reaction to occur in a direction contrary to that prescribed by conventional thermodynamics. Diamond synthesis processes which are characterized by continuous generation of atomic hydrogen or halogens would then fall into the class of nonequilibrium processes. The generation of atomic hydrogen could be by means of microwave or discharge plasma, flame or even a hot filament. This topic is covered more fully in Chapter 4 of Part I1 of this book (C.-P. Klages), but some special cases will be listed here.

508


Table 2. Classification of low-pressure synthesis methods Nominally metastable thermodynamic explanation is possible

Truly metastable: explanations in terms of nonequilibrium thermodynamics, atomic H

Undetermined: perhaps controversial

HPHT: nongraphitic carbon [30], [92] Hydrothermal synthesis nonflow [113, 381 Laser in absence of hydrogen [105, 1151

LPSSS (low pressure solid state source) as by Roy et al. [ 1 16, 1 171 Hydrothermal flow synthesis with flame [116, 1171 Laser in presence of hydrogen [ 1181

Triple laser process using CO, (QQC process) [ 131 PVD using ion beams reviewed by Prins [123] Precipitation from nickel-sodium hydroxide melts [ 1241

Diamonds in interstellar space [102-1041 Electron irradiation of carbon ‘onions’; nanometre scale HPHT [ 1081

Fluidized bed deposition on particles from acetylene flame [ 1191 Discharge electrolysis from water:organics mixtures [120-1221

1.2.6.5 Classification of Low-pressure Methods Using this criterion, it is possible to separate those methods which are nominally metastable, but for which there is an explanation in terms of the additional free energy components discussed in Sections 1.2.6.1 to 1.2.6.3, from those which are truly metastable. There is a third category, undetermined, where the authors have insufficient knowledge of the process to make a judgement. Perhaps these are truly different in their mechanism. 1.2.6.6 Extension of High-pressure Methods In Section 1.2.3 we reviewed the design of high pressure systems for diamond synthesis. The major development in recent decades has been an almost thousand-fold increase in working volumes. The pressures achieved using cemented carbide technology have, however, not increased significantly. The advent of the diamond-anvil-cell DAC [1251has allowed researchers to attain pressures of thousands of kilobars, whilst simultaneously collecting spectroscopic data on phase changes. This is possible only because of the transparency of the diamond anvils. Most research on ultrahigh pressure has been done with DACs. There might be two reasons for industrial interest in pressures of 10GPa. The first is the synthesis of diamond from nonmetallic solvents [24, 1261 such as carbonates at pressures of 7.7GPa and 2000°C. This might offer different and novel types of diamond to those presently commercially made. The second is the possibility of direct conversion, solvent free, of graphite to diamond. Table 3 lists reported conditions for the direct conversion. The direct conversion is dependent upon the crystallinity of the starting carbon [127], and the product, when analysed by XRD, appears to be hexagonal diamond. Figure 20 shows the martensitic transformation, by compression in the c-direction and in-plane buckling, of two crystalline forms of graphite. Hexagonal graphite,


509

Table 3. Direct conversion of graphite to diamond Authors

Apparatus

Bundy and Kaspers [I271 Modified belt Boehler [I281 DAC DAC Xu and Huang [ 1291

Date

Pressure and Temp

Comments

1967 1994 1994

13 GPa and1000"C 10 GPa and 1700°C 6GPa and 1700°C

Hexagonal diamond Raman at 1333 cm-'

Figure 20. Computer-generated perspectives of the martensitic transformation processes, hexagonal graphite to hexagonal diamond, and rhombohedra1 graphite to cubic diamond. The structure on the right is generated by compacting the structure on the left in the vertical direction and puckering those atoms that find a neighbor in an adjacent plane so as to encourage the formation of the new sp3 bonds in the vertical direction [132]. Software designed by Pawel Wzietek, Massy, France.

5 10


which has layers repeating in the order ABAB . . ., transforms to hexagonal diamond, or lonsdaleite. This is a metastable form of diamond and is never seen in diamond synthesized from a solvent/catalyst. Higher pressures and temperatures are required for the conversion of hexagonal graphite to cubic diamond, above line C, in Fig. 7, which suggests that the mechanism requires melting of carbon [130]. Rhombohedral graphite has a layer sequence ABCABC.. . (Fig. 20) and can convert to cubic diamond by a martensitic process of compression and buckling. Rhombohedral graphite is metastable towards hexagonal graphite and rare except in some natural crystals. Shear forces in high pressure or shock synthesis processes could cause rapid conversion of small domains of hexagonal graphite to the rhombohedra1 form, after which the direct conversion to cubic diamond may proceed. It is unlikely that this could happen over domains much larger than the 100 nm crystallite size seen currently in shock-synthesized diamond [13I].

1.2.6.7 Summary and Conclusions There have been many innovations leading to growth of nanometre sized crystals of diamond. A few processes, like the LPSSS and Me-C-H [117] appear to offer the possibility of combining the advantages of low-pressures with the commercial need for a volume-based rather than a surface-based process, and of growing crystals of micron rather than nanometre size. Direct conversion of graphite to diamond at macroscopic sizes will depend upon the synthesis and stabilisation of suitable graphite or mixed sp2/sp3precursors.

1.2.7 Cubic Boron Nitride Crystallization G. J. Davies Cubic boron nitride (cBN) is presently the second hardest material known to man after diamond. The compound boron nitride is unknown in nature. The combination of atoms of boron and nitrogen in stoichiometric arrangements is such that B-N is isoelectronic with C-C. It is not surprising that, by analogy with carbon structures, similar crystalline structures of boron nitride occur. By analogy with graphite, hexagonal boron nitride is a soft, slippery material with an anisotropic, hexagonal structure made up of sheets containing interlinked six-membered rings (3 x B-N). There are some differences however, notably, the n-orbital overlap of graphite is absent, so hBN is an electrical insulator and is white in color. For kinetic reasons, all conventional chemical synthetic routes for making boron nitride result in the hexagonal structure. Again, by analogy with carbon, boron nitride can take up the diamond structure with alternating atoms being boron and nitrogen. This is the common zinc blende or sphalerite structure. This phase of boron nitride has a hardness about half that of diamond; and a bulk modulus of about 367 GPa compared to the 435 GPa of diamond. The value of cBN as an abrasive lies in its much higher oxidation stability compared to that of diamond (1200°C compared to 600°C) and in its reduced chemical


5 11

Figure 21. Phase diagram of boron nitride showing the latest stability fields suggested by Solozhenko [ 1341 compared to the earlier Bundy-Wentorf diagram [ 1381.

interaction with ferrous materials when compared to diamond. It is, therefore, the abrasive of first choice for the grinding of hard, ferrous materials. When the formation of cBN was first reported [133], the pressure and temperature phase diagram was believed to be very similar in form and stability field positions to that of carbon, with the equilibrium line between the hexagonal and cubic phases intercepting the pressure axis when temperature is at absolute zero [134]. The most recent phase diagram for boron nitride is that of Solozhenko, shown in Fig. 21 [135]. This diagram was initially calculated and subsequently partially experimentally substantiated. The hexagonal-cubic-liquid triple point is at about 7 GPa and 3800 K. The hexagonal-cubic equilibrium line intercepts the temperature axis at just above 1500K. However, the exact temperature of the intercept is not well-established [136], although it is now generally agreed that cBN is the stable form of boron nitride at room temperature and pressure. This is where boron nitride differs significantly from diamond. It does not transform to the hexagonal structure until temperatures of at least 1200°C are exceeded. cBN is synthesized using the same general technology as that used for commercial diamond synthesis. A mixture of hBN as source and various catalyst/solvents are subjected to pressures in the range 4.5-5.5 GPa and temperatures between about 1500 K and 2000 K. The elements initially investigated and demonstrated to function as solvent/catalysts were the alkali and alkali earth metals, as well as antimony, tin, and lead [137]. Subsequently, the nitrides of lithium, calcium. and magnesium were shown to be more effective. Single crystals of cBN, in the size ranges appropriate to grinding applications, both black and amber in color, are now readily synthesized by several companies on a commercial basis. The recent developments in the boron nitride phase diagram, particularly with respect to the room temperature and pressure thermodynamic stability of cBN, have stimulated work aimed at producing cBN over a wider range of conditions [136]. In conjunction with this, it is now known that a wide range of nitrogenous

5 12


compounds are capable of aiding the transformation of the hexagonal to the cubic form [138].

1.3 Polycrystalline Diamond and Cubic Boron Nitride S . Ozbayraktar

1.3.1 Natural Polycrystalline Diamond Although the first natural polycrystalline diamond was discovered in the 19th century, it was only investigated on a scientific basis during the 1970s, and was classified into two broad types: Carbonado and Ballas [139]. Carbonado is ‘a porous, randomly polycrystalline diamond aggregate’ [ 1401. Ballas on the other hand is defined as ‘polycrystalline diamond of oriented globular growth’. Its crystallites have (110) directions oriented radially. Experiments so far have clearly indicated that these natural forms of sintered diamond are truly polycrystalline ceramics, exhibiting transgranular fracture as a result of diamond to diamond bonding.

1.3.2 Synthetic Polycrystalline Diamond In 1958, Hall [141] discussed the desirability of preparing a ‘cemented diamond composition analogous to WC’ and hinted that experiments to produce polycrystalline diamonds were underway. But it was not until 1970, when he reported details of his procedures [1421, that he established experimentally practical pressure and temperature fields where pure diamond powder can be sintered within times ranging from several days down to about one second. He mentions hard refractory materials like borides, carbides, nitrides and oxides as suitable binders. In December of the same year, Stromberg and Stephens [143] published a paper titled ‘Sintering of Diamond at 1800-1900”C and 60-65 kBar’. Their manuscript was originally dated 17 December 1969. In their work, natural diamond powders of sizes 0.1-10pm were carefully cleaned in several solvents and then heated at 500°C in a vacuum of torr in order to remove adsorbed gases. The cleaned particles were then loaded, under argon, into a similarly cleaned and outgassed tantalum crucible which was sealed by electron beam welding. The capsule was heated under pressure and temperature for 1 h. They used indium as a ‘cushion’ to protect the diamond specimen on unloading. Small amounts (less than lwt%) of boron, silicon, or beryllium were found to aid the sintering process. Most of the sintered specimens gave hardness values of over 7000 kg mm-*, with a maximum of 8800 kg mm-’.

I .3 Polycrystalline Diamond and Cubic Boron Nitride

5 13

In 1971, Katzman and Libby, from the University of California at Los Angeles [144], described the formation of sintered diamond compacts in the presence of 20 vol% of cobalt binder. The sintering was carried out for about 20 min at 6.2 GPa in a piston and cylinder type high pressure-high temperature apparatus, and at a temperature (159OOC) which was above the cobalt-diamond eutectic, 1570°C at 62 kBar. They claimed their results suggested that the cobalt cleans the diamond surfaces of any adsorbed gases that might prohibit or retard grain growth or fusion, thereby making unnecessary the surface cleaning and degassing procedures described by Stromberg and Stephens. They also claim that the optimum cobalt content was 20% by volume. An attempt to substitute cobalt with nickel under identical sintering conditions also led to a somewhat softer material. Compacts made with diamond grit sizes of 1-5 pm and 0-2 pm were harder than those made with grit sizes of 10-20 pm. Wentorf and Rocco in 1973 [145] described in situ diamond sintering on top of a layer of cemented tungsten carbide (WC). The idea here was that any solvent/catalyst metal (cobalt) which was present in the metal carbide would act as a source for liquid phase sintering. The sintering method involves placing in a cylindrical mold, a 0.51 mm layer of diamond powder on top of a thicker layer (lomm) of WC-lOwt% cobalt powder. The mold is then subjected to high pressures (>5 GPa) and temperatures (1400-1600°C) for at least 1Omin to 1 h. Under these conditions, the cobalt melts and liquid phase sintering of both the carbide and the diamond occur. Veraschagin et al. [146], in 1975 showed that, if the metallic binder was initially in the form of a disc (rather than a powder), better packing densities were achieved and the molten binder would thus be found mainly in the voids rather than in between diamond grains, thus resulting in strong and uniform diamond-to-diamond bonding throughout the compact.

1.3.3 Mechanisms involved in Polycrystalline Diamond Manufacturing Process The pcD manufacturing process can be roughly divided into three stages: cold compaction, hot compaction, and liquid phase sintering of the diamond compact.

1.3.3.1 Cold Compaction Kolomiitsev [147] and Uehara [I481 studied the compaction behavior of diamond compacts under high pressures. Kolomiitsev showed that there are basically three processes taking place during cold compaction, see Fig. 22: particle rearrangement, crushing of diamond particles, and filling of voids by crushed particles. At a given pressure, coarser particles crush more than finer particles. Their average particle size changes drastically and the particle size distribution becomes truly bimodal. At a given pressure, porosity of coarser powders is lower than that of finer powders. The change in porosity with pressure is bigger at lower pressures than at higher pressures for a given starting particle size. Pore size of coarser powders is

5 14


Figure 22. Schematic description of processes taking place during cold compaction.

larger at a given pressure. But the difference between coarse and fine grains gets smaller as pressure is increased. Crushing of diamond particles during this stage dramatically shifts the particle size distribution. The powder, which has a lower number of sharp edges per particle, is more stable to compression forces than the one with many spallings, sharp edges, etc. Since their weakest planes are the planes where they contain inclusions; the density, size and shape of the inclusions may be a critical factor in the crushing of diamond particles during the cold compaction stage.

1.3.3.2 Hot Compaction We describe hot compaction as the stage where temperature is applied together with pressure but the temperatures are not high enough to start infiltration/liquid phase sintering processes. During hot compaction the following processes are believed to be taking place, as shown in Fig. 23: graphitization of the diamond surfaces facing the voids, plastic deformation of diamond grains, and densification of the compact (shrinkage of pores).

1.3 Polycrystalline Diamond and Cubic Boron Nitride

5 15

pressure temper fure

4

Figure 23. Deformation of diamond grains under high pressure and temperature

Once the heat is turned on at high pressures, the densification of the diamond compact proceeds mainly by further crushing and rearrangement of the crushed particles up to 700°C [149]. After 700°C, densification proceeds with plastic deformation [150-152]. The extent of plastic deformation is a function of temperature and pressure. At higher temperatures, the diamond grains become more round, their sharp edges disappear, they start to deform in the zones of contact with each other. Pores get smaller, even diminish at a rate initially determined by their size at the end of the cold compaction stage. Pore size reduction rate is faster for coarser pores than for smaller pores [149]. But the time required for a pore to disappear is constant at a given pressure and temperature to a first approximation, regardless of size [153, 1541. Also at higher temperatures, graphitization of the diamond surfaces, which are not in contact with another diamond surface, takes place. The amount of graphitization is a function of the temperature at a given pressure [150]. The degree of graphitization in diamond compacts can be related to the specific resistivity of the compacts since graphite is a relatively conductive material compared to diamond which is an insulator. That is, the more graphitization takes place, the more conductive the material becomes. At even higher temperatures, bulk graphitization of diamond grains starts to take place and the hardness and wear properties of pcD produced (binderless) start to

5 16


degrade [ 1501. At a given temperature, if the pressure is increased, graphitization seems to decrease and the properties improve. One can conclude from these observations that, at a given plastic deformation, the degree of graphitization determines the change in hardness and wear resistance. But when both plastic deformation and graphitization are taking place, properties are more sensitive to plastic deformation until bulk graphitization starts taking place. Finer particles tend to graphitize more than coarser particles at a given pressure and temperature. This is most probably due to their larger surface area in contact with the pores. 1.3.3.3. Surface Chemistry Effects

Surface chemistry of the starting diamond particles has an effect on the graphitization kinetics of the free surfaces of diamond particles. Surface adsorbed oxygen gas reacts with diamond and speeds up the process of graphitization on these surfaces [149]. This may be due to the fact that the activation energy for the process of oxidation is lower than the activation energy for the process of graphitization without participation of the gas phase. Moreover, such graphitization may begin at lower temperatures. 1.3.3.4 Liquid Phase Sintering

There are a number of binders used for pcD sintering. The most commonly used ones are iron group materials which are known to be good solvent/catalysts for diamond synthesis, namely cobalt, iron, nickel, and manganese or various combinations of these. Utilization of binder materials for the sintering of diamond compacts has threefold benefits: decreasing of sintering temperatures and pressure, cleaning diamond particle surfaces of graphite, and electron discharge cuttability for tool making. There are various methods of adding binder material into diamond compact: mixing it with the diamond powder, coating diamond particles with it, infiltrating from a disc of binder metal, and infiltrating from a substrate containing binder metal. We will concentrate here on the mechanism of infiltration from a substrate because it is the most commonly employed method in the industry. There are basically four stages in the infiltration of diamond compacts [ 155, 1561. During first stage, temperature is just sufficient to cause melting of Co-W-C at the interface due to excessive amounts of carbon which cause the eutectic temperature to drop. The amount that is melted during this stage immediately infiltrates into diamond due to the very high pressure gradient between the porous body and the solid interface. However, this stage is very short and it is rapidly followed by the second stage. The second stage corresponds to melting of the binder in the bulk of the substrate. As the temperature increases, it reaches the eutectic temperature of binder in the substrate and binder becomes liquid and starts infiltrating into diamond compact again under the same pressure gradient which is roughly equal to the capsule internal pressure.

1.3 Polycrystalline Diamond and Cubic Boron Nitride

5 17

As the pores in the diamond layer become filled, the magnitude of the pressure gradient is decreased. Accordingly, the rate of mass transfer is decreased, although its direction is maintained. As this pressure gradient decreases, the effect of the pressure gradient in the reaction volume of the capsule, which was found to be about 1 GPa in the axial direction by Russian researchers in their own capsule [ 1571, starts dominating the infiltration gradient. Whatever the direction of the pressure gradient is, this stage of mass transfer is completed when there is no pressure gradient at the interface between diamond layer and the substrate and when the pores are completely filled. Thus, the rate of mass transfer approximates zero and the binder content in the diamond layer becomes a maximum. The graphitization of the diamond surface, described in Section 1.3.3.2, is important in terms of infiltration of the solvent/catalyst binder. Metals such as cobalt dissolve graphite more readily than diamond at the sintering pressures (Fig. 9b). Hence, the more graphitization that has taken place, the more cobalt is expected to be in the final material. A characteristic feature of the fourth mass transfer stage is a change in its direction: it has been observed [ 155-1 571 that the quantity of liquid phase in the diamond layer is decreased and increased in the substrate. The beginning of mass transfer indicates that a pressure gradient arises in the system, directed from the diamond layer to the substrate. The origin of this phenomenon was explained by Shulzhenko et al. [ 155, 1561 in terms of temperature gradients and their changes resulting from the surface tension of the molten binder-diamond interface. Also, this effect can be explained by the fact that, as the sintering of diamond compact proceeds, the area of the contacts between the diamond grains increases. This accelerates shrinkage and the densification process. This in turn according to Shulzhenko et al. squeezes the excess cobalt, which cannot be accommodated in the ever shrinking pores and cavities in between diamond grains, out of the diamond layer. One destination of this stage of mass transfer is the diamond/substrate interface. As a result of the change in direction of binder mass transfer in the fourth stage, the binder content of the diamond layer is decreased by an amount of the order of couple of percent [155, 1581. The binder content of the diamond layer can be controlled by lowering the sintering temperature and extending the sintering time. Almost complete exclusion of cobalt from large areas where two grains have grown together is observed by TEM studies on polycrystalline diamonds sintered with cobalt binder [159]. No thin, residual layer of cobalt was detected at the grain boundaries. The thickness of regrown diamonds is usually much lower than the original diamond grains it is precipitated onto, and it joins with the regrown diamond precipitated onto a neighboring diamond grain forming a high angle grain boundary. So it can be said that only the surfaces of the diamonds in contact with a pore just before infiltration takes place are potential candidates for reprecipitation of new dislocation free diamond. The areas next to the diamond to diamond contact points or surfaces are the preferred locations for diamond reprecipitation due to lesser surface energy requirements.

5 18


1.3.3.5 Conclusions During cold compaction of diamond powders, coarser powders densify more and have larger size pores than finer powders at a given pressure. During hot compaction, the densification of the diamond compact proceeds mainly by crushing and rearrangement of the crushed particles up to 700°C. Above that temperature, the densification proceeds with plastic deformation. The extent of plastic deformation is a function of temperature and pressure. Pore size reduction rate is faster for coarser pores than for smaller pores. The degree of plastic deformation and of graphitization taking place during hot compaction seems to influence the wear resistance and hardness of the sintered diamond compacts. When both plastic deformation and graphitization are taking place, properties are more sensitive to plastic deformation until bulk graphitization of diamond grains starts. lnfiltration rate of the metallic binder from the substrate is influenced by pressure gradients created by the applied external load, pressure gradients inside the capsule and the pore size of the diamond compact at the onset of infiltration, temperature, and the substrate binder mean free path and chemistry.

1.3.4 Polycrystalline Cubic Boron Nitride As referred to in Section 1.2.7, cubic boron nitride (cBN) was first synthesized in 1957 [133]. Polycrystalline cubic boron nitride is preferable over single crystal cBN because it can be manufactured in bigger sizes and has higher fracture toughness. However, like its counterpart pcD, it can only be sintered at very high pressures and temperatures with the aid of binders because of its strong covalent bonds. pcBN, having a hardness second only to pcD in terms of polycrystalline materials, and being relatively less reactive with ferrous metals, is the cutting tool material of choice for a wide range of applications such as finish and interrupted cutting of hardened steel, gray, and ductile cast iron machining and finds a wide range of applications in the production of automotive parts. These applications will be discussed further in Section 1S.4. Commercially, pcBN tool materials are produced using sintering additives and/or binding materials at pressures of 4-6 GPa and at temperatures of 1200-1500°C. The sintered pcBN has a higher thermal stability than pcD. The first polycrystalline cBN material was sintered by Wentorf and Rocco [I 601 in 1971 on a cemented carbide substrate using alloys of nickel, cobalt and iron with aluminum as binders. cBN to cBN bonding is believed to be achieved by a liquid phase sintering process involving these binder alloys. In February, 1980, Sumitomo from Japan filed the patent ‘Sintered compact for a machining tool and a method of producing the compact’ [161]. This patent basically covers any compact with 1 W O vol% cBN and a balance of binder material that can comprise any carbides, nitrides, borides, or silicides of metals of groups IVa, Va, or VIa. Specifically mentioned are titanium, zirconium, hafnium, vanadium, niobium,

5 19

I .3 Polycrystalline Diamond and Cubic Boron Nitride

Table 4. Properties of the pcBN products and other cutting tool materials.

Density [g ~ m - ~ ] Compressive strength [GPa] Fracture toughness [MPa.m'/2] Knoop hardness [GPa] Young modulus [GPa] Modulus of rigidity [GPa] Bulk modulus [GPa] Poisson's ratio Thermal expansion coefficient [10-6K-1] Thermal conductivity [W.m-IK-']

DBC 50*

Amborite*

SYNDITE WC 010

4.297 3.55 3.64 27.5 607 258 315 0.178 4.7

3.41 2.73 6.36 31.5 653 288 297 0.13 4.9

4.127 4.74 8.39 50 993 453 412 0.102 3.8

44

100

120

A1203 TIC

Sialon

14.7 4.50 10.80 13.0 620 258 375 0.22 5

4.28 4.50 3.31 17.0 370 160 232 0.22 7.8

3.20 3.5 5 13 300 117 227 0.28 3.2

100

16.7

2&25

+

tantalum, chromium, molybdenum, and tungsten (which are, in fact, the elements of groups IVb, Vb and Vlb). In Sumitomo's patent, high heat conductivity together with high hardness are mentioned as paramount properties when selecting a suitable binder material. A rapid rise and fall in temperature in the tool during interrupted cutting often leads to the formation of cracks. Therefore, a binder material that couples high thermal stability and hardness with high thermal conductivity at high temperatures is suitable for this kind of application. The manufacturing process (especially the cold and hot compaction stages) of pcBN is very similar to that of pcD, summarized in Section 1.3.3. However, the types of binders used and liquid phase sintering mechanisms can be quite different. In pcBN production, 'reactive sintering' plays a major role in terms of driving the densification process, whereas in pcD, dissolution and precipitation and/or adhesion/coalescence are the main driving mechanisms. A typical example for pcBN would be the following reaction [162]. 3A1+ 2BN 4 A1B2

+ 2A1N

(4) where the reaction products A1B2 and AlN act as binders [ 1631. However, the chemical reaction to achieve these products is the driving force for sintering this particular pcBN product. The properties of the pcBN products are not only determined by the final microstructure, but also by the phases formed during sintering. The pcBN properties are listed in Table 4 and the properties of the main phases are given in Table 5. A direct conversion process can produce translucent pcBN which has almost theoretical density and very high thermal conductivities. In 1972 and later in 1974, Wakatsuki et al. [164, 1651 sintered a polycrystalline cBN directly from hBN at pressures as low as 55kBar and temperatures of 110&1400°C. Corrigan [1661 observed that thermal conductivity of directly converted compacts increases with increasing grain size. Fukunaga and Akaishi [167] used a small amount of Mg3BN3 to promote conversion from hBN to pcBN at pressures of 5-6GPa and at 1500°C.

Crystal sructure Density [g ~ m - ~ ] Melting point ["C] Knoop hardness [GPa] Young modulus [GPa] Thermal expansion coefficient [10-6K-'] Thermal conductivity [W.m-'K-']

cubic 3.48 2700 41 700-800 3.2 150-700

cBN hexagonal 3.26 2300 12 318 3.9 200

AIN hexagonal 4.52 3197 26.5 434540 8.1 80

TiB2

Table 5. Properties of the main phases found in pcBN products after sintering.

cubic 5.22-5.44 2930 19 370 8 25

TIN cubic 4.92 3065 28-35 450 8.6 3343

TIC

hexagonal 3.96 2050 23 421 8.1 14

A1203

~

57-104 1141 1.54.8 500-2000

cubic 3.51

Diamond (for reference)

2

s.

%2

L L

$

%

a

D

c

6'

a

$

6'

b

0

ul N

1.4 New Ultrahard Materials

521

1.4 New Ultrahard Materials I. Sigalas

1.4.1 Introduction Diamond’s combination of properties make it a unique material. Although hardness is its primary characteristic, thermal conductivity, compressive strength, refractive index, spectral transmittance, and chemical stability are either the highest or among the highest found in nature. This combination of properties can be traced to the same structural characteristics of diamond that give rise to its high hardness. It is therefore reasonable to expect that other ultrahard materials would also exhibit such a suite of properties. This would make them also desirable for a number of industrial applications. New ultrahard materials might possess new attributes other than a higher hardness that would make them more attractive than diamond in some cases. A different chemical composition would give rise to different interaction of cutting tool and workpiece as is the case for cBN, as would a different crystal structure and ultrahard particle morphology. The search for new ultrahard materials is motivated largely by these considerations, but also from purely economic ones. It might be possible to synthesize such materials through routes cheaper than those involved in the case of diamond, thus accessing new applications and new markets. In this paragraph we briefly review the basic ideas behind the search for new ultrahard materials, as well as the latest developments in the search for some of the ‘identified’ cases.

1.4.2 Hardness Hardness is a measure of a material’s ability to resist elastic and plastic deformation. The hardness of non-ideal material is determined by the intrinsic stiffness of the material, as well as by the nature of its defects, be they point defects, dislocations, or macroscopic defects such as microcracks etc. For ideal systems, the hardness of a material will scale with its bulk modulus. Figure 24 shows the Knoop hardness as a function of the bulk modulus for a number of representative materials [168]. Table 6 shows the hardness as well as the bulk modulus of a number of representative materials. The bulk modulus of a solid can be calculated by means of ab initio calculations. A review of this field can be found elsewhere [169]. Such calculations require extensive computer time. Liu and Cohen in 1989 [ 1701 proposed an empirical formula for the bulk modulus of solids with the zinc blende structure: B = (19.71 - 2 . 2 0 h ) p

(5)

where B is the bulk modulus in GPa, d is the bond length in Angstroms, and h is a measure of the ionicity of the compound. For purely covalent compounds (group

522

1 Diamond Materials and their Applications 80oo diamond

7000

6ooo

n

5ooo

N

E E

m ?5

z

I

cubic BN

4000

ZrC

3Mx)

Sic

Y

0

oAL03

2ooo

rn-ZrO, Si

a

1000

0" OMgO

NaCl

%a0

0

a

I

0

50

I

loo

I

150

I

I

I

I

300

350

400

I

250

200

450

6,( G W Figure 24. Knoop hardness (KHN) as a function of bulk modulus (B,) for representative materials.

IV), h = 0 while, for compounds of groups 111-V and 11-VI, h is equal to 1 and 2 respectively. This expression can be modified to allow for nontetrahedrally coordinated compounds as follows [ 171, 1721: B = (Nc/4)(1971 - 220qd-3.5

(6)

Where Nc is the average coordination number. From Eq. (6) it follows that in order for the bulk modulus to be high, the crystal must have: a large coordination number; a high degree of covalency; and short interatomic distances, meaning small atoms. Although a high coordination Table 6. Bulk moduli and microhardness for light covalent ceramics Material

B (GPa)

H (GPa)

H actual GPa [228]

C (diamond) Si3N4 Sic

443 220 21 1

77 35 33

75-100 17 26

1.4 New Ultrahard Materials

523

number may be attainable in metallic materials, where it can be as high as 12, metallic bonds are much weaker than covalent bonds which can allow a maximum coordination number of 4 in naturally occurring materials. Because of these considerations the search for new materials is confined to tetrahedrally coordinated compounds with high degree of covalency and with small atoms. On the basis of this model, it was suggested that, for h = 0.5 and d = 1.47-1.49, a hypothetical tetrahedral compound between carbon and nitrogen would have a bulk modulus of 461-483 GPa, which would exceed that of diamond (443 GPa). However, not all potentially hard or ultrahard materials can be found within the two top periods of the periodic table, nor are they confined to tetrahedral coordination. A number of sp2 coordinated structures have been proposed which, mainly due to their high density, would posses a high bulk modulus and would therefore potentially exhibit high hardness [ 173-1 751. Superdiamond structures have been proposed by Diedrich et al. [176] and Alberts et al. [177], based on polymerized derivatives of tetraethylmethane C(C2H)4and the carbon skeleton of allene CH2=C=CH2 respectively. Leger [ 1681 investigated the feasibility of generating ultrahard materials by producing the high pressure phases of a number of oxides. Such phases would have a higher degree of coordination and bonds with a higher degree of covalency than their ambient pressure allotropes.

1.4.3 C3N4 In 1989, Liu and Cohen reported local density approximation pseudopotential calculations on a hypothetical carbon nitride phase, PC3N4 which suggested that it could be metastable under ambient conditions [170, 1781 and would have a short c-N bond length (0.147 nm), low ionicity ( ~ 7 % and ) a bulk modulus higher than that of diamond. Figure 25 shows the crystal structure of hexagonal PC3N4derived from pSi3N4 [ 1791. Alternative structures have been proposed by Liu and Wentzcovitch [ 1SO], Wentzcovitch and Martins [181] and Wentzcovitch [182], proposing a cubic zinc blende structure with one carbon vacancy per unit cell, and a structure resembling graphitic CN with one carbon vacancy per four nitrogen sites. This study was based on an ab initio molecular dynamics scheme with a variable cell shape algorithm. More recently Teter and Henley [ 1831, using first principles pseudopotential total energy techniques, predicted a cubic form of C3N4with a zero pressure bulk modulus of 495GPa, higher than that of diamond. The same authors predicted that aC3N4and graphite-structure C3N4are energetically more favorable than PC3N4. Figures 26 and 27 show the structures of the proposed cubic C3N4 and aC3N4 compounds respectively. The above theoretical predictions have generated a great deal of experimental activity aimed at synthesising a new ultrahard compound. Several attempts have been made to synthesize thin film C3N4,be it cubic or hexagonal. These include reactive magnetron sputtering [184187], laser ablation [188] ion beam assisted deposition (IBAD) [189], plasma [191, 1921 and plasma-enhanced chemical vapour deposition [190].

524


Figure 25. Hexagonal crystal structure of PC3N4 derived from pSi3N4. Two types of sp2’ bonded N atoms are present. The first type is in a trigonal planar configuration with the trigonal plane perpendicular to the c-axis. The second type forms, together with the carbon atoms, the tunnel at the center [179].

In most efforts, the films produced were either amorphous, or their nitrogen content was less than the target of 57%. D. Li et al. [184] using a magnetron sputtering system produced crystalline carbon nitride/titanium nitride composite coatings with a reported hardness of 55 GPa.

Figure 26. Two adjacent unit cells of cubic C3N4carbon atoms are shown schematically as the black atoms [192].

I .4 New Ultrahard Materials

525

P’C3N4 Figure 27. Predicted crystal structure of aC,N4 [192], compared with that of PC3N4.

Zhang et al. [ 1941 at Harvard were able to synthesize an amorphous CN thin film with composition C2N. This material had exceptional hardness [195-1971. In some cases, tiny crystallises of covalent C-N were observed in the amorphous films [19& 2021. Bhusari et af. in February 1997 [193] were able to synthesise large crystals (750 pm) of Si-containing carbon nitride consisting of a predominantly C-N network, by microwave CVD. More recent efforts involving nitrogen ion beam assisted deposition [203, 206, 2111, hot filament CVD with or without rf plasma [204, 205, 2091, or microwave plasma enhanced CVD [213, 2141, with bias assistance [207, 208, 2121 were able to produce thin films containing C3N4,some with the BC3N4structure and some with the cubic or a-structure. Yan et al. [205] using hot filament CVD were able to synthesize fully crystalline films combining both aC3N4and PC3N4. Xu et al. [210] using a reactive magnetron plasma source claim to have deposited polycrystalline BC3N4films with crystallites as large as 20 pm. Xiao-Ming et al. [211] using argon arc nitrogen-assisted bombardment claim to have deposited completely polycrystalline BC3N4 films. They quote a measured hardness of 52.6 GPa. Similar results were obtained by Wu et al. [212] who used RF plasma enhanced CVD. Finally Chen et al. [213], using microwave plasma enhanced CVD, claim to have grown crystals larger than 10 pm, and propose that the incorporation of silicon from the substrate into the C-N structure promotes BC3N4crystal growth. Efforts to synthesize C3N4at high pressures have been much less extensive than those made to synthesize this material in the thin film form. A number of trials [214] indicate that it may be possible, through the use of pressure to incorporate nitrogen into sp’-bonded carbon nitride.

526


1.4.4 Boron Rich Nitride B3N, B5N and B4N have been synthesized by CVD methods [215-2171. The B4N structure, with a crystal structure analogous to that of rhombohedra1 B4C, may exhibit high hardness. No reported hardness values are available at present.

1.4.5 Boron Carbonitrides In addition to binary compounds made with elements from the 2d period, ternary compounds may also exhibit higher thermal stability that diamond [218]. Such compounds can be considered as solid solutions of carbon in boron nitride. Efforts to synthesize such compounds have been made both at low and at high pressures. Ternary BC,N, films were first produced in 1972 by Badzian et al., using a CVD process [219]. Montasser et al. [220] in 1984 produced films of hardness in the range 4-33GPa. Loeffler et al. [221] studied the influence of substrate temperature on the crystallinity of BC,N, films deposited by PA CVD. Hegermann et al. [222] investigated the influence of carrier gases nitro en, argon and helium and the influence on the applied power density (about 14Wcm-') of the film deposited by the PA CVD process.

1.4.6 Boron Suboxides Boron suboxides with hardness comparable to that of diamond were first reported by A. R. Badzian in 1988 [223]. These suboxides were produced by reacting B2O3 with boron at 1600-2000°C and 7 kbar. Subsequent melting under argon gave a sintered compound with an oxygen content of 4 5 % and a microhardness of 60 GPa. The introduction of oxygen reduces the electron deficiency of the P-rhombohedra1 structure of elemental boron. Filling these voids with atoms results in a cross linking of the boron icosahedra and can contribute to a significant increase of the rigidity of the structure. Recently McMillan et al. [224] reported the synthesis of B60 icosahedra at pressures of 5-6 GPa. ltoh et al. [225] reported the synthesis of B 6 0 powder by reacting B203withboron at 1350-1400°C and the subsequent sintering of that powder with B4C or cBN at 3-7GPa and 1500-1800°C for 10-30min. Although the hardness of the resulting compacts exceeded 40 GPa, the fracture toughness did not exceed 1.5 MPa m1/2. The synthesis of B 6 0 was patented initially in 1992 [226, 2271, but no boron suboxide products have appeared in the market as yet.

1.4.7 Stishovite The search for new ultrahard materials has also been extended into nonboron based oxides. The expectation was that high pressure allotropes would exhibit a high

1.5 Industrial Applications of Diamond and cBN

527

coordination of the metallic ion, as well as more ionic bonding, thus hopefully possessing higher stiffness and, therefore, higher hardness. Leger et al. [166] in 1994 measured the bulk modulus of the high pressure phases of HfOz and RuOz, which they obtained at pressures above 42 and 12GPa. The resulting bulk moduli place the two compounds above diamond in the case of Hf02 and between cBN and diamond for Ru 0 2 . In 1996 Leger et al. [228] reported the synthesis of stishovite from c1 quartz at 20 GPa and 1100°C and measured a hardness of 33 GPa for the polycrystalline compact produced. As the samples obtained were very small, a maximum load of 1.9 N during hardness testing was possible. Synthesis of this material in larger quantities has not yet been possible.

1.5 Industrial Applications of Diamond and cBN M. W. Bailey

1.5.1 Introduction As discussed earlier, the modern industrial diamond business began to expand significantly following the commercial availability of synthetic diamond in the late 1950s. Before then, only natural diamond produced as a by-product of gem diamond mining was available and was, by its nature, limited in volume. Demand was increasing to the point where new application development was a largely selfdefeating exercise since, the more successful they were in terms of using large volumes of diamond, the less feasible they became in terms of total world supply. With the introduction of synthetic diamond, this situation changed dramatically on two important fronts. Firstly, the only constraint remaining on the development of new applications was the cost and technology of producing a suitable product; and secondly the constraints on product characteristics imposed by the nature of a natural raw material were removed: diamond could now be specifically engineered for its intended use. Natural and synthetic diamond are complementary, not competing products. While today a minor player, natural industrial diamond still plays an important role in a variety of special applications where one or other of its particular characteristics will make it the preferred material (refer to Section 1.5.5 for some examples). Industrial applications of diamond have developed over the years as a result of developments by tool makers, machinery manufacturers and the advert of new materials. In addition, new diamond (and cBN) products have evolved, either as a result of technological advances in synthesis or in response to the requirements of a new application, and it is this multi-partnership relationship within the industry which has resulted in the dramatic growth and diversification since the early 1960s.

528


1.5.2 Abrasive Application 1.5.2.1 Bonded Tools One remarkable property of diamond is its hardness and associated resistance to abrasion and it is this which has led to its use as an industrial abrasive. Cutting and polishing stone, glass, and other hard materials were among its early areas of application. Diamond has one limitation in respect of its application, this being its reaction with iron at high temperatures (causing a reversion to graphite and hence high rates of wear), which in general can make it uneconomic in the machining of the ferrous materials by comparison with other conventional abrasives such as aluminum oxide and silicon carbide. Cubic boron nitride (cBN) does not react in this way and hence, although only having 50% of the hardness of diamond, it is still substantially harder than the conventional abrasives and this makes it suitable as a high performance abrasive for use on ferrous workpieces. Figure 28 summarizes the major areas of application for diamond and cBN abrasives. The advantages of using diamond and cBN abrasives are due mainly to their high hardness and abrasion resistance which, when applied correctly, result in longer tool lives, higher levels of productivity, closer tolerances, and many other advantages. The majority of diamond and cBN abrasive is used in so-called ‘bonded tools’, although much polishing and fine finishing is carried out using the abrasive as a loose powder or in a slurry when mixed with an oil or other carrier medium. Diamond and cBN abrasives are normally used in particle sizes of approximately 1 mm down to less than 0.1 pm, and their sizing is subject to international standards, the most widely used being the FEPA standard [229]. In a bonded tool, the abrasive particles are held in a bond or matrix material, usually distributed randomly but in a controlled concentration. The main functions of the bond are: -

to hold the abrasive particles to form a molded profile of the required shape which is usually attached to a hub or carrier

Diamond

Cubic Boron Nitride (cBN)

Non-ferrous materials:

Ferrous materials:

Glass -flat, decorative, optical

Hard ferrous components

Engineering ceramics Cemented tungsten carbides Reinforced plastics Natural stone Semiconductor materials

(Diamond accounts for 75-80%, cBN accounts for 20-25%) Figure 28. Main application areas for diamond and cBN abrasives.


529

Layers of diamond particles in a resin, metal or vitrified bonded tool

Figure 29. Basic principle of bonded tools.

-

to release worn abrasive particles as they reach the end of their useful lives, thus exposing new ones.

It is therefore a general requirement of a bonded tool that the abrasive and the bond material wear at similar rates. The most common types of bond used are as follows: ~

-

resin, normally thermosetting phenolic or polyimide resins vitrified, based on glass frits metal, normally sintered alloys electroplated, single layer or a few layers of abrasive attached to a hub by an electro-deposited metal layer, normally nickel.

Figure 29 shows the concept diagramatically. Figure 30 shows a selection of bonded tools including saw blades and grinding wheels. In the case of the first three bond types, since the rate of bond erosion and rate of wear of the abrasive have to be matched if optimum performance is to be obtained, the selection of the abrasive which has the correct characteristics of size, impact strength, and fracture mode matching the characteristics of the bond is very important. Workpiece material plays a major role in this selection process but also the type of machining operation, the machining conditions and the primary requirement of the operation are important. Maximum tool life, high rates of material removal or high quality of surface finish are common requirements but normally a compromise between one or more is required. Figure 31 shows some of the issues which affect the final machining performance. In the case of the fourth bond type, electroplated (or EP), the abrasive particles are embedded in a metal layer to something less than 50% of their total dimension as shown schematically in Fig. 29. The abrasive particles wear down to the level of the supporting EP layer, at which point the tool is at the end of its life. With this type of tool, the choice of abrasive

530


Figure 30. A selection of bonded tools.

depends primarily upon the workpiece material, the machining conditions and the primary role of the machining operation (e.g. to produce a high surface finish or high material removal rates). The role of the EP bond layer is to support the abrasive and not wear significantly during the life of the tool. When new, EP tools have a high degree of protrusion of the abrasive particles above the bond layer (much higher than the other types of bond) and hence can machine at high material removal rates while generating low cutting forces. The term often used to describe such a

Task Machine Workpiece Coolants

4''"

Wheel type Wheel bond Wheel grit Preparation

Figure 31. Main parameters affecting tool performance.


53 1

General purpose bond Simple to manufacture Easier to use than vitrified or metal (more forgiving) Profile holding not as good as vitrified or metal

Resin bond diamondgrinding the teeth of a cemented carbide cutter

General ouroose for all tvoes of material Sintered‘catbide, Non-oxide ceramics Cermets, Natural stone (polishing)

Figure 32. Main attributes of a resin bond.

characteristic is ‘free-cutting’. As the abrasive particles become worn, so this free cutting character is gradually reduced and the cutting forces increase until the end of the useful life of the tool is reached. 1.5.2.2 Grinding with Diamond and Cubic Boron Nitride Abrasives 1.5.2.2.1 Introduction Grinding with diamond and cBN abrasives is employed by a very wide range of industries and application areas. In dentistry, small EP diamond drills are widely used, spectacle lenses and crystal glass items are machined using diamond, large diamond wheels are used to produce high volumes of cemented tungsten carbide cutting tool inserts to precise tolerances and cBN wheels are widely used in the automotive and bearing industries to produce high volumes of precision ground components. In grinding operations, all four of the main bond types are used. As a general guide they may be considered to have specific advantages and the choice is often dependent upon which characteristic is the more important. Figures 32-35 show a Very high wear resistance +Good profile holding +Difficult to dress + Manufacturing technology more advanced than resin

Decorative, flat, optical glass Aluminium oxide Metal bond diamond grinding a high alumina ceramic seal

Figure 33. Main attributes of a metal bond

532


Single layer of Ni-electroplated abrasive Cheap to manufacture Excellent for profile grinding Limited life High stock removal possible

Electroplated diamond saw blade slicing a block of optical glass

Green carbide, Aluminium oxide Ferrites, Natural stone (profiling) Precious stone, Dental Reinforced plastics, Semiconductors Profile grinding hard ferrous (cBN)

Figure 34. Main attributes of an electroplated bond

typical application for each of the four bond types and some of their particular advantages. The characteristics of the abrasive used can be selected from a wide range available; size, resistance to impact and fracture characteristics being among the more important and Fig. 36 summarizes the main differences and their effects. Grinding operations can take place with the application of a coolant or lubricant (wet grinding) or with no such application (dry grinding). 1.5.2.2.2 Grinding with Diamond Abrasives In grinding applications, the main types of bond used are metal and resin bonds. In resin bond wheels, low strength friable diamond types are normally used, a typical diamond being shown in Fig. 37. Since the resins normally used for the manufacture of diamond tools have relatively low strength and are also sensitive to high Hard and abrasion resistant Up to 30% porosity improves chip clearance independent of grit size Holds its edgelprofile well Longer life than resin Easily dressed More difficult to manufacture than resin

Vitrified bond diamond grinding d a tungsten carbide roll

Non-oxide ceramics Hardened steels (cBN)

Figure 35. Main attributes of a vitrified bond.


~ ~ ~ oblockv n ~orits ~ -r , withstand higher forces greater exposure higher rates of stock removal lower degree of surface finish suited to more ductile, long chipping workpiece materials

-

533

withstand lower forces less exposure lower rates of stock removal higher degree of surface finish suited to hard, brittle, short chipping workpiece materials

Figure 36. Basic effects of abrasive characteristics.

temperatures, a thick metal cladding of about 10 pm of nickel or sometimes copper is applied to the diamond particle such that typically it will comprise 50-60% of the total particle weight (Fig. 37). This cladding functions in two main ways (Figs 38 and 39), firstly by increasing the surface area of the particle and hence aiding its retention by the bond, and second by acting as a heat sink to protect the bond from thermal damage due to the high interfacial temperatures generated between the abrasive particle and the workpiece, particularly under heavy grinding conditions. These temperatures, although of short duration, can reach in excess of 800°C as the particle passes through the arc of cut.

Figure 37. A typical friable diamond abrasive PDA321 used (normally in metal clad form) in resin bonded grinding wheels.

534


Figure 38. The role of metal cladding.

Resin bond tools are used to machine a wide variety of non-ferrous, usually abrasive, workpiece materials. A very small number of ferrous materials can be ground economically with diamond, for example, some carbon rich cast irons which, providing machining temperatures are maintained relatively low, do not chemically react with the diamond, and other difficult-to-machine materials such as some grades of stainless steel which have a tendency to work harden during machining. The major application area for resin bond diamond wheels is the grinding of cemented tungsten carbide (which was also its initial application area in the 1940s). The list below gives the main areas of application in descending order of relative size: -

-

grinding cemented tungsten carbide grinding ceramics and cermets grinding semiconductors grinding polycrystalline diamond (pcD) machining stone/glass others

Although the machining of cemented tungsten carbide represents approximately half the resin bond tools used, other areas are growing, notably semiconductors, ceramics and cermets, pcD and the machining of stone. The bond of an abrasive tool performs various tasks:

* Holds the abrasive particles * Forms a moulded part or segment

(attached to a carrier -blank or hub- by brazing/welding/ adhesive fixinglsintering or electroplating methods)

+ Release worn diamond at a controlled rate

- wear rates of abrasive particles and bond must match for a given workpiece and machining para~eters

Figure 39. The role of the bond.


535

Figure 40. A typical high strength diamond abrasive PDA999 used in grinding wheels.

Metal bond diamond wheels are predominantly used in the machining of glass including windows for the automotive and construction industries, decorative glass such as lead crystal and also optical and electronic components. The diamond types most commonly used in metal bond tools are relatively high in impact strength with regular crystal shapes. Figure 40 shows a typical high strength diamond abrasive. Sintered metal bonds commonly based on bronze are much more abrasion resistant than resin bonds, and therefore better suited to machining workpiece material which produce highly abrasive swarf, glass being one example. The list below summarizes the main application areas for metal bond diamond wheels. grinding glass machining refractory materials - machining semi-precious stones - grinding pcD - grinding ceramics - honing (mostly ferrous materials) ~

Electroplated diamond wheels are used in applications where free-cutting characteristics are required. Fibreglass and other composite materials are in general highly abrasive although not particularly hard. The high abrasion resistance of diamond makes it the ideal tool material for machining these types of material and the very open texture and free-cutting nature of EP tools mean that these materials can be machined quickly. Figure 41 shows an example. Vitrified bond diamond wheels are most commonly used to machine polycrystalline diamond. pcD is difficult to machine and often has to be machined to close tolerances when being used as a cutting tool material as shown in Fig. 42. Both metal and resin bond grinding wheels are also used, but vitrified bond wheels are becoming the most popular, since they provide a good compromise between relatively good wheel life, and hence ease of maintaining dimensional tolerance on the pcD workpiece, and the rate at which the pcD can be ground. Another advantage of vitrified

536


Figure 41. An EP diamond wheel cutting glass.

520mm diameter SYNDITE CTCOO2-tipped saw blade for wood Figure 42. A circular saw blade tipped with pcD.

Flank grinding a PCD-tipped saw


531

bond systems is that a degree of porosity can be built into the bond. Resin and metal bonds are effectively 100% dense and clearance for the swarf produced during grinding is provided by the protrusion of the abrasive particle above the surface of the bond. The pores, or voids, in porous vitrified bonds provide extra clearance volume for the removal of swarf out of the grinding zone and also the transport of coolant into it. In general, all diamond grinding wheels operate at average peripheral speeds in the range 2&30ms-'. If a material is being ground dry, without coolant, the wheel speeds tend to be lower, and also the difficult materials such as cermets and pcD tend to be ground at lower speeds to avoid the excess generation of heat. A small number of specialized applications use much higher wheel speeds, one example being the sawing of glass tubes which, providing the machine is specially designed for the purpose, with particular attention to the coolant application, can be cut much more quickly with a wheel rotating at speeds up to 60-80 ms-' . 1.5.2.2.3 Grinding with Cubic Boron Nitride Abrasives Cubic boron nitride abrasives are used to machine hard ferrous materials. As with diamond, all four of the main bond types are used and, during the early days of its use, in the 1970s and early 1980s, resin bond tools were the majority. Today, vitrified bond tools are the most commonly used type, particularly in high volume production operations, with resin, electroplated, and metal bonds failing in second, third and fourth place respectively. A reason for the growth in the use of vitrified bond cBN tools is that this type of bond system offers a good compromise between high material removal rates, ease of use, and low wear rate resulting in the ability to hold tight tolerances. The growth in the use of cBN since the early 1970s has been due to its penetration into volume production operations, for example in the automotive and aerospace industries. Prior to that time, cBN wheels were used primarily in the toolroom for sharpening and re-sharpening high speed steel drills, milling cutters and other tooling and also in small wheels for precision jig grinding (Fig. 43). The economics of cBN grinding have been improved by the adoption of creep feed grinding techniques and also the use of higher wheel peripheral speeds. Both of these developments required the development of suitable grinding machines. Creep feed (or deep) grinding techniques were originally developed using conventional wheels and the principle of the technique is shown in Fig. 44. In 'normal' or reciprocating grinding mode, a shallow depth of cut is used together with a relatively high table speed to achieve the required material removal rate. Under these conditions, the arc of cut, and hence the area of wheel in contact with the workpiece at any one time, is very small. In creep feed grinding, a significantly deeper cut is made at a slow table speed. Under these conditions, the arc of cut is much longer and hence the number of particles instantaneously in contact with the workpiece is much greater. The specific rate of material removal is a function of the depth of cut and the feed rate of the table. In general, when comparing the two modes of grinding, and assuming equal material removal rates, the main differences for creep feed grinding are that -

the number of particles in contact is higher the load on each particle is lower

538

I Diuniontl Matrrials mu' their. App1iuition.s

Figure 43. Dry grinding with cBN ~

~

the total normal load in the machine is greater (due to the greater number of particles) the life of the grinding wheel is longer surface finish produced is normally improved.

Reciprocating Grinding

DOC = S - SO Micron Table speed = 5 - 15 m/min

Creep Feed Grinding

DOC = 5 - 25 millimetres Table speed = 0.1 - 1 mlmin

Figure 44. Principle of creep feed and reciprocating grinding.

1 .SIndustrial Applications of Diamond and cBN

539

Wheel Speed Figure 45. Effect of wheel speed on grinding wheel performance.

A second development in cBN grinding is the adoption of higher wheel peripheral speeds. In the early days of grinding with cBN, wheel speeds similar to those used for diamond wheels were used. It was found however that increasing the wheel peripheral speed improved the life of the grinding wheel providing coolant was used (Fig. 45). Initially speeds up to 60 ms-' were thought to be optimum but today, wheel speeds in the range 6&150ms-' are common in industry. This has been made possible by improvements in the mechanical design of both the grinding wheels and the machines on which they are used since, at high speeds, safety becomes a major issue which has to be addressed by suitable speed certification for the wheels and safety guarding of the machine. At various academic institutions, much higher wheel speeds have been used, and wheel life has continued to improve. In one such series of tests in Germany, speeds of 500 ms-' were run in one series of experiments. As a general rule in any grinding operation, increasing the rate of material removal results in a decrease in wheel life. Hence, there is usually a balance in the cost of the time taken to grind and the cost of the wear on the wheel (Fig. 46). By adopting either, or both, the technologies of creep feed grinding techniques and high wheel speeds, higher material removal rates (and hence rates of production) can be achieved whilst retaining, or even improving, the wheel life and hence direct machining cost due to wheel wear. The field of application for cBN grinding is almost entirely that of hard ferrous metal machining, although some superalloys are also ground using cBN. The list

T

cost

Material Removal RateFigure 46. Trend of cost versus removal rate.

540


600mm dia.resin bond wheel containing ABN360 Grit size 8126 (120/140) Material removed 5-20 micron

Figure 47. Grinding automotive piston rings with cBN.

below gives the main application areas: -

automotive and aerospace material machining tool and cutter grinding gear grinding machining of bearings general grinding of hardened steel components honing of hardened steel

The largest area of application is in the automotive and aerospace industries, grinding of camshafts, crankshafts and other precision engine parts such as turbine blades and injector nozzles being examples (Fig. 47). Precision grinding of gear tooth profiles is another area where cBN, either in vitrified or EP tools, is used (Fig. 48). The original application area for cBN, tool and cutter grinding, is still a cBN grinding application, the grinding of high speed steel drills and milling cutters, both during manufacture and re-sharpening during use being examples (Fig. 49). In addition to faster metal removal rates and better tolerances compared to grinding methods using conventional abrasives, cBN wheels have the ability to produce lower workpiece temperatures during grinding, which reduces the potential for thermal damage. This can be of significant importance when critical components are being machined.

1.5.3 Machining of Stone and Concrete The largest application area for diamond abrasives in terms of volume consumed is the sawing, drilling and surfacing of natural stone and concrete. This application area grew dramatically during the 1970s and 1980s, due to a number of influences. Industrial diamond became more widely available and also became available in an


541

Figure 48. Electroplated gear grinding wheel

ever increasing range of size and also other characteristics such as strength. Developments by tool manufacturers and machine builders improved the economics of machining hard natural stone types such as the granites, making their processing more economical. There was also a movement towards the use of granite from marble in the construction industry. One of the reasons for this was that it was recognized that granite was far more durable than marble in respect of both wear if used on floors, and also more resistant to chemical attack and discoloration in urban environments if used as a cladding on the exterior of prestige buildings. Its use, therefore, became more widespread and the quantity of diamond abrasives used to process it has continued to increase.

Figure 49. Flute grinding high speed steel.

542


Figure 50. Section of a diamond wire saw.

The types of diamond abrasive used for cutting stone and concrcte are normally those with high impact strength and of a relatively coarse size. The normal production route for stone products includes: -

extraction from the quarry in the form of large blocks squaring and stabbing the rough blocks in the stone yard cutting the blocks into either cladding or tiles surfacing and polishing the finished items.

In all these operations, diamond tools are widely used. For quarrying, diamond wire sawing is established as the standard method of extraction for soft rock (e.g. marble) and the technology for hard rock is being established. Development of diamond wire began in the early 1970s but it came into common use only in the late 1980s. It consists of a high tensile steel wire fitted with beads containing the diamond abrasive, typically 40 beads per metre. The overall diameter of the beads is 6-12mm (Fig. 50). Two intersecting holes drilled in the quarry face and the wire, which can be up to 500m long, is fed through them, joined to form an endless loop and then driven at speeds of 20-50msC' while tension is applied, thus cutting a slot through the rock and enabling a block to be cut (Fig. 51). The resulting block suffers much less damage compared with using explosive techniques and this, in turn, leads to less waste in subsequent processing. A major problem encountered with early designs was erosion of the wire by the fine particles of stone produced by the sawing, leading to breakage. This is now alleviated by protecting the wire with rubber or plastic sleeving bonded to the wire between the beads. When quarrying granite, a relatively low wire speed is typically used, 20-30 ms-' and, for a medium grade of material, a block measuring 30m long x 10m wide x 5 m deep can be separated from the surrounding material in less than 12 h.


543

Figure 51. Wire sawing in a stone quarry.

However, because the block produced is of a regular size, up to a 40% increase in usable material is possible with commensurate similar improvements in the economics. Following extraction from the quarry, the large blocks are either sawn into thick slabs for use in the monument industry or they are squared for subsequent production into tiles or stone cladding. Diamond wire saws are used for these operations together with large circular sawing machines utilizing diamond saw blades of up to 4m diameter (Fig. 52). The advantages of using diamond tools compared to machines utilising silicon carbide abrasives are increased speed of production, improved accuracy of cut and much reduced pollution. Tile production is a major application for diamond saw blades. The sawn blocks are machined on multi-blade machines using up to 32 blades, 1.6 m in diameter on one spindle (Fig. 53). These machines typically cut the granite blocks into a series of slabs 12mm thick, with a thickness tolerance of f0.5mm normally required. A horizontal blade then traverses the block to separate the slabs produced and these

Figure 52. Block sawing of granite.

544


Figure 53. Multi-blade sawing of granite to produce tiles.

slabs in turn are ground to final thickness and sawn to final size, both operations using diamond tools. The final stage in the production of tiles is polishing which is normally achieved using automatic multistage machines (Fig. 54). In this final process, many of the stages are now completed using diamond tools, the early

Figure 54. Polishing natural stone tiles.

1.5 Industrial Applications of Diamond und cBN

545

Figure 55. Resin bond polishing head.

stages in polishing using metal bond tools with high strength diamond to achieve the high material removal rate required and the later stages using resin bond tools incorporating lower strength, more friable, grades of diamond particle. Figure 55 shows a selection of polishing tools. Another major application area for diamond abrasive tools is in the construction, refurbishment and demolition industries. In many instances, concrete, masonry and also roadways need to be cut or drilled and diamond tools provide an economic solution, their advantages being speed, accuracy and also reduced damage to the surrounding structure compared to other methods using percussive hammers or thermal lances. Circular saw blades are used in a wide variety of operations and metal bond tools are most commonly used. On roadways, both asphalt and concrete, diamond saws are used to cut trenches during repairs and also to provide slots to embed sensor wires for traffic management systems (e.g. traffic signals) (Fig. 56). Concrete roadways, runways and pedestrian walkways can, under wet conditions, become slippery and diamond blades, up to 100 spaced along a spindle, are used to cut a series of parallel grooves in the concrete to aid water drainage and improve skid resistance and hence safety in critical areas (Fig. 57). Saw blades are used to cut openings in walls and floors during both building and refurbishment and can also be used during controlled demolition. The advantage of a diamond saw, compared to a saw of conventional abrasive, is its ability to cut both concrete, masonry and steel reinforcement (Fig. 58). The majority of all diamond sawing of concrete is carried out with use of water as a coolant. This has the combined effect of controlling the temperatures at the cutting edge and also removing the swarf produced by the saw. Some modern blades are designed to run dry (Fig. 59). These are now widely used on construction sites for cutting tiles and the minor correction work which needs to be carried out. Wire saw machines, similar in principle to those used for cutting stone, are also used widely for large demolition operations. When using a circular saw, there is a limit on the depth of cut which can be made due to the presence of the drive spindle

546


Figure 56. Roadway sawing.

Figure 57. Roadway grooving to improve skid resistance.

1 .SIndustrial Applications of Diamond and cBN

Figure 58. Sawing in the construction industry.

Figure 59. Dry sawing of concrete.

547

548


Figure 60. Wire sawing of heavily reinforced concrete.

in the centre of the blade. This is a problem if deep sections have to be cut. The wire saw does not have any such limitation and concrete, heavily reinforced with steel as used in large structures such as road bridges, can be cut with modern wire saw machines. Figure 60 shows such an operation. Drilling stone with diamond bits is now widely used and concrete, both plain and steel reinforced, brickwork and other masonry have been drilled for many years using diamond core bits (Fig. 61). These tools can range in size from approximately 20mm to in excess of a meter. They are all metal bonded tools and were until recently designed to operate with water as a coolant. In some areas, the use of water can be inconvenient and a relatively recent development has been the design of core bits which do not need to be cooled with water. These dry drill bits are widely used on construction sites to produce accurate holes in walls for electrical, plumbing and other services (Fig. 62).

1.5.4 Applications of Polycrystalline Ultra-hard Materials 1.5.4.1 Introduction The polycrystalline derivatives of diamond and cBN, normally referred to as pcD and pcBN respectively, provide engineers with materials which have many of the properties of diamond and cBN, notably hardness and abrasion resistance, but in the form of relatively large isotropic pieces, usually in the form of flat discs. These materials have a wide variety of uses, both as a defined edge cutting tool

Figure 61. Core drills ~iseclin the coiistriiclioii inclustrq

element and also as a wear resistant material. As a cutting tool material, the general rules which apply to abrasives also apply to pcD and pcBN. ~

~

pcD is for use in non-ferrous applications. pcBN is for use on ferrous workpieces.

Figure 62. Dry drilling of t i u w n r q

550

1 Diamond Muteriuls and their Applications

cutting 10000 speed [mlmin] 1000

-

100

-

10

Cemented

-

-

0.1

CVD diamond

Carbon tool steel

I

I

I

I

I

I

1.5.4.2 Cutting Tools of Polycrystalline Diamond In most successful applications of pcD cutting tools, the advantages over more conventional tool materials, including sintered tungsten carbide and ceramics, are due to their high hardness and abrasion resistance. This in turn results in longer tool life giving rise to closer tolerances and the ability to work efficiently at high cutting speeds and hence give higher levels of productivity. In addition, workpiece materials which are difficult to machine can be machined efficiently with pcD. Continuous material’s development over the years has played a very significant role in the advances made in machining technology, resulting in higher levels of productivity and, in particular, closer tolerances over long machine runs which are now required by modern manufacturing industries. Figure 63 shows the achievable cutting speeds which have been made possible by advances in cutting tool material technology. Figure 64 shows schematically how current materials relate to each other in terms of two important characteristics, toughness and abrasion resistance/hot hardness. The ideal cutting tool material would feature very high toughness with equally high wear resistance. Although, today, such a material does not exist, it is the objective of many development programmes (see Section 1.4). The main application areas for pcD are the machining of non-ferrous metals, e.g. aluminum alloys, magnesium alloys, brass, copper wood and wood composites - fibre-glass and carbon fibre composites - plastics and rubber - mineral materials -

-

The automotive and aerospace industries provide a large number of applications for pcD, particularly with the movement to more widespread use of advanced alloys

I .5 Industrial Applications of Diamond and cBN

I

PCD

Toughness, transverse rupture strength

Figure 64. Properties of cutting tool materials

55 1

Ideal cutting tool material

-

and reinforced plastics. The main application areas in the automotive industry are summarized in Fig. 65. An example of an engine component commonly machined with pcD is an aluminum alloy engine block (Fig. 66) manufactured from an silicon-silicon alloy Engine components Gearbox components Transmission components Braking systems Interior fittings Body panels Figure 65. pcD applications in the automotive industry

Figure 66. Rough milling A1-9% Si engine blocks with pcD.

552


Figure 67. Face milling A1-12% Si gearboxes with pcD.

which, dependent upon the alloy, can be very abrasive. Similar classes of material are used in the manufacture of gearbox casings and these can also be machined more efficiently with pcD than with any other cutting tool materials (Fig. 67). The cylinder bores of aluminum-silicon alloy engine blocks are also typically machined with pcD. Figure 68 shows such a component from a V-12 engine being finish bored with pcD.

Figure 68. Finish boring of cylinder blocks with pcD


553

Figure 69. SYNDITE 025 machining 20% S i c reinforced Al M M C brake discs.

A material originally used in racing cars and motorcycles is a metal matrix composite (MMC) made of aluminum reinforced with silicon carbide particles. Its light weight and high strength made it ideal for use as a disc brake rotor and also other components. The presence of the silicon carbide makes it a difficult material to machine, pcD being the best cutting tool material. Figure 69 shows a typical MMC machining operation on a brake disc rotor. Machining of reinforced plastics is also a major application for pcD tooling. Figure 70 shows part of a glass reinforced plastic (GFRP) truck body section which is machined with a pcD router. A relatively recent application area for pcD has been in the woodworking industry. Wood composites such as medium density fibreboard (MDF), chipboard, laminated boards with plastic coatings, for use in furniture and on floorings, and also composites, used in wall claddings and in ceilings, all tend to be highly abrasive

Figure 70. pcD milling cutter machining a GFRP truck bumper.

554


Figure 71. Schematic cross-section through typical abrasive particle board.

when being machined. Figure 71 shows a typical cross-section through a wood laminate. Saw blades, routers and other profiling tools now use pcD tips for high volume production. Figure 72 shows a profiled pcD tipped cutter being used to produce furniture panels in MDF, a commonly used material for this application and Fig. 73 shows typical wear rate results obtained when machining such materials and compares high speed steel, cemented tungsten carbide and pcD in terms of tool wear rate as a function of distance machined. 1.5.4.3 Drilling with Polycrystalline Diamond The major advantage of using pcD in a rock drilling application is that pcD drill bits are designed to remove rock by shearing rather than by crushing or grinding, as is the case with most other types of drill. Rock removal by shearing is very efficient, requiring only 15-20% of the energy required by crushing and grinding. This, in

Figure 72. pcD is typically used on medium density fibreboard (MDF), chipboard, and abrasive hardwoods.

1.5 Industrial Applications of Diamond und cBN

555

Flank wear Vb [mm]

Speed Depth Feed Infeed

0.15

1000 d m i n 10mm 0 1 mdtooth = 2 rnmlpass = = ~

f c_tc;;3p

3';

Distance machmed [m]

Figure 73. Wear rates of HSS (high speed steel), WC (tungsten carbide) and pcD routers machining chipboard.

turn, means that pcD drill bits have the potential to drill much faster than conventional bits and, in addition, the high abrasion resistance of pcD results in the cutters remaining sharp and hence extending considerably the life of the drill bit. Drilling for oil and gas is a highly specialized subject: pcD cutters were introduced to this application area in the early 1980s and, since that time, the design of the drill bits and the pcD cutting elements themselves have resulted in ever-increasing performance. A factor which limited the early pcD oil and gas exploration bits was premature failure, either of the tungsten carbide supporting the pcD or of the braze attaching the cutter to the bit body. Today, most of these difficulties have been overcome, and pcD bits offer high penetration rates and long bit lives when drilling medium hard abrasive rock formations. pcD drill bits are also used in mining, although the volume is considerably less than that in the oil and gas exploration industry. One example is in methane drainage, where holes are drilled into the rock surrounding the coal seam and the methane pumped out to prevent it seeping into the mining area and hence causing a fire hazard. Also, in underground mining, holes have to be drilled in large quantities for roof bolts to be inserted to provide roof support, and blast holes have to be drilled into which explosives are packed. Figure 74 shows a selection of typical pcD mining bits. 1.5.4.4 Non-cutting Application of Polycrystalline Diamond

pcD, being hard and abrasion resistant, is potentially ideal for use as a wear resistant material. One traditional non-cutting application is in wire drawing, where pcD is

556


Figure 74. Selection of mining bits.

used as a die to reduce fine wire (Fig. 75). pcD wire drawing dies are most widely used to draw non-ferrous wires such as copper where the life, compared to cemented tungsten carbide, is up to 1000 times longer. A more recent application field has been the use of pcD as a wear part, such as a bearing, a workpiece support rest in a machining operation, a nozzle for abrasive fluids and in high precision gauges. Figure 76 shows a pcD used as a workpiece support rest for an automotive engine component and Fig. 77 shows precision automatic gauging equipment fitted with pcD tipped measuring fingers. In such cases, the high wear resistance of pcD leads to improved accuracy being achieved.

Figure 75. Drawing wire with diamond dies.

1.5 Industrial Applications of’Diamond und cBN

557

Figure 76. SYNDITE pcD wear parts support a crankshaft during machining

1.5.4.5 Machining Ferrous Materials with Polycrystalline Boron Nitride pcBN is used almost exclusively for machining hard ferrous materials. As described earlier, cBN and pcBN do not react with iron at high temperatures in the same way as diamond, and hence can be used to machine hard ferrous metals. To machine a hard ferrous material effectively, the temperature of the workpiece in the zone immediately ahead of the contact between it and the cutting tool (the shear zone) has to be increased such that the workpiece becomes softer and hence easier to machine. Figure 78 shows the hardness/temperature curve for a typical hard ferrous material. By selecting the correct machining conditions of cutting speed and tool geometry, these relatively high temperatures can be achieved, and because pcBN retains its

Figure 77. Measuring diamond honing heads with a pcD-tipped micrometer.

558

1 Diamond Materials und their Applicutions

Figure 78. Workpiece material hardness as a function of temperature.

hardness at these high temperatures, efficient machining can be conducted. The high temperatures generated are limited to the cutting zone only, providing the machining conditions are selected correctly. The bulk temperature of the workpiece remains low and coolant is only required in a minority of cases. pcBN tool materials fall largely into two classes: -

high cBN content (>go% by volume cBN) in a ceramic binder low cBN content ( ~ 6 0 %by volume cBN) in a ceramic binder

High cBN content materials are used in rough machining applications, whereas lower cBN content tools tend to perform better in finishing applications where shallow depths of cut are being taken and a good surface finish is required. Abrasion resistant cast-irons, for example Ni-hard, are used to manufacture pumps for pumping slurry and gravel in the quarrying, mining and associated heavy industries. These materials are difficult to machine with conventional tooling materials but can be machined easily with a high cBN content tool material such an Amborite. Steel rolling mill-rolls, which are normally ground to final size, can also be machined with pcBN, the advantage being higher production rates and easier swarf disposal since most rough machining operations using pcBN are conducted dry, without coolant (Fig. 79). pcBN can also be used in milling, and Fig. 80 shows an example of a milling operation on a machine tool slideway. In the automotive industry, high content pcBN is used to machine a wide range of cast iron components such as brake discs (Fig. 81) and cast iron cylinder blocks (Fig. 82). Using low-content pcBN tools for finishing previous operations often replaces a grinding operation. In many cases, a hardened steel component was finish ground because its hardness meant that it was not possible to machine it in any other way. Many such components may now be turned to final tolerance and surface


559

Ni-HARD Pump Bodies (left) High Chromium Iron Rolls (right) Figure 79. Typical applications of high cBN-content pcBN tools

finish using suitable pcBN tooling. Figure 83 shows a gear component being finished in this way. Low content pcBN can also be used in fine milling. Figure 84 shows a n automotive gearbox component being finish milled with a low content pcBN (DBCSO).

1.5.5 Applications of Single Crystal Diamond Natural diamond crystals were the original diamond tools used hundreds of years ago for cutting and engraving purposes, because the outstanding properties of

Figure 80. AMBORITE milling hardened meehanite slideways.

5 60

I Diamond Materials und their Applicutions

Figure 81. Machining cast iron brake discs with AMBORITE.

hardness and abrasion resistance were recognized, but it was only possible to use relatively large pieces of diamond, several millimeters in size. More recently, other outstanding properties have been recognized, in particular the high degree of optical transparency of type I1 diamonds and the very high thermal conductivity of (the relatively rare) type HA diamond. In recent years, the technology to synthesize

Figure 82. AMBORITE machining a six-cylinder cast iron engine block.

1.5 Induxtrial Applications of Diamond and cBN

561

Figure 83. Finish machining a case-hardened automotive gear with DBCSO.

economically synthetic single crystal diamond of several millimeters in size has been developed and this has contributed to the availability of relatively large diamonds for industrial use. Single diamond crystals are currently used in a wide variety of industrial applications, albeit some highly specialized, and these range from rock drilling to sophisticated applications in micro-surgery and electronics. One advantage of synthetic crystals is that they can be synthesized and processed to specific required sizes and also crystallographic orientations chosen according to the application (Fig. 85).

Figure 84. DBCSO finish milling a transmission component (60HRc).

5 62


i/--g MONODITE

\-Facet

MONODITE Triangle Cube

Whole Stone

4 -

c

MONODRESS MONODITE Triangle

Octaiedron

MONODIE 111

Figure 85. MONOCRYSTAL orientation.

1.5.5.1 Diamond Truing and Dressing Tools Diamond tools are employed for truing and dressing conventions- abrasive w ,.eels, like aluminum oxide and silicon carbide wheels. Truing ensures that newly installed wheels run concentrically on the machine spindle, whilst dressing restores the desired surface topography and texture to grinding wheels that have become dull, glazed, or loaded with grinding debris. The standard types of truing and dressing tools available include single-point, multi-point, chisel, rotary, and indexing tools (Fig. 86). This wide variety caters for all types of conventional abrasive wheel and all types of truing and dressing operations, such as those required to produce both straight and profiled wheel faces, those that have steps, radii and shoulders, and for plunge forming etc. Rotary dressers are used for truing and dressing vitrified-bond cubic boron nitride wheels. Custom-made tools are also produced for specialized applications.

Figure 86. Selection of diamond dressers.

1.5 Industrial Applications of' Diamond and cBN

563

1.5.5.2 Diamond Cutting Tools Diamond cutting-tool applications are many and varied, ranging from rock drilling to the finish machining of nonferrous metals and plastics. In rock drilling with surface-set diamond drill bits, the structural strength, fracture resistance, hardness and wear resistance of diamond are most in evidence because rock is removed by crushing rather than shearing. Its crystal structure and extreme hardness also make it possible for diamond to accept and retain cutting edges which are flawless and unbroken over their entire length when viewed at high magnification. Diamond tools are used for turning, milling and engraving non-ferrous metals, including precious metals, non-metallic materials, abrasive composites etc. (Fig. 87). Typical of applications for which diamond has been successfully used are the turning of aluminum alloy photocopier cylinders and computer memory discs, copper printing rolls, non-ferrous metal optical components and plastic contact lenses, the fly-cutting of multi-facetted scanner mirrors, the production of precious metal jewellery and the restoration of crazed plastic aircraft windows to their original transparency. In such operations, diamond tools give reduced downtime and improved quality, thus making it possible to achieve greater economy in comparison with other types of cutting tool. In many cases, the superior finish obtained with diamond machining means that no subsequent polishing or buffing operations are required, thus improving the economics even further.

Figure 87. Typical diamond cutting tools.

Figure 88. Surgical scalpel for micro-surgery.

In the medical field, diamond blades are used in surgical scalpels (Fig. 88) and in microtome knives (Fig. 89) to prepare tissue samples for microscopic examination. Diamond knives are also used for cleaving fibre optics.

1.5.5.3 Non-Cutting Applications of Diamond Tools

I n comparison with the use of diamond as a cutting tool material, the consumption of diamond for non-cutting applications is small. Nevertheless, high-qualitysingle crystal industrial diamond is often the only suitable material for specific tasks, due to its outstanding mechanical, optical, electrical, chemical or thermal properties.

Figure 89. Diamond microtome knife.


565

Figure 90. Ultra-high pressure diamond anvils.

The types of equipment in which diamond crystals are used include anvils for high-pressure research (in excess of a million atmospheres or Megabar) (Fig. 90), heatsinks, bearings for precision chronometers and other ultra-sensitive electronic meters, styli for audio equipment and surface measuring instruments, hardnesstesting indenters (Fig. 91) and distance stops on machine tools. In addition, diamond is used for metal-forming applications, such as burnishing, engine turning, and the ruling of diffraction gratings, where little or no material is actually removed. The most important of such forming operations is wire drawing. In top-quality wire-drawing dies, the ultimate wear resistance and surface polish rely on the particular qualities of diamond crystals. Such dies enable large quantities of

Figure 91. Diamond hardness indentor.

566


Figure 92. Diamond wire drawing dies.

very fine wire (down to 10pm in diameter) to be produced with the required precision (Fig. 92).

Acknowledgments The authors are indebted to Dr I. Sigalas and Dr R. J. Caveney for the overall editing of the article, and to Mrs Chancellor Teffo for her very competent secretarial help.

References S. Tennant, Phil Trans. R. Soc. 1797,97, 591. H . Liander and E. Lundblad, Arkiv Kemi, 1969, 16, 139-149. F. P. Bundy, H. T. Hall, H. M. Strong, and R. H. Wentorf, Nature, 1955,176 (No. 4471), 51-55. F. P. Bundy, Nature (London), 1973, 241, 116-1 18. G. Davies, Diamond, Adam Hilger, Bristol, 1984, Chapter 4. H. M. Strong, Am. J. Phys. 1989, 57, 794-802. J. W. Harris, Diamond geology, in The Properties of Natural and Synthetic Diamond, J. E. Field (Ed.), Academic Press, London, 1992, pp. 342-393. 8. S. E. Kesson and A. E. Ringwood, Chemical Geology, 1989, 78, 97-1 18.

1. 2. 3. 4. 5. 6. 7.

References

567

9. F. P. Bundy, W. A. Bassett, M. S. Weather, R. J. Hemley, H. K. Mao, and A. F. Goncharov, Carbon, 1996, 34, 141-153. 10. G. J. Davies and T. Evans, Proc. Ro. Soc. Lond., 1972, A328, 413427. 11. M. L. H. Green, M. J. Rosseinsky, and E. S. C. Tsang, 2nd International Interdisciplinary Colloquium on the Science and Technology of the Fullerenes, Keble College and University of Oxford Museum, Oxford, Abstracts, Elsevier Science, Oxford, 1996, p. 186. 12. R. Roy, in Advances in Nav Diamond Science and Technology, S. Saito (eds), MYU, Tokyo, 1994, pp. 17-22. 13. P. Mistry, M. C. Turchan, S. Liu, G. 0. Granse, T. Baurmann, and M. G. Shara, Innovations in Materials Research, 1996, 1(2), 193-207. 14. R. Berman and F. Simon, Zeitschrijt,fur Elektrochemie, 1955, 59, 333-337. 15. C. S. Kennedy and G. C. Kennedy, J . Geophys. Res., 1976,81(14), 2467-2469. 16. R. C. Burns and G. J. Davies, Chapter 10 in The Properties ofNaturalandSynthetic Diamond, J. E. Field (Ed.), Academic Press, London, 1992. 17. H. Kanda, M. Akaishi, and S. Yamaoka, Appl. Phys. Lett. 1994, 65, 784. 18. S. K. Singhal and H. Kanda, J . Cryst. Growth, 1995, 154, 297. 19. N. V. Novikov and A. A. Shul’zhenko, in Science and Technology of”ew Diamond, S. Saito, 0. Kukunaga, and M. Yoshikawa (Eds), KTK Sci, Tokyo, 1990, p. 339. 20. A. V. Andreyev and H. Kanda, Diamond Relat. Mater., 1997, 6, 28-32. 21. H. 0.A. Meyer, Inclusions in Diamond in Mantle Xenoliths, P. H. Nixon (Ed.), John Wiley and Sons, 1987, pp. 501-522. 22. M. Akaishi, H. Kanda, and S. Yamaoka, J . Cryst. Growth, 1990, 104, 578. 23. M. Akaishi, H. Kanda, and S. Yamaoka, Jap. Appl. Phys., 1990, 29, L1172. 24. M. Akaishi, Diamond Relat. Muter., 1993, 2, 183-189. 25. L. Sun, M. Akaishi, S. Yamaoka, H. Yamaoka, and S. Nakano, Special Issue of The Review of High Pressure Science and Technology, AIRAPT-16 1997, 6, 36. 26. M. Arima, K. Makayama, M. Akaishi, S. Yamaoka, and H. Kanda, Geology, 1993, 21, 968. 27. M. Akaishi, Proceedings of 3rd NIRIM International Symposium on Advanced Materials, Tusukuba, Japan, Mar 4-8, 1996,75. 28. H. Yamada, M. Akaishi, and S. Yamaoka, Special Issue of The Review of High Pressure Science and Technology, AI RAPT-16, 1997, 6, 35. 29. S. E. Haggerty, Nature, 1986, 320, 34-37. 30. A. Onodera, K. Terashima, T. Urushihara, K. Suito, H. Sumiya, and S. Satoh, J . Mater. Sci., 1997,32, 430943 18. 31. A. Barti, S. Bohr, R. Haubner, and B. Lux, Int. J . Refract. Metal HardMater., 1996,14,145-147. 32. G. R. Cowan, B. W. Dunnington, and A. H. Holtzman, U. S. Patent, USP34010119, Du Pont de Nemours USA, 1965. 33. R. H. Wentorf Jr., J . Phys. Chem., 1965, 69, 3063-3069. 34. Y. Ma et a/., Appl. Ph)x Lett., 1994, 65, 822. 35. M. N. Rugueiro, P. Monceau, and J. L. Hodeau, Nature, 1992, 355, 237-239. 36. Y. Muranaka, H. Yamashita, and H. Miyadera, Diamond Re/. Mater., 1994, 3, 313-318. 37. H. E. Spear and J. P. Dismukes (eds), Parts I1 and 111 in, Synthetic Diamond, Emerging CVD Science and Technology, John Wiley and Sons, 1994, pp. 21-304. 38. R. Roy, D. Ravichandran, P. Ravindranathan, and A. Badzian, J . Mater. Res., 1996, 11(5), 116441168. 39. P. W. Bridgman, Phys. Rev. 1935,48, 825-832. 40. L. F. Vereschagin, A. A. Semerchan, N. N. Kuzin, and Y. A. Sadkov, Dokl. Akad. Nauk S S R , 1968, 183, 565-567. 41. B. von Platen, in Modem Very High Pressure Techniques, R. H. Wentorf (Ed.), Butterworths, Washington, DC, 1962, pp. 234-239. 42. Yu. N. Pal’yonov, Yu. I. Malinvosky, Yu. M. Borzdov, A. F. Khokhryakov, A. I. Chepurov, A,, Godovikov, and N. V. Sobolev, Doklady Akademii Nuuk S S S R , Earth Science Section 1990,315, 1221-1224. 43. H. Kanda and 0. Fukunaga, in Advances in Earth and Planetary Science, High Pressure Research in Geophysics, V o l I 2 ,S. Akimoto and M. H. Manghnani (Eds), Riedel, Dordrecht, 1982, pp. 525-535.

568


44. Yu. N Pal’yonov, A. F. Khokhryakov, Yu. M. Borzdov, A. G. Sokol, V. A. Gusev, G. M. Rydov, and N. V. Sobolev, Geol. Geojiz, 1997,38, 882-906. 45. H. M. Strong and R. E. Hanneman, J. Chenz. Phys., 1967,46, 3668-3676. 46. G. Will and G. Graf, High Press. Res., 1994, 12, 17-27. 47. I. Sunagawa, Morphology of natural and synthetic diamond crystals, in Material Science of the Earth’s Interior, I. Sunagawa (Ed.), Tokyo Terra Scientific Publishing Co, 1984, pp. 303-330. 48. Y. Wang, R. Tanakabe, and M. Wakatsuki, Proc. 15th AIRAPT Int Conf. on High Pressure Science and Technology, Warsaw, Sept 11-15, 1995, pp. 235-237. 49. M. Wakatsuki and K. Takaro, K. Inone, Inclusions in synthesised crystals of diamond, in 8th AIRAPT Covf. Proc., High Pressure in Research and Industry, Uppsala, pp. 369-372. 50. T. L. McCormick, W. E. Jackson, and R. J. Nemanich, J. Muter. Res., 1997, 12(1), 253-263. 51. G. Davies, Chemistry and Physics ofcarbon, Vol. 11, P. L. Walker (Ed.), Marcel Dekker, New York, 1982, pp. 70-71. 52. R. H. Wentorf, J. Phys. Chem. 1971, 75, 1833-1837. 53. H. M. Strong and R. E. Tuft, US Pat. 4073380, 1978. 54. Y. Suji, et al., Jpn. Pat 2631/1986, 165-167. 55. F. P. Bundy, H. M. Strong, and R. H. Wentorf Jr, Chemistry of Physics of Carbon, Vol. 10, 1973, 213-272. 56. M. Wakatsuki, in Materials Science of the Earth’s Interior I. Sunagawa (Ed.), Terra Scientific Publishing Company, Tokyo, 1984, pp. 35 1-374. 57. Yu. A Kocherzhinskii and 0. G. Kulik, Powder Metal1 Metal Ceramics, 1996, 35(7-8), 470483. 58. H. Kanda, in Advances in New Diamond Science and Technology, S. Saito, N. Fujimori, 0. Fukunaga, M. Kamo, K. Kobashi, and M. Yoshikawa (Eds), MYU, Tokyo, 1994, pp. 507512. 59. M. Wakatsuki, Jpn J. Appl. Phys. 1966, 5, 337. 60. T. Sugano, N. Ohashi, T. Tsurumi, and 0. Fukunaga, Diamond Relat. Muter. 1996,5,29-33. 61. H. M. Strong and R. M. Chrenko, J. Phys. Chem. 1971,75(12), 1838-1843. 62. R. C. Burns, S. Kessier, M. Sibanda, C. M. Welbourn, and D. L. Welch, Proceedings of the 3rd NIRIM International Symposium on Advanced Materials, 1996, pp. 105-1 11. 63. H. Kanda, T. Ohsawa, 0. Fukunaga, and I. Sunagawa, J. Cryst. Outgrowth, 1989,94, 115-124. 1964,86,45404544. 64. P. Cannonand and E. T. Conlin, J. Am. Chem. SOC. 65. R. C. Burns, J. 0. Hansen, M. Sibanda, R. A. Spits, C. M . Welbourn, and D. L. Welch, Diamond Re/. Muter., 1999, 8, 1433-1437. 66. H. Kanda and S. C. Lawson, Indust. Diamond Rev. 1995,2, 56-61. 67. H. Sumiya and S. Satoh, Diamond Re/. Muter., 1996, 5 , 1359-1365. 68. W. Li, H. Kagi, and M. Wakatsuki, Trans. Muter. Res. SOC. Jpn., 1994, 14B, 1451-1454. 69. M. Wakatsuki, W. Li, Y. Gohda, and L. Y. Ding, Diamond Rel. Muter., 1996, 5, 56-64. Muter. Sci. Jpn., 1993, 42(476), 588-592. 70. S. Yazu, J. SOC. 71. H. Kanda, New Diamond, 1990, 58-62. 72. R. A. Chapman, Private Communication, 1996, De Beers Diamond Research Laboratory, South Africa. 73. S. Satoh and H. Sumiya, Jpn. Muter. Res., 1995, 15, 183-195. 74. R. C. Burns, V. Cvetkovic, C. N. Dodge, D. J. F. Evans, M. L. T. Rooney, and C. M. Welbourn, J. Cryst. Growth, 1990, 104, 257-279. 75. S. Satoh, H. Sumiya, K. Tsuji, and S. Yazu, in Science and Technology of New Diamond, S. Saito, 0. Fukunaga, and M. Yoshikawa (Eds), KTK Scientific Publishers/Terra Scientific Publishing Co., Tokyo, 1990, pp. 351-355. 76. H. Kanda, T. Ohsawa, and 0. Fukunaga, in Abstracts 2nd Meeting of Diamond, Tokyo, 1987 pp. 23-24. 77. M. L. T. Rooney, J. Cryst. Growth, 1992, 116, 15-21. 78. A. T. Collins, H. Kanda, and R. C. Burns, Phil. Mag. B, 1990, 61(5), 797-810. 79. S. L. Lawson, H. Kanda, K. Watanabe, I. Kiflawi, and Y. Sato, J. Appl. Phys., 1996,79(8), 110. 80. J. I. Koivula and C. W. Fryer, Gems and Gemology, 1984, 20, 146-158.

References

569

81. J. E. Shigley, E. Fritsch, C. M. Stockton, J. I. Koivula, C. W. Fryer, and R. E. Kane, Gems and Gemology, 1986,22, 192-208. 82. J. E. Shigley, E. Fritsch, C. M. Stockton, J. I. Koivula, C. W. Fryer, R. E. Kane, D. R. Hargett, and C. W. Welch, Gems and Gemology, 1987, 23, 187-206. 83. M. L. T. Rooney, C. M . Welbourn, J. E. Shigley, E. Fritsch, and I. Reinitz, Gems and Gemology, 1993, 29. 3 8 4 5 . 84a. J. E. Shigley and E. Fritsch, J . Cryst. Growth, 1993, 128, 425428. 84b. J. E. Shigley, E. Fritsch, I. Reinitz, and T. M. Moses, Gems andGemology, 1995,31, 256-264. 85. C. M. Welbourn, M. Cooper, and P. M. Spear, Gems and Gemology, 1996, 32(3), 3 156-169. 86. M. Kamo, Y. Sato, S. Matsumoto, and N. Setaka, J . Cryst. Growth, 1983, 62(3), 642-644. 87. J. C. Angus, H. A. Will, and H. S. Stanko, J . Appl. Phys., 1968, 39, 2915. 88. W. G. Eversole, US Patent 3, 030, 187 and US Patent 3, 030, 188 1962. 89. K. V. Ravi, Mater. Sci. Eng., 1993 B19. 203-227. 90. R. Bennan, Solid State Commun., 1996, 99(1), 35-31. 91. K. T. Jacob, Solid State Commun., 1995, 94(9), 763-765. 92. C. J. Brannon and S. L. McCollum, in New Diamond Science and Technology, R. Messier et al. (Eds), Materials Research Society, Pittsburgh, PA, USA, 1991, p. 117. 93. A. Tsuzuki, S.-I. Hirano, and S. Naka, J . Muter. Sci., 1985, 20, 2260-2264. 94. H. Uchikawa, H. Hagiwara, and K Nakamura, in Science and Technology of New Diamond, S. Saito et al. (Eds), Terra Scientific Publishing Co., 1990. 227. 95. Hawtin, J. B, Lewis, N. Moul, and R. H. Philips, Roy Soc. Phil. Trans. Ser A , , 1966,261(116), 67-95. 96. L. F. Vereschagin, E. N. Yakoviev, L. M. Buchnev, and B. K. Dymov, Teplofizika Vysokikh Temperafur, 1977, 1S(2), 316-321. 97. Y. Bar-Yam and T. D. Moustakas, Nature, 1989, 32, 786787. 98. A. A. Abramzon and A. A. Novozhenets, Russ. J . Phys. Chem., 1992, 66(8), 122c1222. 99. J. Abrahamson, Carbon, 1973, 11, 337-362. 100. J. 0. Hansen, R. G. Copperthwaite, T. E. Derry, and J. M. Pratt, J. Colloid Interface Sci., 1989, 130(2), 347-358. 101. R. A. Munson, Carbon, 1967, 5, 471474. 102. J. A. Nuth, Astrophys. Space Sci., 1987, 139(1), 103-109. 103. P. Badziag, W. S. Verwoerd, W. P. Ellis, and N. R. Greiner, Nature, 1990,343(6255), 244245. 104. E. Anders and E. Zimmer, M eteoritics, 1993, 28, 49c514. 105. A. N. Nesterov, Y. I. Merezhko, and V. V. Chernikov, Run. J . Phys. Chem., 1983,57(2), 307-308. 106. D. V. Fedoseev, ColloidJ. USSR, 1978,40(2), 341-342. 107. D. V. Fedoseev, B. V. Deryagin I. G. Varshavskaya, and A. V. Laverentev, Prog. Surf. Sci., 1994,4S, 84-87. 108. F. Banhart and P. M. Ajayan, Nature, 1996. 382,433435. 109. B. Wei, J. Zhang, J. Liang W. Liu Z Gao, and D. Wu, J . Muter. Sci. Lett., 1997, 16,402403. 110. H. Moissan, C. R. Acad. Sci. Paris, 1894, 118, 32G32.5. 111. J. B. Hannay, Nature, 1880, 22, 255-256. 112. Y. G . Gogotsi K. G . Nickel, and P. Kofstad, J . Muter. Chern. 1995, S(12), 2313-2314. 113. M. A. Cappeli, US Patent 5,417,953, 1995. 114. J-Y Wang, Y-Z Wan, D-W Zhang, and Z-J Liu, J . Muter. Res., 1997, 12(12), 3250-3253. 115. F. Davanloo, E. M. Juengerman, D. R. Jander, T. J. Lee, and C. B Collins, J . Muter. Res., 1990, S(l l), 2398-2404. 116. R. Roy, H. S. Dewan, and P. Ravindranathan, Muter. Res. Bull., . 1993, 28, 861-866. 117. R. Roy, K. A. Cherian: J. P Cheng, A. Badzian, C. Langlade, H. Dewan, and W. Drawl, lnnov. Muter. Res., 1996, 1(1), 165-87. 118. P. A. Molian and A. Waschek, J . Muter. Sci.,1993, 28, 1773. 119. M . Horio, A. Saito, K. Unou, H. Nakazono, N . Shibuya, S. Shima, and A. Kosaka, Chem. Eng. Sci., 1996, Sl(1 l), 3033-~3038. 120. Y. Namba, J . Vac. Sci. Technol. A , , 1992, 10(5), 3368-3370. 121. M. Takaya, Japanese Patent Hei 6-25896, 1994. 122. T. Suzuki, T. Ishihara, T. Yamazaki, and S. Wada, Jpn J . Appl. Phys., 1997,36 part 2, No. 4B, L50kL506.

570 123. 124. 125. 126. 127. 128. 129. 130.


J. F. Prins, Diamond Rel. Muter., 1993, 2, 646-655. K. U. Cherian, M. Komath, S. K. Kulkarni, and A. Ray, Diamond Rel. Mater., 1994,4,20-25. A. Jayaraman, Sci. Am., 1984,250,42-54. Y. A. Litvin and L. T. Chudinovskikh, Trans. Russian Acad. Earth Sci., 1997,35SA(6),908-91 1. F. P. Bundy and J. S. Kaspers, J . Chem. Phys., 1967,46(9), 3437-3446. R. Boehier, Private communication, Jun 1994, Max Planck Inst, Mainz Germany. J-A Xu and E. Huang, Rev. Sci. Instrum., 1994, 65(1), 204-207. J. Kleiman, R. B. Heimann, D. Hawken, and N. M. Salansky, J . Appl. Phys. ( U S A ) , 1984, 56(5), 5 1440-1454. 131. 0. Bergmann and N. F. Bailey, in High Pressure Explosive Processing of Ceramics, R. A. Graham and A. B. Sawaoka (Eds), Trans Tech duPont Wilmington, USA, 1987. 132. S. Fahy, S. G. Louie, and M. L. Cohen, Phys. Rev. B, 1986, 34(2), 191-199. 133. R. H. Wentorf, Jr., J . Chem. Phys., 1957, 4, 956. 134. R. H. Wentorf, Jr., New Diamond Sci. Technol., 1990, 1029-1037. 135. V. L. Solozhenko, J . Hard Materials, 1995, 6, 51-65. 136. N. V. Novikov and V. L. Solozhenko, J . Chem. Vap. Dep., 1996,4, 240-252. 137. R. H. Wentorf, Jr., J . Chem. Phys., 1961, 34(3), 809-812. 138. V. L. Solozheuko, Adv. Muter. ‘96: Proceedings of the Third NIRM International Symposium, Tsukuba, Japan, 1996, pp. 119-124. 139. F. P. Bundy and R. H. Wentorf, J. Chem. Phys., 1963,38, 11441149. 140. Y. Moriyoshi, M. Kamo, N. Setaka, and Y. Sato, J. Muter. Sci., 1983, 18(1), 217-224. 141. H. T. Hall, Rev. Sci. Instrum., 1958, 29(4), 267. 142. H. T. Hall, Science, 1970, 169, 868-869. 143. H. D. Stromberg and D. R. Stephens, J . Am. Ceram. Soc., 1970,49(12), 1030-1032. 144. H. Katzman and W. F. Libby, Science, 1971, 172, June 1971, 1132-34. 145. R. H. Wentorf, W. A. Rocco, Patent SAP 7315038, 1973. 146. L. F. Veraschagin, A. A. Semerchan, Patent UKP 1382080, 1975. 147. A. I. Kolomiitsev and V. E. Smirnov, Almazyi Sverkhtverdye Materialy, 1980, 7, 4 5 . 148. K. Uehara and S. Yamaya, Science and Technology of New Diamond, S. Saito (Ed.), Tokyo, KTK Sci Publ., 1990, pp. 203-209. 149. A. A Shulzhenko, V. G. Gargin, V. A. Shishkin, and A. A. Bochechka, Ukrainian Academy of Sciences, Institute for Superhard Materials, printed in Kiev Naukova Dunko, 1989, Chapter 4, 93-173. 150. Yu A. Kocherzhinskii, A. A. Shulzhenko, and V. A. Shishkin, Vliyanie Vysok Davieniya Na Structuru Z Svoistva Materialov, 1983, 34-40. 151. 0. A. Voronov and A. A. Kaurov, Soviet J . of Superhard Materials, 1994, 16(1), 6 1 0 . 152. S. H Robertson, PhD Thesis, University of Reading, 1984. 153. D. V. Fedoseev, G. A. Sokolina, and E. N. Yakoviev, Sov. Phys. J., 1985, 30(5), 425427. 154. D. V. Fedoseev, G. A. Sokolina, E. N. Yakoviev, and A. V. Lavrentev, Sov. Powder Metall. Ceram. Muter., 1984, 23(9), 683-686. 155. A. A. Shulzhenko, S. A. Bozhko, I. A. lgnatusha, and A. N. Vashchenko, Sov. J . Superhard Muter., 1990, 12(5), 14-19. 156. A. A. Shulzhenko, S. A. Bozhko, A. I. lgnatusha, and A. N. Vashchenko, Sov. J . Superhard Muter., 1988, 10(5), 13-17. 157. S. A. Bozhko, A. I. Ignatusha, V. G. Delevi, and A. A. Budyak, Sov. J. Superhard Muter., 1989, 11(5), 32-35. 158. M. Akaishi, S. Yamaoka, J. Tanaka, T. Ohsawa, and 0.Fukunaga, J . Am. Ceram. Soc., 1987, 70(10), C237-C239. 159. J. C. Walmsley and A. R. Lang, J . Muter. Sci., 1988, 23(5), 1829-1834. 160. R. H. Wentorf, Jr. and W. A. Rocco, U. S. Patent 3 767 371, 1973. 161. Sumitomo Electric Industries, Ltd., US Patent 4 334 063, 1982. 162. A. A. Shulzhenko, S. A. Bozhko, A. N. Sokolov, A. Petrusha, N. P. Bezhenar, A. I. Ignatusha, Cubic Boron Nitride Synthesis, Sintering and Properties, The Ukrainian Academy of Sciences, Institute for Superhard Materials, Kiev Naukova Dumka, 1993, 254. 163. P. N. Tomlinson, R. J. Wedlake, Znd. Diamantenrundschau, 1983, 7(4), 234241. 164. M. Wakatsuki, K. Ichinose, and T. Aoki, Muter. Res. Bull., 1972,7, 999-1004.

References

57 1

165. M. Wakatsuki and K . Ichinose, 4th Internutionul Conference on High Pressure, Kyoto, 1974, 441445. 166. F. R. Corrigan, 6th AIRAPT Conference on High Pressure Science and Technology,' 1979, 1, 994-999. 167. 0. Fukunaga and M. Akaishi, High Press. Res., 1990, 5, 911-913. 168. J. M. Leger, J . Haines, and B. Blanzat, J . Muter. Sci. Lett., 1994, 13, 1688-1690. 169. M. C. Payne, M. T. Teter, D. C. Alan, T. A. Arias, and J. D. Ioannopoulos, Rev. Mod. Phys., 1992,64, 1045. 170. A. Y. Liu and M. L. Cohen, Science, 1989, 245, 1981. 171. M. L. Cohen, J . Hard Mater., 1991, 2, 13. 172. P. K. Lou, M. L. Cohen, and G. Martinez, A m . Phys. Soc., 1987, 35, 919&9194. 173. R. Hoffmann, T. Hughbanks, M. Kerttesz, andP. H. Bird,J. A m . Chem. Soc., 1983,105,4831. 174. M. A. Tamor and K . C. Hass, J . Muter. Res., 1990, 5, 2273. 175. A. Y. Liu, M. L. Cohen, K. C. Hass and M. A. Tamor, Phys. Rev. B, 1991,43, 6742. 176. F. Diedrich and Y. Rubin, Angew. Chem. Int. Ed. Engl., 1992, 31, 1101. 177. H. Alberts. T. Sekine, H. Kanda, Y. Bando, and K. Hojou, J . Mater. Sci.Lett., 1990,9, 1376. 178. A . Y. Liu and M. L. Cohen, Phys. Rev. B., 1990,41, 10727-10734. 179. J. V. Badding, Adv. Mater., 1997, 9, 877-886. 180. A. Y. Liu and R. M. Wentzcovitch, Phys. Rev., 1994, 1350, 10362. 181. R. M. Wentzcovitch and J. L. Martins, SolidState Commun., 1991, 78, 831. 182. R. M. Wentzcovitch, Phys. Rev., 1991, B44, 2358. 183. D. M. Teter and R. J. Henley, Science, 1996, 271, 53-55. 184. D. Li, S. Lopez, Y. W. Chung, M. S. Wong. and W. D. Sproul, J . Vuc.Sci. TecAnol., 1995,13, 1063. 185. T. A. Yeh, C. L. Lin, J. M. Sivertsen, and J. H. Judy, IEEE Trans. Laga., 1991, 27, 5163. 186. J. J. Cuomo, P. A. Leary, D. Yu, W. Reuter, and M. Frisch, J . Vac. Sci. Technol., 1979, 16, 299. 187. H. Sjostrom, W. Lanford, B. Hjovarso, K. Xing, and J. E. Sundgren, J . Muter. Res., 1996, 11, 981. 188. C. Niu, Y. Z. Lu. and C. M. Lieber, Science, 1993, 261, 334. 189. K. Ogata, J. F. D. Chubaci, and F. Fujimoto, J . Appl. Phys., 1994, 76, 3791. 190. J. Schwan, W. Dworschak, K . Jung, and H. Erhardt, Diamond Rel. Muter.. 1994,3, 1034. 191. T. Y. Yen and C. P. Chon, Appl. Phys. Lett., 1995, 67, 2801. 192. Y. Guo and W. Goddard 111, Chem. Phys. Lett., 1995,237,12-76; J. E. Lowther, Phys. Rev. B , 1998,57, 57245727. 193. D. M. Bhusari, C. K. Chen, K. H. Chen, T. J. Chuang, L. C. Chen, and M. C. Liu, J . Muter. Res., 1997, 12, 322-325. 194. Z. J. Zhang, S. Fan, J. Huang, and C. M. Lieber, Appl. Phys. Lett., 1996, 68(19), 2639-2641. 195. F. Fujimoto and K. Ogata, Jpn. J . Appl. Phys., 1993, 32, L. 420. 196. D. Liu, Y. W. Chung, M. S. Wong, and W. D. Sproul, J . Appl. Phys., 1993 74, 219. 197. Z. M. Ran, Y. C. Du, Z. F. Ying, Y. X. Qiu, X. X. Xiong, J. D. Wu, and F. M. Li, Appl. Phys. Lett., 1994, 65, 1361. 198. C. Niu, Y. Z. Lu, and C. M. Lieber, Science, 1993, 261, 334. 199. S. Kumai and T. L. Tansley, Solid State Commun., 1993, 88, 803. 200. D. Norton, K. J. Boyd, A/H. Al-Bayari, S. S. Todorov, and J. W. Rabalais, Phys. Rev. Lett., 1994, 73, 118. 201. K. M. Yu, M. L. Cohen, E. E. Hailer, W. L. Hausen, A. Y. Liu, andI. C. Wu, Phys. Rev., 1994, B49, 5034. 202. J. P. Riviera, D. Texier, J. Delafond, M. Jaouen, E. L. Mathe, and J. Chanmond, Muter. Lett., 1995, 22, 115. 203. X. W. Su, H. W. Song, F. Z. Cui, W. Z. Li, and H. D. Li, Surj: Coat. Technol., 1996,84,388-391. 204. Y. Zhang, H. Li, and Q. Xue, Muter. Res. Soc., 1996, 317-322. 205. C. Yan, Liping Guo, and E. G. Wang, Phil. Mag. Lett., 1997, 75, 155-162. 206. K. Yanamoto, Y. Koga, K. Yase, S. Fuziwara, and M. Kubota, Jpn. J . Appl. Phys., 1997,36, L23CL233. 207. E. G. Wang, Chen Yan, and Guo Liping, Physica Scripta, 1997, T69, 108-114

572


208. M. J. Yacaman, J. J. M. Gil, and M. Sarikaya, Muter. Chem. Phys., 1997,47, 109-117. 209. D. W. Wu, W. Fan, H. X. Guo, M. B. He, X. 0. Meng, and X. J. Fan, Solid State Commun., 1997, 103, 193-196. 210. S. Xu, Han-Shi Ki, S. Lee, and Yin-Au Li, IEEE, 1997,213-214. 21 1. Xiao-Ming He, Li Shu, Wen-Zhi Li, and Heng-De-Li, J . Muter. Res., 1997, 12, 1595-1602. 212. Wu Dawei, Fu Dejan, Guo Huaixi, Zhang Zhihong, Meng Xianquan, and Fan Xiangiun, Phys. Rev. B, 1997,56,49494954. 213. Chen, L. L., D. M. Bhusari, C. Y. Yang, K. H. Chen, T. J. Chuang, M. C. Lin, C. K. Chen, and Y. F. Huang, Thin Solid Films,1997, 303, 66-75. 214. J. Badding, Adv. Mater., 1997, 9, 877. 215. K. Ploog, P. Rauh, W. Stoeger, and H. Schmidt, J . Cryst. Growth, 1972, 13/14. 350. 216. K. Ploog, H. Schmidt, E. Amberger, G. Will, and K. H. Kossobutzki, J . Less-Common Met., 1972, 29, 161. 217. H. Saitoh, K. Yoshida, and W. A. Yarbrough, J . Muter. Res., 1993, 8, 8. 218. L. Vel, G. Demazeau and J. Etournea, Mat. Sci. Eng., 1991, BID, 149. 219. A. R. Badzian, T. Niemyski, and E. Ockusmik, Proc. 3rd lnt. Con$ on CVD, F. A. Glaski (Ed.), The Electrochemical Society, Penington, NJ, 1972, p. 747. 220. K. Montasser, S. Hattori, and S. Morita, Thin Solid Films,1984, 117, 311. 221. J. Loeffier, F. Steinbach, J. Bill, J. Mayer, and F. Aidinger, Z. Metailkol, 1996, 87, 170. 222. D. Hegermann R. Riedel, W. Dresier, C. Oehr, B. Schindler, and H. Brunnen, Chem. Vap. Depos., 1997, 3, 257 223. A. Badzian, Appl. Phys. Lett., 1988, 53, 2494. 224. P. F. McMillan, W. T. Petuskey, H. Hubert, K. J. Kingma, L. A. Garvie, A. Grzechnik, and A. Chizmeshya, Adriutico Research Conference, Miramare, Trieste, Italy, July 1997. 225. H. Itoh, I. Maekawa, R. Yanamoto, and H. Iwahara, Rev. High Press. Sci. Technol., 1998,7, 986-988. 226. C. Ellison-Hayashi, M. Zandi, F. J. Csillag, and Shih-Yee Kuo, US Patent 5, 135, 892, 19912. 227. C. Ellison-Hayashi, M. Zandi, D. K. Shetty, P. Kuo, R. Yeckley, and F. Csillag, US Patent 5330,937, 1994. 228. J. M. Leger, J. Haines, M. Schmidt, J. P. Petitet, A. S. Pereira, and J. A. H. da Jomada, Nature, 1996,383,401. 229. Federation Europkene des Fabricants de Produits Abrasifs (FEPA), Standard for Superabrasives Grain Sizes, 2nd Edn. September, 1997.


2 Applications of Diamond Synthesized by Chemical Vapor Deposition R. S. Sussmann

2.1 Introduction The ability of Chemical Vapor Deposition (CVD) technology to synthesize diamond as large plates of high and controlled purity [l-51 is enabling a host of new applications in science and technology that were hitherto hampered by the practical difficulties inherent in the use of natural diamond. The basic principles of CVD for the synthesis of polycrystalline diamond have been discussed in the chapter by Klages et al. [6]. The aim of this chapter is to discuss a selected range of applications to illustrate how CVD diamond is emerging as a key (and in some cases unique) engineering material. The impact that CVD diamond is starting to have on some areas of modern technology is in some way similar to the impact that the invention of the high pressure and high temperature (HPHT) techniques for the synthesis of diamond had on the abrasive industry in the early sixties [7]. At that time the industrial implementation of HPHT technology gave a renewed impetus to an existing but relatively small diamond abrasive industry. HPHT synthesised diamonds have been the foundation of what is now a relatively mature industry in which the extreme hardness and wear resistance of diamond are exploited for numerous abrasive and cutting tool applications [S]. The potential importance of diamond in technology follows from the wide range of technically desirable thermomechanical, optical, and electronic properties exhibited by diamond [9-131. In addition to its extreme hardness and exceptional wear resistance, diamond is the stiffest known material, has a broad transmission spectral range (from the ultraviolet to the far infrared and extending to microwave frequencies), has the highest room-temperature thermal conductivity, one of the lowest thermal expansion coefficients and is radiation hard and chemically inert to all acid and base reagents. What makes diamond so attractive, and in many cases unique, is not only the extreme value of some of the above properties such as hardness, thermal conductivity or the Young modulus, but the combination of two or more of these properties in a high performance product, as will be discussed later in this chapter. The enormous interest that CVD diamond has generated since the mid eighties has been based on the perception that this new diamond synthesis technology could overcome all the limitations previously encountered with the availability of large size natural diamond specimens [ 14,151.As was the case in the abrasives industry, the use of natural diamond in other fields of technology has been relatively small

514

2 Applications of Diamond Synthesized by Chemical Vapor Deposition

and restricted to niche applications in optics [ 161, the thermal management of small electronic devices (such as laser diodes or impatt oscillator diodes) [17,18], and others [19,20]. For these types of applications only the relatively rare high-purity Type IIa form of natural diamond could be used. The most abundant form of natural diamond, Type Ia, contains relatively large concentrations of nitrogen which degrade properties such as optical transmission [21],thermal conductivity [22],and electronic transport [23]. Another major difficulty in the use of natural diamond is the stringent limitation to the size of specimens that are practically available. Most natural Type IIa windows sold commercially are below 5 mm in diameter and typical heat sinks are smaller than 1 x 1 mm [16, 181. The cost of natural single crystal windows of sizes in excess of 19 mm, if available, would be outside the budget range of most practical applications [16]. The largest and the most spectacular natural diamond product ever made was a window, 18.2mm in diameter and 2.8 mm thick, used for the main pressure cell of the pioneer Venus probe launched on August 1978 [24]. Purity and size are also serious limitations in the use of HPHT synthetic diamond in these types of applications because of the high intrinsic single substitutional nitrogen concentration and the practical maximum size of typical HPHT diamond crystals [7]. It is within this perspective that efforts in recent years have been directed to the development of a CVD diamond technology that allows the routine manufacture of large plates of diamond of high and controlled purity. Some of this effort has met with considerable success [l-5,251. In the following sections of this chapter we will briefly review the properties of various types of CVD diamond materials that have been developed to address specific applications. We will illustrate through a few selected examples how the current ability to manufacture CVD diamond (including large area plates and three-dimensional shapes) as a reliable and robust engineering material is enabling the use of diamond in an increasing range of technically demanding applications.

2.2 Properties of Chemical Vapor Deposited Diamond 2.2.1 Material Grades For most practical applications CVD diamond is synthesized as a polycrystalline material. Many of its properties can be expected to be strongly influenced by its grain structure, including factors such as grain size, preferred orientation, intergranular purity, microstress, etc. All of these factors depend strongly on the synthesis technique and process conditions [7] and although there are generic features which are common to all CVD diamond materials, substantial variations in properties such as tensile strength, optical transparency, thermal conductivity and others can be found in material from different origins. It is not yet possible, for instance, to define a generic optical grade of CVD diamond because optically transparent CVD diamond specimens prepared in different laboratories are known to exhibit

2.2 Properties of' Chemical Vapor Deposited Diamond

575

Figure 1. Optical grade CVD diamond window, 100mm diameter, 0.7mm thick

different values of fracture strength, intrinsic bulk absorption, and total forward scatter [26,27]. The majority of the properties described in this section refer to the type of CVD diamond that is currently available as a commercial product sold under the trade name of DIAFILM [28]. With sufficient control of synthesis conditions, the extra degree of freedom conferred by the polycrystalline structure has been used to formulate different 'grades' of CVD diamond in which properties are optimized for specific applications. Optical grades, for instance, exhibit the highest optical transparency and can be made routinely in discs up to 100 mm in diameter. In contrast mechanical grades are dark in color but have a higher mechanical strength and can be made in plates up to 160mm in diameter. Various thermal grades have been developed for thermal management applications which offer a compromise between good thermal conductivity and cost. Most of the applications currently pursued with CVD diamond are relatively new and in certain cases the quality of the corresponding grade is subject to continuous development in which factors such as cost and performance are gradually optimized. Figure 1 shows an example of a l00mm diameter, 0.7mm thick, optical grade CVD diamond window which has been polished on both sides to an optical finish. The as-grown grain morphologies of optical (2 mm thick) and mechanical (0.5 mm thick) grade samples are shown in Fig. 2. Figure 3 shows the Raman spectrum of an optical grade CVD diamond sample measured by a Renishaw instrument with an excitation wavelength of 633nm. The main Raman line is centered at 1332cm-' and the full-width-half-maximum (FWHM) line width is approximately 3.0cmP', which compares well with typical values of 2.8 cm-' to 3.1 cm-' observed in high-quality single-crystal Type IIa natural diamond specimens measured with the same instrument.

576

2 Applications of' Diamond Synthesized by Chemical Vapor Deposition

Figure 2. (a) SEM of the as-grown morphology of an optical grade DIAFILM sample 2mm thick. (b) SEM of the as-grown morphology of a mechanical grade DIAFILM layer 0.5 mm thick.

2.2.2 Optical Properties For a sample of optical grade CVD diamond 1 mm thick in which both faces have been polished to an optical finish the optical transmission from the fundamental cutoff at 220 nm to the far infrared (IR) part of the spectrum is shown in Figs 4 and 5. For comparison, the transmission of a high-quality single-crystal natural Type IIa sample is also shown. For wavelengths beyond the fundamental cut-off at 220nm the transmission rises to reach values close to the theoretical reflection limited value of approximately 71 'YO assuming a refractive index of approximately 2.4 [29,30]. At shorter wavelengths there is a reduction in transmission relative to the Type IIa sample, which can be partly attributed to scattering and, close to the fundamental edge, to a bulk absorption mechanism [311.

2.2 Properties of Chenzical Vapor Deposited Diamond

577

10,000

.-yE

n

8,000

3

4Q

6,000

a

4,000

Y

.w

C

3 0

0

2,000

n (200

1,300

1,400

1,500

1,600

1,700

Raman shift from 633 nm (cm-1) Figure 3. Raman spectrum of an optical grade CVD diamond specimen measured with a Renishaw instrument with an excitation wavelength of 633 nm.

Beyond the intrinsic, multi-phonon absorption bands (between 4000 cm-' and 1500 cm-') [30], the transmission of the CVD diamond specimen reaches the maximum reflectivity limited value (Fig. 5) of 71.4% (using a refractive index value of 2.375 [32,3]) and is identical, within experimental accuracy, to that of high-quality single crystal Type IIa samples measured with the same instrument. 80

Single crystal Type Ha, 0.5 mm thick I

60 C

.-0 .v)

u)

40

E

s 20

n

u

500

1,000 1,500 Wavelength (nm)

2,000

2,500

Figure 4. Transmission spectrum for an optical grade CVD diamond window 1.0mm thick in the UV-visible-near IR spectral range compared to that of a high quality natural Type IIa window 0.5 mm thick.

518


---------.-

1

Optical grade DlAFlLM 1.0 mm thick sample

v) v)

71.4% transmission

L

0

3,000

1,000 Wavenumber cm-I

2,000

-

1

~

500

Figure 5. Transmission spectrum for an optical grade CVD diamond window 1.Omm thick in the infrared spectral range.

Transmission measurements have been recently reported [33] extending the range of the spectrum of Fig. 5 to much longer IR wavelengths (to 20 cm-' or 500 pm) showing that there are no measurable absorption features within that range. The intrinsic absorption at 10.6 pm wavelength of optical grade samples measured by calorimetry is in the range 0.10-.029 cm-' [27,34]. It is thought that some of these values may be affected by surface absorption. This aspect is currently being investigated. As will be discussed below, values of intrinsic absorption below 0.1 cm-' are acceptable for most practical applications, due to the very high thermal conductivity of diamond, coupled with the low change in refractive index with temperature. The low angle scattering of optical grade CVD diamond at ~ 1 pm 0 wavelength has been assessed by measurements of modulation transfer function (MTF) using samples of optical grade CVD diamond with accurately polished surfaces to evaluate one aspect of their performance as imaging windows [4,35]. The MTF is the best parameter available for defining image quality. A modulation transfer function curve effectively shows the contrast in the image as the spatial frequency of that image changes (so that the curve will always drop to zero at a high enough spatial frequency). The MTF of the components was determined from the width of the image of a slit source. Any component which is placed in the light path which has a poor optical performance (whether because of form shape errors or scatter) will increase the broadening of the image of the slit source (the line spread function, LSF). An infrared interferometer was used to determine the LSF due to the form shape error and this can be subtracted from the total broadening as measured in the infrared MTF instrument. What remains is the broadening due to the scatter alone. Figure 6 shows a comparison of the LSF for the system with and without a sample showing that the CVD diamond optical window introduces only a very small broadening. The computer converts these data into an MTF trace such as

2.2 Properties of' Chemical Vapor Deposited Diamond

579

2,500 2,000

e $-= 1,500 b

'i1,000 32

-

FWHM = 115 pm

500 n

-200

-300 -200 -100

0

100

200

300

400

Position (microns)

Figure 6. Comparison of the Line Spread Function measured with and without a sample in the system.

that in Fig. 7 (MTF vs RSF, reduced spatial frequency) which also shows the MTF degradation caused by form shape errors and the M T F of the system with no window in place (the diffraction limited curve). The effect of scatter is therefore reflected in the difference between the lower and middle curves. The conclusion is that optical grade CVD diamond shows minimal scatter and degradation in MTF

diffraction limited 1.s.f. measurement interferometerresults .. . A- - . ... .Q ..

+

Figure 7. Modulation Transfer Function (MTF) measured with a CVD diamond window (lower curve), the M T F degradation caused by form shape errors (middle curve) and the MTF of the system with no window in place showing the diffraction limited case (upper curve).

580


in the far infrared and is therefore a good candidate for windows required in imaging systems. Such an application will be discussed in 2.3.1.

2.2.3 Strength of Chemical Vapor Deposited Diamond C. S. J. Pickles The fracture stress of CVD diamond has been evaluated using the three-point bend geometry [15]. The details of these measurements and its relative merits have been discussed elsewhere [4,35]. The fracture stress is plotted against thickness in Figs 8 and 9 for the growth (coarse grain) and nucleation (fine grain) surfaces respectively, for the optical and mechanical grades of material. All of the samples in this study were in the asgrown state. These strength data suggest several conclusions. First, the measured strength depends strongly on thickness. There is an increase in grain size on the growth side with thickness and it is speculated that the size of the critical flaw is limited by the grain size close to the tested face, as discussed previously [4,25,35]. The variation of the stress level through the thickness is another factor which may influence the thickness dependence of the strength. A more precise analysis has been performed in which the experimental data seem to be in good agreement with the model, as reported elsewhere [35]. Second, the reproducibility of the strength values is high. If the data from samples of thickness between 0.4 and 1.4 mm are assessed, they give the Weibull moduli [ 151 shown in Table 1 (a higher value indicates a more reproducible strength, c.f. sapphire = 2.1, zinc sulphide = 5.4, silicon carbide = lo). These can be corrected to allow for the thickness effect [25,35]. The corrected values for the growth surface are also given in the table. The observations suggest that CVD diamond has bulk (rather than surface)

IGrowth surface strength 1 h

m

Q

8 e c

600

8Optical grade

u)

200 0

0.5

15 2 Thickness (mm)

1

2.5

3

Figure 8. Fracture stress as a function of thickness measured with the growth face in tension.

2.2 Properties of Chemical Vapor Deposited Diamond

58 1

1400 T 1200

-

* loo0

-

2i v)

P

600 8oo

I

400

I

I

I

I

I

strength controlling flaws [26] which are related to the size of the crystal grains.

2.2.4 The Young Modulus Early measurements of the Young modulus on relatively thin specimens (up to 300 pm) gave values in the range 986-1079 GPa [2,36]. More recent results have been obtained on optical and mechanical grade specimens as a function of temperature in the range 20-800°C [37]. For temperature values below 700°C (which is the onset of oxidation in diamond) the Young modulus is relatively insensitive to temperature as shown in Fig. 10. The average value for the Young modulus in optical grade material at room temperature is 1 133 GPa and that of mechanical grades is 1166 GPa. Results published previously using samples of a different origin [38] show that values between 242 and 539 GPa may be exhibited if the CVD diamond contains a high concentration of defects such as micron-size voids or microcracks. This illustrates the great variability in material properties that can be found in CVD diamond of different origins and the need for CVD diamond suppliers to provide precise material specifications to end-users.

Table 1. The Weibull moduli calculated from the data in Figs 8 and 9

CVD diamond type

Growth surface

Growth surface (corrected for thickness variation)

Nucleation surface

Mechanical grade Optical grade

9.8 11.6

11.6 23.1

6.5 11

582


1160

-2 9 -g

1

1140

1120 1100

1080

-& 1060 c

ga 1040 1020 1000

0

100

200

300

400

600

500

700

800

900

Temperature [“C] Figure 10. Young modulus of an Optical Grade CVD diamond specimen in the 20°C to 800°C temperature range [37].

2.2.5 Thermal Conductivity The thermal diffusivity of CVD diamond has been measured over the temperature range 200425 K using the thermal flash technique which measures the thermal diffusivity perpendicular to the plane of the layer (through-the-plane) [l, 31. Figure 11 shows the temperature dependence of the thermal conductivity for different grades of CVD diamond: optical grade, and three grades of thermal I

I

A

Y

4000

E. P

3 3000 u S

U

s 2000

0

150

250

200

300

350

I

I

400

450

Temperature (K) Optical Thermal 1 Thermal 2 Thermal 3 Type 118 natural

0

4

*

-

Figure 11. Temperature dependence of the thermal conductivity for different grades of DIAFILM CVD diamond: optical grade and three thermal grades.

2.3 Optical Applications

583

DIAFILM specimens. These results are compared with literature values of natural Type IIa single crystal diamond [39]. The difference between the thermal conductivity of optical and thermal grade specimens becomes more appreciable at lower temperatures. The optical grade specimen seems to follow the Type IIa trend reaching a value of 4000 W mK-’ at 200 K, probably the highest reported so far for polycrystalline CVD diamond at that temperature. For the ‘thermal-3’ grade, which shows the lowest conductivity and is of dark color, the temperature dependence is not very pronounced, in agreement with previous publications [40,41]. The ‘thermal-1’ grade has a thermal conductivity better than 1800W mK-’ at room temperature which is adequate for most demanding applications, as discussed in 2.5.2. The ability to define different grades of thermal material is an example of the cost-performance optimisation that is possible with CVD diamond.

2.2.6 Dielectric Properties At microwave frequencies in the range 72-145 GHz, the critical parameters for highpower transmission are the dielectric characteristics of the window material: the dielectric loss factor tan 6 and the permittivity E: (or the refractive index n = &:I/*) because they affect power absorption and reflection [42]. The dielectric loss factor tan 6 in low loss samples is usually measured as the decrease in the Q factor of a resonant cavity [43]. Low dielectric loss materials find application as the output windows of high-power microwave tubes. A specific case is that of windows for Gyrotron tubes operating in the 70-170 GHz frequency region with output powers in excess of 1 MW, as will be discussed later. Values of the dielectric loss of CVD diamond have been measured over the past 3 years as a suitable material grade for dielectric window applications was being developed [5]. For open resonant cavity measurements, samples are usually required to be of at least 30 mm in diameter and of thickness in excess of 0.87 mm depending on the measurement frequency and the accuracy required. For recent CVD diamond, values of tan 6 below lop5 have been achieved. A specific example is a window l00mm in diameter and 1.6mm thick which exhibited a tan 6 value of 0.6 (& 0.2) This is the lowest value so far reported for CVD diamond and would enable the material to be used as output windows in Gyrotron tubes of powers in excess of 2 M W as discussed in 2.4.

2.3 Optical Applications In this section, applications will be discussed which illustrate the versatility and advantages of CVD diamond as an infrared and multi-spectral window material. We describe the use of CVD diamond optical elements including CVD diamond domes and flat plates as windows for IR seekers or imaging systems in high-speed flight or other mechanically aggressive environments. Then we describe the use of CVD diamond windows for the transmission of high-power IR laser beams.

584


2.3.1 Chemical Vapor Deposited Diamond for Passive Infrared Windows in Aggressive Environments C. J. H. Wort The extreme hardness and abrasion resistance of diamond makes this material an ideal choice for applications in which the optical components are exposed to aggressive environments. The use of diamond for optical elements for infrared seeking missiles and other military uses constitute probably the best examples of this type of application as discussed in more detail in this section. There are, however, a host of other applications in industry outside the military field in which the attributes of diamond optics are of great advantage. These include windows used in the monitoring of chemical reactions, or in the analysis of fluids which contain abrasive components. Originally conceived during the 1940s to allow armed forces to operate effectively under the cover of darkness, infrared thermal imaging fulfilled a need that could not be met with radar, acoustic, or visible sensors [44]. Thermal imaging relies on sensing the heat emitted by a body, by virtue of its temperature alone; thus thermal imaging is purely passive in its operation. The wavelength of the emitted light is temperature dependent and the IR detector is optimized for maximum sensitivity over a limited wavelength (and hence temperature) range. Infrared detector materials are delicate and require protection from the environment by a transparent window. In many military applications this environment can be extremely harsh, such as the environment encountered by a high-speed, heatseeking missile or a tank window exposed to a sandstorm. The damage caused by a high velocity rain drop impact onto a zinc sulphide window [45] is clearly illustrated in Fig. 12. The erosion/impact damage seriously affects the component strength and also the ability to form an acceptable image behind the window. Owing to the absence of strong absorption by water or carbon dioxide molecules in the atmosphere, two wavebands are basically transparent and are generally used by IR imaging equipment. The mid waveband is between 3 and 5 pm wavelength and the long waveband is between 8 and 14 pm wavelength. For a body at 300 K there is considerably more emission per unit area (termed excitance) in the 8-14 pm waveband than the 3-5 pm waveband (the excitance is about thirty times more). For a body at 600K, the levels of excitance in each waveband are similar but at 2000K the 3-5 pm excitance is a factor of about nine times higher [15]. This is one of the main reasons why the 3-5 pm operating waveband has been favored for heatseeking missiles designed to follow the hot air trail left by a jet engine. The other main reason why the mid waveband seekers are currently preferred is that there are several durable window or dome materials that are transparent in that spectral range, such as sapphire or magnesium oxide [ 151 and that are currently available on a commercial basis. Figure 13 shows typical heat-seeking missiles fitted with 3-5 pm wavelength transparent domes. In contrast, the longer wavelength infrared (LWIR) band is desirable for imaging cooler bodies (such as the aircraft itself rather than just the hot engines, which can be ‘masked’) and although LWIR detector materials and missile seekers have been


585

Figure 12. Example of the damage caused by a high velocity water drop onto a zinc sulphide window (from [45]).

Figure 13. Examples of air-to-air heat-seeking missiles fitted with 3 to 5 pm wavelength transparent domes.

586


Figure 14. Example of a CVD diamond hemispherical dome, 70 mm in diameter and over 1.O mm thick that has been semi-processed on both surfaces.

available for some time, they have not been exploited due to the lack of erosion resistant windows and domes that transmit in the 8-14pm waveband [14]. Materials such as germanium and zinc sulphide, for instance, are transparent in the longer wavelength IR band but even with the best available protective coatings they are not able to withstand the rain erosion encountered during air carriage, let alone a missile flight scenario. Diamond has been recognized to be an ideally suited material for this application [15,14,25] because of its good IR transmission and its extreme mechanical properties. It was only with the introduction of CVD diamond technology and its potential for synthesizing diamond films over large areas that the IR seeker community started to consider the practical possibility of using diamond for this application. One approach has been the use of CVD diamond as a protective coating for other softer IR materials such as ZnS or Ge [46]. However the successful development of thick-film or bulk CVD diamond [2f, 131 motivated active development projects on the synthesis of free-standing CVD diamond domes. This work has met with considerable success [25,47] resulting, for instance, in the synthesis of CVD diamond hemispherical dome geometries 70 mm diameter, 2 1 mm thick that have been semi-processed on both surfaces as shown in Fig. 14. For imaging applications the form shape needs to be very precise and the radii of the two surfaces have been consistently machined to better than fl pm, which is technically difficult to achieve in such a hard material. Figure 15 shows the thermal image of a commercial aircraft through a CVD diamond dome fitted (such as that shown in Fig. 14) to an 8-14 pm thermal imaging camera. There is a small loss of image resolution due to imperfections in both the surface finish (causing scattering) and the form shape of the dome being tested; however, the engines are clearly visible as being hotter than the bulk of the plane, and the plane shape is clearly identified.


587

Figure 15. Thermal image of a commercial aircraft through a CVD diamond dome such as that shown in Fig. 14. The engines are clearly visible as being hotter than the main body and the shape of the aircraft is clearly identified.

Table 2 compares IR window properties of CVD diamond with other, 8-14 pm wavelength transparent materials showing the exceptional optical and thermomechanical properties available with CVD diamond. As in other applications, it is the combination of several of these properties within one window material that makes diamond so special. One example is the remarkable sand erosion resistance of CVD diamond coupled to the high transparency in several spectral regions. The latter property may be important for future military imaging systems since there are considerable advantages in being able to ‘see’ simultaneously in several different spectral regions through a common window. When compared with multi-spectral ZnS (which is the only other truly, multi-spectral material for visible, LWIR, and radar frequencies) it can be appreciated that diamond is not only more transparent but also more resistant to sand erosion by close to a factor of a million times! The extreme sand erosion resistant properties of diamond are graphically illustrated in Fig. 16 which compares the reduction in transmission, as a function of sand erosion time, for CVD diamond and sapphire (which is the most erosion resistant of the multi-spectral, 3-5 pm waveband materials). A potential future application for a multi-spectral window would be on a tank, where both the IR sensing and laser range finding could be performed through the same window as used by the tank driver. Thermal shock of the window (or dome) is an important consideration for the design engineers of future missile systems because of the high thermal gradients generated during rapid in-flight acceleration, and is currently limiting the performance of existing heat-seeking missiles. The combination of a high thermal conductivity coupled with a low thermal expansion coefficient render diamond virtually immune to thermal shock [15]. Using a thermal shock figure of merit as defined in note 3 of Table 2 diamond is approximately 200 times better than sapphire, which is

588


Table 2 Property comparison between high quality CVD diamond and alternative window materials that transmit in the long wave (8-14 pm) thermal imaging region Property

CVD Diamond (high optical quality)

ZnS (FLIR grade)

ZnS (Multispectral)

Germanium

Vicker's hardness (GPa) Fracture toughness (MPa.m'/') Young modulus (GPa) Poisson ratio Tensile strength (MPa) Rain impact DTV (m s-I) 2 mm drop size Sand erosion (mg kg-I) @ 80 m s-' C25/52 sand Sand erosion (mg kg-I) @ 34 m/s C25/52 sand Thermal conductivity @ 300K (W mK) Thermal conductivity @ 750 K (W mK) Thermal expansion coeff. (p.p.m./K) at 273 K Thermal expansion coeff. (p.p.m./K) at 473 k Thermal shock (mild) FOM ( x lo ' w m - ') (3) Refractive index at 10.6 pm dn/dT (K-') (at 300 K) Dielectric const. D (35 GHz) YOincrease in D at 773 K (Yo) Loss tangent at 35 Hz 10.6 pm abs coeff. (cm-') Max. window temperature (K) for extended operation in air at 10.6 pm wavelength ( I ) Useful transmission range (pm)

81 18 >6 1000-1 100 0.1 (assumed) 35&800 525 (>1.25mm thick) 0.18

2.3 0.81 74.5 0.29 103 170

1.5 1 74.5 0.28 60 165

8.5 0.57 103 0.28 90 175

19,000 (2) 156,750 (2)

12,382 (2)

Negligible

950

7,840

1,730

1800-2200

19

27

59

*

1,200

-

0.9

6.6

6.3

6.1

2

7.7

7.5

-

2.6

2.1

6.1

2.19 41 x 10 8.35 -8

2.19 36 x 8.39 -8

4 400 x -17

-

-

-

0.2 573* 873**

0.2 573% 873**

0.02 343***

-

1000

2,376 9.6 x 5.68 0.15 4.3 <0.0002 0.03-0.1 1023**

*

0.3 to 3.0 and 7.0 to >SO0

7.7 to 11.2 0.4 to 11.5

-

1.8 to 23.0

(1) *** Band cdge moves towards shorter wavelength,introducing significant absorption. Mass loss due to oxidation. *** Free-carrier absorption makes material opaque. Calculated sand erosion mass loss values at 80 m/s for ZnS and Ge (to allow a comparison (2) with CVD diamond) are based upon up-scaling theories (using erosion rate velocity scaling factors) from work undertaken at the Cavendish Laboratory [45]. FOM = (Material strength x (1 - Poisson's Ratio) x thermal conductivity/thermal expan(3) sion coeff. x Young modulus) [IS].

generally considered to be the 3-5 pm material which is most resilient to failure by thermal shock. The above considerations illustrate the versatility of CVD diamond technology in the ability to synthesize three-dimensional shapes of high quality diamond and


589

% of rnax. Transmission retained

90 \

CVD diamond, growth surface

70

40

30

I’

0

,

sapphire

I

I

1,000

500

I

1,500

I

2,000

2,500

Exposure time (s) Figure 16. Reduction in transmission as a function of sand erosion time for a CVD diamond window compared to a sapphire window.

explains why CVD diamond is emerging as one of the preferred material options for applications requiring multi-spectral transmission in harsh environments.

2.3.2 Windows for High-power Infrared Lasers Considerable progress has been achieved in recent years on improving the optical properties of CVD diamond and in making this material available in large sizes and with reproducible and consistent properties [4,5,35] so that its use in actual laser systems has become a practical reality [48]. This section reviews the technical issues relevant to the performance of windows for high power lasers comparing the specific case of CVD diamond and ZnSe for the transmission of C 0 2 laser beams. A major requirement for windows used in the transmission of high power beams is the reduction of wave-front distortion which may result in defocusing of the beam and loss of power density. There are two major sources of wave-front distortion: thermal effects and geometrical or form-shape aberrations [49]. 2.3.2.1 Thermal Effects S. E. Coe A window can absorb heat from the laser beam by intrinsic bulk absorption or at the surfaces (through intrinsic surface absorption mechanisms or at the anti-reflection (AR) coatings). The absorbed heat induces radial and axial temperature gradients which in turn result in two distinct forms of optical distortion:

590


Table 3. Comparison of properties between CVD diamond and ZnSe Property at 300 K

Diamond (optical-grade DIAFILM)

ZnSe

Thermal conductivity (W mK-') Thermal expansion (p.p.m. K-') Thermo-refraction (dn/dT) ( T - ' ) Fracture strength (MPa) Weibull modulus Young modulus (GPa) Typical thickness (mm) used for a 25 mm diameter window Bulk absorption coefficient (cm-') Total absorption (AR/AR coated, 0.1 % absorption per surface) (YO)

1800-2100 0.9 1x 400-600 11-23 > 1050 0.7-1.0

16 1.6 5.8 x 1 0 - ~ 55 5 70 6

0.1-0.03 0. 9

-0.0005 0.23

(i) The thermo-refractive effect is the result of changes in refractive index with temperature and is described by the thermo-refractive coefficient dnjdT, where n is the refractive index of the window material. (ii) Thermo-elastic effects which are a caused by bulging (or other form-shape distortions) or by changes in the optical constants by thermal stress gradients through the photo-elastic effect. Materials which have relatively small thermal gradients and do not change geometry with temperature will show the lowest values of thermal distortion. Critical parameters of relevance in these effects are thermal conductivity, thermal expansion coefficient, thermo-refractive constant, bulk absorption coefficient, absorption at the AR coatings and the photo-elastic constants. The total wave-front distortion can be described by the variation in phase (or optical-path-length) of a transmitted beam with radial position. For the thermo-refractive effect, for instance, the phase change will be (dn/dT) x t x A T (where t is the thickness of the window and AT is the temperature change) and is thus smaller for a thinner window. The minimum acceptable thickness of a window will be determined largely by the ability to support vacuum or pressure without breaking or distorting [50]. Critical parameters in these cases are the fracture strength and the Young modulus respectively. The current standard material for optical elements in C 0 2 lasers is ZnSe [27] because of its very low intrinsic absorption at 10.6 pm [14, 15,341. Table 3 compares the critical parameters of CVD diamond and ZnSe showing that CVD diamond, for the reasons discussed above, has the potential to handle much greater beam powers. This has been long recognized [51] and the thermal effects in diamond laser windows have been theoretically modelled in previous work [27,51-531. Some of the earlier results [5 1-53] were derived before reliable data were available on the properties of polycrystalline diamond and were therefore very speculative. The following is an up-date of the predicted thermal effects in optical grade CVD diamond and ZnSe windows where some of the earlier calculations arc revisited. The analysis presented here assumes that bulging and thermo-elastic effects in either diamond and ZnSe can be neglected [27,52-541. The assumed configuration for each case is that of a circular window, 25mm in diameter, which is edge

2.3 Opticul Applications

59 1

Figure 17. Schematic of the axially symmetrical temperature gradients generated in an edge-cooled window fitted with AR coatings when traversed by a laser beam.

cooled at a fixed temperature resulting in an axially symmetrical temperature profile as shown in Fig. 17. An AR coating with a total absorption at 10.6 pm of 0.1 % per surface is assumed for both windows. In the case of ZnSe, the absorption at the AR coating dominates, leading to the highest temperature being at the surface. Because of the difference in mechanical strength and Young modulus (see Table 3), it is expected that the thickness of CVD diamond windows could be made thinner than those of ZnSe. Using the criterion that the thickness should be sufficient to prevent fracture, the thickness-to-diameter ratio must be [35,50,55]: t/D

> 0,5(kjP/S)'/*

(1)

where:

k

constant determined by the boundary conditions and dependent on the Poisson's ratio t = sample thickness D = sample diameter P = pressure differential f = safety factor (depends on the value of the Weibull modulus) S = fracture strength. =

Using the criterion of acceptable deformation imposes the condition [50,52]: t/D

> l.OI{(n

where E is the Young modulus.

