Handbook of Deposition Technologies for Films and Coatings

HANDBOOK OF DEPOSITION TECHNOLOGIES FOR FILMS AND COATINGS Science, Technology and Applications

Second Edition Edited by

Rointan F. Bunshah University of California at Los Angeles Los Angeles, California

np

NOYES PUBLICATIONS Park Ridge, New Jersey, U.S.A.

Copyright © 1994 by Noyes Publications No part of this book may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording or by any information storage and retrieval system, without permission in writing from the Publisher. Library of Congress Catalog Card Number: 93-30751 ISBN: 0-8155-1337-2 Printed in the United States Published in the United States of America by Noyes Publications Mill Road, Park Ridge, New Jersey 07656 10 9 8 7 6 5 4 3 2 1

Library of Congress Cataloging-in-Publication Data Handbook of deposition technologies for films and coatings / edited by Rointan F. Bunshah. -- 2nd ed. p. cm. Rev. ed of: Deposition technologies for films and coatings. c1982. Includes bibliographical references and index. ISBN 0-8155-1337-2 1. Coating processes. I. Bunshah, R. F. (Rointan Framroze) II. Title: Deposition technologies for films and coatings. TP156.C57H38 1994 667' .9--dc20 9 3 -30751 CIP

DEDICATION

This volume is dedicated to Professor John Thornton for his many pioneering contributions to thin film science and technology which have inspired so many of the scientists and engineers working in this field.

vii

MATERIALS SCIENCE AND PROCESS TECHNOLOGY SERIES Editors Rointan F. Bunshah, University of California, Los Angeles (Series Editor) Gary E. McGuire, Microelectronics Center of North Carolina (Series Editor) Stephen M. Rossnagel, IBM Thomas J. Watson Research Center (Consulting Editor)

Electronic Materials and Process Technology HANDBOOK OF DEPOSITION TECHNOLOGIES FOR FILMS AND COATINGS, Second Edition: edited by Rointan F. Bunshah CHEMICAL VAPOR DEPOSITION FOR MICROELECTRONICS: by Arthur Sherman SEMICONDUCTOR MATERIALS AND PROCESS TECHNOLOGY HANDBOOK: edited by Gary E. McGuire HYBRID MICROCIRCUIT TECHNOLOGY HANDBOOK: by James J. Licari and Leonard R. Enlow HANDBOOK OF THIN FILM DEPOSITION PROCESSES AND TECHNIQUES: edited by Klaus K. Schuegraf IONIZED-CLUSTER BEAM DEPOSITION AND EPITAXY: by Toshinori Takagi DIFFUSION PHENOMENA IN THIN FILMS AND MICROELECTRONIC MATERIALS: edited by Devendra Gupta and Paul S. Ho HANDBOOK OF CONTAMINATION CONTROL IN MICROELECTRONICS: edited by Donald L. Tolliver HANDBOOK OF ION BEAM PROCESSING TECHNOLOGY: edited by Jerome J. Cuomo, Stephen M. Rossnagel, and Harold R. Kaufman CHARACTERIZATION OF SEMICONDUCTOR MATERIALS, Volume 1: edited by Gary E. McGuire HANDBOOK OF PLASMA PROCESSING TECHNOLOGY: edited by Stephen M. Rossnagel, Jerome J. Cuomo, and William D. Westwood HANDBOOK OF SEMICONDUCTOR SILICON TECHNOLOGY: edited by William C. O’Mara, Robert B. Herring, and Lee P. Hunt HANDBOOK OF POLYMER COATINGS FOR ELECTRONICS, 2nd Edition: by James Licari and Laura A. Hughes HANDBOOK OF SPUTTER DEPOSITION TECHNOLOGY: by Kiyotaka Wasa and Shigeru Hayakawa HANDBOOK OF VLSI MICROLITHOGRAPHY: edited by William B. Glendinning and John N. Helbert CHEMISTRY OF SUPERCONDUCTOR MATERIALS: edited by Terrell A. Vanderah CHEMICAL VAPOR DEPOSITION OF TUNGSTEN AND TUNGSTEN SILICIDES: by John E. J. Schmitz ELECTROCHEMISTRY OF SEMICONDUCTORS AND ELECTRONICS: edited by John McHardy and Frank Ludwig

v

vi

Contents Series

HANDBOOK OF CHEMICAL VAPOR DEPOSITION: by Hugh O. Pierson DIAMOND FILMS AND COATINGS: edited by Robert F. Davis ELECTRODEPOSITION: by Jack W. Dini HANDBOOK OF SEMICONDUCTOR WAFER CLEANING TECHNOLOGY: edited by Werner Kern CONTACTS TO SEMICONDUCTORS: edited by Leonard J. Brillson HANDBOOK OF MULTILEVEL METALLIZATION FOR INTEGRATED CIRCUITS: edited by Syd R. Wilson, Clarence J. Tracy, and John L. Freeman, Jr. HANDBOOK OF CARBON, GRAPHITE, DIAMONDS AND FULLERENES: by Hugh O. Pierson

Ceramic and Other Materials—Processing and Technology SOL-GEL TECHNOLOGY FOR THIN FILMS, FIBERS, PREFORMS, ELECTRONICS AND SPECIALTY SHAPES: edited by Lisa C. Klein FIBER REINFORCED CERAMIC COMPOSITES: edited by K. S. Mazdiyasni ADVANCED CERAMIC PROCESSING AND TECHNOLOGY, Volume 1: edited by Jon G. P. Binner FRICTION AND WEAR TRANSITIONS OF MATERIALS: by Peter J. Blau SHOCK WAVES FOR INDUSTRIAL APPLICATIONS: edited by Lawrence E. Murr SPECIAL MELTING AND PROCESSING TECHNOLOGIES: edited by G. K. Bhat CORROSION OF GLASS, CERAMICS AND CERAMIC SUPERCONDUCTORS: edited by David E. Clark and Bruce K. Zoitos HANDBOOK OF INDUSTRIAL REFRACTORIES TECHNOLOGY: by Stephen C. Carniglia and Gordon L. Barna CERAMIC FILMS AND COATINGS: edited by John B. Wachtman and Richard A. Haber

Related Titles ADHESIVES TECHNOLOGY HANDBOOK: by Arthur H. Landrock HANDBOOK OF THERMOSET PLASTICS: edited by Sidney H. Goodman SURFACE PREPARATION TECHNIQUES FOR ADHESIVE BONDING: by Raymond F. Wegman FORMULATING PLASTICS AND ELASTOMERS BY COMPUTER: by Ralph D. Hermansen HANDBOOK OF ADHESIVE BONDED STRUCTURAL REPAIR: by Raymond F. Wegman and Thomas R. Tullos CARBON–CARBON MATERIALS AND COMPOSITES: edited by John D. Buckley and Dan D. Edie CODE COMPLIANCE FOR ADVANCED TECHNOLOGY FACILITIES: by William R. Acorn

Contributors

Rointan F. Bunshah Department of Materials Science and Engineering University of California at Los Angeles Los Angeles, California Jan-Otto Carlsson Department of Chemistry Upsala University Upsala, Sweden Joseph E. Greene Coordinated Science Laboratory University of Illinois at UrbanaChampaign Urbana, Illinois Bret L. Halpern Jet Process Corporation New Haven, Connecticut

Donald M. Mattox Society of Vacuum Coaters Albuquerque, New Mexico Gary E. McGuire Microelectronics Center of North Carolina Research Triangle Park, North Carolina Jerome C. Schmitt Jet Process Corporation New Haven, Connecticut Morton Schwartz Electrochemical/Metal Finishing Consultant Los Angeles, California Arthur Sherman Consultant Palo Alto, California

xiii

xiv

Contents Contributors

John A. Thornton* Coordinated Science Laboratory University of Illinois at UrbanaChampaign Urbana, Illinois *

Robert C. Tucker, Jr. Praxair Surface Technologies, Inc. Indianapolis, Indiana

Professor Thornton died unexpectedly in November, 1987.

NOTICE To the best of our knowledge the information in this publication is accurate; however the Publisher does not assume any responsibility or liability for the accuracy or completeness of, or consequences arising from, such information. This book is intended for informational purposes only. Mention of trade names or commercial products does not constitute endorsement or recommendation for use by the Publisher. Final determination of the suitability of any information or product for use contemplated by any user, and the manner of that use, is the sole responsibility of the user. We recommend that anyone intending to rely on any recommendation of materials or procedures mentioned in this publication should satisfy himself as to such suitability, and that he can meet all applicable safety and health standards.

Preface to the Second Edition

A decade after the first edition of this volume was published, a second edition is being brought out partly due to the excellent response to the first edition and also to update the many improvements in deposition technologies, the mechanisms and applications. The entire volume has been extensively revised and contains 50% or more new material. Five entirely new chapters have been added. The organization of the book has also been changed in the following respects: 1. Considerably more material has been added in Plasma Assisted Vapor Deposition Processes. 2. A new chapter on Metallurgical Coating Applications has been added. The chapter in the first edition on Polymeric Coating techniques has been omitted as it deserves a volume by itself. Large topics such as coatings technology in microelectronics, diamond films, etc., have been treated in separate volumes in this series. Although there are some new competing volumes dealing with selected topics on the materials science of thin films, this volume remains the only comprehensive treatment of the entire subject of Deposition Technology. Applications of films and coatings spans the entire gamut of science and technology. Generic application areas include electronic, magnetic, optical, mechanical, chemical and decorative applications. New deposition technologies such as arc evaporation, unbalanced magnetron sputtering, ion beam assisted deposition, and metal-organic CVD have come on stream for critical applications. In this post cold war era, many economic solutions to engineering problems will necessarily involve coatings, e.g., battery materials for the emerging electric car industry.

ix

x

Preface Contents

The core subjects are the basic technologies for the deposition of films and coatings. These are the Physical Vapor Deposition (PVD) Processes consisting of Evaporation, Sputtering, and Ion Plating; Chemical Vapor Deposition (CVD) and Plasma-Assisted Chemical Vapor Deposition (PACVD); Electrodeposition and Electroless Plating; Thermal Spraying, Plasma Spraying and Detonation Gun Technologies. Chapters on other subjects common to the above technologies are included. These are: Adhesion of Coatings, Cleaning of Substrates, Role of Plasmas in Deposition Processes, Structure of PVD Deposits, Growth and Structure of PVD Films, Mechanical and Tribological Properties of PVD Deposits, Elemental and Structural Characterization Techniques, and Metallurgical Coatings. A relatively new development, Jet Vapor Deposition Process, was added as the last chapter in the book during the page proof stage because of its novelty. We hope that this volume will be useful to the multitude of disciplines represented by the workers in this field and provide a source for future developments. University of California Los Angeles, California June, 1993

Rointan F. Bunshah

Preface to the First Edition

Almost universally in high technology applications, a composite material is used where the properties of the surface are intentionally different from those of the core. Thus, materials with surface coatings are used in the entire crosssection of applications ranging from microelectronics, display devices, chemical corrosion, tribology including cutting tools, high temperature oxidation/ corrosion, solar cells, thermal insulation and decorative coatings (including toys, automobile components, watch cases, etc.). A large variety of materials is used to produce these coatings. They are metals, alloys, refractory compounds (e.g., oxides, nitrides, carbides), intermetallic compounds (e.g., GaAg) and polymers in single or multiple layers. The thickness of the coatings ranges from a few atom layers to millions of atom layers. The microstructure and hence the properties of the coatings can be varied widely and at will, thus permitting one to design new material systems with unique properties. (A material system is defined as the combination of the substrate and coating.) Historically, coating technology evolved and developed in the last 30 years in several industries, i.e., decorative coatings, microelectronics and metallurgical coatings. They used similar techniques but only with the passage of time have the various approaches reached a common frontier resulting in much useful cross-fertilization. That very vital process is proceeding ever more strongly at this time. With this background in mind, a short course on Deposition Technologies and their applications was developed and given on five consecutive occasions in the last three years. This volume is based on the material used in the course.

xi

xii

It comprises chapters dealing with the various coating techniques, the resulting microstructure, properties and applications. The specific techniques covered are evaporation, ion plating, sputtering, chemical vapor deposition, electrodeposition from aqueous solution, plasma and detonation gun coating techniques, and polymeric coatings. In addition several other chapters are added. Plasmas are used in many of the deposition processes and therefore a special chapter on this topic has been added. Cleaning of the substrate and the related topic of adhesion of the coating are common to many processes and a brief exposé of this topic is presented. Characterization of the films, i.e., composition, impurities, crystal structure and microstructure are essential to the understanding of the various processes. Two chapters dealing with this area are included. Finally, a chapter on application of deposition techniques in microelectronics is added to give one example of the use of several of these techniques in a specific area. This volume represents a unique collection of our knowledge on Deposition Technologies and their applications up to and including the state-of-the-art. It is hoped that it will be very useful to students, practicing engineers and managerial personnel who have to learn about this essential area of modern technology. University of California Los Angeles, California April 1982

R. F. Bunshah

Contents

xv

Contents

1

Deposition Technologies: An Overview ....................... 27 Rointan F. Bunshah

1.0 2.0 3.0 4.0 5.0

THE MARKET .............................................................................. 27 INTRODUCTION ........................................................................... 28 AIM AND SCOPE ......................................................................... 30 DEFINITIONS AND CONCEPTS ................................................... 31 PHYSICAL VAPOR DEPOSITION (PVD) PROCESS TERMINOLOGY ........................................................................... 32 6.0 CLASSIFICATION OF COATING PROCESSES ........................... 34 7.0 GAS JET DEPOSITION WITH NANO-PARTICLES ....................... 36 8.0 MICROSTRUCTURE AND PROPERTIES ..................................... 38 9.0 UNIQUE FEATURES OF DEPOSITED MATERIALS AND GAPS IN UNDERSTANDING ................................................................... 40 10.0 CURRENT APPLICATIONS .......................................................... 41 10.1 Decorative/Functional Coating ............................................. 41 10.2 High Temperature Corrosion ................................................ 42 10.3 Environmental Corrosion ..................................................... 42 10.4 Friction and Wear ............................................................... 42 10.5 Materials Conservation ........................................................ 43 10.6 Cutting Tools ...................................................................... 43 10.7 Nuclear Fuels ..................................................................... 44 10.8 Biomedical Uses ................................................................. 44

xv

xvi

Contents

10.9 Electrical Uses ................................................................... 44 11.0 “FRONTIER AREAS” FOR THE APPLICATION OF THE PRODUCTS OF DEPOSITION TECHNOLOGY ..................... 44 12.0 SELECTION CRITERIA ................................................................. 46 13.0 SUMMARY ................................................................................... 48 APPENDIX 1: DEPOSITION PROCESS DEFINITIONS........................... 49 Conduction and Diffusion Processes............................................. 49 Chemical processes ..................................................................... 50 Wetting Process........................................................................... 50 Spraying Processes ..................................................................... 51 REFERENCES ...................................................................................... 54

2

Plasmas in Deposition Processes .............................. 55 John A. Thornton and Joseph E. Greene

1.0 2.0

INTRODUCTION ........................................................................... 55 PARTICLE MOTION ..................................................................... 56 2.1 Mean Free Path and Collision Cross Sections .................... 56 2.2 Free Electron Kinetic Energy in a Plasma........................... 58 2.3 Electron Energy Distribution Functions ............................... 59 2.4 Collision Frequencies .......................................................... 61 3.0 COLLECTIVE PHENOMENA ........................................................ 68 3.1 Plasma Sheaths ................................................................. 69 3.2 Ambipolar Diffusion ............................................................. 74 3.3 Plasma Oscillations ............................................................ 75 4.0 PLASMA DISCHARGES .............................................................. 76 4.1 Introduction ......................................................................... 76 4.2 Ionization Balances and the Paschen Relation .................... 77 4.3 Cold Cathode Discharges ................................................... 82 4.4 Magnetron Discharges ........................................................ 84 4.5 RF Discharges .................................................................... 85 5.0 PLASMA VOLUME REACTIONS ................................................. 87 5.1 Introduction ......................................................................... 87 5.2 Electron/Atom Interactions .................................................. 87 5.3 Electron/Molecule Interactions ............................................ 88 5.4 Metastable Species ............................................................ 90 5.5 Applications of Volume Reactions....................................... 92 6.0 SURFACE REACTIONS ............................................................... 93 6.1 Introduction ......................................................................... 93 6.2 Ion Bombardment ................................................................ 93 6.3 Electron Bombardment ..................................................... 100 6.4 Glow Discharge Surface Cleaning and Activation .............. 100 REFERENCES .................................................................................... 103

Contents

3

xvii

Surface Preparation for Film and Coating Deposition Processes ................................................................ 108 Donald M. Mattox

1.0 2.0

INTRODUCTION ......................................................................... 108 CONTAMINATION ...................................................................... 110 2.1 Recontamination ............................................................... 111 3.0 ENVIRONMENT CONTROL ........................................................ 113 4.0 CLEANING PROCESSES .......................................................... 119 4.1 Particulate Removal .......................................................... 120 4.2 Abrasive Cleaning ............................................................. 121 4.3 Etch Cleaning ................................................................... 121 4.4 Fluxing .............................................................................. 122 4.5 Alkaline Cleaners .............................................................. 122 4.6 Detergent Cleaning ........................................................... 122 4.7 Chelating Agents .............................................................. 123 4.8 Solvent Cleaning ............................................................... 123 4.9 Oxidation Cleaning............................................................ 128 4.10 Volatilization Cleaning....................................................... 130 4.11 Hydrogen Reduction Cleaning ........................................... 130 4.12 Electrolytic Cleaning ......................................................... 131 5.0 DRYING AND OUTGASSING ..................................................... 132 6.0 MONITORING OF CLEANING .................................................... 133 7.0 IN SITU CLEANING .................................................................... 134 7.1 Ion Scrubbing .................................................................... 134 8.0 PLASMAS .................................................................................. 134 8.1 Generation of Plasmas ..................................................... 135 8.2 Plasma Chemistry ............................................................ 140 8.3 Bombardment Effects on Surfaces .................................... 141 8.4 Sputter Cleaning and Etching............................................ 143 9.0 STORAGE AND HANDLING ....................................................... 147 10.0 ACTIVATION AND SENSITIZATION ............................................ 148 11.0 SURFACE MODIFICATION ........................................................ 150 12.0 PASSIVATION AND PRESERVATION ....................................... 151 13.0 SAFETY ..................................................................................... 152 REFERENCES .................................................................................... 152

4

Evaporation: Processes, Bulk Microstructures and Mechanical Properties .............................................. 157 Rointan F. Bunshah

1.0 2.0

GENERAL INTRODUCTION ........................................................ 157 SCOPE ...................................................................................... 159

xviii

3.0

Contents

PVD PROCESSES .................................................................... 159 3.1 Preamble .......................................................................... 159 3.2 PVD Processes ................................................................ 160 3.3 Advantages and Limitations .............................................. 165 4.0 THEORY AND MECHANISMS ................................................... 166 4.1 Vacuum Evaporation ......................................................... 166 5.0 EVAPORATION PROCESS AND APPARATUS ......................... 169 5.1 The System ...................................................................... 169 6.0 EVAPORATION SOURCES ....................................................... 172 6.1 General Considerations ..................................................... 172 6.2 Resistance Heated Sources ............................................. 175 6.3 Sublimation Sources ......................................................... 176 6.4 Evaporation Source Materials............................................ 178 6.5 Induction Heated Sources ................................................. 180 6.6 Electron Beam Heated Sources ........................................ 181 6.7 Arc Evaporation ................................................................ 189 7.0 LASER INDUCED EVAPORATION/LASER ABLATION/PULSED LASER DEPOSITION (PLD) ....................................................... 192 8.0 DEPOSITION RATE MONITORS AND PROCESS CONTROL .... 194 8.1 Monitoring of the Vapor Stream ......................................... 194 8.2 Monitoring of Deposited Mass ........................................... 196 8.3 Monitoring of Specific Film Properties ............................... 196 8.4 Evaporation Process Control ............................................. 199 9.0 DEPOSITION OF VARIOUS MATERIALS .................................. 201 9.1 Deposition of Metals and Elemental Semiconductors ........ 201 9.2 Deposition of Alloys .......................................................... 201 9.3 Deposition of Intermetallic Compounds ............................. 205 9.4 Deposition of Refractory Compounds ................................ 209 9.5 Reactive Evaporation Process ........................................... 213 9.6 Activated Reactive Evaporation (ARE) ............................... 213 9.7 Materials Synthesized by Evaporation-based Processes .. 223 10.0 MICROSTRUCTURE OF PVD CONDENSATES ......................... 224 10.1 Microstructure Evolution .................................................... 224 10.2 Texture ............................................................................. 236 10.3 Residual Stresses ............................................................ 237 10.4 Defects ............................................................................. 237 11.0 PHYSICAL PROPERTIES OF THIN FILMS ................................ 241 12.0 MECHANICAL AND RELATED PROPERTIES ............................ 241 12.1 Mechanical Properties ................................................................ 241 13.0 PURIFICATION OF METALS BY EVAPORATION ...................... 256 APPENDIX ......................................................................................... 258 On Progress in Scientific Investigations in the Field of Vacuum Evaporation in the Soviet Union................................................... 258 REFERENCES .................................................................................... 261

Contents

5

xix

Sputter Deposition Processes .................................. 275 John A. Thornton and Joseph E. Greene

1.0

INTRODUCTION ......................................................................... 275 1.1 Sputter Deposition Systems ............................................. 278 1.2 Sputter-Deposition Applications ........................................ 279 1.3 Process Implementation ................................................... 282 1.4 History of Sputter Deposition and Background Reading .... 283 2.0 SPUTTERING MECHANISMS .................................................... 284 2.1 Sputtering Rate ................................................................. 285 2.2 Momentum Exchange ....................................................... 289 2.3 Alloys and Compounds ..................................................... 292 2.4 Sputtering with Reactive Species ...................................... 295 2.5 The Nature of Sputtered Species ...................................... 296 2.6 Energy Distribution of Sputtered Species .......................... 298 3.0 SPUTTER DEPOSITION TECHNIQUES ..................................... 301 3.1 Planar Diode and the DC Glow Discharge ......................... 301 3.2 Triode Discharge Devices .................................................. 305 3.3 Magnetrons ....................................................................... 306 3.4 RF Sputtering ................................................................... 318 3.5 Ion-Beam Sputtering ......................................................... 327 4.0 SPUTTER DEPOSITION MODES ............................................... 328 4.1 Reactive Sputtering ........................................................... 328 4.2 Bias Sputtering ................................................................. 332 REFERENCES .................................................................................... 337

6

Ion Plating ................................................................. 346 Donald M. Mattox

1.0 2.0 3.0

4.0 5.0

INTRODUCTION ......................................................................... 346 PROCESSING PLASMA ............................................................ 351 GENERATION OF PLASMAS .................................................... 351 3.1 DC Diode Discharge.......................................................... 351 3.2 RF Discharge .................................................................... 355 3.3 Microwave Discharges ...................................................... 356 3.4 Electron Emitter Discharge ............................................... 356 3.5 Magnetron Discharges ...................................................... 357 3.6 Plasma Enhancement ....................................................... 358 PLASMA CHEMISTRY ............................................................... 359 BOMBARDMENT EFFECTS ON SURFACES ............................ 360 5.1 Collisional Effects ............................................................. 363 5.2 Surface Region Effects ..................................................... 368 5.3 Near Surface Region Effects ............................................. 369 5.4 Bulk Effects ...................................................................... 369

xx

Contents

6.0

SOURCES OF DEPOSITING ATOMS ........................................ 369 6.1 Thermal Vaporization ........................................................ 370 6.2 Sputtering ......................................................................... 371 6.3 Vacuum Arcs .................................................................... 371 6.4 Chemical Vapor Precursors .............................................. 373 7.0 REACTIVE ION PLATING ........................................................... 373 8.0 BOMBARDMENT EFFECTS ON FILM PROPERTIES ................ 373 8.1 Effects: Adatom Nucleation............................................... 373 8.2 Effects: Interface Formation .............................................. 374 8.3 Effects: Film Growth ......................................................... 374 8.4 Film Adhesion................................................................... 376 8.5 Film Morphology/Density .................................................. 376 8.6 Residual Film Stress ........................................................ 378 8.7 Crystallographic Orientation .............................................. 378 8.8 Gas Incorporation .............................................................. 380 8.9 Surface Coverage .............................................................. 380 8.10 Other Properties ............................................................... 381 9.0 ION PLATING SYSTEM REQUIREMENTS ................................. 381 9.1 Vacuum System ............................................................... 381 9.2 High Voltage Components ................................................ 381 9.3 Gas Handling System ....................................................... 383 9.4 Evaporation/Sublimation Sources ...................................... 383 9.5 Sputtering Sources ........................................................... 383 9.6 Plasma Uniformity ............................................................ 384 9.7 Plasma Generation Near the Substrate Surface ................ 384 9.8 Substrate Fixturing ........................................................... 384 10.0 PROCESS MONITORING AND CONTROL ................................. 385 10.1 Plasma ............................................................................. 385 10.2 Substrate Temperature ..................................................... 385 10.3 Specifications ................................................................... 385 11.0 PROBLEM AREAS .................................................................... 386 12.0 APPLICATIONS.......................................................................... 389 13.0 SUMMARY ................................................................................. 389 REFERENCES .................................................................................... 391

7

Chemical Vapor Deposition ...................................... 400 Jan-Otto Carlsson

1.0 2.0 3.0

INTRODUCTION ......................................................................... 400 IMPORTANT REACTION ZONES IN CVD ................................... 401 DESIGN OF CVD EXPERIMENTS .............................................. 402 3.1 Classification of CVD Reactions........................................ 403 3.2 Thermodynamics .............................................................. 405 3.3 Adhesion .......................................................................... 409

Contents

xxi

3.4 Substrate Cleaning Procedures ......................................... 410 3.5 The CVD system .............................................................. 410 3.6 The Gas Dispensing System ............................................ 411 3.7 The Reactor ...................................................................... 413 3.8 The Exhaust System ........................................................ 415 3.9 Analysis of the Vapor in a CVD Reactor............................ 417 4.0 GAS FLOW DYNAMICS ............................................................ 417 4.1 Gas Flow Patterns ............................................................ 420 4.2 Boundary Layers ............................................................... 423 4.3 Mass Transport Processes Across a Boundary Layer ....... 428 5.0 RATE-LIMITING STEPS DURING CVD ....................................... 428 6.0 REACTION MECHANISMS ........................................................ 436 7.0 NUCLEATION ............................................................................. 438 8.0 SURFACE MORPHOLOGY AND MICROSTRUCTURE OF CVD MATERIALS ............................................................................... 442 9.0 SELECTIVE DEPOSITION .......................................................... 445 9.1 Area-Selective Growth ....................................................... 446 9.2 Phase-Selective Deposition ............................................... 452 10.0 SOME APPLICATIONS OF THE CVD TECHNIQUE ................... 453 11.0 OUTLOOK .................................................................................. 455 REFERENCES .................................................................................... 456

8

Plasma-Enhanced Chemical Vapor Deposition ........ 460

Arthur Sherman INTRODUCTION ......................................................................... 460 REACTOR INFLUENCE ON PLASMA BEHAVIOR ..................... 461 2.1 DC/AC Glow Discharges ................................................... 461 2.2 AC Discharges with Unequal Area Electrodes ................... 464 2.3 Frequency Effects on RF Plasma Reactor Behavior .......... 466 2.4 Adjusting DC Bias for Fixed Electrode Geometry .............. 467 2.5 Plasma-Enhanced CVD (PECVD) Reactors ...................... 467 3.0 FILMS DEPOSITED BY CVD ..................................................... 472 3.1 Silicon Nitride ................................................................... 472 3.2 Silicon Dioxide .................................................................. 478 3.3 Conducting Films .............................................................. 481 REFERENCES .................................................................................... 482 1.0 2.0

9

Plasma-Assisted Vapor Deposition Processes: Overview ................................................................... 485 Rointan F. Bunshah

1.0 2.0 3.0

INTRODUCTION ......................................................................... 485 PLASMA-ASSISTED DEPOSITION PROCESSES ..................... 488 MODEL OF A DEPOSITION PROCESS..................................... 488

xxii

Contents

4.0

MATERIALS DEPOSITED BY REACTIVE VAPOR DEPOSITION PROCESSES ............................................................................. 491 5.0 KEY ISSUES IN PLASMA-ASSISTED REACTIVE VAPOR DEPOSITION PROCESSES....................................................... 492 5.1 Plasma Volume Chemistry ............................................... 492 5.2 Type and Nature of the Bombardment of the Growing Film 493 6.0 PLASMA-ASSISTED DEPOSITION TECHNIQUES IN CURRENT USAGE ...................................................................................... 495 6.1 Plasma-Assisted Chemical Vapor Deposition ................... 495 6.2 Sputter Deposition ............................................................ 496 6.3 Activated Reactive Evaporation (ARE) ............................... 497 7.0 LIMITATIONS OF CURRENT PLASMA-ASSISTED TECHNIQUES 499 8.0 HYBRID PROCESSES ............................................................... 501 9.0 CONCLUSIONS .......................................................................... 501 REFERENCES .................................................................................... 505

10

Deposition from Aqueous Solutions: An Overview ..... 506 Morton Schwartz

1.0 2.0 3.0

INTRODUCTION ......................................................................... 506 GENERAL PRINCIPLES ............................................................ 508 ELECTRODEPOSITION.............................................................. 520 3.1 Mechanism of Deposition .................................................. 520 3.2 Parameters ....................................................................... 526 4.0 PROCESSING TECHNIQUES .................................................... 536 5.0 SELECTION OF DEPOSIT ......................................................... 539 5.1 Individual Metals ............................................................... 539 5.2 Alloy Deposition ................................................................ 543 6.0 SELECTED SPECIAL PROCESSES ......................................... 550 6.1 Electroless Deposition ...................................................... 550 6.2 Electroforming................................................................... 557 6.3 Anodizing .......................................................................... 560 6.4 Plating on Plastics............................................................ 570 6.5 Plating Printed Circuit Boards ........................................... 571 7.0 STRUCTURES AND PROPERTIES OF DEPOSITS ................... 574 8.0 SUMMARY ................................................................................. 596 APPENDIX A - Preparation of Substrates for Electroplating .................. 597 APPENDIX B - Representative Electroless Plating Solution Formulation .................................................... 599 APPENDIX C - Comparison of Aluminum Anodizing Processes (Types I, II and III) ......................................................... 602 REFERENCES .................................................................................... 605

Contents

11

xxiii

Advanced Thermal Spray Deposition Techniques ..... 617 Robert C. Tucker, Jr.

1.0 2.0

INTRODUCTION ......................................................................... 617 EQUIPMENT AND PROCESSES ............................................... 618 2.1 Plasma Spray Process ..................................................... 618 2.2 Detonation Gun Deposition Process ................................. 626 2.3 High Velocity Oxy-Fuel Deposition.................................... 628 2.4 Thermal Control ................................................................ 629 2.5 Auxiliary Equipment .......................................................... 630 2.6 Equipment-Related Coating Limitations............................. 631 3.0 TOTAL COATING PROCESS ..................................................... 632 3.1 Powder ............................................................................. 632 3.2 Substrate Preparation ....................................................... 632 3.3 Masking ............................................................................ 633 3.4 Coating ............................................................................. 633 3.5 Finishing ........................................................................... 635 4.0 COATING STRUCTURE AND PROPERTIES .............................. 636 4.1 Surface Macrostructure and Microstructure ....................... 636 4.2 Microstructure................................................................... 637 4.3 Bond Strength ................................................................... 643 4.4 Residual Stress ................................................................ 644 4.5 Density ............................................................................. 645 4.6 Mechanical Properties ...................................................... 647 4.7 Wear and Friction ............................................................. 653 4.8 Corrosion Properties ......................................................... 660 4.9 Thermal Properties ............................................................ 662 4.10 Electrical Characteristics .................................................. 664 5.0 SUMMARY ................................................................................. 665 REFERENCES .................................................................................... 665

12

Non-Elemental Characterization of Films and Coatings ............................................................ 669 Donald M. Mattox

1.0 2.0 3.0 4.0 5.0

INTRODUCTION ......................................................................... 669 CHARACTERIZATION ................................................................ 671 FILM FORMATION ..................................................................... 677 ELEMENTAL AND STRUCTURAL ANALYSIS ............................ 681 SOME PROPERTY MEASUREMENTS ..................................... 682 5.1 Adhesion .......................................................................... 682 5.2 Film Thickness ................................................................. 689 5.3 Film Stress ....................................................................... 691 5.4 Coefficient of Thermal Expansion ...................................... 695

xxiv

Contents

5.5 Mechanical Properties ...................................................... 695 5.6 Electrical Resistivity.......................................................... 696 5.7 Temperature Coefficient of Resistivity (TCR) ...................... 696 5.8 Electromigration ................................................................ 697 5.9 Density ............................................................................. 697 5.10 Porosity ............................................................................ 698 5.11 Chemical Etch Rate (Dissolution) ..................................... 701 6.0 SUMMARY ................................................................................. 701 REFERENCES .................................................................................... 702

13

Nucleation, Film Growth, and Microstructural Evolution ................................................................... 707 Joseph E. Greene

1.0 2.0

INTRODUCTION ......................................................................... 707 NUCLEATION AND THE EARLY STAGES OF FILM GROWTH.. 708 2.1 Three-Dimensional Nucleation and Growth ........................ 710 2.2 Two-Dimensional Nucleation and Growth .......................... 721 2.3 Stranski-Krastanov Nucleation and Growth ....................... 728 3.0 COMPUTER SIMULATIONS OF MICROSTRUCTURE EVOLUTION ............................................................................... 730 3.1 Film Growth in the Ballistic Aggregation, Low-Adatom Mobility, Limit ................................................................... 732 3.2 Effects of Adatom Migration .............................................. 734 4.0 MICROSTRUCTURE EVOLUTION AND STRUCTURE-ZONE...... 736 5.0 EFFECTS OF LOW-ENERGY ION IRRADIATION DURING FILM GROWTH ................................................................................... 743 5.1 Effects of Low-Energy Ion/Surface Interactions on Nucleation Kinetics ...................................................... 743 5.2 Effects of Low-Energy Ion/Surface Interactions on Film Growth Kinetics.................................................... 750 REFERENCES .................................................................................... 760

14

Metallurgical Applications.......................................... 766 Rointan F. Bunshah

1.0 2.0 3.0 4.0 5.0 6.0

INTRODUCTION ......................................................................... 766 CORROSION .............................................................................. 766 GALVANIC CORROSION ........................................................... 767 3.1 Galvanic Cells ................................................................... 768 EMF AND GALVANIC SERIES .................................................. 770 COATINGS FOR GALVANIC CORROSION ................................ 770 METHODS OF DEPOSITION OF METALLIC COATINGS ........... 772

Contents

xxv

7.0

EXAMPLES OF CORROSION-RESISTANT COATINGS ............. 773 7.1 Preamble .......................................................................... 773 8.0 HIGH TEMPERATURE OXIDATION/CORROSION ...................... 776 9.0 FRICTION AND WEAR ............................................................... 781 9.1 Adhesive Wear .................................................................. 781 9.2 Fretting Wear .................................................................... 781 9.3 Abrasive Wear .................................................................. 782 9.4 Fatigue Wear .................................................................... 782 9.5 Impact Erosion Wear by Solid Particles and Fluids ........... 782 9.6 Corrosive Wear ................................................................. 783 9.7 Electric Arc Induced Wear ................................................ 783 9.8 Solution Wear (Thermodynamic Wear).............................. 783 10.0 COATINGS TO REDUCE FRICTION AND WEAR ....................... 783 10.1 Friction ............................................................................. 783 10.2 Lubrication ........................................................................ 785 10.3 Wear ................................................................................. 785 REFERENCES .................................................................................... 787

15

Characterization of Thin Films and Coatings ............. 789 Gary E. McGuire

1.0 2.0

INTRODUCTION ......................................................................... 789 SURFACE ANALYSIS TECHNIQUES ........................................ 789 2.1 Auger Electron Spectroscopy ........................................... 789 2.2 Photoelectron Spectroscopy ............................................. 797 2.3 Secondary Ion Mass Spectroscopy .................................. 803 2.4 Rutherford Backscattering Spectroscopy .......................... 812 3.0 IMAGING ANALYSIS TECHNIQUES .......................................... 822 3.1 Scanning Electron Microscopy ......................................... 822 3.2 Transmission Electron Microscopy ................................... 828 4.0 OPTICAL ANALYSIS TECHNIQUES........................................... 834 4.1 Ellipsometry...................................................................... 834 4.2 Fourier Transform Infrared Spectroscopy ........................... 838 4.3 Photoluminescence Spectroscopy .................................... 841 REFERENCES .................................................................................... 845

16 1.0 2.0 3.0

Jet Vapor Deposition ................................................ 848 Bret L. Halpern and Jerome J. Schmitt INTRODUCTION ......................................................................... 848 PRINCIPLES AND APPARATUS OF JVD .................................. 849 DISCUSSION ............................................................................. 853 3.1 Jet Structure, Behavior, and Vapor Transport .................... 853 3.2 Substrate Motion .............................................................. 856

xxvi

Contents

4.0

EXAMPLES OF JVD FILMS AND APPLICATIONS ..................... 857 4.1 Cu, Au Multilayer Electrodes; Al, Al2O3 Microlaminates... 857 4.2 PZT: Ferroelectric FRAM Nonvolatile Memories ................ 858 4.3 Electronic Grade Silicon Nitride ........................................ 859 4.4 Fiber Coating for Composite Materials .............................. 859 4.5 Coating of Thermally Sensitive Membranes ....................... 860 4.6 “Ceramic Host–Organic Guest” Films................................ 860 4.7 Polymer Deposition: Parylene ........................................... 861 5.0 SUMMARY ................................................................................. 861 REFERENCES .................................................................................... 862

Index

......................................................................... 864

1 Deposition Technologies: An Overview Rointan F. Bunshah

1.0 THE MARKET Historically, from the late 1950s onward, decorative coatings or aluminum provided the initial thrust for surface-engineered products for toys, textiles, etc. Since then, the uses of deposition techniques in practically all areas of engineering and many areas of science have produced a dramatic growth in sales of equipment and products produced, particularly in the last decade. According to a recent survey (VDI-Technologiezeutrum-FRG), equipment with an estimated value of $6 billion was produced worldwide in 1989 for their film surface technology. Components and devices manufactured with such equipment amounted to $60 billion and the value of the end-products which contained components made possible by surface engineering is estimated at $600 billion. Just one industry, semiconductors, has changed entireproduction lines every 5 to 6 years. It is further estimated that only 10% of all items which can benefit from surface modifications are being processed today. Surface engineering will remain a growth industry in the next decade, because surface-engineered products increase performance, reduce costs, and control surface properties independently of the substrate, thus offering enormous potential due to the following: ! Creation of entirely new products ! Solution of previously unsolved engineering problems ! Improved functionality of existing products—engineering or decorative ! Conservation of scarce materials ! Ecological considerations—reduction of effluent output and power consumption 27

28

Deposition Technologies for Films and Coatings

Research and development expenditures in surface engineering are very extensive. It is reported that Japan is spending $100 to $150 million for R/D in diamond and diamond-like carbon coatings. The payoff is estimated at $16 billion by the end of this decade. In advance thermal barrier coatings by PVD methods for high temperature operation of turbine blades, it is estimated that more than $10 million have been spent in the United States alone. Wearresistant coatings for disc and heads has attracted much more than $10 million in R/D expenditures worldwide. The list continues to expand.

2.0 INTRODUCTION Most materials used in high technology applications are composites, i.e., they have a near-surface region with properties differing from those of the bulk materials. This is caused by the requirement that the material exhibit a combination of various, and sometimes conflicting, properties. For example, a particular engineering component may be required to have high hardness and toughness (i.e., resistance to brittle crack propagation). This combination of properties can be obtained by having a composite material with high surface hardness and a tough core. Alternately, the need may be for a high temperature, corrosion-resistant material with high elevated-temperature strength as is the case with the hot stage blades and vanes in a gas turbine. The solution again is to provide the strength requirement from the bulk and the corrosion requirement from the surface. In general, coatings are desirable, or even necessary, for a variety of reasons including economics, materials conservation, unique properties, or the engineering and design flexibility which can be obtained by separating the surface properties from the bulk properties. This near-surface region is produced by depositing a coating onto it (i.e., overlay coating) by processes such as physical or chemical vapor deposition, electrodeposition, and thermal spraying, or by altering the surface material by the in-diffusion of materials (i.e., diffusion coating or chemical conversion coating), or by ion implantation of new material so that the surface layer now consists of both the parent and added materials. “Coatings” may also be formed by other processes such as melt/ solidification (e.g., laser glazing technique), by mechanical bonding of a surface layer (e.g., roll bonding), by mechanical deformation (e.g., shot peening), or other processes which change the properties without changing the composition.

Deposition Technologies: An Overview

29

As stated above, the coating/substrate combination is a composite materials system. The behavior of this composite system depends not only on the properties of the two components (i.e., the coating material and the substrate material), but also on the interaction between the two (i.e., the structure and properties of the coating/substrate interface) which is integral to the very important factor of adhesion of coatings. In some cases, such as for overlay coatings, this is a distinct region. For others, such as ion implantation or diffusion coatings, it is not a discrete region. Historically, most solid metallic and some ceramic materials were produced by melting/solidification technology. Since the advent of deposition technologies (i.e., production of solid materials from the vapor), the diversity of materials that can be produced has more than doubled because the properties of solid materials produced from the vapor phase can be varied over a much wider range than the same material produced from the liquid phase. This is because melt techniques produce solid materials with properties close to equilibrium properties whereas the deposition conditions may be so chosen as to produce materials from the vapor phase with properties close to equilibrium (similar to their melt-produced counterparts), or properties far removed from equilibrium properties (non-equilibrium properties). Moreover, a much greater variation in microstructure is possible with vapor source materials. For example, a copper-nickel alloy produced by solidification from the melt will always consist of a single phase solid solution, whereas the same alloy produced by alternate deposition from two sources may consist of alternate layers of nickel and copper, i.e., a laminate composite or a solid solution depending on the deposition temperature. A large number of materials are used for coatings today. These may range from the naturally occurring oxide layer which protects the surfaces of many metals such as aluminum, titanium, and stainless steel, to those with very deliberate and controlled alloying additions to the surface to produce specific properties, as exemplified by techniques such as molecular beam epitaxy or ion implantation. Other examples with increasing degree of criticality range from paint coatings applied to wood and metals, electrostatically painted golf balls, the print in the daily newspaper, optical coatings on lenses and other elements, vapor deposited microcircuit elements such as resistors, diffusion or overlay coatings on superalloys used in gas turbines for high temperature corrosion protection, hard overlay coatings of engineering components and machine tools, etc.

30


3.0 AIM AND SCOPE The aim of this volume is to give the reader a perspective on several coating techniques with emphasis on the techniques which are used in critical or demanding (i.e., high technology) applications. Consequently, some of the techniques such as painting, dip coating, or printing will not be emphasized except as they pertain to some special application like thick film electrical components. Nevertheless, a wide variety of techniques and their applications will be covered. The material is intended to present a broad spectrum of deposition technologies to those who may be familiar with only one or two techniques. Hopefully, this will help them to select and weigh various alternatives when the next technological problem involving coatings faces them. The specific deposition technologies to be covered are: 1. Physical Vapor Deposition including evaporation, ion plating and sputtering. 2. Chemical Vapor Deposition and Plasma-Assisted Chemical Vapor Deposition 3. Electrodeposition and Electroless Deposition. 4. Plasma Spraying as well as a very special variant called Detonation Gun Technology. There are some generic areas common to several of the deposition technologies, the most prominent example being the use of plasmas in many of the deposition technologies. Therefore, a chapter on plasmas in deposition processes is included. Another common topic is cleaning of the substrate and adhesion of the coating. A chapter is included on that topic. A further common topic is the characterization of the chemical composition and the microstructure of the coating at various levels of resolution. A chapter is included to satisfy this need. New chapters are added dealing with Metallurgical Applications (Corrosion, Function and Wear), Overview of Plasma-Assisted Deposition Processes, Plasma-Assisted Chemical Vapor Deposition, and Nucleation/Growth of Thin Films. It is realized that all specific applications cannot be satisfied within this framework. For example, specific applications such as coatings for optical or magnetic applications are not addressed per se. At the other end of the spectrum, coatings for the first wall of thermo-nuclear reactors cannot be discussed since the development of the subject is in an embryonic stage.


31

In each of the chapters on deposition technologies, the theory, methodology, advantages, limitations and applications are discussed.

4.0 DEFINITIONS AND CONCEPTS In order to avoid potential problems, it is necessary to clarify certain distinctions which are common and pertinent to deposition technologies. These are as follows: 1. Diffusion vs.Overlay Coatings—Diffusion coatings are produced by the complete interdiffusion of material applied to the surface into the bulk of the substrate material. Examples of this are the diffusion of oxygen into metals to form various sub-oxide and oxide layers, the diffusion of aluminum into nickel base alloys to form various aluminides, etc. A characteristic feature of diffusion coatings is a concentration gradient from the surface to the interior, as well as the presence of various layers as dictated by thermodynamic and kinetic considerations. Ion implantation may be considered to be a special case where the coating material is implanted at a relatively shallow depth (a few hundred angstrom units) from the surface. An overlay coating is an add-on to the surface of the part, e.g., gold-plating on an iron-nickel alloy, or titanium carbide onto a cutting tool, etc. Depending upon the process parameters, an interdiffusion layer between the substrate and the overlay coating may or may not be present. 2. Thin Films vs. Thick Films—Historically, the physical dimension of thickness was used to make the distinction between thick films and thin films. Unfortunately, the critical thickness value depended on the application and discipline. In recent years, a "Confucian" solution has been advanced. It states that if a coating is used for surface properties (such as electron emission, catalytic activity), it is a thin film; whereas, if it is used for bulk properties, corrosion resistance, etc., it is a thick film. Thus, the same coating material of identical thickness can be a thin film or a thick film depending upon the usage. This represents a reasonable way out of the semantic problem.

32

Deposition Technologies for Films and Coatings 3. Steps in the Formation of a Deposit—There are three steps in the formation of a deposit: a. Synthesis or creation of the depositing species b. Transport from source to substrate c. Deposition onto the substrate and film growth

These steps can be completely separated from each other or be superimposed on each other depending upon the process under consideration. The important point to note is that if, in a given process, these steps can be individually varied and controlled, there is much greater flexibility for such a process as compared to one where they are not separately variable. This is analogous to the degrees of freedom in Gibbs phase rule. For example, consider the deposition of tungsten by CVD process. It takes place by the reaction: WF6(vapor) + 3H2(gas)

Heated

———" W(deposit) + 6HF(gas) Substrate

The rate of deposition is controlled by the substrate temperature. At a high substrate temperature, the deposition rate is high and the structure consists of large columnar grains. This may not be a desirable structure. On the other hand, if the same deposit is produced by evaporation of tungsten, the deposition rate is essentially independent of the substrate temperature so that one can have a high deposition rate and a more desirable microstructure. On the other hand, a CVD process may be chosen over evaporation because of considerations of throwing power, i.e., the ability to coat irregularly shaped objects, since high vacuum evaporation is basically a line-of-sight technique.

5.0 PHYSICAL VAPOR DEPOSITION (PVD) PROCESS TERMINOLOGY The basic PVD processes are those currently known as evaporation, sputtering and ion plating. In recent years, a significant number of specialized PVD processes based on the above have been developed and extensively used, e.g., reactive ion plating, activated reactive evaporation, reactive sputtering, etc. There is now considerable confusion since a particular process can be legitimately covered by more than one name. As


33

an example, if theactivated reactive evaporation (ARE) process is used with a negative bias on the substrate, it is very often called reactive ion plating. Simple evaporation using an RF heated crucible has been called gasless ion plating. An even worse example of the confusion that can arise is found in the chapter on ion plating in this volume (Ch. 6) where the material is converted from the condensed phase to the vapor phase using thermal energy (i.e., evaporation) or momentum transfer (i.e., sputtering) or supplied as a vapor (very similar to CVD processes). Carrying this to the logical conclusion, one might say that all PVD processes are ion plating! On the other hand, the most important aspect of the ion plating process is the modification of the microstructure and composition of the deposit caused by the ion bombardment of the deposit resulting from the bias on the substrate, i.e., what is happening on the substrate. To resolve this dilemma, it is proposed that we consider all of these basic processes and their variants as PVD processes and describe them in terms of the three steps in the formation of a deposit as described above. This will hopefully remove the confusion in terminology. Step 1: Creation of Vapor Phase Specie. There are three ways to put a material into the vapor phase-evaporation, sputtering or chemical vapors and gases. Step 2: Transport from Source to Substrate. The transport of the vapor species from the source to the substrate can occur under line-of-sight or molecular flow-conditions (i.e., without collisions between atoms and molecules); alternately, if the partial pressure of the metal vapor and/or gas species in the vapor state is high enough or some of these species are ionized (by creating a plasma), there are many collisions in the vapor phase during transport to the substrate. Step 3: Film Growth on the Substrate. This involves the deposition of the film by nucleation and growth processes. The microstructure and composition of the film can be modified by bombardment of the growing film by ions from the vapor phase resulting in sputtering and recondensation of the film atoms and enhanced surface mobility of the atoms in the near-surface and surface of the film. Every PVD process can be usefully described and understood in terms of these three steps. The reader is referred to Chapter 9 for a more comprehensive treatment.

34


6.0 CLASSIFICATION OF COATING PROCESSES Numerous schemes can be devised to classify or categorize coating processes, none of which are very satisfactory since several processes will overlap different categories. For example, the Appendix contains a list and definitions of various deposition processes based upon those provided by Chapman and Anderson with some additions. These authors classify the processes under the general heading of Conduction and Diffusion Processes, Chemical Processes, Wetting Processes and Spraying Processes. Here, the Chemical Vapor Deposition process falls under the Chemical Processes, and the Physical Vapor Deposition Process (Evaporation, lon Plating and Sputtering) falls under the spraying processes. The situation can easily get confused as, for example, when Reactive and Activated Reactive Evaporation, and Reactive lon Plating are all classified as Chemical Vapor Deposition processes by Yee[3] who considers them thusly because a chemical reaction is involved and it does not matter to him whether evaporated metal atoms or stable liquid or gaseous compounds are the reactants. Another classification of the methods of deposition of thin films is given by Campbell.[4] He considers the overlap between physical and chemical methods, e.g., evaporation and ion plating, sputtering and plasma reactions, reactive sputtering and gaseous anodization.[5] He classifies the Chemical Methods of Thin Film Preparation as follows:

Chemical Methods of Thin Film Preparation Basic Class

Method

Formation from the Medium Electroplating lon Plating Chemical Reduction Vapor Phase Plasma Reaction Formation from the Substrate

Gaseous Anodization Thermal Plasma Reduction


35

In addition, he considers the following as chemical methods of thick film preparation: Glazing, Electrophoretic, Flame Spraying and Painting. In contrast to the chemists’ approach given above, the physicists’ approach to deposition processes is shown in the following classification of vacuum deposition techniques by Schiller, Heisig and Goedicke[6] and by Weissmantel.[7]

Figure 1.1. Survey of vacuum deposition techniques (Schiller[6])

A different classification comes from a materials background where the concern is with structure and properties of the deposits as influenced by process parameters. Thus, Bunshah and Mattox[8] give a classification based on deposition methods as influenced by the dimensions of the depositing specie, e.g., whether it is atoms/molecules, liquid droplets or bulk quantities, as shown in Table 1.1. In atomistic deposition processes, the atoms form a film by condensing on the substrate and migrating to sites, where nucleation and growth occurs. Further, adatoms do not achieve their lowest energy configurations and the resulting structure contains high concentrations of structural imperfections. Often the depositing atoms react with the substrate material to form a complex interfacial region. Another aspect of coatings formed by atomistic deposition processes is as follows. The sources of atoms for these deposition processes can be by thermal vaporization (vacuum deposition) or sputtering (sputter deposition) in a vacuum, vaporized chemical species in a carrier gas (chemical vapor deposition), or ionic species in an electrolyte (electrodeposition). In low energy atomistic deposition processes, the depositing species impinge on the surface, migrate over the surface to a nucleation site where they condense and grow into a coating. The nucleation and growth modes of the condensing species determine the crystallography and microstructure of

36


the coating. For high energy deposition processes, the depositing particles react with or penetrate into the substrate surface. Particulate deposition processes involve molten or solid particles and the resulting microstructure of the deposit depends on the solidification or sintering of the particles. Bulk coatings involve the application of large amounts of coating material to the surface at one time such as in painting. Surface modification involves ion, thermal, mechanical, or chemical treatments, which alter the surface composition or properties. All of these techniques are widely used to form coatings for special applications.

Table 1.1. Methods of Fabricating Coatings

7.0 GAS JET DEPOSITION WITH NANO-PARTICLES One of the chapters in this volume (Ch. 11) deals with Plasma Spraying and Detonation Gun Techniques where a high velocity stream of macroparticles (µm dimensions) impinge on a substrate to form a coating. With the


37

advent of evaporation[9] and sputtering processes[10] to produce nanoparticles (nm dimensions), the same concept can be used to produce coatings by carrying nano-particles in a gas stream and impinging them on a substrate.[11][12] Figure 1.2 shows a schematic of this process where metallic nano-particles produced by evaporation are carried in a gas stream, accelerated through a nozzle and impinged on a substrate to produce a coating. Single nozzles or multiple nozzle configurations can be used, the latter producing an array of dots, for example. The attributes of this process are: 1. Direct write maskless processing to produce dots, lines, and other shapes. 2. High deposition rate, 10 - 20 µm per second over a small area. 3. Low temperature (room temperature) deposition. 4. Metals, alloys, ceramics, and organic materials can be deposiited. 5. Multiphase films with uniform mixing can be produced. 6. The collection officiency is very high, ~90%, i.e. very little waste or scatter. Examples of applications of this technique are: 1. Electrical connecting lines in circuits including the repair aspect. 2. Fabrication of microelectrodes 3. Oxide superconductor contacts. 4. Capacitors 5. Implantation of virus into plants for the bio industry. 6. Cell-gene processing technology.

Figure 1.2. Schematic diagram of gas deposition apparatus.

38


8.0 MICROSTRUCTURE AND PROPERTIES In electrodeposition, typically the growth process involves condensation of atoms at a kink site on the substrate surface, followed by layered growth of the deposit. Adatom mobility is increased by the hydrated nature of the ions and the adatom mobility may vary with crystal orientation. Field ion microscopy stripping studies of copper electrodeposited on tungsten has shown that there is surface rearrangement of the tungsten atoms during the electrodeposition process. Electrodeposited material does not grow in a uniform manner; rather it becomes faceted, develops dendrites and other surface discontinuities. Thus the microstructure of electrodeposited coatings may vary from relatively defect-free single crystals usually grown on single crystal substrates, to highly columnar and faceted structures. In the electroplating process, organic additives may be used to modify the nucleation process and to eliminate undesirable growth modes. This results in a microstructure more nearly that of bulk material formed by conventional metallurgical processes. Electrodeposition from a molten salt electrolyte allows the deposition of many materials not available from aqueous electrolytes. In vacuum processes, the depositing species may have energies ranging from thermal (a few tenths of an electron volt) for evaporation to moderate energies (ten to hundreds of electron volts) for sputtered atoms to high energies for accelerated species such as those used in ion implantation. These energies have an important but poorly understood effect on interfacial interaction, nucleation and growth. Where there is chemical reaction between the substrate atoms and the depositing atoms, and diffusion is possible, a diffusion or compound interfacial region is formed composed of compounds and/or alloys which modify the effective surface upon which the deposit grows. Low energy electron diffraction studies have shown that this interfacial reaction is very sensitive to surface condition and process parameters. If the coating and substrate materials are not chemically reactive and are insoluble, the interfacial region will be confined to an abrupt discontinuity in composition. This type of interface may be modified by bombardment with high energy particles to give high defect concentrations and implantation of ions resulting in a “pseudodiffusion” type of interface. The type of interface formed will influence the properties of the deposited coating. In many circumstances, these interfacial regions are of very limited thickness and pose a challenge to those interested in compositional, phase, microstructural and property analysis.


39

The microstructure of the depositing coating in the atomic deposition processes depends on how the adatoms are incorporated into the existing structure. Surface roughness and geometrical shadowing will lead to preferential growth of the elevated regions giving a columnar type microstructure to the deposits.[13] This microstructure will be modified by substrate temperature, surface diffusion of the atoms, ion bombardment during deposition, impurity atom incorporation and angle of incidence of the depositing adatom flux. The structure zone model of Movchan and Demchishin[14] for vacuum deposited films is discussed in later chapters. In chemical vapor deposition, the chemical species containing the film atoms is generally reduced or decomposed on the substrate surface, often at high temperatures. Care must be taken to control the interface reaction between coating and substrate and between the substrate and the gaseous reaction products. The coating microstructure which develops is very similar to that developed by the vacuum deposition processes, i.e., small-grained columnar structures to large-grained equiaxed or oriented structures. Each of the atomistic deposition processes has the potential of depositing materials which vary significantly from the conventional metallurgically processed material. The deposited materials may have high intrinsic stresses, high point defect concentration, extremely fine grain size, oriented microstructures, metastable phases, incorporated impurities, and macro and micro porosity. These properties may be reflected in the physical properties of the materials and by their response to applied stresses such as mechanical loads, chemical environments, thermal shock or fatigue loading. Metallurgical properties which may be affected include elastic constants, tensile strength, fracture toughness, fatigue strength, hardness, diffusion rates, friction/wear properties, and corrosion resistance. In addition, the unique microstructure of the deposited material may lead to such effects as anomalously low annealing and recrystallization temperatures where the internal stresses and high defect concentration aid in atomic rearrangement. The high value of grain boundary area to volume ratio found in fine grained deposited material means that diffusion processes may be dominated by grain boundary rather than bulk diffusion. The fine grained nature of the materials also affects the deformation mechanisms such as slip and twinning. For thin films, the free-surface to volume ratio is high, and the pinning of dislocation by the free surface leads to the high tensile strengths often measured in thin films of materials.

40


In vapor deposition processes, impurity incorporation during deposition can give high intrinsic stresses or impurity stabilized phases which are not seen in the bulk forms of the materials. Reactive species allow the deposition of compounds such as nitrides, carbides, borides and oxides. Graded deposits can be formed. Vapor deposition processes have the capability of producing unique and/ or nonequilibrium microstructures. One example is the fine dispersion of oxides in metals, where the oxide particle size and spacing is very small (100 - 500 Å). Alternately, metals and alloys deposited at high substrate temperatures have properties similar to those of conventionally fabricated (cast, worked and heat treated) metals and alloys. A more recent example is the nano-scale laminate composites consisting of alternate layers of refractory compounds with unusually high hardness values. 9.0 UNIQUE FEATURES OF DEPOSITED MATERIALS AND GAPS IN UNDERSTANDING It is useful to state at this point some of the unique features of materials produced by deposition technologies. They are: 1. Extreme versatility of range and variety of deposited materials. 2. Overlay coatings with properties independent of the thermodynamic compositional constraints. 3. Ability to vary defect concentration over wide limits, thus resulting in a range of properties comparable to, or far removed from conventionally fabricated materials. 4. High quench rates available to deposit amorphous materials. 5. Generation of microstructures different from conventionally processed materials, e.g., a wide range of microstructures— ultrafine (submicron grain or laminae size) to single crystal films. 6. Fabrication of thin self-standing shapes even from brittle materials. 7. Ecological benefits with certain techniques. The first edition lists some of the areas where our understanding of basic processes and phenomena is lacking and which obviously are the areas where research activities are essential. These are:


41

1. Microstructure and properties in the range of 500 to 10,000 Å— particularly important for submicron microelectronics, reflective surfaces and corrosion. 2. (a) Effect of the energy of the depositing species on interfacial interaction, nucleation and growth of deposit. (b) Effect of “substrate surface condition,” i.e., contamination (oxide) layers, adsorbed gases, surface topography. 3. Residual stresses—influence of process parameters. Considerable progress and understanding has developed in the last decade. 10.0 CURRENT APPLICATIONS The applications of coatings in current technology may be classed into the following generic areas: Optically Functional—Laser optics (reflective and transmitting), architectural glazing, home mirrors, automotive rear view mirrors, reflective and anti-reflective coatings, optically absorbing coatings, selective solar absorbers. Electrically Functional—Electrical conductors, electrical contacts, active solid state devices, electrical insulators, solar cells. Mechanically Functional—Lubrication films, wear and erosion resistant coatings, diffusion barriers, hard coatings for cutting tools. Chemically Functional—Corrosion resistant coatings, catalytic coatings, engine blades and vanes, battery strips, marine use equipment. Decorative—Watch bezels, bands, eyeglass frames, costume jewelry. A few examples are chosen to illustrate them in greater detail. 10.1 Decorative/Functional Coating Weight reduction is a high priority item to increase gas mileage in automobiles. Therefore, heavy metallic items such as grills are being

42


replaced with lightweight plastic, overcoated with chromium by sputtering for the appearance to which the consumer is accustomed. Another extensive application is aluminum-coated polymer films for heat insulation, decorative and packaging applications. A rapidly growing application is the use of a gold-colored wear-resistant coating of titanium nitride on watch bezels, watch bands and similar items. A new application is black wear-resistant hard carbon films. 10.2 High Temperature Corrosion Blades and vanes used in the turbine-end of a gas turbine engine are subject to high stresses in a highly corrosive environment of oxygen-, sulfurand chlorine-containing gases. A single or monolithic material such as a high temperature alloy is incapable of providing both functions. The solution is to design the bulk alloy for its mechanical properties and provide the corrosion resistance by means of an overlay coating of an M-Cr-AI-Y alloy where M stands for Ni, Co, Fe or Ni + Co. The coating is deposited in production by electron beam evaporation and in the laboratory by sputtering or plasma spraying. With the potential future use of synthetic fuels, considerable research will have to be undertaken to modify such coating compositions for the different corrosive environments as well as against erosion from the particulate matter in those fuels. 10.3 Environmental Corrosion Thick ion plated aluminum coatings are used in various irregularlyshaped parts of aircraft and space-craft as well as on fasteners:(a) to replace electroplated cadmium coatings which sensitize the high-strength parts to hydrogen embrittlement or(b) to prevent galvanic corrosion which would occur when titanium or steel parts contact aluminum or (c) to provide good brazeability. New alloy coatings in the micron thickness range have been developed. 10.4 Friction and Wear Dry-film lubricant coatings of materials such as gold, MoS2 , WSe2 and other lamellar materials are deposited on bearings and other sliding parts by sputtering or ion plating to reduce wear. Such dry-film lubricants are


43

especially important for critical parts used in long-lifetime applications since conventional organic fluid lubricants are highly susceptible to irreversible degradation and creep over a long time. 10.5 Materials Conservation Aluminum is continuously coated on a steel strip, 2 feet wide and 0.006 inches thick to a 250 micro-inch thickness in an air-to-air electron-beam evaporator at the rate of 200 feet/minute. The aluminum replaces tin, which is becoming increasingly scarce and costly. The strip then goes to the lacquer line and is used for steel can production. With the change in Eastern Europe, this line has switched to deposition of Cr and Cu on steel. 10.6 Cutting Tools Cutting tools are made of high-speed steel or cemented carbides. They are subject to degradation by abrasive wear as well as by adhesive wear. In the latter mode, the high temperatures and forces at the tool tip promote microwelding between the steel chip from the workpiece and the steel in the high-speed steel tool or the cobalt binder phase in the cemented carbide. The subsequent chip breaks the microweld and causes tool surface cratering and wear. A thin layer of a refractory compound such as TiC, TiN, Al2 O3 prevents the microwelding by introducing a diffusion barrier. Improvements in tool life by factors of 300 to 800% are possible as well as reductions in cutting forces. The coatings are deposited by chemical vapor deposition or physical vapor deposition. Some idea of the importance of such coatings can be assessed from the fact that the yearly value of cutting tools purchased in the U.S. is $1 billion and the cost of machining is approximately $60 billion. The last decade has seen major advances in this area and some of these are: ! Ti alloy nitrides, e.g., (Ti, Al) N ! Ti carbonitrides, e.g., Ti (C,N) ! Multilayer coatings of different nitrides ! Diamond coated tools by CVD and PACVD processes for machining of non-ferrous metals and polymer-matrix composites. A bond layer such as silicon nitride has to be used to attach the diamond coating to the carbide cutting tool.

44

Deposition Technologies for Films and Coatings ! Hard diamond-like carbon for heads and discs ! Cubic boron nitride coatings by plasma-assisted PVD and CVD methods for cutting of hard ferrous materials

10.7 Nuclear Fuels Pyrolytic carbon is deposited on nuclear fuel particles used in gas-cooled reactors by chemical vapor deposition in fluidized beds. The coating retains the fission products and protects the fuel from corrosion. 10.8 Biomedical Uses Parts for implants such as heart valves are made of pyrolytic carbon by CVD techniques. Metal parts are coated with carbon by ion plating in order to obtain biological compatibility. 10.9 Electrical Uses High temperature cuprate superconductors with transition temperatures of 85° to 115°K. This permits the operation of liquid nitrogen cooled devices. Various PVD techniques such as co-evaporation in an oxygen plasma, sputtering from simple or multiple targets and laser ablation have been used to fabricate films, ranging from 1 to 50 cm2 . Microwave devices such as delay lines, quasioptical filters have been fabricated and are being marketed. 11.0 “FRONTIER AREAS” FOR THE APPLICATION OF THE PRODUCTS OF DEPOSITION TECHNOLOGY The following were listed in the first edition published in 1982. 1. Reflective surfaces, e.g., for laser mirrors. 2. Thermal barrier coatings for blades and vanes operating at high temperatures. 3. Corrosion/erosion resistant coatings at high temperatures, e.g., valves and other critical compounds in coal gasification plants. 4. Advanced cutting tools.

Deposition Technologies: An Overview 5. Wear-resistant surfaces without organic lubricants, particularly at high temperatures where lamellar solid state lubricants such as MoS2 are ineffective. 6. First wall of thermonuclear reactor vessels. 7. High-strength/high-toughness ceramics for structural use. 8. Ultrafine powders. 9. Super conducting materials: High transition temperatures >23.2°K. Fabricability of these brittle materials into wire or ribbons. 10. CataIytic materials. 11. Thin film photovoltaic devices. 12. Transparent conductive coatings in opto-electronics devices, photo detectors, liquid crystal and electrochromic displays, solar photo thermal absorption devices, heat mirrors. 13. Biomedical devices, e.g., neurological electrodes, heart valves, artificial organs. 14. Materials conservation. 15. Sub-micron microelectronic devices. In this context, a good question is, How far can dimensions be reduced without running into some limit imposed by physical phenomena? In 1992, new additions to the above list are: 16. Diamond and diamond-like carbon for various applications: ! Tribology, particularly cutting tool ! Heat management–heat sinks of diamond sheet currently several square inches in area are on the market ! Hard protective coatings for infrared applications such as the protection of germanium and sodium chloride optics 17. Cubic boron nitride for various applications: ! High temperature use (up to 1200°) semiconductor devices. Very perfect device quality single crystal films have been grown epitaxially on lattice matched TiC substrates ! Tribological uses for machining of hard steels ! Optical and opto-electronic devices

45

46

Deposition Technologies for Films and Coatings 18. Film deposition using a high velocity gas jet. Hayashi and coworkers[9] have developed a process where ultra-fine powders (~10 nanometer diameter) are carried on a high velocity gas jet and impinged on a substrate to “write” lines of deposited materials, e.g., YBCO superconductors. The usage of material is very high, almost 97% is collected as a deposit. Various applications are envisioned. 19. Unbalanced magnetron deposition—very useful new development where some of the electrons are allowed to escape from the magnetic trap at the sputtering target and from a plasma near the substrate from which ions can be extracted to bombard the growing film.

12.0 SELECTION CRITERIA The selection of a particular deposition process depends on several factors. They are: 1. The material to be deposited 2. Rate of deposition 3. Limitations imposed by the substrate, e.g., maximum deposition temperature 4. Adhesion of deposit to substrate 5. Throwing power 6. Purity of target material since this will influence the impurity content in the film 7. Apparatus required and availability of same 8. Cost 9. Ecological considerations 10. Abundance of deposition material in the world In order to aid the reader in the task of selection, Table 1.2 lists several criteria for each of the processes. It is obvious that there are very few techniques which can deposit all types of materials. It is also impossible to detail the advantages and limitations of each of the techniques. However, in the evaluation of each application, the above factors will lead to a rational choice of the deposition technique to be used.

Table 1.2. Some Characteristics of Deposition Processes

Deposition Technologies: An Overview 47

48


13.0 SUMMARY In the above discussion, we have noted the following: 1. There are a very large number of deposition techniques. 2. There is no unique way to classify these techniques. Depending on the viewpoint, the same technique may fall into fall into one or more classes. 3. Each technique has its advantages and limitations. 4. The choice of the technique to be used depends on various selection criteria which have been given above. 5. More than one technique can be used to deposit a given film as shown in Figure 1.3 below from Campbell’s article on preparation methods in microelectronic fabrication.