-

1)(P/E)2(D/h)}'/2

(2)

592


Using Eq (l), the ratio of the CVD diamond-to-ZnSe thickness (for the same diameter) is: (3) { (kCVDafCVDDSZnSe) / (kZnSefZnSeSCVDD) 1112 Using the parameters listed in Table 3, assuming that the k parameter is similar in both materials and using the same safety factor for ZnSe and CVD diamond results in tCVDD/tZnSe z 1/3. Equations (1) and (2) predict a minimum t / D ratio of 0.017 for CVD diamond (using a safety factor of 4 and a fracture strength value of 400 MPa) or a thickness above 0.4mm for a 25 mm diameter window. In practice, it is found that a better flatness and overall shape stability can be achieved if the thickness is close to 1.Omm. For these calculations it will be assumed that the CVD diamond window is 1.O mm thick and the ZnSe 6.0 mm, which is the typical thickness used for ZnSe windows in the industry [27]. The higher thickness used for ZnSe may be partly justified by the need to use a higher safety factor due to the lower value of its Weibull modulus [35]. The absolute bulk absorption of optical grade CVD diamond windows has been found to be in the range 0.1--0.03cm-' [27].A value of 0.05 cm-' has been assumed for this calculation. The analysis has been done for a continuous wave, 5 kW incident beam that is assumed to have a Gaussian profile with a 14mm diameter at 14% (lie2) of maximum intensity. For this level of power the temperature increases are relatively low and the temperature dependence of parameters such as dn/dT [54] have been ignored, but the temperature dependence of the thermal conductivity for optical grade CVD diamond (Fig. ll), although not very pronounced, has been taken into account. The analysis solves the fundamental heat propagation equations using a twodimensional computational model in which values of the radial and axial variations of the temperature and the refractive index are calculated. The calculated temperature gradients for the two cases are shown in Fig. 18. Figure 19 shows the radial dependence of the average (across the thickness) temperature. The temperature in CVD diamond shows a maximum variation (center to edge) of 3.4"C compared to 23.4"C for the ZnSe case. The temperature variations over the effective diameter of the beam are 1.7"C and 12.4"C for the CVD diamond and ZnSe windows respectively. The phase or optical path length variation A@ = (dn x t ) is a function of radial position in the ZnSe and CVD diamond windows and is shown in Fig. 20. The variation for CVD diamond over the effective beam width is over a factor of 250 smaller. Three factors combine to yield this large difference: the smaller temperature rise in CVD diamond due to the higher thermal conductivity ( A T a factor of more than 7 smaller in CVD diamond), the smaller value of dn/dt for diamond (a factor of 5.8), and the thickness of the window, which for CVD diamond is only 1/6th of that in ZnSe. For an example similar to that shown in Fig. 18 the thermal lensing effect in the ZnSe window was found to be equivalent to a lens of 3-5 m focal length, whereas for the CVD diamond window the effect was negligible [27,34]. As a general rule it can be assumed that the lensing effect can be neglected if the difference in A@is smaller than hjl0 [51],or 1 pm for a C 0 2 laser beam. As shown in

/

tCVDD tZnSe =


593

Figure 18. Calculated temperature gradients in a CVD diamond and a ZnSe windows, edge cooled and with AR coatings.

Fig. 20, A@is approximately 8 pm for ZnSe but under 0.04 pm for CVD diamond. Recent results have been reported using electronic speckle pattern interferometry (ESPI) [56] using a C 0 2 laser of power up to 1 kW on a CVD diamond and a ZnSe windows of 25mm aperture and AR coated on both sides. The ZnSe window starts to deform at moderately low power levels (50W). In the CVD diamond window discernible distortions start at powers over 700 W. These

0

1

2

3

4

5

6

7

8

9 1 0 1 1 1 2 1 3

Radial Position (mm) Figure 19. Radial dependence of the average (across the thickness) temperature for the CVD diamond and the ZnSe windows derived from the calculations shown in Fig. 18.

594


10

8

--...._ -...--. .a

h

*%.

fn C

$ 6

.-

E x 4

--...... *.'

..%. *%.

Y

CI

C

rn 2

-*...

-*...+.-

5.

-... -*..

0

....*-. I

.%

measurements predict a deformation in the ZnSe window for a beam power of 5 kW of approximately 5pm, in rough agreement with our previous calculation. The equivalent deformation measured in that experiment for a 1 kW beam was at least one order of magnitude lower in the CVD diamond window than in ZnSe. Using parameters similar to the ones used in the above modeling, Wild [54] estimated that a CVD diamond window will introduce an optical distortion of 1 pm for a laser beam of at least 160 kW and at that level of incident power, the generated thermal stress in the window should not exceed 100 MPa, a factor of at least four below the fracture strength of optical grade CVD diamond < l m m thick [25,35,54]. From this perspective it could be concluded that the power handling capability of an edge cooled CVD diamond window l.Omm thick is of at least 160 kW before serious thermal degradation is observed. However, this may be an oversimplification because it ignores possible laser damage effects. There is evidence [57] that for CW beams: (i) the laser induced damage threshold (LIDT) is a decreasing function of total power, in fact the power damage threshold scales with diameter rather than with area, and (ii) LIDT can be considerably reduced due to surface imperfections such as pits or digs. The first is a heating effect and it can be expected that, because of its high thermal conductivity, diamond will outperform most other materials. For ZnSe, LIDT values of 3 kW mm-' have been measured [54]. Since temperature excursions in diamond are at least a factor of seven lower than in ZnSe (Fig. 19) a thermal LIDT of at least 20 kW mm-' can be expected for CVD diamond (probably much larger because of the higher mechanical and chemical integrity of diamond compared


595

Figure 21. lnterferometric surface profile of a 25mm CVD diamond window showing a flatness better than one fringe (633 nm illumination).

with ZnSe). For a 160 kW beam, this implies a minimum beam diameter of 8 mm, considerably smaller than actual beam sizes in lasers of this power level. What has not yet been investigated is the potential decrease in LIDT that could be introduced by surface and bulk features typical of polycrystalline CVD diamond [2,35]. For pulsed lasers, considerably higher LIDT values have been reported [58]. Clearly, more experimental work is required before the true power handling capability of CVD diamond is truly understood.

2.3.2.2 Geometrical Distortions: Window Flatness C. S. J. Pickles Beam distortion can be generated when the form shape of the surfaces of a window deviate from perfect flatness. The flatness of optical elements as measured by a interferometer is expressed as the number of fringes (twice the wavelength of the light beam probe) shown in the interference pattern of the optical element placed against a perfect flat surface [59]. Depending on the application, an acceptable number of fringes is specified. Applications in laser optics often require a flatness of < 4 fringes and sometimes < 1 fringe using visible illumination (typically 633 nm wavelength). Figure 21 shows the interferometer pattern of the surface of a CVD diamond Window which has been polished to a flatness better than 1 fringe (633 nm illumination). This is typical of what can be achieved presently for CVD diamond showing that current technology is sufficient to address applications such as exit windows, where a flatness of 1 to 3 fringes is considered adequate. Better results have been achieved using techniques currently under development as shown in Fig. 22 where a flatness of 0.5 fringe has been achieved in a window 48mm in diameter. This

596


Figure 22. lnterferometric surfxe profile of a CVD diamond window showing that by the use of an improved process, a flatness of 0.5 fringe can be obtained over a 48 mm diameter.

suggests that CVD diamond will be able to address a wider range of applications as an optical element in laser systems. The ability to process CVD diamond to controlled form shapes has also allowed the production of samples with wedge angles up to 1 degree, as required to eliminate interference fringes for broad band IR transmission, as is the case in synchrotron beam lines [4,33].

Figure 23. The Rofin Sinar ‘Diffusion cooled’ DC-series COz slab laser with output power up to 3500 W [60].

2.4 Windowsf o r High Power Gyrotron Tubes

597

,Bending mirror

Figure 24. Schematic diagram showing the location of the CVD diamond window in the ‘Diffusion cooled’ Rofin Sinar DC-series COz slab laser [60].

2.3.2.3 Scattering Scattering in the infrared has been discussed in 2.2.2 and by Union et al. [34] con0 a good quality optical grade CVD diamond window does cluding that, at ~ 1 pm, not show a level of scattering that is likely to affect its infrared imaging properties and is capable of transmitting a diffraction limited image. For systems operating at wavelengths in the UV-visible spectral range, scattering effects may be more pronounced, although recent measurements indicate that a good quality CVD diamond window could exhibit values of scattering at 633 nm similar to that of a single crystal Type IIa specimen [54]. CVD diamond windows are now being used in commercial high power COz lasers. An example is the Rofin Sinar Diffusion-cooled C 0 2 slab laser with an output power of 3500 W [48] shown in Fig. 23. In this laser, the CVD diamond window is used to seal the slab cavity, as shown schematically in Fig. 24.

2.4 Windows for High Power Gyrotron Tubes Measurements of dielectric loss in high quality CVD diamond at millimetre wave microwave frequencies, reported in 1993 [2], furnished for the first time clear evidence that the loss tangent of CVD diamond could be comparable or lower than conventional dielectrics such as sapphire or boron nitride. This initial data and subsequent first dedicated dielectric property studies [61] intensified the considerable amount of interest that had already existed in the nuclear fusion community for CVD diamond as a high power window material [62] especially for the development

598

2 Applications of’Diamond Synthesized by Chemicul Vupor Deposition

of Gyrotron tubes [63] for the first ‘International Thermonuclear Experimental Reactor’ (ITER). This section will discuss why, once other advantages of diamond are taken into account (high thermal conductivity, radiation hardness, the relative insensitivity of the dielectric properties with temperature), CVD diamond is being considered not only as an alternative option but probably as the only practical solution for high power Gyrotron tube and Torus windows in ITER.

2.4.1 Window Requirements It is believed that electron cyclotron heating (ECH) at frequencies in the range 7& 170 GHz is one of the major candidates for heating, current drive, and start-up of plasmas in thermonuclear reactors such as the ITER Tokamak or the Stellarator W7-X [64]. The preferred option to supply this source of microwave power is Gyrotron tubes [63] which are currently being developed with an output CW power of at least 1 MW [65]. Figure 25 shows an example of a 1.3 MW, 140 GHz Gyrotron tube being developed at the Forschungzentrum Karisruhe (FZK) [66]. A key component in these tubes is the exit window which should be able to carry high output powers.

Figure 25. Example of a 1.3 MW, 140 GHz Gyrotron tube built at the Forschungzentrum Karlsruhe

WI.

Additionally, the barrier windows in the Plasma Vacuum Torus should be able to withstand high doses of radiation without degradation. The critical parameters for high-power windows are the dielectric characteristics of the windon, material;;he dielectric loss-factor tan 6 and the permittivity E:. (or the refractive index 11 = E:.' ' - ) because they affect power absorption and reflection. The power absorption coefficient 'x is related to tan 6 by [43] 'x =

2x11 tan ijj h

where h is the wavelength of the microwave radiation in free space. To minimize power reflection at a given frequency the thickness t of the window needs to be a11 integer number of half dielectric wavelengths [64] t = Nhj21l

where N is an integer. This requirsment can be relaxed if a Brewster angle window is used. but the penalty is that a much larger diameter window is necessary [67]. The requirement of CW operation puts extremely high demands on the material properties of the dielectric vacuum barrier window that serves as both the primary tritium containment boundary and as the output window of the Gyrotron. The window options available to Gyrotron manufacturers have been discussed in detail by Th~iiiiiiiet ril. [64]. From this analysis it is appreciated that all other options apart from diamond require sophisticated and, in some cases, impractical cooling system involb ing, for instance. cryogenically edge-cooled windows, liquid-surfacecooled double-disc windows or distributed windows i n which segments of dielectric are metal bonded to microchannel cooled metal ribs. Another shortcoming of some of the materials considered. such as sapphire or gold doped silicon, is the pronounced temperature dependence of the loss factor and refractive index, which may lead to thermal runaway and lor detuning of the window resulting in high levels of reflected power. This background explains the keen interest in diamond as an alternative material. Theoretical thcrmal modeling results such as that illustrated in Fig. 26 [66] show that a simple edge cooled diamond window approxiinately 100 mm in diameter and with a value of loss tangent below 1 O-'. should be able to withstand powers in excess of 1 M W with a temperature rise at the centre of the window below 240 C, which is far below the temperature (600-C)at which diamond starts to degrade by oxidation, although thermal stress may be a limiting factor at these temperatures. This relatively low temperature excursion is a direct consequence of the high thermal conductivity of diamond. Table 4 compares properties of diamond relevant to this application with those of other competitive materials.

2.4.2 The Development of Chemical Vapor Deposited Diamond Gyrotron Windows J . R. Brandon The detailed requirements for the use of diamond as a window material in high power Gyrotron tubes can be listed as follows.

600

2 Applications of’ Diamond Synthesized by Chemical Vapor Deposition

Frequsncy= 140 GHz

Powar=I,ZMW l h i i s s = 1.748 mm Waveguide radius = 40 mm C m l i = 293 K: 12000 W,/

Boo

580 5M)

510 520

sw 480 460 440 420 4lm

380 360 340 320

3w 0.00

Figure 26. Results from a theoretical thermal model showing the predicted temperature rise at the center of an edge cooled diamond window for a 1 MW incident microwave beam. Results are shown for different values of dielectric loss and for two different cooling diameters assuming an exposed aperture of 100mm diameter [66].

A reproducible value of loss tangent below lop4 (from modeling results as in Fig. 26) achieved in discs of at least l00mm in diameter and up to 2.2mm thick with reasonable radial and axial uniformity. The thickness requirement is dictated by the need to minimize power reflection as discussed above and by the need to support vacuum with a reasonable safety factor. The variation with temperature of the dielectric constants (tan 6 and n ) needs to be very small in the temperature range of interest (room temperature to approximately 200°C) to minimize the risk of thermal runaway or the detuning of the window for minimum power reflection. The diamond windows need to be bonded to vacuum-tight and high-temperature bakeable metal flanges for mounting to the Gyrotron tube and reactor ports. The dielectric properties need to be relatively insensitive to radiation damage. This requirement is critical for the tritium barrier window that will be exposed to a radiation field of both neutrons and gamma radiation [68]. following summarizes some of the Drotzress achieved to date partly as part of a collaboration between De Beers IndistrTal Diamonds and t h e FZK-ITP, that show convincingly that CVD diamond is indeed a very attractive option for this application.

2.4 Window~sjorHigh Power Gyrotron Tubes

60 1

Table 4. Properties of candidate Gyrotron window materials at room temperature. Property

Diamond Sapphire CVD P.C. S.C.

Silicon BN CVD Au-doped S.C. - p.c.

> 1800 400-600 3.5 0.5 >I050 1.1 5.67 0.8-10 small

150 3000* (170) 2.3 0.7 190 2.5 11.7 0.35 steep

-

Thermal conductivity (WmK-’) Fracture strength (MPa) Density (gcm-3 Specific heat capacity (J gK-’) Young modulus (GPa) Thermal expansion coefficient K-’) Permittivity (at 145GHz) Dielectric loss (at 145GHz) tan 6 Temperature dependence of dielectric properties

40 410 4 0.8 385 5.5 9.4 20 steep

50 80 2.3 0.8 70 3 4.7 1 I5

P.c.: polycrystalline, s.c.: single crystal * The fracture strength of conventional single crystal silicon is 170 MPa [36]. The value of 3000 MPa has been reported after a special surface treatment [76].

2.4.2.1 Dielectric Loss Table 5 shows the dielectric loss values that have been achieved for different types of CVD diamond windows. The four samples DB 1 to DB 4, of diameters between 30 and 40mm, were the result of a development effort to achieve the lowest value of dielectric loss in specimens of the required thickness. The results achieved demonstrate that values of tan F below lop5 are possible in specimens of thickness in excess of 1.7mm. These values of loss are close to the sensitivity of the measuring technique for specimens in this thickness range. When the four samples were stacked together, the more accurate loss value measured was tan 6 = l ( f 0 . 3 ) x lop5 [69]. Samples DB 5 , DB 6, DB 7 and DB 8 are full size (96120mm diameter) windows, synthesized under different conditions. Sample DB 5 (named ‘Super FZK’ elsewhere [70]) is l00mm diameter and demonstrates that the low values of dielectric loss measured in small diameter Table 5. Dielectric loss of CVD diamond window samples. ~

Sample

DB- I DB-2 DB-3 DB-4 DB - 1 to 4 (stacked) DB-5 ‘Super FZK’ [70] DB-6 ‘Star of FZK’ [70] DB-7 DB-8

Diameter (mm)

Thickness (mm)

Dielectric loss tan 6 ( m 5 ) (at 145GHz)

I00

1.8 1.79 1.74 1.72 7.06 1.6

120

2.2

100

1.85 2.23

0.8 0.8 0.8 0.8 1.o 0.6 (10.2) (70 to 370 K) 2 (70 to 370K) 2 13-15

50 30 30 30

96

602


Figure 27. Spatial distribution of the dielectric loss in a 100mm diameter CVD diamond window 1.6mm thick measured with a spatial resolution of 6mm. The isolated points of higher loss at the edge may be caused by scattering losses. Over most of the area the measured dielectric loss is equal which is close to the sensitivity of the measuring technique. This may account or below tan 6 = for some of the measured isolated spurious higher values of loss.

specimens (DB 1 to 4) can also be achieved in full size windows. The spatial uniformity of the dielectric loss in this window is shown in Fig. 27 which shows a map of tan 6 measured with a beam diameter of 6 mm [69]. It can be seen that over most of the active aperture of the window (a central area of approximately 80mm diameter) tan 6 is below lop5 and close to the sensitivity of the measuring technique, which may account for spurious isolated higher values of loss. The temperature dependence of this window over the 70 to 370K temperature range indicates that the loss is largely independent of temperature over this range and that the average [70], which is the lowest loss tangent so tan 6 exhibits a value of 0.6 (* 0.2) far reported for a CVD diamond window. Sample DB 6 is probably the largest Gyrotron window ever made (named 'Star over the same temperature range as of FZK') and exhibits a loss of 2 x DB 5 [70]. The window DB 8, which has a relatively high loss tangent of over 13 x lop5,was the first CVD diamond window to be mounted in a high power Gyrotron tube (shown in Fig. 28) at the Japan Atomic Energy Research Institute (JAERI) [71731. This window was tested in 1998 using the 170 GHz JAERI/Toshiba Gyrotron at output powers of 0.5 MW for 8 s with a water cooled edge showing a temperature increase a t the center of not more than 130°C,consistent with theoretical modelling

2.4 Windo~vs f o r High Power G y o t r o n Tubes

603

Figure 28. The first high-power Gyrotron (JAERI/Toshiba, 170 GHz) fitted with a CVD diamond window. Measurements using beam powers up to 1 MW for 10 seconds have already been performed.

[73]. The second window for the JAERI Gyrotron (DB7) has a lower loss tangent of approximately 2 x lo-’. This window was tested at the end of 1999 using also the 170GHz Gyrotron for powers up to 1 MW for 10s. The temperature increase at the center was less than 3 0 T , also in very good agreement with theoretical modelling [73]. Because diamond reaches almost steady state for times in excess of 3 s the above experiments show convincingly that CVD diamond windows are able to transmit millimeter wave powers of at least 1 MW continuous wave. At the moment of going to press, in excess of 20 full size CVD diamond windows have been delivered to members of the international nuclear fusion community. These exhibit routinely values of loss tangent of 2-3 x lo-’. It is not clear yet if the low values of dielectric loss of samples DB 1 to DB 7 represent a fundamental limit of a loss mechanism in diamond or if they are still affected by impurities or defects in the polycrystalline CVD diamond structure. From measurements in previous samples [74], there is clear evidence that, in samples of relatively low quality (tan 6 greater than 70 x darker regions of the sample have higher values of loss. Even in the highest quality polycrystalline CVD diamond, such as that of the sample shown in Fig. 27, it is possible to observe localized features related to the grain structure of the layer. It can be speculated that these features may be a contributing factor to the dielectric loss of polycrystalline CVD diamond. However, from a practical point of view, this is probably not of great concern in the foreseeable future since it has been estimated that diamond windows with values of tan 6 below 2 x lo-’ (a factor of two above that of the window shown in Fig. 27) should have a power handling capability of at least 2.5 MW [64].

604

2 Applications of' Diamond Synthesized by Chemical Vapor Deposition

10

7 < 0

5

s

2

$

1

* E C

m I-

0.5

v) v)

0

2 A

0.2

Q,

ii

0.1 0.05 1

100 150 200 250

300 350

1

0

TtKl Figure 29. Temperature dependence of tan 6 for a CVD diamond sample measured at 145GHz compared with sapphire [75].

2.4.2.2 Temperature Dependence of Dielectric Constants Preliminary measurements of the temperature dependence of the dielectric properties of polycrystalline CVD diamond are available. Figures 29 and 30 show the temperature dependence of tan 6 and EL measured at 145GHz for a CVD diamond sample compared to sapphire [75].In this temperature range (100-370 K), the dielec-

2.4 Windowsf o r High Powcu Gyrotron Tubes

605

Figure 31. Examples of CVD diamond windows mounted onto vacuum flanges. The two at either end show 100 mm diameter windows mounted to conventional CONFLAT vacuum flanges. The one in the middle illustrates a double flange configuration designed specifically for Gyrotron tube assemblies. All these vacuum assemblies have been tested to be vacuum tight to better than 10-9 mbar I after thermal cycling up to 450°C. s

C

'

tric constants of diamond are largely insensitive to temperature, in contrast to sapphire. In 1999 the FZK team extended the measurements of dielectric loss to temperatures up to 700K [75] showing that in a high quality CVD diamond window the loss tangent at 700K is not more than 5 x lop5 from a value at 300K of 3 x lop5. The results described in 2.4.2.1 indicate that even with transmitted powers of up to 1 MW the temperature increase is likely to be considerably less than 100°C and therefore there is no major risk of thermal runaway.

2.4.2.3 Effects of Radiation Measurements of the effect of radiation on the dielectric properties of CVD diamond have been reported usin neutron irradiation experiments with fast neutron fluences up to at least 10" n m p 2 (energy >0.1 MeV) [65]. Differences before and after irradiation were found to be more pronounced at lower frequencies. At 145 GHz specimens that started with values of tan 6 of 2 x lop5maintained these levels (or even showed a decrease in loss). Further experiments to extend radiation fluencies to lo2*n mp2 are in progress [68].

2.4.2.4 Mounting Techniques A metal-to-diamond bonding technique has been developed to attach CVD diamond windows to vacuum flanges or other assemblies. This is required to mount the windows to the Gyrotron tubes or reactor ports via suitable waveguide sections. Figure 31 shows examples of three l00mm diameter CVD diamond windows mounted to two different types of flanges. The two at either end of the picture are

606


Figure 32. Window disc with slots to reduce hoop stresses and to facilitate bonding alignment

test mechanical grade specimens mounted to conventional CONFLAT vacuum flanges. The one in the middle illustrates a double flange configuration designed specifically for mounting on to a Gyrotron tube allowing for water cooling of an exposed section of the edge of the window. This bonding technique is based on an aluminum alloy and has been demonstrated to withstand thermal cycling at temperatures up to 450°C without degradation of the vacuum integrity (measured at better than lop9 mbar 1 spl [64]. Figure 32 shows a new development in which slots are cut at the edge of the window to reduce hoop stresses and to facilitate bonding alignment. The results described above indicate that CVD diamond windows are very close to meeting all the requirements listed at the beginning of this section, and promise to be the preferred option for high power Gyrotron windows for ITER and other experiments.

2.5 Thermal Management of Laser Diode Arrays Natural Type IIa single crystal diamond heatsinks have been used for the thermal management of microwave and laser diode devices for well over a decade [17, 181. The restriction in the size of available natural diamond has, however, limited the use of diamond to the heat management of small, discrete devices such as single junction laser diodes. The commercial availability of large CVD diamond plates has opened a host of new possible applications in which diamond can be used in the heat management of larger electronic and optoelectronic devices [ 1,77,78].

2.5 Thermal Management of Laser Diode Arrays

607

2.5.1 Laser Diode Arrays: General Issues A specific example is the use of CVD diamond to improve the heat dissipation in laser diode arrays (LDA) [78,79]. In these devices a set of laser diode junctions are monolithically integrated on a single chip. A typical LDA bar is 0.1 mm in height, the resonator length of each junction between 0.3 and 2mm, the junction width between 0.1 to over 0.2 mm and the total length of the array between 5 and 15mm to accommodate 5-75 single emitters side by side along the length of the bar [80]. During operation of a LDA, more than half of the electrical input power is dissipated in heat which needs to be efficiently removed in order to reduce the junction temperature and thus prolong the life of the device. Typical operating junction temperatures are 50-55°C and raising this temperature by only 10°C results in a reduction by a factor of two in lifetime [79]. The thermal resistance of a diode laser bar depends on the area size of the active junctions. Mounting the diode laser bar on a high thermal conductivity material serves to increase the effective area of the junction before contacting the water cooled, copper heat-sink. These submounts are therefore heat-spreaders, rather than heat-sinks [l]. A typical LDA requires a heat-spreader submount over 10 mm in length, which is well within the current capability of CVD diamond technology but would have been almost impossible to implement by the use of natural diamond. The other advantage of using electrically insulating materials, such as CVD diamond, is the possibility to address electrically single emitters of an array individually for applications such as marking and printing [80]. The efficiency of a heat-spreader will be a function of its intrinsic thermal conductivity and of its actual dimensions and is measured by the decrease in the total thermal resistance of the package, equivalent to the total reduction in temperature of the device junction for a given dissipated power. There are a number of issues that need to be addressed before CVD diamond can be used effectively as a heat-spreader submount. (i) (ii) (iii) (iv)

Optimization of submount geometry and thermal conductivity. The flatness of the submount, to ensure uniform thermal contact. Metallization issues such as thickness and thermal stability. Thermal stress and expansion mismatch.

2.5.2 Modelling of Submount Heat Resistance S. E. Coe Numerical simulations of the junction temperature increase of a laser diode array have been performed for a package configuration as illustrated in Fig. 33. Calculations have been performed with and without a CVD diamond heat-spreader. The decrease in thermal resistance has been calculated for different values of the heat-spreader thickness, depth and thermal conductivity. The simulations have been performed for an array of laser diodes which are individually 200pm wide

608


Figure 33. Package configuration analysed in the work consisting of the laser diode array mounted on a CVD diamond spreader which is mounted on a copper heat-sink block.

and separated by a spacing of 200pm. The length of the heat-spreader has been assumed to be constant at 11mm. This is a three-dimensional calculation which simultaneously takes into account heat dissipation parallel and perpendicular to the direction of the LDA bar, in contrast with previous two-dimensional calculations reported in the literature [79]. Figures 34 and 35 show that the thermal resistance decreases as a function of heatspreader thickness and depth. Heat dissipation through the rear (or depth) of the heat spreader is important, because in an array of diodes there is thermal cross-talk between junctions in which the temperature of one junction is affected by that of adjacent junctions. Figure 34 shows that heat-spreader depths between 2 and 3 mm are required for a thickness of 300pm. Also, because of thermal cross-talk, the c

'

0.2 0

I

100

I

1

I

I

I

300 400 500 600 Thickness of CVDD layer (microns)

200

700

Figure 34. Thermal resistance decrease as a function of the thickness of the CVD diamond heat spreader for the configuration shown in Fig. 33.

2.5 Thermal Management of Laser Diode Arrays

8c 0.32 .-4 3 0.28 p?

-m

c

609

~

'

0.24

I-

'

0.2 0

0.5

1

1.5

2

2.5

3

3.5

4

Length of CVDD (mm)

Figure 35. Thermal resistance decrease as a function of the depth of the CVD diamond heat spreader for the configuration shown in Fig. 33.

influence of heat spreader thickness is not as pronounced as would have been expected in a single diode junction and very little gain is achieved for a thickness above 300 pm, as shown in Fig. 35. Figure 36 shows the decrease in thermal resistance as a function of the thermal conductivity of the spreader, showing that, for a total dissipated power of 90 W (or 3.6 W per junction), a reduction close to 5°C is obtained by increasing the thermal conductivity from 1000 W m-' K-' to values close to 2000 W m-' K-I. This reduction in temperature is significant in view of the increase in life of a factor of two expected for a decrease on lO"C, as mentioned above. As discussed in 2.2.5, values of thermal conductivity of 1000W m-' K-' are typical of the lower thermal grade whereas the higher grades exhibit a thermal conductivity close to 2000 W m-' K-l (Fig. 33). The result of Fig. 36 shows, therefore, that for this application it is advantageous to use a higher thermal conductivity CVD diamond grade.

18'

500

1000 1500 Themal Conductivity (WlmK)

2( 10

Figure 36. Decrease of the thermal resistance as a function of the thermal conductivity of the CVD diamond heat spreader.

610


These results indicate that an ideal CVD diamond heat-spreader would reduce the thermal resistance of a 25 element array by approximately 36% from 0.36 K W-' to 0.23 K W-I. Experimental results [79] show that, in a similar array, the measured reduction in thermal resistance is only 22%. This may be a consequence of the extra thermal resistance introduced by the metallization. If a metallization thermal resistance value of 0.12 K W-' is added, the results from the above calculations agree well with experimental results. Using a novel microchannel heat-spreader made of CVD diamond reductions in thermal resistance up to 75% have been predicted [81].

2.5.3 Flatness of Submount The analysis discussed above is an ideal case in which the effects of intermetalic layers is ignored. If the heat spreader is not very flat there will be differences in metal layer thickness between different sections of the array, leading to a nonuniform temperature distribution. This effect is thought to account for differences in thermal resistance observed between different heat-spreaders [79]. It is estimated [80] that the required flatness should be better than 1 pm cm-'. As discussed in 2.3.2.2, it is possible to process CVD diamond plates to accuracy close to one visible fringe over 25 mm. This is equivalent to a deviation from flatness of less than 0.13 pm cm-', well within the requirements for an efficient LDA heat spreader.

2.5.4 Thermal Stress Due to the substantial difference between the thermal expansion coefficient of CVD diamond (1 p.p.m. K-') and laser diode material (6p.p.m. K-' for GaAs), thermal

Figure 37. Example (courtesy of Jenoptik [83]) of a laser diode array bar which has been pre-cracked to reduce effects of thermal stress.

2.6 Cutting Tools. Dressers and Wear Parts

61 1

stress can seriously affect the reliability and life of these devices [79,80,82]. A novel approach has been proposed called stress induced dicing (SID) where the tensile stress in the laser diode bar after soldering and during cooling down is relaxed by cracking the bar at defined lines between emitters [83]. Figure 37 shows an example (courtesy of Jenoptik) of such a laser diode bar mounted on a pre-metallized CVD diamond spreader using a Au:Sn (80:20) solder alloy.

2.6 Cutting Tools, Dressers and Wear Parts J. L. Collins, M. W. Cook and P. K. Sen

2.6.1 Cutting Tools Trends The use of nonferrous materials and composites in modern industry, particularly automotive, is substantial and becoming increasingly important as the demand for stronger, lighter, more wear resistant components increases. Highly abrasive, high performance materials, such as fibre reinforced plastics, high SiAl alloys, metal matrix composites (MMCs) and even abrasive wood composites are steadily increasing in demand Also, there is significant development within the machining industry to increase machining speed and eliminate or reduce the amounts of cutting fluid used for environmental and cost reasons. Components are being manufactured to continually higher levels of precision and consistency. These trends in workpiece material, manufacturing philosophy, and, product quality have led to a strong demand for polycrystalline diamond (pcD) when machining nonferrous materials, and polycrystalline cubic boron nitride (pcBN) when machining ferrous materials. These materials are increasingly providing the best economic solution to high volume, precision manufacturing problems. More recently, CVD diamond has become available and is currently being evaluated and developed in cutting tool type applications. CVD diamond cutting tool materials fall into two categories a thin film conformal coating less than 30 pm in thickness or a thick film solid diamond usually around 0.5mm in thickness. The application of thin film CVD diamond is highly substrate sensitive. The adhesion to tungsten carbide, particularly three dimensional tools like taps and drills, remains a persistent technical obstacle to overcome. CVD thick film can be used free-standing, or has to be brazed directly to a tool body or a substrate that is in turn brazed again to a tool body. The strong acceptance of pcD in machining non-ferrous materials is based on its properties being highly suited to the requirements of modern day manufacturing. CVD diamond has similar properties and it is anticipated that it will gradually become integrated into diamond based cutting tool material selection and application. ~

6 12


Figure 38. Schematic diagram showing the relative merits of various cutting tool materials for the machining of different materials. It can be seen that CVD diamond occupies a specific niche in the machining of materials requiring exceptional wear resistance.

2.6.1.1 Polycrystalline and Chemical Vapor Deposited Diamond An indication of the relative position of various cutting tool materials with respect to their wear resistance and fracture toughness is shown in Fig. 38. Diamond is by far the most wear resistant material, particularly in pure single crystal form, but because single crystal diamond is a brittle material, it does not have the average toughness of the polycrystalline forms both pcD (cobalt-containing) and CVD diamond (which contains no metal phases). CVD thick film diamond complements pcD throughout a wide range of general machining applications. Its main advantage over pcD is its greater thermochemical stability. Potential disadvantages have been its relative brittleness, and its lack of electrical conductivity. But recent advances have resulted in the development of an electrically conductive grade of CVD diamond material, demonstrating that the use of spark erosion cutting and processing techniques (as are used extensively in the diamond tool making industry, and in particular for woodworking tools) could be also adopted for CVD diamond tools. ~

2.6.2 Cutting Tool Application of Chemical Vapor Deposited Diamond The following case studies are typical examples of where CVD diamond’s performance is increasing the range of diamond cutting tool materials available to industry.

2.6 Cutting Tools, Dressers and Wear Parts

6 13

Figure 39. Section through typical HPL wood flooring panel.

2.6.2.1 High Speed Finish Machining of High Pressure Laminate Wood Flooring During the 1990s, the wood working industry has seen a phenomenal growth of the use of PCD cutting tools in sawing, milling, and profiling of high and medium density fibre board, chipboard, particle board - all known to be abrasive materials. In the last five years, high pressure laminate (HPL) - a generic wood product, with its decorative aluminum oxide coated surface for high wear resistance, has emerged as a flooring material. Figure 39 shows the makeup of a HPL wood flooring material. In this application, CVD thick film diamond has recently been tested alongside pcD and single crystal diamond tools for the finish machining of the highly abrasive coated layer of HPL wood flooring material, with very encouraging results [84] as described below. 2.6.2.1.1 Machining Parameters Material machined Cutter diameter No. of teeth Spindle speed Cutting speed Feed rate Feed per tooth

: HPL boards Al-oxide layer : 200 mm

: 8 : 6000 r.p.m. : 63 m s-' : 53 m min-' : 1.1 mm

For CVD diamond materials, the tool edges were prepared by mechanical grinding using a standard cutter grinder and diamond wheels. The cutting edge of synthetic single crystal diamond was prepared by scaife polishing, and the pcD tools' edges were prepared by the standard Electric Discharge Grinding (EDG) method.

6 14

2 Applications of’Diamond Synthesized by Chemical Vupor Deposition

The criterion by which tool life is measured is chipping of the board edge. When chips of an ‘unacceptable’ size are produced, then the tool is changed. ‘Unacceptable’ chip size is determined visually by the machine operator.

2.6.2.1.2 Test Results Figure 40 shows, for each test, the average linear length machined by the diamond tools, before the tools were taken out on the basis of visible edge chipping of the floor panels. The edge wear of CVD tools appears to have been more influenced by grain pull out and erosion at the grain boundaries than wear. All the materials evaluated in these tests are capable of machining HPL flooring, however, it is quite clear that single crystal diamond gives the longest tool life and CVD diamond gives a marginal advantage compared with PCD. Both synthetic single crystal diamond and CVD diamond materials, while providing high tool life, need to be evaluated in the greater cost benefit equation for specific workpiece materials, the requirements of individual HPL flooring products and the manufacturing environment pertaining.

Figure 40. Average linear length machined by different tool materials.

2.6.2.2 Dry Machining of SiCp A1 Metal Matrix Composite MMCs are typically alloys of aluminum reinforced with ceramic particles (usually silicon carbide). This composite material combination has the lightness of aluminum but the strength and temperature resistance of cast iron. Consequently, MMCs have tremendous potential in the automotive and aerospace industries for the replacement of cast iron to considerably reduce weight. MMCs, although readily machinable, are highly abrasive and will dull the cutting edge of conventional tools in a matter of seconds. The performance of CVD diamond thick and thin film and the extremities of pcD grain sizes of 2 and 25pm have been compared when machining an MMC. The machining conditions and parameters are detailed below.


i

0.35

3 E m

g .-8

PCD 002

0.25 0.2

5 0.15

=

PCD 025

0.3

Y

0.1

6 15

CVD thick film

/&.5::

CVD coating

2 0.05 0

2.6.2.2.1 Machining Conditions/Parameters

Material Machined Insert Style Cutting Speed Feed Rate Depth of Cut No Coolant

: 40% Sicp A1 (A356) - MMC : : : :

TNMN 160408 400 m min-l 0.05 mm rev.-’ 0.5 mm

2.6.2.2.2 Test Results Figure 41 shows the flank wear on the respective tool relative to the machining time. In the case of the CVD thin coated tool, initially the rate of tool wear is low but as soon as the ‘thin’ coating is worn through the flank wear increases rapidly. The lowest rate of tool wear is observed on the CVD thick film (0.5 mm) insert. Although testing is limited, the CVD thick film flank wear indicates that the edge has ‘bedded in’ and steady state wear has been established. This suggests that the ultimate tool life could be quite considerable. The machining of MMCs, particularly high volume fraction, > 35%, generates high temperature and involve having to efficiently shear hard silicon carbide particles in the cut zone. The high thermal conductivity/resistance to thermal degradation and abrasion resistance of CVD are particularly suitable for machining high volume fraction MMCs. Table 6 summarizes the relative merits of the three major diamond cutting tool materials - pcD, CVD diamond, and synthetic single crystal as current technology stands, from the perspective of tool performance. It is clear that the various diamond cutting tool formats are largely complementary and, with the development of electrically conductive CVD diamond, the potential application areas of the machining of highly abrasive workpiece materials becomes a reality. ~

6 16


Table 6. The relative merits of different diamond cutting tool materials: tool performance.

Toughness Abrasion resistance Thermo-chemical resistance

PCD

CVD diamond (thick film)

Single crystal

JJJ JJ J

JJ JJ JJJ

J JJJ JJJ

2.6.3 Chemical Vapor Deposited Diamond Dressers Diamond is extensively used for the dressing of precision grinding wheels. The grinding process using conventional, diamond and cBN abrasives is extensively used in the finishing of ferrous, tungsten carbide and ceramic components. Here, the geometrical tolerance, the surface finish of the wheel and its condition is critical and the dresser's function is to achieve and maintain these characteristics. CVD diamond has recently been available in a range of dresser log formats and has shown promising results. CVD diamond potentially has a number of advantages compared to other diamond dresser material formats. The polycrystalline nature of CVD makes it insensitive to orientation. It also has extremely uniform wear characteristics. This means that dresser manufacture is more straight forward and the wear of the dresser is more consistent. The possibility of extracting long sections of CVD from a large disc means that products like blade dressers can be made with continuous diamond pieces compared to numerous discrete stones or pieces that are required to make up the working length of the dresser. Also CVD diamond has increased toughness, meaning that chip-free grinding of the dresser point is easier to achieve. CVD diamond has been compared to single crystal synthetic diamond in single point dressing tests. The conditions and results of these tests are described below. 2.6.3.1 Dresser Test Conditions Wheel Type: Aluminium oxide WA46 KV (300 mm O/D) Wheel Speed: 2300 r.p.m. In-Feed: 0.02 mm wheel diameter-' Feed: 4 mm s f l Flood Coolant 2.6.3.2 Test Results From Fig. 42 it can be seen that CVD diamond exhibits almost identical rates of wear compared to single crystal synthetic diamond. Effectively this means that CVD diamond offers a range of geometric and orientation advantages in dresser design without having to compromise on the wear resistance of the dresser.


’

61 7

Dresser Wear (mm) 0. 5 Dnrarw paramatem Meel Speed In-Feed Feed Caolant

AI-0xde WA46KV (300mm dla x 50 rnm d e ) 2300rpm 0 020 mm I wheel dla 4 rnrnlrec Floodmolant

0.05 0

, 0

200 400 600 Volume of Wheel Removed (cm3)

800

CVD Diamond

-m-

MONZESS

Figure 42. CVD diamond and Monodress single point dresser wear relative to volume of grinding wheel removed.

2.6.4 Chemical Vapor Deposited Diamond Wear Parts The extreme abrasion and erosion resistance of polycrystalline CVD diamond combined with chemical inertness, make it a perfect candidate for use in applications such as wear parts, solid-state sensors, actuators and tools [85,86]. The combination of diamond’s hardness, high Young modulus, very low coefficient m3 N-’ m-’) of friction (less than 0.1 in air) and low wear rates (less than under extreme loading conditions [87,88] potentially give diamond great tribological advantage over most other materials: The advent of polycrystalline CVD diamond deposition technology means that these tribological advantages can be exploited in applications such as large bearings or seals. These can be fabricated from either CVD diamond coated parts or from thick layers of free standing CVD diamond. Examples of these applications areas are pumps, valves and pipelines used in the extremely harsh tribological conditions seen in oil fields, often in remote locations [89]. Limited data is available on the wear rate of CVD diamond [90,91]. A recent study has been conducted in order to investigate the basic tribological properties of bulk and thin-film CVD diamond. The sliding wear characteristics of a number CVD diamond materials have been evaluated using a modified Denison T62 abrasion [92] pin-on-disc test rig, operated in dry and lubricated sliding wear conditions, and compared with a typical hardened steel which is often used for wear part components. Discs and pins of each material were made for testing; the pins all had chamfered edges to prevent edge-galling and were made square to minimise any misalignment thus reducing the contact area. Each material was tested first in dry sliding wear and then in lubricated sliding wear and only tested against components made of the same material.

6 18


The test parameters were identical in each case, with a contact stress of 15 MPa and sliding speed of 150 m min-' . The tests were performed for 48 h where possible (to steady state or until sliding speed diminished through increased friction). The tests all began at nominal room temperature of 20°C and, where required, the lubricant used was a typical heavy white mineral oil; excess lubricant was removed by a repeated solvent wash prior to weighing the pin to determine the specific wear rate. The temperature rise associated with both the dry and lubricated wear tests did not exceed 90°C. Such a temperature rise can be associated with the low coefficients of friction (and carrier mechanisms during the lubricated tests). Such increases are not sufficient to noticeably alter the properties of the CVD diamond or to breakdown the mineral oil lubricant. The coefficient of friction determinations have been made at specific times into the sliding wear tests. These have an associated uncertainty of approximately flo%, predominantly from the (unsteady) change in friction coefficient with time as the sample surface is modified by wear and vibration/noise associated with the friction monitoring transducer. It is worthwhile noting that there are several sources of error in the determination of the specific wear rate which give approximately an order of magnitude uncertainty in the quoted values, in particular, the mass loss through wear of the pins is minuscule - a consequence of the excellent abrasion resistance of CVD diamond and such a low loss is easily affected by any residual lubricant. As can be seen from the data presented in Table 7 the surface finish of the CVD diamond has a considerable effect on both the wear rate and coefficient of friction. ~

Table 7. Wear test results of CVD diamond and hardened steel Material

Hardened steel Hardened steel CVDD (bulk) (Nucleation face) CVDD (bulk) (Nucleation face) CVDD (bulk) (Growth face) CVDD (bulk) (Growth face) CVDD (thin film) (As-grown face) CVDD (thin film) (As-grown face) CVDD (thin film) (Lapped face) CVDD (thin film) (Lapped face)

Surface finish Ra ( w )

Dry or lubricated

Specific Friction wear rate coefficient ( x lo-'* m3 (initially) N-l

,-I

Friction coefficient (at 24 hours)

1

0.07 0.07 0.06

Dry Lubricated Dry

NIA 1 986

Galled at -15 s 0.08 0.115

N/A 0.08 <24 hour test

0.1

Lubricated

0.4

0.038

0.049

0.06

Dry

835

0.15

<24 hour test

0.1

Lubricated

0.7

0.018

0.018

1.52

Dry

61 1

0.124

0.246

1.52

Lubricated

27

0.11

0.08

0.23

Dry

758

0.116

0.196

0.23

Lubricated

84

0.11

0.05

References

619

However, whereas the hardened steel could not be run without a lubricating layer, the CVD diamond does not require lubrication or might be self-lubricating. It is clear that there is a great potential for very low wear rate and low coefficient of friction CVD diamond materials prepared with the required surface finish for wear part applications, where component lifetime is determined by sliding wear in low lubricant or lubricant-free applications.

References 1. R. S. Sussmann, DIAFILM, A New Material for Optics and Electronics, Indust. Diamond Rev., 1993, 53,555, pp. 63-72. 2. R. S. Sussmann, J. R. Brandon, G .A. Scarsbrook, C . G. Sweeney, T. J.-Valentine, A. J. Whitehead, and C . J. H. Wort, Diamond Re/. Muter., 1994, 3, 303-312. 3. C. J. H. Wort, C. G. Sweeney, M. A. Cooper, G. A. Scarsbrook, and R. S. Sussmann, Diamond Re/. Muter., 1994, 3, 1158-1 167. 4. R. S. Sussmann, C. S. J. Pickles, J. R. Brandon, C . J. H. Wort, S. E. Coe, A. Wasenczuk, C. N. Dodge, A. C. Beale, A. J. Krehan, P. Dore, A. Nueara, and P. Calvani, Nuovo Cimento, Section D , 1998, 12 (April), 503-526. 5. R. S. Sussmann, J. R. Brandon, S. E. Coe, C. S. J. Pickles, C . G. Sweeney, A. Wasenczuk, C. J. H. Wort, and C . N. Dodge, CVD Diamond, A. New Engineering Material for Thermal Dielectric and Optical Applications, Proceedings of the Industrial Diamond Association of America meeting on Ultrahurd Materials, Windsor, Ontario 28-30 May 1998, and Indust. Diamond Rev., 1998, 58(578), 69-77. 6. C. P. Klages, Chemical Vapor Deposition of Diamond Films, in Handbook qf Ceramic Hard Materials, R. Riedel (Ed.). Wiley-VCH, Weinheim, 2000, pp. 390-419. 7. G. J. Davies, Section 2 of Diamond Materials and their Applications, in Hundbook of Ceramic Hard Materials, R. Riedel (Ed.), Wiley-VCH, Weinheim, 2000, pp. 485-504. 8. M. W. Bailey, Section 1 of Diamond Materials and their Applications, in Handbook ofceramic Hard Materials, R. Riedel (Ed.), Wiley-VCH, Weinheim, 2000, pp. 479-485. 9. R. Berman Chapter I , J. E. Field Chapter 9, D. Tabor Chapter 10, A. T. Collins Chapter 3, C . D. Clark, E. W. Mitchell and D. J. Parson Chapter 2, and C. A. Brooks Chapter 12, in The Properlies qf Diamond, J. E. Field (Ed.), Academic Press, London, New York, San Francisco, 1979. 10. C. D. Clark, A. T. Collins, and G. S. Woods Chapter 2, R. Berman Chapter 7. J. E. Field Chapter 12. C . A. Brookes Chapter 13. D. Tabor and J. E. Field Chapter 14, J. Wilks and E. M. Wilks Chapter 15, and J. E. Field Chapter 18, in The Properties of Natural andSynthetic Diamond, J. E. Field (Ed.), Academic Press London, San Diego, New York, Boston, Sydney, Tokyo and Toronto, 1992. 11. J. E. Field in Proceedings of the N A T O Advanced Study Institute on Diamond and Diamond-like Films and Coatings 22 July 1990, R. E. Clausing, L. L. Hardon, J. C. Angus, and P. Koidl (Eds), Plenum Press New York, London, 1991, pp. 17. 12. Chapters I , 3, 5 , 7 and 8. in Handbook uf’lndustrial Diamond~sand Diamond Films M. A. Prelas, G. Popovici. and L. K. Bigelow (Eds), Marcel Dekker Inc. New York, Basel, Hong Kong, 1998. 13. Chapters I , 2, and 5. in Properties andGrovvth ofDiamond, G. Davies (Ed.), EMIS Data Review Services, INSPEC IEE, London, 1994. 14. J. Save, in Infrared Optical Materials und Their Antirefection Coatings, Adam Hilger, Bristol, Boston, 1985, p. 119. 15. D. C. Harris in Itfrared Windows and Dome Materials, Tutorial Texts in Optical Engineering, Vol. TTlO SPIE Optical Engineering Press, Bellingham, WA (USA), 1992, Chapter 5. -

620


16. M Seal and W. J . P. van Enckevort, Proceedings of SPZE- Diamond Optics, 1988,969, 144152. 17. M Seal in Industrial Diamond Review, P. A. Daniel (Ed.), 1971, 31 (Nov), 464469. 18. J. Doting and J. Molenadr, Proceedings ofthe Fourth Annual ZEEE Semiconductor Thermal and Temperature Measurement Symposium, Sun Diego California, February. IEEE CH2530418810000-0113, 1988, pp. 113-117. 19. R. J. Caveney in Chapter 20 of Ref. [9]. 20. M. Seal in Chapter 16 of Ref. [lo]. 21. G. S. Woods in Chapter 3 of Ref. [13]. 22. J. W. Vandersande in Chapter 3. of Ref. [13]. 23. A. T. Collins in Chapter 9. of Ref. [ 131; B. A. Fox, L. Harstell, D. M. Malta, H. A. Wynands, G. Tessmer, and D. L. Preifus in Proceedings of the Materials Research Society Symposium, 1996, 416, 319-329. 24. M. Seal, Zndust. Diamond Rev., 1979, 39(461), 115-118. 25. J. A. Savage, C. J. H. Wort, C. S. J. Pickles, R. S. Sussmann, C. G. Sweeney, M. R. McClymont, J. R. Brandon, C. N. Dodge and A. C. Beale. SPZE Proceedings of the Windowand Dome Technologies and Materials Symposium, Orlando, Florida, April, 1997, 3060, 1 4 4 159. 26. D. C. Harris, Development of Chemical Vapour Deposited Diamondjor Infrared Optical Applications, Status Report and Summary of Properties, Report No N AWCWPNS TP 8210, Naval Air Warfare Center Weapons Division, China Lake, CA (USA), 1994. 27. M. Massari, P. Union, G. A. Scarsbrook, R. S. Sussmann, and P. Muys, Proceedings ofthe SPZE Laser-Znduced Damage in Optical Mateilals, Boulder, Colorado, November, 1995, SPZE, 2714, 177-184. 28. DIAFILM is a. trade mark of De Beers Industrial Diamond Division. 29. R F. Potter in Handbook of Optical Constants of Solids, E. D. Palik (Ed.), Academic Press Inc. Orlando. San Diego, New York, London, Toronto, Montreal, Sydney, Tokyo, 1985, Chapter 2. 30. M. E. Thomas and W. J. Tropf, SPZE, 1994,2286, 114&151. 31. V. Vorlicek, J. Rosa, M. Vanacek, M. Nesiadek, and L. M. Stals, Proceedings of the 7th European Conference on Diamond, Diamond-like and Related Materials, Tours France, September 1996, Diamond Rel. Muter., 1997, 6 , 704707. 32. D. F. Edwards and H. R. Philipp, in Handbook of Optical Constants of Solids, E. D. Palik (Ed.), Academic Press Inc. Orlando, San Diego, New York, London, Toronto, Montreal, Sydney, Tokyo, 1985, subpart 3, pp. 668. 33. P. Dore, A. Nucara, D. Cannavo, G. De Marzi, P. Calvani, A. Marcelli, R. S. Sussmann, A. J. Whitehead, C. N. Dodge, A. J. Krehan, and H. J. Peters, J. Appl. Optics, 1998,37 (July), no. 21. 34. P. Union, P. Muys, D. Vyncke, B. Depuydt, and P. Boone, SPIE, 1995,2870, 521-527. 35. C. S. J. Pickles, J. R. Brandon, S. E. Coe, and R. S. Sussmann, Proceedings of Symposium ZV, Diamond Films, 9th Cimtec - World Forum on New Materials, Florence, Italy, June 1998, P Vincenzini (Ed.), Faenza, Italy, 1999, pp. 435454. 36. T. J. Valentine, A. J. Whitehead, R. S. Sussmann, C. J. H. Wort, and G. A. Scarsbrook, Diamond Rel. Muter., 1994, 3, 1168-1 172. 37. F. Sziics, University of Berlin and California Institute of Technology, Pasadena, California. Private communication, 1998, see also: F. Sziics, M. Werner, R. S. Sussmann, C. S. J. Pickles, and H. J. Fecht, J. App. Phys., 1999,86 (December) no. 11, 6010-6017. 38. F. Sziics, C. H. Moelle, S. Klose, H. J. Fecht, S. Fassbender, N. Meyendorf, and M. Werner, in Proceedings of the International Conference on Micro Materials, April I997 (MICRO-MA ’97) B. Michel and T. Winkler (Eds), 1997, pp. 569-572. 39. J. W. V. Vandersande in Chapter 1 of Ref. [13]. 40. J. E. Graebner, M. E. Reiss, L. Seibles, T. M. Hartnett, R. P. Miller, and C. J. Robinson, Phys. Rev. B, 1994, 50(6), 3702. 41. E. Wormer, C. Wild, W. Muller-Sebert, R. Cocher, and P. Koidl, Diamond Rel. Materi., 1996,5, 688-692. 42. See for instance P. A. Rizzi, Chapter 2.6 in Microwave Engineering, Prentice-Hall International, Englewood Cliffs, NJ (USA), 1988, p. 41. 43. A. C. Lynch and R. N. Clarke, IEE Proceedings A , 1992, 139, 221. 44. A. P. O’Leary in Electro-Optics, Special Report on Thermal Imaging Systems, 2nd edn, Jane’s Information Group, Antony Rowe Ltd, Chippenham, Wiltshire, UK, 1994.

References

62 1

45. E. J. Coad, The Response of C V D Diamond and other Brittle Maierials to Multiple Liquid Impacts, PhD Thesis, University of Cambridge, 1996. 46. M. D. Hudson, C. J. Brierley, A. J. Miller, and A. E. Wilson in the Proceedings of the Windou. and Dome Technologies and Materials Synzposium, Orlando, Florida, April 1997, SPIE, 3060, 196202. 47. C. J. H. Wort, J. R. Brandon, B. S. C. Dorn, J. A. Savage, R. S. Sussmann, and A . J. Whitehead in Applications o f Diamond Films and Related Materials: Third International Conference, A. Feldman, Y. Tzeng, W. Yarbrough, M. Yoshikawa and M. Murokawo (Eds), NIST Special Publications, 1995, 885, 569-572. 48. Rofin Sinar Technologies, Annual Report, 1997, pp 6. 7. 49. C. A . Klein, Optical Engineering, 1990, 29(4), 343-350. 50. M . Sparks and M. Cottis, J. App. Phys., 1973, 44, 787-794. 51. S. Singer, Diamond optics, A. Feldman and S. Holly (Eds), SPIE, 1988, 969, 168-177. 52. C. A. Klein in Laser Induced Damage in Optical Materials, 1991, SPIE, 1624, 475492. 53. C. A. Klein, Dianiond and Relaied Materials, 1993, 2, 102&1032. 54. C. Wild, Chapter 10.3 in Lou-Pressure Synthetic Diamond, B. Dischier and C. Wild (Eds), Springer Veriag, Heidelberg, 1998. 55. S. P. Timoshenko and J. N. Goodier, Chapter 12 in Theory of Elasticity, 3rd edn, 1970, McGraw-Hill, NY, pp. 385. 56. B. Depuydt, P. Boone, P. Union, and P. Muys in Proceedings y f t h e 4th International Worshop on Laser Beam and Optics Characterisation, Munich, June 1997, A. Giesen and M. Morin (Ed.), 1997, 657-663. 57. N. Ellis and D. E. Greening Diamond and Related Materials, 1993, 2, 572-581. 58. R. S. Sussmann, C. A. Scarsbrook, C. J. H. Wort, and R. M. Wood. Diamond Rel. Muter., 1994, 3, 1 I73 1177. 59. H.-K. Karow, Fabrication Methods fbr Precision Optics, John Wiley and Sons, 1993, p. 620. 60. Tim Holt, Rofin Sinar 1998, Private communication. 61. R. Heidinger. Digest 19th International Confirence on Injiared and Millimeter Waves,Lausanne ( C H I , SPIR, 1991, 1576,441. 62. L. Redcliffe, Digest 16th International Conference on Injiared and Millinieter Waves.Sindon ( J ) , JSAP Cat No AP 941228, p. 277. 63. M. Thumm, State of the art of high power Gyro-devices and,free-electron Masers, Wissenscliaf~ tliche Berichte F Z K A 5564, Forschungszentrum Karlsruhe GmbH, Karlsruhe 1995, (ISSN 0947-8620). 64. M. Thumm, J. of Infrared and Millimeter Wave, 199X, 19, No. I , pp. 3-14. 65. M. Thumm, Fusion and Engineering Design, 1995, 30, 139-170. 66. M . Thumm, FZK Institute, 1997, Private communication. 67. 0. Braz, G. Dammertz M. Kujntze, and M. Thumm, International Journal of Infrared and Milliineter Waves, 1997, 18(8), 14651477. 68. R. Heidinger, A . Ibarra, and J. Molia, J. Nuclear Materials, 1998, 258-263B, part B, pp. 18821826. 69. R. Sporl, R. Schwab, R. Heidinger, and V. Parshin, C V D Diamondfor High Power Gyrotrons: Characterisutian of Dielectic Properties, 1TG-Pachbericht, Berlin, April 1998. 70. R. Heidinger, R. Sporl, M. Thumm, J. R. Brandon, R. S. Sussmann, and C. N. Dodge, Progranime Commitire, 23rd International Conference on Infrared and Millimeter Waves, Colchester, Essex. UK. ‘CVD Diamond Windows for High Power Gyrotrons’, September 1998, pp. 223-225. 71. 0. Braz, A. Kasugai, K. Sakamoto, K. Takahashi, M. Tsunekoa, T. Imai, and M. Thumm, Int. J . Infrared Millimeter Waves, 1997, 18(8), 1495-1 503. 72. A. Kasugai, K. Sakamoto, K. Takahashi. M. Tsuneoka, T. Kariya, T. Imai, 0. Braz, M. Thumm, J. R. Brandon, R. S. Sussmann, A. Beal. and D. C. Ballington, Rev. Sci. Instrum, 1998, 68(5). 73. K. Sakamoto, Japan Atomic Energy Research Institute, Private communication. 74. R. Heidinger. R. Schwab, R. Sporl, and M. Thumm, 22nd International Conference on Infrared and MiNimeter Waves, Wintergreen. Virginia, July 1997, P. Freund (Ed.). 75. R. Heidinger. FZK Institute, Private communication. -

622

2 Applications of Diamond Synthesized bqi Chemical Vapor Deposition

76. S. M. Hu, Critical stress in silicon brittle fracture, and effect of ion implantation and other surface treatments, J. Appl. Phys., May 1992, 53(5), 35763580. 77. R. C. Eden, in Handbook of Industrial Diamonds and Diamond Films, M. A. Prelas, G. Popovici, and L. K. Bigelow (Eds), Marcel Dekker Inc, New York, Basel, Hong Kong, 1998, Chapter 30, 1073-1102. 78. J. Bonhaus, D. Lorenzen, E. Kaulfersch, A. Denisenko, A. Zaitsev, S. Kwong, K. Man, H. Reichi, and W. R. Fahrner, Proceedings of the International Conference on Micro Materials, Micro Mat '97, Berlin, April 1997, B. Michel and T. Winkler (Eds), 1997, 505-508. 79. D Lorenzen, S. Heineman, J. Bonhaus, Proceedings of the International Conjerence on Micro Materials, Micro M a t '97, Berlin, April 1997 (Ed.: B. Michel and T. Winkler), 1997, 513-516. 80. D. Lorenzen, Jenoptik Laserdiode, 1998, Private communication. 81. K. E. Goodson, K. Kurabayashi. and R. F. W. Pease, IEEE Transactions on Components Packaging and Manujacturing Technology, 1997, Part B. - Advanced Packaging 20-1, 104109. 82. S. Weiss, E. Zakel, and H. Reichi, IEEE Transactions on Components Packaging and Manufacturing Technology, 1996, Part A. Vol. 19-1, 46-53. 83. Patent assigned to Jenoptik Laserdiode, DE 196 44 941 CI (D. Lorenzen, 1998. Private communication). 84. P. K. Sen, M. W. Cook, and R. Achilles, To be published: Proceedings of the Industrial Diamond Association of America Meeting on Ultrahard Materials, Windsor, Ontario 28-30 May, 1998. 85. S. J. Bull and A. Matthews, Diamondjor Wear and Erosion Applications, Diamond and Related Materials, 1992, 1, 1049-1064. 86. B. Lux and R. Haubner, Phil. Trans. R . SOC.Lond. A, 1993, 342, 297-311. 87. T. Le Huu, H. Zaidi, and D. Paulmier, Wear, 1977, 203-204,442446. 88. M. N. Gardos in Protective Coatings and Thin Films,Y. Pauleau and P. B. Barna (Eds), Kluwer Academic Publishers, The Netheriands 1997, 185-196. 89. D. Cooper, F. A. Davies, and R. J. K. Wood, J . Phys. D: Appl. Phys., 1992, 25, A195-A204. 90. A. Edemir and G. R. Fenske, Proceedings of the 51st Annual meeting of the S T L E , Cincinnati Ohio, May 1996, 1-8. 91. R. A . Hay and J. M. Galimbarti, Handbook of Industrial Diamonds and Diamond Films, M. A. Prelas, G. Popovici, and L. K. Bigelow (Eds), Marcel Dekker Inc, New York, Basel, Hong Kong, 1998, Chapter 33, 1135-1 147. 92. M. J. Neale (Ed.), Tribology Handbook, Butterworths, London, 1973 (2nd Edition 1994).


3 Diamond-like Carbon Films C.-P. Klages and K. Bewilogua

3.1 Introduction The term ‘diamond-like carbon’ was coined in the early 70s in order to denote amorphous carbon films, prepared by physical vapor deposition techniques, which were expected to approach genuine diamond, as far as density and mechanical properties are concerned. Today the expression (frequently abbreviated to ‘DLC’) is used for a fairly wide range of amorphous films containing carbon atoms in different states of hybridization and possibly also further elements such as hydrogen, silicon, or metals (‘Si-DLC’, ‘MeeDLC’). There are presently no generally accepted criteria for the use of the term DLC or for a clear distinction between ‘diamond-like’ on the one hand and ‘graphite-like’ or ‘polymer-like’ on the other. This review will therefore pragmatically emphasize those kinds of amorphous carbon-based films which have found the greatest interest from a scientific or an applicational point of view. Aside from hydrogen-free amorphous carbon (a-C) and especially highly tetrahedrally coordinated amorphous carbon (ta-C), deserving the denotation ‘diamond-like’ most of all, hydrogenated films (a-C :H) and metal-containing hydrogenated films (MeC: H) will be treated in more detail, owing to the technical importance for mechanical-tribological applications which these materials have gained. Section 3.2 will present the deposition methods which have been used or are being used for the preparation of DLC films. Microstructures and properties will be treated in sections 3.3 and 3.4 and the final section will deal with applicational aspects of DLC coatings.

3.2 Preparation Methods for Diamond-like Carbon Films 3.2.1 Hydrogenated Amorphous Carbon (a-C:H) In 1976 Holland and Ojha [l] reported the deposition of a-C: H films using a glow discharge in hydrocarbon gases excited by a radio frequency (RF, 13.56MHz) powered substrate electrode. The counter-electrode was the grounded wall of the deposition chamber. Because this technique was relatively simple in operation the work of Holland and Ojha can be seen as the starting point of intensive research and development efforts aiming at practical applications of DLC films. Hydrogenated amorphous carbon films have relatively low electrical conductivities, typically less than lop6 R-’ cm-’. The operation of the substrate with RF allows

624

3 Diamond-like Carbon Films

deposition of a-C: H films with thickness of several micrometers on conducting as well as on insulating substrates without surface charging effects. On the other hand, attempts to excite the substrate electrode in the glow discharge only with DC (direct current) power were reported (e.g. by Whitmell and Williamson [2] and later by Grill and Pate1 [3]). There are indications that these DC deposition processes arc sustained by secondary electrons generated at masks or shield surrounding the substrates [4, 51. These secondary electrons compensate the surface charges. But at least for thicker films the charging of the films becomes an essential problem if their resistivity is higher than about lo8 Rcm. This can be explained by a simple estimation of the voltage drop over the film (see below). Today such twoelectrode processes with only DC excitation have nearly no importance. However, there are intensive efforts to find alternatives to the RF processes, because this technique has only a limited potential for economic up-scaling of the coated parts dimensions. One alternative is the substitution of R F in the MHz range by lower frequencies, especially medium frequencies up to 100 kHz. In 1986 Catherine and Couderc [6] revealed that films prepared with 13.56MHz or with 50kHz, respectively, had nearly the same properties. At present the medium frequency technique is interesting above all from the viewpoints of up-scaling and costs. Following the work of Holland and Ojha several other methods for preparation of a-C:H were developed, for example: sputter deposition [7, 81, ion plating [8, 91, cascade arc plasma jet [lo], and simultaneous operation of a RF and a microwave (2.45 GHz) discharge [l I]. Important parameters of all a-C : H deposition processes are the ion energy, commonly expressed by the negative substrate bias voltage Ub. the power density at the substrate, the pressure of the used hydrocarbon gas or vapor, the ion and neutral flux densities, and the substrate temperature. In contrast to the diamond film deposition the substrates must not be heated additionally. Therefore it is an essential advantage of DLC films that they can be deposited on temperature-sensitive parts. In Table 1 typical parameter ranges of different a-C : H deposition processes are summarized. Besides the technique, the type of precursors, gas pressures and bias voltage, the power density at the substrate and the linear growth rate arc considered. Data on particle flux densities can be found only in few papers. All the mentioned parameters influence structure and properties of the films. It should be noted that before starting the deposition process, the substrates must be cleaned by sputter etching with noble gas ions in order to achieve a good adhesion. The following discussion of characteristic features of a-C :H deposition processes will be concentrated on the R F and the ion plating method. Figure 1 shows a scheme of the R F glow discharge deposition method. The RF power (the mostly applied frequency is 13.56MHz) is capacitively coupled to the substrate electrode, the counterelectrode commonly is the wall of the vacuum chamber having ground potential. In the frequency range of some MHz only the electrons, but not the ions, can follow the R F voltage. Due to the large differences in the mobilities of electrons and ions a high negative DC self-bias U, will be generated at the substrate electrode if the area of the powered electrode is markedly smaller than that of the grounded electrode (asymmetric system, see e.g. [12]). In

3.2 Preparation Method.yfor Diamond-like Carbon Films

625

626

3 Diamond-like Carbon Films Substrate

Process pressure: 1 to 10 Pa Self-bias: -50 to -800 V Power density: 0.05 to 0.5 W cm-2

Deposition rate: - 1 pm h-’

Current density: 0.6 mA cm-’

Figure 1. Schematic drawing of the radio frequency (RF) plasma technique for deposition of a-C: H films with some typical process parameters.

practice this condition is usually fulfilled. The self-bias drops across the plasma sheath in front of the negatively biases substrate electrode down to the plasma potential Up. The maximum ion energy Eimax of the ions arriving at the substrate surface is e( Up - u b ) , where e is the unit electron charge. In asymmetric systems the mean plasma potential is a few volts positive. Therefore u b is a sufficiently good measure for the maximum ion energy. However, the ions can lose energy if their mean free path in the plasma sheath is smaller than the thickness of the sheath. For the pressure range typical for RF de osition of a-C:H ( z 10Pa) Bubenzer et al. [13] derived the relation Ei M U b / P‘ ’ . Catherine and Couderc [6] confirmed this relation for the R F process. The self-bias u b varies with the discharge parameters according to Ub (W/P)’’’ where W is the R F power and P the pressure of the process gas [6]. For the film growth rate the relation R z U b P was found experimentally [ 131. The deposition rate also depends on the type of precursor C,H, used. The highest rates can be realized with benzene (C6H6)and cyclohexane (C6HI2),the lowest with methane. One possible explanation is that the rate decreases with increasing ionization potential of the source gas [14]. The negatively self-biased substrate surface is bombarded with ions which are generated in the discharge (C,H; and noble gas ions if noble gases are added). However, not only the C,H; ions introduce carbon into the film. Comparing the calculated mass deposition rates of major ions (CH; for CH4 gas and C 6 H l as well as C6Ht for C6H6 precursor) with the experimentally determined deposition rates Catherine and Couderc [6] found clear discrepancies in the correlations between rate and R F power and between rate and pressure. In order to overcome these, the adsorption of activated species and their incorporation into the growing film must be taken into account additionally to explain the experimental data.

P

N

3.2 Preparation Methods for Diamond-like Carbon Films

627

Concerning the details of a-C:H deposition mechanisms there is still a lot of uncertainties. A comprehensive discussion of the situation in this field was given by Moller [15]. Besides pressure and power the substrate temperature can influence the growth rate of a-C:H films. Commonly the rate decreases with increasing temperature [15, 161,owing to a declining contribution of adsorbed molecules to the film growth rate [I 51. Thus the observed temperature-dependence of the growth rates is an additional support for the adsorption model. In the R F glow discharge process the plasma generation and the film deposition will be realized by the same electrode. The advantage of the ion plating method is that plasma generation and film deposition are separated. Some typical process parameters for this method are noted in Table 1. The plasma can be generated by fast electrons ( ~ 1 0 eV) 0 from a hot cathode-anode arrangement [S, 91 or by microwaves [17]. Positive ions from the plasma then are accelerated to the negatively biased substrate. This is done with DC voltage only. A simple approximation shows that the use of DC becomes problematic for film resistivities > lo8 Rcm. The voltage drop over the film is estimated as AU = pdfji where p is the resistivity of the film, df the film thickness and the current density at the substrate. For typical values df = 3p m, ji = 0.1 mA cm-ji and p = 1O8 Rcm [S] follows A U = 3 V. For substrate voltages of lOOOV and more such a voltage drop is negligible [8, 51. For higher resistivities or for insulating substrates this deposition method becomes unsuitable. If benzene vapor is used, about 70% of the ions striking the substrate consist of C6H: (x 5 6). As in the case of R F plasma deposition not only ions contribute to the film growth. The contribution of the hydrocarbon ions to the film deposition is not more than 50%, further contributions are due to adsorbed and polymerized species [5]. The properties of the deposited a-C: H films, to be discussed in sections 3.4 and 3.5, strongly depend on the chosen process parameters, especially on pressure, power density, bias voltage and temperature. Therefore the deposition conditions must be carefully adjusted for preparation of films on parts or tools to be utilized in industrial practice. Today both R F plasma and ion plating deposition methods are in use for research and development and for prototype production in the industry. '

3.2.2. Hydrogen Free Amorphous Carbon (ta-C) There is a considerable interest in hydrogen free amorphous carbon films with high sp3 contents (high contents of tetrahedrally bonded carbon atoms), because - compared to a-C :H - such films have properties much closer to diamond. For the preparation of ta-C films a source of energetic carbon species is needed, either with or without mass filtered carbon ions. Already in 1971 Aisenberg and Chabot [18] deposited carbon ions of 40-100eV separated from a special source creating argon and carbon ions by sputtering a graphite target. In their pioneer work they introduced the designation diamond-like carbon for the deposited material. This was the first case of preparation of hydrogen free DLC. At present several different

628


methods to prepare such films are known [19, 201: (i) (ii) (iii) (iv) (v)

Ion beam deposition (with or without mass selection). Sputtering from a graphite target. Arc evaporation from graphite cathodes (with or without filtered beams). Laser ablation from a graphite target. Ion-beam assisted deposition (IBAD: simultaneous bombardment of the growing film with noble gas ions).

Mass selected ion beam deposition is a special method with well defined conditions concerning ion masses and energies. Hofsass et al. [21] for example used I2C-' ions, in the most cases with deposition energies of 100 eV. This method is interesting for basic research, but seems to have no perspectives for industrial uses. From the viewpoint of practical applications especially the sputter and arc evaporation techniques are interesting. Using the sputter deposition from graphite targets with argon and an argon-hydrogen mixture, respectively, Savvides and Window [22, 231 prepared amorphous carbon films with high contents of sp3 bonded carbon. The sp3/sp2ratio in the films increased with increasing energy per carbon atom. The increase of the sp3 content in the case of hydrogen-containin sputter gas was explained by a preferred reaction of atomic hydrogen with sp bonded carbon as known from diamond film deposition. The vacuum arc evaporation technique seems to have a high potential for industrial applications. Arc plasmas are characterized by a high degree of ionization which is especially suitable to obtain high sp3 contents in amorphous carbon films. A disadvantage of arc evaporation is the occurrence of graphitic microparticles. These particles can be avoided by filtering the particle beam [24]. However, such filter systems reduce the deposition rates drastically. As an alternative method which allows to control the arc process and to reduce the microparticle emission the laser arc technique was developed by Scheibe et al. [25]. Recently this technique was presented as a new industrial technology for high rate deposition of hard amorphous carbon [26]. In laser ablation systems a laser beam strikes on a graphite target surface and produces a flux of carbon material from the target to the substrate where the film grows (for a review see [27]). The emitted particles have a broad range of energies and can be utilized to prepare both hydrogen free as well as hydrogenated amorphous carbon films. Ion beam assisted deposition techniques are characterized by a generation of carbon from a target by evaporation, laser ablation or sputtering [28] and a simultaneous bombardment of the growing film with argon ions. The low energy ion bombardment (up to some 100eV) leads to dramatic changes in the film structure and the creation of dense carbon with high concentration of sp3 bonds. Several models are presently under discussion to explain the formation of highly tetrahedrally coordinated carbon ta-C. According to one of these theories the ta-C deposition can be described as a subplantation process [19,29]. The principle of this process is that carbon ions penetrate the first atomic layers and enter interstitial positions, thereby increasing the local density and inducing local compressive

8

3.2 Preparation Methodsfor Diamond-like Carbon Films

629

stress. The densification can cause a change of the hybridization from sp2 to sp3. This process occurs in an energy range which obviously depends on the preparation method used [19]. Considering the experimental data, high sp3 fractions in the a-C films ( > 70%) were found in the energy range between 20 and 600eV. For too low energies the ions can not penetrate, will be stuck at the surface and form sp2-carbon. For too high energies the ion dissipates the energy in a ‘thermal spike’ and relaxation of the densified regions occurs. At present several methods to prepare ta-C films are known, but mostly on a laboratory level. Therefore the deposition technique is still not suitable for broad technological applications as they are already realized in different fields for a-C: H or MeC: H coatings.

3.2.3 Metal-containing Amorphous Hydrocarbon An essential problem to be solved before introducing coatings to practice is a good and reliable adhesion. The very promising properties of a-C :H films for a long time could not be utilized because of their high compressive stress. For thickness of more than 12pm the adhesion, especially on steel, was insufficient. This problem is still more pronounced for ta-C films. Today there are possibilities to improve the adhesion markedly, for example by interlayer systems (for more details see Section 3.5). Another way to overcome the disadvantage of bad adhesion is the incorporation of metals into the a-C: H films (MeC : H). Such films were first prepared by Dimigen and coworkers in 1983 using a reactive R F sputter technique [30, 311. A large number of metals, either forming carbides (like Ti, Ta, and W) or not (Au or Cu), were investigated [32]. For applications in tribology the carbide-forming metals are preferable. MeC: H films with suitably adjusted metal contents turned out to have considerably lower compressive stress than a-C: H, but nearly the same excellent friction behavior. On the other hand the wear resistance of MeC:H coatings is lower than that of a-C:H (see Section 3.5.2). As a function of metal-content, minimum wear rates were measured for metal to carbon ratios of 0.1-0.2 [33]. With respect to industrial requirements, reactive DC magnetron sputter techniques were developed for the MeC: H preparation [34, 351. Magnetron sputtering has a great potential owing to larger deposition rates about one order of magnitude higher than with the R F method and the excellent possibilities for up scaling of the coating devices. An industrial DC magnetron batch coater for the WC:H deposition was presented by Hofmann et al. [36]. Metals (mostly Ti, W) or metal carbides (WC) will be used as target materials. The carbon content can be controlled by the reactive gas content in the sputter gas (e.g. Ar C2H2).Like in the a-C:H deposition the substrates must not be heated additionally. A substrate bias is not necessary to deposit MeC:H films, but a moderate bias voltage (Or, = - 100 to -200 V DC) clearly improves the coating quality. Other typical process parameters to prepare MeC: H coatings with metal to carbon ratios of 0.1 to 0.2 are:

+

-

power density at the target: 10 W cmp2. acetylene concentration (ratio of gas flows [C2H2]/[C2H2+ Ar]): 3 0 4 0 % total gas pressure: 0.3-0.8 Pa

630 -


substrate sputter cleaning: Ar ions ( U , = -600 V DC) substrate current density: > 1 mA cmP2 (unbalanced magnetron mode) substrate holder temperature: <200°C time for deposition process: about 100 min

Concerning the mechanism of MeC: H sputter deposition two components must be considered [35, 371. (i) metal and carbon are sputtered from the target in a Ar-hydrocarbon gas mixture. The carbon material consists of loosely bonded hydrogenated carbon which will be deposited and resputtered continuously at the target surface. Such a mechanism can explain the very high deposition rates (more than 10 pm h-' on substrates stationary in front of the target) reached with the DC magnetron [35]. (ii) Hydrogenated carbon is deposited directly from the plasma surrounding the substrate (a-C:H process similar to that discussed in Section 3.2.1). In the case of magnetron sputter deposition this contribution is small ( ~ 1 0 % ) . Today MeC:H coatings are widely used in industrial practice, for example for machine parts improved with wear resistant surfaces exhibiting low friction. On the other hand the wear resistance of MeC :H has still not reached the level of aC: H. Therefore a further improvement of the MeC :H quality as well as a higher productivity and reliability of a-C: H and ta-C coatings will be necessary.

3.3 Microstructure and Bonding of Diamond-like Carbon 3.3.1 Amorphous Carbon and Hydrogenated Amorphous Carbon Among the Group IV elements carbon takes an exceptional position owing to its ability to exist in bonding configurations formed by different hybridizations of its s and p atomic orbitals [38]: The tetrahedral sp3 hybridization with four o bonds to neighboring atoms is realized in the diamond allotrope, graphite is formed by trigonally sp2-bonded carbon atoms, while polyacetylene ( Q n is an example of a compound with sp-hybridized carbon with two o bonds arranged to form a straight line. The description of DLC structures even if based solely on carbon (a-C), without any hydrogen or metal incorporation - is complicated by the fact, that all kinds of hybridizations will in general be present in the amorphous matrix which may also contain nanocrystalline carbon phases. Additionally, there can be a chemical ordering, such as by clustering of sp2-bonded atoms to longer chains or planar graphite-like segments within a predominantly sp3-bonded matrix [39]. In hydrogen-containing DLC films as grown by plasma CVD or reactive PVD methods in hydrogen atmospheres, different amounts of H atoms can be bonded to the sp, sp2 and sp3 bonded carbon atoms, resp. ~

3.3 Microstructure und Bonding of Diamond-like Curbon

DOS

t

valence bands

63 1

conduction bands

EF

Energy

Figure 2. Schematic drawing of electron density (DOS) of states of amorphous carbon phases. EF = Fermi energy (after [14]).

The most important methods which have been in use to reveal the structure and bonding of a-C and a-C:H films have been summarized in review articles [39, 401. In order to understand these methods and the electrical and optical properties of DLC, a closer look to its electronic structure is necessary. Figure 2 is a simplified schematic diagram of the electron density-of-states (DOS) of amorphous carbon, as it is frequently shown in the literature (e.g. in [ 141). A pair of o and o*bonding and antibonding states is due to the bonds forming the skeleton of the a-C or a-C:H network. Owing to the weaker n: bonds, the n:/n:* couple is located more closely to the Fermi energy E F .(Strictly speaking, the nomenclature of o and n: molecular orbitals would require the presence of a mirror plane within a molecule [38]. Nevertheless, this simplified concept is of certain value for a discussion of bonding in DLC.) The hydrogen content in a-C:H films can be determined by a number of methods, including classical chemical combustion analysis, nuclear reaction analysis (NRA), elastic recoil detection (ERD), nuclear magnetic resonance (NMR) spectroscopy, or an analysis of C-H band intensities in infrared vibrational absorption spectroscopy (IR). The most reliable information about the relative amounts of sp2- and sp3bonded carbon (s bonded C atoms usually do not play an important role) comes from solid state 1 C-NMR spectroscopy [41] and electron energy loss spectroscopy (EELS) [4244] (see also the references to other original papers in [39] and [40]). The results of Weissmantel rt al. already pointed to the existence of a special carbon structure in films deposited either by dual beam sputtering or by condensation of species from a low pressure ionization system operated with benzene. Fink et al. used R F plasma deposition of a-C: H from benzene and measured the EELS spectra for samples grown at different substrate bias voltages before and after annealing up to 1000°C. The o n plasmon energy positions and intensities are significantly dependent on the sp3 and sp2 fractions. The concentration of carbon in sp3 configuration was found to be roughly two third in as-grown films and increased after annealing. The hydrogen content declined strongly from 60 to 25 at-%, when the bias voltage was raised from 200 to 12OOV, accompanied by a significant density . dependence of the sp2 carbon and the hydroincrease (1.49 to 1.75g ~ m - ~A) strong gen content, in a-C:H films grown by R F or DC plasma and ion beam deposition, was also noted in the NMR measurements performed by Kaplan et al. [41]. A compilation of the sp3 percentages and H contents of typical a-C and a-C:H films was given by Robertson [45]. Table 2, containing also density, hardness, and

P-

+

632


Table 2. Structural characteristics and properties of various kinds of carbon, data taken from an earlier comuilation 1451.

Diamond Graphite Glassy carbon Polyethylene a-C, evaporated a-C, sputtered a-C, MSIB (ta-C) a-C : H, hard a-C: H, soft

Yo sp3

at-% H

Density, Hardness, d (g ~ m - ~ ) H (GPa)

Gap (eV)

100 0 -0 100

0 0

1 2-5 90 f 5 30-60 50-80

-

3.515 2.267 1.3-1.55 0.92 1.9-2.0 1.9-2.4 3.0 1.6-2.2 0.9-1.6

5.5 -0.04 0.01 6 0.4-0.7 0.40.7 0.5-1.5 0.8-1.7 1.w

67 -

<9 1040 40-65

100 -

2-3 0.01 2-5 11-15 30-1 30 10-20 <5

~

MSIB = mass selected ion beam deposition.

band gap energy values, compares these special forms of carbon and hydrogenated carbon to a few 'conventional' materials. It is important to note, that a 100% sp3 content, accompanied with a band gap energy in excess of 5 eV, is found in diamond as well as in polyethylene, representing the extremes of the hardness and density scales. In DLC, a high amount of tetrahedral bonding is present in very soft, polymer-like a-C :H films with high H content and low density as well as in amorphous a-C films, grown by mass selected ion beam deposition (MSIB), which have a low or zero H content and densities as well as hardness values approaching those of diamond. The latter, ta-C (tetrahedral amorphous carbon) films, which can also be grown by filtered vacuum arc deposition [46], have in recent years gained considerable scientific interest. Several models have been proposed to explain the formation of these very dense amor hous carbon phases [39], [47]. A prerequisite to obtain maximum density and sp content is a suitable average kinetic energy of carbon ions in the 20-600eV region. If the energies deposited per carbon atom during film growth are much smaller, e.g. during evaporation of graphite ( E M 0.1 eV) or plasma deposition of an a-C:H film from a hydrocarbon at small substrate bias voltages, soft films of low density and hardness will be deposited, consisting predominantly of sp2-bonded carbon (in a-C) and strongly hydrogenated sp3bonded carbon (in a-C : H). Among the methods mentioned above, IR spectroscopy is by far the most inexpensive and - if the analysis of absorption band intensities is done with great care - a powerful tool for structural investigations of hydogenated amorphous carbon films. The interrelation of deposition parameters and structural data of Table 3 which were reported by Dischler [48], are partly based on IR spectroscopic data (namely the ratios of sp"-hybridized carbon atoms and ratios of CH,-bonded segments) and can be considered as typical for plasma deposition of a-C :H. Growth of a dense, hard phase with high refractive index is favored by high ion energies of carbon-containg species deposited on the substrate, which are achieved under conditions of high bias voltages and relatively low pressures. In case of the softer a-C: H film, a large amount of hydrogen is present in methyl groups which do not contribute to the rigidity of the film. For the harder sample, methyl groups are virtually

i:

3.3 Microstructure and Bonding of Diamond-like Carbon

633

Table 3. Deposition conditions and structural properties of hard and soft, polymer-like a-C: H films (after [48]).

~~

a-C :H (hard)

a-C : H (soft)

2.3 1.O 3.2 100 50 68:30:2 0 :40 :60 1.3 2.0 1.65

2.3 0.1 6.1 10 25 53:45:2 25:60: 15 3.0 1.65 1.3

~

Frequency (MHz) Self bias voltage (kV) Benzene pressure (Pa) Estimated kinetic ion energy (eV) Deposition temperature ("C) sp3 : sp2 :sp (%) CH3 : CH2 :CH (%) Optical gap (eV) Refractive index Density (g ~ m - ~ )

absent and all sp3-bonded carbon atoms make at least two bonds to neighboring carbon atoms, thus contributing to the cross-linking of the network. Quantitative investigations of the connectivity in a-C: H random networks were introduced by Robertson [49] and Angus [50], based on the pioneering work on random covalent networks of Phillips, Thorpe and Dohler et al. [51-531. In considerations of this kind, the number of translational degrees of freedom of atoms (= 3 for one atom in three-dimensional space), forming the random network, are balanced out against the number of constraints N,,, on their movement imposed by bonds to neighboring partners. It can be shown that in three-dimensional networks whose bonding is dominated by directed valence bonds to nearest neighbors the number of constraints for one atom is Ncon(r)= r2/2 for Y 5 2 and Ncon(r)= 5 r / 2

3 for r 2 2

(1) r is the coordination number. As long as there are less constraints than degrees of freedom in a random network, atoms can still move in order to minimize strain energy, the network is 'floppy'. As soon as the degrees of freedom are exhausted by an increased number of bonds, the network will be fully constrained, additional cross-linking will then increase the strain energy rapidly. For a monatomic network this happens at an average coordination number r,, = 2.4. For a completely constrained network consisting of sp2- and sp3-bonded carbon atoms as well as hydrogen atoms with molar ratios x3,x4 and x I , Angus [50] deduced an equation for the sp3/sp2ratio (x4/x3)as a function of the hydrogen content xl : xq/x3 =

(8x1 - 3)/(8

13x1)

-

(2) In order to be composed exclusively from sp3 carbon atoms (corresponding to a zero denominator in Eq. (2)) and still be not overconstrained, a network should, according to Eq. (2), contain a hydrogen atom fraction x1 of 8/13 = 0.615 [54]. Modification have to be introduced, however, in order to take account of the clustering of sp2-bonded carbon [55]. In early papers on DLC structure and properties, extended clustering of the sp2-bonded carbon atoms to graphite-like aromatic planes was invoked in order to explain, among other properties, the small band gaps of a-C and a-C:H. For a band gap of, say, 1.2eV, typical for hard a-C:H, -

634


clusters involving up to 25 benzene rings are required according to this assumption [40]. Modern atomistic simulations of the detailed electronic structure of a-C and aC :H, however, point towards much smaller, mainly chainlike clusters, containing only 2-10sp2-bonded atoms. The small band gap is a result of the distortion of these structural elements, as discussed in more detail in [55]. In any case, the fact that ta-C films can be prepared, with sp3 carbon fractions in excess of 80% and virtually no hydrogen, demonstrates, that the range of accessible sp3 contents extends far beyond the limits suggested by the considerations of network connectivity. The highly cross-linked, dense microstructure of amorphous carbon films, grown under the assistance of energetic particles, becomes also evident from diffusion and permeation experiments. As already pointed out by Angus and Jansen, the fact, that even after several years a virtually unchanged content of Ar atoms, entrapped during plasma or low-energy ion beam deposition from Ar-containing gas phases, can be detected in a-C:H films, shows that the Ar diffusion coefficient, at least in certain portions of the film, must be lower than lo-**cm2 s f l . In conventional polymers, the corresponding figures are in the lop9 to lop6cm2s-' region [54]. Using depth profiling of deuterium by SIMS after exposition of a-C:H samples, grown by RF plasma deposition from acetylene, to D 2 0 vapor (160h at 85°C and 8085% relative humidity), Klages et al. [56] showed, that a virtually hermetical sealing against humidity can be achieved: In films of 120 nm to 1.9 pm thickness, deposited with a substrate bias of at least -5OV, D incorporation could be detected only within the outer 20 to 50 nm thick regions. Only in a 150 nm thick film grown at a bias voltage of -1OV, a diffusion of D through the whole film thickness was observed. Conventional organic polymers and coatings as used for corrosion protection, on the other hand, are easily permeable to water, oxygen and ions.

3.3.2 Metal-containing Amorphous Carbon Films Structural investigations of metal-containing amorphous hydrogenated carbon films, prepared by co-evaporation or cosputtering of a polymer and a metal, plasma polymerization of a metal-containing monomer, or reactive sputtering of a metal or metal carbide in a hydrocarbon atmosphere, have generally pointed to a granular structure of the deposits, containing the metal or (in case of carbide formers) a metal carbide as nanocrystalline segregates in an amorphous matrix. A review of the relevant literature until 1989 was published by Klages and Memming [32]. Structural information came from transmission electron micrographs TEM and mainly from X-ray diffraction (XRD) measurements, revealing the crystallographic phases formed and (by application of the Scherrer equation D = h/r cos 8) allowing an estimation of the average size of the particles D from the X-ray wavelength h and the width r of a reflection on the 28 scale of scattering angles. Reactively sputtered MeC :H films containing not too high concentrations of refractory metals Me = Ti, V, W, Mo, Nb were found to contain carbide particles MeC with sizes between about 1 and 10 nm typically. For iron-containing films the presence of the carbide Fe5C2 in superparamagnetic segregated nanocrystals was inferred from magnetic and Moljbauer spectroscopic measurements.

3.3 Microstructure und Bonding of Diamond-like Carbon

635

The microstructure of the amorphous a-C: H matrix surrounding the metal or metal carbide particles in MeC:H, especially its sp3/sp2 ratio, is still largely unknown. The application of IR absorption spectroscopy, for example, is possible only for very small metal contents, owing to the rapidly increasing optical absorption induced by the metal. A dominance of hydrogen bonding to sp3 carbon atoms was indicated, for example, in Ta-C: H films containing 4 at-% Ta. The systematic shift of the G-line in Raman spectra of W-C: H towards larger wavenumbers (1 528 . . . 1564cm-') with increasing W contents (7 . . . 2 5 at-%) was used by Schiffman to argue in favor of a decreasing sp3 content [57]. The hydrogen concentration generally decreases with growing metal content: In Ta-C : H, a H/C ratio of about 0.48 is found by Rutherford backscattering analysis (RBS) for a virtually Ta-free film grown from a methane-containing atmosphere, declining to 0.22 in films with a Ta/C atomic ratio in excess of 0.66 [32]. Using secondary ion mass spectrometry, similar results have been obtained for other metals, too [57, 581. In recent years the application of small angle X-ray scattering (SAXS) working with synchrotron radiation has turned out to be valuable tool for an analysis of the nanostructure of MeC:H films [57, 59-61]. SAXS allows an investigations of structural inhomogeneities of solids, resulting in a modulation of the local electron density, on length scales in the 1-100nm region. Figure 3 shows SAXS curves as obtained from measurements on reactively sputtered WC: H films with a varying content of W [57]. In all cases investigated so far in detail (Me = Au, Pt, W, Fe) the scattering intensity at large values of the scattering vector q (Porod region) (typically q > 3.0nm-') decreases as qP4, indicating a relatively sharply defined, smooth surface of the particles. In the medium q region (Guinier region, typically 10 < qnm < 3), a maximum or shoulder is usually observed, permitting a rough estimate of the inter-particle distance d = 27c/q,,,. The frequently observed increase of intensity towards the lowest q values used points to the existence of a relatively small number of particles having a much larger diameter and therefore a significant W-C:H

0,Ol

q

0,'

Figure 3. Small angle X-ray scattering (SAXS) curves for WC: H films with differing metal contents. Continuous lines = simulation, see text. (From [57] by courtesy of K. I. Schiffmann).

636


volume fraction, among a majority of smaller ones. The continuous lines in Figure 3 are simulated curves, based on a model of hard homogeneous spheres in a homogeneous matrix. In order to represent the particle sizes, a combination of two log-normal distributions was used, one main distribution representing the majority of particles with a relatively small width, plus an additional broad background distribution representing few, much larger particles. Particle radii and distances ranging from 0.6 to 0.8nm and 2 to 3nm, resp. could be deduced from these investigations for W-C:H films with W contents between 7 and 25 at-%. Metal and carbide particle distances and radii in MeC:H films have also been measured by applying scanning tunnel microscopy (STM) imaging. In order to take into account tip convolution effects, leading to an appearant particle enlargement and to particle hiding, statistical methods were proposed by Schiffmann et d., using an off-line tip radius determination by analysing appearant particle radii and a Monte Carlo model to compute fractions of hidden particles. A very good agreement was generally observed for radii and distances as measured by STM, TEM, SAXS and (for the radii only) XRD [57, 601. The anisotropies in two-dimensional SAXS measurements as observed by Fryda [59] using sample tilting with respect to the incoming synchrotron radiation beam are, according to TEM micrographs of film cross sections [57], not due to an elongated shape of the segregated particles, but possibly to a chain-like ordering along the growth direction. The particles themselves are usually nearly spherical and do not exhibit a pronounced shape anisotropy along the growth direction. The average distance d of the particles, at a given volume fraction directly related to their average size, correlates quite closely with the melting point T, of the corresponding bulk metal. The plot against T i 1 in Figure 4 is based on the evaluation of the position of the maximum or shoulder in the Guinier region of SAXS scattering curves for a number of MeC:H films [59] plus the data for Au from [57]. Qualitatively, an increasing mobility of the atoms of lower melting metals 80 70

60

.a 50 \

TJ

40 30

20 10

2

3

4 T~ ;

5 6 I I O - K-? ~

7

8

Figure 4. Plot of the average center of mass distances d (in A) of segregated metal and carbide particles, respectively, in MeC: H films against the reciprocal melting points l/T, of the corresponding bulk metals, according to data taken from [59] and, for Au, [57].

3.4 Physical Properties of DLC Filnis

637

during film growth and segregation of the metal or metal carbide phases can be made responsible for their increasing sizes and distances.

3.4 Physical Properties of DLC Films 3.4. Electrical and Optical Properties The oom temperature electrical conductivity of amorphous carbon varies over R-l ern-', depending on the more than 15 decades from around 1 to growth methods and growth parameters used [49, 621. A fairly exhaustive review, including data, mostly measured over a range of temperatures, for samples prepared by evaporation, sputtering, ionized carbon and hydrocarbon deposition, laser evaporation, arc discharge deposition, and plasma deposition has been published by Frauenheim et al. [63]. The temperature dependence of conductivity has frequently been discussed in terms of Eq. ( 3 ) [64, 6.51:

o ( T )= ooexp{-(To/T)n}; n

=

I . . .1 /4

(3)

The value of n depends on the conduction mechanism: n = 1 represents conduction by activation to extended states as in typical crystalline semiconductors, or by hopping (i.e., phonon-assisted quantum-mechanical tunneling) between localized states on nearest-neighbor sites, frequently found in highly doped crystalline and in amorphous semiconductors at sufficiently high temperatures. For decreasing temperatures, it will become increasingly difficult to find a nearest-neighbor site with a suitable energy separation and it becomes more favorable for the electrons to make larger hops to more distant sites, in order to find a site within a suitable energy distance: The geometry of the most probable path will de end on the temperature, the value of n becoming smaller than 1. Mott's famous TPc4 law will be obtained as a limiting case for a constant DOS around the Fermi energy. In nonhydrogenated amorphous carbon films with small sp3 content, the transition to n = 1 behavior will happen only at relatively high temperatures, n < 1 has therefore been observed up to 300-500K. For hydrogenated films, on the other hand, n = 1 prevails down to about 200-300 K temperatures and only at lower temperatures a transition to variable range hopping occurs [63]. Meyerson and Smith were the first to grow and investigate a-C:H films doped with boron and phosphorus, by additions of diborane or phosphine to the hydrocarbon gas used for deposition [66]. With a dopant concentration of lo%, the room temperature conductivity increased by about five orders of magnitude to lo-' R-' cm-l. However, as shown by Thiele et al., this effect ist not a true p- or n-type doping in the conventional sense (i.e. a shift of the Fermi level towards the valence or conduction band) but an increase of localized states near the Fermi level, supporting a hopping conduction mechanism [67]. The mechanism of electrical conductivity in highly sp3-bonded, virtually hydrogenfree ta-C, having the highest band gap among the dense amorphous carbon phases,

638

3 Diamond-likeCarbon Films

and the question of its dopability is presently under discussion. Data published for undoped ta-C exhibit a large spread in room temperature conductivity, from to 10-12 0-1 cm-l , even for quite similar sp3 fractions [39]. Amaratunga et al. were the first to grow ta-C (filtered arc deposition) with additions of N, P, and B [68]. The conclusions drawn in this paper and in subsequent publications from this group have recently been critically reviewed by Hofsiiss [69] on the basis of their own experimental results. According to his investigations, the conduction in undoped ta-C is mainly due to hopping in band-tails states and (at higher electrical fields) Poole-Frenkel excitation of electrons from traps into the conduction band. Doping with N or B leads to an increase of conductivity by 3 4 orders of magnitude at room temperature, presumably due to a generation of localized states related to three-fold coordinated B or N sites, contributing to hopping as well as PooleFrenkel conduction. This view is supported by completely symmetrical I- V curves measured for layered diode-like structures containing N- and B-doped material. Electrical properties of MeC :H films, deposited by reactive sputtering from metal or metal carbide targets in hydrocarbon atmosphere, have been described elsewhere [32]. In case of Ta as a typical refractory, carbide-forming metal, the electrical behavior can be divided into three categories, depending on the metal content: a 'dielectric' region with thermally activated conductivity according to Eq. (3) at less than about 13 at-% Ta, - a transition region with metallic behavior (i.e. residual conductivity at vanishing temperature) but still a positive or zero slope of log(o)-T curves and o > lo2R-' cm-' at medium Ta contents (14 and 22 at-% Ta) and - a metallic region with with negatively sloped curves log(o)-T and o > lo3R-' cm-' for 44 and 94 at-% Ta.

-

In contrast to Au- and Cu-C:H, for example, showing a percolation-type sudden increase of room temperature conductivity at a medium metal content of several 10 vol-%, the o ( R T ) values in Ta-C:H films tend to increase continuously by about six orders of magnitude right from the lowest concentrations in nearly metal free films up to less than 10 at-% Ta. It is only at low temperatures, that an insulator/metal transition is observable as a four orders of magnitude jump in o(3 K) between 11.5 and 14.3 at-%. A detailed discussion of the electrical properties of MeC :H films is given elsewhere [70]. Similar to the electrical conductivities optical properties like absorption coefficients and refractive indices are critically dependent on the preparation conditions used to grow a-C or a-C :H. Except for absorption coefficients below about lop4 cm-', the photon energy E dependence of the absorption coefficient M can generally be described by the Tauc relation (4) [71]:

EM)"^ = G ( E - Eo)

(4)

The slope G varies relatively weakly among different kinds of amorphous carbon films, the optical gap, however, depends uite strongly on the H-content and sp3/sp2ratio of the films, owing to the fact, that sp -related 71: states (see Fig. 2) are mainly responsible

9

3.4 Physical Properties of DLC Films

639

for absorptions in and near to the visible region. Typical values for a-C and a-C: H are shown in Table 2. While an upper limit of 1.5eV is still quoted in this survey for ta-C, prepared by mass selected ion beam deposition, considerably higher values approaching or even exceeding 3 eV seem now feasible according to the review [39]. Thus, the very hard and wear resistant hydrogen-free ta-C films might gain importance for optical applications in the visible region. Refractive indices for ta-C films are typically close to the value of diamond (n = 2.4) [39]. Incorporation of hydrogen lowers n to values between about 1.8 and 2.3 in a-C:H. In plasma deposition of these films, hydrocarbon pressure P and substrate bias voltage UB are the most important growth parameters allowing a tuning of n. Dense films with high refractive indices are favored by low P and high UB,To a first approximation, holding for n up to 2, constant refractive indices can be achieved for a constant ratio U B / P[62].

3.4.2 Mechanical Properties For tribological applications of diamondlike carbon films, hardness, H, and stiffness, measured as the Young modulus, E, are key physical properties. The hardness of a material is its resistance against permanent geometric deformation under mechanical loading. In indentation testing, it is equivalent or proportional to the average pressure under a sharp indenter and can be calculated from the applied pressure divided through the (projected) contact area between the indenter and the material tested. For hardness measurements on thin films of less than 1 to several pm thickness, conventional microhardness measurements with microsocopic evaluation are difficult due to the small sizes of the indents. Therefore nanoindentation techniques working with a simultaneous registration of force and indentation depth with resolutions of a few pN and nm, resp. have been introduced [72]. Aside from the hardness, Young modulus can be estimated, using these methods, from the slope of the unloading curve [73]. Jiang et al. applied the method for an investigation of a-C:H films grown by R F plasma deposition from methane and acetylene [74, 751. Hardness values between 5 and 20 GPa were typically found, correlating closely with Young moduli of the films, H z 0.1 E . In MeC: H coatings, the metal content is the prime factor determining H a n d E [76,35]: For WC : H and Ta-C : H, H increases from around 20 GPa for films with virtually zero or only a few vol-% of MeC precipitation to about 40 GPa at about 50 vol-%. Over the same range of metal contents E grows from 100 to 200 GPa ( H I E = 0.2). Even higher H and E values, approaching those of diamond (z100-120 and 1 100 GPa) are characteristic for hydrogen-free, highly tetrahedrally coordinated ta-C films as to be obtained by ion beam or vacuum arc deposition, owing to the high average coordination numbers r,, 2 3.6 in these networks [39]. Based on their characteristic chemical bonding and microstructure, as well as mechanical properties, a gross division of technically important materials into the three categories metals, polymers and ceramics is frequently made. It is tempting to compare the spectrum of properties of DLC films to those of these main material groups. In [33] this has already been done for the narrower class of tribologically

640


interesting MeC: H films with low metal contents up to about 20 at-%. However, the essential results basically hold also true for DLC coatings in general. According to these considerations, the exceptional mechanical-tribological properties of DLC can be attributed to ~

-

-

-

hardness values (~10-100GPa) similar or even exceeding those of ceramic inorganic materials (10-20 GPa) and by far larger than for metals (< 8 GPa), elastic moduli (~100-1000GPa) similar or exceeding metals and ceramics (< 600 GPa), combined with elasticities as measured by H / E ratios (0.1-0.2) which are throughout larger than for metals (<0.03) and ceramics (<0.06) and approach the lower limit of a range which is more typical for organic polymers (0.2-0.5), and for hydrogenated film surfaces, surface energies (30-50 x N/m) in a region N/m) and typical for polymers and much less than for metals (300-2000 x ceramics (200-1300 x lop3N/m) (see [33] for references to the sources used).

Friction coefficients, being not a physical property in the strict sense, but depending on the contact partners and many other external parameters, will be reported in the next section. The combination of low friction coefficients, typically 5 0.2, with mechanical and surface properties which are partly polymer-like (elasticity, surface energy) and partly ceramic- or metal-like (hardness, Young modulus) qualitatively explains the outstanding position of DLC films as low-friction, highly wear-resistant coatings on which most of their present applications are based [77].

3.5 Applications of DLC Films Applications of DLC films in several fields like optics, magnetic disks, tribology and machine elements or biomedical purposes have been reported in the literature. A review on this topic was given by Matthews and Eskildsen [78]. The dominating field of application at present is that of tribological coatings on machine parts and tools. This shall be discussed in more detail. However, for all applications of the coatings a sufficient adhesion is required. Therefore this question will be treated first in the next section.

3.5.1 Adhesion of DLC Films The high compressive stresses in hard a-C: H and ta-C films (up to 10 GPa, depending on the deposition conditions, especially the bias voltage) [79] frequently lead to film delamination and thus limit the attainable film thickness. For many years thicknesses of more than 1-2 ym could not be realized on technically relevant substrates like steel, hardmetal or glass. Considerable improvements were achieved by applying intermediate layers consisting of aSi:H [SO], metals (Al, Ti, Ni) [81, 821, or multilayers (Ti-TiCN-Tic) [83], for example. For the characterization of

3.5 Applicutions OfDLC Films

641

0

. =.

0

b

0

o

0

0 ' 0

without

I

I

2

4

6

8

1

0

1

2

1

4

6[ P ~ I Figure 5. Critical load L, against a-C: H film thickness df for deposition with Al- and Ti- interlayers or without interlayers, respectively, on polished steel substrates.

adhesion the critical load, L,, or the crack pattern caused by a Rockwell indentor is frequently used. The critical load is a measure determined by scratch tests [84], utilizing a diamond tip moving over the film with continuously increasing normal load. L, is the load where an intensive acustic emission occurs, the friction coefficient between tip and sample drastically increases or a failure of the film can be observed by a microscope. All three criterions are in use. Therefore a comparison of data from different laboratories is difficult. Figure 5 shows the effect of aluminum and titanium interlayers on the critical load as a function of the a-C:H film thickness df.L, was determined with the second and third criteria. The values determined with both methods were nearly identical. These films were prepared by R F plasma deposition from acetylene after magnetron sputtering of the metal interlayers (thickness about 0.5 pm) onto polished steel substrates in the same coating machine. In particular in the range df > 3 pm a pronounced improvement of the adhesion was found. The hardness of these coatings was about 25GPa. Even films thicker than lOpm having excellent adhesion can be achieved by this method. It should be noted that the effect of interlayers is still more pronounced if substrates with rougher surfaces, like ground substrates are to be coated. However, the reliable and reproducible preparation of well-adhering a-C :H films is commonly not a simple task and requires a good control of several process parameters. Traces of oxygen in the films, to give just one example, can cause a drastic deterioration of the film adhesion. On the other hand, there are very reliable industrial processes for deposition of MeC: H coatings. MeC: H films have lower stresses than a-C: H (< 1 GPa) [34, 851 Therefore their tendency to delaminate is considerably reduced. Furthermore, intermediate layers can be prepared very easily using non-reactive sputtering from

642

-


140[

ap 120

I

E

100

L

a

80Q

>

'=m PI K

6040 -

"

Hard metal

steel (100Cr6)

Ti N

MeC

D LC

Figure 6. Abrasive wear of DLC coatings compared to other coating and substrate materials. As abrasive medium S i c with a particle dimension of 12 Fm was used.

the metal target (see Section 3.2.3.), before the MeC: H deposition process is started.

3.5.2 Tribology of DLC Coatings After satisfying solutions for the adhesion problem had been developed, DLC has become very interesting for tribological applications owing to the very favorable combination of low abrasive and adhesive wear with low friction coefficients. The following discussion concerns typical data on wear and friction of DLC films, but has to be taken into account that these properties can be influenced by the deposition conditions like bias voltage, gas pressure etc. Figure 6 shows the rates of abrasive wear for different types of hard coatings and reference materials. MeC: H and still more pronounced a-C: H exhibit excellent values. In Fig. 7 adhesive wear rates, measured in a pin on disc experiment, are presented. As pin a ball of bearing steel (100Cr6) was used. Figure 7a shows the results for MeC :H and TIN coated discs as well as for a steel disc, Fig. 7b the results for the 100Cr6 counterpart (pin). In the case of MeC:H the extremely low wear of the counterpart is an evidence of the low material transfer between steel ball and coating. It should be mentioned that the wear rates of MeC:H films depend on the metal content. However, for different metals [32, 851 and both for R F and DC magnetron sputter deposition the minimum values of abrasive wear were determined to be in the concentration range of 0.1 to 0.2 (metal to carbon ratio). Friction coefficients, p, against steel (steel 100Cr6) are shown in Fig. 8. While the friction coefficient of titanium nitride strongly increases with growing load due to a material transfer of steel onto the coating, the p values of WC:H (Me/ C ratios in the above mentioned range) even decrease with increasing load. The friction coefficients of metal-free a-C : H are very close to those of MeC :H. Very low friction coefficients (< 0.2 against 100Cr6) were also determined for laser-arc

3.S Applications of DLC Films

643

wear of steel counterpart (b)

wear of coatings and steel disk (a) 1ooc 0 w

looor

w 0

.-: s4-

1oc

?!E L

gF V :g .->' v)

aJ

1c

1

r

-0

a

0.1

Y

AI E Y2

r"

Figure 7. Adhesive wear measured in a ball on disk arrangement: (a) MeC: H and TiN coated discs and a disc of ball bearing steel 100Cr6 and (b) wear of the counterpart (100Cr6 ball). Measurements were performed under ambient conditions.

deposited films [86]. The friction coefficients of DLC films can considerably be lowered further by incorporation of silicon. Minimum p values were reported for films with 1020 atom-% Si [87, 881. Another important factor influencing the friction properties is the humidity. For a-C:H and W C : H (W/C < 0.4) films the friction coefficients

o'2

I

01

0.01

I

0.1

1

10

Normal load "1

Figure 8. Friction coefficients against normal load for MeC:H, TiN, TiAlN and steel measured against ball bearing steel under ambient conditions (up to about 50% relative humidity).

644


drastically declined from p > 0.2 to p < 0.05 with decreasing relative humidity (< 1%) [89,90]. For silicon-containing films very low p values (<0.1) were measured even up to relative humidities of 75% [88, 911. A possible reason for the different friction behavior could be a material transfer from the film to the steel ball (decreasing p) or in the opposite direction (increasing p) [90]. In the case of siliconcontaining films reactions of Si with the atmosphere were assumed. The generated silicon ceramics are known to have extremely low friction coefficients in water [911. However, there are still several open questions concerning the friction mechanisms of DLC films. It should be noted here that any published values of friction coefficients must be considered with caution because measuring conditions like humidity, loads, material of counterparts or sliding distance, which can have a strong influence on the result, can be extremely different [78].

3.5.3 Tribological Applications DLC coatings have been applied successfully for industrially used components and tools to minimize wear and friction and in some cases to avoid the use of lubricants. For example gears, ball bearing races, valves, knives and other parts have been coated with WC:H, see [78]. Roth et al. [17] coated blades for fibreboard material and stamping tools with a-C: H. There is a great application potential for DLC films in the field of forming tools, such as for aluminum- and titanium-based alloys [92]. With coated tools the cost intensive and ecologically harmful employment of forming lubricants can be avoided or at least diminished. By incorporation of certain elements the wetting behavior of DLC can be modified considerably. The surface tension of a-C :H can be decreased by additions of fluorine or silicon and oxygen, while it is enlarged by the incorporation of boron and nitrogen [81]. Low surface tension coatings have been applied successfully to reduce the adhesion of organic material on tools used for polymer fabrication. Another interesting application is the improvement of the heat transfer in steam condensation heat exchangers, thanks to the fact, that the water vapor does not form a heat-insulating water film on the coated heat exchanger surface, but it condenses in form of droplets which easily run down the vertically arranged heat exchanger, exposing a virtually water-free surface.

3.5.4 Other Applications Due to their chemical inertness, high abrasion resistance, low friction and good biocompatibility DLC films are very promising candidates for biomedical applications. For example, femoral heads of hip protheses have been coated successfully with DLC to reduce the production of small wear particles, which can lead to reactions on a cellular level. Furthermore DLC was shown to be a potential material for artificial heart valves where the films must be non-thrombogenic and must have a long-time stability in contact with blood [93]. Mitura et al. [94] investigated DLC

References

645

films prepared by ion beam decomposition, R F plasma and pulsed arc deposition, for coating implants. In spite of different film properties all three materials turned out as very suitable. Ion beam deposited DLC films are used to increase the abrasion and chemical resistance of sunglass lenses, of float-glass windows in supermarket laser scanners and of infrared windows [95].Very thin amorphous carbon films are used to protect magnetic disks and heads. Recently it was found that especially nitrogen containing carbon films (CN,) have an excellent wear resistance (about four-fold value of a-C) [96]. One reason for this difference is presumably the different roughness of the films. Using an atomic force microscope (AFM) roughness values r.m.s. of 0.25 nm and 0.63 nm, respectively, were measured for CN, and for a-C. For the application of coatings, also cost and benefit aspects must be considered. It shall be noted here that Matthews and Eskildsen [78]discussed this topic in detail for four different DLC production methods taking into account prices for deposition equipments to be obtained in the UK.

References 1. L. Holland and S. M. Ojha, Thin Solid Films 1976,38,L17-LI9. 2. D.S. Whitmell and R. Williamson, Thin Solid Films 1976,35,255-261. 3. A. Grill and V. V. Patel, Diamond Films Technol. 1992,4,219-233. 4. L.Holland, J . Vac. Sci. Technol. 1977,14, 5-15. 5. K. Bewilogua and D. Wagner, Vacuum 1991,42,473476. 6. Y.Catherine and P. Couderc, Thin Solid Films 1986, 144, 265-280. 7. C.Weissmantel, K. Bewilogua, K. Breuer, D. Dietrich, U. Ebersbach, H.-J. Erler, B. Rau, and G. Reisse, Thin Solid Films 1982,96, 3144. 8. C. Weissmantel, in Thin Films from Free Atoms and Particles, J. Klabunde (Ed.), Academic, New York, 1985,pp. 153-201. 9. T. Namba and Y. Mori, J . Vac. Sci. Technol. 1983,Al, 23-27 10. M. W. Kroesen, D. C. Schram, and M. J. F. van de Sande, Plasma Chem. Plasma Proc. 1990, 10,49-69. 11. L.Martinu, A. Raveh, A. Domingue, L.Bertrand, J. E. Klemberg-Sapieha, S. C. Gujrathi, and M. R. Wertheimer, Thin Solid Films 1992,208,4247. 12. B. Chapman, Glow Discharge Processes, Wiley, New York, 1980,Chapter 5. 13. A. Bubenzer, B. Dischler, G. Brandt, and P. Koidl, J . Appl. Phys. 1993,54,45904595. 14. J. Robertson, Surf. Coat. Technol. 1992,50, 185-203. 15. W. Moller, Appl. Phys. A 1993,56, 527-546. 16. H. Kersten and G. M. W. Kroesen, J . Vac. Sci. Technol. 1990,A8, 3842. 17. D. Roth, B. Rau, S.Roth, J. Mai, and K.-H. Dittrich, Surt Coat. Technol. 1995,7475,637-641. 18. S.Aisenberg and R. W. Chabot, J . Appl. Phys. 1971,42,2953-2958. 19. Y. Lifshitz, Diamond Re/. Muter. 1996,5, 388400. 20. J. J. Cuomo, D. L. Pappas, J. Bruley, J. P. Doyle, and K. L. Saenger, J . Appl. Phys. 1991,70, 1706171 1. 21. H. Hofsass, H. Binder, T. Klump, and E. Recknagel, Diamond Rel. Muter. 1993,3, 137-142. 22. N.Savvides, Thin Solid Films 1988, 163, 13-32. 23. N . Savvides and B. Window, J . Vac. Sci. Technol. 1986,A4, 504-508. 24. A. Anders, S. Anders, and I. G. Brown, Plasma Sources Sci. Technol. 1995,4, 1-12. 25. H.-J. Scheibe, B. Schultrich, and D. Drescher, Surf. Coat. Technol. 1995,74-75, 813-818.

646


26. H.-J. Scheibe, D. Drescher, B. Schultrich, M . Falz, G. Leonhardt, and R. Wilberg, Surf: Coat. Technol. 1996, 85,209-214. 27. A. A. Voevodin and M. S. Donley, Su$ Coat. Technol. 1996,82, 199-213. 28. F. Rossi and B. Andre, Jpn. J. Appl. Phys. 1992, 31, 872 -879. 29. J. Robertson, Diamond Rel. Muter. 1993, 2, 984-989. 30. H. Dimigen and H. Hubsch, Philips Tech.Rev. 1983/84,41, 186-197. 31. H. Dimigen and H. Hiibsch, R. Memming, Appl. Phys. Lett. 1987,50, 10561058. 32. C.-P. Klages and R. Memming, Muter. Sci. Forum, 1989, 52-53, 609-644. 33. H. Dimigen and C.-P.Klages, Surf. Coat. Technol. 1991,49, 543-547. 34. E. Bergmann and J. Vogel, J . Vac. Sci. Technol. 1987, A5, 70-74. 35. K.Bewilogua and H. Dimigen, Surf. Coat. Technol. 1993,61, 144-150. 36. D. Hofmann, H. Schuessler, K. Bewilogua, H. Hubsch and J. Lemke, SwF Coat. Technol. 1995, 73, 137-141. 37. J. T. Harnack and C. Benndorf, Diamond Rel. Muter. 1992, 1, 301-306. 38. F. L. Pilar, Elementary Quantum Chemistry, MacGraw-Hill, New York, 1968. 39. Y. Lifshitz, in The Physics of Diamond, A. Paoletti and A. Tucciarone (Eds), 10s Press, Amsterdam, 1997, p. 209. 40. J. Robertson, in Diamond and Diamond-like Films and Coatings, R. E. Clawing et al. (Eds), Plenum, New York, 1991, p. 331. 41. S. Kaplan, F. Jansen, and M. Machonkin, Appl. Phys. Lett. 1985, 47, 750-753. 42. C. Weissmantel, K. Bewilogua, D. Dietrich, H.-J. Erler, H.-J. Hinneberg, S. Klose, W. Nowick, and G. Reisse, Thin Solid Films 1980, 72, 19-31. 43. S. Miihling, K. Bewilogua, and K. Breuer, Thin Solid Films 1990, 187, 65-75. 44. J. Fink, Th. Muller-Heinzerling, J. Pfluger, B. Scheerer, B. Dischler, P. Koidl, A. Bubenzer, and R. E. Sah, Phys. Rev. B 1984, 30,47134718. 45. J. Robertson, Diamond Rel. Mater. 1992, 1, 397406. 46. D. R. McKenzie, D. Muller, B. A. Pailthorpe, Z. H. Wang, E. Kravtchinskaia. D. Segal, P. B. Lukins, P. D. Swift, P. J. Martin, G. Amaratunga, P. H. Gaskell, and A. Saeed, Diamond Rel. Muter. 1991, 1, 51-59. 47. H. Hofsass, H. Feldermann, R. Merck, M. Sebastian, and C. Ronning, Appl. Phys. A , 1988, A66, 153-181. 48. B. Dischler, in Proceedings of E-MRS Symposium on Amorphous Hydrogenated Carbon, Symposium C, Strasbourg, France, Les Editions de Physique, Paris, 1987, XVII, pp. 189-201. 49. J. Robertson, Adv. Phys. 1986,35, 317-374. 50. J. C. Angus, in Proceedings of E-MRS Symposium on Amorphous Hydrogenated Carbon, Symposium C , Strasbourg, France, Les Editions de Physique, Paris, 1987, XVII, pp. 179-188. 51. J. C. Phillips, Phys. Rev. Lett. 1979, 42, 153-181. 52. M. F. Thorpe, J. Non-Cryst. Sol. 1983, 57, 355-370. 53. G. H. Dohler, R. Dandoloff, and H. Bilz, J. Non-Cryst. Sol. 1980, 42, 87 96. 54. J. C. Angus, F. Jansen, J. Vac. Sci. Technol. 1988, A6, 1778-1782. 55. J. Robertson, Diamond Rel. Mater. 1997, 6, 212-218. 56. C.-P. Klages, A. Dietz, T. Hoing, R. Thyen, A. Weber, P. Willich, Surf. Coat. Technol. 1996,80, 121-128. 57. K. I. Schiffmann, Thesis, University of Hamburg, published as Fraunhofer-IST-Berichte aus Forschung und Entwicklung, Fraunhofer IRB Verlag, Stuttgart, 1997, Nr. 1. 58. P. Willich and R. Bethke, in Secondarji Ion Mass Spectroscopy ( S I M S X i , A. Benninghoven, B. Hagenhoff and H. W. Werner, Wiley, New York, 1996. 59. M. Fryda, Fortschritt-Berichte VDI, Reihe 5 , VDI Verlag, Dusseldorf, 1993, 303, p. 20. 60. K. I. Schiffmann, M. Fryda, G. Goerigk, R. Lauer, and P. Hinze, Ultramicroscopy 1996, 66, 183-192. 61. K. I. Schiffmann, M. Fryda, and G. Goerigk, Mikrochinz. Acta 1997, 125, 107-1 13. 62. J. C. Angus, P. Koidl, and S. Domitz, in Plasma Deposited Thin Films, J. Mort and F. Jansen (Eds), CRC, Boca Raton, FL, 1986, p. 89. 63. Th. Frauenheim, U. Stephan, K. Bewilogua, F. Jungnickel, P. Blaudeck, and E. Fromm, Thin Solid Films 1989, 182, 63-78. 64. H. Overhof, P. Thomas, Electronic Transport in Hydrogenated Amorphous Semiconductors,

References

647

Springer Tracts in Modern Physics, Vol. 144, G. Hohler (Ed.), Springer Verlag, Berlin, 1989, p. 10.

65. R. Zallen, The Physics ofAmorphous Solids, Wiley, New York, 1983, p. 252. 66. B. Meyerson and F. W. Smith, Solid State Commun. 1982, 41, 23-27. 67. J. U. Thiele, B. Rubarth, P. Hammer, A. Helmbold, B. Kessler, K. Rohwer, and D. Meissner, Diamond Re/. Muter. 1994, 3, 1103-1 106. 68. G. A. J. Amaratunga, D. E. Segal, and D. R. McKenzie, Appl. Phys. Lett. 1991, 59, 69-71. 69. H. Hofsass, in Proc. 1st Int. Specialist Meeting on Amorphous Carbon ( S M A C '97). Amorphous Carbon: State of the Art, S.R.P. Silva et al., (Eds.), Singapore: World Scientific, 1998, p. 296&310. 70. H. Koberle, PhD Thesis, University of Hamburg, 1989. 71. J. Tauc, R. Grigorovici, and A. Vancu, Phys. Stat. Sol. 1966, 15, 627-637. 72. P. Wierenga, A. J. J. Franken, J . Appl. Phys. 1984, 55, 4244-4248. 73. J. L. Loubet, J. M. Georges, 0. Marchesini, and G. Meille, J . Tribol. 1984, 106, 4 3 4 8 . 74. X. Jiang, K. Reichelt, and B. Stritzker, J . Appl. Phys. 1989, 66, 5805-5808. 75. X. Jiang, K. Reichelt, and B. Stritzker, J . Appl. Phys. 1990, 68, 1018-1022. 76. M. Fryda, K. Taube, and C.-P. Klages, Vacuum 1990,41, 1291-1293 77. E. Rabinowicz, Friction and Wear of Materials, Elsevier, Amsterdam, 1987. 78. A. Matthews and S. S. Eskildsen, Diamond Re/. Muter. 1994, 3, 902-91 1. 79. M. Tamor, in Proc. Third Int. Conf on Applications of Diamond Films and Related Materials, A. Feldman et a/. (Eds), NIST Special Publications 885, 1995, p.691. 80. R. Butter, M. Allen, L. Chandra, A. H. Lettington. and N. Rushton, Diamond Re/. Muter. 1995, 4, 857-861. 81. M. Grischke, K. Bewilogua, K. Trojan, and H. Dimigen, Surf: Coat. Technol. 1995,7475,739745. 82. H. Ronkainen, J. Vihersalo, S. Varjus, R. Zilliacus, U. Ehrnsten, and P. Nenonen, Surf. Coat. Trchnol. 1997, 90, 190-196. 83. D. P. Monaghan, D. G. Teer, P. A. Logan, I. Efeoglu, and R. D. Arnell, Surf. Coat. Technol. 1993, 60, 525-530. 84. P. J. Burnett and D. S. Rickerby, Thin Solid Films 1988, 157, 233-254. 85. M. Grischke, K. Bewilogua, and H. Dimigen, Mater. Manuf. Proc. 1993, 8, 407417. 86. H.-J. Scheibe and D. Klaffke, Surf. Coat. Technol. 1993, 57, 111-1 15. 87. S. Miyake, Surf: Coat. Technol. 1992, 5455,563-569. 88. K. Oguri and T. Arai, Thin Solid Films 1992, 208, 158-160. 89. K. Enke, H. Dimigen, and H. Hiibsch, Appl. Phys. Lett. 1980,36,291-292. 90. R. Memming, H. J. Tolle, and P. E. Wierenga, Thin Solid Films 1986, 143, 3 1 4 1 . 91. K. Oguri and T. Arai, J . Mater. Res. 1992, 7, 1313-1316. 92. K. Taube, M. Grischke, and K. Bewilogua, Surf. Coat. Technol. 1994, 68/69, 662-668. 93. R. S. Butter and A. H. Lettington, Proc. Third Int. Con6 on Applications of Diamond Films and Related Muterials, A. Feldman et aI. (Eds), NIST Special Publications 885, 1995, p.683. 94. S. Mitura, P. Niedzielski, D. Jachowicz, M. Langer, J. Marciniak, A.Stanishevsky, E. Tochitsky, P. Louda, P. Couvrat, M. Denis, and P. Lourdin, Diamond Re/. Muter. 1996, 5, 1185-1 188. 95. F. M. Kimock and B. J. Knapp, Surf. Coat. Technol. 1993,56, 273-279. 96. D. Li, E. Cutiongco, Y.-W. Chung, M.-S. Wong, and W. D. Sproul, Diamond Films Technol. 1995,5, 261-272.


4 Ceramics Based on Alumina: Increasing the Hardness for Tool Applications A. Krell

4.1 Recent Trends in the Application of Ceramic Tool Materials New trends in the development and application of hard ceramic grinding and cutting materials will be discussed in this Chapter. For both groups of tools, modern technical demands drive the development of submicrometer microstructures that exhibit significantly increased hardness and reliability. Manufacturing approaches and resulting properties will be described for both advanced single phase sintered alumina materials and for composite ceramics. The fundamental difference between grinding and cutting tools is the existence of a geometrically well-defined cutting edge for the latter group, whereas grinding is performed by a plurality of geometrically not defined edges and tips. Most grinding grits are inorganic nonmetallic materials (diamond, cubic boron nitride cBN, fused or sintered ceramic grinding materials based on alumina A1203 or silicon carbide Sic). For cutting wheels the share of these materials in large-scale manufacturing is reported to be more or less constant at only l4-l6YO, compared with about 80% for hard metals 111. Table 1 reviews the most important ceramic tool materials that are commercially available for the machining of metals, but this contribution will not discuss materials with limited applications (e.g. garnet grits for grinding wood and glass) or with super-hard characteristics (diamond or cBN). In recent years, the intensity of research intended to find basically new solutions was rather different for two main sectors. For grinding metals there was an avalanche of patent applications in the early eighties associated with the development of the first polycrystalline (sintered) grits, and this stream of innovations continues now. Here, among the great number of new applications there is only a small percentage of ‘speculative’ inventions concerning materials with unsure scientific and technical prospects (e.g. an invention claiming diamond reinforced alumina without describing the sintered dense body, thus without determination of technically relevant properties). On the contrary, there is a remarkably large number of inventions which are regarded as so important that an international (European or PCT (Patent Cooperation Treaty)) application is filed without a foregoing national priority. The similarly large number of oppositions on the one hand and of reproaches for the violation of claims demonstrate that such patent applications describe commercially highly sensitive subjects. Most of the recent inventions concern the technology of materials and tools based on polycrystalline sintered corundum (a-A1203),but there are also propositions to manufacture Sic grits with a polycrystalline microstructure (where the importance of S i c grits is

4.1 Recent Trends in the Application of Ceramic Tool Materials

649

Table 1. Commercially available tool materials for machining metals. HVl (testing load 1 kg) and HVlO (10 kg load) hardness values are given for grinding and cutting materials to avoid the misuse of microhardness data Vickers hardness Fracture toughness Bending strength KIC(MPam’”) (MPa) (GPa) Ceramic grinding grits Fused aluminas Brown corundum (ALO,: immrities) White corundum (kl2O3) Ruby (AI2O3:Cr) AI2O3/ZrO2eutectic Sintered corundum (sol/gel) with nearly equiaxed crystal shape with platelet-shaped crystals Sic &

>,

.

HV 1 17.0-18.5 19.0-20.5 18.5-20.5 15.5-17.5

1.O-2.5 2.0-2.2 2.4-3.1 4.0-5.0

-

14.0-22.5 18.7- 19.O 25.5-27.5

2.94.1 3.3-3.7 2.5-3 .O

-

HVlO 18.O-20.0 17.0-1 9.0 19.0-22.0 17.O-20.0 15.0-18.0 27.040.0 > 4500

3.6-4.5 5 .O-6.0 4.0-5.5 6.0-8.0 6.5-9.0 3.5-6.6 6.0-9 .O

500-600 600-750 550-800 650-950 750-1000 600-800 750-1000

-

-

Cutting tool ceramics A1203

+ 4% ZrOz + 15% Zr02 AI2O3+ Ti(C,N) [- 30%] A1203+ Sicwhisker[- 20%] Si3N, cBN cutting tools Diamond (polycrystalline)

generally less compared with alumina because of its poor ability to machine ferrous workpieces). On the other hand, there has been some stagnation in the field of tools with geometrically defined cutting edges after the development of advanced A1203/ ZrO, and A1203/TiC composites at the beginning and of tough Si3N4 cutting tools and Sic-whisker reinforced grades at the end of the eighties (high-strength tetragonal zirconia polycrystals cannot be used for machining metals). At least until 1997 it was the position of most manufacturers that they can satisfy the current requests of their customers with the available ceramics. Consequently, many laboratories are busy performing extended screening tests with commercial grades. A somewhat special position has to be attributed to efforts intended to improve the chemical resistance of Si3N4 by surface coatings. A very recent tendency is the increasing interest in a new class of materials with submicrometer microstructures, a trend which is discussed in Section 4.4. The strong motivation for these developments in grinding materials is the real possibility of new inventions able to surpass the present state of the art by 5&100% (in terms of rate of material removal or maximum time of use of a tool), the chance to find materials that redouble the productivity of the industrial grinding operation and reduce substantially the time-out for the change of tools. Significant recent changes in the market shares of the different kinds of cutting ceramics shown in Table 1 should be noted. There has been an increase in the use of ceramic cutting tools, attributed to increasing use of silicon nitride tools.

650

4 Ceramics Based on Alumina: Increasing the Hardness,for Tool Applications

Two different reasons exist. First, the requirement for high reliability in automated manufacturing lines (without running a tool to the possible maximum time of use) favors the application of tool materials with a high fracture toughness (like Si3N4) tolerating, in the mean time, their lower hardness. Second, a shift in the ratio of machined materials from steel towards globular and vermicular cast iron also requires tougher tools like silicon nitride, especially when high feed rates are necessary. A drawback of nitride tools is their rather high chemical reactivity with iron: against steel and cast iron (like GG25), at feed rates of about 0.3 mm rev-' the wear of Si3N4is usually higher than for A1203-basedceramics. Hence, even for machining cast iron, nitride tools can be applied with advantage only when it is impossible to use the less tough AI2O3/ZrO2or A1203/TiCcomposites due to the required high feed rates. Attempts have been made to overcome this problem by TIN and/or oxide coatings, but real success was obtained only in some very special cases. There is another trend, which has not yet altered market shares, because the new qualities are not commercially available: this is the substitution of certain grinding operations by highly precise cutting approaches, especially in machining very hard materials. It is expected that turning may cut the machining time to 1/3 compared with grinding, and is sometimes associated with significant savings of consumed energy, and even the compromise of first turning with subsequent final grinding reduces the time to 2/3 compared with an all-grinding process. Additional advantages are the less expensive recycling of turning chips compared with the waste disposal of the grinding silt and investment costs of the equipment which can be reduced by a half for precise turning [ 1,2]. On the other hand, investment in new machine tools is imperative for precise cutting of hard materials (the mechanical stability of the most machine tools used today in the automotive industry is probably not sufficient for the new demands). More widespread use of cutting rather than grinding is restricted by the limited lifetime of today's cutting ceramics and the increase in workpiece roughness after rather short cutting paths. Investigations by large-scale manufacturers indicate a lack of real alternatives to cBN for turning hard workpieces with high feed rates, and drilling and milling is the field of hard metal tools, but for precise finishing AI2O3/TiCcomposites exhibit an advantage in the cost-performance ratio [2]. It is expected that such findings will stimulate the development of significantlymore fine grained and harder cutting tool ceramics based on corundum. For example, it was in 1997 that a German company launched a new submicron A1203/Ti(C,N)composite. Indeed, such A1203/Ti(C,N) tools show the fewest problems when compared with other commercial ceramics (which, however, all have not been developed to meet the demands of precise machining of hard surfaces), but an analysis of the technical demands under thls special condition points to better opportunities with other ceramics (cp. Section 4.4.4).

4.2 Technological Essentials for Producing Hard and Strong Tool Ceramics It is beyond the scope of this paragraph to give a review of ceramic technologies, but the development of tools with an advanced technical performance needs an

4.2 Technological Essentialsfor Producing Hard and Strong Tool Ceramics

651

understanding of the required basic mechanical parameters, and of the real chances new technologies offer for their improvement. Also, it must be emphasized that the properties of different new tool materials can be compared only if the effect of their different microstructures is not obscured by processing-related imperfections: the investigation of new materials requires samples that are different in their microstructures but with similarly small frequency and size of defects. Much care has to be taken to perform this recent research (the results of which are described below) because for extremely fine-grained microstructures this demand is not easy to meet.

4.2.1 Typical Defects in Ceramics Tool Materials: The State of The Art It is known that defects reduce the strength of brittle materials due to stress concentrations, which cannot relax by plastic deformation, as in metals. However, there is also evidence that the flaw population has a similarly strong impact on the hardness. The most promising way to develop new tool materials with improved hardness, wear resistance, and reliability is to reduce the grain size in the sintered microstructures: an approach which requires the use of increasingly fine-grained raw materials, be it within the framework of advanced powder technologies or of sol/gel or precursor approaches. Unfortunately, it is just this trend towards extremely fine-grained, nanoscale, raw materials that often prevents the realization of the expected progress in the mechanical properties, simply because with the reduced particle sizes there is an increased risk of defects in the green microstructure.

4.2.1.1 Powder Processing All ‘dry’ powder pressing technologies require the powder to be granulated before compaction, otherwise the specific volume of non-agglomerated particles in or beyond the micrometer range is so great that on compaction the decrease of volume becomes excessive, causing extended elastic relaxation (and associated cracking) on removing the load. The key is to produce granules (by spray drying or other approaches) that are large and strong enough for technical handling but soft enough to be completely destroyed on compaction to avoid relicts of the granules in the pressed and finally sintered bodies (Fig. 1). Cold isostatic pressing can significantly reduce the size and frequency of such defects (provided there is optimized slurry preparation and granulation [3]), but it is impossible to avoid them entirely. Liquid shaping is a promising alternative. Avoiding granulation, a slurry of a dispersed powder is poured into a mold. A problem with liquid shaping is that homogenization and pouring introduces air into the slurry (the more the more fine-grained the powder). The complete elimination of gas from the slurry is difficult, and some porosity may appear in the green bodies that cannot be removed in solid state sintering.

652

4 Ceramics Based on Alumina: Increasing the Hardness for Tool Applications

Figure 1. Relicts of hard powder granules (from spray drying) in pressureless sintered alumina ceramics (solid state sintering).

4.2.1.2 Sol-gel Technologies One of the reasons for developing sol/gel and precursor technologies is the use of raw materials with particle sizes < 30-S0nm, enabling the production of very fine-grained microstructures. However, the raw material exists in a phase composition which is different from the final product, and a complex sequence of phase transformations takes place before or during sintering. Most of these transformations are controlled by the formation of (or pre-existence of artificially introduced) seeds. Without seeding, the transformation starts at a few isolated locations, extends from these nuclei in three dimensions, and may produce a textured microstructure with a preferential crystallographic orientation in rather large subregions and with a porosity that is hard to remove even at high sintering temperatures (promoting grain growth and destroying the benefit of using the nanoscale raw material). But even with seeding the fundamental problem is that a seed is required from a phase formed in a later transformation step, and the seed particles should be as finegrained as possible to be homogenized (at a low concentration) with small distances between them. However, seed particles are always much larger than the particles of the precursor or in the sol (if they were available more fine-grained, people would use them as a raw powder). As a consequence, the coarser character of the seed determines the spatial homogeneity of the transformation and of the resulting microstructure. With a typical size of seed particles between 0.05 and 0.3 ym and seed concentrations between 1 and S%, the homogeneity cannot be better than in the micrometer range. Therefore, even advanced sol/gel derived microstructures contain defects like that in Fig. 2. On the one hand, submicrometer materials with a narrow grain size distribution are easily formed and can be rather perfect over distances of several dozens of micrometers. On the other hand, it is almost impossible to avoid small (0.5-5pm) clusters of very small pores, the spacing of which seems to correlate with the spacing of seeds in the sol/gel process.

4.2 Technological Essentials for Producing Hard and Strong Tool Ceramics

653

Figure 2. Microstructure with typical cluster of micropores in a submicrometer sintered alumina ceramic produced from a boehmite sol with corundum (a-Al,O,) seeds <0.2 pm.

Typical flaws like the small pores in Fig. 2 are too small to be seen with the optics of most hardness testers. Hence, an evasive positioning of indents is impossible, and the testing gives a hardness which will be considerably lower than the inherent hardness of the perfect submicrometer microstructure. In fact, this reduced hardness represents a ‘true’ technical property of the material (because the flaws will affect other hardness-related properties as well, e.g. the tribological behavior), and the real drawback comes from the limitation of assessable properties and not from the influence of measuring procedures.

4.2.2 Recent Trends in Ceramic Technologies Related to Tool Ceramics 4.2.2.1 Technological Background The drawbacks of known industrial technologies have initiated a number of new developments in both powder synthesis and powder processing. Within the first 5-10 years, it is often difficult to distinguish which of the problems of a new technical approach are simply associated with the present, not optimized state of the art, and which are really characteristic of the new technology. Therefore, in the following, the discussion is focused on examples where the original reference work shows that the investigations avoided such technological problems in the fabrication of sintered parts, as far as it is generally possible within the framework of a given (optimized) approach. Of course. disadvantages which are inherent in a technology in the present state of the art can never be excluded. For this reason dzflerent

654

4 Ceramics Based on Alumina: Increasing the Hardness f o r Tool Applications

flaw populations are characteristic of the various technical approaches and cause processing-related property-limits which often cannot be overcome by materials development [31. As to new raw materials, the most significant step has been the development of chemically derived substitutes for materials of fundamental industrial importance like alumina or zirconia (most grades of which were originally manufactured from mineral sources). Since the late eighties, new corundum powders have been commercially available with fine particle sizes (rz 0.2 pm), which offers chances for developing extremely fine-grained microstructures with a high density. The small size and the increased curvature of the particle surface are expected to increase the driving forces of solid state diffusion on sintering, to reduce the required temperature, and to reduce grain growth. However, these potential improvements are often offset by the increased tendency of the fine-grained powders to agglomerate on sintering a body which contains agglomerates, densification proceeds first inside these agglomerates (where the particles are in closer contact), and, as a consequence, with the shrinkage of the agglomerates their boundaries will move away from each other [4]. Then, to eliminate the sinteringrelated new (and rather coarse!) interagglomerate porosity, much higher temperatures are required than expected for the fine-grained powder used. And, of course, the impact of the high temperatures on these fine particles is an undesired, strong grain growth (first starting inside the agglomerates). Therefore, for many years the published progress achieved with using these new raw materials was much less than hoped. Success came from increasing understanding of the relationship between sintering temperature, obtained density, and grain size as a function of spatial homogeneity and of interfaces, similarly to the microstructural dependence of the strength and the hardness of ceramics [3,4]. The general idea concerning the mechanism of sintering has been outlined above. The expertise starts from the simultaneous or sequential application of several dispersal approaches to bring the extremely fine-grained powders into a state that gives the right basis for the adjusted use of different shaping procedures. Successful experiments were reported with direct coagulation casting (DCC) [5] and centrifugal compaction [6] as well as with cold isostatic pressing of freeze dried granules, pressure filtration, and gelcasting [3]. Cold isostatic pressing easily gives dense sintered products with an average crystal size of 0.5 pm. Gelcasting means shaping by casting a (preferentially aqueous) slurry which contains a small percentage of a polymerizable additive. After polymerization, the soft but geometrically stable body is transferred to thermal processing. A bending strength of 700-900 MPa was reported for pressureless sintered single phase alumina bodies (which is about twice the conventional value) [3]. It is interesting to note that it is really the improved homogeneity of compaction of the powder particles in the ‘green’ bodies that is the key to smaller grain sizes with an increased hardness, wear resistance, and strength: in spite of a similar green density (similar starting porosity at the beginning of sintering) of all bodies produced by different shaping, the sequence dry (uniaxial) pressing + cold isostatic pressing + pressure filtration + gelcasting gives a dramatic decrease

4.2 Technological Essentials f b r Producing Hard and Strong Tool Ceramics

655

of the sintering temperature by about 250°C (always using the same raw powder and avoiding any doping additives). Therefore, the improved homogeneity of compacted particles in the given sequence is the one possible reason for the improved sintering behavior. For most machining operations where really hard tools are required (e.g. grinding or cutting of hardened steel), either more complex strength properties than a maximum bending strength of compact bodies are required, or the macroscopic strength is generally less important owing to the applied small feed rates (as in precise cutting). Therefore, advanced principles for obtaining a high macrostrength for sintered shaped bodies manufactured from these tool ceramics [ 3 ] will not be discussed here in detail, whereas special emphasis shall be given to prospects for improved hardness and wear resistance. 4.2.2.2 New Prospects for the Hardness

It seems obvious that this increased homogeneity is also the reason for the (quantitatively evaluated [ 3 , 7 ] ) reduced frequency and size of defects and increased strength. Equally clear is the influence of the reduced sintering temperature on the reduced crystallite size, the increased hardness, and wear resistance. Figure 3 shows the increase in hardness obtained by decreasing crystallite sizes (under the condition of a residual porosity 51%): single phase alumina ceramics can be produced by pressureless sintering with a hardness that exceeds the hardness of hot pressed TIC reinforced composites with a high carbide concentration of

v

656


Figure 4. Microstructure of a single phase submicrometer alumina grinding material manufactured by powder processing (average crystallite size 0.4 pm, hardness HVlO (ground) = 22.4 f0.7 GPa).

45 vol-%; Fig. 4 shows such a submicrometer sintered alumina microstructure. Further improvement is possible with small MgO additives (note that microhardness measurements give artificially higher values related simply to the smaller size of the indents; these microhardness data are, however, not representative for the technical behavior of tool ceramics). Dry Air (4 % r.h.)

Water

Humid Air (50 % r.h.) -

2*1o'6-

1'1 0-l6 -

8'1 0 -1~ 6*1Oei74*10-17-

2*10-j7

0.3

0.6 1.0

D @ml

3.0

0.3

0.6 1.0

D WnI

3.0

0.3

0.6 1.0

3.0

D @mI

Figure 5. Crystallite size and environmental effects on fretting wear of Al2O3-Al2o3couples.

4.2 Technological Essentials for Producing Hard and Strong Tool Ceramics

657

Figure 6. Increasing grain size promotes grain boundary fracture and grain pull-out (fretting wear in dry air; cf. Fig. 5 ) . Average grain sizes: (a) 3 pm; (b) 1.6pm; (c) 0.6pm (bottom) [I I].

658

4 Ceramics Based on Alumina: Increasing the Hardness,for Tool Applications

When comparing data from different sources, the condition of the surface investigated has to be noted: machining introduces high dislocation densities [8], and the hardness of ground ceramic surfaces is usually higher than measured on polished surfaces, an effect that is the more significant the more fine-grained the microstructure is (Fig. 3).

4.2.2.3 A Leap in Wear Resistance The grain size-hardness relationship in Fig. 3, published first in 1995 [9], has opened new prospects for improved tool materials. Until that time it had not been clear whether reducing the grain size of sintered alumina ceramics would cause a technically significant increase in hardness; in the USA it was accepted that the hardness increases down to crystallite sizes of 2 4 pm, whereas smaller grain sizes do not contribute to an improved hardness [lo]. But even people with a different opinion did not have a chance to test it because more fine grained microstructures could not be prepared until the early nineties. Figure 5 shows investigations of the fretting wear of alumina: depending on the environmental conditions, submicrometer ceramics may exhibit a wear rate which is an order of magnitude lower than measured for conventional A1203.It is important to understand that this improvement is not only a simple consequence of the increased hardness, but is also associated with reduced grain pull-out in the wear track of the submicrometer material (Fig. 6) [ll].

4.3 Tool Materials with Undefined Cutting Edge: Sintered Grinding Materials Grinding is an indispensable process in industrial machining. With the exception of special technologies like the grinding of wood or glass with garnet grits, it is generally assumed that high quality grits exhibit a high hardness. The consequences of a high hardness are rather complex, and we will show that a more than proportional increase of the grinding rate with a rising hardness of Al2O3 grits (Fig. 7 [12]) demonstrates only one among several positive aspects of a high hardness. Also, hard grinding materials are advantageously applied not only for grinding metals or very hard surfaces: hard S i c grits are likewise used for machining glass, some heavy metals, lacquers, chipboards, and plastics. For working steel, on the other hand, corundum grits (alumina, a-A1203)are superior to (the much harder!) Sic due to their extreme chemical stability. Alumina grits are manufactured by melting tons of raw powders in an electric arc (‘fused’ or ‘electro-’corundum). Depending on the impurity content and additives, brown, white and Zr-corundum are different types of fused grits where each of the quasi-macroscopic grains (available in the size range 0.01-0.5 mm) has an internal structure close to that of a small single crystal. As to the goal of a high hardness, the advantage of the absence of pores is to some degree offset by the disadvantage of a microplastic deformability that is

4.3 Tool Materials with Undefined Cutting Edge: Sintered Grinding Materials

10

15 20 25 Vickers hardness HV1 [load = 1 kg] (GPa)

659

30

Figure 7. Influence of the hardness of the grinding grit on the rate of removed material on grinding the cross-section of a steel tube (flexible disc, different alumina grits; the results have been normalized for a unified grit toughness of KIc = 3.6MPam”’ to demonstrate the hardness effect without the additional influence of Klc) [12].

lower than in polycrystals owing to the lack of grain boundaries. Zr-corundum, in fact, contains such micro-interfaces because with higher concentration of zirconia (Zr02) the grits are a eutectic of finest A1203 and ZrOz lamellae, but as the hardness of the zirconia subregions is much lower than that of corundum the macroscopic hardness of the eutectic is not improved but even somewhat reduced. On the other hand, the eutectic structure increases the fracture toughness to twice that of grits which are free of zirconia (here, toughness is a parameter which may be understood as a strength related to a unified defect size). On machining cold rolled steel sheets, the increase of the grinding rate by about 20-30%, in spite of the reduced hardness, has to be attributed to this increase of the toughness in Zrcorundum grits. For many applications fused grits are unsatisfactory, and efforts started at the beginning of the eighties to develop much harder (single phase) corundum grades by producing grits with a polycrystalline internal structure. This idea would affect not only the hardness, but also the fracture toughness, the microfracture behavior and the formation of chips. A difficult problem to be solved was that most of the conventionally known approaches for the manufacture of compact polycrystalline sintered ceramics cannot be used for producing irregularly shaped small grits (e.g. uniaxial pressing or cold isostatic pressing of powders). Hence entirely new

660


procedures had to be found to obtain a ‘green’ density of the intermediate (not yet sintered) product that is high enough to enable sintering until a high density with less than 1-2% of residual porosity is reached. Presently approaches that associate minimum crystal sizes with a maximum density of the sintered grits are the primary target of research. Efforts to increase the hardness by using composite materials with harder additives are much less important for grits than in the field of compact ceramics. The reason is the chemical nature of available hard materials (carbides like Sic, Tic, but also nitrides or, in a few experiments, diamond). The inability of Sic to grind ferrous materials is well known, and a similar observation had to be made for carbide reinforced composite alumina grits, in spite of their improved hardness. Further, a carbide additive in an oxide matrix does not always result in an increased hardness because the low diffusion of the carbides (dominated by covalent bonds) often gives increased porosity or (after sintering at a higher temperature) a coarser microstructure. On the other hand, highly pure (and therefore chemically very stable) sintered corundum grades have been developed recently, which after pressureless sintering exhibit a hardness similar to or even higher than the hardest hot-pressed composites with high carbide concentrations (cp. Fig. 5). All these features limit the motivation for new composites and increase the trend towards advanced, thermodynamically stable sintered grits on a pure oxide basis.

4.3.1 Technical Demands for Grinding Materials The following discussion is focused on grinding materials for machining steel and cast iron, because these are most important industrially. Two different requirements are to increase working efficiency and to minimize undesirable interactions of the tool with the ground surface. A frequently used criterion in patent applications is an increased rate of material removal. Another parameter is the quality of the ground surface: the tool should run for a long time with insignificant changes in the roughness of the ground workpiece. With new grinding materials that increase the grinding rate by 30-100% (see paragraph 4.3.2 and 4.3.3) or which give a roughness inside of the tolerance limit for a doubled working path, the economic benefit is even larger than given by the percentage of the reduced time per process (or per workpiece). The increasing output of the manufacturing lines increases the importance of down-times, affected also by the frequency of tool exchanges. The extended life of grinding tools gives an additional contribution to the economic efficiency of a line if, besides the reduced time in the individual grinding step per workpiece, an increasing number of workpieces can be manufactured with one tool. In many grinding operations the development of heat is very important: undesirable changes in the microstructure and the properties of the ground material have to be avoided (e.g. in hardened steel), the introduction of residual surface stresses on machining is tolerable within given limits only. Since grinding tools can give high material removal rates at fairly low contact pressure (minimizing the process


661

temperature), the final grinding of functional surfaces in ball bearings can hardly be substituted by other approaches today. The interest in grinding materials that exhibit a high grinding power at low process temperatures has been further promoted by the need to improve the environmental compatibility of manufacturing, a trend that is manifested by efforts to decrease or to avoid the consumption of coolants (reduced coolant concepts, dry grinding). There are manifold ways to develop such grinding materials. One approach is microstructural engineering: the shape of the crystals in the microstructure governs the fracture behavior of the grinding grain, the shape of the cutting edges and tips, and, finally, the chip formation. One of numerous ideas was the composition of grinding tools with hollow sphere-shaped bodies that, during grinding form always new (sharp) grain edges.

4.3.2 Advanced Commercial Products: Sol/gel-derived Corundum Obviously, to develop grinding grains between 10 pm and 500 pm in size and with a polycrystalline internal structure is possible only if the grit is manufactured as a polycrystal with crystallite sizes much smaller than the dimension of the grinding grain, preferentially < 1 ym. However, the available corundum raw powders have particle sizes of about 0.5 pm, and until now it is impossible to create microstructures with crystallite sizes of less than about 2 pm from such powders by pressureless sintering. Success came with the application of sol/gel procedures that use chemical precursors of the final corundum phase as extremely fine-grained (usually nanoscale) raw materials. Most sol/gel approaches are expensive, but the grinding industry has benefited from the unique situation with alumina where sols can be prepared from boehmite (an aluminum monohydrate A100H) readily available from other chemical processes (the price is about $5 per kg or less). Unfortunately, the sequence of chemical and crystallographic transformations from the precursor to the final phase generates substantial risks in forming inhomogeneous, porous microstructures, running counter to the goal of high hardness. Patent applications suggest different doping [ 131 and seed [14,15] additives intended to control these transformations. As a result it is, in fact, possible to improve the strength, and few examples offer significant advantages in the hardness compared with white fused corrundum. Nevertheless, on grinding rolled steel sheets such grits exhibit an increased power expressed, for example, by a rate of removed material which is 140-150°/~ compared with brown corundum or 115-125% compared with the fused A1203-Zr02 eutectic. The comparison with fused grades in Fig. 8 includes two intermediate qualities of sol/gel derived (sintered) alumina grits together with an advanced material of the sol/gel group the grinding rate of which is increased by about 150% related to fused A1203-Zr02 [16]. In Fig. 8, the improved technical performance of the polycrystalline grinding materials in spite of their almost unchanged hardness seems to disagree with the

662 250


1r

-

HVI

K, (MPadrn)

Wa)

2.83.9

15-20

4.2

16.3

1 .O-2.5

17.5

sintered AI,O, (soVgel)

200-

-

0 *.

* *.

I

0.0

.O fused eutectic AI,OJZrO,

I

0.5 1.0

I

I

1.5

2.0

I

I

I

2.5 3.0 3.5 time (min)

I

I

I

I

4.0

4.5

5.0

5.5

6.0

Figure 8. Comparison of the material removal rate of sol/gel derived sintered and of fused grits on the basis of AI2O3(flexible fiber disc, 6000min-’, contact pressure 30N; workpiece: steel tube [ST W 22 DIN 15431 with diameter 195 mm, wall thickness 1 mm).

concept of developing more powerful grits by increasing the hardness (due to a fine-grained microstructure). Obviously, there are other strong influences, and first of all strength properties, modifications of the microfracture behavior are expected to govern the results in Fig. 8. It has been recognized for a long time that the strength of ceramics depends on microstructural inhomogeneities (‘defects’, ‘flaws’) which locally increase the externally applied mechanical load, causing sudden, ‘brittle’ failure. What is important for grinding grains is, however, not a generally high but an adjustable (depending on the specific grinding process) strength: on the one hand, some global minimum strength under given loading conditions is imperative to run the grit for some time but, on the other hand, originally sharp edges and tips of a very strong grain would be worn and would deteriorate the grinding efficiency within less than a minute: an effect that is readily observed with very strong (e.g. carbide reinforced) ceramic grains. Therefore, the strength must be low enough to enable microfracture processes on a local microscale for re-generating new sharp edges during the grinding process. Since it is impossible to evaluate the strength of a geometrically irregular grinding grain by standard approaches, the fracture toughness is measured by indentation procedures on mounted samples. The method derives the toughness from the length of indentation cracks formed at a given load. The analysis of a larger number of grains gives the statistical distribution of the loading capacity in different grades of grinding materials, and these statistics can be correlated with the technical

663

4.3 Tool Materials with Unde$ned Cutting Edge: Sintered Grinding Materials

0

1

2

3

4

K (MPdrn)

5

6

7

0

1

2

3

4

5

6

7

K, (MPaJrn)

Figure 9. Microstructural dependence of the fracture toughness in grinding grains on the basis of alumina prepared by sol/gel approaches (left) and in fused grits (right).

grinding performance of the grits under different conditions [16]. Figure 9 shows different toughness distributions obtained in sol/gel by producing microstructures with different porosity, crystallite size, and shape (as exemplified by Fig. 10). Obviously, the polycrystalline nature of sintered materials is a strong tool for modifications of the mechanical properties which are impossible in fused corundum grades. From Fig. 9 it can be understood that the sol/gel derived grits owe their increased grinding efficiency (Fig. 8) first of all to their improved global stability (fracture toughness, strength). On the other hand, it is very instructive to understand the grinding performance of the A1203-Zr03 eutectic: with a very high toughness (Fig. 9) and with a hardness which is smaller than observed for zirconia-free fused grits but in no way inferior to the sol/gel grades (Fig. 8), it is impossible to attribute the only intermediate grinding efficiency of this material (Fig. 8) to deficits in its toughness or hardness. On the contrary, the A1203-Zr02 eutectic is a very clear example for the power-limiting effect of too high a strength of grinding grains, decreasing the frequency of microfracture processes which are needed to re-generate sharp cutting edges. Of course, the concept of an adjustable strength that enables an optimum frequency of microfracture events does not mean tolerance of hardness deteriorating defects. From Fig. 2 it can be shown that within the framework of sol/gel approaches the frequency and size of such defects can be minimized but hardly avoided. In spite of the high number of patent applications devoted to improved sol/gel-derived grinding materials, it has been impossible to overcome the fundamental limit associated with this technology: the discrepancy between the unique chances to use nanoscale raw materials and the spatial heterogeneity introduced by the indispensable but always coarser seeds. With the obviously negative impact of flaws like those in Fig. 2 on the hardness and wear resistance even if the porosity is no more than 1-2%, it has to be expected that sol/gel derived products will not be able to exceed some limiting level of properties.

664


(

Figure 10. Microstructure of sol/gel derived sintered alumina grits: (a) platelet shaped microstructure (‘Cubitron’, manufactured by MMM, USA), (b) bimodal grain size distribution after seeding with transitional alumina nuclei, (c) anisotropic grain growth (Ce doping) [ 161.


665

4.3.3 Sintered Alumina Grits Produced by Powder Processing Approaches Sol/gel-derived grinding materials probably have been the first example where aprecursor approach was successfully used in the industry for a manufacturing volume of several tons per year. The scientifically fundamental importance of this comes from the circumstance that it was also the first example of highly dense and submicrometer ceramics with advanced mechanical properties. This progress was first achieved in the industry and only later on investigated in scientific institutes in greater detail. And again, as to the mentioned limitations of the sol/gel approach, it was the requirements of the industry which promoted a new idea: to complete the family of fused and sol/gel derived grinding ceramics by a third group which due to a generally different manufacturing process should exhibit not a stepwise improvement but another class of quality, similarly as the sol/gel grits have shown a generally improved performance compared with the fused materials. Success came with an apparent return to well-known powder processing technologies, but on a rather different level, as described in Section 4.2. Considering that such basic influences like grain size and porosity effects on hardness are often a subject of controversy and are not completely understood, it is not a surprise that only recently has hardness been discussed as a function of defects [17]. Figure 2 shows small defects (2 pm or less) which will significantly deteriorate the hardness parameter which is effective for the grinding process. On the other hand, larger flaws like cracks or pores of about 5-20pm do not necessarily decrease the grinding efficiency: on grinding, such flaws will become nuclei of microfracture processes (which are useful to re-generate sharp cutting edges), and these fracture processes will eliminate thejaws from the microstructure of the mesoscopic cutting edge (the hardness of which is determined by remaining smaller defects). Hence, the 'effective' (local) hardness may be higher than a value measured with the influence of coarser pores. All microstructural understanding, however, can give rise to a new class of grinding materials only if the basic mechanical properties are adjusted to some application-related optimum. Therefore, for grinding as for cutting tools, investigation of technical performance under defined conditions is necessary. Figure 11 shows the results of test series performed by Hermes Abrasives, Hamburg (Germany). The most interesting result is the large progress in the grinding efficiency obtained by the powder derived material compared with the advanced sol/gel grits (all submicrometer microstructures); the difference is of similar magnitude to the advantage of the soligel grits compared with fused brown corundum. Compared with the A1203-Zr02 grinding grit, powder processing gives a single phase alumina grade which exhibits twice the material removal rate of the eutectic. It has to be emphasized that this improvement does not depend on the extremely small crystallite size of the microstructure (e.g. 50.5 pm, for which range Fig. 3 indicated a decreasing grain size influence in the hardness): the result in Fig. 11 relates to a powder-derived material with a grain size of about 1 pm [16] and was equally reproduced with crystallite sizes between 0.5 and 1 pm. An example of a powderderived tool material with a submicrometer microstructure was given in Fig. 4. With a fracture toughness of 3.34MPa'12m for all sintered corundum microstructures independent sol/gel or powder processing, the additional data included in Fig. 11

666


HV1lnhei

250-

sintered alumina (by powder processing)

Wa)

21-25

200 sintered alumina

1

0) v

-

z

.-

$150E -0

a,

>

E 100??

50-

01 0

I

I

I

I

I

1

2

3 time (min)

4

5

6

Figure 11. Comparison of the grinding performance of known (fused and sol/gel derived) grinding materials on the basis of A1203with a new single phase corundum grit produced by a powder processing approach (same test conditions as given for Fig. 8) [16].

demonstrate the importance of the ‘effective’ (local) hardness on technical performance and especially for the stepwise increase of the grinding power in the ranking sequence fused corundum + sol/gel derived sintered alumina -+ sintered alumina by powder processing. This effective hardness (which is inherent to the microstructure) is diminished by small defects like that in Fig. 2, and with their elimination by substituting sol/gel by powder approaches the hardness increases (without decreasing the crystallite size!) and increases the grinding efficiency. With a sufficiently large number of tests, results like that given in Fig. 7 can be obtained which demonstrate a general trend of an increasing grinding power with increasing hardness (under the condition of an always equal chemical interaction of all grits with the ground metal).

4.4 New Trend for Cutting Hard Workpieces: Submicrometer Cutting Ceramics for Tools with Defined Cutting Edge Characteristic for the late nineties was the increasing use of tough grades of silicon nitride, first of all when high feed rates are requested. With their rather low hardness, however, Si3N4 ceramics are outside the scope of the present discussion.

4.4 Neiz, Trend for Cutting Hurd Workpieces

667

Most of the machining of lzavdened steel and hard cast iron, on the other hand, is done by grinding, a process that today is necessary for the final finish of functional surfaces, like in ball bearings. Nevertheless, there are several limitations of grinding which motivate the search for other methods: compared with cutting operations (whenever applicable) grinding is a time-consuming process (up to a factor of three), the equipment is more expensive, and disposal or recycling of the grinding silt is difficult from both ecological and economic points of view. It has been shown that even without a complete substitution the working time can be cut to 2/3 by combining a first turning step with final grinding [2]. If turning operations shall be used for the final finish as well, important tool life criteria are the surface roughness of the machined workpiece and the stability of the geometrical position of the cutting tip (which governs the accuracy of the produced piece). This paragraph deals with the technical demands on cutting tools which might be able to give fresh impetus to such efforts without increasing the costs (as it is the case with tools manufactured of diamond or cubic boron nitride). Therefore, the discussion will be focused on ceramics on the basis of alumina as outlined already in Section 4.1.

4.4.1 Demands for Cutting Materials Used for Turning Hard Workpieces Table 2 illustrates the primary technical demands on cutting ceramics for machining hardened steel or hard cast iron. These demands are a consequence of the required wear resistance in dynamic contact with the abrasive carbide particles in the microstructure of the cut metals. As to more fundamental properties of the cutting plate, there is the requirement for the combination of maximum hardness and sufficiently high microscopic fracture toughness of the grain boundaries to minimize the frequency of grain pull-out at the cutting tip. From the understanding of the influences of the microstructure on hardness and wear properties, an average grain size of substantially less than 1 pm and a low frequency of microflaws are imperative to meet these demands. On the other Table 2. Demands on basic properties of cutting tool ceramics for machining hardened steel: x = important, .xx = very important On precise machining Technical demands Wear resistance (especially against the abrasive impact of carbide particles) Low thermal conductivity Low creep rate (at high temperature) High oxidation stability High chemical stability (preventing solid state reactions with constituents of the iron-based workpieces)

Hard cast iron

Hardened steel

S.X

SX

S

x

.x

xx

x

xx

668


hand, however, special measures may be required to balance the increasing hightemperature creep usually associated with small grain sizes. Reinforcement with whiskers (Sic whiskers are most common [18]) is one way to reduce the creep rate [19], an approach that also increases the thermal shock resistance [20]. Another way to reduce the creep rate is grain boundary engineering. Besides the prevention of impurity associated amorphous grain boundary films it was observed that a small concentration of yttria (0.1%) can reduce the creep rate of sintered alumina [21]. As it is also known from other fields of working metals, the different forms of the chips from machining cast iron and steel cause a much more intense interaction with the cutting plate for the latter material. Therefore, the demands concerning the chemical and oxidation stability are more stringent on working hardened steel than on machining hard cast iron (Table 2). If the wear resistance is increased by decreasing the grain size of the ceramic microstructures, an additional advantage comes from a feature illustrated by Fig. 6. The reduced wear rate is associated with a decreasing amount of grain pull-out at smaller grain sizes, and the same effect can improve the quality of the ground ceramic surfaces. With a constant grinding procedure in finishing alumina ceramics (diamond wheel, grit size 4&50pm), the roughness of a ground 0.6pm alumina ceramic is about one half the value measured on a ground 3.6 pm microstructure [7]. Obviously, the improved quality of the ceramic cutting edge increases the chances for tool life at a small roughness of the machined hardened steel. To prevent spontaneous microstructural changes in the workpiece and to obtain a high efficiency of the machining process, all approaches try to realize conditions with low cutting forces and a low heat input into the finished surface. Whereas this is generally true for both grinding and cutting operations, it is also accepted that the use of cutting ceramics is promoted if local temperatures are high enough for some softening of the (usually brittle) surfaces of hardened steel or hard cast iron [2]. To this end, ceramics with a low thermal conductivity are preferred. Table 3 reveals little differences with the exception of the most expensive tool materials (cBN, diamond) which are on the less favorable side (in fact, the high thermal conductivity of the latter gives an improved thermal shock behavior which, however, is considered as less important when small feed rates are applied on precise machining hardened steel). Much more significant differences exist for the hardness of the different ceramics (Table 1) and for their mechanical resistance at high temperatures. As to its oxidation stability and its chemically almost inert behavior in contact with iron or steel, pure sintered corundum is second to no other ceramic. Table 3. Thermal conductivity of ceramic cutting tool materials. Cutting tool ceramics

Thermal conductivity [W m-' K-'1

A1203+ x 15% Z r 0 2

= 28

A1203

A1203+ x 30% Ti(C, N) Si3N4 Cubic boron nitride (cBN) Diamond

x 30 x 35 30-35 40-120 500-600

669

4.4 New Trend for Cutting Hard Workpieces

4.4.2 Carbide Reinforced Composite Ceramics Based on A1203 Previous investigations of precise cutting operations with hardened steel or hard cast iron were focused on black composite ceramics of A1203 with about 30% Ti(C,N). No positive results are known for grades based on Si3N4 or A1203/Zr02; probably, hardness and chemical stability of silicon nitride and of zirconia do not meet the demands of this machining procedure. For example, a commercial AI2O3/Ti(C,N) tool was used for the final (precise) machining of the inner diameter of a ball bearing ring [2]. The average crystallite size of such cutting ceramics is about 1-1.5 pm. With the large body of evidence for the improvement of hardness and wear resistance by decreasing grain sizes, it is not surprising that most manufacturers have started to develop these tools with a submicrometer microstructure, but the covalent nature of the carbide bonds prevents pressureless sintering of the composites at temperatures of 1600°C or less. With TIC concentrations of 25-35%, the most fine grained of these composites exhibit average sizes of A1203and TIC subregions of about 0.8-1 pm associated with a HVlO hardness up to 23 GPa (measured at a testing load of 10 kgf). Compared with today's commercial grades with HVlO < 21.5GPa, this is an increase of almost 10%. Nevertheless, these tools showed only similar behavior on fine turnin gray cast iron (GG25, lOOOmmin-'), hard cast iron (60-65 Shore, 150mmin- ) and on turning a hardened steel 100CrMo6 for ball bearings (180mmin-', Fig. 12). The latest patent applications try to obtain more fine grained microstructures outside the known compositions A1203/TiC or A1203/Ti(C,N). An alumina composite with 25% (by weight) of a mixed carbide/carbonitride consisting of about 90% Ti(C,N) and 10% WC was hot pressed at 1600°C yielding a microstructure with

F

hard cast iron (60-65Shore)

v = 150 d m i n f = 0.25mm/rev

e

-

hardened steel 100CrMo6 v = 180 m/min f = 0.1 mm/rev

L

Corn

E

S

Y

c

0

E 10-

Corn

S

.c

5

E

z

5-

0

Figure 12. Turning hard workpieces with different ceramic composite tools. Corn = advanced commercial AI20,/Ti(C,N) grade, concentration of Ti(C,N) about 25-30 vol-%. S = laboratory grade: average sizes of A1203 and TIC subregions 0.8-1 .O pm, TiC concentration 35 vol-%.

670

4 Ceramics Based on Alumina: Increasing the Hardness Jor Tool Applications

an average grain size of 0.5 pm [22]. On turning hardened steel (X63CrMoV5) with a discontinuous cutting regime, the comparison with an unspecified commercial tool showed lifetime (or cutting path) that was three times greater. Another recent development uses oxycarbide and oxycarbonitride compositions (Ti(C,O), Ti(C,O,N)) with different stoichiometry instead of TIC and obtains crystallite sizes of about 0.6 pm (after sintering at 1700°C and hot-isostatic post densification at 1435°C); the hardness is HVlO = 22.5-23 GPa [23]. Figure 13 shows the microstructure of these materials which is much more fine grained than an advanced (also claimed submicrometer) commercial material with the same carbide concentration (on the market since 1997). As another example, Fig. 13 includes a new material with an increased concentration of 35 vol-% of the covalent phase which can be manufactured with the same small grain size in spite of the high carbide loading. The cutting performance of these new materials was tested in a direct comparison with similarly hard but single phase (absolutely oxidation resistant) pure alumina ceramics. The results are discussed in Section 4.4.4.

4.4.3 Single Phase Sintered Corundum From Table 2, an advanced cutting ceramic for the precise machining of hard workpieces should associate a maximum hardness with finest grain sizes (to minimize grain pull-out at the cutting edge) and high chemical and oxidation resistance. With today’s knowledge, it is not clear if carbide or carbide reinforced tools will ever be able to meet the thermodynamic and chemical side of these demands: in fact, the (negative) free enthalpy of T i c and TiN is only half of the value of Si3N4 [24], and even the silicon nitride is unable to cut mild steel (not to mention hardened steel). On the other hand, from the data in Fig. 3 there exist pure sintered corundum ceramics with a hardness that even after pressureless sintering is in no way inferior to hot-pressed carbide reinforced composites, with a strength of 800900 MPa [3] (after HIP up to 1300 MPa [25])that equals or even exceeds the strength of the composites, and with the additional great advantage of highest chemical and oxidation resistance. Cutting tests performed with these recently developed alumina grades are rare, some first examples are compared in Section 4.4.4 with the performance of commercial and of submicrometer composite tools.

4.4.4 Comparative Cutting Studies with Submicrometer Ceramics: A1203 and Composites Reinforced with Ti(C,N) and Ti(C,O) 4.4.4.1 Mild Conditions: Machining Cast Iron and Steel First investigations of the cutting performance of submicrometer tools have been performed in Asia [26-281. Unfortunately, all these tests describe only the machining of unhardened steel and cast iron. On milling cast iron, the submicrometer sintered AI2O3 showed almost no advantage and was inferior to 70A1203/30TiC and

4.4 N ~ MTrend,for J Cutting Hard Workpieces

671

Figure 13. Latest tool composites based on A1203:(a) commercial ‘submicronstructured’ Al2O3/ Ti(C,N) composite and new laboratory grades [23] with (b) Ti(C,O;N), (c) and with an increased concentration of about 35 vol-% Ti(C,O).

672


"""1 f = 0.2 mm/rev

a = 1.5 mm

0.0 0

1

2

3

4

5

Cutting time (min) Figure 14. Wear flank width on turning steel (Japanese code S45C). G = sintered corundum with grain size 4 pm. Adapted from Takahashi et al. [26].

60A1203/40Zr02composites, a result which is hardly surprising: in this application, the differences in the tool hardness are less important, and the improvement in the toughness of submicrometer A1203microstructures is so small ( < 4 MPa m1j2[3,29]) that it is technically unimportant. On the contrary, on turning cast iron or steel (Fig. 14) the progress obtained by the submicrometer alumina was significant. It should be emphasized that an advantage of the pure alumina tool compared with carbide reinforced composites and with zirconia toughened alumina was observed just under these conditions where during continuous cutting an increased processing temperature at the cutting edge applies. 4.4.4.2 Machining Hard Cast Iron and Hardened Steel The behavior of the tools described in the last section supports suggestions of a preference of submicrometer alumina ceramics for the precise machining of hardened steel and hard cast iron. First investigations in this field started in 1998 at the Dresden Institute of Ceramic Technologies and Sintered Materials. All turning tests were performed with a feed rate, f = 0.1 mm min-' and a depth of cut, a = 0.2 mm. The size of the cutting inserts was SNGN120412 (i.e. 12.7 x 12.7 x 4.76mm3 with a 1.2mm radius) with a 20" chamfer (width 0.2mm). The plates were positioned with a rake angle y = 6", an inclination angle h = -4", and a large entering angle x = 45" chosen to maximize the selectivity of the tests (as to the tool materials)

4.4 New Trendjor Cutting Hard Workpieces

673

Table 4. Properties of new tool ceramics compared with advanced commercial grades Composition

Composites SH1 (CeramTec) commercial ‘submicron’ AT60A AT60

Average grain size Hardness HVlO (composites: size of (Vickers, 10 kg testing single phase load, ground subregions) surfaces) (w) (GPa)

Al2O3+ N 33 vol-YO Ti(C,N)

1.5

21.16 f 0.23

A1,03+ N 33 vol-YOTi(C,N) Al2O3+ N 33 vol-% Ti(C,O) A1203+ N 37 vol-YOTi(C,O)

<1 0.7

0.67

21.22 f0.54 22.18 f0.77 22.45 f 0.85

0.56

22.08 f 0.90

Pure alumina AC4 1

A1203

by high loads at the cutting tip. A new commercial grade of an A1203/Ti(C,N) tool ceramic was claimed ‘submicronstructured’ by the producer and introduced into the market 1997 especially for machining hard materials; it was compared here with an advanced standard grade SH 1 (CeramTec AG, Plochingen, Germany) and used as a reference to evaluate the cutting power of new laboratory grades (Table 4). Both the submicrometer and the SHl tools are composites based on A1203with about 30% of Ti(C,N). On the turning hardened steel at high cutting velocities > 250 m min-’ there is a significant input of heat not only into the formed chips but also into the cut surface of the shaft. Depending on the state of wear of the tool, cutting forces and process temperatures may increase to an extent that causes softening of the hardened steel, and the hardness was observed to drop from HRC = 57-60 to values of less than 50. Therefore, the decrease of the hardness has to be recorded as an additional parameter in such investigations, and an intermediate machining operation with a low velocity ‘u = 180 m min- was required before every new run to restore the original hardness.

’

4.4.4.2.1 Intermediate Demands: Machining Hard Cast Iron Figures 15-17 show the results for continuous cutting chilled globular cast iron G-X300CrMo 153 (Rockwell hardness HRC = 43-45). The flank wear of the tools is given in Fig. 15. The high hardness of the oxycarbide reinforced submicrometer composite reduces its wear by an amount similar to the progress of the commercial submicrometer tool compared with the standard material SHI. Much more surprising is the qualitative leap in the lower wear of the pure alumina insert which cannot be explained by its hardness: the hardness is only just intermediate between the (lower) hardness of the commercial composites and the (higher) value of AT60A. Also, the grain size of the sintered corundum material is not very different from the microstructural data given in Table 4 for the A1203/Ti(C,0)composite AT60A. Therefore, Fig. 15 suggests that reducing the grain size of TIC, Ti(C,N), or Ti(C,O)-reinforced composites may indeed increase the hardness and reduce the wear in turning hard cast iron, but the effect is limited, probably because of chemical

674


0.3.

0.25. h

E

E !

0.2-

m

> 5 50.15Y

c

m -

c

5 0.1.

3 0.05.

i

p if

0.0.

0

I 10

I

20

I

I

40 Cutting time (min)

30

I

I

50

60

70

Figure 15. Flank wear width of submicrometer ceramic tools on turning hard globular cast iron (Rockwell hardness HRC = 43459, see Table 4 for tool characterization.

interactions between the iron and the carbide phases of the tools. To eliminate this interaction by using a pure alumina tool can greatly improve the wear resistance as far as the process temperature at the cutting tip is not too high to initiate additional wear due the low high-temperature creep resistance of the submicrometer alumina microstructure. The cutting edge displacement (CED) characterizes the obtained precision in turning hard materials. Figure 16 (CED) shows that complex wear processes may give a different ranking of the tool materials compared with Fig. 15 (flank wear). Again, all of the new laboratory submicrometer grades exhibit much less wear than the advanced commercial tools. However, with the criterion of the cutting edge displacement it is the ceramic with the highest hardness which shows the best performance (composite AT60A, Table 4). Another important criterion for the quality of tools in precision turning of hard materials is the obtained roughness of the cut metal surface. The graphical presentation in Fig. 17 gives the average roughness R,. A fixed ratio between the statistically averaged maximum depth of roughness RZD and R, was observed in all the test (Fig. 17). The arbitrary scatter of data was rather large, and no significant difference was observed between the surfaces cut by the SHl or by the commercial submicrometer tools. However, all other inserts show the same ranking as observed for the flank wear in Fig. 15. Improved results are provided by the oxycarbide reinforced composites (AT60), but a more significant progress results from the use of pure submicrometer alumina inserts.

4.4 New Trend f o r Cutting Hurd Workpieces

615

70 cast iron HRC = 42 - 44 v = 250 mlmin = mm f = 0.1 mmlrev a = 0.2 mm K = 45"

E c

c

E

$400) Q

.-

U

g30-

73

g 200,

0 =

.

i

0

AT6OA

(Al,OJri(C,O))

10I)-..

I

..

....4

I

I

I

I

I

0 Cutting time (min)

Figure 17. Surface quality of cut surface. Average grain sizes of the tool microstructures (in composites: sizes of single phase subregions) are D(SH1) = 1.5pm, D(AT60) = 0.67pm, D(AC41) = 0.56 p m (see Table 4).

4 Ceramics Based on Alumina: Increasing the Hardness for Tool Applications 60.0

hardened steel HRC = 57 - 60

40.0

criterion: flank wear width VB = 0.15 mm

'.

h

c

.E20.0 v

e

0 3 c

0

E

-

.-

10.0

8.0 6.0 4.0

, 150

200

250

300

350

v (m/min)

Figure 18. Life time of advanced ceramic inserts on machining hardened steel (see Table 4 for materials characterization). The arrows indicate frequently observed deviations.

Whereas turning of hard materials is usually performed with asymmetrically positioned inserts (entering angle x. = 10-30"), it was mentioned above that K = 45" was chosen in these investigations to maximize the selectivity of the tests (as to the tool materials) by high loads at the cutting tip. For geometrical reasons, a lower roughness is expected for both smaller (10-30") and larger (60-80") entering angles. Therefore, it should be noted that in Fig. 17 the increase of R, towards values between 2 and 4 pm is directly associated with this very special orientation of K = 45" and does not indicate a disadvantage of the tested tool materials. Indeed, in an additional test performed with SH1 and with the commercial submicrometer tool at K = 75" all R, values were < 1.5pm for the whole time of cut (i.e. up to 0.2mm flank wear at least). 4.4.4.2.2 Extreme Conditions: Continuous and Discontinuous Cutting of Hardened Steel Figures 18-19 give the results for continuous cutting of hardened steel 90MnCrV8. Its high hardness (HRC = 57-60) and the choice of tool materials with a low thermal conductivity cause a large input of heat into the formed metal chips. These appear rather flaming than red-glowing, an effect which is optically more spectacular on machining hardened steel with pur corundum inserts than using composite tools the thermal conductivity of which is about 15% higher (see Section 4.4.1, Table 3). On machining hardened steel with composite tools, the lower temperature gives more faceted chips. For the reference tools (SH1 and the commercial submicrometer ceramic), the high temperature is associated with a critical influence of the cutting velocity

677

4.4 New Trendjor Cutting Hard Workpieces (al

70

(bl

hardened steel HRC = 57 - 60

70 hardened steel HRC = 57 - 60

60I

E

150c

E

8 40$

n

g30-

P

5

0I

20-

0

10-

0

I

* I

.

... .-. I

I

I

I

0

5

10

15

20

25

30

Cutiing time (rnin)

Figure 19. Cutting edge displacement on turning hardened steel with advanced ceramic tools: (a) at v = 220mmin-' (above), (b) at 300mmin-' (below); see Table 4 for materials characterization.

between 200 and 250 m min-' . Flank wear is small at v = 220 m min-l , but already at 'u = 235 m min-l a greater crater-wear is observed. At v = 235 m min-I, more than 50% of the tested commercial inserts exhibited a sudden, strong wear increase during the first 10-20min associated with local or global fracture at the cutting edge, and it becomes difficult to record a continuously increasing flank wear up to VB > 0.2mm. Whereas the cutting edge displacement was immeasurably small during the first 5 min, it was overtaken by fracture soon later and could not be determined optically (even at v = 220 m min-' the cutting edge displacement of commercial tools was often shadowed by local features of microfracture). Hence, it was impossible to measure the cutting edge displacement of the reference tools at v = 235 m min-]. Therefore, tests started with v = 220mmin-', but some of the new laboratory grades exhibited extremely few wear at this velocity, and additional experiments had to be performed at higher rates. It should be noted, however, that already at v = 235 m min-' the reference curves for SH1 and the commercial submicrometer material characterize but a minority of inserts standing this condition (about 50% failed at shorter times by cutting edge fracture). Therefore, these data (Fig. 18) overestimate the real behavior of the commercial grades at this velocity. Whereas even at the lower velocity of u = 220mmin-' almost 50% of both commercial grades failed by fracture during the first 5-1 5 min of cutting hardened steel, no one such event happened with the laboratory grades (where crater-wear was also observed but to a lesser extent compared with the commercial ceramics). This behavior correlates with another surprising observation made on preparing the hardened shaft for the tests. This preparation is difficult with shafts of hard cast iron or hardened steel because it associates the high hardness of the counterpart with discontinuous cutting conditions at a changing frequency and power of impacts. Cubic boron nitride inserts (dreborid$/, Lach company, Hanau/Germany) failed on machining the hard (globular) cast iron, and hard metal inserts (WC/ 6% Co, Vickers hardness M 16 GPa, Kc M 9 MPam'/2) were not able to cut the

678


rough outer shell of the hardened steel shaft at velocities between 50 and 100mmin-'. The same observation had to be made with the commercial ceramic composites (SHI and the submicrometer grade fractured after t < 2min). On the contrary, both the pure alumina insert AC41 and the composites AT60A and AT62 were successfully used with w = 120 m min-', feed rate f = 0.3 mm rev-', and a depth of cut a = 1.5mm. These new, submicrometer cutting ceramics did not only not fracture macroscopically during one hour of severe cutting (with one tip) but retained microscopically nearly perfect cutting edges providing a high surface quality of the hardened shaft in the first cut ( R , < 1 pm for t 5 15min [32]). Surprisingly, under these severe conditions the general performance of the pure alumina ceramics was even more prospective than the behavior of the new submicrometer composites [32]. For SH 1, the manufacturer's data sheet gives a strength of 600 MPa (no available data for the submicrometer grade). The bending strength of the Ti(C,O) reinforced composite AT60A is about 800MPa [23],and a lower strength of about 650700MPa was observed for the pure submicrometer alumina ceramic [3]. The fracture toughness of all of the laboratory grades is 3.3-3.8 MPa m1/2, whereas surprisingly high values of 5.5 MPa m'I2 (SHI) and 6.6 MPa rn'" (submicron commercial composite) are given by manufacturer's information for these references. None of these data explain the much smaller fracture risk of the submicrometer laboratory grades on machining hardened steel. Obviously, complex wear-induced flaw-generation processes will have to be considered to understand the improved global stability of the new laboratory grade microstructures. No differences were observed in the roughness of the hardened steel surface when different ceramics were used. At 'u = 235 rnmin-' ( f = 0.1 mm rev-', a = 0.2mm), R, was 5 1 pm for the first 10-15min of cut for all of the tested ceramics, increasing to about 1-1.5 pm after 30 min (K = 45": see comment at the end of Section 4.4.4.2.1!). On the other hand, the tools showed very different flank wear and cutting edge displacements. Figure 18 gives the influence of high cutting velocities on the life of advanced cutting ceramics on machining hardened steel (flank wear criterion VB = 0.15mmin-'). As already mentioned, 'u = 200220 m min-' is the upper limit of reliable use for the latest generation of commercial A1203/Ti(C,N)composites. At this velocity, best results are obtained with the new submicrometer composites with Ti(C,O) which exhibit the highest hardness of all of the tested ceramics (see Table 4). At higher cutting velocities, however, the relationship compared with the pure submicrometer alumina changes, and pure A1203 again gives the greater advantage. A similar smaller effect of the cutting velocity on A1203 inserts compared with composites was noticed on cutting hard cast iron (Fig. 15, Fig. 17). Contrary to the new ceramics, most commercial submicrometer inserts fractured within the first 5-10min when used at w = 300mmin-', and the one continuous wear curve measured up to 15min and displayed in Fig. 19b was obtained at a softer part of the steel shaft (HRC = 52-57). At all cutting velocities the new laboratory grades are clearly superior to advanced commercial inserts. Contrary to a sometimes different behavior of flank wear and cutting edge displacement on turning hard cast iron, both parameters were well

4.4 New Trend,for Cutting Hard Workpieces

679

correlated on machining hardened steel. Comparing the new submicrometer A1203/ Ti(C,O) composite with submicrometer pure corundum, Fig. 19 shows a velocitydependent different ranking of the cutting edge displacement (CED) of these tools in perfect correlation with the data in Fig. 18 (flank wear); the CED curves for the Al,O,/Ti(C,O) composite and for pure A1203were observed close to each other at an intermediate velocity v = 235 m min-' (as suggested by the flank wear data in Fig. 18). Comparing this behavior on machining hardened steel with the turning of hard cast iron, again both the pure alumina tool AC41 and the laboratory grade composite AT60A exhibited less wear than the commercial cutting ceramics, but the changing ranking of AC41 and AT60A ceramics is worth some comments. It is obvious that the technical ranking of both tool ceramics will be affected by some important basic properties: For A1203/TiC composites, chemical interactions cause preferential (local) wear of the covalent carbide constituents of the microstructure even at room temperature and with almost inert conditions (dry air, fretting wear against alumina), with the consequence that at equal crystallite sizes the pure sintered corundum is more wear resistant than the composite [30,31]. The same ranking is observed for the flank wear of ceramic tools on turning hard cast iron (Fig. 15). (ii) On machining hardened steel with v = 220 m min-l, flaming chips indicate much higher process temperatures than on turning hard cast iron at u = 250 m min-' . Whereas pure submicrometer alumina exhibits intense creep at temperatures > 1 100"C, it can be assumed that the continuous network of covalent crystals reduces the creep rates in composites with more than 25 vol% of TIC, Ti(C,N) or Ti(C,O). Indeed, comparing at u 5 250mmin-' the machining of cast iron and steel, the increasing temperature increases the wear of alumina more than the wear of the best composite AT60A and changes their ranking: at a similar submicrometer grain size, AT60A exhibits the lowest flank wear and the smallest cutting edge displacement. (iii) Of course, this advantage of the composite will be lost when at yet higher temperatures oxidation of the Ti(C,O,N) phases or more intense reactions with the steel start. With the results in Fig. 18 and Fig. 19, the commercial composites with Ti(C,N) are more susceptible to such processes than the new laboratory grades with Ti(C,O). In fact, no information about the specific nature of such reactions can be obtained from the present results, but it is obvious that on machining hardened steel an increase of the cutting speed (and of the process temperatures) causes a more intense deterioration of the wear resistance of even the best composite AT60A than observed for the pure alumina tool AC41 (Fig. 18, Fig. 19): already at v = 235 m min-' both ceramics exhibit the same wear, and with u = 300 m min-l the sintered corundum again shows the best performance. (i)

At a velocity of v = 300mmin-', the increasing with time wear deteriorates the cutting edge to an extent that obviously causes an increasing input of heat into

680

4 Ceramics Based on Alumina: Increasing the Hardness j o r Tool Applications

the surface of the machined steel. For example, on machining the hardened steel with the submicrometer composite ceramic AT6OA the hardness of the cut surface decreased from originally HRC = 59-60 to values of 57-59 (25min) and 55-56 (30 min) [32]. This influence of high-velocity turning on the hardness of workpieces is important for different aspects in the development of new machining technologies. On the one hand, the softening of the hardened steel promotes unexpectedly high materials removal rates. In this way, turning of hardened steel with cutting ceramics that are much less expensive than cubic boron nitride (cBN) seems possible now. If, on the other hand, for the final step of precision machining a constant high hardness of the cut surface is required, critical limits exist for the cutting velocity (about 2 5 0 m m K ' in the investigations described above) or for the tolerable tool wear (here at ZI = 300 m min-' : VB 5 0.15 mm). These technical demands are, of course, affected by the choice of the ceramic tool depending on its thermal and wear properties.

4.5 Summary Recent developments show somewhat different trends for sintered ceramic grinding materials and for cutting tools. (i) In the field of grinding grits, extended innovative work is directed towards a substantial further increase of the grinding efficiency (high materials removal rates) and increased tool life within narrow tolerances for the surface quality of the ground workpiece. (ii) For cutting tools, innovation is focused on two aspects: the development of very tough and highly reliable grades for high feed rates and cutting depths (e.g. for globular and vermicular cast iron), and of extremely hard and chemically resistant microstructures for the precise machining of hardened steel and hard cast iron. Apart from temperature dependent transformation toughening (Zr02 additives), high values of the fracture toughness are obtained first of all by anisotropic crystal shapes (whisker reinforcement, rod shaped crystals with high aspect ratio in Si3N4, platelets); generally new mechanism are not likely to be found in the near future. Both the demands for advanced grinding tools and for the precise machining of hard cast iron and hardened steel with cutting ceramics are expected to promote the further development of powder technologies for the manufacture of highly perfect submicrometer microstructures for applications in fields where diamond or cBN tools are too expensive. For grinding materials, corundum (a-A1203) ceramics are dominant at present. Here, contrary to other fields of ceramic technologies, precursor (sol/gel) approaches represent an advanced industrial state of the art. Extended developments are yet in progress, but, on the other hand, it is just the wide industrial use of these procedures

Rejkrences

68 1

which gives insight to some general limitations. New powder routes revealed such a surprisingly large increase of the observed grinding efficiency of the grits compared with sol/gel derived materials, that another leap in the quality has to be expected similar to the progress enabled ten years ago by the introduction of sol/gel approaches. At present, submicrometer cutting ceramics for machining hard workpieces are developed in two classes of materials. In the industry, first tests in Europe are known for carbide reinforced composites based on A1203,whereas more fundamental consideration (Sections 4.4.1 and 4.4.3) and first laboratory results from Asia and Europe (Section 4.4.4) indicate advantages of new grades of pure corundum tools for cutting both mild steel and for machining hard cast iron/hardened steel. Besides of precision machining, first surprising results suggest that pure A1203 cutting tools may be applied for the machining of hardened steel even under conditions where at present only cubic boron nitride can be used (discontinuous cut). On turning hardened steel with pure alumina, high materials removal rates are possible if at high cutting velocities some softening of hardened steel workpieces can be tolerated in a preliminary machining step.

References 1. A. Raab and W. Zielasko, Schneidstoffeinsatz in der Groaserienfertigung, Lecture at the meeting of the VDI-Division Schneidstoffanwendung, 2 I . 1 I . 1996, Siegmaringen. 2. G. Wellein and J. Fabry, Hartdrehen ist mehr als der Einsatz von CBN, Werkzeug-Technik 1996, 45, 7-10. 3. A . Krell and P. Blank, The influence of shaping method on the grain size dependence of strength in dense submicrometre alumina, J . Europ. Ceram. SOC.1996, 16, 1189-1200. 4. F. F. Lange, Sinterability of agglomerated powders, J . Am. Ceram. Soc. 1984, 67. 83-89. 5. T. J. Graule, F. H. Baader, and L. J. Gauckler, Shaping of ceramic green compacts direct from suspensions by enzyme catalyzed reactions, cfi Ber. Dtsch. Keram. Ges. 1994, 71, 317323. 6. S. Tashima, Y. Yamane, H. Kuroki, and N. Narutaki, Cutting performance of high purity alumina tools formed by a high speed centrifugal compaction process, J . Jpn. Soc. Powd. Powd. Metall. 1995, 42, 1464-1468. 7. A. Krell, Fracture origin and strength in advanced pressureless sintered alumina, J . Am. Ceram. SOC.1998, 81, 1900-1906. 8. B. J. Hockey, Observations by transmission electron microscopy on the subsurface damage produced in aluminum oxide by mechanical polishing and grinding, Proc. BY. Ceran?. SOC. 1972, 20, 95-1 15. 9. A. Krell and P. Blank, Grain size dependence of hardness in dense submicrometer alumina, J . Am. Ceram. Soc. 1995,78, 1118-1 120. 10. S. D. Skrovanek and R. C. Bradt, Microhardness of a fine-grain-size A1203,J . Am. Ceram. Soc. 1979,62, 215-216. 11. A. Krell, Improved hardness and hierarchic influences on wear in submicron sintered alumina, Matrr. Sci. Eng. 1996, A209, 156-163. 12. A. Krell, Fortschritte in der spanenden Metallbearbeitung durch Keramiken mit SubmikrometerGefiige, in Hartstofle, Hartstoflychichten, Werkzeuge, Vrrschle&chutz, Proc. of the Hagen Symposium for Powder Metallurgy, 13-14 Nov. 1997, R. Ruthhardt (Ed.), WerkstoffInformationsgesellschaft, Frankfurt, 1997, pp. 57-76.

682


13. H. G. Sowman and M. A. Leitheiser, Non-fused aluminium oxide-based abrasive mineral, a process for its production and abrasive products comprising the said abrasive mineral, European Patent Application EP-24 099 B1, Int. C1.6 C09K3/14, 25.01.1984. 14. T. E. Cottringer, R. H. Van de Merwe, and R. Bauer, Abrasive material and method for preparing the same, U S Patent 4 623 364, C09C1/68, 18.11.1986. 15. G. Bartels, G. Becker, and E. Wagner, Process for the production of a ceramic polycrystalline abrasive, U S Patent 5 034 360, C04B35/10, 23.07.1991. 16. A. Krell, P. Blank, E. Wagner, and G. Bartels, Advances in the grinding efficiency of sintered alumina abrasives, J . Am. Ceram. Soc. 1996,79, 763-769. 17. A. Krell and P. Blank, Al,03 sintering material, process for the production of said material, and use thereof, European Patent 756 586, Int. C1.6 C04B35/111, publ. 19.10.1997. 18. N. G. L. Brandt and Z. D. Senesan, Ceramic cutting tool reinforced by whiskers, US Patent Application 4 867 761, Int. Class. C04B35/117, 10.09.1989. 19. A. G. Evans and B. J. Dalgleish, Creep and fracture of engineering materials and structures, in Proc. 3rd Int. Con$ at University College, B. Wilshire and R. W. Evans (Eds), The Institute of Metals, London, 1987, pp. 929-955. 20. T. N. Tiegs and P. F. Becher, Thermal shock behavior of an alumina-Sic whisker composite, J . Am. Ceram. Soc. 1987,70, C109--C111. 21. J. D. French, J. Zhao, M. P. Harmer, H.-M. Chan, and G. A. Miller, Creep of duplex microstructures, J . Am. Ceram. Soc. 1994, 77, 2857-2865. 22. K. Friedrich, S. Lehmann, M. Fripan, D. Klotz, and H. Ziegelbauer, Mischkeramik auf Aluminiumoxidbasis, European Patent Application 755 904 Al, Int. C1.6 C04B35/117, 29.01.1997. 23. A. Krell, L.-M. Berger, E. Langholf, and P. Blank, Hartstoffverstarkte A1,03-Sinterkeramiken and Verfahren zu deren Herstellung, Germanllnternational Patent Application DE-196 46 344 Al, Int. CL6 C04B35/117, 14.05.1998 / W0-98/21161 A l, C04B35/117, 22.05.1998. 24. M. Fripan and U. Dworak, Keramische Hochleistungsschneidstoffe in der CUR-und Stahlzerspanung, in Schneidstoffe H. Kolaska (Ed.), Verlag Schmid, Freiburg, 1988, pp. 194-214. 25. J. Koike, S. Tashima, S. Wakiya, K. Maruyama, and H. Oikawa, Mechanical properties and microstructure of centrifugally compacted alumina and hot-isostatically-pressed alumina, Muter. Sci. Eng. 1996, A220, 26-34. 26. T. Takahashi, Y. Katsumura, and H. Suzuki, Cutting performance of white ceramics tools having high strength, J. Jpn. Soc. Powd. Powd. Metall. 1994, 41, 33-37. 27. S. Tashima, Y. Yamane, H. Kuroki, and N. Narutaki, Cutting performance of high-purity alumina ceramic tools formed by a high-speed centrifugal compaction process, J. Jpn. Soc. Powd. Powd. Metall. 1995, 42, 464-468. 28. G. K. L. Groh, L. C. Lim, M. Rahman, and S. C. Lim, Effect of grain size on wear behaviour of alumina cutting tools, Wear 1997, 206, 24-32. 29. A. Muchtar and L. C. Lim, Indentation fracture toughness of high purity submicron alumina, Acta Muter. 1998, 46, 1683-1690. 30. A. Krell and D. Klaffke, Effects of grain size and humidity on the fretting wear in fine-grained alumina, AI,O,/TiC, and zirconia, J. Am. Ceram. Soc. 1996, 79, 1139-1 146. 31. V. N. Koinkar and B. Bushan, Microtribiological studies of Al,03. AI2O3/TiC,polycrystalline and single-crystal Mn-Zn ferrite, and Sic head slider materials, Wear 1996, 202, 110-122. 32. A. Krell, P. Blank, L.-M. Berger, and V. Richter, Alumina tools for machining chilled cast iron, hardened steel, Bull. Am. Ceram. Soc. 1999, 78(12), 65-73.


5 Silicon Carbide Based Hard Materials K. A. Schwetz

5.1 Introduction S i c is considered by many as the most important carbide, because of its extreme properties and the variety of present and potential commercial applications. According to Kieffer [l] silicon carbide belongs to the group of nonmetallic hard materials, that is materials whose great hardness and high-melting point result from a high fraction of covalent bonding. Superhard compounds are obviously formed by combination of the four low atomic number elements: boron, carbon, silicon, and nitrogen to form a quarternary system (see Fig. 1); carbon as diamond, boron-nitrogen as cubic boron nitride, boron-carbon as boron carbide, and siliconcarbon as silicon carbide belong to the hardest materials hitherto known.

5.1.1 History Berzelius [2] first reported the formation of silicon carbide in 1810 and 1821. It was later rediscovered during various electro chemical experiments especially by Despretz [3], Schiitzenberger [4], and Moissan [5]. However it was Acheson [6], who first realized the technical importance of the carbide as a hard material and, by 1891, had managed to prepare silicon carbide on a large scale. Originally, Acheson had completely different things in mind: he wanted to produce artificial diamonds by recrystallizing graphite in an aluminum silicate melt; at a time when high-pressure and high-temperature chemistry was unknown. His efforts proved fruitless, but he was rewarded by the discovery of a far more versatile material, silicon carbide. Believing that this was a compound of carbon and corundum, he named the new substance ‘carborundum’ (Fig. 2). Today, silicon carbide has become by far the most widely used nonoxide ceramic material, with an annual world production of about 700 000 tonnes. Owing to its great hardness, heat resistance, and oxidation resistance it has become firmly established as an abrasive as well as a raw material for making refractories such as firebricks, setter tiles and heating elements. Another major use of silicon carbide is as a siliconizing and carburizing agent in iron and steel metallurgy. Self-bonded silicon carbide materials in molded form and of high S i c content in order to make full use of the excellent material properties, have been available only from the late 1960s onwards. They are being increasingly used as refractory materials as well as for structural components in mechanical engineering, and have proved highly successful for use under extreme conditions involving abrasion, corrosion, and high temperatures.

684

5 Silicon Carbide Based Hard Materials N

raphite) (%iamond) (

Figure 1. Nonmetallic hard materials in the system B-C-Si-N.

5.1.2 Natural Occurrence [7] Naturally occurring silicon carbide was first discovered by Moissan [8] in a meteorite from the Diablo Canyon in Arizona, and subsequently several small deposits have been found [9-111. Extraterrestrial occurrences of silicon carbide are even

Figure 2. Crystalline Sic grain from the Acheson process.

5.2 Structure und Phase Relations of S i c

685

rarer [7]. Its existence in interstellar space has been shown by spectral measurements, and it has also been found in samples of moon rock.

5.2 Structure and Phase Relations of S i c As shown in the phase diagram in Fig. 3, silicon carbide is the only binary phase in the siliconxarbon system, with the composition 70.05 weight-% Si, 29.95 weight-% C. Silicon carbide does not have a congruent melting point. In a closed system at a total pressure of 1 bar it decomposes at 2830 f 40°C into graphite and a silicon-rich melt (Fig. 2). This is the highest temperature at which silicon carbide crystals are formed. In an open system, silicon carbide starts to decompose at =2300"C with the formation of gaseous silicon and a residue of graphite. S i c and Si form a degenerate eutectic at 1413°C and 0.02at% C. The solubility of carbon in liquid silicon is 13 at% C at the peritectic temperature [ 121. Silicon carbide exists in several modifications being polymorphic and polytypical and crystallizing in a diamond lattice, like silicon [13]. Polycrystalline silicon carbide obtained by the Acheson process exhibits a large number of different structures (polytypes), some of which dominate. These can be classified in the cubic, hexagonal, and rhombohedra1 crystal systems (Table 1).

4000 3826

Y

3232 3000

L+C

-5 1.5

>99.97

v

2 3 ;2000 L

-0.5

Q)

?

1413

1414 1000

49.4 e 0.6 /

-

Si

0

*1

+ Sic

20

40

Si + S i c

80

)0

Carbon (at. %)

Figure 3. Phase diagram of the silicon-carbon system at 1 bar total pressure after Kleykamp and Schumacher [ 121.

686

5 Silicon Carbide Based Hard Materials

Table 1. Dominant SIC types Modification

Polytype

clSiC (high-temperature modification)

6H 15R 4H

PSiC (low-temperature modification)

3c

Since silicon carbide predominantly exists in its beta form at temperatures below 2O0O0C,this is referred to as the low-temperature modification. Cubic PSiC is metastable and, in accordance with Ostwald' s rule, is formed initially in Sic production from silicon dioxide and carbon. PSiC (polytype 3C) can also be prepared at about 1450°C from simple mixtures of silicon and carbon or by hydrogen reduction of organosilanes at temperatures below 2000°C. Above 2000"C, only the hexagonal and rhombohedra1 types are stable if there are no stabilizing influences which raise the transformation temperature [ 14,151. Thus, in a nitrogen atmosphere at high pressure, PSiC is the stable form, so that above 2000°C aSiC transforms to PSiC. On the other hand, a hexagonal polytype, the wurtzite or 2H modification of Sic, has been obtained at temperatures as low as 1300-1600°C by the decomposition methyltrichlorosilane CH3SiCI3 [ 161. This 2H polytype, which does not occur in commercial Sic, and the most common 3C, 4H, 6H and 15R structures are called short-period polytypes. Long-period polytypes are much less common and consist of blocks of short-period polytypes which are broken by regularly occurring stacking faults. The basic element of the silicon carbide structure is the tetrahedron [17] due to sp' hybridization of the atomic orbitals. This tetrahedron consists of a silicon or a carbon atom at the spatial center, surrounded by four atoms of the other kind. The Sic- bond is 88% covalent. The tetrahedra are arranged in such a way that units of three silicon and three carbon atoms form angled hexagons which are arranged in parallel layers as shown in Fig. 4. The layer sequences can repeat themselves in the cycles ABC, ABC . . . (zinc blende, type 3C) or AB, AB . . . (wurtzite, type 2H), according to cubic or hexagonal close packing. In addition, numerous others stack sequences are formed in the case of silicon carbide, resulting in many similar polytypes. About 180 different polytypes are currently known. They all have the same density, 3.21 g ~ r n - ~ The . formation and stability of the various polytypes are not yet clearly understood. Written polytype nomenclature [ 181 indicates the number of layers in the repeating layer pack by a numeral, while the crystal system is denoted by the letters C, H, or R. The amounts of the most frequently encountered polytypes 3C, 4H, 15R, and 6H can be quantitatively determined by X-ray diffraction techniques [ 191, which can indicate the temperature of formation of the silicon carbide since the stability of a given polytype depends on, among other things, temperature. According to Inomata et al. [20] 2H-Sic seems to be stable below 14OO0C, 3C between 1400 and about 1600"C, 4H between 1600 and 2100"C, 6H above 2100°C and 15R above 2200°C.

5.2 Structure and Phase Relations o j S i C

687

u - S i c l6H)

Figure 4. Crystal structure of the SIC polytypes 3C and 6H.

There is no doubt, however, that impurities also play a part in the formation of the various polytypes, together with the surrounding gas atmosphere and growthkinetic processes. Preferential stabilization of the 4H polytype is achieved with 4 which is thought additions of aluminum up to concentrations of ~ 0 . weight-%, to substitute for silicon in the silicon carbide lattice [21-241. According to Lundquist [21] pure CcSiC mainly crystallizes as 6H, with small amounts of aluminum 15R also occurs, and above 0.1 weight-% aluminum, 4H predominates (see Fig. 5). Also it is well known today that nitrogen can be incorporated into the S i c lattice, whereby it stabilizes not only the cubic PSiC form but also the wurtzite 2H polytype [14,25]. Whereas the nitrogen content obtained in 3C-Sic by gas/solid diffusion is low, e.g. only < 1 atom YO,extremely high nitrogen contents of up to = 14 atom YOwere

A l content. w t %

-

Figure 5. Proportions of polytypes in technical-grade S i c as a function of Al content

688

5 Silicon Carbide Bused Hard Muterials

recently found for nitrogen-stabilized 2H Sic when formed from carbothermal reduction of silicon nitride, Si3N4,in the presence of a liquid phase [26]. Transmission electron microscopy using high resolution lattice imaging techniques allows polytype analysis within single grains in the microstructure of dense S i c bodies [27].

5.3 Production of Sic 5.3.1 The AchesonlESK Process Abrasive and refractory grade silicon carbide is produced industrially according to the following reaction Si02

+ 3C = Sic + 2CO

from high-purity quartz sand (99.5% SO2) and petroleum coke, in electrically heated resistance furnaces [28]. The raw materials should have a maximum grain size of 10 mm. In addition to the reactants, sawdust is sometimes added to increase the porosity of the mix and to facilitate the venting of the carbon monoxide formed. The reaction is strongly endothermic: the heat of formation is 618 kJ mol-' Sic, corresponding to 4.28 kWh kg-I Sic. The process takes place at temperatures of between 1600°C and 2500°C and is far more complex than the above equation because of the intermediate reactions, some of which occur with participation of the gas phase [29], [36]. Acheson Sic furnaces are up to 25 m long, 4 m wide, and 4 m high, consuming up to 5000 kVA, and running for about 130 h. The largest furnaces operate at up to 25000A. The furnace has a rectangular cross section and consists of two watercooled graphite electrodes at the end and two movable side walls, whose purpose is to retain the raw material mixture. When about half of the mixture has been charged to a depth even with the electrodes, a column of tamped graphite powder or a solid graphite rod is inserted to connect the electrodes. This so-called core is then covered with additional reaction mixture. The electrodes are then connected to the transformer, and the current is switched on. The silicon carbide is formed in the hot core in the form of a polycrystalline, compact cylinder, the so-called roll. Temperatures within the furnace reach 2500°C causing the silicon carbide to recrystallize. The surrounding mixture acts as a thermal insulator. The carbon monoxide byproduct escapes through the surface of the reaction mixture and through slits in the side walls, where it burns. After the current is switched of, the furnace is allowed to cool for several days. Then the side walls are removed, and the excess reaction mixture is carried off. The roll of silicon carbide contains not only the original resistance core, but also graphite formed by decomposition of silicon carbide. The boundary between the graphite and the Sic is very sharply defined. The innermost Sic zone is of the highest quality, as the purest and largest crystals can form in this region. Moving

5.3 Production of S i c

30 m

I-

689

4

Resistance core

Reaction mixture

Current conductors

Resistance core

n / Reaction mixture

~

/ Gas collection duct

I

Porous bed

Figure 6. Silicon carbide production by the ESK process [7, 301: (a) longitudinal and (b) radial cross-sections through the furnace

outwards, with decreasing temperature, the crystal growth rate decreases and the S i c roll becomes more and more finely crystalline. The outer zone is composed of very fine crystals of SSiC. While in the case of the classic Acheson furnaces, with their laterally positioned electrodes and side walls, the carbon monoxide burns away on the surface, polluting the atmosphere inside and outside the furnace building, the ESK (Elektroschmelzwerk Kempten) process [30] developed in the 1970s (Fig. 6), is ecologically safe. It operates with floor electrodes, which enter the reaction mass from below a porous floor, and the carbon monoxide is collected by a gastight plastic canopy and transported through gas collecting pipelines to a power station where it is converted into electrical energy after burning. This energy is recycled and amounts to 20% of the energy necessary for the production

690


Figure 7. SIC roll at the ESD facility in Delfzijl, The Netherlands.

of silicon carbide. The silicon carbide cylinders inside the linear or U-shaped resistance furnaces are up to 60 m long and weigh about 300 tonnes after a run of 10 days (see Fig. 7). A comparative review of some Acheson plants has been published [3I]. The lumps of silicon carbide obtained on dismantling the furnace are taken to a sorting area and broken down with hydraulic equipment and compressed air hammers. The finely crystalline outer zone (the so-called amorphous zone) is further size reduced (e.g., in a jaw crusher) and is usually returned to the furnace process by adding it to the coke/sand mixture. Alternatively, this 9&95% Sic material is a valuable raw material for the production of grain refiners for the ferrous metal industry. The coarsely crystalline compact pieces from the inner zone of the Sic roll are sorted according to criteria such as color, structure, etc. This top-grade material forms the largest fraction of the Sic yield from a furnace charge. Further processing of the silicon carbide into the various macro- and microgrits depends upon such factors as the required particle size distribution [32] and shape, and involves breaking, grinding, sieving, air classification, and elutriation. For applications that require a particularly high grade of silicon carbide, such as industrial ceramics, the refractory industry, and ceramic-bound abrasives, the granular product must be washed with alkali or acid to remove adhering traces of elemental silicon, metals, metal compounds, graphite, dust, and silica. Commercial Sic grades vary in composition from 90% to 99.5% Sic. Light green S i c is 99.8% pure, and, as the purity decreases to 99%, the color changes to dark green, and, at 98.5% to black. Before aSiC is converted into molded parts by sintering, it must be intensively milled into the submicrometer range, that is to a mean particle size of less than 1 pm (see Fig. 8b), in attritors [33,34] or ring-gap mills. Adherent impurities which interfere with the sintering process, such as SiOz or wear from milling equipment, are removed by wet chemical purification [35].

5.3 Production of' S i c

69 1

Figure 8. Silicon carbide submicron powders: (a) PSiC powder, specific surface area: 12.6 m2 g-' [40] and (b) CtSiC powder, S = 15.8m2g-' [35].

5.3.2 Other Production Methods 5.3.2.1 PSiC Powder 5.3.2.1.1 Carbothermal Reduction of Silica This process is based on the following equation Si02

-

+ 3C < 1.900'C

PSiC

+ 2CO.

According to the reaction mechanism proposed by Weimer [36] the reaction takes place in several steps: SiO,(S) + C(S)

+ CO(,)

+

+

Si02(,) CO(,) + SO(,) C(S)

SiO(,)

+ C02(g)

+ 2C(,)

+

+ C02(g)

2CO(,)

+

Sic(,) CO(,).

The initial reaction with formation of gaseous SiO at points of direct contact between C and Si02 requires temperatures above M 1600°C at atmospheric

692


pressure (solid/solid reaction). The high rate of SiO(,) formation suggests that the further two gas/solid reactions make the final carbon reduction of the SiO to Sic possible once the C/Si02 contact points are consumed. Carbon crystallite diameter was found to have a substantial influence on the rate of reaction and the size of synthesized Sic. This process has been used in the early 1960s to fabricate pure ultrafine PSiC powders especially suited for the semiconductor and pigment industry [37,38]. During the first step of the General Electric Process [37] a silica gel is formed by hydrolysis of silicon tetrachloride in an aqueous solution of sugar (sucrose). Upon dehydratization at 300°C the sugar pyrolizes and an intimate amorphous carbon/amorphous silica mixture is obtained which is subsequently fired at 1800°C in an inert atmosphere to form PSiC. Later the process was modified for the fabrication of PSiC sintering powder [39, 401, using lower reaction temperatures and as starting materials the combinations tetraethylsilicate/sugar and fumed amorphous silica/sugar, respectively. Particle size control corresponding to powder surface areas in the 10-20 m2 g-' range (see Fig. 8a) was necessary, because ultrafine powders have high oxygen contents (surface silica), and often low green densities result due to their high surface area. Since 1981 a continuous version of the Acheson process has been commercially operated by the Superior Graphite Company [41]. A mixture of silica sand and petroleum coke is fed continuously to the furnace and heated electrothermally to a temperature of 1900°C. The run-off-furnace product is a free flowing grain generally smaller than 3 mm, comprising agglomerates of PSiC crystallites and free carbon. The grain is friable and is readily processed to make sized microgrits and sintering powder by dry milling, wet grinding, classification, and wet chemical purification (removal of 5 weight-% free carbon). 5.3.2.1.2 Synthesis from the Elements Prochazka [39] synthesized submicron PSiC powder with a surface area of 7 m2 g-' using direct reaction of high-purity silicon powder with carbon-black at temperatures between 1500 and 1650"C, but saw little densification with boron and carbon additions, suggesting that free silicon is detrimental to densification. The in situ formation of silicon carbide according to the reaction Si(l,,) + C(S) = SIC(,) is mainly used in the manufacture of reaction sintered Sic components. After shaping a mixture of fine-grained Sic powder and carbon and exposure at high temperatures to liquid or vapor-phase silicon, the carbon is reacted to secondary silicon carbide, which bonds together the primary Sic grains. Excess silicon (about 10vol-Yo) is usually left to fill the pores, thus yielding a nonporous body (Si-Sic). 5.3.2.1.3 Deposition from the Vapor Phase Submicron PSiC can be continuously produced by decomposing gaseous or volatile compounds of silicon and carbon in inert or reducing atmospheres, at temperatures above 1400°C [42]. Particle size

5.3 Production ojsic

693

and morphology depend considerably on the reaction temperature and on the composition in the gas phase. Powders obtained below 1100°C are amorphous. A great variety of reactants and several methods of heating, such as d.c. arc jet plasma [43], high frequency plasma [44],laser [45], and thermal radiation are possible. However, large scale commercialization of these processes has been limited by the very high cost of production. Small particle sizes are produced, 0.01-0.5 pm, with greater than 70% efficiency using silicon tetrachloride (SiC14) and methane (CH4) or methyltrichlorosilane (CH3SiC13)as the gaseous precursors in the presence of hydrogen as the plasma gas:

SiC14(,)

H + CH4(g)--%

+

Sic(,) 4HC1(,).

If boron trichloride is used for the boron dopant together with a small excess of the carbon feedstock, sinterable Sic powders are obtained. The laser synthesis allows laboratory scale production of high-purity S i c powder with controlled particle sizes. Using a C 0 2 laser and a mixture of silane (SiH4) and methane (CH4) or acetylene (C2H2) PSiC particles in the size range 5-200 nm have been obtained. An advantage of using SiH4, rather than SiC14, as a reactant is that it has a strong absorption band near the wavelength of the laser (10.6 pm). Amorphous S i c powders with metal impurity levels below 60 p.p.m. total have been synthesized by thermal decompostion of a methylsilane gas stream at temperatures above 600°C according to the equation CH3SiH3

Sic

+ 3H2

in an induction heated graphite reactor [46]. 5.3.2.2 SIC Whiskers Sic whiskers are discontinuous, monocrystalline hair or ribbon-shaped fibres in the size range 0.1-5 pm in diameter and 5-100 pm in average length (see Fig. lOa).Their aspect ratio (length/diameter) ranges from 20 to several 100 and is an important figure. Sic whiskers are all grown from the gas phase on solid or liquid substrates under conditions that result in a small defect size (0.14.4 pm) in the whisker. Such small defect size results in very high strength, of about 9-17 GPa. Whiskers grown at temperatures below 1800°C are composed of mostly PSiC, those grown above 2000°C are of clSiC. According to Shaffer [47] preparation methods for whiskers can be divided into three routes: (i) vapor-liquid-solid process (VLS), (ii) vapor phase formation and condensation (VC), (iii) vapor-solid reaction (VS). 5.3.2.2.1 The Vapor-Liquid-Solid Process In the VLS process silicon and carbonrich vapors (usually CH4, SiO, or SiC14) react at 1400°C to from S i c on a liquid

694


,,.,. Vapor Feed

Temperature = 1400°C

<

Vapor Feed

sio

c+sI

CSolid Sic Clystals

Liquid Catalyst Supersaturated with Si+C Solid Catalyst (Steel 30vrn)

\1 6%

/////// / / / / / / / / / / / / Substrate (Carbon)

I

77

Figure 9. VLS process for S i c whisker growth (1400°C).

alloy:

+

Si(d) C(d)

1400°C

___t

Sic.

(d) dissolved in liquid iron alloy

Microscopic particles of an alloy are distributed on a substrate (graphite) then exposed to silicon- and carbon-rich vapors. The presence of a liquid catalyst, such as a transition metal or usually an iron alloy, distinguishes this method from all other whisker preparation methods. Although the VLS-method yields near-perfect whiskers and has been investigated since the early 1960s, it has not been developed to economical commercial practice up to now. Figure 9 illustrates the VLS whisker growth sequence. At 1400°C the solid catalyst melts and forms the liquid catalyst ball. Carbon and silicon from the gas phase dissolve into the liquid catalyst, which soon becomes supersaturated and solid S ic precipitates at the interface with the carbon substrate. Continued dissolution of gas species into the liquid catalyst allows the whisker to grow, lifting the catalyst ball from the substrate as additional S i c precipitates. These VLS whiskers are typically larger in diameter (4-6 pm) than those formed by the vapor-solid process (VSwhiskers). They frequently grow to length of tens of millimeters.

5.3.2.2.2 Vapor Phase Formation and Condensation To grow whiskers by this process, bulk S i c is vaporized by heating to very high temperatures (>22OO0C) usually under reduced pressure. Upon cooling aSiC whiskers form on nucleation sites. Sublimation Sichulk

A

Sicwhisker

5.3 Production

of Sic

695

The addition of lanthanum, yttrium, neodymium, or zirconium leads to an increase in grow rates. However, today no whiskers are commercially produced by this sublimation process.

5.3.2.2.3 Vapor-Solid Reaction At present the principal commercial method for SIC whisker production is the carbothermic reduction of low cost silica sources at temperatures of 1500-1 700°C. The reaction for the formation of VS-cubic PSiC whiskers occurs in two steps: SiOz + C SiO

-+

SiO

-

+ 2C <18oo-c

+ CO T,

DSiC + CO

.

Many different VS-processes are known according to the starting raw materials and catalysts species involved. In all cases, a catalyst is used for whisker formation to be favored versus particulate S i c formation. Commercially available VP-whisker grades are typically less than 1 pm in diameter (submicron) and up to 30 pm in length and show strength/Young modulus values of 1-5 GPa/400-500 GPa. During whisker handling and processing, rigorous protection has to be provided since smaller whiskers (< 3 pm diameter) can become lodged in the lungs and represent a health hazard. An inexpensive VS-process utilized calcined rice hulls, which contain both silica and carbon, as a precursor to S i c whiskers. The characteristics of commercial S ic whiskers as claimed by producers are given in Table 2. PSiC whisker have been added to a wide variety of ceramics (A1203,Si3N4,AlN, MoSi2, mullite, cordierite, and glass ceramics) in an effort to achieve increased toughness. In most cases, hot pressing has been required to achieve near-theoretical density of the matrix. A real success story in the advanced ceramic market was the introduction of a 25-30 weight-% S i c whisker reinforced hot-pressed alumina for the high-speed cutting of high nickel content alloys. Thus, by incorporation of S i c whiskers into the alumina matrix a composite with increased strength (by over %YO),fracture toughness (by loo%), increased fatigue and thermal shock resistance was obtained [48]. In this composite, whisker pull out and crack bridging have been identified as the major toughening mechanisms. Crack bridging consists in bridging the crack behind the crack tip by unfractured whiskers or fibers. The effect depends on the strength of the fiber and on its aspect ratio. Inhibiting crack Table 2. Comparison of the properties of PSiC whisker (supplier's data).

Supplier Phase Length (pm) Diameter (pm) Aspect ratio Density (gcm-')

Tokai Carbon

Tateho

P

P

20-50 0.3-1.0

5-200 0.05-1.5 20-200 3.18

3.2

Shin Etsu

C-Axis Technology

a

P

3-20 0.15-0.25 20-50

0.3-1.0 20-50 3.21

Millenium Materials Inc.

ART

P

P

3.21

0.5-1.7 5-20 3.21

30-200 1-3

696


Table 3. Morphological characteristics of commercial S i c platelets. Supp1ier

C-Axis Technology SF grade

Millenium Materials 325 mesh

Carborundum Corporation

Maximum dimension (pm) Thickness (pm) Aspect ratio

5-50 2-5 8-10

5-70 0.5-5 8-1 5

10-500 1-5 10-1 5

opening reduces the stress intensity at the tip of the crack and stops or slows crack propagation. The fiber or whisker pull out has the potential for the greatest toughening effect and consists in a debonding of fiber from the adjacent matrix consuming large energy. This mechanism depends strongly on the aspect ratio of the second phase and is favored by weak interface bonds. Sic whisker also can be combined with metals (MMCs, metal matrix composites) to increase the high temperature strength of the material as well as provide a comparable substitution for heavier traditional materials, such as steel. MMCs can be fabricated by infiltrating a Sic whisker preform with aluminum or by addition of S i c whiskers to molten aluminum.

5.3.2.3 Sic Platelets As a reinforcement material for all matrix materials, ceramics, metals and plastics, Sic platelets offer a similar potential as whiskers but at lower cost and without health hazard. Platelets are single-crystal, plate-like aSiC particles with an aspect ratio of about 8-15. Sic-platelets typically range in size from about 5 to 100 pm in diameter and 1 to 5 pm in thickness (see Table 3 and Fig. lOb).They are commercially produced from inexpensive raw materials (silica and carbon, micron-sized PSiC powders) at temperatures of 1900-2100°C under inert atmosphere [49]. Due to the presence of boron and aluminum dopants added, platelet-shaped crystals are formed. Aluminum enhances the growth in the [OOOl] direction and decelerates the growth perpendicular to the [OOOI] direction. Boron enhances the growth notoriously perpendicular to the [OOOl] direction [50]. As impurity elements in Sic platelets from Millenium Materials, aluminum of 0.0440.45weight-%, boron and nickel of 0.44.8 weight-% (each), and free silicon of 0.3-3.6 weight-% were analyzed [51]. With the aid of high resolution AES most impurity elements were found to be present on the outermost surfaces of the platelets, for example boron was found as about a 3 pm thick layer of B203,as about 1 pm thin layers consisting of boron or silicon borides and as 3-5nm thick segregations near to growth steps [51]. Dense Sic platelet reinforced silicon carbide composites were recently fabricated [52]. The highest sintered densities of 97-98% TD were achieved with 20% platelet contents using hot molding as the thermoplastic forming technique [53]. In addition to reinforcing ceramics, Sic platelets are also used to increase the strength, wear resistance and thermal shock performance of aluminum matrices, and to enhance the properties of polymeric matrices.

5.3 Production of Sic

697

Figure 10. (a) PSiC whisker, and (b) CtSiC platelets (Courtesy S. Weaver, Millennium Materials, Inc., Knoxville, USA).

5.3.2.4 SIC Fibers Together with boron filaments and carbon fibers, Sic fibers are important continuous inorganic reinforcement materials with high modulus. Owing to their high-temperature properties and the resistance to oxidation they are particularly well suited for structural parts where high stiffness at high temperatures is required. Three different types of S i c fibers are now available: substrate-based fibers (CVD-fibers), polymer-pyrolysis derived fibers (PP-fibers), and sintered powder derived fibers (SP-fibers). Defect size in amorphous or polycrystalline fibers is very small so that strength can be high. In Table 4 some commercially available Sic fibers are listed along with their mechanical characteristics.

5.3.2.4.1 Substrate-based Fibers (CVD-Fibers) PSiC filaments of 100-150 pm thickness can be prepared by chemical vapor deposition onto tungsten or carbon Table 4. Mechanical Characteristics of various S i c fibers Supplier

Textron SCS-6

Nippon Carbon Nicalon

UBE Tyranno

Carborundum Sintered-Alpha

Crystal phase Diameter (pm) Tensile strength (MPa) Young modulus (GPa)

PSiC 140 3920 406

PSiC 10-20 > 3000 > 420

amorphous 8-10 > 3000 > 200

CcSiC 20-1 50 up to 1380 400

698


monofilaments of 40 pm thickness, which act as hot substrates during heterogeneous nucleation. Various carbon containing silanes have been used as reactants. Owing to high process costs and the large diameters of the resulting fibers, the method is disadvantageous. It has been found that faster Sic filament production is possible on a carbon filament substrate. The US company Textron produces PSiC fibers on a carbon-fiber-core with a surface layer of pyrolytic carbon, which is itself coated with silicon carbide. These fibers are given the designation SCS. The surface carbon layer provides a toughness-enhancing parting layer in composites having a brittle matrix. 5.3.2.4.2 Polymer Pyrolysis Derived S i c Fibers (PP-fibers) As shown 1976 by Yajima [54], PSiC fibers with a smaller diameter (8-30 pm) and without a central core can be manufactured by solid state pyrolysis of a polycarbosilane (PCS) precursor fiber. The melt-spun PCS fiber is first cured at 200°C in air to produce a thin layer to protect from melting later on, then heated up in inert atmosphere to 1500°C to convert the PCS in crystalline PSiC. The steps leading to the production of Sic can be summarized as follows:

(-Si(CH3) - CH2-),

I

1300°C

PSiC

+ CH4 + H2.

H Dimethyldichlorsilane is dechlorinated with metallic sodium in the presence of xylene to polydimethylsilane (R2Si),, which is converted to polycarbosilane (PCS) by polycondensation with introduction of silicon-carbon bonds into the chain. The pyrolysis of the PCS resin to PSiC is accomplished by a slow heating to 1300°C and is accompanied by an appreciable loss of mass (=60%) and considerable shrinkage. This melt spinning-curing-pyrolysis route has been adopted by Nippon Carbon to produce a fiber called Nicalon [55] which has a composition almost identical to Si3C40thus, it is not pure Sic, but a Si-C-0 composite, in which nanometer-sized PSiC, Si02 and carbon are uniformly dispersed. These impurities affect the thermal stability of the fiber and its strength, which has been reported to degrade above 1200°C. However, these fibers have successfully been used to reinforce aluminum, refractories and Li-Al-Silicate glass ceramics. The coating of fibers, for instance by carbon, allows a decrease in the interface bonding with the matrix, which increases the strength and the toughness by favoring pull-out mechanism. UBE Chemicals synthesized amorphous Si-Ti-C-0 fibers from the PCS-titanium alkoxide compound polymer. These so called Tyranno fibers show excellent properties and can be spun thinner than the Nicalon fibers (see Table 4). However heating above 1000°C results in a crystallization of the fiber. In the mid 1990s the German company Bayer AG synthesized an amorphous Si-B-N-C fiber, by pyrolysis of a polyborosilazane polymer [56]. This SiBN3Cfiber (see Fig. 11) has a tensile strength of 3 GPa and maintains its amorphous character up to 1800°C.The advantage of the production route from liquid to solid to produce S i c has also attracted attention for

5.3 Production

of

sic

699

Figure 11. SEM of SiBN3C fibers made by Bayer AG, Germany (courtesy H.P. Baldus, [56]).

the S i c film production in microelectronics or as protection layers. Amorphous, and polycrystalline films of high purity produced by dip-coating of substrates in PCS solutions and subsequent pyrolysis in inert gas atmosphere have been prepared [57].

5.3.2.4.3 Sintered Powder Derived aSiC Fibers (SP-fibers) Recently, the Carborundum Company developed a single phase polycrystalline aSiC fiber having a diameter of approximately 20-1 50 pm and density of over 96% of theoretical. It is made by first producing green filaments by melt extrusion or suspension spinning of plasticized mixtures of sinterable aSiC powder with organic additives, such as polyethylene or polyvinylbutyral plus novolac, respectively. The green filaments are subsequently debindered during free fall from the extruder in presintering furnaces and finally undergo pressureless sintered at 2 100°C in argon atmosphere. These polycrystalline aSiC fibers are stable up to 1600°C in air and 2250°C in inert gas, far superior than most of the commerciallly available PSiC based fibers [49]. According to Prochazka [%],an increase in the strength of aSiC fibers is obtainable by using refined aSiC powders, having a surface area greater than 20m2gp', a median S i c particle size <0.25pm, and no particles larger than 1.5 pm.

5.3.3 Dense Sic Shapes S i c can be converted into more or less dense parts either by using various bonding phases to bind the SIC particles or grains together a t elevated temperatures without dimensional change of the initial compact, or by sintering (pressureless, sinter/HIP or hot pressing), that is, by mechanisms of mass transport between particles which lead to porosity elimination together with shrinkage of the overall body.

700


5.3.3.1 Ceramically Bonded Silicon Carbide This product group is based on oxide- or nitride-bonded S i c grains and exhibits an apparent porosity of up to 20% by volume. A fracture surface of a silicon nitride bonded Sic is shown in Fig. 12a. Coarse silicon carbide powder fractions are mixed with clay, or pure oxides like Si02 and/or A1203,formed to 'green' shapes by conventional ceramic methods, and fired at about 1400°C.The fired clay or oxidic raw materials bind the silicon carbides particles and, although these are only loosely bonded, this is sufficient for many types of applications such as refractory furnace bricks and abrasive disc. Firing mixtures of silicon carbide particles and elemental silicon in a nitrogen atmosphere, with or without the addition of air and possibly A1203produces silicon nitride-, silicon oxynitride- or SiAlON (silicon aluminum 0xynitride)-bonded Sic products with enhanced high-temperature strength [59-6 11. 5.3.3.2 Recrystallized Silicon Carbide Silicon carbide shapes with a high green density made by slip-casting or press-molding silicon carbide powders of bimodal particle size distribution are fired in an electric furnace at temperatures of up to 25OO0C, under the exclusion of air [62,63]. Evaporation and condensation [64] take place at temperatures above 2 100°C,resulting in a selfbonded structure without shrinkage. The initial (green) density and final density remain the same and solid S i c bonds are produced between the crystals. The final RSiC product, which shows transgranular fracture (see Fig. 12b), consists of 100% Sic and can have densities of up to 2.6gcm-', with a porosity of approximately 20%. 5.3.3.3 Reaction-Bonded Silicon Carbide In reaction sintering or reaction bonding silicon carbide is produced to some extent by chemical synthesis in the molding. A preform consisting of silicon carbide, carbon, and a carbon-containing binder is made by one of the normal ceramic shaping techniques (dry press molding, slip casting, extrusion, or injection molding, see Table 5). The elemental carbon in the preform is then reacted with gaseous or liquid silicon. The process can be controlled in such a way that a porous but pure Sic part is produced, or one whose pores are filled with excess silicon [65-731. Since the latter (SiSiC), which is porefree dense (high density, showing no residual porosity, see Fig. 12c), is much stronger it is this which is most frequently made, usually by infiltration with liquid silicon. Silicon infiltration can also be used to fill the pores of recrystallized Sic moldings at a later stage [74]. The amount of free silicon in infiltrated silicon carbide is usually between 10% and 15%. It is difficult to bring this below 5%. Infiltration is relatively easy, because of the good wettability of the material, although there are problems if the bodies are large or thick, because of the exothermic reaction and the expansion of silicon on freezing, which make perfect impregnation difficult. Reaction sintering can, however, usually be controlled so that the change in volume is negligible, resulting in dimensionally accurate components. This is

5.3 Production of’ S i c

701

Figure 12. Microstructure of various S i c ceramics; (a) KSiC, nitride bonded S i c (AnnaSicon 25); (b) RSiC, recrystallized Sic (AnnaNox CK); (c) SiSiC silicon infiltrated B C (AnnaNox CD), (d) SSiC, BjC doped (Ekasic F), (e) LPSSiC, equiaxed SIC grains (Ekasic T), (f) LPSSiC, platelet S i c grains (ESK), (g) HPSiC,Al-doped (ESK), (h) HIPSIC, undoped (ESK).

102


Figure 12. Continued


703

Figure 12. Continued

achieved by choosing the right S i c and carbon contents in the initial mix, and ensuring correct porosity in the preform and by proper temperature control. 5.3.3.4 Sintered Silicon Carbide 5.3.3.4.1 Solid-state Sintered S i c Once it was thought that silicon carbide was unsinterable. It was only in the early 1970s that it became possible to sinter silicon carbide without applied pressure and thus to achieve densities more than 95% of the theoretical density. This was largely due to the pioneering work of Prochazka [75] at the General Electric Corporation Research Center in Schenectady. The starting materials were submicrometer PSiC [76] or later aSiC [77] with the simultaneous addition of up to 2% carbon and boron. A minimum of 0.3 weight-% boron was required, although larger additions did not improve the density, which Prochazka attributed to the need to exceed the solubility limit of boron in S i c (0.2 weight-% according to Shaffer [78]). Carbon is added in excess to ensure complete deoxidation, that is removal of the Si02 film from the S i c grains, which in turn leads to the presence of carbon inclusions. Additions of excess carbon are observed to inhibit

104

5 Silicon Carbide Bused Hard Materials

Table 5. Shaping techniques for manufacture of S i c preforms (green bodies). Feasible molds, recommended dimensions

Type of tool, tool life

Yield for average parts in pieces/8 h

Automatic or dry pressing

Pellet-like parts in pressing directions not higher than three times the diameter; hole distances not less than 0.2 diameter; hole diameter > 0.1 height of part

Cemented carbide molds more than lo6 pieces

5000 pieces using single mold of average dimensions

Extrusion

Extrusions which can be Extrusion die Approx. cut to size, diameters made of steel, 300 kg of 1. 540 mm wall minimal wear compound thickness > 1 mm, lengths up to 1OOOmm

Slip casting

Complex shapes with unit weights of up to 5 kg and measuring up to 400 mm

Injection molding

Two-part molds, Usually suitable for unit weights multicavity molds made of up to 250 g of steel, up to 50 000 shots

Isostatic pressing and dry grinding

Preferably axially symmetric bodies, maximum diameter 320 mm, maximum length 800 mm

Moulding process

Diagram function scheme

Plaster molds from wooden patterns, up to 40 castings depending on mold

1-10 castings per mold, depending on wall thickness

About 500 pieces per single-cavity mold

Simple rubber About 15 molds isostatically press molded parts

grain growth, and produce a more equiaxed grain structure [79],probably as a result of the energy required from the grain boundaries to overcome the carbon inclusions. Instead of boron and its compounds [80-831 one can also use aluminum and aluminum compounds [84-901, or beryllium and its compounds [91], the net result being the same. Sintering of AljC doped [92] or A1203-doped[90] Sic powders is believed to occur in two stages: transient liquid phase sintering; then solid state sintering during the final stage. This opened up an inexpensive method of producing dense and complex parts consisting of pure silicon carbide (SSiC). The powders (see Fig. 8) can be molded into a green body by any of the methods used in ceramic molding [93,94], depending on the shape required and the number of pieces involved (see Table 5). Sintering is


705

carried out in an inert gas atmosphere or in a vacuum at temperatures between 1900°C and 2200°C. Linear shrinkage exceeds 15% and depends on the green density of the molding. During the sintering process, there is Sic-polytype transformation as well as grain growth, whose extent depends on the type and amount of sintering additives, as well as the sintering temperature. Boron-doped PSiC powders tend towards secondary recystallization (exaggerated grain growth [95] during 3C + 6H polytype transformation) whereas, if one starts with CtSiC powders, a homogeneous, fine-grained microstructure can be obtained (see Fig. 12d). clSiC powder, which predominantly consists of the 6H-type, usually shows 6H + 4H polytype transformation during sintering with aluminum, as well as boron containing dopants. Aluminum promotes 6H + 4H transformation at lower sintering temperatures (2050"C), boron most easily at higher temperatures (2200°C). According to investigations of Schwetz et al. [96] there seems to be no fundamental correlation between Sic-polytype transformation and microstructure development during sintering of S i c (see Table 6). Improved reliability in sliding bearing and rotating seal application have been achieved with microstructure-taylored solid-state sintered S i c materials, which contain pores [97-1001, a combination of pores and graphite [loll, and/or very large grained S i c crystallites [ 1021.

5.3.3.4.2 Liquid-phase Sintered S i c Liquid-phase sintered S i c (LPSSiC) is a very new, dual-phase silicon carbide ceramic, with typically M 95% S i c content and ultra-fine microstructure (grain size M 1 ym, see Fig. 12e). Y203-A1203 [103-1081 or Y2O3-A1N [109-1111 based compositions were found to be the most effective densifying aids for even standard grade S i c powders and a pressureless- or gaspressure sintered material can be produced with properties matching the best hotpressed silicon carbide (see Table 8). Compared with SSiC, LPSSiC exhibits an increase of up to 100% in fracture toughness to 6.0 MPam1I2 and an increase of 70% in flexural strength to 730MPa. Microcracking at the SIC-YAG interface ahead of a propagating crack is proposed as the dominant toughening mechanism. The major part of the secondary phase is YAG (yttrium aluminate, Y3Al5OI2) which crystallizes together with reprecipitated S i c on solidification from the liquid [112]. The fracture toughness of LPS S i c has been recently further improved by anisotropic grain growth, that is growth of larger, platelike S i c grains (in situ toughening) in a matrix of finer equiaxial S i c grains [113, 1141, see Fig. 12f. Although LPSSiC may not be suitable as an universal tribology material like SSiC, the initial assessment of properties suggests that it will be suitable for highstress engineering components and in many wear applications, where strength, reliability, and toughness are demanded. 5.3.3.5 Hot-pressed Silicon Carbide Dense parts made by hot-pressing are generally considered to have a very fine microstructure (see Fig. 12g) and the best mechanical properties. Completely pure silicon carbide does not exhibit any noticeable plastic behavior up to its decomposition temperature and can therefore be densified only under diamond synthesis conditions

0.49 0.45 1.20 1.29 0.76 0.74

-

-

-

1.03 0.38

-

0.44 0.65 1.34

1.o

-

83.5 40.9 85.9 83.2 63.7 48.8 76.1 68.5

0.92 0.93

-

-

-

-

-

-

100 73.2 92.6 100

1.12 0.22 0.21 0.25

-

-

0.86 -

5.6

13.3 3.9

-

-

1.1

-

1.3 14.3 2.1 1.2 5.3 7.0 2.4 3.2

-

-

.-

-

1.5 6.5 2.9

6.4 8.9

15.1 44.8 12.0 15.6 31.0 44.1 21.4 28.3

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

79.1 12.6

3c

26.8 7.4

-

92.9 93.5 97.6

1.2 74.6

4H

Sintered body Polytype analysis (weight-%) 6H 15R

(a) 1 bar argon, graphite-tube-resistance furnace, 15 min hold (b) FG . . . FINE GRAINED AGG . . . ABNORMAL grain growth Characterization of Starting Powders (Specific Surface Area: 15-20 m2 g-I): carbotherm. red. silica, 100 w/o 3 C, 20 p.p.m. Al-content. P a-black Milled Acheson SIC, 89.9 w/o 4 H - 2.5w/o 15 R 7.6w/o 6 H, 0.71 w/o Al a-green-I Milled Acheson - SIC, 100 w/o 6 H; 30 p.p.m. Al a-green-I1 Milled Acheson - Sic, 6.5 w/o 4 H 6.4 w/o 15 R - 87.9 w/o 6 H; 200 p.p.m. Al

12 13 14 15 16 17

-

2050 2200 2050 2200 2050 2200 2050 2200

1B,2C 1 B,2C 1 AIN, 2 C 1 AIN, 2 C 2.5 AlN, 2 C 2.5 AIN, 2 C 1 Al, 2 C 1 Al, 2C

a-green-I1 cc-green-I1 a-green-I1 cL-green-TI a-green-I1 a-green-I1 a-green-I1 a-green-I1

10

-

2050 2050 2050 2050

1.2 B, 2 C 0.2 B, 2 AlN, 2 C 0.2 B, 1 Al, 2 C 0.2 B, 2 C

a-green-I a-green-I cc-green-I a-green-I

6 7 8 9

11

2050 2050 2050

2 c 1 B, 2C 2 AIN, 2 C

a-black a-black a-black

3 4 5

2.5 AIN, 2 C

2050 2050

1B,2C

P P

1 2

ma

Sintering Chemical analysis temperature (weight-%) A1 B

Addition to SIC (weight-%)

T

SiCmodification (colour)

Sample no.

Powder

FG AGG FG FG FG FG FG FG

FG FG FG FG

FG FG FG

FG FG

Microstructureb

Table 6. Microstructure and Sic-polytype phase composition in sintered silicon carbide (B/C, AljC and B/Al/C doped SSiC materials).

6

6'

T

f?

QQ %

Q

e2L

tn

f

5

Q

n

0

ir! 3

b

s?

4


707

(35 kbar, 2300°C) to 100% of the theoretical density [115]. In axial hot-pressing with graphite tools it is, however, necessary to use small amounts of sintering aids: 0.53% boron, aluminum, aluminum oxide, beryllium oxide, yttrium oxide, or tungsten carbide are added to fine uSiC powder before molding [26, 116-1231. The molding temperature is 1900-2000°C, depending on the particle size of the powder and the amount of sintering additives used. The pressure that can be used is limited by the strength of the graphite, the maximum being 50 MPa. The conventional method requires large amounts of energy and mold material and can normally be used only for parts with simple, uncomplicated shapes. Owing to this limitation, precision parts can only be produced by machining with diamond tools [124] these, of course, are expensive. A more convenient method than conventional hot-pressing, also still more expensive, is hot-isostatic pressing (HIP). During the development of this technique [ 125-1271 it proved possible to be prepare high-purity S ic products with silicon carbide contents of more than 99.5% by hot isostatic pressing of S i c powder or SIC preforms in vacuum-sealed casings to a final density exactly equivalent to the theoretical density and with an uniform, fine-grained microstructure (see Fig. 12h). Owing to the higher isostatic pressure of 2 kbar (argon gas), compared with normal hot-pressing, no sintering aids need to be added; that is, the thin film of silica on the silicon carbide particles (which represents residual oxygen impurity of the S i c powder) is sufficient to achieve complete consolidation. The postdensification of SSiC or LPSSiC by hot isostatic pressing is less complicated because no gastight encapsulation is necessary. Not only is it possible to achieve more than 99% of the theoretical density, but the variation of density and strength can also be reduced. Pressureless sintering and hot isostatic pressing can now be performed in the same cycle (see Fig. 13). This new process, called sinter-HIP, offers economic benefits [ 128-1 301. The importance of hot-pressed silicon carbide has decreased considerably since the introduction of pressureless sintering, although it is currently still the most suitable method of obtaining the best mechanical properties for pure, monolithic SIC.

5.3.3.6 Chemical-Physical-Vapor-DepositedSic CVD-SIC is obtained by the chemical reaction of volatile silicon- and carboncontaining compounds in the presence of hydrogen and in the temperature range 1000-1800°C, e.g.

2 SIC + 3HC1.

CH3SiC13

Besides the formation of S i c powders, S ic fibers, and S i c thin or thick films, nowadays monolithic bodies in sections up to 25 mm thick can be prepared by this method. CVD-SIC is of high purity (99.999%), cubic p-modification, very fined grained, and shows the lowest of oxidation rates. Bulk cubic phase CVD-SIC is produced for applications such as electronic packaging, components for wafer handling, kiln furniture for diode manufacture, laser optics for high temperatures, and substrates for computer storage media [131, 1321. It can be polished to variations of less than 0.3 nm. Coatings of CVD-SIC were initially developed in the 1960s

708

5 SiEicon Carbide Based Hard Materials

I

Encapsul. - HIP

I

Shaping

1

Containerless-HIP

Figure 13. HIP processing routes (courtesy R. Oberacker, IKM-Karlsruhe/Germany, [ 1301).

for nuclear fuel particles employed to reduce the diffusive release of metallic fission products from the fuel kernel. Today Sic thin films (< 1 pm thickness) can also be produced by physical vapor deposition (PVD), for example by sputtering, which method allows lower substrate temperatures, but works more slowly. Electrically conductive B/N-doped sintered clSiC with up to 9weight-% free carbon has been developed as target material, [133]. Novel applications for PVDSiC include films for computer storage media, protective coating for lenses, and microwaveable packaging for food. 5.3.3.7 Sic Wafers Since 1990 high quality S i c wafers of 35 mm diameter are commercially available from single crystal 6H-Sic bodes, produced via a seeded-sublimation growth technique [ 1341. In this process, nucleation occurs on a Sic seed crystal located at the top or bottom of a cylindrical growth cavity. As in the Lely process [135], S i c sublimes from a polycrystalline source at temperatures > 2200°C under vacuum to form Si, Si2C, and Sic2 vapors


709

The gases diffuse through a porous graphite retainer and along carefully programmed thermal and pressure gradients. The primary gaseous species of silicon reacts with the graphite walls of the growth cell to from additional Si2C and Sic2 which recombine on the growing crystal according to a ‘double condensation’ reaction 2SiC2

+ 2Si2C =

= 6H Sic.

Hence the silicon plays the role of a carbon-transporting agent. Water-clear boules of the pure 6H polytype having diameters > 50 mm and lengths > 60 mm have been grown by this technique. It is believed that by the year 2001 the size of the boules will have increased to 15 cm in diameter. Based on the growth of epitaxial thin films of single crystal polytypes on boulegrown substrates, S i c is now becoming the material of choice for high-power, high temperature (> 5OO0C), and high-frequency devices [ 1341. As a spin-off, currently colorless S i c gemstones, 1/2 to 1 carat in size, cut from 6H-Sic wafers are entering the jewellery market at about 10-1 5% of the price of diamonds.

5.3.3.8 S i c Nanoceramics Sintering nanosized powders with the aim of flaw avoidence from nanosized sintered bodies (grain size 5 100 nm) is recognized as a promising way of improving mechanical properties and reliability of S i c ceramics. However, nanosized S i c powders are not easy to process [42,45,136], and several difficulties must be overcome: the powder flows badly, exhibits low oxidation resistance [40], has low filling and compaction density [40] and is currently too expensive for large scale use.

5.3.3.8.1 Fabrication by Solid State Sintering In 1991 Vassen et al. [ 1371showed that polycrystalline PSiC bodies with a density of at least 95% of the theoretical density and a fine grain size of 150nm can be prepared by encapsulated HIPing of BjCdoped laser-synthesized powders with particle sizes below 20 nm. Necessary HIPtemperatures were 1500°C, which is 250°C below the temperature needed to densify conventional submicron powders. This fabrication process was later optimized [138] by preheating the shaped Sic bodies in vacuum during an annealing step before encapsulation. Owing to this additional annealing step, residual oxygen contents were minimized and a mean grain size of only 60nm could be obtained in the HIPed Sic shapes. It was found that the reduced final grain size has a strong effect on mechanical properties; as a consequence of grain size reduction from 1 pm to 150nm the Vickers hardness increased from 2000 HVlO to 2500 HV10, whereas fracture toughness decreased from 4 to 3 MPamI”, respectively [139]. However, by adoption of a bimodal grain-size distribution, with introduction of larger sized S ic grains into a nano-sized matrix, an incrase in fracture toughness to 6 MPa m1I2was achieved [140,141]. For a fine grained HIPSiC (300nm) even under a stress of s-’ were measured [142]. 100 MPa at 1600°C very moderate creep rates of 1 x 5.3.3.8.2 Fabrication by Liquid Phase Sintering The preparation of PSiC nanoceramics with an average grain size of 110 nm by liquid phase sintering has been

710

5 Silicon Carbide Based Hard Matevials

demonstrated by Mitomo et al. [143], who were subsequently awarded a patent called ‘Superplastic S i c Sintered Body’ [ 1441. Ultrafine PSiC powder with an average particle size of 90nm was axially hotpressed with additions of A1203,Y203,and CaO at 1750°C. The S i c nanoceramic showed large deformation with high strain rate: 5.0 x lop4sC’ at 1700°C. On the basis of their results [ 1431, the maximum temperature and the minimum deformation rate for nano-SSiC might be defined as 1800°C and 10-4sC’, respectively. The superplastic deformation at temperatures as low as 1700°C is based on the fine-grained microstructure and the presence of a glassy phase at grain boundaries. Thus a new technology has been developed whereby nano-Sic parts can be subjected to plastic deformation as in the case of metals, and can be made into complicated shapes with near net-shape quality, that is without the need for an expensive postprocessing stage such as diamond machining.

5.3.3.9 Sic-based Composites Alloying of S i c is and has been done basically for two reasons: either to improve properties (toughness, wear, etc.) by the formation of tailored composites/solid solutions or to improve processing. Improvements in processing can occur: (1) in solid state sintering, where the second phase acts simultanously as a sintering aid for Sic, accelerating material transport by grain boundary and/or lattice diffusion; (2) in reactive liquid sintering due to reduced sintering temperatures (‘transient liquid phase sintering’). In the latter case S i c and/or additions are reacted to an intermediate liquid which not only provides densification at reduced temperatures but since it is consumed in the reaction yields a Sic-based material without glass at the grain boundaries. Fabrication of S i c composites by second-phase dispersion is widely applied to improve material toughness. The various toughening mechanisms [93] that have high potential to reduce crack extension in Sic- composite materials are: crack deflection, microcrack formation, crack bridging by reinforcement with metallic ligaments (e.g. TIC, TiB,), and crack bridging and pull out by platelet- or fiberreinforcement . Tensile fracture in Sic-based composites will only occur after a large enough load is applied to exceed the compressive stress in the process zone formed by cracking mechanisms along the crack path (see Fig. 14). To achieve increased crack deflection and crack-wake interaction in S i c the microstructure can be modified in various ways: ~

~

-

by addition of a second phase with an elongated or fibrous grain structure (SiCbased composites), by reinforcement with S i c platelets [ 521 or continuous Sic-fibers, by inducing growth of elongated S i c grains (in situ toughening of LPS-Sic [ l l l , 113, 1141).

For optimum toughening by crack deflection Telle et al. [145] pointed out that geometric factors like grain size, volume fraction, orientation and morphology of the added or in situ grown phases as well as the grain boundary strength have to

5.3 Production of S i c

71 1

Figure 14. Potential toughening mechanisms for S i c based ceramics: 1 crack deflection, 2 microcrack formation, 3 crack bridging with metallic ligaments, e.g. TiB2, 4 crack bridging and pullout by platelet or fiber - reinforcement.

be considered (see Fig. 15). In the following sections today’s state of the art in some more or less important Sic-nonoxide composites is reviewed. 5.3.3.9.1 Sic-Tic

Very promising composites have been developed in the SIC-TIC system with S i c as the matrix phase [146-1481. Dispersed TIC particles significantly Grain Boundary Strength

Dispersion, Volume Fraction

Orientation

-

worse-

-better-

Figure 15. Microstructural design for optimized crack deflection in composites (Courtesy R. Telle, RWTH Aachen/Germany [ 1451.

712


Figure 16. SEM micrograph showing crack deflection and crack bridging in Sic-Tic and SiC-TiBz composites: light areas = TIC or TiB2, dark matrix = SIC (Courtesy D.Ly Ngoc, MPI Stuttgart/Germany [149].

improve both the strength and the toughness. Although an addition of T i c does not reduce the densification temperature significantly below 2 1OO'C, the coarsening of S i c is completely retarded which raises the strength to 700-800 MPa [146-1491. The increase in KIc to 6.5-7.5 MPam'I2 is attributed to the misfit of the thermal expansion coefficients of T i c and Sic, introducing considerable radial tensile stresses at the phase boundaries and hoop compressive stresses in the matrix. These stresses enable crack deflection (see Fig. 16), crack branching, and microcracking above a critical particle size of = 3 pm. The optimum volume content of T i c ranges between 20 and 30 vol-%.

5.3.3.9.2 SiC-TiB2 Sic-based composites with transition metal diboride (TiB2, ZrBz, etc.) particulates have been developed for electroconductive applications such as heating elements and ignition 1150,15 I], and also as wear resistant structural parts for high temperatures such as valve-train components and rocker arm pads in super-hot running engines [ 1471. These composites combine the high thermal and electrical conductivity of TiBz and ZrB2 with the oxidation resistance of Sic. Additionally, due to thermal mismatch stresses of the order of 2 GPa toughening mechanisms such as crack deflection and stress-induced microcracking with a pronounced process zone as well as flank friction have been proven to occur. Cai et al. [ 1521 and Faber et al. [ 1531 have presented a detailed analysis of the contributions of

5.3 Production

of'

Sic

7 13

the particular mechanisms to the total fracture toughness, stating that the stress induced microcracking is operational in a process zone of approximately 150 pm width. Typical conditions for densification by axial hot-pressing are 2000-21 OOOC, at a pressure of 20-60MPa for 30-60 min which results in 96-99.8% density. The particle sizes of the matrix and dispersed phases range between 1-5 and 4--8pm, respectively. An optimum volume fraction of reinforcing particles of 25-30 vol-Yo has been reported, yielding a flexural strength of 710 MPa and a fracture toughness of 5.0-5.7 MPam'/2 [149]. Composites with a lower TiB2 content of 15 vol-Yo exhibit a mean strength of 485 MPa combined with a KIc of 4.5 MPa [147]. The strength of SIC based materials with 50 vol-YOZrB2, HfB2,NbB2 or TaB2 particles also ranges between 400 and 500 MPa [ 1501. Similar strength values (480 MPa) combined with a exceptionally higher fracture toughness of 7-9 MPa m1/2have been reported for large scale lots of pressureless sintered 16 vol-YO TiB2 composites [ 1513. Since the sintering was carried out with temperatures exceeding 2000°C yielding 98-99% of the theoretical density and an average particle size of 2.0 pm, it is obvious that the reinforcing phase also acts as a grain growth inhibitor for Sic. The high temperature strength of SiC-TiB2 and SiC-ZrB2 composites was found to remain nearly constant at 480MPa up to 1200"C, and is hence superior to that of many sialons [150,151]. Tani and Wada [ 1541fabricated optimized SiC-TiB2 composites by reactive sintering starting from an intimate mixture of Sic, Ti02, B4C, and C powders. The mixture was either hot pressed or pressureless sintered and post-HIPed at temperatures of > 1900°C. Titanium diboride was formed in situ according to the reaction Ti0 2

+ 0.5B4C+ I .5C

t

TiB2

+ 2C

during an intermediate heating step at 140CL15OO"Cin vacuum or argon atmosphere. Overstoichiometric amounts of B4C and C (1-2 weight-% each) can be adjusted to aid sintering. The primary advantages claimed for this reaction sintering process are the use of water in powder processing due to the disuse of highly reactive, preformed TiB2 powders, and the very small size of reinforcing TiB2 particles formed in situ due to the use of ultrafine Ti02, B4C, and C starting powders. Effective reactive pressureless sintering of SiC-TiB2 composites was recently reported by Blanc, Thevenot, and Treheux [155]. In addition they studied the tribological behavior using a pin on flat configuration (flat: Sic, pin: SiC-TiB2). In dry conditions the composites showed less wear resistance than monolithic Sic, however, with water as lubricant the opposite was the case. In the very recent study of Kuo and Kriven [156]indentation-strength tests were used to determine the retained strength, flaw tolerance, and toughness-curve characteristics of two kinds of SiC-TiB2 composites. BSiC-TiB2 composites which were hot-pressed with an A1203sintering aid, were compared with the well-studied aSiC-TiB2 composites, which were pressureless sintered with boron and carbon additives. TiB2 (I 5 volYo)in the B- and C-doped aSiC only increased the retained strength without a significant improvement in the toughening. On the other hand, TiB2 (3ovol-%) along with the effect of AI2O3 sintering aid for the BSiC-TiB2 composite greatly improved properties with a higher retained strength in long crack regions, better flaw-tolerance behavior and a sharply rising toughness vs. crack size curve. The

7 14

5 Silicon Carbide Based Hard Materials S/HIPS

WHIPS

S/HIPS

S/HIPS

80

t 76.8

58.6

19.2 3.7

37.0 42

39.0 560 4,7

20.0 74,5 5,5

'/. c c a./

SIC

6 '

=/e

Figure 17. Flexural strength of pressureless sintered (= S, unhatched bars) and post-HIPed (=HIPS, hatched bars) SiC-B4C particulate composites [ 1571.

different toughening behaviors for rx and pSiC-TiB2 were related to the weak nature of the Sic-Sic and SiC-TiB2 interfaces as well as the fraction and size of TiB2. 5.3.3.9.3 Sic-B4C In the mid 1980's pressureless sintering and posthipping were developed by Schwetz et al. [157] to produce 100% dense Sic-B4C composite materials having SiC:B4C weight ratios within the range of from 90: 10 to 10:90 and a free carbon content of 4-5 weight-%. These composites combine the good thermal shock resistance and oxidation resistance of silicon carbide, with the hardness, wear resistance, and low specific gravity of boron carbide. In this way a maximum strength of 550 MPa (four point bend) was achieved for a composite of 59 weight-% Sic-37 weight-% B 4 C 4weight-% C (see Fig. 17). The composite can be used in oxidizing atmospheres up to 1200 "C. Its microstructure is characterized by equiaxial B4C and graphite grains of < 1 pm diameter, which were embedded in a matrix of Sic grains with an average grain size of 1.5 pm (see Fig. 18). However, no improvement in fracture toughness was achieved, since the fracture mode was almost 100% transgranular. Similar results on sintered Sic-B4C


7 15

composites were obtained by Thevenot [158] and later by Tomlinson et al. [l 591, who observed a 20% increase in strength when 25 vol-% B4C was added to Sic. Excellent tribological properties for SiC-B4C-C composite materials were recently encountered by Kevorkiijan et al. [ 1601. In this study SiC-B4C-C seal rings for magnetic pumps were prepared by pressureless sintering and characterized by a pin-on-disc method (medium: water, pressure: 16-25 MPa, speed: 35-75 m sC’). The introduction of a lower level of B4C (5-20 weight-%) particles into the S i c matrix resulted in an almost linear decrease of wear rate. Moreover, further addition of B4C ( 2 0 4 0 weight-%) led to an almost parabolic wear rate response. For example, with 40weight-% B4C a decrease of the relative wear volume of seal rings by more than 55% was achieved. These results recommend use of wear resistant SiC-B4C composites for heavily loaded mechanical face seals in the pairing hard/hard against themselves. They may be likewise suitable for the production of shaft protection sleeves and components for sliding bearings whose wear resistance is to be improved. 5.3.3.9.4 SiC-AlN A series of solid solutions between S ic and A1N over the whole composition range was concurrently discovered at the Universities of Utah and Newcastle upon Tyne [ 1611, and has since received considerable attention [I 62-1 651. A 2H wurtzite-type structure is formed by the carbothermal reduction of fine Si02 and A1203with a carbon source under nitrogen atmosphere at 1600°C:

3 sio 2

+ 0.5AI2O3+ 4.5C + N2 1600°C

___i.

3SiC.AlN + 1.5CO

+ 0.5N2.

Kinetically favored is the carbothermal reduction of ct’ SiAlONs or a’ SiAlON precursor mixtures (3Si3N4 3A1N CaO):

+

+

CaSi9A130N15 1OC

+

1800

c

3(3SiC.A1N) + Ca + CO

+ 6N2.

Because the diffusion coefficients in covalent solids are extremely small, solid solution was thought unlikely to be obtained by heating and annealing of the powdered solid components.

716

5 Silicon Carbide Based Hard Materials wt. ‘Yo AIN

Figure 19. Tentative SIC-A1N phase diagram after Zangvil and Ruh [166].

Zangvil and Ruh [166] however, obtained SIC-AIN solid solutions by hot pressing powder mixtures (1 -x)SiC+xAlN+

(SiC),-x(AIN)x

According to the phase diagram proposed by Zangvil and Ruh (see Fig. 19) at temperatures above 2O0O0C, a 6H-4H-2H series of solid solutions appears with increasing amounts of AlN. AlN strongly stabilizes specific polytypes (4H and 2H) at certain composition ranges, enabling the engineering of single-phase Sic materials with discrete physical properties. At temperatures below about 19OO0C, a miscibility gap was first proposed by Rafaniello et al. [163], a suggestion supported by several later studies. Xu et al. [165] obtained strong SIC-AlN materials with flexure strength up to 1 GPa. Several mechanisms of grain refinement resulting from Sic-polytype transformations into a wurtzite (2H) solid solution, were reported. Kuo et al. [167] discovered the formation of modulated structures within the miscibility gap and Lee and Wei [168] reported that pressureless sintering with 2weight-% Y 2 0 3 as sintering aid at 2050°C produced a duplex structure composed of large (Sic), (AIN), grains and small Sic grains. Further solid solution treatment (> 2225°C) followed by annealing within the miscibility gap (1860°C) resulted in spinoidal decomposition, giving various duplex/modulated structures with improved fractured toughness of the alloys. In addition to their possible use as high temperature structural ceramics, materials in the system SIC-AIN have potential as wide band-gap semiconductors and for opto-electronic applications [ 1661. 5.3.3.9.5 SiC-A120C Extensive SiC-Al20C solid solutions have been found by Cutler et al. [161] for 1-100% A4OC. Moreover, they showed that wurtzite 2HSiC can incorporate substantial amounts of AIN and A120C in solid solution and coined the acronym ‘SiCAION to describe these materials by analogy with ‘SiAlON’ceramics. Jackson et al. [169] have sintered Sic at temperatures between 1850°C and 1950°C using a transient liquid phase produced by the carbothermal reduction of A1203by

5.3Production

Qf

Sic

717

A14C3.The resulting ceramic was fine grained (average grain size less than 5 pm) and consisted of S i c (starting polytypes) and A120C as the two major phases. The properties of the hot pressed ceramics varied with the amount of A120C, but at an optimum composition of about 5-10 wei ht Y ' A120C, the strength (660 MPa) - O. and fracture toughness (KIc = 3.1 MPa m I$) obtained were comparable or superior to the corresponding properties of commercial grades of sintered Sic. Huang et al. [170] found encapsulation to be necessary for effective sintering with additions of A1203 and A14C3, the densification occurring above M 1860°C. They attributed densification to a transient liquid phase in the system A1203-A1404C in the route to forming A120C with an eutectic temperature of ~ 1 8 4 0 ° C Four . point bend strength, hardness and fracture toughness for the SiCAlON materials have been reported [170]. The strength decreased with A4OC content in hot-pressed samples, from ~ 6 0 MPa 0 at 10 weight-% A120C to around 250 MPa at M 50% A120C. Most significantly, the fracture toughness of some SiCAlON com ositions appeared to be higher than that of S i c ( ~ 4 . compared 2 with 3.0MPam I f : ), using an indentation technique. Lihrmann and Tirlocq [ 17I] proposed fabrication of sintered or hotpressed S i c based composites containing S i c as well as 5-30weight-YO solid solutions composed of A120C and AIN. Starting from SiC-A1N-Al4C3-Al2O3 powder mixtures, densification was greatly enhanced by occurrence of a transient liquid phase originating in the A1203-A14C3 system at temperatures above 1800°C. The composites with 5-10% (Al2OC-A1N) solid solution exhibit a mean grain size < 2 pm and a mean strength of 620-670 MPa combined with a K,, of 5.1-6.8 MPam'/2. Strength and fracture toughness both retained their values up to 14OOcC, before weakening, thus demonstrating the highly refractory nature of the A120C-A1N second phase. 5.3.3.9.6 S i C S i C and SiC-C (Continuous Fiber Reinforced S i c Matrix Composites) Fabrication by Chemical Vapor Infiltration In chemical vapor infiltration (CVI) silicon carbide is vapor deposited inside a porous preform (4&60% porosity) made of layers of woven cloth, from high strength C- or S i c fibers (CVI-SiCfib,,/ Sicmatrix or CVI-Cfber/SiCmatrix). Isothermal CVI infiltration of the fibers fills the porosity with pure S i c and leads to a composite with up to 90% of the theoretical density (TD), which fractures in a noncatastrophic mode at a typical flexural stress of 300400MPa and a toughness of over 20MPam'I2. Using the forced flow thermal-gradient CVI process developed at Oak Ridge National Laboratory, USA, the infiltration time is reduced from weeks (isothermal CVI at SocietC Europeenne de Propulsion/France) to less than 24 h (ORNL) and final densities for composites of >go% T D 11721 are reached. Material data of CVI-SiCjSiC and CVI-C/SiC composites available from MAN - Technologie, Germany, are listed in Table 7 [173]. Figure 20 shows the fracture surface and fiber pullout of a CVI-SiCjSiC composite. Prior to the matrix infiltration one ore more layers of pyrolytic carbon or boron nitride are usually applied to the fibers to provide a means of fiber debonding and toughening by pull-out and crack bridging.

718


Table 7. Mechanical and physical data for continuous fiber reinforced S i c matrix composites (MAN-Technology, Germany [ 1731. Property

Units

Gradient CVI SiCjSiC

Gradient CVI CjSiC

LPI CjSiC

Fiber fraction Density Porosity Tensile strength Strain E-modulus Flexual strength Compressive strength ILSa CTE~

Vol% g cm-3 YO MPa

4247 2.3-2.5 10-15 280-340 0.5-0.7 190-210 450-550 600-650 45-18 4 4

4247 2.1-2.2 10-15 300-320 0.6-0.9 90-100 450-500 450-570 4448 3 5

4244 1.7-1.8 15--20 240-270 0.8-1.1 60-80 330-370 43W50 35 3 4

%

GPa MPa MPa MPa

K-'

11 I

(a) ILS = Interlaminar strength. (b) CTE = Coefficient thermal expansion.

At high temperatures (> 600°C) the composites degrade in strength and toughness due to oxidation of C fibers and/or these interface layers and prevention of fiber pullout (brittle fracture mode). Studies are in progress to increase oxidation resistance by use of a CVD-SIC overlayer which seals the surface of the porous composites. Fabrication by Liquid Polymer Injiltration ( L P I ) In the first step of the LPI process, a carbon-fiber preform is infiltrated with resin (e.g. polycarbosilane), to bind the fibers together. Then the polymer is pyrolized to form Sic. These process steps are repeated a number of times until the pores are narrow enough that further

Figure 20. Fracture surface of a CVI-SiCjSiC composite showing fiber pullout (Courtesy M. Leuchs, MAN TechnologyjGermany).

5.4 Properties of Silicon Carbide

7 19

infiltration ceases [174]. Finally the body is heated to temperatures between 1000 and 1500°C for crystallization of the S i c matrix (LPI-C/SiC). Fabrication by Liquid Silicon Infiltration (reaction bonding) ( L S I ) A leading candidate for use in industrial gas turbine engine is a Sic matrix composite named ‘toughened Silcomp’ [175]. It is produced by melt infiltration of molten silicon into a porous preform containing carbon as well as BN-coated S i c fibers (e.g. Textron SCS - 6). The composites thus produced consist of a fully dense matrix of SIC Si, reinforced with continuous SIC fibers. Moreover, the melt infiltration process is net shape and fast. Ultimate strength and strain at ultimate strength are 220 MPa and 0.8%, respectively at room temperature (LSI-SiCjSiC-Si).

+

5.4 Properties of Silicon Carbide 5.4.1 Physical Properties 5.4.1.1 Color Pure aSiC is colorless while the cubic [, modification is yellow. The only other elements that can be included in the S i c crystal lattice in amounts > 1p.p.m. are N, Al, and B. Nitrogen gives a green color to 3C and 6H, and a yellow color to 4H and 15R. The presence of the trivalent elements boron and aluminum gives all the modifications and polytypes a blue-black color [ 1761. 5.4.1.2 Optical properties OrSiC is birefringent due to its crystal structure: no = 2.648-2.649, and n E = 2.6883 671 nm) has 2.893 (Na 589 nm, 20°C) [177]. For PSiC, a refractive index of ~ 2 . 6 (Li been reported [7]. 5.4.1.3 Electrical properties Silicon carbide is a semiconductor. The most important electronic properties of Sic are its wide energy band gap of 3.26 eV for 4H-Sic and 3.03 eV for 6HSiC, high breakdown electric field of 2.2 x lo6Vcm-’ for l0OV operation, and high saturated electron drift velocity of 2 x 10’ cm s-’. Doping with the trivalent elements aluminum and boron gives the Sic p character, whde the pentavalent element nitrogen produces y1 character when incorporated into the Sic crystal lattice. The resistivity can be varied between 0.1 R cm and lo’* Ocm, depending on the concentration of the dopant [178]. Whereas compact, homogeneous SIC obeys Ohm’s Law, aggregates of S i c grains show nonlinear current-voltage behavior. At low applied voltages they behave as insulators, but when the applied voltage is increased above a certain value the current increases exponentially. Thus, the points of contact between the grains cause the electrical resistance to be voltage dependent [178,179].

720


5.4.1.4 Thermal and Calorific Properties For a ceramic material, silicon carbide has an unusually high thermal conductivity: 150W mK-' at 20°C and 54 W mK-' at 1400°C [ 1SO]. The high thermal conductivity K-' for 2&1400"C) explain why the material and low thermal expansion (4.7 x has such good resistance to thermal shock. The s ecific heat capacity of Sic is 0.67 J g-' K-' at room temperature, and 1.27Jg-'K1 at 1000°C. The standard enthalpy of formation AH&,, is -71.6 f 6.3 kJmol-', and the entropy 5$98K is 16.50 f 0.13 Jmol-' K-' [181]. 5.4.1.5 Mechanical Properties Silicon carbide is noted for its extreme hardness [ 182-1 841, its high abrasive power, high modulus of elasticity (450 GPa), high temperature resistance up to above 15OO"C, as well as high resistance to abrasion. The industrial importance of silicon carbide is mainly due to its extreme hardness of 9.5-9.75 on the Mohs scale. Only diamond, cubic boron nitride, and boron carbide are harder. The Knoop microhardness number HK-0.1, that is the hardness measured with a load of 0.1 kp (=O0.98N),is ~ 2 6 0 0(2000 for aA1203, 3000 for B4C, 4700 for cubic BN, and 7000-8000 for diamond). Silicon carbide is very brittle, and can therefore be crushed comparatively easily in spite of its great hardness. Table 8 summarizes some typical physical properties of the Sic ceramics. Since the microstructural grain size (Fig. 12a-h), pore content, and chemical composition of the various ceramic products differ considerably, it follows that the properties are also different. Recrystallized Sic is much stronger than the ceramically bonded material, but its high residual porosity imposes limits as far as mechanical strength is concerned [185]. Reaction-sintered SiSiC is still stronger, but only up to 1400°C (Fig. 21), the softening point of the accompanying silicon phase [ 1861. The best mechanical strength is exhibited by sintered S i c and hot-pressed materials [109,187,188]. Solid-state sintered, hot-pressed, and isostatically hot-pressed materials offer considerable advantages over all other ceramic materials in plastic deformation under a sustained load (creep), because of the low content or almost complete absence [189,190] of sintering aids. Compared to solid state sintered silicon carbide (SSiC) and dense alumina (A1203), liquid phase sintered Sic (LPS-Sic) features improved edge toughness (see Fig. 22) close to the edge flaking resistance of sintered silicon nitride (Si3N4). The latter property indicates the sensitivity of edges against mechanical chipping, a quantity which is most important for safer handling of ceramics in grinding, clamping, transportation etc.

5.4.2 Chemical Properties One of the outstanding characteristics of silicon carbide is its chemical resistance, which is due to the high affinity of silicon for oxygen. The reaction of silicon with

5.0 4.3

250 250 400 410

450 430

20 20

<1 <2

0 0

2.60 2.60 3.12 3.15

3.21 3.20

3.21 3.19

100

90

98

95

98

>99.5

98

450

420

4.8

4.5

4.5

4.5

4.9

5.0

5.8

I00

100

20

2.55

up to 95

Ceramic bonded S i c KSiC Recrystallized S i c RSiC Reaction bonded S i c RBSiC Infiltrated Sic SiSiC Sintered S i c (solid state) SSiC Sintered S i c (liquid phase) LPSSiC Hot-pressed S i c HPSiC Hot-isostatic-pressed Sic HIPSiC HIP-post densified SSiC HIPSSiC

Porosity

Thermal expansion 30-1500°C K-I)

(YO)

Young modulus (GPa)

Density (g cm-')

Sic content (weight-%)

Sic material type

Table 8. Physical properties of various S i c ceramics.

50

75

55

50

50

60

25

28

16

Thermal conductivity at 600°C (W mK-')

450

640

640

730

430

350

250

100

30

20°C (MPa)

450

610

650

400

450

200

250

100

20

1400°C (MPa)

Flexural strength

122

I


900

1

SIC-CERAMICS

800 -

600

-

solid st sintered

500 --------------..--

400

_/--

reaction banded (with free Si)

-

1 , \ ,

I

recrystallized

---------1---------' 100 -

0

0

200

ceramic-bonded ~

~

400

L

600

- -

__

-

800 1000 1200 1400 1600 TEMPERATURE "73

Figure 21. Flexural strength of S i c ceramics as a function of temperature.

1 Load P

Distance d

Load P

b Edge Flake

0

0.1

0.2

0.3

0.4

Distance from Edge d, mm

0.5

Figure 22. Edge toughness of LPS-Sic (EKasic T) compared to sintered Al,03 and Si3N4: (a) flaking load against distance from edge, (b) schematic of experimental setup (Courtesy L.S.Sig1, ESK-Kempten/Germany)

5.4 Properties

of Silicon Carbide

723

oxygen in an aqueous medium causes passivity and, if exposed to thermal oxidation, for example, in air, leads to the formation of glassy silica films [191-1931. The oxidation of pure S i c begins at around 600°C, forming a coating of Si02 on the surface of the SIC that prevents further oxidation [ 1941: Sic

+ 2 0 2 + Si02 + C02.

The reaction rate varies with time according to a parabolic law [191]. The kinetics are determined by the diffusion of oxygen through the SiOz layer. The temperature dependence of oxidation follows the Arrhenius equation. The ‘active’ oxidation of SIC is distinguished from the ‘passive’ oxidation reaction described above: Sic

+ O2 + SiO + CO.

Active oxidation takes place under conditions of oxygen deficiency above 1000°C and leads to decomposition of the SIC and formation of silicon monoxide [194, 1921. The two forms of corrosion (active, passive) and the conditions for the boundary have been recently discussed by Nickel et al. [I951 using examples from Sic- and also Si3N4-based ceramics. A likely high temperature boundary for SIC is z 1700-1 8OO0C,where a secondary active-to-passive transition by bubble formation and spalling occurs. Thus silicon carbide is attacked and decomposed by oxidizing agents (e.g., Na2C03 Na202or Na2C03 KN03) if the protective layer of Si02 is removed, thereby enabling the reaction to proceed unhindered. Pure silicon carbide is twice as oxidation resistant, even at 1500”C, than the best current superalloys at their maximum service temperature of 1200°C. Silicon carbide is resistant to most chemicals, resisting acids and alkalis and even aqua regia and fuming nitric acid. A mixture of hydrofluoric adic, nitric acid, and sulfuric acid slowly attacks silicon carbide PSiC being somewhat more reactive than &Sic[196]. However, complete dissolution only takes place if the SIC is very finely divided, under pressure, and at a elevated temperature (e.g., 250°C for 16h). Alkali melts [ 197, 1981will attack it in the presence of oxidizing agents. Oxides, molten metals, and water vapor have a destructive effect on Sic at temperatures of 1000°C and over. Chlorine reacts exothermally with silicon carbide above 800°C with the formation of silicon tetrachloride (Sic&) and carbon. Silicon carbide behaves in various ways towards molten metals. It is not attacked by molten zinc or zinc vapor [199]. Molten aluminum attacks SIC slowly, forming Al,C, and silicon, but as the silicon content increases, the reaction eventually ceases because an equilibrium is established [200]. Molten iron desolves Sic, forming iron carbide and iron silicide.