Electro-

Chemical

Vapor Anodization Thermal Evaporation Sputtering

plating Reduction Phase 123456789012345678901 123456789012345678901 123456789012345678901 Conductors, 123456789012345678901 123456789012345678901 123456789012345678901 resistiors 123456789012345678901 123456 123456 123456 123456 Insulators, 123456 123456 capacitors 123456 Active devices Magnetic materials Superconductors

1234567890 1234567890 1234567890 1234567890 1234567890 1234567890 1234567890 1234567 1234567 1234567 1234567 1234567 1234567 1234567

12345678 12345678 12345678 12345678 12345678 12345678 12345678 12345678 12345678 12345678 12345678 12345678 12345678 12345678 12345678 12345678 12345678 12345678123456789012345678 12345678 1234567890 1234567890 1234567890 1234567890 1234567890 1234567890 123456789012345678 12345678 123456789012345678 1234567890 123456789012345678 12345678 1234567890 123456789012345678 12345678

Figure 1.3. Applicability of preparation methods to microelectronics. Light shading indicates that the component can be prepared by the method; Dark shading indicates that the method is widely used.


49

APPENDIX 1: DEPOSITION PROCESS DEFINITIONS The definitions of various deposition processes are given below. They are grouped as proposed by Chapman and Anderson[1] and many of them are those proposed by these authors. Conduction and Diffusion Processes Electrostatic Deposition is the deposition of material in liquid form, the solvent used then being evaporated to form a solid coating. At the source, the liquid is atomized and charged, and then it can be directed onto the substrate using an electrostatic field. Electrophoretic Coating produces a coating on a conducting substrate from a dispersion of colloidal particles. The article to be coated is immersed in an aqueous dispersion which dissociates into negatively charged colloidal particles and positive cations. An electric field is applied with the article as anode (positive electrode); the colloidal particles are transported to the anode, where they are discharged and form a film. In the case of a paint coating, this requires curing, which further shows that electrophoresis itself is not a very effective transport process, so that electrodeposition may be a better term for the coating process. Electrolytic Depositionis primarily concerned with the deposition of ions rather than of colloidal particles. Two electrodes are immersed in an electrolyte of an ionic salt which dissociates in aqueous solution into its constituent ions; positive ions are deposited onto the cathode (negative electrode). Anodizationis a process which occurs at the anode (hence its name) for a few specific metals. The anode reacts with negative ions from the electrolyte and becomes oxidized, i.e., it forms a surface coating. Gaseous Anodization is a process in which the liquid electrolyte of the conventional wet process is replaced by a glow discharge in a low partial pressure of a reactive gas. Oxides, carbides and nitrides can be produced this way. Ion Nitridingis a gaseous anodization to produce nitride diffusion coating on a metal surface, usually steel. Ion Carburizingis a gaseous anodization to produce a carbide diffusion coating on a metal surface, usually steel. Plasma Oxidationis gaseous anodization to produce an oxide film on the surface of metal, e.g., SiO2 films on Si. Diffusion Coating is produced by diffusion of material from the surface into the bulk of the substrate.

50


Metalliding is a method using electrodeposition in molten fluorides. Spark-hardening is a technique in which an arc is periodically struck between a vibrating anode and the conducting substrate (cathode); material is transferred from the anode and diffuses into the substrate. Chemical processes Conversion and Conversion/Diffusion Coating is a process in which the substrate is reacted with other substances (which may be in the form of solids, liquids or gases) so that its surface is chemically converted into different compounds having different properties. (Anodization could probably be described as an electrochemical conversion process). Conversion coating usually takes place at elevated temperatures and diffusion is often an essential feature. Chemical Vapor Deposition (CVD) is a chemical process which takes place in the vapor phase very near the substrate or on the substrate so that a reaction product is deposited onto the substrate. The deposition can be a metal, semiconductor, alloy or refractory compound. Pyrolysis is a particular type of CVD which involves the thermal decomposition of volatile materials on the substrate. Plasma-Assisted CVD is a process where the reaction between the reactants is stimulated or activated by creating a plasma in the vapor phase using means such as R F excitation from a coil surrounding the reaction vessel. Electroless Deposition is often described as a variety of electrolytic deposition which does not require a power source or electrodes, hence its name. It is really a chemical process catalyzed by the growing film, so that the electroless term is somewhat a misnomer. Disproportionation is the deposition of a film or crystal in a closed system by reacting the metal with a carrier gas in the hotter part of the system to form the compound, followed by dissociation of the compound in the colder part of the system to deposit the metal. Examples are epitaxial deposits of Si or Ge on a single crystal substrate and the Van-Arkel-deBoer process for metal purification and crystal growth. Wetting Process Wetting Processes are the coating processes in which material is applied in liquid form and then becomes solid by solvent evaporation or cooling.


51

Conventional Brush Painting and Dip Coating are wetting processes in which the part to be coated is literally dipped into a liquid (e.g., paint) under controlled conditions of, for example, withdrawal rate and temperature. Hydrophilic Method is a surface chemical process known as the Langmeir Blodgett technique which is used to produce multimonolayers of long chain fatty acids. A film 25 Å thick can be deposited on a substrate immersed in water and pulled through a compressed layer of the fatty acid on the surface of the water. The process can be repeated to build up many layers. Welding Processesare the range of coating techniques all of which rely on wetting. Spraying Processes Printing Process also relies on wetting and is a process in which the ink, conventionally pigment in a solvent, is transferred to and is deposited on a paper or other substrate, usually to form a pattern; the solvent evaporates to leave the required print. Spraying Processes can be considered in two categories;(i) macroscopic in which the sprayed particle consists of many molecules and is usually grater than 10 µm in diameter; (ii) macroscopic in which the sprayed particles are predominantly single molecules or atoms. Air and Airless Spraying are the first of the macroscopic processes. When a liquid exceeds a certain critical velocity, it breaks up into small droplets, i.e., it atomizes. The atomized droplets, by virtue of their velocity (acquired from a high pressure air or airless source) can be sprayed onto a substrate. Flame Spraying is a process in which a fine powder (usually of a metal) is carried in a gas stream and is passed through an intense combustion flame, where it becomes molten. The gas stream, expanding rapidly because of the heating, then sprays the molten powder onto the substrate where it solidifies. Detonation Coating is a process in which a measured amount of powder is injected into what is essentially a gun, along with a controlled mixture of oxygen and acetylene. The mixture is ignited, and the powder particles are heated and accelerated to high velocities with which they impinge on the substrate. The process is repeated several times a second. Arc Plasma Sprayingis a process in which the powder is passed through an electrical plasma produced by a low voltage, high current electrical discharge. By this means, even refractory materials can be deposited.

52


Electric-Arc Spraying is a process in which an electric arc is struck between two converging wires close to their intersection point. The high temperature arc melts the wire electrodes which are formed into high velocity molten particles by an atomizing gas flow; the wires are continuously fed to balance the loss. The molten particles are then deposited onto a substrate as with the other spray processes. Harmonic Electrical Spraying is a process in which the material to be sprayed must be in liquid form, which will usually require heating. It is placed in a capillary tube and a large electrical field is applied to the capillary tip. It is found that by adding an AC perturbation to the DC field, a collimated beam of uniformly sized and uniformly charged particles is emitted from the tip. Sense these particles are charged, they could be focused by an electrical field to produce pattern deposits. Evaporation is a process in which the boiling is carried out in vacuum where there is almost no surrounding gas; the escaping vapor atom will travel in a straight line for some considerable distance before it collides with something, for example, the vacuum chamber walls or substrate. Glow Discharge Evaporation and Sputtering are processes in soft vacuum (10-2 to 10-1 torr) operating in the range 10-1 < pd < 10-2 torr cm where p is the pressure and d is the cathode fall dimension. Molecular Beam Epitaxy is an evaporation process for the deposition of compounds of extreme regularity of layer thickness and composition from well controlled deposition rates. Reactive Evaporation is a process in which small traces of an active gas are added to the vacuum chamber; the evaporating material reacts chemically with the gas so that the compound is deposited on the substrate. Activated Reactive Evaporation (ARE) is the Reactive Evaporation Process carried out in the presence of plasma which converts some of the neutral atoms into ions or energetic neutrals thus enhancing reaction probabilities and rates to deposit refractory compounds. Biased Activated Reactive Evaporation (BARE)is the same process as Activated Reactive Evaporation with substrate held at a negative bias voltage. Sputter Depositionis a vacuum process which uses a different physical phenomenon to produce the microscopic spray effect. When a fast ion strikes the surface of a material, atoms of that material are ejected by a momentum transfer process. As with evaporation, the ejected atoms or molecules can be condensed on a substrate to form a surface coating. Ion Beam Depositionis a process in which a beam of ions generated from an ion beam gun, impinge and deposit on the substrate.


53

Ion Beam Assisted Deposition—two versions are possible. One, an ion beam is used to sputter a target and a second beam is used to bombard the growing film to change structure and properties. This is dual Ion Beam Assisted Deposition. The other version uses an ion beam to bombard the growing film to change structure and properties. In this case, conventional evaporation or sputtering techniques are used to generate a flux of the depositing species. Cluster Ion Beam Deposition is an ion beam deposition in which atomic clusters are formed in the vapor phase and deposited on the substrate. Ion Plating is a process in which a proportion of the depositing material from an evaporation, sputtering or chemical vapor source is deliberately ionized. Once changed this way, the ions can be accelerated with an electric field so that the impingement energy on the substrate is greatly increased, producing modifications of the microstructure and residual stresses of the deposit. Reactive Ion Plating is ion plating with a reactive gas to deposit a compound. Chemical Ion Plating is similar to Reactive Ion Plating but uses stable gaseous reactants instead of a mixture of evaporated atoms and reactive gases. In most cases, the reactants are activated before they enter the plasma zone. Ion Implantation is very similar to ion plating, except that now all of the depositing material is ionized, and in addition, the accelerating energies are much higher. The result is that the depositing ions are able to penetrate the surface barrier of the substrate and be implanted in the substrate rather than on it. Plasma Polymerization is a process in which organic and inorganic polymers are deposited from monomer vapor by the use of electron beam, ultraviolet radiation or glow discharge. Excellent insulating films can be prepared in this way.

54


REFERENCES 1. Science and Technology of Surface Coating,(B. N. Chapman and J. C. Anderson, eds.), Academic Press (1974) 2. Adhesion Measurement of Thin Films, Thick Films and Bulk Coatings, (K. D. Mittal, ed.), Am. Soc. for Testing Materials (1978) 3. Yee, K. K.,International Metal Reviews,No. 226, The Metals Society and American Society for Metals (1978) 4. Campbell, D. S.,Handbook of Thin Film Technology,(L. Maissel and R. Glang, eds.), Ch. 5, McGraw-Hill (1970) 5. Handbook of Thin Film Technology, (L. Maissel and R. Glang eds.), McGraw-Hill (1970) 6. Schiller, S., Heisig, O., and Geodick, K., Proc. 7th Int’l. Vacuum Congress, (R. Dobrozemsky, ed.), p. 1545, Vienna (1977) 7. Weissmantel, C., ibid, p. 1533, 8. Bunshah, R. F. and Mattox, D. M., Physics Today (May 1970) 9. Hayashi, C., Paper presented at the International Vacuum Congress, Hague, Netherlands (Oct. 1992); also: Hayashi, C., J. Vac. Sci. Tech., A5(4):1375 (1987) 10. Suc, T. G., Umarjee, D. M., Prakash, S., and Bunshah, R. F., Surface and Coatings Technology, 13:199 (1991) 11. Hayashi, C., Kashu, S., Oda, M., and Naruse, F., presented at the Int'l Vac. Cong., The Hague, Netherlands (Nov. 1992) - to be published in Mat. Sci. Eng., (1993) 12. Oda, M., Katsu, I., Tsuneizumi, M., Fuchita, E., Kashu, S., and Hayashi, C., presented at Fall Mtg. Mat. Res. Soc., Boston, 1992 13. Thornton, J. A.,Proc. 19th National SAMPLE Symposium, Buena Park, Ca. (April 23-25, 1974) 14. Movchan, B. A., and Demchishin, A. V., Phys. Met Metallogr., 28:83 (1969)

2 Plasmas in Deposition Processes John A. Thornton and Joseph E. Greene

1.0 INTRODUCTION A glow discharge plasma used in deposition processes is a low-pressure gas which is partially ionized and contains approximately equal numbers of positive and negative particles. The character of such a plasma is a consequence of the mass difference between the electrons and the ions. When an electric field is applied to an ionized gas, energy is transferred more rapidly to the electrons than to the ions. Furthermore, the transfer of kinetic energy from an electron to a heavy particle (atom, molecule, or ion) in an elastic collision is proportional to the mass ratio of electrons and heavy particles and therefore very small (~10-5). Consequently, at low pressures (low collision frequencies), the electrons can accumulate sufficient kinetic energy to have a high probability of producing excitation or ionization during collisions with heavy particles. The production of these excited species, and their interactions with surfaces and growing films, is one of the reasons that low pressure glow discharge plasmas are assuming an ever-increasing role in materials processing. Examples of application areas include the following. ! ! ! ! ! !

Sputter deposition Activated reactive evaporation Ion plating Plasma-assisted chemical vapor deposition Plasma-assisted etching Plasma polymerization 55

56


The purpose of this chapter is to review fundamental aspects of glowdischarge plasmas which are of importance in understanding the role of plasma processes in materials processing.

2.0 PARTICLE MOTION 2.1 Mean Free Path and Collision Cross Sections A glow discharge plasma can be viewed as a medium in which electrical energy is transmitted, via an electric field, to a gas. The energetic gas particles are then used to promote chemical reactions or to interact with a surface to produce desirable effects such as sputtering. Thus, the process of energy exchange during collisions involving plasma particles is of fundamental importance. Gas-phase collision probabilities are often expressed in terms of cross sections. A related parameter is the mean free path or average distance traversed by particles between collisions. The mean free pathλ and collision cross sectionσ are generally defined by a simple relationship which treats the particles as impenetrable spheres. Thus, the mean free path for electrons passing through a gas of particle density N is Eq. (1)

λ = 1/(Nσ)

The total collision cross section can be written as Eq. (2)

σt = σel + σex + σion + σ a + σoth

where the subscripts el, ex, ion, a, and oth characterize the particular types of collisions, namely, elastic or momentum exchange, excitation, ionization, attachment, and other processes, respectively. Figure 2.1 shows the cross sections for electrons interacting with Ar gas. The cross sections are typically a strong function of the energy of the colliding species. For the case of electrons colliding with gas atoms, the kinetic energy of the gas atoms is generally much less than that of the electrons and can be neglected. Consequently, only the electron energy is shown in Fig. 2.1. The figure shows that at low electron energies the primary collision process is momentum exchange (σt ≈ σel), while at energies considerably larger than the ionization potential (15.75 eV for Ar), the primary process is ionization (σt ≈ σion ).

Plasmas in Deposition Processes

57

Figure 2.1. Collision cross sections for electrons in Ar gas (from Ref. 1).

Cross sections are most easily measured for reactions involving a species such as an electron or ion which can conveniently be formed as an energetic beam and passed through a stationary gas. Figure 2.2 shows the cross section for energetic O+ ions passing through N2 and producing the reaction O+ + N2 → NO+ + N. Note in comparing Figs. 2.1 and 2.2 that the collision cross sections are typically a few x 10-15 cm2 in magnitude (i.e., a few angstroms in diameter). For collision types that cannot be investigated in beam experiments, the cross sections are often deduced from measurements of macroscopic parameters such as viscosities, diffusion coefficients, and chemical reaction rates.[3] Thus one finds reference to viscosity cross sections, diffusion cross sections, etc. Cross sections are primarily of interest in making comparisons based on kinetic theory. In most plasma calculations, the macroscopic rate parameters are used directly if they are available.

58


Figure 2.2. Cross section for the reaction of O+ ions with N2 to produce NO+ + N (from Ref. 2).

2.2 Free Electron Kinetic Energy in a Plasma r Consider a plasma electron in an electric fieldE . Between collisions with the gas particles, the electron will gain an energy Wf from the electric field that is equal to the force on the electron eE (where e is the electronic charge) times the distance that it moves in the electric field. This distance can be approximated by the mean free path so that, on average, Wf = eEλ. In the steady state case, this energy gain must be balanced against the energy loss in an average collision. We neglect inelastic collisions for the moment and consider collisions with heavy particles in which the electrons lose all of their momentum, i.e., are deflected by 90o . This permits us to use the momentum exchange cross section, as defined in the preceding section, for estimating λ. Application of the conservation of energy and momentum shows that loss of electron energy in such a collision is[4] Eq. (3)

∆W = (2me /mH) (We - WH)

where me and m H are the electron and heavy-particle masses and We and WH are the initial electron and heavy particle energies before the collision. Equating ∆W to the energy Wf gained from the electric field, and using Eq. 1 for λ, yields Eq. (4)

(We - WH) =

½ (mH /me) ( eE/Nσel)


59

In making calculations dealing with plasmas it is useful to note that: ! me = 9.11 x 10-31 kg = 9.11 x 10-28 g ! mH = 1.67 x 10-24 x (atomic mass number) g ! N = 3.2 x 1024 particles/m3 = 3.2 x 1016 particles/cm3 at 1 Torr and 300 K (27oC) ! The electron volt (eV) is the unit of energy generally used in plasma calculations. One electron volt is the energy gained by a particle with unit charge which is accelerated in an electric field produced by a potential difference of one volt (1 eV = 1.602 x 1019 joules = 11,600 K). Consider the case of electrons in an Ar plasma at 1 Torr and 300 K which is subjected to an electric field of 1 V/cm, thus, N = 3.2 x 1016 cm-3 and eE is 1 eV/cm. Using Eq. 4 with σel ≈ 10-15 cm2 from Fig. 2.1 yields (We - WH) ≈ 103 eV. Thus, at steady state, the average electron energy will be much greater than that of the gas atoms (0.03 eV at 300 K). The actual average electron energy will not reach 103 eV, however, because inelastic collisions will become important when We exceeds ≈ 10 eV. Nevertheless, the above analysis shows that even relatively weak electric fields can cause electron kinetic energies in low-pressure glow discharge plasmas to be elevated above gas-atom energies until they are finally “clamped” by losses due to inelastic collisions. Figure 2.3 shows this elevation of electron energy at low pressures for the case of plasma arcs. (The energies here are expressed as temperatures.) At high pressures, electron/gas-atom collisions are so frequent that the gas temperature increases. Such high-pressure arcs are used for a variety of applications. However, the discussion in the following sections will be limited to the low-pressure case where Te > Tg . In this situation, energetic electrons can produce high temperature chemistry in a gas at low temperatures.[6] 2.3 Electron Energy Distribution Functions For most purposes, the state of a glow discharge plasma can be characterized by the densities of heavy particles (N j, where j corresponds to the jth species), the electron density ne , and the electron energy distribution function F e(E). [7] Under conditions of local thermodynamic equilibrium,[8] when the forward and reverse rates for all the electron energy exchange processes are equal (state of detailed balance), [9] the electrons will have a Maxwellian velocity distribution and their state can be defined by

60


an electron temperature Te. Unfortunately, such a state of equilibrium seldom exists in a low-discharge plasma.

Figure 2.3. Electron ( Te ) and gas temperatures (Tg ) in an air arc as a function of pressure (from Ref. 5).

Figure 2.4 schematically illustrates the electron energy distribution function. The equilibrium energy distribution is also shown for comparison. The effect of an electric field is to shift electrons to higher energies and therefore to overpopulate the high-energy region relative to the Maxwellian distribution. The cross section for a representative inelastic collision is shown superimposed (see Fig. 2.1). Electrons undergoing inelastic collisions are transferred from the high-energy to the low-energy end of the distribution. Electron-electron collisions tend to smooth the distribution and drive it toward the Maxwellian form. If these collisions dominate such that a state of detailed balance exists for one dominant process, then Fe(E) can be approximated by a Maxwellian distribution and an electron temperature can be used to describe the state of the electrons. However, even this case seldom occurs in practice. In high-pressure discharges, the electric field perturbation is usually minimal allowing the distribution function to be approximately Maxwellian, although it may be somewhat depleted at high energies by inelastic collisions. In low-pressure discharges, the electric field can generate relatively large numbers of energetic electrons and, in the extreme, produce


61

a bimodal distribution function. This is the case in low-pressure negative glow discharges of the type used in sputtering.[7] Electron energy distribution functions are usually measured by electrostatic analyzer and probe methods. However, interpretation of the results are complicated by plasma/probe interactions. Therefore electron temperatures, although not strictly valid, are generally assumed in making engineering calculations.

Figure 2.4. Schematic illustration of electron energy distribution function and inelastic collision cross section.

2.4 Collision Frequencies The collision frequency is an important plasma parameter. It is defined as the rate at which an average particle undergoes collisions of a specified type. Thus the total electron-atom collision frequency is the rate at which an average electron in a plasma undergoes collisions of all types with gas atoms. The general expression for the collision frequencyν is rather complex and involves the distribution functions of the colliding species.[10] For the electron/ heavy-particle case, the velocity of the heavy particles can be neglected and ν is given by E= ∞

Eq. (5)

νk = N (E/2me)1/2 σk(E) Fe (E) dE E= 0

62


where k is the type of collision (e.g., elastic, excitation, ionization, etc). If the collision cross section σk(E) is assumed to be independent of energy, and the electrons are assumed to have a Maxwellian velocity distribution at an electron temperature Te, then Eq. 5 reduces to Eq. (6)

νk = N σk ve

The quantity ve is the average electron speed, Eq. (7)

ve = (8k Te / πme)1/2

where k is Boltzmann’s constant. It is customary to write kTe in units of eV*. Thus, Eq. 7 becomes Eq. (8)

ve = (6.7 x 107) [kT e (eV)]1/2 cm/sec

For purposes of obtaining rough estimates,σk in Eq. 6 is generally approximated by its value at the electron energy kTe. The electron/electron and electron/ion collision frequencies are of special interest. These are given by[11] Eq. (9)

νee = (3 x 10-6)ne lnΛ /[kTe(eV)]3/2 sec-1

and Eq. (10)

νei = (1.5 x 10-6) ne lnΛ /[kTe(eV)]3/2 sec-1

with Eq. (11)

λee = (4.5 x 1013) [kT (eV)]2 / (ne lnΛ) cm

where lnΛ is a weak function of kT e and ne . The function lnΛ is tabulated in most books on plasma physics, and has a value of approximately 10 for the glow discharge plasmas of interest here. [12] The lnΛ term arises * From kinetic theory, the average particle energy in one dimension is 1/2 kT. The average energy in three dimensions is 3/2 kT. Since T and E are so closely related, it is customary in plasma physics to give temperature in units of eV. To avoid confusion with the number of dimensions involved, it is not the average energy but the energy corresponding to kT that is used to denote the temperature.[11] By a 2 eV plasma, we mean that kT = 2 eV, although the actual average energy in three dimensions is 3/2 kT or 3 eV.


63

because these collisions involve long-range coulomb forces and the cross sections do not cut off as in the hard sphere approximation. The primary use of Eqs. 9 and 10 is in comparing νee and νei with other relevant collision frequencies. A plasma for which νee > νeA, where νeA is the elastic electron/atom collision frequency, is said to be coulomb-dominated. An approximate condition for coulomb domination is easily derived from Eqs. 6 and 9 (see Ref. 13), Eq. (12)

ne

>> αc = (2.23 x 1013)

N

σeA [kT(eV)]2 lnΛ

The term αc in Eq. 12 is known as the critical degree of ionization. Selecting kTe ~ 3 eV andσeA ~ 10-15 cm-3 (Fig. 2.1) yields ne/N ≈ 0.02. Thus a moderatetemperature glow discharge plasma with 2% ionization can be dominated by coulomb collisions. A consequence of coulomb domination can be seen by examining Eqs. 9 and 11. As the electron energy is increased, the electron collision frequency decreases and the mean free path increases. Thus, electrons in an electric field will find that their energy gain is “unchecked” by collisions. Electron runaway is an important consideration in highly ionized plasmas (13) but seldom important in glow discharge plasmas because of inelastic collisions. For the case of a heavy particle of mass m1 in a gas with density N2 of heavy particles of mass m2 , the collision frequency can be approximated by an equation very similar to Eq. 6,[4][14]

Eq. (13)

(

v12 ≈ 2.5 ×105

)σ N (m )

12 12 1 ∗ 2

T  300  

1 2

where the cross section σ12 is assumed to be independent of the velocity of impact and all the heavy particles are at the common temperature T. In Eq. 13, m* is a reduced mass defined as Eq. (14 )

m* =

m1 m2 m1 + m2

where the masses are molecular weights expressed in grams. Reaction Rates: The gas-phase reaction rate R is directly proportional to the collision frequency. For a process k involving electron collisions, Eq. (15)

Rk = ne νk

reactions cm3 -sec

64


If the electrons are assumed to have a Maxwellian velocity distribution at a temperature Te, and if the cross section for a given reaction is approximated by a step function of magnitude σ0 and threshold energy E0 as shown in Fig. 2.5, then the reaction rate is given by Eq. (16)

Rk = ne Nσ0νe [1 + (E0/kTe)] exp(-E0 /kTe )

Figure 2.5. Reaction rate approximation for a Maxwellian velocity distribution.

As a general rule, reaction rate constants rather than actual collision frequencies are measured and used to describe reactions involving heavy particle collisions. Thus, for a reaction occurring via a two-body collision between species A and B in a gas at temperature T, with rate constant κ(T), one has Eq. (17)

R = κ(T) NA NB

Mobilities: Plasma transport properties are dependent on the frequency of elastic (momentum exchange) collisions. The mobility µj relates the electric-field driven drift velocity vd of a given charged particle species j to the strength of the field E: Eq. (18)

vjd = µj E

When the collision frequency is sufficiently large that the drift velocity is small compared to the thermal velocity, Eq. (19)

µj =

1.6 x 10-12 mj ν

cm2 V-sec

where mj is the particle mass in grams.


65

The mobility is generally used to describe the drift of ions through a plasma that is at a sufficiently high pressure to satisfy the collision frequency requirement. Mobilities for several gases of interest are given in Table 1. Table 2.1. Mobilities of Ions in their Own Gas (From Ref. 15) Ion-Gas He+-He Ne+-Ne Ar+-Ar Kr+-Kr H2+-H2 N2+-H2 O2+-O2 CO2 +-CO2

Mobility (cm2/V-sec) 8,000 3,300 1,200 690 10,000 2,000 1,000 730

When a positive ion collides with a gas molecule or atom, two processes can occur. First, the ion and molecule can exchange momentum and energy in a collision in which the particles preserve their identity. Second, an exchange of charge can occur. For example, fast ions moving through a gas can engage in collisions in which the ion extracts an electron from a gas atom with the result that the fast ion becomes a fast neutral atom while the slow atom becomes a slow positive ion. Charge exchange is particularly important for ions of low energy passing through their own gas (resonant charge exchange). Under these conditions, the charge transfer cross section is about one half of the total cross section[16] and therefore contributes significantly in determining the mobility.* Charge transfer is very important in high-pressure sputtering and ion-plating discharges. Electrical Conductivity and Diffusion Coefficients: The electrical conductivity σ is just eNµ, so that Eq. (20)

σj = 1/ρj = 2.6 x 10-31 (Nj /mjν) (Ω-cm)-1

*The charge exchange region surrounding an atom can be considered as a sphere inside of which the probability of charge transfer is ½ and outside it is zero. As the ion approaches the atom, it will simply be deflected by the dipole interaction if the distance of closest approach is greater than the sphere radius. If the ion enters the charge exchange sphere, half the time it emerges as a neutral and half the time as an ion.[17]

66


where Nj is the particle density in cm-3 and mj is the mass of the current carrier in grams. The resistivityρ is often used to avoid confusion with σ, which is the common symbol for both the electrical conductivity and the collision cross section. The diffusion coefficient Dj relates the particle flux to the concentration gradient. Thus, one has Eq. (21)

Nj vjd = Dj (dNj /dx)

where Eq. (22)

Dj =

kT mj ν

= (1.6 x

10-12)

kT (eV) cm2 mj ν

sec

2.5 Particles in Magnetic Fields Charged particle motion in a magnetic field is summarized in Fig. 2.6.

Figure 2.6. Electron paths in static magnetic and electric fields.

r

A charged particle in a uniform magnetic field B will orbit a field line as shown in Fig. 2.6a and drift along the field with velocity v|| that is unaffected by the field, as shown in Fig. 2.6b. The orbiting frequency is called the gyro or cyclotron frequency and is given by Eq. (23)

ωc = eB/m


67

The orbiting radius is called the gyro, cyclotron, or Larmor radius and is given by Eq. (24)

rg = (m/3) (v⊥/B)

Manipulation and confinement of plasma particles by a magnetic field requires that rg be small compared to the apparatus size. Note in Eq. 24 that rg depends directly on the mass of the particle. Thus, very large magnetic fields are required to influence the motions of the plasma ions. When magnetic fields are used with glow discharges, they are generally chosen to be just strong enough to influence the energetic plasma electrons, but not the ions. However, magnetically-confined electrons in a glow discharge will in turn provide considerable confinement for the plasma ions since electrostatic forces prevent the ions from escaping from the electrons. For electrons, Eqs. 23 and 24 become[1] Eq. (25)

ωc = (1.76 x 107 ) B(gauss) rad/sec

and Eq. (26)

rg = 3.37

[W⊥(eV)] 1/2 B(gauss)

cm

Thus, for electrons with an average energy W⊥ of 10 eV and a magnetic field strength B of 100 G, the gyro radius is ≈ 0.1 cm. Magnetic field strengths between 50 and 100 G are typically used with glow discharge devices. An electron that is trapped on a given magnetic field line can advance to an adjacent field line by making a collision, as indicated schematically in Fig. 2.6c. Collisional diffusion of electrons across magnetic field lines is an important consideration in many glow discharge devices. r When an electric field E is present and directed parallel to the magnetic field, the electrons are freely accelerated along the field lines.rHowever, if the electric field has a component E⊥ which is perpendicularrto B , the electrons undergo a drift in a direction perpendicular torbothrE⊥ andB , asrshown r in Figs. 2.6d and 2.6e. This motion is known as the E x B drift. The E x B drift has the cycloidal form shown in Fig. 2.6d if the initial electron energy is small compared to that gained from the electric field; it has the more circular form shown in Fig. 2.6e if the initial electron energy is largely compared to the electric-field-induced variations that occur during the course of the orbit. In both cases, the electron drift speed is given by

68


Eq. (27)

ve = 108

E⊥ (V/cm) B(gauss)

cm/sec

The drift of electrons along a magnetic field line can also be influenced by gradients in the magnetic field. An example of this behavior is shown in Fig. 2.7. Electrons moving in such a field tend to conserve the magnetic moment, µM, defined by[11] Eq. (28)

µM = W⊥ / B

Therefore W⊥ must increase as the electrons move in the direction of increasing field strength. Conservation of energy requires that W|| + W ⊥ be constant. Therefore W|| must decrease, and the electron may be reflected as indicated in the figure. Pinched-field end confinement of this type is frequently used in glow discharge devices

Figure 2.7. Electron reflection in a magnetic field gradient.

3.0 COLLECTIVE PHENOMENA Plasmas differ from non-ionized gases by their propensity for undergoing collective behavior. Three parameters, derived from basic plasma properties, N, ne, and kTe, provide a useful measure of the tendency toward collective behavior.


69

The Debye length, Eq. (29)

λD = 743

kTe (eV) ne (cm3)

1/2

cm

corresponds to the distance over which significant departures from charge neutrality occurs. A plasma cannot exist in a space having a characteristic size less that λD. The plasma frequency, ωp , expressed here as Eq. (30)

fp = ωp /2π = 9000[ne (cm -3 )] 1/2 Hz

provides a measure of the tendency for electrostatic waves to develop. Waves can form if ωp >> νe,el, where νe,el is the electron collision frequency for momentum exchange. The critical degree of ionization αc was defined by Eq. 12. When the degree of ionization α = ne/N >> αc, long range coulomb collisions dominate, and the charged particles behave as though they were in a fully ionized gas. Coulomb domination can occur at degrees of ionization of a few percent for plasmas with low average electron energies (≈1 eV). 3.1 Plasma Sheaths Given a gas of particle density N (cm-3) and temperature T, the flux of particles passing to an adjacent wall is given by Eq. (31)

J = Nv/4 = (N/4) (8kT/πm)1/2

For electrons, this becomes (see Eqs. 7 and 8) Eq. (32)

Je = (1.67 x 107 ) ne [kTe (eV)]1/2 particles/cm2 -sec

which, in units of current density, is equal to Eq. (33)

Je = 2.7 x 10-9 ne [kTe (eV)] 1/2 mA/cm 2

Thus, for a typical glow discharge electron density of 109 cm-3 with an average energy of 1 eV, Je ≈ 3 mA/cm2. For heavy particles such as ions, Eq. 31 can be written in the following useful form:

70


Eq. (34)

Ji = 104 N (40/m)1/2 (T/300)1/2

where T is the gas temperature (K) and m is the species molecular weight. In units of current density, Eq. 34 becomes Eq. (35)

Ji = 1.6 x 10-9 N (40/m) 1/2 (T/300)1/2 µA/cm2

Thus, for an Ar plasma used in sputtering with an ion density of 109 cm-3 at 300 K, the ion current flux to the wall is 1.6 µA/cm2. It is clearly seen by comparing Eqs. 33 and 35 that the electrons tend to flow from a plasma to an adjacent wall at a faster rate than the ions; therefore, a space charge region in which one species is largely excluded forms adjacent to such surfaces. The potential variation between the surface and the plasma is largely confined to this layer, which is called a sheath. Sheaths are typically several Debye lengths in thickness. The nature of the sheath will depend upon the current density passing across it. Except for cases involving very high current densities to anodes, the space charge region will contain primarily the low-mobility ion species. Such sheaths are known as positive space charge sheaths. The function of the sheath is to form a potential barrier, so that the more mobile species, which is the electrons except in the case of a strong magnetic field, are electrostatically reflected. Thus, the height of the potential barrier associated with a sheath adjusts itself so that the flux of electrons to the wall in question just equals the electron current that is drawn from the wall by the external circuit. If the wall is electrically isolated, the electron flux is reduced to the point which is equal to the ion flux. Figure 2.8 shows that a schematic illustration of a typical glow discharge plasma which is in contact with wall surfaces that are either cathodes, anodes, or electrically isolated (floating). The potential Vp is known as the plasma potential. The potential of a floating surface relative to the plasma potential is known as the floating potential Vf. For a Maxwellian velocity distribution, the floating potential is given by[18] Eq. (36)

Vf =

kTe (eV) 2e

ln

π me 2

m

Typical values are -30 to -40 V. When a floating surface is immersed in a plasma, the surface will be bombarded with ions having kinetic energies of up to eVf.


71

Figure 2.8. Schematic illustration of sheaths that form between a plasma discharge and the surrounding apparatus walls for systems having (A) a large anode and (B) a small anode.

Generally, the anodes used in glow discharges are large enough that the current density is less than the thermal current given by Eq. 33. In this case, there is a positive space charge sheath at the anode, as shown in Fig. 2.8a, and the sheath potential drop is between zero and V f . The potential

72


of a plasma locks into the most positive surface, provided that the surface is large enough.[19] If the anode area is so small that the current density must exceed the thermal current, then the anode potential will be above the plasma potential, as shown in Fig. 2.8b. The local electric field surrounding the anode will draw sufficient electrons to the anode to complete the external circuit.* A large potential difference Vs, approximately equal to the entire potential applied by the power supply, occurs in the cathode sheath as shown schematically in Fig. 2.9. The sheath thickness ds is taken to be the region corresponding to Vs over which the electron density is negligible. For the low pressure case where the ion mean free path is larger than ds, the ion current density Ji is related to ds and Vs by the Child-Langmuir law. [11] It is useful to write this relationship as Eq. (37)

Ji = 0.273 (40/mi)1/2 (Vs3/2/ds2) mA/cm2

where Vs is in kV, d s is in cm, and mi is the ion molecular weight in grams. Thus, for an Ar sputtering plasma with Vs = 1 kV and d s = 1 cm, J i = 0.27 mA/cm 2. It is difficult to relate J i to the density N io of ions in the bulk plasma, because there is a quasi-neutral presheath region where a potential drop Vx of the order of 1/2(kTe /e) occurs. As an estimate, the presheath density can be assumed to obey a Boltzmann distribution, [20] such that Nis /Nio = exp(eV x /kTe ) , and Eq. (38)

Ji ≈ (0.6) eNio (kTe/mi )1/2

where mi is in grams and kTe is in ergs. For the high pressure case, where collisions are so frequent that the ion drift velocity is of the order of the thermal velocity, a mobility description is used for the ion motion.[5] Under this condition, Eq. (39)

Ji = 9.95 x 10-5 µi (Vs2/ds3 ) mA/cm2

where µi is the ion mobility in cm 2/V-sec, Vs is the sheath potential drop in kV, and ds is the sheath thickness in cm. For an Ar plasma at 1 Torr, µi = 1,200 cm2/V-sec from Table 1. Taking V = 1 kV and ds = 1 cm yields Ji = 0.11 mA/ cm 2. *The potential rise surrounding a small anode cannot become much larger than the ionization potential of the gas atoms since this potential causes the sheath electrons to be accelerated. If these electrons gain sufficient energy to produce ionization, then the electrons liberated by the ionizing collisions can provide the anode current flow requirement and no additional rise in potential is required.


73

Figure 2.9. Schematic representation of the positive space-charge sheath that develops over a cathode (from Ref. 1).

In a low pressure plasma, the ions will fall through the entire sheath potential and bombard the cathode with an energy about equal to eVs. At higher pressures, where charge exchange is important, the bombarding flux will consist of both ions and neutrals having energies considerably less than

74


eVs as indicated schematically in Fig. 2.10. This is an important consideration in sputtering, ion plating, and reactive ion etching, as discussed in Ch. 5, Sec. 3.0.

Figure 2.10. Schematic representation of charge exchange reactions in the cathode fall region of a glow discharge.

3.2 Ambipolar Diffusion Consider a plasma within a container having electrically isolated or floating walls. A sheath will develop on these walls to reduce the electron flux until it is equal to the ion flux as described in Sec. 3.1. Accordingly, an electric field in the sheath retards the loss of electrons and accelerates the loss of ions. This coupled particle motion is called ambipolar diffusion. The diffusion flux J of electrons or ions to a floating wall is given by Eq. (40)

Je = D a (dne/dx) = Ji = D a (dni /dx)

The term Da is called the ambipolar diffusion coefficient. Noting that µe >> µi (see Eq. 19) permits Da to be approximated as:[11]


Eq. (41)

75

Da ≈ Di (1 + Te /Ti )

where Di is given by Eq. 22. Thus the effect of the ambipolar field is to enhance the diffusion of ions by a factor of more than two, but the diffusion rate of the two species together is primarily controlled by the slower species. In the presence of a sufficiently strong magnetic field perpendicular to the direction of diffusion, the electron mobility, and thus the electron diffusion coefficient can be reduced to the point where it is lower than the ion diffusion coefficient and therefore rate controlling. Under this condition one can write Eq. (42)

Da ≈ {De/[1 + (ωc2/νe)]} (1 + T i /Te)

where De is the electron diffusion coefficient in the absence of a magnetic field. The effect of the magnetic field becomes strong when ωc (given by Eq. 25) is much larger than the electron collision frequency νe , i.e., when the electrons are trapped on magnetic field lines as shown in Fig. 2.6b, and collisional hopping to adjacent field lines is infrequent. It should be noted that Eq. 42 is based on the assumption that electron losses along the lines can be neglected. Attention to these losses should be given when analyzing the performance of an actual plasma device.[11][21] 3.3 Plasma Oscillations The plasma state is rich in wave phenomena when the degree of ionization is large enough to make long-range forces important, particularly when a magnetic field is present.[11] Departures from charge neutrality capable of generating waves can occur in the form of charge bunching and separation over distances of the order of the Debye length, Eq. 29. A general discussion of such behavior is beyond the scope of this chapter. However, one case will be mentioned because of its potential importance in magnetron sputtering devices. Consider the case of a plasma in a uniform electric r rand magnetic field, as illustrated inrthe left side of Fig. 2.11. There is an E x B drift perpendicular r to both E and B , but, in the absence of collisions, simple theory predicts no transport across the magnetic field in the direction of the applied electric field. If charge bunching occurs, as shownr in the right side of r Fig. r 2.11, the perturbation produces an electric field r E p that can result inE x B drift across the magnetic field in the direction ofE . This anomalous collisionless transport across the magnetic field is believed to be an important mechanism in Penning discharges as well as in some magnetron sputtering discharges.[22]

76


Figure 2.11. Schematic representation of a plasma instability resulting in electron transport across a magnetic field.

4.0 PLASMA DISCHARGES 4.1 Introduction A glow discharge plasma is a low temperature, relatively low pressure, gas in which a degree of ionization is sustained by energetic electrons. Glow discharge configurations used in materials processing differ in both their general geometry and in the orientation of the electric field that is used to provide energy to the electrons. In sputtering, simple planar diodes of the type shown schematically in Fig. 2.12a are often used. They may be driven at radio frequencies (RF), as shown in the figure, or by a DC power supply. RF planar diode discharges are also used for sputter etching, plasma etching, and reactive ion etching, as illustrated in Figures 2.13b, 2.13d and 2.13e, respectively. Systems with the configuration shown in Fig. 2.13d are also used for plasma-assisted chemical vapor deposition (CVD). During activated reactive evaporation, a plasma discharge is sustained in a flux of evaporated material and reactive gas that is directed toward the substrates, as shown in Fig. 2.14. The discharge may be driven by DC or RF means, using a variety of electrode configurations. The presence of the plasma has been shown to influence properties such as the chemical composition of the resultant films.[53]


77

Figure 2.12. Schematic illustration of glow discharge devices commonly used in plasma-assisted materials processing.

In ion plating, the discharge is generally sustained in a mixture of the evaporated flux and an inert working gas with the substrate holder biased negatively relative to the plasma potential. Usually this is done by simply making the substrate holder the cathode electrode for sustaining the plasma discharge, as shown in Fig. 2.15. The ion bombardment of the growing coating has been shown to influence its structure.[53] In plasma etching, plasma-assisted CVD, and glow discharge polymerization, discharges are often sustained in glass or quartz reactor tubes by surrounding electrodes which are driven at high frequencies (from 300 kHz to microwave frequencies)[23] Common electrode configurations are a pair of ring electrodes along the tube, clam-shell electrodes as shown in Fig. 2.12b, or a solenoidal coil electrode as shown in Fig. 2.12c. It should be noted that all of these discharges are basically capacitive in nature. Although the coil electrode will introduce considerable inductance into the load seen by the matching network, the capacitive fields generated by the coil-to-coil potential drop dominate over those generated by the time rate-of-change of magnetic flux and therefore act as the primary source of ionization unless special precautions are taken to shield them. In the case of microwave discharges, the reactor tube is generally positioned within the waveguide at a location which places a strong electric field component within the tube.[6][23] 4.2 Ionization Balances and the Paschen Relation The degree of ionization in a glow discharge depends on a balance between the rate at which ionization is produced by energetic electrons and

78


the rate at which particles are lost by volume recombination and by passage to the walls of the apparatus. The rate of ionization depends on a relationship of the form (see Eqs. 6, 8, and 15) Eq. (43)

Figure 2.13. etching.

R ∝ N ne σion (E)1/2

Apparatus configurations commonly used in plasma-assisted


79

SUBSTRATES (S) ELECTRODE

Figure 2.14. Schematic illustration of the activated reactive evaporation (ARE) process (see Ref. 49).

Figure 2.15. Schematic illustration of a typical ion plating apparatus.

80


Thus, the rate of ionization depends on the type of gas (through the ionization cross sectionσion), the gas pressure (through the particle density N), and the electric field strength (through the electron velocity). Wall losses generally dominate over volume recombination. Accordingly, the occurrence of a breakdown, and the resulting formation of a sustaining plasma discharge, in a given apparatus depends on the gas pressure, the electric field strength, and on the surface-to-volume ratio of the plasma. Figure 2.16 shows the interelectrode breakdown voltage as a function of the product of the gas pressure p and the electrode spacing d for plane parallel electrodes in air[5] and Ar.[24] Such curves are determined experimentally and are known as Paschen curves. Relationships of the same general form apply to the conditions under which a steady-state discharge can be sustained. In such cases d may be replaced by a characteristic diffusion length for the plasma vessel.[6][17][25]

Figure 2.16. Paschen curves for breakdown between plane-parallel electrodes in air and argon at 20oC.

The rise in voltage at the low pd side in Fig. 2.16 occurs because the apparatus is small, or the gas density low, such that electrons are lost to the walls without colliding with gas atoms and producing ionization. The rise in the required voltage on the right side happens because the electron energy is becoming too low to produce ionization. This can occur at high pressures, because electron collisions with gas atoms become so frequent that the electrons cannot accumulate sufficient energy to overcome the ionization potential. It can also occur at a given applied voltage in a very large chamber where local electric fields in the plasma are too weak to deliver sufficient energy to the electrons between collisions.


81

The functional form of the curves in Fig. 2.16 provides a useful guide for adjusting the operating conditions within a given device in order to produce a plasma discharge. Conversely, the relation provides guidance for the prevention of discharges on surfaces such as the back of cathodes. One simply places a grounded shield over the surface to be protected ensuring that the spacing d between the shield and the cathode is small enough that the breakdown voltage is larger than the voltage required to form and sustain plasma discharge at the operating pressure of interest. The above considerations are also important in apparatus scaling. A discharge sustained in a small apparatus must have a high average electron energy to counteract wall losses. Such a discharge, with the same electron density but in a larger apparatus size, will be sustained at a lower average electron energy. This can in turn change the active species that are produced. Thus, small-bore discharge tubes are sometimes used in lasers to elevate the average electron energy to a desired value. Typical glow discharge electron densities are in the range of 108 to 10 12 cm -3 with average electron energies of 1 to 30 eV. These conditions are shown in Fig. 2.17 and compared with other forms of discharges.

Figure 2.17. Regions of average electron density and energy representative of various types of plasmas (from Ref. 7).

82


4.3 Cold Cathode Discharges A low-pressure cold-cathode discharge is one which is maintained primarily by secondary electrons emitted from the cathode due to bombardment by ions from the plasma. These secondary electrons are accelerated in the cathode dark space and enter the negative glow, as shown in Fig. 2.18, where they are known as primary electrons. Each primary electron must produce a sufficient number of ions to result in the ejection of another secondary electron from the cathode.[15] The secondary electron emission coefficient is typically about 0.1 for low-energy Ar+ ions (such as are used in sputtering) incident on clean metal surfaces.[26] The coefficient is larger, for example, for oxidized surfaces but still small enough that each primary electron must produce, or lead to the production of, a plurality of ions.[15]

Figure 2.18. Schematic illustration of a cold-cathode discharge.

The negative glow (NG) region of the plasma is where the primary electrons expend their energy, and its extent corresponds to the range of their travel from the cathode.[5][15] The electron energy distribution in the NG is multimodal. It consists of primary electrons, ultimate electrons (primaries that have transferred their energy), and much larger numbers of low-energy ionization products. In the classical glow discharge described in most


83

textbooks, a positive column (PC) extends from the NG to the anode.[5][15][17] The PC is a region in which the electric field is just sufficient to transport the discharge current from the NG to the anode and to produce sufficient ionization to make up for wall losses. In planar-diode material-processing sources of the type shown in Figs. 2.12 and 2.13, the substrate mounting table or anode generally intercepts the NG and there is no PC. A consequence of this small inter-electrode spacing is that the operating pressures are relatively high (see discussion of the Paschen relationship in Sec. 4.2). For example, reasonable operating conditions for DC planar-diode Ar sputtering discharges are: 75 mTorr pressure with a substrate-to-cathode spacing of 4.5 cm, a current density of 1 mA/cm2, and a discharge voltage of 3,000 V. In order for a cold-cathode discharge to operate effectively at low pressures, it is necessary that the primary electrons be preserved and not lost from the system until they have had a chance to expend their energy in ionization. The hollow cathode geometry shown in Fig. 2.19 is effective in this respect. Electrons which are accelerated in the cathode dark space and enter the NG cannot escape once they have lost an amount of energy about equal to their initial ejection energy (which is only a few eV)[26] since they encounter a sheath with repulsive forces whenever they approach the cathode. The only losses are out of the ends, and long hollow cathodes with minimized end losses can be operated effectively at low pressures and voltages. Accordingly, hollow cathodes are often used as ionization sources.[27]

Figure 2.19. Schematic illustration of a hollow cathode discharge.

84


4.4 Magnetron Discharges Magnetron discharge sources are assuming increasing importance for sputter deposition. Therefore, these discharges are discussed in some detail in Ch. 5. It will simply be noted here that magnetrons are cold cathode discharge devices in which magnetic fields are used in concert r r with cathode surfaces to form traps which are so configured that the E x B electron drift currents can close upon themselves.[1] The cylindrical-post configuration shown in Fig. 2.20 provides one of the simplest examples of a magnetron. Primary electrons which leave the cathode barrel and enter the plasma find themselves trapped in an annular cavity which is closed on three sides by surfaces at cathode potential (the hollow cathode effect) and on the fourth side by the magnetic field. The electrons can diffuse across the magnetic field and reach the anode only by making collisions (the process illustrated in Fig. 2.6c) and by plasma oscillations (see Sec. 3.3).[22] Because of the effectiveness of the collisions in producing ionization, these discharges are extremely efficient and operate at pressures of less than 1 mTorr with high current densities (10 - 200 mA/cm2 ) and low voltages (700 - 1,000 V). Planar magnetrons in which plasma rings are magnetically confined on planar cathodes are very important in sputter-deposition technology.[28][29]

Figure 2.20. Cylindrical-post magnetron sputtering source with electrostatic end confinement.


85

4.5 RF Discharges RF-driven planar diode discharge devices of the type shown in Figs. 2.12a, 2.13b, 2.13d, and 2.13e are used for sputter deposition, plasmaassisted etching, and plasma-assisted CVD. Their application to sputtering is discussed in detail in Ch. 5. The operating frequency is generally 13.56 MHz, since this is the frequency in the 10 to 20 MHz range that has been allocated by the FCC for industrial applications. At this frequency, only the electrons can follow the temporal variations in applied potential. Thus the plasma can be pictured as an electron gas that moves back and forth at the applied frequency in a sea of relatively stationary ions. As the electron cloud approaches one electrode, it uncovers ions at the other electrode to form a positive ion sheath. This sheath takes up nearly the entire voltage as in the DC case. The ions are accelerated by this voltage and bombard the electrodes. The RF discharge can be further understood by examining the electrode current flow. These discharges are capacitive in nature, both because of external capacitance which is placed in the electrical circuits and because one or both electrode surfaces are generally nonconducting. Consequently, the total ion and electron charge flow to a given electrode during an RF cycle must balance to zero and a self bias that is negative with respect to the plasma potential develops on any surface that is capacitively coupled to a glow discharge.[51] The basis for this behavior is illustrated in Fig. 2.21, where the current/voltage characteristics are shown for an electrode immersed in a glowdischarge plasma. Because of the mobility difference between the electrons and the ions, much larger currents are drawn when the electrode is positive relative to the floating potential than when it is negative (upper figure). In order to achieve zero net current flow, it is necessary for the DC bias to develop such that the average potential is negative relative to the floating potential, as shown in the lower figure. Thus both electrodes exceed the floating potential (and become anodes) only for short portions of each RF cycle. Most of the time they are cathodes. Because of their inertia, the motion of the ions can be approximated as if they follow the DC potential and flow to both electrodes throughout the cycle. RF discharges in planar diodes can be operated at considerably lower pressures than DC discharges. Typical operating pressures are 5 to 15 mTorr. This is due to two reasons: a reduction in the loss of primary electrons and, at high frequencies, by an increase in the volume ionization efficiency. A fraction of the lower-energy primary electrons are repelled from the electrode toward which they are accelerated and thus remain in the

86


(a)

(b)

Figure 2.21. The formation of a negative bias on a capacitively-coupled surface in an RF glow discharge (from Ref. 51).

discharge longer to make additional ionizing collisions. In addition, electrons can gain energy from the RF field by making in-phase collisions with gas atoms. That is, if an electron, accelerated in one direction during a given halfcycle, makes an elastic collision in which its direction is reversed near the end of the half-cycle, it maintains most of its velocity (due to the large mass mismatch between electrons and ions) and will again be accelerated during the next half-cycle and thus have gained energy during the complete


87

cycle. As the pressure is increased, the volume ionization due to electrons accelerated by the oscillating electric field becomes increasingly important Accordingly, when the planar and cylindrical plasma discharge devices shown in Fig. 2.13 are used for plasma-assisted etching, CVD, and polymerization, the operating pressures are generally high enough that volume-accelerated electrons dominate in producing excitation and ionization. The same is true for high-frequency microwave type discharges.

5.0 PLASMA VOLUME REACTIONS 5.1 Introduction Electron bombardment of atoms and molecules results in excitation ionization, and dissociation, thereby producing a variety of active species and radicals having much different chemical activities than those of the parent gas.[30][31] Thus, although He and Ar atoms are inert, He+ ions with one valence electron are hydrogenic. Ar+ ions are similar to Cl and can react with H2 molecules to form HAr+ ions.[30] Electron ionization processes are obviously important in the sustaining of plasma discharges. The excitation and dissociation processes are important in plasma chemistry and form the basis for plasma-assisted etching, plasma-assisted CVD, and plasma polymerization. 5.2 Electron/Atom Interactions An electron with a kinetic energy which exceeds the ionization energy of an atom has as approximately equal probability of producing either excitation or ionization as it passes in close proximity to the atom. A semiclassical picture of such a collision is shown in Fig. 2.22. The Coulomb force from the electron produces an electric field at the atom. The component of this field which is perpendicular to the direction of electron motion (E⊥) produces a time-varying “impulsive” electric field which can act on the atom. The electric field pulse is equivalent to that which would be produced by a beam of photons having frequencies corresponding to the Fourier components of the pulse.[32] The point is that an electron passing close by an atom does not simply knock an electron out of the atom, but produces a perturbation at the atom which may be approximated as a beam of white light that induces electronic excitation and ionization in proportion to the optical oscillator strengths.

88


Figure 2.22. Virtual photon model of an electron-atom collision (from Ref. 32).

In making plasma calculations, the average energy Wei spent by an electron in creating an electron-ion pair in a given gaseous medium is often used. Values of Wei for various atoms and molecules are shown in Table 2 along with values for the ionization potential I. Note that Wei/I ≈ 2; i.e., there is an almost equal probability of producing either excitation or ionization, although excitation is more probable in molecules. 5.3 Electron/Molecule Interactions Electron interactions with molecules produce excitation and ionization via mechanisms essentially identical to those for atoms as described above. The primary difference is in the fate of the excitation energy. In the atomic case, the excitation energy is lost by radiation unless the transitions are quantum-mechanically forbidden (see Sec. 5.4 below). In the molecular case, it may result in dissociation of the molecules. Consider the case of CF4 , a gas which is commonly used in plasma etching. The threshold for producing excitation is 12.5 eV.[33] The excitation reaction can be written as e− + CF4 → CF4* + e−


89

where the symbol * refers to an excited species. There is evidence that all electronic excitation processes in CF4 produce dissociation. [33] Furthermore, because of the two-step nature of the excitation-dissociation process, one bond is broken, and the primary radicals produced are CF3 and F rather than CF2 and F2.[34] The active F atoms produced in this way play a very important role in many plasma etching processes. Table 2.2. Approximate Energy Spent to Create Electron-Ion Pairs[32] Atom or Molecule He Ne Ar Kr Xe H2 N2 NO CO O2 CO2 C2H2 CH4 C2H4 C2H6 C3H6 C3H8 C6H6

Wei (eV) 46 37 26 24 22 36 36 29 35 32 34 28 29 28 27 27 26 27

I (eV) 24.58 21.56 15.76 14.00 12.13 15.43 15.59 9.25 14.04 12.15 13.81 11.40 12.99 10.54 11.65 9.73 11.15 9.23

W ei / I 1.87 1.71 1.65 1.71 1.81 2.33 2.31 3.13 2.49 2.63 2.46 2.45 2.23 2.65 2.31 2.77 2.33 2.92

The ionization process can also result in dissociation. Thus, one has dissociative ionization reactions of the form e− + CF4 → CF3 + + F + 2e− as well as simple molecular ionization e− + O2 → O2 + + 2e − It has been noted that plasma discharges often contain relatively large numbers of low energy electrons which have expended their energy in making inelastic collisions (this is particularly true in regions of low electric field such as the negative glow). These electrons can attach to electronegative molecules to form negative ions[23] such as e− + O2 → O2−

90


The ion may then dissociate, for example O2− → O− + O Atomic constituents of small molecules such as F2 cannot recombine in two-body gas-phase collisions because the diatomic molecule formed cannot conserve both energy and momentum. Thus the gas-phase recombination reaction requires a third body. Accordingly, the lifetime for such atoms in a plasma reactor can be long except when the working pressure is high. However, when two molecular radicals associate, the energy of dissociation can be distributed within a large number of internal degrees of freedom. Accordingly, the association efficiency is close to unity for simple radicals.[23] Thus, for example, one has CH3 + CH3 → C2 H6 The decay of initial reaction products in cascading reactions, with the development of high molecular weight species, is a well-known characteristic of the radiation chemistry of hydrocarbons and halocarbons in both the gas and solid phases.[31] The general hierarchy for the production of active species in a molecular gas plasma is shown schematically in Fig. 2.23. 5.4 Metastable Species An important consideration in using plasmas for materials processing is the ability of active species to diffuse from the point of production to a point of reaction. Atoms or molecules that are excited into electronic states which can decay radiatively have very short lifetimes (~10−9 s). However, some excited states are forbidden by quantum mechanical considerations from undergoing radiative transitions. Atoms and molecules in these metastable states have sufficiently long lifetimes that they can carry their stored electronic energy from the immediate vicinity of the discharge plasma to other points in a reactor. Atoms or molecules can be excited directly into metastable states, or can arrive in these states by radiative decay after having been excited into states of higher energy. Consequently, a plasma may contain relatively large numbers of metastable species and they can have an important effect on the overall discharge chemistry. Metastable states are depopulated when the atoms undergo collisions. Thus, for example, a metastable atom A* may subsequently pass its excitation energy to another particle, thereby


91

producing ionization or dissociative ionization in atoms or molecules of lower ionization potential, as indicated below.[35] A* + Y → Y+ + A + e− A* + XY → XY+ + A + e− A* + XY → X + + Y + A + e− These reactions are known as Penning ionization processes.

Figure 2.23. Schematic illustration of the production of active species in a molecular plasma.

92


5.5 Applications of Volume Reactions Primary applications of interest here are plasma-assisted CVD,[36] plasma-assisted etching,[37] and plasma polymerization (38). In each of these cases, the advantage of using a plasma is that it can effectively produce reactions at low substrate temperatures. In some cases, the reactions are unique. An example is provided by the plasma-assisted deposition of Si3N4 using a SiH4-NH3 plasma. The plasma chemistry is not understood in detail, however the overall reaction is 3SiH4 + 4NH3 → Si3N4 + 12H2 The important point is that the substrate temperature is typically 300oC or lower. When the same reaction is carried out by conventional chemical vapor deposition, the substrate temperatures are typically between 800 - 1200oC.[39] The lower substrate temperatures in plasma-assisted CVD are particularly important in electronic applications where coatings are deposited onto device structures. The average electron energies in plasma-assisted CVD are typically low, ≈1 - 10 eV, such that the plasma chemistry is dominated by radicals rather than ions.[52] Bond energies are therefore an important criteria in the selection of reactants for a desired process. For example, one of the functions of the plasma during deposition of nitride films is to provide atomic N in the gas phase since the partial pressure of atomic N required to obtain stoichiometric nitride films is much smaller than that of N2. However, 9.83 eV is required to obtain N atoms by cleaving the N2 molecule, N2 → NH2 + H

(∆H = 9.83 eV)

Alternatively, N atoms can be obtained more efficiently through the following steps starting with NH3: NH3 → NH2 + H

(∆H = 4.76 eV)

NH2 → NH + H

(∆H = 3.90 eV)

NH → N + H

(∆H = 3.42 eV)


93

in which no reaction step requires more than 4.76 eV. This explains why NH3 is commonly used in Si3 H4 plasma-assisted CVD deposition. Similarly, the following reactions show why nitrous oxide N2 O, rather than O2, is often used as a parent donor molecule for O atoms during plasma CVD deposition of oxides. N2 O → N2 + O

(∆H = 1.73 eV)

O2 → 2O

(∆H = 4.13 eV)

Plasma-assisted etching is similar to plasma-assisted CVD, except that a volatile rather than an involatile compound is produced at the substrate. Thus, for example, Si etching is accomplished by using a glow discharge to generate active F atoms from an inert molecular gas such as CF4. The F atoms cause etching of the Si by forming volatile compounds such as SiF4 on the Si surface. Plasma polymerization often proceeds in a series of steps.[38] Thus, for example, high molecular weight species can be formed in a glow discharge from low molecular weight starting material by the association processes discussed in the previous section. These high molecular eight species condense on the substrates, where they are cross-linked by plasma radiation and electron bombardment to form a polymer film.

6.0 SURFACE REACTIONS 6.1 Introduction Surfaces in contact with plasmas are bombarded by electrons, ions, and photons. The electron and ion bombardment is important and is used in materials processing, particularly during deposition and etching. Less is known about the influences of the plasma radiation. The relative number of ions and electrons which are incident on a surface depends on whether it is biased as a cathode, an anode, or is electrically isolated. In this section, some of the effects of ion bombardment and electron bombardment, and of plasma bombardment of an electrically floating surface, are discussed briefly. 6.2 Ion Bombardment The momentum exchange associated with ion bombardment can cause rearrangement and ejection (sputtering) of surface atoms. The

94


rearrangement can have dramatic effects on the structure and properties of a growing film[53] and is of importance in the processes of ion plating and bias sputtering. The ejection is important in the processes of sputter cleaning and deposition. Accordingly, these mechanisms are discussed in considerable detail in Chs. 5 and 13. At low working pressures (collisionless ion transport), the energy of ions bombarding a cathode surface will be about equal to the difference between the cathode potential and the plasma potential (approximately equal to the applied cathode-to-anode potential). The current density, bias voltage, sheath thickness, and plasma properties are related by Eqs. 38 and 39. At higher pressures, where ion collisions become important, the bombarding flux consists of both ions and energetic neutrals because of charge exchange collisions (see Fig. 2.10). Thus the average bombardment energies are considerably less than the potential drop across the cathode dark space. This is illustrated in Fig. 2.24 with a histogram showing the cathode arrival energies of 100 Ar+ ions which have crossed a sheath having a voltage Va in Ar gas at 2.5 mTorr. Approximately half (45%) of the ions arrive at the cathode with energies corresponding to less than 10% of the sheath voltage. The sheath parameters for the high pressure case are related by Eq. 39.

Figure 2.24. Calculated ion-energy distribution histogram showing the effect of charge exchange (from Ref. 50).


95

Ion bombardment can greatly influence the processes involved in the adsorption of molecules onto surfaces and their subsequent reactions. The process of molecular adsorption[41] and surface compound formation is illustrated in Fig. 2.25 for the case of gas phase etching. The CVD case with the formation of a nonvolatile product is similar. Any of the steps shown in the figure can be rate-limiting. Physical adsorption is due to polarization (van der Waals) bonding. It is a nonactivated process and occurs with all gas surface combinations under appropriate conditions of temperature and pressure. Adsorption energies are typically less than 0.5 eV. Chemisorption involves a rearrangement of the valence electrons of the adsorbed and surface atoms to form a chemical bond. It involves an activation energy and has a high degree of specificity between gas-surface combinations. Adsorption energies are typically 1 to 10 eV. Molecules may be chemisorbed in their molecular state or may dissociate into atoms. The latter case is known as dissociative chemisorption. Dissociative chemisorption is generally a precursor to compound formation, which is also an activated process. Various types of chemisorption bond sites can exist on a solid surface. Thus both molecular and dissociative chemisorption can occur simultaneously on the same surface. Ion bombardment can influence these processes in the following ways: 1. Ion bombardment can cause adsorbed molecules to dissociate, thereby overcoming the activation energy for this process. 2. Ion bombardment can create surface defect sites which have reduced activation energies for the occurrence of dissociative chemisorption or for the formation of a solid compound. 3. Ion bombardment can remove (by sputtering) foreign species from a surface. Such species may interfere with the dissociative chemisorption of a preferred species. Low-energy ion irradiation during film deposition can have dramatic effects on the microstructure and microchemistry, and hence physical properties, of as-deposited layers as discussed in detail in Chapter 13 and Ref. 53. Applications in which low-energy ion/surface interactions have been used to modify film microstructure include: densification and increased oxidation resistance of optical films; minimization or elimination of columnar microstructure in microelectronic metallization layers; altering the state of stress, average grain size, and preferred orientation; increased film/substrate adhesion; enhanced conformal coverage; controlled magnetic anisotropy in recording layers; and “low-temperature” epitaxy.

96


Figure 2.25. Schematic representation of surface chemisorption and volatile compound formation during dry etching.

While films in most of the above application areas are deposited by bias sputter deposition or plasma-assisted CVD, experiments to isolate ion irradiation effects are often carried out using ion beams. One example is illustrated in Fig. 2.26 showing experimental and calculated (Monte Carlo simulations) densities of CeO2 films deposited at ambient temperature by simultaneous evaporation of Ce and O2+ irradiation from an ion-beam source. The experiments were carried out as a function of ion energy Ei for an ion-tovapor flux ratio of Ji /Jv of unity.[54] The film density initially increased with increasing Ei due primarily to ion implantation, recoil implantation, and, to a lesser extent, sputtering of weakly bound species. However, an optimum Ei for densification was reached as an increasing fraction of the ion energy was lost deeper in the lattice leaving vacancies which could not be filled by arriving vapor species. The optimum ion energy, which depends upon the masses of the collision partners, was ≈200 eV in this case. It should be noted, as discussed in Ch. 13, that while ion irradiation is useful for increasing the density and modifying the microstructure of films deposited at low temperatures, other irradiation-induced effects such as increased defect densities occur simultaneously. This is shown in Fig. 2.27 from the work of Huang et al.[55] who studied the effects of Ar+ ion bombardment during the growth of Ag films at room temperature using a dual ion beam apparatus. They found that the grain size decreased while the dislocation number density increased with increasing average irradiation energy per deposited Ag atom. At elevated growth temperatures, however,


97

low-energy ion irradiation can have the opposite effect and actually reduce residual defect densities in as-deposited films. This has been demonstrated by Hultman et al.[56][57] who used electron microscopy to investigate the dislocation structure in epitaxial TiN films deposited by bias magnetron sputtering of Ti in pure N2 at growth temperatures between 550 and 850oC.

Figure 2.26. Experimental and theoretical values of the density D of CeO2 films deposited at ambient temperature by simultaneous evaporation of Ce and ionbeam acceleration of O2+ as a function of ion energy Ei for an ion-to-vapor flux ratio Ji /Jv = 1. The bulk density of CeO2 is 8.1 g/cm3 (from Ref. 54).

In addition to modifying film microstructure, low-energy ion irradiation is often used during thin-film growth to controllably alter the composition of as-deposited layers. Examples include preferential sputtering from the growing film during deposition of alloys[58]-[61] enhanced reactive gas incorporation during deposition of compounds[62]-[65] and increased dopant incorporation probabilities combined with better control of dopant depth distributions.[66][67] Again, however, ion bombardment can result in potentially deleterious effects, depending upon experimental design, such as rare-gas incorporation in sputter-deposited films.[68]-[71] Mechanisms associated with accelerated-particle/film interactions leading to changes in incorporation

98


probabilities range from purely physical effects such as implantation and recoil processes to irradiation-assisted chemistry.

Figure 2.27. The average grain size and dislocation number density nd in Ag films deposited at room temperature as a function of the average energy 〈E〉 per deposited atom (from Ref. 55).

Reactive ion etching technology also relies heavily on ion-irradiationinduced effects for both stimulating chemical reaction channels and providing anisotropy control. An example of the former is shown in Fig. 2.28 illustrating results for Ar+-ion-assisted F2 /Si chemistry. F2 has a very low probability for dissociative chemisorption on Si.[34] Consequently the etch rate via the formation of volatile SiF2 is low. Ar+ irradiation greatly increases the etch rate by promoting dissociative chemisorption. Fig. 2.28 shows that for the experimental conditions listed, the Si sputter etch rate using 500 eV Ar+ was 2.5 Å/min. The etch rate increased by a factor of approximately 3.5 in the presence of F2 gas. However, the Si etch rate due to F2 itself, in the absence of Ar+ irradiation, was less than 0.1 Å/min.