+

+

5.4.3 Tribological Properties 5.4.3.1 Review on Tribology Work in Sintered &Sic Ceramics

Tribology is the study of adhesion, friction, wear and lubricated behavior of materials in solid state contact. It was in the early 1980s that the advantage of

724


sintered S i c as seal face material was determined [201,202]. Since then, the mechanical engineering industry has focused attention on this material and tribological studies have been simulated to understand better those physical and chemical properties of SSiC that will affect its behavior when in contact with itself, other ceramics, or metals. Model investigations involving Sic have shown the coefficients of frictions against various materials, even itself, to be a function of contact stress [203] and have documented the anisotropic wear behavior of monocrystalline S i c crystals [204]. Lashway et al. [205] found that a controlled amount of porosity improved the ability of sinteredSiC to retain a hydrodynamic film with lower friction. Seal tests also indicated lower power dissipation at varying pressure-velocity values for combinations with sinteredSiC seal face materials. Consistently positive practical experience gathered by Knoch et al. on sinteredSic seal rings [206,207] has helped the material to gain popularity rapidly for use in situations involving wear problems. Excellent results have been achieved in developing sliding bearings for hermetically sealed pumps [208]. All of them are absolutely leak-proof, whether in operation or shut down and they are therefore of great value environmentally. They all have the same design requirement that the sliding bearing of the pump shaft must be flushed and lubricated by the pumped medium. Traditional materials simply could not cope with the harsh conditions of the chemical industry and were quickly destroyed by corrosion and abrasion. Sintered silicon carbide solved the problems. Liquids containing abrasive particles do not restrict the use of sintered Sic. Many sliding-wear problems occurring in the field are attributable to interruption of the ideal (that is properly lubricated) running conditions, in which case the sliding faces of the bearing or seal in question make contact with each other, giving rise to solid-state or dry friction marked by a pronounced increase in the coefficients of friction. Local frictional heat leads to peak thermal stresses that may be of such intensity as to cause a breakout of microstructural constituents. Then, when lubrication (and cooling) is restored, the material is in danger of cracking or fracturing due to thermal shock. Sintered silicon carbide, however, is better able to cope with such situations than other ceramic materials, because it is stronger and has a lower thermal expansion coefficient, and a higher thermal conductivity. Consequently, brief periods of dry running can be survived. The nonlubricated wear behavior of sintered silicon carbide under unidirectional sliding at room temperature was studied by Derby et al. [209] as a function of load and sliding time. At low loads polishing and ploughing mechanisms were observed and the microstructure revealed an etched appearance. With increased loads and sliding times microcracking at the grain boundaries occurred leading to subsequent grain pull-out. In the study of Cranmer [210] it was shown that for sintered silicon carbide sliding against itself also surface plastic deformation, ploughing and cracking were operative as wear mechanisms. Miyoshi et al. [211] carefully studied friction and wear of the pair SiC/iron in vacuum environment at temperatures up to 1500°C. By using X-ray photoelectron spectroscopy it was found that the surface chemistry of sintered silicon carbide (graphite and Si02) as well as the alteration of this chemistry by the temperature are very influential in dramatically changing the friction coefficient. Breznak et al. [212] quantified the role of initial surface roughness on the


725

friction behavior of sintered silicon carbide rubbing against itself in cyclic oscillatory motion. It was shown that the coefficient of friction decreased from an initial value of 0.40 to 0.25 as a result of surface polishing. Further work by Breval et al. [213] indicated that adding graphite to Sic reduces the wear of SiCjSiC couples somewhat, but improving their initial surface finish has the opposite effect: it leads to a greater loss of material during running in. Wear debris of SiCjSiC couples exhibited a bimodal particle size distribution: some particles were micrometer size and the others ranged in nanosize from 5 to 50nm. Using pin on disk, as well as abrasion wheel test, Wu et af. [214] observed that the amount of wear increases with increasing grain size for sintered silicon carbide. Smythe and Richerson [215] conducted experiments in which the dynamic sliding contact behavior at temperatures greater than 1000°Cwas studied. It was found that surface film formation at higher temperatures governs the friction behavior. It is believed that Si02 or Si02 modified by impurities from the SIC or the environment with flow properties contributed to the lowering of the friction coefficient [216]. Tomizawa et al. [217] performed pin-on-disk experiments on friction of SSiC against itself in water at room temperature. The authors found a friction coefficient of 0.26 and noted that wear of sintered silicon carbide occurs by a combination of tribochemical dissolution and the formation of pits by fracture of Sic grains. The amount of material removal varied from one Sic grain to the other, due to a strong dependence of tribochemical wear on crystallographic orientation of Sic grains. Knoch and Kracker [2 181 observed that this anisotropic tribological behavior is most conspicuous with EKasic D sintered silicon carbide, which has a bimodal grain structure: about 3ovol-% of larger platelets of hexagonal a grains (about l00pm long) and about 70vol-Y0 of smaller a grains (about 10pm long) as shown in Fig. 23a. This material shows superior performance particularly if paired against a softer carbon material. Figure 23b shows the surface topography developed in a finely lapped surface under in-service conditions, as water lubricated EKasicj carbon seal. This relief structure, with the depressions in a textured surface, obviously acts as reservoirs for lubricant, thereby improving the emergency running properties of the components. Figure 23c shows a sketch of the process. Boch et al. [219] studied the dry friction behavior of SiCjSiC couples (fixed ball rotating disk) at temperatures from 20 to 1000°C and observed a rapid decrease in the coefficient of friction from 0.45 to 0.16 when temperatures increases over 400°C. From 400 to 700°C, the debris agglomerates in the form of SiOz rolls, which arrange themselves perpendicularly to the sliding direction. These rolls act as minute roller-bearings decreasing the coefficient of friction and preventing the formation of cracks in the wear track. Habig and Woydt [220] studied unlubricated friction and wear of selfmated SSiC sliding couples (EKasic D) at temperatures between 22 and 1000°C using a stationary pin and a rotating disk assembly. As shown in Fig. 24, at room temperature the coefficient of friction decreases with increasing sliding velocity fromf = 0.8 to 0.6. At temperatures >400°C a scatter of friction coefficients between 0.2 and 0.7 is observed. By using small-spot-ESCA (electron spectroscopy for chemical analysis) it was shown that the low friction coefficients of SiCjSiC were due to formation of thin oxide layers formed by tribo-oxidation, while the higher friction coefficients accompanied by higher wear coefficients were related to thicker oxide layers.

126


Figure 23. (a) SEM micrograph of a polished and plasma-etched section of sintered Sic, ‘EKasic D’, note: bimodal grain size distribution. (b) SEM micrograph of EKasic D sliding face after about 4000 h of successful service as a mechanical seal face (SiC/carhon pair, lubricated with water, maximum pressure 50 bar). (c) Schematic of the operational condition of a relief structured sliding surface in the lubricated state and after breakdown of the lubricant film.

These results are in excellent agreement with those of Yamamoto et al. 12211, who observed a decrease in the coefficient of friction when SiCjSiC couples had been heated for 1 h at 1000°C with formation of a thin oxidation layer. However, at a higher oxidation level the coefficient of friction increased rapidly. A low coefficient


T = 22oc

T = 4OOY

T = 800°C

T =

727

looooc

Figure 24. Friction coefficientJ; and wear coefficient, k , of SiC,ISiC sliding pairs as a function of sliding velocity with 10 N for various temperatures (sintered SIC ‘EKasic D’, after Habig and Woydt, BAM Berlin/Germany [220]).

of friction for the SiCjSiC couple was also measured by Martin et al. [222] who studied friction under an oxygen partial pressure of 50 mPa, which should admit only a very thin oxidation layer. Sasaki [223] studied the influence of humidity on friction and wear behavior of SiCjSiC couples. It was shown that humidity decreased the coefficient of friction from 0.5 (dry air) to 0.2 (wet air), simultaneously a decrease in the coefficient of wear was observed from (dry air) to mm3 N-l m-I. Denape and Lamon [224] considered the important contribution of wear particles to the wear mechanisms. The main findings were: polishing at low loads (5 N) is due to fine individual wear particles smaller than 1 pm circulating in the sliding interface, and abrasion and grain pull-out (at high loads of 20N) are associated with large accumulations of particles adherent to the sliding phases. Wear tests performed under water showed again an anisotropic wear of individual S i c crystallites. It was concluded that the circulation of wear debris in the sliding interface is controlling both the wear rate and the friction response. Woydt et al. [225] evaluated the tribological behavior of SIC-TIC and SiC-TiB2 ceramic-matrix composites using a special high temperature tribometer. In an normal ambient environment these composites are able to form ‘lubricious oxide’ reaction layers on the hard substrate; the low friction coefficient of the couple Sic--TiC/SiC-Tic at room temperature with values of f = 0.2. and 0.3 was explained by the formation of self-lubricating layers of TiOz and SiC,O,,. The

728


friction coefficient of self-mated SiC-TiB2 sliding couples was in the range 0.56-0.72 and compared to sintered aSiC materials no improvement was observed in the wear resistance of the SiC-TiB2 composite [226]. In a study on the friction and wear behavior of lubricated ceramic journal bearings it was shown by Maurin-Perrier et d.[227] that both SiCjSiC and SiC/Si3N4 couples provide a significantly better behavior in terms of film stability compared with the classical materials used for water- or hydrocarbon-lubricated bearings. A self-improvement of surface roughness during the running-in period by tribochemical reaction significantly increases the range of stability of the film; consequently it is not necessary to produce a highly expensive surface finish to increase the performance of the SiCjSiC couple. The lifetime of the components is mainly determined by the behavior during the start and the stop phases where boundary lubrication occurs. Sliding wear on sintered silicon carbide leads to a very smooth surface between residual porosity. Further it was found [228] that in dry run situations, sintered Sic of bimodal grain structure (EKasic D) outperforms other Sic materials. The bimodal EKasic D shows the lowest development of heat, and therefore it has the lowest coefficient of friction. In a mayor US study on 'Tribological Fundamentals of Solid Lubricated Ceramics' [229] the bimodal grain size/shape distribution was judged best in terms of wear resistance, when compared with other sintered aSiC materials. Loffelbein et al. [230] tested self-mated sliding couples of SSi3N4,HIP-RBSi3N4, SSiC, SiSiC, MgO-Zr02 and A1203 in different aqueous solutions (H20, NaOH, KOH, NH40H, HN03, H2S04, H3P04, CH3COOH, HCl, HC104) under conditions of boundary or mixed lubrication. The best frictional behavior was observed with couples of SSiC, with steady state friction coefficients of 0.05. The lowest wear coefficients were measured for couples of the two types Si3N4,SSiC and A1203with values of approximately 10-7mm3N-'m-'. If low friction and low wear are required, couples of SSiC seem to be the best choice in most aqueous solutions. A recent study by Anderson and Blomberg [231], based on tests with sintered silicon carbide sliding unlubricated on itself in point (pin-on-disc), line (journal bearing) and plane (mechanical face seal) contacts. Tribo-oxidation and surface fracture were identified as the dominating deterioration mechanisms. The oxidation products formed silicon dioxide and, within narrow operational regimes, silicon monoxide. The highest wear rates occured in the pin-on-disc configuration, while the lowest rates were obtained in the journal bearing tests. Kitaoka et al. [232] investigated the effects of temperature and sliding speed on the tribological behavior of sintered silicon carbide (0.1 weight-% B/l .O weight-% C) by sliding on the same material in deoxygenated water from room temperature to 300°C under high vapor pressures (120°C/2 bar, 3OO0C/85 bar). The wear mechanism appears to consist of hydrothermal oxidation of Sic according to the equations:

+ 2 H 2 0 + Si02 + CH4, S i c + 4H20 Si02 + C 0 2 + 4H2 S i c + 2H20 Si02 + 0.5C2H6+ 0.5H2, S i c + 2H20 Si02 + C + 2H2, Sic

4

4 4


729

and dissolution of reaction products such as silica. Fine mirrorlike worn surfaces were observed without wear debris under all sliding conditions. For description of the tribotechnically used machine-elements and more information about tribology, the book ‘Tribology Handbuch Reibung und Verschleip’ [233] is recommended to the reader. ~

5.4.3.2 Sintered S i c Material Development in Sliding Wear Applications The coefficients of friction of hard and wear resistant ceramic materials and of most surface coatings are always greater than 0.1 under dry run conditions. This value is far too high for dry running bearings to be designed. Frictional heat develops, and if loads are high then very high temperatures are generated. These will not affect the sintered silicon carbide but the housing and the overall structure may well distort and become damaged leading to failure [234]. Since materials research shows that it is impossible to develop hard ceramic material with coefficients of friction below 0.01 (which is the typical value for a lubricated bearing) then the targets for product development must include: ~

~

stabilizing the hydrodynamic lubrication film, preventing dry running, reducing friction and wear if the lubrication film breaks down.

There are obviously two routes towards stabilizing the hydrodynamic film. One is to optimize the design of the component including the most suitable bearing surface characteristics. The other is to optimize the silicon carbide material itself. In following this second route, it was assumed that the tribological performance can be improved by tailoring the microstructure of SSiC. This can be done by introducing isolated pores which act as lubricant pockets or incorporating graphite particles, which act as dry lubricants. Fig. 25 shows typical sliding faces (left) as well as the microstructures (right) of the improved materials in comparison with standard EKasic D. The material containing pores (diameter about 40pm) was named TRIBO 2000, the second material containing pores and graphite particles, in the range 40-60 pm, was named TRIBO 2000-1. Properties of these S i c materials are shown in Fig. 26 in comparison with EKasic D. Both pores and graphite inclusions act as internal flaws, so it is no surprise that strength is reduced. On the other hand since these imperfections are of regular size it follows that the Weibull modulus is increased which is important for design. The Young modulus is slightly reduced. Friction under boundary lubrication conditions decreases as intended, while the corrosion resistance is unchanged. TRIBO 2000 and TRIBO 2000-1 were tested in product lubricated bearings and mechanical seals in comparison with EKasic D as the basis sintered S i c material. The Stribeck testing method was used to evaluate the coefficient of friction as a function of sliding velocity (see Fig. 27). This is an axial bearing test in which a circular segment sits (lubricated) on a rotating ring of the same material. The torque between the segment and ring is measured. The load bearing capacity of the hydrodynamic lubricant film is the key factor. The better tribological material is the one which

730


Figure 25. SEM micrographs of sliding faces (left) and polished sections (right) of sinteredSiC grades: EKasic D (top), TRIBO 2000 (middle), and TRIBO 2000-1 (bottom). The graphite in the TRIBO 2000-1 microstructure was revealed as white particles after coating the polished section with a carbon film.

shows stable hydrodynamics down to slower sliding velocities, before the lubricant film breaks down and the bearing comes into solid state contact. Figure 28 is an example. In this case, perhaps unexpectedly, the TRIBO 2000-1 material shows both a higher hydrodynamic friction and a deviation with increasing velocity when compared with TRIBO 2000 and EKasic D.

5.4 Properties qj" Silicon Cavhidp

-5

73 1

250

200

a -

150 a .->

5

100

00

50

0 Flexural Strength Weibull 4-Point Modulus

Fracture Toughness

Youngs Modulus

~ Corrosion coeffof ~ Resistance khaml Seal

Thermal Thermal Conductlvity Expansion

p.25 bar v=9 4 in's

Figure 26. Comparison of properties of the pore-containing material (TRIBO 2000) and the poresplus-graphite-particles-containing material (TRIBO 2000- 1) with EKasic D

Summarizing all Stribeck test data (load, media, and up to 10000 repeats) the TRIBO 2000 material shows the best performance in terms of stabilizing the hydrodynamic film down to the lowest velocities. This is important for sliding bearings and has been confirmed many times. The poorer performance of the TRIBO 2000-1 material in this typical sliding bearing test is not clearly understood, but it

Lubricant stationary body Surface roughness

R

FOUR MODES OF LUBRICATION 2 3

Boundary friction ( h ---> 0) Mixed friction (h = R ) Elastohydrodynamic (h > R ) lubrication

4

Hydrodynamic lubrication (h >> R )

1

h Film thickness

0 0.1'

f

STRIBECK - CURVE

Coefficient of friction

0,OI Viscosity

.

Rubbing speed v __f

load

Figure 27. Schematic of a Stribeck curve and the four modes of lubrication

~

c

t

~

132


6

a

5

7

=c 4 0 .c

.o 3 IL

o ~ ~ ' ' ' i ' ' " i ' ' " ~ ' " ' ~ ' ' " l

0

5

10 15 20 Sliding Velocity [m/s]

25

Figure 28. Coefficient of friction as a function of sliding velocity for self-mated sliding pairs of sintered S i c materials.

may have something to do with the poor wetting behavior of graphite in water under normal pressure. In mechanical seal applications, under differential pressures of less than 15 bar the same ranking of materials is observed in terms of friction and wear. Under extreme situations, however, like very high differential pressure (producing continuous boundary lubrication) or pump cavitational run (causing breakdown of the hydrodynamic film under frequent violent loads), the TRIBO 2000-1 material showed the best performance (see Fig. 29). To diminish the effects of hydrodynamic grain-boundary corrosion in applications with hot water, such as chipping out of fine grains in hot spots, tribochemical reaction with water S i c + 2H20 = Si02 + CH4

1.2 I

I

.-0

.a

m

Y

a"

-

+ .-

n 0 In

0 Figure 29. Relative wear volume on seal rings of sintered Sic material (self-mated couples).


733

Figure 30. Microstructure of EkasicW: coarse-grained SSiC with platelet structure for mechanical seals and bearings (hard/hard combinations) in hot water applications.

and formation of damaging SiOz layers on the sliding surfaces, another two new modified SSiC materials with a predominantly coarse-grained, bimodal platelet structure have been developed [235]. Catastrophic failure of coponents is avoided since the large S i c platelets near the surface are anchored to a depth at which there is no grain-boundary corrosion. EKasic W silicon carbide is a dense material with a predominantly coarse-grained bimodal platelet structure. The effectiveness of the coarser microstructure (see Fig. 30) in improving the corrosion resistance has been clearly demonstrated in practical tests on a mechanical seal test rig. Even after 500 h of testing (deionized water, 60°C, 6 bar, hard/hard couples) no SO2 layer was formed on the functional surface.

Figure 31. Microstructure of EKasic HW: coarse-grained SSiC with platelet structure and graphite particles (black phase 4 0 4 0 pm in diameter) for mechanical seals and bearings (hard/hard combinations) in hot water applications.

734


EKasic HW silicon carbide, on the other hand, is a dense material with a predominantly coarse-grained, bimodal platelet structure and graphite particles of 40-60pm in size (see Fig. 31). This new tribological material, which is still in its test phase, should be particularly suitable for mechanical seals and sliding bearings in applications in contact with hot water containing solids.

5.5 Quality Control The S i c content of silicon carbide products is now usually determined by measuring the carbon contents. The total carbon content is determined by combustion of the sample in a stream of oxygen at 1050°Cin the presence of lead borate. The COz produced is absorbed in Ba(C104)z solution and determined by coulometry [236,237]. An alternative technique is to oxidize the S i c with oxygen in a high-frequency induction furnace containing a flux metal, and to detect the C 0 2 produced by IR absorption. Free carbon can be directly determined coulometrically by combustion at 850°C without any additives, provided that S i c itself is not appreciably oxidized. If the S i c samples are very finely divided, that is, if they have large specific surfaces, and if long heating times are required due to high free carbon contents, this method can not be used. In such cases, the weight change on combustion is measured and the decrease in total carbon content is determined. This enables the effect of the unavoidable oxidation of the S i c to be eliminated, and the free carbon content can be calculated [236,237]. For the determination of free carbon in S i c sintering powders or in more-or-less contaminated Sic, containing iron or silicon impurities or small amounts of sintering aids etc., the use of a wet chemical method, based on the wet chemical dissolution of free carbon in a hot chromic-sulfuric-acid mixture proved to be very useful [238]. Analyzers with IR detection are also suitable for the direct determination of free carbon provided that they allow precise temperature control during combustion of the free carbon. The S i c content is calculated from the difference between total and free carbon: weight-% S i c = (weight-%C,,,,,

-

weight-%Cfree)x 3.3383

The free silicon content is determined by measuring the volume of hydrogen produced on treatment with sodium hydroxide solution. Alternatively, metallic silver is precipitated from a silver fluoride solution, dissolved in nitric acid, and determined by a standard method [236,237]. In the chemical analysis of S i c abrasive grits, emphasis is placed on the determination of accompanying materials such as carbon, silicon, silicon dioxide, and metallic oxides. The S i c content (plus the free carbon) is found by weighing the residue after treatment with a mixture of hydrofluoric, nitric, and sulfuric acids [237]. For special grades of S i c for refractory and metallurgical applications, volatile components can interfere with the S i c determination. In this case, the sample must be annealed under argon [239].

5.5 Quality Control

735

In order to characterize S i c powders and sintered ceramics the total oxygen and nitrogen content as well as the contents of metallic impurities are analyzed. Total oxygen and nitrogen contents are usually determined by an inert gas fusion method (Leco T C 436) using powdered samples, whereas metallic impurities (Na, K , Ca, Mg, V, Fe, Ti, Al, Cr and Ni) and boron content are determined in acidic solutions by inductive plasma emission (ICP) spectroscopy [240-2421. Besides chemical analysis, physical properties such as particle size, particle size distribution, and bulk density are also important. The particle size distribution of coarse abrasive materials (5 mm-50pm) is determined by sieving. For particle sizes less than 50 pm, sedimentation and laser diffraction methods are used. Both the particle size distributions and the methods for their determination are standardized. The standard for bonded abrasive applications [243] differ from those for coated abrasives [244]. The bulk density depends both on the particle size distribution and on the particle shape. It is measured by weighing a known volume of S i c grains [245]. The packed density is often determined instead of the bulk density [246]. According to the analysis of the microstructure [247] and the physical properties measurements, the reader is referred to the literature [93,248-25 11. The major sequential steps in conducting analysis and properties evaluation for SIC ceramics are shown in Fig. 32.

Ceramic

Powder Processing

Productioi

MixingISpray Drying Bulk Chemical and Physical Analysis

Analysis

Microstructural

Molding

Analysis, SEM

Sintering

TEM, EELS, AES

Raw Materials to Ceramic Materials

Figure 32. Analysis and properties determination for S i c materials.

136

5 Silicon Carbide Based Hard Muterials

5.6 Toxicology and Occupational Health Silicon carbide is nontoxic, and is therefore a nonhazardous material as defined by GefStoffV [252]. For fiber-free finely divided Sic, the MAK value is 4mgmP3 [253]. If the MAK (‘Maximalzulassige-Arbeitsplatz-Konzentration’)is exceeded a dust mask with a PI filter (for inert dusts) must be worn. Recent investigations, however, have shown that these Sic dusts are not fibrogenic [254], so that the need for a MAK value is questionable. Sic whiskers have no known routes of entry into the body except as airborne particles. Conclusive experimental studies on the effects of respirable fibrous S i c dusts (whiskers) are not yet available [255]. In the United States ASTM has published recommendations for safe handling procedures related to all ceramic fibers, including silicon carbide whiskers [256].

5.7 Uses of Silicon Carbide Of the M 700 000 tons Sic produced per year, about 33% is used in metallurgy as a deoxidizing plus alloying agent, and about 50% in the abrasive industry [257]. The remainder is used in the refractory and structural ceramics industries and to a small extent also in electric and electronic industries as heating elements, thermistors, varistors, light-emitting diodes, and attenuator material for microwave devices. In its loose granular form silicon carbide is used for cutting and grinding precious and semiprecious stones and fine grinding and lapping of metals and optical glasses [258-2601. Bound with synthetic resins and ceramic binders S i c grits are used in grinding wheels, whetstones, hones, abrasive cutting-off wheels, and monofiles for machining of metals, ceramics, plastics, coal-based materials, and so on [26 11. Coated abrasives include abrasive paper and cloth in sheet or band form. They are produced by strewing the Sic grains onto a substrate coated with glue or bonding resin and then covering with a second layer of bonding agent [257]. The addition of silicon carbide during the melting of cast iron aids carburization and siliconization, and improves the quality of the cast iron as a result of its seeding action [262]. In the production of steel in an arc furnace, silicon carbide acts as deoxidant and helps in slag melting. The need to control thermal expansion and to increase the strength and Young modulus of aluminum alloys produced new Al-Sic composites alloys containing up to 50 vol-’YOSic particles [263]. The market is poised for rapid expansion in this area over the next years as developmental products move into commercialization stages. For reinforcement applications Sic is also used in the form of whiskers [264], platelets and fibres. The resistance of ceramically bonded and recrystallized silicon carbide to thermal shock, oxidation, and corrosion is utilized in its use as a refractory construction

5.7 Uses of Silicon Carbide

737

material, for example, in linings and skid rails for furnaces and hot cyclones, and as a kiln furniture, especially in saggers [61,265-2691. The good electrical conductivity of the material at high temperatures, coupled with its outstanding oxidation resistance, led to its early use in the electric heating industry [270-2721, which markets its products mainly in the form of rods and tubes that operate up to 1500°C. Recrystallized S i c igniters are used in home gas appliances, replacing pilot lights. High-purity S i c shapes are used in the electronic industry as furnace components for processing of silicon wafers. The thermoelectric properties of S i c suggest the use of sintered S i c rods as high-temperature thermoelements [273] and as Seebeck elements [274] for high-temperature thermoelectric energy conversion. Voltagedependent resistors (varistors) consist of ceramic- or polymer-bonded S i c and are used in overvoltage protection equipment. Silicon carbide is an outstanding material for the construction of electronic equipment. Blue light-emitting diodes having an improved 470 nm peak wavelength are being produced and marketed as the first commercial S i c semiconductor device. The continual development of the deposition of S i c thin films and of large diameter single crystal S i c wafers, the associated technologies of doping, etching and electrical contacts have culminated in a host of new solid state devices including field effect transistors capable of operation up to 650°C [275]. High-density, high-strength SiSiC, SSiC, HP-, and HIPSiC materials, which have been on the market only since the 1970s, have opened up a new field of application [276], namely, in mechanical and high-temperature engineering. SiSiC and, especially SSiC, are displacing the chemically less resistant tungsten carbide (hard metal) and the thermal-shock-sensitive aluminum oxide in modern mechanical seals where they are used in the form of slide rings (see Fig. 33). The excellent

Figure 33. SSiC slide rings (ESK-Kempten, Germany).

738


Figure 34. SSiC sliding bearings (ESK-Kempten, Germany).

wear resistance of sintered silicon carbide, its excellent chemical resistance, and outstanding tribological characteristics ensure that mechanical seals made of this material last longer, resulting in much reduced maintenance and production costs for pump-dependent processes in the chemical industry [277]. SiSiC seal rings (in contrast with those made of pure SSiC) can only be used in acid media because of the accompanying silicon phase, which is attacked by alkalis [207]. For similar requirements involving radial loads, sliding bearings (see Fig. 34) are manufactured from SSiC. The erosion and chemical resistance of S i c enable the designer to position the bearings in the medium to be transported; that is, to eliminate lubrication and sealing problems. Other components include shaft protection sleeves for waste gas exhaust fans, and precision spheres for dosing and regulating valves. Due to the tailored properties of liquid phase sintered silicon carbide (LPSSiC) it is used as dewatering elements in the paper machinery and as rings for highly stressed gas seals. It is a price competitive alternative to silicon nitride materials and outperforms alumina and tungsten carbide materials. In addition, LPSSiC is proposed as neutral matrix in ceramic matrix composites containing plutonium to burn the world’s stockpiles of military plutonium in thermal or fast reactors [278]. Hot-pressed SIC is the preferred material to replace oxide ceramics for rods fixtures, and punches in high-temperature strength testing equipment. In view of the low level of plasma contamination, the low induced radioactivity, and excellent

5.7 Uses of Silicon Carbide

739

high-temperature resistance of silicon carbide, it is the ideal material (especially in its isostatically hot-pressed form) for use in fusion reactors [279]. Combustion tubes made of slip-cast SiSiC [280] have better resistance to corrosion, high-temperatures, and thermal shock, so that they will last far longer than, for example tubes made of heat resistant steel. The heat treatment industry has begun using SSiC radiant tubes in indirect gas-fired heat treating operations. In such systems, the tubes are internally heated by combustion burners and radiant heat to some external work load, such as an ingot of alloy, which is isolated from the combustion atmosphere [281]. SiSiC and SSiC are destined for use in heat exchanger systems because of their high-thermal conductivity and corrosion resistance [282-2863. Sintered silicon carbide bonded to plastic laminate substrates reinforced with glass or Kevlar TM fabrics can be used as ceramic armor to defeat armor piercing projectiles [287]. Sintered and isostatically hot-pressed S i c materials, as well as silicon nitride (Si3N4) are playing an important role in the development of ceramic components (see Fig. 35) for motor vehicle engines and gas turbines [288-2931. Real technical success with sintered S i c components in the field of high temperature gas turbines has not yet been achieved [294,295]. However, S i c and Si3N4gas turbine parts are in field tests and strong development efforts continue in several countries. Developments for the application of continuous fiber reinforced S i c matrix composites (CMCs) have started with and are concentrating on hot components in military and space technology: hot gas ducts and thermal heat shields for space

Figure 35. Monolithic SSiC gas turbine rotor (courtesy A.Lipp, ESK-Kempten/Germany).

740


reentry vehicles [173]. Due to the high thermal shock capability of this class of materials and to the high fracture toughness of some of these composites some civil applications are gaining importance: highly loaded brake discs, for high speed trains, and highly loaded journal bearings; in both applications the conventional materials (metals for brakes, monolithic ceramics for bearings) are not applicable because of the thermal loads and the brittle failure mode respectively.

Acknowledgments Some parts of the text appeared in the earlier review by K.A.S. (Silicon Carbide in: Encyclopedia of Advanced Materials pp. 2455-2461,1994) and Elsevier Science Ltd., Kidlington OX5 16B, UK is kindly thanked for permission to use them.

References I. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

R. Kieffer and F. Benesovsky, Hartstoffe, Springer Verlag,New York, 1963. J. J. Berzelius, Am. Phys. Chem. 1824, 1, 169-230. C. M. Despretz, Compt. Rend. 1849, 29, 709-724. P. Schutzenberger and A. Colson, Compt. Rend. Akad. Sci. 1881,92, 1508. H. Moissan, Der elektrische Ofen, Verlag M. Krayer, Berlin, 1900. E. G. Acheson US Pat. 492 767 (28 Feb. 1893), DE 76629 (1894), DE 85195 (1896). Gmelin Handbook of Inorganic Chemistry, 8th edition, Silicon Supplement Vo.B 3, System No. 15, H. Katscher et al. (eds.), Springer Verlag, Berlin, 1986. H. Moissan, Compt. Rend. Akad. Sci. 1905, 140, 405. A. J. Regis,Land . B. Sand, Bull. Geol. Soc. Am. 1958, 69, 1633. A. P. Bobrievich, V. A. Kalyuzhnyi, and G. I. Smirnov, Proc. Akad. Sci. USSR: Geol. Sci. 1957, 115, 757. J. Bauer, J. Fiala, and R. Hrichovi, Am. Miner. 1963, 48, 620-34. H. Kleykamp and G . Schumacher, Ber. Bunsenges. Phys. Chem. 1993,97, 799-805. A. Dietzel, H. Jagodzinski, and H. Scholze, Ber. Dtsch. Keram. Ges. 1960, 37, 524-537. R. Kieffer, E. Gugel, P. Ettmayer, and A. Schmidt, Ber. Dtsch. Keram. Ges. 1966,43,621-623. N. W. Jeeps and T. F. Page, J. Am. Ceram. Soc. 1981, 64, C177-C178. R. F. Adamsky and K. M. Merz, Z. Kristallogr. 1959, 111, 350-361. H. Ott, Z. Kristallogr. 1925, 61, 515-532; 62, 210-218; 63, 1-19. L. S. Ramsdell, Am. Miner. 1947, 32, 64-82. J. Ruska, L. J. Gauckler, J. Lorenz, and H. U. Rexer, J . Muter. Sci. 1979, 14, 2013-2017. Y. Inomata, Z. Inoue, M. Mitomo, and H. Suzuki, Yogyo-Kyokai-Shi, 1968, 76, 313-319. D. Lundquist, Acta. Chem. Scand. 1948, 2, 177-191. P. T. B. Shaffer, Mater. Res. Bull. 1969, 4, S13-24. M. Mitomo, Y. Inomata, and M. Kumanomido, Yogyo-Kyokai-Shi1970,78, 365-369. M. Mitomo, Y. Inomata, and H. Tanaka, Muter. Res. Bull. 1971, 6, 759-764. M. M. Patience, Silicon Carbide Alloys, PhD. Thesis, University of Newcastle upon Tyne, 1983. D. Foster, Densification of Silicon Carbide with Mixed Oxide Additives, University of Newcastle upon Tyne, 1996. N. W. Jeeps and T. F. Page, J . Microscopy 1979, 116, 159-171; 1980, 119, 177.

References

741

28. Anon, Ind. Heating, 1954, 992-1004. 29. W. Poch and A. Dietzel, Ber. Dtsch. Keram. Ges. 1962, 39, 413426. 30. K. Liethschmidt, Siliciumcarbid in Chemische Technologie, Band2, Anorg. Techn. I , 4 . Aufl. K. Winnacker and L. Kuchler (Eds), Hanser-Verlag, Munchen-Wien, 1982, pp. 626-629. 31. K. H. Mehrwald, Ber. Dtsch. Keram. Ges. 1992, 69, 72-81. 32. A. Graf von Matuschka and G. Schonbrunn, Stand der Mikrokorngropenbestimmung von Elektrokorund und Sic, ZwF 1981, 76, 3 8 4 5 . 33. W. Boecker, H. Landfermann, and H. Hausner, Powder Met. Int. 1981, 13, 37-39. 34. S. Prochazka, Techn. Report 86-CRD-158, General Electric, Schenectady NY, 1986. 35. P. Matje and K. A. Schwetz, Proc. 1st Internat. Conf. Ceramic Powder Processing Science, Nov. 87, Orlando FL, Gary Messing et al. (Eds), Ceram. Trans. IY88, 1, 46W68. 36. A. W. Weimer, Carbothermal reduction synthesis processes, in Carhide, Nitride and Boride Materials Synthesis and Processing, A. W. Weimer (Ed.), Chapman & Hall, London, 1997, pp. 75-180. 37. J. S. Prener, US Patent 3 085 863, 1960. 38. T. L. O'Connor and W. A. McRae, US Patent 3 236 673, 1963. 39. S. Prochazka. Final Report SRD 72-171, General Electric. Schenectady N Y . 1972. 40. K. A. Schwetz and A. Lipp, Radex-Rundschau, 1978, 2, 479498. 41. W. M. Goldberger, A. K. Reed, and R. Morse, Synthesis and characterization of HSC silicon carbide, in S i c '87, J. D. Cawley (Ed.), Ceram. Trans. 1989, 2, 93-104. 42. F. Hahn, G. Rudakoff, and H.J. Tiller, Hermsdocfer Techn. Mitt. 1990, 79, 2546-2550. 43. H. R. Baumgartner and B. R. Rossing, Pressureless sintering and properties of plasma synthesized S i c powder, in S i c '87, J. D. Cawley (Ed.), Ceram. Trans. 1989, 2, 3-16. 44. F. G. Stroke (PPG), US Patent 4 295 890, 1981. 45. W. R. Cannon, S. C. Danforth, J. H. Flint, J. S. Haggerty, and R. A. Marra, J . Am. Ceram. Soc., 1982, 65, 324330. 46. W. Bocker and H. Hausner, Ber. Dtsch. Keram. Ges. 1978, 55, 233-237. 47. P. T. B. Shaffer, Sic Whiskers, in Handbook of Advanced Ceramic Materials, Advanced Refractory Technologies, Buffalo, NY, 1994. 48. T. N. Tiegs and T. F. Becher, Am. Ceram. Sac. BuN. 1987, 66, 339-342. 49. W. D. G. Bocker, S. Chwastiak, F. Frechette, and S. K. Lau, Single phase CtSiC reinforcements for composites, in: SIC '87, J. D. Cawley (Ed.), Ceram. Trans. 1989, 2, 407420. 50. P. A. Kistler-De Coppi, and W. Richarz, Int. J . High Techno/. Ceram. 1986, 2, 99-113. 51. B. Meier, R. Hamminger, and E. Nold, Microchim. Acta 1990,II, 195-205. 52. R. Lenk, J. Adler, Influence of forming technique on the Sic platelet orientation in a liquidphase sintered Sic-matrix, in Fourth Euroceramics, Vol. 2, Part II. Basic Science, C. Galassi (Ed.), Faenca Editrice Italy, pp. 407414. 53. R. Lenk, A. F. Kriwoschepov, and K. Grope, Heipgiepen von drucklos gesintertem Sic, Sprechsaal Ceram. Mater. 1995, 128, 17-20. 54. S. Yajima, J. Heyashi, M. Omori, and K. Okamura, Nature, 1976, 261, 683. 55. T. Ishikawa, S i c continuous fiber, Nicalon, in SIC Ceramics-2, S. Somiya and Y. Inomata (Eds), Elsevier, Amsterdam, 1991, pp, 81-98. 56. H.-P. Baldus and M. Jansen, Angew. Chem. 1997, 109, 338-354. 57. D. Heimann, T. Wagner, J. Bill, F. Aldinger, and F. F. Lange, J . Mater. Res. 1997, 12, 30993101. 58. S. Prochazka, Sintered S i c and method of making, US Patent 566 8068, 1997. 59. K. K. Kappmeyer, D. H. Hubble, and H.W, Powers, Am. Ceram. Soc. Bull. 1966,45, 1060-1064. 60. M. E. Washburn and R. W. Love, Am. Ceram. Soc. Bull. 1962, 41, 447449. 61. A. Fickel, S i c Materials, in Refractory Materials Pocket Manual, G. Routschka (Ed.), Vulkan Verlag, Essen, 1997, pp. 74-80. 62. R. Van der Beck and J. O'Connor, Ceram. Ind. 1957, No. 3 , 9 6 9 8 . 63. R. A. Alliegro, Processing and fabrication of non-hotpressed Sic, in Ceramics,for High Performance Applications, J. J. Burke et a/., (Eds), Metals and Ceramics Inf. Center, Columbus, OH, 1974, pp. 253-263. 64. J. Kriegesmann, Interceram. 1988, No. 2, 27-30 and Powder Metall. Int. 1986, 18, 341-343. 65. P. Popper and D. G. S. Davies, Powder Metall. 1961, 8, 113. -

142


66. K. M. Taylor, Improved K TSiC for high temperature parts, Muter. Meth. 1956, Vol. 44, No. 4, 92-95; US Patent 3 189 472, 1965. 67. R. Kieffer, E. Gugel, and A. Schmidt, Verfahren zur Herstellung eines dichten Formkorpers auf Basis Sic, DE-OS 1 671 092, 1967. 68. P. Kennedy and J. V. Shennan, Engineering applications of REFEL silicon carbide, in Silicon Carbide 1973, R. C . Marhall et a/. (Eds), Univ. of South Carolina Press, Columbia, 1974, pp. 359-366. 69. G. Q. Weaver, H. R. Baumgartner, and M. L. Torti, Special Ceramics 6, P. Popper (Ed.), Brit. Ceram. Res. Association, UK, pp. 261-281. 70. W. B. Hillig, R. L. Mehan, C. R. Morelock, V. J. De Carlo, and W. Laskow, Am. Ceram. Soc. Bull. 1975, 54,1054-1056. 71. T. Hase et al., J . Nucl. Muter. 1976, 59, 4248. 72. Anon. A super ceramic called REFEL, Interceram. 1982, No. 1, 50-51. 73. G. Willmann and W. Heider, Werkstofftechnik1983, 14,158. 74. K. Taylor, Dense silicon carbide, US Patent 3 205 043, 1962. 75. S. Prochazka, The sintering process for Sic, a review, Techn. Report 81-CRD-314, General Electric, Schenectady NY, 1981, pp. 16. 76. S. Prochazka, Sintering of Sic, in Ceramics for High Performance Applications, J. J. Burke et al. (eds.), Metals and Ceramic Info. Center, Columbus, OH, 1974, pp. 239-252. 77. J. A. Coppola and G. H. McMurtry, Substitution of ceramics for ductile materials in design, Nut. Symp. on Ceramics in the Service of Man, Carnegie Institution, Washington DC, 1976. 78. P. T. B. Shaffer, The S i c phase in the system SiC-B4C-C, Muter. Res. Bull. 1969,4, 213-220. 79. R. Hamminger, Carbon inclusions in sintered Sic, J . Am. Ceram. SOC.1989, 72, 1741-1744. 80. Y. Murata and R. H. Smoak, Densification of Sic by the addition of BN, BP and B4C, and correlation to their solid solubilities, in Proc. Int. Symp. Densijcation and Sintering, Oct. 1978, Hakone/Japan, S. Somiya and S. Saito (Eds), Gakujutsu Bunken, Tukyu-Kai, Tokyo, 1979, pp. 382-399. 81. W. Boecker and H. Hausner, Powder Metall. In?. 1978, 10, 87-89. 82. D. R. Stutz, S. Prochazka, and J. Lorentz, 1. Am. Ceram. SOC.1985, 68,479482. 83. T. Fetahagic and D. Kolar, Ceram. Acta 1990, 2, 31-37. 84. K. A. Schwetz, A. Lipp, The effect of boron and aluminum sintering additives on the properties of dense sintered aSiC, Science of Ceramics 10, Berchtesgaden Sept. 1979, H. Hausner (Ed.), Verlag Deutsche Keramiske Ges 1980, pp. 149-158. 85. W. Grellner, K. A. Schwetz, and A. Lipp, Fracture phenomena of sintcred alphasic, in Proc. 7th Symp. on Special Ceramics, Bedford College, London, Dec. 1980, D. Taylor and P. Popper (Eds), Br. Ceram. Res. Assoc., Stoke on Trent 1981, pp. 27-36. 86. W. Boecker, H. Landfermann, and H. Hausner, Powder Metall. Int. 1979, 11, 83-85. 87. H. Hausner, Pressureless sintering of non-oxide ceramics, in 4th CIMTEC, Energy and Ceramics, St. Vincent, Italy 1979, P. Vincenzini (Ed.), Elsevier Scientific Publ. Comp., Amsterdam, 1980, pp. 582-595. 88. H. Tanaka, Y. Inomata, K. Hara, and H. Hasegawa, J . Muter. Sci. Lett. 1985, 4,315-317. 89. K. Suzuki, Pressureless sintering of SIC with addition of AI2O3,Rep. Rm. Lab. Asahi Glass Co. 1986,36, 25-36. 90. K. Suzuki, Pressureless sintering of S i c with addition of AI2O3,in SIC Ceramics - 2, S. Somiya and Y. Inomata (Eds), Elsevier, London, 1991, pp. 163-182. 91. R. H. Smoak, Pressureless sintering beryllium containing SIC powder composition, DE-OS 27 51 851, 1977. 92. A. Mohr, Untersuchungen zur Minimierung der Additivgehalte fur die drucklose Sinterung von aSiC, Diplomarbeit Institut fur Keramik im Maschinenbau, Universitat Karlsruhe, 1989. 93. D. W. Richerson, Modern Ceramic Engineering, Dekker, New York, 1992. 94. F. Thummler and R. Oberacker, Introduction to Powder Metallurgy, The Institute of Materials Series, University Press, Cambridge, 1993. 95. C. A. Johnson and S. Prochazka, Microstructures of sintered Sic, in Ceramic Microstructures 1976, R. M. Fullrath and J. P. Pdsk (Eds), Westview Press, Boulder, CO, 1977, pp. 366-378. 96. K. A. Schwetz, F. Isemann, and A. Lipp, Injection molded sintered turbine components of Al-

References

143

doped alpha-Sic; in Proc. 1st Internat. Symp. Ceramic Comp. For Engine, S. Somiya et a/. (Eds), KZK Scientific, Tokyo, 1984, pp. 583-594. 97. Showa Denko, DE 3 927 300, 1988. 98. Ceramiques et Composites, European Patent 486 336, 1991. 99. Carborundum Company, US Patent 5 589 428, 1996; US Patent 5 635 430, 1997. 100. ESK-GmbH, European Patent 685 437, 1995. 101. Carborundum Company, European Patent 145 496, 1984. 102. J. Greim, L. Sigl, and H. Thaler, Sintered SIC for high-performance bearings and seals, Magazine of Wacker Chernie-GmhH Werk+ Wirken: Int. Edn, 1997, 1, 23-25. 103. M. Omori and H. Takei, J. Am. Ceram. Soc. 1982, 65, C-92; and US Patent 5 439 853, . 1995. 104. E. Kostic, Powder Metall. Int. 1988, 20, 28-29. 105. R. A. Cutler and T. B. Jackson, Liquid phase sintered Sic, in Ceramic Materials and Components for Engines, Proc. 3rd. Int. Symp., V. J. Tennery (Ed.), Am. Ceram. Soc., 1988, pp. 309-318; and US Patent 4 829 027, 1988. 106. W. Boecker and R. Hamminger, Interceram. 1991, 40, 520-525. 107. F. K. Van Dijen and E. Mayer, J. Europ. Ceram. Soc. 1996, 16, 413420. 108. C. Wolf, H. Hubner, and J. Adler, Mechanical behavior of pressureless sintered SIC at high temperature, in 3rd Euro-ceramics, Vol.3, P. Duran and J. F. Fernandez (Eds), Faenze Editrice Iberica, Castellon, Spain, 1993, pp. 465470. 109. K. Y. Chia and S. K. Lau, Ceram. Eng. Sci. Proc. 1991, 12, 1845-1861. 110. K. A. Schwetz, P. Matje, M. Mohr, and K. P. Martin, Development of a pressureless sintering process for ceramics in the system SiC--AlN (0),Final Report on German BMFT Project No. 03-zC-153, 1986 (unpublished). 111. I. Wiedmann, M. Nader, M. J. Hoffmann, and F. Aldinger, in Symposium 7, Materialwissenschaftliche Grundlagen, Werkstqfimhe '96, F. Aldinger and H. Mughrabi (Eds), DGM Informdtionsgesellschaft mbH, Oberursel, 1997, pp. 51 5-520. 1 12. L. S. Sigl and H. J. Kleebe, J. Am. Ceram. Soc. 1993. 76, 773-776. 113. M. P. Padture, J. Am. Ceram. Soc. 1994, 77, 519-523; 2518--2522. 114. S. K. Lee and C. H. Kim, J. Am. Ceram. Soc. 1994, 77, 1655-1658. 115. J. S. Nadeau. Ceram. Bull. 1973, 52, 170-174. 116. R. A. Alliegro, L. B. Coffin, and J. R . Tinklepaugh, J. Am. Ceram. Soc. 1956.39, 386389. 117. S. Prochazka and R. J. Charles, Am. Ceram. Soc. Bull. 1973, 52, 885-891. 118. J. Kriegesmann, Ber. Dtsch. Keram. Ges. 1978, 55, 391-397. 119. F. F. Lange, J. Mafer. Sci. 1975, 10, 314320. 120. J. M. Bind and J. V. Biggers, J. Appl. Phys, 1976, 47,5171-5174. 121. D. Broussaud, Independence of composition of hot-pressed Sic, in Ceramic Microstructures '76, R. M. Fullrath and J. A. Pask, (Eds), Westview Press, Boulder, CO, 1976, pp. 679-688. 122. T. Iseki, K. Arakawa, H. Matsuzaki. and H. Suzuki, Yogyo-Kyokai-Shi1983, 91, 349. 123. K. Nakamura and 0. Asai, Kagaku Kogyo 1982, 33, 977. 124. H. Kessel and E. Gugel, Industrie Diamanten Rundschuu 1978, 12, 180-185. 125. J. Kriegesmann, K. Hunold, A. Lipp, K. Reinmuth, and K. A. Schwetz, European Patent 71241. 1981. 126. T. J. Whalen, R. M. Williams, and B. N. Juterbock, HIP of Sic and Si3N4structural ceramics, in IOth Plansee Seminar Proceedings, Reutte, Tyrol, 1985, p. 783. 127. K. Hunold, Powder Metall. Int. 1984, 16, 236-238; 1985, 17, 91-93. 128. K. Hunold, Sinter/HIP of Sic, in Proc. Adv. Mat. Technology Ceramic Workshop No. 4 , Advances in Materials, Processing and Manufacturing Science, Nagoya Japan, March 3 4 , Japan Fine Ceramics Center, 1988, pp. 49-62. 129. R. Oberacker, A. Kiihne, and F. Thummler, Powder Metall. Int. 1987, 19, 43-50. 130. F. Fetahagic, R. Oberacker, and F. Thummler, Process development for sinter-HIPing of Sic, in Ber. KFA-Jiihlich 1989, Juel. Conf: -77, Emerging Mat. Adv. Process., 9th German-Yugoslavian Meeting on Materials Science Development, April 1 6 1 9 , 1989, pp. 313-325. 131. T. Hirai and M. Sasaki, S i c prepared by CVD, in S i c Ceramics-I, S. Somiya and Y. Inomata (Eds), Elsevier, London, 1991, pp. 77-98. 132. Anon, Material Innovations: CVD scaled up for commercial production of bulk Sic, Am.

744


Ceram. Soc. Bull. 1993,12,74-78. 133. M. Tenhover, I. Ruppel, S. S. Lyle, and L. J. Pilione, DC-magnetron sputterable Sic, in Proc. 36th Annual Techn. Conf: Society of Vacuum Coaters, 1993,pp. 362-365. 134. Anon, S i c electronic materials and devices, in M R S Bull. 1997,22,No. 3 (Special Issue); Cree Research Inc., World Patent 97/28297,Growth of colorless S i c crystals, 1997. 135. J. A. Lely, Ber. Dtsch. Keram. Ges. 1955,32, 229-231. 136. R.Fantoni, E. Borsella, S. Piccirillo, R. Ceccato, and S. Enzo, J. Muter. Res. 1990,5,143-150. 137. R. Vassen, D. Stover, and J. Uhlenbusch, Sintering and grain growth of ultrafine amorphous Sic-Si-Powder mixtures, in Euro-Ceramics II, Vol. 2 , Structural Ceramics and Composites, Proc. 2nd European Ceram. SOC.Conf. Augsburg, Sept. 1991, G. Ziegler and H. Hausner (Eds), Deutsche Keram. Ges., Koln, pp. 791-797, 1993. 138. R. Vassen, H. P. Buchkremer, and D. Stover, Verfahren zum Herstellen feinkristalliner Siliciumkarbidkorper, DE 196 42 753 (1999). 139. R. Vassen and D. Stover, Philos. Mag. B 1997,76, 585. 140. J. Forster, R. Vassen, and D. Stover, J . Muter. Sci. Lett. 1995, 14, 214-216. 141. R. Vassen, J. Forster, and D. Stover, NanoStructured Muter. 1995,6, 889-892. 142. A. Kaiser, R. Vassen, T. Stover, and H. Buchkremer, NanoStructuredMater. 1997,8,489497. 143. M.Mitomo, Y.-W. Kim, and H. Hirotsuru, J . Muter. Res. 1996, 11, 1601-1604. 144. M. Mitomo, H. Hirotsuru, and Y.-W. Kim, US Patent 5 591 685,7 Jan. 1997. 145. R. Telle, R.J. Brook, and G.Petzow, J . Hard Muter. 1991,2, 79-114. 146. G. C.Wei and P. F. Becher, J . Am. Ceram. SOC.1984,67,571. 147. M. A. Janney, Am. Ceram. SOC.Bull. 1987,66,322. 148. D. J . Jiang, J. H. Wang, Y. L. Li, and L. T. Ma, Muter. Sci.Eng. 1989,A109, 401. 149. D.Ly Ngoc, Gefiigeverstarkung von Sic Keramiken, doctoral thesis, University of Stuttgart, Germany, 1989. 150. R.Jimbou, K. Takahashi, and Y. Matsushita, Adv. Ceram. Muter. 1986, 1, 341. 151. C.H. McMurtry, W. D. G. Backer, S. G. Seshadri, J. S. Zanghi, and J. E. Garnier, Am. Ceram. SOC.Bull. 1987,66,325. 152. H. Cai, W. H. Gu, and K. T. Faber, in 5th Techn. Con$ On Composite Materials, Proc. Am. SOC.Comp., 1990,pp. 892-901. 153. K. T. Faber, W. H. Gu, H. Cai, R. A. Winholtz, and D. J. Magleg, in Toughening Mechanisms in Quasi-Brittle Materials, S. P. Shah (Ed.), Kluwer, Dordrecht, 1991, pp. 3-17. 154. T. Tani, S.Wada, European Patent 303192, 1988. 155. C. Blanc, F. Thevenot, and D. Treheux, Elaboration and characterization of submicronic SiCTiB2 composites by reactive pressureless sintering, in Key Eng. Muter. 1997,132-136,968-971. 156. D. H.Kuo and W. M. Kriven, J . Europ. Ceram. SOC.1998,18,51-57. 157. K. A. Schwetz, K. Reinmuth, and A. Lipp, World Ceramics, 1985,2, 70-84. 158. F.Thevenot, Sintering of B4C and B4C-SiC two-phase materials and their properties, in Proc. 9th Int. Symp. on Boron, Borides and Related Compounds, H. Werheit (Ed.), University of Duisburg Germany,l987, pp. 246-256. 159. W. J. Tomlinson and J. C. Whitney, Ceram. Int. 1992,18,207-21 1. 160. V. Kevorkiijan, A. Bizjak, J. Vizintin, F. Thevenot, G. Interdonato, and C. Reimondi, B4CS i c based material for wear applications, in 4th Euro Ceramics Vol.4, Basic Science, A. Bellosi (Ed.), Faenza Editrice, 1995,pp. 209-216. 161. I. B. Cutler, P. D. Miller, W. Rafaniello, H. K. Park, D. P. Thompson, and K. H. Jack, Nature 1978,215,434-435. 162. W. Rafaniello, K. Cho, and A. V. Virkar, J. Muter. Sci. 1981, 16,3479-3488. 163. W. Rafaniello, M. R. Plichta, and A. V. Virkar, J . Am. Cerm. SOC.1983,66, 772-76. 164. R. Ruh and A. Zangvil, J . Am. Ceram. Soc. 1982,65,260-265. 165. Y. Xu,A. Zangvil, M. Landon, and F. Thevenot, J. Am. Ceram. Soc. 1992,15,325-333. 166. A. Zangvil and R. Ruh, Alloying of S i c with other ceramic compounds: A Review, in Silicon Carbide '87, J. D. Cawley and L. E. Semler (Eds), Am. Ceram. SOC.,Westerville, OH, 1989, pp. 63-82. 167. S. Y.Kuo, A. V. Virkar, and W. Rafaniello, J. Am. Ceram. SOC.1987,70, C-125. 168. R. R. Lee and W. C. Wei, Ceram. Eng. Sci. Proc. 1990, 11, 1094-1121. 169. T. B. Jackson, A. C. Hurford, S. L. Brunner, and R. A. Cutler, Sic-based-ceramics with

References

745

improved strength, in Silicon Carbide '87, J. W. Cawley and C. Semler (Eds), Am. Ceram. Soc., Columbus, OH, 1989, pp. 227-240. 170. J. L. Huang, A. C. Herford, R. A. Cutler, and A. V. Virkar, J . Muter. Sci. 1986,21, 1448-1456. 171. J . M. Lihrmann and J. Tirlocq, Process for producing dense products based on S i c and composite products thus obtained, World Patent, PCT, WO 97j06119, 1997. 172. D. P. Stinton, R. A. Lowdon, and R. H. Krabill, Mechanical property characterization of fiber-reinforced S i c matrix composites, ORNL/TM- 11524, April 1990. 173. H. Wurtinger and A. Miihlratzer, Cost effective manufacturing methods for structural ceramic matrix composite (CMC) components, A S M E Paper 96-GT-296, 1996. 174. E. Fitzer and R. Gadow, Fiber reinforced Sic, Am. Ceram. Soc. Bull. 1986, 65, 326335. 175. G. S. Corman, M. K. Brun, P. J. Meschter, K. L. Luthra, and R. Eldrid, Toughened silcomp ceramic composites for gas turbine applications, Proc. 39th Internat. S A M P E Symposium, April 1 1-1 3 1994, pp. 2300-23 13. 176. Ullmanns Encyclopedia of Technical Chemistry (E. Bartholome et al., eds.), 4th edn., 21, Verlag Chemie, Weinheim, 1982, 43 1438. 177. P. Wecht, Feuerfest-Siliciumcarbid, Applied Mineralogy, Vol. 11, Springer Verlag, New York, 1977. 178. K. Ziickler, in Halbleiterprobleme 111, W. Schottky (Ed.), 1956, pp. 207-229. 179. S. H. Hagen, Ber. Dtsch. Keram. Ges. 1970, 47, 630-634. 180. E. Gugel, P. Schuster, and G. Senftleben, Stahl Eisen 1972,92, 144-149. 181. Anon, JANAF Thermochemical Tables, 2nd edn., NSRDS-NBS37, Washington DC, June 1971. 182. K. Niihara, Am. Ceram. Soc. Bull. 1984,63, 1160-1164. 183. T. F. Page, Proc. Brit. Ceram. Soc. 1978, 26, 193-208. 184. W. Kollenberg, B. Mossner, and K. Schwetz, VDI-Berichte 1990, Nr. 804, 347-358. 185. M. E. Washburn and W. S. Coblenz, Ceram. Bull. 1988,67, 356-363. 186. F. Thummler, Sintering and high temperature properties of Si3N4 and Sic, in Sintering Processes, G. C. Kuczynski (Ed.), Plenum, New York, 1980, pp. 247-277. 187. J. Kriegesmann, A. Lipp, K. Reinmuth, and K. A. Schwetz, Strength and fracture toughness of SIC, in Ceramics for High Performance Appl. 111, E. M. Lenoe et al. (Eds), Plenum, New York, 1984, pp. 737-75 1. 188. R. Hamminger, R. G. Grathwohl, and F. Thummler, Microchemistry and high temperature properties of sintered Sic, Proc. 2nd Int. ConJ Science of Hard Materials, Rhodos Sept. 1984, Inst. of Physics Conf. Series No. 75, Hilger, Bristol, 1986, pp. 279-292. 189. G. Grathwohl, Th. Reets, and F. Thiimmler, Creep of hot-pressed and sintered S i c with different sintering additives, in Sci. Ceram. 1981, 11, 425431. 190. J. L. Chermant, R. Moussa, and F. Osterstock, Rev. Int. Hautes Temp. Refract. 1981,18,5-55. 191. G . Wiebke, Ber. Dtsch. Keram. Ges. 1960, 37, 219-226. 192. J. Schlichting, Ber. Dtscli. Keram. Ges. 1979, 56, 196199 and 256261. 193. J. Schlichting and K. Schwetz, High Temp. High Press. 1982, 14, 219-223. 194. B. Frisch, W. R. Thiele, R. Drumm, and B. Miinnich, Ber. Dtsch. Keram. Ges. 1988,65,277-284. 195. K. G. Nickel, Z. Fu, and P. Quirmbach, Trans A S M ;J. Eng. Gas Turb. Power 1993,115,7682. 196. K. Konopicky, I. Patzak, and H. Dohr, Glus Email Keramo. Tech. 1972, 23, 81-87. 197. D. W. McKee and D. Chatterji, J . Am. Ceram. Soc. 1976, 59,441444. 198. N. W. Jepps and T. F. Page, J . Microscopy 1981, 127, 227-237. 199. P. T. B. Shaffer, Ceram. Age 1966,82, 4244. 200. T. Iseki, T. Kameda, and T. Maruyama, J . Muter. Sci. 1984, 19, 1692-1698. 201. T. Labus, Lubr. Eng. 1981,37, 387-394. 202. J. H. Eisner, Chemie-Anlagen + Verfahren, 1982, p. 46, p. 51, p. 54. 203. D. W. Richerson, Muter. Sci. Res. 1981, 14, 661-676. 204. K. Miyoshi, Proc. Int. Con$ Wear of Mat., ASME, San Francisco 1981. 205. R. W. Lashway, S. G. Seshadri, and M. Srinivasan, Various forms of Sic and their effects on seal performance, 38th Ann. Mtg. of the ASLE, Houston, Texas, 1983. 206. H. Knoch, J. Kracker, and A. Schelken, Chemie-Anlagen + Verfahren, 1983, 28-30. 207. H. Knoch, J. Kracker, and A. Schelken, Cliemie-Anlagen + Veufahren, 1985, 101-104. 208. H. Knoch, J. Kracker, and A. Schelken, World Ceramm 1985, Vol. 2, 96-98.

746


209. J. Derby, S. G. Seshadri, and M. Srinivasan, Fracture Mechanics of Ceramics, Vol. 8, R. C. Bradt et a/. (Eds), Plenum, New York, 1986, pp. 113-125. 210. D. C. Cranmer, J. Muter. Sci. 1985, 20, 2029-2037. 211. K. Miyoshi, D. H. Buckley, and M. Srinivasan, Ceramic Bull. 1983, 62, 494500. 212. J. Breznak, E. Breval, and N. H. Macmillan, J . Mat. Sc 213. E. Breval, J. Breznak, and N. H. Macmillan, J . Mat. S 214. C. C. Wu, R. W. Rice, B. A. Platt, and S. Carrir, Ceram. Sci. Eng. Proc. 1985, 6, 1023. 215. J. R. Smythe and D. W. Richerson, Ceram. Sci. Eng. Proc. 1983, 4, 663473. 216. L. J. Lindberg and D. W. Richerson, Ceram. Sci. Eng. Proc. 1985, 6, 1059-1066. 217. H. Tomizawa and T. E. Fisher, A S L E Tans. 1987,30,4146. 218. H. Knoch and J. Kracker, cfi Dtsch. Keram. Ges. 1987,64, 159-163. 219. P. Boch, F. Platon, and G. Kapelski, J . Europ. Ceram. Soc. 1989, 5,223-228. 220. K. H. Habig and M. Woydt, in Proc. 5th Znt. Congress Tribology, Vol. 3, K. Holmberg and I. Nieminen (Eds), Lansi-Savo Oy, St Michel, Finland, 1989, 106. 221. Y. Yamamoto, K. Okamoto, and A. Ura, in Proc. 5th Znt. Congress Tribology, Vol. 3, K. Holmberg and I. Nieminen (Eds), Lansi-Savo Oy, St Michel, Finland, 1989, 138. 222. J. M. Martin, T. LeMogne, H. Montes, and N. N. Gardos, in Proc. 5th Znt. Congress Tribology, Vol. 3, K. Holmberg and I. Nieminen (Eds), Lansi-Savo Oy, St Michel, Finland, 1989, 132. 223. S. Sasaki, Wear 1989, 134, 185-200. 224. J. Denape and J. Lamon. J . Muter. Sci. 1990, 25, 3592-3604. 225. M. Woydt, A. Skopp, and R. Wasche, Proc. 4th Internat. Symp. Ceramic Materials and Components for Engines R. Carlsson eta/. (Eds), Elsevier, Amsterdam, 1992, pp. 1219-1239. 226. 0.0.Ajayi, A. Erdemir, R. H. Lee, and F. A. Nichols, J . Am. Ceram. Soc. 1993,76,511-517. 227. P. Maurin-Perrier, J. P. Farjandon, and M. Cartier, Wear Muter. 1991, 2, 585-588. 228. P. Maurin-Perrier, BRITE-EURAM Project Proposal P-2231, Contract No. RJ-lb-295, 1992. 229. N. N. Gardos, Determination of the tribological fundamentals of solid/lubricated ceramics, WRDC-TR-904096, Hughes Aircraft, El Segundo, CA 90245, 1990. 230. B. Loffelbein, M.Woydt, and K. H. Habig, Wear 1993, 162-164, 220-228. 231. P. Andersson and A. Blomberg, Wear 1994, 174, 1-7. 232. S. Kitaoka, T. Tsuji, T. Katoh, Y. Yamaguchi, and K. Kashiwagi, J . Am. Ceram. Soc. 1994, 77, 1851-1856. 233. H. Czichos and K.-H. Habig, Tribologie Handbuch Reibung und Verschleip, Vieweg Verlag, Braunschweig/Wiesbaden, 1992. 234. H. Knoch, L. Sigl, and W. D. Long, Product development with pressureless sintered Sic, 1 Oct. 1990, 37th Sagamore Army Mat. Res. Conf. AMTL-Watertown MA. 235. J. Greim, H. Thaler, and L. Sigl, Gesintertes Siliciumcarbidfur hochbeanspruchte Komponenten in der Lager- und Dichtungstechnik, in VDZ- Bericht 1331, Dusseldorf, 1997, pp. 153-159. 236. DIN 51 075 part 1-5, 1982, Beuth Verlag GmbH, Berlin/Germany. 237. I S 0 9286, 1997, Beuth Verlag GmbH, Berlin/Germany. 238. K. A. Schwetz and J. Hassler, J . Less Common Metals, 1986, 117, 7-15, and DIN 51079-3, 1997. 239. DIN 51 076, part 1, 1991, Beuth Verlag GmbH, Berlin/Germany. 240. F. F. van Dijen, Interceram 1993, 42, 92-94. 241. J. A. C. Broekdert, R. Brandt, F. LeiP, C. Pilger, D. Pollmann, P. Tschopel, and G. Tolg, J . Anal. Atom. Spectr. 1994,9, 1063-1070. 242. G. Zaray, F. Leis, T. Kantor, J. Hassler, and G. Tolg, Analysis of silicon carbide powder by ETV-ICP-AES, Fres. J . Anal. Chem. 1993,346, 1042-1046. 243. DIN-IS0 8486, part 1-2, 1997, Beuth Verlag GmbH, Berlin/Germany. 244. DIN 69 176, part 1-3, 1985. 245. FEPA 44-D-1986, Fachverband Elektrokorund-und Sic-Hersteller, Fr.a.M. 246. I S 0 787, part 11, 1991, Beuth Verlag GmbH, Berlin/Germany. 247. V. Carle, U. Schafer, U. Taffner, F. Predel, R. Telle, and G. Petzow, Ceramography of high performance ceramics, Part I1 silicon carbide, Pruct. Met. 1991, 28, 420-434. 248. W. Kollenberg, Prufverfahren keramischer Hochleistungswerkstoffe und deren Grundlagen, KfA-Bericht Nr. 470, Zentralbibiothek der Kernforschungsanlage Julich GmbH, 1988. 249. L. L. Hench and R. W. Gould, Characterization of Ceramics, Dekker, New York, 1971. -

747

References

250. 0. Van der Biest, Analysis of High Temperature Materials, Applied Science Publishers, London and New York, 1983. 25 1. H. J. Hunger, Werkstqffinalytische Verfahren, Deutscher Verlag fur Grundstoffindustrie, Leipzig/Stuttgart, 1995. 252. FEPA-ESK, S i c material safety data sheet, according to 9ljl55 EEC and ISO-Standard 11014, Edition 1, issued 22-05-97, 5 pages. 253. D. Henschler (Ed.), Gesundheitsschadliche Arbeitsstoffe, Toxikologisch-arbeitsmedizinische Begriindung von MAK-Werten, VCH, Wcinheim, 1987. 254. J. Bruch, Institut fur Hygiene und Arbeitsmedizin, Universitatsklinikum Essen, 1993. 255. I. D. Birchall et al., J . Muter. Sci. Lett. 1988, 7, 350-352. 256. ASTM, Standard Practice for Handling S i c whiskers, E 1437-91, Subcommittee E 34.70, American Society for Testing and Materials, Philadelphia, USA, I99 1. 257. L. Coes, Jr., Abrasives, in Applied Mineralogy, Vol. I , Springer Verlag, Wien, 1971. 258. K. Martin, Neue Erkenntnisse iiber den Werkstoffsabtragsvorgang beim Llppen, Fachberichte Oberflachentechnik 1972, 10, 197-202. 259. G. Spur and I. Sabotka, ZwF 1987,82, 275-380. 260. G. Spur and D. Simpfendorfer, ZwF 1988, 83, 207-212. 261. H. B. Britsch, Ber. Dtsch. Keram. Ges. 1976, 53, 143-149. 262. T. Benecke, S. Venkateswardn, W. D. Schubert, and B. Lux, GiePerei 1993,80, 256-662. 263. S. V. Nair, J. K. Tien, and R. C. Bates, Int. Met. Ref. 1985, 30, 275-290. 264. A. Lipp, Feinwerktechnik 1970,74, 150-154. 265. E. Gugel, Ber. Dtsch. Keram. Ges. 1966, 43, 354359. 266. A. F. Fickel, Sprechsaal 1980, 113, 517-531; 1980, 113, 737-747. 267. R. Rasch and H. Maatz, Maschinenschaden 1978, 51, 145-147. 268. Z. Stavric and M. Hue, Crystar, Keram. Z. 1975, 27, 125-128. 269. G . Bierbauer, Keram. Z . 1972, 24, 142-145. 270. 0. Rubisch and R. Schmitt, Ber. Dtsch. Keram. Ges. 1966, 43, 173-179. 271. E. Buchner and 0. Rubisch, Corrosion behavior of S i c heating elements, in Silicon Carbide 1973, R. C. Marshall et al. (Eds), Univ. South Carolina Press, Columbia, SC, 1974, pp. 428434. 272. Y. Nakamura and S. Yajima, Am. Ceram. Soc. Bull. 1982, 61, 572-573. 273. W. Huether, Thermoelement zur Temperaturmessung und Verfahren zur Herstellung desselben, European Patent 72 430, 1982. 274. K. Kuomoto, M. Shimohigashi, S. Takeda, and H. Yanagida, J. Muter. Sci. Lett. 1987, 6, 1453-1 455. 275. R. Davis, Recent advances regarding the definition of the atomic environment, film growth and microelectronic device development in Sic, in The Physics and Chemistry of’ Carbides, Nitrides and Borides, R. Freer (Ed.), Kluwer, Dordrecht, 1990, pp. 589-623. 276. M. Srinivasan, The silicon carbide family of structural ceramics, in Structural Ceramics Treatise on Materials Science and Technology, Vol. 29, J. B. Wachtman (Ed.), Academic, San Diego, CA, 1989, pp. 100-159. 277. D. Zeus, How the use of advanced ceramics as tribomaterial has effected the evolution of mechanical seals, cfi Ber. Dtsch. Keram. Ges. 1991, 68, 3-5. 278. V. D. Krstic, M. D. Vlajic, and R. A. Verrall, SIC ceramics for nuclear applications, in Advanced Ceramic Materials, H. Mostaghaci (Ed.), Key Eng. Muter. 1996, 122-124, 387-396. 279. F. Porz, G. Grathwohl, and R. Hamminger, J . Nucl. Muter. 1984, 124, 195-214. 280. G. Heuschmann and G. Willmann, Interceram. 1986, 1, 24-29. 281. D. P. Butt, R. E. Tressler, and K. E. Spear, Discontinuous phase formation and selective attack of S i c materials exposed to low oxygen partial pressure environments, in Corrosion of Advanced Ceramics, K. G. Nickel (Ed.), Kluwer, Dordrecht, 1994, pp. 153-164. 282. R. A. Penty and J. W. Bjerklie, S i c for high-temperature heat exchangers, in Ceram. Eng. Sci. Proc. 1982, 3, 120-127. 283. J. Heinrich, J. Huber, S. Foster, and P. Quell, Advanced ceramics as heat exchangers in domestic and industrial appliances, in High Tech Ceramics, P. Vincenzini (Ed.), Elsevier, Amsterdam, 1987, pp. 2427-2440. 284. M. C. Kerr, Advanced ceramic heat exchangers utilizing Hexology-SA, single phase S i c tubes, in High Tech Ceramics, P. Vincenzini (Ed.), Elsevier, Amsterdam, 1987, pp. 2441-2449. -

-

748


285. W. Heider, Verwendung von SiSiC im Apparatebau, Proc. Techkeram. '87, WiesbadenlFRG, S . Schnabel and J. Kriegesmann (Eds), Demat Exposition Managing, 1987, pp. 18.01-18.19. 286. W. Hof, Hexoloy S i c SA ein neuer Werkstoff fur Rohrbiindel-Waremetauscher, Chemi. Techn. 1991,20, 18-22. 287. B. Matchen, Applications of ceramics in armor products, in Advanced Ceramic Materials, H. Mostaghaci (Ed.), Key Eng. Mat. 1996, 122-124, 333-342. 288. R. S. Storm, R. W. Ohnsorg, and F. J. Frechette, Fabrication of injection molded sintered alpha-Sic turbine components, Trans. ASME: J. Eng. Power. 1982, 104, 76. 289. K. A. Schwetz, W. Grellner, K. Hunold, A. Lipp, and M. Langer, HIP-treated sintered Sic turbocharger rotors, in Proc. 2nd Intern. Symposium Ceramic Materials and Components for Engines, W. Bunk, H. Hausner (Eds), DKG Verlag, 1986, pp. 1051-1062. 290. R. W. Ohnsorg and M. 0. Ten Eyck, Fabrication of sintered a-Sic Turbine Engine Comp., in Silicon Carbide '87, J. D. Cawley and C. E. Semler (Eds), Ceram. Trans., Vol. 2, Am. Ceram. SOC.,Westerville, OH, pp. 367-386. 291. R. Westerheide, T. Hollstein, and K. A. Schwetz, Tension-compression testing of HIP-treated sintered Sic for gas turbine applications at temperatures between 1400 and 1600"C, Proc. 6th Intern. Symp. on Ceramic Materials and Components for Engines, October 1997, Arita, Japan, (K. Niihara et al., eds.), 253-258, 1998. 292. Ch. Gutmann, A. Schulz, and S. Wittig, A new approach for a low-cooled ceramic nozzle vane, ASME-Paper 96-GT-232, 1996. 293. M. Dilzer, Ch. Gutmann, A. Schulz, S. Wittig, Testing of a low-cooled ceramic nozzle vane under transient conditions, paper presented at 43th ASME Gas Turbine and Aeroengine Technical Congress, June 2-5 1998, Stockholm/Schweden, Paper No. 98-GT-116, ASME Atlanta/ Georgia. 294. G. Andrees, Entwicklung eines keram. Werkstoffes zur Auskleidung thermisch hochbeanspruchter Brennraume und Heipgasfiihrungen, Final Report, Project BMFT 03M-2028, MTU Motoren- und Turbinen-Union Miinchen GmbH, p. 114. 295. L. S. Sigl and K. A. Schwetz, Fabrication and properties of HIP-treated sintered S i c for combustor liners of stationary gas turbines, in Symposium 3, Werkstoffefiir die Energietechnik, Werkstoffwoche '98/Miinchen, A. Kranzmann and V. Gramberg (eds.), Wiley-VCH, Weinheim, NY,1999, pp. 15-24. -


6 Silicon Nitride Based Hard Materials M. Herrmann, H. Klemm, Chr. Schubert

6.1 Introduction Silicon nitride ceramic inaterials have been intensively studied for many years because of their great potential for use in structural applications at room and high temperatures. This is due to their excellent mechanical properties in combination with good corrosion and thermal shock resistance. Over the past decade a continuous increase in the number of application fields has been observed. Besides parts for the automotive industry (cam rollers, valve plates for coninion rail systems). which have not been produced in the volume that was predicted 10 years ago, applications such as cutting tools. ball bearings, wear parts. applications at high temperatures, and application in the electronic industry are becoming increasingly important [1-4]. A new cooking system based on a silicon nitride plate is penetrating into the market at the moment [51. This wide range of applications leads to an increase in the variety of materials with different microstructures (see Fig. I ), making precise control of the microstructure and propertics of the materials necessary (Table 1). The materials differ in the production technology used to make them, their phase content and properties. The different production technologies used are summarised in Table 2. A inore detailed overview is given in [ l . 61. The reaction-bonded materials (RBSN) produced by the nitridation of compacts of silicon powder (see Table 2) are not dense. whereas the other kinds of materials in the table are dense. The range of applications for RBSN and sintered reaction-bonded silicon nitride has shrunk over the last few years. This is connected with the decreasing cost of silicon nitride powders. reducing the production costs of SSN and the lower strength and lifetime of RBSN in comparison to SSN in inany applications. For this reason, these materials will not be explained in detail in this work; reference is made to special reviews [ 6 , 7 ] . The following paragraphs concentrate on dense silicon nitride materials made from silicon nitride powders. The most common sintering technology used for silicon nitride materials is gaspressure sintering. This technology lcads to improved reliability and strength in comparison to norinal gas-pressure sintering and is accompanied by only a moderate iiicrease in cost. The HIP processes are only used for special applications (e.g. balls for ball bearings) due to the high cost of these technologies. Hot pressing is used for evaluation of materials or for some special applications requiring simple geometries and only ;I very limited number of parts. Besides the different production technologies, the materials differ in their composition and microstructure (see Fig. I ) . The wide variation of the microstructure and

750

6 Silicon Nitride Based Hard Materials

Figure 1. Microstructures of different gas pressure sintered dense Si,N4 materials.

6.1 Introduction

75 1

Table 1. Overview of the microstructure-property relationships for dense Si,N, materials. Property

high strength up to 1000°C high strength at T > 1200°C high fracture toughness high hardness high fatigue strength (cyclic mechanical load) high heat conductivity

Microstructure grain sizejshape

grain-boundary phase

fine-grained/needle-like grains

median content

fine-grainedjneedle-like grains

AI2O3-freewith special compositions

large, needle-like grains, or large, needle-like grains in a fine matrix fine-grained or d-SiAION fine-grained/needle-like grains

low AI2O3and SiO, content

large grains

no components which can be incorporated into the silicon nitride crystal lattice, nitrides (e.g. Al, Be), low impurity content A1203-freewith special compositions or no sintering additives AI2O3-freewith special compositions or no sintering additives special compositions depending on the corrosive media

high creep resistance

large, needle-like grains, composiles with S i c or refractory silicides

high oxidation resistance at T > 1200°C high corrosion resistance

large, needle-like grains, composiles with S i c or refractory silicides

good wear behaviour

fine-grained microstructure

low additive content

homogeneous distribution

phase content, which is possible in all of the above-mentioned production technologies, is the reason for the wide variation of material properties produced by each of these production technologies. Dense silicon nitride materials are not monophase materials because, for densification, additives, which remain as an amorphous or partially crystallised grain-boundary phase in the material, are necessary (see Section 6.3). This grain-boundary phase has both positive and negative consequences for the material properties (see Section 6.5). The softening of the grain boundary has a negative influence on the high-temperature properties, but on the other hand the grain-boundary phase is a main factor causing the high strength and fracture toughness of Si3N4. The liquid-phase-sintering process and the wide range of possible additives make the tailoring of the microstructure and the adaptation of the microstructure to a given application possible. The aim of this paper is to show how the different microstructures and compositions influence the properties at room and elevated temperatures.

metallic Si

Starting components Heat treatment

low strength (1 50-350 MPa)

refractories

Main disadvantages

Applications

Relative density 7&88% Main advantages no shrinkage

1250-1450°C up to 100 h

Reaction bonded silicon nitride (RBSN)

Material type

Sintered silicon nitride (SSN)

the same as SSN/GPSN

expensive nitridation

Lower strenght than GPSN, good sinterabillity necessary wear parts

Si3N4,+ additives nitridation as for 170&I80O0C RBSN, Sintering as for GPSN 95-99% 95-99% low shrinkage, low sintering low cost raw cost, different shapes material possible

Si + additives

Sintered RBSN (SRBSN)

Table 2. Different production technologies and resulting materials.

cutting tools, wear parts, ball bearings, seals, valves, turbo charger rotor

Si3N4 + additives 1750-2000°C nitrogen pressure up to 10 MPa 98-100% different shapes possible, better reliability than SSN; lower additive content can be used

Gas pressure sintered silicon nitride (GPSN) Si3N4, + additives 1750-2000°C gas pressure up to 200 MPa 100 % different shapes possible, sintering of materials with low additive content (low sinterability), high reliability high sintering costs

Sinter HIP silicon nitride (Sinter-HIPSN)

mainly used for balls for ball evaluation of bearings, wear materials, parts prototypes with simple geometries

only simple shapes, low productivity

Si3N4 additives 150&I80O0C uniaxial pressure in a graphite die 100% good densification, high reliability,

+

Hot-pressed silicon nitride (HPSN)

used for materials with very low additive content or without additives

high costs for encapsulation

Si3N4 + additives 1750-2000°C gas pressure up to 200 MPd >99% Materials without additives can be produced

HIP-ed silicon nitride HIP-SN

6.2 Crystal Structure and Properties qf the SijN, Modifications

753

6.2 Crystal Structure and Properties of the Si3N4 Modifications Silicon nitride has three modifications. The a and p modifications can be produced under normal nitrogen pressure. Recently a high-pressure modification was produced under 15 GPa pressure by the technique of laser heating in a diamond cell [8] (see Table 3 ) . The a-phase is metastable under sintering conditions (e.g. at 1400-2000°C and 0.1-100 MPaN2), but is the main phase in the starting silicon nitride powders. The a and p modifications are based on SiN4 tetrahedra connected at the corners. Every N belongs to 3 tetrahedra, i.e. the structure is made up of a three-dimensional network of tetrahedra. Only one layer of SiN4 tetrahedra exists in p-Si3N4 (see Fig. 2), whereas two layers shifted with respect to each other exist in a-Si3N4.This leads to a doubling of the c-axes in the a-Si3N4crystal lattice in comparison to the p-Si3N4 lattice. The high-pressure phase has a completely different structure. On the basis of powder diffraction patterns, it is found that this phase has a spinel-type structure, in which one silicon atom is coordinated by four nitrogen and two silicon atoms by six nitrogen atoms (octahedra; see Fig. 2d). The Si and N atoms in p-Si3N4can be replaced by Al and 0 atoms to form the socalled p’-SiAlONs with the formula: Si6-ZAl,N8-,0,. The range of the p’ solid solution extends from z = 0 to 4.2 at 1750°C [I, 31, i.e. it includes pure Si3N4 (see Fig. 3 ) . That is why a clear distinction between SSN and P-SiAlON materials cannot be made. Every Si3N4 material containing A1203 as a sintering additive (i.e. nearly every commercial material) is some kind of p-SiAlON. Most commonly the terminology P-SiAlON is used for z values >0.5. A similar solid solution can be Table 3. Crystal structures of the modification of the Si,N, modification and the resulting materials names. Property

Space group Lattice parameter a, nm c, nm Hardness Coefficient of thermal expansion 0-1000”C, 10-6/K materials based on the modification

Modifications P-Si3N4

cc-Si,N,

high pressure modification [8]

P63:m No. 176 [9, 101

P31c No. 159 [9]

Fd3, or Fd3 m

0.7586 0.2902 (1 00) plane: 2 100 (001) plane: 1326 HV 0.025 [12] 3.39 1131

0.766 0.5615 (1 10) plane: 2250 (001) plane: 2200 HV 0.300 [ 1 11 3.64 [I31

0.7738

nearly all SSN-, GPSN-materials,

cc’-SiAION cc’/P’-SiAION

P'-Si AlON

20-40 GPa

?

754


SI (OCt)

SI (tetr).

c7 b

(4

a

Figure 2. Crystal structures of the Si3N4modifications

formed with the addition of BeO/BeN [13]; however, materials made from these solid solutions have no practical use due to the toxicity of Be. Analysis of the a-Si3N4crystal structure shows that there is an empty position with a coordination number of eight (7 1). This position can be partially occupied

+

6.3 DensiJicntion

755

2iR207) A /

/ \‘

i

/

/

_’

3(Si02)e

‘,

\

4/3(AIN A1,03)

,,

>2(A1203)

‘

-

\ \ \ \

‘4

3/2(Sl*N20) ‘\

i

S-SlAlON

Si3N4

I I

Ct-SiAION

I

SALONPolytypolds

Figure 3. Stability range of a’- and 0-SiAIONs in the system Y -Si-AI-O-N (see also Section 6 4.2 [ I , 14,151)

by ions with an atomic radius of about 0.1 nm. Additionally, Si and N must be replaced by A1 and 0 to obtain electroneutrality. The resulting phase is the socalled a-SiAION with the formula: R,Si12-n-mAlm+nN16-nOn. Occupation of the (7 1)-coordinated position leads to stabilisation of the metastable a-Si3N4 phase. The lowest x value for trivalent cations is 0.33, for Ca 0.3 [14-161. This means that the a’-SiAlON solid solution does not include the composition of pure a- Si3N4. The different kinds of the materials based on the different modifications are given in Table 3 . The hardness values of the crystals are given in Table 3. These data show that the p modifications have much lower hardnesses than the a modifications. N o data for the chemical stability of the high-pressure modification have been published up to now. First measurements of the microhardness show that this structure must have a high hardness which is similar to that of diamond and c-BN. The microhardness was determined for the different crystal planes of B-Si3N4.The data differ widely in different orientations.

+

6.3 Densification The high energy of the covalent chemical bonds in the Si3N4 crystals, which is the basis for the excellent properties, presents a disadvantage in the fabrication of the materials. The self-diffusuion coefficient is very low in comparison to other

756


ceramic materials. The mass transport necessary for sintering can only be achieved at temperatures where the decomposition of silicon nitride occurs to a great extent even at high pressure (i.e. >2O0O0C). Dense materials without sintering additives can only be produced by encapsulation-HIP processes. The green bodies which are thus formed are sealed in a dense glass container and densified at 1800-1900°C at 2000 bar [l]. The resulting material consists of Si3N4,with some Si02 at the grain boundary. These materials have good oxidation and creep resistance (see Section 6.5.2) but low fracture toughness and strength (see Section 6.5.1). Usually the densification of Si3N4 is achieved through the acceleration of mass transport by the formation of a liquid phase during sintering. In the liquid-phasesintering process, silicon nitride can be sintered in a temperature range of 15501900°C. The liquid phase used for sintering is an oxynitride liquid formed by the reaction of the sintering additives with the Si02 existing on the surface of the Si3N4powder particles. The most commonly used additives are A1203 in combination with Y203, La203, lanthanoids or MgO. For high-temperature applications, pure rare-earth additives are used. AlN additives are necessary for the production of a’-or P’-SiAlON. Under sintering conditions stable oxides or nitrides, which form a liquid phase with Si3N4 and Si02 during sintering, can be used as sintering additives. For example, alkalis and alkaline earths beside MgO can be used as sintering additives. The problem in using these elements is their high vapour pressure under sintering conditions. Besides Y203,Sc203and La203,the oxides of the d-elements, Zr02 or HfO,, can be used as sintering aids. Under sintering conditions, Ti02 reduces to TIN; hence Ti02 can only accelerate the sintering during the initial stage of sintering. The delements in the 51hto gth group form silicides under sintering conditions, forming separate inclusions. The amount of liquid phase formed and the resulting grain-boundary phases can be predicted by the phase diagrams (see Fig. 3). Detailed information about the phase diagrams are given in [l, 3,13,17]. Densification is improved when the liquid is formed at low temperatures and has a low viscosity. For this reason, the densification behaviour worsens in the order MgO/A1203< MgO/R203;MgO; R203/A1203 << R 2 0 3(R = Y, Sc, La and lanthanoids). The mechanisms of liquid-phase sintering of ceramic materials are explained in detail in [19]. Figure 4 shows typical shrinkage curves of Si3N4 materials with different amounts of liquid phase of the same composition. Densification starts at 1200-1300°C. At these temperatures, the chemical reaction involving the additives starts and the formation of the liquid is initiated. The first maximum in the densification rate is connected with the rearrangement of the Si3N4 particles. At 1500-18OO0C, densification by the solution/diffusion/precipitation mechanism takes places. The phase transition of a-Si3N4 to P-Si3N4 occurs simultaneously. In [18] it was shown that the a/P-Si3N4 ratio has practically no influence on the densification when the grain sizes of the a and P phases are the same. This is an indication that the solubilities of the a- and P-Si3N4 modifications in the liquid do not differ significantly [20].

6.3 DensiJication

151

dLILo, 9 21.0 18.0

15.0 12.0 9.0 6.0

3.0 0.0 1200 1300 1400 1500 1600 1700 6 12 18 Temperature, "C Time, min

dLIdt, %bin

1200 1300 1400 1500 1600 1700 6 12 18 Temperature, "C Time, min

Figure 4. Typical densification curves for Si3N4 materials (1-3.5 vol.% additives, 2-6 vol.% additives; 3-8.5 vol.% additives; AI2O3/YzO3weight ratio 1 : 2).

The sintering behaviour is improved with increasing additive amount (see Fig. 4). However, for additive contents higher than 15-20 vol.%, formation of gas bubbles hinders densification. The bubbling is connected with the low solubility rate of Si3N4 and the increased formation of gaseous SiO [21,22 1. After sintering, the additives are situated at the grain boundary (3-20 vol.%). The majority of the grain-boundary phase is concentrated in the triple junctions. The thickness of the grain boundary films between two grains is in the range of 0.81.5 nm [41]. The grain-boundary phase significantly influences the material properties. Therefore, one usually attempts to reduce the amount of additives. However, a lower additive content also leads to a lower sinterability, a lower reliability and a broader distribution of the properties. Thus for materials for room-temperature applications

758


'f

LT

/\ ........................... ...............

1

I

P

'dp dt

t

Figure 5. Time pressure sintering regime for gas-pressure sintering

(e. g. wear parts), the additive content usually lies between 6 and 15 vol.%, representing a compromise between reliability and properties. For high-temperature applications, special refractory grain boundary compositions are used (see Section 6.5.2). The most common method used for sintering silicon nitride materials is gaspressure sintering. The principle of gas-pressure sintering is the same as that of the sinter-HIP process (see Fig. 5). The material is sintered to closed porosity ( > 95% theoretical density) at a low gas pressure. Then an outer gas pressure of up to 10 MPa (GPS: gas-pressure sintering) or 100-200 MPa (sinter-HIP) is applied. The difference between the outer pressure and the pressure in the closed pores is an additional driving force for densification, leading to a better healing of defects. The increase in cost for the GPS process in comparison to sintering under 1 atm (SSN) is relatively low. In many cases these costs can be offset by the better reliability and properties of the materials.

6.4 Microstructural Development 6.4.1 Microstructural development of P-Si3N4materials The adaptation of silicon nitride materials to various applications leads to an increase in the variety of microstructures (see Fig. 1). This was the impetus for

6.4 Microstructural Development

759

the large number of investigations into the mechanisms of microstructural formation and controlling of the same in recent years. The most important commercial silicon nitride powders have u-Si3N4, i.e. the phase which is metastable at high temperatures, as the main constituent [ 11. The consequence of this is that during densification, a phase transformation takes place, having a great influence on the subsequent formation of the microstructure. It was shown that high fracture toughness and strength can be achieved when the grains have a needle-like shape. Characterisation of the shape is usually done using two parameters: the thickness of the grains and the aspect ratio, i.e. the ratio of needle length to thickness [23-261. For this reason the influence of different parameters such as additive content, grain size of the powder, p- and oxygen content on the sintering and microstructural formation was investigated [2439]. In basic works from the 1980s [26-291, it was formulated that a needle-like microstructure, which is necessary for high fracture toughness, can only be achieved using starting powders which contain a high amount of ct-SijN4. The reason for this was suggested to be the needle-like grain growth of a limited number of p nuclei [28] or the homogeneous nucleation of P-Si3N4 [26,27]. Later it was shown that needle-like microstructures can also be produced from fine p-Si3N4 powders [30,32-341. Despite intensive research, no direct evidence of homogeneous nucleation in the oxynitride liquid or heterogeneous nucleation of P-Si3N4 on a-Si3N4could be found [30,36-381. The only process of epitaxial growth of one modification on the other was found in SiAlON systems with a high degree of substitution [40]. The lattice constants differ less in these materials than in pure a/P-Si3N4 due to the incorporation of other ions. The reason for the absence of formation of P-Si3N4nuclei during sintering seems to be connected with the minimal differences in the free energies of formation of pand a-Si3N4, leading to a quite high critical size of nuclei which can grow (60600nm [38]). Thus the critical nucleation size is the same or even larger than the grain size of the P-Si3N4 crystallites in the powder. In recent works, it could be shown experimentally that the addition of small P-Si3N4 nuclei to a commercial u-Si3N4powder does not result in a change in the microstructure, whereas the addition of P-Si3N4nuclei of the same size as the nuclei existing in the starting a-Si3N4 powder alters the microstructure significantly (see Fig. 6). The result is a fine structure which is very similar to the structure made from the P-Si3N4powder. In [20], it was shown that at low sintering times and temperatures, the microstructure correlates with the crystallite size in the starting powder. This provides indirect evidence of the absence of nucleation during sintering. The measurement of the p crystallite size in materials sintered at different temperatures shows a continuous increase in the p crystallite size, indicating that no nucleation takes place (see Fig. 7). A detailed analysis of different microstructures [29-3 1,331 shows that they can be explained by the number of [3 nuclei in the starting powder, even at very low P-Si3N4 contents. In Fig. 8 the calculated volume fractions of growing nuclei which are necessary in the starting composition to produce two different fine microstructures in the sintered materials are given. This calculated content is the lowest (j-Si3N4content in the starting powder required to explain the microstructure. For crystallite sizes between 50 and 100 nm [20,35], the necessary amount of the p fraction in

760


Grain thickness in the material, pm

Figure 6. Change in the grain size distribution of materials with different p-Si3N4 amounts and crystallite sizes (material 1 produced from cc-Si3N4 powder with 4% S-Si3N4;crystallite size pSi3N4 46nm; material 2 produced from a mixture of 60% cc-Si3N4 powder and 40% p-Si3N4 powder (the p-powder contain 33% p-Si3N4; crystallite size 18nm); material 3 made from 60% a-Si3N4powder 40% p-Si3N4 powder (the p-powder contain 80% p-Si3N4;crystallite size 55 nm); material 4 produced from p-Si3N4 powder (crystallite size 18 nm) (according to [62]; corrected device function of the XRD-equipment).

the starting powder must be in the range >0.5-1%. This is the case for commercially available a-Si3N4 powders. In [39], it was found that at normal heating rates the nitrogen content in the oxynitride liquid during heating is lower than at low heating rates and low additive contents. This means that for normal heating rates, the liquid phase is undersaturated with respect to (j-Si3N4and no nucleation can take place. These experimental data suggest that (j-Si3N4nucleation has no influence on the microstructure of silicon nitride materials under common sintering conditions and using common powders. The consequence of this is that the different microstructures are a result of the different amounts and sizes of pre-existing p nuclei as well as the growing conditions. The amount of growing (j-Si3N4nuclei depends on the (j content in the starting powder, the crystallite size and the sintering conditions. In Fig. 9 the dependence of the equilibrium concentration in the oxynitride liquid during sintering on the

:I

0

vl

53

c

6.4 Microstructurul Development

,

;J:, /

/+ __.--

10

0 1200

76 1

7

1300

1400

1500

1600

1700

Temperature, "C

100 j

3

E

C

'1

20 i

1200

1300

1400

1500

1600

1700

Temperature, "C

Figure 7. Change in crystallite size and phase content as a function of sintering temperature, amount of sintering additives and starting powder 1. diimid powder 5% Y2O,/Al2O3; 2. gas phase synthesised powder 5% Y2O3/Al2O3;3. gas phase synthesised powder 11.6% Y2O3/AI2O3.

crystallite size of a- and P-Si3N4is shown. Nuclei with a crystallite size lower than a critical value have a higher equilibrium concentration than the Si3N4concentration in the liquid. These crystals dissolve in the liquid and cannot growth. During the a/P transformation under isothermal conditions, the critical size of P-Si3N4 is determined by the crystallite size of a-Si3N4(r(crtl), see Fig. 9). During heating, especially for high heating rates, the oxynitride liquid is undersaturated (see Fig. 9, lines 2 and 3). As a result of the undersaturation, the critical crystallite size becomes infinitely large and only the largest nuclei survive because they require more time to dissolve than smaller nuclei. The resulting microstructure of materials heated at high heating rates in the temperature range of 1200-1600°C is coarser than for materials heated at low heating rates [20,29,39]. After the alp transformation, the critical crystallite size is determined by the crystallite size distribution. The critical crystallite size increases with increasing sintering time and temperature, i.e. only the largest crystallites which were present in the starting powder determine the microstructure.

162


Crystallite size, prn

Figure 8. Calculated volume content of growing nuclei as a function of the size of the nuclei, if the resulting microstructure has a mean grain thickness of 0.2 pm (I) and 0.3 (2) and an aspect ratio of 5.

The needle-like grain shape seems to be kinetically determined and therefore not the equilibrium crystal shape [30,38]. The estimated equilibrium aspect ratio (ratio of grain length to thickness) is 1.3, not as high as 20, as was found to be the case for the silicon nitride materials. The microstructural formation must therefore be explained on the basis of growth mechanisms. The explanations for the microstructure given in the ’80s were based on the assumption that needle-like grain growth can take place only during the ct/p transformation. Later it could be shown that anisotropic grain growth could also take place after the a l p conversion [31,32,42] (see Fig. 10).

low heating rate (2)

----------r( crit3) = =

high heating rate (3) -

in the liquid 0.6 0

0,025 0.05 0,075

0.1

0,125 0,15 0,175

0,2

Crystallite size, pm Figure 9. Dependence of the equilibrium concentration in the oxynitride liquid during sintering on the crystallite size of a- and P-Si3N4,according to [63].

6.4 Microstvuctuvul Developnzent

763

Figure 10. Microstructure of a material hot pressed at 1800°C (starting powder: fine P-Si3N4)and the same material after additionally heat treatment at 1900°C. It is observed that anisotropic grain growth takes place after a/P conversion.

Anisotropic grain growth occurs due to the different growth mechanisms in the different directions of the B-Si3N4 crystals. The (001) basal planes of the needlelike (j-Si3N4 crystals are atomically rough; therefore, their growth is diffusioncontrolled [4244]. Transport occurs by two mechanisms: diffusion through the oxynitride liquid and surface diffusion in the (100) planes [44]. The (100) prismatic planes are atomically flat and therefore the growth is interface-controlled. The growth rate of the (001) plane has been assigned values of 30 [38] and greater than 27 [35]; thus, it is at least ten [42] times higher than the growth rate of the (100) plane. The first value seems to be the more exact one because it was determined by growth observation on crystals in oxynitride glasses. The different growth mechanisms are supported by TEM observations [43,44], theoretical calculations [38] and by the observed crystal habits [4547]. The relationship between growth rate and oversaturation is shown schematically in Fig. 11. For a high degree of oversaturation, the (1 00) planes become atomically rougher, causing the growth rates to approach the rate of the other planes, the growth to be isotropic and the grain morphology to be equiaxed. At the lower saturations found in common sintering conditions, extended anisotropic grain growth occurs. The boundary between the two regimes shifts to lower oversaturations with increasing temperature. For a given crystal, the oversaturation depends on the size and is different for the different crystal planes. Using the differences in surface energy calculated by the broken bond method [38], the oversaturation as a function of grain thickness is given in Fig. 12. The equilibrium concentration above the basal planes depends only on the thickness of the grains, whereas the equilibrium concentration above the (100) planes depends on both the thickness and the length of the grains. Using these data and the dependencies of the growth rate on the oversaturation, it can be shown that the largest crystals with the highest aspect ratios grow anisotropically, whereas the small crystals with a low aspect ratio dissolve. Thin crystals with a high aspect ratio grow stably whereas crystals of the same thickness but a low aspect ratio dissolve. This behaviour was also determined experimentally [30,48]. At first these results and the growth mechanisms discussed seem to be in contradiction to the results of the works from the 1980s showing a reduction in the aspect

164


*

region of anisotropic grain growth

.-.-

.-.-

region of isotropic,/' grain growth // /

, /

/

//

A

,

0

c

e

$e

I

!

W

/

1

i

i

/interface controlled Oversaturation o(1OOyD;

o(100YD + o(oo1)n

Figure 11. Growth rate of the basal plane (diffusion-controlled) and the prismatic planes (interfacecontrolled) as a function of oversaturation in the oxinitride liquid. The oversaturation after the alptransformation depends on the surface energy (o), the grain thickness (2D) and the length of the needle like grains (2L).

1.1:

(C(100) A-I 1.1

0

Y 1.05

1

Grain thickness, pm

Figure 12. Calculated dependence of the oversaturation of the basal (001) plane and the prismatic (100) planes on the thickness of the grains, according to [30] using data [38,63] (A = aspect ratio).


765

ratio with increasing p-Si3N4content and at longer soaking times. These differences can be explained by taking the crystallite size of the P-Si3N4in the starting powder into consideration. The powders used for these investigations were produced by direct nitridation of silicon. The higher p content of these powders was a result of the faster nitridation and larger crystallite size. The larger crystallite size has two consequences: 1. the oversaturation, which is inversely proportional to the grain size, decreases and consequently the grain growth rate also decreases; and 2. the thicker crystallites must undergo more growth to become needle-like. Both processes lead to a reduced aspect ratio. Additionally, the powders had a wide grain size distribution. With increasing sintering time, the critical crystallite size (see Fig. 9) increases and only the largest crystals survive. However, the largest crystallites had lower aspect ratios. The aspect ratio is the result of two competing processes: anisotropic grain growth (preferred growth in the direction of the needle), and the dissolution of small elongated grains. When the second process is the dominating process, the overall aspect ratio decreases. An additional effect influencing the different microstructures which are formed is connected with the different grain sizes of the materials. The oversaturation depends on the mean grain size. With increasing grain size, the curve characterising the dependence of the oversaturation on grain size is flatter (see Fig. 9); for a given mean grain size, the oversaturation cannot compensate for the difference in surface energy between the (001) and (100) planes. As a result, anisotropic grain growth cannot take place and the grain shape approaches the more equiaxed equilibrium grain shape. Based on the investigations of microstructural formation, it can be concluded that these processes are only significant for mean grain thicknesses of 0.7-1.5 pm or larger. The materials investigated in the 1970s and early 1980s were characterised by mean grain thicknesses of 0.7-2 pm, whereas today’s materials are based on finer powders having normally finer grain sizes. Recent investigations [49-521 show that needle-like microstructures can be achieved from p-Si3N4 powders when the crystallite size of the starting powder is small enough or the sintering temperature and soaking time are large enough. Materials with high fracture toughnesses of up to 10-12 MPam1/2[48] are observed. Low-cost powders with higher P-Si3N4 contents, produced by special nitridation processes which allow the exothermic nitridation to be controlled [53], have relatively small crystallite sizes. Materials with strengths higher than 800 MPa and fracture toughnesses similar to materials based on a-Si3N4powders can be produced from these powders [49]. At high sintering temperatures, abnormal grain growth can take place [35,54,55]. The reason for this is the presence of large p-Si3N4 nuclei (seeds) which can exist in the starting powder, formed during densification [35,56,62] or added as seeds [57-591. The formation of large seeds during sintering can be intensified by low heating rates or dwell times at temperatures higher than 1500°C but lower than the temperature at which rapid densification occurs [56]. The amount of additional seeds needed to produce a reinforced material ranges from 1-5 or 10 vol.%. The seeds are usually between 10 and 100 times larger than the nuclei in the starting powder [56-58, 60, 611. The addition of larger seeds is intended to increase the fracture toughness without drastic decreasing the strength (see Section 6.5.1).

166

6 Silicon Nitride Based Hard Materidv

The mechanism of accelerated grain growth can be explained by adapting the ideas developed for metals and hard materials [4546]. According to [45], abnormal grain growth can take place under sintering conditions when the grain growth is interface-controlled (two-dimensional nuclei on an atomically flat suface) and the degree of supersaturation is high, sufficient to realise accelerated growth. The region of necessary oversaturation is given in Fig. 11. This grain growth is anisotropic and leads to the formation of large P-Si3N4 needles. With increasing temperature, the probability of formation of two-dimensional nuclei on the (100) surface (rate determine step of the growth) at a given oversaturation increases. As a result, the accelerated grain growth is more pronounced at higher sintering temperatures than at lower temperatures. The oversaturation above a large nucleus depends on the mean grain size of the matrix (see Fig. 12); therefore, abnormal grain growth occurs to a greater extent in materials with a fine matrix. When the density of large nuclei is too high, the mean oversaturation is reduced and no abnormal grain growth takes place. Thus the addition of more than 1&15 vol.% of large p-Si3N4 seeds to a starting a-Si3N4 powder does not result in a bimodal microstructure but in a coarse microstructure consisting of large grains with low aspect ratios. If the seeds added have the same size as the crystallites in the starting powder, the resulting microstructure consists of fine P-Si3N4grains. If the size is smaller than in the starting powder, there is no change in the microstructure. The microstructural changes resulting from the addition of P-Si3N4 seeds are shown schematically in Fig. 13. The microstructural formation depends on the

mean crystallite size in the starting powder (r,,)

1

10

50

100

amount of added seeds, Vol %

Figure 13. Microstructural changes resulting from the addition of P-Si3N4seeds (schematically).


767

density of p nuclei. The density is proportional to the ratio of amount of p-Si3N4 to the mean crystallite size. Fig. 13 is based on the crystallite size of the common aSi3N4 powders, which lies in the range of 50 to l00nm. A change from one type of microstructure to an other can be achieved by changing the sintering conditions. To example a fine grained homogeneous microstructure can be changed in a bimodal microstructure by high temperature sintering (Fig. 10). The scheme shown in Fig. 13 can be used for the design of the microstructure in p-Si3N4. The use of p powders with small grain size distributions allows the microstructure of the P-Si3N4 materials to be controlled more precisely. Presently such fine P-Si3N4 powders are only available in limited quantities and are not widely used [63]. By hot pressing or using special shaping technologies (tape casting), an anisotropic orientation of the needle-like grains can be achieved. This results in anisotropy of the fracture toughness, strength and thermal conductivity. The microstructure, i.e. grain thickness and aspect ratio, depends on the type and amount of sintering additives used. The influence of additive type on the morphology was determined from a large number of investigations; only some important tendencies can be given as a result of these investigations. With increasing A1203 content and constant overall additive content, the Si3N4 grains become finer and more equiaxed. [24,26-27,641. At the same time, partial incorporation of A1 in the Si3N4 grains occurs. The incorporation of A1 in the grains can be expressed by the reaction: 0 . 5 ~A1203+ (2 - 0.25z)Si3N4@ Si6p_Al,N8-,0,

+ 0.252 Si02

This process leads to a reduction of the A1203 content in the grain-boundary phase and to an increase in the Si02 content; i.e. AI2O3 functions as a buffer of the grain-boundary phase composition. The more Si02 evaporates during sintering, the more A1 dissolves in the silicon nitride grains and reproduces Si02 [64]. The incorporation takes place according to the local equilibrium in the liquid. The diffusion in the silicon nitride grains is slow. Therefore A1 gradients in the silicon nitride grains can be observed [64,67]. The incorporation of A1 in the silicon nitride crystal lattice leads to a reduction of the thermal conductivity of the materials [65,66]. This means that the A1203/A1N content in the sintering additives can be used for the tailoring of the thermal conductivity; values between 10 and 150 W/mK are possible. Investigations of the growth of the p-Si3N4 grains in oxynitride liquids with different rare-earth additives show that with decreasing melt viscosity (Yb to La), the aspect ratio of the grains increases and the thickness of the grains decreases. This was found in materials with rare earths and alumina as sintering additives and in materials with only rare-earth additives [68-691. Very fine, needle-like microstructures were obtained for Sc203-containingadditives. [70]. CaO and MgO were found to accelerate the densification and the anisotropic grain growth [71,72], but were found to be detrimental to the high-temperature properties. The use of rareearth additives or mixtures with MgO results in the formation of microstructures consisting of grains with high aspect ratios [ 13.

168


6.4.2 Microstructural development of a’-SALON materials a’-SiALONs are formed during liquid-phase sintering by the reaction of Si3N4, AlN, A1203 and an appropriate cation which can enter the structure of a’SiALON (R,Si12-n-mAlmtnN16-nOn). The amount of liquid available for densification is quickly reduced due to the formation of a-SiAlON solid solutions (Fig. 14). This reaction starts during heating at temperatures above 1450°C 4/3 (AINxA1203)

4

Si3N4

m=3

m=2

RN x3A1N

4/3 (AINxA1203)

area of elongated grain growth R203 x 9A1N

7

Si3N4

m=1

m=2

m=3

RN x3AlN

b) Figure 14. Stability area of the a’-SiAlON solid solution R,Si,,_,~_,A1,+,N16_nO~ a) in the systems with R = Nd, Sm, Dy, Y and Yb at 1800-1900°C (after [14]) and b) in the area where elongated grain growth takes place in the a’-SiAlON/P’-SiAlON-plane(after [14,15,77,79]).


769

[73,74]. When the a’-SiAlON formation is completed, no more liquid exists for densification. This is the reason why pure a’-SiAlON materials with low degrees of substitution n and m are difficult to densify to full density. Better densification can be realised at higher n and m values (Fig. 14) [75-771 or in composite dip’SiALON materials [78]. The resultant materials had microstructures consisting of equiaxed grains, leading to reduced fracture toughness and strength. This was a reason for the limited use of these materials [I]. In the past few years materials with elongated grains have been produced by different research groups. The materials with elongated a-SiAION-grains have relatively high n and m values (greater than 1.2) (Fig. 14) or mixed cations (rare earth’s and Sr or Ca [14,75]). In early studies it was suggested that elongated grain growth occurs only when p-Si3N4 powder is used [76]. However, later investigations have shown that the use of a-Si3N4 powder also leads to elongated grain growth [75,14, 15,791. The reason for the anisotropic grain growth of a’-SiAlON is not completely clear because it was not investigated in the same manner as the p-Si3N4 grain growth. The a’-SiAlON materials with low n and m values and equiaxed grains possess a low grain size. This is the consequence of the fast disappearance of the liquid. In contrary, the microstructures of the a’-SiAlON materials with elongated grains show a coarser microstructure; i.e. more intensive grain growth takes place. This must be connected with the amount of liquid existing during sintering, which is higher when higher substitution levels of the d-SiAIONs are used. Recently it could be shown that materials with elongated a’-SiAlON and low n and m values can be produced by gas pressure sintering using an appropriate liquid phase (see Fig. 15) [80]. This suggests, that the needle like grain shape is governed by the grain growth kinetics. The stability of a’-SiAlON at elevated temperatures (1 100-1 500°C) depends on the nature of the cations R entering the a’-SiAION structure. For the trivalent rare-earth cations (i.e. La-Yb; Y), the stability increases with decreasing atomic radius [14,81]; i.e. the temperature below which a phase transformation a’ -+ p’ can thermodynamically take place decreases. The stability also increases with increasing substitution level. Thermodynamic stabilisation of the a’-SiAlON with respect to the p’-SiAION can be achieved by the use of mixed cations [14]. Besides the nature of the cation which is incorporated in the a’-SiAlON, the rate of a’-$-transformation depends on the existence of $-SiAlON nuclei and the amount and viscosity of the amorphous grain-boundary phase. The transformation between a’-SiAlON and D’-SiAlON is reconstructive and must occur via a liquid in which one phase can dissolve and the product precipitate. Therefore the transformation can be retarded by the absence of a glassy phase or by a very high viscosity [14,81]. Better understanding of the microstructural formation of the a’-SiAlON materials in the future will be necessary for the improvement of the mechanical properties, especially fracture toughness and strength, and the wider application of this group of materials, having higher hardnesses than the B-Si3N4 and pSiAlON materials (see Fig. 16).

770


Figure 15. Polished and etched micrograph of an a’-SiAION material with n,m < 1; a) plasmachemically etched material (dark area: P-Si3N4;grey area: d-SiAION; bright area: grain-boundary phase); b) chemically etched material.

6.5 Properries

of Si, N4 Materials

771

HV10

110

cl

-

1000

3

iE Ei

-900

5

, '' , '

-3

t20 3

-800 .

E

to

2

--2"

$

-700

I

c 5

m

;a .-

I I

-

600

- 500

- 400

T

B-Si3N4 materials

Conventional a '-SiAlON materials

High toughness a '-SiAION materials

Figure 16. Comparison of the properties of the conventional a'-SiAION materials with P-Si3N4 materials and with high toughness a'-SiAION materials with elongated grains [80].

6.5 Properties of Si3N4Materials 6.5.1 Mechanical properties at room temperatures The strength of commercial silicon nitride materials lies in the range of 800 to 1400 MPa. The strength-determining defects are usually pores, inclusions such as iron silicides or agglomerates of glasses especially for low medium strengths. Besides these defects, large, elongated grains can be strength-determining defects for higher strengths or in specially developed, high-fracture-toughness materials. For strength levels higher than lOOOMPa, special surface finishing of the samples is necessary because surface defects play a decisive rule in determining the strength. The strength of a brittle material is proportional to the fracture toughness and indirectly proportional to the square root of the defect size. The defect size of the materials can be reduced by optimised processing. [55,82]. The highest measured mean three-point-bending strength for silicon nitride was 2000 MPa [85]. This corresponds to a defect size of about 5 pm. Materials with a strength level of 1400 to 1500 MPa usually have a defect size of 10 pm [55,83], i.e. these materials have

772


[871

[551

" .0

seeds oriented

5

10

1

Fracture toughness, MPa rnA1/2 Figure 17. Dependence of the tree point bending strength on the fracture toughness.

to have a grain size lower than that of the defect, because larger grains can act as strength-determining defects. This means that high-strength materials must be fine-grained. On the other hand, the fracture toughness increases with the square root of the grain size [36,69,87,88]. These two adverse dependencies produce the relationship between strength and fracture toughness shown in Fig. 17. For low fracture toughness values, flaws other than the largest grains are the strengthlimiting defects. In this region of KIC, further improvement of the strength is possible. For high KIc values, the strength is determined by large elongated grains. Thus the structure itself determines the strength, not processing-dependent defects. Materials with large, needle-like grains in the matrix are usually called insitu-reinforced materials. Attempts to improve the strength of such materials by seeding a matrix with large, needle-like grains with a narrow size distribution are underway [57,59,60]. Results for tape-cast silicon nitride materials with large, oriented P-Si3N4needles represent a possible approach [60] (Fig. 17) to increase the strenght and fracture toughness. However, such materials exhibit anisotropic properties, e.g. low strength in the direction of orientation of the needles. Additionally, the insitu-reinforced silicon nitride materials exhibit an increased Weibull modulus [86]. This has two reasons: firstly, the concentration of the large, strength-limiting grains is so high that the defect size has a narrow size distribution; and secondly, these materials exhibit pronounced R-curve behaviour, which makes the material tolerant of larger cracks. The high thermal shock resistance of these materials also seems to be connected with this fact [30]. The subcritical crack growth rate in silicon nitride materials is relatively low in comparison to other ceramic materials. The growth exponent is in the range of

6.5 Properties of Si3N, Materials

773

N = 30-300 for static loads and n = 20-50 for cyclic loading [90,91,97-1001. Up to now, no clear correlation between the microstructure and the growth exponent could be made. For cyclic loading it can be shown that materials with the lower grain size have the higher growth exponent, i.e. have a lower degradation of strength during loading and a higher lifetime at a given strength level [91,88]. The fracture toughness depends strongly on the microstructure. Two main factors influence the fracture toughness: 1. the grain shape and size, and 2. the composition of the grain-boundary phase. The high fracture toughness of silicon nitride materials in comparison to other ceramic materials is connected with toughening mechanisms which are similar to those in whisker-reinforced composite materials: grain bridging and pull-out, crack deflection and grain branching around large, elongated grains [88,89,94]. Due to these mechanisms, the fracture toughness increases with increasing volume fraction and square root of the mean grain thickness of the elongated grains (i.e. grains with aspect ratio >4) [37,70,88,89]. Recent investigations show that fine-grained materials have a higher toughness than coarse-grained materials in the small crack region but lower toughness in the large crack region [82,84,90]. This can be important for the use of such materials under conditions in which high local stresses exist (e.g. for ball bearings). The dominant toughening mechanism depends on the grain thickness of the elongated grains. Elastic bridging and pull-out were observed for thin, needle-like grains (with thickness < 1 pm). Crack deflection was mainly observed for thick, elongated grains (with thickness > 1 pm), whereas frictional bridging was detected independently of the grain size [89]. The toughening mechanisms described above can only operate when the dominant fracture mode is intergranular. The ratio of transgranular to intergranular fracture depends on the relative strengths of the grain boundary and the grains. For a material with a high toughness, the grain boundary must be weak in comparison to the grains (see Fig. 18). The strength of the grain boundary is connected with two different mechanisms: local residual stresses [92,101] and special chemical interactions between the grainboundary phase and the silicon nitride grains [93,94]. The amorphous or partially crystallised grain-boundary phases generally have different thermal expansion coefficients than silicon nitride. When the thermal expansion coefficient of the grain boundary is higher than that of the silicon nitride, the grain-boundary phase is under tensile stress and the fraction of intergranular fracture increases. As a consequence the fracture toughness increases (see Fig. 19). Similarly, a material with a grain-boundary phase under compression (e.g. HIPSN without sintering additives) shows a low fracture toughness [ 1021 due to a high percentage of transgranular fracture (see Fig. 18a). The residual stresses can be influenced by the sintering cycle. This can be brought on by a change in the composition of the grain boundary (e.g. by evaporation of S O 2 ) during sintering, by crystallisation of the grain boundary or by partial relaxation of the stresses (e.g. through slow cooling). These changes are usually outweighed by the grain size and shape or the starting composition of the material, which have a greater influence on the stress state and KIc.

114


Figure 18a and b. Crack path in silicon nitride materials a) HIPed material (only S O 2 sintering additive, mainly transgranular fracture; chemical etched), b) material with Y20,/A120,additives (mainly intergranular fracture, plasma etched).

Figure 19 shows the additional influence of the grain size (see curves 1-3). The composition of the grains also influences the fracture toughness by determining the special chemical interactions that occur between the grains and the grain boundaries. The formation of P-SiAlON layers on the Si3N4 grains was found to result in an increase in the amount of transgranular fracture and a decrease in the fracture toughness [93, 1001. This can be a reason for the lower fracture toughness of the P-SiAlONs with a high degree of substitution, in addition to the lower aspect ratio (see Fig. 18d). The recently developed a-SiAlON materials have a microstructure consisting of relatively coarse, elongated grains (see Fig. 15) and show a high percentage of intergranular fracture; thus, these materials exhibit a high fracture toughness (see Fig. 16). Summarising the fracture toughness data, it can be concluded that the fracture toughness of the silicon nitride materials depends strongly on the microstructure. For high fracture toughness values, elongated grains and a grain-boundary phase which promotes intergranular fracture are necessary. The hardness of different silicon nitride materials is given in Fig. 20. Materials with a high a-Si3N4 content, i.e. materials with a high amount of metastable aSi3N4which did not transform during sintering (normally produced by hot pressing or HIP’ing [l, 59]), show a hardness which is as high as that of the d-SiAlON materials. Analysis of the hardness values of different P-Si3N4 materials reveals that the hardness increases with decreasing amount of grain-boundary phase and decreasing

6.5 Properties q f S i 3 N 4 Mureriuls

775

Figure 18c and d. Fracture surfaces of silicon nitride materials c) fracture surface of material with Y2O7/AI2O1additives (mainly intergranular fracture), d) fracture surface of a P’-SiAION material with a high degree of substitution z (z = 4) (mainly transgranular fracture).

grain size; i.e. the dependence of hardness on the microstructure is opposite to that of the fracture toughness. Usually the hardness values of b-Si3N4 materials (HVlO) are in the range of 12 GPa (coarse grain size, high additive content, residual porosity) to 16 GPa (fine grain size, low additive content). The hardness of the mixed a/b-SiAlONs change lineary with the phase content [ 1311. The hardness at elevated temperatures depends additionally on the softening of the glassy grain-boundary phase. Materials with MgO/Al2O3as sintering additives show a faster degradation of the hardness than materials with more refractory grainboundary phases (Fig. 20).

716


/

Common sintering additives’ Glass Si02 ain boundary

i

Silicon nitride

---___ ----_--__

/

/

,i h e r m a l expansion coefficient,

10-6/K

,

SiO2-grain

boundary phase

Si3N4

I

2J

0

2

4

6

8

Thermal expansion coefficient, 10-6/K Figure 19. Residual stresses of grain-boundary phase and the silicon nitride grains as a function of the thermal expansion coefficient of the grain-boundary phase (calculated after [92]); used constants for Si3N4:E = 320, v = 0.27, s( = 3.39 x K; and for the glass: E = 140, v = 0.29, volume fraction of glass = 0.15; a) and dependence of the fracture toughness on the thermal expansion coefficients (data 1-3 after [92]) 1. mean thickness of the grains 0.55; 2. 1-1.3; 3. 1.5-1.7 (data 4 after [64]).

6.5 Properties of Si,Nd Materials

777

25

20

15

10

5 500

loo0

1500

Temperature, "C

Figure 20. Hardness of different Si3N4materials; 1 . additive free material with remaining a-Si3N4 [59], 2. hot pressed with 10% Ce02 and 6.6% AIN contain 20% a-SiAION [84]; 3. sintered p-silicon nitride material with 3% Y 2 0 3and 2 % Al2O3;4. d-SiAlON.

6.5.2 High-temperature properties of silicon nitride materials Silicon nitride is among the most promising materials for high-temperature applications because of its combination of excellent mechanical properties at room and elevated temperatures, oxidation resistance, low coefficient of thermal expansion and low density in comparison to refractory metals. However, for the majority of Si3N4materials these properties can only be obtained by a purposeful tailoring of chemical composition and microstructure. Special emphasis needs to be placed on the intergranular grain-boundary phase which is formed during the sintering process. As already mentioned, Si3N4 materials are usually densified by the addition of sintering aids. They form a glassy silicate phase with the silica always present as an impurity in the Si3N4raw powder; this silicate phase remains as an amorphous or crystalline grain-boundary phase between the Si3N4grains after sintering. While the properties of the Si3N4 grains do not change up to temperatures of 16OO0C,as a consequence of the high degree of strong covalent bonding in Si3N4, the grain-boundary phase starts to become weak at elevated temperatures. Depending on the amount, composition and condition of the grain-boundary phase between the silicon nitride grains, various processes (diffusion, creep, slow crack growth, oxidation, corrosion) occur at elevated temperatures. The consequence of these high-temperature processes is the generation of a new defect population which determines the failure behaviour and ultimately limits the lifetime of the silicon nitride material. The extent to which these processes occurs is mainly influenced by the softening point and viscosity of the amorphous grain-boundary phase.

778


Table 4. High-temperature properties of various Si3N4 materials (01400’~ bending strength at 1400°C; i - creep rate at 1400”C, 200MPa; Am,, - weighted gain during oxidation at 1500°C; o,, - residual strength after oxidation). ~

1

2 3 4 5 6 7

HIP-Si3N4 SN 5% Y203/3% A1203 SN 8% Y203/0.6%Ai203 cL’/D’-Sialon (Y) SNS%Y203 Si3N4-SiC (SY” Y203) Si3N4-MoSi2 (So/, Y203)

500 450 720 380 730 750 730

9x ~

6x 5 x lo-* 2 x lo-’ 2 x lo-’ 2x

0.7 (2500 h) 4 (100h) 3.7 (1000h) 10 (100h) 2.2 (2500 h) 3.3 (2500 h) 3.1 (25ooh)

490 < 200 210 < 200 340 640 630

1 < 0.2 0.2 < 0.2 0.4 0.8 0.8

In the following chapter fundamentals of the high-temperature behaviour of silicon nitrides and silicon nitride-based composites will be presented. Depending on the sintering additive used, silicon nitride materials with different high-temperature properties can be obtained. Some typical materials are chosen from the broad variety of Si3N4ceramics that exist and summarized in Table 4. The first Si3N4ceramic (Table 4) is a material without additional sintering aids. The only liquid phase appearing during sintering is the silica from the high-purity Si3N4 powder. Because of the small amount and the high softening point of the grain-boundary phase, these materials can only be densified by encapsulation hot isostatic pressing (HIP). The materials exhibit excellent behaviour at elevated temperatures as the consequence of a clean grain-boundary phase consisting only of silica [102]. The oxidation resistance is used to demonstrate the excellent hightemperature properties of this material. Figure 21 shows the microstructure of a HIP’ed Si3N4 without additives after 2500 h oxidation at 1500°C. The oxidation behaviour of silicon-based nonoxide ceramics is mainly influenced by the protective layer which is formed at the surface of the material as a result of the oxidation

Figure 21. SEM images of polished cross sections of the Si3N4material 1 (Table 1) after 2500 h of oxidation treatment at 1500°C.

6.5 Properties of' Si3N4 Muteriuls

779

process and the ability of this layer to prevent oxygen from diffusing into the material. In this case the formation of a surface layer of pure silica causes the rate of oxygen diffusion into the material to be very low [103, 1041. A more refractory grain boundary can only be obtained by removing the silica from the grain boundaries and triple junctions. Each addition of only a small amount of impurities or sintering additives will weaken the material by producing a change in the chemistry of the grain-boundary phase [105-1071. The result is degradation of the high-temperature behaviour of the materials. Besides the costly fabrication, the main disadvantage of these material is their inadequate mechanical properties (especially toughness) at room temperature. The microstructure, which is mainly globular, and the strong bonding of the intergranular phase do not allow toughening mechanisms like crack deflection or bridging to be active. With the addition of sintering aids these disadvantageous properties can be improved; however, the improved densification, the microstructure with elongated grains, and the improved toughness are the consequence of a changed grain-boundary phase. Depending on the kind of sintering additive used, a silicate phase will be formed during sintering and will remain as an amorphous or partially crystallized phase in the grain boundaries and triple junctions between the Si3N4grains. Besides the chemical composition and the amount, special emphasis is placed on the crystallinity of the intergranular silicate phase. With a high amount of crystallized silicates in the grain boundary, the amorphous phase is minimized; this is found to be especially beneficial to the creep and slow crack growth behaviour of the Si3N4 materials. Glass-forming and stabilizing sintering additives which, with the silica of the Si3N4 powder, form a silicate phase with a low softening point and viscosity such as MgO, AlzOi or AlN are principally not useful in the fabrication of high-temperature silicon nitride materials. More refractory silicates that form an intergranular phase with a high crystalline content were obtained by using yttria or the rare-earth oxides as sintering additives. Examples for these materials are given in Table 4. When A13+ions are present, the softening point and viscosity of the grain boundary are lowered [ l , 31, with the consequence of degradation of the mechanical properties at elevated temperatures. Figure 22 summarizes the creep curves of the alumina-containing silicon nitride, material 3 (Table 4), in comparison with the additive-free material and the material with yttria as a sintering aid. The HIP'ed material exhibits the highest creep resistance. Superior creep behaviour is also found for Si3N4with Y203 as a sintering additive. This is the consequence of the high degree of crystallized grain-boundary phase having the apatite structure. The creep resistance of the Si3N4materials is found to be lowered by an amount which depends on the amount of A13+ions present. With the addition of a small amount, the crystalline grain-boundary phase does not change significantly; however, a degradation of the creep resistance is observed, caused by the concentration of A13+ ions in the residual amorphous grain-boundary phase. In the material with 5% Y 2 0 3and 3% A1203,the grain-boundary phase is found to be mainly amorphous, with a lowered softening point and viscosity. The result is a very poor high-temperature resistance: the material failed after only a few minutes during creep testing at 1400°C and 200 MPa. A similar tendency is found by comparison of the oxidation behaviour of the materials. With the addition of the sintering aid the composition of the protective

780


0,Ol 0,008 0,006 0,004 0,002

1/

0 0

20

40

60

80

100

Figure 22. Comparison of the creep behaviour of the materials (1 in Table 4) HIP’ed Si3N4without additives, (3 in Table 4) Si3N4with 8% Y203/o.6%A1203and (5 in Table 4) Si3N4with 8% Y203as sintering additives. The tests were conducted at 1400°C in air at 200 MPa.

oxidation layer formed during oxidation changes, with the consequence of a higher oxygen diffusion into the materials. The weight gain of the Si3N4with Y2O3 is about three times higher than that of the HIP’ed Si3N4without sintering aids, and that of the material with the small amount A1203about one order of magnitude higher than that of the HIP’ed Si3N4without sintering aids. Figure 23 shows the microstructure of the materials 3 (8% Y2O3 0.6% A1203) and 5 (8% Y203)after oxidation at 1500°C. Oxidation processes are observed in the bulk of both materials and lead finally to alteration of the microstructure. However, due to the addition of A1203, the oxidation of material 3 (in Table 4) occurs faster. Already after 1000 h oxidation the microstructure is found to be strongly degraded. The same damaging processes occur with the Y203-containingSi3N4material; however, this material exhibits, due to its more refractory grain-boundary phase, a higher oxidation resistance (see

+

Figure 23. Microstructural damage in the surface region of the Si,N4 materials with Y2O3/Al2O3 (3 in Table 4) after 1000 h (A) and Y 2 0 3(5 in Table 4) after 2500 h (B) oxidation at 1500°C.

6.5 Properties of Si3N4 Materia1.r

78 I

microstructure in Fig. 23 after 2500 h). As a result of the oxidation processes, degradation of the mechanical properties is observed; this is found to be dependent on the extent of oxidation as represented by the residual bending strength after oxidation (see Table 4). Interesting materials with respect to high-temperature behaviour are mixed a’/p’ SiAlON [3,78]. Due to the ability of the a’-SiAlON grains to incorporate cations from the sintering aids into their lattice, it is possible to modify the grain boundary of the silicon nitride material. In a material with a high amount of a’-SiAlON, a skeleton from strong grain boundaries between the a’-SiAlON is formed due to the incorporation of the ions of the sintering aids into these grains, with the consequence of an improved creep resistance. The oxidation resistance cannot be improved by the increase in the d-SiAlON content. The high amount of A13’ ions in the material is the reason for the formation of a weak surface layer of alumosilicates during oxidation, allowing a high rate of diffusion of oxygen into the material 1781. One goal in recent studies in the field of high-temperature Si3N4materials was the improvement of the behaviour at elevated temperatures by the formation of composites. Typical examples which have been reported in the literature are Si3N4-Sic micro- or nanocomposites or Si3N4-MoSi2 composite materials. In the case of Si3N4-SiC nanocomposites, a structural synergism between the matrix Si3N4and the nano-Sic was supposed to be responsible for the high level of the properties of these composites at elevated temperatures [ 1091. Recent studies, however, showed that the improvement of the mechanical properties at high temperatures was the consequence of a chemical modification of the composition of the grainboundary phase and a significant improvement of the oxidation resistance [104, 108,110,ll I]. Similar behaviour was found for Si3N4-MoSi2 composite materials [112]. After long-term oxidation tests at 15OO0C, a considerably less damaged microstructure was observed. Figure 24 shows the comparison of the material 5 (Si3N4 with 8 % Y203) with a Si3N4-MoSi2 composite (material 7). Apart from the 10% MoSi2 in the composite, both materials had the same composition. The microstructural development of the materials was found to be different. A significantly lower amount of damage was observed in the Si3N4-MoSi2 composite, although the oxidation processes also occurred in the bulk of the composite.

Figure 24. Comparison of the microstructure of (A) the Si3N4material ( 5 in Table 4) and (B) the Si3N4-MoSi2 composite (7 in Table 4) after 5000 h oxidation at 1500°C in air.

182


bending stress / MPa

1-1

900

700 500

i

300

I00

time I h

Figure 25. Comparison of time to failure behaviour of silicon based nonoxiode materials at 1400°C in bending (curve 1: material 7; curve 2: material 6; curve 3: material 5; curve 4: SBC).

This behaviour was the consequence of different processes in the surface region of the materials, resulting in a changed oxidation mechanism in the composites. The important feature of the Si3N4--MoSi2and Si;N,-SiC composite was the increased formation of Si20N2 instead of SO2, producing an additional layer between the oxidation surface and the bulk of the silicon nitride composite materials. The significantly reduced defect size gives an idea of the potential of these composite materials in terms of stability and time-to-failure behaviour at elevated temperatures (Fig. 25). Principally, the amount, composition and condition of the grain-boundary phase are the key factors which must be considered for the successful development of Si3N4materials for applications at elevated temperatures. Materials without sintering additives exhibit the highest stability at elevated temperatures due to the high refractory grain-boundary phase of pure SO2. The addition of sintering aids ultimately leads to a degradation of the properties at elevated temperatures. However, by a purposeful design of the grain-boundary phase and the formation of high temperature stable composites which are stable at high temperatures, it is possible to obtain silicon nitride materials which can be used up to temperatures of 1500°C.

6.5.3 Wear resistance of Si3N4 materials The excellent wear properties of silicon nitride materials are the basis for many applications of these materials, such as ceramic or hybrid rolling bearings, cutting tools, wear parts in automotive engines (valves, cam rollers, valve plates for common rail systems), and parts for metal shaping. Wear is a property of the system, not only of the material. Therefore a few special applications will be presented to show the influence of the microstructure on the wear properties.

6.5 Properties of Si3N4 Materials

783

Figure 26. Typical cracks on the track of ball bearings (dry running under 1.75GPa Hertzian turns). pressure. 18

Silicon nitride is the preferred ceramic material for use in ceramic and hybrid ball bearings due to its high strength, fracture toughness and high resistance to subcritical crack growth. The advantages of ceramic ball bearings in comparison to metal ball bearings are: reduced wear, higher hardness, better temperature and corrosion resistance, lower specific weight and higher electrical resistance. This combination of properties makes lubrication of such bearings either almost or completely unnecessary [2, 113, 1 15, 1171. The field of application of such bearings is fast growing. The bearings show a good performance and a high reliability. To improve the behaviour a development of silicon nitride materials for ball bearing was started in the last few years [ I 13-1 16,1181. The main failure mechanism of ceramic ball bearings are connected with local and temporal changing high tensile and compressive stresses during operation. These local and temporally changing stresses result in the formation of cracks perpendicular to the rolling direction of the balls in the ball bearing (see Fig. 26) [I 141. The lifetime and load capacity of a ball bearing can be increased by optimisation of the geometry of the track [ 1 13-1 15,1181, by reduction of the material damage due to cyclic Hertzian contact, and by reduction of the friction coefficient [I 14,1161. The influence of the microstructure on the Hertzian contact damage was investigated in detail [119, 1201. Coarse-grained silicon nitride materials are subjected to an accelerated fatigue under cyclic Hertzian pressure, unlike fine-grained materials. Therefore materials for ball bearings have to be fine-grained. Apart from that, the materials must exhibit high strength (i.e. >800 MPa) and a high Weibull modulus because every defect on the track can be a starting point for failure. The influence of the friction coefficient on the load capacity is connected with the fact that, besides rolling, sliding always exists in the ball bearing [114, 1151. This friction of the ball on the track is the reason for an increase in the local tensile stress in the track during operating. Therefore reduced friction will increase the load capacity and the lifetime. The load capacity of ball bearings operating in wet conditions is more than 2 times higher than in dry conditions. This is due to the lubricating effect of the water [I 151. The development of Si3N4 composites with small amounts of submicro-TiN or BN inclusions leads to an increase in the load capacity of up to 60% under dry run (see Fig. 27) [115]. Further improvements can be expected through surface modification by ion implantation [I 171, the use

184


Self lubricant SN-based composites Friction Figures p 0.02

Idry run in air

water lubrication 0,015

0,Ol

0.005

0

Standard Silicon nitride

Si3NUBN

Si3N3/TN

Standard Silicon nitride (grease lubriicated )

Figure 27. Friction figure (maximal value) of angulare contact bearings 7006 E under dry running in air and in water at 500N preload [115].

of special thin coatings or by the development of Si3N4materials with high strength and low grain sizes < 100 nm with improved wear resistance [63]. The friction of silicon nitride materials depends strongly on the wear conditions. At low loads the grain boundary is mainly affected (see Fig. 28a), whereas at higher loads silicon nitride grains can break (see Fig. 28b). The wear products can form lubricating layers or rolls that can reduce the friction (see Fig. 28c) [1211241. Additions of TIN and BN can further reduce the friction and wear under special conditions. An overview of the wear properties under sliding conditions of different Si3N4 materials is given in [121-124,631. Up to now, the silicon nitride materials have not been optimised for use under different wear conditions. Optimisation of the grain size and composition and amount of the grain boundary can lead to improved wear behaviour of silicon nitride materials in the future. Silicon nitride materials are used as cutting tools for cutting cast iron and superalloys [ 125-1 3 11. Ceramic tools and particularly silicon nitride cutting tools exhibit higher hardness at high temperatures than cemented carbides. That allows the use of ceramic tools for machining at higher cutting speeds than for cemented carbides. The typical increase in metal removal rate due to the use of Si3N4in comparison t o cemented carbide tools is more than 200% [129]. The high temperatures and mechanical load that are generated at the cutting edge cause wear by a variety of mechanisms: -

abrasive wear due to hard inclusions in the workpiece or chemical interaction with the workpiece or the ambient atmosphere,

6.5 Properties

of' Si,N4 Materials

785

Figure 28. Surface of silicon nitride materials after oscillating sliding tests. -

thermal deformation or degradation by thermal shock, subcritical crack growth and, consequently, fracture of the tool. Tool failure due to fracture must be avoided because it leads to an unpredictable behaviour of the tool. Therefore the cutting materials must have high strength, fracture toughness and thermal shock resistance.

The first two mechanisms lead to more or less continuous wear of the cutting tool over its lifetime. The temperature of the cutting tool edge can reach about 800°C for turning cast iron [ 130, 1341; temperatures of up to 1350°C were measured for the cutting of steel [134]. The enhanced chemical interaction and softening of the grain-boundary phase due to the high temperatures during steel cutting are the reasons why silicon nitride cutting tools cannot be used effectively for the cutting of steel. Chemical interaction is also a dominant wear mechanism during turning and milling of cast iron [ 127,1301. On the surface of the cutting tool, different silicates, which have a low hardness and which can therefore be easily removed during machining, can be formed. Chemical interaction seems to be the reason for the relatively rapid rounding of the cutting edge at the beginning of the cutting process. This produces an increase in the cutting forces and is the reason for using silicon nitride tools only for rough machining. The combination of the different wear mechanisms during cutting makes understanding of the cutting behaviour difficult. Up to now, many open questions exist and the cutting behaviour cannot be predicted using knowledge of the mechanical properties and microstructure. However, some overall relationships between cutting behaviour and material properties can be given. The cutting materials must exhibit

786

6 Silicon Nitride Based Hard Mutevials

high strength, hardness and fracture toughness. These properties are necessary but not sufficient for a good cutting behaviour: there are several Si3N4materials which fulfil these conditions, but which have a bad cutting behaviour due to enhanced chemical interaction with the workpiece. A wide range of substances in different ratios are used as sintering additives for cutting tools: CaO, MgO, A1203,Y 2 0 3 ,Zr02 and Hf02 [2,171,125-1311. The use of sintering additives that reduce the high-temperature strength is surprising, but is less so when one considers that the softening of the glassy grain-boundary phase starts at temperatures above 800”C, whereas the temperature at the cutting edge in the machining of cast iron is lower than 800°C. An increased Si02 content in the grain-boundary phase reduces the lifetime of the cutting tools [133]. This seems to be connected with the acceleration of the chemical interaction. A decrease in the amount of sintering additives improves the cutting behaviour [125]. The problem of optimising the cutting tool material is that on the one hand it is necessary to minimise the amount of the grain-boundary phase and on the other hand it is necessary to attain a reliable densification to achieve full density, high strength and high reliability. One possibility to improve the densification of the materials with a low additive content is to use MgO-containing additives [71,129]. Another way to improve the sintering behaviour without increasing the amount of the residual grain-boundary phase is to form p’- or P’/a’-SiAlONs [130]. The substitution grade z in the P-SiALON must be small because high z values cause the fracture toughness and strength to be reduced. In [ 1321 an optimum for Si6-,Al,0,N8_, is found with z = 1-1.5, but most of the commercial cutting tools have an lower substitution degree. p’/a’-SiAlON composite materials are used especially for cutting superalloys. They have a slightly reduced strength and fracture toughness [128, 1311. A low grain size is favoured for cutting tools because the hardness increases with decreasing grain size. Nevertheless excellent cutting performance is found for selfreinforced Si3N4materials with large, elongated grains [7 11. The examples given above show that it is necessary to find a compromise between the different material properties. The compromise strongly depends on technological possibilities (e. g. improvement of the sintering behaviour of the powder) and on changes in the metal machining. The chemical stability of the cutting tools can be increased through the use of coatings, e. g. A1203/TiN[129].

6.5.4 Corrosion resistance of Si3N4 Silicon nitride-based ceramics are promising engineering materials for application under corrosive and wear conditions [135-142]. However, the use of silicon nitride materials in hot acids under hydrothermal conditions and in bases is often limited by degradation of the materials. This is the impetus for current efforts for better understanding of the influence of the microstructure on the corrosion behaviour of silicon nitride materials and to improve the materials’ behaviour on its basis. Up to now, only a small amount of data about the mechanisms of corrosion and pitting formation exists.

6.5 Properties o f S i 3 N 4 Muterials

787

Table 5. Classification of corrosion conditions and main features of the corrosion behaviour Conditions

Corrosion

Literature

Organic components (oil, hydrocarbons)

Work as lubricants, reducing the wear

113, 115

Acids (HCI, H2S04, H N 0 3 . . . )

Main attack at the grain boundary phase Corrosion resistance can be improved significantly by tailoring the composition

135-146, 148

Media-intensive solving SiOz protective layers, H F alkaline melts; concentrated alkaline solutions at high temperatures > 100 to 150°C; hydrothermal conditions at 2250°C

Dissolution of the grain boundary and the Si3N, grains

135, 144-146, 149

Bases at medium temperatures

135, 138, 140, 144, 149, 150

Hydrothermal conditions at T 5 200°C


136, 141, 151--155

Intensive corrosion

Nevertheless, some correlation’s between the microstructure and the corrosion behaviour are known. The corrosion behaviour of silicon nitride materials in liquids is mainly controlled by the stability of the grain-boundary phase. Therefore the corrosion resistance of the silicon nitride materials in different media can be altered by more than two orders of magnitude by changing the composition of the material. The corrosion behaviour can be organised into a few main classes (see Table 5). Standard silicon nitride materials with Y203/Al2O3sintering additives degrade quite strongly in hot acids of medium concentration (see Figures 29-31). With increasing temperature the corrosion resistance decreases, whereas with decreasing additive content the corrosion resistance increases. The best corrosion resistance in acids was obtained for a HIP’ed Si3N4material without additives. Si3N4materials are more stable in concentrated acids (>5 N) than in solute acids [135, 136, 138, 1481. The MgO-containing materials possess better corrosion resistance than the Y 2 0 3 / A1203-containing materials. The reasons for this are not clear [136, 1411. The corrosion behaviour is different in H 3 P 0 4than in other acids due to formation of a protective phosphate layer [138, 1441. Corrosion in HF-containing solutions is much more intensive than in other acids. This is connected with the ability of H F to dissolve the SiO, protective layers. Therefore H F solutions will not only attack the grain-boundary phase but also dissolve the Si3N4grains.

788


Figure 29. a) SEM micrograph of an corroded in H2S04 100 h 90°C material with Y203/A1203 additives and b) SEM micrograph of an material with optimized grain boundary corroded in H2S04 500 h 90°C material (plasmaetched).

Recent investigations show that, besides the amount of the additives, the main parameter governing the corrosion resistance of the silicon nitride materials is the Si02 content in the grain-boundary phase (see Figures 29-31). A material with a high S O z content in the grain-boundary phase is more stable in acids than one with a low Si02 content. This correlates with the stability of the grain-boundary phase. The acid corrosion of glasses shows a similar behaviour [147,155]. The network modifier is leached from the glass network by the acids and the glass network hydrated. Therefore a network with a low amount of network modifiers has a higher stability [ 1471. Additional to the factors influencing the corrosion of glass, the corrosion of silicon nitride is affected by the fact that corrosion takes place only in small channels between the Si3N4 grains. During corrosion the Si3N4 grains also produce hydrated Si02, which can reduce the corrosion rate and change the corrosion mechanism. This seems to be one of the reasons for the low corrosion rate of Si3N4 materials with a low amount of sintering additives [141].

6.5 Properties of Si,N4 Muterials

I OY 203/A1203- 1

789

P

I

m

10 MgO/Y203/A1203

.

5Mg0

*-

*-

1

< N

5

;;4

E

v1

0

0

-

-

100

200

3,2 Vol.% 300

400

Time, h b)

105 1

751

\ 3,2 Val.%,

65 0

100

200

300

400

Time, h C)

Figure 30. Weight loss of different Si3N4-materialsin H2S04 at 60°C (additive composition is given at the curves) a) and weight loss b) and residual strength c) of different materials with different amount of Y203/A1203sintering additives in HCI at 60°C.

190

6 Silicon Nitride Bused Hard Muterials

0,li 25

'

,

30

'

"

'

35

'

'

8

40

45

50

55

60

65

70

Si02-content, Mol%

Figure 31. Weight loss of silicon nitride materials with different Si02 content in the grain boundary (200 h corrosion in H2S04 at 60°C).

Additionally, the concentration of the other constituents in the glassy grainboundary phase has an influence on the corrosion behaviour more detailed investigations are necessary to confirm this, The kinetics of the corrosion of Si3N4materials in acids and the corrosion mechanisms are not clear. Data about diffusion and reaction-controlled mechanisms exist. It is possible that the mechanism changes, depending on the corrosion depth, the acid concentration, temperature and microstructural parameters, but further investigations are necessary to confirm this [141]. The reduction of strength after corrosion in acids for a short time (pitting formation) cannot be correlated with the weight loss. Acid corrosion have an influence on the subcritical crack growth of the materials [138]. The corrosion resistance of selected Si3N4materials in NaOH is given in Fig. 32. The silicon nitride materials are less attacked by bases than by acids. The extent of corrosion increases with increasing temperature and concentration of the bases [150]. Materials which are less stable in acids are more stable in bases and under hydrothermal conditions. The most stable materials under hydrothermal conditions are materials with a low Si02 content in the grain-boundary phase (see Fig. 33). The HIP'ed materials without additives are, besides the MgO-containing materials, less stable than the Y 2 0 3 / A1203-containing materials. Under hydrothermal conditions the grain-boundary phase of these materials is dissolved completely, leading to removal of the silicon nitride grains from the surface. This behaviour is different to the corrosion in acids, where the grain boundary is not completely dissolved and the corroded structure is quite strong and stable. This is connected with the stability of the glass network. Under corrosion conditions in acids mainly the network modifiers are dissolved, whereas under hydrothermal corrosion conditions and in strong basic solutions the SiO, network dissolves. Therefore materials with a high SiOz content

6.5 Properties of Si3N4 Muterials

79 1

7%Y203/A1203

Time. h

7%Y 203/Al203

I O%Y203/A1203

0

50

100

I50

200

Time , h Figure 32. Weight loss and residual strength of silicon nitride materials in 1 N NaOH at 60°C.

in the grain boundary are less stable under hydrothermal conditions than materials with a low Si02 content in the grain-boundary phase. Under hydrothermal corrosion at 270"C, a significant dissolution of the Si3N4 grains takes place. In materials with Y203/A1203as sintering additives, the dissolu-

192

6 Silicon Nitride Bused Hard Mutevials

Si3N4 MgOiA1203 r4

‘E

1-

HIPSN without additives

Sl

E y:

-0

0,014 10

100 Time, h

1000

Figure 33. Weight loss of silicon nitride materials under hydrothermal conditions at 210°C.

tion rate of the grains is higher than that of the grain-boundary phase (see Fig. 34). The weight loss due to hydrothermal corrosion depends linearly on time. Only materials with high corrosion rates (MgO-containing and Si02-rich materials) show an increase of the corrosion rate with time. Investigations of the corrosion behaviour of silicon nitride materials in acids, bases and under hydrothermal conditions show that a main parameter governing the corrosion resistance is the composition and amount of the grain-boundary phase. Materials with a high Si02 content in the grain-boundary phase are stable in acids and less stable in bases and under hydrothermal conditions. Materials with a low Si02 content in the grain-boundary phase are stable in bases and under hydrothermal condition but less stable in acids. Up to now no materials which are simultaneously stable in acids, bases and under hydrothermal conditions exist. A more detailed understanding of the corrosion mechanisms and the influence of the medium on the lifetime of the materials is necessary before silicon nitride materials can be used in a broad range of applications under corrosive conditions. The results of the investigations indicate that tailoring of the microstructure and the grain boundary composition is necessary for the application of silicon nitride ceramics in corrosive environments.

6.6 Conclusions/Further potential of silicon nitride materials Silicon nitride ceramic materials exhibit a high potential for structural applications at room and elevated temperatures. This is due to their excellent mechanical properties in combination with good corrosion and thermal shock resistance.

6.6 ConclusionslFurther potential

of silicon nitride rnateriu1.y

793

Figure 34. Corrosion layer of Si3N4 materials a) MgO/AI2O3-additive system, hydrothermal corrosion conditions: 210'C, 200 h under at 250°C; b) Y2O3/AI2O3-additivesystem, hydrothermal corrosion conditions: 270"C, 200 h (the grain-boundary phase solves less than the Si,N4 grains and are visible as bright areas).

The development of the silicon nitride materials in the future will be divided in two main directions; the increased penetration in the market of the state of the art silicon nitride materials by improving the technology and reliability and decreasing the production costs and the development of new materials with improved wear, corrosion, high temperature properties.

794


In the last decade a continuously increase in the number of application fields was observed. The technical feasibility and the reliability of the materials and components were shown in many applications, such as valves and other parts for the automotive industry, bearings, household applications. For the wider penetration into the market there is a engineering work necessary to optimise the production, stabilise the reliability and reduce the production cost. First steps in this direction were made by using low cost powders and optimising processing, sintering and finishing. The majority of applications can be realised with strength levels lower 10001200 MPa. Higher strength level need a special more expensive finishing procedure increasing the costs of parts. Additionally a strength level of 1200 MPa corresponds to a defect size of 10-15 pm (at the existing KTClevels) which is in the range of defects created during operating of the materials. Whereas the direction of further material development mentioned above is connected with optimising the technology, the second direction is connected with adapting of the microstructure and properties of silicon nitride materials on different applications. As like as exists different kinds of steel, which are well designed for the given applications, as well in the future will exist different classes of silicon nitride materials aimed for applications under different conditions e.g. wear, corrosion, operation at high temperature and so on (see Table 1). These differentiation processes stay only at the beginning and need as a bases a controlled formation of the microstructures and the knowledge of the relation between the microstructure and properties. The main factors and processes influencing the microstructure of Si3N4materials are known and the microstructure can be tailored in a wide range. In the future the use of different p-silicon nitride powders as seeds for a precise tailoring of the microstructure will be used more intensively. New aspects can be expected by the development of fine grained materials. The correlation between microstructure, strength and toughness are well understood, whereas the understanding of the correlation between microstructure on the one side and wear, corrosion properties and long term behaviour on the other side need further intensive interdisciplinary investigations. An intensive development of high fracture toughness a’-SiAlON-materials can be expected. The better understanding of the microstructure formation of these materials and a further improvement of strength and toughness by a reduced grain size are main topics of the investigations. Frequently, there is a need to make a compromise between the different properties due to the opposite dependencies on the microstructure. How such compromise looks like depends on the technological possibilities and on the powder properties, which developments offer new possibilities. An improvement of the materials behaviour can be expected by the development of homogeneous composite materials based on silicon nitride, as it was shown for Si3N4/MoSi2or Si3N4/SiCcomposites for the high temperature long term applications. Especially in this area the use of metallorganic precursors can lead to an improvement of the materials. The reduction of sintering temperature and in

References

795

some modifications the shrinkage (concept of active filler) can allow the production of composites of thermodynamic unstable components. The development in this direction is only at the beginning.

Acknowledgements The authors are grateful to the colleges of the IKTS which take part in the technical completion of the paper especially A. Bales and D. Hermannutz. They are indebted to the coworkers in the IKTS especially to Dr. I. Schulz, Dr. T. Reich, A. Bales and G. Michael involved in research projects in the field of silicon nitride materials which results were included in this paper. The work is based on different projects supported by the BMBF, AiF and the DFG in the field of silicon nitride materials.

References 1. S. Hampshire, “Nitride Ceramics in Structure and Properties of Ceramics”, ed. R. W. Cahn, P. Haasen and E. J. Kramer, Materials Science and Teclznology, 11, 121-168. 2. R. N. Katz, “Application of Silicon Nitride based Ceramics in the U.S.”, Mat. Res. Soc. Symp. Proc., 287, 1993. 197. 3. K. H. Jack, “SiALON Ceramics: Retrospect and Prospect”, Mat. Res. Soc. Symp. Proc., 287, 1993, 15-18. 4. J. G. Heinrich and H. Kruner, “Silicon Nitride Materials for Engine Applications”, M. J. Hoffmann and G. Petzow (eds.), Tailoring of’ Mechanical Properties of Si,N4 Ceramics, Kluwer Academic Publishers, 1994, 1941. 5. “Der Kochherd der Zukunft arbeitet mit Si3N4-Platten”,cfi/Ber. DKG, 76, 1999, D 18-20. 6. J. Heinrich, “Siliciumnitridkeramik”, editor: J. Kriegsniann, DKG Technisclze Keramisclze Werkstofe, 1999, 4.3.1 .O. 7 . H. M. Jennings, “Review on Reactions between Silicon and Nitrogen, Part 1 Mechanisms”, Journal of Material Science, 18, 1983, 951 -967. 8. A. Zerr, G. Miehe, G. Serghiou a. o., “Synthesis of Cubic Silicon Nitride”, Nature, 400, 1999, 340-344. 9. W. Ching, Y. Xu, J. Gale and M. Riihle, “Ab-Inito Total Energy Calculation of cc- and p-silicon nitride and the Derivation of Effective Pair Potentials with Application to Lattice Dynamics”, J . Am. Ceram. Soc, 81, 1998, 3189-3196. 10. I. Kohatsu and J. W. McCauley, “Re-Examination of the Crystal structure of a-Si3N4”,Mat. Res. Bull, 9, 1914, 9 17-920. 11. H. Suematsu, J. J. Petrovic and T. E. Mitchell, “Plastic deformation of silicon nitride single crystals” Materials Science und Engineering, A 209, 1996, 97-102. 12. J. Ihsza, T. Eschner and K. Rundgren, “Hardness anisotropy in bimodal grained gas pressure sintered Si3N4”,Journal of Materials Science Letters, 16, 1997, 16641667. 13. Andrievskii, Spivak I. I. , “Nitrid kremnija i materiali na evo osnove”, Moskau, Metallurgijr, 1984, 19-20. 14. H. Mandal, D. P. Thompson and K. H. Jack, “cc H f3 Phase Transformations in Silicon Nitride and Sialons”, preprint. 15. L.-0. Nordberg, “cc-Sialon Ceramics and Y-cc-Sialon Composites; Composition, Microstructures and Properties”, Doctorial Dissertation, 1997, 1-124.

196


16. K. H. Jack, “a-SiAION Ceramics”, Nature, 274, 1978, 880-882. 17. T. Y. Tien, G. Petzow, L. J. Gauckler and J. Weiss, “Phase Equilibrium Studies in Si3N4Metal Oxides Systems”, Progress in Nitrogen Ceramics, 1983, 89-99. 18. D. D. Lee, S. J. Kang, G. Petzow and D. N. Yoon, “Effect of c( to p Phase Transition on the Sintering of Silicon Nitride Ceramics”, J. Am. Ceram. Soc., 73, 1990, 767-9. 19. M. N. Rahaman, “Ceramic Processing and Sintering”, Marcel Dekker Znc., Department of Ceramic Engineering University of Missouri, 1995. 20. M. Herrmann und Chr. Schubert, “Grundlagen der Gefugeausbildung in 0-Siliciumnitridwerkstoffen”, editor: J. Kriegsmann, DKG Technische Keramische Werkstoffe,Sep. 1999, 5.1.3.1. 21. P. Greil, “Einflup der intergranularen Glasphase auf die Hochtemperatureigenschaften von PSiALONen” Dissertation Stuttgart 1982. 22. M. Herrmann, G. Putzky, S. Siege1 and W. Hermel, “Einflup von Zersetzungsreaktionen auf die Sinterung”, c$/Ber DKG, 69, 1992, 375-382. 23. P. Obenaus and M. Herrmann, “Methode zur quantitativen Charakterisierung von Stengelkristalliten in Siliziumnitridkeramik”, Prakt. Met., 27, 1990, 503-5 13. 24. H. Bjorklund, L. K. Falk, K. Rundgren and J. Wasen, “a-Si3N4Grain Growth, Part I: Effect of Metal Oxide Sintering Additives” Journal of the European Ceramic Society, 17, 1997, 1285-1299. 25. H. Bjorklund and L. K. L. Falk, “a-Si3N4 Grain Growth, Part 11: Intergranular Glass Chemistry”, Journal of the European Ceramic Society, 17, 1997, 1301-1308. 26. G. Wotting, B. Kanka and G. Ziegler, “Microstructural Development, Micccrostructural Characterization and Relation to Mechanical Properties of Dense Silicon Nitride”, Nonoxide Ceramic, 1986, 83-95. 27. G. Ziegler, J. Heinrich and G. Wotting, “Relationship between Processing, Microstructure and Properties of dense and Reaction-Bonded Silicon Nitride”, J. of Mat. Sci., 22, 1987, 3041-3086. 28. F. F. Lange, “Fracture Toughness of Si3N4as a Function of the Initial a-Phase Content”, J. Americ. Ceram. Soc., 62, 1979, Nr. I , 428429. 29. G. Woetting, H. Feuer and E. Gugel, “The Influence of Powder and Processing Methods on Microstructure and Properties of Dense Silicon Nitride”, Mat. Res. SOC.Symp. Proc., 287, 1993, 133-146. 30. S. Kessler, M. Herrmann and Chr. Schubert, “The a-P-Transformation and anisotropic growth of 0-grains during sintering of Si3N4”, Mat. Sci. Forum, 94-96, Trans. tech. Publ., 1991, 821-827. 31. G. Petzow and M. J. Hoffmann, “Grain Growth Studies in Si3N4-Ceramics”, Mat. Sci. Forum, 113-115, 1993, 91-102. 32. M. Herrmann, S. Kessler and Chr. Schubert, “Microstructural Design of Dense Si3N4”,EuroCeramics IZ, V01.2,Edited by G. Ziegler and H. Hausner, 1991, 847-851. 33. W. Hermel, M. Herrmann and Ch. Schubert, “Have the limits of the development of P-Si3N4Materials Already been reached”, In: Third Euro-Ceramics, V3, edited by P. Duranand, J. F. Fernandez, Faenca Edititrice Iberica S. L., 1993, 391-396. 34. W. Hermel, M. Herrmann and I. Schulz, “Sintering and Microstructure of Si3N4-Materials”, Silicon Nitride 93, Edited by M. J. Hoffmann, P. F. Becher, G. Petzow, Trans tech Publications, 1993, 181-186. 35. W. Dressler, H.-J. Kleebe, M. J. Hoffmann, M. Ruhle and G. Petzow, “Model Experiments Concerning Abnormal Grain Growth in Silicon Nitride”, J. qf’ the Europ. Ceram. Soc., 16, 1996, 3-14. 36. M. Kramer, M. J. Hoffmann and G. Petzow, “Grain Growth Studies of Silicon nitride Dispersed in an Oxynitride glass’’, J. Am. Ceram. Soc., 76, 1993, 2778-84. 37. W. Dressler and R. Riedel, “Progress in Silicon Based Non-Oxide Structural Ceramics”, Znt. J. of Refractory hard Materials, 15, 1997, 1347. 38. M. Kramer, “Untersuchungen zur Wachstumskinetik von P-Si,N4 in Keramiken und Oxinitirdglasern”, Dissertation, Stuttgart, 1992. 39. G. Ziegler, W. Lehner and H.-J. Kleebe, “Nucleation affecting microstructure development of Si3N4ceramics”, Brit. Ceram. Proc., 60, 1999, 5-7. 40. S.-L. Hwang and I.-W. Chen, “Nucleation and Growth of a’-SiAION on a-Si3N4”, J . Am. Ceram. SOC.,77, 1994, 1711-18.

RLferences

797

41. H.-J. Kleebe, “Structure and Chemistry of Interfaces in Si,N4 Ceramics Studied by Transmission Electron Microscopy”, Journal of’the Ceramics Society of Japan, 105, 1997, 453475. 42. T.-Y. Tien, “Silicon Nitride Ceramics-Alloy Design”, Mut. Res. Soc. Symp. Proc., 287, 1993. 43. T. Y. Tien and C. J. Hwang, “Microstructural Development in Silicon Nitride Ceramics”, Mater. Sci. Forum, 47, 1989, 84109. 44. L. L. Wang and T . Y. Tien, “Morphology of Silicon Nitride Grown from a Liquid Phase”, J. Am. Ceram. Soc, 81, 1998, 10, 2677-2686. 45. D. Y. Yoon, Korean Advanced Institute of Science and Engineering, Report 16.6. 1999 in the IKTS. 46. Y.-J. Park. N.-M. Hwang and D. Y. Yoon, “Abnormal Growth of Faceted (WC) Grains in a (Co) Liquid Matrix”, Metal/. Mater. Trans., 27A, 1996, 1-1 1. 47. D.-D. Lee, S.-J. L. Kang and D. N. Yoon, “Mechanism of Grain Growth and a-p’ Transformation During Liquid-Phase Sintering of ff-Sialon”, J. Am. Ceram. Soc., 71, 1988, 803-806. 48. N. Hirosaki, Y. Akimune and M. Mitomo, “Microstructural Design by Selective Grain Growth of a-Si3N4”, Mat. Res. Soc. Symp. Proc., 287, 1993, 405410. 49. M. Herrmann, I. Schulz and J. Hintermayer, “Materials From Low Cost Silicon Nitride Powders”, Proc. 4th EcerS Conf ., Riccione, Ed. by C. Galassi, Gruppo Editoriale Faenza Editrice, 1995, 21 1-216. 50. N. Hirosaki, M. Ando, Y. Akimune and M. Mitomo, “Gas-Pressure Sintering of p-Silicon nitride Containing Y z 0 3 and Nd z0 3 ”,J. of Ceranz. Soc. ofJapan, 100, 1992, 817-820. 51. M. Mitomo, M. Tsutsumi, H. Tanaka, S. Uenosono and F. Saito, “Grain Growth During Gas-Pressure Sintering of p-Silicon Nitride”, J . Am. Ceratn. Soc., 73, 1990, 244145. 52. M. Mitomo, H. Hirotsuru, H. Suematsu and T. Nishimura, “Fine-Grained Silcon Nitride Ceramics Prepared from P-Powder”, J . Am. Ceram. Soc., 78, 1995, 21 1-14. 53. J. Hintermayer, G. Ernst, W. Gmohling, G.Schrol1 and W. Kobler, “Verfahren zur Herstellung von Siliziumnitrid” EP 0377132B1, 1994. 54. E. Hidseyuki and M. Mitomo, “Control and Characterization of Abnormally Grown Grains in Silicon Nitride Ceramics”, Journal of the European Ceramic Society, 17, 1997, 797-804. 55. M. Mitomo, “In-situ Microstructural Control in Engineering Ceramics”, Key Engineering Materials, 161-163, 1999, 53-58. 56. P. Branvall and K. Rundgren, “Grain growth mechanisms in in situ reinforced Si3N4”,Brit. Ceranz. Proc., 60, 1999, 7-9. 57. K. Hirao, T. Nagaoka, M. E. Brito and S. Kanzaki, “Microstructure Control of Silicon Nitride by Seeding with Rodlike 8-Silicon Nitride Particles”, J . Am. Ceram. Soc., 77, 1994, 1857-62. 58. K. Hirao, T. Nagaoka, M. E. Brito and S. Kanzaki, “Mechanical Properties of Silicon Nitrides with Tailored Microstructure By Seeding”, Journal of the Ceramic Society of Japan, 104, 1996, 55-59. 59. R. A. Andrievski, “Some high-temperature properties of silicon nitride”, High Temperature High Pressure, 26, 1994, 451455. 60. H. Emoto and H. Hirotsuri, “Microstructure Control of Silicon Nitride Ceramics Fabricated from a-Powder Containing Fine 8-Nuclei”, Key Engineering Materials, 161-163, 1999, 209212. 61. K. Hirao, H. Imamura, K. Watari, M. E. Brito, M. Toriyama and S. Kanzaki, “Seeded Silicon Nitride: Microstructure and Performance”, Key Engineering Materials, 161-163, 1999, 469474. 62. M. Herrmann, I. Schulz, Chr. Schubert and W. Hermel, “Silicon Nitride Materials with Low Friction Coefficients”, Key Engineering Materials, 161-163, 1999, 599-602. 63. M. Herrmann, I. Schulz, Chr. Schubert, I. Zalite and G. Ziegler, “Ultrafine Si3N4-Material with Low Coefficient’s of Friction and Wear Rates”, c j / B e r . DKG, 75, No 4, 1998, 3845. 64. G . Riedel, H. Bestgen and M. Herrmann, “Influence of Sintering Additives with Differing Proportions of Y2O3/AI2O3on the Sintering and Material Properties of Si,N4 Ceramics”, cJ, Ber. DKG, 75, No. 1-2, 1998, 30-34. 65. K. Watari, K. Ishizaki, S. Cao and K. Mori, “The Relationship between Thermal Conductivity and Microstructure in Si3N4 Ceramics”, Key Engineering Materials, 161-163, 1999, 2 13-2 16. -

798


66. N. Hirosaki, Y. Okamoto, M. Ando, F. Munakata and Y. Akimune, “Effect of Grain Growth on the Thermal Conductivity of Silicon Nitride”, Journal of the Ceramic Society ofJapan, 104, 1996, 50-54. 67. D. A. Bonnell, M. Riihle and T.-Y. Tien, “Redistribution of Aluminum Ions During Processing of Sialon Ceramics”, J. Am. Ceram. Soc., 69, 1986, 623-27. 68. M. Kitayama, K. Hirano, M. Toriyama and S. Kanzaki, “Anisotropic Ostwald Ripening in 0Si3N4With Different Lantanide Additives”, G . L. Messing, F. F. Lange, S.-I. Hirano: Ceramic Processing Science, 5 17--24. 69. M. J. Hoffmann, “Analysis of Microstructural Development and Mechanical Properties of Si3N4Ceramics”, M . J. Hoffmann, G . Petzow, Tailoring of Mechanical Properties of Si3N4 Ceramics, Kluwer Academic Publishers, 1994, 59-72. 70. H. J. Kleebe, G. Pezzotti and G. Ziegler, “Microstructure and Fructure Toughness of Si,N4 Ceramics”, J. Am. Ceram. Soc., 82, 1999, 1857-1860. 71. A. R. Prunier and A. J. Pyzik, “Self reinforced Silicon Nitride for Cutting Tool Application”, Key Eng. Mat., 89-91, 1994, 129-134. 72. A. J. Pyzik and D. R. Beaman, “Microstructure and Properties of Self-Reinforced Silicon Nitride”, J. Am. Ceram. Soc., 76, 1993, 273144. 73. M. Menon and I.-W. Chen, “Reaction Densification of a’- Sialon: I, Wetting Behavior and Acid-Base Reactions”, Journal of the American Ceramic Society, 78, 1995, 545-552. 74. M. Menon and I.-W. Chen, “Reaction Densification of d-Sialon: 11, Densification Behavior”, Journal of the American Ceramic Society, 78, 1995, 553-559. 75. H. Mandel and M. J. Hoffmann, “Novel developments in a-SiALON ceramics”, Brit. Ceram. Proc., 60,1999, 11-13. 76. Schutzrecht USA WO 98/23554 (1998-06-04). 77. I-W. Chen and A. Rosenflanz, “A tough SiAlON ceramic based on cc-Si3N, with a whiskerlike microstructure”, Nature, 389, 1997, 70 1-704. 78. H. Klemm, M. Herrmann, T. Reich, C. Schubert, L. Frdssek, G. Wotting, E. Gugel and G. Nietfeld, “High-Temperature Properties of mixed a’/$-Sialon Materials”, Journal of the American Ceramic Society, 81, 1998, 1141-1 148. 79. T . 4 . Sheu, “Microstructure and Mechanical Properties of In Situ 0-Si3N4/d-SialonComposite”, Journal of the American Ceramic Society, 77, 1994, 2345-2353. 80. M. Herrmann, annual report 1998, IKTS, 1998. 81. N. Camascu, D. P. Thompson and H. Mandal, “Effect of Starting Composition, Type of Rare Earth Sintering Additive and Amount of Liquid Phase on a’/p-Sialon Transformation”, Journal of the European Ceramic Society, 17, 1997, 599-613. 82. K. Urashima, Y. Ikai and S. Iwase, “Features of a Superior Strength Si3N4 Ceramic and its Applications”, 6th International Symposium on Ceramic Materials and Components for Engines, 1997, 167-112. 83. G. Riedel and H. Kruner, “Si3N4-material with high strength (1400MPa)”, Third Euro Ceramics, Madrid, 1993, 453-458. 84. R. F. Silva and J. M. Vieira, “Hot hardness of Si3N4-basedmaterials”, Journal of Material Science, 30, 1995, 5531-5536. 85. M. Yoshimura, T. Nishioka, A. Yamakawa and M. Miyake, “Grain Size Controlled HighStrength Silicon Nitride Ceramics”, J. Ceram. SOC.Japan, 1995, 407. 86. T. Nishioka, K. Matsunuma, T. Yamamoto, A. Yamakawa and M. Miyake, “Development of High Strength Si3N4Material for Automobile Parts”, Sumitomo Electric Technical Review, 36, 1998, 77. 87. N. Hirosaki, Y. Akimune and M. Mitomo, “Effect of Grain Growth of 0-Silicon Nitride on Strength, Weibull Modulus and Fracture Toughness”, J. Am. Ceram. Soc., 76, 1993, 1892-94. 88. P. F. Becher, H. T. Lin, S. L. Hwang, M. J. Hoffmann and I. W. Chen, “The Influence of Microstructure an the Mechanical Behaviour of Silicon Nitride Ceramics”, Mat. Res. Soc. Symp. Proc., 281, 1993, 147-157. 89. P. Sajgalik, J. Dusza and M. J. Hoffmann, “Relationship between Microstructure, Toughening Mechanism, and Fracture Toughness of Reinforeced Silicon Nitride Ceramics”, J. Am. Ceram. Soc., 78, 1995, 2619-24.

References

799

90. C. W. Li, C. J. Gasdaska, J. Goldacker and S. C. Lui, “Damage Resistance of in Situ Reinforced Silicon Nitride”, Mat. Res. SOC. Symp. Proc., 287, 1993, S. 473-480. 91. B. Speicher, G. A. Schneider, W. Drepler, G. Lindemann, H. Boder and V. Knoblauch, “Reliability of Ceramic Valve plates for Common-Rail Injection Pumps”, preprint Materialica 1999, 1999. 92. I. Peterson and T. Tien, “Effect of the Grain Boundary Thermal Expansion Coefficient on the Fracture Toughness in Silicon Nitride”, J . Am. Crram. Soc., 78, [9], 1995, 2345-52. 93. E. Y. Sun, P. F. Becher and K. P. Plucknett, “Microstructural Design of Silicon Nitride with Improved Fracture Toughness: 11, Effects of Yttria and Alumina Additives”, J . Am. Ceram. SOC.,81, 1998, 2831-2840. 94. A. G. Evans, “Perspective on the Development of High-Toughness Ceramics”, J . Am. Ceram. SOC.,73, 1990, 187-206. 95. H. Kessler. H. Kleebe, R. Cannon and W. Pompe, “Influence of Internal stresses on Crystallization of Intergranular Phases in Ceramics”, Acta metall muter., 40,1992, 2233-2245. 96. I. Tanaka, G. Pezotti, Y. Miyamoto and T. Okamoto, “Fracture Toughness of Si3N4and its Si3N4 Whisker Composite Without Sintering Aids”, J . of Mat. Sci., 26, 1991, 208-210. 97. J. Zhihao and Q. Guanjun, “Fatigue Behavior of some Advanced Ceramic Materials”, 6th International Symposium on Ceramic Materials and Componentsfor Engines, 1997, 760-765. 98. T. Ogasawara and Y. Mabuchi, “Effect of Humidity and Cyclic Loading Condition on Fatigue behavior of Silicon Nitride”, J . Ceram. Soc. Jap., Int. Edition, 101, 1993, 1122-1 127. 99. D. Jacobs and I. Chen, “Mechanical and Environmental Factors in the Cyclic and Static Fatigue of Silicon Nitride”, J . Am. Ceram. Soc., 77, 1994, 1153-61. 100. P. F. Becher, E. Y. Sun, K. P. Plucknett, K. B. Alexander, C. H. Hsueh, H. T. Lin, S. B. Waters and C. G. Westmoreland, “Microstructural Design of Silicon Nitride with Improved Fracture Toughness: I, Effects of Grain Shape and Size”, J . Am. Ceram. SOC., 81, 1998, 2821-30. 101. R. Schober, M. Herrmann, W. Krehler, H.-J. Richter, W. Hermel and G. Naumann, “Verdichtungsverhalten und Eigenschaften von nachgesintertem reaktionsgebundenem Siliziumnitrid (SRBSN)”, editor: J. Kriegsmann, DKG Technische Keramische Werkstofe, 3. Erg.-Lfg. 1999. 102. I. Tanaka, G. Pezzotti, T. Okamoto, Y. Miyamoto and M. Koizumi, “Hot Isostatic Press Sintering and Properties of Silicon Nitride at Elevated Temperatures”, J . Am. Ceram. SOC., 74, 1992, 752-59. 103. H. Klemm, M. Herrmann and Chr. Schubert, “High Temperature Oxidation and Corrosion of Silicon-Based Nonoxide Ceramics”, AS ME Turbo Expo ‘98,Stockholm, Sweden 1998, ASME Paper 98-GT-480. 104. H. Klemm, M. Herrmann and Chr. Schubert, “High Temperature Oxidation of Silicon Nitride Based Ceramic Materials”, Proc. 6th Internat. Symp. on Ceram. Muter. & Comp.for Engines, Arita, Japan, 1997, 57681. 105. I. Tanaka, G. Pezzotti, K. Matsushita, Y. Miyamoto and T. Okamoto, “Impurity Enhanced Cavity Formation in Si3N4at Elevated Temperatures”, J . Am. Ceram. SOC.,74, 1992,752-59. 106. H. Klemm and G. Pezzotti, “Fracture Toughness and Time-Dependent Strength Behavior of Low- Doped Silicon Nitrides for Applications at 1400C”, J . Am. Ceram. Soc., 77, 1994,553-61. 107. G. Pezzotti, K. Ota and H.-J. Kleebe, “Viscous Slip along Grain Boundaries in ChlorineDoped Silicon Nitride,” J. Am. Ceram. SOC.,80, 1997, 234148. 108. M. Herrmann, Chr. Schubert, A. Rendtel and H. Hiibner, “Silicon Nitride/Silicon Carbide Nanocomposite Materials: I, Fabrication and Mechanical Properties at Room Temperature”, J . Am. Ceram. Soc., 81, 1998, 1095-108. 109. K. Niihara, “New Design Concept of Structural Ceramics - Ceramic Nanocomposites”, The Centennial Memorial Issue of the Ceramic Society of Japan, 99, 1991, 974. 110. M. Herrmann, H. Klemm, Chr. Schubert and W. Hermel, “Long-Term Behavior of Sic/ Si3N4-Nanocomposites at 1400-1500”C”, Key Engineering Muter., 132-136, 1977-80, 1997. 11 1. G. Pezzotti and M. Sakai, “Effect of a Silicon Carbide ‘Nano-Dispersion’ on the Mechanical Properties of Silicon Nitride”, J . Am. Ceram. SOC.,77, 1994, 303941. 112. H. Klemm. K. Tangermdnn, Chr. Schubert and W. Hermel, “Influence of Molybdenum Silicide Additions on High-Temperature Oxidation Resistance of Silicon Nitride Materials”, J . Am. Ceram. Soc., 79, 1996, 2429-35.

800


113. R. Sternagel and M. Popp, “Innovationen fur Gleitlager, Walzlager, Dichtungen und Fuhrungen”, VDI-Ber., 1331, 1997, 131-138.

114. M. Rombach, R. Schofer and W. Pfeiffer, “Tragfahigkeit und Lebensdauer fur Wdzlager aus Hochleistungskeramik”, Report of a BMBF- project 03 M2107B, 1996. 115. M. Popp, R. Sternagel and G.Wotting, “Hybride- and Ceramic Rolling Bearings with modified Surface and low Friction Rolling Contact”, Euromat 1999, 1999. 116. Ch. Schubert and M. Herrmann, “Materials development of silicon nitride for ceramic ball bearings”, Report of the BMBF- project Nr. 03 M2107B4, 1996. 117. M. Herrmann, I. Schulz, Chr. Schubert and W. Hermel, “Siliziumnitridwerkstoffe mit niedrigem Reibkoeffizienten und hohem Verschleipwiderstand”, Werkstqfwoche 98, Vol. V , Symposium 6 , 1998, 241-247. 118. M. Popp and R. Sternnagel, “Neue Entwicklungen bei Walzlagern aus Hochleistungskeramik’, Werkstoffwoche 98, Vol. V , Symposium 6 , 1998, 247-253. 119. B. R. Lawn, S. K. Lee and K. S. Lee, “Fracture and Deformation Damage Accumulation in Tough Ceramics”, Key Engineering Materials, 161-163, 1999, 3-8. 120. S. K. Lee and B. R. Lawn, “Role of Microstructure in Hertzian Contact Damage in Silicon Ntride: 11, Strength Degradation”, Journal qf’ the American Ceramic Society, 81, 1998, Nr. 4, 997-1003. 121. S. W. Lee, “Tribological Characterization in Ceramic Based Composite”, Key Engineering Materials, 161-163, 1999, 593-598. 122. K. H. Habig, “Tribologisches Verhalten von Ingenieur-Keramik”, Ingenieur- Werkstoffe, I, 1989, 78-83. 123. A. Skopp, “Tribologisches Verhalten von Siliziumnitridwerkstoffen bei Festkorpergleitreibung zwischen 22°C und lOOO”C”, Dissertation 1993. 124. S. K. Lee and B. R. Lawn: Role of Microstructure in Hertzian Contact Damage in Silicon Ntride: 11, Strength Degradation. In: Journal of ihe American Ceramic Society, 81, 1998, 997-1003. 125. H. Tanaka, “A Recent Tendency of Si3N4Cutting Tools”, Advanced Materials: Ceramics, Powders, corrosion and Advanced Processing, 14A, 1994, 541-545. 126. H. K. Tonshoff and B. Denkena, “Wear of Ceramic Tools in Milling”, Journal of the Society of Tribologists and Lubrication Engineers, 47, 1991, 772-778. 127. H. K. Tonshoff, Chr. Blawit, M. Rodewald and H.-G. Wobker, “Entwicklung und VerschleiPverhalten von Si3N4-Schneidkeramik”, Mat.-wiss. u. Werkstofftech., 26, 1995, 255-262. 128. G. Brandt, “Development of ceramic cutting tools”, Maieriaux & Techniques, 85, 1997, 3-12. 129. R. D. Nixon and P. K. Mehrotra, “Optimization of Sintering Aids in Silicon Nitrides for Cutting Tool Application”, Ceramic Engineering & Science Proceedings, 18, 1997, 457464. 130. H. K. Tonshoff and C. Blawit, “Wear mechanism of silicon nitride in dependence of workpiece material”, BMFT- Project, Final Report No. 03T0028A, 1993, 1-96. 131. H. Miao, L. Qi and G. Cui: Silicon Nitride Ceramic Cutting - Tools and their Applications”, Key Engineering Materials, 114, 1995, S. 135-172. 132. G. Brandt, “Wear- and thermal shock-resistant SIALON cutting tool material”, EP 97-850038 9703 1 1. 133. US- Patent 432 325. 134. Y. R. Liu, J. J. Liu, B. L. Zhu, Z. R. Zhu, L. Vincent and P. Kapsa, “Wear Maps of Si3N4 ceramic cutting tools”, J . Mater. Eng. and Perform., 6, 1997, 671-675. 135. Gmelin Handbook, Si-Suppl, b5d2., 11&117; 1995, 184-210. 136. K. Komeya, T. Meguro, S. Atago, C.-H. Lin, Y. Abe and M . Komatsu, “Corrosion Resistance of Silicon Nitride Ceramics”, Key Engineering Materials, 161-163, 1999, 235-238. 137. M. Herrmann, A. Krell, J. Adler, G. Wotting, T. Hollstein, W. Pfeifer and M. Rombach, “Innovationen fur Gleitlager, Walzlager, Dichtungen und Fuhrungen”, VDZ-Ber., 1331, 1997,251-258. 138. T. Hollstein, T. Graas, K. Bundschuh and M. Schutze, Keramische Zeitschrijt, 50, 1998, 416 421. 139. T. Sato, Y. Tokunaga, T. Endo, M. Shimada, K. Komeya, et al., “Corrosion of Silicon Nitride Ceramics in Aqueous Hydrogen Cloride Solutions”, J. Mat. Science, 23, 1988, 3440-3446.

RL.firences

80 1

140. S. Iio, A. Okada, A. Akira, A. Tetsuo and Y. Masahiro, “Corrosion of silicon nitride ceramics in solutions. Part 3. Corrosion behavior in hot sulfuricacid and microstructure of the corroded layer”, Journal oftlie Ceran?ic Society of Japan, h t . Edition, 100, 1992, July, 954-956. 141. M. Herrmann, Chr. Schubert and G . Michael, “Korrosionsstabile keramische Werkstoffe fur Anwendungen in Walzlagern und im Anlagenbau”. Fortschrittsberichte der DKG, 14, 1999, 13~151. 142. M. Herrmann and G. Michael, “Corrosion Behaviour of Engineering Ceramics in Acids and Basic Solutions”, Britidl Ceramic Proceedings, 60, extended Abstracts, Vol I, 1999, 455457. 143. A. Okada and M. Yoshimura, “Mechanical Degradation of Silicon Nitride Ceramics in Corrosive Solutions of Boiling Sulphuric Acid”, K q Engineering Materials, 113, 1996, 227-236. 144. Y. G. Gogozi and V. A. Lavrenko, “Corrosion of High Performance Ceramics”, Springer Verlag, Berlin, 1992, 7 6 7 8 . 145. M. Shimada and T. Sato, “Corrosion of Silicon Nitride Ceramics in H F and HCI Solutions” Ceram. Trans., 10, 1989, 355-365. 146. Q. Fang, P. S . Sidky and M. G . Hocking, “The Effect of Corrosion and Erosion on Ceramic Materials”, Corrosion Science, 39, 1997, 51 1-527. 147. A. Paul, “Chemical Durability of Glasses: a Thermodynamic Aproach”. J . Mat. Sci., 12, 1977, 2246-2268. 148. K. Kanbara, N. Uchida, K. Uematsu, T. Kurita, K. Yoshimoto and Y. Suzuki, “Corrosion of Silicon Nitride Ceramics by Nitric Acid”, Mat. Res. Soc. Symp. Proc., 287, 1993, 533-538. 149. T. Sato, Y. Tokunaga, T. Endo, M. Shimada, K Komeya, K. Nishida, M. Komatsu and T. Kameda, “Corrosion of Silicon Ceramics in Aqueous H F Solutions”, J . q f M a t . Sci., 23, 1988, 3440-3446. 150. T. Sato, S. Sato, K. Tamura and A. Okuwaki, “Corrrosion Behaviour of SiliconNitride Ceramics in Caustic Alkaline Solutions at High Temperatures”, Br. Ceram. Trans. J., 91, 1992, 117-120. 151. T. Sato, T. Murakami, E. Shimada, K. Komeya, T. Komeda and M. Komatsu, “Corrosion of Silicon nitride ceramics under hydrothermal conditions” J . Mat. Science, 26, 1991, 1749-1754. 152. M. Yoshimura and S. Yamamoto, “Corrosion and alteration of ceramic materials in high temperature-high pressure water”, Serumikkusu, 30, 1995, 995-998. 153. K. Oda, T. Yoshio. Y. Miyamoto and M. Koizumi, “Hydrothermal corrosion of pure, hot isostatically pressed silicon nitride”, J . A m . Cerum. Soc., 76, 1993, 1365-1368. 154. K. Oda and T. Yoshio, “Properties of YzO3-Al2O3-SiO2Glasses as a Model System of Grain Boundary Phase of SijN4 Ceramics (Part 2)”, J . Ceraniic Soc. qfJupan, 99, 1991, 11 10-1 112. 155. S. Iio, A. Okada, T. Asano and M. Yoshimura, ”Corrosion Behavior of Silicon Nitride Ceramics in Aqueous Solutions (Part 3)”, J . o j t h e Ceramic Soc. ofJc~pan,100, 1992,965-967.


7 Boride-Based Hard Materials R. Telle, L. S. Sigl, and K. Takagi

7.1 Introduction Materials based on boron compounds have been explored for many decades because of their exceptional properties in respect to chemical bonding, crystal structure, and phonon and electron conduction. Especially in the field of energy conversion, electron emission, and neutron absorption, borides occupy many niches of application for which no other material can be employed. Until approximately 1980, the main interest in borides always came, however, from basic research aimed at the understanding of their electronic structure, being either responsible for the unique transport properties or the peculiarities in chemical bonding. It is, therefore, no wonder that the most information about borides was at that time created from the viewpoint of physicists and chemists. Although even some of the interesting thermal and mechanical properties of borides, e.g., the generally high melting point and high hardness, have already been exploited and led to application as wear-resistant parts or grinding grits a long time ago, the interest in borides increased dramatically together with the fundamental understanding of technical ceramics. The exploitation of materials with exceptional mechanical, chemical, electrical and thermal properties yielded boron compounds as potential candidates for “high-tech” applications besides the well-developed oxide ceramics, silicon nitride, silicon carbide, and hard metals or cermets. For a successful economic application, these advanced ceramics should be available without any limitations; prepared by relatively simple methods, they should exhibit a low specific gravity, high reliability, long lifetime and of course they should be available at low costs. Unfortunately, most boron-based materials do not satisfy all of these requirements and, hence, disadvantages and disappointments have to be tolerated in development and application. In particular, densification of powder-derived parts is extremely difficult due to their high amount of covalent bonding resulting in low diffusion coefficients and, therefore, costly. Furthermore, oxide impurities on surfaces and grain boundaries create undesired and unexpected effects on the properties whereas the affinity of boron to oxygen limits its use at higher temperature in an oxidizing atmosphere. Additionally, phase diagrams and crystal structures of the particular compounds are the subject of permanent revision since improvement of the purity of raw materials, more sophisticated synthesis procedures and characterization methods, such as precursor-based processing, advance chemical analysis, atomic-resolution transmission electron microscopy, and high-temperature X-ray diffraction, yielded more precise but also more confusing information about the true characteristics of the boride structures and phases.

7.2 Chemical Bonding and Crystal Chemistry of Borides

803

Besides the relatively high costs a lack in understanding is the main reason why boride-based materials are not yet in as widespread use as the other ceramics materials, although some properties such as extreme hardness make them superior to oxides and nitrides. Thus, some “eras” of boride exploration can be distinguished, one supported by the needs of the nuclear energy and weapon industry around 1960, another between 1970 and 1980 for thermo-electric power generation and coming up again at the present, the next driven by the need for exceptional wear properties between 1980 and 1990, and finally between 1985 and 1992 triggered by aerospace and military research. The interest in borides rising from a political and strategic point of view has contributed to the fact that information about borides is still difficult to obtain and even more difficult to interpret since most of the data have been restricted for a long time and their sources and quality cannot be ascertained in all cases. Thus, sometimes contradictory data are available that cannot be easily discerned as being right or wrong. Besides the “high-tech” applications of novel materials, it should not be forgotten that there was and still is a tremendous market for borides in the metallurgy of steel and iron, e.g., for antioxidizing additives in refractory linings or as alloying ingredients for the metals. Thus, this contribution is aimed at the state of the art in boride ceramics with their problems in densification, microstructural peculiarities and exceptional mechanical properties. Starting with the unique interaction of metallic, covalent and ionic types of bonding and the crystal structures of technically important compounds, phase diagrams will be presented as far as they are of technical interest. The major part consists of the description of the synthesis and properties of ceramics and cermets, reflecting the development of suitable sintering procedures and the consequent improvement of the thermal and mechanical properties.

7.2 Chemical Bonding and Crystal Chemistry of Borides The nature of the chemical bonding is the key to the physical and chemical behavior of matter. Borides possess exceptional properties due to a high amount of covalent bonding in combination with small band gaps or even metal-like transport properties. The unique interaction of metallic, covalent and ionic types of bonding results in a high melting point combined with semiconductivity or metallic electric and thermal conductivity and excellent wetting by metallic melts. The formation of many unique crystal structures found only in borides reflects again the outstanding nature of these materials.

7.2.1 Chemical Bonding of Borides The nature of the chemical bonding in boron compounds is governed by the wellknown two-electron-three-center bond, i.e., three boron atoms share two common

804

7 Boride-Based Hard Materials

electrons. These electrons are thus more or less delocalized. The resulting sp2 hybridization leads to the plane B3X3 hexagon as the main structural element in BN, B203,H3B03 and related compounds, and to the B3 triangle as a fraction of the typical five-fold symmetric icosahedron of elemental boron, the group of boranes and their derivatives. Depending on the saturation of the electron deficiency, soft and non-conducting, salt-like compounds or semimetallic to metallic materials of exceptionally high melting point and hardness and excellent electrical conductivity exist. As pointed out in the following section, the latter boron compounds may contain ionic, metallic and covalent fractions of bonding forming very stable compounds due to the well-balanced electron transfer between metal and boron sublattice.

7.2.2 The Crystal Structure of Borides Similar to silicates, the crystal structures of borides can easily be classified according to the arrangement of the boron atoms. Boron may occur as an isolated atom or form B-B bonds with an increasing degree of interconnection in the chains, double chains, layers and frameworks and combinations thereof (Fig. 1). Due to the strong covalent bonding between the boron atoms and the electron deficiency of the three-center bond a number of complex and unique structures result which

1Single

x 0-0

8 0-0

p7GG-j

Multiple

i

LUnits helping to form 3-D frameworks]

sp2-Type

Figure 1. Structural units in borides (after [64]).

B,-TYPe


805

have been the subjects of investigation for many years [1,2]. In general, compounds with a boron-to-metal ratio of less than 1.O are built up of isolated boron atoms or pairs with a low B-B interaction (e.g., Ni3B, Ru7B3,Fe2B, Cr5B3),in zigzag chains with additional, isolated B (e.g., o-Ni4B3).At a ratio of 1.0 to 1.3, infinite chains are formed which may be parallel to one or even two crystallographic axes (e.g., mNi4B3, FeB, CrB, MOB), whereas in M3B4borides double chains are predominant (e.g., Cr3B4). With increasing boron content, two-dimensional nets are stable, yielding preferential stoichiometries between M2B3 and MB4. The most important structure type group thereof is the A1B2 type, which is covered in more detail later on. Three-dimensional frameworks exist in so-called higher borides with typical stoichiometries of MB4, MB,, MBI2,and MB,,. Channels with rectangular crosssections were found in, e.g., CrB4 and MnB4, which is unique for the three-center bond of boron [3]. A rigid boron skeleton consisting of B, octahedra is a characteristic of the CaB6 structure group (important member: LaB,), whereas the UB12 structure contains BI2cubo-octahedra. Other borides of MB6 and MB12 stoichiometry or a higher boron-to-metal ratio, especially the main group element borides, can be derived from the trigonal rhombohedra1 a-boron or (3-boron structure with the B12 icosahedron as the most important structural unit. The SiB6 structure consists of a special type of boron arrangement built up by 18 boron icosahedra (BIZ units), 4 icosihexaedra (BI5 units) and 8 single atoms where some of the boron positions are occupied by Si [4]. A structural curiosity are the yttrium borides with B/Y > 25. The member being richest in boron, YB66crystallizes in cubic symmetry and contains 1584 boron atoms and approximately 24 yttrium atoms (Fig. 2). The boron is arranged in 27 (B12)13 units made up by 13 interconnected icosahedra and in 8 Bso clusters being occupied by approximately 42 B atoms. The yttrium is arranged in Y-Y pairs of most probably Y3+ state, providing all the electrons necessary to stabilize this framework [5, 61. Even this compound is of high technical importance for monochromators of synchrotron radiation. Some general systematics on chemical bonding and crystal chemistry have been published by Matkovich and Economy [7], and Aronsson et al. [8, 91, who also refer to the structural similarities in silicides and phosphides. 7.2.2.1 A1B2-Type Structures The transition metal borides of the AlB2 structure type group are of great technical interest for ceramics, as are the ternary a,p and T type borides as compounds for cermets and coatings. The A1B2 structure is conveniently described as a sequence of alternating metal and boron layers of hexagonal symmetry. The metal layers are close-packed and stacked in an A-A-A sequence, resulting in a basal-centered unit cell. The boron atoms are six-fold coordinated and situated in the center of trigonal prisms of metal atoms ( H position). Hence they generate a planar primitive hexagonal, two-dimensional, graphite-like network (Fig. 3). The total stacking sequence is then A H A - H A H . .. and belongs to the space group P6lmmm. The unit cell contains one formula unit, MB2. Since this structure is very versatile at accommodating metal atoms of various sizes and electron configurations, M could be Mg, Al, group IVa, Va, VIa, actinide or lanthanide elements.

806


: Y-Y pair

a

Figure 2. Structural units in YB66 (after [ 6] ).

Furthermore, other transition metal borides of various stoichiometries can be derived from the AlB2 structure type by introducing the metal layer positions B and C in analogy to close packings and the boron layer types K and K’, which are slightly puckered. By allowing stacking sequences such as AHAK-BHBKCH CK . . . (“Mo2B5”-type), AHAK‘-BHBK‘. . . (“W2B5”-type), or AH‘AK‘BH‘BK‘. . . (Ru2B3-type)and AK‘BK‘. . . (ReB2-type), and vacancies in the boron K layers, other structures and symmetries can be generated (Fig. 4) [9, 101. The particular molybdenum and tungsten borides have formerly often been denoted M02B5 and W2B5,respectively, but there is evidence that the homogeneity range is narrow and close to the 1 : 2 stoichiometry [9-12,1411. High-resolution TEM micrographs of WB2-containing ceramics show, however, that the puckered B layer really exists regularly alternating with the plain boron layer of the A1B2 structure (Fig. 5), yielding the space group P6/mnzc. Thus, the stoichiometry range from WB,,, [ 131 to WB2.27[ 141 arises from a boron deficiency in both kinds of layers. As


co

807

I 0 @.

Metal

Boron

A H

C

Figure 3. The AIBz structure type

an exception, A1B2-type WB2 has been produced under non-equilibrium conditions by chemical vapor reaction of WCI6 with a boron wire at 800°C [I51 with entirely different unit cell dimensions. Calculation of the band structures of A1B2-type compounds shows that no band gaps are present, and all the compounds are predicted to be electron conductors, which is in agreement with experimental results. For the main-group element

a1

-c

Figure 4. The Mo2BS structure type (after [ l I]).

K’

A, B: metal

A

H, K’: boron

808

7 Boride-Bused Hard Muterials

Figure 5. High-resolution TEM micrograph of WlB, stacks, view of [110] zone axis, distance of stacks: 13.9 A (courtesy, B. Freitag [368]).

diborides the boron 2p0 and 2p7t orbitals are the main constituents of the states at the Fermi edge, while for the transition metal diborides it is the localized metal 3d orbitals which are the predominant component of the valence and conduction bands. Since the boron sublattice is electron deficient all diborides exhibit an electron transfer from the metal atom to the boron, which gives rise to a strong ionic contribution to the bonding. In the transition metal diborides, the charge transfer decreases from 2.28 electrons in ScB2 to 1.09 electrons in MnB2 [16]; lower values have been presented by Samsonov and Kovenskaya [17, 181. The additional electrons occupy the 2p7t orbitals of the boron where the electrons are involved in both the B-B bonding as well as the metal-boron interactions. Cluster calculations of main group element diborides show that the metal-metal bonds are weak, the metal-boron interaction is significant and the boron-boron interactions are very strong. In the transition metal diborides the metal-metal bonds within the layers are considerably stronger than in main-group diborides and reaches a maximum for VB2. This internal bonding within the layers is clearly of a metallic type and is thus responsible for the metallic transport properties. The metal-metal interlayer bonding, as well as the metal-boron interactions also increase from ScB2 to MnB2, whereas the contribution of the boron-boron bonding decreases in this order. Due to the existence of vacancies in the boron layer and the possible occupation of interstitial sites by additional boron atoms, the boron sheets may also exhibit some metallic or semimetallic conductivity. The considered metallic fraction does not, however, account very much for the transport properties. In contrast, the interaction between metal and boron layers contains a more efficient metallic portion, which explains the electric conductivity along the c-axis [19]. In ideal boron

7.2 Chemical Bonding and Crystal Chemistrv of Borides

809

layers, the donor capability of the metal governs the extent of electron localization in the sp states of the boron atoms. Thus the covalent character of the B-B bonds decreases from group IV to group VI metal diborides [20]. As it has been established that the boron network is rather rigid and governs the lattice expansion in the a direction whereas the lattice dilatation perpendicular to the metallic layers strongly depends on the metal species, it seems likely that the metal atoms are distorted in some cases [21, 221. The c / a ratio is thus a function of the Y ~ ~ ratio ~ and ~ depends ~ / furthermore Y ~ ~on the ~ valency ~ ~ electron concentration ~91. Since the layers of the A1B2 type structure consist of very distinct kinds of material, namely metals and boron, exhibiting entirely different binding characteristics, more attention was recently paid to the (0001) terminating surface layers of single crystals. Hayami et al. [23] and Souda et af. [24] have studied HfB2, TaB2, and WB2 single crystals by impact-collision ion scattering spectroscopy at 5 x lo-' Pa inert gas pressure, which gives information on the ultimate surface structure by atomic shadowing effects of scattered He+ and Lif ion beams at low incident angles. As a result, the HfB2 basal plane is shown terminated by metal atoms whereas B is entirely absent even in some layers down further. In contrast, TaB2 and WB2 basal planes exhibit boron surface layers emerging from boron diffusion to the surface at above approximately 500°C, and compensating for the boron defects due to ion bombardment. This effect is attributed to the comparatively stable graphite-like boron layers and the higher affinity of Hf to oxygen compared to Ta and W. 7.2.2.2 Crystal Structure of Boron Carbide and Isotopic Compounds Boron carbide, referred to as (B4C) in brackets indicating the solid solution in contrast to B13C2,for example, meaning the stoichiometry, crystallizes in the trigonal-rhombohedra1 space group R3m. The unit cell is shown in Fig. 6. The structure may be described as a cubic primitive lattice elongated in the direction of the space diagonal with almost regular icosahedra at the corners. Parallel to the space diagonal, which becomes the c-axis in hexagonal notation, a linear chain consisting of three atoms interconnects the adjacent icosahedra. Thus the unit cell contains twelve icosahedral sites and three sites on the linear chain. If B atoms are attributed to the icosahedral positions and C atoms are considered to be situated in the linear chain, a stoichiometry of BI2C3,i.e., B4C, results. The icosahedra exhibit two topologically different positions, first the B1 position (also known as 6 h l ) which consists of a planar array of three atoms perpendicular to the linear chain around the outer atoms. Thus, this position occurs six times in the unit cell. The second distinguishable icosahedral site is the B2 (or 6h2) position, which is situated in the middle of the edges of the rhombohedra1 unit cell and accounts for a further six atoms. A special position is the centrosymmetric B3 (or 16) site in the linear chain, which is considered to be preferentially occupied by larger atoms such as A1 and Si that form solid solutions [25-291. For the binary solid solution series the question is whether the linear chain is formed by a C-C-C, a C-B-C, a C-C-B, a B-C-B, or a B-B-C array throughout the homogeneity range and how to explain the variations in the

8 10


Atom Symbol -Site B,C: @ 6h, B(1) B,C,Si :

! !

trigonal k)

OXIS

0

B,C?,A1:

Figure 6. Crystal structure of boron carbide.

stoichiometry [3042]. At that time, the majority of the authors agreed that the linear chain is of the C-B-C type and does not change with the C content, which fits the stoichiometry of the most stable compound in the system, B13C2.Accordingly, the linear chain thus contains a closed shell of ten valence electrons, which is achieved by a charge transfer to the B12icosahedra to which 38 valence electrons are formally assigned. The charge difference between B and C within the chain results in even stronger electrostatic binding forces [35], but the energetic differences estimated by density-of-state calculations are very small [43]. Since there is a deficiency of one electron in the icosahedron, additional C as an electron donor preferentially replaces B in one of the icosahedral sites [42, 44, 451. The B2 site was established as the most favorable position for such a replacement [41, 43, 46491. Thus the total structure can be written as (BI1C)-(CBC)+. To re-investigate the carbon-rich limit of solid solubility, boron carbide being in eutectic equilibrium with graphite was prepared by arc fusion or diffusion couples and subsequently analyzed with a microprobe [50] yielding a composition of B4.3C (i.e., a maximum carbon concentration of 18.87 at.-%). Optical absorption spectroscopy of this material indicated that 8 1.4% of the linear chains had a C-B-C structure

7.2 Chernicul Bonding and Crystal Chemistry of Boridrs

8 11

Table 1 Phase

Sublattice structure

Reference

and 18.6% consisted of a C-B-B array; the chains were statistically distributed [51]. Concluding from Fourier transform infrared (FTIR) spectroscopic data, Kuhlmann rt al. [51] argue that in contrast to the generally accepted continuous substitution of B I 2 for B l l C icosahedra and C-B-B for C-B-C chains with decreasing carbon content, the structure consists of statistical mixtures of these units. With decreasing C content, a growing portion of unit cells without any central linear chain is formed. Earlier, the boron-rich side of the homogeneity range was established at 8.7 mo1.-% C (i.e., B : C = 10.4) by Bouchacourt and Thevenot (1981) [49] who assigned the structural composition B12_\-C\.(C-B.yCI - ,-C) with 15.33 atoms per unit cell to the resulting formula B10.4C.The density of boron carbide at the boron-rich end is with 2.465 g cmP3lower than that at the carbon-rich corner with 2.51 g cmP3. Together with the assessment of the binary B-C phase diagram the homogeneity range of boron carbide and the other compounds was modeled by means of the Compound Energy Formalism using the following sublattice models (Table 1) where vacancies are denoted as V: In comparison to Lim and Lukas [52] who still consider the existence of a linear C-C-C chain, Kasper [53] favors the model taking vacancies in the chain into account. His calculation reveal a prediction of the occupation state of the icosahedral and chain position (Fig. 7) being in accordance with experimental results on IR absorption bands and heat capacity. As already mentioned, the linear chain can accommodate other main group elements such as Al, Si, P, As, and 0 without a change in the structure type. Solid solutions formed with (B4C), however, are only known for A1 and Si, which can partially occupy one of the positions within the C-B-C chain, e.g., A1 is placed in the centrosymmetric B3 site which causes a slight kinking of this linear array, whereas Si replaces up to one third of the carbon sites at the ends of the chain. The total solid solubility of both species is comparatively small (max. 2.5 at.-%) [28, 541, although a complete solid solution series between BI2(C,B, Si), and the silicon boride “SiB4” was considered [55]. Due to the large size of the Si atom, however, the formation of a two-atom chain is favored, as in the cases of P, As, and 0. The isotopic binary borides of these elements thus fit into the general stoichiometry B12(XVX). This arrangement favors a charge transfer to the BI2 icosahedron. The ideal structure of BI2O2 may be written as (B12)2-(00)2+, which exactly yields a filled-band configuration. In reality, for all these compounds a considerable disordering has to be taken into account, which results in the compositions B12P1,8, B12A~2.0, and B1201.82 (corresponding to B6.60)[25]. In the case of = SiB2.89,which is attributed to “SIB4”, the real stoichiometry is (B10,4Sil.6)(SiSi) a Si-Si chain and two sites in the icosahedron being partially occupied by Si [7, 21, 561. Sublattice structure modeling carried our by Kasper [53] to establish

812


XC

Figure 7. Calculated occupation of sites in the icosahedra and the chain of B4C solid solution [53].

the ternary B-C-Si system considered that the Si-Si chain may also at least partially be the structural unit for the Si-containing boron carbide solid solution [Eq. (l)]: B4+8C(Si) = (B, C)12(C2B,CB2, B2V, Si2) (1) The solid solubility limit was then simulated by a statistical combination of all the kinds of chain occupancies known.

7.3 Phase Systems Knowledge of the phase diagrams for compounds of technical interest and of the environmental phases in contact with these compounds is the key for materials development and for the understanding of materials behavior in application. Not only can the thermal stability of particular phases be calculated by means of thermodynamic data, but suitable sintering procedures can also easily be considered, and decomposition in aggressive media can be predicted. Generally recommended data books on binary and ternary systems are, e.g., those by Hansen [57], Elliott [58], Shunk [59], Mofatt [60, 611, Massalski [62], and Petzow and Effenberg [63].Nowadays, the thermodynamic data of most of the important phases are available in publications or databases and can be readily used for thermochemical calculations. In the following paragraphs, phase diagrams of the most important boroncontaining phases will be presented and discussed, starting with the binary systems,

7.3 Phase Systems

8 13

then selected ternary systems which are of technical interest. In the subsequent sections particular phase systems will be treated in respect to sintering of B4C and TiB2 or in the context of microstructural design and mechanical strengthening.

7.3.1 Binary Phase Diagrams of Technically Important Systems Many attempts have been made to correlate the binary metal boride phase equilibria, the boride crystal chemistry, and the ranking of the elements in the Periodic Table [64]; larger metals and those with unfilled d-shells favor the formation of boron-rich phases with two- or three-dimensional boron frameworks. The smaller metals having a high number of d-electrons prefer the formation of metal-rich phases with only a few boron-rich bonds, whereas more noble metals such as Ru, Rh, and Pd tend to generate defect structures. Systematic work on crystal structures and stabilities has been performed by Kiessling [l], Aronsson et al. [8], and Lundstrom [2, 651. Spear [64, 661 presented a systematic compilation of binary phase diagrams resulting in predictions of phase relations. Guillermet and Grimvall [67,68] systematized thermodynamic data of transition metal diborides with emphasis on the enthalpy of formation, the vibrational entropy, and the melting temperature, in order to account for the transition from stable to metastable phases with increasing atomic number. As a result, an entropy-related free energy term was introduced which correlates linearly with cohesive energies and melting points. By this means, stabilities of boride, carbide, and some nitride phases have been successfully predicted. In the following sections some binary systems of technical interest are presented.

7.3.1.1 The B-C System In contrast to early publications by Samsonov and Schuravlev [69] and Schuravlev and Makarenko [70] considering several boron carbide phases it is generally accepted today that only one binary phase B13C2* exists with a wide homogeneity range of 8.8 to 20.0 at.-% C, depending on temperature. This phase melts congruently at 2450°C [58] at the composition B13C2(18.5 at.-% C, 20.2 weight-% C). For the B-rich corner of the phase diagram, Bouchacourt and Thevenot [71] proposed a degenerated peritectic with elemental boron at 2075"C, according to measured element distribution coefficients. In this diagram the melting point of boron is placed at 2020°C. Since the melting point of B accepted today is 2092°C the resulting reaction with boron carbide should be an eutectic one, assuming that the non-variant equilibrium at 2075°C is correct. The maximum carbon content is usually given as 20.0 at.-%, corresponding to the stoichiometry of B4C. Beauvy [72] suggested a carbon content steadily increasing with temperature from 21.4 at.-% (20°C) to 23.1 at.-% (2375°C). Recent microprobe analyses by Schwetz and Karduck [50] indicated, however, that the maximum carbon content of fused boron carbide being in equilibrium with graphite is only 19.2 at.-% at the eutectic with carbon, and 18.5 at.-% at IOOO"C, corresponding to the formula B4.3C. The eutectic with

8 14


2600 2400

- 2200 2 - 2000 a, L

3

3 1800 L

a,

1600 a,

I-

1400 1200 1000

B

10

30

20

40.0C

-C (atom-%)-

(a) 2800

t

/

2400

I;) ! ! l 2000

a

E

Boron + boron

F 1600

carbide

1200 8.6

B (b)

18.81 I

1

10

20

I

30 Carbon, atom% ---+

40

Figure 8. The B-C phase diagram (a) calculated, (b) according to Schwetz and Karduck [50].

carbon is given at 2375 f 5°C and 29-31 at.-% C, which is in a good agreement with thermodynamic calculations [53, 731 stating 29.1 at.-% as the eutectic composition. The phase diagram is presented in Fig. 8 comparing the calculated version (Fig. 8a) with the one (Fig. 8b) suggested by Schwetz and Karduck [50]. The reaction of boron carbide with elemental boron was modeled as a eutectic at 0.2 at.-% C and 2073"C, i.e., 2 K below the melting point of boron.

7.3 Phase Systems

8 15

Table 2. Experimentally observed and calculated reactions in the B-Si system. Reaction type

Temperature “C Calculated

-+-

SiB, 0-B + liquid SIB, SIB, + liquid SIB, (Si)sol,dH liquid SiB, SiBb + liquid SiB, + (Si) H liquid SIB3 (Si),,l,, SiB6 ++

+

Experimental

Diirner [78]

Lim and Kasper Lukas [52] [53]

Armas [82]

Olesinski and Telle Abbaschian [80] [83]

2020 1898

2020 1850 1385

2054 1850 1384

1929 1989

2020 1850 1385

1377 1340

-

-

-

1270

1198

-

1377 1340

2060 1900 1385

-

1270

I358

7.3.1.2 The B-Si System The B-Si system is of particular importance for the understanding of Si as a sintering aid for boron carbide because of its chemical relationship to B and C since hot pressing of Si bearing boron carbide results in a significant reduction of the grain size and, therefore, in an improved fracture toughness and strength of 500600 MPa [74, 751. Moreover, the silicon borides have been periodically studied for their thermoelectric and thermomechanical properties. Uncertainties in the interpretation of the system arise from the three silicon borides, SiB2.89-3.65, SiB6, and SiB12-14.Both the homogeneity range and the decomposition temperatures have not yet been completely established (e.g., Elliott [58]; Ettmayer et al. [76]; Lugscheider [77]; Dorner [78]; Olesinski and Abbashian [SO]). Experimental problems in both binary and ternary phase studies are related to the comparatively high vapor pressure of Si at temperatures exceeding 1400°C. The Si-richest SiB2.89-3.65, often also denoted SiB, or SiB4, can be derived from the B4C structure by an arrangement of B12p\-Si,y icosahedra and a Si-Si chain. It is not known for certain whether “SiB3” decomposes peritectoidally at 1270°C to form SiB6 and solid Si or shows a eutectic reaction with Si at 1340-1385°C and decomposes peritectically to SiB6 and a boron-rich Si melt at 1377°C. SiB6 crystallizes in an own structure type of space group Pnnm, which is built up by 18 boron icosahedra (BI2units), 4 icosihexaedra (BI5units) and 8 single atoms where some of the boron positions are occupied by Si [4]. In spite of this versatility in B-Si exchange, its solid solubility for Si and B is generally considered less than 1 at.-%. It is agreed upon that SiB6 reacts peritectically to the next higher boride SiB12-14and a Si liquid containing 65.3 at.-% B. SiB, with n = 12-14.. .23 comprises a homogeneity range between 93.3 and 97 at.-% B and can structurally be derived from the P-B structure. The close similarity to elemental boron led formerly to the conclusion that SIB, is in fact a Si-rich solid solution of boron. SIB, undergoes a peritectic reaction with B containing 3 at.-% Si and liquid containing 9.3 at.-% Si at 2020 f 15°C as compiled by Olesinski and Abbashian [80] and calculated by Lim and Lukas [52]. Table 2 presents the experimental and calculated data of the various reactions, Figs. 9a,b show the corresponding types of phase diagrams.

816


2200

1200-

- SiB,

SiB,

-

1000

,

I

'

I

'

I

+

'

Si I

'

I

'

I

I

I

'

Si

BlMal X &Via

X 60Bro

[3

A BOVia

0

.2

B

.4

.6

Molar Fraction

.0

1.0

Si

Figure 9. (a) The binary B-Si system according to experimental results with powders of technical purity. Homogeneity fields of the silicon borides according to Ettmayer et al. [76], Lugscheider et al. [77], Armas et al. 1981 [82]. (b) The binary B-Si system according to recent calculations by Lim and Lukas [52].

7.3.1.3 The TCB System The most recent compilation on the Ti-B system was published by Murray et a/. [84]. The assessed phase diagram (Fig. lo), being in good agreement with thermodynamic calculations, consists of three intermediate phases, orthorhombic TiB (FeB type structure), orthorhombic Ti3B4 (Ta3B4 structure), and hexagonal TiB2

7.3 Phase Systems

8 17

(AlB2 structure). While TiB and Ti3B4 decompose peritectically at 2180 and 22OO0C, respectively, TiB2 melts congruently at 3225 f 25°C [85]. TiB has a narrow homogeneity range of about 49-50 at.-% B [86] and reacts eutectically with Ti solid solution at 1540°C and 7 at.-% B. The existence of the Ti3B4 phase was proven by Fenish [86] and its peritectic reaction with TiB2 and liquid was placed at 2020°C. Rudy and Windisch [85], however, omitted this phase from their binary diagram, probably because it could not be observed in meltderived samples close to the decomposition point of TiB. In 1981, Ti3B4 was reinvestigated by Neronov et al. [87] in reaction layers between Ti and B; later, in 1986, it was confirmed by Spear et al. [88] by arc melting and annealing studies. TiB2 reacts eutectically with elemental boron at 2080 f 20°C and approximately 98 at.-% B. Batzner [89] re-calculated the system with slightly different liquidus lines and omitting the little solid solubilities of TiB and TiB2, respectively, reported by Murray et al. [84]. The congruent melting point of TiB2 was set to 3216"C, the eutectic between TiB2 and B to 2059°C instead of 2080 20°C after Murray et al.

*

7.3.1.4 The Zr-B System and Other Transition Metal-Boron Systems Similar to the Ti-B system, ZrB2 is an important phase having an A1B2-type structure and a melting point of 3250°C (Fig. 11) [62]. It reacts eutectically with elemental Zr at approximately 1680°C and 86 at.-% Zr. In contrast to the Ti-B system, no ZrB or Zr3B4phases exist, but there is a ZrB12phase with a UB12structure which melts incongruently at 2030°C forming ZrB2 and liquid. ZrBlz forms a eutectic with elemental boron at approximately 1990°C which is not yet firmly established.

Figure 10. The Ti-B phase diagram (after [84]).

818


03050

---

0 B

70 80

10 20 30

m a s s % Zr 90

100

LO 50 6 0 7 0 80 90 100 at.% Zr Zr

Figure 11. The Zr-B phase diagram (after [62]).

In the other transition metal boron phase systems of groups 111, IV, and V, the MB2 phase is also the dominating compound with respect to the high melting point. Exceptions are the Y-B, lanthanide metal-B and actinide metal-B systems, which possess very stable MB4- and MB6-type compounds. In the Y-B system additional higher borides of the stoichiometry YBI2, YBZ5,and YB6, exist. In group V the number of known phases with a B/M ratio < 2 increases, and the tendency for the MB phase to be more stable than MB2 is obvious on advancing from V to Ta. Destabilization of the A1B2 structure to the benefit of the MB structure is also evident in group VI where the MB2 phase forms an individual structure type that can be derived from the A1B2 structure by the introduction of variations in the stacking sequences. In the VIIth and VIIIth group metal-boron systems the melting points of the borides decrease becoming significantly lower than the melting points of the elements. A summary of the known boron-containing binary systems is given by Spear [64].

7.3.2 Ternary and Higher Order Systems Many ternary, quaternary and higher systems containing borides have been intensively investigated basically for three reasons: firstly, to elaborate suitable sintering systems for these high-melting and thus difficult to densify compounds; secondly, to avoid probable chemical complications such as phase changes and decomposition during application; and thirdly, to investigate ways of optimizing materials properties by the fabrication of tailored composites or solid solutions. Most of these investigations have been concerned with military-, nuclear- or aerospacerelated research for new high-temperature materials, fabrication of cutting tools (e.g., transition metal borides for hard metals and cermets), or wear resistant

7.3 Phase Systems

8 19

parts (high-strength and high-toughness structural ceramics based on composites). For the last decade, more systematic studies related to the edition of alphabetic volumes on ternary systems, or to the investigation of peculiarities of chemical bonding in multicomponent phases, have also been started. Hence the data available can be divided into three groups: boron-carbon-metal/semi-metal systems (basically ceramics: sintering systems, composites), transition metal boride systems with low melting metals (for densification of hard metals, cermets and other cutting tools), and systems with two transition metals and boron (development of solid solutions of exceptional properties or with emphasis on the substitution of tungsten and other strategic metals). In the following sections some ternary systems are selected as technically important examples of these three categories. 7.3.2.1 Boron-Carbon-Metal Systems Aluminum is an effective sintering aid for B4C and Sic ceramics if combined with elemental boron and carbon. Phase relations in the B-C-A1-Si system may hence indicate suitable procedures to initiate transient liquid phase or enhanced solid-state sintering. Furthermore, liquid A1 may be used to infiltrate porous B4C bodies acting as a reinforcing phase. Although the binary boundary phase diagrams of the AI-B-C system are rather well established there is only limited experimental information on the ternary equilibria [73, 901. Six ternary phases have been discovered, B4,,A1C4 and B48A12C8, which have a B4C structure and thus are probably B12(B,C,A1)3solid solutions, orthorhombic BS1A12C8,hexagonal and orthorhombic B4*AI3Cs,and hexagonal B4A18C, denoted T. The temperature stabilities of these phases are not known. The ternary solid solubility of B12(B,C,A1)3was discovered by Lipp and Roder [27] and described in more detail by Neidhard et al. [32]. An isothermal section calculated by Lukas [73] neglecting all boron-rich ternary phases except T is presented in Fig. 12, stating that B12(B,C,A1)3is in equilibrium with a boron-rich A1 melt at above 1000°C. The B-C-Si system was first treated by Kieffer et al. [91]. The experimental data indicated a ternary equilibrium between B4C, SiB6 and SIC up to temperatures exceeding 1900°C. Thermodynamic calculations by Darner [78], Lukas [93] and Lim and Lukas [52], however, clearly demonstrated the existence of a binary phase equilibrium of boron carbide and a Si- and B-containing melt above 1560°C. The theoretical results were confirmed by hot pressing, liquid phase sintering and infiltration experiments by Lange and Holleck [75], Telle [83], Telle and Petzow [94], and Telle [54], which also yielded more details on the extension of the homogeneity field of boron carbide towards the Si-rich corner of the system B-C-Si. Discrepancies between previous and recent experiments as well as calculations arise from the binary B-Si system, in particular from the various plausible equilibria at the Si-rich corner. Furthermore, experimental problems in the ternary phase studies are related to the high Si vapor pressure at > 1400°C. The characteristics of the B-C-Si system as assessed today, comprise the stability of a BI2(B,C,S& solid solution with a maximum of 2.5 at.-% Si [28, 82, 951 being

820

7 Boride-Based Hard Materials C

Mol-% A1

Figure 12. The B-C-A1 system at 1400°C (after [73]).

in equilibrium with a boron-saturated silicon melt [54, 751. No ternary phase exists. The unlimited solid solubility between B4C and SiB2.89,as postulated by Meerson et al. [55], could not be verified. According to Secrist [96] and Shaffer [97], the B4C-SIC section is of a quasi-binary type with a eutectic equilibrium between 2250 and 2420°C and 3&35mol.-% Sic. This section is, however, of a real ternary type since the eutectic melt postulated by Secrist and Shaffer is in fact a binary equilibrium between solid carbon and Si-B-C liquid. The formation of a solid solution of B12(B,C,Si)3is accompanied by the precipitation of P-SiC which melts eutectically with SiB6 and residual solid Si above 1380°C. As a Si incorporation into the boron carbide lattice always results in the release of C or the simultaneous formation of Sic if Si is present in excess, it may be concluded that Si substitutes for C assuming that there is no carbon in boron carbide on interstitial sites. This assumption is supported by X-ray analysis and electron energy loss spectroscopy in TEM. Lim and Lukas [52], and Kasper [53] have refined again the B-C-Si system by thermodynamic calculations. A consistent data set is now available which reproduces the maximum solid solubility of boron carbide for silicon and the peritectic decomposition of the particular silicon borides. The calculated phase diagram shows some differences to that in [54], arising from the new information about the B-Si system where now the eutectic reaction SiB6 Si H liquid is preferred rather than the SiB3 Si H liquid (Fig. 9b and 13). This means that SiB3 is formed peritectoidally by

+

+

7.3 Phase Systems

821

-. W m+

@--

;

w

SiB, + Si

B I

z

W

w

SiB3 + Si

Figure 13. Two possible view of the B-Si system (after [53]).

+

SiB3 H SiB6 Sisolid.Fig. 14 shows a calculated isothermal section of the system at 2000K according to Lim and Lukas [52]. The ternary B-C-Ti system was intensively studied by Rudy et al. [99] and has been considered for the fabrication of ceramic cutting tools [loo-1031. The most recent critical assessment was published by Duschanek et al. [104, 1051 who, in contrast to the previous authors, take the re-established Ti3B4phase into account. No ternary phases have been discovered. The ternary solid solubility of the particular binary compounds except TiC0.81B0.17 is generally less than 1 at.-%. TiB2 coexists with

822

7 Boride-Based Hard Muteriuls

S Ji

C (B4C)

Molar Fraction C

Figure 14. Isothermal section of the B-C-Si system at 2000 K (calculated by Lim and Lukas [52]).

T i c l --x in a quasi-binary eutectic equilibrium at 2620 f 15°C (calc.: 2643°C) and 57 f 2mol.-% Ticl --x (Fig. 15, [99]; calculated by Duschanek et al. [104, 1051). TiB2 also forms a quasi-binary eutectic with B4C at 2310°C (calc.: 2381°C) and 88 f 3mol.-% B4C, as well as with C at 2507°C (calc.: 2456°C) and 33 f 3 at.-% C. This means, that T i c is not stable in the presence of B4C but reacts to form TiB2 C. On the other hand, TiB2-B4Ccomposites can be fabricated from TIC

+

Table 3. Experimentally observed and calculated reactions in the B-C-Si system. Reaction type

Temperature "C Calculated Dorner [78] ~~

B4C + S i c H C + liquid B4C SIB, ++ B + liquid B4C + SiB6 H SiB, + liquid S i c + SiB, ti B4C + liquid B4C + Sisolldti Sic + liquid Si + SiBs + Sic ++ liquid

+

si + SiB6

H

SiB,

+ sic

~

2166 2058 1896 1560

Kasper [53] ~~

~~

2286 2038 1840 -

1396

~

Experimental Secrist [96]

Telle [83]