99

Figure 2.28. The results of beam experiments designed to investigate ionstimulated interactions between F2 and Si (from Ref. 42).

100


6.3 Electron Bombardment Electron irradiation is a primary source of substrate heating during film deposition by DC and RF diode sputtering.[72][73] Energetic electron, as well as photon, irradiation of ionically bonded substrates has also been shown to strongly affect film nucleation kinetics through the creation of charged surface vacancies which act as preferential adsorption sites.[74]-[77] Reduced epitaxial temperatures have been reported for many film/substrate combinations including Si and Ge on NaCl[78] and PbTe on CaF2 .[76] Electron irradiation can also give rise to surface chemistry during film growth through, for example, excitation and ionization of adsorbed molecules into states leading to dissociation, bond rearrangement, or desorption. Adsorbed organic molecules can be polymerized by electron irradiation. An example of electron-stimulated surface chemistry during plasma etching is shown in Fig. 2.29. XeF2 dissociatively chemisorbs on SiO2 but etching does not occur due to a high activation barrier for the reaction channel leading to the formation of SiF4. Electron bombardment alone has been observed to remove O from the surface of SiO2 and produce elemental Si,[43][44] but it does not cause etching. However, when SiO2 is subjected to electron bombardment in the presence of XeF2 , etching occurs at relatively high rates, ≈ 200 Å/min in the example given in Fig. 2.29.[42] 6.4 Glow Discharge Surface Cleaning and Activation Glow discharge cleaning, in which electrically isolated parts are immersed in a low-pressure plasma, has been used for many years,[45] particularly for glass and other non-conducting materials that cannot be subjected to simple DC sputter etching. The process, although highly empirical, often provides an effective final cleaning step prior to vacuum deposition. Working gases are typically air, O2, or Ar. Recent work on damage production and sputter cleaning of substrate surfaces prior to epitaxial growth[79]-[82] suggests that low-energy ionirradiation-induced damage can be continuously annealed out at elevated temperatures. Yu[82] used low-energy electron diffraction (LEED) to show that the temperature required to maintain a Si(111)7x7 surface reconstruction during Ne+ ion irradiation decreased from≈ 450 to 150 oC as the ion energy was decreased from 500 to 80 eV. In sputter cleaning experiments employing cross sectional transmission electron microscopy, Gaverick et al.[81] used a low power RF plasma with an acceleration potential of 100 V to etch Si(100) substrates at 750oC immediately prior to Si deposition bylow-


101

Figure 2.29. The results of beam experiments designed to investigate electronstimulated interactions between XeF2 and SiO2 (from Ref. 42).

102


pressure CVD. Rutherford backscattering spectroscopy combined with planview and cross sectional transmission electron microscopy analyses showed the film and substrate to be defect free. Corona discharges, operated at atmospheric pressure, have long been used to prepare polymer surfaces for processing. More recently, low-pressure glow discharges are being used to modify surface chemistry and promote adhesion with vacuum-deposited metal overlayers. X-ray photoelectron spectroscopy (XPS) studies of the effects of O2 plasma treatments on ABS, polypropylene,[46] and polystyrene[47] surfaces showed the formation of both single and double C-O bonds. This, in turn, led to stronger metal overlayer adhesion through the formation of oxygen bridge bonds between C and metal atoms. Bodo and Sundgren[83] obtained similar increases in metal overlayer adhesion for Ti on polyethylene using an Ar+ bombardment pretreatment to remove low molecular weight impurities, promote cross-linking, and allow the formation of a carbidic Ti-C interfacial layer as observed in XPS. Both Ar+ ion irradiation and O2 plasma pretreatments also increased the adhesion of T on polydimethylsiloxane (a silicone rubber) due to the formation of Ti-C and Ti-O bonds.[84]

ACKNOWLEDGEMENTS The authors gratefully acknowledge the support of the Joint Services Electronics Program and the Materials Science Division of the Department of Energy over the course of several years.


103

REFERENCES 1. Thornton, J. A. and Penfold, A. S. in:Thin Film Processes, (J. L. Vossen and W. Kern, eds.), p. 75, Academic Press, New York (1978) 2. McDaniel, E. W., Cermak, V., Dalgarno, A., Ferguson, E. E. and Friedman, L., Ion-Molecule Reactions, p. 345, Wiley-Interscience, New York (1970) 3. Hirschfelder, J. O., Curtiss, C. F. and Bird, R. B., Molecular Theory of Gases and Liquids, p. 523, Wiley, New York (1954) 4. Sutton, G. W. and Sherman, A.,Engineering Magnetohydro-dynamics, McGraw-Hill, New York (1965) 5. Cobine, J. D., Gaseous Conductors, Dover, New York (1958) 6. The Applications of Plasmas to Chemical Processing, (R. F. Baddour and Robert S. Timmins, ed.), MIT Press, Cambridge, Mass. (1967) 7. Thornton, J. A., J. Vac. Sci. Technol., 15:188 (1978) 8. Griem, H. R., Plasma Spectroscopy, p. 129, McGraw-Hill, New York (1964) 9. ter Haar, D., Elements of Statistical Mechanics, p. 381, Holt, Rinehart and Winston, New York (1960) 10. Rose, D. J. and Clar, M., Jr.,Plasmas and Controlled Fusion, p. 80, MITWiley, New York (1961) 11. Chen, F. F., Introduction to Plasma Physics, Plenum Press, New York (1974) 12. Spitzer, L., Jr., Physics of Fully Ionized Gases, Interscience, New York (1956) 13. Delcroix, J. L. Introduction to the Theory of Ionized Gases, p. 128, Interscience, New York (1960) 14. Chapman, S. and Cowling, T. G., The Mathematical Theory of NonUniform Gases, p. 90, Cambridge Univ. Press, Cambridge, England (1960) 15. von Engel, A.,Ionized Gases, Clarendon Press, Oxford, England (1965) 16. McDaniel, E. W., The Mobility and Diffusion of Ions in Gases, p. 132, Wiley, New York (1973)

104


17. Brown, S. C., Basic Data of Plasma Physics, MIT Press, Cambridge, Mass. (1959) 18. Chen, F. F., in: Plasma Diagnostic Techniques, (R. H. Huddlestone and S. L. Leonard, eds.), p. 113, Academic Press, New York (1965) 19. Mittleman, M. H., in: Plasma Dynamics, (F. H. Clauser, ed.), p. 54, Addison-Wesley, New York (1960) 20. Bohm, D., Burhop, E. H. S. and Massey, H. S. W., in:The Characteristics of Electrical Discharges in Magnetic Fields, (A. Guthrie and R. K. Wakerling, eds.), p. 13, McGraw-Hill, New York (1949) 21. Glasstone, S., and Louberg, R. H.,Controlled Thermonuclear Reactions, p. 459, Van Nostrand, New York (1960) 22. Thornton, J. A., J. Vac. Sci. Technol., 15:171 (1978) 23. McTaggart, F. K., Plasma Chemistry in Electrical Discharges, Elsevier, New York (1967) 24. Ganger, B., Der Elecktrische Durchschlag, Springer-Verlag, Berlin (1953) 25. Brown, S. C. and MacDonald, A. D., Phys. Rev., 76:1629 (1949) 26. McDaniel, E. S., Collision Phenomena in Ionized Gases, Ch. 13, Wiley, New York (1964) 27. Williams, D. G., J. Vac. Sci. Technol., 11:374 (1974) 28. Fraser, D. B., in: Thin Film Processes, (J. L. Vossen and W. Kern, eds.), p. 131, Academic Press, New York (1978) 29. Waits, R. K., Ibid, p. 131 30. Libby, W. F., J. Vac. Sci. Technol., 16:414 (1979) 31. Transfer and Storage of Energy by Molecules,(G. M. Burnett and A. M. North, eds.), Wiley-Interscience, New York (1969) 32. Christophourou, L. G., Atomic and Molecular Radiation Physics, p. 6, Wiley-Interscience, New York (1971) 33. Winters, H. F., Coburn, J. W. and Kay, E.,J. Appl. Phys.,48:4973 (1978) 34. Coburn, J. W. and Winters, H. F., J. Vac. Sci. Technol., 16:392 (1979) 35. Muschlitz, E. E., Jr., Science, 159:599 (1968)


105

36. Hollahan, J. R. and Rosler, R. S., in: Thin Film Processes, (J. L. Vossen and W. Kern, eds.), p. 335, Academic Press, New York (1978) 37. Melliar-Smith, C. M. and Mogab, C. J., Ibid, p. 497 38. Yasuda, H. Ibid, p. 361 39. Kern, W. and Ban, V. S., Ibid, p. 257 40. See Ch. 7 41. Chemisorption and Reactions on Metallic films, (J. R. Anderson, ed.), Academic Press, New York (1971) 42. Coburn, J. W. and Winters, H. F., J. Appl. Phys., 50:3189 (1979) 43. Thomas, S., J. Appl. Phys., 45:161 (1974) 44. Carriere, B. and Lang, B., Surface Science, 64:209 (1977) 45. Holland, L.,Vacuum Deposition of Thin Films, Ch. 3, Chapman and Hall Ltd., London (1966) 46. Burkstrand, J. M., J. Vac. Sci. Technol., 15:223 (1978) 47. Burkstrand, J. M., Appl. Phys. Lett., 33:387 (1978) 48. Hansen, R. H. and Schonhom, H., Polymer Lett., 4:203 (1966) 49. Bunshah, R. F. and Raghuram, A. C., J. Vac. Sci. Technol., 9:1385 (1972) 50. Davis, W. D. and Vanderslice, T. A., Phys. Rev., 131:219, (1963) 51. Butler, H. S. and Kino, G. S., Phys. Fluids, 6:1346 (1963) 52. Gorczyca, T. B. and Gorowitz, B., in: VLSI Electronics: Microstructure Science, 8:69, (N. G. Einspruch and D. M. Brown, eds.), Academic Press, New York (1984) 53. Greene, J. E., Barnett, S. A., Sundgren, J. E. and Rockett, A., in: IonBeam Assisted Film Growth, p. 101, Elsevier, Amsterdam (1988) 54. Muller, K. H., Applied Physics, A40:209 (1986) 55. Huang, T. C., Lim, G., Parmiagiani, F. and Kay, E., J. Vac. Sci. Technol., A3:2161 (1985) 56. Hultman, L., Helmersson, U., Barnett, S. A., Sundgren, J. E. and Greene, J. E, J. Appl. Phys., 61:552 (1987)

106


57. Hultman, L., Barnett, S. A., Sundgren, J.-E. and Greene, J. E.,J. Crystal Growth,92:639 (1988) 58. Winters, H. F., Ramondi, D. L. and Horne, D. E., J. Appl. Phys., 40:2996 (1969) 59. Tarng, M. L. and Wehner, G. K., J. Appl. Phys., 42:2449 (1971) 60. Cuomo, J. J. and Gambino, R. J., J. Vac. Sci. Technol., 12:79 (1975) 61. Zilko, J. L. and Greene, J. E., J. Appl. Phys., 51:1549 (1980) 62. Brett, M. J. and Parsons, R. R., Can. J. Phys., 63:819 (1985) 63. Shimizu, S., Tsukakoshi, T., Komiya, S. and Makita, Y., GaAs and Related Compounds, Inst. Phys. Conf. Series 79:91 (1985) 64. Harper, J. M. E., Cuomo, J. J. and Hentzell, H. T. G., J. Appl. Phys., 58:550 (1985) 65. Sundgren, J. E., Johansson, B. O., Rockett, A., Barnett, S. A. and Greene, J. E., Physics and Chemistry of Hard Coatings, American Inst. Phys. Series Conf. Proc. 149:95 (1986) 66. Hasan, M. A., Knall, J., Barnett, S. A., Sundgren, J. E., Markert, L. C., Rockett, A. and Greene, J. E., J. Appl. Phys. 65:172 (1989) 67. Fons, P., Hirashita, N., Markert, L. C., Kim, Y. W., Greene, J. E., Ni, W. X., Knall, J., Hansson, G. V. and Sundgren, J. E., Appl. Phys. Letters, 53:1732 (1988) 68. Winters, H. F. and Kay, E., J. Appl. Phys., 38:2928 (1967) 69. Pan, A. and Greene, J. E., Thin Solid Films, 78:25 (1981) 70. Hoffman, D. W. and Thornton, J. A.,J. Vac. Sci. Technol., 20:355 (1982) 71. Hultman, L., Markert, L. C., Sundgren, J. E. and Greene, J. E., Appl. Phys. Letters, 53:1175 (1988) 72. Ball, D. J., J. Appl. Phys., 143:3047 (1972) 73. Lau, S. S., Mills, R. H. and Muth, D. G., J. Vac. Sci. Technol., 9:1196 (1972) 74. Stirling, D. J., Appl. Phys. Letters, 9:326 (1966) 75. Palmberg, P. W., Todd, C. T., and Rhodin, T. N.,J. Appl. Phys., 39:4650 (1968)


107

76. Jordan, M. R. and Stirland, D. J., Thin Solid Films, 8:221 (1971) 77. Lord, D. G. and Prutton, M., Thin Solid Films, 21:341 (1974) 78. Shimaoka, G., J. Cryst. Growth, 31:92 (1975) 79. Ronsille, R., Boch, R., Destefanis, G. L., and Tissot, J. L., Appl. Phys. Letters, 44:679 (1984) 80. Comfort, J. H., Gaverick, L. M., and Reif, R., J. Appl. Phys., 62:3388 (1987) 81. Gaverick, L. M., Comfort, J. H., Uyeh, T. R., Reif, R., Baiocchi, F. A. and Luftman, H. S., J. Appl. Phys., 62:3398 (1987) 82. Yu, M. L., Appl. Phys. Letters, 40:986 (1982) 83. Bodo, P. and Sundgren, J. E., J. Vac. Sci. Technol., A2:1498 (1984) 84. Bodo, P. and Sundgren, J. E., Thin Solid Films, 136:147 (1986)

3 Surface Preparation for Film and Coating Deposition Processes Donald M. Mattox

1.0 INTRODUCTION

The termsurface preparation has many interpretations depending on the application. For instance, atomically clean surfaces are of particular interest in some studies and these surfaces may be prepared by cleaving a crystal (in an ultrahigh vacuum), or other very careful surface preparation in ultra-high vacuum.[1] Deposition techniques that are extremely sensitive to surface preparation include molecular beam epitaxy (MBE) where great pains are taken to clean the surface before the deposition of the epitaxial layer, and surface chemical reaction studies where submonolayer coverages are important. Other deposition techniques such as ion plating are less sensitive since surface preparation is integral to the deposition process. Substrate preparation for our purposes may be defined as the conditioning of the substrate surface prior to film/coating deposition in order to obtain desirable processing and film/coating properties.[2] Substrate preparation may involve the reduction of the type and amount of “contaminants” to an acceptable level (cleaning), modification of the physical or mechanical properties of the surface,activation of a surface species to enhance reactions, or the addition of desirable species to the substrate surface to aid in nucleation and reaction (sensitization). In the extreme case, surface preparation may mean forming a “new” surface by adding a primer or glue layer. Substrate preparation determines the surface properties and these are directly or indirectly related to the film formation stages of adatom nucle–

108

Surface Preparation

109

ation, interface formation, and film growth. These, in turn, affect film properties such as adhesion, pinhole density, porosity, film microstructure, morphology and mechanical properties. Often local surface properties determine film properties such as pinholes which determine the product yield. Surface treatments that do not influence the product in a desirable way are unnecessary and expensive. Surface preparation is an integral part of any film/coating deposition process. The objective of surface preparation processes is to allow the fabrication of an acceptable product in the most reproducible and economical way. In many cases there are allowable trade-offs between surface preparation and subsequent processing. For example, an increase in the deposition temperature may decrease the surface cleaning requirements. As the technological demands on films and coatings increase, the need for better and more reproducible surface preparation techniques also increases. There is a wide variety of approaches to surface preparation and each film-substrate couple, deposition process, and function requires specific techniques and development. Typically, surface preparation processes are developed empirically and controlled by good processing specifications. Process specifications and travelers are the key to obtaining reproducible surface preparation processing, fabrication processes, and thus product reproducibility. Specifications define the materials, equipment and procedures that are to be used. Travelers define what has been done to each individual part or lot. Specifications are the end-product of a surface preparation development program. Travelers should contain a response by the operator (e.g., time, meter reading, temperature, etc.). An important factor in surface preparation is the condition of the initial surface. A process developed for one surface condition may not be satisfactory for another surface condition. The initial substrate material, condition and history (contamination) should be known, and its condition and properties should be specified where possible. Monitoring of the surface preparation is often difficult since any testing of the surface usually contaminates the surface. Generally, processing relies on following specifications and possibly monitoring and testing samples from each lot of surfaces. In addition to the surface preparation process, the handling and storage of prepared surfaces is an important part of the fabrication process. If the prepared surfaces are used immediately or if the final step of the cleaning process is done as part of the deposition process, the problems of

110


maintaining the desired surface may be appreciably different than if the surfaces are exposed to the environment and recontamination, or when surface changes may occur with time. Some film deposition processes and material combinations are more sensitive to surface preparation processes than others. Some deposition processes may have harmful surface effects such as in CVD where hot corrosion of the substrate surface by reaction products may give poor adhesion of the deposited material. In others, the deposition process may aid in surface preparation; in CVD for example, hydrogen firing can clean the substrate surface before the film precursor gases are injected. In many deposition processes the surface preparation is a separate step from the film deposition, but in some cases the deposition process includes a surface preparation step (e.g., ion-plating/sputter-cleaning, hot dip galvanizing/fluxing, electroplating/off-plating). When surface preparation is separate from the deposition process, the preparation of high quality films in many cases requires a final in situ surface preparation step in the deposition system.[3][4] An example of in situ surface preparation is the plasma cleaning of glass prior to deposition of optical coatings and mirror surfaces. This chapter covers a broad range of surface preparation techniques and gives the reader an appreciation of the factors involved in developing a reproducible surface preparation procedure for a specific application.

2.0 CONTAMINATION A contaminant is any material on a surface that interferes with the processing or performance of the surface. Contaminants may be reacted layers such as oxides, adsorbed layers such as hydrocarbons, segregated surface layers, or particulates. The contaminant may originate from:(i) natural reaction with the ambient (oxides, sulfides), (ii) adsorption from the ambient (hydrocarbons, water), (iii) processing steps (oils, fingerprints), (iv) handling and storage (polymers, oils), (v) settling from the ambient (particulates), (vi) electrostatic attraction in the ambient (particulates), (vii) outgassing or outdiffusion from the bulk (plasticizers, water, solvents -plastics) or (viii) contact with contaminated surfaces (silicone oils have a very high creep rate). Some of this recontamination is unavoidable but some is avoidable with proper fabrication, handling, and storage techniques.

Surface Preparation

111

Many contaminants can be predicted by knowing something about the material in general and the way that it is normally fabricated and handled. Examples are: (i) plastics absorb water and solvents easily, (ii) metals are machined and deformed using oil lubricants, (iii) plastics are molded using plasticizers to make the material fluid, etc. The presence of contaminants can be detected without necessarily identifying the composition of the contaminant. For instance, if a glass surface is contaminated with a hydrocarbon (hydrophobic), the wetting angle of a fluid drop will be high (doesn’t wet).[5] However, this type of test must be used with caution since soap residue (hydrophilic) on the surface will make the surface wettable like a clean surface. Adsorption of a tracer such as a radioactive material may also be used to detect the presence of many contaminants. Particulates originate from a variety of sources including: (i) wear mechanisms,(ii) vaporization,(iii) vapor phase nucleation,(iv) evaporation of aerosols, and (v) shedding of particles (skin, paper, cloth etc.). Particulates adhere to the surface by weak chemical bonds (van der Waals),[6] but for small particles, the most important adhering mechanism is condensation of water in the “crack” between the particle and the surface. The evaporation of aerosols and vapor phase nucleation are the most important sources of ultrafine particles (10 -100 nm). Surface adsorption can be very dependent on the surface and the adsorbing species. For instance, most oxide surfaces do not adsorb O2 while conducting and semiconducting surfaces do so easily.[7] 2.1 Recontamination Recontamination of surfaces that have been cleaned is a major concern. The recontamination rate and amount is a function of time, temperature, and environment. For example, the oxidation of reactive materials begins immediately on exposure to oxygen. On materials such as aluminum and silicon, 10 Å of oxide will re-form within seconds then slowly increase in thickness. Recontamination can also occur by adsorption of vapors from the environment. Figure 3.1 shows the recontamination rate of cleaned gold surfaces in various environments, as determined by coefficient of adhesion measurements.[8][9] The contaminants are assumed to be condensed hydrocarbon vapors. Note that recontamination begins immediately. Recontamination can come from a number of other sources such as poor environmental control, poor handling and storage, contamination by subsequent processing, etc.

112


Figure 3.1. Recontamination of clean gold surfaces in various environments as measured by an Au-Au adhesion tester.[8][9]

Recontamination can also occur in the cleaning process itself. Complete rinsing is necessary; otherwise residues from the processing chemicals will recontaminate the surface. For example, in the final rinse, if the part is submerged in the rinse tank then drawn up through the liquid surface on which particles have accumulated, the particles will be painted on the surface and must be removed before they are allowed to dry. During storage and handling, the type and degree of recontamination is dependent on: (i) time, (ii) temperature, (iii) environment and, (iv) surface condition. Many contaminants “harden” with time and become more difficult to remove, so after exposing the surface to a contaminating process or environment it is best to clean the surface as soon as possible. Recontamination can occur in the processing system and during the processing. Reactive gas contamination (such as oxygen or water vapor) may come from residual gases, gases desorbed from surfaces, real leaks and virtual leaks. Heating and plasma-surface interactions enhance gas

Surface Preparation

113

desorption from surfaces. Outgassing from virtual leaks is time/temperature dependent and processing should be designed to allow for desorption from these sources. Often, in plasma processes, gas throughput is decreased and contaminants may build up in the system, and proper gas throughput or cleanflush-pump cycles should be employed to reduce contaminant levels. Examples of processing recontamination include: (i) plasma desorption and activation of contaminants in plasma processing, (ii) outgassing of thermal vaporization source material, (iii) particulate generation in the deposition system, (iv) particulate deposition due to turbulence in a vacuum pumping system, etc. Vacuum and plasma deposition systems may have their contaminant gas levels lowered by using the proper construction materials and techniques, and conditioning their internal surfaces. Conditioning may be done by: (a) Heating (bake-out, thermal desorption)[10] (b) Oxidizing techniques (UV/O3), [11][12] (NO at 200°C)[13] (c) Pump/plasma-discharge/pump to desorb wall species (ion scrubbing: chamber is a grounded anode of the discharge) using an inert gas, oxygen[14] or hydrogen plasma (d) Physical or chemical sputtering of the walls using an inert or reactive plasma species such as hydrogen[15] or compounds containing chlorine or fluorine (chamber is cathodic to the plasma) Recontamination is controlled by controlling the processing and storage environments.

3.0 ENVIRONMENT CONTROL A key aspect of surface preparation is the control of the processing environment to avoid contamination during processing and in subsequent handling, storage, and processing. Environmental factors include:(i) particulates,(ii) ambient gases,(iii) processing gases,(iv) condensable vapors, (v) fluids, and (vi) contacting solids. Particulates come in all sizes. Metal smokes, aerosols (for example sneezes and sea spray), viruses and tobacco smoke provide some of the smallest particle sizes.

114

Deposition Technologies for Films and Coatings Particulate contamination may be minimized by: (a) Minimizing dust and particulate generating activities and materials (e.g., clothing/skin/cosmetics, soldering, aerosols) (b) Low velocity air currents, little turbulence (c) Elimination of electrostatic charging of insulator surfaces (d) Air filtration—“clean” rooms and stations

Particulates on smooth or patterned surfaces (semiconductor) can be detected by operators using optical microscopes (slow and costly), or by using a scanning laser microscope which detects scattered light. Ultraviolet fluorescence can be used to detect some types of particles. Commercial surface particulate detection systems are available. Airborne particulate contamination may be effectively controlled by filtration of air (90 - 100 ft/min), through directional(laminar flow) dry fiber filters (HEPA—High Efficiency Particle Air). HEPA filters can be made from a variety of materials, and filters compatible with the environment should be used. For instance, it has been reported that salt particles on some filter fiber materials absorb water and degrade the filter to the point that the filter produces particulates. Filters allow the fabrication of clean rooms, clean benches, etc.,[16] and must be utilized with care in order to maintain a low particle count.[17] It should be noted that air filtration doesnot remove vapor contamination. In the United States, GSA- Federal Standards 209b utilize the number of particles per cubic foot of volume with a size greater than 0.5 microns as the standard (no particles larger than 5 microns). Air filtration with proper flow patterns can provide a Class 100 or better environment (100 particles per ft3 ). In 1986, a Class 100 clean room cost an estimated $400 - 500 US per ft2 (some say $1000) to construct and $30 US per ft2 per day to operate. Continuous care, maintenance and personnel training are necessary for a properly functioning clean room! Airborne particles larger than 0.5 microns are typically counted by light scattering. Below 0.5 microns the particles are counted by first condensing a vapor on the surface (like a contrail from a jet) and then using light scattering, or by electrostatically charging the particle then counting it, or a combination of the two methods. Particles may be selectively attracted to charged surfaces. It is therefore important to prevent electrostatic charging of critical surfaces. When blowing with an air nozzle, the air should be ionized to prevent

Surface Preparation

115

electrostatic charge buildup. Permanently charged surfaces (electrets) can be used to preferentially attract particulates. (Note: electret materials have a permanent surface charge. They are mainly plastics that have been heated and stretched in a DC electric field. Electret materials may be used in brushes, filters, or as surfaces.) Humans, their clothing, and behavior are a major source of contamination. In clean rooms, particulate generation is minimized by using special body covering and other techniques. Ultimately robots may be used to eliminate one of the major sources of particles—man. In order to attain Class 1 and 10 environments and to control particles smaller than 0.5 microns, it is proposed that substrate handling and processing will have to be done in small compartmentalized units where the substrates will not be exposed to the ambient environment. An example of such a system is the completely-contained processing for metallizing and assembling quartz crystal oscillators, where vapor and particulate contamination are eliminated to prevent frequency shift due to contamination of the crystal surface during use. In the future, more use is expected to be made of containers and processing equipment that can be mechanically mated so as to only need small volumes of Class 1 environments. Clean rooms may be less important in the future! Particulate contamination from processing gas supplies may be controlled by filtration. Filtration at the point-of-use is often done with 0.2 micron filters. Teflon filters should be used in oxygen lines. Particulate contamination in flowing gases may be monitored by the scattering of a laser beam.[18] Particulates generated in gas piping may be due to:(a) flaking from walls, oxides, fluxes, polymers, (b) wear particles from mechanical equipment, (c) contamination from opening system, (d) leaks; and affected by: (i) wear (valves, pumps),(ii) mechanical vibration, (iii) thermal cycling, and(iv) changes in flow velocity. Contamination of gas supplies by unwanted reactive gases can be a problem. In order to prevent gas contamination, one can: (i) use ultrapure gases from tanks,(ii) use vapors from liquid gases (LN 2), (iii) purify the gases, and (iv) be careful to have non-contaminating plumbing. Gas purification can be used to remove some gaseous contaminants from gas supplies. Purifiers use hot reactive beds (chips) (Ti, U, Cu) for removal of oxygen, or diffusion— Pd for H2 , Ag for O2. Commercial purifiers will purify silane, ammonia, hydrogen and the inert gases to less than 10 ppb of O2, H2O, CO2 , and chlorinated compounds.

116


Particulate contamination from fluids can be avoided by filtration.[19] Care should be taken that the filter does not contaminate the fluid by extracting (leaching) material from the filter (use Teflon™ or ceramic filter material). Particulate contamination in fluids can be measured directly by light scattering. Overflow tanks should be used in rinsing operations. Particulates from the air tend to float on the surface of fluids (like water spiders) and paint on the substrate surface as it is withdrawn through the fluid surface giving extensive particulate recontamination of a cleaned substrate surface. One ambient “contaminant” that should be controlled is electrostatic charging. This is done by controlling the humidity (typically 40 - 45% relative humidity) and using ground straps, antistatic coatings, and conductive clothing on personnel who handle sensitive electronic devices. Electrostatic charging of insulator surfaces contributes to particulate contamination by attracting and holding particles. Electrostatic charging of surfaces can result from blow-off with dry air. The dry air should be ionized before being used for the blow-off operation. The humidity in a clean room is normally controlled by dehumidifying using cold surfaces (air conditioning, or air compression which is more costly) then re-humidifying using steam or “foggers”. It has been proposed that the humidifying operation is a major source of fine particulate contamination in the clean room environment since the evaporation of aerosols is a major source of fine airborne particulates. Condensable vapor contamination is generally not controlled in the processing environment except by venting and segregation of vapor producing processes (soldering, electroplating, etc.) from “clean” areas. Hydrocarbon vapors are the most common vapor contaminants and are controlled in the small volumes used for handling and storage by selective absorption (freshly oxidized aluminum), or by continuous oxidation in a ultraviolet/ozone atmosphere (UV/O3—see cleaning section), or by condensation on cold surfaces. Contaminant pick-up from surfaces is controlled by (control of) surface materials, good housekeeping, smooth surfaces, use of coverings (finger cots, lint-free cloth), high molecular weight organics (nylon and Teflon™) or metal for holders and tools, and the use of vacuum tools for handling wherever possible. Vacuum tooling for holding is preferable to other types of handling tools since it minimizes abrasive transfer of material. Special low-contaminant materials have been developed for semiconductor processing applications; unplasticized polyethylene seems to be

Surface Preparation

117

best,and gloves of this material—furnished on paper rolls—is the recommended handling material. However abrasive transfer of organic materials from soft plastics can be a problem. Woven nylon gloves prevent direct contact between skin and surfaces but do allow sweat and body oils to wick through; rubber finger cots should be worn under the nylon gloves. Alcohol, acetone, and many other solvents which are used in cleaning processes will leach organics from vinyl gloves. When using these solvents, unplasticized polyethylene gloves should be used. Some vinyl (and about all latex) gloves may have powder on them and, of course, this is a source of particulates. Processing chemicals may be contaminated when received so ultrapure chemicals (semiconductor grade) should be used. Improperly rinsed surfaces which have impure chemicals on them (solvents, etchants) may leave residues on drying. A chemical may become contaminated by being in contact with a material which it dissolves or attacks: alcohol in contact with many plastics— vinyl (use polyethylene - no plasticizers); Tygon™ removes phthalate plasticizers (use Teflon™). Hydroscopic materials such as anhydrous chemicals (alcohols) will pick up moisture from the atmosphere on exposure. Chlorinated solvents may react with water vapor and become contaminated with HCl, thus becoming corrosive. If impure fluids are allowed to dry on a surface, they leave residues. These residues are then very difficult to remove. Residue analysis consists of allowing a volume of the chemical to evaporate and analyzing the residue which remains (ASTM Method D1353-78), or analyzing the particulate residue from a sprayed droplet (Wen). Often residue can be detected by the “fogging” of what should be a clean glass surface on evaporation of some of the solution. Residues can be minimized by rinsing in copious amounts of ultrapure water or other appropriate solvent. Wet surfaces should not be allowed to dry without rinsing with a low residue solution! Chemicals can be contaminated by “carry-over” from a previous process. Carry-over can be minimized by good rinsing between cleaning/processing steps. Metallic contaminates in electrolytes may result in surface contamination by displacement plating from solution (Zn and Sn)—don’t use galvanized parts or soldered plumbing for transferring ultrapure chemicals such as water. Sodium contamination is of major concern in silicon technology. Sodium can come from leaching of soft glass, and fingerprints, as well as chemicals, furnace liners, etc.

118


The most common rinsing technique is to use successive rinses (cascading rinses) in ultrapure water until the rinse water has a high resistivity (> 15 megohm). This is called “rinse to resistivity”. Ultrapure water (18 megohm - cm resistivity) is a widely used chemical for cleaning and rinsing since it leaves a minimum of residues. Water purity is typically measured using a conductivity cell that measures the ionic concentration in the water. The semiconductor industry standards call for detection of ionic impurities to 5 ppb NaCl equivalent. Specific ion content may be measured using ion chromatography. Conductivity measurements do not measure the organic or biological contamination and some type of residue analysis should be used to measure these impurities. Typical industrial specifications of ultrapure water for endpoint use are: 1. Resistivity—18 megohm continuous at 25°C 2. Particle count—less than 500 particles (0.5 microns or larger) per liter 3. Bacteria count—less than one colony (cultured) per cc 4. Organics—less than one part per million 5. Total electrolytes—less than 5 parts per billion 6. Quantity requirements 7. Peak-level usage The ultrapure water is made by: 1. Pretreatments—pH adjustment, coagulation, filtration 2. Reverse osmosis—semipermeable membrane (pore size 10-3 to-4 microns) rejects salts, dissolved solids (90 - 98%) and organics (99%)—400 to 600 psi feedwater[20] 3. Degasification—remove dissolved CO2 4. Ion exchange resins (anion & cation)—remove ions by exchanging H+ for cations and OH- for anions. 5. Absorption materials (activated carbon)—remove organics 6. Filtration—remove particulates and biological matter, 0.2 microns for bacterial, 1.0 microns general 7. Ultraviolet radiation—kills bacteria on filters 8. Endpoint filtration PVC plumbing should be used with ultrapure water since the pure water is rather corrosive to metals (particularly to Cu, Zn). A high volume, ultrapure water facility can be very large and expensive.

Surface Preparation

119

Bacteriological contamination can penetrate porous filters and has been correlated to reduced device yields. Ultraviolet radiation or dissolved ozone may be used to kill the bacteriological contaminate agents (Nebel). End-point filtration is often used to make sure that bacteriological contamination does not get on a part and leave a residue. Activated carbon is an amorphous material with a high surface area (5001500 M2/gram). For use in fluids it has a pore size of about 1000 Å. For use in gases it has a pore size of 12 - 200 Å. Activated carbon has a high affinity for the absorption of organic molecules (better for non-polar than polar). Catalytic agents (Cu, Ag, Cr) can be added to improve the absorption of complex molecules (e.g., gas masks). Activated carbon filters do not remove biological agents effectively. An important part of the rinsing operation is the drying of the surface to prevent particle pickup and adherence—see Sec. 5.0 on drying and outgassing.

4.0 CLEANING PROCESSES “Cleaning” is the reduction of surface contamination to an acceptable level. As a practical matter, a “clean” surface is one that contains no significant amounts of undesirable material; thus what constitutes a clean surface (degree of cleaning) depends on the requirements. The requirements range from those concerned with monolayer coverages and atomically clean surfaces to crude cleaning such as used for fusion welding. The economics are such that unnecessary cleaning is to be avoided. Cleaning processes should be as simple and effective as possible in order to meet the requirements of the processing. Elaborate cleaning processes are often expensive and self-defeating. Often there is a tradeoff between the various stages of the cleaning process, handling/storage, and subsequent processing, such that simple changes in one stage make complex changes in another step unnecessary. Effective cleaning generally consists of two or three stages. The first is removal of gross contamination by fluxes, etchants, or abrasion. In the second stage, the cleaning steps are designed to remove specific types of contaminates such as particulates and organics, by solvents, saponifiers, emulsifiers, oxidation techniques, etc. Cleaning solutions may have several actions to attack specific contamination, such as detergents, solvents, wetting agents and mild etchants. Next, the surface is rinsed, dried and outgassed, (if necessary). Lastly, a final or insitu cleaning step may be used

120


in a very controlled environment such as in a vacuum or plasma deposition chamber or electrochemical solution. 4.1 Particulate Removal Particulate contamination may be removed by detergent washing, liquid spray (high pressure), blow-off, brush-off (in liquid or air), flow-off (liquid or condensing vapor), or spin-off (copious fluids) techniques. The most effective techniques seem to be detergents (with wetting agents) and mechanical rubbing in a fluid. High pressure spray, brush-off under liquid, and flow-off using condensing vapor are less effective. When using any mechanical rubbing technique, care should be taken to prevent contamination by abrasive transfer from the rubbing media. Use gentle pressures. Blow-off techniques have the advantage that they can be done after the substrates have been placed in fixtures and even in a deposition system. The best means of blow-off is to use filtered (0.2 micron) gas from a liquid nitrogen tank. The gas is filtered in the nozzle and some nozzles allow ionization of the gas with a radioactive source. Ionized gas should be used when blowingoff insulator/organic surfaces to prevent electrostatic charge buildup which will attract particles. When using high velocity gases for blow off one should be careful not to entrain particles in the gas stream which could impinge on the surface and stick. An interesting technique studied at the University of Arizona Center for Microcontamination Control is the use of high purity carbon dioxide “snow” formed and blown from a gaseous carbon dioxide cylinder. Apparently the snow scrubs the particles from the surface without leaving residuals or harming the surface. Blow-off of particulates is often done with dusters using canned pressurized gases. One common duster uses dichlorodifluoromethane (DuPont Freon™ 12—CCl2F2, BP: 30°C) which liquifies under pressure. Residuals from the blow-off gases should be checked. Also check for residuals with the spray can in an inverted position (liquid comes out) while spraying. Caution: when using Freon™ dusters, make sure the gas canister is not intended for recharging air conditioning systems—these canisters contain oil lubricants which spray out, particularly when the can is inverted. In optics, it is common to remove particulates from optical surfaces by applying a film that is stripped from the surface—leaving hydrocarbon contamination, no doubt.

Surface Preparation

121

4.2 Abrasive Cleaning The removal of gross contamination by abrasive cleaning involves the use of abrasive pads (sandpaper, emery paper, etc.), impacting particles (glass beads, alumina/silica grit) in air or fluid streams (vapor honing), or mechanical rubbing of particles in a fluid suspension. Grit blasting uses grit (fractured cast iron, alumina) of varying sizes and shapes, accelerated in a gas stream, to deform and gouge the surface. In addition to removing gross contamination, grit blasting roughens and “activates” the surface, and the surface should be coated as soon as possible after grit blasting (less than 2 hours). The Society of Automotive Engineers (SAE) has specifications on grit size and type. Particle bombardment places the surface in compressive stress and may give unacceptable distortion of the part. Glass bead blasting (dry) is a commonly used cleaning technique[21] but may leave shards of glass embedded in soft surfaces. Particles may be entrained in a high velocity gas steam by using a siphon system or a pressure system (sand blasting equipment). Water soluble particles may be used for abrasive cleaning (example: the Prophy-jet™ dental abrasive unit uses 5 micron baking-soda–magnesium-carbonate particles) and allows easy removal of embedded particles. Bead blasting in a fluid (honing) is also used to clean surfaces of gross contamination. 4.3 Etch Cleaning Chemical etching may be used to remove some of the surface material along with the contaminants. This is a very useful technique for getting the surface into a “known” condition, removing surface layers (oxides), and removing difficult-to-remove contaminates. Etchants may change the surface chemistry! Common etchants for glass are sodium or ammonium bifluoride and hydrofluoric acid. Note: when using HF extreme care should be taken to prevent the HF from getting on the skin—bad chemical burns can result. (First aid: flush with water then use magnesium sulfate to neutralize. A commercial magnesium-sulfate/glycerin creme is available as Acid-Aid™.) Acid “pickling” is a common technique for cleaning metal surfaces. [22][23] Acid cleaning of metals may have the detrimental effect of introducing hydrogen into the surface and embrittling the metal. If hydrogen embrittlement is a problem, either don’t use an acid (best) or give the etched part a high temperature vacuum fire after etching. When using etchants for cleaning, care must be taken to prevent selective removal (leaching) of

122


surface constituents that are important to further processing (e.g., etching glass bonded Al2O3 in HF results in selective removal of the glass [Ca-Mg-AlSi-O] which can weaken the surface and give problems with adhesion).[24] Sometimes chemical etching does not remove some constituents from a surface and leaves a “smut” that must be removed by another etching step. For example, etching aluminum alloys with NaOH leaves a copper smutand/ or a silicon smut. These may be removed with HNO3 or HNO3/HF respectively. In some cases an etchant can be devised that will etch all the constituents uniformly; for instance, in etching aluminum containing silicon (1%) IC metallization, concentrated nitric acid plus ammonium bifluoride (100 cc:6.8 gr) may be used. The etching mechanism is oxidation of the aluminum and the silicon, then etching of the resulting oxides—the etchant actually etches silicon more rapidly than the aluminum. 4.4 Fluxing Fluxes remove oxides by dissolving them or by undercutting and floating the surface layers away.[25][26] 4.5 Alkaline Cleaners Alkaline cleaners are saponifiers which convert organic fats to watersoluble soaps. Saponifiers are alkaline and are often in the form of hot solutions. Strong alkaline cleaners have a pH of about 11. When using alkaline cleaners, the surface should be neutralized by an acid prior to the water rinse since alkali salts adhere strongly to surfaces. Clean oxide surfaces strongly adsorb hydrocarbons which detergents and solvents normally will not completely remove. These hydrocarbons must be removed by alkaline cleaners or oxidants. 4.6 Detergent Cleaning In detergent cleaning, the detergent (soap) surrounds the particle taking it into suspension without actually dissolving the material. This action is helped by wetting agents which loosen the particles. Many detergents contain phosphates. Liquid dishwasher soap is an excellent detergent for many applications (also laboratory green soap). Alconox™ is also a widely used laboratory cleaning solution though it is somewhat difficult to remove

Surface Preparation

123

from the surface and changes the surface pH. Ajax™ cleaner contains abrasives and care must be taken to eliminate large chunks which can scratch surfaces. A major problem with soaps is that metal ions such as the calcium and magnesium which are found in hard water (high content of ionic material) make the soaps insoluble and leave a residue. Therefore de-ionized water should be used for detergent cleaning. There is a tendency for people to use too much soap in a solution giving problems with rinsing and residues, particularly if the solution is used cold. About 1 tablespoon of detergent per gallon of water is generally sufficient. The author has been told that a slurry of carbon black (from burning acetylene) in de-ionized water mechanically abraded on a glass surface is very effective in removing absorbed organic contaminants—I have no first hand experience with this technique. 4.7 Chelating Agents Chelating agents keep the normally insoluble phosphates that are formed in hard water detergent cleaning in solution. Glass cleaning solutions often use chelating agents such as ethylene diamine tetra-acetic acid (EDTA). 4.8 Solvent Cleaning Hydrocarbon contaminants may be removed from surfaces by solvents which dissolve the contaminants. Solvents may vary greatly as to their ability to dissolve(solvate) contaminants, and their effectiveness needs to be known by determining the “solubility parameter” for specific contaminants (if contaminate is known).[27] Polar solvents such as water are used to dissolve polar contaminates (ionic material) while non-polar solvents such as the chlorinated hydrocarbon solvents, are used to remove non-polar contaminates (grease, rosin solder flux, etc.). Often a mixture of solvents is used to solvate both polar and nonpolar contaminates. Chlorinated hydrocarbon solvents are often preferred to hydrocarbon or petroleum based solvents because of their low flammability (flashpoint), though there is concern with the toxicity and carcinogenic properties of some of these materials. Chlorinated solvents may react with water to form acids. The acids react with metals causing corrosion. Often stabilizers are added to the chlorinated solvents to reduce their tendency to react with water (hydrolyze)

124


and form acids. Examples of such stabilizers in trichloroethylene (TCE) are: (i) 1,2 butylene oxide,(ii) cyclohexene oxide,(iii) para-tert-butyl phenol and(iv) 1-propanol. If stabilizers are not used, then the pH of the cleaner should be monitored to keep a pH of 6 - 7 (IPC Test Method No. 2.2.30; ASTM-D-2989 “Acidity/Alkalinity of Halogenated Organic Solvents”). If there is a possibility of solvent trapping which prevents complete rinsing, particularly in a stressed metal joints, chlorinated solvents should not be used since residues will enhance stress corrosion in those areas. Cleaners containing chlorine-based oxidants may present the same problem. Chlorinated halogen solvents are coming under increasing scrutiny as to their toxicity. Stringent exposure levels are being imposed by OSHA/EPA and it is anticipated that they will get even more stringent. Solvent properties to be considered include: 1. Suitability for application technique (spray, vapor degrease, recycling, etc.). 2. Selective solvency (solubility parameter)—ability to solvate the contaminants of interest. 3. Wetting characteristics—depends on viscosity and surface tension. Allows the solvent to wet surfaces and displace soils 17.2 to 21.4 dynes/cm3 for Freon™ solvents. 4. Miscibility with other solvents (to generate solvents for particular applications)—azeotropes = constant boiling point mixture of two or more components, i.e., composition of vapor is the same as the liquid. 5. Safety and environmental concerns—flammability, toxicity (breathing, contact) carcinogenicity, effect on the ozone layer, OSHA and EPA regulations present and future. 6. Stability—thermal and chemical, nonreactive with parts to be cleaned (chlorocarbon and alcohol solvents may react with Al, Mg, Be, Zn [white metals] to form inorganic salts which give residues etc). Photochemical stability. Solvents may leach materials from some container and piping materials. 7. Low energy requirements—low boiling points to give vapors without high energy requirements (vapor degreasers), parts may be handled immediately after cleaning. 8. High density—solvents displace soils and float them to the surface of the cleaning system (e.g., 9.6 to 13 lb/gal for Freon™ solvents)(ASTM-D- 2111 “Specific Gravity of Halogenated Organic Solvents and their Admixtures”).

Surface Preparation

125

Typical solvent systems are: 1. DuPont Freon™ TF (trichlorotrifluoroethane) 2. Azeotrope mixture of TF with methylene chloride (50%) = Freon™ TMC (metal degreasing) 3. TF with ethanol (4%) and nitromethane (1%) = Freon™ TES (rosin fluxes and ionic contaminates from solvent sensitive assemblies) 4. TF with ethanol (4%) = Freon™ TE (defluxing) 5. TF with acetone (11%) = Freon™ TA (broad range of solvency) 6. Blends of TF with methanol (6%) and nitromethane (0.25%) = Freon™ TMS (deflux) 7. TF with anhydrous isopropanol (35%) + stabilizer = Freon™ T-P 35 (cold cleaning ) 8. TF with ethanol (35%) = T-E 35 (organic and polar solvents). This data is taken from DuPont solvent formulation data bulletin no. FST5. Other equivalent solvents and solvent blends are available. Caution: Freon™ with water (or alcohol which takes up water) will corrode aluminum, zinc, and cadmium (white metals) if left in contact for a period of time; aluminum will take fluorine from the molecule. Aluminum parts should be dried immediately, preferably by vacuum bake, but at least hot-air-dried to minimize corrosion. There is also a safety concern: extended breathing of halogenated solvents can cause liver damage (like glue sniffing). These solvents must be used in a well ventilated area such as a chemical hood. Elevated temperatures are often used to increase detergent, solvation, and etching activities. This is often done using immersion heaters (materials must be compatible) or externally heated tanks. Abrasives may also be used in conjunction with solvents to loosen contaminants from the surface. Application methods of solvent and fluid type cleaning techniques include: (i) soaking,(ii) mechanical scrubbing,(iii) mechanical agitation,(iv) spraying (low and high pressure), (v) vapor condensation (vapor degreasing), (vi) hydrosonic agitation 2 Hz - 20 kHz),(vii) ultrasonic (20 - 60 kHz) agitation (cavitation) and(viii) megasonic agitation (850 - 900 kHz) (pressure wave). In mechanical scrubbing, lint-free, de-sized cloths make good toweling (sizing can be removed by multiple washings). For brushes, there is a variety of materials including: camel hair, mohair, polypropylene, Teflon™ and nylon. In semiconductor technology, mechanical scrubbing combined

126


with high pressure fluid jets (2000 - 3000 psi) is a standard cleaning procedures. Spraying may be performed at low pressure (50 psi) or at high pressure (1000 psi). Spray systems often use copious amounts of material so the liquid is usually recycled. This means that after the fluid becomes contaminated above a certain level it must be replaced. With increasing concern about solvent vapors, many of the newer systems are self-contained with condensers to trap the vapors and allow them to be recycled. Some systems allow the purification of the solvents by distillation. Vapor degreasers operate by putting a cold part in hot vapor above a vapor degreaser “sump”. The solvent condenses on the surface and flows off into the sump. Cleaning action only occurs during the condensation process, and when the part reaches a temperature where the solvent doesn’t condense, cleaning stops and the part should be removed. Parts should never be immersed in the sump fluid. Fluid in the sump should be changed when it becomes contaminated. Figure 3.2 shows a schematic of a typical old-style industrial degreaser for cleaning large parts either by spraying or by vapor degreasing. This type of system allows the escape of vapors and is becoming increasingly undesirable.

Figure 3.2. Industrial vapor degreaser with spray wand.

Surface Preparation

127

Ultrasonic cleaning[28] relies on the jetting action of collapsing cavitation bubbles to give a high pressure jet of fluid against a surface. The cavitation bubbles are formed by the tension wave portion of an ultrasonic wave in a fluid media. The ultrasonic wave is produced by a transducer typically operating at 20 - 40 kHz at about 100 watts/gal of fluid. The cavitation nature (size of bubbles) of the fluid depends on its vapor pressure and temperature (e.g., 3 microns for water at 60°C at 40 kHz). The jet pressure may be as high as 300 psi. The colder the media, the more energetic is the cavitation jetting. The bubbles nucleate in the fluid or on a surface. With a fixed frequency transducer, nodes and antinodes are formed (standing waves) which give variations of cavitation in the fluid. In order to overcome this effect, swept frequency generation is used with one system at 40 kHz ± 2 kHz. (Frequency modulation at full amplitude is best for sweeping frequency). If frequency sweeping is not used, the parts should be moved from one region to another in the tank. Variables in ultrasonic cleaning include: Nature of the transducer fluid (density, vapor pressure) Temperature of fluid Gas content of the fluid (function of degassing of fluid and entrainment with parts) Energy of cavitation implosion (temperature, pulse height of ultrasonic wave) Average cavitation density (volume or surface) with time Average cavitation density with position in tank Shape of the ultrasonic pulse Nature of ultrasonic cycle train (“quiet time”, “degas time”, cycles per train) Ultrasonic cleaning has to be used with care since the jetting action caused by the collapsing gas bubbles on the surface can cause erosion and introduce fractures in the surface of brittle materials, leading to poor adhesion. For example: in high power laser applications it has been shown that improper ultrasonic cleaning increases the light scattering from the surface, indicating surface damage or possibly surface roughening. Also ultrasonic agitation has been shown to create particles by erosion of the container surface, with stainless steel giving 500 times as many particles as Pyrex™ glass. In all cases studied, particles of the container were produced. Resonance effects may also damage some parts in an ultrasonic

128


cleaner.[29] Surface damage can be controlled by adjusting the energy density of the jets or controlling the time of application. The ultrasonic cavitation may be generated by magnetostrictive or electrostrictive transducers. The power may be from 500 watts for a small model (5 gallon) on up to very high powers. Ultrasonic erosion of aluminum foil (or an aluminum metallized glass surface) may be used as an indication of the cavitation power to which a surface is exposed in the ultrasonic solution.[30] A general rule is that ultrasonic cavitation will generate 10 holes in a 1 X 2 inch aluminum foil of 2 mils thickness in 10 sec. The cavitation ability is dependent on how well the energy is coupled to the fluid. Fixturing is very important in ultrasonic cleaning to insure that all surfaces are cleaned. Parts should be held parallel to the stress wave propagation direction. Energy absorbing containers, such as polyethylene or TeflonTM beakers and fixtures, should not be used since they absorb the ultrasonic energy. Hydrosonic cleaning utilizes hydrodynamic rather than electric generation of the fluid pressure waves.[31][32] The megasonic agitation system is applicable to smooth surfaces, particularly for removing particles, but doesn’t work on configured surfaces since the pressure wave is easily shadowed. 4.9 Oxidation Cleaning Oxidation cleaning relies on the formation of volatile or soluble oxidation products. If non-soluble products result from oxidation (e.g., silicone to silica) then a residue may be left on the surface. Oxidation cleaning may be used for surfaces that are normally oxides (glass, ceramics, metals that form coherent oxides) or that don’t oxidize (gold). High temperature oxygen or air fire is an excellent way to clean surface that can withstand high temperatures. For instance, to clean Al2O3, air fire the material to 1000°C then remove it while still warm (to prevent moisture condensation) and place in container. In thermal oxidation, the type of contaminate may be determined by monitoring the selective oxidation products as a function of temperature. Oxygen (or air) plasmas are very effective in removing hydrocarbons and absorbed water vapor from surfaces.[33] However the oxygen plasma may oxidize materials, which may be undesirable. Where oxidation is a problem, hydrogen plasmas may be used to remove hydrocarbons and adsorbed water from surfaces.

Surface Preparation

129

The use of oxidation by ultraviolet radiation, which generates ozone and causes bond scission of the hydrocarbon contaminates (UV/O3 cleaning) has greatly simplified the production, storage and maintenance of hydrocarbonfree surfaces.[11][12] UV/O3 exposure also allows the controlled thin-layer oxidation of surfaces such as silicon and silicon-germanium alloys. In a typical UV/O3 cleaning/storage chamber, the UV is provided by a mercury vapor lamp in a quartz envelope so that both the 1849 Å and the 2537 Å radiation is transmitted. The radiation intensity is 1 - 10 milliwatts/cm2 at the substrate surface. The chamber is of aluminum with no organic seals, and in a correctly operating system, ozone can be smelled when the chamber is opened (10 ppm ozone). The temperature in the chamber is typically 150°F during the cleaning operation. A heater may be used to decrease the possibility of moisture condensation when the chamber is open. Typical exposure times for cleaning are from a few minutes to remove a few monolayers of hydrocarbon contamination to hours, days, or weeks for storage of cleaned surfaces. The UV/O3 cleaning technique is also useful for cleaning holes (vias) in surfaces.[34] Caution: when there are corrosive agents (or materials that can decompose into corrosive agents, e.g., Freon™) in the atmosphere, we have found that the UV/O3 greatly enhances the corrosion rate. For instance, a little chlorine in the atmosphere causes stainless steel to rapidly corrode. Hot (115°F) concentrated sulfuric acid plus ammonium persulfate is an excellent oxidizing cleaner. The addition of the ammonium persulfate (solid) to the hot sulfuric forms an unstable compound that decomposes releasing ozone. The ammonium persulfate should be added just prior to the immersion of the substrate into the solution. This treatment is sometimes followed by a brief dip in a 10:1 solution of water and HF or immersion for 20 minutes in a solution of hydrogen peroxide and ammonium hydroxide. H2 O : H2O2 (30%) : NH4 OH (29%) at 80°C A hot chromic-sulfuric acid cleaning solution prepared from potassium dichromate and sulfuric acid provides free oxygen for cleaning but has a tendency to leave residues unless rinsed very well. K2 Cr2O7 + 4H2SO4 → K2SO4 + Cr2(SO4)3 + H2O + 3O Boiling hydrogen peroxide (30%) is a good oxidizing solution. Unstabilized H2 O2 must be used, and it should be stored in a refrigerator to slow

130


decomposition. Hydrogen peroxide is sometimes used with ammonia with a ratio of 8 (H2O2) : 1 (NH3) : 1 (H2O). Caution: 30% H2O2 is extremely reactive so it must not contact oxidizable materials such as organics. Oxidation cleaning may be performed using chlorine-containing chemicals. For example, a slurry of sodium dichloroisocyanurate (pool chlorine— 63% available chlorine) in water may be used to scrub an oxide surface to remove hydrocarbon contamination. 4.10 Volatilization Cleaning Heating volatilizes some surface contaminates such as water. This technique can often give problems because it may pyrolyze hydrocarbons into carbonaceous forms which are then very difficult to dissolve. The temperature may also cause changes in the surface composition and morphology. The surface composition may change due to volatilization of a constituent or by segregation of a bulk constituent to the surface. Thermally driven surface segregation can be greatly influenced by the nature of the environment (vacuum or reactive gas). Ga from GaAs surfaces may be thermally etched to give improved electronic properties at the resulting film-substrate interface.[35] In the case of some glasses, high temperatures tend to cause particles of oxidized glass constituents to form on the surface. Thermal treatment of silicon to >700°C removes the oxide but the surface begins to vaporize and form surface features.[36] Thermal cleaning is used to clean porous surfaces by increasing the surface diffusion of the contaminate from the subsurface regions to the surface where it can be removed. 4.11 Hydrogen Reduction Cleaning Hydrogen reduction of oxide layers may be used to clean surfaces in a furnace environment. Figure 3.3 shows the stability of a number of metal oxides at various temperatures and varying dew points of the hydrogen. Note that, depending on the dew point and the temperature, a hydrogen furnace can be either reducing or oxidizing! In some cases forming gas (90% N2 , 10% H2) is used instead of hydrogen since it is less explosive. Hydrogen reduction has been used to clean the oxide from silicon surfaces at 900°C.

Surface Preparation

131

Figure 3.3. Metal-metal oxide equilibria diagram for hydrogen plus water as a function of temperature.

4.12 Electrolytic Cleaning Electroetching may be used to anodically remove metal from a surface (along with contamination) and usually roughens the surface. The higher the current density, the more roughening occurs. For stainless steel, the surface is passivated by oxides (hydrated on the surface) at low potentials, while at higher potentials, the surface is etched.[37] Carbon fibers often have a weak surface layer and this layer may be removed by anodically electroetching (oxidizing) the surface followed by hydrogen firing. This treatment increases the strength of the carbon fiber and improves the bond when the fiber is used as part of a composite material.

132


Electropolishing removes material and smooths the surface.[38][39] The smoothing action is due to protection of the flat areas by a deposited material (usually a phosphate) and the preferential erosion of the peaks. Electropolishing leaves a surface film (phosphate) which has to be removed (hot water scrub) to obtain a clean surface.

5.0 DRYING AND OUTGASSING After fluid cleaning and rinsing it is important to dry the surface quickly in order to prevent the liquid film from collecting particles. Drying may be done by blowing the surface with filtered gas (from a liquid nitrogen tank) or by displacing the water by a high vapor pressure solvent such as anhydrous alcohol which dries rapidly. The best technique is an “alcohol vapor dry” where the cold surface is immersed in the vapor above a heated anhydrous alcohol sump. The cold surface condenses the alcohol vapor which flows off into the sump taking water and particulates with it. When the surface becomes hot condensation ceases and the hot part, when withdrawn, will rapidly dry. Spin drying tends to leave liquid along the outside edges of the substrate which may result in contamination of this area. If spin drying is used the part should be flooded with copious amounts of ultrapure water during spinning. Anhydrous alcohol, which displaces water and dries quickly, is one of the best materials with which to wipe and flush surfaces—it leaves the least residue; however it is not a very good solvent. Alcohol should only be used with polyethylene gloves. Isopropyl alcohol (IPA) is most commonly used since it requires no denaturant. Ethyl alcohol is generally more pure but requires the use of denaturants. Alcohol is denatured to avoid tax and accountability. Denaturants range from ethyl ether to kerosene (over 200 denaturants allowed). Low residue denaturants include methanol (5% by vol.) and acetone (10% by vol.). It is best to use pure (undenatured) alcohol if possible. Anhydrous alcohols can take up water from the atmosphere and lose their ability to displace water in the drying operation Drying and outgassing is especially important for polymers and porous materials which absorb solvents and water. It is often easier to dry and outgas prior to placing the materials in a deposition chamber. The usual technique is to heat the material (to some temperature that doesn’t degrade it) in a vacuum (vacuum bake) or desiccated environment. A common mistake is to vacuum bake the material for an insufficient time—often many

Surface Preparation

133

hours are necessary. The time-temperature-vacuum conditions necessary to outgas the material can be determined by weight loss measurements. Microwave energy may be used to heat polar molecules such as water (also alcohols, aldehydes, ketones, amides, amines, nitrate, cyanides, proteins, unsymmetrical halogenated hydrocarbons, and ionic solutions) as long as there are no electrical conductors present. Microwave heating and drying of such materials may be more effective than conventional thermal heating.[40]

6.0 MONITORING OF CLEANING The best monitoring techniques monitor those elements of the process which are critical to providing a surface that can be further processed. The testing of surface preparation such as cleaning will invariably result in contamination of the surface, so tested surfaces can not be used for subsequent processing. In some cases, sample surfaces may be tested for certain properties in order to determine surface conditions. These tests include(i) contact angle of a water drop (wetting angle),(ii) sheeting behavior of a fluid draining over a surface, (iii) nucleation of moisture on a surface and (iv) friction and adhesion tests. A common check on the cleaning of a glass surface uses the contact angle of a water drop on the surface of the cleaned glass. If the surface has no hydrophobic contamination (oil, hydrocarbons, silicones, etc.) the water will wet and spread over the surface giving a contact angle of <5° as measured with a contact angle goniometer. This technique must be used with some care since, if a hydrophilic contaminant such as a soap residue is present, the contact angle will be low even though the surface is contaminated. If a glass surface is clean, water will sheet over the surface without breaking up to avoid areas of contamination (water break test). Observation of this sheeting during the rinsing operation is a check that an experienced operator can use in the cleaning process. If the water film breaks up, then the surface is not clean. This breaking up into “legs” is how a wine taster judges the viscosity of a wine (i.e., they need a dirty glass). If you breathe on a clean glass surface, the moisture will condense uniformly over the surface giving the “black breath figure”.[41] You can see this effect in your bathroom where condensing moisture shows up the dirt swipes on the mirror. A clean glass surface has a high coefficient of friction (“squeakyclean”).