>2250

n.d. n.d. 1900 I560

~~~~

~

1380

-

1380

1345

-

1345

7.3 Phase Systems 3500

-\

3000

-

I

I

I

823

I

2500 -.

s TiB2+TiC ,.x

2000 -

1500

-

Ti, B, +TIB,+TiC,., 1000

A o

I

I

I

I

0.1

0.2

0.3

0.4

TiB,

TIC

xc

Figure 15. IsoplethTiB2-TiC1-,(calculated by Duschanek ef al. [IOS]).

and B by reaction sintering. Composites of T i c , --x and TiB2 have been investigated for coherent grain boundaries [103]. Furthermore, Tic, --y is stable together with TiB up to approximately 2160°C (calc.: 2113°C) and then decomposes to Ti3B4 and liquid. The previously established ternary peritectic at 2160°C (TiB2 T i c l p.y + TiB liquid, [99]) has also to be replaced by three transition equilibria due to the existence of Ti3B4. Table 4 presents the ternary reactions accordingly [ 1051.

+

+

Table 4. Ternary Reactions in the B-C-Ti system [104,105]. Reaction Type

Reaction

quasibinary eutectic quasibinary eutectic quasibinary eutectic eutectic eutectic transition transition eutectic eutectic transition

liquid TiB, + TIC, TiB2 + C tf liquid liquid H TiB2 + B4C liquid ++ TiB, + TIC, -, + C liquid TiB2 + B4C + C liquid + TiB2 tf Ti3B4+ T i c l liquid + Ti3B4 tf TiB + TIC, -, TiB2+ B4C,,+ B tf liquid TiB + T i c , ~~z+ PTi u liquid PTi + T i c , -,tf TiB + aTi ~

-

1;

Experimental temperature ["C]

Calculated temperature ["C]

2620 2507 2310 2400 2240 2180 2160 2016 1510 890

2643 2456 2381 2394 2246 2121 21 13 2058 1535 927

824


1.0

pTitTiBtTiC

*

l.x

Figure 16. Isothermal section of the B-C-Ti system at 1400°C [105].

Figures 16 and 17 show isothermal sections of the Ti-B-C system at 1400°C and 21OO0C, respectively [104, 1051. The diagrams based upon the work of Rudy et af. [99] are wrong below 2100°C as they neglect the Ti3B4phase. Technically interesting isopleths are shown in Fig. 18 (TIC-B), Fig. 19 (TiB2-B4C),Fig. 20 (B4C-Ti), and Fig. 21 (B4C-Tic). The equilibria of the various titanium borides with TIC1-, have been studied in more detail by Brodkin and Barsoum [106]. The quaternary B-C-Si-Ti system has not yet been established completely. Evidently TiB2 and Sic form a eutectic and can be cast after arc-melting [107]. TiB2-BI2(B,C, Si)3-SiC composites are chemically stable and may form a quasiternary eutectic. Composites of these three hard materials can be prepared either by direct hot pressing or by reactive infiltration of porous B4C bodies with an Si-TiSi2 eutectic melt [108, 1091. Of technical interest is the ternary boundary system C-Si-Ti studied by Brukl [I lo], Borisova et af. [l 1 I], and Holleck [112], calculated by Touanen et al. [ 1 131 and experimentally re-examined by Wakelkamp et af. [114]. 7.3.2.2 Ternary Systems with Boron and Metals High-temperature equilibria of the extraordinarily hard borides with metallic melts bring about the opportunity for a pressureless liquid phase sintering and the fabrication of hard and simultaneously tough composites similar to hard metals but are also of interest in ceramic systems or coatings. In this section emphasis is put on binary and ternary borides which are in equilibrium with transition metals.

7.3 Phase Systems L B

A

0

0.2

TIC,.,

0.6

0.8

1.o

Mole fraction C

Ti

C

Figure 17. Isothermal section of the B-C-Ti system at 2100°C [105].

3500

I

I

0.2

I

I

0.6

0.4

Tio.5Co.5

Figure 18. Isopleth along Tic-B [105].

X.B

I

0.8

1 .o B

825

826

. +


2500 -

-

Y

2000 -

-

TiB, +B4C

1500 -

-

I

I

A o

0.2

I

0.4

0.6 Ti

Figure 20. Isopleth along B4C-Ti [lo51

0.8

1 .o Ti

7.3 Phase Systems

3500

i

827

7 L+Bd C

Figure 21. Isopleth along B,C-Tic [105].

The Ti-Fe-B System The fabrication of TiB2-based cermets was recently achieved by using Fe as a binder phase [I 15-1 191. The ternary Ti-Fe-B system was studied first by Federov and Kuzma [120] who established that no ternary phases exist. Below llOO°C, TiB2 is in equilibrium with FeB, Fe2B, a- or y-Fe, Fe2Ti, and FeTi, whereas TiB is stable together with FeTi and a-Ti (Fig. 22). Although there are still some controversies concerning the phase diagram at higher temperatures [ 1211241, TiB2 is in an eutectic equilibrium with liquid Fe at 1340°C (eutectic concentration 6.3 mo1.-% TiB2), which enables liquid phase sintering. Discrepancies exist for the phase equilibria at temperatures between 1100 and 1300°C because of the problem of whether the observed, undesired brittle Fe2B is an equilibrium phase or results from impurities in the starting materials used (Figs. 23 and 24). It is, however, obvious that oxygen and carbon contaminants introduced by the manufacturing processes of the starting powders significantly affect the composition of the liqyid phase by the precipitation of Ti-rich oxides and TIC, respectively. Since the solid solubility of B in 6-Fe is less than 0.5 at.-%, any slight Ti deficiency of the liquid phase composition will move the overall composition from the binary TiB2-Fe equilibrium into the ternary TiB2-Fe-Fe2B field. Both constituents, oxygen and carbon, do, therefore, indeed cause dramatic changes of the sintering kinetics [119].

828


B

Figure 22. Isothermal section of the Fe-Ti-B system at 1050°C (calculated by Golczewski and Aldinger [320]).

The Ti-Ni-B System In order to achieve lower sintering temperatures, also many other metallic additives such as Ni, Co, and Cr, or borides of these elements have been used at higher concentrations to allow liquid phase sintering of TiB2. These transition metals 1500

L

2

E

1300

a,

E a, I-

1200

1100

0

20

40

60

Concentration TiB,, Mol0/o Figure 23. Isopleth of the Fe-TiB, section [121].

80

100

7.3 Phase Sq’stems 1500

1400

0

1300

1200

1100

0

20

40

60

80

Concentration TiB,, Mol% Figure 24. Isopleth of the Fe-TiB, section [124]. both 1992 refs?

T: Ni,,Ti,B,

Ti Figure 25. Isothermal section of the Ti-Ni-B system at 800°C in at.-% (after [125]).

100

829

830


I ' ' TiB,

'

I

20

'

1

'

I

40

I

1

'

a t % Ni

1 " ' 1 ' 1

60

I

80

'

I

I

Ni

Figure 26. Isopleth of the Ni-TiB, section (after [127]).

react forming various metal borides with a low melting point (approx. 90&1 100°C) and a suitable wetting behavior. In contrast to the very convenient Ti-Fe-B system, the Ti-Ni-B phase diagram contains a congruently melting ternary phase, denoted z, with a composition of Ni21Ti2B6and a Cr23C6structure (Fig. 25) [125-1271. At 800°C, the z-phase is in equilibrium with Ni, Ni3B, Ni3Ti, and TiB2. A pseudo-binary eutectic with Ni exists at 1077 5"C, whereas the relationships in the TiB2-rich corner are more complicated. An isopleth (Fig. 26) across the line Ni-TiB2 reveals a solid-state equilibrium below 980°C involving TiB2, z,and Ni3B [127]. A liquid phase forms above that temperature because of the decomposition of z. Above llOO°C, Ni3B also decomposes completely in the presence of TiB2, which increases the amount of liquid phase and thus accelerates the densification. The Boron-Transition Metal Systems of Group (IV-VI) and Group VIII Elements Cemented borides with a metallic matrix have also been fabricated successfully from the ternary transition metal borides of so-called z-, cp- and a-types since these composites can easily be liquid phase sintered with metallic melts. The z-phase with a general composition of Mi1Mi'B6 or (MI, Mr1)23B6 has been observed in ternary systems with M' = group VIIIB elements such as Fe, Co, or Ni and M" = group IVB-VIB elements such as Ti, Zr, Hf, Nb, Ta, Mo or W, and with M' as the liquid phase [125-1281. It forms in coatings or particulate reinforced metals as a consequence of the reaction between diborides such as TiB2, ZrB2,

7.3 Phase Systems

831

Figure 27. Isothermal section of the B-Co-Mo system at 1000°Cwith indicated field of liquid phase formation [128].

HfB2, NbB2, and TaB2 and Ni-Co-Cr based alloys. The refractory and extraordinarily hard MB2 phases are hence in equilibrium with a comparatively soft and brittle ternary compounds. The ternary systems of these kinds of metals with boron reveal a more complex structure because of the presence of many other ternary phases denoted to as pand o-phases. The stoichiometries of these p- and o-phases are M'MI'B and M;M"B2, respectively, where M' represents Cr, Mo, Ta, or W and M" holds for Fe, Co, or Ni and solid solutions thereof. Other ternary phases have the composition and M;M1'B2\-, e.g., TaNiB2, Mo2FeB4,and Mo3CoB6.As an example, an isothermal section of the B-Co-Mo system is shown in Fig. 27 in which both the t- and the p-phases are linked with Co as the binder [128]. However, in systems with Fe replacing Co, a p-phase does not exist. Hence o is in equilibrium with liquid metal and is thus likely to form a cermet material with Fe like the z-phase Mo2Fe13B5(Fig. 28). Phase compositions located in the pseudo-binary equilibria with a metal can easily be pressureless liquid phase sintered at temperatures between 1500°C and 1700°C. Wear-resistant parts have been developed from Mo2FeB2-Fe cermets with Ni or Cr additives [129-131, 3071. Figure 29 presents an isothermal section of the Ni-Ta-B system at 950°C [126] with three ternary phases where only z is in equilibrium with metallic Ni. Quasibinary Systems of Ti, Cr, and W Diborides

Boron-based ternary phase diagrams with two transition metals have been investigated for exploitation of the diborides with an A1B2 structure, which have high hardness and a high melting point. Due to the identical crystal structure, most of the transition metal diborides have been considered of high mutual solubility

832

7 Boride-Based Hard Materials B

Fe

Mo,Fe,

Mo

Figure 28. Isothermal section of the Mo-Fe-B system at 1000°C [128].

B A

Ni

8 "O \

p"

m-Ni4B3

o-Ni, B,

0

Ni

20

,

J

TaNi,TaNi,

40

TaNi At

YoTa

60

TapNi

80

100

Ta

Figure 29. Isothermal section of the Ta-Ni-B system at 950°C (after [126]).

7.3 Phase Systems

833

[ 132-1 361. Precise experimental data are, however, rare or are not readily available. Makarenko [137] simply mentions that most of the transition metal diborides of group IV-VI are fully soluble between 2000 and 3000°C and probably also at room temperature. As the only limiting factor the difference in atomic radii was considered which should not exceed 15% like in case of V and Cr, and Hf and Zr [132]. This general statement should be taken with care since low-temperature miscibility gaps have been observed for (Cr,Nb)B2 and (Cr,Ta)B2 [133, 1381 and other quasibinary systems were proved of eutectic character. Complete solid solubility with an miscibility gap at lower temperatures was also proven for the TiB2-CrB2 system, limited low-temperature boundary solubility and large homogeneity ranges at high temperatures have been observed for the TiB2-WB2, and CrB2WB2 systems which will be discussed in more detail. Studies in the TiB2-CrB2system are difficult due to the comparatively high vapor pressure of Cr and the little densification of powder blends at temperatures below 2000°C. Hot-pressing at 2000°C for 3 w 2 0 min yields a homogeneous (Ti,Cr)B2 solid solution [136]. Koval’chenko et al. [139] have also studied the sintering behavior of pre-reacted powders and the physical properties of a (Ti,Cr)B2 solid solution containing 20 mo1.-YOCrB2. Klimenko and Shunkowski [ 1401 have investigated the activated sintering of a “titanium-chromium mixed boride” at 2000 and 2200°C being obviously a completely single phase. Unfortunately, no X-ray nor chemical analyses have been reported for the particular annealing steps. In spite of the problems upon obtaining equilibrium conditions in the overall material chemical and X-ray analyses by Telle et al. [I361 yielded evidence for the existence of an immiscibility gap below 2000°C in the TiB2-rich corner. In the temperature-concentration range between 1500°C and 2000°C and correspondingly 0 to 55mol.-% CrB2 no solid solutions possessing an intermediate CrB2 concentration were observed whereas homogeneous (Cr,Ti)B2 particles with CrB2 contents exceeding 55 mo1.-% at 1500°C could be detected in contact with (Ti,Cr)B2 phase of a CrB2 content less than 2mol.-Yo. Samples with overall concentrations of 35, 75, and 90mol.-% CrB2 have been treated by dilatometry and hot-pressing up to 2200°C to prove the formation of a liquid phase but no melting could be monitored at temperatures lower than the melting point of CrB2. This indicates that the quasi-binary system does not contain a liquidus temperature minimum nor is of eutectic character. From the equilibrium limits a room temperature solubility of TiB2 in CrB2 to 1520mol.-%0 is estimated. The melting points of TiB2 and CrB2 in Fig. 30 have been adopted from Rudy and Windisch [85] and Liao and Spear [142]. These results confirm the existence of a continuous solid solubility above 2100°C as stated by Post et al. [132]. Zdaniewski [I351 also reported on properties of a continuous series of (Cr,Ti)B2 solid solutions, however, without giving details on the temperature range. Post et al. [132] have investigated the CrB2-WB2 system at 2 1 0 0 f 100°C but note that no exact limits of the homogeneity range could be determined. Telegus and Kuz’ma [ 1431 studied 15 compositions in the quasi-binary CrB2-WB2 section represented as a part of an isothermal section of the B-Cr-W system at 1500°C. They report a maximum solid solubility of 6 at.-% Cr in W2B5 corresponding to 3mol.-% CrB2 in WB2 whereas the published section shows 16.7mol.-% CrB2 being dissolved in the W2B5 structure. This work allows an extrapolation of

834


3 500

..

0

30

20

10

Ti62

40 50 60 70 80 Mole - O h CrBz

90

100

CrBz

Figure 30. The quasi-binary TiB2-CrB2system.

the homogeneity range of (W, Cr)B2 to about 89 mo1.-Yo CrB2 or less at 1500°C. Moreover, in contrast to Rudy [104, 1441 the binary homogeneity range of the “W2B5” phase is more extended. In the CrB2 corner Telegus and Kuz’ma report less than 2at.-% W in solution which could not be confirmed in the materials investigated by Telle et al. [ 1361who synthesized the system up to 2070°C. Unfortunately, the liquidus equilibria were not established, so the accurate composition of the eutectic is not determined yet. At temperatures between 1750°C and 2000°C samples with overall compositions between 40 and 86 m0l.-YOWB2 consist of two phases, namely (Cr,W)B2 solid solutions with WB2 concentrations up to 3840mol.-% and (W, Cr)B2 phases containing always less than 9-14mol.-% CrB2. These results allow the formulation of the boundaries of the homogeneity fields 2500 Liquid

-- -

-_____ ; ; 2200 -----==--------__ 0

W

-

*.-;: I

,, I

2365

, ’

2000 -

2030250

L

3

c

z W

a

E,

1500

..

-

(Cr,W)B, + (W,Cr)B,

1-

10000

CrB2

I

10

,

20

I

I

30

40

I

.

50

60

I

70

Mole -YOWB,

Figure 31. The quasi-binary CrB2-WB2system.

I

80

,

90

100

WB2

7.3 Phase Systems

835

3500 3225

0 -

3000

0

f 2500

,2365

I

Y 2000

+W 1500 1000

/ .L

(Ti.W)B, AIB,-type

+

(W,Ti)B, W2B,-type

*

0 1 0 20 30 10 50 6 0 70 80 90 1 0 0

TIB,

Mol% WB,

WB,

Figure 32. The quasi-binary TiB2-WB2system.

of (Cr, W)B2 and (W, Cr)B2 phases with AlB2 and W2B5 structure, respectively. As shown in Fig. 31, the CrB2-WB2 system is of a simple eutectic type (T, = 2030 f 50°C) with a solid solubility of CrB2 in WB2 of 1Omol.-% and WB2 in CrB2 of 37mol.-% at eutectic temperature. The tungsten content in CrB2 decreases strongly with decreasing temperature whereas the chromium content in the W2B5 structure is almost temperature independent. The quasibinary TiB2-WB2 system as a part of the Ti-W-B system has been treated by Telle et al. [136] up to 2250°C. The borders of the (Ti, W)B2 homogeneity range have been intensively studied between 1500°C and 17OO"C, at 2000°C and around the quasi-binary eutectic temperature. Thus a huge solid solubility up to approximately 63mol.-YOWB2 could be established. In the W-rich corner of the phase system, the TiB2 concentration in (W, Ti)B2 is with 1-2 mo1.-% almost constant up to 1700°C. At higher temperatures the solid solubility of the W2B5-type structure for TiB2 increases slightly to approximately 3 mo1.-% at the eutectic temperature. A further decrease in TiB2 along the supersolidus equilibrium could be established up to 2250°C as expected according the phase rules. Taking inaccuracies in the temperature measurement of the dilatometric experiments into account, TiB2 and WB2 react eutectically at T, = 2230 40°C and 90 f 3 mo1.-% TiB2 (Fig. 32). The solid solubility of WB2 in TiB2 at that temperature is approximately 63 mo1.-% whereas the solid solution of (W,Ti)2B5 type contains only 3mol.-% TiB2 at the eutectic equilibrium. The homogeneity range of the (Ti, W)B2 solid solution narrows significantly with decreasing temperature and is 46-49 mo1.-Yo at 2000°C and 810mol.-% WB2 at 1500°C. Post et al. [I 321 have treated this system before and found a homogeneity range of the A1B2 structure type to at least 50mol.-Yo WB2. The preparation of a pure single phase WB2 of AlB2 type structure has failed. Pastor [134] also states that the WB2 concentration in (Ti, W)B2 may exceed 50mol.-%. Yasinskaya and Groisberg [145] have examined the interaction of TiB2 with metallic W up to 2700°C. Although no data are given in the text a maximum solubility of 5 at.-% metallic W can be taken

*

836


B

Figure 33. The ternary Ti-W-B system at 1400°C (after [146]).

from the given pseudo-binary section TiB2-W. The corresponding points are shown as a quadrangles in Figs. 32 and 33. The isothermal section of the B-Ti-W system presented by Kuz’ma et af. [146] reveals a maximum of 6-8 mo1.-% WB2 dissolved in (Ti, W)B2 and 34mol.-% TiB2 in (W, Ti)B2 at 1400°C (Fig. 33). These data are more or less consistent with the phase diagram presented by Telle et al. [136] and are thus shown as triangles in Fig. 32. Kosterova and Ordan’yan [147] have retreated the ternary B-Ti-W at 1400°C in order to specify more accurately the phase equilibria in the metal-rich region but have basically found the same situation in the TiB2WB2 section as Kuz’ma et al. According to their isothermal section the TiB2 content in (W,Ti)B2 is considered in the range of 10mol.-%, which is certainly overestimated. Contrary to Kuz’ma et al. [146], Kosterova and Ordan’yan [147] take the existence of a WB4 phase into account being in equilibrium with a W-rich TiB2 and WB2. The presentation of a metastable B-Ti-W phase diagram at room temperature by Ariel et al. [148] notes a maximum W content in (Ti, W)B2 of 2 at.-%, which was obviously measured on samples of the TiB2-W section and thus corresponds to the data of Kosterova and Ordanyan [147]. Unfortunately, the more recent publication on the formation of A1B2 structure type solid solutions [135] does not give any data on temperaturexoncentration relations in this system. A partial re-treatment of the Ti-W-B system by Ahn et af. [149] for the development of TiB2-W cermets was based on the isothermal section of Kuz’ma et al. [146] but proved the existence of a distinct WTiB2 phase in the WB-TiB monoboride section. Hot isostatic pressing of TiB2 with 1-30vol.-% W at 1900°C resulted in TiB2-WTiB2 composites whereas a higher amount of W yielded WTiB2-W cermets with little residual TiB2. The WTiB2 possesses P-WB structure and contains 25.7 at.-% W, 24.8 at.-% Ti, and 49.3 at.-% B which is close to the stoichiometry 1 : 1 :2. In spite of this, the isothermal section given by Ahn et al. [149] show an extension of the homogeneity range to WB.

7.4 Boron Carbide Ceramics

837

Current EXAFS- and HR-TEM studies in the TiB2-WB2by Pohl et al. [150,151], Mitra and Telle [ 152, 1531, and Schmalzried and Telle [ 1541 indicate, however, that annealing of (Ti, W)B2 solid solutions in the TiB2-“W2B5”two phase region at lower temperatures results not only in the precipitation of ‘‘W2B5’’ particles but, besides segregation of metallic glasses and elemental boron, also of metastable pWB-type phases and spinodal ex-solutions of W-rich A1B2-type phases which have formerly been erroneously interpreted as “W2B5” [136, 1551. This is attributed to the much faster diffusion of boron compared to Ti and W [I501 and shows that equilibrium conditions are difficult to obtain. The quasi-ternary CrB2-TiB2-WB2 system is of special interest in respect to anisotropy effects of the thermophysical properties of the diboride solid solutions and has been studied between 1900°C and 2100°C by Telle et al. [I361 up to now. It is obvious that homogeneous ternary (Ti, Cr, W)B2 solid solutions of A1B2 type structure are formed in a wide range of composition. The miscibility gap emerging from the TiB2-CrB2edge extends to towards the TiB2-WB2edge but disappears at < 80 mo1.-% CrB2, in the WB2 rich corner a W-rich ternary liquid was observed. It is concluded that the quasi-ternary system exhibits a ternary eutectic at 1800 f 50°C. Close to the TiB2-WB2edge, a relatively small amount of chromium stabilizes a (Ti, Cr, W)B2 solid solution with a high tungsten content exceeding by far that of the quasi-binary compound. Also the extend of TiB2 solid solubility in “W2B5”is increased by the presence of CrB2. But also in this system, the previously mentioned metastable phases have been observed which make the interpretation of the entire equilibria difficult.

7.4 Boron Carbide Ceramics Boron carbide was first prepared by Joly [ 1561 and labeled as B6C by Moisson [ 1571. In 1934, Ridgeway [158] suggested the composition B4C, which is still under controversial discussion. As discussed in Section 7.2.2.2 in connection with the crystal structure, the composition as assessed today ranges from B4,3Cto B10.4C.Due to the fabrication process where carbon is used to reduce boron oxide, the composition of the commercially available boron carbide is close to B4C.

7.4.1 Preparation of Boron Carbide Boron carbide is prepared for most purposes according to a technical-scale process providing comparatively cheap powder which is used for grinding and lapping grits or for the boronization of steel, superalloys and other non-ferrous metals. For the production of ceramics this material must be processed further on by intensive milling, favorably ball-milling, and subsequent cleaning from impurities. Other more sophisticated methods, especially vapor phase reactions and synthesis starting with metallo-organic precursors are very costly and, therefore, restricted to scientific applications or special purposes of high profit gain.

838


7.4.1.1 Technical Scale Production Boron carbide powder is produced on a technical scale by the carbothermic reduction of boron oxide with graphite or petroleum coke 2B203

+ 7 C + B4C + 6CO

(2) The process is carried out in huge electric arc or resistance furnaces and is comparable to the Acheson process. The reaction takes place between 1500 and 2500°C, is strongly endothermic and requires 1812 kJ/mol, i.e., 9.1 kWh/kg [159, 1601. Since large quantities of carbon monoxide (approximately 2.3 m3/kg) are formed, the reaction of Eq. (2) is accelerated to the benefit of B4C. Both volatilized boron oxides and carbon monoxide generate an internal Boudouard equilibrium within the raw material mixture and thus contribute to a self-propagating purification process, which can be expressed by B203+3CO-+2B+3CO2 2 co * c02 c

+

4B

+ C + B4C

(3) (4) (5)

The furnace is usually cooled externally to limit the loss of volatile materials and hence the outer mantle stays unreacted. The core contains blocky boron carbide of relatively high purity (total metallic impurities < 0.5 mass-%), reproducible stoichiometry (B/C ratio = 4.3) [50], and several percent of residual graphite. The chunks are crushed and milled to the final grain size. A similar process with lower productivity is used for the synthesis of high-purity B4C of controlled stoichiometry. At temperatures of 1600-1800°C, hydroboric acid reacts with acetylene black, high purity sugar, or ethylene glycol in a vented tube furnace: 4H3BO3

+ 7 C + B4C + 6 H 2 0 + 6CO

(6)

Powders of 0.5-5 pm particle size are obtained [161]. Boron oxide can also be converted to boron carbide by exothermic magnesiothermic reduction in the presence of carbon black at 1000-1 800°C [ 1621. 2B203

+ 6Mg + C

B4C

+ 6MgO

(7) The process is performed by single point ignition (thermiteprocess) or in a carbon tube furnace in a hydrogen atmosphere. The problem is the removal of magnesia, magnesium borides, and unreacted magnesium metal which are usually extracted by hydrochloric or sulfuric acid. Since MgO acts as a grain growth inhibitor, submicron powders with Mg compounds as the only impurities are produced [163, 1641. Further chemical refinement by high-temperature vacuum treatment, however, induces an undesirable coarsening of the particles. The 1990 total annual production of boron carbide in the western world is estimated at approximately 500-600 t. -+

7.4.1.2 High-Purity Material In laboratory-scale production, boron carbide can also be synthesized in the form of high-purity powders or coatings (e.g., [165]):


839

(i) from the elements by arc melting at 25OO0C, or self-propagating synthesis above 1100°C: 4B

+ C + B4C

(8) (ii) by chemical vapor deposition reducing boron trichloride in the presence of carbon in a hydrogen atmosphere: 4 BC13

+ 6 H2 + C -+ B4C + 12 HCI

(9) (iii) by pyrolysis of boron trihalides with methane or carbon tetraiodide as carbon carriers, in high-frequency furnaces: 4 B(C1, Br)3 + CH4 + 4 H2 4 BI3

+ CI4

900- 1800°C 900-11OO’C

+ 12 HCl(HBr) B4C + 8 12 B4C

(10)

The latter methods yield boron-enriched solid solutions with a maximum of 20.4 mass-% carbon. Very fine boron carbide powders of spherical shape and 20-30nm in size have been prepared by chemical vapor deposition according to (iii). In an Ar-H2-CH2BC13 atmosphere a radio frequency plasma produces stoichiometries between BIS& and B3.& [33, 1661. Also laser-induced pyrolysis of similar gas mixtures with or without acetylene has been employed for the preparation of nano-sized particles [167]. With similar success, composites of B4C and S i c have been produced by the pyrolysis of boron-containing polysilanes [ 1681. The general problem associated with the production of submicron powders by pyrolysis is the comparatively low yield of these highly expensive procedures and the excess of free carbon which cannot usually be avoided. The advantages of high purity and well-defined composition are limited due to the pick-up of oxygen by the large and hence extremely reactive surface area of the particles when exposed to air.

7.4.2 Sintering of Boron Carbide Sintering of covalently bonded materials is generally much more difficult than densification of oxide ceramics or metals. This is not only due to the low selfdiffusion (poor tendency towards grain boundary and volume diffusion), high ratio of grain boundary-to-surface energies and high vapor pressure of particular constituents (strong tendency towards surface diffusion and evaporation recondensation), but also due to their extreme sensitivity to environmental factors such as sintering atmosphere, traces of contaminants, particle size and shape distribution, temperature gradients, etc. The phenomenon of a “terminal” density, i.e., the density obtained after sintering which is far below the theoretical density for pore closure, above which neither an increase in the temperature nor a prolonged sintering time would assist further densification, was frequently observed for B4C, S i c and Si3N4.The reason for this, as proposed by DeHoff et al. [169], Greskovich and Rosolowski [170], and Prochazka [171], is that upon sintering the decrease in

840


Density I%]

Figure 34. De-Hoff diagram showing the loss of specific surface area during densification of various ceramics (after [ 1711).

the specific surface area (driving force for densification) is consumed to a much greater extent for pore and particle coarsening (Ostwald ripening) than for grain boundary movement and pore removal. Figure 34 shows a so-called DeHoff diagram correlating the specific surface area and the fractional density on which the path of an ideally densifying material is illustrated by the diagonal line. The lines plotted for A1203, Sic and B4C powders make it obvious that in the first step of sintering, the surface energy is dissipated very fast due to coarsening [171]. Since a doubling in particle size corresponds to a decrease in the densification rate by a factor of ten, it is no wonder that densification comes to an end before pore closure is achieved. As pores are favorably removed by grain boundary movement, it is essential to generate a pore size distribution below a critical size above which pores are stable or even tend to grow (i.e., the driving force for pore shrinkage is 5 0) and to induce grain growth at moderate rates so that vacancies may be suitably removed from the surface of the pores. Another reason for the poor sinterability is the extraordinarily high vapor pressure of boron oxides and suboxides. Since boron carbide powders are generally coated by a B203 layer [172] which quickly reacts to form boric acid, H3B03,in humid atmosphere, vapor phase reactions are active at higher temperatures, in particular above 15OO0C, providing a fast transport of boron compounds. Redox reactions such as B2°3

vap.

+ B4Csolid *

Bovap.

+ covap.

may be shifted to the benefit of the left or right side of Eq. (12) depending on the local chemical potential of the particle surface being defined by the local curvature. The Gibbs-Thomson equation correlates not only the particular chemical potential p with the local surface radii but also the local vapor pressure. As a result, B 2 0 3 may pick up boron from B4C particle areas with convex surface curvatures and move it

7.4 Boron Curhide Ceramics

84 1

as BO to areas of concave curvature where B4C is deposited again due to the lower vapor pressure compared to the starting point. This reaction releases B203 again, which starts the process cycle another time. Depending on the oxygen vapor pressure also other boron suboxides such as B202and B 2 0 may be involved. This mechanism, denoted as evaporation/reprecipitation in the science of sintering, is known to contribute significantly to grain growth without any shrinkage (e.g., [173]). In conclusion, the sintering of boron carbide requires (i) oxygen-removing additives such as graphite, carbon black or organic deoxidation agents such as formaldehyde resin, (ii) very fine powders of high surface area and therefore high driving force, (iii) high temperatures to enable grain boundary and volume diffusion, and, if everything fails, (iv) high pressures.

7.4.2.1 Pressureless Sintering Densification of pure stoichiometric boron carbide is extremely difficult. Due to pore eliminating mass transport the high fraction of covalent bonding (> 90 YO), mechanisms such as grain boundary and volume diffusion become effective at temperatures above 2000"C, i.e., at temperatures close to the melting point. At lower temperatures, surface diffusion and the already discussed evaporation-recondensation reactions are the favored mechanisms, resulting in neck formation (increase of contact area), pore coalescence and particle rounding (decrease of specific surface area), or euhedral growth of particles by vapor phase reactions, respectively. Grabchuk and Kislyi [ 1741 proved that the regime of predominant surface diffusion extends from 1500°C to 1800°C, whereas sublimation occurs above 1800°C with boron being the more volatile species. Only the latter sintering mechanism causes an enhanced shrinkage of the ceramic body. However, a poor tendency towards plastic deformation, a high resistance to grain boundary sliding, and low surface energies hinder considerable particle rearrangement or shape accommodation before grain boundary or volume diffusion is effective. Even submicron powders cannot thus readily be densified completely by pressureless sintering if they are not mechanically or chemically activated. The general preconditions for the densification of pure stoichiometric boron carbide are to start with very fine powders (preferably << 3 ym) of low oxygen content and to use temperatures in the range of 2250-2350°C. Above 2000"C, a rapid coarsening occurs, which usually results in unremovable, entrapped residual porosity. Sintering parameters and obtained densities have been presented for many compositions in the homogeneity range of boron carbide by, e.g., Adlassnig [175] (2250-2300°C; 80-87% density; 2450°C: >goo/,) and Grabchuk and Kislyi [176] (2300°C: 99-99.5% density) who used finer powders. A considerable reduction in temperatures can be achieved by microwave sintering using 2.45 GHz radiation. After a 12 min treatment at 2O0O0C, 95% of the theoretical density was obtained [177]. The energy conservation compared to hot-pressing is, however, rather low. Starting from powders with a smaller particle size, e.g., < 1 ym, results in lower sintering temperatures and higher final densities. Boron carbide, however, becomes pyrophoric with increasing specific surface area and is hence strongly oxygen-loaded or even dangerous in handling.

842


Activation of grain boundary and volume diffusion and thus densification at lower temperatures is possible by increasing the density of point defects or dislocations: (i) mechanically by high-energy milling (attrition milling); (ii) by doping with trivalent ions, which substitute for carbon and thus introduce electron deficiencies and vacancies, e.g., by adding boron or aluminum; (iii) by introducing sintering additives, which remove oxide layers on the surface of the boron carbide particles and thus increase the surface energy, e.g., by adding carbon, aluminum carbide, silicon carbide or related compounds, which also inhibit exaggerated grain growth [178]. Other methods make use of additives which possess a comparatively low melting point and have a suitable wetting behavior on boron carbide to provide a rapid path for mass transport via the melt and thus to initiate liquid phase sintering. Dense bodies of boron carbide have also been obtained by liquid phase infiltration of highly porous powder compacts or presintered ceramic bodies. Kislyi and Grabchuk [179] reported that volume diffusion is enhanced in the boron-rich area of the homogeneity range of boron carbide due to the generation of point defects. In fact, pressureless sintering with boron additives results in an onset of shrinkage at temperatures which are 300 K lower than those required for stoichiometric B4C. Since aluminum also substitutes for carbon, a similar mechanism may be activated. Accordingly, 95-99.2% of the theoretical density is obtained at 2100-2200°C with 3-15%, preferentially < 1% A1 additive (e.g., [ISO]). Other Alproviding sintering additives are A14C3,A1203 and AlF, [180-1821, which also use carbon or fluorine as deoxidizing agents. The use of metallic additives is limited due to the low thermodynamic stability of boron carbide reacting with metals to form metal borides and free carbon, except in the case of Cu, Zn, Sn, Ag, and Pb. Nevertheless, Mg, Cr, Co, and Ni have been used [181,183,184], with minor success. Stibbs et al. [185] have proposed additions of 5-10 mass-% Al, Mg or TiB2 to obtain > 99% density between 21 50 and 2250°C. TiB2, CrB2, and W2B5 additives inhibit grain growth by grain boundary pinning or, as in the case of W2B5 at > 2220°C, initiate liquid phase sintering if a eutectic reaction occurs [186]. Sintering of submicron powder with an addition of 1 mass-% Be2C resulted in 94% density when sintered between 2200 and 2280°C [187]. Kriegesmann [188] made use of the previously discussed vapor phase reactions and considered a “recrystallization” process to produce intentionally porous boron carbide resembling “recrystallized” Sic. Since the vapor pressure of the boron species increases with higher surface area an smaller radii of curvature the starting material consists of powders with two maxima of grain size distribution. The smaller portion of the particles serves as a feeding material for the volatile boron species, which are deposited at the necks between the coarser particles. The process is carried out at >2000°C and a highpurity ceramic of defined pore size and grain size is produced. The only technically important sintering additive for boron carbide is carbon, as discovered almost simultaneously by Schwetz and Vogt [189], Henney and Jones [190], and Suzuki et al. [191], which today allows the routine production of dense parts. An amount ranging from 1 to 6 mass-% is sufficient to obtain almost theoretical density. Schwetz and Grellner [ 1921 added phenolic resin (corresponding to 1-3 mass-% C) to a submicron B4C powder and obtained >98% density at 2150°C. The resins are pyrolized at temperatures up to ~ 1 0 0 0 ° C and they leave a


843

homogeneous layer of amorphous carbon on the B4C surface, which promotes sintering. The sinter activation was attributed to an increase in surface energy due to the removal of oxide layers. Moreover, residual graphite particles which have been observed at the grain boundaries may inhibit surface diffusion and evaporation and may also control the grain boundary movement [178,193]. Firing of B4C with 6 mass-% C additive at 2220°C results in a 97% dense microstructure of 1-5pm equiaxed particles, i.e., almost no coarsening has occurred. Abnormal growth of individual grains to 10-30 pm starts above 2235°C; at 2250°C, extensive Ostwald ripening and twinning is observed. The local growth of faceted grains exceeding 500 pm in size was attributed to liquid phase sintering processes due to the presence of low-melting impurities [ 1781. The method of in-situ pyrolysis of organic additives such as novolac-type resins to amorphous carbon was also studied by Bougoin et al. [194]. The advantage of the precursor method is the improved homogeneity of the carbon distribution and the extraordinarily fine resulting average grain size of 2 pm and less. Furthermore, the resin may act as a molding aid upon cold isostatic pressing, or may even be the plasticizer for injection molding. Thus complicated parts can be fabricated easily and subsequently pyrolyzed and pressureless sintered. The drawbacks of the resin route, i.e., the fairly complex handling of the powders and environmental problems during pyrolysis, have triggered research towards suitable resin substitutes. The direct blending of amorphous carbon with B4C has been suggested as an alternative, but until recently rather porous microstructures were observed when B4C is sintered with carbon black, the suspected reason being the insufficient distribution of the sintering aid. An improved route of dispersing amorphous carbon in B4C powders has been developed recently by Matje [195] and studied further by Sigl and Schwetz [116,197]. Boron carbide powder was doped with carbon black resulting in 5, 6, and 7 mass-% of free carbon in the as sintered bodies. Sintering was carried out in a 1Ombar Ar atmosphere for 2 h at temperatures between 2050 and 21 75°C. All compositions sinter at surprisingly low temperatures and approach nearly full density at around 2150°C. These sintering temperatures are significantly lower ( ~ 1 0 0 K )than those reported for the phenolic resin as a dopant. As shown in Fig. 35 densification is hardly affected by the amount of free carbon. This observation agrees well with the results of Dole and Prochaska [178] who found that the density of B4C becomes independent of the carbon addition above 733 mass-% C. A typical microstructure of B4C doped with 5 mass-% free carbon and sintered at 21 50°C is shown in Fig. 36. In general the microstructure is very uniform and fine grained, but some grains have already entered the regime of discontinuous growth. This process is more clearly reflected in Fig. 37: While the grain size increases rather slowly below 2150"C, a rapid rise of the mean intercept length is observed beyond that temperature. Combined additives consisting of carbon and a metal carbide or horide make use of both the deoxidizing effect of carbon and the diffusion enhancement by the metal, or the grain growth inhibiting and reinforcing effect of nonreacting phases, e.g., B C, S i c + C, S i c + Al, or TiB2 + C (e.g., [194, 198,1991). Weaver [200,201] sintered relatively coarse (average size 9 pm) boron carbide powders with 2 4 0 mass-%

+

844

7 Boride-Based Hard Materials 1

5 Q

s c. .rn

s C

i

0.98 0.96 free carbon, wt%

a, 0.94

.->

3

2

0.92

2000

2050

2100

2150

2200

2250

Sintering Temperature, "C Figure 35. Densification of boron carbide with 5-7 mass-% C (after 11971).

S i c and 0-10 mass-% A1 additives to > 85% density. In similarity to the decomposition of an A14C3 addition, metallic A1 is dispersed very homogeneously by evaporation and condensation in the still porous ceramic body [180,202]. Starting from submicron powders, Schwetz et al. [203] prepared composite materials consisting preferentially of 9-10 mass-% Sic and 1-3 mass-% C with 97-99.7% density at 2000-2100°C. Residual porosity was removed completely by a post-HIP (hot isostatic pressing) treatment at 1950-2050°C. Both C and S i c may also be introduced in the form of organometallic precursors, e.g., by infiltration of a porous B4C body with polycarbosilane and phenolic resin, or by coating of boron carbide powders followed by pressing and subsequent pyrolysis. [107,204]. Bougoin and Thevenot

Figure 36. Microstructure of boron carbide pressurelessly sintered with carbon black (after [ 1971).


0

L-

2050

'

1

2100

1

8

1

845

I

8

2150

2200

Sintering Temperature, 'C Figure 37. Grain growth In boron carbide pressurelessly sintered with carbon (after [197]).

[205] reported on the fabrication of composite bodies containing 5 mass-% S i c residue but no free graphite. Sintering for 15 min at 2175°C results in a density of > 92%. The microstructure of 7.5 mass-% polycarbosilane material exhibits relatively large, faceted B4C particles (20-50 pm) with entrapped pores and local enrichments of S i c implying that liquid phase sintering may be active. Increasing the amount of polycarbosilane to 17.5 mass-% results in a more uniform microstructure that is characterized by B-to-a transformed S i c platelets of 50 pm size. Pressureless sintering with liquid phases was studied in the B4C-A1 and B4C-Si systems. Since A1 melts below 600°C and exhibits a significant vapor pressure at only slightly higher temperatures, the equilibrium between 1000 and 188O"C, at which liquid A1 is stable with an Al-saturated B12(B,C,A1)3 solid solution (Fig. 12) [73], cannot readily be utilized for liquid phase sintering. Moreover, problems in wetting due to oxide layers on the surface of both A1 and B4C powder particles have to be overcome. As shown by Halverson et al. [206], it is more effective to infiltrate compacted or presintered porous B4C bodies with liquid Al. Since the resulting material is a metal-reinforced B4C cermet rather than a liquid-phase sintered B4C ceramic, it will be treated in detail in one of the following sections. According to the B-C-Si phase diagram, liquid phase sintering of B4C should generally be possible above 1560°C with a B-rich Si liquid [54,94]. Starting from powder mixtures of B4C, B, and Si, the first unit is generated at 138OoC,which is in equilibrium with SiB6 and S i c and thus may cause the partial decomposition of B4C. Above 1560"C, however, a B12(B,C,Si)3solid solution is in equilibrium with the liquid. Besides the complications due to iterative changes of the wetting behavior due to dissolution and precipitation reactions upon heating, a strong limitation on the final densification arises from the continuous evaporation of Si, which may cause degassing channels and thus even open porosity. Recently, boron carbide was successfully densified with TIC by pressureless sintering [207]. Since TIC reacts with B4C by the formation of TiB2 and free carbon,

846


’05 In

1

+ 35 vol%

lo4

I

U 25.1 vOI% ----.f- 13.7 VOl% 4.5 vol%

m

n E

v

103

u

A

.-

c v)

lo2

0

0

.-In

>

10’

loo 0

2

4

8

6

10

12

14

PH Figure 38. Viscosity as a function of pH and volume fraction of B4C in aqueous suspension (after [2081).

densification is enhanced by the same mechanisms described before. Details about the resulting B4C-TiB2composites will be discussed later on. Since pressureless sintering to relatively high density allows the fabrication of complex parts compared to hot-pressing research on appropriate molding procedures for boron carbide has increased since 1990. Slip casting was studied by, e.g., Williams and Hawn [208] using aqueous solutions. Surface contamination by hydroboric acid was removed by washing the powders with water or alcohol. Electrophoresis indicated that B4C is negatively charged in water above pH 1. Stable dispersions with > 30 v0l.-YOsolid fraction and low viscosity were obtained and successfully slip-cast at pH > 6 but boric acid destabilized suspensions above pH 7. A slight reoxidation after storing the powders in air for 6 months caused an increase of viscosity at pH > 9 but yielded > 58% green density compared to 45% of as-received powders. The highest casting density of 63% was obtained after washing of the powder. Fig. 38 shows the viscosity as a function of pH and volume fraction of solids whereas Fig. 39 presents the influence of surface conditions to the viscosity for a 35 v01.-YOB4C suspension. The zeta potential of B4C is generally negative above pH 1, indicating the isoelectric point is at pH << 1 (Fig. 40). The results by Williams and Hawn [208]correspond to similar findings by Pyzik et al. [209] who have prepared slips at pH values between 9 and 10. This behavior is attributed to the strong shielding of surface charges by dissolved boric acid or, at higher pH, by the formation of polyborates. Injection molding of boron carbide with 2-5 mass.-% carbon black was developed by Schwetz et al. [210]. Like in conventional processes known for oxide and nitride ceramics, the spray-dried powder blend was mixed with 18 mass-% organic binder and molded at 120°C and 45 MPa. Dewaxing was accomplished by heating in an atmosphere-controlled furnace at 100 mbar. The binder components decomposed thermally by cracking and evaporated within four days and temperatures up to


-sr

847

105

a

a

E

4

.-I*n L

lo4 lo3

_C_

As Recvd

_f_

6 Months Old freshly Dried freshly Washed

__o_

0

0 In

.-

7

lo2 10'

4

5

7

6

8

9

1

0

1

1

PH Figure 39. Influence of surface conditions to the viscosity for a 35vol.-% B4C suspension (after [2081).

450°C. Vacuum sintering between 2125 and 2225°C for 2 h yielded >96% of theoretical density which was improved to > 99.5% by subsequent hot isostatic pressing at 2050°C at 200 MPa. 7.4.2.2 Hot Pressing and Hot Isostatic Pressing Since pressureless sintering allows the fabrication of complex shapes but results in coarse microstructures and some residual porosity, this process is only applicable --t- As Recvd

> .-

+ Washed

0.2M B(OH),

U

E m

-20

-

c

C aJ

0

a

-40-

m

c

aJ

N

-60 -I

L

- s o ! .

I

0

2

'

I

4

'

1

.

6

'

8

'

1

10

'

1

12

.

1

14

PH Figure 40. Zeta potential of B4C (0.01 mass-% suspension) as a function of pH [208]

848


70

99-100 97-98

60 h

2

Q,

95-96 90-94 o 80-89 A 70-79 v 60-69 A

CI

L 5040-

L

3

O \

A

o

"i*

30-

2! 0-

% th. 0.

2010 0 '

I

1600

'

'

'

I

~

'

'

1800

Temperature

I

2000 (O

'

"

I

'

2200

C)

Figure 41. Hot-pressing parameters and densities obtained from undoped boron carbide (literature survey).

for wear parts or shieldings which are not subjected to high stresses because these materials exhibit a low strength ( C T ~< 300MPa) and a low fracture toughness (Klc< 3MPa m1I2). Hence, for high densification at reasonable temperatures a hot-pressing treatment that causes particle rearrangement and plastic flow is required. Grain boundary sliding, strain-induced twinning, creep and, at a later stage, bulk diffusion combined with recrystallization were identified as the mechanisms of mass transport [211-2131. Densification maps and diffusion diagrams of Brich boron carbide and C have been established by Beauvy and Angers [214], and Bouchacourt et al. [215]. Suitable preconditions are (i) the use of submicron powders, (ii) temperatures in the range 2100-22OO0C, (iii) pressures of 2 5 4 0 MPa, (iv) 15-20 min hold, and (v) a vacuum or an argon atmosphere. To resist the high pressures at these temperatures and to provide carbon as a sintering aid, the use of boron nitride-coated, graphite crucibles is favored. In Fig. 41, literature data on the obtained fractional densities of undoped B4C are related to the particular hot-pressing conditions. It is obvious that both high temperatures and high pressures are required to achieve a density of >95%. Only formation of sinter necks is obtained at 20 MPa pressure and 2000°C (Fig. 42). However, a strong coarsening has to be taken into account at higher temperatures and average particle sizes of > 100 pm in commercial ceramics are not rare. Similar to pressureless sintering, additives may be used for hot pressing of boron carbide to reduce temperatures required for grain boundary and bulk diffusion and to retard grain growth. Figure 43 shows a typical microstructure of a hot-pressed Bdoped material with strain-induced polysynthetic twinning. Suitable dopants are B [216,217], almost doubling the strength, C [192], Mg, Al, Si, Ti, V, Cr, Fe, Ni, and Cu [94,183-185,2161. As demonstrated by Telle and Petzow [218], combined B-Si or B-Si-Ti additions lubricate the grain boundary sliding and prevent coarsening by


849

Figure 42. Formation of sinter necks in undoped boron carbide after hot-pressing at 20 MPa pressure and 2000°C.

forming a thin Sic or TiB2 grain boundary phase (Fig. 44), which pins the grain boundary movement and controls the surface diffusion. Compounds used as additives are various glasses, alumina, sodium silicate with Mg(N03)2,and Fe203,which may reduce the hot-pressing temperature down to 1750°C [219]. MgF2, AlF3 [181] and ethyl silicate [220] are other additives which are active at particle surfaces and grain boundaries: Hot pressing with 1-5 mass- % of the above-mentioned additives usually requires a temperature of 1750-1900°C to obtain >95% density. In most cases, grain size refinement and distributed second phases result in improved mechanical properties such as strength and fracture toughness.

Figure 43. Hot-pressed B-doped boron carbide with strain-induced polysynthetic twinning.

850


Figure 44. S i c intergranular phase in hot-pressed B,C-Si-B powder blends.

Hot isostatic pressing (HIP) of boron-containing ceramic powders creates special difficulties due to the choice of the canning material. In general, containers made from metals or usual glasses cannot be used because of reactions with the sample material. In the presence of metals, boron carbide decomposes forming metal borides plus graphite, which embrittles the capsule. In the case of silica glass, boron diffusion from the outer layers of the specimen into the glass strongly changes the viscosity and the glass transformation temperature. Hence the softening of the container and the pressure transfer to the specimen cannot be controlled reliably. Moreover, boron oxide gas may be released from both the capsule and the sample and hence result in blowing of the container. Promising techniques have been developed by Asea Cerama AB, Sweden, and Elektroschmelzwerk Kempten, Germany, using diffusion barriers and a special type of boron oxide glasses [221]. These methods are also applicable to silicon nitride and silicon carbide ceramics and make the fabrication of complex parts, e.g., injection-molded sand blasting nozzles, in large-scale production feasible. In the case of boron carbide, this treatment was applied to additive-free submicron powder obtained by sedimentation of commercial powder in an aqueous suspension by changing the pH value from 10 for dispersion to 3 for flocculation. The sedimented powder exhibited a particle diameter << 3 pm and yielded a final density of 100% after HIP above 1700°C for 60 min at a hydrostatic pressure of 200 MPa. In Fig. 45, a fracture surface is shown which illustrates that no grain growth has occurred during the heat treatment. A three-point bending strength of 714 MPa was reported with a Weibull modulus m of 8.3 [221]. The increase in strength compared to normal hot-pressed material is almost three-fold. The fracture tou hness was, however, not influenced at all since values ofKrc = 2.5-3.2MPam” have been measured in all cases by the indentation method. Generally, all presintered or hot-pressed materials with closed porosity (i.e., > 95% density) can be fully densified by a post-sintering HIP-treatment [210,222]. Best results for C-Sic-doped B4C are obtained at 2000°C and 200MPa isostatic pressure. Post-HIP of injection molded B4C-C powder mixtures yielded > 99.6% density. The four-point bending strength of 470435MPa for 2% C samples sintered between 2150 and 2175°C was improved to 560-580 MPa. At higher sintering or post-HIP temperatures the strength decreases due to the exaggerated grain

B


85 1

Figure 45. Fracture surface of pure B4C hot isostatically pressed at 1700°C

growth to 350MPa and less. The fracture toughness as studied by Schwetz et al. [2 101 by means of the bridge method (indentation and subsequent quasi-threepoint loading) was found to be grain size dependent and ranges between 2.5 and 3 MPa m1/2 at particles sizes < 2 p m and between 3.7 and 4 MP a m1l2 at 10m mean grain size, which can be attributed to grain size dependent residual stresses in the microstructure causing crack deflection.

7.4.3 Properties of Boron Carbide The physical and chemical properties of boron carbide have been reviewed by Lipp [159], Thevenot and Bouchacourt [256], Thevenot [164,165], and Schwetz (1999) [223]. Special problems while presenting the physical properties arise from the large homogeneity range of boron carbide. Furthermore, its poor sinterability requests additives that are usually unspecified and results in residual porosity and various grain sizes which are often also not considered in the publications. Most variation and discrepancies in the properties reported come from the undefined composition of the materials studied. The microstructure and dopant-dependent mechanical properties have been addressed in the previous section. The strong variability of Young's modulus, strength, hardness and toughness become more comprehensive if compared with the sintering temperature and, e.g., free carbon content and grain size (Figs. 4 6 48). The intrinsic mechanical properties given in Table 5 are based on the most reliable measurements essentially by Schwetz et al. [210]. An excellent overview of hardness measurements as a function of sample preparation and testing load was presented by Bouchacourt and Thevenot [71] stating that the hardness values rise exponentially with decreasing load. A more or less constant hardness is measured at loads > 20 N, which results in a large indentation diameter integrating across the average microstructure. Depending on porosity and free carbon content, however, parasitic cracks may be generated by this and higher loads which make the evaluation of the indents again impossible. At very low loads, e.g., at 0.25 N hardness values in the range of 58 GPa have been measured with, however, a high standard deviation. Figure 49 compares the data by Bouchacourt and

852


2

500

rn

400:

I

350 1-

3

9 300

1

0

1

2

3

4

5

6

7

Free Carbon in HIPS Boron Carbide, wt% Figure 46. Young's modulus vs. free carbon in post-HIP B4C sintered at 2175°C (after [210]).

Thevenot [7 11 with those of Si-doped both single crystalline and polycrystalline B4C [83]. At room temperature the hardness of B4C is only inferior to diamond and cubic boron nitride, which tend to weaken above 500-600°C due to the beginning of the transformation from the diamond structure into the graphite structure. Above

2.100

2.150

2.200

2.250

Sintering Temperature, "C Figure 47. Four-point bending strength vs. sintering temperature and free carbon content (after [2101).


"1

853

o

5 Wi-%C

0

0

2

4

6

8

10

12

Mean Grain Size , prn

Figure 48. Fracture toughness (Bridge method) vs. mean grain size and free carbon content (after [2 101).

11OOOC, and in a non-oxidizing atmosphere, B4C is the hardest compound known up to now. B4C is thus used for wear-resistant parts and inserts for mortars and ball mills, wear plates, sand blasting nozzles, dressing tools for grinding wheels, lightweight armor plates for helicopters, tanks, and in composites of glass fiberreinforced plastics as bullet-proof protection for personnel. Depending on the B : C ratio boron carbide possesses remarkable conduction properties. It is a high-temperature p-type semiconductor with a forbidden band width of 0.8 eV. The electric resistivity is with 0.1-100 Rcm in the range of Sic. The temperature coefficient of electric resistivity is negative but it shows a high increase of thermoelectric power (e.g., [224]). The extraordinarily high thermoelectric power of boron carbide was subject of research for many years for direct power conversion plants. The application of this unique property is today restricted to high-temperature thermocouples in connection with a graphite electrode. Temperatures up to 2300°C can reliably measured by this device. Thermal, optical, and electrical properties have been intensively studied in order to understand the nature of atomic bonding and the homogeneity range by, e.g., Bouchacourt and Thevenot [225], Werheit et al. [226], Wood et al. [227], Emin [79], and Aselage et al. [39,228] in hundreds of papers since 1980 so that no general view can be given in this book. The data listed in Table 5 have been taken from these few publications listed above. The large cross section for thermal neutrons makes boron carbide an interesting candidate for absorption or retardation of neutron radiation in power plants and as first-wall coating in fusion reactors. The cross section for 'OB is approximately 4000 barn, which is naturally present in boron carbide at 19.9%.

854


Table 5. Physical properties of boron carbide. Property

Unit

High carbon

High boron

Formula Density Composition B, C Crystal structure

-

B4.3C 2.52 81.1, 18.9 R3m

B10.4C 2.465 91.3, 8.7 R3m

g/cm3 at.% Space group nm nm3 "C "C 10-'/K

Thermal conductivity X

W1m.K

Thermal diffusivity CL

cm2 s-'

Electric resistivity p Thermal coeff. of electric conductivity Seebeck coefficient of thermoelectric power

Rcm

0.5607, 1.2095 0.5651, 1.2196 0.32894 0.33938 2380 (eutectic with C) 2250 (liquid + C-rich) 2450 2.6 + 4.5 . T (25-8OO0C) 4.5 (600°C) 4.6 (25HOO"C) T - 9.18. lo-' T~ 3.016 + 4.3. 35 (25°C) 4.0 (20°C) BqC 28 (200°C) 4.7 (200°C) 23 (400°C) 5-7 (400°C) 19 (600°C) 5-6.9 (600°C) 16 (SOO'C) 5-6.7 (80O0C) 1 .O x lo-' (400 K) 3.0 x lo-' (600 K) 1 x lo-' (600K) 2.1 x (800K) 0.8 x (800K) 0.1-100

K-'

3.2

PVIK

100 (400 K)

Lattice constants a, c Cell volume Melting point Congruent melting point Thermal expansion coefficient CL

10-~ 220 (400 K) 233 (750 K) 256 (1000K)

140 (750 K) 196 (1000K) 225 (1300K) Self diffusion coefficient: boron neutron absorption cross section Fracture energy ys Hardness HKo,, Strength (4-pt. bending) Fracture Toughness (bridge method) Young's modulus E

Porosity dependence of E Poisson ratio Sound velocity

Cm-*

5.18 x lo-'' (lO0OT)

barn mJ kp/mm2 MPa MPam'I2

400-750 at 0.025 eV 5.45 f0.44 2950 450-550 2.5-3 for < 2 pm grain size 3 . 5 4 for > 2 pm grain size 440460 (25°C) 430 (250C) 420 (500°C) 415 (750'C) 400 (lO0OoC) Ep = E, [(l - P)/(1 2.999P)], E, 0.15 14.000

GPa

GPa -

ms-l

+

= 460 GPa


- 167

60

855

-

50 40 30 20 I

B12(C,6 ,Si)3- single crystal B12(C,6,SiJ3

10

B,C,

0

i

'

'

4

+ Si B,

+Si " " "

7

lo0

i01

2

i02

Load (N) Figure 49. Hardness vs. test load (after [71,83]).

7.4.4 Chemical Properties and Oxidation of Boron Carbide The chemical properties are characterized by a pronounced stability in acids or alkali liquids. Boron carbide reacts slowly with HF-H2S04 or HF-HN03 mixes, i.e., in strongly oxidizing environment. Due to this stability, impurities of metal or other boron compounds can readily be removed by chemical leaching. In contact with molten alkalis boron carbide reacts to form borates. A disadvantage of the application of boron carbide ceramics is their instability against metals a t high temperatures, in particular with metallic melts. Depending on the affinity of the metal to boron or carbon, the particular metal borides and carbides are formed. If no stable metal carbide exists, free carbon is released by the reaction. Boron carbide is also capable of reducing many oxides to form metal borides and carbon monoxide under reducing conditions. In air, the particular metal borates are generated. In hydrogen it reacts slowly above 1200°C to form borane and methane, which prohibits sintering in hydrogen atmosphere. In nitrogen a decomposition to BN occurs above 1800°C. Boron halides are evolved by the reaction with gaseous chlorine and bromine above 600-C and 8OO"C, respectively. Boron carbide in air is immediately, i.e., within seconds, coated by a B203 layer, in presence of water by a hydroboric acid layer [172,229]. Oxidation starts at 500600°C and accelerates significantly above 800-IOOO'C, depending on the humidity. The weight gain of boron carbide powder being surface-cleaned in glove boxes of 1 ppm O2 and 1 ppm H 2 0 and subsequently Si-sputtered, Ar' ion implanted, or

856


46 C/g Ar

_____----t untreated I

r-.

---*---

Iu 1 i 0

100

200

300

400

2000

1000

Exposure to Air [hJ Figure 50. Weight gain of ion-implanted B4C powders at 20°C.

left untreated at 20°C in air was monitored by Heuberger [230] and reported by Telle [231,232]. Similar experiments were carried out by Matje and Schwetz [172]. Controlled exposure of Ar-sputtered and implanted powders resulted in a strong, immediate weight gain which was considerably higher for the treated powder compared to the untreated (Fig. 50). After approximately 20 h, the untreated and the Si-sputtered material was stable whereas the implanted powder exhibited a continuous weight gain with a final oxygen content of almost 4 mass-% compared to the contamination of 1.6 mass-% in the as-received powder. The studies by Matje and Schwetz [172] proved a similar parabolic weight-gain in atmosphere of 92% humidity, whereas the increase in weight was linearly dependent on time in air of 52% humidity (Fig. 51). The tremendous oxygen pick-up by Si-sputtered and

-f

4

t

B4C-

powder

92 % rel.humidity

3t El

x

B

O'

0

/ '

I00

'

200

'

300

'

400

"

I000

'

2000

Exposure to Air [h] Figure 51. Weight gain of B4C powders in dry and humid air [172].

'


857

Ar-implanted materials at room temperature is attributed to the enlarged surface area as well as due to the formation of structural defects which aid oxygen diffusion.

7.4.5 Boron Carbide-Based Composites Since boron carbide is brittle and susceptible to oxidation in air a combination with other materials such as Sic, TiB2, ZrB2 was considered beneficial. As discussed earlier, boron carbide tends to grow exaggeratedly at the temperatures required for high densities. According to the theory of sintering isolated inert particles dispersed in the boron carbide matrix would also inhibit coarsening by pinning the grain boundaries.

7.4.5.1 Boron Carbide-Silicon Carbide Ceramics In boron carbide-based composites, silicon carbide can be dispersed as isolated particles, e.g., by simple powder mixing 12031, mechanical alloying or as a grain boundary phase which is formed in situ by liquid phase reactions [233]. Another method of coating B4C with S i c which was mentioned in the previous section is the deposition of a polysilane precursor on powder particles prior to sintering which can be converted to S i c by a pyrolytic heat treatment [204]. In all the examples, the presence of S i c retards the strong coarsening of the matrix at temperatures above 1900°C (Figs. 44 and 52). In general, B4C and S i c matrix ceramics can be toughened by the incorporation of S i c whiskers, but polytype changes are encountered because of the high temperatures required for complete densification, and the decomposition temperature of 2160°C may easily be reached in the B4C-SIC system. Pressureless sintering of B4C-SiCfibre composites is difficult because of back-stresses which cause porosity in the vicinity of the fibers. Moreover, the toughening and strengthening effect is not very large since the thermal expansion coefficients of matrix and inclusions are about the same. Thus, misfit stresses are small, and only load transfer mechanisms due to differences in the Young’s modulus may be operational rather than crack deflection. Thus the combination of B4C and S i c

Figure 52. Microstructure of a B4C-SiC composite derived from polysilane-coated B4C powders.

858


are, for the time being, only useful for the fabrication of corrosion and oxidation resistant parts with, however, comparatively small fracture toughness. The oxidation behavior of single-phase B4C and B4C-Sic composites in dry and in humid air was studied by Telle [2311 in more detail depending on the microstructure. B4C-SIC composites were prepared for this purpose (i) by the conventional powder route yielding isolated Sic particles in B4C matrix and (ii) by the precursor route [204] giving Sic layers between the B4C particles with the same volume fraction. Figure 53 illustrates the weight change for the heating period and the

.-m

5 - 0.5 - 1.0

I

0

-E 4 -

~

I

I

100 200 300 Time at 700 O C Iminl

,

~

400

BJ+SiC lprec.)

U

o

2 -2 .-m

5

-4

-6 0

0

200 400

600

Temperature

800 1000 1200

100

200

Mo

400

300

1

Time at 1000°C Iminl

LOCI

0

100

200

Time a t 1200 OC [rninl

Figure 53. Oxidation behavior of B4C and B,C-SiC composites in humid air.

3

7.4 Boron Carbide Cerumics

&-

E

8

-

;

\

ISI

E

I

859

----A: Bulk untreated B . Bulk Si-implanted C . Powder Si-implanted 0 : Powder Al- implanted E : Liquid phase reacted .........

-

-

U L

L

o .-

2 r -

so&0

.

500 Temperature

1000

1500

["[I

Figure 54. Weight gain of Si-sputtered, ion-implanted, and liquid-phase sintered B,C in air.

subsequent isothermal annealing procedures between 700°C and 1 2 0 0 T in humid air (dew point 0°C). Upon heating to 700"C, B4C and powder-derived B4C-SIC composites react with a slight sinusoidal weight gain and weight loss followed by a strong weight loss at higher temperatures. Precursor-derived B4C-SiC composites exhibit a slight increase in weight starting from the very beginning but a decrease in rate is observed above 1000°C. The isothermal treatment at 700°C shows clearly that B4C and B4C-SiCpowder ceramics behave similarly, undergoing a parabolic weight gain whereas the B4C-SiCpre,,,,or material exhibits less increase in weight. At 1000°C, the initial weight loss of single phase B4C during heating is compensated again by a slight weight gain approaching steady-state conditions. B4C-SiCpowder samples again show almost the same reaction of stabilizing weight change whereas precursor-derived composites undergo a continuous weight gain. At 12OO0C, all materials suffer from a slight weight loss after 50-l00min of exposure. Monitoring the weight change of continuously heated single-phase and Siccontaining boron carbide ceramics in dry air (dew point -20°C) yielded again a step of weight gain at 650°C for all materials except one with continuous S i c grain boundary phase prepared by liquid phase sintering, followed by a strong increase in mass at above 1200-1250°C (Fig. 54). A XPS analysis of the energies of Sizp and B,, bonds in oxidation layers from humid air compared to untreated reference materials revealed that in B4CSicpowder composites the binding energy of Si stays constant between 700°C and 1200°C and is somewhat lower than that of crystalline silica, but a peak splitting is observed at 1200°C indicating that a boro-silica glass is additionally formed. The B binding energies are close to that of glassy and crystalline boron oxide. Precursor-derived B,C-SiC composites show a decreasing binding energy of Si with increase of oxidation temperature, indicating that a boro-silica glass forms at low temperatures becoming rich in Si with increasing temperature and time. Finally at 1200°C, the binding energy is close to that of silica, and, indeed, isolated cristobalite and tridymite crystals can be found on the surface.

860


The B bonds decrease only slightly in strength confirming that a boro-silica glass of changing composition is present. X-ray analysis of oxidized single-phase B4C and B4C-SiCpowder composites reveals also the existence of H3B03and HB02 after cooling if treated at 700°C and IOOO'C, respectively, whereas no derivatives of boric acid are found at room temperature after heating to 1200°C except of crystals formed newly during storage. In contrast, B~C-SiC,,,,,rs,, did not exhibit any indication of the presence of boric acid. It may be concluded that in the initial stage the oxidation of all materials is governed by the build up of a boron oxide layer, which in the case of single-phase B4C and B4C-SiCpowderimmediately undergoes hydrolysis in humid air yielding boric acid, which is removed by evaporation and hence is responsible for the subsequent weight loss according to the reactions. Only in dry systems a stable liquid B203 layer may be formed before volatile suboxides are generated at locally low oxygen partial pressures and temperatures exceeding 1000°C:

dry air B203(g1ass)

+

B2°3(glass) B2°3(glass)

B202(vap) + 1/202 B2°(v2p)

---$

humid air

B203(liq)

-k O2 +

B2°3(vap)

+ 3 H20 B203 + H20

B203

+

4

B2°3

+ H3B03

2 H3B0qliq)

2 HB02(liq) liquid f

H2°

(14%b) (15a,b) b,

In the case of B4C-SiCpOwd,, material, the B4C matrix is exposed to air like pure B4C, thus oxidizes according to Eqs. (13)-( 16) and hence governs the weight change of the composites whereas a Si02,,,,,,, passivation layer is deposited on the isolated S i c particles. Thus, the oxygen attack is not retarded at all (active oxidation) unless a boro-silica glass is formed at 1200°C which tends to release B2O3 as a vapor phase. In contrary, B4C particles which have been coated with precursor-derived Sic form a continuous boro-silica glass layer at low temperatures (400°C), which results in a progressive weight gain (passive oxidation) although B2O3 may also evaporate from the melt and cause a relative enrichment of SO2. At 1200"C, B203and Si02 eventually segregate in all materials and form distinguishable Si02 and borosilica glass areas. The oxidation then is accompanied by increasing loss of CO at 1200"C), generating small fumaroles in the protecting cover. Sur(pC0 = prisingly, thermodynamic calculations predict the occurrence of elemental boron as a stable reaction product. At temperatures above approximately 14OO0C, SiO evaporates at the expense of the boro-silica layer whereas Sic is newly formed due to active carbothermic reactions close to the oxidation front involving Si02 and CO. The rate-controlling mechanism after the formation of boro-silica glass layers is molecular oxygen diffusion through that layer. The activation energy for that diffusion in silica is reported to range between 83 and 125kJ/mol[234] whereas that in B203Si02 glass is 139 kJ/mol. The activation energy of B4C oxidation in dry air was determined by Telle [231] and Litz [235] to be 134 and 108 kJ/mol, respectively, which is considerably higher than that determined for humid air in that work (51 kJ/mol).


861

As expected, B4C-Sic composites exhibit an intermediate activation energy of 77 kJ/mol, whereas precursor-derived B4C-SiC composites with 112 kJ/mol are close to oxygen diffusion in S i c or boro-silica glass, which confirms this mechanism to be the rate-controlling step in dry systems. For comparison, the formation of Si-0H . . . -0-Si bonds (i.e., H2 diffusion in Si02(glass))requires 71-75 kJ/mol for activation [236] and may become important for humid environments. The microstructural appearance of S i c in B,C-SIC composites clearly influences the oxidation kinetics. The formation of a continuous boro-silica layer at rather low temperatures is helpful in preventing active oxidation during which always new material is exposed to oxygen attack by the evaporation of volatiles. Humid atmosphere is generally more detrimental to the oxidation behavior of B4C-containing materials than a dry one. A uniform boro-silica glass layer is only formed if S i c is homogeneously distributed as an intergranular phase, which can be accomplished by (i) liquid phase reaction sintering, (ii) coating of B4C powder with polysilanederived Sic, and (iii) ion beam assisted sputtering of Si or S i c on B4C powders if the powders are processed under inert gas atmosphere. 7.4.4.2 Boron Carbide-Transition Metal Diboride Ceramics Boron carbide-based composites with transition metal diborides - in particular with TiB2 - have been extensively studied for cutting tools and wear parts [71, 98, 100, 103, 218, 237-2401. Since both phases are thermodynamically stable up to 2300°C composites can be prepared either by pressureless sintering with an Fe additive at 2175°C [241], or by hot-pressing and HIP without additives. Nishiyama and Umekawa [240] have obtained full density by pressureless sintering blends of 20-60 v01.% TiB2 at 2100°C for 1 h in a vacuum. A maximum three-point bending strength of 620 MPa was measured on composites with 35 v01.-% TiB2 combined with hardness HRA of 93.8. During cutting of a A125Si alloy, the composites exhibited the same performance as cBN and K15 hard metal tools and were clearly superior to Ti(C,N)- and alumina-based materials. Another method of densification makes use of reaction hot-pressing or selfpropagating combustion sintering of MC-B powder mixtures under pressure: MC

+ 6 B + MB2 + B4C

(17)

Since sintering of MB2-B4Cpowder mixtures yields similar complications to the sintering of the pure compounds due to favored surface diffusion and evaporationrecondensation reactions, the combustion route is more likely because heat is generated inside the sample due to the exothermic conversion, the bulk diffusion is significantly enhanced and a grain size refinement occurs as the carbide phase decomposes. A certain risk is the evaporation of volatiles such as CO, C 0 2 , B203 as deoxidation products, or even of Bgas, which may be formed because of the high heat release. Temperatures exceeding 2300°C have been reported during fabrication of TIC and TiB2 from the elements [242-2461. The reaction velocity can be retarded by the addition of the final conversion product to the starting powder which then behaves as if inert. Thus in the case of TiC/B mixtures, TiB2 is added

862

7 Boride-Based Hard Matevials

Figure 55. SEM micrograph of a reaction hot-pressed Tic-B powder blend. Light: TiB2, dark: B4C.

or B4C, which also takes part in the reaction: 2MC

+ B4C -+ 2MB2 + 3 C

3C+12B+3B4C

(18) (19)

In this case, B4C also undergoes a grain size refinement which is very beneficial for the mechanical properties. In Fig. 55, a micrograph of a reacted TIC-B powder mixture is presented, which still exhibits TiB2-B2C agglomerates of the size of the initial T i c particles. Note that the average particle size of both reaction products is approximately 1 pm. Generally, this reaction can be employed for most of the transition metal boride-boron carbide composites since the borides are usually more stable than the particular carbides [92]. Dense composites of MB2 and B4C, in particular of TiB2 and B4C, regardless of their fabrication technique exhibit improved mechanical properties compared to the particular single-phase materials. The increase in strength of hot-pressed or HIPed materials to ob = 600-800MPa is mostly attributed to a retardation of the grain growth, whereas the improved tou hness is due to crack deflection around TiB2 particles. K,, values of 5-7.3 MPa m' have been reported for B4C-based composites with LaB6, TiB2, ZrB2, NbB2 and W2B5 [75]. At 2150°C, pressureless sintered B4C/TiB2 composites with 1 mass-% Fe additive exhibit a maximum bending strength of ob= 420 MPa at an optimum volume fraction of 20% TiB2. The lower strength compared to the hot-pressed material is mainly attributed to the embrittling FeB intergranular phase. With increasing sintering temperature and amount of additive, the strength even drops to 100-250 MPa due to the exaggerated coarsening of the B4C matrix by one order of magnitude. Another example of successful materials development is B4C-TiB2-W2B5composite ceramics prepared by reaction hot pressing [218,232,237]. The initial powders consisted of B4C, B, and Si and were mechanically alloyed with WC, TIC, and Co. During sintering or hot pressing, the carbides react with B4C and free boron to the desired transition metal borides. This reaction is accompanied by a strong bloating. Above the eutectic a boron- and metal-containing Si melt is generated which promotes liquid phase sintering depending on the volume fraction. Generally, hot pressing is, however, required to overcome the swelling. The reaction was

,g


- 100 I

863 1

,\"

v

x 8o

.-

4-

In

C

-,C

60

H

I

Y

8

a

LO

Q)

.->

20

4-

0 a,

+

L

O 0

200

LOO

800

600

1000

1200

ILOO

1600

1800 2000

Temperature ("C ) Figure 56. X-ray analysis of sintering reactions during hot pressing of B4C-Si-B-TiC-WC-Co blends. Holding time at the particular temperature step 1 min.

monitored by heat treatments at temperatures ranging from 600°C up to 2100°C with pressure of 47 and 65MPa and isothermal sintering times ranging from 1 min to 8 h. As expected from the phase diagrams, WC and TIC react with elemental boron and B4C-forming borides such as WB4, W2B5,and TiB2.As shown by the relative X-ray intensities of the phases in Fig. 56, this reaction takes place in the temperature range from 1000°C to 1200°C. Si coming from the additive and Co resulting from the binder material of the hard metal milling balls also react forming a B-rich Si-Co melt at approximately 1100°C. At the same temperature, unknown phases denoted as U, and U2 appear which are possibly silicides or ternary borides and decompose at 1600°C. Above 1600"C, the dominating phases are B4C, W2B5, and TiB2. Individual Si or Co phases are not detectable any more. Si and Co segregate at or close to the grain boundaries between the transition metal borides and the boron carbide matrix where they dissolve into TiB2 and W2B5 particles. A solid solubility of TiB2 for Si of at least 8 mass-% was documented in model experiments. Free carbon can be observed only in powders which were milled for 3 h if the initial boron content was not increased. The strong weight loss observed during heating is due to the evaporation of Si, Co, and B species as well as due to deoxidation products such as H3B03, B203,CO, and CO2. The following chemical reactions may occur: 2WC+ 1 TIC

+6B

8 WC + 7 B4C

900-1 IOO'C

3

B W2B5+2B4C ~

900-1 100°C

1000-1 100°C

(20)

TiB2 + B4C

4W2B5

+ 2 B4C + I3 C

(22)

864

7 Boride-Bused Hard Materials

Figure 57. Microstructure of hot-pressed material. Dark matrix: boron carbide, gray: TiB2-solid solution, white: W2BS.

2TiC+B4C-

1000-1100°C

> 600°C

4B+C-B4C

2TiB2+3C

(23) (24)

These reactions are strongly exothermic and belong to the type of materials synthesis denoted as “self-propagating high-temperature combustion synthesis”. The standard Gibbs Free Energy of the reaction (21) is with AGO = -200 kJ/mol B4Cinsensitive to the MC compound involved and fairly independent of temperature. Furthermore, W2B5 tends to decompose above 1600°C in the presence of TiB2 forming a solid solution (Ti,W)B2 with increasing solubility of W by increasing temperature: nTiB2

-

+ mW2B2> 1600°C (Ti,W,)B2

with n = 1 - m and mmax= 0.63 at 2230°C

(25) The microstructure of a hot-pressed sample is characterized by a homogeneous distribution of B4C, W2B5, and (Ti,W)B2 (Fig. 57). The average grain size is 0.81.O pm. The final phase composition consists of 72 v01.-YO B4C, 20 v01.-YO (Ti,W)B2, and 8vol.-% W2B5. The porosity is less than 3vol.-%. Boron carbide and titanium diboride particles are of more or less equiaxed morphology whereas it is a characteristic of the W2B5 phase to grow in an elongated shape. Hot pressing at temperatures above 1900°C results in an exaggerated grain growth of boron carbide and the transition metal borides. The average grain size may reach more than 10pm if fired at 2000°C. Boron carbide grows most probably by transient liquid film sintering, consuming smaller particles, whereas the transition metal borides grow by the dissolution of W2B5 in the liquid film and precipitation as (Ti,W)B2 solid solution. The liquid phase consists possibly of a Si-Co-Ti-W-B alloy which is consumed by the formation of B12(B,C,Si)3and (Ti,W,Co,Si)B2 solid solution in the areas close to the grain boundaries.


865

Table 6. Mechanical properties of B4C-TiB2-W2B5composites hot pressed at standard conditions. Property

Value

Unit

Hardness HV,, 4-pt. bending strength Fracture toughness ISB ICL,lON Young’s modulus Shear modulus Poisson’s ratio

32.1 3.2 830 rt 113

GPa M Pa

4.2 f 0.4 5.0 f0.8 425 zt 10 185 5 0.15 zt 0.01

MPa ml’*

*

+

GPa GPa ~

The hardness (HV,,) reaches 32 GPa as a maximum. The maximum bend strength of 830 MPa is obtained by hot pressing at 1820°C. Compared to single-phase boron carbide, the strength is increased more than two times. In comparison to other TiB2reinforced boron carbide ceramics, the increment in strength is still 150%. Higher hot-pressing temperatures result, however, in a decrease to 600 MPa due to coarsening. Fracture toughness was also improved from 2.5 to 4.2-5.0 MPam’/2 by crack deflection. The properties are summarized in Table 6. Figure 58 shows the microstructure of the composite in Fig. 57 failed by dynamic fracture toughness measurement applying the multiple impact test (courtesy of B. Ilschner and R. Zohner). The microstructure of hot isostatically pressed samples does not differ so much from that of hot-pressed material. The average grain size is slightly smaller than

Figure 58. Fracture surface after dynamic fracture toughness test.

866


Table 7. Mechanical properties of B4C-TiB2-W2B,composites HIPed at 17OO0C,200 MPa, 60min.

Silicon content [mass-%]

Porosity

Hardness

Fracture toughness [MPa

Young's modulus [GPa]

Poisson ratio

HVI [GPd]

4-pt. Bending strength [MPa]

[vol.-%,I

0.0 3.0 1.5

3.4 1.7 0.2

17.8 f 2.4 37.3 f 0.2 29.8 f 2.4

102 f 965 894 73 1129 f 85

4.5 f0.2 3.5 f 0 . 7 5.2 rtO.8

384 412 431

0.17 0.17 0.17

*

that of hot-pressed material, i.e., in the range of 0.8pm if treated at 1600°C and 1.2pm if HIPed at 1900°C. Porosities of <0.55% at 1600°C have been achieved. The bending strength is further enhanced to 1129 i 85 MPa. The fracture toughness exhibits a remarkable maximum of 5.2 MPam'I2 at 1700°C HIP temperature. The entire data set is listed in Table 7. A further improvement of both hot-pressing cycles and additives, in particular concerning the transition metal diborides resulted in a microstructural optimization with in-situ grown W2B5 particles. After annealing, the average of the diameter of the W2B5 particles was about 4.5-7 pm (Fig. 59). According to the proceeding growth of W2B5platelets an increase of toughness with increasing sintering temperature and time was observed. As predicted by the theory of particle reinforcement of brittle matrices introduced by Faber and Evans [247] the increasing aspect ratio makes crack deflection a more efficient toughening mechanism (Fig. 60). In Figs. 61 and 62 the data of high-temperature strength and K,, measurements are shown. At 1000°C the strength is only half of the room temperature strength. The toughness at this temperature is still more than 60% of the room temperature toughness and stays constant for higher temperatures. Since crack deflection responsible for the relatively high fracture toughness is mainly caused by internal stresses emerging from the misfit of thermal expansion between matrix phase and

Figure 59. SEM micrograph of annealed B4C-TiB2-W2B5sample. Note the elongated W2B5 particles.

7.4 Boron Carbide Cerumics

867

7

6n

E0

a 52E U U

U

Y

4

3

1.0

1!5

2.’0

2

Aspect Ratio a/c Figure 60. Dependence of the toughness of B4C/diboride composites on the aspect ratio of platelets.

dispersed particles, it is evident that during an increase in temperature the stresses are reduced to zero at 1000°C. At that temperature no stresses contribute to the deviation of the crack path, and the composite possesses the same toughness as the unreinforced boron carbide matrix. Simple stress calculation according to Selsing [248] in the B4C-TiB2-WB2-CrB2 system have shown possible stress variations from -10 to 2500 MPa if the thermal 20/7mm,51rin

air

500

-a Ll

f.

400

300 _ _ _ _ _ _ _ _ _ _ _ - - _\ _ -4_ _ _ _

b” 200

unreinforced matrix

+-\*

\I\

100 200

400 600 800 1000 1200 Temperature I”C1

Figure 61. High-temperature strength of hot isostatically pressed B4C/TiB2/W2B5composite.

868


ISB, Srnin air

5.0

% O { ,

. . 200

,

,

,

,

,

,

,

600 800 1000 Temperature P C I

400

,

1200

Figure 62. High-temperature fracture toughness of hot isostatically pressed B4C/TiB2/W2B5composite measured by the ISB method.

expansion of the diborides is changed by the formation of solid solutions [232,249,250]. Since measurements show that the thermal expansion of (Ti, Cr)B2 solid solutions do not obey the linear rule of mixture and that the anisotropy of a may even change in sign [251], the radial matrix stresses around these kinds of particles may be tensile or compressive in nature. Calculations of the stress levels indicate, however, that microcracking is not likely to occur due to the comparatively small grain size of the hot-pressed and HIPed materials [250]. As introduced above, the reaction of boron carbide with metal carbides can be used to fabricate metal borides or metal boride/boron carbide composites in a controlled way during densification if boron carbide or free boron is used in excess, or if carbon is bonded by another additive. Although the incompatibility of B4C and metal carbides is well known, many attempts have been undertaken to produce composites or coatings thereof but failed as soon as equilibrium conditions were approached. Physical or chemical vapor deposition of B4C on hard metal substrates, or WC coatings on boron carbides are typical problems (e.g., [252]). In both cases, interlayers of graphite form and hence result in an unsatisfactory adhesion of the deposited coating to the substrate. Recently, Sigl [207] made use of this conversion and reacted B4C with TIC to fabricate a B4C-TiB2composite while using the emerging free carbon for the reduction of oxide layers and thus for activated pressureless sintering of boron carbide powders. The most striking advantage associated with the processing of MC sintering aids is their chemical similarity to boron carbide. Thus, unlike resins or amorphous carbon, metal carbides exhibit significantly fewer problems with longterm stability, dispersability of the sintering agent, or flow behavior of spraydried granules into die cavities. Furthermore, the in-situ synthesized TiB2 is anticipated to support sintering similar to the beneficial effects such as grain growth inhibition that have been observed in SiC-TiB2 or B4C-TiB2 composites. Submicron boron carbide powder was doped with various amounts of fine TIC to yield samples with nominal contents of 1.5, 3, 4.5, and 6 mass-% free carbon after the reaction. The B4C-Tic reaction took place in a 60min holding step at 1250°C. The specimens were finally sintered in a 10 mbar Ar atmosphere for 2 h


869

Figure 63. Microstructure of B4C-TiC sintered with 3 mass-% C. White: TiB2, black: free C (after ~071).

at various temperatures (2125,2150,2175, and 2200°C). Samples with closed porosity were further densified in a post-HIP cycle at 2050°C (30 min., 200 MPa Ar). Figure 63 shows representative microstructures of the composite sintered at 2150°C with 3 mass-% C. It is notable that the 1.5 mass-% C composites consist of a small fraction of pores, two major solid phases, i.e., TiB2 grains in a B4C matrix, and traces of the dark graphite. Significantly more graphite becomes visible in the high carbon materials. Figure 63 shows, that TiB2 particles and most graphite are in direct contact, indicating the simultaneous formation of both compounds consistent with the reaction considered. The mean grain size of B4C in all compositions is plotted in Fig. 64. As expected, there is a pronounced tendency for grain

/ -

B,C sintered with

4

7

0 2050

1

1 2100

2150

1 2200

1 2250

Sintering Temperature, "C Figure 64. Dependence of the grain size of B4C on free carbon content and sintering temperature (after [207]).

870


Table 8. Mean strength and toughness of sintered and post-HIPed B,C-TiB2 composites as a function of sintering temDerature; n.d. = not determined. Residual C content [mass-%]

Flexural strength [MPa] Sintering temperature ["C]

1.5 3.0 4.5 6.0

Fracture toughness [MPaJmI

21 50°C

2175°C

2200°C

21 50°C

2 175°C

2200°C

292 368 454 502

286 n.d. n.d. n.d.

266 n.d. n.d. n.d.

4.2 3.6 3.2 2.9

4.0 n.d. n.d. n.d.

3.7 n.d. n.d. n.d.

growth in B4C with increasing sintering temperature. However, the coarsening is suppressed by an increasing amount of sintering aid as documented in Fig. 64. This plot, together with the observation that the number of second phase particles at grain boundaries increases with initial TIC content, provides evidence that both TiB2 and graphite precipitates are effective in pinning moving grain boundaries. The TiB2 particles coarsen with increasing sintering temperature as well. This coarsening happens most likely by coalescence [253]. All compositions, except the 1.5 mass-% carbon material, exceed a relative density of 95% at temperatures above 2150°C. In high carbon materials, density increases further as the sintering temperature is raised to 22OO0C, whereas the density of materials with 1.5 and 3 mass-% carbon declines at 2200°C. Composites exceeding 95% of the theoretical density were post-HIPed to full density, virtually independent of the initial sintering temperature. The range of strength and toughness in B4C-Tic-derived composites is generally comparable to B4C materials sintered on the resin route. Data for fracture toughness, K,, and flexural strength, of, are disclosed in Table 8 as a function of the nominal carbon content. An inverse trend becomes, however, evident from Fig. 65

1

a I

B,C-TIC

L

0

L 0

200

400

600

800

Flexural Strength, MPa Figure 65. Fracture toughness and strength of B,C-TiC materials (after [207]).


4 t

1

871

I*-/

/

F

i

1

t

Figure 66. Fracture toughness-grain size relation (after [207]).

where the flexural strength is plotted versus toughness: This plot reveals a general trend for the strength of boron carbide materials to drop with increasing toughness. Figure 66 shows a plot of the asymptotic fracture toughness of B4C materials versus grain size. The fracture resistance data of B4C-TiC materials closely follow a scatter band, which is also characteristic for B4C doped with amorphous carbon. KI, starts at a toughness of 2 MPaJm, reflecting the inherent fracture resistance of B4C, and increases up to a shallow maximum of = 4 MPaJm at a grain size of about 10-15 pm, until KI, gradually drops again. From a microstructural point of view, this behavior is accompanied by a continuous switch from inter- to transgranular fracture. This trend is outlined in Fig. 67, which summarizes the dependence of flexural strength on grain size for B4C materials doped both with TIC and amorphous carbon. As in many ceramic materials, the strength data could be fitted to a grain size dependence, where strength scales with the inverse square root of grain size, d-'I2. Since toughening mechanisms by changing the crack path from transgranular to intergranular fracture depend on both internal stresses and interfacial strength it is worth while studying the grain boundary structure in B4C-TiB2 materials in more detail as well as to explore the possibility for microcrack formation. In contrast to Telle and Petzow [218] who have used transient liquid film hot pressing providing a relatively strong B4C-MB2 interface, Sigl and Schwetz [254] and Sigl and Kleebe [255] have exploited the opportunity for crack deflection and microcracking by weakening the B4C-TiB2 phase boundary by the incorporation of free carbon. It was suggested that microcracking accounts most for the toughness increment. Sintering and post-HIP of B4C blends 20 and 40vol.-% TiB2 and 0, 1.2, and 4.8 mass-% free carbon derived from phenolic resins yielded full density samples which have been examined by SEM and TEM. A SEM micrograph of a 20vol.Yn TiB2 composite is shown in Fig. 68. Despite the fact that 1.2 mass-% free

872

>


800

7 1 1 7

(d

n

z

0

600

Strength/Grain-Size Dependence of B,C sintered with amorphous C

L

Q, L

B4C-Tic

400

5 3 X

200

a,

LL

0

I

0

J

-

I

10

5

15

20

Grain Size, pm Figure 67. Strength-grain size relation (after [207]).

carbon should be present according to the chemical analysis, only two phases, i.e., TiB2 (white) and boron carbide (gray), are visible. Particulate carbon precipitates are obviously absent in B4C-TiB2composites with < 2 mass-% free carbon. Discrete graphite particles appear in 4.8 mass-%-free carbon composites

Figure 68. SEM micrograph of post-HIPed 1.2% free carbon containing B4C-TiB2 composite. Arrows indicate circumferential microcracks after cooling from sintering temperature (after [255]).

7.4 Boron Carbide Crrumics

873

Figure 69. TEM micrograph of microcracked B4C-TiB2grain boundary (after [255]).

(Fig. 69). Additionally, the analysis of hetero-phase boundaries by TEM yields evidence of thin carbon interlayers (Fig. 70) with a thickness between z 5 and 10 nm. Composites with free carbon contain microcracks which have formed spontaneously upon cooling from sintering temperature. Such cracks are predominantly detected at B4C/TiB2phase boundaries (Fig. 68), and in rare cases also at TiB2/TiB2 grain boundaries. Notably, microcracks are not present at all phase boundaries but restricted to a few interfaces feeling residual tension above a critical threshold stress. Remarkably, microcracking coincides with the position of carbon interlayers which is in accordance with the fact that microcracks were not observed in composites without free carbon. Another origin of microcracking is occasionally observed inside graphite particles, which are usually located along B4C-TiB2

Figure 70. TEM micrograph of C interlayer between B4C and TiBz particles (after [255]). The inserted diagram presents EELS-analyses of the corresponding points 1-3 of the micrograph.


8 1 BT 80120 7

P plain Boron Carbide

0

0

2

4

6

8

Relative Carbon Content, Cfre$B,C, wt% Figure 71. Fracture toughness-free carbon relation (after [255]).

phase boundaries. Figure 7 1 compares the fracture toughness of B4C-TiB2composites with plain boron carbide as a function of free carbon content. Two effects obviously control the toughness of these composites: (i) Particle effect: The toughness of plain boron carbide varies around 2.2-2.5 MPam'l2. Upon the addition of 20 and 40vol.-% TiB2, the fracture resistance increases to 3.0 and 3.6 MPam'/*, respectively, with the trend to increase with rising volume fraction of particles. (ii) Free carbon effect: A further increase in toughness is observed upon the addition of free carbon such that the fracture resistance exceeds 6.0 MPam'/2 at elevated carbon contents. At still higher carbon quantities, toughness tends to remain constant. In conclusion, strong B4C-B4Cinterfaces with comparatively low tensile stresses coexist together with weak but highly stressed B4C-TiB2phase boundaries with carbon interlayer or graphite precipitates which allow not only crack deflection but also microcracking. Cracking and the relatively large graphite particles obviously contribute to a decrease in strength since no data are provided by Sigl and Kleebe [255].

7.5 Transition Metal Boride Ceramics Transition metal boride ceramics are mainly based on TiB2 or ZrB2 due to their high hardness and relatively high toughness. Because of their poor sinterability, additives have been employed consisting of other transition metal diborides, carbides or irongroup metals and their borides. Since the introduction of a second phase as usually beneficial effects to the mechanical properties, a large variety of composites has been created to reinforce TiB2-based ceramics but, vice versa, TiB2 was also used to strengthen other matrix phase materials. The fabrication of cemented borides is presented separately in Section 1.8.

7.5 Trunsition Metal Boride Ceramics

875

7.5.1 Preparation of Transition Metal Borides Large-scale production of metal borides occurs preferentially in electric furnaces by the following high-temperature reactions: (i) Carbothermic reduction of the metal oxide, graphite or carbon black MOz

+ B203 + 5 C

-+

MB2

+ 5CO

(26)

The carbothermic method yields carbon contaminated powders and is suitable for materials in which a C content of up to 3 mass-% can be tolerated. For instance, TiB2, ZrB2, and the technically important hexaboride CaB6 are synthesized by this method. (ii) Reduction of metal oxides with carbon and/or boron carbide, known as the boron carbide process 2MO2

+ B4C + 3C + 2MB2 + 4 C O

M203+3B4C+MBs+3CO where M = rare earth elements. The boron carbide process can also start from blends of metal carbides, metal hydrides, boron oxide, boron carbide and carbon black:

3 MO2

+ B4C + B203 + 8 C + 3MB2 + 9 C O T + MO;! + B4C + 2MB2 + 2 C O T

MC

(29) (30)

This material usually contains only small amounts of residual carbon or boron carbide but no metals, and is thus the favored process for the technical synthesis of less contaminated borides. The process is carried out in tunnel furnaces under hydrogen or in a vacuum at 1600-2O0O0C, i.e., below the melting point of the boride. It is thus a reaction sintering procedure yielding a high-porosity product which can easily be crushed and milled. Additional refinement is obtained by multiple vacuum treatments with metallic or B4C additives to compensate nonstoichiometries. The final product is then called “vacuum quality”. (iii) Aluminothermic,silicothermic, magnesiothermic reduction of mixtures of metal oxides and boric acid MOz

+ B 2 0 3+ AI(Si, Mg) + MB2 + A1203(Si02,MgO)

(31)

The yield is usually contaminated by residual metals or oxides and thus has to be purified by subsequent leaching, or a high-temperature vacuum treatment. High-quality borides of the transition metals with-defined stoichiometry and crystal structure are synthesized by the following laboratory-scale methods: (i) From the elements or metal hydrides by fusion in an arc or resistance furnace, or by diffusion during sintering or hot pressing M + 2 B + MB2 MHz

+ 2 B + MB2 + H2

(32) (33)

876

7 Boride- Based Hard Materials

(ii) Borothermic reduction of metal oxides MO2

+ 4 B + MB2 + B202 t

(34)

(iii) Conversion of metal carbides with boron and/or boron carbide yielding powder mixtures or, carried out during powder metallurgical densification, i.e., sintering or hot pressing, composites

+ 2 B + MB2 + C MC + 6 B MB2 + B4C 2 M C + B4C 2MB2 + 3 C MC

4

(35) (36)

(37) (iv) Electrolysis of fused salts containing metal oxides, boron oxide or hydroboric acid plus alkaline borates and fluorides. (v) Molten metal/boron dissolved in Al, Cu, Sn, or Pb melts (auxiliary-metal bath method). This procedure is based upon the growth of large particles at the expenses of small particles of high surface area which are dissolved in the melt. Furthermore, solid solutions may be precipitated starting from a mixtures of the pure materials. The grain size can readily be controlled by the ultimate temperature of soaking and the cooling rate. Upon cooling down the solubility of the feedstock material gets smaller, which is hence preferentially precipitated. The auxiliary bath method is well-suited for the fabrication of single crystals of borides. (vi) Chemical vapor reaction of metal halides and boron halides in a hydrogen atmosphere under plasma conditions. This method, however, yields material of varying stoichiometry and crystallinity. 4

Limiting factors for commercial fabrication are the relatively high costs of elemental boron and the low production rate in the reactors.

7.5.2 Densification of Transition Metal Borides The densification of single-phase and pure ceramics of transition metal diborides is complicated by two characteristics of these compounds, the high melting point and the comparatively high vapor pressure of the constituents. As a rule, sintering temperatures exceeding 70% of the absolute melting temperature have to be applied.

7.5.2.1 Pressureless Sintering Titanium diboride, TiB2 (T, = 325OoC),requires firing temperatures of the order of 1800-2300°C to initiate grain boundary and volume diffusion, and thus to obtain more than 95% of the theoretical density. One disadvantage is that the borides undergo a similar abnormal grain growth at high temperatures to B4C or Sic. Furthermore, at lower temperatures evaporation of B and boron suboxides enhance the grain growth without shrinkage by evaporation and recondensation mechanisms. Since TiB2 is strongly anisotropic due to its layered structure well-faceted particles are formed and, together with pore trapping inside large grains, the final

7.5 Transition Metal Boride Ceramics

877

densities of pure TiB2 do not exceed 90% even if heated between 1900 and 2500°C (e.g., [257,258]). Thus it is nearly impossible to achieve completely dense bodies by pressureless sintering, as no shrinkage between 1900°C and 2100°C occurs, and further densification by volume diffusion and plastic flow, is accompanied by exaggerated grain growth. This behavior was attributed by Coble and Hobbs [259] and Kislyi and Zaverukha [260] to the competing mechanisms of evaporation-recondensation and volume diffusion exhibiting the same rate of mass transport, whereas gas-transport reactions are favored due to the lower activation energy. The powder size-dependent sintering behavior of TiB2 was studied by Kislyi et al. [26 I]. Starting with high-purity submicron-size powder, synthesized from TiC14 and BC13 in hydrogen in a plasma-arc heater, Baumgartner and Steiger [262] achieved densities of 98.4-99.4% at 2000-2100°C combined with a comparatively fine microstructure (average grain size 1-18pm) due to TIC and T i 0 inclusions. Further heating or a prolonged holding time generate exaggerated grain growth to 80pm, whereas the density does not improve. This is attributed to entrapped porosity, which can only be removed by volume diffusion. In contradiction to reports of the other authors, no significant weight loss was observed even after several hours hold at 2250"C, which could be related to active evaporation reactions. Since the small grain size of 1 pm could be retained up to 2O0O0C, the porosity was preferentially removed by grain boundary diffusion. Thus, contrary to carbothermically produced TiB2 powder, high densities have been obtained below the critical temperature promoting rapid grain growth. This implies that both the initial particle size and the presence of impurities significantly influence the densification kinetics. Baik and Becher [263] have studied the effect of oxygen contamination of submicron TiB2 powders and concluded that in the case of hot-pressing between 1400 and 1700"C, oxygen promotes grain coarsening by enhanced evaporation-recondensation of B203. Upon pressureless sintering between 1700 and 2050°C oxygen remains primarily as titanium oxides and suboxides, which increase the surface diffusivity and thus the pore and particle coarsening rather than the densification. A maximum total amount of oxygen of less than 0.5 mass-% in the powder or reducing additives such as carbon is recommended. Sintering is usually carried out under vacuum in a resistance furnace with a graphite, tantalum or tungsten resistor or in a high-frequency furnace with a graphite susceptor. Ar or H2 gas atmospheres can also be used. If carbon crucibles are used, boron nitride diffusion barriers have to be inserted to prevent eutectic melting of the borides and carbon in the temperature range of 20O&250O0C. The considerable losses of volatile boron or boride species may be reduced by powder-bed sintering [ 1341.

7.5.2.2 Hot Pressing and Activated Sintering Densities above 95% have been achieved by axial hot-pressing at pressures exceeding 20 MPa and temperatures above 1800°C. The microstructures consist typically of particles of > 20 pm in size. Another problem is related to the hexagonal layered structure of the A1B2-type borides. Because of the strong anisotropic behavior of the physical properties, especially of the coefficients of thermal expansion, the

878

7 Boride-Based Hurd Materials

coarsening can be very detrimental to the mechanical properties, by producing spontaneous microcracking and residual strains. Pressureless sintering and hot-pressing of transition metal borides can be generally activated either physically by starting from submicron powders or by extensive milling, i.e., by increasing the specific surface area and introducing defects, or chemically by doping with small additions (0.3-3 mass-%) of transition metals such as Fe, Ni, Co, Cr, Pt, or their halides. The mechanisms of sinter activation by doping are not yet readily understood but an increase of driving forces and volume diffusion and a retardation of evaporation seems to be likely. Crystallographic studies on Co-, Nb-, Cr-, and Re-doped ZrB2 by Czech et al. [264] indicate that the metals substitute for Zr in the metal sublattice, which is also confirmed for Mo and W [265]. Other borides such as VB2, NbB2, TaB2 or W2B5 and Mo2B5 could not be satisfactorily densified by pressureless sintering. For an extensive survey on powder molding, compaction and sintering of various transition metal borides, containing detailed descriptions of additives, the reader is referred to [ 1341. Reaction sintering starting from chemically incompatible compounds may also lead to high densities, especially if combined with hot-pressing, since that synthesis is strongly exothermic and provides high internal temperatures [242]. The so-called self-propagating high-temperature combustion synthesis was used for TiB2, ZrB2, NbB2, and TaB2 starting from the elements, B4C mixtures with metallic Ti, Zr, Cr, or Nb [266], and Ti-B-TiB2 blends [244,267]. In the case of other reactants, e.g., blends of metal oxides and boron carbide or Al-Ti02-B203mixtures [243], the formation of gaseous by-products often prevents complete densification but may result in bodies of well-defined porosity. The kinetics of the combustion synthesis of TiB2 from the elements have been studied by Holt et al. [245]. Ouabdesselam and Munir [268] investigated the sinterability of directly synthesized TiB2 powder but could not find any significant difference to carbothermically produced powders.

7.5.3 Properties of Transition Metal Borides Ceramics Transition metal borides are mainly explored for their mechanical properties. Since they exhibit metallic transport properties such as high electric and thermal conductivity with a negative temperature coefficient they are also of interest as electrode materials, for heating elements and sensors. 7.5.3.1 Single Phase Ceramics

Single phase TiB2 ceramics with high density have been prepared almost exclusively by hot-pressing. Applying temperatures between 1800 and 2300°C densities of >95% may be achieved [134]. Small amounts of additives may, however, significantly improve the sintering behavior. While between 1950 and 1965 the fabrication of single phase TiB2 was aimed at applications in the nuclear industry, multiphase ceramics produced since then have basically been aimed at wear applications. But also the electric properties have been studied extensively and led to highly developed electrode materials ready for use in, e.g., aluminum production.


879

Table 9. Physical properties of TiBz and ZrB2 Property

Unit

TiBz

ZrB,

Density Crystal structure Lattice constants a, c Standard enthalpy of formation Melting point Thermal expansion coefficient a,, Thermal expansion coefficient a,, Thermal conductivity h Electric resistivity p

g/cm3 Space group nm

4.52 C6/mmrn 0.3028, 0.3228

6.09-6.17 C6jmmm 0.3167, 0.3529

kJ/mol “C

279.49 322 5 520

308.78 324 f 518

10-6/K 10-6/K 25-1600°C W/m.K pf2 cm

5.107+ 1.997 7.443 + 2.261

Thermal coeff. of electric conductivity Debye temperature 6 Hardness HK,,, ceramics, >95% dense

Young’s modulus E Shear modulus G Poisson ratio

K-’ K kp/mm2

GPa GPa -

T

1-3

T

24-59 20.4 (25°C) 26 (200°C) 36 (400°C) 46 (700°C) 56 (1000’C) 68 (1300°C)

23-24 9.2 (25°C) 10 (200°C) 11 (400°C) 13.5 (700°C) 17 ( 1000°C) 20 (1300‘C)

4.76 x 807-820 2600 (25°C) 2400 (200°C) 1800 (400°C) 1050 (600°C) 700 (1OOO°C) 560 490 0.327

6.32 x 585 2100 (25°C) 1850 (200°C) 1000 (400°C) 900 (600°C) 800 (IOOO’C) 490 220 0.3

The physical properties of diborides have been reviewed in many papers, e.g., by Clougherty and Pober [269], Samsonov et al. [270], and Castaign and Costa [271]. The reported value depend, however, on purity, final porosity, grain size, and other factors that are usually not well-documented. Therefore, a comparison of the data is difficult. Especially the mechanical properties differ very much as the anisotropy of thermal expansion eventually results in microcracking during cooling from sintering temperature, or at least to high stress concentrations at grain boundaries. Hardness and fracture toughness, for instance, vary therefore with grain size and testing conditions. Table 9 presents a data set for TiB2 and ZrB,. Only data from polycrystalline materials with densities > 95% and preferentially small grain size, if available, have been used. Thermal expansion of some diborides have been measured by high-temperature X-ray analysis by Lonnberg [272] and by Fendler et al. [250] and Telle [232]. The electric resistivity in the solid solution system TiB2-ZrB2 was studied by Rahman et al. [273]. Billehaug and 0 y e [274] present a study of several transition metal diborides for cathode materials in Hall-H&-ould cells and come to the conclusion that TiB2 should be an excellent candidate because of its stability against the

880


aluminum-kryolithe melt and high thermal and electric conductivity. The sensitivity to thermal shock, the infiltration of liquid phase along grain boundaries and, finally, the high price are, however, the limiting factors. In combination with other, cheaper or properties-adjusting materials, there are excellent chances for this kind of application. Ssrlie and 0ye [275] favor today a dense TiB2 coating on graphite to reduce costs and to gain full advantage of the corrosion resistance and energy saving of TiB2. In connection with BN and AlN the excellent electric conductivity of TiB2 and the perfect wetting by A1 has been used for the manufacturing of resistanceheated evaporator boats for the metal deposition on plastic sheets. Data about the corrosion behavior have been presented by Bannister and Swain [276]. 7.5.3.2 TiB2-TiC Composites As shown in the phase diagram in Fig. 19, the combination of TIC and TiB2 is thermodynamically stable up to 2500°C undergoing a quasi-binary eutectic reaction [I011 In that system, excellent wear-resistant materials have been produced by hotpressing or even by pressureless sintering of eutectic compositions at 1600-1 700°C [103]. A Vickers hardness of HV2 = 23 GPa was measured at room temperature, which is lower than that of the pure materials with values of 27.5GPa for TIC and 28.5GPa for TiB2. At 600”C, however, the hardness of the composite, 8.3 GPa, far exceeds the hardness of monolithic TIC and TiB2, which decrease to 6.8 GPa and 7.8 GPa, respectively. The fracture toughness is notably improved to 7.1 MPa m’I2. The significant decrease in wear during turning or milling of steel compared to the monolithic materials was mainly attributed to “phase boundary toughening” due to the favored occurrence of common coherent (1 1 (0001)TiB2particle interfaces [103]. Besides this very sophisticated toughening effect, mechanisms which influence the crack propagation such as crack deflection or crack impediment due to thermal misfit effects between the boride and carbide phases certainly contribute to the increase in toughness, whereas grain growth retardation due to the pinning of grain boundaries by incorporated particles affects the strength positively. As another example, WC is used for grain size refinement of TiB2, and, vice versa, TiB2 is used as an additive for WC-based materials [257,277]. High-temperature reinforcement by in situ precipitation of TIC and TiB, from supersaturated solid solutions has already been used with interesting results. In the Tic-TiB2 system, either TIC or TiB2 can be the host crystal for the corresponding minority phase or the precipitate [278-2801. The addition of a small fraction of boron to T i c can increase the critical resolved shear stress at 1600°C by a factor of six if TiB2 precipitates are formed at the (1 1 I) slip plane of TIC. 7.5.3.3 TiB2-B4C Composites Kang and Kim [281] have investigated the improvement of TiB2 with a dispersion of B4C particles. Using 1 mass-% Fe as a reactive additive, hot-pressing at 1700°C for 60 min at 35 MPa resulted in 99% dense composites with a clear maximum in strength of 700MPa at IOvol.-% B4C and in KIc of 7.6MPa m1/2at 20~0l.-% B4C. This optimizing effect was attributed to both grain growth inhibition and


88 1

change in fracture mode from transgranular to intergranular by the B4C addition. Since studies on the B4C-rich side of this system also indicate optimum properties at approximately 60-70 v01.-YO B4C a change in strengthening and toughening mechanisms most occur at a composition between 40 and 50vol.-% B4C. The total system was investigated by Nishiyama and Umekawa [240] by pressureless sintering of ultrafine B4C and TiB2 powders. Besides other properties such as oxidation and wear resistance, they report a maximum in strength of 650 MPa at 35 v01.-YO TiB2 and an optimum hardness of HRA = 94 at 2Ovol.-% TiB2. Pressureless sintering of B4C with additions up to 16.7 mass-% TIC to initiate the reaction to TiB2 and the simultaneous release of C for deoxidation was studied by Sigl [207] and discussed already in Section 7.4.4.2 like all other B4C-TiB2 composites with B4C as the majority compound. 7.5.3.4 Transition Metal Diboride-Sic Composites Reinforcement of TiBz by dispersed S i c particles is generally possible since both materials are chemically compatible. As a result, crack impediment is obtained but the increase in strength and toughness is small [282]. This composite material has, however, not yet been studied extensively. Silicon carbide-based composites with transition metal boride particulates have been developed for electroconductive applications such as heating elements and igniters [283-2851 but also as wear-resistant structural parts for high temperatures such as valve-train components and rocker arm pads in super-hot running engines [286,287]. These composites combine the high thermal and electric conductivity of, e.g., TiB2 and ZrB2 with the oxidation resistance of Sic. Additionally, due to thermal mismatch stresses of the order of 2 GPa, toughening mechanisms such as crack deflection and stress-induced microcracking with a pronounced process zone, as well as crack flank friction have been proven to occur. Cai et al. [288] and Faber et al. [289], have presented a detailed analysis of the contributions of the particular mechanisms to the total fracture toughness, stating that stress-induced microcracking is operational in a process zone of approximately 150 pm width. Typical conditions for densification by axial hot-pressing are 2000-2 1OO'C, at a pressure of 2&60MPa for 30-60min which results in 96-99.8% density. The particle sizes of the matrix and dispersed phases range from 1-5 and 4-8 pm, respectively. An optimum volume fraction of reinforcing particulates of 25-30 v01.-YOhas been reported, yielding a flexural strength of 710MPa and a fracture toughness of 5.0-5.7 MPa rn'/*, as shown in Fig. 72 [282]. Composites with a lower TiB2 content of 15 v01.-YOexhibit a mean strength of 485 MPa combined with a K,, of 4.5 MPa m1I2[287]. The strength of Sic-based materials with 50vol.-% ZrB2, HfB2, NbB2 or TaB2 particles also ranges between 400 and 500MPa [283]. Similar strength values (480 MPa) combined with an exceptionally higher fracture toughness of 7-9 MPa m'l2 have been reported for large-scale lots of pressureless sintered l6vol.-% TiB2 composites [284]. Since the sintering was carried out with temperatures exceeding 2000°C (no details given) yielding 98-99% of the theoretical density and an average TiB2 particle size of 2.0 pm, it is obvious that the reinforcing phase also acts as a grain growth inhibitor for Sic. The high-temperature

882


A

0

TIC 1,5um

-

o T i c 8,Oym

T 70 -

TiB 4,Oum

3 60 aJ & 50 3 to

!-

.I-

0

2 2

=3.LMPa\liii ICO

-

30 -

W

a

K

20

-

10 -

c-(

0

c

0

20 30 LO TiB and Tic-Content in Sic [vol.-%] 10

50

Figure 72. Volume-dependent mechanical properties of SiC-TiB2composites [282].

strength of SiC/TiB2 and SiC/ZrB2 composites was found to remain nearly constant at 480MPa up to 1200”C, and is hence superior to that of many sialons [283,284]. SiC-TiB2 composites have been prepared a in situ synthesis by Ohya et al. [253] by adding TIC and B powder to Sic. If the dopants are well-balanced, these incompatible compounds react between 1000 and 1600°C to form isolated and homogeneously distributed TiB2particles. The reaction is, however, accompanied by a strong expansion, which can be overcome by isothermal soaking at 1500°C and final pressureless sintering at 2200°C for 30 min. The fracture toughness of the >98% dense samples is around 3.7MPa m112 at lOvol.-% TiB2. A systematic study on the rule of mixtures for the mechanical and electrical properties of TiB,-TiC-SiC composites was carried out by de Mestral and Thevenot [290]. They modeled “iso-property” curves in the quasi-ternary phase diagram for mechanical parameters by fitting 20 independent coefficients of a third-order polynomial developed by Phan-Tan-Luu et al. [291] to the results of experimental test points. Calculated iso-bend strength curves as well as tests on hot-pressed materials indicate a maximum of 1 100 MPa close to the TiB2-Tic binary edge of the system (Fig. 73). The best fracture toughness value of 6.4MPa m112was obtained on the binary SIC-TIC edge (measured and calculated), on .the binary TiB,-Sic edge and in the ternary region close to the TiB2 phase (67mol.-% TiB,, and 16.5mol.-% S i c and Tic, respectively) (Fig. 74). The calculated rule of mixtures could also be confirmed in the case of hardness measurements. Sic-based composites with W2B5have been discussed by Telle [232]. The fabrication method was similar by starting with mechanically alloyed WC-B-containing S i c powders. Due to the strong tendency of W2B5 to grow anisotropically in a platelet shape crack deflection and crack bridging was efficiently applied, Several kinds of diboride additives have been studied by Tanaka and Iyi [292] to reinforce Sic. Pressureless sintering of P-SiC composites with 15-17 vo1.-YO NbB,, TaB,, TiB2, and ZrBz at > 2000°C resulted in > 99% density. Surprisingly, the additives strongly influenced the p-cl transformation of Sic during sintering. TaB2 addition


883

Ti 8

Si C

Ti C

Figure 73. Iso-bend strength curves in SiC-TiC-TiB2composites [290].

clearly stabilized the 3C polytype up to 2200°C whereas ZrB2 addition extended the stability of the 15R structure. NbB2 doping results in a very smooth transition from 3C to 6H between 2000 and 2200°C. The 4H polytype was only found in TaBz containing materials at a larger content. Although the grain size and shape of the S i c matrix is heavily influenced the increment in toughness reaches only 20%. The highest vaIue was obtained for ISvol.-% TaB2 with KI, = 4.751MPam”~.

7.5.3.5 TiB2-TransitionMetal Diboride Composites Combinations of diborides of different transition metal borides have been studied, especially in the TiB2/CrB2 and TiB2/W2B5systems, for wear applications and to a minor extent for electrodes in Hall-Herould cells [135, 139, 140,293-2951. Since the transition metal diborides crystallize in the same structure type, namely the AlB2

884

7 Boride-Bused Hard Muterials Ti B ,

Si C

Ti C

Figure 74. Iso-toughness curves in SiC-TiC-TiB2composites [290].

layered structure, the formation of solid solutions has been extensively investigated and used for hardening effects. As an example, the quasi-binary system CrB2-TiB2 exhibits a continuous mutual solid solubility approximately between 2000°C and 2100 f 50°C (Fig. 30) [132, 1361, but there is evidence of a solubility gap below 2000°C where the solubility of TiB2 in CrB, is about 40mol.-% at 1500°C and the solubility of CrB2 in TiB2 is less than 1 mol.-% below approximately 1800°C. The presence of CrB2 aids the densification of TiB2 due to its higher diffusion coefficient. Above 21OO0C, CrB2 containing materials partially melts, which is due to an almost horizontal solidus line between approximately 40 mol.-% CrB, and pure CrB,. This fact enables liquid phase sintering of TiB2 but with the risk of exaggerated grain growth and evaporation of chromium and chromium borides, since the vapor pressure of Cr is four orders of magnitude higher than that of Ti. Pre-reacted and hot-pressed materials of that system exhibit a flexural strength of 350-500MPa [139,140].


885

Figure 75. SEM micrograph of large W2B5 precipitates white and thin TiWB, layers in (Ti,W)B, solid solution matrix.

In the TiB2-W2B5system, the borders of the (Ti,W)B2 homogeneity range have been intensively studied between 1500°C and 1700"C, at 2000°C and around the quasi-binary eutectic temperature. TiB2 and WB2 react eutectically at T, = 2230 f 40°C and 90 f 3 mo1.-% WB2 (Fig. 32). The solid solubility of WB2 in TiB2 at this temperature is approximately 63 mo1.-%, whereas the solid solution of the (W,Ti)2B5-typecontains only 3 mol-% TiB2 at the eutectic equilibrium. The homogeneity range of the (Ti,W)B2 solid solution narrows significantly with decreasing temperature and is 46-49 mol-% at 2000°C and 8-10 mol-% WB2 at 1500°C. A high-temperature treatment of TiB2-W2B5 powder mixtures inside the solid solubility range of (Ti,W)B2 at above, e.g., 2000"C, for 30-720 min results theoretically in a uniform, single phase microstructure. Subsequent annealing at, e.g., 15001700°C, causes the epitaxial precipitation of very fine platelets of the metastable WTiB2 monoboride phase with p-WB structure onto the prism plains of the host crystal (Fig. 75). After 30-240 min annealing, these precipitates measure 0.5-5 pm in length and 0.01-0.2pm in thickness and can be aged by prolonged heating or by the choice of a higher temperature (Fig. 76). The growth of W2B5 platelets can also be initiated by heterogeneous nucleation close to a grain boundaries of the host crystals [81, 152,1531. The precipitate is then able to grow across the grain boundary into a neighboring W-rich (Ti,W)B2 grain and thus create an

Figure 76. Aged in-situ reacted (Ti,W)B2-WB2composite with grown WB, particles.

886


Figure 77. Interlocking grain boundaries bridged by WB2 particles. Note the thin TiWB, stacks in the host crystals.

interlocking microstructure, as shown in Fig. 77. Crack propagation studies confirm that crack deflection is operational around the W2B5-typephases. This process is assisted by differences between the Young's moduli of the particular phases, the differences in the thermal expansion coefficients and their anisotropic behavior generating residual misfit stresses. Of similar importance for crack interactions are the grain boundaries of the W-depleted host crystals and neighboring W-rich solid solutions. Here, an active crack deflection was observed, which indicates that both the elastic constants and thermal misfit stresses of TiB2-type solid solutions vary significantly with composition. High-temperature X-ray diffraction measurements of the lattice constants of (Ti,W)B2 solid solutions confirm this observation. Hot-pressed com osite materials developed from the more complex systems of the type TiB2-MTB2-M with M' being Hf, V, Nb, Ta, Mo, or Mn and MI' being sintering additives such as Co and Ni, exhibit bending strengths between 850 and 1000 MPa which are due to the grain growth inhibiting influence of the 1-5 mass-% of M"B2 particulates (Fig. 78) [293-2961. During liquid phase sintering in a Co- or Ni-boride melt, both TiB2 and MI are partially dissolved and reprecipitated as a solid solution. The effect of grain growth retardation as well as of strength and hardness increments is attributed to stresses at the TiB2/(Ti,M')B2 phase boundaries generated by the mismatch of the lattice parameters between the unreacted TiB2 acting as a nucleus and the epitaxially precipitated (Ti,M')B2 solid solution. In the case of a TiB2-5 massYOW2B /TaB2 material with 1 mass-% COBbinder the lattice strain ranges between 9 x 10- d and 14 x depending on the hot-pressing temperature [293]. The addition of 1.7% TIC to the above-mentioned base composition reduced the porosity from 0.3-0.7 to 0.1-0.2vol.-% after hot-pressing at 1500°C and a pressure of 20MPa for 1 h. The improved sintering behavior was achieved by intensive ball milling resulting in an average particle size of 1 pm, but increased oxygen contamination. Watanabe and Shoubu [297] reported the formation of a (Ti,Ta)(C,O) solid solution which is considered to initiate the improved densification resulting in a flexural strength of l000MPa. In a similar multiphase system, transition metal carbides were used as additives for pressureless sintering of TiB, yielding composites of binary and ternary borides [218,296]. Attrition milled powder mixtures of TiB, with 3-10 mass-% Co or Ni and 20-35 mass-% WC have been sintered in a vacuum at temperatures between

g

7.5 Transition Metal B o d e Ceramics

1200

887

I

600 400 -

"\ "

t

o

-

200 ,

r

l

,

,

,

l

l

l

l

,

I

I

I

~

Figure 78. Strength-grain size relation of various TiB2-MBZ-Mcomposites (data from [293]).

1500 and 1700°C for 60 to 120min yielding 98-99% relative density. Densification starts above 980°C due to the formation of a liquid phase in the Ti-B-Co/Ni system (Fig. 25). At this early stage, a rigid skeleton of TiB2 and WC develops. Due to dissolution and reprecipitation, a (Ti,W)B2 solid solution grows on the residual TiB2 particles. Subsequently, crystals of w-phase (see Section 7.8.3) form with compositions of WCoB or W2NiB2,respectively. Upon cooling, the residual liquid phase crystallizes as C-and Ti-enriched Co3B or Ni3B solid solution. A typical microstructure is shown in Fig. 79. Sintering at 1700°C for 2 h yields an average particle size of

Figure 79. SEM micrograph of a (Ti,W)B2-WzNiB2-Ni3Bcomposite. Light: W2NiB2,intermediate: Ni,B, dark: TiB2.

888


0.8 m resulting in a flexural strength oh= 600-680 MPa. The KIc of 6.5-7.5 MPa mIP is mainly attributed to crack deflection because of the weak Co- or Ni-boride intergranular phases.

7.6 Multiphase Hard Materials Based on CarbideNitride-Boride-Silicide Composites Ternary composites of Ti(C,N)-TiB2-MoSi2 were studied by Shobu and Watanabe [298] in order to improve the oxidation resistance of Ti(C,N) TiB2 materials. Full density was obtained after sintering composites with less than 80 mass-% TiB2 and less than 60 mass-% MoSi2 at 1750°C. The oxidation resistance above 1000°C was good for small Ti(C,N) concentrations, i.e., when all the carbonitride particles were surrounded by a phase of either TiB2 or MoSi2. The formation of rutile (Ti02) and silicate glass was observed and considered to prevent further oxygen diffusion. TiB2-20 mass-% MoSi2 composites sintered at 1800°C in a vacuum exhibited a flexural strength of 600MPa and a hardness of HV 2100, whereas the fracture toughness was only 3.7 MPa m'I2. Composites of 70 mass-% Ti(C,N)-30 mass-% TiB2 showed a three-point bending strength of 800MPa and a KI, of 5MPa m1/2.With a hardness of HV > 2500, the material was tested as a cutting tool and exhibited a longer lifetime upon machining plain carbon steel at 300 m/min than conventional hard metals or cermets [298].

7.7 Boride-Zirconia Composites The most important toughening strategy for oxide ceramics, namely the so-called transformation toughening by dispersed tetragonal zirconia particles, is not applicable to borides and carbides to a similar extent. Chemical interactions between Zr02, with its pronounced tendency for oxygen loss, and, in presence of oxygen, the thermodynamically less stable borides lead to the formation of boron oxides or carboxides, respectively, which in some cases result in the total degradation of the composite. This is particularly the case under reducing conditions and at high temperatures, which are both required for a successful densification of hard materials. For example, boron carbide decomposes in the presence of zirconia according to Lange and Holleck [75]: B4C

+ 2 Zr02

2 ZrB2 + CO + B203

(38) A pronounced bloating of the samples is observed due to the release of gaseous compounds. Eq. (38) resembles the so-called boron carbide route for the production of the particular transition metal borides. Another limiting factor is the chemical, geometrical and mechanical destabilization of tetragonal zirconia if combined

7.7 BorideeZirconia Composites

889

Figure 80. SEM micrograph of a (Ti,Zr)B2-(Zr,Ti)02material. Note the core-rim structure of the dark TiBz particles.

with transition metal diborides. Stabilizing additives such as MgO or Y z 0 3 tend to migrate into grain boundaries since these bivalent or trivalent cations in the zirconia lattice are substituted by the more favored, such as Ti4+.A geometrical destabilization results from the strong coalescence of ZrOz causing a particle coarsening due to the high sintering temperatures. Large crystallites exceeding a critical size cannot be retained in the tetragonal modification upon cooling to room temperature and hence transform spontaneously to the monoclinic modification. Moreover, if associated in clusters, a transforming zirconia particle may trigger the transformation of all the other crystals by an autocatalytic reaction. This mechanical destabilization results from the anisotropy of the thermal expansion of the diborides, which introduces radial tensile stresses in the vicinity of the zirconia inclusions. This initiates the spontaneous tetragonal-to-monoclinic transformation or at least reduces the contribution of the ZrOz volume expansion during stress-induced transformation to toughening [299]. TiBz-ZrOz ceramics have been studied intensively as possible candidates for active transformation toughening [239,300-3051. Composites with ZrOz additives show an improved densification behavior and a grain growth inhibiting effect for the TiB2 (Fig. SO). Hot-pressing of composites with 22-60 mass-% ZrOz between 1700°C and 1900°C at 20 MPa yields densities exceeding 99.8% [300,305], whereas 98% of the theoretical density is obtained by pressureless sintering at 2100°C [239]. Volume fractions of unstahilized ZrOz between 15 and 30% result in a significant increase in both the strength and the toughness. Depending upon the microstructure and the density, a maximum oh of 700-800MPa is measured at 22 or 35vol.-%, respectively, and the maximum KI, varies between 5 and 9.5MPa m'/* (Figs. 81 and 82). The hardness decreases linearly with the amount of Zr0,additive and is thus of the order of 16-18 GPa at 20-30 mass-% Z r 0 2 being optimum for both strength and toughness (Fig. 83). This improvement in the mechanical properties is attributed to enhanced sintering and grain size refinement of TiB2, active transfor-

890

-

7 Boride-Bused Hard Muteriuls

900

0

Q

E

800

5

700

u

C

2”v, 2 3

600 500 400

LL

300 200

I

I

I

I

1

0

10

20

30

40

50

Z r 0 2 [voI.-% ] Figure 81. Strength of TiB2-Zr02composites (data from [218,300]).

mation toughening [300],crack deflection and microcracking [239,301]. Miiller [305] quantified the transformable amount of tetragonal Zr02 by X-raying of as-sintered and as-ground samples. Depending on the processing and sintering conditions, the remaining tetragonal fraction ranges between 0 and SO%, which could almost be entirely transformed to monoclinic during machining. Between 20 and 30 mass-% ZrOz, there is, however, about 15 mass-% untransformable tetragonal Zr02. The presence of these untransformable tetragonal ZrOz particles without an yttria addi12.0

11.0 10.0

-

9. 0

E

c a

3 ::

Y

8. 0 7. 0

6.0

5. 0 4. 0

3. 0

I 0

1

1f I+ t

10

20

30

40

a

Vol% ZrO,

Figure 82. Facture toughness of TiB,-ZrO, composites (dots: data from [300],squares: data from P181).

7.7 Boride-Zirconia Composites

89 1

Figure 83. Hardness vs. Z r 0 2 content [305]

tion can be explained by a pronounced mutual Ti and Zr interdiffusion, resulting in (Ti,Zr)B2 and (Zr,Ti)02 solid solutions. The (Zr,Ti)02 may contain 14-16 mol-YO Ti02 at 1700°C in outer layers of the Z r 0 2 particles and hence result in a stabilization of the tetragonal modification. The maximum ZrB2 content in TiB2 at 1700°C was found to be 3.2mol-%. Moreover, after sintering between 2000°C and 2100°C there is evidence of an intergranular phase between adjacent TiB2 and Z r 0 2 particles consisting of zirconium titanate (Zr,Ti)*04which forms peritectically from a Ti02Z r 0 2 melt [239,306]. This embrittling phase can be avoided by the substitution of TiB2 by ZrB2 since zirconium zirconates do not exist. These composites are, however, also very sensitive to a spontaneous tetragonal-to-monoclinic transformation resulting in extensive microcracking (Fig. 84). If yttriu-stabilized Z r 0 2 is applied, the contribution of stress-induced transforma-

Figure 84. TEM micrograph of spontaneously cracked ZrB2-Zr02composites.

892


1.400

-

0

a

z 1200 f

m C

-E 1000VI

m

5 C

800-

m

600

-

80 100 20 LO 60 Tetragonal ZrO, content (mass%)

0

Figure 85. Mechanical Properties of TiB,-YTZP composites sintered at 1500°C [301].

tion to the toughening is more likely. An average bending strength of 1250MPa (maximum ob= 1500MPa) is obtained a 40 mass-% of 1.94mol-% Y203-doped Zr02, whereas the fracture toughness is, however, only 4 6 MPa m1/2but increases with higher Y-Zr02 contents (Fig. 85) [301]. The reason for this comparatively small increase in toughness was attributed to the high stiffness of the matrix phase, which reduces the dilatational strain associated with the tetragonal-to-monoclinic transformation of ZrOz [302,308]. In contrast to these results Muller [305] observes a parabolic increment of toughness as a function of the 3Y-stabilized tetragonal zirconia content (Fig. 86). The obtained maximum fracture toughness was 7.5-8 MPa m1/2at 25-40 mass-% Zr02 if the sample was hot-pressed at least 1700°C. Muller reports a maximum strength of 750 MPa at 20-25 mass-% Zr02. At higher 3Y-TZP fractions the strength decreases drastically, at > 40 mass-% the samples ruptured spontaneously during cooling from hot-pressing temperature. X-ray analysis of as-sintered and as-ground samples proved that during cooling 20-90% of the YTZP transforms to the monoclinic state which explains the severe cracking of the material. The residual tetragonal phase can almost be fully transformed to monoclinic by grinding [305]. The reasons for the spontaneous transformation are the diffusion of Y2O3 from the Zr02 solid solutions into the grain boundaries where, especially at temperatures > 18OO0C,a Y203-B203eutectic is found. On the other hand, the formation of Zr02 clusters with increasing sintering temperature and

7.7 Boride-Zirconia Composites

893

8.5

7.5

6.5

B

u

4.5

3.5 0

10

20

30

40

50

60

Mass -% ZrO,

Figure 86. Parabolic increase in toughness in TiB2-3YTZP-composites [305].

time is obvious [239,305]. Composites of comparably fine grain size and homogeneous phase distribution, i.e., isolated ZrO2 particles of z 1-2 p size, undergo a continuous phase redistribution and coarsening if hot-pressed up to 1900°C. Zirconia is then arranged in clusters with favored Zr02-Zr02 interfaces and has grown to > 5 pm size. As in the case of unstabilized Zr02, Ti is found in the surface-near areas of the Z r 0 2 particles and, vice versa, Zr is also present in the outer rims of the TiB2 particles. Thus, a grain boundary diffusion of both Zr and Ti, and presumably also B and 0 is most likely. Plasma etching makes an intergranular phase clearly visible. This kind of coalescence resulting in Z r 0 2 clusters incorporated by a contiguous TiB2 matrix with an optimized number of TiB2-TiB2grain boundaries is considered the reason for geometrical and mechanical destabilization of the tetragonal phase. The spontaneous t-m transformation is assisted by the large ZrOz grain size and the radial tensile stresses at the contact to TiB,. Since these interfaces are the weakest anyway, microcracking occurs here preferentially, and the residual tetragonal Zr02 particles in the cluster transform by autocatalysis. At higher temperatures, coarsening continues up to 2100°C where the material decomposes by internal oxidation [239]. Both boron and oxygen are then very mobile and exhibit a high vapor pressure so that the material is blown up drastically. Volume changes of more than 200% can be observed. The ceramic then consists of Ti02, titanium borates, and oxygen-deficient Zr02-Ti02 solid solutions whereas B 2 0 3 vapor creates round entrapped pores causing the strong bloating. Rupture was not observed, most probably because the grain boundary phase is already liquid or at least viscous at this temperature. The high enthalpy of formation of both Z r 0 2 and transition metal borides can be used to enhance densification by a chemical driving force starting from, e.g., Ti02 and ZrB, [303,304]:

a94


Ti02 + ZrB2

+ TiB2

+ Zr02

(39) This reciprocal salt-couple reaction yields a tremendous grain size refinement since all powder particles are involved in the conversion. After reaction the microstructure is characterized by an average grain size of 1-2 pm, approximately, and the preformation of solid solutions. Thus Eq. (39) has to written more precisely as follows:

Ti02 + ZrB2 + (Ti, Zr)B2 + (Zr, Ti)02

(40) The mechanical properties of the reaction product have not been exploited as yet. Besides TiB2, also WC, ZrC, TIC and mixtures thereof have been transformationtoughened with Y-stabilized Zr02 [309]. Hot-pressing of WC-TZP blends with up to

0TI (C, Nj-50 wt % T I 8,-a%

Zr 0,

c]TI (C, N)-50 wt % TI 8, -a% 2Y Zr 0, m

Q

E c

---

F

65 ._ U c a l

600

m 400

I

200

'

I

10

I

I

L

I

20 30 40 Zr 0 , content (wt%)

I

50

Figure 87. Mechanical properties of Ti(C,N)-50 mass-% TiBz-ZrOz composites (after [3lo]).

7.8 Cemented Borides

895

5Ovol.-% ZrOz at 1600°C yields almost dense composites with a flexural strength of 2300MPa and a hardness of HV 1800 (HRA = 93.4). A reaction between zirconia and tungsten carbide has not been observed but the appearance of a liquid phase is reported. A similar behavior of the mechanical properties has been observed for composites of Ti(C,N) containing Y-doped tetragonal zirconia composites where a significant increase in the strength and hardness with Y-Zr02 content is, however, accompanied by a modest increment in the toughness. In Ti(C,N)-50 mass-% TiB2-Zr02a decrease in hardness and strength combined with a parabolic increase in toughness was measured with increasing Z r 0 2 content [310]. The best performance was achieved by 12% ceria-stabilized ZrOz (Fig. 87). A characteristic of all the Zr02-reinforced borides is the comparatively high scatter of the mechanical properties measured.

7.8 Cemented Borides Due to their remarkable hardness, borides are interesting candidates for the development of metal-matrix composites resembling the hard metals based on cemented carbides or cermets such as WC-Co or Ti(C,N)-Ni composites. However, since metal-boron systems usually contain many binary and ternary borides a suitable choice of materials is difficult and requires full knowledge of the particular phase diagrams and compound properties. Moreover, to gain full advantage of the chosen metal-boride system, problems in densification such as wetting have to be overcome by appropriate dopants, and, finally, the comparatively little corrosion and oxidation resistance of borides and matrix phase at high temperatures have to be taken into consideration for an application.

7.8.1 Boron Carbide-Based Cermets The fabrication of metal matrix cermets with boron carbide as a dispersed phase is very limited under equilibrium conditions since B4C reacts with all metals, except Ag, Cu, Sn, and Zn, forming metal borides and graphite or metal carbides (e.g., [31 11). In systems with slow reaction kinetics, however, complex low-temperature materials with interesting mechanical properties have been investigated. A development from Kiew makes use of a Ti-containing bronze as a binder phase in which the reaction of Ti with B4C to give TiB2 is employed for active brazing and improvement of the wetting behavior. The use of pure Cu, Sn, or Zn, or alloys thereof for the infiltration of B4C power compacts usually fails since the wetting behavior is rather poor (wetting angle >90"), but this can be improved by adding Cr or other metals, which may react with the B4C when approaching equilibrium conditions. Other metal matrix composites with B4C particulates have been obtained using aluminum because of slow reaction kinetics. The process is based on a n infiltration

896


of liquid A1 into a porous body of B4C at temperatures between 700°C and 1200°C. Since A1 melts at 600°C and exhibits a significant vapor pressure at only slightly higher temperatures, the equilibrium between 1000 and 1880°C at which liquid A1 is stable with an Al-saturated B12(B,C,A1)3-solidsolution [73] cannot readily be utilized for liquid-phase sintering with small volume fractions of liquid. As shown by Halverson et al. [206], it is more effective to infiltrate compacted, or presintered, porous B4C bodies with liquid Al. The resulting material is a metal-reinforced B4C cermet rather than a liquid-phase sintered B4C ceramic. The wetting behavior is strongly influenced by oxidation layers formed on the surface of the B4C particles [312], but can be improved by superheating the melt. Between the melting point of A1 and approximately 1000°C, wetting angles of 100-150" are observed, which decrease to reasonable values with prolonged soaking for thousands of hours [206]. Hence, in that temperature range, only hot-pressing or hot isostatic pressing result in high-density cermets. Above 1000-1200°C, a suitable wetting behavior is obtained within minutes of annealing. Due to capillary forces and phase reactions both densification and adhesion of the metal-ceramic interface are excellent. During infiltration, reactions of A1 with B4C occur. Below 1200"C, A14BC, A1B2, A1B12, and AIBI2C2are formed within tens of hours whereas above 1200°C the generation of A14C3,A1B12,and A1B24C4is more favored [206]. If the composite is prepared by fast heating, infiltration, and rapid cooling, most of the aluminum matrix is retained unreacted. The matrix can then be hardened by a subsequent heat treatment at 800°C for 20 hours due to the precipitation of aluminum carbides and borides. Since the mechanical properties are determined by the A1 matrix, a K,, of 5-16 MPa m1I2and a flexural strength of 200-680 MPa can be obtained, depending on the quality and volume fraction of the metallic binder. The Vickers microhardness of 15.7 GPa for a 3 1 v01.-YOA1 composite is improved by annealing to 19.4 GPa. A re-investigation of the Al-B-C system with special emphasis on B4C-A1cermets was published by Pyzik and Beaman [90]. The composites consisted of both isolated and interconnected B4C particles and A1 matrix and only a small amount of ternary aluminum carbo-borides by fast firing and a post-sintering heat treatment between 600 and 1050°C. Cermets with a high B4C contiguity had the highest strength of 550 MPa if treated at < 600°C. Then the strengths drops to 400420 MPa and remains constant at higher heat treatment temperatures while structures with isolated B4C particles show a maximum strength of 420 MPa after a 600°C treatment with a similar decrease at higher temperatures. As expected, the latter composite with the continuous Al-matrix phase gives the higher fracture toughness of 11.7 MPa m1/2 after a 600°C annealing for 20 hours compared to 8.5 MPa m1I2 for the contiguous B4C microstructures, being best without any heat treatment (Fig. 88). Reactions of B4C with metals have also been employed for the bonding of boron carbide-based ceramics to metallic substrates. Nishiyama and Umekawa [3131 have studied the intermediate phases in B4C-Fe couples with A1 and Al/Mo interlayers. Depending on the treatment temperature and thickness of the foils, B12C2A1and intermetallic phases are formed that allow reactive brazing with a tensile strength of 55-60 MPa. In a similar way, porous boron carbide of several B:C ratios was infiltrated by Si-TiSi2alloys with a 1330°C melting point [94,109,233]. After wetting and


0

200 400 600 800 1000 Heat-Treatment Temperature, "C

897

1200

Figure 88. Effect of heat-treatment temperature on fracture toughness of B4C-A1composites (after

POI). infiltration, the liquid phase reacts with boron carbide to form TiBz and SIC. Therefore, these composites were discussed in Section 7.4.4.2.

7.8.2 Titanium Diboride-Based Cermets TiB2 is a candidate material for wear-resistant parts and cutting tools because of its high hardness, very high Young's modulus, considerable high-temperature strength and remarkable chemical stability. The liquid-phase sintering of TiB2 with suitable metallic binder phases is intended to produce cermets which combine the desired properties with sufficient toughness. Although TiB2 is stable together with liquid A1 and Cu, these systems have not been studied seriously for cermets but extensively for particle hardening of aluminum and copper alloys by TiB2. In the Al-Ti-B system it is, for instance, still uncertain whether a continuous (Ti,Al)B2 solid solution exists which is in equilibrium with A1 [89,3 141 or whether AlB2 and TiB2 coexist as separate phases [315]. Pastor [I 341 summarizes the results of other systems based on the iron group and other transition metals such as Cr, Zr, Mo, W, and Re. Fe, Co, and Ni have, however, the highest potential for this purpose. 7.8.2.1 TiB2-Fe Composites

In various studies Fe was investigated as a binder phase for TiB2 [316,317], but Samsonov [3181 and Kieffer and Benesovsky [ 3 191mentioned, that Fe2Bis generated during the liquid phase sintering of TiB2-Fe powder blends. Therefore, the fabrication of suitable TiB2-Fe cermets was considered impossible. Other studies by

898


I

I

I

I

I

I

I

-

h

-

1400

L + Ti&

0

-

v

9 2

3 +

-

1300

1268 'C

a,

Q

-

E

L + TiB,

+a 1183'C -

1200

L + TiB, + y

1167 "C -

y + Fe,B 1100

I

0

y + TiB,

I 20

I

I

I

40

+ Fe@ I

I

60

I 80

I

100

Concentration TiB,, (Mol0/o) Figure 89. Isopleth of TiB,-Fe section (after [320]).

Federov and Kuzma [ 1201 and Shurin and Panarin [ 1211, however, showed evidence for the existence of a pseudo-binary Fe-TiB2 equilibrium, which is characterized by a eutectic point at a temperature of 1340°C and 6.3mol.-% TiB2 (Fig. 23). After Smid and Kny [122] the two phase equilibrium is limited at 800°C by the tie-lines TiB2-Fe2B and TiB2-Fe (5 at.-% Ti). Ottavi et al. [123,124] finally proved that not pure Fe is in equilibrium with TiB2 but a Fe solid solution with Ti. The boundary tie-line interconnects TiB2 with Fe 0.5 at.-% Ti. Accordingly, a small excess of boron by result in the formation of FeB2. The first liquid phase in the pseudo-binary section appears at 1170°C coming from the Fe-FeB2 eutectic, the ternary eutectic is set to 1240°C (Fig. 24). The contradictions in these experiments have been attributed by Sigl and Schwetz [116,119,196] to the instability of the system TiB2-Fe in the presence of C or B4C impurities, which originate from the TiB2 synthesis by the carbothermic reduction of Ti02-B203or Ti02-B4C mixtures. These impurities react with Fe and TiB2 and form Fe2B and TIC. Thermodynamic calculations by Golczewski and Aldinger [320], however, take the y-a transition of iron more precisely into account and indicate that TiB2 is compatible with Ti-bearing ferrite only below 900°C (Figs. 89 and 90). The pseudo-binary section TiB2-pure Fe therefore contains FeB2 below 1167"C, and between 1 167 and 1268°C solid y- or a-Fe instead plus liquid. At above 1268"C, the two phase equilibrium TiB2-liquid is entered (Fig. 89) whereas in the TiB2-Fe (2 at.-% Ti) section FeB2 and y-Fe are absent (Fig. 90). Although there are still some uncertainties on the phase diagram, which the synthesis of pure two-phase cermets is based upon, the authors agree that the presence of oxygen and carbon impurities is detrimental to the wetting behavior

+

899

7.8 Cemented Borides 1600

I

1500

L + Ti& 1400

1300

'-

L + TiB,

1100

'

0

I

I 5

I

I

10

+a

I

I

I

I1

15

Concentration TiB,, (Mol%)

2o

I

Figure 90. Isopleth of TiB2-(Fe + 2 at.-% Ti) section (after [320]). Note the difference to Figure 89 concerning the stability of CL 4 y Fe and the appearance of the liquid phase.

and responsible for the presence of the embrittling but hard Fe2B phase, which also controls the sintering behavior and thus the properties. After the development of cemented borides with a Fe matrix by Funke et al. [316] and Funke and Yudkovskii [32 11, this cermet system was recently intensively studied again by Yuriditsky [115], Sigl and Schwetz [116, 1961, and Jiingling et al. [322,323], and, in connection with an enhanced European research activity, independently by Ottavi et al. [123, 1241, Ghetta et al. [117], Pastor et al. [324,325], and Sanchez et al. [326].

Liquid Phase Sintering of TiB2-Fe Composites

A disadvantage of TiB2-Fe composites is their delicate densification behavior, which has been attributed to the oxygen impurities present in commercial TiB2 powders. At solidus temperature, oxide impurities may give rise to evaporation and recondensation reactions involving volatile boron suboxides, which result in an exaggerated grain growth of TiB, without densification. Coarsening by means of vapor phase or surface reactions consumes surface energy and, therefore, sinter activity. Consequently, much effort was put into the removal of oxide contamination from TiB2 powders, e.g. by reduction with BC13 [263,327]. As already mentioned before, oxides and carbon impurities are also responsible for the wetting behavior of the liquid. Figure 91 illustrates both complete and imperfect wetting of TiB2 by liquid iron. Thin Ti02 layers covering parts of the TiB2 grains are assumed to account for the poor wetting [116,196]. Experiments

900

7 Boride-Bused Hard Materiuls

+

Figure 91. SEM micrograph of an 8Ovol.-% TiB, 2Ovol.-% Fe cermet sintered at 1600°C. Note both the incomplete and complete wetting of TiB, by liquid Fe.

by Ghetta et al. [328] have demonstrated that the spreading of liquid Fe in fact worsens with increasing oxygen content of TiB2 substrates (Fig. 92) whereas powders with < 0.26 mass-% oxygen show a wetting angle of < 5" after 10 min soaking time. Generally, transition-metal diborides should be wetted well by liquid metals due to their missing gap between the valency band and the conduction band whereas oxides exhibit bad wetting behavior due to their partially ionic and partially covalent character in bonding [329-33 11. Although an improvement of wetting by liquid Fe may be obtained by addition of Cr or Ni [332], breaking up the Ti02layer is thus considered the key to well-processible TiB2 powders. 80

-1

7 7 - r

Oxygen content in TIB,

70 0

a- 60 50 40

30 20 0

500

1000

1500

2000

Time, sec. Figure 92. Oxygen contamination affecting the wetting behavior of Fe liquid on sintered TiB, at 1300°C (redrawn after [117]).

90 1


Figure 93. SEM micrograph of a TiB2-Fe cermet with a nominal composition of 80voI.-% TiB, + 20vol.-% Fe. Note the existence of Fe,B in the binder phase (after [119]).

It also appears that the impurities in commercial TiB2 promote the formation of iron boride, Fe2B. A TiB2-Fe alloy with a typical two-phase microstructure of the binder, i.e. with both Fe and Fe2B making up the matrix between the TiB2 grains, is shown in Fig. 93. Fe2B is a hard but fairly brittle compound [333] and its formation consumes a substantial fraction of the ductile Fe phase such that Fe2B is believed to embrittle the composite seriously. Avoiding that phase has consequently been a major goal for developing tough TiBz cermets. Commercial TiB2 powders are typically produced by the carbothermic reduction of Ti02 with B2O3 or B4C. Both reactions Ti02

+ B 2 0 3+ 5 C + TiB2,+5

CO

(41)

and 2Ti02

+ B4C+ 3 C + 2TiB2 + 4CO

(42) are being industrially utilized. A typical analysis of the as-received TiB2 is listed in Table 10. Since this powder originates from the B4C-processingroute, i.e. Reaction (41), it is not surprising that about 90% of the total carbon is present as boron carbide and only 10% is free carbon. Assigning the oxygen content to particular compounds is not unambiguous, but it is estimated that approximately 1/2 of the oxygen is present as Ti02 and B203,respectively. Nitrogen is completely bonded as TIN, as BN is usually not found. It is therefore concluded that B4C, B2O3, free carbon, and Table 10. Chemical composition of the as-received TiB2 and Fe powders Total C Total N B4C Free C TiO, BZ03 Powder Spec. Total 0 surf. [weight-%] [weight-%] [weight-%] [weight-%] [weight-%] [weight-%] [weight-%] Im2/nl TiB Fe

3.2 0.5

2.1 0.2

0.20 0.05

0.05 -

0.85 ~

0.02 ~

2.6 -

1.5 ~

902


Table 11. Chemical compatibility of the compounds in TiB2-Fe powder mixtures Fe

C

B4C

TiOp

B2°3

Ti02, generally make up the critical impurities in the TiB2 powder. More carbon is introduced to the powder blend when organic binders are added and it should be noted that the oxygen content increases significantly upon milling [118]. The key questions for a successful liquid phase sintering of TiB2 with iron are, therefore, (i) how to break up the Ti02-films on the TiB2 grains and (ii) how to avoid the formation of Fe2B. Sigl and Jiingling [119] have shown that powder compacts of TiB2 20 v0l.-% Fe heated to 1000°C contained Fe2B and Ti203.Notably, their concentration did not increase significantly at temperatures up to 1600"C, suggesting that the major part of the reaction had been completed at that time already. Since Fe2B and Ti203 already appear at low temperature, solid-state reactions are considered responsible for the formation of this compound. Of particular interest are the reactions of Fe with the boron-containing impurities, i.e. with B4C and B203.According to Table 11 the following reactions are candidates to develop Fe2B:

+

+ 8 Fe --+4Fe2B + C 2 B203+ 7 Fe 2Fe2B + 3 FeO B4C

(43)

(44) The change in Gibbs Free Energy, AG, has been calculated for Eqs. (43) and (44) with the data from Janaf [334]. The results are plotted versus temperature in Fig. 94. Obviously a reaction between Fe and B203is unlikely in the whole temperature range of interest, which has been confirmed by corresponding model reactions. It is, however, well known that Fe and B4C react heavily even at low temperatures forming Fe2B and free carbon [335]. Due to the favorable properties of Fe2B, this fact is commercially utilized for depositing thin layers on steel surfaces [336]. As mentioned above, borides can be synthesized by the carbothermic reduction of B203 with carbon and a metal or a metal oxide [337], e.g. according to Eq. (40). Since free carbon is present in the as-milled powder mixture and also develops during Reaction (43), the following process must also be considered a potential source of Fe2B: ---f

+

B203 4Fe

+3C

2Fe2B

+ 3 CO.

(45) Although the AG, which is plotted as a dashed line in Fig. 94 appears to favor Reaction (45) above 800"C, it is still considered unlikely. According to Sigl and Jiingling [119], the following steps rather than the gross Reaction (45) are proposed to operate instead t


903

a,

,f

-400

-800 L 600

800

1000

1200

1400

1600

Temperature, "C Figure 94. The change in Gibb's free energy for reactions yielding Fe2B at 0.1 mbar total pressure.

B20

+ 3CO

2B+4Fe

2B

+ 3C02

+ 2Fe2B

3COz+3C+6CO Given the CO/CO2 equilibrium according to Eq. (48), the Gibbs Free Energy for Reactions (45) and (47) can be calculated and is plotted in Fig. 94 at a total pressure of 0. I mbar. Though Fe and B would favorably react, CO is unable to reduce B 2 0 3 below 1000°C and therefore the carbothermal formation of Fe2B is operating only beyond 1000°C. This hypothesis has been confirmed experimentally: While a B203Fe-C powder blend with a composition according to Eq. (45) does not change its phase composition after annealing the mixture at 1000°C for 1 h, the X-ray diffraction of identical samples heated up to 1600°C shows ample evidence for Fe2B. This suggests that processes involving liquid iron can play an important role during the formation of Fe2B as elaborated below. In the regime of liquid-phase sintering TiB2 is dissolved in the Fe melt according to (49) TiB2 {Ti} + 2 {B} where the brackets { } denote the dissolved state of an element in liquid Fe. Obviously impurity compounds such as T i0 2 , Ti203, C, B203, TIN, and Fe2B will also dissolve in the liquid. During cooling some of these compounds reprecipitate from the melt in their original composition, e.g. TIN or T i2 0 3and hence do not influence the equilibrium concentration of Ti and B in the liquid. Species such as and {B} but only little carbon, Ti02, and B2 0 3which introduce a lot of {C}, {0}, {Ti} into the melt do, however, precipitate as compounds that are richer in Ti than the previously dissolved species, e.g. as

904


B

Fe

Fe,Ti

Ti

FeTi -Ti [at.-%] +

Figure 95. Isothermal section of the Fe-Ti-B system at 1000°C. The point indicates the cermet composition and the arrow the change in composition if the titanium content in the liquid phase is decreasing.

+ 2 {B} + TiB2 {Ti} + { C} + TIC 2 {Ti} + 3 {0}+ Ti203 {Ti}

(50) (51)

(52) Thus, these processes extract a lot more {Ti}from the melt than has been introduced during the dissolution. Consequently, the solubility product (uTi a;) being constant at a given temperature forces the composition of the liquid to shift towards low {Ti) and high {Bj concentrations upon cooling. Finally, a {Bj-rich Fe melt is left behind. This conclusion is supported by the observation that a liquid phase is stable down to 1 150°C[335], i.e. the eutectic temperature of Fe-Fe2B,although the TiB2-Feeutectic solidifies at 1350°C. Eventually, the remaining melt crystallizes as a mixture of Fe and Fe2B (Fig. 93). An experimental confirmation of this conclusion was obtained by Sigl and Jungling [119]. These processes shift the composition of the cermets from their intended place in the (TiB2 + Fe) two-phase field into the (TiB2 Fe Fe2B) three-phase equilibrium of the Fe-B-Ti ternary system, either by introducing B or by extracting Ti (Fig. 95). Thus the Ti concentration in TiB2 cermets is a variable to be strictly controlled during sintering, very similar to the C content of cemented carbides, which must be adjusted well to avoid graphite or q-phase formation. Figure 91 suggests that liquid Fe cannot penetrate in between the TiB2 grains, unless the Ti02 layer is broken up. Total wetting is nevertheless feasible. Figure 96 indicates that this process is closely related to the precipitation of particulate Ti203. It is generally accepted that the reduction of Ti02 involves various titanium suboxides, i.e. Ti305,Ti203,and TiO, until Ti is finally obtained [334]. The compound that actually evolves from this process depends on how well the reducing agent can overcome the affinity of oxygen in the Ti oxides. Since the TiB2 powder of this study contains only free carbon as a suitable reducing agent, reactions such as

-

+ +


905

Figure 96. Optical micrograph of a TiB2-Fe cermet containing Ti203emerging from powder contaminations.

+

2 T i 0 2 + C 4Ti203 CO

(53) need to be studied. The Gibbs free energy for Reaction (52) is plotted in Fig. 97 as a function of temperature at a total pressure of 0.1 mbar. Though carbon should be able to reduce T i 0 2 at temperatures above 850°C according to the thermodynamic calculations, a reaction between a thin oxide film and solid carbon is considered unlikely. Instead, carbon monoxide rather than free carbon is believed to deoxidize T i 0 2 according to 2Ti02

+ CO

4

+

Ti203 C02

-

(54)

since the driving forces for Reactions (52) and (53) are identical. A further reduction

7 Y

)r

F a

5

a

9

LL

B 0 (3

Temperature, "C Figure 97. The change in Gibb's free energy for the reduction of titanium oxides by carbon monoxide at 0.1 mbar total pressure.

906


of Ti203according to

+

Ti203 CO + 2TiO

+ C02

(55)

is also supported by thermodynamic arguments (Fig. 97), but CO seems unable to reduce T i 0 to Ti at a pressure of 0.1 mbar. Since experiments have shown that TiB2 powders with more than 2 weight-% oxygen can be sintered to full density [116, 196,3221 and that Ti203 particles precipitate during sintering, it is considered that the free carbon in the as-received TiB2 powder and the carbon being set free from the reaction of Fe with B4C generates CO, which eventually breaks up the oxide layers on the TiB2 grains. The result that Ti203appears instead of TiO, as may be inferred from Fig. 97, is attributed to the small quantity of carbon normally available. Principally, a carbon quantity sufficient to form either Ti203 or T i 0 should promote the wetting and consequently the sintering behavior of TiB2 powders. This conclusion is supported by Fig. 96, which suggests that Ti203 particles hardly disturb the wetting by liquid Fe. Excess carbon can, however, lead to the formation of Fe2B via the liquid-phase processes described above. Since small quantities as in the previous case will be used up during the reduction of Ti02, the carbothermic formation of Fe2B from Fe and B203according to Eq. (45) cannot occur. Hence it is concluded that Fe2B originates mainly from the reaction of Fe with B4C and that the gross reaction B4C

+ 2 Ti02 + 8 Fe + 4Fe2B + Ti203 + CO 7'

(56)

operates at temperatures below 1000°C.Though a good part of Fe2B is due to solidphase reactions, it is anticipated that B203 will also give rise to both Fe2B and Ti203. However, the role of B203is as yet not fully understood. In conclusion, deoxidizing the thin TiOz films appears to be an absolute prerequisite for the sintering of TiB2 with liquid iron. Though carbon monoxide cannot fully reduce Ti02 at the sintering conditions (0.1 mbar), the formation of particulate lower oxides such as Ti203 seems to be sufficient for supplying a good wetting behavior. Kinetics of Liquid-Phase Sintering of TiBz with Fe

The sintering kinetics and phase development during sintering were investigated by Jungling et al. [322]who demonstrated that Fe2B already forms at low temperatures, i.e. before shrinkage starts (Eq. (55)). The deoxidation of TiB2 grains and the simultaneous formation of Fe2B contribute significantly to the densification behavior of TiB2-Fe-materials, i.e. nearly full densities (>98.5% T.D.) were achieved by pressureless sintering. However, the formation of Fe2B consumes a considerably high amount of the Fe binder, which deteriorates the fracture toughness of the composite seriously. For these reasons Jungling et al. [322] and Sigl and Jungling [338] suggested the addition of Ti to balance the excess boron in order to obtain materials with a ductile binder: 2 Ti02

+ 2 Ti + B4C + 2 TiB2 + Ti203 + CO

(57)


9

1800

4

1400

0

1000

-4

2i 3

5 a,

907

8 A-

d

Q

-8

600

-12

200

0

100

200

300

Time, min Figure 98. Densification behavior of model alloys with 30 v01.-% binder with and without Ti addition (after [340]).

This reaction as well as Eq. (56) were confirmed by thermodynamic considerations. Ghetta et al. [328] and Ottavi et al. [339] concluded from wettability studies that the addition of Ti deoxidizes the TiB2 grain surface by the formation of a parasitic Ti(O,C,N) phase. Nevertheless, the sintered density of a material with Fe2Ti addition was rather poor ( ~ 8 9 %of theoretical density). The addition of NdNi5 enhanced the density to 96.7% of theoretical density and resulted in a parasitic N d 2 0 3phase. Jungling et al. [323] report on model experiments to study the sintering kinetics of TiB2 with 20 v01.-YOcarbonylic Fe and 20 v0l.-YOcarbonylic Fe 7 mass-% Ti up to 1700°C in 0.1 mbar Ar. Sintering parameters were optimized for materials with 15 and 20 v01.-YOFe and 20 vol.-% Fe-Cr-Ni binder and for materials with 5-20 v01.-YO Fe-Cr-Ni binder phase with Ti addition to prevent the formation of the Fe2Bphase. Materials with Ti addition were pressurelessly sintered at 1650°C to a density of about 95% T.D. followed by a hot isostatic pressing step at 1460°C with a pressure of 100 bar Ar. Figures 98 and 99 illustrate the shrinkage behavior and densification rates of the alloys with and without Ti additions. Soon after the spontaneous start of densification, the sintering rate decreases drastically, but eventually begins to increase again. Without Ti addition shrinkage starts at 1140°C but in the alloy containing Ti, sintering begins at 1250°C. After this temporary decrease, the densification rate increases in both materials above 1320°C. Densification comes to an end after 2 h isothermal sintering at 1700°C. Unlike the material with pure Fe, the alloy with F e t - 7 mass-% Ti does not develop any Fe2B. Again Ti203is formed at 1000°C but the intensity ratio Ti203/ TiB2 is smaller than without Ti addition and increases with increasing temperature. Additionally, X-ray diffraction presents evidence for the formation of very small amounts of TIC in the sintered sample. Figure 100 displays the density as a function of the sintering temperature without holding time. Samples with 15 or 20vol.-% Fe show similar densities in

+

908

.-r


1800

2

1400 $I

-. s E

n! 3 2

1000

a,

Q

-1

50

0

100

200

150

250

200 350

300

Time, min Figure 99. Densification rate of model alloys with 3Ovol.-% binder with and without Ti addition (after [340]).

the temperature range of 1500-1700°C and reach nearly 99% T.D. The addition of 7 mass-% Ti results in a strong decrease in the density of as-sintered materials, but at > 1650°Ca density of >95% T.D. and closed porosity are obtained. Further densification can be achieved by subsequent hot isostatic pressing. Materials with 5, 10, and 20 v0l.-YOFe-Cr-Ni-binder with 7 mass-% Ti addition nearly approach their theoretical densities by a post-HIP treatment at 1460°C with a pressure of 100 bar Ar (Table 13). The spontaneous shrinkage for materials with Fe2B at 1140°Cis explained by the formation of a liquid phase and a subsequent rearrangement of particles. This conclusion is compatible with the eutectic temperature of 1177°Cin the system Fe-Fe2B 100

I

,

--I

,

,

r

Grade F + carbanyllc Fe

d 95c

c 90

--

"1 i

-

+20 VOl% 1

1450

1500

1550

1600

' 1650

1700

1750

Sintering Temperature, OC Figure 100. Densification behavior of TiB2

+

15 or 20 v01.-% Fe cermets (after [340]).


909

Table 12. Solid-state and vapor-phase reactions during sintering. Without Ti additive B4C + 8Fe

+ 4Fe2B

With 7 mass-% Ti

+C

-

c + c02 + 2 c o 2 TiOz

+ CO

+

+

Ti203 C 0 2

~

B4C + 2Ti

--i

2TiB2 + C

c + co*+ 2co 2 Ti02

+ CO

+

Ti203+ COz

Eq .

(57) (58) (59) (60)

[341]. In this way Fe2B contributes a lot to the excellent densification behavior of these materials but it also consumes a major part of the ductile Fe-binder phase. In materials without Fe2Bthe beginning of shrinkage is delayed. A first maximum in densification rate appears at 125OoC,about 90 K below the eutectic temperature of Fe-TiB2. Again, this densification peak is attributed to the formation of a liquid phase, which coincides with the eutectic Fe-Fe2Ti at 1289°C rather than with the Fe-TiB2 eutectic [342]. This can be explained by the solution of some Ti in the binder during heating. As proved by X-ray analysis, the addition of 7 mass% Ti fully prevents the formation of Fe2B. Smaller amounts of Ti yield corresponding quantities of Fe2B. Table 12 compares the particular reactions of materials with and without Ti additions. After initial particle rearrangement the densification rate decreases in both materials. It is accelerating again when a sufficient amount of liquid phase is formed due to the eutectic Fe-TiB2 at about 1320°C. During cooling the liquid phase finally solidifies at 1140°Cfor samples with and at 1250°Cfor samples without Fe2B as proved by exothermic reactions producing the small expansion peak in Fig. 99. The last expansion peak at 800°C is due to the a 4 y transformation of Fe. As expected, this peak is much higher for the composite that does not contain Fe2B. In conclusion, the formation of FeB2 can be avoided by the control of powder impurities, sintering atmosphere, and compensation of excess boron and carbon by titanium addition. The wetting behavior depends strongly on oxide surface layers and is excellent if surface contamination of TiB2 is entirely removed. The sintering temperature and Fe addition must ensure that a suitable amount of liquid phase is available at the lowest possible vapor pressure. Since the reducibility of T i 0 2 and the vapor pressure of liquid Fe determine the atmospheric pressure, the interval (0.01-0.1) mbar < p < 1 mbar is considered best suited. The optimum amount of liquid phase is generated at 1450 < T < 1650°C if grain growth is taken into account as another limiting factor. Accordingly, a feasibility diagram of suitable sintering conditions can be constructed (Fig. 101) showing the optimum environmental parameters. Microstructure and Properties of TiB2-Fe Composites

Since the eutectic concentration in the quasi-binary TiB2-Fe system with 14 v01.-% TiB2 is considerably closer to the metal corner than in the similar WC-Co system (32 v01.-YO WC), a much smaller amount of liquid phase is generated upon sintering,

97.8

Density of pressureless sintered material [% T.D.] Density of post-HIP [Yo T.D.] 3-point bend strength [MPa] Young's Modulus [GPa] Hardness [HV,,] 25°C Hot hardness [HV2]200°C Hot hardness [HV,] 400°C Hot hardness [HV,] 600°C Hot hardness [HV2]800°C K,,/ICL [MPam] KI,/NB [MPam] 5.9 5.5

-

-

-

610 510 2260

-

5

Binder content [vol.-'Yo]

Binder composition

Starting powders

6.9 9.3

760 465 1830

98.8

15

7.9 10.0

-

720 465 1510

99.1

20

cc-Fe + Fe2B

TiB2 + Fe

790 465 1610 1130 870 730 480 8.2-8.7 11.4-1 1.9

99.2

20

+ Fe2B

+ Fe-Cr-Ni

y-Fe-Cr-Ni

TiB,

-95 100 900 515 2040 1560 1290 1030 660 7.3-8.1 7.6

5

-95 99.1 1010 510 1800 1270 990 810 580 8.1-9.5 9.0

10

%95 99.2 950 465 1450 1150 850 740 500 12.4 14.0

20

+ 7 mass-% Ti y-Fe-Cr-Ni

TiBz + Fe-Cr-Ni

Table 13. Properties of optimized TiB2-hard metals. ICL: Indendation-crack-length method, NB: notch-beam method.

b

2

6.

.

% 2

3%.

Q

6.

g

Ptx

2 -.

7.8 Cemented Borides 10'

-ki E

Y

1

'

1

'

1

'

1

91 1

2

reduction of Ti0,

1oo

grain growth

10'

-

-

lo3 1300

1400

1500

1600

1700

1800

Temperature [ "C] Figure 101. Optimum parameters for the pressureless sintering of TiBz-Fe cermets limited by grain growth, evaporation, suboxide formation, and suitable liquid phase content.

which makes densification more difficult. A simple increase in temperature cannot satisfactorily balance the lack of liquid because it is accompanied by accelerated coarsening of TiB2 due to Ostwald ripening [I 161. The volume fraction of binder phase thus ranges between 10 and 30%. A typical microstructure is very similar to that of WC-Co hard metals. Euhedral TiB2 particles are embedded in a continuous Fe matrix. The densification mechanisms are typically dissolution and reprecipitation as well as coalescence, i.e., rearrangement and intergrowth of particles with common faces of the same orientation. The latter mechanism is active if the volume fraction of liquid exceeds 30% but may result in the growth of elongated platelets. The residual porosity after pressureless sintering between 1500°C and 1800°C depends upon the initial liquid phase composition. At 15OO0C, 88% of the theoretical density has been obtained for the TiB2-Fe system at lSOO°C), whereas at 1450°C Ti addition results in 98% and combined Ti-Nb additives result in 96.7%. Hot-pressing and hot isostatic pressing yield densities >98% at a lower binder content. Figure 102 shows the microstructure of a plasma etched sample with 2Ovol.-YO binder. Besides TiB2 with its hexagonal grain shape and Ti203a binder consisting of Fe and Fe2B is present. A Fe-Cr-Ni binder phase diminishes the grain sizes notably. In materials with 7 mass-% Ti addition to the Fe-Cr-Ni binder the FeB2 phase is absent. The total carbon content of samples without Ti addition strongly decreases during sintering from 0.11 YO(starting powder) to 0.0 1YO(sintered compact). The oxygen content slightly decreases from 2.8% to 2.4%. Table 13 summarizes some properties of TiB2 with 15 and 2Ovol.-% Fe-Fe2B binder, 20 vol.-% Fe-Cr-Ni-Fe2B binder, and 5-20 v01.-YO Fe-Cr-Ni binder. Samples with a two-phase binder reach densities of about 99% T.D. by pressureless sintering. However, alloys with a single-phase binder approach no more than 95% T.D. by pressureless sintering, yet can be densified to near theoretical density by a

912

7 Boride-Rased Hard Materiuls

Figure 102. Plasma-etched microstructure of a cermet with 20 v01.-% cc-Fe-Fe2B-binder.

post-HIP treatment. As expected, samples with a low binder content are harder than those with high binder fraction, provided the binder composition remains unchanged. The observation that materials with Fe-Cr-Ni-Fe2B binders are harder than materials with Fe-Fe2B binders is attributed to the smaller grain size of the former materials (Fig. 103). Simultaneously, the fracture toughness increases. Thus avoiding the formation of Fe2B strongly increases the fracture toughness but decreases the hardness only moderately, such that the combination of hardness and fracture toughness is clearly improved [343] compared to WC-Co-based hard metals. The bending strength does not vary significantly with increasing binder content (Fig. 104) but the presence of FeB2 results in considerably lower values. This not only attributed to the brittleness of this phase but is also due to the residual 2500

s>

1

2000

r

ui u)

a,

c

E 1500

m

I conventionalcemented carbi

10001'

0

" " " " I

5

10

-

'

"

'

I

'

15

" ' 1

20

Fracture Toughness hC,M P a h Figure 103. Hardness and fracture toughness of TiB2-Fe cermets in comparison to conventional hard metals (after [340]).


a a

913

1200

.- 1000 a

?t

800 600

a,

L

400

1

1 I

3

Fe-Cr-Nt-Fe,B

I

0

5

10

15

20

25

Binder Content, vol% Figure 104. Flexural strength dependence on binder volume of TiB2-Fe cermets with and without FezB (after [343]).

2 % porosity. A post-HIP treatment may overcome this problem and also yields a lower scattering of data. Additions of metals such as Mo, Cr, Ni, and Co to the Fe matrix may be used to fabricate composites with improved mechanical and corrosion properties. Figure 105 shows significant variations in the bending strength with increasing amounts of Mo in the binder phases of different volume fractions [115]. Figure 106 illustrates the hot-hardness of composites with various binder content, again in comparison to WC-Co and WC-Fe-Co-Ni hard metals [343]. It is evident that TiB2-based materials with a small amount of binder are superior in the entire temperature range measured.

Figure 105. Variation of strength with Mo content in the binder phase of TiB,-Fe cermets (after [115]). Binder volume: A, 12.5vol.-%; M, 15vol.-%; 0 , 17.5vol.-%.

914


25001

1

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

I

2000

ujv) 1500 a, C

E ca

T

1000

0

200

400

600

800

Temperature, "C Figure 106. Room-temperature hardness HV,, and high-temperature hardness HV2 of TiB2-Fe cermets and conventional hard metals (after [340]).

7.8.2.2 TiBz-Ni, Co Composites Transition metals such as Co and Ni are useful for liquid-phase sintering of TiB2type borides causing rapid densification and grain growth by Ostwald ripening, but they react chemically to form M,B,-type phases or even more complex ternary phases that are very brittle and possess a comparatively low melting point < 1500°C. No metal-reinforced composites containing solely TiB2 as the hard phase can thus be produced except where reactions can be at least partially avoided by fast heating during hot-pressing. In contrast, Co- and Ni-based alloys can be successfully improved for wear resistance by the incorporation of TiB2 and CrB2 particles if reaction layers of lower hardness can be tolerated. In the case of Ni-bonded TiB2, a ternary 7-phase with the composition Ni21Ti2B6 forms by the dissolution of TiB2. At 800°C, the z-phase is in equilibrium with Ni, Ni3B, Ni3Ti, and TiB2, as shown in the isothermal section in Fig. 25 [125]. During sintering or hot-pressing, the 7-phase is generally not obtained. The formation of 7 may be suppressed either by the presence of T i 0 or Ti02 [344], or for kinetic reasons by fast firing and quenching. Hence the residual matrix phase consists mainly of Ti-containing Ni3B solid solution [218,344,345]. Since other Ni-borides such as Ni2B and Ni3B4 and even metallic Ni are found after hot-pressing at 1600°C [94], equilibrium conditions are obviously not easy to obtain. Typical metal contents required for a successful liquid phase hot-pressing of TiB2 are 5-25 mass-% (i.e. 2-12 at.-%o) Ni or Co. In order to avoid reactions consuming TiB2, the borides of Ni or Co have also been used. By this method, the sintering temperatures have been decreased from 2100°C to 1400°C [134, 277, 294, 295, 3461.


915

Figure 107. Optical micrograph of a TiB2-Ni composite, pressurelessly sintered at 1600C. Note the incomplete wetting of TiBz by liquid Ni and the formation of Ni borides in the matrix phase.

The liquid phase intensifies the mass transport but causes an accelerated grain growth. The microstructures of composites prepared by liquid-phase sintering are similar to those of hard metals. The TiB, particles form a rigid skeleton of faceted crystals whereas the binder, e.g., Ni3B, Ni2B, Ni3B4,or comparable compounds of Fe, Cr, or Co, is the matrix phase. The TiB2 grain size usually exceeds 20pm (Fig. 107). Depending upon the wetting behavior, which is influenced by the surface oxidation of the hard material phase, round pores may accumulate at particle/matrix interfaces or close to triple junctions that have not been completely infiltrated by the liquid phase. Moreover, the evaporation of Fe-, Co-, or Ni-borides may cause entrapped gas pores. Hence, hot-pressing is still required for a homogeneous distribution of the liquid phase, particle rearrangement, and complete removal of the residual porosity. In contrast to hard metals, the matrix phase is very brittle, e.g., the K,, of Ni3B equals 1.4-1.9 MPa m"2 [347], and hence does not improve the mechanical properties. Nishiyama and Umekawa [348] report on sintering and wear application of a TiB, cermet with Ni7Zr2 binder. Pressureless sintering between 1450 and 2100°C in hydrogen results in an incomplete melting of the alloy added with wetting angles of around 90". The phases detected are Ni4B3, Ni7Zr2,and solid solutions thereof with Ti. Since TiB2 forms a rigid skeleton the hardness HV, reaches 30-33 GPa at binder contents of < 8 mass-%, dropping almost linearly to 17 GPa with 30 mass-% Ni7Zr2 addition. Due to the toughness of the matrix phase, the 3-point bending strength improves from 500-600 MPa at < 5 mass-% binder to 850-900 MPa with > 15 mass-% of additives. 7.8.2.3 Transition-Metal Diboride Cermets with Co, Ni, Cr, Mo, and W

The research for high-strength/high-toughnesscomposites in the TiB2-transition metal systems resulted in many attempts to avoid brittle matrix phases and to introduce hardening particles while making use of the opportunity of liquid-phase

916


-

: 1 0 0 0 - :;9 x

I

. 0

0

0

TiB,+lrn%CoB TtBz+5rn%TaB,+lrn%CoB

TiB +, 5 rn% W B ,

st

1m % C o B

m 2 200-1 0

5 10 15 A v e r a g e groin size (prn)

20

Figure 108. Strength-grain size relation in TiB,-COB-based composites (drawn according to data in [2931).

sintering. Pastor [I341 reports on the manifold of diborides that have been sintered with Fe, Co, Ni, Cr, Mo, W, and even Re additives, yielding closed porosity or almost total density. Unfortunately no phase compositions were presented in most cases. In order to avoid brittle ternary phases, many liquid-phase sintering systems started with combinations of binder metals, sometimes even with their low melting boride phases, which will be reviewed in the following paragraph. Hot-pressed composite materials developed from the more complex systems of the type TiB2-M1B2-M" with M' being Hf, V, Nb, Ta, Mo, or Mn and MI' being sintering additives such as Co and Ni, exhibit bending strengths between 850 and 1000 MPa, which are due to the grain growth inhibiting influence of the 1-5 mass% of M"B2 particulates (Fig. 108) [293-2961. During liquid-phase sintering in a Co- or Ni-boride melt, both TiB2 and M' are partially dissolved and reprecipitated as a solid solution. The effect of grain growth retardation as well as of strength and hardness increments is attributed to stresses at the TiB2/(Ti,M')B2 phase boundaries generated by the mismatch of the lattice parameters between the unreacted TiB2 acting as a nucleus and the epitaxially precipitated (Ti,M')B2 solid solution. In the case of a TiB2-5 mass-% W2B5/TaB2material with 1 mass-% COB binder the and 14 x depending on the hot-pressing lattice strain ranges between 9 x temperature [293]. An improved sintering behavior of this material was obtained by the addition of 1.7% TIC, which reduced the porosity from 0.3-0.7 to 0.1-0.2 vo1.-YO after hot-pressing at 1500°Cand by intensive ball milling resulting in an average particle size of 1 pm, but increased oxygen contamination. Watanabe and Shoubu [297] reported the formation of a (Ti,Ta)(C,O) solid solution, which is considered to initiate the improved densification resulting in a flexural strength of l000MPa. The strength-grain size correlation in Fig. 108 can be interpreted in the way that the fracture is flaw-controlled at average grain sizes < 7-8 pm and microcrack-controlled at > 8 pm. In a similar multiphase system, transition metal carbides were used as additives for pressureless sintering of TiB2, yielding composites of binary and ternary borides [218,296]. Attrition milled powder mixtures of TiBz with 3-10 mass-% Co or Ni and 20-35 mass-% WC have been sintered in a vacuum at temperatures between


917

Figure 109. SEM micrograph of a (Ti,W)B2-W2NiBz(w)-Ni,B composite. Light areas: W2NiB2; intermediate: Ni3B; dark: (Ti,W)B,.

1500 and 1700°C for 60 to 120min yielding 98-99% density. Densification starts above 980°C due to the formation of a liquid phase in the Ti-B-Co/Ni system (Fig. 26). At this early stage, a rigid skeleton of TiB2 and WC develops. Due to dissolution and reprecipitation, a (Ti,W)B2 solid solution grows on the residual TiB2 particles. Subsequently, crystals of o-phase form with compositions of WCoB or W2NiB2.Upon cooling, the residual liquid phase crystallizes as C-and Ti-enriched Co3B or Ni3B solid solution. A typical microstructure is shown in Fig. 109. Sintering at 1700°C for 2 h yields an average particle size of 0.8 pm, resulting in a flexural strength ob= 600-680 MPa. The K,, of 6.5-7.5 MPa m1/2is mainly attributed to crack deflection because of the weak Co- or Ni-boride intergranular phases. As already shown, an interesting characteristic of the mixed diborides is the socalled core-rim structure, which is well known from complex cermets based on (Ti,Ta . . . )(C,N) compounds. During sintering or hot-pressing of the particular carbonitride powder blends solid solutions are formed at the phase contacts but the annealing time is insufficiently long to obtain equilibrium conditions. Consequently, interdiffusion is stopped, an unreacted core of, e.g., Ti(C,N) remains whereas an outer layer, the rim, consists of, e.g. (Ti,Ta)(C,N) solid solution with a very distinct boundary to the residual host crystal. Exactly the same effect is observed in borides of all kinds of composition if sintered for a not appropriately long time (Fig. 1 lo). In case of liquid-phase sintered borides, this layer is formed by epitaxial precipitation of the thermodynamically more stable solid solution onto a nucleus of undissolved starting material. Recently, Telle et al. [349] observed dislocations at the interface of TiB2-(Ti,W)B2solid solutions in hot-pressed TiB2WB2 composites (Fig. 11 l), indicating that stresses are generated at the epitaxial interface as considered by Watanabe and Kouno [293]. Another approach to fabricate metal-matrix-based boride and carbide composites according to the Lanxide process starts with reactive blends of B4C and Ti or Zr metal. Upon conversion to TiB2 or ZrB2, respectively, a large release of heat is observed, which can easily lead to partial melting of the composites. Depending upon the starting composition, residual metallic Ti or Zr, or B4C may be found after reaction. Interesting microstructures can also be obtained if TIC or ZrC are added as fillers [350].

918

7 Boride-Bused Hurd Material5

Figure 110. Core-rim structure of (Ti,W)B2 solid solution on a TiB, host crystal.

Similar metal-matrix composites with B4C, Sic, and TiB2 as fillers have been fabricated by the so-called Lanxide- or Dimox-process (direct metal oxidation) where an Al- or Ti-based liquid mixed with ceramic particles -preferably of whisker or platelet shape - is slowly converted in air, oxygen, or nitrogen to alumina or titanium nitride, respectively [351-3531. This self-propagating reaction yields columnar crystals of the oxide or nitride phase, with B4C, Sic, or TiB2 inclusions and residual metal-filled channels, which contribute significantly to the strength and tou hness. SiC-A1203compositeshave an excellent fracture toughness of 8-1 5 MPa rn1Ig and a flexural strength of 500-800 MPa. They have also been demonstrated to be highly resistant against erosive wear [354].

Figure 111. Dislocations at the boundary (Ti,W)B,/TiB,.


919

7.8.3 Cemented Ternary Borides As discussed before, hard metal-like composites can be prepared by pressureless sinterin of ternary borides with Fe, Ni, or Co melts. Materials with z-phase (M:,M2I? B6, where MI = Fe, Ni, or Co, and M" = Zr, Hf, Nb, Ta, or W with MI as the matrix phase) have not been developed for technical use but Ni-based alloys with z are in applications as wear- and corrosion-resistant coatings on steels [355]. The z phase is also used for the improvement of the creep resistance of Ni-based superalloys. The other kind of ternary hases, cp and o,have M'M'IB and MiM"B2 stoichiometry, respectively, where M = Cr, Mo, Ta, or W and MI' = Fe, Ni, Co, or Cr and solid solutions thereof. In particular, the o-type borides have been developed extensively by Takagi et al. [129-1311 and Komai et al. [356], focusing on molybdenum rather than W or Ta due to its corrosion resistance.

P

7.8.3.1 Technically Important Systems and Structures The crystal structure of the o-type ternary borides M:M"B2was described first by Rieger et al. [357] as a distinct kind of structure similar to the U3Si2structure. The lattice is orthorhombic, space group Immm, and contains two formula units, i.e. ten atomic sites. The cell is built up by two metal prisms containing four atoms of M' and two atoms of M". These double prisms are piled-up along the b-axis. The boron atoms are arranged in pairs at the edges and in the center of the cell with their axes parallel to the b-axis. If Cr is inserted into the structure it is considered to preferentially replace Fe, Ni, or Co and thus occupies the MI1-position. Rietveld analysis of powder data by Ozaki er al. [358]proves that this kind of solid solutions creates a distortion of the orthorhombic structure, which then can be indexed according to a tetragonal lattice where the former c-axis of the orthorhombic cell becomes the tetragonal c-axis. The M' sites are now arranged in a planar quadrangle surrounded by a tetragonal-face centered MI' coordination shell. Since the edges of the MI quadrangle are not parallel to the axes of the cell, the boron pairs are also out of the b-alignment. Thus the space group P4/mbm is of comparatively low symmetry (Fig. 112). Three kinds of ternary borides with metallic binders have been exploited for future applications in detail, the WCoB-Co cermet for its high heat resistance, the Mo2NiB2-Ni cermet because of its excellent corrosion resistance, and the Mo2FeB2-Fecomposite for its wear resistance. In all cases, these borides are compatible with a melt consisting of M" and alloys thereof. According to the particular phase diagrams, which have been treated in Section 7.3.2.2 (third part), the corresponding liquid phase must, however, contain some portion of M' as well. The two phase equilibria cp- or o-liquid, respectively, are limited by a tie line with the brittle z-phase or, in systems in which (does not exist, with the particular binary M" borides, e.g., with Fe2B, Co3B, etc., which have already been treated as a problem in TiB2-Fe cermets. The other tie line interconnects the ternary borides of M',Mj' type intermetallic phases, which may even be beneficial for the hardening of the consolidated binder phase.

920


(a) Orthorhombic-MsBz: a=0.70945 b=0.45746 ~=0.31733nma=By=90'

Mo Ni

.B

(b) Tetragonal-M3B2 : a=b=0.58042 C=O.31367nm a+y=90'

Figure 112. Crystal structure of w-type ternary borides MiM"B2 with M' = Mo and M" = Ni or Cr. Depending on the Cr content, the orthorhombic structure (a) or the tetragonal structure (b) are stable (after Ozaki et al., 1994 [358]).

7.8.3.2 Liquid-Phase Sintering and Phase Reactions

Since ternary borides are commercially not readily available the preferred sintering route is not a simple liquid-phase sintering, making use of powder blends of the equilibrium compounds. Takagi [359] introduced a reaction sintering process denoted as reaction boronizing sintering. This method has several benefits, (i) the starting powders are comparatively cheap, available in fine grain size with suitable purity and can easily be handled; (ii) the process takes advantage of chemical activation, i.e. the enthalpy of formation being released for sintering; (iii) the liquid phase forms after a solid-state reaction and is therefore homogeneously distributed throughout the microstructure; and (iv) the previous reaction brings about in-situ grain size refinement and removal of oxide impurities, which enables an almost perfect wetting by the liquid phase. The process starts with a pressed powder mixture of one or more binary borides of MI-type metals and the metals required for the M" constituent of the ternary phase and for the binder phase. This method enables the fabrication of tailored MI-MIr-solid solutions in the ternary boride as well as allows the introduction of high-performance Fe- or Ni-based alloys with exceptional properties. The metal powders are usually atomized alloys but may also consist of powder blends of the particular elements of interest. In Fig. 113 [360] the reaction process is schematically summarized showing the M" alloy denoted as "metal". In principle, the first step is the boronization of the M" powder particles by boron diffusion


92 1

(1

B o r o n s o u r c e powder

0

0

0 Metal

powder

T e r n a r y complex b o r i d e Liquid

i n i t i a l stage mixed powders

solid state diffusional

l i q u i d phase s i n t e r i ng

final stntered compact

(3

Metal m a t r i x

b o r o n i z ing reaction

Figure 113. Schematic illustration of the "reaction boronizing sintering process" (Takagi, 1993) [359].

from the M' boride across the contact points. This solid-state diffusion is enhanced by the concentration gradient in boron and both M' and M" metals as well as by the chemical driving force for the formation of the ternary M;M"B2 phase. Both diffusion and heat release result in the growth of sinter necks between the starting particles and the precipitation of very small ternary boride particles being homogeneously arranged around the M" grains. The source powder for boron is used up completely by this reaction, and, upon further heating, the liquid phase is created first on sites where the eutectic composition exists. Since the ternary borides have been precipitated in situ, no oxide layer is present, preventing the wetting. During spreading of the melt particle rearrangement by capillary forces is active and causes shrinkage and a more homogeneous distribution of all phases. Due to the higher temperature and the enhanced diffusion via the liquid phase, the residual metal powder dissolves in the melt, filling the residual pores and allowing grain growth of the stable ternary borides. Finally, after cooling, the cermet consists of the hard boride phase dispersed in a metallic matrix phase or with a binder phase at the triple points, depending on the volume fraction of binder. The microstructure resembles that of ordinary hard metals based on WC-Co with some peculiarities, however, which will be discussed later. To enable the solid-state reaction at the beginning and to control wetting at higher temperatures, oxide impurities on the starting metal powders have to be removed. This can be performed by reducing agents such as carbon or by evaporation of boron oxides and suboxides during vacuum sintering. A hydrogen atmosphere can also be used for sintering. Pressureless sintering is carried out slightly above the quasi-binary eutectic temperature, usually between 1200 and 1350°C. The sintering time is comparatively short, e.g. 20-30 min to prevent grain growth by dissolution and reprecipitation. A detailed study of the sintering phenomena of a Mo2FeB2-(Fe-alloy)cermet is presented by Sivaraman et al. [361] who focused on the commercially available KH-C5O material by Toyo Kohan Co., Ltd. consisting of 42.71 mass-% Mo, 39.28% Fe, 10.20% Cr, 4.96% B (as FeB), and 2.85% Ni. Sintering was followed by DTA, TG, dilatometry, and X-ray diffraction; the microstructure was observed after quenching from particular heating steps. As a result, the sintering reactions could easily be monitored by both shrinkage effects and exo- and endothermic effects, The first solid-state reaction is the formation of Fe2B at 486°C followed by the precipitation of tetragonal Mo2FeB2, the reduction of metal oxides, the

922


Table 14. Sintering reactions of ternary boride cermets. Temp. "C

Reaction

48G500 700-1000 10o(r1100

Fe

1132

yX

Remarks, Microstructure

+ FeB Fe2B 2 Mo + 2 Fe2B --+ Mo2FeB2+ 3 Fe 4

1150

+ M,O, x M + y XO t Mo2FeBz+ Cr Mo2(Fe,Cr)B2

1150-1 180

Fe + Fe2B + Mo2(Fe,Cr)B,

1200 1225 1245

--+

4

+ Liquid

Solid-state reaction Sinter neck formation Solid state reaction, enhanced growth of sinter necks, formation of spherical particles and clusters Reduction of oxides, X = reducing agent, e.g., C Solid-state reaction, later also in liquid state; high-density clusters with rounded surfaces, initial growth of faceted grains by surface and volume diffusion Ternary eutectic, wetting, spreading, cluster fragmentation and particle rearrangement Redistribution of phases, filling of pores Grain growth of faceted particles Continuous homogenization, formation of boride framework with high contiguity

formation of the liquid phase, and the incorporation of Cr into the Mo2FeB2.Table 14 summarizes the reaction steps and the resulting intermediate microstructures. The sintering as monitored by dilatometry is illustrated in Fig. 114 [361]. Evidently, the solid-state reaction to form Fe2B as well as the initial state sintering by neck growth and particle center-approach yield a small contribution to shrinkage. The most dramatic effect is, however, the generation of a liquid phase at above 1150°C which yields a shrinkage rate of 6%/min, i.e., after 5-10min the final density of 99.6% of theoretical density is achieved. This fast sintering can be attributed to the sudden rearrangement of the boride particles of less than 1 pm in size. At temperatures exceeding 1200-1220°C only coarsening is observed by dissolution and reprecipitation mechanisms but the particle size remains around 1 pm. The change in particle size in Mo2NiB2cermets with Cr-Ni binder was investigated in more detail by Matsuo et al. [362]. They started from MOBas a boron source to compare the influence of Cr addition on structural changes. Table 15gives the composition of the boride phase aimed at whereas the binder content was kept constant at 25 mass-% (Ni f 10% Cr). Sintering at 1260-1300°C for 20min revealed that MOBwas consumed totally by the formation of Mo2NiB2, which exhibited an orthorhombic crystal structure, while the addition of Cr yielded Moz(Ni,Cr)B2with a tetragonal structure. The microstructures consist of blocky faceted crystals of somewhat elongated habit in all cases. It is, however, evident that Mo2NiB2tends to coarsen strongly (mean particle size (M 8-10 pm) compared to Moz(Ni,Cr)B2 (mean particle size (M 3-5 pm), Fig. 1 15. Another peculiarity is that the first phase forms euhedral crystals with sharp edges and corners whereas the latter phase shows rounded edges and corners under exactly the same conditions. This observation is attributed to the higher surface energy ani-

923

7.8 Cemented Boridees

0

400

800

1200

1600

temperature(%)

Figure 114. Sintering of MozFeBz with Fe,Cr,Ni-binder (after [361]).

sotropy of Cr-free orthorhombic boride by which the formation of equilibrium lowenergy facets are favored, as predicted by Warren [363]. The lower anisotropy in case of tetragonal Mo2(Ni,Cr)B2is therefore also considered the reason for the observed retardation of grain growth by Ostwald ripening [358,362]. 7.8.3.3 Microstructures and Properties A further study of Mo2NiB2-based cermets was dedicated to the influence of other transition-metal additives on the structure and habit of the ternary boride phase [365]. Taking the previously developed 35.4 mass-% Ni 6 mass-% B 58.6 mass-% Mo cermet as a starting point, 10 mass-% of Cr, V, Fe, Co, Ti, Mn, Zr, Nb, and W were added at the expenses of Ni. In another series of experiments, V was added in steps of 2.5 mass-% up to 12.5 mass-%. The mechanical properties

+

Table 15. Composition of Mo2NiBzwith Cr-Ni binder. Boride composition Mo2NiBz Moz(Ni,Cr)Bz

mass- % at.-% mass-% at.-%

B

Mo

Ni

Cr

7.9 40 8.0 40

70.5 40 71.4 40

21.6 20 10.6 9.7

0 0 10.0 10.3

+

924


Figure 115. Microstructure of Mo2NiB2and Mo2(Ni,Cr)B2with Ni and Ni-Cr binder.

obtained clearly indicate that some of the elements have no effect on strength and hardness at all, namely, Fe and Co, whereas Ti, Mn, Zr, Nb, and W yielded even lower strength than the cermet without additives. Cr and V, however, caused more than 40% increase in strength with a slightly higher hardness. The addition of V gives rise to a steep increase in both strength and hardness if only 2.5 mass-% are added (Figs. 116-1 18). It was proved that all boride solid solutions except the Cr- and V-containing one crystallize in the orthorhombic structure type. As mentioned before, the lower anisotropy of the surface energy of the tetragonal phase is the reason for the formation of more spherical particles. The inhibition of grain growth was accordingly found in Cr- and V-containing cermets but also in Nb- and W-bearing materials (Fig. 119). In the latter cases, however, the microstructure is very inhomogeneous, which may be the reason for the comparatively low strength. In case of Co and Mn addition the particles have grown considerably, which again explains the low strength since no other brittle phase was detected. If Ti and Zr are added, third phases have been observed but could not be identified. A variation of the Cr content similar to the studies of the influence of V on the properties was carried out by Komai et al. [356].Also in this case, an improvement of both strength and hardness was achieved as a function of the additive amount.


925

Figure 116. Strength of Mo2NiB2-based cermets with various additives (after [365]).

However, the strength shows a pronounced maximum at 2300 MPa at 10 mass-% and a decrease beyond the level of the cermet without Cr at 20 mass-%, while the hardness steadily increases on the order of 5% (Fig. 120). Although the ternary phases are hard but brittle, the improved commercial1 available cermets exhibit excellent toughness ranging from 18 to 30MPa m l i : and strength ( ( J = ~ 1.6-2.2 GPa). Young's modulus reaches 290-350 GPa, and the hardness 83-89 RA, depending on additives and binder content [359]. The coefficient of thermal expansion ranges between 8.5 and 1 1 x lop6K-I. The cermets

7

p

Figure 117. Hardness of Mo2NiB2-basedcermets with various additives (after [365]).

926


92

2.5

2

-

2 2.0

ln tn

. 0

2+

88

g

E I

1.5 86

1.0

-

c

a4

0

5

2.5

V

7.5

10 12.5

(mass%)

Figure 118. Transverse rupture strength (TRS) and hardness of V-containing Mo2NiB2 (after [365]).

Figure 119. Back-scattered SEM micrographs of a Ni-6 mass-%, B-58.6 mass-%, Mo-10 mass-% metal cermet. Note the change in grain size and shape due to the metal additive (after [365]).

7.8 Cemented Borides 50-2.

927

92 ~

90 88

2

-86

\

-a4

g

In [z

I

82 -

0.501 0

5

10

15 20 Cr content (mass%)

80

I 78 25

Figure 120. Transverse rupture strength and hardness as a function of C r addition to Mo2NiB, (after [356]).

of the Mo2FeB2 group exhibit an excellent corrosion resistance against organic and inorganic acids and alkalis [ 1311 being comparable to or better than those of cemented carbides and stainless steels. Furthermore, the cemented borides are resistant in various molten metals such as Zn and Al. Consequently, these kinds of materials can also be used as anti-corrosion layers on other metals [361].

7.8.4 Potentials and Applications Most of the developments in cemented borides aim at the high wear resistance due to the superior hardness compared to carbide-based hard metals. One of the most important applications are, therefore, the cutting tools where cemented boride have to compete with tungsten carbide, titanium carbonitride, and ceramic materials as well.

7.8.4.1 Cutting Tools Bearing the potential application as a wear-resistant material in mind, the mechanical properties of the new cermets were assessed by measuring key mechanical properties such as fracture toughness and hardness both at ambient and elevated temperatures. Obviously, boride-based cermets integrate both excellent hardness and a considerable fracture toughness such that their hardness-toughness profile is superior to that of commercial hard metals. Of particular importance is the fact that TiB2-Fe alloys are able to cover the high hardness regime beyond 2000 HVlO. A test program to evaluate the potential of TiB2 cermets as a cutting tool material was thus initiated by the European Community and many companies between 1990

928


Table 16. Mechanical properties of tool materials.

Tool material

Hardness HV 10

Fracture toughness [MPaJmI

Young’s modulus [GPaI

TiB2-cc(Fe)-Fe2B TiB2-y(Fe-Cr-Ni) A1203-Zr02 Ti(C,N)-Ni

1830 1800 1700 1550

9.3 9.0 4.5 8.1

490 510 380 440

and 1993. The tests were aimed at (i) screening the cutting properties of TiB2-Fe and (ii) providing useful information for further development of these materials. The properties of TiB2 cermets suggest that workpiece materials that demand tools of the lower P, M, and K qualities would be most adequate for cutting tests. Thus a medium carbon steel Ck 45, a globular cast iron GGG 50, and a hypo-eutectic cast G-AlSil2CuMgNi alloy were selected by Sigl et al. [366] for continuous turning experiments having a Brine11 hardness of <210, <220, and < 115, respectively. Two series of TiB2 cutting inserts with ferritic iron binders and austenitic steel binders (nominal composition 18 weight-% Cr, 10 weight-% Ni, bal. Fe) were tested. Due to a Ti deficiency, the ferritic TiB2 grades with a nominal binder fraction of 15 v01.-YO contained approximately 1Ovol.-% cc-Fe and 5vol.-% Fe2B. Ti was added to prevent the formation of Fe2B in the steel cutting grades, such that the second series of TiB2 cermets could be prepared with 10vol.-YOof fully austenitic binder. Uncoated ferrite-binder cermets were used to machine the A1 and cast iron alloys, whereas the Ck 45 steel was treated with both uncoated and TiNcoated (3 pm CVD-TIN layer) inserts with austenite binder. Both composites represent “generation-I” TiB2 cermets, in the sense that they feature a much coarser grain size (on the order of 5 pm) than the composites available today. The tests were carried out in comparison to conventional cutting tool materials. Thus the cast iron grade was machined with a commercial alumina-zirconia material (10 weight-% Zr02), while the carbon steel and the hypo-eutectic A1-Si alloy were treated with a commercial Ti(C,N)-Ni cermet. Important mechanical properties of the cutting tool materials are compared in Table 16. The actually tested combinations of workpiece and cutting tool materials are compiled in Table 17. Continuous turning tests were performed. The SNGN 120408 tool geometry and parameters summarized in Table 18. Cutting was interrupted after a cutting length of 500m for the GGG 50 and after l000m for the Ck 45 and G-AlSil2 alloys, Table 17. Tested combinations of workpiece and cutting tool materials.

Tool materials

G-AlSi 12

GGG 50

Ck 45

TiB2-a(Fe)-Fe2B TiB2-y( Fe-Cr-Ni) A1203-Zr02 Ti(C,N)-Ni

X

X X X

X

X

X

929


Table 18. Cutting parameters. Cutting parameters

G-AlSi 12 finishing

GGG 50 light roughing

Feed ,f [mmjrev.] Cutting depth a [mm]

0.2 1 .o

GGG5O finishing

Ck 45 finishing

0.3

0.1

1.5

0.5

0.1 2.0

respectively. The flank wear, wB, the depth of cut notch depth, 6,the surface roughness of the workpiece material, the cutting forces, and the power consumption of the machine were monitored at these intervals. The test was performed until the failure criterion or a total cutting length of 8000 m was reached. In all cutting materials the flank wear and the cutting forces increased continuously up to the ultimate lifetime. With very few exceptions of cutting edge fracture, all inserts faded due to flank wear. Failure due to excess depth-of-cut notch depth was not observed. Cratering at the rake face did occur, but was in no instance the reason for insert breakdown. The surface roughness of the workpiece materials was rated acceptable in all cases. Thus a flank wear, wB of 0.4mm was defined as lifetime criterion. The cutting results using this criterion are summarized in Figs. 121-124 for the various combinations of tool and workpiece materials. Ranking the cutting tool materials, TiB2 with a ferritic binder is superior to the Ti(C,N) cermet and the alumina-zirconia ceramic for light cutting operations such as finishing of GGG 50 and G-AlSil2. However, increased thermal loads favor cutting materials with enhanced high-temperature endurance, i.e., the Ti(C,N) cermets and the alumina-zirconia ceramics. Figures 121- 124 illustrate the details of these general observations: TiB2 cermets outperform commercial Ti(C,N) tool materials 100

.-c

E

10

1

0.1

Cutting Speed v,, d m i n Figure 121. Life-time diagram for finishing G-AlSil2.

930

7 Boride-Based Hard Materials I000

100

10

1 10

100

1000

Cutting Speed v, m/min Figure 122. Life-time diagram for finishing GGG 50.

in A1 cutting over the whole range of cutting speeds up to 700 m/min, and they work well at medium cutting speeds up to 200 m/min during finishing operations of gray cast iron. Tool life decreases dramatically, however, upon increased cutting loads, i.e., for light roughing of GGG 50 and for steel finishing operations (Figs. 123, 124). In particular, the Taylor slope of TiB2-Fe in cast iron cutting is much steeper than for alumina-zirconia tools (Figs. 122, 123). A similar, yet less pronounced difference in Taylor slopes is observed between Ti(C,N) and TiB2 steel cutting grades (Fig. 124). The TIN coating significantly improves the life of TiB2 tools, but it cannot fill the gap of the exceptional quality of Ti(C,N) cermets.

100

10

1

10

100

Cutting Speed v,, m/min Figure 123. Life-time diagram for light roughing GGG 50.

1000


10

100

93 1

1000

Cutting Speed v,, m/min Figure 124. Life-time diagram for finishing Ck 45

The following issues must be considered in a discussion of the cutting results: TiB2 is a very creep and plasticity resistant compound, particularly a t elevated temperatures [367]. (ii) The ferrite binder in the A1 and cast iron grades contains only Ti and B in solid solution and its high-temperature strength is expected to be rather limited. The austenite binder with 18% Cr and 10% Ni should perform significantly better at high temperatures, but is probably less creep resistant than Ni(Mo) binders. (iii) Coarse microstructures are known to favor erosive wear, even in very hard and tough materials. 0)

According to (iii), a generally improved cutting performance of TiB2 cermets, both at high and low temperatures, is expected from a reduction of the boride grain size. A further enhancement, specifically at high cutting loads, is anticipated from binders with improved creep resistance since both high cutting speed and increasing feed and depth of cut tend to raise the cutting tool temperatures. Thus the reduced cutting performance at rising loads is primarily attributed to the limited hightemperature strength of the ferrous binders. This conclusion is supported by the improved tool life of TiN-coated inserts because the coating reduces the friction between tool and chip and consequently keeps the temperature level of the cutting insert low. In conclusion, TiBz is an excellent hard compound for continuous cutting operations already matching or surpassing conventional tool materials in applications where limited temperature loads evolve. Specifically, ferritic binders perform well in the cutting of cast A1 alloys and cast iron. TiB2 austenite cermets do show potential for the cutting of steel, but as yet they cannot match the exceptional performance of Ti(C,N) cermets.

932


An optimization of the microstructure towards fine grain size of TiB2 and binders with sufficient high-temperature strength will further improve the cutting capacity of TiB2 cermets specifically for steel machining operations. The major disadvantage that has prevented the production of TiB2-Fe composites on a commercial scale is their delicate processing behavior. Thus the cost-performance relation is still the critical point for the current market conditions. 7.8.4.2 Wear Parts and Molding Tools The applications of cemented ternary borides aim at other industrial branches but also make use of the unique combination of hardness, toughness, and corrosion resistance. Sliding wear tests carried out by Takagi [359] in a rotating-ring-onblock arrangement have shown that the wear rate of a Mo2FeB2 (Fe-alloy) cermet is one order of magnitude less than that of WC-1OCo. The characterization of the wear debris yielded a small amount of Moo2, B203,and Fe304 at sliding velocities >0.6m/s, whereas chips produced by the WC-Co hard metal did not give evidence for any oxide formation. The higher wear rate of WC-Co is attributed to WC-particle pull-out, providing abrasive grains at the interface between the wear couples. Wear-resistant parts such as injection-molding machine parts for fabrication of fiber-reinforced polymers, metals, and ceramics, drawing dies for metallic wires, bearings, liners, and other parts for protection against sliding wear have been developed from Mo2FeB2-Feor Mo2NiB2-Nicermets (Fig. 125). They are commercially available and already widely used in Japan. Moreover, ternary borides have been successfully applied as coatings on steel for surface hardening and improvement of the corrosion resistance [360]. Thus this new class of materials has to

Figure 125. Injection molding machine parts composed of MolFeB2-based alloys: (a) cylinder; (b) auger; (c) check valve; (d) seal ring; (e) auger head.

7.9 Future Prospects and Fields of Application

933

compete with all kinds of hard metals and cermets with respect to performance and costs.

7.9 Future Prospects and Fields of Application At the time this article was written, optimism concerning the application of highperformance advanced ceramics was still considerably low after the boom between 1980 and 1990, especially in respect to structural parts for automotive engines or substitutes for other high-strength, high-toughness materials. This tendency was and is not restricted to Europe or the USA., but is also valid for Japan and can easily be attributed to the following problems: (i) difficulties related to reliable large-scale production at reasonable cost and quality (problems of “zero-flaw” processing and part testing), (ii) difficulties in a material-appropriate design of parts (problems of substitution and new construction taking the relative brittleness of ceramics into account), (iii) the still comparatively high price of high-quality starting materials and manufacturing processes (problems of affordable highpurity and fine grain size), and (iv) the current worldwide economic recession in production stopping any activities in materials research and development. Considering these facts, the following paragraph will give some basic ideas on the actual and most probable applications, which have already been demonstrated up to now to be feasible. Besides the well-known application of grits (e.g., as grinding, lapping, and polishing particles, as oxidation inhibitors in refractories, and for alloying of special metals), boride ceramics are already widely used in mechanical, chemical, thermal, and electrical areas. Most of the applications of borides related to mechanical aspects make use of their extraordinarily high hardness and wear resistance. Also, their high Young’s modulus (stiffness) and excellent high-temperature behavior are very attractive properties. The high-temperature hardness of the most important ceramics and hard materials is shown in Fig. 126. It is obvious that, besides carbides, borides are superior to most oxide ceramics. At high temperatures, however, the metallic transition-metal borides suffer from a strong decrease in hardness due to enhanced plastic deformability (thermal initiation of slip systems), which is basically due to their comparatively primitive crystal structure and their high metallic portion in bonding. Additionally, oxidation has to be taken into account at temperatures above 600°C. The poor corrosion resistance of borides may be partially overcome in composites with S i c if the microstructure allows passive oxidation kinetics. In combination with S i c the borides retain their high electrical and thermal conductivity and thus suitable thermal shock resistance. SiC-TiB2 composites have been extensively developed for wear parts in machinery such as sliding rings, valves, valve seats, roller and ball bearings, plungers, and rocker arm pads. Besides the well-known cemented carbides and cermet cutting tools, boron carbide- and titanium diboride-containing cermets have been proven suitable for

934


8000

6000

-

N

E E m

\

-.Y

2 1000 Ln

In a,

C

? I

2000

0

0

LOO

800 Temperature 1°C)

1200

Figure 126. High-temperature hardness of ceramic materials in comparison to diamond and cubic boron nitride.

machining aluminum alloys, whereas ternary boride cermets have been successfully developed for injection molding nozzles, bearings, wire drawing cones, and other wear resistant parts that are not exposed to high temperatures. Composites of TiB2/TiC, TiB2/TiN, TiB2/ZrB2/TaN, as well as B4C/TiB2, B4C/SiC, and B4C/ SiC/TiB2 have been exploited as cutting tools for brass, bronze, and A1 alloys, drilling tools for rocks, and concrete due to their comparatively high toughness. An obvious conclusion from the various efforts in strengthening and toughening of boride-based materials is that the maximum strength obtained by crackpropagation influencing mechanisms operational in composites is about 11.2 GPa and the fracture toughness usually obtained in optimized microstructures is about 6-7 MPa m1I2.It is also clear that simple two-phase composites are generally superior to materials of more complex composition. Metal-matrix composites may exhibit a much higher toughness than ceramics but are limited in hightemperature use due to their sensitivity towards creep and corrosion. With the possible exception of the very promising TiB2-Fe-based and ternary boride-based cermets, borides are thus currently of more interest for strengthening metals similar to oxide-dispersed superalloys. A certain challenge is still the transformation toughening of borides by Y-stabilized Zr02, increasing the toughness by a factor of two

7.9 Future Prospects and Fields of Application

935

compared to simple particulate reinforcement. In competition with oxide and nitride ceramics the borides have thus been developed to a similar excellent mechanical performance but still suffer from their considerably high price and poor oxidation resistance at high temperatures. Generally speaking, the high reactivity of the boron compounds is the most important, and sometime the most limiting factor. Although the wear behavior is very promising due to the high hardness and the high thermal conductivity, boron compounds, except for some transition-metal diborides, cannot be used in contact with iron-based alloys but for wear contacts with aluminum and, in some cases, nickel-based materials. On the other hand, reaction sintering can be widely employed for the fabrication of unique microstructures, which has not yet been exploited entirely. Although mechanical applications are currently by far the most important aims for the development of materials and parts based on carbides and borides, this may easily change to the benefit of thermal and electrical applications if there is an improvement in the understanding and control of the semiconductivity of B4C and a breakthrough in the manufacture of suitable parts. Moreover, thermal applications may increase considerably in importance as more efficient energy conversion or recovery becomes a significant requirement. Excellent opportunities for applications of transition-metal borides such as TiB, are thus more a result of their high electrical conductivity and resistance against metallic melts combined with an excellent wetting behavior. Well-known examples are TiB,-AlN-BN evaporator dies (e.g., for the evaporation of Al, Cd, Ge, Pb, Bi, Cr, Cu, Ag, Au, Mg, Fe, Zn, and Sn for the coating of plastics, paper, and parts) and electrodes for the aluminum electrolysis in Hall-Herould cells. In the latter case, there is a very high demand for such a material having a high electric conductivity and extraordinary corrosion resistance against kryolithe melts, but some undesired grain boundary reactions (infiltration of the melt along impurity-loaded grain boundaries) and the poor thermal shock resistance (strongly anisotropic thermal expansion) are the limiting factors for a successful replacement of graphite electrodes. Other diborides, such as ZrB,, have been studied for use as structural parts, such as bearings, nozzles, and molds for injection molding, valves, and sealings, but further development is still necessary for an industrial breakthrough. For example, ZrB2-MoSi2 composites exhibit remarkable corrosion resistance up to 2000"c. Functional applications of borides are also related to the high cross-section of boron for the absorption or retardation of thermal neutrons. B4C sintered products have been used as moderating elements and radiation-protecting components in nuclear reactors of several designs. Of course, this market is declining with the decreasing political support for nuclear power and nuclear research. In conclusion, boride ceramics are materials of high potential for use in mechanical and electric fields. Due to the stagnating materials markets, however, most research on borides is focused on the basic questions of atomic bonding and physical transport properties rather than on application-related problems. Only a few highly specialized research groups are active worldwide and promote the science rather than the technology of borides. Depending upon the economic situation and the

936


demand for future key technologies, there is a chance that these materials may increase in importance.

References 1. 2. 3. 4.

5. 6. 7.

8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37.

Kiessling, R. Acta Chem. Scand. 1950, 4, 209. Lundstrom, T. Arkiv Kemi 1969, 31(19), 227. Anderson, S. and Lundstrom, T. Acta Chem. Scand. 1968, 12, 3103. Vlasse, M., Slack, G. H., Garbauskas, M., Kasper, J. S., and Viala, J. C. J . Solid State Chem. 1986, 63, 31. Richards, S. M. and Kasper, J. S. Acta Crystallogr. 1969, B25, 237. Higashi, I., Kobayashi, K., Tanaka, T., and Ishizawa, Y. J. Solid Stute Chem. 1997, 133, 16. Matkovich, V. I. and Economy, J. in Boron and Refractory Borides, Matkovich, V. I. (Ed.), Springer, Berlin, 1977, pp. 77-94. Aronsson, B., Lundstrom, T., and Rundqvist, S. Borides, Silicides and Phosphides: A Critical Review of Their Preparation, Properties and Crystal Chemistry, John Wiley & Sons Inc., New York, 1965. Aronsson, B., Lundstrom, T., and Engstrom, I. in Anisotropy in Single Crystal Refractory Compounds, Plenum Press, New York, 1968, Vol. 1, pp. 3-22. Lundstrom, T. Arkiv Kemi 1969,31(19), 227. Higashi, I. and Takahashi, Y. J . Less-Common Met. 1986, 123, 277. Okada, S., Atoda, T., Higashi, I., and Takahashi, Y., J . Mater. Sci. 1987, 22, 2993. Lundstrom, T. Arkiv Kemi 1968, 30(11), 115. Romans, P. A. and Krug, M. P. Acta Crystallogr. 1966, 20, 313. Woods, H. P., Wawner, F. E., and Fox, B. G. Science 1966, 151, 75. Armstrong, D. R. in Proc. 9th Int. Symp. on Boron, Borides, and Related Compounds, Werheit, H. (Ed.), University Press, Duisburg, 1987, pp. 125-131. Samsonov, G. V. and Kovenskaya, B. A. in Boron and Refractory Borides, Matkovich, V. I. (Ed.), Springer, Berlin, 1977, pp. 5-18. Samsonov, G. V. and Kovenskaya, B. A. in Boron and Refractory Borides, Matkovich, V. I. (Ed.), Springer, Berlin, 1977, pp. 19-30. Aravamudham, R. Z. Metallkd. 1967, 58, 179. Samsonov, G. V., Goryachev, Yu. M., and Kovenskaya, B. A. Izv. Vuz Fiz. 1972, 6 , 37. Hoard, J. L. and Hughes, R. E. in The Chemistry of Boron and Its Compounds, Muetterties, E. L. (Ed.), Wiley, New York, 1967, pp. 25-154. Spear, K. E. J. Less-Common Met. 1976, 47, 195. Hayami, W., Souda, R., Aizawa, T., and Tanaka, T. Surf. Sci. 1998,415, 433. Souda, R., Asari, E., Kawanowa, H., Otani, S., and Tanaka, T. Surf. Sci. 1998,414(1-2), 77. La Placa, S. and Post, B., Planseeberichtefiir Pulvermetallurgie 1961, 9, 109. Lipp, A. Ber. Deutsche. Keram. Ges. 1966, 43(1), 60. Lipp, A. and Roder, M. Z . Anorg. Allg. Chem. 1966, 343, 1. Lipp, A. and Roder, M. Z . Anorg. Allg. Chem. 1966, 344, 225. Matkovich, V. I. and Economy, J. in Boron and Refractory Borides, Matkovich, V. I. (Ed.), Springer, Berlin, 1977, pp. 98-106. Clark, H. K. and Hoard, J. L. J. Am. Ceram. Soc. 1943,65, 2115. Allen, R. D. J. Am. Chem. Soc. 1953,753582. Neidhard, H., Mattes, R., and Becher, H. J. Acta Crystallogr. 1970, 26B, 315. Ploog, K. J . Less-Common Met. 1974, 25, 115. Will, G. and Kossobutski, K. H. J . Less-Common Met. 1976, 47, 43. Kirfel, A., Gupta, A,, and Will, G. Acta Crystallogr. 1979, B35, 1052. Werheit, H. and de Groot, K. Phys. Status Solidi 1980, 97, 229. Conard, J., Bouchacourt, M., and Thevenot, F. J . Less-Common Met. 1986, 117, 51.

References

937

38. Tallant, D. R., Aselage, T. L., and Campbell, A. N. Phys. Rev. B 1989, 40, 5649. 39. Aselage, T. L., Emin, D., and Wood, C. in Trans. 6th Symp. Space Nuclear Power, Albuquerque, NM, January 8-12, 1989, pp. 430433. 40. Aselage, T. L., Tallant, D. R., Gieske, J. H., Van Deusen, S. B., and Tissot, R. G. in The Physics and Chemistry of Carbides, Nitrides, and Borides, Freer, R. (Ed.), NATO AS1 Ser. E, Vol. 185, Kluwer, Dordrecht, The Netherlands, 1990, pp. 97-1 12. 41. Aselage, T. L. and Emin, D. in Boron-Rich Solids, Proc. 10th Int. Symp. on Boron, Borides, and Related Compounds, Emin, D., Aselage, T. L., Switendick, A. C., Morosin, B., and Beckel, C. L. (Eds), AIP Conf. Proc. 231, American Institute of Physics, New York, 1991, pp. 177-185. 42. Morosin, B., Aselage, T. L., and Emin, D. in Boron-Rich Solids, Proc. 10th Int. Symp. on Boron, Borides, and Related Compounds, Emin, D., Aselage, T. L., Switendick, A. C., Morosin, B., and Beckel, C. L (Eds), AIP Conf. Proc. 231, American Institute of Physics, New York, 1991, pp. 193-196. 43. Kleinman, L. in Boron-Rich Solids, Proc. 10th Int. Symp. on Boron, Borides, and Related Compounds, Emin, D., Aselage, T. L., Switendick, A. C., Morosin, B., and Beckel, C. L. (Eds), AIP Conf. Proc. 231, American Institute of Physics, New York, 1991, pp. 13-20. 44. Armstrong, D. R., Bolland, J. and Perkins, P. G. Acta Crystallogr. 1983, 39, 324. 45. Emin, D. in Boron-Rich Solids, Emin, D., Aselage, T, Beckel, C. L., Howard, I. A,, and Wood, C. (Eds), AIP Conf. Proc, American Institute of Physics, New York, 1986, pp. 189-205. 46. Silver, A. H., Bray, P. J. J . Chem. Phys. 1959, 31, 247. 47. Hynes, T. V., Alexander, M. N. J . Chem. Phys. 1974,54, 5296. 48. Becher, H. J., Thevenot, F. Z. Anorg. Allg. Chem. 1974, 410, 274. 49. (a) Bouchacourt, M. and Thevenot, F. J . Less-Common Met. 1981,82,219. (b) Bouchacourt, M. and Thevenot, F. J . Less-Common Met. 1981, 82, 227. 50. Schwetz, K . A. and Karduck, P., J . Less-Common Metals 1991, 175, I . 51. Kuhlmann, U., Werheit, H., and Schwetz, K. A. J . Alloj's Compd. 1992, 189, 249. 52. Lim, S.-K. and Lukas, H.-L., in Hochleistungskeramiken Herstellung, AU!fbau,Eigenschaften, Petzow, G., Tobolski, J., and Telle, R. (Eds), VCH, Weinheim, and Deutsche Forschungsgemeinschaft, 1996, pp. 605 -616. 53. Kasper, B. Dr. rer. nut. Thesis, University of Stuttgart, 1996. 54. Telle, R. in The Physics and Chemistry of Carbides. Nitrides, and Borides, Proc. Freer, R. (Ed.), NATO AS1 Ser. E, Vol. 185, Kluwer, Dordrecht, The Netherlands, 1990, pp. 249-268. 55. Meerson, G. A., Kiparisov, S. S., and Gurevich, M. A. Sov. Powder Metall. Met. Ceram. 1966, 5, 223. 56. Magnusson, B. and Brosset, C. Acta Chem. Scand. 1962, 16,449. 57. Hansen, M. Constitution of Binary Alloys, McGraw-Hill, New York, 1958. 58. Elliott, R. P. Constitution qf Binary Alloys, 1st suppl., McGraw-Hill, New York, 1965. 59. Shunk, F. A. Constitution of Binary Alloys, 2nd suppl., Mc-Graw-Hill, New York, 1969. 60. Mofatt, W. G. The Handbook of Binary Phase Diagrams, General Electric Co., Schenectady, NY, 1976. 61. Mofatt, W. G. The Index to the Binary Phase Collections, General Electric Co., Schenectady, NY, 1979. 62. Binary Alloy Phase Diagrams, Massalski, T. B. (Ed.), ASM Int., Materials Park, OH, 1990, Vol. 1: Alloys, Vol. 2: Phase Diagrams. 63. Ternary Alloys, Petzow, G., Effenberg, G. (Eds), VCH, Weinheim, 1988ff, Vols. 1 ff. 64. Spear, K. E. in Application of Phase Diagrams in Metallurgy and Ceramics, National Bureau of Standards, Gaithersburgh, 1977, Special Publication SP-496, pp. 744-762. 65. Lundstrom, T. in Boron undRefractory Borides, Matkovich, V. I. (Ed.), Springer, Berlin, 1977, pp. 351-376. 66. Spear, K. E. in Phase Diagrams, Materials Science and Technology, Vol. IV, Alper, A. M. (Ed.), Academic Press, New York, 1976, Ch. 1 I , pp. 91-159. 67. Guillermet, A. F. and Grimvall, G. Phys. Rev. B 1989, 40, 1521. 68. Grimvall, G. and Guillermet, A. F. Boron-Rich Solids, Proc. 10th Int. Symp. on Boron, Borides, and Related Compounds, Emin, D., Aselage, T. L., Switendick, A. C., Morosin, B., and Beckel, C. L. (Eds), AIP Conf. Proc. 231, American Institute of Physics, New York, 1991, pp. 423-430. -

938


69. Samsonov, G. V. and Schuravlev, N. N. Fiz. Met. Metullovad. Akad. Nauk, S S S R Ural Filial 3 1956, 109. 70. Schuravlev, N. N. and Makarenko, G. N. Ed. Acad. Sci. USSR,OTN, Metallurgia i Toplivo I , 1961, 133. 71. Bouchacourt, M. and Thevenot, F. J. Less-Common Met. 1979,67, 327. 72. Beauvy, M. in Conf. Abstract, 8th Int. Symp. on Boron, Borides, Carbides and Related Compounds, Tsagareishvili, G. V. (Ed.), Acad. Sci. Georg. S.S.R., Tbilisi, 1984, p. 25. 73. Lukas, H. L. in Constitution of Ternary Alloys 3, Petzow, G. and Effenberg, G. (Eds), VCH, Weinheim, 1990, pp. 140-146. 74. Ekbom, L. B. Sci. Ceram. 1977, 9, 183. 75. Lange, D. and Holleck, H., Proc. 11th Int. Plansee-Conf., Planseewerke, Reutte, 1985, HM 50, 741. 76. Ettmayer, P., Horn, H. C., and Schwetz, K. A. Microchim. Acta, 1970, Suppl. IV, 87. 77. Lugscheider, E., Reimann, H., and Quaddakers, W.-J., Ber. Deutsche Keram. Ges. 1979, 65(10), 301. 78. Dorner, P. Dr. rer. nut. Thesis, University of Stuttgart, 1982. 79. Emin, D. in The Physics and Chemistry of Carbides, Nitrides and Borides, Freer, R. (Ed.), Kluwer, Dordrecht, The Netherlands, 1990, pp. 691-704. 80. Olesinski, R. W. and Abbaschian, G. J. Bull. Alloy Phase Diagrams 1984, 5, 478. 81. Mitra, I., Biittner, D., Telle, R., Babushkin, O., and Lindback, T. in Proc. 8th CIMTEC, Ceramics Charting the Future, Advances in Science and Technology, Vol. 3B, Vincenzini, P. (Ed.), Techna Srl., Milano, 1995, pp. 787-792. 82. Armas, B., Male, G., and Salanoubat, D. J . Less-Common Metals 1981, 82, 245. 83. Telle, R. Dr. rer. nut. Thesis, University of Stuttgart, 1985. 84. Murray, J. L., Liao, P. K., and Spear, K. E. Bull. Alloy Phase Diagrams 1986, 7, 550. 85. Rudy, E. and Windisch, S. in Ternary Phase Equilibria in Transition Metal-Boron-CarbonSilicon Systems, Part I, Vol. VII, Technical Report No. AFML-TR-65-2, Wright Patterson Air Force Base, OH, 1966. 86. Fenish, R. G . Trans. A I M E 1966, 236, 804. 87. Neronov, V. A,, Korchagin, M. A,, and Aleksandrov, V. V. J . Less-Common Met. 1981, 82, 125. 88. Spear, K. E., McDowell, P., and McMahon, F. J . Am. Ceram. Soc. 1986,69, C4. 89. Batzner, C. Dr.rer.nat. Thesis, University of Stuttgart, 1994. 90. Pyzik, A. J. and Beaman, D. R. J . Am. Ceram. Soc. 1995, 78, 305. 91. Kieffer, R., Gugel, E., Leimer, G., and Ettmayer, P. Ber. Deutsche. Keram. Ges. 1972,49(2), 44. 92. Mogilevsky, P., Gutmanas, E. Y., Gotman, I., and Telle, R. J. Eur. Ceram. Soc. 1995, 15, 527. 93. Lukas, H. L. Final Report, Deutsche Forschungsgemeinschaft Project, Contract No. Lu 283/ 2-2, 1992. 94. Telle, R. and Petzow, G. in High Tech Ceramics, Vinzencini, P. (Ed.), Material Science Monographs, Elsevier, Amsterdam, 1987, pp. 961-973. 95. Werheit, H., Kuhlmann, U., Laux, M., and Telle, R. J . Alloys and Compounds 1994, 209, 181. 96. Secrist, D. R. J. Am. Ceram. Soc. 1964, 47, 127. 97. Shaffer, P. T. B. Muter. Res. Soc. Bull. 1969, 4, 213. 98. Nishiyama, K., Mitra, I., Momozawa, N., Watanabe, T., Abe, M., and Telle, R. J . Jpn. Res. Inst. Muter. Technol. 1997, 15, 292. 99. Rudy, E. and Windisch, S. Ternary Phase Equilibria in Transition Metal-Boron-Carbon-Silicon Systems, Part 11. Related Tenary System Vol. VIII, TI-B-C System, Technical Report No. AFML-TR-65-2, Wright Patterson Air Force Base, OH 1966. 100. Nowotny, H., Benesovsky, F., and Brukl, C. Monatsh. Chem. 1961,92,403. 101. Ordan’yan, S. S. and Unrod, V. V. Sov. Powder Metall. Met. Ceram. 1975, 14, 729. 102. Holleck, H. Planseeber. Pulvermetall. 1982, 2, 849. 103. Holleck, H., Leiste, H., and Schneider, W. in High Tech Ceramics, Proc. 6th CIMTEC, Vincenzini, P. (Ed.), Elsevier, Amsterdam, 1987, pp. 2609-2622. 104. Rudy, E. Ternary Phase Equilibria in Transition Metal-Boron-Carbon-Silicon Systems, Vol. V, Compendium of Diagram Data, US Atomic Energy Comm. Publ. AFML-TR-65-2, Wright Patterson Air Force Base, OH, 1969.

References

939

Duschanek, H. and Rogl, P., Lukas H. L. J . Phase Equilibria 1995, 16, 46. Brodkin, D. and Barsoum, M.W. J . Am. Ceram. Soc. 1996,79, 785. Telle, R., Brook, R. J., and Petzow, G. J . Hard Muter. 1991, 2, 79. Taffner, U., Telle, R., and SchHfer, U. Z. Prukt. Metallogr. 1988, 21, 17. Klingler, H., Franz, E.-D., and Telle, R. Spreclisaal 1995, 128, 11. Brukl, C. E. Ternary Phase Equilibria in Transition Metal-Boron-Carbon-Silicon Systems Part 11; Ternary Systems, Wright, Patterson Air Force Base, OH, 1965, Vol. VII, Tech. Rep. No. AFML-TR-65-2. 11 1. Borisova, A. L., Borisov, Yu. S., Polyanin, B. A., Shvedova, L. K., Kalinovskii, V. R., and Gorbatov, I. N. Sov. Powder Metall. Met. Ceram. 1986, 25, 769. 112. Holleck, H. in Proc. 12th Planseeseminar ofPoivder Metallurgy 3, Bildstein, H., Ortner, H. M. (Eds), Planseewerke, Reutte, 1989. pp. C3, 1-12. 113. Touanen, M., Teyssandier, F., and Ducarroir, M. in Proc. 7th Eur. Con6 on Chemical Vupour Deposition, Ducarroir, M., Bernard, C., and Vandenbulcke, L. (Eds), J . Phys. 1989, 50, 105. 114. Wakelkamp, W. J. J., van Loo, F. J. J., and Metselaar, R. J . Eur. Ceram. Soc. 1991, 8, 135. 115. Yuriditsky, B. Y. Refractory Muter. Hard Muter. 1990, 3, 32. 116. Sigl, L. S. and Schwetz, K. A. in Boron-Rich Solids, Proc. 10th Int. Symp. Boron, Borides, and Related Compounds, Emin, D., Aselage, T L., Switendick, A. C, Morosin, B., and Beekel, C. L. (Eds), AIP Conf. Proc. 231, American Institute of Physics, New York, 1991, pp. 468472. 117. Ghetta, V., Gayraud, N., and Eustathopoulos, N. Solid State Phenomena 1992, 25/26, 105. 118. Jiingling, T., Oberacker, R., Thiimmler, F., Sigl, L. S., and Schwetz, K. A. Powder Metall. Int. 1991, 23(5), 296. 119. Sigl, L. S. and Jiingling, T. J . Hard Muter. 1992, 3, 39. 120. Federov. T. F. and Kuzma, Y. B. Izvest. Akad. Nuuk S S S R - Neorg. Muter. 1967, 3, 1489. 121. Shurin, A. K. and Panarin, V. E. Izvest. Akad. Nauk. S S S R Metally 1974. 4, 235. 122. Smid, I. and Kny, E. I n f . J . Refractory Met. Hard Muter. 1988, 8, 135. 123. Ottavi, L., Chaix, J. M., Allibert, C., and Pastor, H. Solid State Phenomena 1992,25/26, 543. 124. Ottavi, L., Saint-Jours, C., Valignant, N., and Allibert, C. Z. Metullkd. 1992, 83, 80. Tech. 1965, 19, 715. 125. Schobel, J. D. and Stadelmaier, H. H. M~~tallwissenschuft 126. Lugscheider, E., Reimann, H., and Pdnkert, R. Z. Metullkd 1980, 71, 654. 127. Lugscheider, E., Reimann, H., and Pankert, E. Metall. 1982, 36, 247. 128. Haschke, H., Nowotny, H., and Benesovsky, F. Monatsh. Chem. 1966, 97, 1459. 129. Takagi, K., Ohira, S., and Ide, T. in Modern Developments in Powder Metallurgy, Vol. 16: Ferrous and Non-ferrous Materials, Aqua, E. N., and Whitman, C.I. (Eds), MPI & APMI Publ., Princeton, NJ, pp. 153-166. 130. Takagi, K. et al. Int. J . Powder Met. 1987, 23, 157. 131. Takagi, K., Ohira, S., Ide, T., Watanabe, T., and Kondo, Y. Met. Powder Rep. 1987,42,483. 132. Post, B., Glaser, F. W, and Moskowitz. D. Acta Mefall. 1954, 2, 20. 133. Kuz’ma, Y. B., Telegus, V. S., and Kovdlyk, D. A. Poroshkovuya Mefallurgiya 1969,5(77), 79; Sov. Powder Metall. Met. Ceram. 1969, 4, 403. 134. Pastor, H. in Boron and Refractorj, Borides, Matkovich, V.I. (Ed.), Springer, Berlin, 1977, pp. 457493. 135. Zdaniewski, W. A. J . Am. Ceram. Soc. 1987, 70, 793. 136. Telle, R., Fendler, E., and Petzow, G. J . Hard Muter. 1992, 3, 21 1. 137. Makarenko, G. N. in Boron and Rcrf).actoryBorides, Matkovich, V. I. (Ed.), Springer, Berlin, 1977, pp. 310-330. 138. Kuz’ma. Y . B., Telegus, V. S., and Marko, M. A. Poroshkovaya Metallurgiya 1972,4(66), 215 139. Koval’chenko, M. S., Ochkas, L. F., and Vinokurov, V. B. J . Less Common M e t . 1979,67,297. 140. Kliinenko, W. N. and Shunkowski, G. L. Planseeher. Pulverrnetall. 1981, 2, 405. 141. Okada, S., Satoh, M., and Agata. T. Nippon Kagaku Kuishi 1985, 685. 142. Liao, P. K. and Spear, K. E. Binar?, Alloy Phase Diagrams Vol. I , ASM Internat., Metals Park, OH, 2nd Ed. 1990,471474. 143. Telegus, V. S. and Kuz’ma, Y. B. J . Sov. Powder Metallurgy and M e t . Ceram., 1968, 2, 133. 144. Rudy, E. and Windisch, S. Ternary Phase Equilibria in Transition Metal-Boron-CarbonSilicon Systems, Part I, Related Binary Systems, Technical Report No. AFML-TR-65-2, Wright Patterson Air Force Base. OH, 1966. 105. 106. 107. 108. 109. 1 10.

-

940


145. Yasinskaya, G. A. and Groisberg, M. S. Sov. Powder Metall. Met. Ceram. 1963, 2, 457. 146. Kuz’ma, Yu. B., Svarichevskaya, S. I., and Telegus, V. S. Soviet Powder Metall. Met. Ceram. 1971, 10,478. 147. Kosterova, N. V. and Ordanyan, S. S. Zvest. Akad. Nauk U S S R Neorg. Muter. 1977, 13, 1411. 148. Ariel, E, Barta, J. and Niedzwiedz, S. J. Less-Common Met. 1970, 29, 199. 149. Ahn, D.-G., Kawasaki, A,, and Watanabe, R. Muter. Trans. JIM 1996, 37, 1078. 150. Pohl, A., Kitzler, P., Telle, R., and Aldinger, F. Z . Metallkunde 1994, 85, 658. 151. Pohl, A., Telle, R., and Petzow, G. 2. Metallkunde 1995, 86, 148. 152. Mitra, I. and Telle, R. in Werkstoflwoche 1996: Symposium 7: Materialwissenschaftliche Grundlagen, Aldinger, F. and Mughrabi, H. (Eds), DGB Informationsgesellschaft, Oberursel, 1996, pp. 521-525. 153. Mitra, I. and Telle, R. J. Solid State Chem. 1997, 133, 25. 154. Schmalzried, C. and Telle, R. in Proc. Werkstoffwoche 98:\Band 7: KerumiklSimulation Keramik, Heinrich, J., Ziegler, G., Hermel, W., and Riedel H. (Eds), Wiley-VCH, Weinheim, 1999, pp. 377-382. 155. Telle, R. in Materials Science and Technology, Cahn, R. W., Haasen, P., and Kramer, E. J. (Eds), Verlag Chemie, Weinheim, 1993, Vol. 11, Ch. 4, pp. 175-266. 156. Joly, A. C. R. Acad. Sci. 1883, 97, 456. 157. Moisson, H., C. R. Acad. Sci. 1894, 118, 556. 158. Ridgeway, R. R. Trans. Electrochem. Soc. 1934,66,117; Norton Co. US Patent I8972I4, 1933. 159. Lipp, A. Tech. Rundschau 1965, 57(14), 5; 57(28), 19; 57(33), 5. 160. Lipp, A. Tech. Rundschau 1966, 58(7), 3. 161. Dufek, G., Wruss, W., Vendl, A,, and Kieffer, R. Planseeber. Pulvermetall. 1976, 24, 280. 162. Gray, E. G. Eur. Patent Appl. 1 IS2 428, 1980. 163. Schwetz, K. A. and Lipp, A. in Ullmanns Encyclopedia of Industrial Chemistry, VCH, Weinheim, 1985, A4, pp. 295-307. 164. Thevenot, F. J . Eur. Ceram. Soc. 1990,6, 205. 165. Thevenot, F. in The Physics and Chemistry of Carbides, Nitrides, and Borides, Freer, R. (Ed.), NATO AS1 Series E, Vol. 185, Kluwer, Dordrecht, The Netherlands, 1990, pp. 87-96. 166. McKinnon, I. M. and Reuben, B. G. J . Electrochem. Soc. 1975, 122, 806. 167. Knudsen, A. K. Adv. Ceram. 1987, 21, 237. 168. Walker, B. E., Rice, R. W., Becher, P. F., Bender, B. A,, and Coblenz, W. S. Ceram. Bull. 1983, 62, 916. 169. DeHoff, R. T., Rummel, R. A., LaBuff, H. P., and Rhines, F. N. in Modern Development in Powder Metallurgy, Hausner, H. H. (Ed.), Plenum, New York, 1966, pp. 31&330. 170. Greskovich, C. and Rosolowski, J. H. J. Am. Ceram. Soc. 1976, 59, 336. 171. Prochazka, S. WhyIs It DifJult to Sinter Covalent Substances? General Electric Corp. Res. and Dev. Center Technical Information Series, Report No. 89CRD025, Schenectady, NY, 1989. 172. Matje, P. and Schwetz, K. A. in Proc. 2nd Int. Conj: on Ceramic Powder Processing Science, Hausner, H., Messing, G. L., and Hirano, S. (Eds), Deutsche Keramische Gesellschaft, Koln, 1989, pp. 377-384. 173. Shaw, N. J. Powder Met. Int. 1989, 21, 16; 31. 174. Grabchuk, B. L. and Kislyi, P. S. Sov. Powder Metall. Met. Ceram. 1976, 15, 675. 175. Adlassnig, K. Plunseeber. Pulvermetall. 1958, 6, 92. 176. Grabchuk, B. L. and Kislyi, P. S. Poroshkovaya Metallurgiya 1974, 8, 11. 177. Katz, J. D., Blake, R. D., and Petrovich, J. Met. Powder Rep. 1988, 43, 835. 178. Dole, S. L. and Prochazka, S. in Ceramic Engineering and Science Proc. 6(7/8), Smothers, W. J. (Ed.), American Ceramic Society, Westerville, OH, 1985, pp. 1151-1160. 179. Kislyi, P. S. and Grabchuk, B. L. in Proc. 4th Eur. Symp. Powder Metallurgy, SOC.Franqaise Metall., Grenoble, 1975, p. 10-2-1. 180. Kriegesmann, J. Ger. Patent DE 37 I I 871 C2, 1989. 181. Lange, R. G., Munir, Z. A., and Holt, J. B. Muter. Sci. Res. 1980, 13, 311. 182. Kanno, Y., Kawase, K., and Nakano, K. J. Ceram. Soc. Jpn. 1987, 95, 1137. 183. Glasson, D. R. and Jones, J. A. J. Appl. Chem. 1969, 19(5), 125. 184. Janes, S. and Nixdorf, J. Ber. Deutsche. Keram. Ges. 1966, 43, 136. -

References

941

Stibbs, D., Brown, C. G., and Thompson, R. US Patent 3 749 571, 1973. Zakhariev, Z. and Radev, D. J . Muter. Sci. Lett. 1988, 7, 695. Prochazka, S. U S . Patent 4005235, 1977. Kriegesmann, J. in Teclznische Keramische Werkstoffe, 4th Suppl., Kriegesmann, J. (Ed.), Deutscher Wirtschaftsdienst, Koln, 1991. 189. Schwetz, K. A. and Vogt, G. Ger. Patent 2 751 998, 1977; U S Patent 4 195 066, 1980. 190. Henney, J. W. and Jones, J. W. S. Br. Patent Application 20141934, 1978. 191. Suzuki, H. and Hase, T. in Proc. Conf. Factors in Densification of Oxide and Nonoxide Ceramics, Somiya, S. and Saito, S. (Eds), Japan, 1979, p. 345. 192. Schwetz, K. A. and Grellner, W. J . Less-Common Met. 1981, 82, 37. 193. Dole, S. L., Prochazka, S., and Doremus, R. H. J . Am. Ceram. Soc. 1989, 72, 958. 194. Bougoin. M., Thevenot, F., Dubois, F., and Fantozzi, G. J . Less-Common Met. 1985, 114, 257. 195. Matje, P., internal report, Elektroschmelzwerk Kempten GmbH, Germany, 1990. 196. Sigl, L. S. and Schwetz, K. A. Powder Metall. Int. 1991, 23, 221. 197. Sigl, L. S. and Schwetz, K. A. Euro-Ceramics II 1991, 1, 517. 198. Grabchuk, B. L. and Kislyi, P. S. Poroshkovaya Metallurgiya 1975, 7(151), 27. Engl. Transl.: Sov. Powder Metall. Met. Ceram. 1975, 14, 538. 199. Oh, J. H., Orr, K. K., Lee, C. K., Kim, D. K., and Lee, J. K. J . Korean Ceram. Soc. 1985,22, 60. 200. Weaver, G. Q. U S Patent 4320204, 1982. 201. Weaver, G. Q. UK Patent GB 2093481 A , 1982. 202. Borchert, W. and Kerler, A. R. Metall 1975, 29, 993. 203. Schwetz, K. A,, Reinmuth, K., and Lipp, A. Sprechsaal 1983, 116, 1063. 204. Lorcher, R., Strecker, K., Riedel, R., Telle, R., and Petzow, G. in Solid State Phenomena 8&9, Proc. Int. Conf: Sintering of Multiphase Metal and Ceramic Systems, Upadhyaya, G . S. (Ed.), Sci-Tech Publications, Vaduz, Gower Publ. Co., Brookfield, VT, 1990, pp. 479492. 205. Bougoin, M. and Thevenot, F. J . Muter. Sci. 1987, 22, 109. 206. Halverson, D. C., Pyzik, A. J., Aksay, I. A,, and Snowden, W. E. J . Am. Ceram. Soc. 1989,72, 775. 207. Sigl, L. J . Europ. Ceram. Soc. 1998, 18(11), 1521. 208. Williams, P. D. and Hawn, D. C. J . Am. Ceram. Soc. 1991, 74, 1614. 209. Pyzik, A. J., Aksay, A,, and Sarikaya, M. Muter. Sci. Res. 1987, 21, 45. 210. Schwetz, K. A., Sigl, L. S., and Pfau, L. J . Solid State Chem. 1997, 133, 68. 21 1. Kuzenkova. P. S., Kislyi, P. S., Grabchuk, B. L., and Bodnaruk, N. I. J . Less-Common Met. 1979, 67, 217. 212. Ostapenko, I. T., Slezov, V. V., Tarasov, R. V, Kartsev, N. F., and Podtykan, V. P. Poroshkovaya Metallurgiya 1979,197,38. Engl. Transl.: Sov. Powder Met. Met. Ceram. 1979,19,312. 213. Brodhag, C., Bouchacourt, M., and Thevenot, F. in Ceramic Powders, Material Science Monographs 16, Vincenzini, P. (Ed.), Elsevier, Amsterdam, 1983, pp. 881-890. 214. Beauvy, M. and Angers, R. Sci. Ceram. 1980, 10, 279. 215. Bouchacourt, M., Brodhag, C., and Thevenot, F. Sci. Ceram. 1981, 11, 231. 216. Ekbom, L. B. and Amundin, C. 0. Sci.Ceram. 1980, 10, 237. 217. Champagne, B. and Angers, R. J . Am. Ceram. Soc. 1979, 62, 149. 218. Telle, R. and Petzow, G. Muter. Sci. Eng. 1988, A105/106, 97. 219. Vasilos, T. and Dutta, S. K. Ceram. Bull. 1974, 53, 453. 220. Furukawa, M. and Kitahira, T. Nippon 7ungsten Rev. 1979, 12, 55. 221. Larker, H. T., Hermansson, L.. and Adlerborn, J. Ind. Ceram. 1988, 8, 17. 222. Schwetz, K. A., Grellner, W., and Lipp, A. in Science of’Hard Materials, Proc. 2nd Int. Conf. Science of Hard Materials. Almond. E. A,. Brookes. C. A,. and Warren. R. (Eds’l. Inst. Phvs. Conf. Series, Elsevier, London, 1986, pp. 41 5426. 223. Schwetz, K. A,, “Boron Carbide, Boron Nitride, and Metal Borides”, in Ullmann’.~Encyclopedia of Industrial Chemistry, 6th ed. on CD-ROM, Wiley-VCH, Weinheim, Germany, 1999. 224. Schwetz, K. A,, de Groot, K., and Malkemper, W. J . Less-Common Metals 1981, 82, 153. 225. Bouchacourt, M. and Thevenot, F. J . Muter. Sci. 1985, 20, 1237. 226. Werheit, H. Prog. Crystal Growth and Charact. 1988, 16, 179. 185. 186. 187. 188.

\

942


227. Wood, C., Emin, D., and Gray, P. E. Phys. Rev. B 1985, 31, 6811. 228. Aselage, T. L. and Emin D. Proc. 10th Int. Symp. Boron, Borides, and Related Compounds, Albuquerque, NM, Boron-Rich Solids, AIP Con$ Proc. 1990, 231, 177. 229. Heuberger, M., Telle, R., and Petzow, G. Powder Metallurgy 1992, 35, 125. 230. Heuberger, M. thesis, University of Stuttgart, 1995. 231. Telle, R. in Boron-Rich Solids, Proc. 10th Int. Symp. Boron, Borides, and Related Compounds, Emin, D., Aselage, T. L., Switendick, A. C, Morosin, B., and Beekel, C. L. (Eds), AIP Conf. Proc. 231, American Institute of Physics, New York, 1991, pp. 553-560. 232. Telle, R. Ceramics for High-Tech Applications, Materials Science, European Concerted Action COST 503 Powder Metallurgy-Powder Based Materials, Vol. V, Valente, T. (Ed.), The European Communities, Brussels, 1997. 233. Telle, R. and Petzow, G. in Proc. 9th Int. Symp. on Boron, Borides, and Related Compounds, Werheit, H. (Ed.). University Press, Duisburg, 1987, pp. 234-245. 234. Efimenko, L. N., Lifshits, E. V., Ostapenko, I. T., Snezhko, I. A,, and Shevyakova, E. P. Sov. Powder Met. and Met. Ceram. 1987, 26(3), 318. 235. Litz, L. M. and Mercuri, R. A. J. Electrochemical Soc. 1963, 110(8), 921. 236. Motzfeld, K. Acta Chem. Scand. 1964, 18, 1596. 237. Hofmann, H. and Petzow, G. J. Less-Common Met. 1986,117, 121. 238. Telle, R. and Petzow, G. in Horizons of Powder Metallurgy 11, Proc. 1986 Int. Powder Metall. Conf. Exhibit., Kaysser, W. A. and Huppmann, W. J. (Eds), Verlag Schmid, Freiburg, 1986, pp. 1155-1158. 239. Telle, R., Meyer, S., Petzow, G., and Franz, E. D. Muter Sci. Eng. 1988, A105/106, 125. 240. Kim, D. K. and Kim, C. H. Adv. Ceram. Muter. 1988,3, 52. 241. Nishiyama, K. and Umekawa, S. Trans. Jpn. Soc. Ceram. Met. 1985, 11(2), 53. 242. Rice, R. W., Richardson, G. Y., Kunetz, J. M., Schroeter, T., and McDonough, J. in Proc. 10th Annu. Con$ Composites and Advanced Ceramic Materials, Messier, D. R. (Ed.), American Ceramic Society, Westerville, OH, 1986, p. 737. 243. Richardson, G. Y., Rice, R. W., McDonough, W. J., Kunetz, J. M., and Schroeter, T. in Proc. 10th Annu. Con$ Composites and Advanced Ceramic Materials, Messier, D. R. (Ed.), American Ceramic Society, Westerville, OH, 1986, p. 760. 244. McCauley, J. W., Corbin, N. D., Resetar, T., and Wong, P. in Proc. 10th Ann. Con$ Composites and Advanced Ceramic Materials, Messier, D. R. (Ed.), American Ceramic Society, Westerville, OH, 1986, pp. 538-554. 245. Holt, B., Kingman, D. D., and Bianchi, G. M. Muter. Sci. Eng. 1985, 71, 321. 246. McCauley, J. W. Am. Ceram. Soc. Bull. 1988, 67, 1903. 247. Faber, K. T. and Evans, A. G. Acta Metallurgica 1983, 31(4), 565. 248. Selsing, J. J . Am. Ceram. SOC.1961, 44,419. 249. Fendler, E., Babushkin, O., Lindback, T., Telle, R., and Petzow, G. in Proc. IntSymp. Ceramic Materials and Components,for Engines, Gothenberg, Sweden, June 10-12, 1991. 250. Fendler, E., Babushkin, O., Lindback, T., Telle, R., and Petzow, G. J . HardMater. 1993,4, 137. 251. Lonnberg, B. J. Less-Common Met. 1987, 141, 145. 252. Rey, J. and Male, G. in Proc. 9th Int. Symp. on Boron, Borides, and Related Compounds, Werheit, H. (Ed.), University Press, Duisburg, 1987, pp. 419420. 253. Ohya, Y. and Hoffmann, M. J., Petzow, G. J. Am. Ceram. Soc. 1992, 75, 2479. 254. Sigl, L. S. and Schwetz, K. A. Jap. J. Appl. Physics 1994, 10, 224. 255. Sigl, L. S. and Kleebe, H.-J. J . Am. Ceram. Soc. 1995, 78(9), 2374. 256. Thevenot, F. and Bouchacourt, M. L’lndustrie Ceramique 1979, 732, 655. 257. Binder, F. Radex-Rundschau 1975, 4, 531. 258. Fitzer, E. Arch. Eisenhiittenwesen 1973, 44(9), 703. 259. Coble, R. L. and Hobbs, H. A. in Investigation of Boride Compounds for Very High Temperature Applications, NTIS Report AD 428006, Kaufman, L. and Clougherty, E. V. (Eds), Clearinghouse for Federal Scientific and Technical Information, Springfield, VA, 1973, pp. 82-120. 260. Kislyi, P. S. and Zaverukha, 0.V. Poroshkovaya. Metallurgiya 1970,7, 32. Engl. Transl.: Sov. Powder Metall. Met. Ceram. 1970, 7, 549. 261. Kislyi, P. S., Kuzenkova, M. A,, and Zaveruha, 0. V. Phys. Sintering 1972, 4, 107.

References

262. 263. 264. 265. 266. 267. 268. 269. 270. 271. 272. 273. 274. 275. 276. 277. 278. 279. 280. 281. 282. 283. 284. 285. 286. 287. 288. 289. 290. 291. 292. 293. 294. 295. 296. 297. 298. 299. 300. 301. 302. 303.

943

Baumgartner, H. R. and Steiger, R. A. J . Am. Ceram. Soc. 1984, 67, 207. Baik, S. and Becher, P. J . Am. Ceram. Soc. 1987, 70, 527. Cech, R., Olivierus, P., and Seijbal, J. Powder Metall. 1965, 8(15), 142. Rasskazov, N. I. in Proc. 3rd Znt. Powder Metallurgy Conf. 2, Karlovy Vary, Czechoslovakia, 1970, pp. 228-238. Krylov, Yu. I., Bronnikov, V. A., Krysina, V. G., and Prislavko, V. V. Sov. Powder Metall. Met. Ceram. 1976, 15, 1000. Munir, Z. A. Am. Ceram. Soc. Bull. 1988, 67, 342. Ouabdesselam, M. and Munir, Z. A. J . Muter. Sci. 1987, 22, 1799. Clougherty, E. V. and Pober, R. L. Nucl. Met. 1964, 10, 422. Samsonov, G. V., Goryachev, Y. M., and Kovenskaya, B. A. Izv. Vuz Fiz. 1972, 6, 37. Castaign, J. and Costa, P. in Boron and Refractory Borides, Matkovich, V. I. (Ed.), Springer, Berlin, Heidelberg, New York, 1977, pp. 390412. Lonnberg, B. J . Less-Common Metals 1988, 141,145. Rahman, M. and Wang, C. C., Chen, W., Akbar, S. A. J . Am. Ceram. Soc. 1995, 78, 1380. Billehaug, K. and 0ye, H. A. Aluminium 1980, 56, 642. Serrlie, M. and 0ye, H. A. Cathodes in Alurniniuni Electrolysis, 2nd ed., Aluminium-Verlag, Dusseldorf, 1994. Bannister, M. K. and Swain, M. V. Ceram. Znt. 1989, 15, 375. Murata, Y., Julien, H. P., and Whitney, E. D. Ceram. Bull. 1967, 45, 643. Venables, J. D. Philos. Mag. 1967, 16, 873. Williams, W. S. Trans AZME 1966, 236, 211. Ramberg, J. R. and Williams, W. S. J . Muter. Sci. 1987, 22, 1815. Kang, E. S. and Kim, C. H . J . Muter. Sci. 1990, 25, 580. Ly Ngoc, D. Dr. rer. nut Thesi.s, University of Stuttgart, 1989. Jimbou, R., Takahashi, K., and Matsushita, Y. Adv. Ceram. Muter. 1986, 1, 341. McMurtry, C. H., Bocker, W. D. G., Seshadri, S. G., Zanghi, J. S., and Garnier, J. E. Am. Cerani. SOC. Bull. 1986, 66, 325. Takahashi, K., Jimbou, R. Commun. Am. Ceram. Soc. 1987, C-369. Janney, M . A. Am. Ceram. Soc. Bull. 1986, 64(5), 357. Janney. M. A. Am. Ceuam. Soc. Bull. 1987, 66(2), 322. Cai, H., Gu, W-H., and Faber, K. T. in Proc. Am. Soc. Composites, 5th Techn. Conf. on Composite Materials, 1990, pp. 892-901. Faber, K. T., Gu, W.-H., Cai, H., Winholtz, R. A,, and Magleg, D. J. in Toughening Mechanisms in Quasi-Brittle Materials, Shah, S. P. (Ed.), Kluwer, Dordrecht, The Netherlands, 1991, pp. 3-17. de Mestral, F. and Thevenot, F. in The Physics and Chemistry of Carbides. Nitrides, and Borides, NATO AS1 Series E, Vol. 185, Freer. R. (Ed.), Kluwer, Dordrecht, The Netherlands, 1990, pp. 457-482. Phan-Tan-Luu, R., Mathieu, D., and Feneuille, D. Mbthodologie de la Recherche Expirimentale, Fascicules de cours, L.P.R.A.I., Universite d’Aix-Marseille, 1989. Tanaka, H. and Iyi, N. J . Am. Ceram. Soc. 1995,78, 1223. Watanabe, T. and Kouno, S. Ceram. Bull. 1982, 61, 970. Watanabe, T. J . Am. Ceram. Soc. 1977, 60, 176. Watanabe, T. Am. Ceram. Soc. Bull. 1980, 59, 465. Petzow, G. and Telle, R. in Advanced Ceramics, Somiya, S. (Ed.), Terra Scientific Publ. Co, Tokyo, 1987, pp. 131-144. Watanabe, T. and Shobu, K. Yogo-Kyokai-Shi1988, 96(7), 778. Shobu, K. and Watanabe, T. Yogo-Kyokai-Shi1987, 95(1), 991. Telle, R., Meyer, S., Petzow, G., and Franz, E. D. Marer. Sci. Eng. 1988, A105/106, 125. Watanabe. T. and Shobu, K. J . Am. Ceram. Soc. 1985,68, C34. Shobu, K., Watanabe, T., Drennan, J., Hannink, R. H. J., and Swain, M. V. in Proc. 4th Int. Conf: Sci. Techn. Zirconia Advanced Ceramics, Zirconia 86, American Ceramic Society, Westerville. OH, 1987, pp. 1091-1099. Swain, M. J . HardMater. 1991, 2, 139. Franz, E.-D., Tumback, M., and Telle, R. Sprechsual 1992, 125, 55.

944 304. 305. 306. 307.


Tumback, M., Franz, E.-D., and Telle, R. Sprechsaal 1992, 125,415. Miiller, R. thesis, University of Stuttgart, 1995. McHale, A. E. and Scott, R. S. J. Am. Ceram. SOC.1986,69, 827. Takagi, K., Komai, M., Ando, T in Sintering ’87, Vol. 2, Proc. 4th Inst. Symp. Science and Technology of Sintering, Somiya, S., Simada, M. (Eds.), Elsevier, Amsterdam, 1988, pp. 1296-1 301. 308. McMeeking, R. M. J. Am. Ceram. SOC.1986, 69, C-301. 309. Ogata, T. Toray Zndustries Internal Report, private communication, 1989. 310. Watanabe, T., Shobu, K., Tani, E., and Nakanishi, T. in Proc. 11th Znt. Symp. Boron. Borides and Related Compounds, JJAP Series 10, Uno, R. and Higashi, I. (Eds), Publication Office, Jap. J. Applied Physics, Tokyo, 1994, pp. 208-21 1. 3 11. Hamjian, H. J. and Lidman, W. G. J. Am. Ceram. SOC.1952, 35, 44. 312. Halverson, D. C., Pyzik, A. J., and Aksay, I. A. Ceram. Eng. Sci. Proc. 1985, 6, 736. 3 13. Nishiyama, K. and Umekawa, S. in Achievements in Composites in Japan and the United States, Kobayashi, A. (Ed.), Proc. Vth Japan-US Conference on Composite Materials, Tokyo, 1990, pp. 371-378. 314. Hayes, F. H., Lukas, H. L., Effenberg, G., and Petzow, G. Z . Metallkunde 1989,80, 361. 315. Zupanic, F., Spaic, S., and Krizman, A. Mater.Sci. Technol. 1998, 14, 1203. 316. Funke, V. F., Yudkovskii, S. I., Samsonov, and G. V. Russ. J. Appl. Chem. 1961,34,973. 317. Tangermann, I. Neue Hutte 1963,8, 973. 318. Samsonov, G . V. Metallov. i. Obrab. 1958, 12, 35. 319. Kieffer, R. and Benesovsky, F., Hartstoffe, Springer, Vienna, 1963. 320. Golczewski, J. and Aldinger, F., poster presentation at Annu. Mtg. Germany Society ,fbr Materials (DGM), Gottingen, 1994. 321. Funke, V. F. and Yudkovskii, S. I. Zhur. Fiz. Khim. 1963, 37, 1557; translation into English: Russ. J. Phys. Chem. 1963,37, 835. 322. Jiingling, T., Oberacker, R., Thiimmler, F., Sigl, L. S., and Schwetz, K. A. J. Hard Muter. 1991, 2(34), 183. 323. Jiingling, T., Oberacker, R., Thiimmler, F., and Sigl, L. S., in Proc. Znt. Conf.Advances in Hard Materials Production, 1992, pp. 15/1-15/14. 324. Pastor, H., Allibert, C. H., Ottavi, M. L., Albajar, M., and Castro, F. French Patent Application No. 91 08030, 1991. 325. Pastor, H. et al., in HMP’Y2 - Advances in Hard Materials Production, Proc. 4th Int. Conf., MPR Publishing Services, Bellstone, UK, 1992, pp. 23/1-23-11. 326. Sinchez, J. M., Barandika, M. G., Gil-Sevillano, J., and Castro, F. Scripta Metall. Muter. 1992, 26, 957. 327. Brynestad, J., Bamberger, C. E., Land, J. F., Fandinch, C. B. J. Am. Ceram. SOC.1983,66, C215. 328. Ghetta, V., Gayraud, V., and Eustathopoulos, N. presented at Sintering ‘ Y I , Vancouver, August, 1991. 329. Li, J. G. J. Am. Ceram. SOC.1992,75, 3118. 330. Li, J. G. Rare Met. 1992, 11, 177. 331. Savov, L., Heller, H.-P., and Janke, D. Metall. 1997, 51, 475. 332. Panasyuk, D. and Umansky, J. J. Less-Common Met. 1986, 117,336. 333. Schwetz, K. A., Reinmuth, K., and Lipp, A. Radex Rundschau 1981, 3, 568. 334. Janaf, Thermochemical Tables, 2nd ed., NBS, Washington, DC, 1971. 335. Kubaschewski, 0. and Alcock, C. B. Metallurgical Thermochemistry, Int. Series on Materials Science and Technology, Vol. 24, Pergamon, Oxford, 1979. 336. Matuschka, A. V. Borieren, Carl Hanser Verlag, Munich, 1977. 337. Schwetz, K. A., Reinmuth, K., and Lipp, A. Radex Rundschau 1981, 3, 568. 338. Sigl, L. S., Jiingling, T. Proc. 4th Int. Conf. Sci. Hard Materials (ICSHM4), Funchal, Madeira, NOV.10-15, 1991, p. 321. 339. Ottavi., L., Chaix, J. M., Allibert, C., and Pastor, H. presented at Sintering ‘91, Vancouver, August, 1991. 340. Jiingling, T., Dr.-Zng. Thesis, University of Karlsruhe, Germany, 1992, IKM-Series 007. 341. Liao, P. K. and Spear, K. E. in Binary Alloy Phase Diagrams, Vol. 1, Massalski, T. B., et al. (Eds.), American Society of Metals, Metals Park, OH, 1986, p. 355.

References

945

342. Murray, J. L. in Binary Alloy Phase Diagrams, Vol. 1, Mas-salski, T. B., et a/. (Eds.), American Society of Metals, Metals Park, OH, 1986, p. 1118. 343. Jiingling, T., Sigl, L. S., Oberacker, R., Thiimmler, F., and Schwetz, K. A. Int. J . Refractory Met. Hard Muter. 1993194, 12, 71. 344. Angelini, P., Becher, P. F., Bentley, J., Brynestad, J., Ferber, M. K., Finch, C. B., and Sklad, P. S. in Science of Hard Materials, Proc. 2nd Int. Cohf. on Science of Hard Materials, Almond, E. A,, Brookes, C. A,, and Warren, R. (Eds), Inst. Phys. Conf. Ser. 75, Adam Hilger, Bristol, 1986, pp. 1019-1032. 345. Sklad, S. and Yust, C. S. in Proc. 1st Int. Con$ Science ofHardMaterials, Hagenmiiller, R. and Thevenot, F. (Eds), Bordeaux, I98 1. 346. Takatsu, S. and Ishimatsu, E. in Proc. 10th Plansee-Seminar on Powder Metall, Planseewerke, Reutte, 1981, Vol. 1, p. 535. 347. Finch, C. B., Cavin, 0. B., and Becher, P. F. J . Cryst. Growth 1984, 67, 556. 348. Nishiyama, K. and Umekawa, S. in Proc. Composites '86: Recent Advances in Japan and in the United States, Proc. 3rd Japan-US Conf. on Composite Materials, Tokyo, Kawata, K., Umekawa, S., Kobayahi, A. (Eds.), 1986, pp. 433440. 349. Telle, R., Schmalzried, C., and Schunck, B, unpublished report to the Volkswagen-Foundation, 1999. 350. Johnson, W. B., Nagelberg, A. S., and Breval, E. J . A m . Ceram. Soc. 1991, 74, 1093. 351. Newkirk, M. S., Urquhart, A. W., and Zwicker, H. R. J . Muter. Res. 1986, 1, 81. 352. Antolin, S., Nagelberg, A. S. J . Am. Ceram. Soc. 1992, 75, 447. 353. Nagelberg, A. S . , Antolin, S., and Urquhart, A. W. J . Am. Ceram. Soc. 1992, 75(2), 455 354. Weinstein, J. in Proc. Int. Symp.Advances in ProcessinE andCharacterization of Ceramic Metal Matrix Composites, CIMjICM; Vol. 17: Mostaghaci, H. (Ed.), Pergamon, Oxford, 1989, p. 132. 355. Lugscheider, E. and Eschnauer, H. in Proc. 9th Int. Symp.on Boron, Borides, and Related Compounds, Werheit, H. (Ed.), University Press, Duisburg, 1987, pp. 202-212. 356. Komai, M., Yamasaki, Y., Takagi, K., and Watanabe, T in Advances in Powder Metallurgy and Particulate Materials 8, Capus, J. M. and German, R. M. (Eds), Metal Powder Industries Federation, Princeton, NJ, 1992, pp. 81-88. 357. Rieger, W., Nowotny, H., and Benesowsky, F. Monatsli. Chem. 1966, 97, 378. 358. Ozaki, S., Yamasaki, Y., Komai, M . , and Takagi, K. in Proc. 11th Int. Symp. Boron, Borides and Related Compounds, JJAP Ser. 10, Uno, R., Higashi, I. (Eds.), Publication Office, Jap. J. Applied Physics, Tokyo, 1994, pp. 220-221. 359. Takagi, K. in Proc. 11th Int. Symp. Boron. Borides and Related Compounds, JJAP Series 10, Uno, R. and Higashi, I. (Eds), Publication Office, Jap. J. Applied Physics, Tokyo, 1994, pp. 200-203. 360. Sivaraman, K., German, R. M., and Takagi, K. in Proc. Int. Con/ Powder Metallurgy PM'96, Toronto, 1996, pp. 9/117-9/129. 361. Sivaraman, K., Griffo, A,, German, R. M., and Takagi, K. in Proc. Int. Confi Powder Metallurgy PM'96, Toronto, 1996, pp. 11/67-11/79. 362. Matsuo, S., Ozaki, S., Yamasaki, Y., Komai, M., and Takagi, K. in Sintering Technology, German, R. M., Messing, G. L., and Cornwall, R.G. (Eds), Marcel Dekker, New York, 1996, pp. 261-268. 363. Warren, R. J . Muter. Sci. 1968, 3, 471. 364. Schmalzried, C. and Telle, R. in Proc. Int. Conf: on Boron, Borides and Related Compounds ( I S B B '99). Dinard, France, October 1999, to be published in J . Solid State Chem. 365. Takagi, K., Yamasaki, Y., and Komai, M. J . Solid State Phys. 1997, 133, 243. 366. Sigl, L. S., Schwetz. K. A., and Dworak, U. Int. J . Refractory and Hard Materials, 1994, 12, pp. 95-99. 367. Ramberg, J. R., Wolfe, C. F., and Williams, W. S. J . Am. Ceram. Soc. 1985, 68,C78. 368. Freitag, B. and Mader, W. J . Microsc. 1999, 194, 42.


8 The Hardness of Tungsten Carbide-Cobalt Hardmetal S. Luyckx

8.1 Introduction WC-Co hardmetal is a sintered material consisting of brittle tungsten carbide (WC) crystals bonded by a tough cobalt-based binder. Figure 1 shows the microstructure of a typical commercial WC-Co hardmetal. WC-Co is used mostly on account of its hardness which is high for a metallic material, although other outstanding properties, such an extremely high Young modulus, a high thermal conductivity, and a low coefficient of thermal expansion, contribute to its success in a wide range of technical applications. This chapter reviews the information available on the dependence of the hardness of WC-Co on microstructural parameters, composition and external conditions, such as temperature. It also reviews information available on the effect of microstructure and composition on the relationships between hardness and toughness and hardness and abrasive wear resistance. The hardness of industrial hardmetal grades is characterized preferentially by the Rockwell A scale [l] in the USA [2] and by the Vickers scale [3] in Europe. Most research work, however, has been carried out, also in the USA, using the Vickers scale. Most of the results reported below were obtained using standard pyramidal Vickers indenters [ 3 ] .

Figure 1. Microstructure of a typical WC-6 weight-% Co alloy.

8.2 The Hardness of the Two Component Phases

947

Table 1. The hardness of three crystallographic faces of a WC single crystal [5].

WC crystal plane

Hardness, HVI

(0001) (1 100) (1 101)

2100 1080 1060

8.2 The Hardness of the Two Component Phases 8.2.1 The Hardness of Tungsten Carbide WC is present in hardmetal in the form of single crystals. These have a hexagonal structure and are highly anisotropic [4]. Table 1 lists the Vickers hardness of three WC crystal faces and shows that the hardness can vary from face to face by up to 100% [5]. On account of the superior hardness of the (0001) faces, attempts have been made to produce WC-Co parts with a (0001) texture [6]. The literature contains contradictory results on the hardness of WC because it has been measured mostly on sections of hardmetal samples, where the grains are randomly oriented, rather than on single crystals. An additional source of contradictory results is the dependence of hardness on the load applied to the indenter when microhardness is measured (and very low indenting loads must be used when measuring the hardness of WC grains in hardmetal samples, because the grains are typically between 1 and 10 pm in size). Figure 2 [7] is a

34 32

-3

5

28

In m 0)

f

26

m

r 24 22 I

201

1 0

Figure 2. Plot of the microhardness of a (0001) plane of a WC single crystal against the indenting load.

948

8 The Hardness of Tungsten Carbide-Cobalt Hardmetal

15!

10

15

20

25

30

35

40

l/d112(mm-’12)

Figure 3. Plot of the hardness of hot-pressed polycrystalline WC against d - ‘ I 2 , where d is the mean size of the WC grains. Adapted from reference [9].

plot of the microhardness of the (0001) planes of WC crystals versus the indenting load. It shows that the microhardness of WC decreases with increasing load, which means that when measuring the microhardness of WC on sections of WC-Co samples one obtains different results at different loads. The hardness of WC has also been measured on polycrystalline samples [8]. Figure 3 reports results from several investigators [9] which satisfy the following Hall-Petch relationship:

+ KOWCd-1’2,

HWC= HOWC

(1)

where Hwc = hardness of polycrystalline WC, d = mean grain size of the randomly = average hardness of a WC crystal (over all possible oriented WC grains, HOwc crystallographic planes), and KOwc= Hall-Petch coefficient, related to the ease of slip transfer across WC-WC grain boundaries. From Fig. 3 it has been calculated that Hwc = 1330 kgfmmP2 and Kowc = 24 kgf mm-3/2. It must be noted that the results reported in Fig. 3 were obtained at room temperature, thus the values of HOwcand KOwcare expected to be those given above only at room temperature.

8.2.2 The Hardness of Cobalt The binder in WC-Co hardmetal is not pure cobalt but a solid solution of carbon and tungsten in cobalt. When cooling the material from the sintering temperature

8.2 The Hardness of the Two Component Phases

949

where X is the Figure 4. Plot of the hardness of the cobalt binder in WC-Co alloys against mean width of the binder layers, or binder mean free path. Adapted from reference [9].

to room temperature only part of the cobalt transforms from the high temperature (above 417°C) f.c.c. structure to the low temperature h.c.p. structure, and the transformed regions are present as hexagonal lamellae in a predominantly cubic material [lo]. The hardness of pure cobalt has been reported as being between 140 and 240 HV, depending on the method of sample preparation [l I]. However, a solid solution of C and W in cobalt can be up to 100% harder than pure cobalt [12], depending on the composition. It has been difficult to measure directly the hardness of the binder in WC-Co because the width of the binder regions in the hardmetal is typically of the order of 0.1 pm. However, Fig. 4 summarizes measurements by a number of investigators of the in situ hardness of the cobalt binder [9]. The hardness is plotted versus X-”2, X being the thickness of the cobalt layer where the hardness was measured, usually called ‘the cobalt mean free path’. The results in Fig. 4 satisfy the following HallPetch type relationship Hco = Hoco

+ KocoX-”2

(2) where Hco = the in situ hardness of the cobalt layers, X = thickness of the layers, Hoco = mean hardness of the binder and Koco = Hall-Petch coefficient. From Fig. 4 it has been calculated that Hot,, = 149kgfmmP2 and Koco = 16 kgfmmp’/2. Koco is related to the ease of slip transfer across Co-WC interfaces. The contribution of the Co-Co grain boundaries to the hardness of the material can be neglected because in WC-Co the size of the cobalt grains is of the order of lo3 pm while the size of the WC grains is of the order of 1 pm. Thus the number of Co-Co boundaries is negligible compared to the number of Co-WC interfaces.

950


8.3 Factors Affecting the Hardness of WC-Co Hardmetal 8.3.1 Cobalt Content and Tungsten Carbide Grain Size The two main factors affecting the hardness of WC-Co are the cobalt content and the tungsten carbide grain size. It is well established that the hardness of WC-Co decreases with increasing cobalt content at all carbide grain sizes and decreases with increasing grain size at all cobalt contents [13]. Figures 5 and 6 confirm this trend for the most comprehensive range of cobalt contents tested to date [14,15]. As a result, by selecting appropriate combinations of cobalt contents and carbide grain sizes, the hardness of WC-Co can be varied from below 800 HV to more than 2000 HV. The most appropriate combination is determined by the properties which, besides hardness, are required for a specific application. The recent introduction of grain sizes of the order of lo2 nm has extended the range of possible hardness well above 2000HV [16]. However, nano-grade hardmetals are not included in the present review because they always contain grain refiners such as VC or Cr2C3, while this review is limited to two-phase WC-Co alloys. Lee and Gurland [8] attempted to express the relationship that exists between the hardness of WC-Co, its microstructure and its composition by using a simple law of mixtures of the type

Hwc-co 2200

= Hwc vwc

+ HCO(1

-UF+F

2000-

(3)

- VWC)

*M

-++C

1800.................................. ............................................................................................................................. ......................................................................................

1000 - ...................................

800- ............................................. 600- .....................................................................................................

400

0

I

I

I

I

I

I

I

10

I

I

5

15

20

25

30

35

40

45

50

Cobalt content (wt%) Figure 5. Plot of the hardness of WC-Co alloys against their cobalt content at four different carbide grain sizes. The grain sizes are indicated in the key where: U F = 0.7 pm; F = 1 .O pm;M = 3.0 pm; C = 5.0 pm. Reproduced by permission of the Int. J. Refract. Met. & Hard Muter.

95 1

8.3 Factors Affecting the Hardness of WC-Co Hardmetal 2200

I

I

I

I

I

I

I

I

I

i

Figure 6. Plot of the hardness of WC-Co alloys against carbide grain size at the cobalt contents indicated in the key. Reproduced by permission of L. Makhele.

where Hwcpc0= the hardness of the alloy, Hwc and Hco = the hardness of the component phases as expressed by Eqs (1) and (2), and Vwc = volume fraction of the WC phase. However, Lee and Gurland found that this simple law does not agree with the measured hardness values of WC-Co alloys. In order to reproduce the measured values, they had to introduce the concept of ‘continuous carbide volume’ or ‘volume fraction of the carbide skeleton’, V,, which is defined as the volume fraction occupied by that part of the WC grains which are involved in forming infinitely long chains of connected particles while those parts which are surrounded by the cobalt binder do not contribute to V,. Lee and Gurland derived that V, is related to the volume fraction of the carbide phase, Vwc, via the contiguity, C:

v,

=

cvw,

(4)

where the contiguity, C , is defined as the ratio of the grain boundary area and the total surface area of the carbide phase. By replacing Vwc in Eq. (3) with V,, Lee and Gurland obtained the following equation: Hwc-co = CVWCHWC + (1 - CVWC)HCO which is in excellent agreement with experimental results (see Fig. 7).

(5)

952

8 The Hardness of Tungsten Carbide-Cobalt Hardmetal 1800

/c-

1600

.

iii

E E

I

m

& 1400

I

v)

(I)

5 c m t Y

1200 1000

0

5

-0

5 0 2

800

600 400 4 0

600

800

1000

1200

1400

1600

1 DO

Calculated hardness (kg/mma)

Figure 7. Plot of the hardness of WC-Co alloys, calculated from Eq. ness. Adapted from reference [8].

(9,against measured

hard-

By using Eq. ( 5 ) it is possible to calculate that in commercial WC-Co grades (i.e. grades with cobalt content ranging from about 6 to about 20 weight-%) the first term of the equation accounts for up to 80% of Hwcpc0,which means that the continuous carbide volume accounts for up to 80% of the hardness of the alloy.

8.3.2 Grain Size Distribution and Cobalt Mean Free Path A wide carbide grain size distribution obviously leads to variations in the microhardness of the hardmetal, but it has not been established what effect it has on the macrohardness. However, if variations in grain size distribution lead to variations in the binder mean free path the effect on hardness can be substantial, as shown in Fig. 8 [17]. Figure 8 shows that at equal mean free path (and so equal binder hardness, according to Eq. (2)), finer grades have lower hardness than coarser grades.

8.3.3 Binder Composition and Carbon Content Although the hardness of hardmetal is determined mostly by the carbide phase (see Section 8.3.1), the hardness of the binder does contribute to the overall hardness of the material. Therefore the composition of the binder, and specifically the amount of W dissolved in the binder, affects the overall hardness. The composition of the binder depends on the total carbon content of the material since, according to Exner [ 181, there seems to exist a reciprocal relationship between

8.3 Factors Affecting the Hardness qf WC-Co Hardmetal

0

2

4 6 a mean free path (um)

10

953

12

Figure 8. Plot of the hardness of WC-Co alloys of four different grain sizes against the binder mean free path U F = 0 7pm, F = 1 Opm, M = 3 Opm, C = 5 Opm Reproduced by permission of the Int J of Refract Met & Hard Mater

the dissolved W and the dissolved C. The higher the total carbon, the lower is the amount of W dissolved in the binder and the lower is the hardness of the binder (Fig. 9) [19].

8.3.4 Porosity The hardness of materials produced via powder metallurgy is known to decrease with increasing porosity level. However, hardmetals are almost pore-free materials and variations in their extremely low level of porosity does not affect their hardness appreciably.

8.3.5 Effect of Temperature In most applications WC-Co is subjected to temperatures which can be as high as 900-1000°C (as in metal cutting). Therefore knowledge of the effect of temperature on hardness is essential for a correct selection of the grades suitable for a specific application. The hardness of both WC and cobalt decreases with increasing temperature. Figure 10 shows the rate of hardness decrease with temperature for hot pressed polycrystalline WC and for WC single crystals [20] and Fig. 11 shows the rate of hardness decrease in cobalt [I 11.

954


90.5 go - ......................................................

89.5 - ......................................................

~

LII v) v)

E

89 ...................................................... . 88.5 _.....................................................

I

_ 87.5 _ ...................................................... 88 ......................................................

---High

87 7 0.1

carbon

-

Low carbon 1

mean free path in Co (um)

Figure 9. Plot of the hardness of WC-Co alloys of different carbon contents against the binder mean free path. Adapted from reference [19].

The most comprehensive work on the effect of temperature on the hardness of WC-Co has been carried out by Milman and co-workers [21] and has involved WC-Co grades ranging in grain size from 0.5 to 2.3pm and in cobalt content from 6 to 15 weight-%. These grades have been tested at temperatures ranging from 20 to 900°C and the results are summarized in Figs 12 and 13. Figure 14 gives the same results in Hall-Petch coordinates, and shows that at all temperatures and cobalt contents a linear relationship exists between the hardness of WC-Co, Hwc-c0,and d-'I2, d being the mean grain size of the WC grains. The linear relationships shown in Fig. 14 can be explained by combining Eqs (l), (2), and (5) and by expressing A, the mean free path in cobalt, in terms of d using the

22

.;a 20 n

c 2 In In

18

a,

5

16

0

.c

8

14

0

4

12

10 8 100

200

300

500 600 temperature ( 'C)

400

700

800

900

Figure 10. Plots of the hardness of WC against temperature. Adapted from reference [20].

8.3 Factors Affecting the Hurdness of' WC-Co Hardmetal

955

200 180 160 140 120 100 80 60 40

20

0

100

200

300

400

500

600

700

800

temperature (OC)

Figure 11. Band including possible values of the hardness of cobalt at various temperatures. Adapted from reference [I 11.

following equation due to Lee and Gurland [8]:

By combining the above equations one obtains

where

and

with

Ho and K y can be considered constant at constant carbide volume fraction, Vwc, since the contiguity, C, varies with cobalt content but does not vary appreciably with grain size [8]. Hence the relationship between Hwc-co and d-'I2 in Eq. (7) is

956


24

22

--

d = 0.5 pm

h

m a

c7

\ 6 - r ~ 15 wt% co 4 0 100 200 300 400 500 600 700 800 900 Temperature ("C)

0

0

100 200 300 400 500 600 700 800 Temperature ("C)

100 200 300 400 500 600 700 800 900 Temperature ("C)

Figure 12. Plot of the hardness of WC-Co alloys against temperature. The cobalt content of the alloys and the grain sizes are indicated.

linear at any Vwc, in agreement with the results in Fig. 14. Figure 14 shows that the above relationship is linear also at all temperatures, which implies that the relationships between Hwc and d and Hco and X (see Eqs (1) and (2)) are linear at all temperatures. Figures 15 and 16 show that Ho in Eq. (7) decreases rapidly with increasing temperature up to about 600"C, while K,, decreases substantially with temperature only above 600°C. Since Ho is a combination of the intrinsic hardness of WC and Co (see Eq. (8)) and Ky a combination of the Hall-Petch coefficients of Eqs (1) and (2) (see Eq. (9)), Figs 15 and 16 suggest that the softening of WC-Co with increasing temperature is due to the intrinsic softening of the component phases up to about 600°C but is controlled by the ease of slip transfer across grain boundaries and interfaces above that temperature. Milman and co-worker? results [21] have established that the hardness of finer grained material decreases with increasing temperature at a lower rate than the hardness of coarser grained material and thus the advantage of using finer grained material increases with increasing temperature.

8.3 Factors A.fecting the Hardness of' WC-Co Hardmetal

951

24 22 20

0

c2

g

18 16 14

C

$ r

12 10 8 6 4

0

100

200

300

400 500 600 temperature (C)

700

800

900

20

10 wt% c o 18

-I? g

16 14 12

C

$ c

10

-

8 6 4 I

d = 0.5 urn +- d = 1.2 urn

100

200

300

+++

d = 2.3 urn

400 500 600 temperature ( C)

18

700

800

15 wt% Co

C

I

16

-I?

9.

g

14 12 10

C

$ 8 c 6

4

2

0

temperature ( C)

Figure 13. Plot of the hardness of WC-Co alloys against temperature. The grain size of the alloys and the cobalt contents are indicated,


-

25 .........

20 -

m

5

'5-

In) u acl

2 m

10-

I 5~

15

20

30

25

35

40

d (rnm)i'nr ll.l^l"lllll ................................ _I___

45

"

"

50

I __ *20 ' C

20 -

0200 ' C

A300 ' C

A400 ' C 0500 ' C 0 600 ' C

m700 ' C 0800 'C XJOO ' C ~

I

i j 20

15 wt%

~

co A300 "C

1 A400

"C

i 0 6 O O "C

15

20

25

30

35

40

45

50

d (mrn)4'n)

Figure 14. Plot of the hardness of WC-Co alloys against d-''2, where d is the carbide mean grain size, at temperatures in the range 20-900°C. The regression coefficients are all higher than 0.95 and mostly higher than 0.99.

8.3 Factors Affecting the Hardness of WC-Co Hardmetal

959

700

600 500 h

.

E 400 E

Y 0

Y

0

300

I 200 100

04

o

I

I -

loo

200

WC-10 vol% Co

300

-+

700

500 600 temperature ("C) 400

aoo

I 900

WC-16 vol% Co -+- WC-24 vol% Co

Figure 15. Plot of the parameter Ho of Eq. (8) against temperature for WC-Co alloys of different cobalt content. The cobalt contents are indicated in the key. Reproduced by permission of the Znt. J . of Refract. Met. & Hard Mater.

40

I

35 30 A

E E

25

m

& 20

y" 15 10 5

I

-D--

100

200

WC-10 vol% Co

300

400 500 600 temperature ("C)

t-WC-16

vol% Co

700

-

aoo

WC-24 vol% Co

I goo

I

Figure 16. Plot of the parameter K y of Eq. (8) against temperature for WC-Co alloys of different cobalt contents. The cobalt contents are indicated in the key. Reproduced by permission of the Int. J . of Refract. Met. & Hard Mater.

960

8 The Hardness of Tungsten cur bide^ Cobalt Hurdmetul

-

L "

20

6 wt% CO

19

cc

18

10 Wt%

3 a 17

w

E E 2

c

! d = 0.5 pm 14.1 -200

3

-50 -100 Temperature ("C)

-150


15 14

12 1 I

-150

16

I

0

50

I 0

50

11 -200

d = 1.2pm -150


0

50

- 16 m

& 15 u)

$ 2 E

14 13 12 11 104 -200

I

Figure 17. Plots of the hardness of WC-Co alloys of different cobalt contents against low temperatures. The grain size of the alloys is indicated in the keys.

Milman and co-workers tested WC-Co alloys of grain size ranging from 0.5 to 2.3pm and cobalt content from 6 to 15 weight-% also at temperatures ranging from -196°C to 20°C. The results are shown in Figs 17 and 18 [21].

8.4 Relationships between Hardness and Other Hardmetal Properties In most hardmetal applications it would be desirable to use the hardest possible grade. However, properties often have reciprocal relationships, i.e. an improvement in one leads to a deterioration in another. Therefore, in order to select the most appropriate grades for a specific application, it is desirable to know quantitatively the relationships between the various properties. So far, extensive work has been

8.4 Relationships betn3een Hurdness and Other Hardmetal Properties

124 -200

I

96 1

I

-150

-100 -50 temperature ( C)

0

50

-150

-100

-50 temperature ( C)

0

! I

-150

-100 -50 temperature ( C)

0

50

24 22

-

20

cr

18

n u)

I?

16

; c

14 12 10 -200

18 17 16

: 15 cr

3

14

-0 c

c

13 12

11

10 -200

Figure 18. Plots of the hardness of WC-Co alloys of different grain size against temperature. The grain sizes and the cobalt content are indicated.

962


-t22pm

--yt

6pm

~-

1000 I 8

I

10

12 14 16 18 Fracture Toughness (MPafi)

20

22

Figure 19. Plot of the hardness of WC-Co alloys of constant grain size against toughness. The grain sizes of the alloys are 2.2 pm and 6 pm. Adapted from reference [22].

done to establish the relationships that exist between hardness and toughness and hardness and abrasive wear resistance.

8.4.1 Relationship between Hardness and Toughness It is well established that a reciprocal relationship exists between the hardness and the toughness of WC-Co. However, it has proved difficult to explain this relationship in terms of the material’s microstructural parameters until it was observed that the relationship between the two properties is linear when the grain size is kept constant [22]. Figure 19 shows the particular case of alloys having grain size 2.2 and 6pm, but the linearity of the relationship has been confirmed for a wide range of grain sizes [23]. Therefore it appears that, in general: Hwc-co = 4 W I C + b ( 4 (1 1) where KIc = toughness of WC-Co, and a(d) and b ( d ) = functions of the WC grain size.

8.4.2 Relationship between Hardness and Abrasive Wear Resistance Figure 20 shows the results of a comprehensive investigation into the relationship between the hardness and the abrasive wear resistance of WC-Co, with the wear resistance being measured according to the ASTM Standard B611-85 for hardmetal,

Acknowledgments

963

1000

7

........................

.-0IE n E

-

U

l ................................... ............................

..................................

0.1 400

600

800

-UF

+F

"M

*c

1000 1200 1400 1600 1800 2000 Hardness fKglrnrn2J

Figure 20. Plot of hardness against abrasive wear resistance for a wide range of WC-Co alloys. U F = 0.7 pm; F = 1.0 pm; M = 3.0 pm; C = 5.0 pm. Reproduced by permission of Int. J . Refract. M e t . & Hard Mater.

using alumina particles as abradors. The alloys tested ranged in cobalt content from 3 to 50 weight-% and in grain size from 0.6 to 5.1 pm [17]. Figure 20 shows that up to a hardness of approximately 1000HV there was a one-to-one correspondence between hardness and abrasion resistance, which suggests that the main wear mechanism was plastic deformation. Above 1000 HV the wear resistance of coarser grained alloys was higher than that of finer grained alloys of equal hardness, up to approximately 1600 HV. This suggests a wear process controlled by microfracture, since in that range of hardness coarser alloys are tougher than finer alloys. Therefore, the abrasive wear resistance of hardmetal increases with increasing hardness but, in the hardness range which is of interest to most applications, there is not a one-to-one correspondence between hardness and abrasive wear resistance since resistance to wear is determined also by other properties, such as toughness.

8.5 Conclusions The knowledge of the factors affecting the hardness of WC-Co alloys has increased considerably in recent years, which has allowed a more precise selection of grades. However, much work remains to be done to quantify the relationships that exist between hardness and other properties which affect the performance of the material.

Acknowledgments The author wishes to acknowledge the contribution of several colleagues to the research work carried out over the years and reported here. Professor Yu. V. Milman,

964


Mr D. G. F. O’Quigley and Mr I. T. Northrop must be mentioned in particular. The financial support of the Foundation for Research Development, THRIP and the Boart Longyear Research Centre, over the years, is also gratefully acknowledged.

References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.

ASTM Standard B 294-92. J. J. Oakes, Teledyne Advanced Materials, Private Communication. I S 0 Standard 3878. S. Luyckx, F. R. N. Nabarro, Siu Wah Wai, and M. N. James, Acta Metall. Muter. 1992, 40, 1623. T. Takahashi and E. J. Freise, Philos. Mug. 1965, 12, 1. S. Luyckx and J. Katzourakis, Muter. Sci. Technol. 1991, 7,472. S. Luyckx and L. C. Demanet, Unpublished Results. H. C. Lee and J. Gurland, Muter. Sci. Eng. 1978, 33, 125. L. S. Sigl and H. E. Exner, Muter. Sci. Eng. 1989, A108, 121. J. Freytag, PhD Thesis, University of Stuttgart, 1977. W. Batteridge, Cobalt and its Alloys, Chichester: Ellis Horwood, 1982. B. Roebuck and E. A. Almond, Muter. Sci. Eng. 1984, 66, 179. H. E. Exner and J. Gurland, Powder Metall. 1970, 13, 13. D. G. F. O’Quigley, MSc Thesis, University of the Witwatersrand, Johannesburg, South Africa, 1995. L. Makhele, University of the Witwatersrand, Unpublished results. D. F. Carroll, 14th International Plansee Seminar, Plansee AG, Reutte, Austria, 1997, 2, 168. D. G. F. O’Quigley, S. Luyckx, and M. N. James, Int. J. Refract. Met. Hard Muter. 1997,15,73. H. E. Exner, Int. Muter. Rev. 1979, 4, 149. T. Sadahiro, J. Jpn SOC.Powd. Powd. Met. 1979, 26, 33. M. Lee, Metall. Trans. A , 1983, 14A, 1625. Yu. V. Milman, S. Luyckx, and I. T. Northrop, Int. J . Refract. Met. Hard Muter. in press. D. G. F. O’Quigley, S. Luyckx, and M. N. James, Muter. Sci. Eng. 1996, A209, 228. S. Luyckx, V. Richter, D. G. F. O’Quigley, and L. Makhele, Proc. Int. Con$ Deformation and Fracture in Structural Materials, Institute of Materials Research, Slovak Academy of Science, Kosice, Slovakia, 1996, 2, 109.


9 Data Collection of Properties of Hard Materials* G. Berg, C. Friedrich, E. Broszeit, and C. Berger

9.1 Introduction Hard materials are used as thin hard coatings of some microns thickness for wear protection of tools and machine parts because of their high abrasive wear resistance. For the selection of the coating material the physical, mechanical, and technological properites of these coatings, required by the application, are decisive. The following data collection presents fundamental and available material properties for approximately 130 hard materials as a result of a literature search on carbides, nitrides, borides, silicides, and oxides.

9.2 Profile of Properties Each technical application demands a special profile of properties concerning the material. In tribological applications the material selection of the system components plays a dominant role. The existing profile of material properties must, if possible, go beyond the demanded characteristics. For a functional selection of materials basic data must be available. A ‘profile’ does not exist if only one property is known, it is built up from all relevant material characteristics, which could be called ‘mechanical-technological’, ‘mechanical-tribological’ or ‘physical-chemical’ for mechanical components. Only the combination and comparison of numerous characteristic data of one material make it possible to sketch a profile of properties (columns in the following table). First if there is a full characteristic profile for this material available, it makes sense to compare it with other materials in complex applications to recognize all positives and negatives (rows of the table). Investigations of thin films show that the material properties are not constant. Responsible for this are different deposition technologies and procedures which result in different structure properties of the coatings, having various textures or various densities of defects. For this reason the evaluation of material properites by different references is of special interest (several data for one characterisic value). * This contribution is the revised and translated version of the publication Datensammlung zu Hartstoffeigenschaften, published in Materialwissenschaft und Werkstofftechnik 1997, 28, 57-76, 0VCH Verlagsgesellschaft mbH, D-69451 Weinheim

966

9 Data Collection of Properties of Hard Materials

The presented data are mainly of a physical-mechanical character. For applications technological/tribological data (i.e. load bearing capacity of hard coatings in different tribological systems under corrosive attack or under high temperature applications) are of special interest. For these uses, in practice physical data may only give first qualitative hints for decisions concerning a successful selection.

9.3 Organization and Contents of the Data Collection The following data collection provides an overview of basic properties of hard materials found in the literature. The list is structured according to the chemical composition of the materials. To define ‘hard materials’ a minimum hardness of HV = 1000 is demanded (definition of Schedler [57]). Values of bulk materials and coatings are listed together. A differentiation would be useful but is very difficult to achieve because basic information on design and structure of the layers is not available. For this reason multilayer and multiphase coatings like (Ti, A1)N or Ti(C, N) have not been taken into account. Physical values like crystal structure and the Young modulus are cited, which may play a dominant role in mechanical/tribological loadings. Additionally values for microhardness and oxidation resistance are given. More than others these two values are a function of the test system, the test parameters, and the evaluation method. At this point very strong deviations appear (e.g. hardness of Sic: HV = 1400 - 6000). In many cases the boundary conditions of the investigations are not documented (e.g. no load specification for hardness testing). For this reason values for hardness and oxidation resistance have a qualitative character more than others. This points out the necessity of further standardization efforts, to give the results of the system tests a transferrable and absolute character. In the field of thin films, the standardization is in progress, including special test procedures, but they are not yet established and for this reason not used and documented. Giving a guarantee of properties is only possible if firm regulations exist. For series investigations the realization of this demand is absolutely necessary and already realized for other test methods (e.g. macro hardness testing according to different DIN or I S 0 standards). To reduce the size of the table, the given data are as far as possible compressed. One line in the data collection includes data from different authors. Mostly in one reference only single properties of the hard materials are given, so that only the collection of these properties leads to a profile of characteristic properties for a single material. The references are quoted in brackets. The influence of the deposition procedure and exactness o f the test methods on the profile of properties of hard coatings leads to very different values presented by the different authors. Especially the properties depending on crystal structure like thermal conductivity, electrical resistivity or the Young’s modulus (see e.g. electrical conductivity for NbC or the Young modulus for TIN) differ very much.

Acknowledgement

967

Good correspondence is found for crystal structure, lattice parameters, density and melting point for one material. The data collection (Table 1) does not claim to present all existing investigations on hard materials. The aim of the presented table is to give a representative overview on a special material. The number of quotaions gives a first impression on the number of published values (see e.g. TIN which is well known as layer for tribological applications). Looking on the five material groups, carbides and nitrides are dominant. The database for the data collection comprises actual textbooks, theses, and publications in technical journals and from conferences in recent years. All in all approximately 3000 references were searched through. One result is that in new publications single phase materials are seldom studied. The quotation of older literature in these publications leads to the repetition of single values by different authors.

Acknowledgement Many thanks to Mrs. R. Kurth for listing all the data on PC and thanks to Mr. U. Petzel as well as Mr. T. Eid for their help in increasing the data base.

I 146/0.552/0.2821 I 147/0.5545/0.2830 1.147/0.5545:0.2830 1.147:0.553j0.282 I 147/0.5545/0.2839 I. 148310.5531/0.2827 I. 1488/0.552710.2829 1.14883,O 5527310 28286

0 5631/1.2144 0.5804,1.2079 0.5599/1.20?4 0 56003/1.2086

[52]

161 [lo] [Ill [25]

[I121

[55] [SS]

2.52 2.5 2 52 2.52 2.52 2.52 2.52

20 21 22 23 24 25 26 27 28 29 30 31

19

18

[I121

[52] [74] [I121

1281

[6l [7] 191 [lo] [XI [391 [53] [57, 1071 1741 1771 [YO. 1061

[97l U061 11081

6.68 6.68 6.68 6 68 6.68 6.68 6 66 67 6.68 6.68 6 68

Wl

2.52 2.5 2.45

[I] 171 1141 [251 [?I] [35] [38] 1391 1431 [48] [52] [57] [70] [771

2.52 2.5

2.52 2.52 2.51

2.51

[6] [8] [lo] [Ill [25] 1281 [46] [52] [57] [74] [76]

[XI [31] [43] [SZ] [I121

2.52 2.52

ortho ortho ortho ortho ortho ortho ortho ortho ortho ortho ortho ortho

rhom rhom rhom rhom hex

8 9

Cr,C:

B4C

(c: cm-’)

Density

cub: cubic cub-B1: cubic, NaCI-type rhom: rhombohedral/trigonal

10 11 12 13 14 15 16 17

6 7

5

3 4

7

I

No. Symbol Crystal Structure

Lattice parameters inm)

stmcbre: orthorhombic hexagonal hexagonal close-packed

Carbides

Crystal ortho: hex: hcp:

Legend materials: carbides, nitrides, borides, silicides, oxides m. 0, y. S, 6, TI, T2: phases HV: Vickers hardness, value after HV gives load in kp according to DIN HK: Knoop hardness, value after H K gives load in k p according to DIN HU Universal hardness. value after HU gives load in N according to DIN

Table 1. Data collection of properties of hard materials

[9] [lo] [40] [57] [76] [77] [87] [I061

10.3

[XI

117 10.3 10.3 10.3 10.3 11.7 10.3

[I] [8] [30] [31] [35] [381 [431 [48] [571 ~701 [77l [97] [I061

10 3

[1.48,57.7716 [5] 4.5 [7] 7.32,’s 33 [XI 4.5 [I41 45 [25] [30] 4.5 [31] [38] [39] [43] 4.5 4.5 [55] 4.5 [88] [I061

Linear thermal expansion, 3 K-’)

19

18 8

28 35 28 27 2

27.63 29.3 10.42 29

[9]

(81

[38] [43] [48] [57] [I061

1351

[I] [XI

Thermal conductivity, X (W m-’ K - ’ )

fcc: face-centered cubic bcc body-centered cubic let’ tetragonal

75 75 75 75 75 75

75

4 x 10’ 101-10’ 4 x 10’ lo6 5 x 10’ 5 x 10’

5 x 10’

lo6

[I061

[87

[7] [9] [40] [75] [77]

[I41 [311 [38] [48] 1571 [681 I771 [Y71

Electrical resistivity ncm)

88.8 88 23 109.7 85.4 94.2

72 40 71 6 57.7 71.6

[6] [Ill [46] [I021 [73]

PI

(51 [43] 1461 [48] [73]

(kl mol-’)

Enthalpiy

mono: monoclinic tri: triclinic

monocrystalline polycrystalline incongruent melting point (formation of liquified material and a second solid phase) peritectic melting point (several solid phases and liquified material in equilibrium)

1900 (per) [Sl [7] I890 [XI 1810 [9. 10,401 176.77.871 1810 I895 [I31 1895 [ZS] [39,46.90] 1890 1895(d) I531 18Y5(d) I571 1850 [I061

1850

2450 2350

2500

2450 2470 2350 2350 2450 2450 2430 2450 2450 2447 2447

Melting point i’C)

(mo) (PO) (i) (per)

37 4 3.7 3.73

3.284

4 3.86 3.7 3.73

4.5 4.48 4.4 2.96 4.41 45 4.6 4.5 4.48 42 441 4.41 2.9 4.4-4.7

191 [lo] I401 [57] 1741 [76] 1771 [87] [I061

[I061 [lox]

[97

[I1 [lo] I141 [XI [31] [38] [43] [48] [S7] [70] [77]

Young modulus (LO’ N mm-’)

2150HV

2280 1350

1500-2000 1300 HV 2100 1300 100&1400 2280 HV 1300 HV 0.05 1350 HV 2280 HV

4000-5500 30W3500 HV 0 I 300&7000 30w-4000 HV 3500 HV 0.2 3200 HV

3000 490&5000 HV 2940HK0.05 3700 HV 0.05 300M000HV 2800 HK 0 I 3700-4700 HV 4950 HV 2800 HK 3700 HV 35w-4500 HV 0 I 2940 HK 0. I

3700

Micro hardness (10 Nmm?)

12 11-14 11-12

[8,871 [91 16, 10.40.761 [I 11 [I31 ~251 139.901 1531 [571 1591 [771

x

’

11-14 7-8(P)

[25] [53]

[71

[ZS] [35] L431 [701

(IOO’C)

Oxidation resistance

I71

[58] 1701 [781 [77] [881 [I081

[I. 331 I71 IS] [I41 1251 [3I] [33l [35] 1391 [44l [13.44. 57,681 [48] I581

The crystal systems of single materials may he explained by different crystal stuctures e.g. hexlrhom for AIzO,)

density as determines by X-rays, determined by lattice parameters and atomic mass (d) decomposition (p) “heat proofness” (I) “thermal stability”

(x)

[I] [6] [lo] [Ill [25] [46.76] [52] ortho [74]

0.428

Mo2C-n ortho [ l o ] 0 7244jO 600410 5199 ortho [ I l l 04733l06034'05206 ortho (28. 1121 04732.0 6037 05204

[I121 hex fcc [Ill cub-BI 1251

hex hex hex

72 73 74 75

0 29'0 284 0 2901/02785 o 2896 o 2x09 0.2896\0 2809 0,2932 I 0 9 7 0427

0.464 0 464 0.464 0.464 0 4646 0.446 0 463765

1.0659 I066 1.06599

I065

I 06 I0638 I0655 I 06&1 066

[lo] [Ill [??I [321 [52] [I091 [I121

[Hi]

1.398 0.4523 I401 0.4525 I401~04?32 I 398,O 453 I4006 0.4532 0.7015,1.?153 0.4532 070149'1 2153 045320

MoC-0

[2] 161 [in] 1521

1112)

[8] IIO] [I I ] 1251 [30] [52] 1571 1741 1761 [98] [I091

[I] [6] [lo] [Ill (251 1521 174. 7x1 (1121

[I1

181 [lo]

14/ll.45

70 71

hex

fcc fcc cub-Bl cub-BI cub-Bl cub-BI fcc fcc cub-Bl fcc cub-BI cub

fcc fcc fcc fcc fcc cub

cub fcc

ortho Ill!]

hex her hex hex rhom

hex

hex

MoC-.,

HfC

Cr2,C6

Cr7Ci

65 66 61 68 69

64

63

53 54 55 56 57 58 59 60 61 62

49 50 51 52

46 47 48

45

44

41 42 43

39 40

77 38

36

33 34 35

32 [I1 [61 1521

[7] (91 I 2 6 7 ( x ) [lo] 12 2 I251 I?7 1251 12 3 [38] -12 [39] 12 7 [52] 12.2 1741 I2 1901 I22 [I061 122 [I081 12.7 [I071

122 1267

69 6.85 69

2670 2400 2410

2700 2677 2692 2600

3900

3387 3830 3890 3890 3928 3890

3950 3890 3890

[I31 [39]

(71

[6] [lo]

IS]

[I]

i62j [76] [YO] 198. 1061

1571

[32] 1381 1391 1451 [46]

j25] [30]

3890 3927

151 [7,8.9.13] [lo]

[I]

3830 3890 3928

1520

1780Id) [I] 1780 [46] 1600-1790 [I021

IlOh]

6 I3

29.31

63

[XI [lo] 1301 [32] 1381 [45] (621 [76] 1981

[I]

[I]

49:./x2 [lo] 8.5'4 5 5 7 [28]

6.6 67 659

6

673 66 678 66 6.4 6.9

101

106

[8] [PI [32] [38]

49

37 39 37 40 37 37

[6]

[7] [9] [32] [38] [57] [lO6]

190? 178

126 10

11.7

209 7 205 219 2303

396 5

68 7 68 6

228 2 I62 181 3

iloi]

[5] [I!] [73]

[II] [45] [4h] [73]

[I/ [8] [73]

[731

161 1461

[I] [53] [74]

228

[lo]

[9] [lo] [32] [38] 3 5 4 6 0 8 [74] 4 64 [76] 3 52 [I021 3 55 3 174.21 4.71 4

3.726

3.5

1479-1800 1950 H V

1500

1600

2600 180@2500 ?913HV 2750 H V

3200 180&2500 H V

2600 2700 H V 2800 2700 2276 HK 0.1 2200 1932-2900 H V 0.05 1000 H V 2913 H V

2200 H V 2

[I31

[101 [I11

171

(581 [681

[ I 11

IROOHV l8OOHV

111

1650

[72] [1271

[641 1681

[SBj [61]

1531

[I1 [Ill

1000

1200 H K 0 01

I600 1900-2200 1200-1600 H V 2200-2400 H V 0 05 1910HV003 1200-1600 H V I500 H V 10.025

2200

5-8

7

[2?]

[53]

113

110 SIC Ill II2

102 103 Nh,C-P I04 105 106 Nb?C-y I07 108 109

Ir.hcx pfcc a:hex B:cub

[22] [22] 1311 1311

B.0.4360 n:0.3-7.3,1-1.5 0:0.43 0 0 4358

03128104974 0.3127i0.4972 0.311910.4Y59

1.092/0497/0.308 1.092010.4974~0.309 0.540j04974 0.3116,’04958

ortho ortho hex hex hex hcp hcp hcp hcp

[lo. 641 [I121 161 110. 1121 1281 [6] [lo] 1281 1521

045 0.4171 0447 0.4471 0.4461 0.447 0.4449 0.447 044698

1251

[6] 0.2994/0.4722 [10,112] 0.3005/0 4755 [Ill 0.3004/0 4722 1141 0 300210.4724 1251 0.414 128,461 0 3012 0.47352 (521 1571

Lattice parameters (nm)

fcc [I] cub-BI 161 fcc 1x1 cub-BI [lo] cub-81 (111 cub-Bl 1251 cub-BI 1301 cub-BI 1461 cub-BI 1521 fcc 1571 ccc 114) cub [I121

MolC-$ hcp hex hex hcp hcp hex hen hcp fcc

85 NbC 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 Nb,C-a

84

82 83

76 77 78 79 80 81


Table I. Continued

1521

[61 [22] (221

1521

1101

I61

3.2 3.17 3.2 3.22

111 16) 171 191

7 85 (XI 161

1521

785(x)

[lox]

[I] 161 171 [lo] I251 1391 1571 1741 1771 1901 11061 [I071

[6] 1251 1571 1771 1901 1971 [I061 (107] [I081

11121 UOI 11121

7.8 76 7.78 7 56 7.6 7.9 7.9

7.78 7 51-7.82 7.82 7.79(x) 7 82 7.56

8.Y 8.9

89 9 18 9.2 9.18 8.9 9 18 8.9

[lo]

I251 1521

[6] [lo]

(gcm-7

Density

(61

197) 1106l [73l

(771

[lo] [I31 125) 1301 1321 1391 (571 j73,901

181

[I] [S] 161 I71

(61 1251 [46] 1571 1731 1771 1901 1971 (1061

2200 27W 2760fd) 2300(d) 1141

191

[I] [7]

-3500 (per) 161

3100

3490 3600 3477-3900 3500 3480 3600 3500 3500 3500 3600 3480 3500 3480 3613 3613 3480 3090 (per)

2687 2690 2430(per) 2400Id) 2430 2517 2410 2517 2697

Melting point (“c)

1301 I321 (571 1771 197) (1061

110)

161 181

111

1281 1571 1771 1971 [I061

4.8 a:4.5

53

5 68

[I] 191 (141 [22]

6.67.0,8 7 (61 6.816.9 1281

4.1UIl.OU7 1281

6.65 6.2 6.84 66 52 6.6 67 7.2 7.2 6.65

5.6/7 5 7.8 7.8-9 3 7.8-9 3 7.8

K-’)

Linear thermal expansion, a

15.49 63-155 42 60

14.24 11.25 14.2 14 24 14 24 14 24 I4

(221

191 1141

111

[I] [6] 181 [lo] 1251 I321 (1061

Thermal conductivity, X (W m-‘ K - ’ )

1321 1571 166) (771 1971 11061

1251

[6]

[?I

[6] (251 is71 I771 [97] [I061

[9] 1141 10’-10’? 1221 1311 10’

10’ 10‘

19 35

19

3.C74 34 74 51 35 19

97.5 97 133 57 57 71

Electrical resistivity (lo-‘ Qcm)

1

80 70.0

p73.3

a: 71 6

167-195

1863 191.5

139 8 140.7 140.7 138.2 140.7

46 46 46.1

161 161 1221 (431

161

1461 1731

IS] [6] [Ill 1461 1731

[Ill (461 (731

(kJ mol-’)

Enthalpiy

(571 1771 1971 IlO6l

4.8 3.94 I 4.8 4.7

[I] 161 [9] 1141

3.4 [I] 2.763.45 161 3.38 [lo] 3 45 1321 3 38 1571 3 193-5 10 1741 58 [7?] 5.8 1971 3.38 [IOh]

533 5.4 5.4 533

Young modulus (10’ N mm-2)

3500 1400HV 2169-2428 2600

1924-2322

1800 1961 HV 1265 2000 2004 HV 2400 HV 0.025 1800 HV 1961 HV 1800 HV

2400 195k2700 2000-2400 2400 HV 2055-2400 2400 HV 0.05

ISOOHV 1950 I500 HV 1500 1499 HV I660 HV

iaoo HK

15OOHVU1410.05 1499 HV

Micro hardness (10 N mm-’)

[I. 331 141 161 17. 451

161

(441

1251 p91

1617 1251 I3-14iP) 1351 1431 16 1701 1: 16

Oxidation resistance (IOWC)

TrC

Ta,C-Z

159 TIC I60

I58

151 Ta2C-P 152 153 154 I55 I56 157

149 I50

148

I45 146 I47

142 143 144

141

I40

I39

138

137

136

I34 I35

131 132 133

I30

122 123 I24 I25 I26 127 I28 I2Y

II 4 115 116 II 7 118 119 I20 I21

[6]

[ I , 1121 [lo] [Ill [I41 (251 [52]

[25] (301 [46] [52] [57] [74] [98] [I091 [I121

[Ill

[lo]

[XI

[I] [6]

[I121

[43] (431 [47] [47j [52] [52] [76] [76] 198)

fcc [I] cub-BI [3]

hex

hex

hex

hex hex hex hcp

fcc cub-BI fcc cub-BI cub-BI cub-BI cub-Bl cub-BI cub-BI fcc fuc kc cub-BI cub

?:hex flub xhex Fcub ?.hex Pub ?.hex Fcub hex xhex

3n73!1 508

0.429~~33

n 433

0 31110.495 o.3103/0 49378 0.3102/0 494 0.3106:0.4945 0.3091:n.493 0.31037/0 49394 0.31042:O 4941 0.3 1046’0.49444

044 0 4454 0 4456 0 4455 0 4455 0 4456 0.4456 0.4456 0.44 0.445 0.44547

X:O

[I] (61

u121

4.93 4.93

141

[I]

[I]

I 5 05 (XI [6]

[in71

[I081

[90] [98] [I061

[MI

[7] 1101 [I41 [25] [38] (391 [57] (741 (771

6 2,4 8

[28]

2093 17-235

[I] [R]

40

[6]

71.2

3067 3150

[?I

I41

7.42 7.4

[31

[I]

52 68

[41 171

1796

I838

104.3 203.1 203.1

188.4

Ill I61

4 3.22

2.85

3400 (d) [I] 3500 (per) [73]

2 85

3880

56 291 56 56

141 [6]

3200 HV 3200 HV

1000 HK

I700 1000- I720

179nHVO2 I550 HV 2500 HV

inoo H V 2500 1547-1952 1790 HV I550 HK 0 05 1800 H V o 05 1600H V loon H V I599 HV I790 I599 2.91 29 2 85 2 85 3 04 5 34 2 85

5.5

2600 HV 2500 HK 0 I 3nw3500 HV 20W6000 HV 0 05 2585 HK 3000 HV 0.2 200&60no 2580 HK 0.1 2500 -4000 H V 3000 2500 H V 30 ~4on-3nonHV 0. I 2400-3500 H V

3500 HV 0.05 2600 HV

350M000

3nowoon

20W3000 3500 HV 2580 H K 0 05

1790 1490 I800

[77] [80] (971 1981 1871 [I061

[22] [22] [31] [38] [43] [45] [57] [671 1701 [77] [97] [98] [I081

29 2.91 3.65

2. 440 fl. 4.0 4.8 48 :I 3 0 4 . 0 48 48 4 4 48 48 48 3544

3985 2985 3983

4000 3880 dono 3985 3780 3880

3980 3919

3780 3877 38x0 3983

2760 2760 2700

“I

21x0

[77] [XR] (971 [98] [lox] [I061

2760 2300 -2227 2827 (per) 2760 7200 2986 (d) 2760 (dl

-2500

2700 -2200

[70]

[57] [67]

I451

[39] [43]

[31] [35] [38]

1141 [22] [ZSj

[6]

14.5

5 14 65 14.5 (XI 14.5 14.48 14.5 14 3 14 5 13 9 1448 145 14.3 I4 5 139 13.9

3 22 3.22 25-34 32

3 22 32 3.15 3.21 3.22 32

3 21 32 3.21

3.22 3.2 32 3.21 3.2

PI

(251 1321 [52] [74] [98] [I091 [I121

[Ill

[h] [lo]

[I1

(1121

[1.13.35.44] [58.68.95]

134j

12 II 14

[7] [25]

[9Ll

0092 SZOO AH 0095 AH motmooz B H 099Z AH b60Z SO0 AH 008Z AH 0562

000E-OSPZ

0562

AH 0085 0065 WH O9LZQPEZ

SO0 AH 00LE-0085 z o AH oozr AH 8865 5500 AH 007s SO0 AH OOOS SOOAHSESI SO0 AH OOZZ DOZS-00PE SO0 AH OOZS-OWE s o 0 AH ommoor 1.0AH wzr AH ooomrs

008Z

006r WH OLPZ AH 000f OS AH mzt 1'0 B H OLPZ

omz

AH 0085 SO0 AH OOZE

SO 0 AH OOLfHM8Z SO 0 WH 0085

E'O AH 008Z-0091 00Sf-0085 AH o o x r - 0 ~ 6 7 0067 OOZE AH 0062 5 AH 0085

PE'P 87

ZZP (9011 1161 0 ILL1 Z'P b L I ZEW ' YZ [LSI ZZ'P

loll "1

I91

[Ill

[ELI I9Pl

[sl [I1

[91

6'001

[LLI

L90l1 IL6l

09

[91

[Sl ZP 61'P

[a1

I9011 [LLI [LSI [or]

Z'L r'L

Z'L SSY 5'9

[LLI ILSl

8p9Z

o w

0185

triz

I6El

lor1

OERZ OLLZ OSLZ OLLZ

om

8P9Z

OW OE87 [SZl [Ell 1011 [XI

[LI L91 [Sl

"1

oLir 090s

1161 1061 ILL1 [PLl [LSI [6E1 [SZl [Oll [Ll

E6b

6P

80P X P

[PLI

[6011

SIVO

riwo zvo

91VO Z81P'O 91P'O 991P'O

[zsl

[SZI

1011

[91

"1

3A

IOZ

zoz

ooz

161

861

66 I

961 561 P6 I E61

561 161

06 I 681

881 181

9R I 581

P81

EX1 281 181

081 61 1 811 LLl 91 I SLl

6P 6'P

PLI

E6'P E6'P

LPlE oLor

f6'P r6b

ELI

L~OE

wt

6P

Z6f

190s L90E

89

091s osir

t o

1561 [861 [La]

55

LPit

r6P

PL L 9'8-8

ZS ZS

WIE

[9011 [861

5s

wit [9011 1191 ILSl

62

oz

6P

Ebb

E6 P E6'P

ZS

[Ed

ft 89

16')

16P 16P

80 P

om HOE L90E

E6'P

[srl

OL SP ZS 5'89

0x1s L6l

zr-iz

OLlf

ZS

5s

55 ZS

IZ lwpz

osir

89

P6'P E6b E6P E6b ILEl

zs

190s 190s 190s OblE oLor

IZ

(X)

ZLI ILl OLI 691 891 191 99 I 891 09 I $91 Z91 191

65

6'001

1191 [LSl [OPI [8E1

[zrl

19011 (Z61 [861 [L61 [LXI

65 65 09 951 OS1-09

9'101 6101 1'501 Z'Lll

L6tZ

"El [rrl [ril

AHOOli

5 0 AH OROZ AH O S f i Si0 0 AH OOCZ OOSi AH 0Wi SO 0 AH 000Z-00f I 000Z-00i1 AH 0011 1lZl AH 000; AH OOPZ YH onRi AH ORLl AH OROZ AH 00Oi-oOY1

OOPi IOYH0081

A H nooi 00x1 a o AH o o ~ i

lox1

[X61 [Lhl [Sh] [LR]

[YO11

iL

ii'L

iL S6 Y

96 Y

iL ii

LLLl

111

I61 [PI

[nil

[PI1

[Xi1 [iil [ti1 [El

iL

iL L

69 Y

L

LO L Li L

ir~

[I1

[PI

191

[Ll

1x1

[OII [61

[Ell

SXOZ

OOti

[XOll "011

[YO11

[xi

[861 [Lhl [Oh]

EY 51 8 51 SI Ll ZL'SI ZL'SI L'SI L SI iL'SI EY S I

L'SI L'SI L'SI

LSl L'SI

i'jl

1E l LSI LLSl

iRil

[LLl [PLI [LYI [LSI [is1 [SPI [hfl

[SCI

lit1

ktl

lit1

LL5 X'S PS

[Sil

YIP0

YSLP Oi8PhhiO 9liLPO'hOXlS 0 I LP 0;Xhi 0 SSLp'O~iOOtO i L P 01662 0

X ~ L P0'10oi'n

mi o Z Y O ~ Zn 16i n 8t8i 0 Y06i 0 iixz n ~ 0 6 io

X i 0 6i 0

LlRZ 0 Y06i 0 ltXi 0 OOhi 0 6iRi 0 YO62 0 L t R i 0 Y06i 0

[Sil L'SI [PI1 L 51 [Oll i x l R S 1 161 iLSl lLl L'SI [YI L $1 i L 51 L 51

IXOlI [LO11

lpO6i 0

it05 0 SPLS 0 LLSP 0 SSP 0 YO0 I 6P1 1 ios o PLS n LSP o

LhLSP 0

L6SP 0 Y06i 0 LLSP o i o h i n LLSP 0 ZOhi 0

bl

19011

S'P

[I1

OLLi XPYi

[YI

OlRi

SYIi

LYXZ SXLZ YLLZ OOYi

YLLZ OtLi

OOYi YLLi

[iYI OtLi [ESI (PlOOLi [Sil OELi ILl LSXi 191 0XRi*iLi (51 ooxi

nxoi

AH OWZ OOOi 110Ii 00 IZ AH O O U

1911 26 Y [Pi1 LOZ Y-Lil Y [hY] i'L[LYI 96 Y [LS! 96'9 56 Y [SP] IOPI 56 Y S'L

x-c It1

AHO81i

Its1 ILS 'I1

so o YH O S E ~

[jtl

8

idly; x-S

161 181 [YL'SP'OPI l i t '(11 'il

[Ll

[L61 [061

[YO11

[ti1

lSi1

,AH 9605 AH 0062

2x1 282 283 284 285 286

280

277 2?8 279

268 269 270 27 1 272 273 274 275 276

265 266 267

BN

AIN

Nitrides

264

262 263

hex hex hex hex hex hex hex cub cub cub

hex hex hex hex hex

[78] [Y] (141 1351

1521

111 130) (391 (431 1461

[6] 1431 1521 (741 (I121

fcc [571 fcc 1741 fcc (981 mb-BI IlOS] cub (1121

259 260 261

1461

fcc [I] cub81 [6,76] fcc 181 cub-BI [lo] cub-Bl (11,251 cub-Bl 130.521

1251

tet

zrc

fa

258

252 253 254 255 256 257

251

250


Table 1. Continued

6.73 6.63 6 63 6.73 651 6.7

0.4693 0.47 0 467 0.4693

0.251/0.669 0 2504/0.6661 0.2504/0.666 I 0 362 0.36158

3420

3535 3400 3532 3530 3530 3445

2795

Melting point ("c)

3000 3000 3000 3000 (d) 2997 2727 2300(d) 2730 2730 2730

>2200 1900 2230(d) 2300(dj 2397 2227(d) 2400 2517 2250 (d)

[74,97] 3530 (97,981 3443 [9,45,771 3420 11081 3530

[39]

[I] 161 [?I [lo] I251 [38]

2.25 2.25 2 25 2.27 2.34 3 48 3.48 2.2 3.4b3.49 (mo)(351 3.3-3.4 (p0)[35]

3.05 3 09 3.25 3.05 3.26 3.26 3 26 3.26 3 26 3.09 3.2 3.25

6.73

0.4694

0.311/0.498 0.31 IjO 494 0.31 10/0.4975 0.311\0.4979 0.311 14/0.49792

6.5 6.51 (xj 6.51 6 59 (x) 6.9 66

(gem-')

Denskty

0.47 0.46894.476 0 4698 0.4698 04685 04698


57

6.73 6.93 7.0-74 6.7 8.3 6.7

1251

[I] [5] [30] 1381 (391 [43] (71 [9] 1141

[6] [7] 1381 1391 [43] [47, 571 1731 (77,971

[5, 106)

[38] (431 [471 1571 (771 [97] [lox]

1771 197) I981 [I061

1451

[1,6] [XI 191 (101 I301 (32,481

(381 3.8 -2.8/41.5 [30] 1431 4.8 [XI2 . 5 4 7 [9] 2.5-4.7 [35] [43] 3.5 1571 14.0 (IOX]

5.7 5.3

7 57

6 5.4 48

[39. 57,901 7 a 7 . 4 I301 7-7.4 [32.38.461 7-7.4 [13.25, 1061 6.73

IS] [6] [?I [XI [9,45. 77, 97, 9x1 1101

Ill

1731

Linear thermal expansion, a (10.' K-I)

(106]

1981

[I] [XI 191 (251 1321 (381

[I] 138) 25 1431 Is0-2Ol[8142 [9] 41.9 [35] 200 [36] 42 [43] 15-33 [I081 20 [I061

25

284.7

10 1381 Il&170[43] 30 [57] 165 [I081

20.5

209

20.93 20.5 I2 20 52 20 52 19

Thermal conductivity, X (W m-' K-')

11

1661 (771 [97.981 [I061

1571

(2Sl 1321 (381

[9]

(61

1381 1571 [66] 10" [771 10" (971 1 0 ~ ~ - 1(ioq 0~~

(261 138) 1661 191 (141 lozo 1261 10'8 [?7] 10"-10!~ [IOX] 3 x lo" 10" 10'' 10" 10''

10"

10" 10"

42 42 42 42

42

50

63-156 42 42 75 43


252.5 119.3 225 252.5 252.5

288.9 290 3186 318 318 2

263.8 181 7 196.8 202.1 196.8 184.6

(51 [6] 1431 1461 [73]

[S] (43) 1461 I651 1731

(981

[I] 151 1111 (461 1731

(kJ mol-')

Enthalpiy

8 6.8 6.6

6.6 5.9 6.8 7.2

3

0.9 09

3.15 -3 3 3 437 35

3. .32& 4 02 4 4 4 3 48

38 4 4.07 3 88 3.9 4

[I] (381 1431 [9] (141 1361 [36] 1431 (571 (771

1431 1.571 [74] 1771 138,971

(771 (971 (981 [I061

[74]

[I] 191 [lo] (321 (381 [45]

Young modulus (los N miK2)

4700 HK 0 05 210(t3600 9OOC-9500 HV (mo) 7000-8000 HV (PO) 4050HK 4700HK

5000

4400HV 4700 4700HV

1230HV 25OOHV 1230HK 1 1200 1200 HV

I200

2925HV 2700 lOo(t2800

2600 HV 0.05

260C-2900 2560HV

2500

2560 2600 HV 2600

2000-2500 1900 HV

Mxro hardness (10 N mm-')

174

139,901 1591

(251

[L,571 [6, 81 [7, 10,32.45] 191 1131 113. 17.981

(621 P51 12 11-14

Oxidation resistance (100°C)

171 [25]

hcp

333 334

330 33 I 332

326 327 328 329

324 325

123

[Sl]

[II?]

fcc

[57] [74] 1761 [YR]

1521

cub

fcc

Ec cub-Bl cub-BI fcc fcc fcc cub-BI

HI%

320 321 322

[8] [lo] 1251

[25]

[6. 52.751 13.74.761 [I041 [110.112]

hex

hex

hex hen

Cr:N

[I121 [lo] [ZS] [46] 1521 [57]

l ] 13.61

[77, 781 [I121

[58, 761

[36] [39] [43] 1571

0415 04149

0.451 0 4513 04518 0 452 0 4525 0 451 0 452 0 45257

0 452 0 4526

0 4760,’O 4438 0 4759 0 4438 0 48 I I 0 4484 04759 0.4438 02750447 0 481 I3 0.44841

0414 04149 0.4148 04149 04148 fcc 0.414 fcc [60] 0.4149 fcc 174. 751 0416 cub-Bl 176. 1091 0415 fca [XI] 04148 cub [l04. 1101 0414

f c c j cub-Bl cub cub-Bl cub-BI cub-Bl cub.Bl

cub cub cub cub

cub

cub cub

317 318 319

314 315 316

313

308 309 310 31 I 312

296 CrN 291 298 299 300 30 I 302 303 304 305 306 307

287 288 289 290 291 292 293 294 295

.

I

[3] 161 [I01 [25] 1321 [52I 174. 1091

[I1

[lox]

[57] [70.77] [I061 [I071

[43.88]

1361 1361 1391

13 8 ( x ) I I 97 13 84 1197 138 13 8

13 8

59 58 6 51

59

59

6 I !x) 6I 5.9 6 12 6 12

6.1 5Y

6 I !x!

77

[XI] [I061 [IO7]

1981

[74]

ilbl

61 Ill 5 39-7 75 161

3 48 34 3 45 3 48 345 3 48 2 34 22 18

3310 3300 2700 3387 3300 1387 3928 3928 3330 3300 3300 3928 3200 3305

I500 1590

-

I050 I500

loso

I050 1450 (d) I500 I500 I500 (d)

3027 3000 2730 3000 2300

[lo] [32] 1601 23 [971 2.3 [XI] 23 [I101 23 [Ill] 7 S(XS0C- [ I l l ] -1040C)

66 6.9 6.9 6.9

66

6.9 69 6.9 69 6.6

94 9 41

23 23

23

2177 21 77

II 3

II 72 1 I 72

I101 1321

[XI

{lo] 1321

26 I5 33 56. 5

640 640 640 640

[71 (321 1571 11061

I571 1661 (771 [97]

369 4 369 369 3 300 369.3

252

308

118-124 123.1 I24 8 123 I 118 I

1121 [I71 [461 1651 [73]

[Ill] [Ill]

[Ill]

1461 [lo?] [73]

[61

2.45

33341 4 64 4.64 3.8 4.64 3.8

1811

[291 1401 1451 1741 1871

[Ill]

1741

illlj

[I201

4 4

3 138 27

[I041 11101

[60] 1741 [77] 1971 [XI]

1971

2 65-2.9

4 3.236 4 4 33

6.6

1700 2000 160!-35W 2700 H V 2030 1300 H U 0 I

2000 H V 1600 1500-3500 H V 150&8590 H V 1600HVOI 2700 1640

170CL2000

1570HV005

IIOOHUOI

2250 H V 0 01 2500HV lLCU2330 HK

810HV 175P2650 H V 0 25 2600-2900HK 0 0 5

-

1890HV001 IIOOHV 2300 HV 13284140 HVOOOOO5 70rMI00 H U 0 I 680-780 HU 0 I 100&1700 H V 0 0 I5 1300 H V 0.05

1090 1x06-2200 H V 0 05 1965~2295HV

1751 1791 1851 [1031 [1111

[I251 11251

[1201

[I041 [II51 [I181

[I031

[791 [Xh]

1771

[10.32.591 [23] [45] [571 [hO] 1641 j75j 2006-2200HV005 2606-2900 HKOO5

171

1806-2100

[881

1571 1581 [70] [761 [771

1100

8000 H V 0.2

5000 H V

200&4000

4000 HK 4700 HK 0.1 6006-8000 H K 0 I

[I251

1277P

[lo?]

8 6 3 1103 161 114 3 [46. 731

7-7.5

365 366 361 368 369 370 371 372 373 374

364

~0.78~0.56 ?: 0.75810.5623 p: 0.7610 30 1:0 77541/0 56217 p: 0.76044,O 29075

[6] (221

[25] (461

021,0.496 0.3052,’0.4964 o 3058~04961 0.305610.4948

z.P.hex

hex

0.44 0.439 0.4392 0.4375 02951,’1.1271 0.2958/1.1272 0.2958/1.1272 0.295211.125 0 298610.5548 0.2980:1.1270

0.41394.4160 0.4128 0.41554.4160 0.4163 04163 0.4184.42110.8-0.820 0.421010.8060

[I] [6. 521

1761 11121

a$ hex 143)

hex hex

357 358 359 360 361 362 363

SIN,

hex hcp hex oriho

NbN

353 Nh,N 354 355 356

352

NbN-E

[I, 81 [6] [25] [57] (61 [lo] 114.461 (74,761 1521 [I121

fcc cub-Bl cub41 fcc hex hex hex hex hex-Bl hex

NhN-6

342 343 344 345 346 341 348 349 350 351

[6, 101 [I121 I251 1521 [57,76] [6] 176, 112)

335 Mo?N-y fcc 336 cub fcc 337 338 fcc 339 Icc 340 Mo2N-B tet 341 let

No. Symbol Crqstal structure Lattice parameters (nm)

Table 1. Continued

1521 Ill21 [6]

[25l

[lo] 1101

[6]

[6]

3.2 1881

[1.106] [6] (71 [Y] 1141 3.2 1221 3.2 [38] 344 139) 3.19 [43] 2.75-2.95 [47] I 8-3 I 147) [SZ] 3.44 3.15(x) [52] 3.25 1571 32 167) 3 21 1701 3.19 (71.971

3.44 3.44 3.44 3.19 3 I8

[I] 8.08-8 62 [6] 8.33 [52] 8 31 (x) [52]

8.3

8.4 7.3 (x) 8 8.43 843 84 73

2300(d)

700(d) 527

Melting point

1900 1900

1900

1900 1900 1900 1900 1900(d) 1900(d) 1900(d) I897 1897(d) 1900

2430 2400

2300 2050 2630(d) [7] 2570 [lo] 2300 1741 2204 1771 2630 I971 2204(d) 1106, 1081 2573 [I071

[I] 8 . 2 6 8 4 [6] 8.4 125)

83

93

9.46

(gem-')

Density

~

[I061

[W

I91 1141 [221 (381 1391 1431 (571 [771

(11 [61 I71

121 [73]

[I] 18) 1251 1571 (71 191 110. 32,971 1461 i771 [I061

1571 (731

~~

-1.5 2.5 2.5 33

2.4 2.5 3 2.7-3 5 28 2.7-3.3 4.2 24 3.3 3

3.26

10.1

10.1

10 I 10 I

10.1 10 1

3.77 3.77

[lo]

1018

1018

p. 2.90

1’3.20

2.06

21

2. I 1.7-3.0

1.63 0

2.2

2. I

[6] [22] (431 [46]

4 814 4.8 4.8

[lox]

750.5 760 630 745 2

[6] [12.73] 1461

[I]

[I21 1461 (731

151

[I] 161

[741 1771 1971

~

Young modulus (10’ N rnm-’)

-3 3.1 3

1141 [22]

255.8 255.8 253.3 248 7

21 1.9 2366 234 236.6

237.8 237.8

70 3-81.6 161 69.5 (461 69 5 1731

(kl mol-’)

tnthalpiy

138) [57] 1661 (671 1771 1971

[Y] 10’8 in“-io18 10” lo6 lo1* >lo’’

[I061

1321 [66] (771 1971

1571

10”

10”-10~’

[Y] 1221 1381 1431 [57] 1671 [I081

200

58

60 58 58

[67l

10-13

20-24 17-30 I2 1&15 30 32

4 27-8.71 161

1321

[lo]

-200

[TO] (761 [77] [97] (1081

[I] [9] 1141 1221 1381 1431 (471 1571

[6]

[lO6]

[32) 1771

3 74

[8]

2 9 N . 6 1 [6]

[6]

18)

17.88

Thermal Electrical conductivity. X resistivity (W m? K-’) Ikm)

161

K-’)

Linear thermal expansion, cz ~~~

8: 2000-3000 2300 HV 0.05 3340 HV 2 0 W 5 0 0 HK 1800 HV cl.1700HKOI 25WiW0 HV 3300 1$00 HV 30 2200-3200 HV 0. I 5@3900 I720 HV 1400 HV 0.2

1410 HV 1700 1700 H K 0.05 1700 HK a 28OCL4000

2120 1620-1 820

1400 1400 HV

1400 I4on

1378-1422 1400 HV 1400

630

Micro hardness ( I 0 N mm-’) (571

Oxidation resistance ( I 00°C)

[I121

[I] [3) [6]

fcc cub-BI cub-BI

392 TIN 393 394 395 396 397 398 399 400 40 I 402 403 404 405 406 401 408 409 410 41 I 412 413 414 415 416 417 418 419 420 421

cub

[lo] 1231 [25] [41] [46]

19x1 ISl] [I091 [I121

[loll

[57] [60] [74] 1761

[52]

1x1

rcc

cub-BI fcc cub-BI fcc cub-BI cub-BI Pc fcc fcc cub-Bl fcc fcc fcc cub-BI

0 423 0.4238 0.423 0 424 0 423 0.424 0 424 0 4245 0 423 0 424 0 4241 0 424 0 42 0 42 0 422 0.424173

0 3010 493 0 30476 0 49187 0.3048/0.4919 0.3048 0 4919 0 3044510 49141

[6.10l 146) [14,52]

hex hcp ortho hcp hex

387 Ta2N 388 389 390 391

[21

0 52 0.29 0 5185,10.2908 0.5185/0 2908 0.495/0.305 0 518510 2908 0 5 18510 2908 0.519210.2908 0 51Y18\0 29081

377 T d N 378 379 380 381 3x2 383 384 385 386

375 376

[2Y] 1321

[I] [3] 161 [lo] [25]

[In]

[2] [6] [lo] 1521

[Il>]

[6] [lo] (251 [32] 1521 [74]

[I1

5.21 5.4 52 5 43 5 21 54 52 5.4 5.4 5 43 5.21 5.21 54 5.22

54 4 73 5 43 5 21 5.3

5 21 5 21 5 21 54 5.39 (x) 5 21 5 21

1 8 I5 46 15.46 15 86(x)

13.613 8 14 35 (x) 16.3 14 34(x) 14.1 13.6 13.8 163 I6 3

[97] [98] [92] [XI] [I061 [I071 [lox]

1881 [90] [91] [93) 1961

1521 1531 [57] [60] [74] [77]

(451

171 [9] [lo] 1251 [37]

[I] 161

2950 2950 2950 2950 2949 2950 2950 2900 3205 2930 2950 2950 2950

2930-2950 2949 2950 2950 3205 2900 7.950 2950

2950 2950 3220 2950 2930 2950 2949

3000 2700 2727

[I] [6]

1521 [52]

3000 3090 3360 2090 3093 3093 2950 W30(d) 3360

[11 [6] [6] (7.741 (lO,52[ [25,107] [52] (881 [lO6] [I081

34 [lo?] 2.4-3.44 [lox]

[92]

[Y8]

187) [90] [91] (931

[88]

I571 [60] [61] [62] 1761 1771

[46]

(291 132) [37] 1381 (391 [40] (451

[9) [lo]

[6] [7] 181

[?I

[I]

121 [46] 1131

111 [5,25.88] [l] [8] [lo] 1321 (461 (571 [I061

9 35 94 9.35 9.4 9.35 9.35 9 35 93 94 93 93 9 35 9.35 8.3 8.3 9 35 94 93 94 94 8 339.4 9 35 83 9 35 9 35

36 5 3.6 36

29.2 38 30 19 70

29.31 28 9 30 19.26 19.26

10 05

8 58 96 8 79 8 79

Ill [XI 19) [lo] 1321 [37] [38] 1571 [9l] [96]

[6]

[lo] [32]

[S]

[6l 161 [25] [32] 1571 [84]

25 25 21 7 25-30 75-500 25 30 25 25 17 25 25 25 25 25 25 I7 22 [27] 1321 [38l [40l (571 161) [66] [77] [87] I911 [97] (981 [92] (1061

[??I

[7l 191 [ZS]

263 161 19&250 [84]

128 135 128 I28 13&250

336 6 303 I 336 2 338 I 336 336 6 337 333 338.1 336.6

270 9 270 9 270 5 273

225.7 243 1 25 I 252 5 252 5

[I] 191 [I?] (321

1741

1371 [38] 1401 6 1451 25 [53] 59 1601 2 56 [61] 256 1621 2 51-6.40 [741 59 [77] 2 56 [SO] 2 . 7 9 4 5 1 [82] 59 [87] 2.66640 [XY] 5.9 [91] 5.9 [93] 2.6 [95] 59 [Y7l 59 [98] 25 [92] 4 1811

25 59 2.51136 2 56 2 56 2.6 59

5 756

2450 HV 180~2800 2400 HV 2500-2800 2000 2000-2500 HV 0.05 1-6000 HV 0 001 2000 HV 0 015 220&2800 HV 0 05 1770 HK 0 I 2300-2600 1170HK0.1 2450 HV 30 1994 HV 2100 2200 HV 3300 HK 0.05 2300 160&2000 HV 2000-2700 HV 1900-2400 2450 HV0.I I050 HV 0.1 200&2400 HV 0 05 170C-2800HV 0 05 15OC-2700 HV 0 05 1725-1825 HV 0.02 1775-1935 HV 1800 HV 2160 HV

3wo I220

3240 1060 1500-3000 1300 HV 1000 1060 1150-3000 3230 HV 0.2

80&2000 HV

5-8

[25]

[6,52,57] [10,83] [25,74] [I121

fcc

W,N

ZrN

448 449 450 45 I

452 453 454 455 456

fcc [I] cub-BI [6] fcc [8] cub-BI [lo] cub-BI [25]

cub fcc cub

[6] [10,52] [I121

hcp hex hex

[1,8] [6.25] [46,52] [57,74] [3,10] [76,109] [I121

[6] [41,112]

WN-6

fcc cub-BI cub-BI fcc cubdl cub-BI cub

tet

let

445 446 447

444

436 VN 437 438 439 440 441 442 443

434 Ti,N-E 435

422 423 424 425 426 427 428 429 430 43 1 432 433


Table 1. Continued

0.46 0.458 0.459 0.4577 0.456

0.4126 0.4118 0.4128 0.4126

0.2893/0.2826 0.2893i0.2826 0.2893/0.2826

0.41 0.4137 0.414 0.4126 0.4169 0.4139 0.413 0.413916

0.494/0.303 0.49452/0.30342


[I] [3] [6] [lo] [25]

6.93 7.09 7.32 7.3 (x) 6.93

[10,52,74] 12 ~251 [831 [I121

[I] 6.13 [6] 5.62 [3,10,32] 6.04 (x) PI 6 [52] 6.04 [74] 6.11 [lo91 6.13 [I121 6.13 6.1

[I011 [I121

(g cm-’)

Density

[25]

[lo]

[61 [7] [9]

ilOSl

.

[25,52] 177.971

[lo]

[7.74] [6] 161 ..

3000 2980 2980 2980 2982

2050 (d) 2050 2177 2350 2350 (d) 2177 2050

2947

Melting point (“C)

[Il [5.7,8] [25.38] [39,90] [6.9.10]

6 7.9 7.2 7.24 7.24

8.1 [5,6,7,8,25,73]8.1 [10,32] 8.1 (461 9.2 9.2 [571 I77.971 8.1 SO61

[!I

[I061

[32]

[lo]

[I] [8] [9]

[32] [77] [97] [I061

[lo]

[XI

Linear thermal expansion, c( K-‘)

16.75 10.9 28 20.52 20.52

11.3 11.3 11.3 11.3

[32]

[lo]

[I] [8] [9]

[lo] [32]

[XI

[6]

Thermal conductivity, X (W m-’ K-’)

13.6 21 21 13.6 21.1

IOl&180

200

85

85-100

[6] [7] [9] [25] 1321

[83]

[6] [32] [57] [66] [771 [97l [I061


365.5 336.2 365.5 368.4 366

147.8 217.3 217.3 217.3

[I] [5] [I21 [46] [65]

[6] [46] [73]

[S]

(kJ mo1-l)

Enthalpiy

5.1 5.1 3.93-1.60 5.1 5.1

4.6 4.6

[9l I451 [74] [77] [93]

[77] [97]

Young modulus (10’ N mm?)

2000 1300-2000 1900 HV 1600 I500

2 5 W 0 0 0 HV 0.05

1520 1560 HV

1500

1520 140CL1600 1500 HV

3 0 W W O HV 240k2700 HV 0.01

1666-3098 HV 153k2730 HV 0.01 2450 HV 0.2 2100 HV 1994 HV 2300 HV 0.03 2500-2800 HV 0.05 2200 HV 0.05 180k2800 HV 0.05 500-1200HU 0.1 20W3350 HK 750-1200 H U 0.1 1640 HV 0.05 2225-2290 HK 0.01 22W2400 HV 0.01


111 [7] [XI 19.591 [10.32]

[23l

1 2 [ 7 1 11-14 [25]

(100°C)

Oxidation resistance

hex

478 CrBl 479 480 48I 482 483 484 485

493 Fe,B 494 495

490 FeB 49I 492

[I] (61 1521 11121

[I] [6] [25] I521 I571 [I121

[6] [25] 1521 [I121

tet

let let

ortho

I521 [I091 [I121

[I 121

ortho [52] ortho [I091

tet

tet tet let

hex

hex

hex hex

hex

ortho ortho ortho ortho

486 Cr$, 487 488 489

1521 [I121

[52] [I 121

mono 1521

tet

tet

hex

hex

cub

[52] 1571 1741 [76] [I091 [I121

cub-BI 1461

cub-BI fcc fcc cub-BI cub-BI

474 CrB 475 476 477

469 AIB,, 470 47I 472 473

467 AIB? 468

Borides

461 462 463 464 465 466

4Ml

457 458 459

0.279I0.307 0.296910.3066 0.296910.3066 0.296910.3066 0.29730/0.30709

0.296910.7858/0.~932 0.296910.785810.2932 0.296910.785810.2932 0.29663/0.78666, 0.29322

1521 1771

(521

[38] 1391 1451 i74;106] [77] [90] [93] [97] [in81 [in71

-7.0

6.3

6.1 6.12 6.14

5.6 5.6 5.6 5.6 5.6 5.58 5.58

1521

[52]

[I] [6] 1251

[1.106] [6,107] [7] 1251 1571 1771 191

[6] [25,52] 6.11(x) 1521

6.05 6.05

2.58

2.57

3.17(x)

0.3@86/0.3252 0.30054;0.325276

O.lZ58jO.lO20 0.1030/0.1433 1.01611.428 1.884 i.704~1.100~ 0.8522~1.100/0.7393

7.3 7.09 7.32 7.09 7.32 7.09 7.32 7.32 7.09 7.4

0.4575 0.4577 0.4562 0.4578 0.463 0.457756

1389

1550

1650

1900 1960

2170 18OC-2300

[46,61.73]

[5,46,73] 1611

[I] 1251

[I] [5] [6] 2150 [7] 1850 [25.106] 2200 [571 1850-1900 1731 2188 177.971

2200

I550

1550

[6] [25] 1731

[77]

2150(d)

I550

[5] 1461

1461

[32,45] [77.93,97] 1291 (463 157) [I061

2070 (per) 2070(per)

1975

2982 2982 2930-2982 2950 2955 2980

7.1

10.5 10.5

11.1 11.1

6.5 7.24 7.2 7.2 7.2 7.24

[I] 1571 1771 1971 [I061

1381 1451 [77] 1931 1971 11061

19

[38]

1661

1381 1571 1661 1771 1971 [I061

18 I8 21

56 21 56 I8

[7] [25] 1571 [66] 1771 1971 [I061

1251

2 x 10’2 1771

10”

14

21

21

21

20 21

71.2

71.2

125.6

94.6 129.8 -94.2 94.2

-75.4 75.4

201

67

365.5

46.73

5.46.73

[I]

[5] [6] 1461 [73]

1461 1731

[46]

[46]

1731

I

2.15 5.4

2.15

4.3

5

[I] 1571 1771

1771

[97]

14M1190

18on~~o.i 1800 2250 HV 2100

2200

1700

2250

2600 HV

a: 2400 HK

1250HV 0.1 1500-2600 I600HV 2500 HV 0.05 2130 HV0.05

1520

2300

I510 HK 0.1 1520 HV

171

177.971 [591

WI

[71 [251

161

(1.571

1771

1581

1 6 1 8 [25]

cub cub

LaB6

hex hex hCP hex

518 MOB> 519 520 52 I

hex hex hex hex

tet

Mo,Br

Mo,B2

526 527 528 529 530

531

tet

te1 let

tet

Mo,B

522 523 524 525

[I]

[ I I?]

[52]

[I] 1251

[I] (251 152, 1091 (1121

[25, 521 1571 [I 121

[I]

1251 1521 [I121

te1

tet let

[52] [55]

cub cub

ortho [25,52] ortho [57] ortho [I121

[IOY] [I121

125) 1521 [57] [I121

508 LaB, 509 510 511 MOB-6 512 513 514 515 MOB-p 516 517

505 506 507

hex hex hex hex

498 HIB? 499 500 %I 502 503 504

0.60/0.31

0.3/2.1 0.3011/2.093 0.3011/2.093 0.30 I I74/2.09369

0.55/0.47 0.5543/0.4735 0.5547/0.4739 0.554/0.474

0.3/0.31 0.305/0.3113 0.306/0.310 0.304/0.307

0.4156 0.4145 0.4156 0.31 10/1.695 0.3110/1.695 0.3105/1.697 0.3108/1.697 0.316!0.861/0.308 0.316/0.861/0.30X 0.316/0.845/0.308

0.415 0.41569

0.3141/0.3470 0.3141/0.3470 0.314245/0.347602

0.462 0.6517/0.3218/0.4919

cub-Bl [52] ortho [I121

496 497

HfB



Table 1. Continued

Ill

[I1 1251 1521 [I 121

Ill 125.521 [I121 11091

ill21

[I1 [251 1521

1521 1551 1551 1251 ~521 1521 [I121 [25] (521 [I 121

11121

IW

1251 1521 [I121

[521 [I121

[571

8.5

9

7.2 7.48 (x) 7.45 7.12 7.8

9.2 9.1 9. I 9.31 (x)

[I]

[52] (77,971 [I061 “071

[I]

[I] [25] [521 [521

[I] [7] [25]

[7] 1391 [77] [ZS] [52] [52]

7.8 7.12 8

3200 3380 -3060 3370 3250 2147 3240


2240

2300 2140 2140 2100

2140 1850

2100 2300 2250

2530 2200 2770 -2500 1930 2350

[YO] 2530 [l06,lO7l 2200 I1081

[7] (25,1071 [52,108] 1521 1571 [I061

4.73 4.76 4.73 8.3 8.77 (x) 8.3 8

4.76 4.7 2.61

10.96

10.5 11.2(x) 11

11.2

11.01

(g cm-’)

Density

8.6 8.6 5

5.84 6.4

5.3 5.7 5.3

IW

[77] [97]

130) [77]

[571 19x1 [I061

K-’)

Linear thermal expansion, u.

430

1981

Thermal conductivity, X (W m-’ K-’I

25

18

18

40

45 30

50

50

17 15

28

I0 15.8 12

1771 [97] [ID61

[25]

[57l

[25]

[25] (571

171 1771

(1061

1251 1571 (1061


175.8

209.3

106.8

96.3

336 6 336.2 336.2

[I]

[I]

[I]

[I]

[5] [46] (731

(kl mol-’)

Enthalpiy

6.7 6.7

(771 [97]

Young modulus (lo5 N mm-*l

2300

3200 2350 HV 2350 HV

2500 1660HVO.l 166C

l38DHV D.l 3000

2500

1400-2500 2770 HV 2530 HV 1570 HV 0.1 1570 25w

Z770 HV

2800 2900 2900 2850 400M6000 HVO.O1


Ill

1771 P71

“I

~ 5 1 1331

111

[571

(27

[591

I71 1-19] 1771 1251 P31 1571

Pol

[ill

171 [571 [59 PSI

11-14 125)

11-17 1251


Td!B

Ta,B?

Ta,B,

TiBl

561 562 563

564 565 566

567 568 869 570 571 572

[I] [b. 251

[52]

11091 [I I?]

tet

11121

hex hex hex hex hex hex

16) [25] [30] 1311 .. [43]

[I]

ortho [6. 251 ortho [52] ortho [I121

[I] 161

tet tet

tet

let

!el

tet

[I] [6] [25] [52] 1571 [I121

[82l [57] [I121

let


ortho [I121

SiBh

TaB,

hex her hen hex hex hex

NbBz

[I] [a] 1251

ortho [6.25] ortho [52] ortho (1121

NhB

556 557 558 559 560

553 554 555

547 548 549 550 551 552

541 544 545 546

542

540 54 I

539

536 537 538

535

534

513

532

0 310 32 0 302810 3228 0.3026’0 3213 0.3028’0 3228 0.303034 0 322953 03

0 329 I 4010 313 032914010313

0.62’0 33 0 6184101284 (I 6184 0 3286

058048 0 5778’0 4864 0 577810 4846 0 578 0 486 0.5783:O 4866

0.31 0 33 0 3085 0 3249 0 3078/0.3265 0 309803 0.322660

1.4470 I835009946

031’033 0.3099,0.3271 0.3086 0 3306 0 3096,0.3306 0.311133~032743

031883

0 3298 0.X724 0 3166 032Y2!0.8713.03165 0 32973 0 87229

[I] [6] [25] [52] [74] [YX]

4.5 4.38 4 38 45 44 45

21 35 10 89 109

[6l [25] [I061

[73]

[5] [6] [46]

3.7 56 5.6 37 4.8

68

2 62

3.3 33

-2.6 63 63

[I] [9] [31] [38] 1401

161 1771

[77] [97]

[6] [77] [97]

2200 2615 HK 0 I 2500 210&2400

1910 245&?800HV 2300 H K 0 I 2400-2800 H V 0 I m m HV

2600 130&2600 1800-1900 2600 H V

[1.107] [h. 1061 [71 [9. 1081 [25] [31]

[6. 521 16.521

2900 2900 2920 3230 3225 3225

6 39 78 4,8419.83 7.8 7.4

54

[I] 191 [30] [31] 1381

27

16) [9[ [25] 1381

7

9 7 7 10

1401

[71 191 [31] [38]

305

1507 324 I 280 324.1

[I] 151 [43] [46] 1481

300&3400 250&3500 3400HVO05 3000 H V

3480

3350

2770 2770

2200 2430HV

2500 H V 2000-2700

2200

209 3 193 8 209 3 209 3

[731

(I] IS] [46]

1920

[I]

150.7 175.8

251 2 251 2

[I]

1251 1571 1771 [I061

[hb] [77] [Y7]

1151 1571 1771 (971 [I061

2200

2lOOHV

82

2596 26 13 26-39 26.13

68 ?I 14 68

82

51 7 88

10’

10’

32

I?

32 I? I?

10’

[I] [6] 1571 [771 [I061

1675 [I] 16 75-35 2 [6] 16 75 [?5] 16 7 (1061

54

83

[6] [77] [97] (1061

[6]

3100 3150 3037 3000

3150 3100 3040 -3000

1950 1947 1900 1900

h?

7 1’96

I67 5

[25] [521 (521 [771 [106. 1081 [I071

171

[1,87] [h]

[I] 16. 1081 [3Y] (701 1771

3000 3000 3036 (35,521 -2900 [52] 3036 [77.97] 3036 1106, 1081 2900

(I] [6] [7. 571

1518(x) [bl

I52

1?.6(x) 12.58 II I5 I2 6

I2 I (x) I2 58 11.7 11.7

243 243 2.43

247

243

6.98 697

7.21 in)

66

68 72

(6.25. 521 13 5 [I121 I36O(x)

[I121

(521 [Il?]

I61 [28]

“I

[h. 251 [5?) /I121

191 1251 131.97.981

171

[I]

1571

[11 [61

161 1471

“I

[771

[1.57] [61 [7. 591 [281 [471 [761

[70] (77. 971

15x1

[11 [39]

1251

11-17 [25] 13 [43l

11-14 [25]

[1.6.57.59]11-14 [71 [281 1771

[61

609

605 606 607 608

W,B

tet tet tet tet tet

1251 1521 (1091 [ I 121

[I]

[I] 1251 i 5 i 1121 p: ortho 1251 ortho 1521 onho (57)

te1 6: tet

let

598 WB 599 600 601 602 603 604

[I] 161 (251 1521 [57] 11121

1521

tet

her hex hex hex hex hex

p51

let

589 VB, 590 89 I 592 593 594 595 596 597

[I]

tet

586 Ti,B 587 588

1521 [57] (741 1981 11121

hex hex hex hex hex

573 574 575 576 577 578 579 580 581 582 583 584 585

0.56047 0 556410.4740 0.5564/0.4740 O.S56!0.474 0 556810.474

031L7 0.311511.693 0.3115/1.693 0.319/0.840/0.307 0.319/0 840/0.307 0.311655/1.69101

0.310.31 0.2998/0.3057 0.2'99810.3057 0.2998jO.3057 0.29976110.305620

0.61'0.46 0.611/0.456 0.61010.453

0 303034,0.322953 0.3018-0.3023/ ... .../0.3194 322

No. Symbol Crystal structure Lattice parameters (nm)

Table 1. Continued

[I 121

IW

1251 1521

Ill

1251 152) [I121

Density

16.5 16 16.72 (x) 10.77

15.5 15.3 15.3 16.0 (x) 15 5 15.7 10.77

5 10 (x) 5 5.05 4.92

48 4.56 5.1 5. I 4.61

4.5 4.5 4.52 4.5 4.52 4.38 4.5 4.5 4.5 4.4

(g cm-3)

[52l (1061

[I] 125,521

i52j [52] 1571 [I071 I1081

2770 2670-2780 2710 2900

2860 2685-2920 2860(d)

2400 2450 2100 2400 2747

2200

3225 2980 2980 2980 3197 3325 2790

Melting point ("C)

[I] 161 1251 [I061

[I. 251 161 .. 1571

L5,1 161 [25] (571 177,971

171

[77,97,98] [39,40.76] [87,90] 1401 [43] 1451 [I061

4.7

5.3 5.3 7.6 76 6.1

8 7.8 6.618.6 6.39 8 7.8 8 7.8 7.8 5.9

(10.'

[I061

IW

[431 1451 (481 [571 1761 1771 1871 1971 1981

K-I)


24-56 26 9 26 I

[48] 1981 (1061

Thermal conductivity, X (W m-' K-')

21-43

9 I 7 7 7 7 15, 5

[I061

1511 (66) (771 [871 1971 1981 [I061

Electrical resistivity (lo-' Qcm)

203 9

324.1 150.7

(731

1731 1981

(kJ mol-')

Enthalpiy

5.1 5.1

4.8 5.6 3.74 4.8 5.6 4.8 5.6 5.6

1771 1971

[I61 171 1x71 1971 [98]

1571

I451 1481

Young modulus (10' N mm-')

[I]

111 1571

3750 3750

2350

[I, 571 (61 171 i5Yl 1771 1971

111

1591 (871 1881 (1231 [I191 [I271

157

1331 (39,90,77] 140. 761 I451 1471 1481

2080 2077 13@2100 2100 2150HV 2120 HV

2500

3400 3300 HV 3370 3000 2900-3700 HV 25W3000 3480 3300 3370 HV 34x0 HV 0.2 164631 IS HV 0.05 IR43HV0.05 2 2 W 9 0 0 HVO.O1

Micro hardness ( I 0 N rnm-,)

1251

I71

8-14[25]

13 8-14

Oxidation resistance (100T)

WB:

ZrB

cub

cu h

Cr1Si

650

648

649


CrSi?

cub cuh cub

cub cuh cu h

447

646

635

644

640 64 I h42 443

635 CrSi 636 637 638 639

cub cub

cuh

Silicides

634

cub

fcc cub

ZrB,?

hen ha hex hex hex hex hex hex

cub-Bl

fcc cub-61

hex hex hcp hex

631 632 633

620 62 I 622 623 624 625 626 627 628 629 630

619

618 ZrB,

615 616 617

611 W?B< 612 613 614

610

[II?]

“I

I11

IM

[I1

0 455W.4564 0 4558

0.455 0.45W 0.4555

0 44?:0 655 0.443 l/0.6364 0.442210.635I o 4427’0 6375 0 4420,O6349 0.44281 0.63691

0 462

0.46074 4629

0 462

0 4619

0.462

o 7408

0 14 0 7408 0 7408

0 32)O35 0.3170 o 3531 0.316910 3530 0 1170~0.7533 0 32 03l6870~0353002

0 J7 0 468 0.465

03’1.3Y U.?YX? I S87 0 2982 1.387 O:YX~?O~~S

6.7 in)

1391 (471 1521 177. 971 1981

6.17

6.45

6.52

1521

(61 1251

IW

6.52

(1071

5.5

[SZl

1251

[61

[6]

lW

4.91 4.4

491 4.4

5 38

3.7 3.63

3.7

(1)

IW

3 65

I901

61 6 0x5 5 64 (1081

PXl

6.11 6

6 17

-6 6 09 ( X I 6 II

ps. 521

1710 1770 1750 1770

1630 1451 1570 1457

1550 1550 1475 1457 (pa)

3245

2992 3250 2990 3200

[I, 38. 571 3000 [6. 71 3040

6.17

61 5 64

[I] I521 LQI

3000

2365

[J7,9?)

65 57

29x0 2200-2980

(521 [52]

[I]

13 I II 13 1 IX) 13.03

2920 2300

[I)

1513

[I] 151 [61 146. 63. 73)

1731

I251

161 146)

1731

(61 [63]

[I]

6.83 6 I2 “4 6.8 171 59 [25. 38. 57.881 6 83 (461 59 [77,97. 981 59 5.9 55

(1.30. 1061 [S. 39.901

177. 97

78

PI 161

78

[I,571

[7]

I 2 I4

1251

!I 19 IY [9i]

1571 1771

105.5

92.1

105 5 105.5

100.5 80 80.1

54.8

53.2

502.4

1633

146) (731

IS]

[I]

[I] 1461 1731

[73]

1461

[I]

54

54 54

3.5 3.5 3.5

77 7.7

1771 [97]

1005

90&980

1100 xw110 I IS0 HV 0.1

1000 95&1050

2500 2500

2200 2600 22Ml HV 0 05 2252 HV 2200 HV 965 H V 0 I 2300 2300-3000 2300 H V 2200 H V 0 2

3600 3600

2700HV

2700 2700

2600

171

0.96410.49 0.96210.490 0.96483/0.49135

[I.1121 1461 1521

Mo,Si,

NbSi,

Nh&

Nh&

670 67 1 672

673 674 675 676 677 678

679 680

681 682 683 684

tet

tet

p: tet

3:tet

p: tet

01:

hex hex

hex

hex hex hex hex

cub tet

[I] 161

p: l.OOl0/0.5070

a: 0.658311.1884

p l.Ml0/0.5070

3:0.6570/1.1884

0.36j0.50 0.35910.446

[28, 521 [I121

161 [6] 152,721 152,711

0.4810.66 0.4785/0.6576 0.4797/0.6590 0.479506589 0.47971/0.6592

[I] 161 1251

0.489 0489 0.489 0.489

[I] 1251 1461 152, 1121

cub cub cub cub

MolSi

666 667 668 669

7.8

1521 [I121

III

5.29

1251

B: 7.?O(xJ [6]

a: 7.09 (x) 161

[6] [6]

111 161

[losl [I071

7.8 8.01

5.37 5.6

161 171

Ill

Ill

1521

111 [25] (521

[1,107] 17, 1081 1221 1251 [3Y, 901

111 161

1521 [lli]

161 251 1281

[I1

11121

5.5 5.69(x) 5.7

8.4 84 8.97

1251

1521

8.8

Ill

11121

IIW

6.3 6.31 6.2 6.12 6

161 1521 1521

[I]

151 [63]

2480 2480

[I] [7] 1251 (521

1950 1950 1950 1930

1950

[I] [SO] 1731

[I] [25] 1731

2100 2180 2190

2150 2050 2025

2190 (per)

10.78/17.05 (301

Linear thermal expansion, 01 K-‘)

0: 7.3/4.6

8.4

161

161

[I,7,22.30] 8.4 [I] 1221 [25,39. 73,901 8 1501 5.32115.55 1301 1631

2050 2030 2020

[I] 122) 1251 [28] 1521

tet tet

tet

tet

tet

0.32/0.786 0.32/0.65 1).320/0.786 0.3205/0.7848 0.3203:0.7886 0.320/0.786 0.32047/0.78449

[I] 1221 [25.28] [30.52] 1461 [I091 [I121

tet tet

MoSi2

659 660 66 I 662 663 664 665

[30] 163)

-1520 > 1500

11121

0.4281/1.375

1301 11121

tet tet

LaSi2

657 658

1461 [73]

1647 I647

5.9 5.6 5.9(xJ

1251 1521

0 9170/0.4636 0.9 I70,’0.4636

[6] 1521

tet tet

Cr&

654 655 656

[I]

111

1560

5.6

111

0.919/0465

[I]

tet

Cr,Si

653

(61

151&1610 I550

(g cm+)

[6] 171


Density

5.6 4.7


Cr3Si2

~

651 652

No Symbol Crystal structure