134


7.0 IN SITU CLEANING In situ cleaning is done in the deposition system and is intended to remove the small amount of contamination that has developed since the primary cleaning process. Some of the cleaning processes that have already been described are applicable to in situ processing but others are more desirable. In situ cleaning in an electrolytic environment may be done by etching or “offplating” by making the surface an anode of an electrolytic cell. In situ cleaning techniques for vacuum or plasma processing include: ! Oxygen plasma cleaning ! Hydrogen plasma cleaning ! UV/O3 cleaning ! Volatilization ! Ion scrubbing ! Sputter cleaning ! Reactive plasma etching (RPE) ! Reactive ion etching (RIE) 7.1 Ion Scrubbing Ion scrubbing of a surface occurs when the surface is in contact with a plasma and the plasma sheath potential accelerates low energy ions to the surface with sufficient energy to desorb absorbed gases. The technique is often supplemented with a reactive gas to give a version of reactive plasma cleaning. The technique is widely used in the optical coating business for substrate preparation in the vacuum deposition system (in situ cleaning) using an air discharge.[41]

8.0 PLASMAS Plasmas are gaseous media which contain enough ions and electrons to be electrically conductive and generally volumetrically neutral. Energy is introduced into the plasma by the acceleration of electrons in a DC, RF or microwave field. These electrons then fragment, excite, and ionize particles by collisions. A processing plasma is one that is used in processing a material. In a processing plasma, the volume density of the various gaseous

Surface Preparation

135

species and their energies depend on a number of factors including: method of generating the plasma, processing parameters, and system geometry. In a processing system, the local plasma densities and properties may vary significantly due to electrode configurations, presence of surfaces, and other geometrical factors. In cleaning processes, the substrate may be in plasma generation chamber or may be exposed to the plasma in a “downstream” location(remote plasma processing). In the plasma chamber the substrate may be on a driven electrode or placed where it is only exposed to the plasma, hence acts like a wall, though it may be exposed to a variety of plasma-related effects such as induced bias, electron bombardment, and energetic neutral bombardment. The downstream location avoids many of these plasma-related effects. Plasma discharges may also be used as a source of ion beams where ions are extracted from the plasma chamber and accelerated to a high energy by using a grid extraction system. Beam intensities are limited by the extraction grid. 8.1 Generation of Plasmas In plasmas used for plasma processing, the electron energy is increased by acceleration in electric field gradients. The most typical configurations for generation of plasmas are: (i) DC diode discharge, (ii) RF (radio frequency) discharge,(iii) electron emitter sustained discharge,(iv) magnetron enhanced discharge,(v) microwave discharge,(vi) vacuum arcs, and (vii) plasma arcs. Figure 3.4 shows a schematic of some of these plasma generation configurations. DC Diode Discharge. The DC diode configuration consists of an anode and a cathode immersed in a low pressure gas. At the cathode, the cathode potential (-) attracts ions from the edge of the plasma region, and they are accelerated across the cathode fall region to impinge on the cathode (target). The cathode fall region, which surrounds the cathode, is where most of the potential drop in a DC discharge is to be found. The region between the edge of the cathode fall region and the anode is the plasma region where there is little potential drop. In the DC discharge, energetic particles (ions and neutrals) impinging on the cathode (target) cause the ejection of secondary electrons which are then accelerated across the cathode fall region and create ions which sustain the discharge process. The secondary electron emission coefficient of a surface depends on the chemical nature and morphology of the surface. Oxides typically have higher electron emission coefficients than do metals. The secondary

136


Figure 3.4. Plasma generation configurations: (a) DC diode, (b) DC diode with permanent magnets giving a planar magnetron (c) RF plasmas with planar electrodes immersed in the plasma, electrodes external to a dielectric wall and a coil immersed in the plasma, (e) electron emitter (thermoelectron) with magnetic confinement and (e) microwave cavity.

electrons can be accelerated to high energies and impinge on the anode or other surfaces in the system. This can give rise to extensive heating of surfaces (substrates) in the system. The DC discharge requires a relatively high gas pressure (> 10 microns argon). In the cathode fall region, some of the ions may be neutralized by charge exchange processes which give rise to energetic neutral particles which are not affected by the applied electric field. The result is fluxes of

Surface Preparation

137

energetic ions and neutrals with a spectrum of energies that bombard the cathode. In order to sustain a discharge, the secondary electrons must create enough ions to compensate for losses. If the anode or ground surface is brought too close to the cathode, the discharge is extinguished. This effect can be used to confine the DC discharge to areas of the cathode surface where bombardment is desired—other areas may have the bombardment prevented by having a ground shield in closed proximity to the surface. The Paschen curve gives the relationship between breakdown voltage and the minimum anode-cathode separation in a gaseous environment. Insulator surfaces cannot be used as cathodes in a DC diode configuration since charge buildup on the surface will prevent ion bombardment. In addition to causing the ejection of secondary electrons, high energy ions and neutrals which impinge on the target (or other surfaces) cause the physical ejection of surface atoms (physical sputtering) by momentum transfer processes. The sputtered particles leave the surface at higher-thanthermal energies but may be rapidly thermalized by collisions in the gas phase. The sputtered particles may be scattered back to the target surface; this effect is more prominent the higher the gas pressure. Some of the energetic ions that bombard the cathode may be reflected as high energy neutrals. The electrical current measured in the DC cathode circuit is the sum of the charge due to the ion flux to the target and the secondary electron flux away from the surface. Therefore, the cathode current density and cathode voltage do not specify the flux and energy of the impinging ions. However these measurements (along with gas pressure and gas flow) are typically used to specify the plasma parameters in DC diode plasma processing. Typically a DC diode discharge plasma is “weakly ionized” with many more neutral particles than ions (104-107 : 1). It will also have a low electron temperature and an even lower ion temperature. If molecular gas species are present in the discharge, many radical species will be formed in the plasma and they will generally greatly outnumber the ions. Any surface in contact with the plasma will be subjected to a flux of ions, neutrals and electrons. A sheath potential will be developed because of the greater mobility and energy of the electrons as compared to the ions. This wall potential (typically 3 - 10 volts) will accelerate ions from the plasmas, giving rise to ion scrubbing of the surface. In plasma processing, the DC diode configuration has many advantages: (i) a rather uniform plasma can be generated over large areas; (ii) power

138


input (watts/cm2) can be very high;(iii) the power supplies are rather simple, inexpensive and powerful; (iv) process reproducibility can be attained by controlling the geometry, gas pressure, and target power (current and voltage); and(v) sputtering of surfaces may be used as a source of depositing material. It also has some disadvantages: (i) surface geometries can result in focusing effects giving non-uniform bombardment; (ii) electron heating of surfaces can be extensive; and(iii) insulating surfaces cannot be bombarded. RF Discharge: At high frequencies in a capacitively-coupled discharge, the electrons oscillate in the changing field thus gaining energy, and by collision with atoms, create ions and more electrons. Typical RF power supplies operate at 13.56 MHz (USA industrial frequency) with peak-to-peak voltages of greater than 1000 volts. The plasma acts as a low density electrical conductor and the RF field penetrates quite some distance into the plasma. When the driven RF electrode is a conductor, the surface is bombarded by ions from the plasma during the half-cycle that the electrode is negative. If the surface of the RF electrode is an insulator (backed by a conductor), the metal-insulator-plasma acts as a capacitor and the surface potential that appears on the insulator surface alternates between a low positive potential (because the electrons have a high mobility) and a higher negative potential (because the ions have a relatively low mobility). Ions are extracted from the RF plasma during the negative portion of the cycle and bombard the insulator surface. The RF potentials in the plasma can be determined using capacitive probes. The ion energies bombarding a surface may be determined using a sampling orifice, a retarding grid and a mass spectrometer. In capacitively-coupled RF discharges, the plasma potential, hence the sheath potential at the electrodes, can have a time-varying value of tens to hundreds of volts. When the electrodes have different effective areas, the plasma potential can also have a large DC potential with respect to one or more of the electrodes. These factors affect the distribution of ion energies incident on the electrode surfaces in an RF discharge. Small area electrodes will attain higher voltages than large area electrodes and the electrode potentials can be varied using external capacitance in the circuit. The amount of energy that is coupled into the RF discharge depends on the impedance matching (reflected power) and coupling losses to other surfaces (stray losses). In RF plasma processing, it may be important to determine just how much energy is actually being coupled into the plasma.[42] RF power may be coupled to the plasma using metal electrodes

Surface Preparation

139

external to a dielectric wall, or the RF plasma may be excited using immersed electrodes. Very high plasma densities and ionization can be attained in RF driven plasmas and the discharge may be established at lower pressures than the DC diode discharge. Surfaces immersed in an RF plasma will assume a self-bias. This bias depends strongly on the surface areas and configurations. Often the value of the self-bias is not known (or controlled) and can vary within the system, giving anisotropic bombardment effects. In plasma cleaning, the RF discharge has the advantage that insulating surfaces or insulating films can be bombarded by applying an RF potential. Disadvantages are: (i) high power inputs (heating) to insulating materials cause cracking; (ii) electrode geometries can cause problems with coupling to the RF power;(iii) there are many sources of RF power loss in systems;(iv) plasma uniformity is difficult to obtain over complex surfaces; and (v) the bias conditions on surfaces in the RF plasma are variable and often difficult to control. Microwave Discharges. Plasmas can also be excited at much higher frequencies, 300 MHz to 10 GHz, where electron cyclotron resonance coupling gives more efficient ionization.[43] Ionization can be as high as 20% in a such a microwave discharge. Microwave plasmas are most often used in the downstream processing configuration since substrates in the microwave cavity can “detune” the system. Electron Emitter Discharge. In the DC diode and RF plasma configurations, the electrons necessary to sustain the plasma are produced in the plasma. When using electron emitters, the electron source is independent of the plasma processes. Common electron emitters are hot thermoelectron emitting cathodic surfaces and hollow cathodes. For example: LaB6 surfaces can give an electron emission of >20 A/cm 2 at 1700°C. [44] Often the electrons are confined by a magnetic field (50 - 500 gauss) directed along the anode-cathode axis. The magnetic field increases the electron path length in its movement from the cathode to the anode by causing the electron to spiral in the magnetic field. This increases the ionization efficiency of the electron. The ions in the plasma may be extracted using an electrode at a DC or RF potential to give bombardment of a surface (triode configuration). The triode configuration suffers from a nonuniform plasma density along its axis which gives nonuniform bombardment and a density variation in activated species over a large biased surface. The thermoelectron emitter system is very amenable to forming dense plasmas and for application to downstream processing. By applying highmagnetic

140


fields, the plasma (ions and electrons) may be confined and steered into a processing chamber. In plasma processing, some of the advantages of the electron emitter configurations are:(i) the flux of electrons is independent of other plasma and electrode processes; (ii) very high plasma densities can be attained; iii) the plasma properties can be controlled by controlling the electron emission; and (iv) the electron beam can be used as a source for thermally vaporizing material. Disadvantages are:(i) need for well controlled and long life electron emitting sources, and(ii) plasma non-uniformity over large areas and complex surfaces. Low strength (50 - 500 gauss) magnetic fields may be used to confine the electrons and increase their path length in any plasma system. There are a number of ways to establish magnetic fields in plasma chambers including: (i) internal permanent magnets,(ii) external permanent magnets, (iii) external electromagnets, and (iv) moving magnets. Permanent magnets have the advantage that they may be placed in such a way as to position the field lines in a desirable manner; however, getting a uniform magnetic field over a large or complex surface is difficult. Magnetron enhanced plasma configurations have many advantages including: (i) confining the plasma to a small region, (ii) increased ionization and plasma density, (iii) may be operated at low pressures where gas phase collisions are reduced. Disadvantages include:(i) non-uniform magnetic fields give non-uniform plasma generation; (ii) isolation of the plasma to a small region of the processing chamber requiring auxiliary plasma sources in some applications; and (iii) low pressure processing can give rise to a flux of high energy reflected neutral which may affect the processing in an undesirable manner. 8.2 Plasma Chemistry Plasma is a very energetic environment and many chemical processes can occur.[45]-[47] The principal chemical processes are: (i) electron impact ionization, (ii) dissociation (fragmentation) of molecules (formation of radicals), (iii) Penning ionization (metastable collision),(iv) dissociative electron attachment, (v) electron attachment, (vi) excitation,(vii) momentum transfer collisions, (viii) de-excitation of excited species, and (ix) recombination (neutralization). As an example of the complexity of plasma chemical processes, consider that there can be 24 reactions and 16 species formed by the decomposition and reaction of CF4 in a plasma.[48]

Surface Preparation

141

As an example of Penning ionization, consider argon which has metastable excited states of 11.55 and 11.75 eV and copper which has an ionization energy of 7.86 eV. Thus a copper atom colliding with a metastable argon atom is easily ionized. Metastable atoms may be very effective in ionizing other species by collision. Many of these processes are characterized by “cross sections” for collision processes and threshold energies for attachment processes. For example, CF3Cl has a high cross section and low threshold energy (2 - 3 eV) for electron dissociative attachment. CF4 has a low cross section and high threshold energy (5 - 6 eV) for electron dissociative attachment and CCl4 is not activated at all by electron attachment. Therefore CF3Cl is much more easily fragmented and ionized in a plasma than is CF4 or CCl4 . The degree of ionization, dissociation and excitation of the species depends strongly on the gaseous species, electron energy, and density in the plasma. Generally there is much more dissociation than there is ionization of molecular species. Many of these plasma processes serve to activate gas species, i.e., to make them more chemically active by dissociation, ionization, or excitation (plasma activation). Plasma discharges are very effective in desorbing contaminates (e.g., H2 O) from surfaces in a plasma processing chamber. These impurities are activated in the plasma and may contaminate the depositing material. A number of techniques may be used to determine plasma properties.[49] Optical emission is the most common. Actinometry compares the emission interactions of the excited states of reference and subject species to obtain the relative concentrations of the ground states. Optical absorption techniques may also be used to characterize the gaseous species and temperature in a gas discharge. Electron and ion densities in a plasma may be measured by the use of small area Langmuir probes. 8.3 Bombardment Effects on Surfaces The physical effects of energetic particle bombardment on surfaces and depositing films is very dependent on the mass, flux, and energy of the bombarding particles, the flux of non-energetic particles (i.e., depositing or absorbing species), and the atomic mass and chemical nature of the bombarded surface. In many cases the fluxes of impinging particles are not determined or controlled except by the processing parameters. Figure 3.5 depicts the effects of bombardment by energetic species (not electrons) on the surface and the subsurface region. Surface effects

142


include: (i) desorption of weakly bonded surface species, (ii) ejection of secondary electrons, (iii) reflection of the energetic species as high energy neutrals, (iv) sputter ejection (physical sputtering) of surface atoms by momentum transfer through collision cascades, (v) sputtering and redeposition of sputtered species by collisions in the gas phase, ionization and acceleration back to the surface and byforward sputter deposition due to the ejection angle on a rough surface, (vi) enhanced surface mobilities of atoms on the surface, and (vii) enhanced chemical reaction of impinging and adsorbed species to produce condensed species (“reactive deposition”) or volatile species (etching). In the subsurface region:(i) the impinging particles may be physically implanted (ii) the collision cascades cause displacement of lattice atoms and the creation of lattice defects, (iii) surface species may be recoil-implanted into the subsurface lattice, (iv) mobile species may be trapped at lattice defects, and (v) much of the particle kinetic energy is converted into heat. Lattice channeling processes can carry these effects deeply into the surface.

Figure 3.5. Schematic depiction of the energetic particle bombardment effects on surfaces and growing films.

The desorption of weakly bound surface species is important to plasma cleaning and may be used to reduce the incorporated contaminants in deposited films. The desorption may also be useful in desorbing unreacted species in reactive deposition processes giving rise to more stoichiometric and chemically stable deposits.

Surface Preparation

143

The physical sputtering of a surface may lead to surface texturing to give a roughened surface (e.g., Ref. 50, 51). Preferential crystallographic sputtering will result in some crystalline orientations being etched at faster rate than others (sputter etching). Preferential atomic sputtering can cause changes in the chemical composition of alloy and compound surfaces.[52][53] If a reactive species is used for bombardment, the surface may be etched if the resulting chemical species is volatile (reactive ion etching, chemical sputtering), or the surface may be converted to a compound if the chemical species is not volatile. Most of the bombarding energy goes into heating the bombarded surface.[54] The incorporation of bombarding species into the surface gives rise togas charging which increases the chemical potential between this region and the interior and thus the diffusion of the gas into the material. In hydrogenbombardment cleaning of a hydrogen-sensitive metal, the hydrogen must be desorbed while the surface is hot. 8.4 Sputter Cleaning and Etching Sputter cleaning uses physical sputtering to remove some of the surface layer which includes contaminates. Sputter cleaning has been called the universal etch since everything can be removed by the sputtering process at approximately the same rate. Sputtering from a plasma environment has disadvantages: ! Contaminates in the plasma become activated and can react with the surface being cleaned. ! Sputtered species can be returned to the surface by scattering (redeposition). ! Surface species can be recoil implanted into the surface. ! Sputtering may develop undesirable surface features. ! High voltages are used in the process. ! Bombardment from the plasma may electronically damage semiconductor materials. ! Special equipment and fixturing may be required. Low energy ion bombardment can be used to clean surfaces without electronic damage.[3][4][55][56] The low energy ion bombardment can be obtained from high pressure plasmas, downstream processing with low biases, and with low energy ion beams.

144


Ion beam sputter cleaning may be done in a vacuum environment where the sputtered species are not redeposited on the substrate surface. Reactive plasma cleaning/etching (RPE) uses a reactive species in the plasma which reacts with the surface to form a volatile species which will leave the surface (no concurrent ion bombardment of the surface).[57] Plasmas containing reactive species are used in plasma etching (dry etching) and “reactive plasma cleaning”.[58] Fluorine (from CF4, CHF3, C2F6 , C3F8 and SF6) and chlorine (from Cl2 , CCl4 and BCl3 ) are the most widely used reactive gases. Oxygen is often added to the fluorine system to promote the formation of atomic fluorine and thus increase the etch rate. One of the most common gas mixtures is 96% CF4 with 4% O2. Helium is often added as a diluent and to increase the thermal conductivity of the plasma hence reducing the temperature rise of the surface. The reactive plasma technique is typically specific and may be selectively used to take the oxide from the substrate materials and then etch the substrate material at a low rate. Numerous gases and gas mixtures are available for RPE.[59]-[61] Examples of plasma etching (cleaning) of aluminum with various gases are: Al etched with Cl2, BCl3, CCl4 and SiCl4.[62] The BCl3 removes the oxides, others don’t do very well on oxides. (BCl3 is a good scavenger of H2O and O2 in the plasma system; it produces condensible material—B2O3). If Cu is present in the aluminum there will be a CuCl2 residue which may be volatilized by heating above 200°C. Most metals are most easily cleaned using fluorine gas (because the products are more volatile) rather than chlorine. Caution: etching and cleaning with compound gases should be done with caution since the decomposition products (B,C,Si) may react with or deposit on the surface, thereby changing the chemical composition or contaminating the surface. When using a carbon containing chemical (e.g., CCl4, CF3) in the plasma, a residual carbon contaminate remains—using of HCl or SF6 avoids this problem. Exposure to reactive plasmas may leave a reacted/chemisorbed layer of halogen species. This layer may be very important to the sensitization of the surface to atomic nucleation, or the wettability of organic species to a surface. For instance, the NH4 plasma treatment of Ti gives good adhesion when coated with an amine epoxy. Reactive plasma etching of silicon in CCl4 plasmas has been reported to give a very thin fluoride layer that passivates semiconductor surfaces to oxidation. Oxygen plasmas can be used to reactively remove materials that have volatile oxidation products (e.g., C, hydrocarbons). Hydrogen plasmas can be used to remove materials with volatile hydrides (e.g., C, Si, hydrocarbons).

Surface Preparation

145

In reactive ion etching (RIE), ion bombardment of the surface is used to add energy and secondary electrons to the depositing/etching surface environment. It has been shown with RIE of silicon that carbon residue limits the rate of etching; when etching oxides, the oxygen prevents the formation of the carbon layer and higher etch rates result.[63] In RIE of silicon, the residue that remains on the surface must be removed by a postdeposition treatment of low temperature oxygen annealing.[64] Carbon residuals, when using carbon-containing etchant gases, have also been found in the reactive plasma cleaning of metals where the problem was avoided by using HCl as the etching gas.[58] Typically RIE introduces less surface damage in semiconductor materials than does sputter etching[65] but more than does RPE. The use of ion bombardment with a molecular beam of the etchant gas in vacuum (bombardment enhanced chemical etching) allows reactive cleaning to be used in a vacuum environment.[66] It has been shown that bombardment does increase chemical reactivity at a surface although the mechanism is not well defined.[67] For instance the role of absorbed reactive species, which are subjected to the bombardment, has not been determined and this effect will be different in a plasma environment than in a vacuum environment. Secondary electron emission may play an important role in chemical reactions on a bombarded surface. Plasma etching is used for pattern delineation in semiconductor wafer fabrication, particularly VLSI fabrication. In RIE, electric fields direct ions normal to the surface and etch anisotropy can be obtained. This anisotropy can be used to etch steep-walled features but can be a problem on nonplanar surfaces where off-normal surfaces are etched slowly. A major concern in plasma etching is the etch selectivity which determines the ability to stop or significantly slow down the etching process when materials change (SiO2 on Si). Plasma etching was introduced into the semiconductor industry in the mid-70s in the form of batch reactors, namely barrel reactors (or volumeloading or tubular reactors)[68] and parallel-plate reactors (diode, Reinberg reactors, surface-loading reactors), and more recently, the “HEX” reactors.[69] These reactors hold a number of wafers and are available in a large number of configurations and plasma generation techniques, ranging from planar diodes to triodes to magnetrons. In barrel reactors the etching is due to the activated species, and electric fields (with their associated bias) are often eliminated by the addition of a conducting etch tunnel around the wafers which confines the plasma generation to the region between the

146


tunnel and the reactor walls. The etching in a barrel reactor with an etch tunnel depends on long-lived activated species and the etching is isotropic. To achieve high reaction rates, one needs high plasma densities and a large number of reactive species at reasonable power densities. These characteristics can be increased by increasing the electron-atom collision probability by: ! Short mean free paths (diode)—“high” pressures (<1 torr) ! Auxiliary electron source (triode)—low pressures (0.01- 0.2 torr) ! Increased path length (magnetron)—very low pressures(<0.01 torr)(Hill) ! Microwave plasma excitation Flow uniformity is extremely important to etch/cleaning uniformity. Gas is typically introduced through a series of orifices or in some cases porous diffusers (though these may clog up easily). Large orifices allow high pressure regions and electron trapping that give local high density plasma that affect uniformity. In high pressure reactors, the electrode spacing is small and plasma uniformity is difficult to obtain. Various techniques are used to hold the plasma over the driven electrode. The use of guard rings on the edge of the driven electrode is the most common technique. In low pressure plasma reactors, an auxiliary electron (or plasma) source is used to sustain the plasma and allow the decoupling of the plasma source and the driven electrode. This allows more process variation to be used. Electrode spacing is larger but the non-symmetry of the plasma may make plasma uniformity difficult to achieve, particularly when varying process parameters. The higher plasma densities may also increase the plasma sheath potential giving rise to increased radiation damage of the wafer, but it provides more directional ion bombardment therefore better directional etching. In very low pressure reactors, the use of the magnetic fields at high plasma densities results in a lowered plasma sheath potential so a biased electrode may be useful. The reactive etching/cleaning process gives volatile species which may be deposited in other parts of the system under different conditions and may have a detrimental effect on the gas handling/pumping system, and may be a source of particulates in the etching system.[70]

Surface Preparation

147

9.0 STORAGE AND HANDLING An integral, and often neglected aspect of surface preparation is that of handling and storage before the next processing step or usage. Handling and storage during processing and after cleaning is a major source of recontamination. It is not unusual to see someone put a carefully cleaned substrate into a plastic bag where it is contaminated by the polymer. Storage should be in a non-recontaminating (particles, vapors) environment. Non-contaminating environments may be passive or active. Passive environments are those such as carefully cleaned glass containers—possibly evacuated and desiccated. Active environments are ones where the contaminants are continually removed by adsorption or oxidation. Adsorption can be on freshly oxidized aluminum or activated carbon (particulate problem). Oxidation may be done using the UV/O3 cleaning cabinet described in Sec. 4.9. This is by far the best technique for storing surfaces where surface oxidation is not a problem. The materials used for storage are very important. Storage should generally not be done in a plastic container since recontamination can occur from vapors, physical contact, abrasive transfer or the diffusion of moisture through the polymer. If polymer bags are to be used, and moisturepermeation is a problem or potential problem, the best ones have a metal foil laminated between two layers of polymer. Most paper products are acidic and can be corrosive. Paper products may have also absorbed corrosive gases from the environment. Many polymer sheet products have antistatic coatings on the surface; these antistatic coatings are often hydrophilic electrolyte materials which may be corrosive. Polymers may have plasticizers in them that will volatilize and contaminate surfaces. Polyvinyl chloride (PVC) can breakdown and form hydrochloric acid. Most adhesives have corrosive components. Cloth has “sizing” on the surface of the fibers to aid in weaving and this may be transferred to parts wrapped or handled with the cloth. Aluminum foil generally has a layer of oil on the surface. Materials that may be suitable for storage of surfaces that are extremely sensitive to corrosion are: ! Cleaned glass containers ! Cleaned metal containers ! Cleaned aluminum foil ! Desized cloth (desize by multiple washing) ! Acid-free paper products—buffered or non-buffered

148

Deposition Technologies for Films and Coatings ! Polyester polymers (e.g., Mylar™) (no plasticizers, no antistatic coating), Polyester laminate (aluminum foil between plastic sheets - reduces moisture permeation) ! Polypropylene (no plasticizers, no antistatic coating) ! Polyethylene (no plasticizers, no antistatic coating) ! Methyl cellulose glue (neutral pH, high water content)

10.0 ACTIVATION AND SENSITIZATION Activation of a surface means making the surface more reactive without the addition of material to the surface. For example, activation in electroplating may be the removal of oxide layers by chemical or electrolytic treatments just prior to insertion into the electroplating bath. Such activation is used for plating: nickel-on-nickel, chrome-on-chrome, gold-on-nickel, silver-on-nickel, nickel-on-Kovar™. For example: the acid cleaning of nickel by immersion into an acid bath (20% by volume sulfuric acid) then transferring through the rinse into the deposition tank, keeping the part wet at all times and minimizing the transfer time. Methods for activating polymer surfaces include:[71] corona discharges (air), glow discharges (radiation, ion bombardment) (oxygen, fluorine, argon),[72] x-ray irradiation,[73] electron irradiation,[73] low energy electron bombardment, ion bombardment,[74][75] ultraviolet radiation, and mechanical abrasion Plasma activation of a polymer involves taking an existing member of a polymeric chain (atom, molecule etc.) and making it more reactive by breaking bonds and leaving dangling bonds which are capable of reacting with depositing species. The resulting chemical bonding contributes to the adhesion of the deposited film to the polymeric substrate. The number of active sites generated by the plasma treatment determines the nucleation density and the strength of the chemical bond contributes to the adhesion strength. The plasma treatment of polymers is sometimes called CASING (Crosslinking by Activated Species of Inert Gas).[76]-[78] Sometimes plasma activation is followed by exposure to ammonia before the surface is used for bonding. The following are some bond strengths between fluorine (the most electronegative element) and common depositing atomic species.

Surface Preparation

Bond

Dissociation energy

Ag-F Al-F Fe-F C-F Ni-F Ti-F Au-F

135.5 Kcal/mole 163.8 135.0 116.8 148.1 145.5 Unstable

149

The more electronegative the element, the higher the chemical bond strength one would expect. Active bonding sites may become saturated (“poisoned”) by reacting with molecules in the ambient atmosphere (oxygen, OH, etc.) so activation of a surface will degrade with time. The activation process may also permanently change the polymer surface by increasing crosslinking. Plasma treatment of polymer surfaces with inert gas species may give surface texturing, and the improved bond strengths are then attributed to mechanical interlocking. Mechanical activation of metal surfaces by mechanical brushing just prior to film deposition is a technique that gives improved adhesion of vacuum deposited coatings on steel.[79][80] Plasma deposited polymer films (plasma polymerization) have high concentrations of dangling bonds and are adherent to many surfaces and may act as good coupling layers (primer) for subsequent metal deposition if used before the unsaturated bonds become saturated. Activation of ionically bonded solids may be by exposure to radiation which creates point defects which may act as bonding sites. Electrons and photon radiation of insulator and semiconductor surfaces prior to film deposition have been used to enhance the adhesion of the film,[81] probably by changing the nucleation behavior. Sensitization of a surface means adding a small amount of material to the surface to act as nucleation sites for adatom nucleation. This may be less than a monolayer of material. In electroplating, the addition of nucleating agents (Sn - stannate, Zn - zincate) to the surface of difficult-to-plate metals such as Ti, Al, Zr, and U[82] allows the deposition of adherent metal coatings (example, zincate process for Al, ASTM -B-253). In electroplating, the sensitization process often involves simultaneous etching and displacement plating.

150


Various materials are used to sensitize polymer surfaces for bonding and electroplating (Teflon™—napthelenides or alkali metals, Nylon™— iodine). In plasma processing, polymer sensitization is the addition (chemisorption) of a surface species which acts as a coupling agent to react with the depositing atoms. One sensitization technique is plasma activation with the addition of a coupling agent (usually oxygen) to the active site; this coupling agent in turn reacts with other organics (amine epoxy treated with oxygen plasma then coated with urethane) or with depositing metal atoms. Oxygen plasma treatment of polymer surfaces has been shown to form an oxygen complex with the carbon in the plastic.[83]-[85] The deposition of copper allows the copper to react with the oxygen giving improved adhesion.[86]

11.0 SURFACE MODIFICATION Surface modification may involve changing the surface chemical composition by: (i) conversion to a chemical compound (nitride, carbide, oxide); (ii) changing the chemical composition by selective loss, e.g., hydrogen ion bombardment of carbide surfaces results in carbon depletion to a depth corresponding to the physical penetration of the hydrogen into the surface,[87] hydrogen firing of a carbide surface results in the decarburization of the surface;[88] (iii) the addition of a surface layer which is compatible with the substrate material and forms a new surface on which to deposit the film. In thin film metallization this layer is sometimes called the “glue” layer (e.g., Ti on oxides under Au, Ni on metals, Cr on polymers,[89]). Surface modification may be done by changing the physical properties of the surface such as roughness or hardness. Surface roughening may be used to give more mechanical bonding. Roughening may be accomplished by grit blasting, mechanical abrasion, chemical etching (grain boundaries, crystal orientation, phase), oxidation then reduction or etching, electrochemical etching, or sputter texturing[51] Examples of surface roughening are: ! AlN ceramics with NaOH[90] ! Al2O3 ceramics with molten NaOH[91] ! ABS copolymer: etch to remove one phase and give porous surface ! Chemical etching of Kovar™ by ferric chloride

Surface Preparation

151

! Sputter texturing of copper[92] ! Oxidation and reduction of a molybdenum surface to make it porous for gold deposition ! Sputter texturing of medical implants ! Sputter texturing of polymers ! Anisotropic chemical etching of silicon[93]

12.0 PASSIVATION AND PRESERVATION Clean surfaces (or those freshly prepared) may be passivated or protected by the addition of a layer of a material that is easily removed. Examples are: ! Au flash on metallization surface to prevent oxidation and make the surface solderable ! Strippable organic films on metal surfaces to prevent corrosion ! Strippable coating on optical surfaces to prevent particulate contamination ! Water-wetted surface after activation (oxide strip) in electroplating to prevent re-oxidation ! I on HF-cleaned Si to prevent oxidation—desorbed at 500°C in vacuum[94] Cleaned parts may also be stored under liquids to exclude reactive gaseous agents. Metals may be stored in anhydrous liquids such as alcohol or acetone until needed. Storage of material in deoxidized water (boiled) will decrease the oxidation of the surface compared to air or oxidated (cold) water. Nitridation of UHV cleaned surfaces minimizes recontamination and makes subsequent cleaning easier.[13] Coating of UHV surface with gold makes them less likely to adsorb contamination. The UV/O3 oxidation of GaAs [95][96] has been used to form a passivating (sacrificial) layer that can subsequently be vaporized as an in situ cleaning technique that leaves no residue and doesn’t damage the crystal surface. This has been reported to be better than thermal or air oxidation since thermal oxidation selectively oxidizes one constituent and when removed, leaves a poor surface composition.

152


Some silicon surfaces have been preserved by coating with a metal such as Ga,[97] In[98] or I[99] onto the clean surface and then evaporating the species from the surface as an in situ technique. 13.0 SAFETY Chemical manufacturers and distributors are required to provide “Material Safety Data Sheets” (MSDS’s) for hazardous materials with or before shipment of the materials. OSHA has mandated that employees must be given information and training as to the hazards of the materials that they are using (Hazard Communication Standard 29 CFR 1910.1200) It should be recognized that chemicals in combination can generate a safety hazard where the separate chemicals may not. Examples are: ! Organics with oxidants = fire, explosion ! Cyanide compounds with acids = poison (hydrogen cyanide) Various industry organizations have formulated guidelines for using of industrial chemicals safely. For instance the Institute for Interconnecting and Packaging Electronic Circuits has issued a guideline entitled “Guidelines for Chemical Handling Safety in Printed Board Manufacture” (IPC-CS-70).

REFERENCES 1. Musket, R. G., McLean, W., Colmenares, C. A., Makowiecki, D. M., and Siekhaus, W.J., Appl. of Surf. Sci., 10:143 (1982) 2. Mattox, D. M., Thin Solid Films, 53:81 (1978) 3. Vossen, J. L., Thomas, J. H., III, Maa, J-S., and O’Neill, J. J.,J. Vac. Sci. Technol., A2:212 (1984) 4. Vossen, J. L., Thin Solid Films, 126:213 (1985) 5. Jones, W. C., Met Finish, 83(10):13 (1985) 6. Bowling, R. A., J. Electrochem. Soc., 132:2208 (1985) 7. Henrich, V. E., Rep. Prog. Phys., 48:1481 (1985) 8. Cuthrell, R. E. and Tipping, D. W., Rev. Sci. Instrum., 47:595 (1976) 9. Cuthrell, R. E., Surface Contamination, (K. L., Mittal, ed.), 2:831, Plenum Press (1979) 10. Comsa, G., and David, R., Surf. Sci. Repts., 5:145 (1985)

Surface Preparation

153

11. Sowell, R. R., Cuthrell, R. E., Bland, R. D., and Mattox, D. M., J. Vac. Sci. Technol., 11:474 (1974) 12. Vig, J. R., J. Vac. Sci. Technol., A3:1027 (1985) 13. Grunze, M., Strasser, G., and Elshazly, O., J. Vac. Sci. Technol., A4:2396 (1986) 14. Holland, L., Vacuum, 26:97 (1976) 15. Bouwman, R., van Mechelen, J. B., and Holscher, A. A., J. Vac. Sci. Technol., 5:91 (1978) 16. Newhouse, R. D., Microelectronic Manuf. Test, 9:1 (1986) 17. Meeks, R. F.,Contamination Control Training Manual, General Electric report 74ND-3 (GEPP-121) available from NTIS 18. Malczewski, M. L., Borkman, J. D., and Vardian, G. T., Solid State Technol., 29(4):151 (1986) 19. Goldsmith, S. H. and Grundelman, G. P., Solid State Technol., 28:219 (1985) 20. Grant, R. D., Mat & Design, 9:22 (1988) 21. Balcar, G. and Woelfel, M., Met Finish, 83(12):13 (1985) 22. Bibliography on Chemical Cleaning of Metals, Vol. 1 (#52135), Vol. 2 (#52129), available from NACE (National Association of Corrosion Engineers), PO Box 218340, Houston, TX 77218 23. Cherepnin, N. V., Treatment of Materials for Use in High Vacuum, Ordentlich, Isreal (1976) 24. Sundahl, R. C., J. Vac. Sci. Technol., 9:181 (1972) 25. Manko, H. H., Solders and Soldering, Ch. 2, McGraw-Hill (1981) 26. Brazing Manual, Ch. 4, American Welding Society (1975) 27. Jackson, L. C., Adhesives Age, p. 23 (Dec.1974) 28. Physical Principles of Ultrasonic Cleaning, Vol. 1, (L. D. Rozenberg, ed.), Plenum Press (1973) 29. Cieslak, W. R., Proc. ASM Third Conf. on Electronic Packaging: Materials and Processes & Corrosion in Microelectronics, Minneapolis, MN (April 28-30, 1987) 30. Fredrick, J. R., Ultrasonic Engineering, Wiley (1965) 31. Walker, R., Treatise on Clean Surface Technology, (K. L., Mittal, ed.), 1(3) Plenum Press (1987) 32. Walker, R., TSF 119, 223, 84 33. Holland, L., The Properties of Glass Surfaces, Ch. 6, Wiley (1964)

154


34. Norstrom, H., Ostling, M., Buchta, R., and Petersson, C. S., J. Electrochem. Soc., 132:2285 (1985) 35. Saito, J., Ishikawa, T., Nakamura, T., Konagai, M., and Takahashi, K., Jpn. J. Appl. Phys., 25:1216 (1986) 36. Ishikawa, Y., Ikeda, N., Kenmochi, M., and Ichinokawa, T., Surf. Sci., 159:256 (1985) 37. Olefjord, I., Brox, B., and Jelvestam, J., J. Electrochem. Soc., 132:2854 (1985) 38. Faust, C. L., Metal Finish, 82:29 (1984) 39.

Shigolev, P. V., Electrolytic and Chemical Polishing of Metals, Freund Pub. (1974) from the Russian

40. Smith, F. J., R & D Mag, 30, 54 (1988) 41. Holland, L., The Properties of Glass Surfaces, Ch. 5, Wiley (1964) 42. Horwitz , C. M., J. Vac. Sci. Technol., A1:1795 (1983) 43. Dahimene, M. and Asmussen, J., J. Vac. Sci. Technol., B4:126 (1986) 44. Goebel, D. M., Hirooka, Y., and Sketchy, T. A., Rev. Sci. Instrum., 56:1717 (1985) 45. McDaniel, E. W., Collision Phenomona in Ionized Gases, Wiley (1964) 46. Hollahan, J. R., and Bell, A. T., Techniques and Applications of Plasma Chemistry, J. Wiley (1972) 47. McTaggert, F. K., Plasma Chemistry in Electrical Discharges, Elsevier (1967) 48. Kushner, M. J., J. Appl. Phys., 53:2923 (1982) 49. Thornton, J. A., J. Vac. Sci. Technol., 15:188 (1978) 50. Berg, R. S. and Kominiak, G. J., J. Vac. Sci. Technol., 13:403 (1976) 51. Kowalski, Z. W., J. Mat. Sci. Lett., 6:69 (1987) 52. Betz, G., Surf. Sci., 92:283 (1980) 53. Malherbe, J. B., Hofmann, S., and Sanz, J. M., Appl. Surf. Sci., 27:355 (1986) 54. Mathews, A. and Gethin, D. T., Thin Solid Films, 117:261 (1987) 55. Achard, B., Gruzza, B., and Pariset, C., Surf. Sci., 160:L519 (1985) 56. Fonash, S. J., Solid State Technol, 28(4):201 (1985) 57. Sawin, H. H., Solid State Technol, 28(4):211 (1985) 58. Kominiak, G. J. and Mattox, D. M., Thin Solid Films, 40:141 (1977)

Surface Preparation

155

59. Baker, W. A., and Mossman, A. L., The Matheson Gas Data Book, Matheson Co. 60. Webber, J.,Microelectronic Manufacturing and Testing, p. 40 (Jan 1985) 61. Boyd, H. and DeBord, D., Microelectronic Manufacturing and Testing, 8:1 (April 1985) 62. Choe, D. H. G., Knapp, C., and Jacob, A.,Solid State Technol, 28(3):65 (1985) 63. Nordstrom, H., Buchta, R., Runovc, F., and Klund, P. W., Vacuum 32:737 (1982) 64. Oehrlein, G. S., Clabes, J. G., and Spirto, P., J. Electrochem. Soc., 133:1002 (1986) 65. Pang, S. W., J. Electrochem. Soc., 133:784 (1986) 66. Geis, M. W., Lincoln, G. A., Efremow, N., and Piacentini, W. J., J. Vac. Sci. Technol., 19:1390 (1981) 67. Winters, H. F., Coburn, J. W., and Chuang, T. J., J. Vac. Sci. Technol., B1:469 (1983) 68. Poulsen, R. G., J. Vac. Sci. Technol., 14:266 (1977) 69. Broydo, S., Solid State Technol, 26(4):159 (1983) 70. Poll, H. U., Meichsner, J., and Steinrucken, A., Thin Solid Films, 112: 369 (1984) 71. Kelber, J. A.,Plasma Treatment of Polymers for Improved Adhesion To be published in Vol 119 of MRS Proceedings (D. M., Mattox, J. E. E., Baglin, R. Gottschall, and C. D. Batich, eds.) 72. Bodo, P. and Sundgren, J.-E., Thin Solid Films, 136:147 (1986) 73. Wheeler, D. R. and Pepper, S. V., J. Vac. Sci. Technol., 20:442 (1982) 74. Suzuki, K., Christie, A. B., and Howson, R. P., Vacuum, 36:323 (1986) 75. Bodo, P. and Sundgren, J.-E., J. Appl. Phys., 60:1161 (1986) 76. Schonhorn, H., Ryan, F. W., and Hansen, R. H., J. Adhesion, 2:93 (1970) 77. Sowell, R. R., DeLollis, N. J., Gregory, H. J., and Montoya, O., Recent Advances in Adhesion, (Lieng-Huang Lee, ed.), pp. 77–89, Gordon & Breach (1973) 78. Ouellette, R. P., Barbier, M. M., and Cheremisinoff, P. N., Lowtemperature Plasma Technology Applications, Technomic Publishing 79. Schiller, S., Foerster, H., Hoetzsch, G., and Reschke, J., Thin Solid Films, 72:351 (1980)

156


80. Schiller, S., Foerster, H., Hoetzsch, G., and Reschke, J., Thin Solid Films, 83:7 (1981) 81. Gazecki, J., Sai-Halasz, G. A., Alliman, R. G., Kellock, A., Nyberg, G. L., Williams, J. S., Appl. Surf. Sci., 22/23:1034 (1985) 82. Dini, J. W. and Johnson, H. R., ASTM Spec. Pub. 830, (R. T. Webster and C. S. Young, eds.), p.113-123 (1984) 83. Burkstrand, J. M., J. Vac. Sci. Technol., 16:363 (1979) 84. Burkstrand, J. M., J. Vac. Sci. Technol., 15:223 (1978) 85. Burkstrand, J. M., Appl. Phys. Lett., 33:387 (1978) 86. Hosokawa, T. and Hosokawa, N., J. Vac. Sci. Technol., 16:348 (1979) 87. Sharp, D. J. and Panitz, J. K. G., Surf. Sci., 118:429 (1982) 88. Sproul, W. D. and Richman, M. H.,J. Vac. Sci. Technol., 12:842 (1975) 89. Mattox, D. M., Thin Solid Films, 18:173 (1973) 90. Osaka, T., Nagata, H., Nakajima, E., and Koiwa, I., J. Electrochem. Soc., 133:2345 (1986) 91. Elmore, G. V. and Hershberger, R. F., J. Electrochem. Soc., 121:107 (1974) 92. Berg, R. S. and Kominiak, G. J., J. Vac. Sci. Technol., 13:403 (1976) 93. Campbell, P. and Green, M. A., J. Appl. Phys., 62:243 (1987) 94. Liberman, R. and Klein, D. L., J. Electrochem. Soc., 113:957 (1966) 95. Ingrey, S., Lau, W. M., McIntyre, N. S., J. Vac. Sci. Technol., A4:984 (1986) 96. McClintock, J. A., Wilson, R. A., and Byer, N. E., J. Vac. Sci. Technol., 20:241 (1982) 97. Wright, S. and Kroemer, H., Appl. Phys. Lett., 36:210 (1980) 98. Yang, H. T. and Berry, W. S., J. Vac. Sci. Technol., B(2):206 (1984) 99. Liberman, R. and Klein, D. L., J. Electrochem. Soc., 113:957 (1966)

4 Evaporation: Processes, Bulk Microstructures and Mechanical Properties Rointan F. Bunshah

1.0

GENERAL INTRODUCTION

Physical Vapor Deposition (PVD) technology consists of the techniques of evaporation, ion plating and sputtering. It is used to deposit films and coatings or self-supported shapes such as sheet, foil, tubing, etc. The thickness of the deposits can vary from angstroms to millimeters. The wide variety of applications of these techniques ranges from decorative to utilitarian over significant segments of the engineering, chemical, nuclear, microelectronics and related industries. Their use has been increasing at a very rapid rate since modern technology demands multiple, and often conflicting, sets of properties from engineering materials, e.g., combinations of two or more of the following: high temperature strength, impact strength, specific optical, electrical or magnetic properties, wear resistance, ability to be fabricated into complex shapes, biocompatibility, cost, etc. A single or monolithic material cannot meet such demands in high technology applications. The solution is, therefore, a composite material, i.e., a core material and a coating each having the requisite properties to fulfill the specifications. PVD technology is very versatile, enabling one to deposit virtually every type of inorganic materials—metals, alloys, compounds and mixtures thereof, as well as some organic materials. The deposition rates can be varied from 10 to 750,000 Å (10 -3 to 75 µm) per minute, the higher rates having come about in the last twenty years with the advent of 157

158


electron beam heated sources. For zinc and aluminum, deposition rates as high as 25 µm per second have been reported using electron beam evaporation sources. The thickness limits for thin and thick films are somewhat arbitrary. A thickness of 10,000 Å (1 µm) is often accepted as the boundary between thin and thick films. A recent viewpoint is that a film can be considered thin or thick depending on whether it exhibits surface-like or bulk-like properties. Historically the first evaporated thin films were probably prepared by Faraday[1] in 1857 when he exploded metal wires in a vacuum. The deposition of thin metal films in vacuum by Joule heating was discovered in 1887 by Nahrwold[2] and was used by Kundt[3] in 1888 to measure refractive indices of such films. In the ensuing period, the work was primarily of academic interest concerned with optical phenomena associated with thin layers of metals, researches into kinetics and diffusion of gases, and gasmetal reactions.[4][5] The application of these technologies on an industrial scale had to await the development of vacuum techniques and therefore dates to the post World War II era, i.e., 1946 and onwards. This proceeded at an exponential pace in thin films and is covered in an excellent review by Glang[6] on evaporated films and in other chapters of the Handbook of Thin Film Technology[7] as well as in the classic text by Holland.[8] A more recent reference on the Science and Technology of Surface Coatings[9] includes material on PVD techniques as well as the other techniques for surface coatings. The work on mechanical properties of thin films has been reported in several review articles.[10]-[15] The work on the production of full-density coatings or self-supported shapes by high deposition rate PVD processes started around 1961 independently at two places in the U.S.A. Bunshah and Juntz at the Lawrence Livermore Laboratories of the University of California produced very high purity beryllium foil,[16]-[21] titanium sheets,[22] and studied the variation of impurity content, microstructure and mechanical properties with deposition conditions, thus demonstrating that the microstructure and properties of PVD deposits can be varied and controlled. At about the same time, Smith and Hunt were working at Temescal Metallurgical Corporation in Berkeley on the deposition of a number of metals, alloys and compounds and reported their findings in 1964.[23][24] The development of evaporation processes in the U.S.S.R is described in the Appendix kindly supplied to the author by Dr. A. V. Demchishin of the Paton Electric Welding Institute, Kiev. In the years between 1962 and 1969, there was considerable effort on the part of various steel companies to produce Al and Zn coatings on steel using HRPVD techniques on a production scale.[25][26] In 1969, Airco

Evaporation

159

Temescal Corp. decided to manufacture Ti-6Al-4V alloy foil in pilot production quantities for use in honeycomb structures on the SST aircraft. The project was eminently successful but the patient, the supersonic transport aircraft “SST,” died. The results of this work were published in 1970.[122a] To give some idea of the production capability, 1,200 ft/run of Ti-6Al-4V foil, 12" wide, 0.002" thick was produced at the rate of 2 to 3 ft/min. The stated cost at that time was about one-fifth of the cost for similar material produced by rolling (i.e., $60/lb for HRPVD vs. $300/lb for rolled material). It is very difficult to roll this alloy because it work-hardens very rapidly and therefore needs many annealing cycles to be reduced to thin gauge (A. B. Sauvegot, TMCA Tech. Report AFML-TR-67-386, Dec. 1967). The work on thick films and bulk deposits has matured later than the work on thin films and reviews on it have been given by Bunshah[114][116] and by Paton, Movchan and Demchishin[122] who summarized the work done at the Paton Electric Welding Institute up to 1973. In addition, the Soviet literature in the 1960s has numerous references to the extensive work on thin and thick films by Palatnick and coworkers of the Kharkov Polytechnic Institute (see Appendix). Note should also be made of a recent book in German on electron beam technology by Schiller, Heisig, and Panzer in which many of the PVD aspects are treated.[27]

2.0

SCOPE

The scope of this chapter will be to review the evaporation technologies, theory and mechanisms, processes, deposition of various types of materials, the evolution of the microstructure and its relationship to the properties of the deposits, preparation of high purity metals, current and future applications, and finally cost analysis as far as possible.

3.0

PVD PROCESSES

3.1

Preamble

In general, deposition processes may principally be divided into two types: (i) those involving droplet transfer such as plasma spraying, arc spraying, wire-explosion spraying, detonation gun coating, and (ii) those involving an atom by atom transfer mode such as the physical vapor

160


deposition processes of evaporation, ion plating and sputtering, chemical vapor deposition, and electrodeposition. The chief disadvantage of the droplet transfer process is the porosity in the final deposit which effects the properties. There are three steps in the formation of any deposit: 1. Synthesis of the material to be deposited: a. Transition from a condensed phase (solid or liquid) to the vapor phase. b. For deposition of compounds, a reaction between the components of the compound, some of which may be introduced into the chamber as a gas or vapor. 2. Transport of the vapors between the source and substrate. 3. Condensation of vapors (and gases) followed by film nucleation and growth. There are significant differences between the various atom transfer processes. In chemical vapor deposition and electrodeposition processes, all of the three steps mentioned above take place simultaneously at the substrate and cannot be independently controlled. Thus, if a choice is made for a process parameter such as substrate temperature (which governs deposition rate in CVD), one is stuck with the resultant microstructure and properties. On the other hand, in the PVD processes, these steps (particularly steps 1 and 3 can be independently controlled and one can therefore have a much greater degree of flexibility in controlling the structure and properties, and deposition rate. This is a very important consideration. 3.2

PVD Processes

There are three physical vapor deposition processes, namely evaporation, ion plating, and sputtering. Ion plating is a hybrid process. In the evaporation process, vapors are produced from a material located in a source which is heated by direct resistance, radiation, eddy currents, electron beam, laser beam or an arc discharge. The process is usually carried out in vacuum (typically 10-5 to 10-6 torr) so that the evaporated atoms undergo an essentially collisionless line-of-sight transport prior to condensation on the substrate. The substrate is usually at ground potential (i.e., not biased). Figure 4.1 is a schematic of a vacuum evaporation system illustrating electron beam heating. It may be noticed that the deposit thickness is greatest directly above the center-line of the source and decreases away from

Evaporation

161

it.[28] This problem is overcome by imparting a complex motion to substrates (e.g., in a planetary or rotating substrate holder) so as to even out the vapor flux on all parts of the substrate; or by introducing a gas at a pressure of 5 to 200 µm into the chamber so that the vapor species undergo multiple collisions during transport from the source to substrate, thus producing a reasonably uniform (±10%) thickness of coating on the substrate. The latter technique is called gas-scattering evaporation or pressure plating. [29][30]

Figure 4.1. Vacuum-evaporation process using electron beam heating.

162


In the ion-plating process, the material is vaporized in a manner similar to that in the evaporation process but passes through a gaseous glow discharge on its way to the substrate, thus ionizing some of the vaporized atoms (see Fig. 4.2). The glow discharge is produced by biasing the substrate to a high negative potential (-2 to -5 kV) and admitting a gas, usually argon, at a pressure of 5 to 200 mTorr into the chamber. In this simple mode, which is known as diode ion-plating, the substrate is

Figure 4.2. Ion-plating process.

Evaporation

163

bombarded by high-energy gas ions which sputter off the material present on the surface. This results in a constant cleaning of the substrate (i.e., a removal of surface impurities by sputtering) which is desirable for producing better adhesion and lower impurity content. This ion bombardment also causes a modification in the microstructure and residual stresses in the deposit. On the other hand, it produces the undesirable effects of decreasing the deposition rates since some of the deposit is sputtered off, as well as causing a considerable (and often undesired for microelectronic applications) heating of the substrate by the intense gas ion bombardment. The latter problem can be alleviated by using the supported discharge ion-plating process[31a,b] where the substrate is no longer at the high negative potential; the electrons necessary for supporting the discharge come from an auxiliary heating tungsten filament. The high gas pressure during deposition causes a reasonably uniform deposition of all surfaces due to gas-scattering as discussed above. In the sputtering process, illustrated schematically in Fig. 4.3, positive gas ions (usually argon ions) produced in a glow discharge (gas pressure: 20 to 150 mTorr) bombard the target material (also called the cathode) dislodging groups of atoms which then pass into the vapor phase and deposit onto the substrate. Alternate geometries of importance in various processing applications are shown in Fig. 4.4. For example, hollow cathode sputtering would be the ideal geometry for coating the outer surface of a wire. Sputtering is an inefficient way to induce a solid-to-vapor transition. Typical yields (atoms sputtered per incident ion) for a 50 eV argon ion incident on a metal surface are unity. Thus the phase change energy cost is from 3 to 10 times larger than evaporation.[32] Thornton[32] has provided an excellent review on sputtering as applied to deposition technology. The reader is also referred to the proceedings of a special conference on "Sputtering and IonPlating."[33] The deposition rates for the various processes are indicated in Table 4.1. The deposition rates of the evaporation and ion-plating processes are much higher than those of the sputtering process. Recently, Schiller and Jasch,[228] reported on large scale industrial applications of deposition of Al on strip steel continuously at a deposition rate of 20 µ/min. It should be noted that sputter deposition rates at the high side (approximately 10,000 Å/min) with diode sputtering can only be obtained for target materials of high thermal conductivity like copper, since heat extraction from the target is the limiting parameter. For most materials, it is much lower, i.e., 50 to 1,000 Å/min. With magnetron sputtering, much higher deposition rates are obtained (see Ch. 5 in this volume).

164


Figure 4.3. Basic sputtering process.

Figure 4.4. Cylindrically symmetric sputter-coating systems.

Evaporation

165

Table 4.1. Deposition Rates for Various PVD Processes Evaporation, Å/min Ion Plating, Å/min Sputtering, Å/min

100 - 250,000* 100 - 250,000 25 - 10,000

*In special cases to 500,000 Å/sec

3.3

Advantages and Limitations

There are several advantages of PVD processes over competitive processes such as electrodeposition, CVD, and plasma spraying. They are: 1. Extreme versatility in composition of deposit. Virtually any metal, alloy, refractory or intermetallic compound, some polymeric type materials and their mixtures can be easily deposited. In this regard, they are superior to any other deposition process. 2. The ability to produce unusual microstructures and new crystallographic modifications, e.g., amorphous deposits. 3. The substrate temperature can be varied within very wide limits from subzero to high temperatures. 4. Ability to produce coatings or self-supported shapes at high deposition rates. 5. Deposits can have very high purity. 6. Excellent bonding to the substrate. 7. Excellent surface finish which can be equal to that of the substrate. 8. Elimination of pollutants and effluents from the process which is a very important ecological factor. The present limitations of PVD processes are: 1. Inability to deposit polymeric materials with certain exceptions. 2. Higher degree of sophistication of the processing equipment and hence a higher initial cost.

166


4.0

THEORY AND MECHANISMS

4.1

Vacuum Evaporation

Reference to the various steps in the formation of a deposit enumerated in the previous section shows that the theory of vacuum evaporation involves thermodynamic considerations, i.e., phase transitions from which the equilibrium vapor phase pressure of materials can be derived, as well as the kinetic aspects of nucleation and growth. Both of these are of obvious importance in the evolution of the microstructure of the deposit. The transition of solids or liquids into the gaseous state can be considered to be a macroscopic or an atomistic phenomenon. The former is based on thermodynamics and results in an understanding of evaporation rates, source-container reactions and the accompanying effect of impurity introduction into the vapor state, changes in composition during alloy evaporation, and stability of compounds. An excellent detailed treatment of the thermodynamic and kinetic bases of evaporation processes is given by Glang.[6] He points out that the application of kinetic gas theory to interpret evaporation phenomena resulted in a specialized evaporation theory. Such well known scientists as Hertz, Knudsen and Langnuir were the early workers in evaporation theory. They observed deviations from ideal behavior which led to refinements in the theory to include concepts of reaction kinetics, thermodynamics, and solid state theory. From the kinetic theory of gases, the relationship between the impingement rate of gas molecules and their partial pressure, p, is given by

Eq. (1)

dNi = (2πmkT )−½ p A e dt

where Ni is the number of molecules striking a unit area of surface, and Ae is the area of the surface. Hertz,[34] in 1882, first measured the evaporation rate of mercury in high vacuum and found that the evaporation rate was proportional to the difference between the equilibrium vapor pressure of mercury, p*, at the evaporant surface and the hydrostatic pressure, p, acting on the surface, resulting from the evaporant atoms or molecules in the gas phase. Thus, the evaporation rate based on the concept of the equilibrium vapor pressure, (i.e., the number of atoms leaving the evaporant surface is equal to the number returning to the surface) is given by:

Evaporation

Eq. (1a)

167

dN e = ( 2 πmkT ) −½(p* − p ) cm2 sec −1 A e dt

such that dN e, the number of molecules evaporating from a surface area A e in time dt, is equal to the impingement rate of gas molecules based on the kinetic theory of gases with the value of p* inserted therein, minus the return flux corresponding to the hydrostatic pressure p of the evaporant in the gas phase. In the above equations, m is the molecular weight, k is Boltzmann’s constant, and T is the temperature in °K. The maximum possible evaporation rate corresponds to the condition p = 0. Hertz measured evaporation rates only about one-tenth as high as the theoretical maximum rates. The latter were subsequently measured by Knudsen[35] in 1915. Knudsen postulated that some of the molecules impinging on the surface were reflecting back into the gas phase rather than becoming incorporated into the liquid. As a result, there is a certain fraction (1 - αν) of vapor molecules which contribute to the evaporant pressure but not to the net molecular flux from the condensed phase into the vapor phase. To this end, he postulated the evaporation coefficient, αν, which is defined as the ratio of the real evaporation rate in vacuum to the theoretically possible value defined by Eq. (1a). This then results in the well-known Hertz-Knudsen equation Eq. (2)

dN e = α ν ( 2 πmkT ) −½(p* − p ) A e dt

The value of αν is very dependent on the cleanliness of the evaporant surface and can range from very low values for dirty surfaces to unity for clean surfaces. In very high rate evaporation with a clean evaporant surface, it has been found that the maximum evaporation given by Eq. (2) has been exceeded by a factor of 2 to 3 for the evaporation of a light metal such as beryllium[21] using electron beam heating. The reason for this is that the high power input results in considerable agitation of the liquid evaporant pool resulting in a real surface area much larger than the apparent surface area. The directionality of evaporating molecules from an evaporation source is given by the well-known cosine law. Figure 4.5 shows a small surface element dAr receiving deposit from a small area source Ar. The mass deposited per unit area

168


Eq. (3)

dMr ( σ, θ) Me = cos φ cos θ dA r πr 2

where Me is total mass evaporated.

Figure 4.5. Surface element dA r receiving deposit from a small-area source dAe. (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Company.)

For a point source, Eq. (3) reduces to:

Eq. (4)

dMr M = e cos φ dAr πr 2

For a uniform deposit thickness, the point source must be located at the center of the spherical receiving surface such that r is a constant and cosθ = 1. In high rate evaporation conditions, e.g., using a high power electron beam heated source, the thickness distribution is steeper than with a point or small area source discussed above. This has been attributed by some

Evaporation

169

authors[28][36] to the existence of a virtual source of vapor located above the molten pool. On the other hand, at high power, the electron beam impact area on the surface of the molten pool is not flat but pushed down into an approximate concave spherical segment which as Riley shows[37] can equally well account for the steeper thickness distribution. The above discussion points out one of the problems with evaporation technology, i.e., the variation in thickness of the deposit on a flat substrate. Numerous solutions are possible which involve either moving the substrate in a manner so as to randomly sample the vapor flux, the use of multiple sources, or sources of special shapes. These have been discussed in some detail by Holland[8] as well as by Bunshah and Juntz.[38] Models have also been presented for calculating the deposit temperature[39] and thickness distribution[40] during high-rate evaporation and verified against experimental data. In a more recent paper, Szekely and Poveromo[41] have given a more general formulation describing the net rate of vapor deposition from a molten source onto an initial cold surface, making allowance for both molecular transport and diffusion effects.

5.0

EVAPORATION PROCESS AND APPARATUS

5.1

The System

A schematic of the evaporation apparatus has been illustrated in Fig. 4.1. It consists of the following: chamber, vacuum pumps, vacuum gages, including total and partial pressure gages on sophisticated systems, evaporation sources, substrate holders, rate monitors, process controller, etc. Vacuum Chamber: This ranges from a simple bell jar or rectangular box for experimental or batch type production to more complex gear for production applications. The latter may consist of a deposition chamber with loading and unloading chambers attached to the deposition chambers by manifolds with isolation high vacuum valves. These are called fast cycle coaters. Alternate approaches are semi-continuous in-line systems where a strip substrate stored in the vacuum chamber can be fed continuously over the source (Fig. 4.6) or a continuous system where the strip or sheet substrate is inserted and removed from the deposition chamber through airto-air seals[4][42] as shown in Figs. 4.7 and 4.8.

170


Figure 4.6. A schematic representation of a 24 inch continuous high vacuum strip processing line.

Figure 4.7. A three-high roll seal arrangement for stripline.

Evaporation

171

Figure 4.8. Vacuum seal using steam jet or curtain.

Vacuum Pumping System: The gas loads in evaporation processes are fairly high due to outgassing from chamber walls promoted by the heat load from the evaporation source and substrate heaters, particularly for high deposition rate conditions. Therefore the pumping system is usually based on a diffusion pump with a liquid nitrogen cooled anti-creep type bafflebacked with a mechanical pump or a Roots blower/mechanical pump combination for large systems. For very high purity, low deposition rate, low heat flux conditions, ion pumped systems backed with cryosorption rough pumping are used, since a base pressure of 10-9 to 10-10 torr is needed. More recently, turbomolecular and cryogenic pumps are used instead of diffusion pumps where desired (e.g., oil-free systems). This is particularly true for molecular beam epitaxy where extreme control over composition and layer thickness are essential and deposition rates can be quite low. In such cases, the chamber and pumps are to be baked as with any other ultra-high vacuum operation. Pressure Measurement: The vacuum gages used depend again on specific applications. A combination of high pressure gages (such as the Pirani or Thermocouple Gage) for monitoring the roughing of the system in

172


combination with high vacuum gages (such as the hot cathode ionization gage and/or capacitance manometer). A partial pressure gage is highly desirable particularly for ultra-clean applications as well as for leak hunting. Evaporation Sources: These are discussed separately in Sec. 6). Substrate Holders and Heaters: Substrate holders may be very simple for stationary flat substrates or can incorporate quite complex motions as illustrated by planetary or rotating devices. The reason for this is to ensure deposition thickness uniformity and control over a large number of small parts such as lenses or silicon wafers. Substrate heating can be accomplished by radiant heaters with refractory wires or quartz lamps acting as the heat source. Occasionally, substrates are directly heated by a scanning or diffuse electron beam. Deposition Rate Monitors: These are discussed in Sec. 8. 6.0

EVAPORATION SOURCES

6.1

General Considerations

Evaporation sources are classified by the mode of heating used to convert the solid or liquid evaporant to the vapor phase. Thus one talks of resistance, arc, induction, electron beam, arc imaging, lasers, and exploding wire types of sources. A very important fact to be noted is that we cannot evaporate every material fromany of the types of sources listed above for the following reasons. 1. Chemical interaction between the source material and the evaporant which would lead to impurities in the deposit. For example, evaporation of titanium from a MgO source would cause oxygen and magnesium contamination of the deposit; the titanium would reduce the MgO. Therefore, for the evaporation of reactive metals like titanium, zirconium, etc., we use water cooled copper crucibles. 2. Reaction between metallic source (such as a W or Ta boat) and evaporant (Ti) could occur. In many cases at high temperatures two metals can mutually dissolve in each other leading to a destruction of the source. 3. The power density (i.e., watts per sq. cm) varies greatly between the various heat sources. Table 4.2 from Ref. 6 from the article by Glang lists the temperature and support materials to be used in the evaporation of elements. Similar tables are found in the literature of many of the manufacturers. Evaporation of alloys and compounds pose additional problems and they are considered in Sec. 9.

Evaporation

173

Table 4.2. Temperatures and Support Materials Used in the Evaporation of the Elements

174


Table 4.2. (Cont'd)

Evaporation 6.2

175

Resistance Heated Sources

The simplest vapor sources are resistance heated wires and metal foilsof various types shown in Fig. 4.9.

Figure 4.9. Wire and metal-foil sources. (A) Hairpin source. (B) Wire helix. (C) Wire basket. (D) Dimpled foil. (E) Dimpled foil with alumina coating (F) Canoe type. (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

They are available in a variety of sizes and shapes and at sufficiently low prices so that they can be discarded after one experiment if necessary. They are usually made from the refractory metals, tungsten, molybdenum, and tantalum which have high melting points and low vapor pressure so as not to contaminate the deposit. Their properties are given in Table 4.3. Platinum, iron or nickel are sometimes used for materials which evaporate below 1000°C. The capacity (total amount of evaporant) of such sources is small. The hairpin and wire helix sources are used by attaching the evaporant to the source in the form of small wire segments. Upon melting, the evaporant must wet the filament and be held there by

176


surface tension. This is desirable to increase the evaporation surface area and thermal contact. Multistrand filament wire is preferred because it increases the surface area. Maximum amount held is about 1 gram. Dimpled sources and basket boats may hold up to a few grams.

Table 4.3. Properties of Refractory Metals

Since the electrical resistance of the source is small, low voltage power supplies, 1 to 3 kW, are recommended. The current in the source may range from 20 to 500 amps. In some cases, the evaporant is electroplated onto the wire source. The principal use of wire baskets is for the evaporation of pellets or chips of dielectric materials which either sublime or do not wet the wire on melting. In such cases, if wetting occurs, the turns of the baskets are shorted and the temperature of the source drops. The rate of evaporation from such sources may vary considerably due to localized conditions of temperature variation, wetting, hot spots, etc. Therefore, for a given thickness of film, the procedure is to load the source with a fixed weight of evaporant and evaporate to completion or use a rate monitor and/or thickness monitor to obtain the desired evaporation rate and thickness. 6.3

Sublimation Sources

For materials evaporating above 1000°C, the problem of non-reactive supports may be circumvented for materials such as Cr, Mo, Pd, V, Fe and

Evaporation

177

Si which reach a vapor pressure of 10-2 torr before melting. Hence, they can sublime and produce a sufficiently high vapor density. The contact area between the evaporant and the source crucible is held to a minimum. Figure 4.10 shows such a source designed by Roberts and Via.

Figure 4.10. Chromium sublimation source after Roberts and Via. The electric current flows through the tantalum cylinder (heavy lines). (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

A different type of sublimation source is used for the vaporization of thermally stable compounds such as SiO which are commonly obtained as powders or loose chunks. Such source material would release large quantities of gases upon heating thus causing ejection of particles of the evaporant which may get incorporated into the film. Figure 4.11 shows two sources which solve this problem by reflection of the vaporized material.

178


Figure 4.11. Optically dense SiO sources. (A) The Drumheller source. (B) Compartmentalized source. (After Vergara, Greenhouse and Nicholas.) (From Handbook of Thin film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

6.4

Evaporation Source Materials

We have already discussed the potential problems concerned with the reaction between metal sources and evaporants. Oxides and other compounds are more stable than metals. Table 4.4 gives the thermal stability of refractory oxides in contact with metals. There are many metals not listed in Table 4.4 which can be evaporated from refractory oxide sources. Note that there is no such thing as an absolutely stable oxide, nitride or other compound. Reaction is controlled by kinetics, i.e., temperature and time. Oxide crucibles have to be heated by radiation from metal filaments or their contents can be heated by induction heating. This is illustrated in Fig. 4.12 and 4.13 for resistance heated sources. Other source materials are nitrides such as boron nitride. A 50% BN50% TiB2 is also well established as a crucible material. This material (HDA composite, Union Carbide) is a fairly good electrical conductor and hence can be directly heated to evaporate materials. It can be readily machined to shape. Pyrolytic BN and carbon are also used.

Evaporation

179

Table 4.4. Thermal Stability of Refractory Oxides in Contact with Metals*

Figure 4.12. Oxide crucible with wire-coil heater. (From Handbook of Thin Film Technology. C 1970, McGraw-Hill. Used with permissionof McGraw-HillBookCo.)

180


Figure 4.13. DaSilva crucible source. (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

6.5

Induction Heated Sources

Figure 4.14 shows the induction heated sources using a BN-TiB2 crucible. Figure 4.15 shows an induction heated evaporation sublimation source using a water cooled copper crucible.[19] This is suited to the evaporation of reactive metals such as Ti, Be, etc., which will react with all the refractory oxides, nitrides, etc.

Figure 4.14. RF heated aluminum source with boron-nitride/titanium-diboride crucible. (After Ames, Kaplan and Roland). (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

Evaporation

181

Figure 4.15. Schematic representation of the distillation setup.

6.6

Electron Beam Heated Sources

Electron beam heated sources have two major benefits. One is a very high power density and hence a wide range of control over evaporation rates from very low to very high. Two, the evaporant is contained in a water-cooled copper hearth thus eliminating the problem of crucible contamination. The evaporation rate for pure metals like Al, Au, Ag, which are good thermal conductors, from water-cooled copper crucibles decreases due to heat loss to the crucible walls. In such cases, crucible liners of carbon and other refractory materials are used. Any gun system must consist of at least two elements—a cathode and anode. In addition, it is necessary to contain these in a vacuum chamber in order to produce and control the flow of electrons, since they are easily

182


scattered by gas molecules. A potential difference is maintained between the cathode and the anode. This varies from as little as a few kilovolts to hundreds of kilovolts. In melting systems, a normal operational range is of the order of 10 - 40 kV. In the simple diode system, the cathode emits electrons, which are then accelerated to the anode across the potential drop. Where the anode is the workpiece to be heated, this is termed a workaccelerated gun. It is shown schematically in Fig. 4.16a. In a selfaccelerated gun structure, an anode is located fairly close to the cathode, electrons leave the cathode surface, are accelerated by the potential difference between the cathode and anode, pass through the hole in the anode and continue onward to strike the workpiece. Self-accelerated guns have become the more common type in use and offer more flexibility than the work-accelerated gun.

Figure 4.16. Simple electron beam guns. (a) Work-accelerated gun. (b) Selfaccelerated gun.

Evaporation

183

Electron beam guns may be further subdivided into two types depending on the source of electrons: (i) thermionic gun and (ii) plasma gun. Thermionic Gun: In thermionic guns, the source of electrons is a heated wire or disc of a high temperature metal or alloy, usually tungsten or tantalum. Such guns have the limitation of a minimum operating gas pressure of about 1 x 10-3 torr. Higher pressures cause scattering of the electron beam as well as a pronounced shortening of the cathode life (if it is a wire or filament) due to erosion by ion bombardment. Figure 4.17 shows examples of thermionic electron beam heated work-accelerated sources. The close cathode gun shown in Fig. 4.17A is not a desirable configuration since molten droplet ejection from the pool impinging on the cathode will terminate the life of the cathode due to low melting alloy formation. Thus cathodes are hidden from direct line-of-sight of the molten pool and the electron beam is bent by electrostatic fields (Figs. 4.17B and 4.17C) or magnetic field (Figs. 4.18 and 4.19) generated by electromagnets. The latter is a preferred arrangement since variation of the X and Y components of the magnetic field can be used to scan the position of the beam on the molten pool surface.

Figure 4.17. Work-accelerated electron-bombardment sources. (A) Pendant-drop method. (B) Shielded filament (Unvala). (C) Shielded filament (Chopra and Randlett). (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

184


Figure 4.18. Bent-beam electron gun with water-cooled evaporant support. (With permission of Temescal Metallurgical Co., Berkeley, CA). (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

Figure 4.19. Transverse electron beam gun.

Evaporation

185

Figures 4.17, 4.18 and 4.19 show linear cathodes (i.e., wires or rods) and are referred to as transverse linear cathode guns. Figure 4.20 shows a disc cathode which is characteristic of a high power Pierce type electron beam gun. Low power Pierce type guns may have a hair pin filament or awire loop as the cathode. In either case the beam geometry of the Pierce gun is different than that of the transverse linear cathode guns. In some instances, the electron emitter assembly is located at a distance from the crucible in a separately pumped chamber to keep the pressure below 1 x 10 -3 torr, with a small orifice between the emitter chamber and the crucible chamber for the passage of electrons.

Figure 4.20. Schematic representation of a Pierce gun.

186


Plasma Electron Beam Gun:A plasma is defined as a region of hightemperature gas containing large numbers of free electrons and ions. By a proper application of electrical potential, electrons can be extracted from the plasma to provide a useful energy beam similar to that obtained from thermionic guns. There are two types of plasma e-beam guns: (a) The Cold Cathode Plasma Electron Beam—The plasma electron beam gun has a cylindrical cathode cavity made from a metal mesh or sheet (Fig. 4.21) containing the ionized plasma from which electrons are extracted through a small aperture in one end. The cathode is maintained at a negative potential, e.g., -5 to -20 kV, relative to the workpiece and remainder of the system, which are at ground potential. After evacuation of the system, a low pressure of ionizable gas in the range of 10 -3 to 10 -1 torr is introduced. Depending upon the high voltage level, a long path discharge between the cathode and other parts of the system will occur in the gas at a particular pressure. Ionizing collisions in the gas then produce positive ions which are accelerated to the cathode, causing electrons to be released from the cathode surface. Although the cathode may heat up slightly due to ion bombardment, no heating is required for electron emission. Upon proper adjustment of cathode voltage and gas pressure, a beam mode of operation is established, since interaction between the plasma inside and outside of the cathode and the electric fields between cathode surface and plasma boundary will largely confine electron emission to the end of the cathode and its interior. In argon, a beam mode is supported at about 10 -2 torr with 5 - 10 kV. Beam currents range up to 3 A for a 3 inch diameter cathode in argon at 20 kV. With lighter gases, e.g., helium, higher pressure to about 10 -1 torr will yield a beam mode in this same voltage range. Beam current will vary with voltage and pressure control, also. More specific information is given by Cocca and Stauffer.[44] The beam is self-collimating because of the focusing effect of positive ions in the beam path and the electrostatic lensing action of the aperture since it separates regions of different potential gradient. The beam is well collimated, having a cross section equal to that of the cathode aperture. Adjustment of focus can be achieved to some extent by varying pressure and voltage,

Evaporation but external focusing may also be used if desired, with magnetic or electrostatic lenses, as with conventional election beams.

Figure 4.21. Cold cathode plasma electron beam gun.

(b) The Hot Hollow Cathode Discharge Beam—The hollow cathode discharge beam applied to vacuum processing has been reported by Morley[45] and differs in a number of respects from the plasma beam. A schematic of the hollow cathode discharge beam is shown in Fig. 4.22. Here the cathode must be constructed of a refractory metal since it operates at elevated temperature. An ionizable gas, usually argon, is introduced into the system through the tubular-shaped cathode. A pressure drop across the orifice in the cathode provides a sufficient amount of gas inside the cathode to sustain the plasma, which generates the beam.

187

188

Deposition Technologies for Films and Coatings A low voltage, high amperage DC power source is utilized. When RF power from a commercial welding starter is coupled to the gas, it becomes ionized and the plasma is formed. Continued ion bombardment of the cathode results in heating of the cathode and increased electron emission. Ultimately, a high current “glow discharge” will occur, analogous to that experienced in vacuum arc melting at higher pressures. At this point, the discharge appears as a low power density beam “flowing” from the cathode aperture and fanning out in conical shape into the chamber. However, a parallel axial magnetic field is imposed on the beam (as seen in Fig. 4.22) which then forms a high power density, well-collimated beam. The hollow cathode discharge beam is operationally stable and efficient over the pressure range from 10-4 to 10-1 torr. A more detailed description of physical aspects, operational characteristics, and cathode design has been given by Morley.[45]

Figure 4.22. Schematic of the hot hollow cathode electron beam gun.

Evaporation

189

Comparisons:Thermionic as well as the plasma e-beam guns can be used equally well for evaporation. Focusing of the beam spot is easier for the thermionic guns. The plasma guns have the advantage of being able to operate at higher pressures which can be important for gas scattering evaporation, reactive evaporation, and ion plating. 6.7

Arc Evaporation

The definitions of arcs are: Karl T. Compton, Princeton University: “A discharge in a gas or vapor that has a voltage drop at the cathode of the order of the minimum ionizing or exciting potential of the gas or vapor.” Lafferty: “The arc is a self sustained discharge capable of supporting large currents by providing its own mechanism of electron emission from the cathode.” Berghaus[46] describes the use of arcs to form refractory compounds by reactive evaporation. Since 1940, consumable and nonconsumable vacuum arc melting processes have been developed to melt and refine various reactive metals such as Ti, Hf, Zr, etc. More recently, arc techniques have been used to deposit metals[47][48] and refractory compounds, and even for extraction of ions from the vacuum arc plasma for the deposition of metal films.[49] Wroe[50] in 1958 and Gilmour et al.[51] suggested vacuum arcs as a source for metallic coatings. The US patents to Snaper[52][53] in 1971 and the Russian patents to Sablev[54][55] in 1974 set the stage for the commercial production of arc coatings which were achieved in the USSR around 1977 - 1978. The first commercial use of the arc evaporation-deposition method was for TiN coatings deposited at low temperatures, particularly for high speed steel cutting tools by arc evaporation of titanium in a nitrogen plasma. This follows on the heels of the Activated Reactive Evaporation (ARE) process developed in 1971 for deposition of refractory compounds such as TiN using electron beam evaporation techniques and discussed in Sec. 9.6. There is very extensive Russian literature on vacuum arc coating technology and the reader can find a convenient source in recent reviews by Sanders[56] and by Martin.[57] There are two types of cathodic arc systems—pulsed and continuous. In the pulsed devices, the arc is repeatedly ignited and extinguished using a capacitor blank to supply the arc power.[51] Pulsed arcs have the advantage of letting the target cool between the pulses. The disadvantage is the decrease in steady state coating rates.

190


The continuous cathodic arc can be random in nature or controlled. By the use of an insulating ring, a random arc source can be constrained at the edge of the target, but allowed random motion within that constraint. Random arc sources have the advantage of simplicity and excellent target utilization because the entire target (except near the very edge) is utilized in the arc of very large parts. The main disadvantage of random arcs is the formation of macroparticles which may cause the resulting coating to be unsuitable in some applications. Figure 4.23 shows that macroparticles are ejected at small angles with respect to the target surface, and can therefore be minimized using appropriate shielding. Such a strategy has made possible arc-produced decorative coatings where surface finish and optical specularity are of concern.

Figure 4.23. Phenomena occurring at a discrete cathodic arc spot.

Evaporation

191

Magnetic fields can be used to control the trajectories of the arcs. These fields can be used to discourage the arc from leaving the desired portion of the target surface or can actually be used to define a well controlled path for the arc to follow in the so called “steered arc” devices. While the mechanism is still the subject of some debate, it is clear, at least in the case of ceramic coatings based on refractory metals, that steered arcs can produce coatings having extremely low or no measurable macroparticle component. Macroparticles can also be removed by the use of suitable filters as discussed by Sanders[56] and by Martin.[58] This is the so-called “filtered arc evaporation process.“ Other strategies for macroparticles involve the production of diffuse arcs. In one case, the cathode is contained in a crucible which is allowed to heat up to a temperature where the target material has a substantial vapor pressure.[59] This causes a decrease in the arc voltage and current density, the discharge becomes diffuse and macroparticles no longer form. The other approach is the so-called "anodic arc,“[60]-[62] (see Fig. 4.24). In this process the cathode initially supplies electrons as well as ions until the anode heats up. At this point, with sufficient electron emission, a diffuse arc forms on the hot anode target material which supplies the ions necessary to sustain the discharge. The cathode material is not evaporated and the coating material now emanates from the anode. There are no macroparticles formed. High deposition rates (several µm per minute) are obtained for a variety of metals including Al, Ti, V, Ca, Mn, Fe, Ni, Cu, Pd, Ag, Au and Pt.[60] Since the substrate is left relatively cool, the process makes it possible to produce adherent coatings on plastics at temperatures less than 70°C which makes this relatively new process a competitor for sputter deposition. Alloy coatings such as stainless steel can be readily deposited with good stoichiometric transfer. For example, Ni, Al, and stainless steel coatings less than 1 µm thickness impart excellent corrosion protection to iron.[63] One of the main advantages of arc deposition processes is the relatively high level of ionizing atoms in the plasma. This makes it convenient to extract ion beams from the plasma and deposit macroparticle free coatings entirely from the ion beam.[56][58]

192


Figure 4.24. Schematic of the anodic arc evaporation process.

7.0 LASER INDUCED EVAPORATION/LASER ABLATION/PULSED LASER DEPOSITION (PLD) This technique with many names was first used by Smith and Turner[64] in 1965 to deposit thin films in a vacuum chamber using a pulsed ruby laser. Systematic studies in the 1970’s were performed to provide a better understanding of the physics of laser-solid interactions and the related issues of deposition mechanisms and film quality. More recently the process has been extensively used for growing highly crystalline dielectric films,[65] compound semiconductor epitaxial layers, layers for band-gap engineering,[66][67] and very extensively for high T c superconducting films.[68][69] The reader is referred to an excellent review by Cheung and Sankur.[70]

Evaporation

193

In this technique, material is vaporized and ejected from the surface of a target as it irradiated by a laser beam. Films are formed by condensing the material ablated from the target onto a solid substrate. Absorption characteristics of the material to be evaporated determine the laser wavelength to be used. To obtain the high power density required in many cases, pulsed laser beams are generally employed. Pulse width, repetition rate, and pulse intensity are selected for specific applications. In some studies on YBCO film deposition, the laser version of a plasma-assisted reactive evaporation process was used. Oxygen was bled into the system and a plasma was created in the target-substrate space by the use of a positively biased electrode placed some distance above the target. This is the ARE process geometry developed earlier and described in Sec. 9.6 Although laser evaporation is an attractive approach for synthesis of high purity metal alloys and compound films, it suffers from the following limitations: 1. Complex transmitting and focusing systems need to be employed to direct the beam from the laser located outside the vacuum system onto the evaporant placed inside the system. This involves special optical path designs and increases the cost of the set-up. Also, a window material which efficiently transmits the wavelength band of the laser must be found and mounted in such a way that it is not rapidly covered up by the evaporant flux. 2. It is not always possible to find a laser with wavelength compatible with the absorption characteristics of the material to be evaporated. 3. Energy conversion efficiency is very low—usually around 1 to 2%. 4. The size of the deposited film is small (10 to 20 mm, or 0.4 to 0.8 in., diameter), resulting from the small size of the laser impact spot. 5. The “splashing effect,”[69] which involves the production of microparticles between 0.1 and 10 µm in size, diminishes film quality. The main advantages of this technique are: 1. the production of high-energy species which enhances film quality.

194

Deposition Technologies for Films and Coatings 2. Excellent transfer of stoichiometry between the target and the film, e.g., the deposition of hydroxyl apatite thin films for biomedical applications such as implants.

The macroparticle density can be decreased by lowering the power level at the expense of deposition rate. The latter may not be important for many thin film applications. The question of large area deposition has been recently addressed by Greer.[71] He has constructed a vacuum deposition system in which the laser beam is scanned on a rotating YBCO target and the substrate is itself rotated. This rather complex apparatus is capable of depositing YBCO films onto two or three inch diameter substrates.

8.0

DEPOSITION RATE MONITORS AND PROCESS CONTROL

The properties of deposits are dependent on the control exercised during the process. The thinner the deposit, the more critical is the control of the operation. 8.1

Monitoring of the Vapor Stream

Ionization Gauge Rate Monitor: This device is very similar to a hot cathode ionization gauge and monitors the atom density in the vapor phase by ionizing the vapors, collecting and measuring the ion current. Several arrangements are shown in Fig. 4.25. Particle Impingement Rate Monitors: The gauge which is a cylinder suspended by a wire or riding on a bearing is imparted a momentum by the impinging particles which can be measured by the torsional forces. They are illustrated in Fig. 4.26. Ion Current Monitor for Electron Beam Heated Source:An electron beam heated molten pool has a plasma sheath above it. Positive ions from the plasma follow a very similar trajectory as the electrons with a slightly larger radius of curvature, due to their higher mass, and are beamed away from the molten pool by the same magnetic field which bends the electrons towards the pool. Therefore an ion collector can be placed so as to intercept this ion beam and the resultant ion current can be used in a feed-back loop to control the evaporation rate. Two manufacturers of electron beam guns have offered this option.

Evaporation

195

Figure 4.25. Ionization rate monitor designs and arrangements. (A) After Schwarz. (B) After Giedd and Perkins. (C) After Perkins. (D) After Dufour and Zega. (From Handbook of Thin Film Technology. © 1970, McGraw-Hill, Used with permission of McGraw-Hill Book Co.)

Spectroscopic Methods: Monitoring and control of the deposition rate can be done on the basis of mass spectrometry, atomic absorption spectrometry and electron emission impact spectrometry. Each of them involves the choice of an appropriate materials-selective sensor. The principles, advantages and limitations of each of these are presented in a good review paper by Lu in Thin Solid Films 45:487 (1977). The reader is referred to this paper and the references cited therein.

196


Figure 4.26. Particle-impingement-rate monitors. (A) Torsion-wire device. (After Neugebaur.) (B) Pivot-supported device. (After Beavitt.) (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

8.2

Monitoring of Deposited Mass

Microbalances: There are various types of devices which measure a change in mass due to condensed atoms based on elongation of a thin quartz-fiber helix, the tension of a wire or the deflection of a pivot-mounted beam. Examples are shown in Figs. 4.27 and 4.28. Crystal Oscillators: The crystal oscillator monitor utilizes the piezoelectric properties of quartz. The resonance frequency induced by an AC field is inversely proportional to crystal thickness. In practice, the change in frequency of a crystal exposed to the vapor beam is compared to that of reference crystal. An example is shown in Fig. 4.29. 8.3

Monitoring of Specific Film Properties

In preparing thin films, often only one property is of interest, e.g., optical or electrical. Optical Monitors: They measure phenomena such as light absorbence, transmittance, reflectance or related interference effects during film deposition. An example is shown in Fig. 4.30. Resistance Monitors: The film thickness can be continuously monitored using in situ resistance measurements as shown in Fig. 4.31.

Evaporation

197

Figure 4.27. (A) Schematic drawing and (B) circuit diagram of a microbalance constructed from a micro-ammeter movement. (Hayes and Roberts.) (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

Figure 4.28. Microbalance with torsion-fiber suspension and electromagnetic force compensation at beam end (Mayer et al.) (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

198


Figure 4.29. Oscillator crystal holders for deposition monitoring. (A) After Behrndt and Love. (B) After Pulker. (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

Figure 4.30. Schematic of an RF sputtering system (after Davidse and Maissel) with optical-thickness monitor. (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

Evaporation

199

Figure 4.31. Wheatstone-bridge circuit for resistance monitoring. (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGrawHill Book Co.)

8.4

Evaporation Process Control

Thickness Control: Usually monitoring of an evaporation process is combined with means to control film deposition. Frequently, the only requirement is to terminate the process when the thickness or a thicknessrelated property has reached a certain value. The simplest way is to evaporate a weighed amount of source material to completion. Knowing the emission characteristics of the source will allow the film thickness to be calculated. Alternately, monitoring devices discussed earlier can be calibrated to measure thickness directly. Rate Control: Rate control is a more complex task and involves measuring the signal from a rate monitor and using it in a feedback loop to control the power to the source and hence its temperature and evaporation rate. Table 4.5 illustrates the pros and cons of various evaporation process control methods.

200

Table 4.5. Evaporation Process Control


From Handbook of Thin Film Technology. Copyright © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Company.

Evaporation 9.0

201

DEPOSITION OF VARIOUS MATERIALS

The family of materials which are deposited by evaporation include metals, semiconductors, alloys, intermetallic compounds, refractory compounds (i.e., oxides, carbides, nitrides, borides, etc.) and mixtures thereof. An important point is that the source material should be pure and free of gases and/or inclusions to forestall the problem of molten droplet ejection from the pool commonly called spitting. Let us consider each of the materials. 9.1

Deposition of Metals and Elemental Semiconductors

Evaporation of single elements can be carried out from a variety of evaporation sources subject to the restrictions discussed above dealing with melting point, reactions with container, deposition rate, etc. A typical arrangement is shown in Fig. 4.1 for electron beam heating. As discussed above, either heating method can also be used. These are the simplest materials to evaporate. Fortunately, at this time, it is estimated that 90% of all the material evaporated is aluminum! 9.2

Deposition of Alloys

Alloys consist of two or more components, which have different vapor pressures and hence different evaporation rates. As a result, the composition of the vapor phase, and therefore the deposit, has a constantly varying composition. There are two solutions to this problem—multiple sources and single rod-fed or wire-fed electron beam sources. Multiple Sources: This is the more versatile system. The number of sources evaporating simultaneously is equal to or less than the number of constituents in the alloy. The material evaporated from each source can be a metal, alloy or compound. Thus, it is possible to synthesize a dispersionstrengthened alloy, e.g., Ni-ThO2. On the other hand, the process is complex because the evaporation rate from each source has to be monitored and controlled separately. The source-to-substrate distance has to be sufficiently large (15 inches for 2 inch diameter sources) to have complete blending of the vapor streams prior to deposition, which decreases the deposition rate (See Fig. 4.32). Moreover, with gross difference in density of two vapors, it may be difficult to obtain a uniform composition across the width of the substrate due to scattering of the lighter vapor atoms. Some examples are given in Table 4.6.

202


Figure 4.32. Two-source evaporation arrangement yielding variable film composition. (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

It is possible to evaporate each component sequentially, thus producing a multi-layered deposit which is then homogenized by annealing after deposition. This procedure makes it even more difficult to get high deposition rates. A multiple source arrangement for production of alloy deposits at high rates is not known. Single Rod-Fed Electron Beam Source: The disadvantages of multiple sources for alloy deposition can be avoided by using a single source.[72][73] It can be a wire-fed or rod-fed source; the latter is shown in Fig. 4.33. There is a molten pool of limited depth above the solid rod. If the components of an alloy, A1 B10, have different equilibrium vapor pressures, then the steady-state composition of the molten pool will differ from the feed rod, e.g., A1 B10. Under steady-state conditions, the composition of the vapor is the same as that of the solid being fed into the molten pool. One has the choice of starting with a button of appropriate composition A1B10 on top of a rod A1 B1 to form the molten pool initially or one can start with a rod of alloy A1B1 and evaporate until the molten pool reaches compositions A1 B10. Precautions to be observed are that the temperature and volume of the molten pool have to be constant to obtain a constant vapor composition. A theoretical model has been developed and confirmed by experiment. Ni20Cr, Ti-6Al, Ag-5Cu, Ag-10Cu, Ag-20Cu, Ag-30Cu, Ni-xCr-yAl-xY alloy deposits have been successfully prepared. To date, experimental results indicate that this method can be used with vapor pressure differences of a factor of 5,000 between the components. This method cannot be used where one of the alloy constituents is a compound, e.g., Ni-ThO2 .

Table 4.6. Two-Source Evaporation, Experimental Conditions, and Types of Films Obtained

Evaporation 203

204

Table 4.6. (Cont'd)


Evaporation

205

Figure 4.33. Schematic of direct evaporation of an alloy from a single rod-fed source.

In a recent paper, Shevakin et al.[74] investigated the relationship between the composition of the evaporant material and the condensates for alloy evaporation using electron beam evaporation techniques. They used this method to determine thermodynamic activities of the components of binary alloys at temperatures above the melting point of the alloy. 9.3

Deposition of Intermetallic Compounds

Intermetallic compounds which are generally deposited such as GaAs, PbTe, InSb, etc. have as their constituents elements with low melting points and high vapor pressures. These compound semiconductors need to have a carefully controlled stoichiometry, i.e., cation:anion ratio. Therefore, they can best be prepared by flash evaporation or sputtering.

206


In flash evaporation, powder or chips of the two components are sprinkled onto a superheated sheet to produce complete evaporation of both components. Various possible arrangements are shown in Fig. 4.34. Table 4.7 gives examples of the use of this technique.

Figure 4.34. Flash-evaporation mechanisms. (A) Belt feeder. (Harris and Siegel.) (B) Worm-drive feeder with mechanical vibrator. (Himes, Stout, and Thun, Braun and Lood.) (C) Disk feeder (Beam and Takahashi.) (D) Disk magazine feeder. (Marshall, Atlas, and Putner.) (E) Mechanically vibrated trough and cylinder source. (Richards.) (F) Electromagnetically vibrated powder dispenser. (Campbell and Hendry.) (From Handbook of Thin Film Technology. © 1970, McGraw-Hill. Used with permission of McGraw-Hill Book Co.)

Table 4.7. Flash Evaporation of Materials

Evaporation 207

208

Table 4.7. (Cont'd)


Evaporation 9.4

209

Deposition of Refractory Compounds

Refractory compounds are substances like oxides, carbides, nitrides, borides, and sulfides which characteristically have a very high melting point (with some exceptions). In some cases, they form extensive defect structure, i.e., exist over a wide stoichiometric range. For example, in TiC, the C:Ti ratio can vary from 0.5 to 1.0, demonstrating vacant carbon lattice sites. In other compounds, the stoichiometric range is not so wide. Evaporation processes for the deposition of refractory compounds are further subdivided into two types: (i) Direction Evaporation [75] where the evaporant is the refractory compound itself; and(ii) Reactive Evaporation[76] or Activated Reactive Evaporation (ARE)[77] where the evaporant is a metal or a low-valence compound, e.g., where Ti is evaporated in the presence of N2 to form TiN or where Si or SiO is evaporated in the presence of O2 to form SiO2. Direct Evaporation: Table 4.8 gives the experimental conditions for the direct evaporation of refractory compounds. Evaporation can occur with or without dissociation of the compound into fragments. As seen from Table 4.8, the observed vapor specie show that very few compounds evaporate without dissociation. Examples are SiO, MgF2, B2 O3 , CaF2 and other Group IV divalent oxides (SiO homologs like GeO and SnO). In the more general case, when a compound is evaporated or sputtered, the material is not transformed to the vapor state as compound molecules but as fragments thereof. Subsequently, the fragments have to recombine, most probably on the substrate, to reconstitute the compound. Therefore, the stoichiometry (anion:cation ratio) of the deposit depends on several factors including the deposition rate and the ratios of the various molecular fragments, the impingement of other gases present in the environment, the surface mobility of the fragments (which in turn depends on their kinetic energy and substrate temperature), the mean residence time of the fragments of the substrate, the reaction rate of the fragments on the substrate to reconstitute the compound, and the impurities present on the substrate. For example, it was found that direct evaporation of Al2O3 resulted in a deposit which was deficient in oxygen, i.e., which had the composition[78] Al2O3-x. This O2 deficiency could be made up by introducing O2 at a low partial pressure into the environment. In other cases, for example the direct evaporation of TiB2 and ZrB2 , the deposit contains both the monoboride and diboride phases.[79]

210


Table 4.8. Direct Evaporation of Inorganic Compounds

Evaporation Table 4.8. (Cont'd)

211

212


Table 4.8. (Cont'd)

Evaporation 9.5

213

Reactive Evaporation Process

The difficulties involved in direct evaporation processes due to fragmentation of the vaporized compounds are overcome in reactive evaporation where a metal is evaporated in the presence of the reactive gas; the compound is formed by reaction of the evaporated metal species with the molecules of the reactive gas. Though this technique has been extensively used to deposit a variety of oxide films for optical applications, it is generally observed that the films are deficient in oxygen. It is also observed in some cases, especially in the synthesis of carbide films, that the deposition rate becomes a limiting factor governing the growth of the films. In such cases, stoichiometric TiC films could only be deposited at very low rates (~1.5 Å/sec max).[80] This limitation of deposition rate in the case of the reactive evaporation process is due to the reaction kinetics of the compound formation by this process. The presence of a “plasma” in the ARE process influences the reaction kinetics by providing activation energy to the reactive species, thereby making it possible to synthesize compound films at considerably higher rates[82]-[84] and lower temperatures. 9.6

Activated Reactive Evaporation (ARE)

The ARE process generally involves evaporation of a metal or an alloy in the presence of the plasma of a reactive gas.[81][82] For example, TiC and TiN coatings are deposited by this process by evaporating Ti in the presence of C2H2 and N2 plasma respectively. The two basic variants of the ARE process are shown in Figs. 4.35, 4.36. For more information on the ARE process, please refer to a review by Bunshah and Deshpandey.[83] The role of the plasma in this process is two-fold: 1. To enhance the reactions that are necessary for deposition of compound films. 2. To modify the growth kinetics and hence the structure/ morphology of the deposits. In the following section we discuss the above two aspects. Thermodynamic and Kinetic Considerations In Plasma Assisted Deposition Processes. For the formation of a compound by any chemical reaction, the corresponding thermodynamic and kinetic constraints must be satisfied which also apply to the deposition of refractory compound films by reactive evaporation. In order to understand the role of plasma in enhancing the chemical reactions essential for the formation of a particular compound, one has therefore to consider the kinetics of these reactions.

214


Figure 4.35. Schematic of the Activated Reactive Evaporation Process.

Let us consider the reactions involved in the synthesis of some oxides, carbides, and nitrides by reactive evaporation. Given below are the reactions for forming Al2O3, TiC, and TiN. 2Al + 3 /2 O2 → Al2O3

∆G° = -250 kcal (mol O2)-1 at 298 K

2Ti + C2H2 → 2TiC + H2

∆G° = -7.65 kcal (mol C 2H2 )-1 at 298 K

2Ti + N2 → 2TiN

∆G° = -73.5 kcal (mol N2 )-1 at 298 K

Evaporation

215

Figure 4.36. The Activated Reactive Evaporation (ARE) Process[182] using resistance-heated evaporation source.

As can be seen from the above reactions, the thermodynamic criterion of free energy of formation is satisfied for the respective compounds. The reaction kinetics in reactive evaporation process can be treated in exactly the same manner as for reactions occurring in heterogeneous systems of condensed phases. The model for heterogeneous metallurgical kinetics involves: (i) transport of reactant to the reaction interface; (ii) transport of reaction products away from the reaction interface; (iii) the chemical reaction at the chemical interface;(iv) the nucleation of new phase; and (v) heat transfer to or away from the reaction interface. For reactive evaporation, this model may be depicted as follows (e.g., for TiC formation):

216

Deposition Technologies for Films and Coatings Reactants

Products

Ti (metal atoms) C2 H2 (gas)

TiC (deposit) H2 (gas)

Reaction Interface On the basis of the above model, the rate-controlling steps in the reactive evaporation process are: (i) adequate supply of reactants; (ii) adequate collision frequency; (iii) the rate of chemical reactions at the interface; and (iv) the rate of removal of the reaction products from the interface. It is easy to satisfy (i), (ii), and (iv) above for a reactive evaporation process. However, condition (iii), i.e., the rate of reaction, becomes the rate governing step. The “plasma” in the ARE process influences this step, i.e., the rate of reaction, by providing the necessary activation energy to the reactive species. The effect of plasma on rate of reaction can be clearly demonstrated by considering the results of Abe et al.[80] and Bunshah and Raghuram[84] on deposition of TiC coatings. Abe et al. found that titanium carbide with a carbon-to-titanium ratio of 1 could be formed by a reaction between Ti and C2H2 or C2 H4 molecules on a substrate at 300 - 500°C only if the deposition rate is 1 to 1.5 Å/sec. At higher deposition rates, no TiC was formed. Clearly the activation barrier could not be overcome at the higher deposition rates. Bunshah and Raghuram[84][85] have similarly reported that the deposition of TiC by reactive evaporation at higher deposition rates (150 - 200 Å/sec) required a very high substrate temperature, exceeding 1000°C. However, in the presence of plasma, these authors reported that it was possible to deposit TiC at a high rate at a relatively low substrate temperature. The plasma imparts sufficient energy to the reacting species to overcome the activation energy barrier, and hence condition (iii), i.e., the rate of reaction, no longer remains the rate-governing step. Basic Variants of the ARE Process. The two basic variants of the ARE process are activated reactive evaporation with an electron-beam evaporation source[82] and the ARE processes with a resistance-heated source.[86] 1. ARE processes with an electron-beam-heated evaporation source are illustrated in Fig. 4.35. In this process, the metal is heated and melted by a high-acceleration-voltage electron beam that produces a thin plasma sheath on top of the melt. The low energy secondary electrons from the plasma sheath

Evaporation are pulled upwards into the reaction zone by an electrode placed above the pool biased to a low positive DC or AC potential (20 to 100 V), thus creating a plasma-filled region between the electrode and the electron-beam gun. The lowenergy electrons have a high ionization cross section, thus ionizing or activating the metal and gas atoms and increasing the reaction probability on collision. Charge-exchange processes between positive ions and neutral atoms take place in the plasma. In addition, as suggested by Yee, [87] transient highly excited compound species are formed. The formation of the compound is completed most probably on the substrate from these energetic and excited transient species. The synthesis of TiC by reaction of Ti metal vapor and C2H2 gas atoms with a carbon-to-metal ratio approaching unity was achieved with this process.[82][84] Moreover, by varying the partial pressure of either reactant, the carbon-to-metal ratio of carbides could be varied[84] at will. The ARE process has also been applied to the synthesis of all five different Ti-O oxides.[88] These authors noted that in the ARE process (i.e., with a plasma) as compared to the RE process (i.e., without a plasma), a higher oxide is formed for the same partial pressure of O2 , thus demonstrating a better utilization of the gas in the presence of a plasma. The same observation was noted by Bunshah and Raghuram,[84] as well as by Granier and Besson,[89] for the deposition of nitrides. 2. A variation of the ARE process uses a resistance-heated evaporation source. The basic ARE process uses electronbeam-heated sources, which are expensive and inconvenient for the evaporation of low-melting-point high-vapor-pressure materials. Nath and Bunshah[86] modified the ARE process for resistance-heated sources, as shown in Fig. 4.36. The metal vapors are generated from the chamber; the reaction is enhanced by a plasma generated by injecting low energy electrons from a heated thoriated tungsten emitter towards a low-voltage anode assembly. A transverse magnetic field is applied to cause the electrons to go into a spiral path, thus increasing the probability of electron/atom collision and subsequent ionization.

217

218


Modifications of the Basic ARE Process. The ARE process has substantial versatility since the substrate can be grounded, positively or negatively biased, or it can be allowed to float electrically. There are several modifications of the basic ARE process, as illustrated in Fig. 4.37. 1. The Enhanced ARE Process.[90] This is the conventional ARE process using electron-beam heating with the addition of a thermionic electron emitter (e.g., a tungsten filament) for the deposition of refractory compounds at lower deposition rates as compared to the basic ARE process. The low-energy electrons from the emitter sustain the discharge, which would otherwise be extinguished since the primary electron beam (used to melt the metal) is so weak that it does not generate an adequate plasma sheath above the molten pool from which low energy electrons can be extracted by positively biased interspace electrode. The substrate may be biased, grounded or floating. 2. Low-Pressure Plasma Deposition (LPPD) Process. Using electron-beam evaporation sources, the electric field may be generated by biasing the substrate positively instead of using a positively biased interspace electrode. In this case, it is called low-pressure plasma deposition (LPPD) by Nakamura et al.[91] However, this version has a disadvantage over the basic ARE process since one does not have the freedom of choice to ground the substrate, let it float, or bias it negatively (the BARE process—see #4 below). 3. ARE Using Plasma Electron-Beam Guns. The plasma electronbeam gun, instead of the thermionic electron-beam gun, can be used to carry out the ARE process. The hot hollow cathode gun has been used by Komiya et al.[92] to deposit TiC films, whereas Zega et al.[93] used a cold cathode discharge electronbeam gun to deposit titanium nitride films. The plasma e-beam sources produce an abundant supply of low-energy electrons for the ARE-type process. 4. Reactive Ion Plating (RIP) Processes. If the substrate is biased in the ARE process, it is called biased activated reactive evaporation (BARE). This bias is usually negative to attract the positive ions in the plasma. The BARE process has been reinvented and called reactive ion plating by Kobayashi

Evaporation

Figure 4.37. Basic “ARE” process and later variations.

and Doi.[94] Reactive ion plating (RIP) is very similar to the reactive evaporation process in that metal atoms and reactive gases react to form a compound aided by the presence of a plasma. Since the partial pressure of the gases in reactive ion plating are much higher (> 10-2 torr) than in the ARE process (> 10-4 torr), the deposits can become porous or sooty. The plasma cannot be supported by lower pressure in the simple diode ion plating process; therefore, Kobayashi and Doi[94] introduced an auxiliary electrode biased to a positive low

219

220

Deposition Technologies for Films and Coatings voltage (as originally conceived for the ARE process) to initiate and sustain the plasma at low pressure (~10-3 torr). This is no different than the ARE process with a negative bias on the substrate reported[81] much earlier by Bunshah, which was designated by him as the biased ARE (or BARE) process. 5. Another variation of reactive ion plating using a triode configuration[95] involves injection of electrons into the reaction zone between the electron-beam-heated evaporation source and the negatively biased substrate from a heated tungsten filament transversely to the metal vapor path. These lowenergy electrons are pulled across the reaction zone by a positively biased anode located opposite to the cathode. The arrangement is very similar to that shown in Fig. 4.27 except for the use of an electron-beam-heated evaporation source, and is also very similar to the triode sputtering. This adds versatility as well as complexity to the process through the addition of another process variable. 6. Murayama[96] uses an electron-beam-heated source with a negatively biased substrate and RF activation of the reactants by means of a coil electrode of aluminum wire in the reaction zone to deposit oxide and nitride films. 7. ARE Process Using an Arc Evaporation Source. Evaporation of metals using a low-voltage arc in the presence of a plasma and a negatively biased substrate is used by Snaper[52][53] and Dorodnov[97] to deposit nitride and carbide films, with N2 and hydrocarbon reactive gases, respectively.

Recent Developments in the ARE Process. In the last few years, new techniques based on ARE are being developed for synthesis of novel and unique materials. The emphasis of such developments is generally on two aspects: i) new approaches to produce the vapor species, and ii) new plasma excitation and confinement techniques and development of modified plasma excitation geometries. New Approaches to Produce the Various Species. The basic process involves evaporation of the constituent metal alloy or compound using e-beam or resistance/induction heated sources. However, it is difficult to use this approach with certain materials such as boron and carbon. Two possible solutions can be used to overcome these difficulties: i) use a low melting point compound of the respective element, and ii) use

Evaporation

221

a pulsed laser beamwherethe pulse rate and pulse width can be appropriately adjusted to control the rate of material generation and fragmentation. Moreover, in many cases, the energy of the laser beam can also be used as source for plasma excitation. Both of these approaches have been explored. A process developed by Bunshah et. al.[98] for the synthesis of cubic boron nitride involves boric acid as an reactant, which can be easily evaporated from a resistanceheated tungsten boat. In addition to the ease of evaporation, this process also excludes the toxicity problems associated with fine boron particles which can be produced during e-beam evaporation of boron. A similar approach can be extended to evaporation of carbon using a low melting point carbon compound such as adamantine. It is likely that many new materials hitherto difficult to synthesize may possibly be deposited using this routine. Moreover, this novel approach may contribute to further development in reactive MBE processes and other vapor deposition processes involving organometallic compound reactants. The use of pulsed laser beams in an ARE type of process has been demonstrated in recent literature on high Tc superconducting films. Films with high Tc (90 K) and high critical current density (0.7 x l06 A·cm-2 at 77 K) have been produced.[99] It is claimed that pulsing of the laser beam avoids fractionation of the compound and hence good control of film stoichiometry is achieved. It is also suggested that the photon energy is sufficient to activate the reactive gas/metal species thereby increasing their reactivity, leading to increase in oxygen concentration in the deposited films. New Plasma Excitation Modes and Geometries. As discussed earlier, the attributes of the ARE processes are due to the possibility of controlling the plasma parameters independently of the deposition process. However, improvement in excitation and confinement of the plasma, as well as control and optimization of plasma parameters in the ARE processes, are likely to enhance the process capabilities. Recent developments include (i) the use of inductively coupled RF with parallel plate RF geometries, and (ii) the use of multiple filaments and anodes with magnetic confinement. These enhancements have led to substantial improvements in film properties as well as process control. Examples are high rate deposition of a-Si-H films,[100] transparent conducting films on polymeric substrates(101) and TiSx and MoSx[102][103] films with variable x values. Two additional modes of ionization are being explored. Currently an auxiliary RF excitation source similar to that reported by Oeschner[104] is being developed for use in ARE. It is believed that the high electron density

222


and energy selectivity offered by this source is likely to enhance advantages of the ARE processes for compound synthesis. Also, work is underway to integrate Electron Cyclotron Resonance (ECR) excitation at microwave frequencies with the ARE process. ECR plasmas are characterized by a very high level of ionization and excitation, and may greatly enhance the use of ARE for the deposition and synthesis of films. Mechanism of the ARE Process. A reactive evaporation process can be simply written as a reaction between the reactants giving rise to the products. Illustrating this for the deposition of TiC films, one may write xTi (vapor) + CxHy (gas) → xTiC (solid) + yH (gas) In a plasma-assisted deposition process, the reactants dissociate into fractions/radicals and ionic species are produced. Therefore a multiplicity of reaction paths are possible and the overall reaction becomes more complex. Deshpandey, O’Brien, Doerr, and Bunshah[105][106] studied the synthesis of TiC and TiN films, evaporating Ti in a plasma of CxHy gases for the synthesis of TiC films and N2 or NH3 with Ti for the synthesis of TiN films. Several spectroscopies were used to carry out diagnostics on the plasma in the source-substrate volume to determine the species present and the potential reaction paths leading to film formation. Neutral mass spectrometry (MS), plasma mass spectrometry (PMS), and optical emission spectroscopy (OES) were used to examine the nature and relative concentrations of neutral, excited and ionized species present in the process. The main results of these investigations are as follows: 1. Polymerizing reactions producing higher molecular weight hydrocarbon species are dominant in the case of methane. Polymerization increases with increasing flow rate of CH4 for a given electron beam current. The above reactions lead to the formation of relatively soft films containing TiC and graphitic phases. 2. Hard, single-phase TiC films are formed at flow rates of about 50 standard cm3 min-1 C2 H2 for beam currents in the range of 0.2 - 0.3 A. Polymerization reactions do not take place when C2 H2 is used as a reactive gas. Species such as carbon, CH, and CH2 formed in the plasma from the dissociation of C2H2 react with titanium to form TiC. The PMS and MS data indicate the following possible routes for formation of TiC:

Evaporation

223

a. formation of TiC in the plasma volume through reactions such as: Ti + C → TiC Ti + CH2 → TiC + 2H

Ti + CH → TiC + H Ti + CH3 → TiC + 3H

followed by condensation of TiC molecules on the substrate; or b. formation of TixCy or Ti2 Cy or Ti2CyHz complexes in the plasma volume followed by condensation on the substrate to form TiC according to: TixCy → TiC + C

TixCyHz → TiC + CyHz

Present data are not sufficient to determine which of these two schemes is dominant in the formation of TiC. PMS and MS sampling of the arriving flux on the substrate as well as studies with a biased substrate are necessary to resolve this issue. Similar studies on the deposition of TiN films revealed the following: 1. Evaporation of Ti in a N2 plasma showed that the predominant species leading to hard stoichiometric TiN films is 2Ti+ + N2+ → 2TiN. The ratio of Ti+/N2+ in the plasma was 1.05, i.e., close to unity. When this ratio was increased to 1.5, soft films with excess Ti in the deposit were produced. Yee[87] also proposed the same reaction path based on his optical emission spectrographic studies. 2. Evaporation of Ti in an NH3 plasma showed similar results. Under conditions where the Ti+/N2+ ratio was high, the films were soft and titanium rich. With a higher flow rate of NH 3, the N2+ concentration in the plasma was higher and the films were hard. 9.7

Materials Synthesized by Evaporation-based Processes

A variety of metals, alloys, and compounds (oxides, nitrides, carbides, sulfides) have been deposited using evaporation and related processes. In particular, the plasma-assisted variant of the evaporation process, such as activated reactive evaporation, has been successfully used for deposition of a variety of compounds for tribological as well as opto-electronic applications. Recently, a modified process basedon the ARE technique has also proved

224


to be successful in synthesizing c-BN.[98][107] A representative list of the compounds synthesized by the ARE process is given below. In a very recent development, the ARE process has been able to deposit Al2O3 films at very high deposition rates (8 to 12 µm/hr); these rates are 10 to 30 times higher than those by sputter deposition.[108] The compounds synthesized by the ARE process include: Carbides: TiC, HfC, ZrC, VC, W2C, TaC Carbonitride: Ti (C,N) Nitrides: TiN, HfN, ZrN Oxides: TiO2, ZrO2, Al2O3, SiO2 Sulfides: TiS2, MoS2, MoS3 Superconductors: Low Tc: Nb3Ge, CuMo6 S8 High Tc: YBa2CU3O7-8 Photovoltaic Materials: a-SiH, CuInS2 Opto-electric Materials: In(Sn)O2, ZnO Novel Materials: c-BN, Diamond, i-C, a-C

10.0 MICROSTRUCTURE OF PVD CONDENSATES 10.1 Microstructure Evolution PVD condensates deposit as single crystal films on certain crystal planes of single crystal substrates, i.e., by epitaxial growth,[109] or in the more general case, the deposits are polycrystalline. In the case of films deposited by evaporation techniques, the main variables are: (i) the nature of the substrate;(ii) the temperature of the substrate during deposition; (iii) the rate of deposition; (iv) the deposit thickness; (v) the angle of incidence of the vapor stream; and (vi) the pressure and nature of the ambient gas phase. Contrary to what might be intuitively expected, the deposit does not start out as a continuous film one monolayer thick and grow. Instead, threedimensional nuclei are formed on favored sites on the substrates, e.g., cleavage steps on a single crystal substrate; these nuclei grow laterally and in thickness (the so-called island growth stage) ultimately impinging on each other to form a continuous film. Figure 4.38 shows the growth of gold film on rock-salt. The average thickness at which a continuous film forms depends on the deposition temperature and the deposition rate (both of which influence the surface mobility of the adatom) and varies from 10 Å for Ni condensed at 15°K to 1000 Å for Au condensed at 600°K.

Evaporation

225

This familiar model of island growth of a polycrystalline film during the initial stages of deposition illustrates the case where there is limited interaction between depositing atoms and the substrate. This is not always the case.

Figure 4.38. Sequence of micrographs illustrating the effect of increasing deposit thickness of gold on rock salt (x 8000). (After Pashley,[101] with permission.)

Important differences have been observed. Namba and Mori[237] found that by converting a significant fraction (~ 10%) of the vapor flux of Ag to positive ions, epitaxial growth of a single crystal Ag film on a single

226


crystal Ag film on a single crystal NaCl substrate biased to -3,000 V was observed, whereas with vacuum evaporation, the Ag film was polycrystalline. No clear explanation is possible except to note that the mobility of the deposited species is much greater when partially ionized than for neutral vapor specie. The effective surface temperature of the growing film is much higher due to ion bombardment, thus permitting greater surface mobility and resulting in epitaxial growth. Taylor(238) used low energy electron diffraction (LEED) techniques to study the epitaxial deposition of Cu onto a single crystal[196] face of tungsten under ultra-high vacuum conditions. This represents the case where there is appreciable bonding between depositing atoms and the substrate. The deposit on a clean tungsten surface was a uniformly thin[166] Cu film, i.e., no island growth prior to the formation of a continuous film even at thicknesses of 1½ atomic layers. He further observed that chemisorption of even a half monolayer of oxygen severely inhibited epitaxial growth. Sherman, Bunshah, and Beale[119] studied the deposition of thick Mo films onto a rolled Mo sheet substrate as a function of deposition temperature. They observed polycrystalline deposits at all temperatures except in the range of 973° to 1188°K, where the surface oxide MoO3 is unstable and evaporates rapidly, thereby leaving behind a “clean” Mo surface on which epitaxial growth can readily occur aided by the high surface mobility at the elevated deposition temperature. Once a continuous film has formed, the subsequent evolution to the final structure of the thin film is poorly understood at present. It undoubtedly depends on the factors mentioned above which in turn influence the primary variables of nucleation rate, growth rate, and surface mobility of the adatom. The problem has been tackled by Van der Drift[110] and is also the subject of a paper by Thornton.[111] The microstructure and morphology of thick single phase films have been extensively studied for a wide variety of metals, alloys and refractory compounds. The structural model was first proposed by Movchan and Demchishin,[75] Fig. 4.39, and was subsequently modified by Thornton as shown in Fig. 4.40. Movchan and Demchishin’s diagram was arrived at from their studies on deposits of pure metals and did not include the transition zone of Thornton’s model, Zone T, which is not prominent in pure metals or single phase alloy deposits, but becomes quite pronounced in deposits of refractory compounds or complex alloys produced by evaporation, and in all types of deposits produced in the presence of a partial pressure of inert or reactive gas, as in sputtering or ion plating processes.

Evaporation

227

Figure 4.39. Structural zones in condensates. (Movchan and Demchishan.)

Figure 4.40. Structural zones in condensates. (Thornton.)

The evolution of the structural morphology is as follows: At low temperatures, the surface mobility of the adatoms is reduced and the structure grows as tapered crystallites from a limited number of nuclei. It is not a full density structure but contains longitudinal porosity of

228


the order ofa few hundred angstroms width between the tapered crystallites. It also contains a high dislocation density and has a high level of residual stress. Such a structure has also been called “Botryoidal” and corresponds to Zone 1 in Figs. 4.39 and 4.40. As the substrate temperature increases, the surface mobility increases and the structural morphology first transforms to that of Zone T, i.e., tightly packed fibrous grains with weak grain boundaries, and then to a full density columnar morphology corresponding to Zone 2 (Fig. 4.40). The size of the columnar grains increases as the condensation temperature increases. Finally, at still higher temperatures, the structure shows an equiaxed grain morphology, Zone 3. For pure metals and single phase alloys, T1 is the transition temperature between Zone 1 and Zone 2 and T2 is the transition temperature between Zone 2 and Zone 3. According to Movchan and Demchishin’s original model,[75] T1 is 0.3 Tm for metals, and 0.22 - 0.26 Tm for oxides, whereas T2 is 0.45 - 0.4 (Tm is the melting point in °K). Thornton’s modification shows that the transition temperatures may vary significantly from those stated above and, in general, shift to higher temperatures as the gas pressure in the synthesis process increases. It should be emphasized that: 1. The transition from one zone to the next is not abrupt but smooth. Hence the transition temperatures should not be considered as absolute, but as guidelines. 2. All zones are not found in all deposits. For example, Zone T is not prominent in pure metals, but becomes more pronounced in complex alloys, compounds, or in deposits produced at higher gas pressures. Zone 3 is not seen very often in materials with high melting points. The reader is referred to a more extensive description given by Greene in this book in Ch. 13, which includes a discussion of the effects of substrate surface roughness and pressures. Most thick deposits exhibit a strong preferred orientation (fiber texture) at low deposition temperatures and tend towards a more random orientation with increasing deposition temperature. Figure 4.41 shows the evolution of a large-grained columnar morphology in a Be deposit from a much larger number of fine grains which were originally nucleated on the substrate. As growth proceeds, only those grains with a preferred growth direction survive, presumably due to considerations of the minimization of surface energy.

Evaporation

229

Figure 4.41. Photomicrograph of a Be deposit showing the evolution of large columnar grains.

Elegant proof of the importance of surface mobility was also provided by Movchan and Demchishin.[75] Plots of the log of the grain diameter versus the inverse of deposition temperature in Zones 2 and 3 yield straight lines from which activation energies can be computed. It was found that the activation energy for Zone 2 growth corresponded to that for surface selfdiffusion and for Zone 3 growth to volume self-diffusion. The morphological results reported by Movchan and Demchishin for nickel, titanium, tungsten, Al2O3 and ZrO2 have been confirmed for several metals and compounds. The data are given in Table 4.9.[114][124][115][116] Bunshah and Juntz [117] studied the influence of condensation temperature on the deposition of titanium. Their microstructures, shown in Fig. 4.42, agree substantially with those of Movchan and Demchishin for Zones 1 and 2 and T1, the transition temperature between Zones 1 and 2. However, they failed to observe Zone 3 at the temperatures above 700°C found by Movchan and Demchishin.[75] The structure was columnar up to 833°C, which is the α:β phase transformation temperature for titanium. At deposition temperatures above 833°C, the deposit crystallizes as the β phase and on cooling to room temperature, should transform to theα phase, resulting in the typical “transformed-beta” microstructure shown in Fig. 4.42 (900°C deposit), which could be mistaken for an equiaxed microstructure. Hence, the claim of such a transition in structure from Zone 2 to 3 by Movchan and Demchishin for titanium deposits is confusing.

230


Table 4.9. Transition Temperatures between Various Structural Zones

Kane and Bunshah[118] observed the change in morphology in deposited nickel sheet. At 425°C deposition temperature, the deposit showed a Zone 2 morphology, whereas, at 554°C, the deposit showed a Zone 3 morphology. Chambers and Bower[123] studied the deposition of magnesium, copper, gold, iridium, tungsten, and stainless steel. Of the photomicrographs presented, gold and magnesium showed Zone 2 columnar morphology at the appropriate substrate temperatures. Figure 4.43 shows surface and cross section photomicrographs of a Ni-20Cr sheet deposited by Agarwal, Kane and Bunshah.[124] At 950°C, 760°C, 650°C, and 427°C deposition temperatures, the surface and cross section showed an equiaxed Zone 3 morphology. Mah and Nordin[121] found that the Movchan-Demchishin model was obeyed by beryllium also. They observed structures corresponding to all three zones with transition temperatures as predicted by the model.

Evaporation 231

Figure 4.42. Structure of titanium deposits at various substrate temperatures (Bunshah and Juntz).

232


Figure 4.43. Photomicrographs of typical Ni-20Cr deposits at various substrate temperatures. (Agarwal, Kane, and Bunshah.)[124]

Evaporation

233

Neirynck, Samaey and Van Poucke[125] studied the influence of deposition rate and substrate temperature on the microstructure, adhesion, texture, and condensation mechanism of aluminum and zirconium coatings on steel substrates and wires in batch and continuous-coating methods. Kennedy[120] showed a change in morphology from columnar to equiaxed in Fe and Fe-10Ni alloy with higher deposition temperature. Deposits of Fe-1%Y which is a two phase alloy, showed columnar morphology only, the structure becoming coarser at higher deposition temperature. The second phase appears to nucleate new grains so that the grain size in Fe1%Y alloys is much finer than that of iron. The microstructure of copper-nickel alloys[122] produced by codeposition from two sources showed a single phase, as might be expected for this system, which shows a complete solid solubility. On the other hand, sequential deposition of Cu and Ni from two sources shielded from each other onto a rotating substrate produced a microlaminate structure in the deposit where the laminate size can be varied from 0.01 to 40 µm by adjusting the deposition parameters.[239] Similar structures were also developed in the Fe-Cu[239] and in the Ti-B4C system.[239] ln alloy systems showing the presence of several phases, e.g., Ni-B and Cr-Si, the deposits showed the phases present corresponded to those expected from the diagram.[122] Smith, Kennedy, and Boericke[122a] studied the deposition of the two phase (α+β) type Ti-6Al-4V alloy deposited from a single rod-fed source. The microstructure was very similar to wrought material with the same characteristic α+β morphology present on a finer scale in the deposited material. Dispersion-strengthened alloys produced by co-deposition from multiple sources have also been produced. Paton et al.[122] produced Ni-TiC, Ni-NbC and Ni-ZrO2 alloys. The particle size increases from 100 to 1000 Å by changing the deposition temperature from 350° to 1000°C. The size of the dispersed carbide phase particles increased on annealing at 1000° to 1100°C due to their slight solubility in nickel. On the other hand, the size and distribution of ZrO2 dispersion remained constant even after exposure at 1300°C for 5 hours as shown in Fig. 4.44. Movchan, Demchishin, and Kooluck[126] produced Fe-NbC and Fe-NiNbC dispersion strengthened alloys by co-evaporation. The microstructure exhibited columnar morphology, with the inclusion of a fine dispersion of NbC particles.

234


Figure 4.44. Microstructure of dispersion strengthened Ni-ZrO2 alloy before and after exposure at 1300°C for 5 hours (Paton, Movchan, and Demchishin).[122]

Evaporation

235

Raghuram and Bunshah[127] studied the microstructure of TiC deposits from 500° to 1450°C shown in Fig. 4.45. They observed the transition from the tapered crystallite (Zone 1) to columnar structure at 973°K, or 700°C (0.3 Tm). The highest deposition temperature (1450°C) used by these investigators was not sufficient to produce an equiaxed structure although this temperature corresponds to 0.51 Tm.

Figure 4.45. Structure of TiC deposits at various substrate temperatures (Raghuram and Bunshah).[127]

236


The energy of the depositing beam of atoms can be increased if some of them are ionized. It has been shown by Smith[23] that a small fraction of the vaporized species from an electron beam heated source is ionized due to collisions with electrons in the plasma sheath above the molten pool. Bunshah and Juntz[128] biased the substrate to -5,000 V during the deposition of beryllium at 570°C and found that the columnar grain size was markedly refined by the ion bombardment as compared to the grain size produced without biasing the substrate at the same deposition temperature. It may be postulated that the ion bombardment causes a localized increase in temperature at the surface where deposition is occurring, thus causing a higher nucleation rate and a finer grain size. Similar results have been reported for tantalum.[129] The use of hollow cathode gun intensifies the degree of ionization of the vapor species, resulting in a marked increase in kinetic energy of the vaporized atoms.[130] The effects of substrate bias are, therefore, easier to observe. Increasing the substrate bias results in a change in morphology from columnar to fine, equiaxed grains for silver deposited on beryllium and stainless steel,[131] and for silver and copper deposited on stainless steel.[132] On the other hand, the presence of a gas at high pressures (5 to 20 µm) results in a net decrease in kinetic energy of the vaporized atoms due to multiple collisions during the transverse from source to substrate. This degrades the microstructure to loose columnar grains[132] and eventually to an agglomerate of particles. (This, in fact, is a way to produce fine powders by evaporation and subsequent gas-phase nucleation and condensation.) The negative effects of the presence of a high gas density on the kinetic energy and the mobility of adatoms on the deposit surface can be overcome by either biasing the substrate[132][133] and/or heating the substrate to a higher temperature.[134] 10.2 Texture The texture of evaporated deposits is, in general, dependent on deposition temperature. At low deposition temperatures, a strong preferred orientation is generally observed: {211} in iron,[120] {220} in TiC,[127] and {0002} in Ti.[135] As deposition temperature increases, the texture tends to become more random. In the case of beryllium,[114] the texture changed to a {110} orientation at high deposition temperatures. The presence of a gas tends to shift the preferred orientation to higher index planes.[136] For silver, increasing the substrate bias changes the preferred orientation from {111} to {200} and back to {111}.[121]

Evaporation

237

10.3 Residual Stresses Residual stresses in deposits are of two types. The first kind arises from the imperfections built in during growth (the so-called growth stresses). An increase in deposition temperature produces a marked decrease in the magnitude of this stress.[127][137] The other source of residual stress is due to the mismatch in the coefficient of thermal expansion between the substrate and the deposit. Its magnitude and size depend on the values of the thermal expansion coefficients as well as the thickness and size of the substrate and deposit. The influence of a negative bias on the substrate produces a compressive stress in the deposit, which reaches a maximum value at -200 to -300 V DC bias and then decreases.[133] High residual stresses can cause plastic deformation (buckling or bending), cracking in the deposit or the substrate, or cracking at the substrate-deposit interface. The latter can be minimized by grading the interface, i.e., producing the change in material over a finite distance instead of producing it abruptly at a sharp interface. A graded interface can be produced by gradually changing the deposition conditions or by interdiffusion, which is enhanced by higher substrate temperature or increased kinetic energy of the vapor species. 10.4 Defects Let us next consider the “defects” found in vapor-deposited materials. The first one is classified as a spit, or small droplet ejected from the molten pool, which lands on the substrate and is incorporated into the coating.[138] An example is shown in Fig. 4.46. The composition of the droplet is different from that of the coating in the case of an alloy and can therefore be the site of corrosion initiation. The bond between the droplet and the surrounding material is usually poor. Hence, corrosion attack can proceed down the boundary to the substrate or undermine the coating. The spit may also fall out, leaving a pinhole behind which can act as a stress concentrator and limit the ductility or the uniform elongation of a sheet material. Spits or pinholes do not affect the yield strength or reduction of area in a ductile material, but they can be stress raisers and sites for fatigue-crack initiation. Both spits and foreign particles on the substrate surface induce preferential growth of the deposit in that area because of higher exposure to the vapor flux than the general growing interface. This region of preferential growth is termed a flake; typical flakes are shown in Fig. 4.47. There is marginal bonding between the flake and the deposit, which can lead to formation of a pit or crack, or to nucleation of corrosive attack.

238 Deposition Technologies for Films and Coatings Figure 4.46. Vapor source droplet (spit) defect in M-Cr-Al-Y coatings. (a) and (b) show defects overcoated with additional material. (c) fatigue crack initiated at spit (Boone et al.). (Courtesy of Amer. Inst. of Physics.)

Evaporation

239

Figure 4.47. Flake defects in (a) and (b) produced by accelerated coating deposition on foreign particles. Glass bead peening incorporates flake into the coating (c) or knocks it out and forms a pit (d) (Boone et al.). (Courtesy of Amer. Inst. of Physics.)

Spits can be suppressed by eliminating porosity, oxide inclusions and compositional inhomogeneities in the evaporant source material, since spitting can be caused by included-gas release or by the release of bound gas through thermal decomposition. In electron-beam evaporation, the beam of electrons dissipates energy over a path extending as much as a mil (25 µm) or more into the melt. If this energy is delivered at a rate faster than the coating material can accommodate by evaporation, conduction, or radiation, a pocket of vapor forms and spitting occurs. Spits are also caused by gas pockets included in the evaporant rod that suddenly expand when rapidly heated by the beam. Nonmetallic inclusions also can trap pockets of superheated vapor below them, which can erupt in a shower of molten droplets. Spits can be avoided by using a high purity vacuum melted rod as the evaporant. Flake formation can be avoided by avoiding the presence or impingement of foreign particles on the

240


substrate (primarily by substrate surface cleaning and good housekeeping of the deposition apparatus). Deep grooves or ridges on the substrate can also produce flake-type defects by shadowing adjacent regions of the specimen surface. Another type of defect occurs in complex alloys[138] such as M-Cr-AlY (where M can be nickel, cobalt or iron), where even at deposition temperatures of 955°C, the deposit morphology corresponds to the fibrous transition zone between Zone 1 and Zone 2. The grain boundaries in this morphology are weak, causing intergranular corrosive attack (see Fig. 4.48). The problem can be obviated by increasing the adatom mobility through the use of a higher substrate temperature or specimen bias of about -200 V, or by using a post-coating process that consists of a room temperature high intensity glass bead peening followed by a high-temperature anneal in hydrogen. Compound rotation of the specimen, which exposes higher surface irregularities to varying angles of impingement of incoming vapor atoms, produces a significant decrease in the number and size of open, columnar defects.

Figure 4.48. SEM photomicrograph of impact fracture surface of as-deposited overlay coating. Fracture is intercolumnar indicating weak boundaries (Boone et al.). (Courtesy of Amer. Inst. of Physics.)

Another problem in deposits of complex alloys is due to the variation in deposit chemistry attributable to segregation in the ingot and large pool temperature variations caused by the finite size of the electron beam.[138][139] Improved ingot quality, development of improved electron beam sources,

Evaporation

241

and decrease in the temperature gradient at the crucible walls by using crucible liners or coolant of lower heat capacity, such as NaK, instead of water would minimize this problem. In a more recent investigation on the origin of defects and continuing on the above,[140] it was found that spits in M-Cr-Al-Y type alloys consist of ejected pool material exhibiting enrichment in impurity elements of low vapor pressure as a result of superheating of non-metallic particles (carbides or oxides) in the melt initiating the ejection of pool material. Flakes, generally cone shaped, were found to originate at non-metallic particles loosely attached to the surface. Leader formation was found to be weakly dependent on the angle of incidence of the arriving vapor flux. Both flakes and leaders seem to be enhanced by preferential growth and shadowing phenomena.

11.0 PHYSICAL PROPERTIES OF THIN FILMS The Handbook of Thin Film Technology[7] contains an extensive section on the electrical and electronic conduction, piezoelectric and piezoresistive, dielectric and ferromagnetic properties of thin films. The reader is referred to it.

12.0 MECHANICAL AND RELATED PROPERTIES 12.1 Mechanical Properties Mechanical Property Determination:A number of testing techniques have been used to determine the strength properties of thin films. They include the high speed rotor test,[141] the bulge test,[142]-[146] microtensile testing machines of the soft[147]-[150] and the hard categories[142]-[146] and even fixtures which can be operated in the electron microscope.[155][156] Hoffman[157][158] has reviewed the test techniques and the reader could do no better than to read Hoffman’s article or the original references. The basic handling problem encountered with the preparation and mechanical property testing of thin film specimens is much less severe with thick films for which many of the standard test specimens, machines, and techniques can be readily used. Therefore, the spectrum of mechanical properties measured on thick films is much broader than with thin films.

242


Tensile Properties of Thin Films:The tensile properties of thin films have been reviewed.[154][157]-[159] As Hoffman concludes,[157] the data reported are not very consistent even on the same material. The reader is advised to consult the references for details. In general, the observed strength of vapor-deposited metal films consists of three parts: σOBS = σBulk + σImperfections + σThickness whereσBulk is the inherent strength level of bulk polycrystalline material in the annealed state,σImperfections is the contribution due to point defects in excess of those normally found in the bulk annealed state and σThickness is the contribution arising from the smallest dimension of the film and its limiting effect on grain size such that dislocation multiplication and migration are impeded.[149] Table 4.10 gives the strength properties of thin films of some metals and compares them to bulk values.[158] In many cases, the strengths are about 200 times those of annealed bulk samples and 3 to 10 times those of hard drawn samples. The tensile strength values are given numerically as well as by fractions of the shear modulus. The ductility of the high strength films is very limited, which is similar to the behavior of high strength fibers or whiskers. A principal point of contention is whether the ultimate tensile strength is a function of the film thickness or not. The discrepancy also appears to be dependent on the test method used, i.e., between the bulge test and tensile test. In many cases, it appears that the strength decreases as the film thickness increases from approximately the 200 - 300 Å range to about 2000 - 4000 Å range. At the greater thickness, the strength is about the same as that of heavily worked bulk material. There are several papers relating the strength properties of thin films to the “crystallite size” and “block structure” as influenced by the deposition temperature, stress, recovery, and recrystallization process.[160]-[169] One manifestation of this is the phenomenon of creep or plasticity in room temperature tensile tests as exhibited by an irreversible initial loading curve but almost reversible unloading and reloading curves as long as the previous stress level is not exceeded. An example of this is shown in Fig. 4.49 from Neugebauer[148] as the change in slope of the stress-strain curve. The possibility that this change in slope is related to an elastically soft measurement or to creep in the cementation of the grips cannot altogether be discarded.

Evaporation Table 4.10. Strength of Properties of Thin Films

Figure 4.49. Typical stress-strain curve for thin film.

243

244


Long term creep rates have been measured and for gold they vary from 10-7 to 10-4 min-1 depending on load, dimensions and the amount of prestrain.[148] The estimates of the relative elastic and plastic extension at fracture vary from completely elastic to an almost even mixture of elastic and plastic deformation. Fracture in ductile gold single crystal films[154] results from a localized plastic deformation with resultant thinning of the film and a rise in stress level. Eventually the smaller cracks formed in this manner join to cause fracture. The dislocations—necessary for the deformation—are not the grown-in dislocations but those which nucleate and multiply in discontinuous regions. Most observations show no necking prior to fracture. The maximum stress appears to correspond to that needed to propagate cracks from flaws existing in the specimen. In polycrystalline nickel, the fracture is the “cleancleavage” type.[149] Mechanical Properties of Thick Condensates and Bulk Deposits: Table 4.11 lists the mechanical properties of thick deposits of metals, alloys, refractory compounds, and laminated structures. In many cases, the mechanical test data are quite extensive showing yield strength, ultimate tensile strength, hardness, and ductility as a function of grain size, deposition temperature, and test temperature. One of the features of the data is that the properties of thick deposits of metals and alloys are very similar to those of wrought materials which are produced by the conventional processes of melting, casting, mechanical working, and heat treatment. We consider each type of material separately since the behavior of metals and alloys is vastly different from that of refractory compounds. The early work in this area was that of Bunshah,[17][18] Bunshah and [22] Juntz, and Smith[23] who deposited thick films of Be, Ti, and Cu, respectively, and measured mechanical properties. In 1965, Palatnik and coworkers published a paper on mechanical properties of Al condensates.[160] It is impossible to review in detail all the papers. The pertinent data are shown in Table 4.11 and the discussion below concentrates on the highlights. Tensile Properties and Hardness of Metal and Alloy Deposits. Movchan and Demchishin studied the tensile properties and microhardness of Ni, Ti, and W condensates produced at various deposition temperatures. No tensile tests were performed on specimens deposited in Zone 1 (Fig. 4.49). Tests on specimens deposited in Zone 2 showed high strength and low ductility at low deposition temperature. The strength decreased and the ductility increased with deposition temperature. The strength and ductility values of specimens deposited in Zone 3 showed approximately the same

Table 4.11. Mechanical Properties of Thick Films or Bulk Condensates

Evaporation 245

246

Table 4.11. (Cont'd)



Evaporation 247

248



Evaporation

249

values as for recrystallized specimens produced from wrought material. The microhardness variation with deposition temperature for Ni, Ti, and W is shown in Fig. 4.50. The tapered crystallite morphology in Zone 1 showed a high hardness much greater than that of annealed metal. The hardness decreased rapidly with increasing deposition temperature to a fairly constant value for Zone 3 morphology which corresponds to the hardness of recrystallized metals.

Figure 4.50. Variation of microhardness with deposition temperature of metals.

Bunshah and coworkers studied the effect of deposition temperature on the grain size, tensile properties, and hardness of Ti,[22][117] Ni,[118] Nb, V, Mo,[119] and Ni-20Cr[124] alloys for deposits made in Zones 2 and 3. They found that increasing deposition temperature produced larger grain size, lower strength, higher ductility, and lower hardness. Even at the lowest deposition temperature in Zone 2, the ductility was good (>20% RA for 1 µm grain diameter Ti at a yield strength of 56,000 psi). Moreover, they found that both the yield strength and hardness varied as the inverse square root of grain diameter, i.e., followed the Hall-Petch relationship[192][193] which is σys = σo + kd-½ where σys is the yield strength, d is the grain diameter, and σo and k are constants. Figure 4.51 shows an example of this relationship for Ni-20Cr alloy.

250


Figure 4.51. Yield stress vs. inverse square root of average diameter for Ni-20Cr alloy at 25°C. ∆ - wrought; O - deposited; # - Wilcox et al.; $ - Webster. (J. Vac. Sci. Technol., Vol. 12, No. 2:662 (1975), Refs. 12 and 13).

For all these metals and alloys, the yield strength, ductility and hardness values correspond to those of the same materials produced by casting, mechanical working, and recrystallization. The variation of yieldstrength and hardness with grain size, i.e., Hall-Petch type relationships, were also very similar between the deposited and wrought materials, small variations being ascribable to differences in grain morphology and preferred orientations. The Ni-20Cr alloy showed good strength at 1000°C and also obeyed the Hall-Petch relationship. The Hall-Petch relationship is also obeyed by thick films of Cu and Ag to grain-sizes as small as 0.05 µm as shown by Nenioto, Jumbou and Suto.[194] Thus, these thick deposits behave as true engineering materials. Chambers and Bower[195] studied the mechanical properties of 18-8 stainless steel, gold, and magnesium, and showed that their tensile properties were very similar to their wrought counterparts. Smith, Kennedy, and Boericke[122a] studied the (α + β) type Ti-6Al-4V alloy. They showed that the tensile properties are very similar to the wrought material except for a much smaller value in percent elongation due to

Evaporation

251

premature onset of plastic instability in a tensile test at pinholes in the deposited samples. The bend ductility was, however, superior to the wrought material. Shevakin et al.[74] studied the strength and hardness of aluminum and copper condensates as a function of the deposition parameters. They found that the mechanical properties varied widely with changes in process parameters. The deposited materials also showed higher strength and plasticity than the same materials conventionally fabricated, i.e., casting followed by the neo-mechanical treatments. They also found that the hardness values obeyed the Hall-Petch relationships. Paton, Movchan, and Demchishin[122] showed that it is possible to produce thick deposits of all the alloys across the Cu-Ni system and that the mechanical properties vary systematically with composition as would be expected. Dispersion-Strengthened Alloy Deposits. The first data on dispersion-strengthened alloys produced by evaporation methods was reported by Paton, Movchan, and Demchishin[122] who showed that Ni-ZrO2 alloys produced by co-evaporation from two sources contained ZrO2 particles in the size range of 150 - 3000 Å by changing the deposition temperature from 650° to 1100°C. They also showed that the creep strength at 1000°C increased with volume fraction of zirconia. These alloys showed remarkable stability in the microstructure and mechanical properties even after creep exposures of 5 hours at 1300°C. Subsequently, Movchan and coworkers studied the structure and properties of Ni-ZrO2 alloys,[196] and Fe with Al2 O3, ZrO2, ZrB2, TiB2, NbC, or TiC second phases.[197] The alloys were produced by co-evaporation of the constituents from electron-beam heated evaporation sources. One of the very striking effects of the incorporation of a dispersed phase in an evaporated metallic coating is a very pronounced refinement in grain size, often by a factor of 10 to 100 or more, and the inhibition of grain growth at elevated temperatures. This was first reported by Kennedy[120] for the incorporation of Y2O3 dispersions in Fe condensates. It was also observed by Majumder[204] for Cu-Al2O3 deposits and by Jacobson et al.[224] in Ni-Al2 O3 deposits. In a very recent paper, Movchan et al.[75] show the grain size reduction in the Ni-Al2O3, Fe-ZrO2, Fe-ZrB2, and Fe-NbC deposits. The most intense grain refining effect is observed at low volume fractions (0.5 vol.%) of the second phase. Of particular interest to this topic is a subsequent paper by Majumder[205] showing the strong effect of alumina content in increasing creep strength,

252


which confirms the model proposed by Mott[225] who suggested that the ideal creep-resistant material is one with a fine grain size in which the grain boundaries are filled with some substance, say a refractory oxide, to inhibit the motion of grain boundaries. Perhaps the most interesting result from Movchan’s work[196][197][226] is that the dispersed phase alloys show a maximum in room temperature ductility in the W-ZrO2 system at 1 vol.% ZrO2, in the Fe-Al2 O3 system at 0 3 vol.% Al2O3, and in the Fe-NbC system at 0.1 vol.% NbC. The yield strength and tensile strength do not show such a maximum but monotonically increase with volume fraction of the oxide phase. The significance of this observation lies in the possibility of increasing the ductility of MCrAlY coatings which, in turn, would result in increased resistance to spalling, thermal shock and fracture, thus improving the performance of the coating. One might speculate on reasons for this effect including strain-relaxation sites at particle matrix interface, or at grain boundaries due to the greatly increased grain boundary area, favorable changes in residual stress distribution in the coating possibly due to changes in elastic modulus or strength, increased toughness or crack propagation resistance conferred by the dispersed phase particles, change in crystallographic texture, etc. Movchan, Badilenko, and Demchishin[227] have recently presented a very detailed treatment on the regulation of microstructure and mechanical properties of thick vacuum condensates with the help of dispersed phases. They give a detailed theoretical model of(i) the influence of dispersed phases on grain size; (ii) the size and shape of dispersed particles as affected by deposition parameters; (iii) strength and ductility of two phase condensates as influenced by the grain size, particle size, mean free path, nature of the particle (deformable vs. nondeformable) and particle-matrix adhesion energies; (iv) steady-state creep behavior. The model is then confirmed by the experimental results. As a good illustration of one of these points, Fig. 4.5 shows the difference in strength and ductility vs. volume fraction of second phase when the latter is deformable or nondeformable. For both types of particles, there is a ductility maximum at a particular Dg / l ratio, but the strength behavior is diametrically opposite showing a monotonic increase for a nondeformable particle and a minimum for the deformable particle. Dg is the grain size in the plane perpendicular to the vapor flux direction and λ is the interparticle spacing. This model forms an excellent basis for design of experiments to study the effect of dispersed phases on the structure and properties of MCrAlY alloys.