~~~~

Table 1. Continued

221.9 50

122)

[I]

Thermal conductivity, X (W m-’ K-’)

6.3

21 20 21.5

1420

1251

125)

[71 1221

[7]


452.2 452.2

35.6-67.8 138.2 138.2

50.2

[I]

280.5 310.2 309.8

151 (731

[I] [6] (461 (731

146. SO] 1731

[I] [5] (461 I731

[I] 1461 [SO] 1731

1461 1731

100.5 116.4 116.4 116.5

108.9 131.9 131.5 131.8

211.4 223.2

(kJ rno1-l)

Enthalpiy

3.84 4.4

[I] [22)

Young modulus (10’ N mm-*J

550

700 66h1320 H V lOSOHV0.1

1170

1310 1310HV0.1

1581

17,221 [251 (39. YO1

III

“I 1280

1290 1300 1290 H V 0 I 1200 H V 900HV0.1

I61 I71

10Sc-I2W 1000


17171 > 17 I251 17 I1001

Oxidation resistance (IOO’C)

1521

[ll?]

tet

tei

V5Si,

WSil

721 722

723 724

[I]

(61

let

tet

03210788 0.32 0 78

TI 094?8104750 0.94276'047555

1521 [ll?]

cub cub cub cuh

717 VISi 718 719 720

lei

047 0 4712 0 4721 047253

[I] (61 (25. 521 [I121

hex

0.46!0 64 04562,06359 0 457310.6374 0 4571/0 6372 0 45723010.63730

0 75 0 52 074650516? 0 7465 0 5162 0?44805141 07444,05143

[I] 161 [25] 128. 521 [I121

[I] (251 130.521 (I001 [IIZ]

(221

..

0823604773'08523 (h.251 0 8263'0 4800'0 8553 12x1 0 8253,O 4783'0.8540 1521 0 8267 0 4800,O 8551 11001 0 82687 0 85534 0 47983

hex hex hex hex

VSI!

hex hex

hex hex

her

urtho (521 ortho [IOO] ortho [IIZ]

ortho (251 ortho 1281

TI~O9820,05010 T206513,I 1864 0 6516, I 1873

TZ.0 6516 I 1873

712 713 714 715 716

TI,%

TiB!

707 708 709 710 711

706

702 703 704 705

701

700

1521 Ill21

tet tei

let

(61 [6] (461

let let

o 9880:~5060

[a]

let tet

TI

061,050 0615705039 06157,05039 06157~05039

[I]

iei

Ta,Si

692 693 694 695

696 T a , S 697 698 699

0 4773'0.6552 0 4773,0.6552 0 4783106567 0.4781,O 6564 0 47835110.656980

161 [25] (28.521 146) (I 121

hex hex hex

686 687 688 689 690 69 I

hex her

I1 6569Rll.lX87

]I121

tet

TaSi2

685

L

1

[loo]

[I]

?I20 2120 2130 -2150

1480

-Is50

(1001

125. 391 [I071

1520 1470 1540 -1527

[90] 17. 1081

9.5 9.4

[7: 1081

Ill

55 [I] 5 33-5 67 161 567 [52]

[??I

45

1221

154 9

95 I25 6 125 6

577 8 579 579 9 579. I 579 5

1344 I34 3 134.4

335 4 334 9 334 9

83.7 126 4 125 6

1507 1193 1192

2165 2170 [7]

11. 391

10

125

92 I 92 9

462 2 462 6

97

95

18 123

I08

I24

8.5

38 46

2150 2010

6.5

7.03 15 61 1301 9.2 [I001

T2 5 5'8 0 (61

161

1507 I50 7

(5.631 1461

1631 1731

8 Y18.8

2060 2030 (di 1935

45 Ill I650 4 3 4 4.42 [h] 1680 57 171 1759 471 (25. 1071 442 jl08l

43 4 32

4 39 4 4 39 44 407

2500 2500

2450

[7] 2200 (61 2400 (251 2200 152. 1081 (52] [107]

13 5 [I] 13 54 (xi (6. 521 I? 4

8.4 (A) 9.1

9.14

883

Y.O?(x)

92

1090 1100

I500 1410-1560

960 89G960 IZO&II?OHV 1090 HV 0 I

986 986 HV 0 I 968HV I

892 HV 700 870 HV 0 I 870 HV I

I500

I200 1560 HV 0 I I560

1410

171

(35. 901

111

1251 1331

161 17)

II [?I 8-1 I 125)

11 171 11-14 1251

tet tet tct tet

hex onho ortho

ortho ortho ortho ortho

tct tet

bl

hen hex

733 ZrSi 734 735

736 ZrSi, 737 738 739

140 Zr2Si 74 I 742

743 Zr5Si, 744

745 Al?O,-a hex 746 hex 741 rhom 748 rhom 149 rhom 750 hex 751 hex 752 rhomihex 753 hex 754 hcx/rhom 755 756 757 758 759 760

Oxides

tet

729 W& 730 73 I 732

0.5127 - a = 55'16.7' 0.512 0.513 --u = 55.3' 0.554410.9024 0.47588i/1.2992

(52, 1121 0.7885j0.5558

0.7886/0.5551

0.66/0.54

0.661210.5294 0.66oS/0.5298

ju~j

I1091 I1121

1571

1521

[52] [I121

1521 [I121

Ill

0.372/1.41610.367 16,251 0.3698/1.4761/0.3665 (281 0.3721/1.468/0.3683 152, 1001 0.36958/1.4751/0.36654 [I121

[46, 521 [I121

[46]

(521 Ill21

Ill

1281 1521 [IIZ]

1251

0.7005/1.2772 1521 0.6982/0.3786/0.5302 [52] 0.6981!0.3785/0.5301 [I121

[I]

[6] [25,28] [52, 1001 [I121

[ I 121

[SZ]

146,521

[I] 161 [5?1 [I121

0.561,10.496 0.96410.497 0.960110.4972

0.3212/0.7880 0.321310.7829 0.321 1/0.7868 0.3211/0.7829

I221 [25,28] 146,521 [IIZ]

725 126 727 128

tet let let


No. Symbol Crystal structure

Table 1. Continued

Density

3.99 3.98 3.9 3.99 3.8-3.9 3.98 4 3.8 399 3.96 (x) 3.98 3.96 3.97 3.95 3.99 3.96

6 6.22

4.87 4.88 4.83 4.9

5.65 (x)

14.56 13.06

9.8 9.3

(g cm-9

[I,731 (99,631

2090 (461 2107(per) (731

2320 2350

[I] 152)

?: 8.5

7.2 8.4 8 8.5

9

2015

8 8.6 7.2-8.6 7-9 9.5110 8.3 9 8 7.7 6.8

2047

1461

9.7

1561 .rj7i. 1621

1531

I91 [I61 (301 1381 1401 1431 [ q

[a]

[31

[loo]

K-')

Linear thermal expansion, u

2040 2050 2050 2054

2043 2045 2030 2047 2015 2050 2050 2050 2041 2300

2150

2110 11,461 2107(per) I731

[6] 1700 [25] 125, 1081 1520 [I001 [lo01 1517(per) 1731 11071

[52]

[I] (100]

1221 203W2W I221 125, 107) 2150 [25.46] 2160 1731 2210 1901

Melting point ("C)

35 27-36

30.1

30. I 27-36 4.2-16.7 25 27 2s 27 28

[I211

(1061

I381 1431 [57] [h3] 167) 19x1

1161

[XI (Y]

Thermal conductivity, h (W m-' K ' )

161

33.4

[25]

1251


Enthalpiy

1676.4

1690

1580.I 1590 1678.5

614.2 576.1

309.8 339.1 208.5

159.4 151.1 159.5

141.8 154.9

125.6 134.4 135.2 134.6

92.8

151 (431

I731

1461

(731

[46]

[I]

(IOO] 1631 173)

1461 1731

1131

146)

151

[I]

[73]

(kJ mol-I)

4 3.7 37 4

4 4 4 X.? 2.5 5.2 4

4 4 3.6 3.5 4. I

2.348

(h71 [hY] 1701

16-31

(h?]

ph]

(531

[.I31 [A51 IS31

1401

2020HK 18WHV 2100 ZOSOHV 1200-1600HV 250&3000 HV

28WHV 2100

1800

IXW?XK! 250W3000 HV 0.05 ?Ow HK 0.1 2800

I251 1381

2100HV0.05 2200 HV ZIOOHV

1230

1030HV0.1 942 HV 1

770

I200 1090HV0.1 1074 HV

Micro hardness (10 N m K 2 )

[7] (91 1161

[IMI]

Young modulus (10' N mm-')

8-11

[ZS]

> 17 1251 18 1431 [8. 77,87,98] 20 1701

1331 133. 691 1341

I91

141

PI

111

IW

(251

Ill

[221 1251 1391


Cr,O,

800

802

801

Hf02

mono

tet

tet

0 512

0 87210.750 0.6145 0.755

1521

1521 [II?]

1521

9.7

0 4544

cub

Cr,O

799

[I121

[7]

[I061

1521 [52] 1531 1601

[6] [7] [39]

5.21 5.2 5.21 5.22 5.25 (x) 5.41 5.4 5.21

I521

[6l

0.536-%=55’ 0.5361 -rr=55’ 0.495876/1.35942

rhom rhom hex

Cr,O,

791 792 793 794 795 796 797 798

[6] 1521 [52]

2.81 2.7 2.82 (x)

0.573/0.852/0.474 [6] 0 574310.8557/0.4789 1521 0 57494,’O 8556’0.4796 [I121

ortho ortho ortho

CrO,

788 789 790

[52]

tet

tet

Cr02

186 787

(521

0.441,0?91 04421’0 2916

hex hex

Be0

780 781 782 783 784 785

4.8 490(x)

[I211

1521 [I121

iiosi

[52l

0 269910.4401

[77]

188) (981 (92) “5] [lO6] [I071

(251 L39.771 1431 1521 1521

lll2l

1521

0.3958 0.79 o 774.8

3.99 3.9 3.9

4

3.98 3.9 398 3.98

3 3 03 301 3.01 3 00(xj

A120,-? cub (svtnellj [521 777 fcc 778 cub cub-El 779

761 762 763 764 765 766 767 768 769 770 771 772 173 774 775 776

-2900

2440 2300 2400 2343 2343 2708

170-198 185

2450 2530 2570 2567 2580 2550

2047 2046 2045 2045

2050

2050 2300 2047 23M)

15.471

1106)

[a]

[7] [39] [46] [53]

[46]

[6]

[?5] 1391 [43] [45] 1771

IS]

10

5.6

6.7 6.7

9 9

[43l

[60] [I061

[53]

1431 1771

3

264

[43l

(431

(661 (771

5 x 10’’

[I061

10” [6] 1.3 x lo9 [I061

10’’

lo?’

1053.4

1130.4 1140.5 1130.4

579 9 589.9 578.6

582 8 582

608.8 599.1

580

569

PI

[43] (461

[S]

3 3.9 3.9

4 4 3.8 4 38 4 2.5 294 400

[43] [77]

[?5]

(121)

I761 1771 [82] [871 [95] 1981 [92] [I061

ZIOOHVOI

2000

900

1000 HV 2915 HV 2300 HV 12W1700HV 12W1700HV0.1 2000 1200-1700 HV0.05 I100 HV 0.02

1230-1490 HV IS00 HV

2200

1510HK

2600-2800 HV

2000 HV

20W2300 HV 2080 H K 0 I 2000-2500 HV 1200-1600 HV 0 1 2300 20w2500 855-975 HV 0 02 2000 HV 30 15OLL2200 HV 30 2000-3000 HV 0 I 23OO HV 0 2

[591

[64]

[60]

1591

1531 IS81

PI

1341 1531

S10

817 818 819 820

821

Ta,O,-u

tet tKt

cub

t ~ t

CriStoballtK:

trigonal

quartr-

mono

cub-Bl cub

838 839

Tho,

fcc cub

830 83 I 832 TdlOr-p ortho 833 ortho 834 ortho 835 Pd!Ol ortho 836 837

829

828

822 823 824 825 826 827

SiO,

Nb!O,

813 814 815 816

806

cub-El cub-El

cub

MkQ

807 808 809 810 811 812

805

mono cub mono

[6] 143)

161 (461 [5?] [II?]

[6] [52]

I61 [IOU]

[I091

[I121

[I121

[IOU]

[43] [461 [52]

[46] [IW] [I121

Symbol Crystal structure

803 804

No

Table 1. Continued

[6]

[52]

05595

[I121

0 6198 4 029 0 3888

05859

[6] (521

[52]

[6] [IOY]

[6] [I091

[I121

1521 [IOY] [I121

0 619 0 367 0 389 0 6192 4 4019 0 I898

0.3808 0.3567

cristobalite 0.70,0.69 0.704

quartr 0.4093,’O 5393 0 421.’0.539

2893/03827;1.758

0.4208 0.42 04213

0.51 [I091 0 5281~0.51819~0.5115[I121

Lattice parameters (nm) [43] I771 [I061 [I071 [7]

2850

1460-1520 1510 IS12 1785

2837 2825 2827 3073

[I051 (1061 [I071

[6] [7] 1521 [99] [6] [52]

[6] [7.105] [47] (471 11061 [I071 [I081

3250 2997-3250

1880 1880

1785 1900

1880 1887 1880

1983

1722 1373-1473 1713

1713

1703-1729

17, 1071 1750 [I051 1610 [106,108] 1705 1975

[7] [I061 [I071

[391 [43. 531 1771 (106. 1081

[25, 1071 2800

10.05 (x) [6] 9.7 [7]

83 7.53 8.7

8.37-9.48 75 8.53 (x) 82 8 18-891 8.30(xJ

162.7 2 1-2.2 2.2 2.6 2.18

2.2

233

2.13

21

2.1

4.5 447 4.6

3.6 3.5 3 65 3.58 3.77 3 58

2790 2897 2810 2790

(“C)

(g cm-7

9 68 10.2 9.63 96

Melting point

Density

6.5

[5] [a]

11061 (IOS]

[6] (461

(51 [h] [71

[6] [7] [46] [73] [IOS] [I061

[71 [46] [I051 [I061

[61 171 (731 [I061

9.3

10

2

2

05-075

04

1.76

[5] [7.25.39. 531 (431 II 2 [46. 731 [77] [lob]

[7] [43] [46] [I061

1431 [77]

[ 1061

161

161

[I061 [I081

[6]

[43] [771 [I061

1771

K-’)


10 10

I38 12-1.4

36 36

1431 [I061

[I081

[I061

[43] [I061

Thermal conductivity, X (W m-’ K - ’ )

10’‘

1016

10”

10.’

10;’

10” 10” 10”

[6h] 1771

[h]

[hh] [I061 H IIIS]

[66] [77] [I061

ncm)

Electrical resistivity

I I71 I 11711

2045 3 2047 3 ?047 3

1953 I

911 911.5

98.4

1905.8 1903 3

568 6 570 601.6 601.6

11138

1010 1113.7

[5] (431

[h] [46] [73]

IS]

[46] [73]

[461

161 [73]

[5] I431 [46] [73]

[43] 1461 [731

(kl mol-’)

Enthalpiy

I 3X 14

1.114 0 654.75

0.5-10

3.2 3.2 2.4

[h] I251

[82] [I051 [I081

[43] [771 [I061

Young modulus (10’ N mm-’)

950 H V

66&1030

113&1260 1200 IOW1?00H V

740 700

745 H V 520HV -400HV 520 H V 0. I 750 H V

780 H V

Micro hardness (10 N nim-’)

1991

IS?]

1591

161

1771


tet tel

cub cub

878 879 880

a76 877

bcc

Ill

tet

mono

111

b€C

Y?O;

wo,

874 875

870 87 I 872 873

mnnn

1461 I521 [IOY] [I 121

[6] [46. 521 1521 11121

1.06 1.04041

1.0601

0.382 0.748)O 728 0.3835/0.7517/0 7285 0 525010 39 I 5 07309107522:07678

0 55610.555 0.55754~048995.' 0 55608

0 4870,0.2776

[521 [I091 [I121

[6] [52] [52] [I121

[52] 11091 [I121

(521 [IOY] 11121

tet tet

867 868 869

[52]

16.46, 521 0 5560,O 4884,O 5546

WO?

866

[II21

mnno

0.9828'0.3776,0.9898

1461 11121

mnno mnno

ThoI

864 865

0 5454 - x = 59 5' 0 5139,l 3659

[I121

rhom rhom rhom

TI~O,

1461 [52]

[II?]

IW

861 862 863

860

859

856 857 8%

cub-BI

854 855

let

0.449,'o 2959 045933'0?9592

0 4543:O 2959

[4h] 1521 [25]

tet

Ti0:

851 852 853

850

0417 0418 0424 04177

16.461 [II?] [52,76] [IOY]

cub-Bl Cub cub-BI cub-Bl

Ti0

847 848 R4Y

0.521 0.5597

[46] [52] [ I091 [II?]

fcc fcc cub cub

840 841 842 843 844 R45 846

5 4.5 5.03 4 84

7 16 7 16

11.05

1913

1840 1867 1860 1860

IXbO

1900 1855 1867

1700

1750 2020

3300 3323

2990 3050 3217 3370 3220

[6]

[7. 1051 1381 [I081 [I061

[hl [7. 1061

2465 2415 2450 2704 2420 2410 2683

1473 1473 1472

1724 (per) 1724(d)

[7] [38] [46] [73] [I051 [I061

151

161 17. 1061 [46]

I461 [?)I

8 81

[6. 7. 32. 1051 7 6 I461 11061

1500-1600 161

1780

[5?, 1051 2130 [52] 1842 [I081 1760

[7] [Yl 1411 [53] [77.98] [88] [I061 [105.107] [I241

[6l

161 [7. 1051 [I061

[I071

1251 [43] [??I [I061

1521 1082(x) [52]

I? I

4.05

4.6 3.58 1x1

4 19 4I 425 2.40-2.49 4 16 425 42 424 42 3.02-3.92

493

4 88 49

Y7 10 10 YY 10

[iXl [lox]

[I061

14 XI?

0.6

8 5

II

[I081

1381

[I061

[Y] [lox]

[I061

[32]

[I061

[lox]

(1-1. 5 ) x [hi 10"

IO'Y-IO:'

I.2 x lo"' [I061

2 x 10"

1906 7 IY067

843 4 843 5

1731

1461

1461 1731

I461 I731

590 I

590 I

[61

[73]

1461

[I] [46.73]

I461 I731

[??I

1461 1731

587

2461 24608

1433 I IS?I 9

945 4 945 4

520

1227 5 1227 2

I8

2 05-2 80 2 05 0 8-20 205 065 I 4 9

24 24 148

[38]

11241

[98]

[Y] 1771 [X?]

[431 [771 [I061

YXO HV

600 7on 1000 HV 1) 2 II50 HV

767 I000 HK 7oo I Ino 1~IOOHV 600 1000 HV 600- 1000 HV I1 I I100 HV

1100 1800 I OIIO

888 889 890

tet

mono

1521 [I091

[52]

1521 161

0.51463iO.52135iO.5311 [IlZ] 0.364/0.521 [IIZ]

0.521/0.5209,'0.5315 0.522Oj0.5211;0.53XI 0,508410.5165 0.508

0.511

1391 1521 (39,431 146,521 1521 [I091 16, 1121 [I121

881 Zr02 882 883 884 885 886 887

cub cub mono mono tef cub


No. Symbol Crystal structure

Table 1. Continued

5.6 17, 1051 5.7 19, 881 5.8 (251 5.56 138,431 5.56 (mo)[39] 6.21 (cub) 1391 6.1 153, 1081 6.01 1671 5.76 1771 5.6 [I061

(B cm-7

Density

2150 2681 2611 2690 2680 2700 2681 2700

Melting point ("C) [S] [7] (9,43,53,77] 125, 391 138,461 I881 11061 [IOS]

7.5- 10.5 I0 9-10.5 10 10 17.2

[9] 1381 (431 1611 1771 [I061

Linear thermal expansion. c( (lo-' K - ' ) 0.7-2.4 2 2 2.Y 1.7

[9] 1381 (431 1671 [I061

Thermal conductivity. X (W in-' K - ' )

10"

loi6

10"

10l6

(91 1381 (661 [77]


1101.5 1098.2

1035 1040

IS] [43] (461 1731

(kJ mol-')

Enthalpiy

1.63 1.8 1.7 2.4 2.05 2.1 1.9

[38l 1431 [67] [771

1251

[h] [9]

Young modulus (10' N mm ')

150&1650

1200 I200 HV ISOOHV IOOOHV 1500 HK 0.1 1600 1300 HV 30 400-1000 I100 HV 0.2

Micro hardness (10 N m m - * )

PI 139.771 1531 1581 1581 I591 1671 1761 1881 11161

17

1431

> 11 [ZS]


References

99 1

References I . R. Kieffer and F. Benesovsky, Hartstoffe. Springer, Vienna, 1968. 2. H. Holleck, Binare und ternare Carbide und Nitride der Ubergansmetalle und ihre Phasenbeziehungen. Gebr. Borntrager Verlag, Berlin, Stuttgart, 1984. 3. U. Konig, Untersuchung von kathodenzerstaubten binaren und ternaren Hartstoffschichten zur VerschleiBminderung von Hartmetallen. Tribologie Reibung, Verschle$, Schmierung. Band 9, Springer, Berlin 1985, pp. 282, 287, 296, 306, 31 I , 316. 4. H. Kolleck, Die Bedeutung von Phasengleichgewichten bei der Entwicklung VerschleiBfester Werkstoffe. Chem. 2. 1982, 106, 216. 5. M. Riihle, Zum technischen Stand der Dispersionshartung - Teil I . Metall. 1982,36, 128 1-1285. 6. Gmelin: Handbuch der anorganischen Chemie. Verlag Chemie: Weinheim 1960. Angaben in den jermeiligen Bande der Elemente (z.B. Cr, Ti, 7. R. A. Haefer, OberJachen- und Diinnschichttechnologie Teil I: Beschichtung von Oberjfachen. Springer, Berlin, 1987, pp. 282-294. 8. H. Frey and G. Kienel, Diinnschichttechnologie. VDI-Verlag, Diisseldorf 1987, p. 121. 9. B. Rother and J. Vetter, Plasma-Beschichtungsverfahren und Hartstoflschichten. Deutscher Verlag fur Grundstoffindustrie, Leipzig, 1992, p. 171-172. 10. L. E. Toth, Transition Metal Carbides and Nitrides. Academic, New York, 1971, pp. 5-7, 8081,92,95-97 and 148-150. 11. E. K. Storms, The Refractory Carbides, Academic, New York, 1967. 12. H. L. Schick, Thermodynamics of Certain Refractory Compounds. Vols. 1 and 2, Academic, New York, 1966. 13. E. Hornbogen, Werkstofe: Aufbau und Eigenschaften von Keramik, Metallen. Polymer- und Verbundwerkstoflen. Springer, Berlin, 1987, p. 286. 14. H.-J. Bargel and G. Schulze, Werkstoffe.CDI-Verlag, Diisseldorf, 1988, pp. 293-294. 15. V. Demarne and E. Bergmann, Verwendung von Hartstoff- und Hartmetallschichten fur Schneidwerkzeuge. In I . Int. PVD-Tagung an der TH Darmstudr am 15.116. Murz 1983, Vol. 20, ISBN 3-88607-027-1, THD Schriftenreihe Wissenschaft und Technik 20, 1983, pp. 269-277. 16. K. H. Kloos et a/. Herstellung und Eigenschaften gesputterter Aluminiumoxidschichten. In 2. Int. PVD-Tagung an der TH Darmstadt am 11.112. Marz 1986, Vol. 30, THD Schriftenreihe Wissenschaft und Technik, 1986, p. 100. 17. H. Jehn et a/. Beeinflussung der Schichteigenschaften mangetrongesputterter Nitridschichten durch Substrattemperatur und Sputteratmosphare. In 2. Int. PVD-Tagung an der T H Darmstadt am 11,112. Murz 1986, Vol. 30, THD Schriftenreihe Wissenschaft und Technik, 1986, p. 138. 18. J.-E. Sundgren et a/. Structure and Properties of HfN Coatings grown by Reactive Sputtering. In 2. Int. PVD-Tagung an der T H Darmstadt am 11,112. Marz 1986, Vol. 30, THD Schriftenreihe Wissenschaft und Technik, 1986, p. 185. 19. P. Johannsen, Beschichtete HSS- und Hartmetall-Werkzeuge in der GroBserienfertigung. In 2. Int. PVD-Tagung an der TH Darmstadt am 11.112. Marz 1986, Vol. 30, THD Schriftenreihe Wissenschaft und Technik, 1986, p. 192. 20. U. Kopacz and H. Jehn, Harte- und Haftfestigkeitsmessungen an Nitridschichten auf Schnellarbeitsstahl. In 2. Int. PVD-Tagung an der T H Darmstadt am 11.112. Marz 1986, Vol. 30, THD Schriftenreihe Wissenschaft und Technik, 1986, pp. 217-218. 21. H. R. Stock and P. Mayr, Hartstofheschichtung mit dem Plasma-CVD-Verfahren. In 2. Int. PVD-Tagung an der T H Darmstadt am 11.112. Marz 1986, Vol. 30, THD Schriftenreihe Wissenschaft und Technik, 1986, p. 250. 22. A. Petzold, Physikalische Chemie der Silicate. Deutscher Verlag fur Grundstoffindustrie, Leipzig, 1991, p. 182. 23. D. Kammermeier, Charakterisierung von binaren und ternaren Hartstoffschichten anhand von Simulations- und Zerspanungsuntersuchungen. Fortschritt-Ber. VDI Reihe 2 Nr. 271, VDI-Verlag, Diisseldorf, 1992, p. 2. 24. R. F. Bunshah, Deposition Technologies for Films and Coatings. Noyes Publications, New Jersey, 1982, p. 149. -

992 25. 26. 27. 28. 29. 30. 3I . 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54.


R. Kieffer and P. Schwarzkopf, Hartstoffe und Hartmetalle. Springer, Vienna, 1953. S. u. Z. Marinkovic et al. Boron Nitride Coatings. Kernforschungsanlage Julich GmbH, 1989. S. u. Z. Marinkovic et al. Titanium Nitride Coatings. Kernforschungsanlage Jiilich GmbH, 1988. B. Lonnberg, Synthesis, Structure and Properties of some technologically important Carbides, Borides and Silicides. Acta Univ. Uppsala, 1988, pp. 17-21, 26 and 30. T. Wolf, Herstellung und Charakterisierung von TiN, ZrN und HjN. Dissertation Universitat Karlsruhe, 1982. H. Bleckmann, Rontgenographisclze Bestirnmung der therrnischen Ausdehnung an verschiedenen metallischen und nichtmetallischen Hartstoffen. Dissertation RWTH Aachen, 1971. G. Hilz, Zum EinfluJ innerer Grenzflachen auf Aufiau und Eigenschaften mehrphasiger Hartstoffschichten. Kernforschungszentrum Karlsruhe GmbH, KfK 5022, 1992. G. D. Brundiers, Herstellung, Aujbau und Eigenschaften von Hafniumverbindungen irn System Hf-C-N-0. Kernforschungszentrum Karlsruhe GmbH, KfK 2161, 1975, p. 7. P. Schwarzkopf and R. Kieffer, Cemented Carbides. Macmillan, New York, 1960, pp. 138 and 214. A. R. Lansdown and A. L. Price, Materials to Resist Wear. Pergamon Press, Oxford, 1986, pp. 7 and 99. R. Reinhold and S. Becker, Superharte Schneidwerkstoffe der Spanungstechnik. VEB Verlag Technik, Berlin, 1982, p. 17. W.-E. Borys, Vergleichsuntersuchung zum Einsatz hochharter polykristalliner Schneidstoffe beim Frusen. Dissertation Universitat Hannover, 1984, p. 3. W. Konig and H. U. Schemmel, Untersuchung moderner Schneid.ytoffe - Beanspruchungsgerechte Anwendung sowie VerschlezJlursachen. Forschungsbericht des Landes NordrheinWestfalen, Nr. 2472, 1975, p. 54. F. Loffler, Eigenschaften von keramischen Hartstoffschichten auf Silicium- und Aluminiumbasis. Fortschritt-Ber. VDI Reihe 5 Nr. 180, VDI-Verlag, Dusseldorf, 1990, p. 23. R.-J. Peters, Beschichten mit Hartstoffen. VDI Technologiezentrum Physikalische Technologien. VDI-Verlag, Dusseldorf, 1992, pp. 2-3. H . Freller and H. P. Lorenz, Kriterien fur die anwendungsbezogene Auswahl von Hartstoffschichten. VDI-Verlag, Dusseldorf, 1992, p. 22. C. Ribeiro, Silbernes TiN durch Magnetronsputtern und seine Anwendungsbeispiele. VDIVerlag, Dusseldorf, 1992, p. 124. E. Erturk and H.-J. Heuvel, Neue Schichtsysteme mit dem Arc-PVD-Verfahren. VDI-Verlag, Dusseldorf, 1992, p. 254. F. Thummler et al. Fortschritte mit neuen Werkstoffen: Keramik fur den Maschinenbau. Keram. Z. 1988,40, 158-159. K. Dettling, VerschleiDfeste Werkstoffe fur Extrusionswerkzeuge. Keram. Z . 1988, 40, 802. R. Riesenberg et al. Mechanische Eigenschaften von Schichten. Wissenschaftliche Schriftenreihe TU Chemnitz, Heft 9/1990, p. 53. 0. Kubaschewski et al. Metallurgical Thermochemistry. Pergamon,. Oxford, 1979, pp. 268ff. C. E. Morosanu, Thin films by chemical vapour deposition. Thin Films Sci. Technol. 1990,7, pp. 225,229, 381, 391 and 424. R. Telle, Boride eine neue Hartstoffgeneration? Chemie in unserer Zeit 1988, 22, 95. P. Hedenqvist, Evaluation of vapour-deposited coatings for improved wear resistance. Acta Univ. Uppsala, 1991, p. 13. T. C. Chou and T. G. Nieh, Anisotropic grain growth of d-NiMo. Thin SolidFilms 1991,219,61. G. Berg; E. Broszeit, C. Friedrich, W. Herr and K. H. Kloos, Grundlageneigenschaften und VerschleiBverhalten von HfB,- und Hf(B,N)-Schichtsystemen. Mat.-wiss. u. Werkstoftech. 1994.25, 175-179. W. B. Pearson, Lattice Spacings and Structures of Metals and Alloys. Pergamon, London, 1958. DD. 219ff. H. Simon and M. Thoma, Angewandte Oberjlachentechnik f u r metallische Werkstoffe.Hanser, Munchen, 1985, pp. 103, 108 and 131. G. Berg, C. Friedrich, E. Broszeit and K.-H. Kloos, Comparison of fundamental properties of r.f. sputtered Ti,N-and Hf,N coatings on steel substrates. Surf. Coat. Technol. 1995, 74-75, 135-1 42. -

Rejerences

993

55. H. Bolmgren, Synthesis and structural studies of some refractory borides. Acta Univ. Uppsala 1992, p. 19. 56. N. P. Cheremisinoff, Handbook qf Ceramics and Composites. Dekker, New York, 1990, p. 24. 51. W. Schedler, Hartmetali fur den Praktiker. VDI-Verlag, Dusseldorf, 1988, pp. 3, 8-9, 12 and 207. 58. K.-H. Habig, VerschlegundHurte von Werkstofen. Hanser, Munchen, 1980, pp. 125, 164,211 and 263-268. 59. H.-D. Steffens and W. Brandl. Moderne Beschichtungsverfahren. Lehrstuhl fur Werkstofftechnologie der Universitat Dortmund, 1992, pp. 242 and 261. 60. J. H. Mittendorf, Einsatz von Werkzeugheschichtungen,furnusgenlahlte Umjormverfahren. Dissertation RWTH Aachen D82, 1992, p. 123. 61. H. Benninghoff, Industriereife CVD-Verfahren. Metalloberfliiclie 1976, 30, 476. 62. E. Erben et a/. CVD ein modernes und vielseitiges Beschichtungsverfahren. Metall 1981, 35, 1256. 63. M. Ruhle, Zum technischen Stand der Dispersionshartung Teil 2. Metall 1985, 39, 520, 522 and 528. 64. K.-H. Habig, VerschleiB-Schutzschichten - Entwicklungstendenzen zur Optimierung von Eigenschaften und Verfahren. Metall 1985, 39, 915. 65. H. Holleck, Neue Entwicklungen bei PVD-Hartstoffbeschichtungen. Metal1 1989, 43, 622. 66. S . Hofmann, Charakterisierung keramischer Werkstoffe mit der Auger-Elektronenspekund Werkstofftechnik 1990, 21, troskopie: Moglichkeiten und Grenzen. Material~~,i~~senschaft 99. 67. H. Kolaska and K . Dreyer, Hartmetalle, Cermets und Keramiken als verschleiBbestandige Werkstoffe. Metall 1991, 45, 227 and 232. 68. W. Wahl, VerschleiBfeste metallische GuBwerkstoffe. VerschleiBfeste Werkstoffe. Vortruge der VDI-Taguiig Stuttgart 1973, VDI-Berichte Nr. 194 (1973), p. 82. 69. G . Schumacher, Hartmetall fur den VerschleiBschutz. Vortruge der VDI-Tagung Stuttgnrt 1973, VDI-Berichte Nr. 194 (1973) p. 113. 70. E. Gugel, Nichtoxidkeramische Werkstoffe fur die VerschleiBtechnik. Vortruge der VDITagung Stuttgart 1973, VDI-Berichte Nr. 194 (1973), p. 139. 71. B. Finnern, Konstitution, Verfahrenstechnik und Anwendung verschleiBfester Diffusionsschichten. Vortriige der VDI-Tagung Stuttgart 1973, VDI-Berichte Nr. 194 (1973), p. 219. 72. K.-H. Kloos, WmkrtqfJkzmdeIV, Vorlesungsumdruck TH Darmstadt, 1994, pp. 2.4.2.5f. 13. I. Barin and 0. Knacke, Thermocliemical Properties of Inorganic Substances. Springer, Berlin, 1971 and 1973. 74. H. Jehn, G. Reiners et a/. Charakterisierung dunner Schichten. DIN-Fachbericht 39. Beuth, Berlin, 1993, pp. 134, 155 and 171. 75. K. Kashigawi et a/. Chromium nitride films by radio-frequency reactice ion-plating. Vuc. Sci. Technol. 1986, A4, 210 and 214. 76. J.-E. Sundgren et ul. A review of the present state of art in hard coatings grown from the Vapor phase. Vac. Sci. Technol. 1986, A4, 2260ff. 77. H. Holleck, Material selection for hard coatings. Vac. Sci. Techno/. 1986, A4, 2663. 78. W. Gissler et a/. Mixed phase nanocrystalline boron nitride films: preparation and characterization. Thin Solid Films 199I , 199, 1 15. 79. R. A. Andrievski et ul. Structure and microhardness of TIN compositional and alloyed films. Thin Solid Films 1991, 205, 173-174. 80. N. Zhang and V. Wang, Dislocations and hardness of hard coatings. Thin Solid Films 1992, 214, 5. 81. C. Friedrich, G. Berg, E. Broszeit and K.-H. Kloos, X-ray diffractometry analysis of r.f. sputtered hard coatings based on nitrides of Ti, Cr, Hf. Surf. Coat. Technol. 1995, 74-75, 279-285. 82. E. Vancoille et a/. Mechanical properties of heat treated and worn PVD TiN, (Ti, Al)N, (Ti, Nb)N and Ti(C, N) coatings as measured by nanoindentation. Thin Solid Films 1993,224, 170 and 175. 83. S. D. Marcus and R. F. Foster, Characterization of low pressure chemically vapor-deposited tungsten nitride films. Thin Solid Films 1993, 236, 331 and 332. -

-

994


84. Sun, Xin et a/. Properties of reactively sputter-deposited Ta-N thin films. Thin Solid Films 1993,236, 350. 85. W. Heinke, A. Leyland, A. Matthews, G. Berg, C. Friedrich and E. Broszeit, Evaluation of PVD nitride coatings, using impact, scratch and Rockwell C adhesion tests. Thin Solid Films 1995, 270,431438. 86. A. Schroer, Synthese und chemische, mikrostrukturelle und mechanische Charakterisierung von Hartstojschichten ayf Stiihlen. Dissertation Heidelberg, 1992. 87. J. Berger Beschichten mit Hartstoffen. VDI-Verlag, Diisseldorf, 1992. 88. E. Dorre, Nichtmetallische Hartstoffe, in Abrasion und Erosion, H. Uetz (Ed.), Hanser, Miinchen, 1986,433437. 89. C. Biselli and L. Chollet, Young’s modulus of TIN and TIC coatings, in Mechanics of Coatings. Tribology Series 17 D. Dowson, M. Godet and C. M. Tayloer (Eds), Proccedings of the 16th Leeds-Lyon Symposium on Tribology, Lyon, Amsterdam, Elsevier, 1990. 90. P. Mayr and H. Vetters, Gefiige, Oberflachenzustand und technologische Eigenschaften von Hartstoffschichten. H T M 1993, 48, 281-287. 91. R. Wild, Hartstoffschichten leisten Widerstand. Metalloberflache 1990, 44, 355-361. 92. R. SaB, and P. Thienel, Oberflachenbehandlungen von SpritzgieRwerkzeugen. Metallober,fliiche 1991, 45, 313-322. 93. R. Rochotzki and J. Vetter, Festkorperreibung von PVD-Hartstoffschichten: TIN,, ZrN,, TIC,, TiC,/i-C. Neue Hiitte 1989, 34, 272-271. 94. R. S. Bonetti and M. Tobler, Amorphe diamantartige Kohlenstoffschichten. Metallobery%iche 1990,44, 209-211. 95. K. Keller and F. Koch, CVD-Beschichtung von FlieRwerkzeugen. VDZ-Zeitschrift.1989, 131, 42-50. 96. M. Fieber and W. Konig, Kontaktbedingungen beim Einsatz beschichteter Werkzeuge. Tribologie + Schmierungstechnik 1995,42, 135-140. 97. G. Kienel (Ed.): Vakuumbeschichtung 4 und 5, Anwendungen Teil 1 und 2. VDI-Verlag, Diisseldorf, 1993. 98. B. W. Matthes, Abscheidung, anwendungsbezogene Prufung und Optimierung von PVDHartstofieschichtungen des Systems Ti-B-N/Al2O3aus der Hochfrequenz-Kathodenzerstaubung. Dissertation, TH Darmstadt, 1993. 99. P. J. Martin et al. Properties of thin films of tantalum oxide deposited by ion-assisted deposition. Thin Solid Films 1994, 238, 183-184. 100. R. Rosenkranz, Strukturen. physikalische und mechanische Eigenschaften der hochschmelzenden Ti5Si3-,TiSi2- und ZrSiz-Phasen. Fortschritt-Ber. VDI Reihe 5 Nr. 291, VDIVerlag, Diisseldorf, 1993. 101. E. D’Anna et a/. Synthesis of thin films of semiconductor and refractory metal nitrides by laser irradiation of solid samples in ambient gas. Thin Solid Films 1992, 218, 224-225. 102. J. L. Davis et al. Thin metal films on polyether imide. Thin Solid Films 1992, 220, 221. 103. G. Berg, C. Friedrich, E. Broszeit and C. Berger, Development of chromium nitride coatings substituting titanum nitride. Surf. Coat. Technol. 1996, 86-87, 184191. 104. C. Friedrich, G. Berg, E. Broszeit and C. Berger, Measurement of the hardness of hard coatings using a force indentation function. Thin Solid Films 1996, 290-291, 216-220. 105. A. G. Balzers, Coating materials and sources selector guide. Company information. Balzers AG, FL-9496 Balzers, Fiirstentum Liechtenstein. 106. A. G. Leybold, Berechnungsgrundlagen fur Diinnschicht-Technik. Company information 01 017.9.70.61.02.21/KT 5. 107. Cerac Inc.: Advanced special inorganic materials. Company information 1995. Cerac Inc., PO Box 1178, Milwaukee, WI 53201-1178, USA. 108. Goodfellow: Firmenkatalog zu Metallen, Keramiken und anderen Materialien. Edition 19951 96. Deutsche Niederlassung: Goodfellow GmbH, Postfach 1343, D 61213 Bad Nauheim. 109. K. Schafer and C. Synowitz, Taschenbuchfur Chemiker und Physiker. Bd. 3,3. Aufl. Springer, Berlin, 1970. 110. W. Herr, B. Matthes, E. Broszeit, M. Meyer and R. Suchentrunk, Influence of substrate material and deposition parameters on the structure, residual stress and adhesion of sputtered Cr,Ny hard coatings. SUTJCoat. Technol.,1993, 60, 428433.

References

995

1 1 I. M. Atzor, Aspekte des Magnetronsputterns zur Herstellung verschleifl- und korrosionsbestiindiger Schichten auf Chrombasis. VDI-Fortschrittsbericht Reihe 5 , Nr. 156, VDI, Diisseldorf, 1989. 1 12. International Center for Diffraction Data Ponder Diflraction File, ASTM, Swarthmore, PA 19081, USA, CD-ROM 1992. 113. S. Luridiana and A. Miotello, Spectrographic study of oxide growth of arc evaporated TiN and ZrN coatings during hot air oxidation. Thin Solid Films 1996, 29k291, 289-293. 114. P. J. Kelly, 0. A. Abu-Zeid, R. D. Arnell and J. Tong, The deposition of oxide coatings by reactive unbalanced magnetron sputtering. Sur$ Coat. Techno/. 1996, 86-87, 28-32. 1 IS. C. Gautier. H. Moussaoui, F. Elstner and J. Machet, Comparative study of mechanical and structural properties of CrN films deposited by d.c. magnetron sputtering and vacuum arc evaporation. Surf. Coat. Technol. 1996, 8 6 8 7 , 254262. 116. M. S. Wong, W. J. Chia, P. Yashar, J. M. Schneider, W. D. Sproul and S. A. Barnett, Highrate reactive d.c. magnetron sputtering of ZrO, coatings. Sur-: Coat. Techno/. 1996, 86-87, 381-387. 117. M. H. Staia, E. S. Puchi, D. B. Lewis, J. Cawley and D. Morel, Microstructural characterization of chemically vapor deposited TIN coatings. Surb Coat. Technol. 1996, 86-87, 432&l37. 118. A. Igartua, J. Lauricirca, A. Aranzabe, T. Leyendecker, 0. Lemmer, G. Erkens, M. Week and G . Hanrath, Application of low temperature PVD coatings in rolling bearings: tribological tests and experiences with spindle bearing systems. Surf: Coat. Techno/. 1996,86-87, 460-466. 119. S. Heck, T. Emmerich, I. Munder and J. Steinebrunner, Tribological behaviour of Ti-Al-BN-based PVD coatings. Surf. Coat. Technol. 1996, 86-87, 467471. 120. K.-D. Bouzakis, N. Vidakis, T. Leyendecker, 0. Lemmer, H.-G. Fuss and G. Erkens, Determination of Fatigue behaviour of thin hard coatings using the impact test and FEM simulation. Surf. Coat. Technol. 1996, 86-87,467A7 1. 121. F. Fietzke, K. Goedicke and W. Hempel, The deposition of hard crystalline A120, layers by means of bipolar pulsed magnetron sputtering. Szrb Coat. Technol. 1996, 86-87, 657-663. 122. G . DErrico, R. Chiara and E. Guglielmi, PVD coatings of cermet inserts for milling applications. Surf. Coat. Technol. 1996, 86-87, 735-738. 123. J. Steinebrunner, T. Emmerich, S. Heck, I. Munder and R. Steinbuch, A novel impact tester operating at elevated temperatures for characterising hard coatings. Surf: Coat. Techno/. 1996, 86-87, 748-752. 124. 0. Anderson, C. R. Ottermann, R. Kuschnereit, P. Hess and K. Bange, Density and Youngs modulus of thin TiOz films. Fresenius J A n d Chem 1997, 358, 3 15-3 18. 125. B. Navinsek, P. Panjan and I. Milosev, Industrial applications of CrN (PVD) coatings, depopited at high and low temperatures. Surf. Coat. Techno/. 1997, 97, 182-191. 126. J.-R. Park, Y. K. Song, K.-T. Rie and A. Gebauer, Hard coating by plasma-assited CVD on plasma nitrided stellite. Surf: Coat. Techno/. 1998, 98, 1329-1335. 127. E. Kelesoglu and C. Mitterer, Structure and properties of TiBz based coatings prepared by unbalanced D C magnetron sputtering. Surf. Coaf. Techno/. 1998, 98, 148-1489.


Index

a-modification, silicon nitrides 753 a-SiAION structure 755 ff, 768 ff ah initio pseudopotential approach 256 f Aboudi model 72 abrasive applications, diamond materials 528 ff abrasive wear - silicon nitrides 784 - tungsten carbide-cobalt hardmetal 962 abrasives 5 ff - silicon carbides 688,736 absorption bands, CVD diamond 577 absorption coefficient, diamond-like carbon 638 acetylene 629 Acheson process 683,688 f acid stability, boron carbides 855 acidic solutions - CVDdiamond 415 - silicon carbides 735 activated sintering 877 activation energy, self-propagating synthesis 357 active corrosion 141 ff additives - boroncarbides 843 - cemented ternary borides 925 - diamond synthesis 498 adhesion - diamond-like carbon films 640 f - silicon carbides 722 f adiabatic combustion temperature 323 agglomerates, alumina-based ceramics 654 aggregates - diamond 380 - directed metal oxidation 307 aggressive environment 584 alkali halide gas 43 alkali stability, boron carbides 855 alkaline earth nitrosilicates 29 alkoxides - polymer-ceramic transformations 447 - thin film processing 463 allenes LI allotropes XLIV, 27 1-285,485 - diamond synthesis 390 f allylhydropolycarbosilanes 463 alumina 3, 292 - bulkcomponents 69 - directed metal oxidation 29 1 f - elasticmoduli 71

- silicon carbides 695 alumina addition, titanium carbide SHS 362 alumina-aluminum DMO composites 3 14 alumina-based ceramics 184, 192 - tool applications 6 4 8 4 8 2 aluminothermic reduction 875 aluminum - cemented borides 896 - diamond synthesis 498 aluminum borates 164 aluminum boride type structures 805 f aluminum interlayers, diamond-like carbon films 641 aluminum kryolite melt 880 aluminum lithium alloys 289 aluminum-magnesium systems, directed metal oxidation 295 aluminum-magnesium-zinc system, directed metal oxidation 292 aluminum-nitride-silicon carbide system, selfpropagating synthesis 354 aluminum nitrides 715 aluminum oxides 787 aluminum titanate 3 18 AMBORITE milling 559 ammonia, carbide synthesis 2 10 amorphous carbon XLIX, 272 - films 6 2 3 4 4 7 amorphous composites, nanocrystalline 112 amorphous covalent ceramics (ACC) 446 ff amorphous hard materials 36 ff amorphous structures, carbon nitrides 261 amorphous zone, silicon carbides 690 anatase structures, titanium oxides 32 anisotropic growth, silicon nitrides 763 annealing, nanostructures 130 anthraxolite 376 antireflection coatings, CVD diamond 589 anvils, diamond synthesis 490,565 apatite XL applications 477-995 - alumina-based ceramics 6 4 8 4 8 2 - borides 802,933 - cemented borides 927 ff - CVD diamond 410,573422 - diamond materials 479-572 - diamondboron nitrides 527 - diamond-like carbon films 640 ff

998

Index

- directed metal oxidation 3 16 - polycrystalline ultrahard materials 548 ff - silicon carbides 736 ff - silicon nitrides 751 ff, 782 ff, 792 - single crystal diamond 559 ff - TM carbideshitrides 202 ff, 238 ff arc evaporation, diamond-like carbon films 627 Archimedian three connected nets 275 argon 43 argon sputtering - boroncarbides 8.56 - ncNiCr 111 armor, directed metal oxidation 3 17 armor modules 94,97 Arrhenius behavior - self-propagating synthesis 323 - silicon carbides 450 aspect ratio - silicon carbides 695 - siliconnitrides 767 ASTM, silicon carbides 736 AT60 composite 674 atomic force microscope (AFM) 54,405,440 atomic hydrogen, diamond synthesis 507 attack modes, corrosion 141 f attrition milled powder mixtures 916 Auger electron spectroscopy (AES), 2 11 augmented-plane wave (APW) calculations 15 austenitic phases, carbides 16 automated pressing, silicon carbides 704 P-modification, silicon nitrides 753 P-phases, silicon carbon nitrides 265 ballas 512 ballistic properties 94,97 ff band gaps, borides 803 barriers, coulombic 94 BC(8/C4/T4/T8) series, carbon allotropes 272 ff bell jar reactor 398 belt devices, diamond synthesis 490 bending strength - alumina-based ceramics 649 - boron carbides 852,865 - silicon nitrides 772,778 bending tests 73 benzene - carbide synthesis 208 - diamond-like carbon films 627,633 Berkovich model 86,195 Berman-Simon line 374 beryllium oxides 30 bias voltage, diamond-like carbon films 624 bias-induced heteroepitaxy 404 f binary carbides/nitrides 202 ff binary systems, borides 81 3 ff

binders bulkcomponents 68 - carbides 12 - cemented ternary borides 9 19 - polycrystalline diamond 516 - self-propagating synthesis 357 - titanium boride-iron composites 910 - tungsten carbidexobalt hardmetal 952 - tungsten carbides 165 binding energies, boron carbide-silicon carbide ceramics 859 birefringence, silicon carbides 7 I9 black body radiation 45 black composites, alumina 669 block sawing, diamond abrasives 543 blocking, corrosion 146 boehmite - alumina-based ceramics 653.661 - Vickers hardness 184 Boltzmann constant 45,93, 322 bonded tools, diamond/boron nitrides 528 bonding 3 f, 198 - borides 803 - boron nitrides 420 f - carbon allotropes 274 - carbon nitrides 258 - diamond-like carbon 630 ff - TM carbides/nitrides 204 ff borane-dimethylamine 60 boride-based hard materials 802-945 boride-zirconia composites 888 ff borides - bulkcomponents 69 - crystal structures 8 f - data collection 965-995 boron - diamond synthesis 500 - silicon carbide-based materials 683 boron-based hard materials LVIII boron-boron interactions, borides 808 boron carbide-based cermets 895 ff boron carbide-based composites 857 ff boron carbide ceramics 837 ff boron carbide process 875 boron carbide-silicon carbide ceramics 857 boron carbide-transition metal diboride ceramics 861 boron carbides XL - ballistic properties 94 - corrosion 161 ff, 176 - crystal structures 8 f - silicon carbides 7 14 boron-carbon-aluminum system 820 boron-carbon-metal system 819 boronxarbon-silicon system 8 19,822 boronxarbon system 8 13 boron-carbon-titanium system 823 ff -

Index

boron carbonitrides LXII, 526 boron doping - carbon allotropes 276 - diamond-like carbon 638 boron halides 855 boron nitrate 406 boron nitride/silicon carbide double layers 309 boron nitrides XL-LXXI, 6f, 253 ff - alumina-based ceramics 648 - bulkcomponents 69 - corrosion 171 ff - crystallization 5 10 ff - cubic polycrystalline 479 f - CVD diamond 597 ff - high pressure melting 49 - nanostructures 104, 125 f - self-propagating synthesis 340 - vaporphase deposition 42M45 boron-rich nitrides 526 boron-silicon system 815,821 boron suboxides XL, LXV, 526 boron-transition metal systems 830 boro-silica glass 860 borothermic reduction 876 boundary conditions, corrosion 142 boundary sliding, nanocomposites 121 Bragg angle - crystal stuctures 36 - nanocomposites 119 breakdown - corrosion 147 - spinel layers 300 f Bridge method 853 Bridgmen anvils 490 Brillouin zone 276 Brine11 hardness XLI, 85, 107 brittle failure 67 brittleness 20, 74 ff bromine reactions 855 bronzes 895 brookite structures 32 brown corundum 658 bubble formation, silicon carbides corrosion 160 bulk components, mechanical properties 68 ff bulk modulus XLII - carbon allotropes 277 - carbon nitrides 263 - polycrystalline boron nitride 519 - silicon carbon nitrides 265 bulk properties, TM carbides/nitrides 203 Burgersvector 93, 110, 115

C,, see: fullerenes calcite XL calcium silicate, melting temperatures calcium sulfate coatings 307

51

999

calibration, diamond anvil cell 43 CALPHAD (calculation of phase diagrams) 213 carbide conversion 876 carbide nitride-boride-silicide composite-based hard materials 888 carbide-reinforced composite ceramics, alumina-based 669 ff carbides - bulkcomponents 69 - corrosion 155ff - data collection 965-995 - superstoichiometric 127 - transition metal diboride cermets 916 - transition metals 12 ff carbon - corrosion 154ff - diamond-like XLIX - hydrothermal synthesis 374 - self-propagating synthesis 338 - silicon carbide-based materials 683 carbon allotropes XLIV, 485 - diamond synthesis 390 f - doping 271-285 carbon-based hard materials XLIV ff carbon black - boroncarbides 843 - directed metal oxidation 293 - titania-aluminum system 364 carbon black reduction 875 carbon-boron system 813 carbon clusters 470 carbon content, tungsten carbide-cobalt hardmetal 952 carbon dioxide, diamond synthesis 490 carbon dioxide laser heating technique 4 1 4 5 carbon filaments, silicon carbides 697 carbon films, diamond-like 6 2 3 6 4 7 carbon-h ydrogen-oxygen system, hydrothermal synthesis 376 ff carbon phase diagram 485 carbonado 5 12 carbonitrides LIII, 202-270,523 - corrosion 173ff - nanostructures 104 ff carborundum 7,683 carbothermic reduction - boronoxides 838 - metaloxides 875 - silica 691 - silicon carbides 7 15 carburization rate, diamond synthesis 394 carrier gas hot extraction 21 1 cascade arc plasma jet 624 cast iron machining 660, 667,670 cast self-propagating high temperature synthesis 342 catalysts, diamond synthesis 498 f

cathodic arc deposition 422 caustic alkalis 155 caustic oxidizing media 415 cavities 92 cemented borides 895 ff cemented carbides 12,238 f cemented carbonitrides 238 f cemented diamond compositions 5 12 cemented ternary borides 9 19 ff centrifugal compaction 654 ceramic alumina 3 ceramic bodies processing 446-476 ceramically bonded silicon carbides 700,72 1 cermets 202,345 charge carrier density - boronnitrides 438 - carbon nitrides 260, 266 charge coupled devices (CCD) 45 chemical analysis, TM carbides/nitrides 210 f chemical bonding - borides 803 - carbides 15 - nitrides 25 chemical compositions see: compositions chemical properties - boron carbides 851 f, 855 f - silicon carbides 720 chemical vapor deposition (CVD) - boronnitrides 421 - carbide synthesis 210 - diamond films 390-419 - hydrothermal synthesis 374,386 - nanostructures 116 f - silicon carbides 697,707 - thin films 463 - transition metal borides 876 - transition metal carbides/nitrides 24 1 chemical vapor deposition (CVD) diamond XLVII, 484 - applications 573-622 chemical vapor infiltration (CVI) 3 10,717 chevron-nothed threepoint bend beam 84 chlorides 208 chlorination 374 chlorine reactions 855 chromium - diamond synthesis 498 - transition metal diboride cermets 915 chromium binder 923 f chromium carbides 17,208 chromium diborides 83 1 chromium-nitrogen system 219 cluster fragmentation, cemented ternary borides 922 clusters - alumina-based ceramics 653 - carbon allotropes 272

titanium boride-yttria stabilized zirconia composites 893 coagulation, alumina-based ceramics 654 coarse grain, CVD diamond 580 coarsening see: Ostwald ripening coarser powders, polycrystalline diamond 5 13 coated abrasives 736 coatings 66,70 f - boroncarbides 838 - boronnitrides 429 - diamond-like carbon films 640 - directed metal oxidation 307 - nanocomposites 121 f - self-propagating synthesis 366 - TM carbides/nitrides 242 f cobalt - diamond synthesis 498 - hardness 948 - transition metal diboride cermets 915 cobalt hard metals 238 cobalt-iron-system 488 cobalt-tungsten-carbon system 5 16 coesite-stishovite phase boundary 55 cold compaction 5 13 cold isostatic pressing 65 1 colloid size range growth, diamond 504 color - boronnitride 510 - diamond synthesis 498 - silicon carbides 690,7 19 f - TM carbides/nitrides 224 combustion analysis - diamond-like carbon 63 1 - TM carbides/nitrides 21 1 combustion tubes compact nitridation, silicon powders 749 compact tension test 84 compaction homogeneity, alumina-based ceramics 654 complex kinetics, corrosion 148 ff composite ceramics, whisker-reinforced 669 composites - aluminum-magnesium 300 - boron carbide-based 857 ff - nanocrystalline 112 - silicon carbides 7 10 - transition metal carbides 880 compositions - cemented ternary borides 923 - titanium boride-iron composites 901 compounds - borides 809 - isoelectronic 6 f - transition metals 203 compressive strength - polycrystalline boron nitride 519 - silicon carbides 7 18 -

Index compressive stress, sapphire 186 concrete machining 540 ff conducting target sputtering 430 conduction band, nanostructures 132 conductivity - alumina-based ceramics 667 - boroncarbides 854 - boron nitrides 421 - CVD diamond 412,582 - data collection 965-995 - directed metal oxidation 289.3 16 - polycrystalline boron nitride 519 - silicon carbides 721 - silicon nitrides 75 1 - TM carbides/nitrides 203,225 CONFLAT vacuum flanges 606 continuous fiber-reinforced silicon carbide matrix composites (CMCs) 717 ff, 739 controlled atmosphere, thin film processing 464 conversion process - self-propagating synthesis 328 f - thinfilms 464 copper - diamond synthesis 488 - directed metal oxidation 291,305 cordierite, silicon carbides 695 core-mantle composite model 110 core drills 549 core-rim structures, transition metal diboride cermets 918 cores, silicon carbides 688 comer sharing - carbides 19 - nitrides 26 - titaniumoxides 33 corrosion 14&182 - directed metal oxidation 3 16 - silicon nitrides 749,786 ff corundum XL - alumina-based ceramics 648 f, 653,658 - crystal stuctures 30 - silicon carbide based materials 683 - single phase sintered 670 Coulomb barriers 94 Coulombic forces, carbon/silicon-nitirides 254 covalent bonding XL, 253 - borides 803 - boron carbides 841 - boron nitrides 420 f - carbides 197 - hydrothermal synthesis 374 crack branching 113 crack bridging 96 - silicon carbides 695,710 f crack deflection - boride-zirconia composites 890 - silicon carbides 7 10 f

1001

titanium boride-titanium carbide system 880 crack growth rate, silicon nitrides 772 f crack propagation 67 f, 75 ff, 86 cracking, directed metal oxidation 291,298 creep behavior 92 ff - alumina-based ceramics 667 - directed metal oxidation 313 - nanocomposites 121 - silicon nitrides 75 1,778 creep feed grinding 537 crosslinking, polymer-ceramic transformations 447 crushing - boroncarbides 838 - polycrystalline diamond 513 crystal structures 1-285 - borides 804 - boroncarbides 854 - boron nitrides 420 f - carbon allotropes 277 - carbonitrides 258 - cemented ternary borides 9 I9 - data collection 965-995 - polycrystalline boron nitride 520 - silicon carbides 685 f - silicon nitrides 753 ff - TM carbides/nitrides 204 ff - zirconium/titanium borides 879 crystallite size effect, hydrothermal synthesis 316.382 crystallites - alumina-based ceramics 655,661 - boride-zirconia composites 889 - diamond synthesis 400,503 - silicon nitrides 759 crystallization, diamond 485 cubic boron nitride XL-LXXI see also: boron nitride cubic diamond - hydrothermal synthesis 374 - synthesis 509 cubic structures - borides 809 - boron nitrides 6,42@445 - cobalt 949 - diamond Sf - silicon carbides 685 - silicon carbon nitrides 266 - TM carbides/nitrides 205 Cubitron 664 curing 448 cutting edge displacement (CED) 674,679 cutting edges 658,666 cutting tools - alumina-based ceramics 648 ff - carbonitrides 202f - cemented borides 927 -

1002

Index

- CVDdiamond 411,611ff - material selection 68 f - polycrystalline boron nitride 5 19 - polycrystalline diamond 550 - silicon carbides 736 - silicon nitrides 784 - single crystal diamond 563 - temperature effects 199 - TM carbides/nitrides 241 cyclohexadiene 273 Czochralski growth 190 damage creep 94f - intergranular 199 - silicon nitrides 780 f - tolerance 67 data collection, hard material properties 965-995 DBC5O milling 561 De Hoff diagram 840 Debye temperature 879 decomposition reactions, CO, laser heating 54 ff deep grinding 537 defects - borides 813 - boroncarbides 842 - carbides 12 - ceramic tool materials 65 1,662 - silicon nitrides 771 f definded cutting edge 666 deformation XLI - indentation test 87 - modes 76,94 f, 185 - polycrystalline diamond 514 dense materials, self-propagating synthesis 342 ff dense shapes, silicon carbides 699 densification - boroncarbides 844 - silicon nitrides 755 ff - titanium boride-iron composites 906 f - transition metal borides 876 ff density XLII - alumina-based ceramics 654 - boroncarbides 854 - CVDdiamond 601 - data collection 965-995 - diamond-like carbon 632 - polycrystalline boron nitride 519 - silicon carbides 695,718, 721 - titanium boride-iron composites 910 - TM carbides/nitrides 203 - zirconium/titanium borides 879 density effect, self-propagating synthesis 339 -

density functional based tight-binding (DF-TB) 273,276 density of states (DOS) - carbon allotropes 279 f - TM carbides/nitrides 206 deposited layers, TM carbides/nitrides 241 deposition mechanisms 66 deposition methods - boronnitrides 421 f see also individual types (CVD, PVD, etc.) deposition temperature, diamond-like carbon 633 design, superhard materials 109 ff detector materials, CVD diamond 584 DIAFILM 575 ff, 583 diamond XL-LXXI, 5 ff - alumina-based ceramics 648 f - carbon allotropes 271 ff - classifications 498 f - corrosion 154ff - hydrothermal synthesis 374-389 diamond abrasives, synthesis 491 diamond anvil cell (DAC) 508 - laser heating 41-65 diamond films 390-419 diamond incorporation 367 diamond materials 479-572 diamond single crystals 379 diamond synthesis 481 ff - carbonnitrides 257 diamond-like carbon films (DLC) 623-647 diboride systems 831 dielectric loss, CVD diamond 601 dielectric properties, CVD diamond 583 dies, diamond synthesis 490 differential thermal analysis (DTA) 36, 151,446 diffusion, directed metal oxidation 291 diffusion-controlled growth, silicon nitrides 763 diffusion-cooled CO, laser 596 diffusion layers, TM carbides/nitrides 246 f diffusional techniques, carbide synthesis 2 10 diffusivity 229 ff dilution effect, self-propagating synthesis 335 diodes 606 direct coagulation casting (DCC) 654 direct current plasma jet deposition XLVIII directed metal oxidation (DMO) 92,289-321 dislocations 183, 188, 196 - boroncarbides 842 - creep 94 - CVD diamond 412 - nanocomposites 121 - transition metal diboride cennets 918 displacements - control 77 - Voigt bounds 71 dissolution, nanostructures 130

Indexdissolution temperatures, diamond 380 distortions, CVD diamond 589,595 doping - alumina-based ceramics 661 - boroncarbides 852 - diamond-like carbon 638 - directed metal oxidation 305 - sp’ bonded carbon allotropes 271-285 double bonds, carbon allotropes 278 double layers, directed metal oxidation 309 double torsion test 84 dressers, CVD diamond 61 1,616 f dressing tools 562 drilling - carbonitrides 202 f - polycrystalline diamond 554 - polycrystalline ultrahard materials 549 dry grinding - abrasives 532, 538 - silicon carbides 704 dry machining 6 14 dry powder processing 65 1 dry pressing 704 drysawing 547 ductile binders 68 ductile-to-brittle transition - self-propagating synthesis 356 - titanium carbide 235 duplex layer, directed metal oxidation 292,298 duplex techniques, titanium nitride coatings 342 q-carbides 18 earth’s lower mantle materials 5 1 f edge sharing - nitrides 27 - titaniumoxides 33 edge toughness 722 Einstein equation 130 EkasicT 722ff elastic modulus 71 ff, 81 ff - directed metal oxidation 3 10 - SiAlON 460f - silicon carbides 718 f - siliconaluminocarbosilanes 454,457 f elastic properties, TM carbides/nitrides 23 1 f elastic recoil detection (ERD) 63 1 elastic recoil detection analysis (ERDA) 436 electric discharge grinding 613 electric resistivity - boroncarbides 854 - data collection 965-995 - zirconium/titanium borides 879 electrical conductivity 2 2 5 3 16 electrical properties - boron nitrides 441 - CVDdiamond 413

1003

diamond-like carbon 637 f silicon carbides 719 electrochemical applications, CVD diamond 415 electro-corundum 658 electrolysis, transition metal borides 876 electron cyclotron heating (ECH) 598 electron cyclotron resonance (ECR) 422 electron energy loss spectroscopy (EELS) 63 1 electron mobility, CVD diamond 414 electron paramagnetic resonance (EPR) 379 electron probe microanalysis (EPMA) 2 11 electron spectroscopy for chemical analysis (ESCA) 725 electron transport, directed metal oxidation 290 electronic applications, CVD diamond 413 electronic properties, carbon allotropes 279 electronic speckle pattern interferometry (ESPI) 593 electroplated tools 529,532 Elektroschmelzwerk Kempten (ESK) process 688f element synthesis, silicon carbides 692 energy-dispersive analysis of X-rays (EDX) 119 energy-volume relationship, carbon nitrides 264 engineering requirements 70 enstatite 488 enthalpy - data collection 965-995 - self-propagating synthesis 323 enthalpy increase, diamond synthesis 505 epitaxial growth 291 equilibrium volumes XLII erosion tests 8 1, 89 f Eshelby treatment 31 1 etching - diamond synthesis 404 - silicon nitrides 770 - titanium boride-yttria stabilized zirconia composites 893 eutectic concentrations, titanium boride-iron composites 909 eutectic reactions, borides 823 evaporation - carbide synthesis 210 - diamond-like carbon films 627 even/odd phases, carbon nitrides 259 evolutionary selection, diamond synthesis 401 experimentals - CO, laser heating 44 f - corrosion 1SO ff - diamond anvil cell 43 f - self-propagating synthesis 33 1 ff - SiAlON 460f - silicon carbide-water system 383 - siliconaluminocarbosilanes 452 - thin film processing 463 - ncTiN/aSi,N, deposition 118 -

1004

Index

extended X-ray absorption fine structure spectroscopy (EXAFS) 264,446 external corrosion 141 f extinction coefficient, boron nitrides 440 extrusion, silicon carbides 704 failure 67,94 ff - cemented borides 929 fatigue, mechanical 95 f fatigue strength, silicon nitrides 75 1 feldspar XL Fermi level - borides 808 - carbon allotropes 274 - diamond-like carbon 637 - nanostructures 132 - TM carbideshitrides 206 femtic binders, cemented borides 929 ferrous material machining 557 fiber-reinforced directed metal oxidation composites 314 fiber-reinforced plastics 61 I fibers - directed metal oxidation 307 f - silicon carbides 697,7 18 Fick diffusion 142 ff field-activated self-propagating high temperature synthesis 348 ff figures-of-merit, CVD diamond 413 film adhesion 427 fine grain, CVD diamond 580 finish machining 613 f flaking 70 flat faces, diamond synthesis 401 flatness - CVD diamond windows 595 - diamond submount 610 flaws 88,184 - alumina-based ceramics 653,662 - creep 95 - nanostructures 111 f flexural strength 82 - boron carbide ceramics 870 - polyaluminocarbosilanes 456 - silicon carbides 714, 718, 721 - titanium boride-iron composites 913 fluorescence spectroscopy 48 fluorite XL flute grinding high-speed steel 541 Forschungszentrum Karlsruhe (FZK), CVD diamond 598 forsterite 488 Fourier heat transfer 322,350 Fourier transform infrared spectroscopy (FT-IR) 36,437,464 - borides 811

fractional parameters, carbon/siliconnitirides 255 fracture corrosion stress 158 fracture strength 81 f, 601 fracture stress - CVDdiamond 580 - nanostructures 105, 109 fracture toughness 67 f, 74,83 f - alumina-based ceramics 649,659,663 - borides 928,934 - boron carbides 853 f, 865,870 - CVDdiamond 588 - directed metal oxidation 289 - polycrystalline boron nitride 5 19 - SiAlON 460f - silicon nitrides 75 1,771 - titanium boride composites 890,912 fragmentation, directed metal oxidation 29 1 Frank-Kasper polyhedron 18 Frenkel excitations 638 fretting wear 656 friction coefficients - boronnitrides 440 - CVD diamond 618 - diamond-like carbon 640 f - nanocomposites 127 - silicon carbides 722 f, 732 - silicon nitrides 783 full-width-at-half-maximum (FWHM) - CVDdiamond 575 - diamond synthesis 405 - silicon carbide-water system 386 fullerenes - carbon allotropes XLVI, LIII, 27 1 ff - complexes 253ff - hydrothermal synthesis 374 furan coke 505 fused corundum 658 fused grinding materials 648 ff y-alumina 291 y-point sampling, carbon allotropes 276 garnets, phase diagrams 56 gas corrosion - boron carbides 163, 172 - silicon carbides 159 - silicon nitrides 167 gas fusion methods 735 gas phase diagnostics 395 gas pressure 624 gas pressure sintered silicon nitrides (GPSN) 752,758 gas temperatures, diamond synthesis 392 gaseous nitridation 290 gasification, graphite 307 gaskets, diamond synthesis 490

Index GefStoffV 736 gelcasting 654 gemstons ruby 30 General Electric process 692 generalized gradient approxiamtion (GGA) 255 geometrical distortions, CVD diamond 595 germanium 488 gettering 498 GG25 cast iron 669 Gibbs-Thomson equation 840 Gibbs free energy - diamond synthesis 407 - titanium boride-iron composites 902 Gibbs free enthalpy, nanocomposites 120, 124 f girdle devices 490 glas-reinforced plastics 553 glassy phases 92 glow discharge deposition - diamond-like carbon films 623 - nanocomposites 117 f grades - CVDdiamond 574ff - silicon carbides 687 grain boundaries - boroncarbides 841 - corrosion 141 - deformation 183-20 1 - junctions 91 - magnesia 292 - nanostructures 110, 130 - silicon nitrides 75 1 grain boundaries interlocking 886 grain size 183-20 1 - alumina-based ceramics 654 - boron carbides 838,871 - silicon nitrides 75 I - tungsten carbides 950 f grains - alumina-based ceramics 664 - CVD diamond 575,580 - polycrystalline diamond 514 granite block sawing 543 granules, alumina-based ceramics 65 1 graphite XLIV ff, 253 ff - borides 813 - boroncarbides 838 - cemented borides 895 - diamond films 390 ff - diamond synthesis 486 ff graphite4iamond transformation 407 graphite reduction, transition metal borides 875 graphitic components, directed metal oxidation 293 graphitic structures 259 graphitization - diamond corrosion 155 - polycrystalline diamond 514

1005

graphyne 272 gray cast iron 669 green bodies, polyaluminocarbosilanes 452,458 green density - alumina-based ceramics 660 - self-propagating synthesis 360 - silicon carbides 700 Griffith model 74 f, l l 0 f grinding - alumina-based ceramics 648 f, 6.58 ff - boron nitride abrasives 530,538 - CVD diamond 613 - diamond abrasives 530 ff - dislocations 194, 199 - hardness XLI - silicon carbides 704,736 grits - alumina-based ceramics 658, 665 - borides 802 group (IV-VI) elements 830 group IV carbide system 21 3 group IV nitride system 216 group IVB carbonitrides 202-252 group IVB transition metal carbide layers 241 group V carbide system 2 13 group VB nitride system 21 7 group VIB transition metal carbide system 21 3 group VIII elements - borides 830 - diamond synthesis 498 growth - boron nitrides 433 ff - CVDdiamond 580 - diamond synthesis 392,407 ff, 492,496 - diamond-like carbon films 625 - directed metal oxidation 300 ff - patterns 147 ff - silicon nitrides 760 gypsum XL gyrotron tube windows 583,597 f

H3, carbon allotropes 273 f hafnium nitrides 23 1 halides 876 Hall-Petch relation 93 - nanostructures 110 ff - TM carbides/nitrides 235 - tungsten carbide-cobalt hardmetal 948 ff halogenated hydrocarbons 377 hard materials XLI-LXXI - structures 1-285 hard tool ceramics 650 hardened steel machining - alumina-based ceramics 667,672 ff - CVD diamond 618

1006

Index

hardness XLI-LXXI, 79,84 ff alumina-based ceramics 648-682 - boride-zirconia composites 891 - borides 802 - boron carbides 851,854,865 - boron nitrides 421,440 - cemented ternary borides 925 - cobalt 948 - CVDdiamond 410 - diamond-like carbon 632,639 - polycrystalline boron nitride 5 19 - SiAlON 460f - silicon nitrides 75 1 ff, 775 f - titanium boride-iron composites 910 - tungsten carbide-cobalt hardmetal 946-964 - tungsten carbides 947 - ultrahard materials 521 - zirconium/titanium borides 879 Hashin-Shtrikman model 72 health 73.5 heat resistance 607 f heat sinks 318 heated filament-assisted chemical vapor deposisiton (Hl-CVD) XLVII Hertzian contact damage 783 Herzian indentation test 84,98 heteroepitaxial films 400,404 heterostructures, nanocrystalline 114 ff hexagonal-cubic transition 433 f hexagonal graphite - diamond synthesis 508 - hydrothermal synthesis 374 hexagonal structures - boron nitrides LVIII, 6,420 ff - cobalt 949 - diamond 5 ff - silicon carbides 7,685 - titanium borides 816 - TM carbides/nitrides 205 high frequency glow discharge 117 f high power gyrotron tubes 593 high power infrared lasers 589 high pressure decomposition 54 ff high pressure diamond synthesis 487 f high pressure high temperature phase diagrams 54 ff high pressure high temperature (HPHT) synthesis - boronnitrides 420 - diamond XLVII,508 - hydrothermal 374,380 high pressure laminates (HPL) 613 high pressure microwave source reactor (HPMS) 398 high pressure modifications, silicon nitrides 753 high pressure process, diamond synthesis 504 high pressure stability, CO, laser heating 54 ff high purity materials 838 -

high resolution transmission electron microscopy (HR-TEM) 119 high speed finish machining 6 13 high speed steel (HSS) 541,555 high temperature expansion, TM nitrides 230 high temperature hardness 934 high temperature properties - silicon nitrides 777 ff - polyaluminocarbosilanes 457 f high temperature strength 3 13 highest occupied molecular orbitals (HOMO), carbon allotropes 279 ff highly ordered pyrolytic graphite (HOPG) 416 holemobility 414 hole transport 290 homogeneity - alumina-based ceramics 654 - corrosion 141f Hookelaw 71,81 - nanostructures 104, 108, 129 Hopkinson bar test 97 host lattice, TM carbides/nitrides 205 hot compaction 5 13 f hot corrosion 151 - boroncarbides 163 - boronnitrides 172 - directed metal oxidation 3 16 - silicon carbides 157 f - silicon nitrides 167 f hot filament chemical vapor deposition (HF-CVD) 39 1 ff hot isostatic pressing (HIP) - boron carbides 847 f, 861 - self-propagating synthesis 345 - silicon carbides 707,721 - silicon nitrides 166,749 ff, 778 f hot pressing - boron carbides 847 f, 861 - silicon carbides 705,721 - silicon nitrides 749 ff - transition metal borides 877 hot target sputtering boron nitrides 431 hybridization XLIV - diamond-like carbon 630 - nanostructures 104 - silicon carbides 686 hydroboric acid layers 855 hydrocarbons - CO, laser heating 60 - diamond synthesis XLVIII, 489 - hydrothermal synthesis 377 hydrochloric acid 3 16 hydrogen, diamond synthesis 390 f, 395, 507 hydrogen-free amorphous carbon (ta-C) 627 f hydrogenated amorphous carbon (a-C:H) 623,630 ff

Index

hydrothermal corrosion boroncarbides 162 - silicon carbides 156 - silicon nitrides 166 hydrothermal synthesis, diamond -

374-389

icosahedral structures borides 809, 815 - boroncarbides 8 ignition, boron carbides 838 ignition temperature 322,359 ilmenite structures 57 immersion tests 150 impurities - CVDdiamond 412 - diamond synthesis 501 incubation - corrosion 150 - directed metal oxidation 295 indentation crack length method 910 indentation hardness 107 indentation size 183-201 indentation tests 80 f, 86 - hardness XLI - SiAlON 460f inductive plasma emission (ICP) spectroscopy 735 industrial applications - diamondboron nitrides 527 - TM carbidestnitrides 238 ff inelastic deformation 183 ff inert atmosphere pyrolysis 452 f infrared lasers 589 infrared seekers 583 initiation process, directed metal oxidation 289,295 injection molding - boroncarbides 846 - silicon carbides 704 injection molding machine parts 932 interface controlled growth, silicon nitrides 763 interfacial energy, diamond synthesis 505 intergranular damage 199 intergranular microcracks 91 interlaminar strength 71 8 interlocking grain boundaries 886 intermediate phases, titanium borides 8 16 internal corrosion 141 f international thermonuclear experimental reactor (ITER) 598 interstitial compounds 15 interstitials, TM carbides/nitrides 205 ion-ion core repulsion, carbon allotropes 276 ion beam-assisted deposition (TBAD) 523,627 ion energy distribution, diamond synthesis 406 ion irradiation, thin film processing 467 f -

1007

ion plating 624 f ion sputtering 111 ionic bonding, boron nitrides 420 ionic diffusion 291 iron, diamond synthesis 498 iron binder 923 iron catalysis 41 1 iron-nickel-system, diamond synthesis 488 iron oxides/silicates, melting temperatures 5 1 isoelectronic compounds 6 f isostatic pressing 704 isothermal microwave plasma chemical vapor deposition 397 isotopic labelling 409 isotropic compounds 809 Jagodzinski notation 7 f, 13 Japan Atomic Energy Research Institute (JAERI) 602 kimberlite 377,488 kinetics - corrosion 142ff - titanium boride-iron composites 906 kinked faces 401 Kjeldahl analysis 21 1 Knoop hardness XL, 86, 193 - boronnitrides 440 - data collection 965-995 - nanostructures 107 - polycrystalline boron nitride 5 19 - SiAlON 460f - ultrahard materials 521 Koehler model 1 14 Koks model 110 laminate wood flooring manufacturing 613 Langmuir probe 438 Lanxideprocess 917 laser ablation - carbon nitrides 257 - diamond-like carbon films 627 laser diode arrays (LDA) 606 ff laserheating 41 65 laser-induced damage threshold (LIDT) 594 laser irradiation, carbide synthesis 210 laser synthesis, silicon carbides 693 lasers, CVD diamond 589 lattice defects 185, 188 lattice dislocations 93 lattice parameters - boronnitrides 420f - carbon allotropes 274 f, 277 - silicon nitrides 7.53

TM carbidednitrides 203 zirconium/titanium borides 879 - data collection 965-995 layers 241 - corrosion 141 Lee-Gurland equation 955 Lehoczky model I 14 Lely process 708 lifetime diagrams, cemented borides 929 ff light covalent ceramics 522 light emitting diodes 737 line spread function (LSF) 578 linear rate constant, corrosion 144 liquid media corrosion - boron carbides 161 - boron nitirides 17 1 - silicon carbides 156 - silicon nitrides 166 liquid phase sintering (LPS) - boroncarbides 845 - cemented ternary borides 920 - polycrystalline diamond 5 13,516 - silicon carbides 705,709, 721,738 - silicon nitrides 756 - titanium boride-iron composites 899 liquid polymer infiltration (LPI) 718 f liquid shaping, alumina-based ceramics 65 1 liquid silicon infiltration (LSI) 719 f lithium 289 lithium aluminosilicate (LAS) 3 15 lithium doping - carbon allotropes 27 I - directed metal oxidation 304 load control 76, 187 ff local density approximation (LDA) 255 localized corrosion 141 f long range order (LRO) 16 longer wavelength infrared (LWIR) band 584 lonsdaleite XLII, 5 - carbon allotropes 273 - diamond synthesis 5 10 - hydrothermal synthesis 375 loss tangent, CVD diamond 588,600 low angle scattering 578 low pressure solid-state source (LPSSS) - diamond synthesis 490,508 - hydrothermal synthesis 374 low pressure sustained growth, diamond 504 lubrication, silicon carbides 722 f -

-

machining - alumina-based ceramics 649 ff, 660 - diamond abrasives 540 ff - laminate wood flooring 613 magic angle spinning nuclear magnetic resonance (MAS-NMR) 36

magnesia see: magnesium oxides magnesiothermic reduction 875 magnesium - diamond synthesis 488 - directed metal oxidation 289 ff, 304 magnesium oxides 779,787 f - crystal stuctures 31 - directed metal oxidation 292 magnesium oxides/silicates, melting temperatures 5 1 magnetron sputtering - boron nitrides 430,435 f - carbonnitrides 524 main group element nitrides 24 f main group element oxide ceramics 30 f MAK (maximal zulassige Konzentration) value, silicon carbides 736 manganese 498 manufacturing - alumina-based ceramics 650 - polycrystalline diamond 5 13 martensitic phases, carbides 16 mass selected ion beam deposition (MSIB) - boronnitrides 422 - diamond-like carbon 632 mass spectroscopy (MS) 36,446 mass transport, directed metal oxidation 290 material grades, CVD diamond 574 f material properties 1-285 - boron carbides 851 ff - boronnitrides 421 - cemented ternary borides 923 - CVD diamond 410,574 ff - data collection 965-995 - diamond-like carbon 632 - ncMnN/Si,N, composites 119 ff - polycrystalline boron nitride 5 19 - silicon carbides 719 ff - silicon nitrides 753 ff, 771 ff - titanium boride-iron composites 909 - TM carbides/nitrides 224 ff - transition metal borides 878 materials 477-995 - corrosion 154ff mean free path 949 f meandering 113 see also: cracking measurement, corrosion 150 ff measurement models 193 f mechanical grade, CVD diamond 575 mechanical properties-microstructure relation 6 6 1 0 3 mechanical properties - borides 802 - boron carbide ceramics 865 - boronnitrides 440 - cemented borides 923,928

lndex

- data collection 965-995 - diamond-like carbon 639 f - directed metal oxidation 3 10 - self-propagating synthesis 327 ff - silicon carbides 720 - silicon nitrides 771 ff - polyaluminocarbosilanes 456 - titanium boride-yttria stabilized zirconia composites 892 - transition metal borides 878 medium density fireboard (MDF) 553 melonophlogite 274 melting, directed metal oxidation 295 melting points 3 f - borides 802,814 - boroncarbides 854 - cemented borides 897 - creep 94 - data collection 965-995 - diamond synthesis 498 - high pressure 49 - polycrystalline boron nitride 520 - self-propagating synthesis 331,342 - silicon carbides 685 - TM carbides 202 ff, 224 - TM nitrides 224 - zirconium/titanium borides 879 melts corrosion - boron carbides 163 - boronnitrides 172 metal-bonded tools 529,534 metal borides, cemented 895 ff metal-boron interactions 808 metal carbides - cemented borides 895 - diamond-like carbon films 629 - transition metal borides 876 metal containing amorphous hydrocarbon 629 f, 634 f metal hydrides 875 metal matrix composites (MMC) - cemented borides 895 - CVDdiamond 611 - polycrystalline diamond 553 - silicon carbide whiskers 696 metal oxidation, directed 289-321 metal oxide silicon field-effect transistor (MOSFET) 415 metallic binders, cemented ternary borides 919 metallic impurities, silicon carbides 735 metallic melts 855 metallic nanoclusters 132 metals corrosion - silicon carbides 156 - silicon nitrides 167 metastable processes, diamond synthesis 504 methane 60,208

1009

diamond synthesis 395,406 methane hydrogen plasma sustained growth 504 methyl radicals 408 methylhydroxylsiloxanes 463 Meyer hardness 85 microanalysis, physical 21 1 microcracking - boride-zirconia composites 890 - boron carbide ceramics 873 - coatings 70 - nanostructures 110, 113 microcrystalline diamonds 379 microhardness XL, 84, 194 - carbides 215 - data collection 965-995 - light covalent ceramics 521 - TM carbides/nitrides 203,234 f - tungsten carbides 947 see also: Knoop hardness micropores, alumina-based ceramics 653 microstructure-mechanical properties relation 66- 103 microstructures - cx-SiAlON materials 768 ff - alumina-based ceramics 656 - p-silicon nitrides 758 - boroncarbides 844 - boron carbide-silicon carbide ceramics 857 - cemented carbonitrides 240 - diamond-like carbon 630 ff - directed metal oxidation 302 ff - Ekasic W 733 - nanocrystalline 128 - silicon carbides 706 - silicon nitrides 749 - titanium boride-iron composites 909 f - tungsten carbide-cobalt hardmetal 946 microwave-plasma chemical vapor deposition (MWP-CVD) 391,397 migration - directed metal oxidation 304 - high pressure melting 49 - self-propagating synthesis 343 military applications, CVD diamond 584 milling - cast iron 670 - single crystal diamond 559 mining bits, polycrystalline diamond 556 modeling procedures - carbon/silicon-nitirides 254 ff - heat resistance 607 modifications, silicon nitrides 753 modulation transfer function (MTF) 578 moduli - carbon nitrides 263 ff - elastic 71 ff, 81 ff - polycrystalline boron nitride 5 I9 -

ultrahard materials 521 - zirconium/titanium borides 879 modulus of rupture (MOR) see: rupture modulus Mohs hardness XL, 80, 107 molding - boroncarbides 846 - silicon carbides 704 molding tools 932 molybdenum - diamond-like carbon 634 - directed metal oxidation 291 - transition metal diboride cermets 915 molybdenum boride-type structures 807 molybdenum-iron-boron system 832 molybdenum-nitrogen system 219 molybdenum silicides 20 f, 330 monolithic ceramics 97 monolithic components processing 450 morphology - boron nitride films 423 f - diamond synthesis 495 Mottlaw 637 mounting techniques, CVD diamond 605 mullite - directed metal oxidation 293 - silicon carbides 695 multi-anvil devices, diamond synthesis 49 1 multiblade sawing 544 multilayer coatings 242 multiphase hard materials 888 multiphase systems, transition metal diboride cermets 916 multiphonon absorptions bands, CVD diamond 577 multiple cracking 29 1 -

nanoceramics 709 nanocrystalline composites 116 ff nanocrystalline diamond 379 nanocrystalline materials 110 f nanoindentation 108 nanostructured superhard materials 1 0 4 139 nanotubes 271 natural occurrence, silicon carbides 684 natural polycrystalline diamond 5 12 Nd-YAG laser 41 near cubic structures 266 neck formation - boroncarbides 849 - cemented ternary borides 922 negative electron affinity (NEA) 415 Newkirk process 289 ff Nicalon - polymer-ceramic transformations 448 - reinforced lithium alumiosilicate 3 15 - silicon carbides 695

nickel 915 nickel binders 923 f nickel-carbon phase diagram 491 nickel chromides 111 nickel doping 305 nickel-manganese system 488 nickel silicides 325 niobium - diamond synthesis 498 - diamond-like carbon 634 niobium-nitrogen system 218 nitridation, aluminum 290 nitrides - bulkcomponents 69 - corrosion 166 ff - crystal structures 23 ff - data collection 965-995 - superlattices 114 nitridosilicates 29 nitrogen - diamond anvil cell 43 - diamond synthesis 498 ff - silicon carbide based materials 683 nitrogen doping - carbon allotropes 271,276 - diamond-like carbon 638 nitrogen pressure - nitrides 212 - silicon nitrides 753 nitrogen-rich boron nitrides LXII noble gases 43 nodule spacing, directed metal oxidation 305 nomenclature, crystal structures 7 f noncutting applications - polycrystalline diamond 555 - single crystal diamond 564 nondestructive evaluation (NDE) 91 non-equilibrium thermodynamics, diamond synthesis 507 nongraphitic carbons 505 non-isothermal plasma deposition 400 notch-beam method 910 novolac type resins 843 nuclear magnetic resonance (NMR) 63 1 nuclear reaction analysis (NRA) 63 1 nucleation - boron nitrides 433 ff - CVDdiamond 580 - diamond synthesis 400 ff, 492 - magnesia 292 - siliconnitrides 760 numerical simulations, heat resistance 607 occupational health 736 occurrence, silicon carbides octahedral structures 205

684

Index odd/even phases, carbon nitrides 259 Ohm’s law 71 9 onions, carbon 374 - diamond synthesis 507 optical applications, CVD diamond 575,583 ff, 589 optical properties - boronnitrides 440 - CVD diamond 575 f, 589 - diamond-like carbon 637 f - silicon carbides 719 orbitals 279, 808 ordered defects 12 organic compounds 60 ff oriented globular growth 512 ornamental layers 243 orthorhombic structures - borides 10 - cemented ternary borides 920 - titanium borides 8 16 Ostwdd-Volmer rules 489 Ostwald ripening - nanocomposites 123, 129 f - titanium boride-iron composites 91 1 Ostwald rule 686 output windows, gyrotron tubes 583 oversaturation, silicon nitrides 761 ff oxidation - boron carbides 163,855,858 - directed 289-321 oxidation resistance 965-995 oxide ceramics - bulkcomponents 69 - crystal structures 30 ff oxides 965-995 oxyacetylene 391 oxycarbide composites 674 oxygen acetylene torch XLVIII oxygen attack, corrosion 141 oxygen contamination, titanium boride-iron composites 900 oxygen pressures 159 oxynitride solution 760 ff ~

Palmqvist cracks 87 parabolic rate constant, corrosion 145 paracyanogen 104 Paris law 95 ParthC-Yvon rule 13 partial pressure, directed metal oxidation 291 particle size, self-propagating synthesis 338 passivation - directed metal oxidation 291 - spinels 295f passive corrosion 141 f, 145 f

1011

passive infrared windows 584 Pauling rules 33 PDA diamonds 533 f Pearson symbol 4,203 Peierls-Nabarro stress 110 Peierls barrier 359 Peierls stress 72 penetration depth, ballistic 79 periclase 302 periodic bond chains (PBC) 401 permittivity 598 perovskite oxides 59 perovskite structures 27 Petchlaw 92 petroleum coke 838 phase diagrams - borides 812 ff - carbonitride systems 221 f - CO, laser heating 54 ff - silicon carbide- aluminum nitride 716 - silicon carbides 685 - transition metal-nitrogen systems 216 ff - transition metal carbides 15 - tungsten-carbon-cobalt system 239 phase equilibria, carbide systems 213 ff phase properties, polycrystalline boron nitride 520 phase reactions, cemented ternary borides 920 phase segregation, nanocomposites 1 16, 125 phase transitions, laser heating 4 1 4 5 phases, silicon nitrides 755 phenolic resins 842 phosphorus doping 27 1 physical boundary conditions, corrosion 142 f physical microanalysis, TM carbides/nitrides 211 ff physical properties 66 f - boron carbides 851 ff - boron nitrides 421,440 - cemented ternary borides 923 - CVD diamond 410 ff, 574 - data collection 965-995 - diamond-like carbon 632,637 ff - polycrystalline boron nitride 5 19 - silicon carbides 695,719 ff - silicon nitrides 753 ff, 771 ff - titanium boride-iron composites 909 - transition metal borides 878 f - transition metal carbides/nitrides 203, 224 ff physical vapor deposition (PVD) - boronnitrides 421 - carbide synthesis 210 - silicon carbides 708 - thin films 463 - TM carbides/nitrides 241 n-n bonding 274 pileup models 110

1012

Index

pitting 158 f, 786 Planck formula 45 plasma-assisted chemical vapor deposisiton (PA-CVD) XLVII plasma decomposition carbonitrides LV plasma deposition 624 f plasma-enhanced chemical vapor deposition (PE-CVD) 523 plasma etching - diamond synthesis 404 - silicon nitrides 770 - titanium boride-yttria stabilized zirconia composites 893 - titanium boride-iron composites 912 plasma vapor deposition 116 f plastic deformation see: deformation plastic indentation 108 platelets - Ekasic W 733 - silicon carbides 696 platelets alignement 92 point defects - boroncarbides 842 - diamond svnthesis 501 poisoning agents, directed metal oxidation 290,307 Poisson ratio 72 - boron carbides 854,865 - CVDdiamond 588 - nanostructures 108 - polycrystalline boron nitride 5 19 - TM carbides/nitrides 232 - zirconium/titanium borides 879 polishing, diamond abrasives 544 polyaluminocarbosilanes (PAIC) 45 1 ff polybenzene 274,278 polyborosilizanes 174 polycarbosilane (PCS) - boroncarbides 845 - polymer-ceramic transformations 446 f - silicon carbides 698 - thin film processing 463 pol ycrystalline boron nitrides 421,512,518 ff, 611 polycrystalline diamond 482 f, 512 ff, 611 polycrystalline grits 648 polycrystalline silicon carbides 685 polycrystalline superlattices 114 polycrystalline tungsten carbides 948 polyhedron structure, carbides 18 polymer to ceramic transformations 446-476 polymeric networks, carbon allotropes 272 polymer-pyrolysis derived (PP) fibers 697 f polyphase ceramic composites 289 ff polysilanes 446 f polysiloxanes 446 f polytitanocarbosilane (PTC) 447

Poole-Frenkel excitations 638 pore sizes, polycrystalline diamond 5 13 porosity 84,91 - boroncarbides 841 - residual 184 - self-propagating synthesis 339,346 - polyaluminocarbosilanes 455 - tungsten carbide-cobalt hardmetal 953 potentials, cemented borides 927 powder processing - alumina-based ceramics 65 1,665 f - carbides 12 - directed metal oxidation 293 - silicon carbides 691 precursors - diamond XLVIII - diamond-like carbon films 624 - polymer-ceramic transformations 447,453 f preforming, directed metal oxidation 307 preheating, self-propagating synthesis 366 preparation - boron carbides 837 f - carbonitrides LVI - diamond synthesis 391 ff - diamond-like carbon films 623 - SiAlON 460 - polyaluminocarbosilanes 452 ff - transition metal borides 875 - transition metal carbides/nitrides 207 f pressing 704 pressure, directed metal oxidation 291 pressure conditions, diamond anvil cell 43 pressure effects, self-propagating synthesis 347 pressure-temperature conditions, diamond films 390 pressure-temperature diagram, diamond synthesis 493 pressure-temperature regime, CO, laser 42 pressureless sintering - boroncarbides 841 - transition metal borides 876 process-related properties, alumina-based ceramics 654 processing 287-476 processing defects 91 processing zones, temperature effects 199 production - silicon carbide nanoceramics 709 - silicon carbides 709,717 production routes - alumina-based ceramics 650 - boroncarbides 838 - silicon carbides 688 ff protective scales, corrosion 141 pseudo-hot-isostatic pressing 345 f pseudopotential approach, carbon/silicon nitrides 256

Index purity boroncarbides 838 - diamond synthesis 501 - silicon carbides 690 pyrolysis - amorphous materials 36 - boroncarbides 843 pyrope glass 57 pyrophyllite gaskets 490 -

QQC deposition 490,508 quality control, silicon carbides quantum tunneling 29 1 quartz 377 Quasam 274 quaternary systems - borides 818 - carbonitrides 222 ff - diborides 831

734 ff

R8, carbon allotropes 273 f radiation effects, CVD diamond 60.5 radiofrequency diode sputtering 430 radiofrequency glow discharge deposition 623 rain drop impact, CVD diamond 584,588 Raman spectra - CVD diamond 412,575 - diamond anvil cell 48 - diamondfilms 394 - diamond XLVIII - silicon carbide-water system 383 - thin films 469 ff - tungsten carbides 165 Ramsdell notation 7 f random aggregates 5 12 rare earth hexaborides 1 I rare earth nitridosilicates 29 reactants - diamond synthesis 408 - self-propagating synthesis 342,366 reaction-bonded aluminum oxide (RBAO) 290,293 f reaction-bonded mullite (RBM) 293 reaction-bonded silicon carbides 700,72 1 reaction-bonded silicon nitride (RBSN) 290 ff, 749-801 reaction-boronizing sintering 920 reaction control, scale formation 145 reaction-sintered silicon carbide (RSSC) 290,293 reactive atmosphere pyrolysis 460 f reactive plasma techniques 210 reactor, microwave-plasma chemical vapor deposition 39 1 rebound hardness 107

1013

rebound test 81 recession rate, corrosion 152 reciprocating grinding 538 recrystallization - boroncarbides 842 - silicon carbides 700, 721 REFEL process 290,294 refractive index - boron nitrides 421,440 - CVD diamond 588,599 - diamond-like carbon 633,639 - silicon carbides 7 19 refractories, directed metal oxidation 289 refractory carbides 3 refractory grade silicon carbides 688 refractory materials, self-propagating synthesis 334 refractory metal wires 392 refractory metals 634 refractory nitrides 1 13 reinforced dies 490 reinforced grades, alumina-based ceramics 649,668 relative stability, carbon/silicon nitrides 255, 263,268 residual porosity 91, 184 residual stress 66 resin bond polishing head 545 resin-bonded tools, diamondboron nitrides 529 resins, boron carbides 842 f resistance curves 78,89 resistivity - boron nitrides 421 - data collection 965-995 - TM carbides/nitrides 203 - zirconium/titanium borides 879 Reuss-Voigt average 3 10 Reuss model 72 rhenium filaments 394 rhombohedral graphite 509 rhombohedral graphitic structures, carbon/silicon nitrides 256 rhombohedral structures XLIV - boron nitrides LVIII, 420 f - corundum 30 - silicon carbides 7,685 rigidity 104 roadway sawing 546 Rockwell hardness X I , 85 - titanium carbide reinforced alumina 674 f - tungsten carbide-cobalt hardmetal 946 room temperature - grain boundaries 195 ff - silicon nitrides 771 f - Youngmodulus 234 ruby fluorescence scale 43 rule-of-mixtures, elastic moduli 72

1014

Index

rupture modulus 72 directed metal oxidation 3 10 - polyaluminocarbosilanes 456 rupture strength 926 Rutherford backscattering (RBS) 21 1,395,46,4 rutile structures - nitrides 26 - titaniumoxides 32 -

o-bonds carbon allotropes 27 1 - diamond-like carbon 630 salt baths 210 salt melts corrosion - silicon carbides 157 - silicon nitrides 167 sample preparation, diamond anvil cell 42 sand erosion 588 sapphire 30, 186 ff, 597 ff saturation, silicon nitrides 761 ff sawing 542 SC24, carbon allotropes 275 scales - corrosion 141 - directed metal oxidation 305 f scanning electron microscopy (SEM) 54 - thin film processing 464 - cemented carbonitrides 240 - cemented ternary borides 926 - diamond 379,393,404 - silicon nitrides 777 f - titanium boride-iron composites 900 - titanium carbide-boron powder blend 862 - tungsten borides 885 scanning tunneling microscopy (STM) 636 Schemer equation - crystal stuctures 36 ff - nanocomposites 119 Schrodinger equation 276 scratch hardness XLI, 80, 89, 107 secondary ion mass spectrometry (SIMS) 21 1 - boronnitrides 434 - diamond synthesis 396 - thin film processing 464 secondary neutron mass spectrometry (SNMS) 211 Seebeck coefficient 854 Seebeck elements 737 seeding, alumina-based ceramics 652,661 selection criteria, diamond synthesis 401 self-propagating high-temperature synthesis (SHS) 322-373 self-propagating synthesis, boron carbides 839 self-scavenging reactions, diamond synthesis 408 -

self-sustaining high-temperature synthesis (SHS) 207 semi-empirical approaches, carbon/silicon nitirides 254 f semiconductor properties - bluediamond 500 - CVDdiamond 414 - silicon carbides 7 I9 SH1 composites, titanium carbide reinforced alumina 673 ff shaping techniques, silicon carbides 699,704 f shear modulus - boron carbide ceramics 865 - nanostructures 115 - titanium carbide 235 - zirconium/titanium borides 879 shear stress 105 shock wave production XLVlI shock wave tests - hydrothermal synthesis 374 - sapphire 186 - methane 60 short range order (SRO), carbides 16 shrinkage - boron carbides 841 - cemented ternary borides 92 1 - polycrystalline diamond 5 14 - silicon carbides 705 - polyaluminocarbosilanes 455 - titanium boride-iron composites 906 f Si3N4,nanostructures 104 SiAlON see: silicon aluminum oxynitride Sicalon 716 see also: siliconcarbides silica carbides, thin film processing 463 silica formers, corrosion 142 silica reactor tube 397 silicides 20 ff, 965-995 silicon, diamond synthesis 404 silicon aluminum oxynitride, pyrolysis 460 ff silicon aluminum oxynitride-bonded silicon carbides 700 silicon based materials LXV silicon borides 815 silicon carbide-titanium carbide-titanium diboride composites 883 f silicon carbide A1,OC 716 silicon carbide-based hard materials 683-748 silicon carbide-carbon systems 717 f silicon carbide metal matrix composites 614 silicon carbide-silicon carbide systems 717 f silicon carbides - alumina-based ceramics 648,668 - boron carbide ceramics 857 - corrosion 141, 156ff - crystal structures 6 f - directed metal oxidation 308 ff

fibers 446 f hydrotherinal treatment 3x2 ff - wlf-propugating synthehis 35 I - temperature effect\ I98 - thin film proce\\ing 3h3 silicon carbon nitrides 204 ff' corrosioii I73 f silicon carbon nitrogen slstem\ 253-270 silicon containing hystems. polymer-ceramic transformations 446 f vlicon nitride ba\ctl hard materials 739--801 silicon nitrides XL. 253-270 - bulk component\ 69 corrosion I60 ff cry\ial structures 23 - nanostructures I13 ff silicon oxides 55 silicon sputtering 856 polyaluminocarho~ilanes 452 ff silicothcrinic reduction 875 bingle crystal data I 8 8 single crystal diamond, applications 559 ff sinsle layers. self-propagating \ynthesis 366 single-phase ceramics X7X single-phase sintered corundum 670 single-point ignition. boron carbides 838 sintered grinding materials 638 f. 658 ff sintered pohder derived (SP)fibers 697 f \intcrecl reaction-honded \ilicon nitride (SRBSN) 749-801 sintered silicon carbides 700 ff. 721 sintered silicon nitrides (SSN) 749-801 sin tering - alumina ceramic\ I84 - alumina-based ceramic\ 6-5 1 - borides X I S . 828 boron carbides 839 ff cemented ternary boridcs 920 ff polycrystalline diamond 5 12,s I 6 titanium boridc-iron compo\ites 899 - transition metal borides 876 f slide rings 737 sliding tests 785 sliding wcar 723.729 ff slip casting 700. 704. 739 slip dislocations 95 \mall angle diffraction (SAD) pattern. silicon carbide-water system 385 small angle X-ray scattering (SAXS) 635 sodium 289 1 softening. nanostructures 1 I1 w - p e l derived corundum 66 I sol-gel technology 652 f solid-state laser3 31 solid-state reactions. titanitnn horide-iron composites 009 soliti-\tate sintering 703. 709. 72 I -

-

-

-

-

-

-

snlvcnt\. diainond synthc\i\ 398 f Soret effect, self-propagating rynthesis 341 +sp' bonding 274 \p' bonded carbon allotropes 271-285 sp' hybridization 104 space groups 4 ff borides 805, XI5 carbon nitrides 158 ccnicntcd ternary borides 9 I9 silicon nitrides 753 spacing, directed metal oxidation 30.5 .;palling - coatings 70 - corrmion 147 sphalerite 5tructure 173. 5 10 spin propagation 325, 336 \pincl\ crystal stuctures 31 - directed metal oxidation 292. 295 ff. 305 f - silicon nitrides 753 spruy drying 65 1 \pringh;ick. polyaluininocarbosilanes 452 sputtering - boron carbides 856 - boron nitrides 429 ff carbon nitrides 524 diamond-like carbon film\ 624 f hydrotherinal synthesis 374 ncNiCr 1 1 1 thin films 463 ST12. carbon allotropes 273 f stability - alumina-based ceramics 667 - cwbidcs 212 - carbon nitrides 255, 263. 268 CO, laser heating 54 ff diamond 380 perovjkite oxides 59 silicon nitridcs 255. 268 stable processes. diamond synthesis SO4 stacking faults 412 stacking sequences XLV borides 805 - boron nitrides 420 - diamond 5 , 510 - nitrides 2X - silicidcs 20 silicon carbides 7,686 TM carbidesinitrides 205 \tar of FZK 602 static properties, carbon allotropes 278 steel, elastic moduli 71 steel machining 660,667 ff \tepped faces, diamond synthesis 40 I stiffiie5s 66 JCB tr/.so: Young modulus stishovite structures XL. 33. 5.5 526 -

-

-

-

-

-

-

-

-

-

-

1016

Zndex

Stone-Wales transformations 282 stone machining 540 ff strain energy release rate (SERR) 76 strength - alumina-based ceramics 649 - boron carbides 852,865 - cemented ternary borides 926 - CVDdiamond 580ff - directed metal oxidation 31 1 - mechanical 66 f, 72 - polycrystalline boron nitride 519 - silicon carbides 7 18 - siliconnitrides 751, 771 - titanium boride-zirconia composites 890 - titanium boride-iron composites 913 strength-grain size relation, diboride composites 887 stress - CVD diamond 610 - residual 66f - silicon nitrides 776 stress intensity factor (SIF) 72,77,95 stress rupture 94 stress-strain characteristics, nanostructures 129 stress testing 186 ff Stribeck test data 729 ff strong tool ceramics 650 structural application, silicon nitrides 792 structural chemistry, hard materials 3 4 0 structures 1-285 - borides 804 - boroncarbides 854 - carbon allotropes 277 - carbon nitrides 261 - cemented ternary borides 919 - cobalt 949 - diamond-like carbon 632 - silicon carbides 685 ff - silicon carbon nitrides 265 - TM carbides/nitrides 204 ff - zirconium/titanium borides 879 see also: crystal structures subgranular residual porosity 91 sublattice structures, borides 811 submicrometer cutting 666 submount heat resistance 607 suboxides, boron LXV, 526 substrate based CVD fibers 697 substrate bias voltage 624 SuperFZK 601 superabrasive tools 479 ff supercritical water 379 superequilibrium atomic hydrogen 507 superhard materials 253-270 - nanostructured 104-139 Superior Graphite Company, silicon carbides 692

superlattices 114 superplastic silicon carbide sintered body 710 supersaturation, diamond synthesis 492 superstoichiometric carbides 127 surface adsorption, diamond synthesis 493 surface chemistry, polycrystalline diamond 5 16 surface damage 67 surface graphitization, diamond corrosion 155 surface states 183-201 SYNDITE 025 553 synthesis 287476 - boron carbides 838 f - boronnitrides LX - diamond XLVII, 504 ff - polycrystalline diamond 5 12 f - silicon carbides 692 - TM carbides/nitrides 207 f talc gaskets 490 talcum XL tantalum 498 tantalum carbides 12 tantalum filaments, diamond synthesis 394 tantalum-nickel-boron system 832 tantalum-nitrogen system 219 target materials, diamond-like carbon films 629 Tauc relation 638 technical applications - borides 813 f - cemented ternary borides 919 f - TM carbidednitrides 202 ff technical demands, alumina-based ceramics 660,667 technical grades, silicon carbides 687 technical scale production, boron carbides 838 technological properties 965-995 temperature strength 92 temperatures 183-201 - boroncarbides 841 - CO, laser heating 45.48 - CVDdiamond 582 - diamond-like carbon 633 - dielectric constants 604 - self-propagating synthesis 325 f - silicon carbides 688 - silicon nitrides 765 - tungsten carbide-cobalt hardmetal 953 tensile fracture, silicon carbides 7 10 tensile strength - CVDdiamond 588 - silicon carbides 697,718 ternary composites, borides 888 ternary systems - borides 818 ff - boron-metals 824 ff, 836 - phase diagrams 56 f

Index ternary transition metal carbides 213 ternary transition metal nitrides 221 Tersoff potential 255,261 testing 80ff tetragonal structures, cemented ternary borides 920 tetragonal zirconia 888 tetrahedral amorphous carbon 428 textural changes, melting temperatures 49,54 textured films 400 thermal barriers, directed metal oxidation 3 18 thermal conductivity - alumina-based ceramics 667 - boroncarbides 854 - boron nitrides 421 - CVD diamond 412,582 - data collection 965-995 - directed metal oxidation 289 - silicon carbides 721 - TM carbides/nitrides 225 thermal effects, CVD diamond 589 thermal emission, silicon carbides 46 thermal expansion coefficient (CTE) - alumina 69 - boroncarbides 854 - CVD diamond 588,610 - data collection 965-995 - polycrystalline boron nitride 5 19 - silicon carbides 718,721 - silicon nitrides 753 - TM carbides/nitrides 203,228 - zirconium/titanium borides 879 thermal grade, CVD diamond 575 thermal management, laser diode arrays 606 ff thermal migration, self-propagating synthesis 343 thermal properties, silicon carbides 720 thermal resistance, CVD diamond 607 f thermal shock - CVDdiamond 587 - directed metal oxidation 3 13 thermal shock resistance - silicon carbides 714 - silicon nitrides 749 thermal spikes, boron nitrides 427 thermal stress, CVD diamond 610 thermite process 838 thermochemistry, diamond synthesis 407 ff thermodynamic stability, diamond 380,504 thermodynamics - hydrothermal synthesis 378 - self-propagating synthesis 349 - silicon carbide-water system 382 - TM carbides/ni trides 2 12 ff thermoelastic effect, CVD diamond 589 thermogravimetric analysis (TGA) 36,446 - corrosion 151

1017

- directed metal oxidation 295 thermorefractive effect, CVD diamond 589 thin films processing 446,463 ff three dimensional carbon phases L tight-binding - carbon allotropes 273,276 - carbon/silicon nitrides 255 f time dependence, self-propagating synthesis 33 I titanium - diamond synthesis 498 - diamond-like carbon 634 - directed metal oxidation 305 titanium boride-iron composites 897 titanium boride-yttria-stabilized zirconia composites 892 titanium boride-zirconium boride system 879 titanium borides, physical properties 879 titanium-boron-nitrogen films 125 titanium-boron system 333,816 titanium carbide-boron system 825 titanium carbide-reinforced alumina 673 titanium carbides XL, 12 - alumina-based ceramics 669 - corrosion 174 - self-propagating synthesis 324,356 titanium-carbon-nitrogen system 222 titanium carbonitrides, corrosion 174 titanium-chromium mixed borides 833 titanium diboride-boron carbide system 880 titanium diboride-nickel, cobalt composites 914 titanium diboride-titanium carbide system 880 titanium diboride-transition metal boride composites 883 titanium diboride-based cermets 897 ff titanium diborides XL, 83 1 - boron carbide ceramics 861 - corrosion 176 titanium dicarbides, silicon carbides 71 1 titanium interlayers, diamond-like carbon films 641 titanium-iron-boron system 827 titanium-mol ybdenumxarbon-nitrogen system 222 titanium-nickel-boron system 827 titanium nitrides - corrosion 174 - diamond synthesis 500 f titanium-nitrogen system 216,323 titanium oxides, crystal stuctures 32 titanium silicides 330 titanium-tungsten-boron system 836 titanium-tungsten-carbon-nitrogen system 222 tools - applications 648482 - bonded 528 - superabrasive 479 f topaz XL

1018

Index

toroidal anvils 490 toughness - directed metal oxidation 31 1 - tungsten carbide-cobalt hardmetal 962 toxicology, silicon carbides 736 transformation processes, diamond synthesis 509 transformation toughening 92 transgranular fracture, silicon nitrides 774 transgranular microcracks 91 transition carbide systems 213 transition metal alloys 488,498 transition metal boride ceramics 874 ff transition metal borides, crystal structures 9 transition metal-boron system 8 17 transition metal carbides 202-252 - corrosion 174 - crystal structures 12 ff transition metal diboride cermets 915 ff transition metal diboride-silicon carbide composites 881 ff transition metal diborides 71 1,861 ff transition metal doping 305 transition metal nitrides 15, 202-252 transition metal nitrides - corrosion 174 - crystal structures 25 f - nanostructures 113 ff - wetting 131 transition metal-nitrogen systems 216 f transition metal oxides 32 f transition metal silicides 20 ff transition pressures, carbon/silicon nitrides 255 transition temperatures, silicon carbides 159 transmission electron microscopy (TEM) 185,446,464 - CVD diamond 576 f - diamond XLVIII - diamond-like carbon 634 - silicon nitrides 763 - tungsten borides 808 transport properties, borides 803 transverse rupture strength - cemented ternary borides 926 - directed metal oxidation 3 10 TRIBO 2000/-1, silicon carbides 729 ff tribological properties - boronnitrides 440 - diamond-like carbon films 639,642 ff - silicon carbides 723 ff trigonal structures 7 - prismatic 205 triple layers - directed metal oxidation 309 - self-propagating synthesis 366 triple point, boron nitride 5 11 tripling, tungsten silicides 21

truing tools 562 tungsten - diamond-like carbon 634 - transition metal diboride cermets 915 tungsten carbide-cobalt hardmetal 946-964 tungsten carbide-nickel, bulk components 68 tungsten carbides 12, 512 - bulkcomponents 68 - cobaltmetals 238 - corrosion 165 - diamond-like carbon films 629 - elasticmoduli 71 - grain size effects 197 - hardness 947 - self-propagating synthesis 324 - thermal expansion 228 - Youngmodulus 233 tungsten diborides 83 1 tungsten filaments - diamond synthesis 392 - silicon carbides 697 tungsten nitrde-aSi,N, system 124 tungsten silicides 20 turbostatic layers, boron nitride films 423 twins 183, 188, 196 - boroncarbides 849 type I/II CVD diamond 574 type Ib/IIab diamonds 498 type IIa single crystal diamonds 606 Tyranno fibers 383,697

ultrahard materials 479 ff, 521 ff ultrahigh pressure anvils 565 ultraviolet spectroscopy (UPS) 21 1 undefined cutting edges 658 unit cell - borides 804 - carbonnitrides 258 - crystal structures 8 ,2 0

vacancies 130 vacancy complexes 272 vacancy ordering 205 vacuum arc evaporation 627 vacuum hot gas extraction 21 1 vacuum sintering, boron carbides 847 valence bonds 633 valence electron concentration (VEC) 206,234 valence electrons 8 10 vanadium 634,924 vanadium borides 808 vanadium carbides 68,324 vanadium nitrogen system 217 vapor-liquid-solid process, silicon carbide whiskers 693

Index vapor-solid process, silicon carbide whiskers 693 f vapor phase deposition 42&445,692 vapor phase formation/condensation, silicon carbide whiskers 693 vapor phase reactions, titanium boride-iron composites 909 vented tube furnace 838 Vickers hardness XLI, 86, 183-201 - alumina-based ceramics 649,655 - CVDdiamond 588 - data collection 965-995 - nanostructures 105 - SiAlON 460f - titanium boride-titanium carbide system 880 - tungsten carbide-cobalt hardmetal 947 viscosity, boron carbides 846 vitrified bonded tools 529,532 voids - carbides 13 - nanostructures 110, 113 - polycrystalline diamond 5 13 - TM carbides/nitrides 205 Voigt bounds 71 volatile chlorides 208 Volmer equation 488

wetting - borides 803 - carbonitrides 202 - cemented borides 895 - diamond-like carbon films 644 - directed metal oxidation 307 - self-propagating synthesis 347 - titanium boride-iron composites 900 whiskers 92 - alumina-based ceramics 649,668 - silicon carbide-boron carbide ceramics 857 - silicon carbides 693,736 white corundum 658 wide band gap 441 Williams expansion 77 wire sawing 542 wood flooring manufacturing 613 work hardening model, nanostructures 110 working steels 658 wurtzite XLIV - berylliumoxide 31 - boron nitrides LVIII, 6 - silicon carbides 685,715 Wykoff parameters 255 X-ray diffraction (XRD) 9 1,446 boron carbides 163 - diamond 380,394 - diamond-like carbon 634 - self-propagating synthesis 353 - silicon carbides 686 X-ray spectroscopy (XPS) 36,211 -

wafers, silicon carbides 708 Wagner model 159 Warren-Averbach analysis 119 water vapor corrosion 162 wear parts - cemented borides 932 - CVDdiamond 611,617f wear properties 89 f - boronnitrides 440 - CVD diamond 4 10 - diamond-like carbon films 643 - directed metal oxidation 3 14,3 17 - silicon carbides 722 f - transition metal borides 878 - transition metal carbideshitrides 242 wear resistance - alumina-based ceramics 655,658,667 - silicon nitrides 75 I , 782 ff - tungsten carbide-cobalt hardmetal 962 Weibull modulus 73 , s I - CVD diamond 580 f, 590 - silicon nitrides 772,783 weight gain - boroncarbides 856 - directed metal oxidation 297 weight loss - silicon nitrides 789 f - polyaluminocarbosilanes 455 wet grinding 532

1019

yield stress XLI Young modulus 7 1,80 f, 93 - boron carbides 85 I , 854, 865 - boron nitrides 421 - cemented borides 928 - CVD diamond 581 f, 588 - data collection 965-995 - diamond-like carbon 639 - diboride composites 886 - directed metal oxidation 31 1 - nanostructures 110 - polycrystalline boron nitride 519 - silicon carbides 697,721 - polyaluminocarbosilanes 4.58 - titanium boride-iron composites 910 - TM carbides/nitrides 203,232 - zirconium/titanium borides 879 yttria-stabilized zirconia 892 yttrium-aluminum-garnet (YAG) 41 yttrium borides 805 yttrium oxides 787

Zeldovich criterion 336 zeta potential 846 Zhadanov notation 7 f zinc - diamond synthesis 488 - directed metal oxidation 289 zinc sulfide window 584 zincblende structures XLIV - boron nitrides LVIII, 6,420 ff, 510 - carbon nitrides LVI, 258 - ultrahard materials 521 zirconia 92

zirconium diamond synthesis 498 - directed metal oxidation 305 zirconium borides 879 zirconium-boron system 8 17 - self-propagating synthesis 333 zirconium-carbon-nitrogen system 222 zirconium corundum 658 zirconium dioxide 35 zirconium oxide/aluminumoxide/siliconoxide layers 309 ZnSe windows 589 f -

Handbook of Ceramic Hard Materials

Recommend Documents