Evaporation

253

Another fascinating observation by Movchan et al.[227] applies to two phase alloys with deformable particles having a high adhesion to the matrix. The ductility of the alloys exceeds that of the pure matrix material at room temperature by a factor of 1.5 to 2 at a strain rate of 1.67 x 10-3 sec -1 (0.1 min1). At high temperatures, the elongation at fracture exceeds 100%, i.e., superplasticity is developed. Laminate Composites. Laminate composites are attractive and preferable over fibrous composites because of their uniform properties in the plane of the sheet. In comparison to mechanical methods of producing laminate composites, e.g., bonding of sheets or foil, physical vapor deposition techniques are very suited to the production of such composites, particularly if each lamellae is to be very thin (0.01 to 1 µm thickness) in order to improve the strength and toughness of the composite. From theoretical considerations, it may be expected that the mechanical properties of microlaminate composites would follow an adaptation of the well known Hall-Petch relationship.[192][193] (Yield strength or hardness =αd½ where d is a characteristic microstructural parameter such as grain diameter, sub-grain diameter, laminae thickness, etc.). This correlation will be explored later. In another approach, Koehler[241] proposed that a laminate structure which is formed of thin layers of two metals, A and B, where one metal,A, has a high dislocation-line energy and the other metal, B, has a low dislocationline energy, should exhibit a resistance to plastic deformation and brittle fracture well in excess of that for homogeneous alloys. If the dislocation-line energies are so mismatched, the termination of the motion of dislocations in metal B is energetically favored over dislocation propagation across the layer interface into metal A. In the case of thick layers, the dislocations generated in either of the layers will pile-up in B at the A-B interface and thereby provide the stress concentrations needed for premature yield. Therefore, to suppress the generation of new dislocations in the layers, the thicknesses ofA and B must be small. Thus, there is a critical minimum layer thickness required for the generation of dislocations. This model does not take into account a high imperfection content in the laminate layers but assumes that their mechanical properties are similar to bulk annealed materials. Most of the prior work on microlayer condensates was investigated in condensates produced at low deposition temperatures[240][242]-[249] (T <0.3 Tm) thus resulting in a high imperfection content. Moreover, the deposits were very thin (<25 µm in thickness), which makes it very difficult to measure the mechanical properties (particularly ductility) and draw good

254


correlations with theory. The systems investigated were Ge/GaAg, Al/Mg, Be/Al, Al/Cu, Al/Ag, Ni/Cu, Mg/Cu, Al/Al2 O3 . Recently Lehoczky[249] studied the layer thickness dependence of the yield strength of Al-Cu and Al-Ag laminates of thin specimens prepared by alternate vapor-deposition. Below the critical layer-thicknesses required for dislocation generation in the layers, the experimental results are in good agreement with Koehler’s predictions. For layer thicknesses greater than those required for dislocation generation, he has extended the theoretical model to include dislocation pile-up groups. A very recent investigation, on the other hand, by Bunshah et al.[239] used high deposition temperatures (T ≅ 0.4 - 0.45 Tm) where equilibrium structures are formed, and thick specimens (200 to 1,000 µm thickness) containing a very large number of microlayers were produced such that mechanical properties can be easily measured on standard test specimens. Fe-Cu and Ni-Cu microlaminate composites were prepared by sequential deposition from two evaporation sources. Very marked increases in strength were observed, by as much as a factor of 10 as compared to the pure metals and a factor of 5 as compared to the solid solution Cu-Ni alloy of the same composition. The ductility decreased somewhat but was still appreciable (5% elongation) for the highest strength alloys. The strength and hardness values followed the Hall-Petch relationship. Superplastic behavior was observed in Fe-Cu microlaminates when the average grain size of the metal equals the interlammellar spacing (approximately 0.45 - 0.50 µm) at a test temperature of 600°C at a strain rate of 0.005 min-1 . High temperature creep properties of thick Fe/Cu and Ni/Cu microlaminate condensates were studied at 600°C as a function of layer thickness. Steady state creep rate has been found to increase with a decrease in microlayer thickness. Microstructural study of the specimens after creep tests revealed the disintegration of iron and nickel layers in Fe/ Cu and Ni/Cu condensates respectively with the formation of separate inclusions of an oval shape. The creep rate variation in the microlayer condensates is explained with the help of a structural model of high temperature creep. Refractory Compounds. Deposits of refractory compounds, oxides, nitrides, and carbides are very important for wear resistant applications in industry. Their structure and properties are strongly dependent on the deposition process. Their behavior is very different from metals and alloys. It is also very hard to measure the mechanical properties of ceramics by tensile tests similar to those used for metals and alloys because of their brittle nature. A very good test to measure the fracture

Evaporation

255

stress of such brittle coatings is the Hertzian fracture test which measures the fracture stress and the surface energy at the fracture surface.[198] Colen and Bunshah[182] used this test to measure the fracture behavior of Y 2O3 deposits of various grain sizes. Figure 4.52 shows the variation in microhardness with deposition temperature for Al2O3 and ZrO2 from the work of Movchan and Demchishin,(75) showing that the behavior of these oxide deposits is quite different in one respect from that of metals (Fig. 4.47). The hardness falls when the structure changes from tapered crystallites (Zone 1) to columnar grains (Zone 2) as with metals. However, unlike metals, the hardness increases markedly as the deposition temperature rises from 0.3 Tm to 0.5 Tm. The authors attribute this to a more “perfect” material produced at the higher deposition temperature due tovolume processes of sintering. A similar hardness curve was obtained for Y2 O3 deposits. [182]

Figure 4.52. Variation of microhardness with deposition temperature for Al 2O3 and ZrO 2.

Figure 4.53 from the work of Raghuram and Bunshah[127] also shows a very marked increase in microhardness of TiC deposits on going from 0.15 Tm (500°C) to 0.3 Tm (1000°C). The hardness increases for the oxides and TiC with increasing deposition temperature. Both sets of results may be explained by the following concept. Since the strength of ceramics is very adversely effected by growth defects and at the higher deposition temperature, the occurrence of these defects is markedly reduced, the hardness (or strength) increased very significantly. However, it should be noted that the absolute value of the hardness of the oxides is much lower than that of the carbides. Thus thepossibility of a different explanation for the “similar”

256


behavior of these materials, i.e., the hardness increase with the deposition temperature needs to be investigated. The hardness data on sputtered TiC and TiN coatings are quite similar to those produced by evaporation techniques.[186]

Figure 4.53. Variation of microhardness with deposition temperature for TiC.

13.0 PURIFICATION OF METALS BY EVAPORATION Impurities in the deposit can be classified into two types, metallic and nonmetallic. Knowing the composition of the evaporant, the experimental conditions (temperature and time), certain thermodynamic data (vapor pressure and activities in solution), the composition of the vacuum environment during the experiment, and the types of melt-crucible reactions, if any, it is possible to estimate the impurity content of the distillate. The amount of impurity transfer to the vapor phase and hence in the deposit (assuming a sticking coefficient of unity) depends directly on the partial pressures of the impurity and the basis metal. For metallic impurities, one assumes that each impurity behaves independently of the other and, using Rayleigh’s equation, the metallic impurity content of the distillate may be estimated. Experimental verification has been demonstrated by Bunshah for beryllium.[20] The amount of nonmetallic impurity (C, O2, N 2, and H2) is estimated as follows: for example, for oxygen,

Evaporation

ppm (atomic) O2 =

257

∑ υG υM

where ∑υG = sum of the impingement frequencies (number of atoms/cm2/ s) of the various gases and vapors present in the vacuum environment containing oxygen, such as H2, CO, CO2 and MO (metal suboxide), on the substrate, andυM = impingement frequency of metal atoms on the substrate, an experimentally determined parameter. Implicit in this treatment is the assumption that the sticking coefficient for all the species is unity. This assumption is good for reactive gases such as CO, CO2 and H2O but poor for gases such as H2, as has been shown by Bunshah and Juntz[21] for beryllium; they also demonstrated a satisfactory agreement between computed and experimentally observed values for the nonmetallic impurities. Table 4.12 shows the production of very high purity beryllium in sheet form by vacuum melting followed by vacuum distillation. The oxygen content of the distillate is due to suboxide vaporization (Be2 O) from the melt and consequent contamination of the substrate, since the suboxide has a higher vapor pressure than the evaporating species. The oxygen content of refractory metal deposits produced by vacuum evaporation can also be substantially increased by suboxide vaporization from the melt. The suboxide can be that of the deposit itself, e.g., MoO in the case of molybdenum deposition; or that associated with an impurity in the evaporant, e.g., MoO in the evaporation of vanadium. Table 4.12. Purification of Beryllium by Vacuum Melting and Distillation (in parts per million atomic)

258


APPENDIX On Progress in Scientific Investigations in the Field of Vacuum Evaporation in the Soviet Union A. V. Demchishin E. O. Paton Electric Welding Institute Kiev, Ukraine, U.S.S.R

The first investigations dealing with the problems of evaporation and condensation were carried out by Soviet scientists as early as the twenties. Y. I. Frenkel[A1] found theoretically that there exists a critical temperature of reflection of metal atoms from a substrate.[A1] Y. B. Kharitonov and N. N. Semenov have shown experimentally that this phenomenon actually took place.[A2] The problem of formation of chemical compound with a simultaneous condensation of molecular beams of cadmium and sulfur was studied by A. I. Shal’nikov and N. N. Semenov.[A3] Structural studies of condensates of gold-copper alloys by electron and x-ray diffraction were carried out by M. M. Umanskii and V. A. Krylov.[A4] At the beginning of the forties, S. A. Vekshinskii and his colleagues performed a lot of work on a development of methods for production of specimens of condensates, on experimental verification of condensate distribution law, on studying physical and chemical properties of condensed metal films of pure metals and binary alloys.[A5] S. A. Vekshinskii suggested the use of a method of co-condensation of vapor mixtures of several components for producing the films of variable composition thus enabling the structure and properties of an entire n-component system or its part to be studied at once without recourse to production of a great number of separate samples of constant composition alloys. In the middle of the fifties, investigation of condensates was conducted by L. S. Palatnik and his collaborators at the Kharkov Polytechnic Institute towards the following trends: • structure and substructure of thin and massive condensed films; • mechanism of formation and kinetics of growth of continuous and island films; • physical properties of films (mechanical, electrical, semiconductive, magnetic, thermal and other properties);

Evaporation

259

• studying the correlation between structure (substructure) and physical properties of films; • the effect of physical and technological variables of evaporation processes and vacuum condensation on structure (substructure) and physical properties of continuous (thin and massive) and island films. The main results of these investigations are published in Refs. A6, A7, and A8. In addition to the said studies, in the sixties and seventies, the characteristics of macro-, micro- and submicroporosity of condensed films depending on substrate temperature, angle of incidence of molecular flow, condensation rate, film thickness, pressure and composition of residual gas atmosphere were investigated. Mechanisms of porosity formation processes were established and relationships between the porosity characteristics and physical-mechanical properties of films have been studied.[A9][A10] In the middle of the sixties, B. A. Movchan and his collaborators developed an electron-beam technology for production of preparations of condensed systems and commenced the study of thick (up to 1 mm) condensates. In the sixties-seventies the effect of condensation conditions on structure and physical-mechanical properties of thick condensates of pure metals, refractory oxides, carbides, borides and their mixtures, ceramicmetallic materials and dispersion strengthened compositions were investigated. Their main results were published in Refs. A11 to A15.

REFERENCES (for Appendix) A1. Frenkel, J. I., Zeitschr. f. Physik, 26:117 (1924) A2. Chariton, J. B. and Semenoff, N. N.,Zeitschr. f. Physik, 25:287 (1924) A3. Shal’nikov, A. I., Semenov, N. N.,The Journal of Russian Physical and Chemical Society, 60:33 (1928) A4. Umanskii, M. M., Krylov, V. A.,The Journal of Exp. and Theor. Physics, 6:691 (1936) A5. Vekshinskii, S. A., A New Method of Metallographic Studies of Alloys, Bostechizdat, Moscow-Leningrad (1944) A6. Palatnik, L. S., Papirov, I. I., Epitaxial Films, Nauka, Moscow (1971) A7. Palatnik, L. S., Fux, M. Y., Kosevich, V. M., Mechanism of Formation and Substructure of Condensed Films, Nauka, Moscow (1972)

260


A8. Palatnik, L. S., Sorokin, V. K., Fundamentals of Film Semiconductive Materials Technology, Energia, Moscow (1973) A9. Palatnik, L. S., Fux, M. Y., Cheremskoi, P. G., Transactions of the Academy of Sciences of the U.S.S.R., 203(5):1058 (1972) A10. Fux, M. Y., Palatnik, L. S., Cheremskoi, P. G. Toptygin, A. L., Physics of Metals and Physical Metallurgy, 46(1):114 (1978) A11. Movchan, B. A., Demchishin, A. V., Physics of Metals and Physical Metallurgy, 28, No. 4:653 (1969) A12. Paton, B. E., Movchan, B. A., Demchishin, A. V., Proceedings of the Fourth Int'l. Conf. on Vac. Metallurgy, p. 251, Tokyo, (June 4-8, 1973). Published by the Iron and Steel Institute of Japan, Tokyo (1974). A13. Movchan, B. A., Demchishin, A. V., Kooluck, L. D., Thin Solid Films, 44:285 (1977) A14. Movchan, B. A., Demchishin, A. V., Badilenko, G. F.,Strength Problems, No. 2:61 (1978) A15. Movchan, B. A., Malashenko, I. S., Pap, P. A., Problems of Special Electro Metallurgy, Naukova Dumka, Kiev, No. 8:78 (1978)

Evaporation

261

REFERENCES 1. Faraday, M., Phil. Trans., 147:145 (1857) 2. Nahrwold, R., Ann. Physik, 31:467 (1887) 3. Kundt, A., Ann. Physik, 34:473 (1888) 4. Soddy, F, Proc. Roy. Soc. London, 78:429 (1967) 5. Langmeir, I., J. Am. Chem. Soc., 35:931 (1913) 6. Glang, R., Handbook of Thin Film Technology, (L. I. Maissel and R. Glang, eds.), pp. 1-7, McGraw-Hill (1970) 7. Handbook of Thin Film Technology, (L. I. Maissel and R. Glang, eds.), McGraw Hill, (1970) 8. Holland, L. Vacuum Deposition of Thin Films, Chapman & Hall (1956) 9. Science and Technology of Surface Coatings, (B. N. Chapman and J. C. Anderson, eds.), Academic Press (1974) 10. Allen, J. A., Rev. Pure Appl. Chem., 4:133 (1954) 11. Bassett, G. A. and Pashley, D. W., J. Inst Metals, 87:449 (1958) 12. Hoffman, R. W., Thin Films, p.99, Am. Soc. for Metals, (1964) 13. Hoffman, R. W., Physics of Thin Films, 3:246, Academic Press, New York (1966) 14. Buckel, W., J. Vac. Sci. Technol., 6:606 (1969) 15. Kinosita, W., Thin Solid Films, 12:17 (1972) 16. Bunshah, R. F., Physical Metallurgy of Beryllium, Conf. No. 170, Oak Ridge National Laboratory (April 1963) 17. Bunshah, R. F., Materials Science and Technology for Advanced Applications, 2:31, Am. Soc. for Metals (1964) 18. Bunshah, R. F., Metals Engineering Quarterly, p. 8, (Nov. 1964) 19. Bunshah, R. F. and Juntz, R. S., Beryllium Technology, 1:1 Gordon and Breach Science Publishers, (1966) 20. Bunshah, R. F.,Proc. Int’l. Conf. on Beryllium, p. 63, Press Universitaires de France, Grenoble, France, (1965) 21. Bunshah, R. F. and Juntz, R. S., Trans. Vac. Met. Conf, p. 209, Am. Vac. Soc. (1966) 22. Bunshah, R. F. and Juntz, R. S., Trans. Vac. Met. Conf., p. 200, Am. Vac. Soc. (1965)

262


23. Smith, H. R., Materials Science and Technology for Advanced Applications, 2:569, Am. Soc. for Metals (1964) 24. Smith, H. F., Jr. and Hunt, C. D’A., Trans Vac. Met. Conf., p. 227, Am. Vac. Soc. (1964) 25. Meyers, R. F. and Morgan, R. P., Trans. Vac. Met Conf., p. 271, Am. Vac. Soc. (1966) 26. Butler, J. F., J. Vac. Sci. Tech. 7:S-52 (1970) 27. Schiller, J. and Heisig, U., “Evaporation Techniques” (in German), Veb Verlag Technik, Berlin, (1975) 28. Graper, E. P.,J. Vac. Sci Tech.8:333 (1971);J. Vac. Sci. Tech.10:100 (1973) 29. Kennedy, K. D., Schevermann, G. R., Smith, H. R., Jr., Res. Dev. Mag., 22:40 (1971) 30. Beale, H. A., Bunshah, R. F., Proc. 4th Int'l. Conf. on Vac. Met, p. 238, Iron and Steel Institute of Japan, Tokyo, Jpn (June 1973) 31a. Wan, C. T., Chambers, D. L., Carmichael, D. C., ibid, p. 231 31b. Baum, G. A., Report No. RFP-686, Dow Chemical Co., Golden, CO. (Feb. 6, 1967) 32. Thornton, J. A., SAE Transactions, (1973) 33. “Sputtering and Ion Plating,” NASA SP-511 (1972) 34. Hertz, H., Ann. Physik, 17:177 (1882) 35. Knudsen, M., Ann. Physik, 47:697 (1915) 36. Smith, H. R., Proc. 12th Ann. Tech. Conf., pp. 50-54, Soc. of Vac. Coaters, Detroit, MI (1969) 37. Riley, T. C., “The Structure and Mechanical Properties of Physical Vapor Deposited Chromium”, Ph.D. Thesis, Stanford University (Nov. 1974) 38. Bunshah, R. F. and Juntz, R. S., Trans. Vac. Met Conf., p. 799, Am. Vac. Soc. (1967) 39. Chow, R. and Bunshah, R. F., J. Vac. Sci. Tech. 8, VM 73 (1971) 40. Nimmagadda, R., and Bunshah, R. F., J. Vac, Sci. Tech. 8, VM 85 (1971) 41. Szekely, J. and Poveromo, J. J., Met. Trans. 5:289 (1974)

Evaporation

263

42. Smith, H. R. and Hunt, C. D’A., Trans. Vac. Met. Conf., p. 227, Am. Vac. Soc. (1965) 43. Hayashi, C. and Oku, Y., Trans. Vac. Met Conf., p. 257, Am. Vac. Soc. (1966) 44. Cocca, M. A. and Stauffer, L. H., Trans. Vac. Met. Conf., p. 203, Am. Vac. Soc. (1963) 45. Morley, J. R., ibid, p. 186 46. Berghaus, B., German Patent No. 683,414 (1939) 47. Sabalev, L. P., et al., US. Patent 3,783,231 (Jan. 1, 1974); 3,793,179 (Feb. 19, 1974) 48. Dorodnov, A. M., Soviet Phys. Tech. Phys., 23:1058 (1978) 49. Osipov, V. A., et al., Soviet Rev. Sci. Inst., 21:1651 (1978) 50. Wroe, H., Br. J. Appl. Phys., 9:488-491 (1958) 51. Gilmour, A. S., Jr., Lockwood, D. L., Proc. IEEE, 60:No. 8:977-991 (1972) 52. Snaper, A. A., “Arc Deposition and Apparatus,” US. Patent No. 3,625,848 (1971) 53. Snaper, A. A., “Arc Deposition and Apparatus,” US. Patent No. 3,836,451 (1974) 54. Sablev, L. P., “Apparatus for Vacuum Evaporation of Metals under the Action of an Electric Arc,” US. Patent No. 3,783,231 (1974) 55. Sablev, L. P., “Apparatus for Metal Evaporation Coating,” US. Patent No. 3,793,179 (1974) 56. Sanders, D. M., Handbook of Plasma Processing Technology, (S. Rosnagel, J. J. Cuomo, and W. D. Westwood, eds.), p. 419, Noyes Publications (1990) 57. Martin, P. J., et al., Thin Solid Films, 153:91 (1987) 58. Martin, P. J., Netterfield, R. P., and Kinder, T. J., Thin Solid Films, 193/ 194:77 (1990) 59. Vasin, A. I., Dorodnov, A. M., et al., Sov. Tech. Phys, Lett., (Engl. Trans. of Pis’ma Zh. T Fiz.), 5: No. 23-24, (1979) 60. Ehrich, H., Hasse, B., et al., Proc. 8th Intl. Conf. Discharge Appl., 591592, 596, Essen Univ. (1985)

264


61. Dorodnov, A. M., Kunetsov, A. N., et al., Sov. Tech. Phys. Lett., Engl. Trans. of Pis’ma Zh. T Fiz., 5: No. 418-419 (1979) 62. Ehrich, H., J. Vac. Sci. Technol., A6: 134-138 (1988) 63. Meassick, S., Chan, C., and Allen, R., “Thin Film Deposition Techniques using the Anodic Arc,” to be published 64. Smith, H. M. and Turner, A. F., Appl. Opt., 4:147 (1965) 65. Sankur, H., DeNatale, J., Gunning, W., and Nelson, J. G., J. Vac. Sci. Technol., A5:2869 (1987) 66. Cheung, J. T. and Madden, J., J. Vac. Sci. Technol., B5:705 (1987) 67. Cheung, J. T., Chen, J. S., and Otsuka, N., Proc. IRIS IR Detector Specialty Meeting, Seattle, WA (Aug. 1987); This work was followed by several other similar investigations presented at the 34th Nat’l. Symp. of Am. Vac. Soc., Anaheim, CA (Nov. 1987) 68. Dijkkamp, D., Venkatesan, T., Wu, X. D., Shaheen, S. A., Jisrawi, N., Min-Lee, Y. H., Mclean, W. L., and Croft, M., Appl. Phys. Lett., 51:619 (1987) 69. Wu, X. D., Dijkkamp, D., Olgale, S. B., Ina, A., Chase, E. W., Miceli, P. F., Chang, C. C., Tarascon, J. M., and Venkatesan, T., Appl. Phys. Lett., 51:861 (1987) 70. Cheung, J. T. and Sankur, H., Solid State and Materials Sciences, 15:63 (1988) 71. Greer, J. A., J. Vac. Sci. Technol., 10(4):1821 (1992) 72. Nimmagadda, R., Raghuram, A. C., and Bunshah, R. F., J. Vac. Sci. Tech., 9 (1972) 73. Santala, T. and Adams, M., J. Vac. Sci. Tech., 7:s22 (1970). 74. Shevakin, Y. F., Kharitonova, L. D., and Ostrovskaya, L. M.,Thin Solid Films, 62:337 (1979) 75. Movchan, B. A. and Demchishin, A. V., Fizika Metall, 28:653 (1969) 76. Auwarter, M., US. Patent 2,920,002 (1960) 77. Bunshah, R. F. and Raghuram, A. C., J. Vac. Sci. Technol., 9:1385 (1972) 78. Hoffman, D. and Liebowitz, D., J. Vac. Sci Technol, 9:326 (1972)

Evaporation

265

79. Bunshah, R. F., Nimmagadda, R., Dunford, W., Movchan, B. A., Demchishin, A. V., and Chursanov, N. A., Thin Solid Films, 54:85 (1978) 80. Abe, T., Inngawa, K., Obusa, R., and Murakami, Y., Proc. 12th Symp. on Fusion Technol., Julich (1982) 81. Bunshah, R. F., Thin Solid Films, 107:21 (1983) 82. Bunshah, R. F., U.S. Patent 3,791,852 (1972) 83. Bunshah, R. F. and Deshpandey, C., Physics of Thin Films, (J. L. Vossen and M. H. Francombe, eds.), p. 60, Academic Press, New York (1987) 84. Bunshah, R. F. and Raghuram, A. C., J. Vac. Sci. Technol., 9:1385 (1972) 85. ibid., p. 1389 86. Nath, P. and Bunshah, R. F., Thin Solid Films, 69:63 (1980) 87. Yee, K. K., Proc. Conf. Chem. Vap. Deposition, 5th Int'l. Conf., p. 238 (1975) 88. Grossklaus, W. and Bunshah, R. F., J. Vac. Technol., 12:593 (1975) 89. Granier, J. and Besson, J., Proc. Plansee. Sem., 9 (1977) 90. Yoshihara, H. and Mori, M., J. Vac. Sci. Technol., 16:1007 (1979) 91. Nakamura, K., Inagawa, K., Tsuroka, K., and Komiya, S., Thin Solid Films 40:155 (1977) 92. Komiya, S., Unezu, N., and Narasawa, T., Thin Solid Films, 54:51 (1978) 93. Zega, B., Korrmann, M., and Amiquet, J., Thin Solid Films, 54:57 (1977) 94. Kobayashi, M. and Doi, Y., Thin Solid Films, 54:57 (1978) 95. Matthews, A. and Teer, D. G., Thin Solid Films, 80:41 (1981) 96. Murayama, Y., J. Vac. Sci. Technol., 12:818 (1975) 97. Dorodnov, A. M., Sov. Phys. Tech. Phys., 40:211 (1977) 98. Bunshah, R. F., Chopra, K. L., Deshpandey, C., and Vankar, V. D., U.S. Patent No. 4,714,625 (1987) 99. Inam, A., Hegde, M. S., Wu, X. C., Venkatesan, T., England, D., Miceli, P. F., Chase, E. W., Chang, C. C., Taraskaw, J. M., and Watchman, J. B., Appl. Phys. Lett., 53(10):908 (1988)

266


100. Chen, C. Y., Thesis, M. S., Department of Materials Science and Engineering, University of California, Los Angeles, (1987) 101. O’Brien, B. P., Thesis, M. S., Department of Materials Science and Engineering, University of California, Los Angeles, (1987) 102. Zender, D., Deshpandey, C., Dunn, B., and Bunshah, R. F., Proc. of the Fifth Int'l. Conf. on Solid Ionics, Part 1, (J. B. Boyle, L. C. DeJognhe, and R. A. Huggins, eds.), P. 813, North Holland, Amsterdam (1986) 103. Shin, H., Doerr, H. J., Deshpandey, C., Bunshah, R. F., and Dunn, B., Surface and Coating Technol., 39/40:683 (1989) 104. Oeschner, H., personal communication 105. Deshpandey, C., O’Brien, B. P., Doerr, H. J., and Bunshah, R. F., Surface and Coating Technol., 33:1 (1987) 106. Deshpandey, C., O’Brien, B. P., Doerr, H. J., and Bunshah, R. F., to be published, Thin Solid Films 107. Lin, P., Deshpandey, C., Doerr, H. J., Bunshah, R. F., Chopra, K. L., and Vankar, V. D., Thin Solid Films, 153:487 (1987) 108. Yoon, J. S., Deshpandey, C., Doerr, H. J., and Bunshah, R. F.,Surface and Coating Technol., 43/44:213 (1990) 109. Pashley, D. W., Adv. Phys., 5:1973 (1956) 110. Van der Drift, A., Phillips Res. Rep., 22:267 (1967) 111. Thornton, J. A., Ann. Rev. Mater. Sci., p. 239 (1977) 112. Thornton, J. A., J. Vac. Sci. Technol., 11:666 (1974) 113. Bunshah, R. F., Proc. 4th Int’l. Conf. on Vacuum Metallurgy, p. 17, Iron and Steel Inst., Japan (1973) 114. Bunshah, R. F., J. Vac. Sci. Technol., 11:633 (1974) 115. Bunshah, R. F., J. Vac. Sci. Technol., 11:814 (1974) 116. Bunshah, R. F.,New Trends in Materials Processing, p. 200, Am. Soc. for Metals (1976) 117. Bunshah, R. F. and Juntz, R. S., Met Trans., 4:21 (1973) 118. Kane, N., and Bunshah, R. F., Proc. of the Fourth Int'l. Conf. on Vacuum Metallurgy, pg. 242, Iron and Steel Inst. of Japan, Tokyo, Jpn (June 1973)

Evaporation

267

119. Sherman, M., Bunshah, R. F., and Beale, H. A., J. Vac. Sci. Technol., 11:1128 (1974) 120. Kennedy, K., Trans. of the Vacuum Metallurgy Conf., p.195, Am. Vac. Soc. (1968) 121. Mah, G. and Nordin, C. W., Proc. of the 16th Ann. Conf., p. 103, Soc. of Vac. Coaters, Chicago (1973) 122. Paton, B. A., Movchan, B. A., and Demchishin, A. V., Proc. 4th Int'l. Conf. on Vacuum Metallurgy, p. 251, Iron and Steel Inst. of Japan (1973) 122a. Smith, H. R., Jr., Kennedy, K., and Boericke, F. S., J. Vac. Sci. Tech., 7:S48 (1970) 123. Chambers, D. L. and Bower, W. K.,J. Vac. Sci. Technol.,7:S63 (1970) 124. Agarwal, N., Kane, N., and Bunshah, R. F., Nat’l. Vac. Symp., New York (1973) 125. Neirynck, M., Samaey, W., and Van Poucke, L., J. Vac. Sci. Technol., 11:647 (1974) 126. Movchan, B. A., Demchishin, A. V., and Kooluck, L. V., J. Vac. Sci. Technol., 11:640 (1974) 127. Raghuram, A. C. and Bunshah, R. F., J. Vac. Sci. Technol., 9:1389 (1972) 128. Bunshah, R. F. and Juntz, R. S., J. Vac. Sci. Technol., 9:404 (1972) 129. Mattox, D. M. and Kominiak, G. J., J. Vac. Sci. Technol., 9:528 (1972) 130. Morley, J. and Smith, H. R., J. Vac. Sci Technol., 9:1377 (1972) 131. Mah, G., McLeod, P. S., and Williams, D. G., J. Vac. Sci. Technol., 11:663 (1974) 132. Stowell, W. R. and Chambers, D.,J. Vac. Sci. Technol.,11:653 (1974) 133. Bland, R. D., Kominiak, G. J., and Mattox, D. M., J. Vac. Sci. Technol., 11:671 (1974) 134. Bunshah, R. F., unpublished research. 135. Turk, C. F. and Marcus, H. L., Trans. AIME, 242:2251 (1968) 136. Beale, H. A. and Bunshah, R. F., unpublished research

268


137. Thornton, J. A.,New Industries and Applications for Advanced Materials Technology, 19:443, SAMPE, (1974) 138. Boone, D. H., Strangman, T. E., and Wilson, L. W., J. Vac. Sci Technol., 11:641 (1974) 139. Krutenat, R. C., J. Vac. Sci. Technol., 11:1123 (1974) 140. Grossklaus, W., Ulion, N. E., and Beale, H. A.,Thin Solid Films,40:271 (1977) 141. Beams, J. W., Breazeale, J. B., and Bart, W. L., Phys. Rev., 100:1657 (1955) 142. Beams, J. W., Structure and Properties of Thin Films, p. 183, Wiley, New York (1959) 143. Catlin A., and Walker, W. P., J. Appl. Phys., 31:2135 (1960) 144. Jovanovic, S. and Smith, C. S., J. Appl. Phys., 32:121 (1961) 145. Krukover, P. I. and Buravikhin, V. A., Fizika Metall., 22:144 (1966) 146. Brandon, D. G. and Bauer, Z., Israel J. Technol., 8:247 (1970) 147. Marsh, D. M., J. Sci. Instrum., 38:229 (1961) 148. Neugebauer, C. A., J. Appl. Phys., 31:1096 (1960) 149. D’Antonio, C., Hirschorn, J., and Tarshis, L., Trans. AIME, 227:1346 (1964) 150. Blakely, J. M., J. Appl. Phys., 36:1756 (1964) 151. Kuhlmann-Wilsdorf, D. and Raghaven, K. S., Rev. Sci. Instrum., 33:930 (1962) 152. Lawley, A. and Schuster, S., Rev. Sci. Instrum., 33:1178 (1962) 153. Orowan, E., Z. Phys., 82:235 (1933) 154. Menter, J. W. and Pashley, D. W., Structures and Properties of Thin Films, p. 111, Wiley, New York (1959) 155. Wilsdorf, H. G. F., Rev. Sci Instrum., 29:323 (1958) 156. Pashley, D. W., Proc. R. Soc. Lond., A225:218 (1960) 157. Hoffman, R. W., Thin Films, p. 99, Am. Soc. for Metals (1964) 158. Hoffman, R. W., Physics of Thin Films, 3:246, Academic Press, New York (1966) 159. Neugebauer, C. A., Physics of Thin Films, (Edited by G. Hass and E. Thun), Vol. 2, Academic Press, New York. (1964)

Evaporation

269

160. Palatnik, L. S., Ill’inski, A. I., Federov, G. V., and D’yachenko, V. S., Izvestra VUZ, Fizika (Soviet Physica Journal) 1:122 (1966) 161. Grunes, R. L., D’Antonio, C., and Kies, F. K., J. Appl. Phys., 36:2735 (1965) 162. Ya Fuks, M., Belozerov, V. V., and Boyko, Y. F., Fizika Metall.,33:571 (1372) 163. Kinosita, K., Maki, K., Nakamizo, K., and Takenchi, K., Jpn. J. Appl. Phys., 6:42 (1967) 164. Ya Fuks, M., Velozero V. V., and Boyko, Y. F., Fizika Metall., 33:571 (1972) 165. Doljack, F. A. and Hoffman, R. W., Thin Solid Films, 12:71 (1972) 166. Ya Fuks, M., Palatnik, L. S., Belozerov, V. V., Zolotnitsky, Y. V., and Roschchenko, S. T., Fizika Metall., 36:316 (1973) 167. Pines, B. Y. and Tan, N. S., Fizika Metall., 19:899 (1965) 168. Aleksanyan, I. T., Fizika Metall., 25:947 (1968) 169. Hoffman, R. W., Thin Solid Films, 34:185 (1976) 170. Palatnik, L. S., Federov, G. V., Prokhavulov, A. I., and Federenko, A. I., Fizika Metall., 20:574 (1965) 171. Palatnik, L. S., Ya Fuks, M., Boiko, B. T., and Pugacheu, A. T., Soviet Phys. Dokl., English Translation, 8:713 (1964) 172. Oding, A. and Aleksanyass, I. T., Soviet Phys. Dokl., 8:818 (1964) 173. Bunshah, R. F., Vacuum, 27(4):353 (1977) 174. Yamamoto, H. and Kamoshita, G., Trans. Jpn. Inst. Metals, 12:12 (1971) 175. Palatnik, L. S., Ya Fuks, M., Ill’inski A. I., and Alaverdova, O. G., Fizika Metall., 22:744 (1966) 176. Henning, C. A. O., Boswell F. W., and Corbett, J. M., Acta Met.,23:177 (1975) 177. Ziling, K. K., Pkrovskiy, L. D., and Pohelkin, V. Y., Fizika Metall., 29:1112 (1970) 178. Palatnik, L. S., Ill’inski A. I., and Ravlik, A. G., Fizika Metall., 19:310 (1965) 179. Henning, C. A. O., Boswell, F. W., and Corbett, J. M.,Acta Met.,23:187 (1975)

270


180. Palatnik, L. S., Ill’inski, A. I., Ravhk, A. G., Nechitayls, A. A., and Lyakh, G. Y., Fizika Metall., 27:1114 (1969) 181. Bunshah, R. F. and Schramm, R. J., Thin Solid Films, 41 (1977) in press 182. Colen, M. and Bunshah, R. F., J. Vac. Sci. Tech., 13:536 (1976) 183. Furuuchi, S., Sakata, H., and Aiwaka, K., Japan J. Appl. Phys., 13:1905 (1974) 184. Grossklaus, W. and Bunshah, R. F., J. Vac. Sci. Technol., 12:811 (1975) 185. Wasa, K., Nagai, T., and Hayakowa, S., Thin Solid Films, 31:235 (1976) 186. Mah, G., Norden, C. W., and Fuller, J. F.,J. Vac. Sci. Technol., 11:371 (1974) 187. Henning, C. A. O., Boswell, F. W., and Corbett, J. M.,Acta Met.,23:193 (1975) 188. Palatnik, L. S., Ill’inski, A. I., and Sapelkin, N. P.,Soviet Phys. Solid St., 8:2016 (1967) 189. Solonovich, I. I. and Startsev, V. I., Problemy Prochn., 1:28 (1973) 190. Palatnik, L. S., Ill’inski, A. I., Biletchanko, N. M., and Sinel’nikova, R. I., Fiz. Metall., 32:199 (1971) 191. Bunshah, R. F., Gupta, Y. D., and Raghuram, A. C., unpublished data 192. Hall, E. O., Proc. Phys. Soc, Lond., B64:747 (1951) 193. Petch, N. J., J. Iron Steel Inst., 174:25 (1951) 194. Nenioto, M., Jumbou, R., and Suto, H.,Trans. Jpn. Inst. Metals,12:113 (1971) 195. Chambers, D. L. and Bower, W. K.,J. Vac. Sci. Technol.,7:S62 (1970) 196. Movchan, B. A., Demchishin, A. V., and Badilienko, G. F., Thin Solid Films, 40:237 (1977) 197. Movchan, B. A., Demchishin, A. V., and Kooluck, L. D., Thin Solid Films, 44:285 (1977) 198. Frank, F. C. and Lawn, B. R., Proc. R. Soc. Lond., 229A:291 (1967) 199. Boone, D. H. and Sullivan, C. P., STP, 520:401, Am. Soc. for Testing and Materials (1973)

Evaporation

271

200. Goward, G. W., J. Metals, 22:31 (1970) 201. Clough, P. J., New Types of Metal Powders, (H. H. Hausner, ed.), p. 9, Gordon and Breach (1964) 202. Hayashi, C., Jpn. J. Appl. Phys., 12:1675 (1973) 203. Bunshah, R. F., unpublished research 204. Majumder, K. S., Thin Solid Films, 42:327 (1977) 205. Majumder, K. S., Thin Solid Films, 42:343 (1977) 206. Chi, K. C., Dillon, R. O., Bunshah, R. F., Alterovitz, S., Martin, D. C., and Vollam, J. A., Thin Solid Films (1978) 207. Zubeck, R. F., King, C. N., Moore, D. F., Barbee, T .W., Hallak, A. B., Salem, J., and Hammond, R. H., Thin Solid Films, 40:249 (1977) 208. Martin, P. L., Bunshah, R. F., and Dymond, A. M., J. Vac. Sci Tech., 12:754 (1975) 209. Agarwal, P. L., Bunshah, R. F., and Crandall, P. H., unpublished research, UCLA (1978) 210. Sinha, A. K., Giessen, B. C., and Polk, D. E., Treatise on Solid State Chemistry, (N. V. Hannay, ed.), 3:1, Plenum Press, New York (1976) 211. Keung, P. K. and Wright, J. G., Phil. Mag., 30:995 (1974) 212. Hughes, J. L., Metals Eng. Quart. 14, No, 1:1 (1974) 213. Hill, R. J., Hughes, J. L., and Harker, H. R., Proc. of the 4th Int'l. Conf. on Vacuum Metallurgy, p. 248, Iron and Steel Institute of Japan, Tokyo, Japan (June 1973) 214. Harker, H. R., and Hill, R. J., J. Vac. Sci. Technol., 9:1395 (1972) 215. Bunshah, R. F., U.S. Patent No. 3,971,582 (Feb. 12, 1974) 216. Nakamura, K., Inagawa, K., Tsuruoka, K., and Komiya, S., Thin Solid Films, 40:155 (1977) 217. Kodama, M., Bunshah, R. F., and Shabaik, A. H., Thin Solids Films, (1978) 218. Bunshah, R. F., and Shabaik, A. H., Res./Dev., 26:46 (1975) 219. Bunshah, R. F., Shabaik, A. H., Nimmagadda, R., and Covey, J., Thin Solid Films, 45:1 (1977) 220. Hewig, G. H. and Bloss, W. H., Thin Solid Films, 45:1 (1977) 221. Boer, K. W., Annual Progress Report, NSF/RANN/SE/G134872., University of Delaware (Jan. 1974)

272


222. Brody, T. P. and Shirland, F. A.,Proc. of NSF Workshop on Photovoltaic Conversion of Solar Energy for Terrestrial Applications, p. 63, Cherry Hill, NJ (Oct. 1973) 223. Laegreid, N., in ibid, p. 63 224. Jacobson, B. E., Springarn, J. R., and Nux, W. D., Thin Solid Films, 45:517 (1977) 225. Mott, N. F., Phil. Mag., 44:742 (1953) 226. Movchan, B. A., Soviet Physics Doklady, 20(7):575 (1975) 227. Movchan, B. A., Badilenko G. F. and Demchishin, A. V., Thin Solid Films, 63:67 (1979) 228. Schiller, S. and Jasch, G., Thin Solid Films, 54:9 (1978) 229. Kobayashi, M. and Doi, Y., Thin Solid Films, 54:57 (1978) 230. Komiya, S., Umezu, N., and Narusawa, T., Thin Solid Films, 54:51 (1978) 231. Zega, B., Kornmann, M., and Amiguet, J., Thin Solid Films, 45:577 (1977) 232. Berghaus, B., German Patent No. 683,414 (1939) 233. Sabalev, L. P., et al., U.S. Patent 3,783,231 (Jan. 1, 1974); 3,793,179 (Feb. 19, 1974) 234. Dorodnov, A. M., Soviet Phys. Tech. Phys., 23:1058 (1978) 235. Osipov, V. A., et al., Soviet Rev. Sci. Inst., 21:1651 (1978) 236. Nath, P. and Bunshah, R. F., Thin Solid Films, 69:63 (1980) 237. Namba, Y. and Mori, J., J. Vac. Sci. Technol., 13:693 (1976) 238. Taylor, N. J., Surface Science, 4:161 (1966) 239. Bunshah, R. F., Nimmagadda, R., Doerr, H. J., Movchan, B. A., Grechanuk, N. I., and Dabizha, E. V., Thin Solid Films, 72:261 (1980) 240. Hordon, M. J., Titanium Science and Technology, (R. I. Jaffee and H. M. Burte, eds.), 4:2347-57, Plenum Press (1973) 241. Koehler, J. S., Phys. Rev., B 2:547 (1970) 242. Henning, C. A. O., Boswell, F. W., and Corbett, J. M. Acta. Met., 23:193. 243. Palatnik, L. S. and Ill’inski, A. I., Soviet Phys. Dokl., 9(1):93 (1961)

Evaporation

273

244. Palatnik, L. S., Ill’inski, A. I., and Sapelin, N. P., Soviet Physics Solid State, 8(8):2016 (1967) 245. Palatnik, L. S., Ill’inski, A. I., Biletchenko, N. M., and Sinel’nikova, R. I., Fiz. Met. Metalloved., 32(6):1312 (1971) 246. Solonovich, I. I. and Startsev, V. I., Problemy Prochnosti, 1:28-30 (1973) 247. Hordon, M. J. and Wright M. A., in Metal-Matrix Composites, Symp. of the Metallurgical Soc. AIME,DMIC Memorandum 243, pp.10-12 (May 1969) 248. Springer, R. W. and Catlett, D. S., Thin Solid Films, 54:197 (1978) 249. Lehoczky, S. L., J. App. Phys., 49:5479 (1978)

SUGGESTIONS FOR FURTHER READING Books Berry, R. W., Hall, P. M., Harris, M. T., Thin Film Technology, D. Van Nostrand Co. (1968) Bhushan, B. and Gupta, B. K., Handbook of Tribology, McGraw-Hill (1992) Chopra, K. L., Thin Film Phenomena, McGraw Hill Book Co., (1969) Handbook of Thin Film Technology, (L. I. Maissel, and R. Glang, eds.) McGraw Hill Book Co. (1970) Holland, L., Chapman and Hall, Vacuum Deposition of Thin Films, (1968); The Bible Ohring, M., Materials Science of Thin Films, Academic Press, (1992) Physics of Thin Films, Vols. 1-6, Academic Press (1963-1971) Science and Technology of Surface Coatings, (B. N. Chapman and J. C. Anderson, eds.) Academic Press (1974) Techniques of Metals Research, Vol. 1, part 3, (R. F. Bunshah, ed.) John Wiley & Sons (1968) Thin Films, Am. Soc. for Metals (1964) The Use of Thin Films in Physical Investigation, (J. C. Andrews, ed.) Academic Press (1966)

274


Journals Applied Surface Science Japanese Journal of Applied Physics, Japan Journal of Applied Physics, USA Journal of Electrochemical Society, USA Journal of Materials Research Journal of Materials Science, England Journal of Materials Synthesis and Processing Journal of Vacuum Science and Technology, Am. Phys. Society Materials and Manufacturing Processes Processing of Advanced Materials Review of Scientific Instruments, USA Surface and Coatings Technology, Elsevier, S. A. Thin Solid Films, Elsevier, S. A., Switzerland Vacuum, England

5 Sputter Deposition Processes John A. Thornton and Joseph E. Greene

1.0 INTRODUCTION The process of sputtering may be defined as the ejection of particles from a condensed-matter target due to the impingement of energetic projectile particles. The use of sputtered species as source material to deposit thin films was first reported in the literature in 1852[1] and has since enjoyed several periods of scientific and commercial interest interspersed with periods of disrepute. However, it is only relatively recently that sufficient understanding of the complex processes occurring during ion bombardment of solid surfaces has been developed to allow the reproducible and controllable use of sputter deposition for growing high-quality single crystals, complex alloys, superlattices, and materials with tailored microstructures. The evolution of the branch of science concerned with ion/surface interactions has been facilitated by the parallel development of ultra-high vacuum technology and highly sensitive microanalytic techniques for identifying the state of scattered particles, sputtered species, and implanted material. Sputter-ejected species have kinetic energies considerably greater than thermal. In addition, depending on the experimental configuration, the substrate and growing film may also be subjected to low-energy particle bombardment from accelerated host lattice species, dopants, inert-gas ions, and energetic particles backscattered from the target. Thus, ion/ surface interactions are not only important at the target, but they can also

275

276


play a decisive role, as discussed in Ch. 13, in determining film nucleation and growth kinetics, microstructure evolution, dopant incorporation probabilities, and hence the physical properties of as-deposited layers. The controlled use of ion bombardment effects allows a considerable enhancement in the ability to tailor film properties. Sputter deposition is inherently a vacuum coating process. In operation, the source of coating material, termed the “target,” is mounted opposite the substrates in a vacuum chamber which is then evacuated to a base pressure which typically ranges from 10-6 to 10-10 Torr*, depending upon the process. The most common method of providing the ion bombardment necessary for sputtering is to backfill the evacuated chamber, using a continuous flow of a gas such as Ar, to a pressure of from 1 to 100 mTorr, and establish a glow discharge. A negative potential, typically between 0.5 and 5 kV, is applied to the target in order to initiate positive-ion bombardment while the counterelectrode (the substrate) is grounded. A sputtering apparatus in which the target and substrate are opposing parallel plates, shown schematically in Fig. 5.1, is termed a diode system. The discharge in such a device is commonly operated in the abnormal negative-glow mode.[2] The most striking characteristic of the sputtering process is its universality. Since the coating material is passed into the vapor phase by a physical momentum-exchange process, rather than a chemical or thermal process, virtually any material is a coating candidate. DC discharge methods are generally used for sputtering metals, while an RF potential must be applied to the target when sputtering nonconducting materials. In some applications, rather than immersing the target in a plasma, it is more convenient to use a separate ion-beam source consisting of a self-contained discharge with ionacceleration optics. Sputter-deposition technology includes many variations of the basic process described above. For example, coatings may be formed by: 1. Employing a target which is a mosaic of several materials. 2. Employing several different targets simultaneously to obtain an alloy film. 3. Employing several targets sequentially to create a compositionally layered coating.

*The pressure unit of Torr (1 Torr = 1 mm Hg) is a carryover from the time when pressure was measured with a Hg manometer. Most commercial pressure gauges are still calibrated in Torr or microns (1 µm Hg = 1 mTorr). Therefore, Torr will be used in this chapter although both Torr and Pa (SI units) are given in some of the figures. 1 Torr = 133 Pa = 1.33 mbar = 1.316 x 10-3 atm.

Sputter Deposition Processes

277

Figure 5.1. Schematic representation of a parallel-plate diode sputtering system.

4. Electrically biasing the substrate to provide ion bombardment of the growing film during deposition in order to modify the film microstructure and/or microchemistry. 5. Employing a gas (e.g., O2, N2 , H2 S, etc.) to introduce one of the coating materials into the chamber. This process is known as reactive sputtering.

278 1.1

Deposition Technologies for Films and Coatings Sputter Deposition Systems

Sputtering systems can assume an almost unlimited variety of configurations, depending on the application. The simplest is the parallel-plate diode shown schematically in Fig. 5.1. Such systems have played a major role in the development of sputtering technology over the past twenty years and are still widely used. Figure 5.2 shows a planar-diode sputtering installation of a type commonly used in research and for small production runs.

Figure 5.2. Planar-diode sputtering system of the type used for research end small production runs. The system can be used for both DC and RF sputtering. (Photo courtesy of CVD Products, Inc., Rochester, NY)

The substrates in a planar-diode system are in contact with the plasma. This makes it relatively easy to carry out the processes of substrate sputter cleaning and bias sputtering. It is partly due to the effects of these processes that sputtering has long enjoyed a reputation for providing coatings with superior adhesion. However, the heating associated with plasma and electron bombardment often prohibits the use of planar diodes for coating thermallysensitive substrates. It is difficult to sustain an intense plasma discharge in the planar-diode electrode geometry. Thus, working pressures are necessarily relatively high at 20 to 75 mTorr and current densities are low, ≈1 mA/cm2 . The high


279

pressure causes the transport of coating material from the target to the substrate to be primarily diffusive rather than ballistic and sputtered material is lost to the walls of the container by scattering. This, coupled with the low current density, leads to deposition rates which are generally less than 75 nm/ min (4.5 µm/h). Triode devices, in which additional electrons are injected into the discharge by thermionic emission from a third electrode, can be used to produce intense sputtering discharges at low pressures. The deposition rates that can be achieved with triode devices are also higher than with planar diodes. For example, high-rate triode sputtering has been used to fabricate a free-standing 1.3 kg deposit of a Cu-alloy in the form of a cylinder 15 cm in diameter.[3] However, the complexity of triode designs for obtaining uniform deposition has, in general, limited their use to special applications. The recent development of a class of sputtering sources with magnetic plasma confinement, called magnetrons, has greatly enhanced the capabilities of the sputtering process. There are many forms of magnetrons. They vary from small ring sources—often referred to as Sputter-gunsTM (Sloan Technology, Santa Barbara, CA) and S-gunsTM (Varian Associates, Palo Alto, CA)— to long rectangular planar magnetrons and cylindrical magnetrons with post or hollow cathodes. Magnetrons can be used for both DC and RF sputtering but are particularly effective for DC sputtering, where deposition rates can be more than an order of magnitude larger than those obtained with planar diodes. Planar and cylindrical magnetrons can be scaled to large sizes to provide uniform deposition over very large areas (many m2). In addition, well-designed magnetrons virtually eliminate substrate heating caused by electron bombardment. 1.2 Sputter-Deposition Applications The enormous range of sputtering applications reflects the universality of the process. Films containing essentially every element in the periodic table have been prepared by sputtering. Alloys and compounds can generally be sputter-deposited while preserving their compositions. For example, PTFE (Teflon ) has been sputtered to produce lubricous films having many of the properties of the starting material. The ability to control composition has caused sputtering to become widely used in the electronics industry. Typical applications are aluminum alloy and refractory metal microcircuit metallization layers, microcircuit insulation layers, transparent conducting electrodes, amorphous optical films for integrated optics devices, piezo-electric transducers, photoconductors and luminescent films

280


for display devices, optically addressed memory devices, amorphous bubble memory devices, thin film resistors and capacitors, video-discs, solid electrolytes, thin film lasers, and microcircuit photolithographic mask blanks. Figure 5.3 shows a multisource sputtering system designed for wafer processing. Sputter deposition is also beginning to replace evaporation for depositing high performance optical components and is commonly used for depositing magnetic alloys with strong preferred orientation in magnetic recording devices. In addition, one finds applications ranging from coating razor blades to depositing wear-resistant coatings for machine tools.

Figure 5.3. Multi-source sputtering system designed for wafer processing. Wafer batches are passed into and out of the coating chamber through vacuum interlocks. (Photo courtesy GCA Corporation, Vacuum Industries Division, Somerville, MA)


281

Planar diodes are still widely used, particularly for depositing materials requiring RF power. However, recent trends find magnetrons replacing planar diodes for many DC, and some RF, applications. In addition, the magnetrons have opened up new applications because of their large-area capability and reduced substrate heating. For example, large in-line systems with vacuum interlocks use planar magnetron sources to coat 2 m x 3.5 m architectural glass plates at three-shift production volumes of about 106 m2/yr.[4] Sputtering is being investigated as a means for depositing selective absorber coatings for solar heating and for manufacturing photovoltaic cells for direct solar-toelectrical energy conversion. Because of the reduced substrate heating, magnetrons are used on a production basis to deposit chromium decorative coatings on automobile grilles and other exterior trim. Figure 5.4 shows an automated load-lock sputtering system designed for metallizing plastic automotive parts.

Figure 5.4. Automated load-lock magnetron-sputtering system designed for metallizing plastic automotive parts. System throughput is 46.5 m2/h of platen surface on which substrates may be mounted. (Photo courtesy of Varian Associates, Inc., Palo Alto, CA)

282


The selection of a sputtering apparatus for a given application depends on the substrate size, shape, and sensitivity to heat and plasma irradiation. It also depends on the nature of the coating—e.g., single layer or multilayer, thickness, types of materials involved, and critical parameters such as hardness, porosity, resistivity, semiconductor charge-carrier lifetimes, and magnetic anisotropy—as well as the production volume. Planar targets of an almost unlimited range of materials, including the new high temperature superconducting oxides, are available from many suppliers. Thus planar diodes are attractive for depositing thin coatings of complex materials onto planar substrates for research studies or for small production volumes. However, the substrates must be capable of withstanding the plasma environment, particularly electron bombardment. Triode devices are attractive when thicker coatings are required. However, for large production volumes, thick coatings, complex substrate shapes, or thermally sensitive substrates, magnetron type sources should be considered. The selection of a particular type of magnetron will depend on the nature of the coating and substrate and the availability of sputtering targets of the required material in the desired geometry. The procurement of high-quality targets is an important consideration for all sputtering systems. Sputtering, like other vacuum coating processes, suffers from the disadvantage that the equipment is expensive. In addition, high-rate sputtering equipment generally incorporates large, nonstandard, power supplies and automatic control systems. As a general rule, sputtering is most effective when production volumes are sufficient to permit the equipment cost to be amortized over a large number of parts. An advantage of sputtering is that it is reliable and lends itself to automatic control. 1.3 Process Implementation Almost any vacuum chamber capable of evacuation into the 10-6 Torr range can be used for sputtering. Provisions are usually required for throttling the pumping system so that the desired working gas pressure can be sustained with the pumps in operation. Small planar and gun-type magnetrons are particularly easy to install. Some forms of cylindrical magnetrons require special chamber geometries. Chambers used in large production runs generally include substrate-loading interlocks so that the target surface is not exposed to the atmosphere between deposition cycles. Examples are shown in Figs. 5.3 and 5.4.


283

Pre-deposition pumping and the importance of achieving low base pressures will depend on the application. It is important to remember that sputtered coatings are deposited in an atmosphere which contains outgassing flux from the substrates and chamber walls. This flux can have a significant influence on the growth and properties of the coatings. Special problems may be encountered if the substrates themselves undergo severe outgassing.[5] Generally, pre-deposition pumping is continued until the total outgassing flux from the chamber walls and substrates has decreased to a value that is significantly less than the total sputtering flux that will be used. The working gas is then injected into the chamber with the pumps throttled and sputtering is initiated. New, or air-exposed, targets should be “pre-sputtered,” with the substrates shielded, prior to deposition in order to clean and condition the target and chamber surfaces. The selection of deposition conditions is generally determined empirically. The primary control parameters are the deposition rate, target voltage, working gas species and pressure, and the substrate temperature and plasma bombardment conditions. The available selection range for the deposition parameters is determined largely by the apparatus. In planar diodes, many of the parameters are interrelated and unavailable for independent control. Much greater control is possible with magnetrons. However, other variables become important. For example, in many magnetron geometries, along with operation at low pressures where the sputtered atoms can pass to the substrates while making few collisions, coating-flux angle-of-incidence considerations become important in determining coating properties.[6] Thus, in all applications where large production volumes are anticipated, it is wise to perform development work using an apparatus of the same type and geometry anticipated for the production facility. Scale-up increments should generally not exceed a factor of three in apparatus size. 1.4 History of Sputter Deposition and Background Reading Several review papers written over the last twenty years permit the interested reader to follow the developments in sputtering technology. An extensive review of the basic process was published by G. K. Wehner, one of the most prominent of the early workers, in 1955.[7] Film properties obtained in early experiments were discussed by E. Kay in a 1962 review.[8] Sputter deposition processes were reviewed by L. Maissel[9] and Kay[10] in 1966. The Handbook of Thin Films , published in 1970, contains reviews by Wehner and Anderson [11] and by Maissel. [12] A review article by Thornton

284


in 1973 discusses sputtering equipment.[13] Process considerations in glow discharge sputtering were reviewed by Westwood in 1976.[14] A book edited by J. L. Vossen and W. Kern contains several chapters reviewing magnetron sputtering[15] and B. Chapman’s book[2] provides an introduction to glow discharges used in sputtering and plasma etching. The growth and properties of semiconductors deposited by sputtering have been reviewed by J. E. Greene.[16]-[18] Harper and co-workers [19]-[21] have written review articles on ion-beam sputter deposition. Finally, the role of low-energy ion/surface interactions in controlling the microstructure and microchemistry of vaporphase deposited films has been discussed in detail in a number of review articles by Greene and co-workers.[22]-[25] Developments in the science and technology of sputtering are most commonly reported in the following journals: Journal of Vacuum Science and Technology, Thin Solid Films, Journal of Applied Physics, Vacuum, Surface Science, Applied Surface Science, and the Journal of the Electrochemical Society.

2.0 SPUTTERING MECHANISMS Sputtering is a statistical process which occurs as a result of a momentum-exchange collisional cascade process initiated near the target surface by an incident energetic projectile. Figure 5.5 shows a computer simulation of such a process resulting from a single bombardment event. It is immediately clear that sputtering cannot result from a single binary collision since the momentum vector of the struck target atom must be altered by more than 90o . In the simulated collision sequence of Fig. 5.5, the incoming projectile (depicted as a solid circle) strikes target atom 1 driving it deeper into the lattice. The collision is elastic and the subsequent path of the initial projectile towards atom 2 can be calculated from conservation of energy and momentum considerations. The glancing collision with atom 2 causes the projectile to hit atom 3 which is displaced and collides with surface atom 4 imparting sufficient momentum to allow atom 4 to overcome the surface energy barrier and be ejected. The initial projectile as well as atoms 1 and 2 displace other lattice atoms in subsequent “knock-on” collisions but, in this simulation, fail to lead to any further sputtering events. The statistical nature of the sputtering process is evident from the above example. Computer simulations of Cu bombardment by 600 eV Ar+ ions [26] have shown that the radius of a collision cascade under such


285

conditions is of the order of 10 nm and that the fraction of collision sequences which actually intersect the surface and transfer sufficient momentum to result in sputtering is quite low. Statistical analyses show that sputter ejection very rarely occurs due to collision cascades initiated more than five atomic layers below the surface. Most of the energy transferred to the lattice during ion bombardment is lost as heat. The time associated with a particular collision event is short with respect to the projectile time of flight between collisions. Thus, under normal sputter deposition conditions, the probability that overlapping lattice regions will be excited simultaneously by individual bombardment events is small.

Figure 5.5. Computer simulation of a portion of a collision sequence initiated by a single ion-bombardment event in a solid lattice.

2.1 Sputtering Rate The sputtering process is quantified in terms of the sputtering yield, defined as the number of target atoms ejected per incident particle. The yield depends on the target species and the nature, energy, and angle of incidence of the bombarding species. It is relatively insensitive to the target temperature. [11] (At sufficiently high temperatures, of course, the evaporation rate becomes of the order of, or larger than, the sputtering rate). The yield is also independent of whether or not the bombarding species is ionized. In fact, incident ions have a high probability of being neutralized by

286


a field-emitted electron prior to impact.[11][26]-[29] Molecular bombarding species behave as if the atoms of the molecule arrived separately with the same velocity as the molecule and initiated their own sputtering events.[11] The sputtering yield tends to be greatest when the mass of the bombarding particle is of the same order of magnitude or larger than that of the target atoms. The use of inert-gas ions avoids chemical reactions at the target and substrate. Accordingly, Ar is often used because of its mass compatibility with materials of engineering interest and its low cost. Sputtering yields are determined experimentally. Figure 5.6 shows yield versus ion-energy data for several materials under normal ion incidence. Additional data are given in Table 5.1. The dependence of the yield on the bombarding-ion energy exhibits a threshold of 20 - 40 eV,[11] followed by a nearly linear region which may extend to several hundred eV. At higher energies, the yield vs ion-energy dependence becomes sublinear. The sputtering process is most efficient from the standpoint of energy consumption when the ion energies are within the linear range.

Figure 5.6. Variation of the sputtering yield of several materials as a function of Ar+ ion energy at normal angle of incidence. Data from R. V. Stuart and G. K. Wehner, J. Appl. Phys. 33, 2351 (1962); D. Rosenberg and G. K. Wehner, J. Appl. Phys. 33, 1842 (1962); and R. Behrisch, Ergeb. Exakt. Naturw. 35, 295 (1964).


287

Table 5.1. Sputtering Yields for Various Materials under Argon Ion Bombardment. Ion energy in eV. Data from Ref. 36. Target

200

600

Ion Energy (eV) 1,000 2,000 5,000

Metal Ag Al Au C Co Cr Cu Fe Ge Mo Nb Ni Os Pd Pt Re Rh Si Ta Th Ti U W Zr

1.6 0.35 1.1 0.05* 0.6 0.7 1.1 0.5 0.5 0.4 0.25 0.7 0.4 1.0 0.6 0.4 0.55 0.2 0.3 0.3 0.2 0.35 0.3 0.3

3.4 1.2 2.8 0.2* 1.4 1.3 2.3 1.3 1.2 0.9 0.65 1.5 0.95 2.4 1.6 0.9 1.5 0.5 0.6 0.7 0.6 1.0 0.6 0.75

Sputtering Yields (Atoms/Ion) 2.0 3.6 5.6 7.9 3.2 4.3 5.5 1.4 2.0** 2.5** 1.5 2.0 3.0 1.1 1.5 2.1 0.6 0.9 1.4 1.05 1.1 1.7 1.1 -

0.5 0.4 0.4 0.4 0.25 0.6 -

1.2 0.9 1.0 0.9 0.55 1.40 0.45 -

Sputtering Yields (Molecules/Ion) 1.2 0.13 0.4 0.04 0.11 -

Compound CdS(1010) GaAs(110) GaP(111) GaSb(111) InSb(110) PbTe(110) SiC(0001) SiO2 Al2O3 *Kr+ ions

**Type 304 stainless steel

10,000

Heat of Sublimation

8.8 6.6 2.2 2.1 -

(eV/atom***) 2.94 3.33 3.92 7.39 4.40 4.11 3.50 4.13 3.98 6.88 4.45 8.19 3.90 5.95 8.06 5.76 4.68 8.10 5.97 4.86 5.00 8.80 6.34

***From Ref. 240

Note that for typical ion acceleration energies, the sputtering yields of most metals are near unity and within an order of magnitude of one another. This is in contrast to evaporation where the rates for different materials at a given temperature can differ by several orders of magnitude. In addition, the evaporation rate for a given material varies exponentially with temperature while the sputtering yield is essentially independent of temperature. The general dependence of the sputtering yield on the ion angle of incidence is indicated in Fig. 5.7.[30] In glow-discharge sputtering devices, the ions generally approach the target in a direction normal to the target

288


surface. Thus, the relationship shown in Fig. 5.7 is of particular significance when the target surface is highly irregular or for ion-beam sputtering where the ion-incidence angle can be controlled.

Figure 5.7. Schematic diagram showing variation of the sputtering yield with ion angle of incidence for a constant ion energy.

Sputtering systems are generally calibrated to determine the deposition rate under a given set of operating conditions. However, yield data of the type described above are often used in estimating rate changes when changing coating materials and in estimating the amount of material removed during sputter cleaning and bias sputtering. The erosion rate is given by Eq. (1)

R = 6.23

JSMA δ

(nm/min)


289

where J is the ion current density in mA/cm2, S is the sputtering yield in atoms/ ion, MA is the atomic weight in grams, andδ is the density of the target material in g/cm3. The reader should be cautious about using Eq. 1 in attempting to predict absolute sputtering rates, especially in planar diode systems where the average energy of ions striking the target may be considerably less than Ei = eVT (VT is the applied target potential), due primarily to inelastic charge exchange collisions[31] between accelerated ions and neutral sputtering gas species. S(Ei ) data, on the other hand, are usually obtained from ion beam experiments carried out in low pressure (long mean-free path) environments where the ion energy is given by the accelerating energy. The apparent lower yield in the glow discharge sputtering case (due to the lower average ion energy) is, however, partially offset by the flux of energetic charge-exchanged neutrals which are incident at the target. 2.2 Momentum Exchange Consider a particle of mass Mi and velocity v i which impacts on a line of centers with a target particle of mass Mt that is initially at rest, as shown in Fig. 5.8a. Three simple observations can be made. First, as noted above, the momentum imparted to the target particle drives it into the lattice. Secondly, from a simple line-of-centers atomic collision calculation, a fraction Eq. (2)

ε =

4 Mi Mt (Mi + Mt)2

of the kinetic energy of the incident particle is transferred to the target particle. An expression for the yield, which can be written in the form shown in Eq. 3 below, has been derived by assuming perpendicular ion incidence onto a target consisting of a random array of atoms (a good approximation for a smallgrain polycrystalline material) with a planar surface.[32]-[34] Eq. (3)

S = (constant) ε

E U

α(Mt/M i)

The relationship is useful for illustrating the functional dependences of the important parameters and provides reasonably good agreement with measurements for medium mass (Ar, Kr) bombardment of a wide variety of

290


materials. The yield is seen to depend directly on the energy transfer coefficient ε. The term α(M t/Mi ) is a near-linear function of Mt/Mi, E is the kinetic energy of the incident ion, and U is the heat of sublimation for the target material. The mass dependence ofεα does not vary greatly from one material to another. The primary material-sensitive factor is the heat of sublimation, and this is only a first power dependence. This is in contrast to chemical and thermal processes that depend exponentially on an activation energy. It is this relative insensitivity to the properties of the target material that gives sputtering the universality referred to previously.

Figure 5.8. Schematic diagram showing momentum exchange processes that occur during sputtering; Mi and vi are the ion mass and velocity, Mt and vt are the target-atom mass and velocity, and the prime superscript denotes velocities after collision.

When the ion mass is lower than that of the target atom, it may be reflected backward in a single collision with a kinetic energy that is still a significant fraction of its initial energy. For a 180o reflection, this fraction is Eq. (4)

f =

(Mi - M t)2 (Mi + Mt)


291

If Mi > Mt, reflection requires more than one collision and the reflection coefficient is low.[35] Since the ions have a high probability of being neutralized prior to impact, they are reflected as energetic neutrals which are therefore not influenced by the electric field over the target surface.[36] The flux of reflected species contributes to substrate heating,[37] particularly in devices operating at low pressures where the reflected neutralized ions may reach the substrates with little loss of kinetic energy by gas-phase collisions. Consequently, the reflected species bombard, and can become entrapped in, the growing film.[38]-[41] The energy flux which leaves the cathode via backscattering can be estimated using the sputtering efficiency[42]-[44] which is defined as the fraction γ of the bombarding ion energy incident on the target surface, Ein, which leaves the surface in the backward direction, Eout, in the form of sputtered atoms or backscattered ions.

Eq. (5)

γ =

Eout Ein

=

Esputtered + Ebackscattered Ein

The energy of the sputtered atoms is discussed in a subsequent section. Theoretical calculations[42] for a target consisting of a random array of atoms in which the surface binding energy was neglected indicates that the sputtering efficiency is independent of the energy of the incident ion and is simply a monotonically increasing function of the target-atom/ion mass ratio. This dependence, which has been confirmed for both low and high ion energies, is shown in Fig. 5.9. Momentum exchange processes also provide an explanation for the angular dependence of the sputtering yield shown in Fig. 5.7. An ion which is incident on the target surface at an angle θ will, to first order, have its path length increased by a factor secθ before it passes through depth d. At larger angles of incidence, ion reflection dominates and the yield decreases. Another question of interest is the ultimate fate of the inert gas ions that bombard the target. The probability of their becoming trapped in the target increases with ion energy above a threshold of ~ 50 - 100 eV.[45] Thus, a concentration of inert gas, which depends on a balance between the rates of implantation and release, will develop in the near-surface region of the target. The amount of gas entrapped in the target can be large enough to influence the sputtering yield.[46]

292


Figure 5.9. Sputtering efficiency versus target-to-ion mass ratio. The solid curve is from the theoretical work of Sigmund (Ref. 42). The experimental data is from substrate heating experiments with cylindrical-post magnetrons (Ref. 37).

2.3 Alloys and Compounds An important advantage of the sputtering process is that the composition of a sputter-deposited film tends to be the same as that of the target, provided that: (i) the target is maintained sufficiently cool to avoid bulk diffusion of the constituents, (ii) the target does not decompose, (iii) reactive contaminants are not present,(iv) the gas-phase transport of the components is the same, and (v) the sticking coefficients for the components on the substrate are the same. Targets can be formed by casting or by hot pressing powders. In addition, composite targets can be formed by placing wires, strips, or discs of one material over a target of another material. The details of ion/surface interactions with multicomponent materials are complex[47]-[49] and poorly understood. Consider the case of a homogeneous starting material composed of species having different individual sputtering yields or masses. When sputtering is first initiated from such a


293

target, the sputtered flux will, in general, be rich in one of the constituents. The composition of this altered surface layer continues to change until the product of the partial sputtering yield times the surface concentration for each species is proportional to its concentration in the target. The process is indicated schematically in Fig. 5.10. Once a steady-state altered layer is formed, the composition of the sputtered flux is equal to the bulk target composition.

Figure 5.10. Schematic illustration of the surface composition modification which occurs during sputtering of a single-phase alloy.

The thickness and composition of the altered layer will depend on the target material and sputtering conditions. Typical altered-layer thicknesses are 3 - 10 nm for single-phase alloys[49][51] and up to several µm for multiphase alloys.[52] A change in sputtering conditions will in general require an adjustment of the altered layer. It is important to note that the partial sputtering yield of a constituent in an alloy or compound will not be the same as for that constituent by itself because of the difference in binding energies and the different atomic masses involved in the collision sequence within the alloy or compound. In an alloy for which the constituent species have similar binding energies, the low mass species can be expected to have higher partial sputtering yields. If the masses are similar, the weakly-bound species will have higher partial sputtering yields.[34][47]-[48] Thus, in the sputtering of most oxides, the altered layer becomes deficient in the flow-mass oxygen component.[47]

294


Sputtering of two-phase alloys in which the phases have significantly different sputtering yields results in the development of an irregular surface topography.[30][53]-[55] The sloping surfaces that survive tend to be those that make an angle with the sputtered flux such that the sputtering yield is maximized. If the second phase, or any included impurity particles, have very low sputtering yields, the surface may develop into a forest of cones with side walls at the maximum sputtering angle[52][56]-[59] as shown in Fig. 5.11. The cones will eventually be sputtered away; however, the receding target surface will expose new second-phase regions and impurity particles (if they are distributed throughout the bulk) and new cones will form. Thus a steady-state surface topography will develop. Surface diffusion on the target will, in general, make this situation more complex than the picture described above. The important point is that, following an incubation period, the composition of the sputtered flux leaving the target will become identical to that of the target. Nevertheless, the irregular surface topography may cause the overall yield to be considerably lower than what might be expected on the basis of the yields of the primary target constituents.

Figure 5.11. Schematic representation showing stages of cone formation during ion irradiation of a contaminated or two-phase target.

Topographical evolution such as cone formation can also influence the performance of composite sputtering targets.[60][61] When such targets are used in sputtering systems that operate at high pressures (greater than about 20 mTorr), some of the sputtered material will be backscattered by the working gas. Thus mixing of low and high yield materials can occur on the target segments. When atoms of a low-yield material are deposited on a highyield target surface, the low-yield material can agglomerate into islands capable of protecting the material underneath and cones will form. [58]


295

Relative sputtering yields as well as the target temperature appear to be important in predicting this behavior. An example is the formation of Moprotected cones on Cu surfaces.[58] The resultant sputtering rate from the cone-covered surface has been found to be very close to that for the low-yield material, Mo. Thus the composition of films deposited from composite targets can be much different than that estimated using the individual sputtering yields and the relative areas of the target segments. Special care should be exercised when using hot-pressed targets. Hotpressed Au-Ni and Au-Co targets composed of powders in the 50 to 130 µm range were found to yield deposits with compositions that matched those of the target after a transition period during which a layer≈20 µm thick had been sputtered from their surfaces.[62] However, the overall yield dropped to a value equal to that of the low-yield constituents (Ni or Co), even when the volume fraction of that constituent was only ≈30%. Contamination can present a particular problem with hot-pressed targets because of the large surface area contained in the starting powder. Such contamination may be present throughout the target and will provide a continuous virtual leak as the target is used.[63][64] Particular caution must also be exercised when using targets composed of compounds having poor electrical and thermal conductivities. Cracking often limits allowable current densities. The problem is especially egregious in planar magnetron systems where concentrated heating occurs under the plasma ring.[65] Poor thermal conductivity leads to high surface temperatures and may also result in the loss of volatile constituents by evaporation or sublimation. The high electric field in a poorly conducting target can act in concert with the high temperature and promote diffusion within the target. Thus the requirements listed at the beginning of this section for obtaining films with the same composition as the target are violated. It is not uncommon for films sputter deposited from such targets to be deficient in the more volatile constituents.[66][67] 2.4 Sputtering with Reactive Species The most complete data on the dependence of the sputtering yield on the ion species are those collected by Almen and Bruce, shown in Fig. 5.12.[68] Although the ion energies were considerably above those generally used for sputter coating technology, they do illustrate trends. Sputtering yields increase with the mass of the ions and, for a given row in the periodic table, the rare-gas ions have the highest yields. Of particular interest is the fact that yields vary much more with ion species (factor of 100 or more) than

296


they do with target atom species (factor of 10).[11] This is believed to result from the bombarding ions forming alloys or compounds with high binding energies on the surface of the target. Note that the yields for the three target materials examined in Fig. 5.12 are particularly low for reactive species such as Be, C, Mg, Si, Ti, and Zr.

Figure 5.12. Sputtering yields for various ions impacting at normal incidence on Ag, Cu, and Ta surfaces at high energies (45 keV). Data from Ref. 68.

Reduced yields are commonly observed in reactive sputtering (see Sec. 4.1 in this chapter) and attributed to compound formation on the target surface. Such surface interactions can also significantly influence the surface topography that develops on the target. Thus, 20 keV O2+ bombardment of an Fe target yielded a considerably smoother surface than 20 keV Ar+ bombardment.[69] 2.5 The Nature of Sputtered Species Under typical metal or semiconductor thin-film deposition conditions, most sputtered material is ejected in the neutral atomic state. The fraction of charged particles sputtered from clean metal and semiconductor surfaces is on the order of 10-4, becoming larger for surfaces contaminated with


297

strongly electropositive or electronegative species.[70]-[73] In glow discharge sputtering, the target is biased negatively and hence ejected positive ionswill be recaptured. The fraction of negative ion emission from pure semiconducting targets such as Si, GaAs, or GaP is typically less than 10-4 .[74] However, the negative ion yield can be quite large for targets composed of elements, one of which has a low ionization potential, while another has a high electron affinity. Examples are TbF3 ,[75] SmAu,[76] and YBa2 Cu3Ox.[77] Glow discharge sputtering of such materials results in acceleration of the negative ions, via the cathode fall potential, to the substrate. Relatively little experimental data is available on the probability of material being sputtered as molecules or clusters. Investigation by Oechsner and Gerhard[78][79] and Gerhard[80] using mass spectrometric analyses of post-ionized sputtered neutral particles has shown that with 1 keV Ar+ bombardment, the maximum fraction of sputtered dimers is 0.1 for Ag, Au, and Cu and about 0.03 for other metals. The fraction of trimers is about 0.001. Molecular-dynamic computer simulations by Winograd et al.[81] indicate that the fraction of sputtered Cu multimers varies strongly with crystal orientation, being largest for the (111) face. While such simulations are useful for predicting trends, the number of individual events sampled is too small to expect reliable statistics. Nevertheless, the predicted yield fractions of sputtered dimers and trimers agree reasonably well with the measured results of Gerhard and co-workers. The mechanism for the sputtering of molecular species is not well established. So-called “statistical models” have been proposed in which sputtered neutral atoms resulting from nearly-simultaneous ejection events agglomerate above the surface if their ejection is properly correlated in space and time and their relative kinetic energy is less than the dissociation energy of the molecule formed.[79][82]-[84] Können et al.[82] used such a model to describe the energy distribution of sputtered K2 and KI molecules. Winograd et al.[81][84] invoked a similar mechanism to obtain sputtered clusters in their computer simulations. However, Prigge and Bauer[85] reported experimental results which may indicate that, at least for the case of dimer and trimer ions, species comprising sputtered molecules were originally vertically displaced nearest neighbors in the lattice. In their experiments, they used 1 keV Ar+ ions to sputter thin Cu or Pd layers which had been deposited onto (110) W. The metal layers ranged in thickness from less than a monolayer to greater than three monolayers and sputtered species were detected by secondary-ion mass spectrometry. No sputtered Cu2+ or Pd2 + ions were observed emanating from targets with less than one

298


monolayer coverage and no trimer ions from samples with less than two monolayers. The sputtering yield of monomer, dimer, and trimer ions all saturated at a coverage of ≈3 monolayers. In the case of compounds, most of the information available on molecular sputtering is for alkali halides and oxides where clusters can account for a significant percent of the total sputtered flux. Coburn et al.[86] showed that the relative fraction MO/(M + O) of sputtered species from MxOy metal oxide targets increased with increasing M-O bond energy. Rare gas matrix isolation spectroscopy was used by Gruen et al.[87][88] to identify Al2 O and AlO molecular species sputtered from Al2 O3 targets under Ar + bombardment and TiO and ZrO species from metal targets sputtered with 2 keV O2 + ions.[89] The only published work on compound semiconductors is for GaAs. Using 140 eV Ar+ bombardment, comparable to the average impact energy in many glow discharge deposition experiments, Comas and Cooper[90] found from post-ionized mass spectroscopy measurements that molecular species (GaAs, Ga2, As2) amounted to less than 1% of the total sputtered flux. However, for 6 keV Ar+ ion sputtering, Szymonski and Bhattacharya[91] observed that at room temperature, sputtered GaAs and As2 molecules accounted for ~14% and 11% (data uncorrected for the variation in the detector efficiency as a function of mass), respectively, of the flux. The fraction of sputtered GaAs molecules was found to increase rapidly for target temperatures above 250oC. This latter effect was explained as being due to enhanced sputtering from collisional spikes. 2.6 Energy Distribution of Sputtered Species An important distinction between sputtering and other vapor-phase deposition techniques is that sputtered atoms can have quite high kinetic energies. For example, the average ejection energy of Ge atoms under 1.2 keV Ar+ bombardment is ≈15 eV[92] compared to only≈0.1 eV for evaporated Ge. In sputter deposition systems for which the target-substrate separation is less than a few mean free paths, the energy distribution of sputtered species impinging on the substrate will be approximately the same as the ejected species energy distribution. The most probable ejection energy is typically of the order of one half the surface binding energy, but because of the extended high-energy tail the average ejection energy is considerably higher (see Fig. 5.13)[92] and, in general, is found to increase with the atomic number of the target.[93]-[95]


299

From collision cascade theory,[96] the energy distribution of sputtered species is expected to be Eq. (6)

∆N/∆E ∝

E (E + U)3

where ∆N/∆E is the differential flux of sputtered particles with energy E and U is the surface binding energy. In practice, the high-energy tails of experimentally determined sputtered-atom energy distributions for rare-gas bombardment energies from≈1 to 10 keV generally follow a E-2 dependence in agreement with Eq. (6).

Figure 5.13. (Top) Energy distribution of sputtered Cu atoms ejected by Kr + ions at various bombarding energies. (Bottom) Comparision of velocity distributions of sputtered and evaporated Cu atoms. Data from Ref. 92.

300


Sputter-ejected atoms from metals with atomic number Zt > 20 have ejection velocities which lie in a relatively narrow range (see Fig. 5.14b). Average ejection energies therefore increase with increasing Zt as shown in Fig. 5.14a for neutral metal atoms sputtered from polycrystalline targets with 1.2 keV Kr+ ions.[11]

Figure 5.14. Average energies (top) and velocities (bottom) of sputtered atoms ejected by 1.2 keV Kr+ ion bombardment. Data from Ref. 11.

Atoms sputtered from polycrystalline or amorphous targets under perpendicular-incidence bombardment by medium-mass ions with energies of 1 - 3 keV are ejected in nearly random directions, as a consequence of multiple collisions within the target, and therefore have near cosine distributions.[11] At low ion energies (≈1 keV), the distribution may be slightly under


301

cosine (more emission at large angles) while at higher energies (≈3 keV) it is over cosine.[10] Under oblique-incidence bombardment, the target atoms are sputtered in the forward direction from smooth surfaces. However, the roughness of most practical targets causes the emission to be random. This is particularly true for polycrystalline targets, where the difference in yield for different crystallographic directions can lead to an increase in surface roughness as sputtering proceeds. Thus, a cosine distribution is often a reasonable approximation for estimating deposition profiles.[11]

3.0 SPUTTER DEPOSITION TECHNIQUES The fundamental problem in implementing the sputtering process is to provide a uniform and copious supply of ions over the surface of the target. The low-pressure glow discharge has proven to be the most cost-effective source of ions. A wide range of glow discharge apparatus geometries have been used in attempts to: (i) increase the ion supply and thus the sputtering rate, (ii) increase the target area and thus the available deposition area, (iii) reduce plasma heating of the substrates, (iv) permit operation at lower working-gas pressures, and(v) facilitate the coating of particular substrate shapes. In the following discussion, the essential features of the glow discharge and several of the more commonly used apparatus types are reviewed. 3.1 Planar Diode and the DC Glow Discharge The planar diode shown schematically in Fig. 5.15 is the simplest and probably the most widely used sputtering configuration. Cathode diameters are typically 10 to 30 cm and the cathode-to-anode spacing≈5 to 10 cm. Such systems are operated with both DC and RF power supplies. In DC diodes, the cathode serves a dual capacity. It is the target or source of coating material as well as the cathode electrode for sustaining the glow discharge and is generally water-cooled. Often the target consists of a disc of the material to be sputtered which is attached with solder or conducting epoxy to a backing plate which serves as part of the cathode-cooling channel. A low-pressure glow discharge of a type known as an abnormal negative glow[97] is maintained between the cathode and an adjacent anode which may also serve as the substrate mounting table, as shown in Fig. 5.15.

302


Figure 5.15. Schematic representation of the plasma in a planar diode sputtering source.

A grounded shield is used to promote a uniform erosion rate over the target surface and to prevent sputtering from the sides and the rear surface of the target. The discharge current is carried, in the vicinity of the negatively biased cathode, primarily by positive ions passing out of the plasma volume, and, in the vicinity of the anode, by electrons passing from the plasma volume to the anode. Thus, a necessary condition for sustaining the discharge is that the plasma volume be a suitable source of electrons and ions. Because of the relatively low mobility of the ions compared to the electrons, most of the electrical potential that is applied between the anode and cathode by the power supply is consumed in a cathode dark space, or sheath region.[97] Dark-space thicknesses are typically 1 to 4 cm, depending on the pressure and current density.[99] Accordingly, strong electric fields are formed, and ions passing from the plasma volume to the cathode are accelerated by these fields to impact the cathode. However, these ions also cause a small number of secondary electrons to be emitted from the surface (approximately one for every ten ions in the case of Ar+ ions impacting on a metal cathode).[100][101] These electrons are accelerated in the cathode dark space to energies approaching the applied potential and enter the plasma volume (negative glow) where, known asprimary electrons, they collide with gas atoms and produce the volume ionization necessary to sustain the discharge.[97][102] The requirement for sustaining such a discharge is that each primary electron must produce sufficient ions to release one further electron from the


303

cathode. Therefore, the interelectrode spacing must be large compared to the electron mean free path. The cross section, σ, for Ar ionization by the impact of 500 eV electrons is about 10-16 cm2.[103] Thus, for example, at an Ar pressure of 1 mTorr (gas density N = 3.2 x 1013 atoms/cm3) the electron mean free path (1/Nσ) for the production of ionization is 300 cm; i.e., much larger than the cathode-to-anode spacing. Consequently, discharges of the form shown in Fig. 5.15 can be sustained only at relatively high working pressures (50 - 100 mTorr), where a high density of Ar collision partners is provided for the primary electrons. (The grounded shields shown in Fig. 5.15 prevent the discharge from forming on the sides of the cathode because the electrode separation is too small to support the ionization mechanisms described above at the operating pressures of interest.). Attempts to increase the discharge current in a planar diode by increasing the applied voltage are thwarted to a large degree by the fact that the ionization cross-section decreases with increasing electron energy for energies greater than about 100 eV.[103] The current, and thus the sputtering rate, can be effectively increased at a given voltage by increasing the Ar pressure. However, if the pressure is too high, the deposition rate starts to decrease since the motion of both ions and sputtered atoms is impeded by the working gas atmosphere, as discussed below. These conflicting requirements result in an optimum operating pressure for producing the maximum deposition rate in a given apparatus. Typical operating conditions for metal deposition in a DC planar diode sputtering source are listed below. Cathode current density - 1 mA/cm2 Discharge voltage - 3,000 V Ar pressure - 75 mTorr (10 Pa) Deposition rate - 40 nm/min (2.4 µm/h) At typical planar diode operating pressures, the motion of the ions across the dark space is disrupted by collisions with gas atoms. In such collisions, there is a high probability of charge exchange, particularly when noble gas ions are passing through an atomic gas of their own species (resonance charge exchange).[97] A fast ion extracts an electron from a slow gas atom. The fast ion then becomes a “fast” neutral atom, while the “slow” atom becomes a positive ion, as indicated schematically in Fig. 5.16. Thus, instead of being bombarded by a current of ions having an energy equal to the potential drop across the cathode dark space, the target is bombarded with a much larger number of ions and fast neutrals having an average energy that is often less than 10 - 20% of the applied potential.[31][104]

304


Figure 5.16. Schematic illustration of the charge-exchange process that affects ion transport across the cathode dark space.

The deposition rate in planar diodes is further reduced by gas-scattering of sputtered atoms during transport between the target and the substrate. Optical emission measurements have confirmed that at typical sputtering pressures, sputtered-atom transport within the negative glow region is largely by diffusion.[105][106] The combination of charge transfer processes and diffusion transport make it necessary to determine deposition rates experimentally for each set of operating conditions. Another consequence of the collision-dominated transport of the sputtered atoms is a reduction in their kinetic energy at the substrate.[14][107][108] Figure 5.17 shows the results of an approximate calculation of the maximum distance required for sputtered atoms with various initial energies to have their kinetic energy reduced to the thermal energy of the gas atoms (≈0.025 eV).[109] At typical planar diode operating pressures, the equilibration distances are short compared to targetto-substrate spacings. Even under the relatively high-pressure conditions that yield the maximum deposition rates, the planar-diode discharge is inefficient. Many of the high-energy primary electrons fail to transfer their energy in the plasma volume and are incident at the anode and substrates while still possessing considerable energy. Ions and electrons are also lost from the edges of the discharge. Note also that the substrates are in contact with the plasma and are therefore also subjected to bombardment by plasma electrons and ions. This irradiation precludes the coating of many heat-sensitive materials such as plastics.[110]


305

Figure 5.17. Maximum distance from the target at which sputtered Al and Ta atoms of different initial energies are thermalized in Ar at various pressures. Data from Ref. 109.

Planar diodes are widely used despite substrate heating and low deposition rates. The reason is their simplicity and the relative ease with which planar targets can be fabricated from a wide range of materials. Sputter cleaning of the substrate and bias sputtering are easily accomplished by adding an auxiliary anode and applying a negative bias to the substrate holder. 3.2 Triode Discharge Devices Triode discharge devices utilize an additional electrode, independent of the target, to sustain the glow discharge.[13] The most common configuration is the hot-cathode triode shown schematically in Fig. 5.18. Electrons are emitted from the cathode surface thermionically rather than by ion bombardment. This relaxes the volume ionization requirement for sustaining the discharge. Consequently, hot-cathode triodes can be operated at low pressures (0.5 to 1 mTorr). The driving voltage is only 50 - 100 V, although the current may be several amperes. Radial plasma losses are often minimized through the confining effect of an axial magnetic field as shown in the figure. However, such a field produces a distortion of the current distribution over the target.

306


Figure 5.18. Schematic drawing of hot-cathode assisted discharge device (triode) From Ref. 13.

Triodes permit high deposition rates (several hundred nm/min) to be achieved, even at low pressures (mTorr range).[3] Although thick coatings have been deposited,[111] use of triodes has been limited by difficulties in scaling and the vulnerability of the thermionic emitter to reactive gases. Consequently, magnetron sources (next section) are assuming primary importance as high-rate sputtering devices. 3.3 Magnetrons The development of high performance magnetron sputtering sources that provide (i) relatively high deposition rates,(ii) large deposition areas, and(iii) low substrate heating, revolutionized the sputtering process by greatly expanding the range of feasible applications.[112] Magnetron sputtering sources can be defined as diode devices in which magnetic fields are used in concert with the cathode surface to form electron ¯xB ¯ electron drift currents close upon traps which are so configured that the E themselves.[113][114] Magnetrons can be configured in a variety of forms. Examples include the planar magnetrons shown in Fig. 5.19a, the S-gun type shown in Fig. 5.19b, and the cylindrical type shown in Fig. 5.20.


307

Figure 5.19. Magnetrons with magnetic end-confinement: (a) planar magnetron, (b) gun type. From Ref. 110.

Figure 5.20. Cylindrical-post magnetron sputtering source with electrostatic endconfinement. From Ref. 110.

The magnetron configuration shown in Fig. 5.20 has been termed the “cylindrical-post magnetron.”[113]-[115] It provides the simplest geometry for explaining the principles of magnetron operation.[113]-[118] The cathode consists of a cylindrical barrel with end plates, all composed of the material ¯ to be sputtered. It is mounted in a chamber with a uniform magnetic field B directed parallel to the cathode axis. The magnetic field is of such strength

308


(a few hundred gauss or less) that it affects the plasma electrons but not the ions. Figure 5.21 shows a typical chamber configuration for cylindrical magnetrons. A set of solenoidal field coils is positioned surrounding the cylindrical vacuum wall, which is constructed of a nonmagnetic material. A magnetic steel shell surrounds the coils and makes contact with the chamber top and bottom plates which are also fabricated from a magnetic material such as low-carbon steel. Thus, a low reluctance return path is provided for the solenoidal flux, as indicated in the figure, with the consequence that the coil system efficiently provides a uniform magnetic field within the chamber.

Figure 5.21. Chamber and magnetic field coil configuration used for cylindricalpost magnetron sputtering sources.

Secondary electrons which are emitted from the cylindrical-magnetron cathode barrel due to ion bombardment find themselves trapped in an annular cavity which is closed on three sides by surfaces at cathode


309

potential and on the fourth side by the magnetic field. Anode rings are located adjacent to the end plates on one or both ends of the cathode. Therefore, electrons emitted from the cathode must migrate radially across the cavity in order to reach the anode. Electron collisions of the type required to sustain a plasma discharge play an essential role in allowing this migration to occur. Thus, in contrast to the conventional discharge, the electrons are forced to make the required collisions, and an effective sputtering discharge is maintained in the cavity. The electron motion can be understood as follows.[119][120] When an electron is in a uniform magnetic field, its motion perpendicular to the field lines can be pictured as an orbit around a field line, as shown in Fig. 5.22a. Its motion along the field is unimpeded, so that if it has a component of velocity along the field line its net motion is a spiral as shown in Fig. 5.22b. Such electrons can be considered to be trapped on magnetic field lines and can advance to adjacent field lines by making a collision, as indicated schematically in Fig. 5.22c. An electron will also undergo a drift motion across the magnetic field if an electric field E is present. However, this ¯xB ¯ drift, is not in the direction of the electric field but in motion, known as E ¯xB ¯ a direction perpendicular to both the electric and the magnetic fields. E drift has the cycloidal form shown in Fig. 5.22d if the initial electron energy is small compared to that gained from the electric field, and the more circular motion shown in Fig. 5.22e if the initial electron energy is large compared to the electric-field-induced variations during the course of an orbit. Referring back to Fig. 5.20, a radial electric field also exists in the annular cavity. The field will be strong in the sheath region adjacent to the cathode but relatively weak at larger radii. Electrons emitted from the cathode will therefore undergo motions of the type shown in Fig. 5.22d and will become trapped in orbits revolving around the cathode. They will be able to advance radially only by making collisions or by the action of plasma oscillations which produce azimuthal electric fields and radial drifts.[113][114] Since the electrons leave the cathode sheath with energies of several hundred eV, electron collisions with gas atoms have a high probability of ¯ xB ¯ drift motions of the primary electrons and resulting in ionization. The E the products of ionization collisions produce an intense azimuthal current sheet of trapped electrons adjacent to the cathode. Because of the free axial movement of the electrons along the field lines, the sheet tends to be uniform along the cathode length. Large numbers of ions are produced which give rise to uniform sputter-erosion and high sputtering rates along the cathode barrel.

310


Figure 5.22. Electron motion in static magnetic and electric fields.

As the electrons give up energy by collisions and become ultimate electrons,[102] they move into regions of weak electric field at larger radii, ¯xB ¯ drift and their motion becomes more like that shown in Fig. 5.22e. The E velocity is relatively small, and the electrons move primarily up and down the field lines, reflecting off the end plates as indicated in Fig. 5.20.[121] When they reach the anode radius R, they immediately pass into the anode. Therefore, the high mobility of the electrons along the magnetic field lines causes the anode ring to be projected as a virtual anode sheet which surrounds and terminates the plasma discharge but is transparent to the sputtered flux.[113][114] Thus the electrons are trapped within the annular cavity throughout their lifetimes. The ions are constrained electrostatically to stay with the electrons and are therefore largely confined to this region as well. Consequently, there is virtually no plasma bombardment of substrates located beyond the anode radius.[37] Low-energy ion irradiation of the growing film (in order to controllably alter microstructure and/or microchemistry)can be induced, however, by "unbalancing" the magnetic confinement as discussed at the end of this chapter in Sec. 4.2.


311

Because of the efficiency of the ionization mechanisms in the magnetron cavity, intense plasma discharges capable of providing high sputtering rates can be maintained at moderate and near-constant voltages, even at low pressures. Deposition rates will depend on the radial position of the substrates. Typical operating conditions for cylindrical magnetrons of the type shown in Fig. 5.20 are:[114] Cathode current density - 20 mA/cm2 Discharge current - 1 to 50 A Discharge Voltage - 800 V Argon Pressure - 1 mTorr (0.13 Pa) Cathode erosion rate - 1.2 µm/min Substrate position - radius equal to 6 cathode radii Deposition rate - 200 nm/min (12 µm/h) While these deposition rates are approximately 10x lower than those of planar magnetrons, the deposition areas are proportionally larger since the substrates surround the cylindrical source. An important attribute of cylindrical magnetrons is their capability of being scaled through a range of sizes while retaining common operating characteristics.[114] Cathodes which range in length from 0.1 m to 2.1 m have been used. Figure 5.23 shows a 2.1 m cylindrical magnetron designed for depositing decorative coatings.[122] Such long cathodes provide a large substrate placement area around the circumference. Substrates can also be passed on each side of a cylindrical post magnetron in systems that operate continuously or semi-continuously. However, the most common application is batch processing in which the substrates are arranged around the cathode as shown in Fig. 5.24. Post cathodes have been used to coat the insides of tubes up to 1.8 m long.[123] Cylindrical magnetrons can also be configured in the inverted or hollow cathode form shown in Fig. 5.25.[114][116][124] Long hollow cathodes have the property that the coating flux at all points within the cathode is approximately equal to the erosion flux at the wall. This makes hollow cathodes particularly effective for coating objects with complex shapes.[125] Cylindrical magnetrons can also be designed in an arrangement whereby the magnetic field lines are bent such that they intersect the cathode barrel as shown in Fig. 5.26. The annular cross section of the electron trap is now closed on three sides by the magnetic field and on the fourth side by the surface at cathode potential. The plasma has the form of a ring rather than a sheet. Therefore such systems are generally configured with several such electron traps along the cathode cylinder.[116][117][126] Magnetrons of this

312


type are often referred to as having magnetic end-confinement, as opposed to those shown in Figs. 5.20 and 5.25 which are referred to as having electrostatic end-confinement.[112][113] Magnetic end-confinement devices have also been operated in the inverted or hollow cathode form.[126]

Figure 5.23. Large cylindrical-post magnetron sputtering source with a length = 2.1 m. See Ref. 122.

Cylindrical magnetrons with electrostatic end-confinement can be configured with large-diameter cathodes which provide a large inventory of coating material. Furthermore, the material is used very efficiently because of the uniform sputter erosion along the cathode length. However, a potential disadvantage with cylindrical sources is that target fabrication may be difficult. Plasma rings can be confined over planar surfaces or within cylindrical surface cavities. This is the basis of the planar magnetron[127][128] and sputter or S-gun[129] configurations shown in Fig. 5.19. At the present time, these devices are the most widely used form of magnetrons. Like cylindrical magnetrons, planar magnetrons are attractive because of their ability to be scaled to large sizes. Elongated planar magnetrons are particularly useful


313

for coating large substrate surfaces that are transported in a direction perpendicular to the long axis of the cathode. In this arrangement, close cathode-substrate spacing and deposition rates of µm/min or more may be used. Proper cathode design, aided by minimal aperturing, can provide deposition uniformities of better than ± 5%.[130]

Figure 5.24. Typical arrangement of substrates for batch processing with a cylindrical-post magnetron sputtering source.

Figure 5.25. Cylindrical-hollow magnetron sputtering source with electrostatic end-confinement. From Ref. 110.

314


Figure 5.26. Cylindrical magnetron with magnetic end-confinement. From Ref. 110.

Figure 5.27 shows a schematic drawing of an in-line system, with vacuum interlocks, which is typical of the apparatuses that are used to achieve high production volumes with planar magnetrons. Large systems of this type are used to coat architectural glass panels several square meters in size at production volumes of 106 m 2/yr. The planar magnetrons are typically 2 to 3 m long and are driven by currents in the 100 to 200 A range. Cathodes as long as 6.5 m have been considered for architectural glass coating.[131] Large circular planar magnetrons with arrays of concentric plasma rings 0.6 m in diameter have also been reported.[132] Patents have been granted for various configurations of the sputtering sources shown in Fig. 5.19.[133]-[137] Gun-type magnetrons do not have the scaling capabilities of the other forms of magnetrons. However, arrays can be used to coat large areas. The system shown in Fig. 5.4 uses an array of twenty-four S-gun sources in an inline configuration in which the substrates are transported relative to the sputtering source. Magnetron sputtering sources can be used to deposit magnetic as well as nonmagnetic materials. However, when a sputtering target composed of magnetic materials is used, it must be saturated magnetically so that its magnetic behavior is suppressed and a field of the desired shape can be maintained over its surface.[114]


315

Figure 5.27. Schematic illustration of an in-line system, with vacuum interlocks, used to achieve high production volumes with planar magnetrons.

DC magnetrons are typically operated at discharge currents in the range of 1 to 50 A. Current densities are generally higher in the plasma-ring devices (Figs. 5.19 and 5.26) than in the plasma sheet devices (Figs. 5.20 and 5.25). Total coating material fluxes (which are dependent on the total discharge current) can be comparable at high currents to those obtained with evaporation systems. The current-voltage characteristic reveals a great deal about the ionization processes in a plasma discharge. The more efficient the discharge, the lower the voltage for a given cathode current density. Discharges operating in the magnetron mode obey an I-V relationship of the form I proportional to Vn, where n is an index to the performance of the electron trap and is typically in the range 5 to 9. Typical I-V curves for various types of magnetron sputtering sources are shown in Fig. 5.28 and compared with an I-V curve for a planar diode. A basic disadvantage of the plasma-ring devices is that sputtering occurs only under the plasma rings. Troughs are eroded into the cathode, and the source material is used relatively inefficiently. Relative motion between the cathode and the magnetic field pattern is sometimes provided to improve the target usage.[127] A further disadvantage of ring devices is that the complexity of the magnetic-field shape makes effective anode placement more difficult than for cylindrical magnetrons.[114][127] Some field lines will intersect the substrates, as shown in Fig. 5.26, thereby allowing electron bombardment of the substrates. However, the bombardment intensity is much less than in planar diode sputtering sources. All plasma-ring magnetrons offer the advantage that the required magnetic field can be produced by permanent magnets located within the cathode rather than by magnetic field coils located at or beyond the chamber walls, as is required for cylindrical magnetrons.

316


Figure 5.28. Typical current-voltage characteristics for a planar diode sputtering source and for various types of planar and cylindrical magnetrons. All sources were operated with Al targets at the Ar working-gas pressures indicated.

At typical operating pressures (≈1 mTorr), the sputtered flux from magnetron sources passes to the substrate while undergoing very little gas scattering. Thus the deposition flux can be predicted with reasonable accuracy by assuming a cosine emission of sputtered material from the erosion area and collisionless passage to the substrates.[114][127] Figure 5.29 shows calculated and experimental profiles for the 2.1 m long cylindrical post magnetron in Fig. 5.23.[122] Figures 5.30 and 5.31 show typical deposition flux profiles for planar magnetrons of the ring and rectangular types, respectively.


317

Figure 5.29. Comparision of experimental deposition profile with calculated profile for long (2.1 m) cylindrical-post magnetron with electrostatic end-confinement. The profile was measured parallel to the cathode axis at a radius of 0.86 m. Data from Ref. 122.

Figure 5.30. Deposition-rate profile for a ring-type planar-magnetron sputtering source at various distances from the cathode surface.

318


Figure 5.31. Deposition rate profile for a rectangular-type planar-magnetron sputtering source along the long axis (A-A) at various distances from the cathode surface.

3.4 RF Sputtering DC methods cannot be used to sputter nonconducting targets because of charge accumulation at the target surface. This difficulty can be overcome by using radio frequency (RF) sputtering.[2][66][138]-[142] A single RF sputtering apparatus can be used to deposit electrically conducting, semiconducting, and insulating coatings. Consequently, RF sputtering has found wide application in the electronics industry. Nonconducting and semiconducting materials which have been deposited by RF sputtering include elemental semiconductors: Si[143] and Ge;[144] III-V compounds: GaAs, [145] GaSb,[146] GaN,[147] and AlN;[148] II-VI compounds: CdSe [149] and CdS;[67] IV-VI compounds: PbTe;[150] refractory semiconductors: SiC;[151] ferroelectric compounds: Bi4Ti3 O12;[152] oxides: In2O3,[153] SiO2 ,[154][155] Al2O3 ,[156][157] Ta2 O5,[158] Y2 O3 ,[159] TiO2 ,[160] ZrO2,[161] SnO2,[162] PtO,[163] Bi2O3,[164] ZnO,[165] and CdO;[166] pyrex glass; [167] and plastics.[168][169] Often several targets are placed within a common vacuum enclosure so that multilayer coatings can be deposited without breaking vacuum.


319

The usefulness of RF methods for sputtering nonconducting materials is based upon the fact that a self-bias voltage, negative with respect to the plasma floating potential, develops on any surface that is capacitively coupled to a glow discharge.[170] The basis for this potential, which forms as a consequence of the difference in mobility between electrons and ions, is illustrated schematically in Fig. 5.32. The current-voltage characteristic for an electrode immersed in a plasma is given in Fig. 5.32a. The floating potential is negative relative to the plasma potential by an amount that depends upon the gas species and plasma electron energy distribution function, but is typically -20 to -50 V and therefore too low to produce significant sputtering of most materials. When an alternating voltage is applied to such an electrode, more electron current flows when the electrode is positive relative to the floating potential than ion current flows when the electrode is negative relative to the floating potential (Fig. 5.32b). Capacitive coupling requires that there be no DC current flow; i.e., the net current to the electrode in each RF cycle must be zero. Accordingly, a negative bias must form such that the electron current on the positive side of the cycle becomes equal to the ion current on the negative side. The negative bias is approximately equal to half the peak-to-peak voltage of the RF signal and therefore can be made large enough to produce sputtering. The behavior illustrated in Fig. 5.32 applies strictly to the case where the electrode is passive; i.e., is not responsible for sustaining the plasma discharge. The planar diode shown schematically in Fig. 5.15 is the most commonly used apparatus for RF sputtering. The electrodes sustain the discharge and therefore have slightly different current-voltage characteristics than the one shown in Fig. 5.32, particularly at negative voltages. However, the overall effect when an RF potential is superimposed on the I-V characteristic is essentially identical. Figure 5.33 shows a schematic drawing of a typical RF planar-diode sputtering configuration in which a nonconducting target is placed over one electrode and substrates are placed on the other one. The electrodes reverse cathode-anode roles on each half cycle. The discharge is operated at a frequency that is sufficiently high that significant ion charge accumulation does not occur during the cycle time when an electrode is serving as a cathode.[11] Frequencies in the low MHz range are required. Most systems are operated at a frequency of 13.56 MHz, since this has been allocated by the Federal Communications Commission for industrial-scientific-medical purposes. Operation at other frequencies requires careful shielding to assure compliance with FCC regulations on radio interference.

320


Figure 5.32. Schematic illustration of the development of a negative bias when an RF potential is capacitively coupled to a probe immersed in a plasma. From Ref. 170.


321

At MHz operating frequencies, massive ions cannot follow the temporal variations in the applied potential. However, the electrons can. Thus the cloud of electrons in the negative glow plasma can be pictured as moving back and forth at the applied frequency in a sea of relatively stationary ions. As the electron cloud approaches one electrode, it uncovers ions at the other electrode to form a positive ion sheath. This sheath takes up nearly the entire applied voltage, the same as in the DC case.

Figure 5.33. Schematic drawing of a planar RF diode sputtering device. From Ref. 13.

A non-conducting target constitutes a capacitor in the electrical circuit between the electrodes (an external capacitor would have the same effect). Thus there can be no DC component to the current flow. The total ion and electron charge flow to a given electrode during an RF cycle must balance to zero, as discussed previously. However, a large electron current flows to a given electrode as the electron cloud makes contact. Thus the electron cloud need approach a given electrode for only a small fraction of a half cycle for purposes of supplying sufficient electrons to fulfill the anode requirement; i.e., to balance the entire ion flux through the cycle. Accordingly, in the steady state both electrodes develop a negative DC bias relative to the plasma potential, such that the electrodes approach or exceed the plasma potential (and become anodes) for only very short portions of their RF cycle as indicated in Fig. 5.34.

322


Figure 5.34. Approximate representation of target voltage waveforms relative to the plasma potential for a balanced RF system with two equal-area sputtering electrodes. Vf is the floating potential.

The motion of the ions, because of their inertia, can be thought of as responding to the DC potential and passing to both electrodes throughout the cycle. The electron cloud spends most of its time near the center position between the electrodes. Visually, the discharge appears as a DC discharge with a cathode dark space over each electrode. Functionally, sputtering occurs continually at both electrodes. RF discharges in planar diode systems can be operated at considerably lower pressures than can DC discharges. Typical operating pressures are 5 to 15 mTorr. There are two reasons: a reduction in the loss of primary electrons, and at high frequencies, an increase in the volume ionization efficiency. A fraction of the lower-energy primary electrons are repelled from the electrode toward which they are accelerated and thus remain in the discharge longer to make additional ionizing collisions. In addition, electrons can gain energy from the RF field by making in-phase collisions with gas atoms. That is, if an electron, accelerated in one direction during a given halfcycle, makes an elastic collision in which its direction is reversed near the end of the half-cycle, it maintains most of its velocity (due to the large mass mismatch between electrons and ions) and will again be accelerated during the next half-cycle and thus have gained energy during the complete cycle.


323

The versatility of RF sputtering is not achieved without drawbacks. Implementation of the process is complicated.[66] A typical electrical circuit is shown schematically in Fig. 5.35. It consists of an RF power supply, an inductive coupling to the load, and a matching network.

Figure 5.35. Schematic circuit of a single-ended RF discharge system including an equivalent circuit for the plasma discharge. See Ref. 142 and 173.

An equivalent circuit for an RF glow discharge is also shown in Fig. 5.35. The equivalent circuit assumes that both electrodes and the chamber walls are in contact with the plasma, and that the impedance is dominated by the plasma sheaths. The sheath capacitances result from the charge separation across the dark space. These capacitors are shunted to the electrode surface by a resistor to account for the ion current, and by a diode to account for the high electron current that can flow from the plasma to an electrode that is biased positive relative to the plasma potential. The capacitor Ct accounts for capacitance of the target. Cb is a blocking capacitor that is added to make the system independent of variations in the target capacitance. The RF current through the plasma is principally an electron current caused by the relative motion of the electron cloud. To the extent that there is no volume power transfer from the oscillating electrons to the gas, this current is out of phase with the applied voltage. The primary power transfer

324


occurs via the relatively small ion and electron current components that are in phase with the voltage. Thus, in the equivalent circuit approximation, the power transfer to produce sputtering occurs as the ion currents pass through the sheath resistances. Efficient power transfer requires that the RF power supply operate into a resistive load. Therefore a matching network is used to introduce inductance, and often capacitance, into the circuit in such a way that, in combination with the load, they form a resonant circuit.[171] When the variable matching network components are tuned to resonance, high circulating currents flow within the resonant circuit. However, the power supply sees only the resistive component of the load, the current passing from the power supply to the resonant circuit is in phase with the load and represents the power passing to the load. Many commercial sputtering sources monitor the reflected power from the load as an index of how effectively the matching network is adjusted. The reflected power should be minimized.[32][141][172] The ion current, and thus the sputtering rate at a given electrode, is determined by the average difference in potential between the electrode and the plasma. Thus it is useful to consider the plasma potential as a zero-point reference voltage in examining the performance of RF sputtering systems. The electrical character of RF sputtering systems can be classified in general as being either balanced or single-ended. In a balanced system, both electrodes are configured as identical sputtering targets and their potentials are 180o out of phase. The average sputtering voltage is about equal to half the peak-to-peak applied RF potential. The link center trap is placed at ground potential to stabilize the system and the chamber walls and substrates are connected to the center tap ground, as shown in Fig. 5.36. Since this point is at zero potential relative to the RF voltage, no RF current will flow to these elements. Furthermore, because of the capacitance in series with each of the electrodes, there is no DC current path from the plasma to the wall and substrates and then back through the electrodes into the plasma.[173] Thus a charge will develop on the capacitors such that the substrates float at a potential slightly negative with respect to the plasma, just like a floating electrode in a DC plasma. In an unbalanced RF system, the electrode on which the substrates are placed is made considerably larger than the target electrode.[142] This makes the sheath capacitance large, and the RF voltage drop across the substrate electrode small, as shown schematically in Fig. 5.37. The chamber and one side of the link are generally grounded (Fig. 5.35). Again, the capacitance in both electrode circuits prevents a DC current flow to the chamber, and a negative bias develops relative to the plasma potential. A


325

potential relative to the plasma potential will exist on the substrates and chamber wall unless the substrate electrode area is large enough to reduce this potential to essentially zero and to move the RF balance point to the grounded end of the link. It is important that these voltage drops be small so that sputtering from uncontrolled surfaces does not introduce contamination into the coatings. An impedance may be added to the substrate electrode circuit so that the potential of this electrode relative to the plasma can be controlled for purposes of bias sputtering.[173]

Figure 5.36. Schematic representation of an equivalent circuit for a balanced RF system with two equal-area sputtering electrodes and center-tap ground. The matching network is not shown.

The above discussion has been presented in the context of planar diode sputtering systems. Magnetron sputtering sources can also be used for RF sputtering. Cylindrical-post,[114][116] cylindrical-hollow,[114][174] planar, [127] and gun-type[129] magnetrons have all been successfully operated with RF power. However, some problems are encountered. Magnetron sputtering technology is basically a DC concept. The cathodes are shaped such that, in concert with the magnetic field, they form electron traps with specific symmetry. Anodes are placed to collect electrons which diffuse out of the trap.

326


Figure 5.37. Approximate representation of voltages (as functions of time) relative to the plasma potential for a single-ended RF sputtering system in which the wall area is much larger than target area. Vs is the substrate ion bombardment potential. See Ref. 142 and 173.

Effective double-ended RF magnetrons can be provided for some geometries. These configurations provide independent traps for both electrodes but allow magnetic coupling between them so that the electrons leaving one trap can diffuse freely to the vicinity of the other.[116] However, most magnetron configurations must be operated with single-ended arrangements. The magnetic confinement produces gradients in the plasma density, so that special care is required to minimize the voltage and therefore the sputtering rate at the counter electrode. Furthermore, in the planar magnetron case, the current-density concentration under the plasma ring requires that the power level be limited to avoid cracking when using targets with low thermal conductivities. When magnetron sources are driven single-ended, they generally operate in hybrid modes with current-voltage characteristics which are not representative of true magnetron behavior. Nevertheless, they provide deposition rates that are typically a factor of three greater than those achieved with RF planar diodes. (This is to be compared to the factor of twenty-to-thirty improvement in deposition rate which DC magnetrons provide over DC diodes when sputtering metals). Reduced electron bombardment and substrate heating are other advantages of magnetrons, as opposed to planar diodes, for RF sputtering.


327

3.5 Ion-Beam Sputtering Glow discharge sputtering technology is limited in the sense that the target current density and voltage cannot be independently controlled except by varying the working-gas pressure. An exception is the cylindrical magnetron where the voltage can be varied at a fixed current and pressure by varying the magnetic field strength. Ion-beam sputtering permits independent control over the energy and the current density of the bombarding ions.[175] A sputtering target is arranged to obliquely intersect an ion beam of given energy and flux density that is created by an independent ion source. Substrates are suitably placed to receive the coating flux, as shown in Fig. 5.38. In addition to the independent control over the ion current and voltage, ion beam sources permit sputtered coatings to be deposited at very low inert working-gas pressures (≤ 0.1 mTorr) onto substrates that are not in contact with a plasma.

Figure 5.38. Schematic representation of ion-beam sputtering showing relative locations of target and substrate.

Early ion sources were of the duoplasmatron type, where an ion beam was extracted through an aperture from a low pressure arc.[176] Hollowcathode ion sources were also used.[177] These devices were limited for practical deposition because of the small ion-beam sizes (≈1 cm). The recent adaptation of ion thruster technology has provided distributed ion sources with ion beams of relatively large diameter (≈10 - 30 cm). [175]

328


Although these devices cannot compete as deposition sources with the very large substrate areas that are provided by magnetrons, they are attractive for ion beam etching and for special deposition applications and research studies. Sources designed for etching have provided 500 eV, 1 mA/cm2, ion beams that are 30 cm in diameter with a uniformity of ±5% over 20 cm.[178] Reviews of the fundamentals of ion beam deposition are provided in Ref. 21 and 179.

4.0 SPUTTER DEPOSITION MODES Sputter deposition, in any of the configurations discussed above, can be carried out in a variety of modes developed to provide better control over film chemistry and/or microstructure. The most important of these modes of operation are reactive sputtering and bias sputtering. 4.1 Reactive Sputtering Reactive sputtering is a process in which a fraction of at least one of the coating species enters the deposition system in the gas phase. The target is typically either a pure metal (or metal alloy) or a compound containing volatile species. In the former case, the high vapor-pressure species, e.g., N in TiN,[180] S in CdS, [181] or O in VO2 ,[182] is provided entirely in the gas phase (via N 2, H 2 S, and O2 , respectively) while in the latter case (e.g., GaAs in As4 )[145] a considerably smaller partial pressure of the reactive gas is added to the discharge to account for the less than unity sticking probability of that species at the growing film surface. Bibliographies and reviews of early work covering a wide variety of compounds including oxides, nitrides, sulfides, carbides, etc. may be found in Refs. 98 and 183. The advantages of reactive sputtering are: (i) compounds can be formed using relatively easy-to-fabricate metallic targets, (ii) insulating compounds can be deposited using DC power supplies, and (iii) films with graded compositions can be formed. The difficulty in the reactivesputtering process is the complexity which accompanies its versatility. Chemical reactions occur at the target, at the substrate, and in cases of very high working pressures, in the gas phase. When sputtering with a reactive-gas/Ar mixture, the relationship between film properties and the reactive gas injection rate is generally very nonlinear. The condensing films can be considered as an additional pump for the reactive gas. The


329

nonlinearity occurs because the sticking probability (or getter-pump speed) of the condensing coating depends in a complex way on its growth rate, composition, film structure, and temperature. The composition dependence is shown in Fig. 5.39 for N2 incident on a growing Ti film.[184] Note that as the number of N2 molecules adsorbed per Ti atom deposited approaches 0.5 (i.e., a stoichiometric TiN film), the sticking probability α drops by more than two orders of magnitude. The decrease inα occurs as the number of unoccupied surface adsorption sites decreases.[185] Thus, for example, when sputtering in an N2 /Ar mixture at low reactive-gas injection rates, virtually all of the injected gas can react with the film. Consequently, the nitrogen is largely removed from the working gas, and the cathode process becomes primarily one of simple Ar sputtering of a metal. The coatings deposited under such conditions are generally metallic in nature.

Figure 5.39. Sticking coefficient of N2 measured during the continuous deposition of Ti as a function of the ratio of the getter-pumped nitrogen flux to the Ti deposition flux. Data from Ref. 184.

330


As the reactive-gas injection rate approaches that required to produce a stoichiometric coating, there is an increase in the reactive-gas partial pressure in the sputtering system because of the reduced getter-pumping rate of the depositing coating. This change in the composition of the sputtering gas greatly changes the processes which occur at the cathode surface. The result is that, for most metal/reactive-gas combinations, the sputtering discharge undergoes a transition into a mode in which the metal sputtering rate, and therefore the reactive-compound deposition rate, is reduced. The cathode surface reactions in this mode produce an energetic flux of highly reactive gas atoms and molecular fractions which accompany the sputtered metal atoms to the substrate. This large flux of reactive species makes the reactive sputtering process so effective for producing a wide range of compounds. The variation in discharge voltage and relative deposition rate during a typical transition is shown in Fig. 5.40. The voltage decrease at high oxygen injection rates is the metal-to-compound transition. The voltage increase at low injection rates is the compound-to-metal transition. The reduction in sputtering rate shown in Fig. 5.40 results primarily from compound formation on the cathode surface and the reduced sputtering yield of the reactive-gas molecules. The compounds often have higher electron secondary emission coefficients which give rise to a reduction in both the discharge voltage and the ion component in the cathode current for discharges driven at constant currents. The hysteresis effect, which is shown for discharge voltage but also applies to the deposition rate, occurs since the target compound layer, once formed, will remain until the sputtering gas is made sufficiently lean in the reactive species that a net sputter removal of the layer can occur. A cathode on which such a layer has formed is often referred to as being “poisoned.” The effect of cathode poisoning on the reactive sputtering process depends on the metal/reactive-gas combination and the properties of the cathode surface layer. Thus the very pronounced poisoning effect shown in Fig. 5.40 occurs for the oxygen reactive sputtering of materials such as Al, Cr, Ti, and Ta that form strong oxides. The decrease in deposition rate is generally less for other reactive gases such as N2. No poisoning occurs for Au, where the sputtering rate with pure O2 is not much different from that with Ar. The poisoning effect introduces two practical problems. One is the loss in deposition rate. The second is that during the transition, the material being deposited often passes abruptly from a metal to a nearly stoichiometric compound. Intermediate materials such as suboxides therefore become difficult to deposit. Consequently, considerable work has been directed toward trying to operate sputtering sources at, or very near, the transition


331

point. Transition behavior has been observed in planar diodes,[185] planar magnetrons,[186] and cylindrical magnetrons.[114] Many papers have been written concerning the transition mechanism.[185][187]-[192] Most are incomplete, however, because they concentrate on the cathode processes and do not consider the total system. It is important to realize that the reactive sputtering process is dependent on the total system; i.e., its geometry, the accumulation of coating on walls and fixtures and the positions of gas injection. All these parameters must be carefully controlled in order for reactive sputtering to be effectively used on a production basis.

Figure 5.40. Transitions in the steady-state operating mode of a Cr cylindrical-post magnetron sputtering source due to injection of oxygen.

332


Berg and co-workers[193]-[195] have recently developed nonlinear models of the reactive sputtering process, which do account for both internal getter pumping and external pumping, to predict stability conditions in order to investigate hysteretic behavior. They showed that with a sufficiently high overall pumping speed, there is a smooth transition between the metal and compound sputtering modes. Critical pumping speeds required to eliminate hysteretic behavior can be estimated based upon their work. Several techniques have been developed to increase the deposition rate during reactive planar magnetron sputtering which take advantage of the nonuniform cathode current densities in these devices.[196]-[198] Some of these are illustrated schematically in Fig. 5.41. The reactive gas flux, and therefore the tendency for reactive gas adsorption, is relatively uniform over the cathode surface. However, the ion flux is nonuniform and causes sputter removal of adsorbed reactive species, thereby reducing their surface coverage under the plasma ring (Fig. 5.41b). At higher current densities (Fig. 5.41c), the sputtering rate is adequate to maintain a fresh metal surface which in turn yields a high rate of sputtering. By exerting control over the total system— i.e., by arranging the reactive gas injection adjacent to the substrate and the Ar injection adjacent to the target—it is possible, with the assistance of suitable getter surfaces, as illustrated in Fig. 5.40d, to maintain a gradient in the composition of the reactive gas in the sputtering atmosphere. If the gradient is adequate, the target surface under the plasma rings can remain unpoisoned and yield a high flux, even when the reactive-gas flux to the substrate is adequate to produce a stoichiometric compound.[196] Baffles have also been used between the target and substrate to maintain a gradient in reactive gas partial pressure.[198] For reactive sputtering in N2 , where the hysteretic behavior is more gradual than in O2, (see, for example, Ref. 199), Sproul [200]-[203] has developed feedback control techniques which allow film deposition rates of transition-metal nitrides such as TiN, ZrN, and HfN at values very nearly equal to those of the pure metals. The feedback controls maintain constant target power, total pressure, and N2 partial pressure. 4.2 Bias Sputtering Bias sputtering, in which the substrate is biased negatively with respect to the plasma potential, is often used to provide low-energy ion bombardment of the growing film. The analog of glow-discharge bias sputtering can be carried out in an ion-beam deposition system using a second ion gun to irradiate the substrate.[17] Low-energy ion/surface interactions during film


333

growth is a subject of study unto itself and recent reviews may be found in Refs. 22 - 25. Ion irradiation has been shown to strongly affect film nucleation and growth kinetics, adhesion, film microstructure and chemistry, and hence film properties. Mechanisms by which ion/surface interactions modify film nucleation, growth, microstructure evolution, and film properties are discussed in Ch. 13 while reviews on adhesion[204][205] andion-induced changes in film chemistry (including trapping, secondary implantation, and preferential sputtering) [24][206] have been published recently.

Figure 5.41. Schematic illustration of various elements of the planar magnetron reactive sputtering process.

Recently, Window and Savvides[207][208] demonstrated that substrate ion currents during magnetron sputter deposition can be influenced by stray magnetic fields B leaking from the target magnet assembly. B can be varied intentionally over a limited range by changing the relative strengths of the inner and outer target magnetic poles, i.e., by "unbalancing" the magnetron. A similar concept was employed by Petrov et al.[209] and Adibi et al.[210] in developing an ultra-high-vacuum DC planar-magnetron

334


(PM) sputter deposition system with an external variable axial magnetic field Bext superimposed on the permanent magnetic field of the PM as shown in Fig. 43. A pair of Helmholtz coils, located outside the vacuum chamber, produces Bext which is uniform along the axis orthogonal to both target and substrate surfaces.

Figure 5.42. Schematic diagram of an ultra-high-vacuum reactive-magnetron DC sputter-deposition system with both variable external and permanent internal magnets.


335

As shown schematically in Fig. 5.43, the value and sign of Bext has a strong effect on the plasma density near the substrate, and hence on the ion flux, Ji, incident at the substrate, with only a minor effect on the target-atom flux. For a Ti target sputtered in pure Ar at 20 mTorr with a target-substrate separation of 6.5 cm, changing Bext from -50 G (opposing the field of the outer PM pole) to +600 G (reinforcing the field of the outer PM pole) varied the ionto-Ti flux ratio Ji/JTi incident at the substrate by a factor of sixty from 0.1 to 6 with the bias held constant at any desired negative value between ~ -15 V (limited by the difference between the floating Vf and plasma Vp potentials) and the highest values examined, -100 V. For reactive sputter deposition in N2 (where the primary ion is N2+) under the same conditions, Ji/JTi varied by a factor of fifty from 0.7 to 35. Vp was negative with Bext set to positive values and ranged form ~ 0 (Bext = 0 G) to -13 V (Bext > +200 G) in Ar and 0 to -20 V in N2. Using an N2+ ion energy of 20 eV (10 eV per N) to bombard the growing film, Adibi et al.[210] showed that by varying Ji/JTi from 1 to ≥ 5.2, the microstructure of metastable NaCl-structure Ti0.5Al0.5N alloys deposited at 250°C could be controllably altered from a porous columnar structure with a complete (111) texture to a dense completely (002)-oriented structure with no residual ion-induced defects observable by high-resolution plan-view and cross-sectional transmission electron microscopy.

ACKNOWLEDGEMENTS The authors gratefully acknowledge the support of the Joint Services Electronics Program and the Materials Science Division of the Department of Energy over the course of several years.

336


Figure 5.43. Schematic diagram illustrating the effect on the plasma during sputter deposition of superimposing an external magnetic field B ext which (a) opposes and (b) reinforces the field of the outer permanent magnet in the planar magnetron target.


337

REFERENCES 1. Grove, W. R., Trans. Royal Soc. 142:87 (1952) 2. Chapman, B., Glow Discharge Processes; Sputtering and Plasma Etching, Wiley, New York (1980) 3. Dahlgren, S. D.,Proceedings 3rd. Int. Conf. on Rapidly Quenched Metals, Univ. Sussex, Brighton, England (1976) 4. Grubb, A. D., Airco Temescal, Carleton, Michigan (private communication) 5. Thornton, J. A.,Proceedings 20th Tech. Conf. Society of Vacuum Coaters, p. 5, Atlanta, GA (April 1977) 6. Thornton, J. A., Tabock, J. and Hoffman, D. W., Thin Solid Films 64:111 (1979) 7. Wehner, G. K., Advanced Elec. and Elec. Phys. 7:239 (1955) 8. Kay, E., Advances Elec. and Elec. Phys. 17:245 (1962) 9. Maissel, L. I., Physics of Thin Films, 3:61 (1966) 10. Kay, E., in Techniques of Metals Research, (R. F. Bunshah, ed.) p. 1269, Vol. 1, Part 3, Interscience, New York (1968) 11. Wehner, G. K. and Anderson, G. S., inHandbook of Thin Film Technology, (L. Maissel and R. Glang, eds.), p. 3-1, McGraw Hill, New York (1970) 12. Maissel, L., in Handbook of Thin Film Technology, (L. Maissel and R. Glang, eds.), p. 4-1, McGraw Hill, New York, (1970) 13. Thornton, J. A., SAE Transactions, 82:1787 (1974) 14. Westwood, W. D., Progress in Surface Sci., 7:71 (1976) 15. Thin Film Processes, (J. L. Vossen and W. Kern, eds.), Academic Press, New York (1978) 16. Greene, J. E., in Handbook of Semiconductors, (S. Keller, ed.), Vol. 3, p. 499, North Holland Pub. Co., Amsterdam (1980) 17. Greene, J. E., CRC Critical Rev. Sol. St. and Matrl. Sci. 11:47 (1983) 18. Greene, J. E., CRC Critical Rev. Sol. St. and Matrl. Sci. 11:189 (1984) 19. Kaufman, H. R., Cuomo, J. J. and Harper, J. M. E., J. Vac. Sci. Technol., 21:725 (1982) 20. Harper, J. M. E., Cuomo, J. J. and Kaufman, H. R., J. Vac. Sci. Technol., 21:737 (1982) 21. Harper, J. M. E., Sol. St. Technol., 30:129 (April 1987) 22. Greene, J. E. and Barnett, S. A., J., Vac. Sci. Technol., 21:285 (1982)

338


23. Greene, J. E., Motooka, T., Sundgren, J. E., Lubben, D., Gorbatkin, S. and Barnett, S. A., J. Nucl. Instr. Methods, B27:226 (1987) 24. Greene, J. E., Sol. St. Technol., 30:115 (April, 1987) 25. Greene, J. E., Barnett, S. A., Sundgren, J. E. and Rockett, A., in IonAssisted Film Growth, (T. Itoh, ed.), p. 101, Elsevier, Amsterdam 26. Harrison, D. E., Kelly, P. W., Garrison, B. J. and Winograd, N., Surf. Sci., 76:311 (1978) 27. Hagstrum, H. D., Phys. Rev., 104:317 (1956) 28. Hagstrum, H. D., Phys. Rev., 123:758 (1961) 29. McCracken, G. M., Rep. Prog. Phys., 38:241 (1975) 30. Navinsek, B., Progress in Surface Sci., 7:49 (1976) 31. Davis, W. D. and Vanderslice, T. A., Phys. Rev., 131:219 (1963) 32. Sigmund, P., Phys. Rev., 184:383 (1969) 33. Townsens, P. D., Kelly, J. C. and Hartley, N. E. W., Ion Implantation, Sputtering and Their Applications, p. 111, Academic Press, New York (1976) 34. Sigmund, P., J. Vac. Sci. Technol., 17:396 (1980) 35. Eckstein, W. and Biersack, J. P., Z. Phys., B63:471 (1986) 36. Vossen, J. L., J. Vac. Sci. Technol., 8:S12 (1971) 37. Thornton, J. A., Thin Solid Films, 54:23 (1978) 38. Winters, H. F. and Kay, E., J. Appl. Phys., 38:3928 (1967) 39. Brodie, I., Lamont, L. T. Jr. and Jepson, R. L., Phys. Rev. Lett., 21:1124 (1968) 40. Lee, W. W. and Oblas, D., J. Vac. Sci. Technol., 7:129 (1970) 41. Lee, W. W. Y. and Oblas, D., J. Appl. Phys., 46:1728 (1975) 42. Sigmund, P., Can. J. Phys., 46:731 (1968) 43. Anderson, H. H., Rad. Effects, 3:51 (1970) 44. Oechsner, E. and Gesang, W. R., Phys. Lett., 37A:235 (1971) 45. Kornelsen, E. V., Can. J. Phys., 42:364 (1964) 46. Blank, P. and Wittmaack, K., J. Appl. Phys., 50:1519 (1979) 47. Coburn, J. W., Thin Solid Films, 64:371 (1979) 48. Anderson, H. H., in Symp. on the Physics of Ionized Gases 1980, (B. Cobic, ed.) Boris Kidric Aust. Nucl. Sciences, Belgrad, Yugoslavia (1981) 49. Eltoukhy, A. H. and Greene, J. E., J. Appl. Phys., 51:4444 (1980)


339

50. Coburn, J. W., J. Vac. Sci. Technol., 13:1037 (1976) 51. Gillam, E., J. Phys. Chem. Solids, 11:55 (1959) 52. Greene, J. E., Natarajan, B. R. and Sequeda-Osorio, F., J. Appl. Phys., 49:417 (1978) 53. Nobes, M. J., Colligon, J. C. and Carter, G., J. Mater. Sci., 4:730 (1969) 54. Carter, G., Colligon, J. C. and Nobes, M. J.,J. Mater. Sci., 6:115 (1971) 55. Sigmund, P., J. Mater. Sci., 8:1545 (1973) 56. Stewart, A. D. G. and Thompson, M. W., J. Matl. Sci., 4:56 (1969) 57. Witcomb, M. H., J. Matl. Sci., 9:551 (1974) 58. Wehner, G. K. and Hajicek, D. J., J. Appl. Phys., 42:1145 (1971) 59. Tarng, M. L. and Wehner, G. K., J. Appl. Phys., 43:2268 (1972) 60. Oohashi, T. and Yamanaka, S., Jpn. J. Appl. Phys., 11:1581 (1972) 61. Shimizu, R., Jpn. J. Appl. Phys., 13:228 (1974) 62. Dahlgren, S. D. and McClanahan, S. D., J. Appl. Phys., 43:1514 (1972) 63. Vossen, J. L., J. Vac. Sci. Technol., 8:751 (1971) 64. Wheeler, D. R. and Brainard, W. A., J. Vac. Sci. Technol., 15:24 (1978) 65. Cistola, A. B., J. Vac. Sci. Technol., 17:420 (1980) 66. Vossen, J. L. and O’Neill, J. J. Jr., RCA Rev., 29:149 (1968) 67. Fraser, D. B. and Melchior, H., J. Appl. Phys., 43:3120 (1972) 68. Almen, O. and Bruce, G., Transac. 8th National Vac. Sym., p. 245, Pergamon Press, New York (1962) 69. Tsunoyama, K, Suzuki, T. and Ohashi, Y., Jpn. J. Appl. Phys., 15:349 (1976) 70. Anderson, C. A. and Hinthorne, J. R., Science, 175:853 (1972) 71. Benninghoven, A. and Mueller, A., Phys. Letters, 40A:169 (1972) 72. Wener, H. W., Surface Sci., 47:301 (1975) 73. Benninghoven, A., Surf. Sci., 53:596 (1975) 74. Williams, P., Materials Research Laboratory, University of Illinois, private communication 75. Hanak, J. J. and Pellicante, J. P., J. Vac. Sci. Technol., 13:406 (1976) 76. Cuomo, J. J., Gambino, R. J., Harper, J. M. E., Kuptsis, J. D. and Weber, J. C., J. Vac. Sci. Technol., 15:281 (1978) 77. Sandstrom, R. L., Gallagher, W. J., Dinger, T. R., Laibowitz, R. B., and Gambino, R. J., Appl Phys. Lett., 53:44 (1988)

340


78. Oechsner, H. and Gerhard, W., Surf. Sci., 44:480 (1974) 79. Gerhard,W. and Oechsner, H., Z. Physik, B22:41 (1975) 80. Gerhard, W., Z. Physik, B22:31 (1975) 81. Winograd, N., Harrison, D. E. Jr., and Garrison, B. J., Surf. Sci., 78:467 (1978) 82. Können, G. P., Tip, A. and deVries, A. E., Radiat. Eff., 21:269 (1974) 83. Können, G. P., Tip, A. and deVries, A. E., Radiat. Eff., 26:23 (1975) 84. Garrison, B. J., Winograd, N. and Harrison, D. E. Jr., J. Chem. Phys., 69:1440 (1978) 85. Prigge, S. and Bauer, E., in Proc. 2nd Int. Conf. Secondary Ion Mass Spectrometry, (A. Benninghoven, C. A. Evans, R. A. Powel, Jr., R. Shimizu, and H. A. Storms eds.), p. 133, Springer-Verlag, New York (1979) 86. Coburn, J. W., Taglauer, E. and Kay, E., Japn. J. Appl. Phys. Suppl., 2:501 (1974) 87. Gruen, D. M., Guadioso, S. L., McBeth, R. L. and Kerner, J. L.,J. Chem. Phys., 60:89 (1974) 88. Gruen, D. M., Finn, P. A. and Page, D. L.,Nucl. Technol., 29:309 (1976) 89. Steinbruchel, C. and Gruen, D. M., Surf. Sci., 93:299 (1980) 90. Comas, J. and Cooper, C. B., J. Appl. Phys., 38:2956 (1967) 91. Szymonski, M. and Bhattacharya, Appl. Phys., 20:207 (1979) 92. Chopra, K. L., Thin Film Phenomena, McGraw Hill, New York (1969) 93. Stuart, R. V., Wehner, G. K. and Anderson, G. S.,J. Appl. Phys., 40:803 (1969) 94. Oechsner, H., Z. Physik 238:433 (1970) 95. Bernhardt, F., Oechsner, H. and Stumpe, E., Nuc. Inst. Methods, 132:329 (1976) 96. Sigmund, P., Rev. Roum. Phys., 17:1079 (1972) 97. von Engle, A., Ionized Gases, Oxford University Press, London (1965) 98. Vossen, J. L. and Cuomo, J. J., in Thin Film Processes, (J. L. Vossen and W. Kern, eds.) Academic Press, pp. 1-73, New York, (1978) 99. Westwood, W. D. and Boynton, R., J. Appl. Phys., 43:2691 (1972) 100. McDaniel, E. S., Collision Phenomena in Ionized Gases, Chapter 13, Wiley, New York (1964)


341

101. Hagstrum, H. D., Ion-Surface Collision, Chapter 1, Academic Press, New York, (1977) 102. Thornton, J. A., J. Vac. Sci. Technol., 15:188 (1978) 103. Christophorou, L. G.,Atomic and Molecular Physics, p. 379, Wiley, New York (1971) 104. Houston, J. E. and Uhl, J. E., Sandia Report, Sc-RR-71-0122 (1972) 105. Stirling, A. J. and Westwood, W. D., J. Appl. Phys., 41:742 (1970) 106. Greene, J. E., J. Vac. Sci. Technol., 15:1718 (1978) 107. Abril, I., Gras-Marti, A. and Valles-Abarca, J. A., J. Vac. Sci. Technol., A4:1773 (1986) 108. Gras-Marti, A., Valles-Abarca, J. A. and Bersaoula, A., J. Vac. Sci. Technol., A5:2217 (1987) 109. Westwood, W. D., J. Vac. Sci. Technol., 15:1 (1978) 110. Thornton, J. A., Metal Finishing, 74:46 (1976) 111. Busch, R. and McClanahan, E. D., Thin Solid Films, 47:291 (1977) 112. Thornton, J. A., Metal Finishing, 77:45 (1979) 113. Thornton, J. A., J. Vac. Sci. Technol., 15:171 (1978) 114. Thornton, J. A. and Penfold, A. S., inThin Film Processes, (J. L. Vossen and W. Kern, eds.) p. 75, Academic Press, New York (1978) 115. Penning, F. M. and Moubis, J. H. A., Proc. Ned. Akad. Wet., 43:41 (1940) 116. Penfold, A. S. and Thornton, J. A., U.S. Patents 3,884,793 (1975) 117. Hosokawa, N., Tsukada, T. and Misumi, T., J. Vac. Sci. Technol., 14:143 (1977) 118. Korov, K. I., Ivanov, N. A., Atanasova, E. D. and Minchev, G. M.,Vacuum, 26:237 (1976) 119. Spitzer, L. Jr., Physics of Fully Ionized Gases, Interscience, New York (1956) 120. Chen, F. F., Introduction to Plasma Physics, Plenum Press, New York (1974) 121. Thornton, J. A., J. Vac. Sci. Technol., 16:79 (1979) 122. Penfold, A. S., Metal Finishing, 77:33 (1979) 123. Penfold, A. S., Telic Corporation (unpublished) 124. Penning, F. M., Physics, 3:873 (1936) 125. Thornton, J. A. and Hedgcoth, V. L.,J. Vac. Sci. Technol., 12:93 (1975)

342


126. Heisig, U., Goedicke, K. and Schiller, S., Proceedings 7th Intl. Symp. Electron and Ion Beam Science and Technology, Washington, DC, p. 129, Electrochemical Society, Princeton, NJ (1976) 127. Waits, R. K., J. Vac. Sci. Technol., 15:179 (1978) 128. Waits, R. K., in Thin Film Processes, (J. L. Vossen and W. Kern, eds.) p. 131, Academic Press, New York (1978) 129. Fraser, D. B., in Thin Film Processes (J. L. Vossen and W. Kern, eds.) p. 115, Academic Press, New York (1978) 130. Aronson, A. and Weinig, S., Vacuum, 27:151 (1977) 131. Van Vorous, T., Optical Spectra, p. 30 (November, 1977) 132. Smith, H. R. Jr., Proceedings 20th Annual Tech. Conf., Society of Vacuum Coaters, p. 1, Atlanta, GA (1977) 133. Corbani, J. F., U.S. Patent 3,878,085 (1975) 134. Clarke, P. J., U.S. Patent 3,616,450 (1971); U.S. Patent 3,711,398 (1973) 135. McLeod, P. S., U.S. Patent 3,956,093 (1976) 136. Rainey, R. M., U.S. Patent 4,100,055 (1978) 137. Chapin, J. S., U.S. Patent 4,166,018 (1979) 138. Vossen, J. L., J. Vac. Sci. Technol., 8:S12 (1971) 139. Davidse, P. D., Vacuum, 17:139 (1967) 140. Probyn, B. S., Vacuum, 18:253 (1968) 141. Jackson, G. N., Thin Solid Films, 5:209 (1970) 142. Koenig, H. R. and Maissel, L. I., IBM J. Res. Develop., 14:168 (1970) 143. Brodsky, M. H., Title, R. S., Weiser, K. and Pettit, G. D., Phys. Rev. B, 1:2632 (1970) 144. Messier, R., Takamori, T. and Roy, R., J. Vac. Sci. Technol., 13:1060 (1976) 145. Barnett, S. A., Bajor, G. and Greene, J. E., J. Appl. Letters, 37:735 (1980) 146. Eltoukhy, A. H. and Greene, J. E., J. Appl. Phys., 50:6390 (1979) 147. Hovel, H. J. and Cuomo, J. J., Appl. Phys. Lett., 20:71 (1972) 148. Shuskus, A. J., Reeder, T. M. and Paradis, E. L., Appl. Phys. Lett., 24:151 (1974) 149. Glew, R. W., Thin Solid Films, 46:59 (1977) 150. Corsi, C., J. Appl. Phys., 45:3467 (1974)


343

151. Wasa, K., Nagai, T. and Hayakawa, S.,Thin Solid Films, 31:235 (1976) 152. Takei, W. J., Formigoni, N. P. and Francombe, M. H., J. Vac. Sci. Technol., 7:442 (1970) 153. Wickersham, C. E. and Greene, J. E., Phys. Status Solidi (a) 47:329 (1978) 154. Morrison, D. T. and Robertson, T., Thin Solid Films, 15:87 (1973) 155. Hickmott, T. W., J. Appl. Phys., 45:1050 (1974) 156. Salma, C. A. T., J. Electrochem. Soc., 117:913 (1970) 157. Pratt, I. H., Solid State Technol., 12:49 (1969) 158. Young, P. L., Fehler, F. P. and Whitman, A. J., J. Vac. Sci. Technol., 14:176 (1977) 159. Goldstein, R. M. and Wigginton, S. C., Thin Solid Films, 3:R41 (1969) 160. Wu, L. C. and Greene, J. E., J. Appl. Phys., 50:4966 (1979) 161. Greene, J .E., Wickersham, C. E., Zilko, J. L., Welsh, L. B. and Szofran, F. R., J. Vac. Sci. Technol., 12:72 (1976) 162. Takao, T., Wasa, K. and Hayakawa, S.,J. Electrochem. Soc., 123:1719 (1976) 163. Westwood, W. D. and Bennewitz, C. D., J. Appl. Phys., 45:2313 (1974) 164. Titchmarsh, J. G. and Toombs, P. A. B., J. Vac. Sci. Technol., 7:103 (1970) 165. Paradis, E. L. and Shuskus, A. J., Thin Solid Films, 38:131 (1976) 166. Lakshmanan, T. K., J. Electrochem. Soc., 110:548 (1963) 167. Shimomoto, Y., Matsumaru, H. and Nishimura, T., Jpn. J. Appl. Phys., Suppl. 2, Pt. 1:701 (1974) 168. Robertson, T. and Morrison, D. T., Thin Solid Films, 27:19 (1975) 169. Biederman, H., Ojha, S. M. and Holland, L., Thin Solid Films, 41:329 (1977) 170. Butler, H. S. and Kino, G. S., Phys. of Fluids, 6:1346 (1963) 171. Terman, F. E.,Electronics and Radio Engineering,p. 458, McGraw Hill, New York (1955) 172. Logan, J. S., Mazza, N. M. and Davidse, P. D., J. Vac. Sci. Technol., 6:120 (1969) 173. Coburn, J. W. and Kay, E., J. Appl. Phys., 43:4965 (1972) 174. Thornton, J. A. and Chin, J., Ceramic Bulletin, 56:504 (1977) 175. Kaufman, H. R., J. Vac. Sci. Technol., 15:272 (1978)

344


176. Chopra, K. L. and Randlett, M. R., Rev. Sci. Instr., 38:1147 (1967) 177. Reader, P. D., White, D. P. and Isaacson, G. C., 14th Symp. Elect. Ion Photon Beam Tech., Palo Alto, CA (May 1977) 178. Robinson, R. S., J. Vac. Sci. Technol., 15:277 (1978) 179. Harper, J. M. E., inThin Film Processes (J. L. Vossen and W. Kern, eds.) pp. 175-206, Academic Press, New York, (1978) 180. Johansson, B. O., Sundgren, J. E., Greene, J. E., Rockett, A. and Barnett, S. A., J. Vac. Sci. Technol., B3:303 (1985) 181. Thornton, J. A., Cornog, D. G. and Anderson, W. W., J. Vac. Sci. Technol., 18:199 (1981) 182. Kusano, E., Theil, J. A. and Thornton, J. A., J. Vac. Sci. Technol., A6:1663 (1988) 183. Vossen, J. L., in Physics of Thin Films, (G. Haas, M. H. Francombe, and R. W. Hoffman, eds.) 9:1, Academic Press, NY (1977) 184. Harra, D. J. and Haywood, W. H.,Supplemento al Nuovo Cimento, 5:56 (1967) 185. Heller, J., Thin Solid Films, 17:163 (1973) 186. Maniv, S. and Westwood, W. D., J. Appl. Phys., 51:718 (1980) 187. Holland, E. and Campbell, D. S., J. Mater. Sci., 3:544 (1968) 188. Harvey, J. and Corkhill, J., Thin Solid Films, 6:277 (1970) 189. Stirling, A. J. and Westwood, W. D., Thin Solid Films, 30:19 (1975) 190. Abe, T. and Yamashina, T., Thin Solid Films, 30:19 (1975) 191. Shinoki, F. and Itoh, A., J. Appl. Phys., 46:3381 (1975) 192. Donaghey, L. F. and Geraghty, K. G., Thin Solid Films, 38:271 (1976) 193. Berg, S., Blom, H. O., Larsson, T. and Nender, C.,J. Vac. Sci. Technol., A5:202 (1987) 194. Berg, S., Larsson, T., Nender, C. and Blom, H. O.,J. Appl. Phys., 63:887 (1988) 195. Larsson, T., Blom, H. O., Nender, C. and Berg, S.,J. Vac. Sci. Technol., A6:1832 (1988) 196. Schiller, S., Heisig, U., Goedicke, K., Schade, K., Teshner,G. and Henneberger, J., Thin Solid Films, 64:455 (1979) 197. Schiller, S., Heisig, U., Steinfelder, K. and Strümpfel, J., Thin Solid Films, 63:369 (1979) 198. Este, G. and Westwood, W. D., J. Vac. Sci. Technol., A2:1238 (1984)


345

199. Natarajan, B. R., Eltoukhy, A. H., Greene, J. E. and Barr,T. L., Thin Solid Films, 69:217 (1980) 200. Sproul, W. D., Thin Solid Films, 107:141 (1983) 201. Sproul, W. D., Thin Solid Films, 118:279 (1984) 202. Sproul, W. D. and Tomashek, J. E., U.S. Patent 4,428,811 203. Sproul, W. D., U.S. Patent 4,428,812 204. Baglin, J. E. E., Matrl. Sci. Engr., B1:1 (1988) 205. Mattox, D. M., Baglin, J. E. E., Gottschall, R. E. and Batich, C. D., Adhesion in Solids, Vol. 119, MRS Symposium Proceedings (1989) 206. Greene, J. E., Barnett, S. A., Sundgren, J. E. and Rockett, A., inPlasmaSurface Interactions and Processing of Materials, (O. Auciello, A. Grass-Marti, and D. L. Flamm, eds.), p. 281, NATO Advanced Study Institute, Klewer, Boston (1990) 207. Window, B. and Savvides, N., J. Vac. Technol., A4:196 (1986) 208. Window, B. and Savvides, N., J. Vac. Technol., A4:453 (1986) 209. Petrov I., Adibi, F., Greene, J. E., Sproul, W. D., and Münz, W.-D., J. Vac. Sci. Technol., A10:3283 (1992) 210. Adibi, F., Petrov, I., Greene, J. E., Hultman, L., and Sundgren, J.-E., J. Appl. Phys., 73:8580 (1993)

6 Ion Plating Donald M. Mattox

1.0

INTRODUCTION

Ion plating is a generic term applied to atomistic film deposition processes in which the substrate surface and the growing film are subjected to a flux of energetic bombarding particles sufficient to cause changes in the film formation process and the properties of the deposited film. This broad definition does not specify the source of the depositing film particles, the source of bombarding particles, nor the environment in which the deposition takes place. The principal criterion is that energetic particle bombardment is used to modify the film formation process and film properties.[1][2] Figure 6.1 shows a simple ion plating system using a DC diode arrangement with the substrate as the cathode electrode and a thermal vaporization source as the source of the depositing material.[1] A description of the ion plating process was first published in the technical literature in 1964[3] and a U.S. patent was granted for the process in 1974.[4] Recently it has been found that a patent for a very limited version of a similar process was granted in Europe in 1938[5] but was not published in the technical literature nor pursued commercially. In 1965 it was reported that the use of a bias on the substrate during sputter deposition (bias sputtering) decreased the contaminant level in sputter-deposited films.[6] The term ion plating is generally accepted [7] and is often used without any description or reference, however there does not seem to be a universally accepted definition and many other terms are applied to ion plating-like processes such as bias sputtering, bias sputter deposition, “ion

346

Ion Plating

347

vapor deposition” (“Ivadizing”),[8][9] ion beam enhanced deposition (IBED), accelerated ion deposition,[10] plasma-enhanced vapor deposition,[11] ionassisted deposition (IAD),[12] biased activated reactive deposition (BARE),[13] plasma surface alloying, etc.

Figure 6.1. An ion plating configuration using a DC diode discharge and a thermal vaporization source.[1]

Early work on the ion plating process was concerned with thedeposition of atoms originating from thermal evaporation/sublimation, sputtered surfaces (sputter ion plating), or from chemical vapor species in the gas (chemical ion plating).[4] Bombardment was by ions extracted from a plasma by applying a negative potential to the surface to be coated which was immersed in the plasma. The plasma could be of an inert gas species, contain reactive species, or contain gaseous chemical compound species. In the latter cases, the chemical species could be activated in the plasma to either become more reactive with depositing species (for example, nitrogen to form the nitrides) or could be decomposed to deposit a coating from the constituents of the gaseous chemical compound (e.g., Ti from TiCl4).[14] The latter process might be considered an early form ofplasma enhanced CVD[15] with ion bombardment in addition to the plasma activation.

348


Most recently, the termion plating has been applied to processes where the surface to be coated is in contact with a plasma and the term ion assisted deposition (IAD) or ion beam enhanced deposition (IBED) is used where the substrate is bombarded by an ion beam in a vacuum environment during deposition.[16]-[18] There are several other modifying terms which are sometimes used with ion plating such as: sputter ion plating and chemical ion plating which specify the origin of the depositing species (sputtered material or chemical vapor precursor gases respectively),vacuum ion plating which is done in a vacuum environment,[19] and reactive ion plating,[20] used for the deposition of films of compound materials. Initially the film property of most interest was the adhesion of the film to the substrate and the ability to have an in situ substrate surface cleaning process, as well as to introduce thermal energy directly into the surface region without having to heat the bulk of the material. In the early 70’s a number of studies were made on the modification of film microstructure,[21][22] stress,[23] composition[24] and properties[25][26] by the concurrent bombardment during deposition, and the excellent coverage obtained by gas scattering and sputtering-redeposition during ion plating. Bombardment during reactive deposition was shown to improve the stoichiometry of the deposited film material. In the latter part of the 70’s and the early 80’s, sources of low energy (100 to 10,000 eV) ion beams of gaseous species became more generally available and bombardment could be done in a vacuum environment. This led to studies of bombardment effects under more controlled conditions. Studies included both those of a physical nature such as the effect of bombardment on film stress,[27] and those of a chemical nature (reactive deposition).[28][29] Recently it has been shown that bombardment during vacuum deposition greatly improves the properties of vacuum deposited optical coatings[16] by increasing the index of refraction (density) and the environmental stability. Recently ion implantation accelerators have been used to give high energy (50 to 100 keV) particle bombardment of surfaces to give recoil implantation of previously deposited material or “mixing” of the interfacial region between a film and the substrate. [30][31] This high energy bombardment may be concurrent with the atomistic film deposition or may be used as a post-deposition treatment. The development of ion sources for fusion reactor applications has lead to large area, high current, ion sources that allow higher fluxes and larger areas of bombarding particles. [32][33] Also during the early and mid 80’s, equipment and techniques were further

Ion Plating

349

developed that allowed ions of the film material to be used as the bombarding species. [34][35] Some of these sources were derived from ion sources developed for isotope separation.[36] The basic rules for ion plating are: 1. Bombardment must be over the whole surface to be covered—the more uniform the bombardment (species, number [ratio of bombarding species to depositing species] and energy) the better. 2. A minimum bombardment ratio and particle energy must be established in order to attain the desired property and structural modification of the surfaces and deposited film. 3. Bombardment must be continued from surface conditioning through interface formation for good adhesion. 4. Bombardment may be continued through the film formation stage in order to modify film composition, morphology, microstructure, and properties. 5. Contamination in the plasma should be low in order to reduce reaction with the substrate and with the depositing material (reactive species are activated in the plasma—good for reactive deposition, bad for contamination). 6. More material must be deposited than is removed by the bombardment process; however, all of the deposited material may react with the substrate giving a type of surface modification coating (ex., Pt on Si giving a PtSi coating). Ion plating may be divided into several stages, namely: (a) surface preparation, (b) nucleation and interface formation, and (c) film growth. It is important that the bombardment be continuous through the various stages for the ion plating process to work properly. The surface preparation stage allows in situ sputter cleaning of the surface prior to the beginning of deposition. This “cleaning” portion of the process allows good interfacial reactions for adhesion[37] and the generation of ohmic contacts to semiconductor materials.[37] In addition to cleaning, surface preparation may also be in the form of roughening the surface morphology or changing the surface composition (surface reaction or preferential sputtering). Bombardment may also make the surface more “active” by the generation of reactive sites and defects. For example: an unbombarded silicon surface metallized with aluminum shows no interdiffusion but a bombarded surface gives rapid diffusion.[39]

350


The interface formation stage allows the formation of a desirable diffusion or compound type interface on the “clean” surface if the materials are mutually soluble, or the formation of a “pseudo-diffusion” type of interface due to the energetic particle bombardment during the initial deposition if the materials are insoluble.[37] Interface formation is aided by defect formation and the deposition of energy (heat) directly into the surface without the necessity for bulk heating. In some cases the temperature of the bulk of the material may be kept very low (ex., liquid nitrogen cooling) while the surface region is heated by the bombardment.[4] This allows the development of a very high temperature gradient in the surface region which limits diffusion into the surface. The ion bombardment along with a high surface temperature may cause all of the depositing material to be diffused into the surface giving an alloy or compound coating. In ion plating the bombardment may or may not be continued during the film formation stage. If the bombardment is not maintained, the process, through the interface formation stage, may be considered as astrike for further deposition by another technique (vacuum deposition, sputter deposition, electroplating, etc.). If the bombardment is maintained during the growth stage, it is usually with lower energy bombarding particles and a higher ion flux than is used in the surface preparation and interface formation stages. This is to reduce the gas incorporation and compressive stress in the resulting deposited material. Generally energetic particles for bombarding surfaces and growing films are of gaseous ions and arise from:(a) biasing (DC or RF) a surface in contact with a plasma so that it is bombarded by ions from the plasma, (b) extraction of ions from a confined plasma and accelerating them to a high energy through a grid system into a vacuum environment (ion beam),[32][40] or (c) reflected high energy neutrals which arise from ion bombarding a surface in a low pressure environment[41]-[43] such that the reflected neutrals are not thermalized by collisions in the gas phase. The energetic particles (ions) may also be of a condensible film species and arise from:(a) sources such as are used for isotope separation,[36][44]-[46] (b) acceleration of negative ions from a negatively biased compound or alloy sputtering target,[47] (c) ions from vacuum or plasma arcs,[35][48] or (d) special ion sources.[49] The most general source of energetic particles is the extraction of ions from a plasma to bombard a surface which is at a negative potential with respect to the plasma. The bombarded surface may be located in the plasma generation region or at a downstream location in the plasma.

Ion Plating 2.0

351

PROCESSING PLASMA

Plasmas are gaseous media which contain enough ions and electrons to be electrically conductive (e.g., fifty). Energy is introduced into the plasma by the acceleration of electrons in a DC, RF, or microwave electric field. These energetic electrons then fragment, excite, and ionize atoms and molecules by collisions. A processing plasma is a plasma that is used in materials processing.[51]-[53] In many, if not most, cases of film deposition the processing plasma is a weakly ionized plasma such that there are many more neutral particles than ions in the gas phase and there is a large number of radical species compared to ions when a molecular gas is used. In a processing system, the local plasma densities and plasma properties may vary significantly due to electrode configurations, presence of fixturing, and other geometrical factors. Typical properties of a weakly ionized plasma are: Ratio of neutrals to ions 107 - 104:1 (100 times as many radicals as ions when using a molecular gas)

3.0

Gas pressure

10-3 - 10-1 torr

Electron temperature

1 - 10 eV

Ion temperature

0.025 - 0.035 eV

GENERATION OF PLASMAS

In plasmas used for plasma processing, the electron energy is increased by acceleration in electric field gradients. The most typical configurations for generation of plasmas are: (a) DC diode discharge, (b) RF (radio frequency) discharge,(c) electron emitter sustained discharge,(d) magnetron enhanced discharge, (e) microwave discharge, (f) vacuum arcs, and (g) plasma arcs. Figure 6.2 shows a schematic of some of these configurations. 3.1

DC Diode Discharge

The DC diode configuration consists of an anode and a cathode immersed in a low pressure gas. In ion plating, the substrate may be the cathode of the DC diode discharge. At the cathode, the negative potential (-) attracts positive ions from the edge of the plasma region and they are

352


Figure 6.2. Plasma generation configurations: (a) DC diode; (b) DC diode with permanent magnets giving a planar magnetron; (c) RF plasmas with planar electrodes immersed in the plasma, electrodes external to a dielectric wall and a coil immersed in the plasma; (d) electron emitter (thermoelectron) with magnetic confinement; and (e) microwave cavity.

accelerated across the cathode fall region to impinge on the cathode (target). The cathode fall region, which surrounds the cathode, is where most of the potential drop in a DC discharge is to be found.[54] The plasma region is located between the edge of the cathode fall region and the anode where there is little potential drop. In the DC discharge, energetic particles (ions and neutrals) impinging on the cathode (target) cause the ejection of secondary electrons which then accelerate across the cathode fall region, collide with gas species, and create ions which sustain the discharge process. The secondary electron emission coefficient of a surface depends on the chemical nature and morphology of the surface. Oxides typically have a higher electron emission coefficient than metals, and rough surfaces have a lower secondary emission coefficient than smooth surfaces. The secondary electrons can be accelerated to high energies and impinge on the

Ion Plating

353

anode or other surfaces in the system. This can give rise to extensive heating of surfaces (substrates) in the system. The DC discharge requires a relatively high gas pressure to sustain the discharge (>10 microns argon). In the cathode fall region, some of the ions may be neutralized by charge exchange processes which give rise to energetic neutral particles[55][56] which are not affected by the applied electric field. The cross-section for charge exchange is much larger than that for physical collision. The charge exchange process is dependent on the gas pressure, and at low pressures the accelerated ions will arrive at the cathode with the full cathode fall energy since the collision probability is low. Because of these charge exchange processes there is, at the cathode, a flux of energetic particles consisting of ions and neutrals with a broad spectrum of energies. In order to sustain a DC discharge, the secondary electrons must create enough ions to compensate for losses by recombination. If the anode or ground surface is brought too close to the cathode the discharge is extinguished. This effect can be used to confine the DC discharge to areas of the cathode surface where bombardment is desired—other areas may have the bombardment prevented by having a ground shield in closed proximity to the surface. The Paschen curve gives the relationship between breakdown voltage and the minimum anode-cathode separation in a gaseous environment. Typically argon gas pressure of about 10 microns is used to sustain the DC diode discharge. At this pressure the width of the cathode dark space is about 1 cm. Insulator surfaces cannot be used as cathodes in a DC diode configuration since charge buildup on the surface prevents ion bombardment. Electrically insulating films deposited in a DC discharge will build-up a surface charge which will cause arcing through the film. In addition to causing the ejection of secondary electrons, high energy ions and neutrals which impinge on the target (or other surfaces) cause the physical ejection of surface atoms (physical sputtering) by momentum transfer processes. The sputtered particles leave the surface at higher than thermal energies but may be rapidly thermalized by collisions in the gas phase.[57][58] The sputtered particles may be scattered back to the target surface—this effect is more prominent the higher the gas pressure. Some of the energetic ions that bombard the cathode may be reflected as high energy neutrals.[40]

354


The electrical current measured in the DC cathode circuit is the sum of the charge due to the ion flux to the target and the secondary electron flux away from the surface. Therefore the cathode current density and cathode voltage do not specify the flux nor the energy of the impinging ions. However these parameters (along with gas pressure and gas flow) are typically used to specify the plasma parameters in DC diode plasma processing. Typically a DC diode discharge plasma is weakly ionized with many more neutral particles than ions. Any surface in contact with the plasma will be subjected to a flux of ions, neutrals, and electrons. A sheath potential will be developed because of the greater mobility and energy of the electrons as compared to the ions. This wall potential (typically 3 - 10 volts) will accelerate ions from the plasmas giving rise toion scrubbing of the surface. In ion plating, the surface in contact with the plasma may be biased to accelerate ions from the plasma and bombard the surface at higher energies. In plasma processing, the DC diode configuration has many advantages including: (a) a rather uniform plasma can be generated over large cathodic areas;(b) power input (watts/cm2) can be very high;(c) the power supplies are rather simple, inexpensive and powerful; (d) process reproducibility can be attained by controlling the geometry, gas pressure, and target power (current and voltage); and (e) the sputter erosion of cathodic surfaces may be used as a long-lived, stable source of depositing material. Disadvantages include:(a) surface geometries can result in focusing effects giving non-uniform bombardment;(b) heating of substrates by secondary electrons accelerated away from the cathode can be extensive; and (c) inability to bombard electrically insulating surfaces or films. Typical conditions for ion plating in a DC plasma environment are: Bombarding current: (typically ions, inert, reactive)

0.5 mA/cm2

Applied accelerating potential: (DC diode)

50 - 5000 volts

Ratio of bombarding particles to depositing atoms

1:10 - 1:100

Typical plasma/system parameters that are controlled or monitored in DC diode ion plating where the substrate is the cathode of the discharge are:

Ion Plating

355

System geometry Substrate:

Power input (voltage and current) Plasma uniformity Applied potential Pre-sputtering time Deposition rate/uniformity

Plasma:

Gas pressure Gas purity

Deposition source geometry 3.2

RF Discharge

At high frequencies, in a capacitively coupled discharge, the electrons oscillate in the changing field thus gaining energy, and by collision with atoms create ions, radicals and more electrons. Typical RF power supplies operate at 13.56 MHz (U.S. industrial frequency) with peak-to-peak voltages of greater than 1000 volts. The plasma acts as a low density electrical conductor and the RF field penetrates quite some distance into the plasma. When the driven RF electrode is a conductor, the surface is bombarded by ions from the plasma during the half-cycle that the electrode is negative. If the surface of the RF electrode is an insulator (backed by a conductor), the metal-insulator-plasma acts as a capacitor and the surface potential that appears on the insulator surface alternates between a low positive potential (because the electrons have a high mobility) and a higher negative potential (because the ions have a relatively low mobility).[59] Ions are extracted from the RF plasma during the negative portion of the cycle and bombard the insulator surface. During a portion of this half-cycle the bombardment energy may be sufficient to cause physical sputtering of the insulator surface. The RF potentials in the plasma can be determined using capacitive probes.[60] The ion energies bombarding a surface may be determined using a sampling orifice, a retarding grid, and a mass spectrometer.[61][62] In capacitively coupled RF discharges, the plasma potential, and hence the sheath potential at the electrodes, can have a time-varying value of tens to hundreds of volts.[63] When the electrodes have different effective areas, the plasma potential can also have a large DC potential with respect to one or more of the electrodes.[64] These factors affect the distribution of ion energies incident on the electrode surfaces in an RF discharge.[65] Small

356


area electrodes will attain higher voltages than large area electrodes, and the electrode potentials can be varied using an external capacitance in the circuit. The amount of energy that is coupled into the RF discharge depends on the impedance matching (reflected power) and coupling losses to other surfaces (stray losses). In RF plasma processing, it may be important to determine just how much energy is actually being coupled into the plasma.[66]-[69] RF power may be coupled to the plasma using metal electrodes external to a dielectric wall [70] or the RF plasmas may be excited using immersed electrodes (see Ref. 71, for example). Very high plasma densities and ionization efficiencies can be attained in RF driven plasmas and the discharge may be established at lower pressures than the DC diode discharge.[72] In plasma processing, the RF discharge has the advantage that insulating surfaces or insulating films on conductive surfaces can be bombarded by applying an RF potential. Disadvantages are: (a) high power inputs (heating) to insulating materials cause cracking, (b) electrode geometries can cause problems with coupling to the RF power, (c) there are many sources of RF power loss in systems, (d) plasma uniformity is difficult to obtain over complex surfaces, and (e) the bias conditions on surfaces in the RF plasma are variable and often difficult to control. 3.3

Microwave Discharges

Plasmas can also be excited at much higher frequencies—300 MHz to 10 GHz—where electron cyclotron resonance coupling gives more efficient ionization.[73] Microwave discharges differ in many essential respects from DC and RF discharges. Namely: (a) there is an increased amount of excitation in the discharge and a lot of vacuum-UV is produced,(b) high degree of ionization (as high as 20%), (c) the electron densities are higher, 1013 vs. 109 - 1010, and (d) the particle temperatures are higher (factor of ten or more). Microwave plasmas are most often used in the downstream processing configuration since substrates in the microwave cavity can detune the system. 3.4

Electron Emitter Discharge

In the DC diode and RF plasma configurations, the electrons necessary to sustain the plasma are produced in the plasma. When using electron

Ion Plating

357

emitters, the electron source is independent of the plasma processes. Common electron emitters are hot thermoelectron emitting cathodic surfaces and hollow cathodes.[74] For example: hot LaB 6 surfaces can give an electron emission of >20 A/cm2 at 1700°C.[75] These discharges may have very high electron densities (1012). Often the electrons are confined by a magnetic field (100 - 500 gauss) directed along the anode-cathode axis. The magnetic field increases the electron path length in its movement from the cathode to the anode by causing the electron to spiral in the magnetic field. This increases the ionization efficiency of the electron and allows the discharge to be sustained at a low gas pressure. The ions in the plasma may be extracted using an electrode at a DC or RF potential to give bombardment of a surface (triode configuration). The triode configuration suffers from a nonuniform plasma density along its axis, thus giving nonuniform bombardment and a density variation in activated species over a large biased surface. The thermoelectron emitter system is very amenable to forming dense plasmas and for application to downstream processing. By applying magnetic fields, the plasma (ions and electrons) may be confined and steered into a processing chamber.[33] Steering (bending) of the plasma beam occurs since the electrons follow the magnetic field lines and the ions follow the electrons. The electron emitter configuration may also be used to melt and vaporize material for film deposition and at the same time as producing the plasma.[76]-[81] In plasma processing some of the advantages of the electron emitter configurations are:(a) the flux of electrons is independent of other plasma and electrode processes; (b) very high plasma densities can be attained; (c) the plasma properties can be controlled by controlling the electron emission; (d) the plasma may be steered from the plasma generation chamber; and(e) the electron beam can be used as a source for thermally vaporizing material. Disadvantages are:(a) need for well controlled and long life electron emitting sources, and (b) plasma non-uniformity over large areas and complex surfaces. 3.5

Magnetron Discharges

Low strength (100 - 500 gauss) magnetic fields may be used to confine the electrons and increase their path length in plasma systems by causing the electrons to spiral around the magnetic lines of flux (magnetron configurations). There are a number of ways to establish magnetic fields in plasma chambers including: (a) internal permanent magnets, (b) external permanent magnets, (c) external electromagnets, and (d) moving magnets.

358


Permanent magnets have the advantage that they may be placed in such a manner as to position the field lines in a desirable manner. However, getting a uniform magnetic field over a large or complex surface is difficult with any magnetic field configuration. Magnetron enhanced plasma configurations have many advantages including: (a) confining the plasma to a small region, (b) increased ionization and plasma density, and (c) may be operated at low pressures where gas phase collisions are reduced. Disadvantages include: (a) non-uniform magnetic fields give non-uniform plasma densities,(b) isolation of the plasma to a small region of the processing chamber requiring auxiliary plasma sources near the substrates in some applications, and(c) low pressure processing can give rise to a flux of high energy reflected neutrals which may affect bombarded surfaces and growing films, and affect their properties in an undesirable manner.[43] 3.6

Plasma Enhancement

Plasma enhancement techniques may also be used to locally increase the plasma density. This plasma enhancement may be done by using local RF fields,[82] thermoelectron emitting surfaces,[33] hollow cathode electron emitters,[74][81][83][84] deflection of secondary electrons in e-beam evaporation, localized higher gas pressure, etc. The plasma density may also be increased by the use of magnetic fields which cause the electrons to spiral around the magnetic field lines thus increasing their path length (magnetron configurations).[85] Some of the most dense plasma sources have been developed for the magnetic fusion community. Many of these sources use RF power input or thermoelectron emitting surfaces along with confining magnetic fields. In some film deposition processes, ions of the film material (condensible or non-condensible) may be used to bombard the substrate. Ions of reactive gaseous species may be formed in plasmas by conventional techniques. High concentrations of ionized condensible film species may be formed: (a) in plasmas having a high density of low energy (100 eV) electrons,(b) in vacuum arc-plasmas on solid cathodes,[86][87] or (c) above molten anodes in vacuum arc-plasmas.[88][89] In many cases, these species may be multiply ionized. The addition of a reactive gas to the plasma allows the deposition of compound materials.[90] When using plasmas and bombardment effects in ion plating, many processing variables are unknown. Processing unknowns include: (a) the

Ion Plating

359

portion of the substrate current that is due to secondary electron emission,(b) the flux and energy spectrum of the ions and electrons, and (c) the flux, adsorption, and surface coverage of the neutral gaseous species. Generally no attempt is made to determine these process variables during the processing but rather they are controlled by controlling other processing variables such as: (a) system geometry, (b) vaporization rate, (c) gas pressure, (d) gas composition, (e) gas flow rate(s), (f) substrate and system temperatures, (g) contaminants in the plasma, and (h) substrate power input per unit area (voltage and current).

4.0

PLASMA CHEMISTRY

The plasma is a very energetic chemical environment and many chemical processes can occur (e.g., Refs. 91-93). The principal chemical processes are: (a) electron impact ionization,(b) dissociation (fragmentation) of molecules (formation of radicals), (c) Penning ionization (metastable collision), (d) dissociative electron attachment (e) electron attachment, (f) excitation, (g) momentum transfer collisions, (h) de-excitation of excited species, and (i) recombination (neutralization). As an example of the complexity of plasma chemical processes consider that there can be twentyfour reactions and sixteen species formed by the decomposition and reaction of CF4 in a plasma.[94] As an example of Penning ionization, consider argon which has metastable excited states of 11.55 and 11.75 eV, and copper which has an ionization energy of 7.86 eV. Thus a copper atom colliding with a metastable argon atom is easily excited or ionized. Metastable atoms may be very effective in ionizing and exciting other species by collision. Many of these plasma processes are characterized by cross-sections for collision processes and threshold energies for attachment processes. For example CF3 Cl has a high collision cross-section and low threshold energy (2 - 3 eV) for electron dissociative attachment. CF4 has a low cross-section and high threshold energy (5 - 6 eV) for electron dissociative attachment, and CCl4 is not activated at all by electron attachment. Therefore CF3 Cl is much more easily fragmented and ionized in a plasma than is CF4 or CCl4. The degree of ionization, dissociation and excitation of the species depends strongly on the gaseous species, electron energy, and density in the plasma. Generally there is much more dissociation than there is ionization of molecular species.

360


Many of these plasma processes serve to activate(plasma activation) the gas species, i.e., to make them more chemically active by dissociation, ionization, or excitation. Plasma activation may partially decompose chemical precursor species and make them easier to thermally decompose. This type of activation is used in plasma-enhanced and plasma-induced processes such as plasma-enhanced chemical vapor deposition (PECVD).[95] Plasma discharges are very effective in desorbing contaminates (ex. H2 O) from surfaces in a plasma processing chamber. These impurities are activated in the plasma and may contaminate the depositing material. A number of techniques may be used to determine plasma properties.[96] Optical emission is the most common.[97][98] Optical absorption techniques may also be used to characterize the gaseous species and temperature in a gas discharge.[99] Electron and ion densities in a plasma may be measured by the use of small-area Langmuir probes.[100] In film processing utilizing plasmas, the depositing (condensible) species usually traverses the plasma before condensing on the substrate. In doing so, some of the species may be fragmented and/or ionized in the plasma. However in the usual ion plating configuration (low density, weakly ionized plasma) little ionization of the condensible species is to be expected.[101]

5.0

BOMBARDMENT EFFECTS ON SURFACES

The physical effects of energetic particle bombardment on surfaces and depositing films is very dependent on the mass, flux, and energy of the bombarding particles, the flux of non-energetic particles (i.e., depositing or absorbing species) and the atomic mass and chemical nature of the bombarded surface. In many cases the fluxes of impinging particles are not determined or controlled except by the processing parameters. Bombardment can be from ions accelerated to the surface under:(a) an impressed bias,(b) an induced bias, or (c) due to the development of a sheath potential. Bombardment can also be from energetic neutrals formed by charge exchange processes or by the neutralization and reflection of energetic ions from a surface. Many studies have shown that intentional and/or unintentional particle bombardment may affect the various stages of film growth (substrate preparation, nucleation, interface formation, and film growth). The parameters (flux, energy, ratio) which are important are usually poorly defined, and

Ion Plating

361

in many instances, the importance of bombardment is not recognized. An example is in magnetron sputtering where the stress in the deposited films may be correlated to the gas pressure in the sputtering chamber and thus to the bombardment by energetic particles reflected from the sputtering target.[102] At low pressures, high energy particles, which are reflected as neutrals from the sputtering target, provide the particle bombardment of the growing film, giving high compressive stresses. At higher gas pressures,the high energy particles are thermalized by collisions before they can bombard the growing film and tensile stress is developed in the growing film.[43] High energy reflected particles are more prevalent at high angles to the surface normal so substrates in these areas will be more affected by bombardment than substrates positioned normal to the surface. Also the post cathode or rod cathode target configurations will be more sensitive to bombardment than planar targets. The nature of the stress can also be correlated to the angle-of-incidence of the deposited material[103][104] and the sputtering current density.[105] It has also been shown that anisotropic bombardment gives rise to anisotropic stresses in films.[43] Particle bombardment allows one means of in situ preparation of a substrate surface prior to film deposition. In situ surface preparation may be necessary to generate the high quality interface necessary for the fabrication of some semiconductor devices.[38][39] Figure 6.3 shows several regions affected by particle bombardment and the regions are defined as follows: Surface: Interface between solid and gas (vapor or vacuum). Surface-region: Region of physical penetration by the bombarding particles in which there is a collision cascade. Near-surface region: Region beyond physical penetration but which is affected by the bombardment (heating, diffusion) Bulk region: Region of the material which is not significantly affected by the bombardment (or can be made so by cooling) Figure 6.3 also depicts the effects of bombardment by energetic species (not electrons) on the surface and the subsurface region. Surface effects include:(a) desorption of weakly bonded surface species,(b) ejection of secondary electrons, (c) reflection of the energetic species as high energy neutrals,(d) sputter ejection (physical sputtering) of surface atoms by momentum transfer through collision cascades, (e) sputtering and redeposition of sputtered species by collisions in the gas phase, ionization, and acceleration back to the surface and byforward sputter deposition due

362 Deposition Technologies for Films and Coatings Figure 6.3. Schematic depiction of energetic particle bombardment effects on surfaces and growing films. See text for discussion.

Ion Plating

363

to the ejection angle on a rough surface, (f) enhanced surface mobilities of atoms on the surface, and (g) enhanced chemical reaction of impinging and adsorbed species to produce condensed species (reactive deposition) or volatile species(etching). In the subsurface region:(a) the impinging particles may be physically implanted,(b) the collision cascades cause displacement of lattice atoms and the creation of lattice defects, (c) defects coalesce, (d) surface species may be recoil-implanted into the subsurface lattice,(e) mobile species may be trapped at lattice defects, and (f) much of the particle kinetic energy is converted into heat.[106] Lattice channeling processes can carry these effects deeply into the surface. Film growth may be considered to be layered growth where each layer is a surface (surface-region) which is covered by another surface layer. Thus particle bombardment effects on surfaces and on growing films are closely related. 5.1

Collisional Effects

Particles striking other particles transfer momentum (billiard ball effect). Particles striking surfaces also transfer momentum to the surface atoms which results in lattice atom displacement and vibration (heating). In many cases there will be some penetration of the bombarding particle into the surface lattice structure. The amount of this penetration will depend on the relative masses of the bombarding and “target” atoms, and the crystallographic orientation (penetration will be greatest along open planes). Particles with energies too low to give collisional lattice displacement (<20 eV, depending on masses) may enhance chemical reactivity on the surface thereby influencing: reactive deposition processes, reactive plasma etching, ion enhanced chemical etching, reactive plasma cleaning, chemical sputtering (volatile species), plasma polymerization, etc. These low energy particles may also enhance the removal of weakly bonded surface atoms (desorption). For instance, low energy bombardment by species accelerated across the plasma sheath are used inion scrubbing to clean optical surfaces before film deposition, and hydrogen ion scrubbing is used to remove surface contaminants in Tokamak-type fusion reactors. Chemical sputtering occurs when the bombarding species (H, Cl, Fl) chemically reacts with the surface to form a volatile species which leaves the surface with thermal energies.[107] Chemical sputtering is often a synergistic effect when both a chemically reactive species and an energetic species bombard the surface simultaneously (bombardment enhanced chemical etching).[108][109]

364


For particle energies large enough to create collisional damage, many things can occur. Figure 6.3 depicts the various processes that can occur when these particles strike a surface and generate a collision cascade in the near-surface region. These processes include: 1. Reflection from the surface with some loss of energy— these particles may be used for surface analysis (ISS, Ion Scattering Spectrometry) or give bombardment-induced changes in the deposited material. 2. Physical sputtering of the surface atoms by momentum transfer processes. 3. Subsurface implantation of the bombarding species. 4. “Knock-on” implantation of surface species (recoil implantation). 5. Point defect formation along the collision cascade track. 6. Localized high temperatures along the collision cascade track, the heating from which diffuses through the surface region. 7. Produce secondary electron from the surface which may be accelerated away from a surface that is at a negative potential. Surface effects: Particle bombardment processes/effects may be classed as: Prompt processes (<10-12 sec) - collision effects Cooling effects (>10-12 to <10-10 sec) - thermal spikes Delayed effects (>10-10 sec to < 1 hour) diffusion, segregation Persistent effects (> 1 hour) - compressive stresses, amorphorization, phase change Physical sputtering (often just called sputtering)[110] is the physical ejection of a surface atom by momentum transfer in the collision cascade where it intersects the surface (sputter erosion). This process is not a thermal process so the ejected particles have energies greater than thermal and a distribution (ejection pattern) that depends on the crystallographic orientation of the surface atoms. Only a small amount of the energy of the bombarding particles appears in the energy of the sputtered particles; the rest (70% or more) goes into heating the bombarded surface. The sputtering yield, the ratio of ejected atoms to incident particles, is a function of incident particle energies (normal incidence), and is a

Ion Plating

365

function of the mass of the incident particle and the mass of the target atoms. The sputtering process only begins after a threshold energy is reached where there is enough momentum transfer to give ejection (> 20 eV). The sputtering yield is a function of crystallographic orientation of the surface material, giving differing yields for differing planes and allowing the delineation of crystallographic structure of a surface by sputter etching. The sputtering yield is also a function of the incident particle energy, increasing from some threshold value to some maximum above which most of the energy is deposited too far below the surface to affect the collision cascade where it intersects the surface and the sputtering yield decreases. There have been some reports of an equilibrium time for the sputtering yield to stabilize, possibly due to saturation of the surface with bombarding gas. For surfaces bombarded at normal incidence, the ejected particles will come off with a cosine distribution at low bombarding energies,[111] and an over-cosine distribution as the energies become higher,[112] and with some dependence on mass when sputtering alloys or isotopic mixtures.[113] This angular dependence may change with texturing of the sputtered surface. If the bombarding flux is off-normal, the ejected flux will be skewed in a forward direction. (The ejection from a single crystal surface will depend on the orientation of the crystal planes). The energy of the ejected particles will depend on the bombarding angle with oblique bombardment giving higher energy ejected particles. The sputtering yield is also a function of the angle of incidence of the impinging particle. For this off-normal bombardment, the sputtering yield, as a function of incident particle bombardment angle, initially increases to a maximum then decreases rapidly above some angle as the bombarding particles are reflected from the surface. The maximum generally occurs at about 70 degrees off-normal but this varies with the relative masses of the particles. This property may be used to give forward sputtering of materials where the impingement angle on the target (inclined plane with ion beams, edges of cylinders) is high. This angular dependence of sputtering also results in the formation of cones on a sputtered surface. The taper of the cones is determined by the angle-of-incidence dependence of the sputtering yield. For a single component surface, the material will be removed from the surface with the bulk composition. If the surface is an alloy or compound, preferential sputtering may occur with some surface species being removed more easily than others (may actually extend several monolayers into the

366


surface—altered region). This will give rise to a surface composition that is different from the bulk (enriched, depleted);[114]however, if no diffusion occurs, the sputtered species will have the same composition as that of the bulk material at equilibrium. Preferential sputtering of compounds decreases with the off-normal incidence angle of the bombarding ions.[115] During deposition with concurrent bombardment, preferential sputtering can cause loosely bound species to be removed (contaminates, or nonreacted compound-forming species),[6] or in the case of alloys, one of the constituents may be preferentially sputtered.[116] If diffusion in the target does occur, the sputtered species will have a composition that differs from the bulk and may be continuously variable. Alloy sputtering can give some mass separation with the flux normal to the surface being enriched in the lighter element and the flux off- normal being enriched in the heavier element.[113] The sputtering process from a negatively biased elemental surface gives neutral species. Sputtering from a grounded surface gives varying amounts of ionized species which may be used for SIMS (secondary ion mass spectroscopy) analysis or, if ionized in a plasma, may be used for GDMS (Glow Discharge Mass Spectroscopy) or endpoint optical analysis of the plasma. Sputtering of a target containing several species may give negative ions of the species having the lesser electronegativity (ex., O-, Au- from AuCu alloy). At relatively high gas pressures (DC diode sputtering conditions) a portion of the sputtered species may be scattered back to the surface. A portion of the sputtered species may be ionized in the plasma and accelerated back to the target surface giving self-sputtering. On non-planar surfaces, some of the sputtered species may be forward-sputtered so that they are deposited on the target surface. All of these affect the apparent sputtering yield of surfaces. Other sputtered species may condense on substrate surfaces giving “sputter deposition” often just called sputtered films (poor terminology). Controlled sputter erosion in conjunction with surface analysis techniques is used to depth-profile from the surface in order to allow in-depth determination of elemental composition.[117] Particles on the surface or inclusions in the surface region may protect the local area from sputter erosion, giving a texturing of the surface morphology (cone formation). [118] Enhanced surface mobility may also create whiskers on the bombarded surface.[119] Texturing of surfaces may

Ion Plating

367

give a long term change in the sputtering yield (2 - 3X) from the surface since the surface morphology changes (changing angle of incidence and redeposition processes).[120] Texturing is used to generate optical “trapping” surfaces,[121] to treat surfaces for medical implants to improve bone growth[122] and adhesion, and to reduce secondary electron emission from surfaces. If the surface species is a “foreign” atom the process of removal may be termeddesorption and may be calledion-induced desorption or cleaning.[123] It has been shown that the process of recoil implantation is an important parameter in the desorption (sputter cleaning) of monolayers of chemisorbed species from a surface. Surfaces may be etched byreactive plasma etching where a chemically active species (Cl, F) is formed in a plasma and reacts with the surface to form a volatile species. Typically a volatile chemical species is formed on the surface and volatilization removes the surface species. The process is somewhat difficult to control since the plasma composition and the flux of impinging particles from the plasma may vary. If the particles are accelerated to the surface, chemical reaction is enhanced and the process becomes reactive ion etching or reactive plasma cleaning.[124] The desorption of weakly bound surface species is important to plasma cleaning and may be used to reduce the incorporated contaminants in deposited films.[6][125] The desorption may also be useful in desorbing unreacted species in reactive deposition processes giving rise to more stoichiometric and chemically stable deposits. In ion-assisted chemical etching,[126] a molecular beam of the chemically reactive species and a beam of inert gas ions simultaneously bombard the surface to be etched. Using this technique, very high etching resolution can be obtained at high etch rates.[127] The bombardment-assisted chemical processes that occur on surfaces are very poorly understood.[109][128]-[130] On one hand the increased chemical activity may be due to the increased “temperature” due to momentum transfer from the bombarding ions. On the other hand, secondary electrons from the bombardment of the surface may play an important role. In some cases, surface changes due to the bombardment may provide sites for chemical reactions. The etching process is sensitive to the amount of adsorbed reactive gases.

368 5.2

Deposition Technologies for Films and Coatings Surface Region Effects

The surface region is affected by the penetration of the bombarding species. A major portion of the energy goes into atomic vibration and appears as heat. The resulting collision cascades generate point defects and collisional mixing by atomic displacement. The point defect densities may be as high as 1 - 20 atomic percent and may combine to form large-scale defects and, in the extreme, disrupt the crystalline material into an amorphous form. The defects introduced into the surface region can affect the electronic properties of the region.[37] For silicon, these electronic effects have been studied by bombarding the surface, then fabricating Schottky barriers.[131] The lattice defects may also allow the trapping of normally mobile species.[132] Trapping studies of ion-bombarded surfaces show trapping site densities of 1 - 20 atomic percent.[133][134] The bombarding species may be implanted to quite high atomic concentrations if they are not lost by diffusion to the surface. Typically 1 - 10% of entrapped bombarding species is found in the surface region of argon-ion-bombarded sputtering targets. In the argon bombardment of silicon, it has been shown that the argon content rises very sharply from about 225 eV bombarding ion energies and approaches a plateau around 2200 eV. Films of gold grown under bombardment conditions have been deposited with 40 atomic percent helium[135][136] which is normally insoluble in gold. Other gases have been incorporated into more complex sputter-deposited amorphous materials to high concentrations.[137] Incorporated gases may embrittle materials, precipitate into bubbles, or outgas at elevated temperatures. The incorporation of radioactive krypton into surfaces (by pressure or ion bombardment) for subsequent thermal desorption and wear studies has been called kryptonation.[138]-[142] The materials thus formed have been called kryptonates. In the case of hydrogen ion bombardment of carbides, carbon depletion is noted to depths far greater than the penetration depth of the hydrogen ions.[143] Hydrogen ion bombardment also acts to hydrogen charge the surface region giving a chemical potential which enhances the hydrogen diffusion into the material. The same is true for nitrogen where nitrogen ion bombardment, diffusion, and reaction results in nitriding of the surface region. Outgassing of the incorporated gases may present adhesion problems when a film has been deposited on the bombarded surface.

Ion Plating

369

The combination of high defect densities and heat allow rapid diffusion (similar to radiation enhanced diffusion[144]) and the generation of analtered layer whose composition may differ from the bulk (alloys and compounds). The implantation of the bombarding species and the recoil implantation of the surface species results in the compaction(peening) of the near-surface region and the formation of a compressive stress in that region. The implantation process also incorporate bombarding (and surface species recoil implantation) species into the near surface region often in amounts that are above the normal solubility limits. Both the peening action and the incorporation can affect the stress in this region film. 5.3

Near Surface Region Effects

The near-surface region is affected by the proximity of the surface region. Compositional changes in the surface-region generate a chemical potential which, along with the temperature and defect profile, may allow rapid diffusion from this region to the surface and vice versa. Compressive stresses generated in the surface-region are offset by tensile stresses in the near-surface region and these may cause subsurface fracturing. In hydrogen ion bombarded silicon, subsurface defects and fracturing are attributed to the tensile stresses generated beneath the compressive surface region giving fracturing. Temperature-rise studies by the crystallization of metallic glass substrates during the ion plating of gold has shown that the temperature rise in the near-surface region (10 - 15 microns) approached 500°C.[145] 5.4

Bulk Effects

The principle bulk effect is that of heating. Particularly if the substrate is not actively cooled then the bulk temperature can rise significantly.

6.0

SOURCES OF DEPOSITING ATOMS

In ion plating, the depositing (condensible) species may have thermal energies or may be all or part of the energetic bombarding flux. The source of the depositing species may be:

370

Deposition Technologies for Films and Coatings Thermal evaporation or sublimation sources Sputtering sources (sputter-ion plating) Vacuum arc sources Chemical vapor species (Chemical Ion Plating)

6.1

Thermal Vaporization

Thermal evaporation or sublimation of a material is performed by heating the material to the point that it has an appreciable vapor pressure. Heating may be done by: Resistively heating the material directly Material in contact with a resistively heated surface Bombardment with low energy non-focused electrons Bombardment with high energy focused electrons Radiant heating from a high intensity source Heating with a laser beam RF inductive heating Resistively heated sources are typically used to vaporize materials which have appreciable vapor pressures below about 1500°C. Low energy electrons may be produced by hollow cathodes[34] and deflected in a low magnetic field. The electron energies are typically around 100 volts which is also the energy for the maximum cross-section for ionization. Therefore as the vapor leaves the evaporating surface and passes through the impinging electron cloud, the possibility for ionization is high. Such a system has been used to ionize silver for deposition on beryllium. The silver ions are used first to sputter clean the surface and then, by lowering the acceleration voltage, a film of silver is allowed to form.[146] High energy electron beam evaporation is done by generating a small electron current and accelerating the electrons to very high energies, on the order of 10,000 eV. The electrons are deflected by a magnetic field and are moved over the surface in a variable electric field. This technique allows a very high energy density spot to be rastered over the surface to be evaporated thus allowing the evaporation of refractory materials. This type of evaporation source can be used in a vacuum environment, along with an ion gun, to supply inert gas ions for bombardment of the growing film to do ion plating. A great deal of interest has been shown in optical coatings

Ion Plating

371

formed this way since the resulting films have higher index of refraction (more dense) and are more environmentally stable than are vacuum deposited films.[16] The adhesion of optical films may also be improved by the concurrent bombardment.[147] By bombarding the growing film with reactive ions, compounds of materials such as TiN can be formed.[148] High energy electron beam evaporation may also be done in a plasma environment as long as the hot filament is not exposed to the plasma. This may be done by differentially pumping the chamber below the electron beam evaporator where the electron emitting filament is located.[1] 6.2

Sputtering

Sputtering of a surface may be done in a plasma by extraction of ions from the plasma under an impressed electric field and bombarding the target surface with energetic ions to cause physical sputtering of the surface. This source of depositing material is used in sputter ion plating.[149] Sputtering may also be done in a vacuum environment using ion beams formed in a separate plasma chamber and extracted into the vacuum chamber using a grid system. The sputtered material may then be deposited in a vacuum environment and an ion beam source may be used to bombard the deposit with an inert or reactive beam.[150] Figure 6.4 shows some of the configurations that may be used. 6.3

Vacuum Arcs

Vacuum arc vaporization sources can also provide film-ions in vacuum ion plating. Arc evaporation (vacuum arc) occurs when a high current, low voltage arc passes between electrodes in a vacuum. The arc is confined to spots which have a very high energy density (1011 W m-2). The vaporized material may be deposited as a film, often with a bias applied to the substrates. Arc vaporization is an old method of vaporizing carbon to form thin carbon films. By using a reactive gas atmosphere, compounds may be deposited. In arc evaporation, a large fraction (0.5) of the vaporized material is ionized.[151]-[153] The potential distribution in the arc gives a “hump” near the cathode allowing ions on the anode side of the hump to be accelerated to a high potential (80 eV).[48] Problems with this deposition technique include stabilization of the arc and the formation of globules of the ejected material. Arc steering over the cathode surface and plasma deflection have been applied to reducing theglobule problem. The ions

372


Figure 6.4. Configurations for using an ion beam in a vacuum system to allow concurrent bombardment of a depositing film.

formed in the vacuum arc may be accelerated to a biased substrate, and thus constitutes a source of bombardment and deposition in the ion plating process. The vacuum arc melting process[154] utilizes a vacuum arc to melt materials and the melting equipment may also be used as a source of ions (from the molten anode) for deposition.

Ion Plating 6.4

373

Chemical Vapor Precursors

In chemical ion plating, a precursor gas containing the material to be deposited is injected into the plasma where it is totally or partially decomposed, ionized, and deposited (accelerated) to the substrate surface. This technique is similar to plasma enhanced CVD with the addition of the acceleration of the ionized particles. In some cases, the chemical vapor is introduced into an ionizing source chamber then the disassociation products are extracted and accelerated into the deposition chamber( under vacuum) as a beam to impinge on the substrate surface. This process has been used to deposit i-carbon (hydrogenated carbon, i C-H)[155]-[157] and BN[158] films.

7.0

REACTIVE ION PLATING

In reactive ion plating, co-depositing species or surface species react to form a non-volatile (compared to reactive ion etching where a volatile species is formed) condensed species. The concurrent bombardment tends to activate the reaction process (activate the species in the plasma, ion enhanced reactions on the surface) making the reaction easier, and tends to resputter the non-reacted species giving a more stoichiometric deposit.[159] A large number of compounds have been deposited by reactive ion plating.

8.0

BOMBARDMENT EFFECTS ON FILM PROPERTIES

Many surface and film properties may be modified by particle bombardment.[160] These properties include: adhesion, composition, grain size, crystallographic orientation, growth morphology, surface morphology, impurity content, electrical properties, magnetic properties, optical properties, film stress, density, and mechanical properties. Particle bombardment may be an important process variable and, if not controlled, may give unacceptable product variability. 8.1

Effects: Adatom Nucleation

Particle bombardment may be used to sputter-clean surfaces in order to remove barrier layers and contaminates. This allows adatoms to interact directly with the substrate and, if there is chemical bonding, the nucleation density is increased and interfacial reactions will be promoted.

374


Ion bombardment during adatom deposition may increase the nucleation density in many instances.[161][162] This increase may be due to the formation of active sites due to the defect formation[163] and/or due to the recoil implantation of adatoms into the surface region where they act as nucleation sites. Conversely, ion bombardment may enhance surface mobilities. (Ex: Ion bombardment enhanced surface diffusion during cone formation.[164]) The presence of a plasma has been shown to influence the nucleation either due to bombardment or to electrical effects.[165][166] It has been shown that the nucleation density of gold deposited on oxide substrates by sputter deposition in an oxygen plasma is much higher than in an inert plasma. This allows the deposition of adherent gold films on substrates where normally the adhesion is poor.[167] The reason for this dependence on the oxygen plasma is not understood but may be due to chemical or electrical charging effects. 8.2

Effects: Interface Formation

The nature and type of interfacial region that is formed during film deposition is important to adhesion and the functionality of the film-substrate couple. Interfacial regions may be classed as: (a) mechanical, (b) abrupt, (c) compound,(d) diffusion,(e) pseudo-diffusion or combinations thereof.[36] The type of interfacial region formed during deposition depends on the film/ substrate materials, chemical interactions, energy available, nucleation behavior, contamination, and surface morphology. Particle bombardment prior to and during film deposition affects many of these factors. Of particular interest is the ability to clean a surface, influence nucleation, and to provide energy to the surface region to enhance diffusion and chemical reactions. 8.3

Effects: Film Growth

Film growth may be considered to be layered growth where each layer is a surface (surface-region) which is covered by another surface layer. Thus particle bombardment effects on growing films are much the same as those for the surface and surface-region effects. In addition there are sputtering/ redeposition effects on film morphology/microstructure and annealing effects on the film structure due to the heating during deposition and local thermal spikes during the bombardment.[168]

Ion Plating

375

Film microstructure, morphology, and properties that can be modified by ion bombardment during growth include: Stress Stoichiometry Microstructure Morphology (surface and bulk) Grain size Grain orientation Epitaxial growth Hardness Abrasion Resistance Optical properties (index of refraction) Density Pinhole density Adhesion By using a partially ionized and accelerated beam of depositing particles it has been shown that the “epitaxial temperature” can be lowered compared to the deposition of a non-ionized beam.[169] The important parameters in structure modification are: Substrate temperature during deposition—all sources Angle of incidence of depositing species Resputtering during deposition—indication of forward sputtering Redeposition (of sputtered species) during deposition In the DC diode configuration (where there is appreciable redeposition of sputtered material) studies have shown that an apparent resputtering rate of 0.2 to 0.3 is necessary to achieve appreciable modification of the columnar microstructure. In the magnetron configuration (where there is much less redeposition) apparent resputtering rates of 0.6 to 0.7 are necessary to give appreciable modification of the columnar microstructure. Concurrent energetic particle bombardment during atomistic film deposition may modify many film properties (e.g., Refs. 23 - 26, 170). The amount of modification will depend on both the mass, energy, and flux of the bombarding species and the mass and flux of depositing species. In the case of reactive deposition, the availability of activated species and the

376


effect of adsorbed surface species may also be important. The following are some of the film properties that can be modified by controlled concurrent bombardment during deposition. 8.4

Film Adhesion

The adhesion of a deposited film to a surface depends on the deformation and fracture modes associated with the failure.[171][172] Energetic particle bombardment prior to and during the initial stages of film formation may enhance adhesion by: removing contaminant layers, changing the surface chemistry, generating a microscopically rough surface, increasing the nucleation density by forming nucleation sites (defects, implanted and recoil implanted species), increasing the surface mobility of adatoms, and by creating lattice defects and introducing thermal energy directly into the surface region, promote reaction and diffusion. These effects will also improve surface coverage and thus decrease the number of interfacial voids which result in easy fracture and poor adhesion. Film adhesion may be degraded by the diffusion and precipitation of gaseous species at the interface. The adhesion may also be degraded by the residual film stress, due either to differences in the coefficient of thermal expansion of the film and substrate material in high temperature processing, or the residual film growth stresses developed in low temperature processing. 8.5

Film Morphology/Density

Physical sputtering and redeposition, increased nucleation density, and increased surface mobilities of adatoms on the surface under bombardment conditions may be important in disrupting the columnar microstructure that develops during low temperature atomistic deposition processes.[26][173][178] Figure 6.5 shows the fracture cross-section and surface morphology of RF sputter deposited chromium films at zero bias and a -500 volt bias during deposition. Note that the bombardment completely disrupted the columnar microstructure. Bombardment-related effects may also improve the surface coverage and decrease the pinhole porosity in a deposited film. This increased film density is reflected in film properties such as: better corrosion resistance, lower chemical etch rate, higher hardness, lower electrical resistivity (metals), and the increased index of refraction (optical coatings). However, it has been found that if the bombarding species is too energetic and the substrate temperature is low, high gas incorporation gives rise to

Ion Plating

377

voids (e.g., Ref. 179). Some investigators have used the parameter resputtering rate (deposition rate with and without an applied bias) as the parameter for disruption of the columnar morphology; however this parameter does not take into consideration the backscattering from the gas phase which will be greater with higher gas pressure, and so must be used with caution.

Figure 6.5. Fracture cross-section (bottom) and surface morphology (top) of a thick RF sputter deposited chromium deposit: (a) without bias (no bombardment), and (b) with concurrent bombardment (-500 V bias on the substrate).[25]

378 8.6

Deposition Technologies for Films and Coatings Residual Film Stress

Invariably atomistically deposited films have a residual growth stresses which may be tensile or compressive in nature and may approach the yield or fracture strength of the materials involved. The origin of these stresses is poorly understood although several phenomenological models have been proposed.[180] Generally, vacuum-deposited films and sputter-deposited films prepared at high pressures have tensile stresses which may be anisotropic with off-normal angle-of-incidence depositions. In low pressure sputter deposition and ion plating, energetic particle bombardment may give rise to high compressive film stresses due to the recoil implantation of surface atoms.[181]-[184] This effect is sometimes called atomic peening. Studies of deposited films with concurrent bombardment have shown that the conversion of tensile stress to compressive stress is very dependent on the ratio of bombarding species to depositing species.[27][185] In plasma processing, the residual film stress may be very sensitive to the substrate bias and gas pressure[43] during deposition in a plasma environment. High intrinsic film stresses may lead to long-term film stability problems such as roomtemperature grain growth[186] and void formation.[187] Figure 6.6 shows the residual stress and gas content in sputter deposited chromium films as a function of substrate bias.[25] Figure 6.7 shows the anisotropic residual stresses in post-cathode magnetron sputter-deposited molybdenum films as a function of sputtering gas pressure and orientation.[43] Where rather thick films of high modulus materials are involved, these stresses must be controlled or spontaneous failure (adhesion, cracking, blistering) will occur.[171] The stresses may be controlled by controlling the film thickness, materials involved, film morphology, bias during deposition, deposition temperature, and/or sputtering pressure.[43] The lattice strain associated with the film stress represents stored energy and this energy, along with a high concentration of lattice defects, may lead to:(a) lowering of the recrystallization temperature in crystalline materials,(b) a lowered strain point in glassy materials,(c) a high chemical etch rate, (d) electromigration problems, (e) void growth in metallization lines by creep, and (f) other such mass transport effects. 8.7

Crystallographic Orientation

Under proper bombardment conditions, the crystallographic orientation of the deposited material is developed such that the more dense

Ion Plating

379

Figure 6.6. Residual stress and gas content of an RF sputter-deposited chromium deposit as a function of substrate bias during RF sputter deposition.[25]

Figure 6.7. Stress and stress anisotropy in post-cathode magnetron sputterdeposited molybdenum films as a function of orientation and sputtering gas pressure. [43]

380


crystallographic planes are parallel to the bombarding direction.[188][189] This effect is attributed to the channeling of the bombarding species into the film thus decreasing the sputtering rate under this orientation. Under more energetic bombardment condition, however, the crystallographic orientation is disrupted due to the formation and consolidation of defects. 8.8

Gas Incorporation

When a depositing film is bombarded during deposition by energetic gaseous particles, the incorporated gas content is dependent on the particle energy, substrate temperature, film material, and bombarding species. Generally, low mass bombarding particles are more easily incorporated than are large mass particles. The gas incorporation increases with energy of the bombarding species to the point that heating causes gas desorption. Under some conditions, very high concentrations of normally insoluble gas may be incorporated into the depositing film by concurrent bombardment during deposition. An example is the incorporation of 20 - 40 atomic percent hydrogen and helium in gold[135][136][190] and the incorporation of krypton in amorphous metals films.[137] This incorporation is probably due in part to the high lattice defect concentration in the bombarded material which trap mobile species. At very high gas contents, the gas will precipitate into voids. Gas incorporation can be minimized by using low-energy bombarding species (i.e., less than 100 eV), an elevated substrate temperature during deposition (300 - 400°C), and/or using higher atomic mass bombarding species (Kr, Xe, Hg). 8.9

Surface Coverage

The macroscopic and microscopic surface coverage of a deposited film on a substrate surface may be improved by the use of concurrent bombardment during film deposition. The macroscopic ability to cover complex geometries depends mostly on scattering of the depositing material in the gas phase.[1][191][192] The gas scattering by collision in the gas phase may be aided bygas pumping[193]-[195] in the discharge, which will give a directed velocity to the film-atoms toward the substrate surface. If gas scattering is extensive then gas phase nucleation will occur and the resulting deposit will be poorly consolidated. If a plasma is present and the substrate is at a negative potential, the gas phase nucleated materials will become negatively charged and repelled from the substrate. In addition, bombard-

Ion Plating

381

ment will heat, densify, and consolidate the deposited material into a high quality film over the whole surface. On a more microscopic scale, sputtering and redeposition of the depositing film material will lead to better coverage on micron and submicron sized features,[24][179][196]-[199] and reduce pinhole formation. On the atomic scale, the increased surface mobility, increased nucleation density, and erosion/redeposition of the depositing adatoms will disrupt the columnar microstructure and eliminate the porosity along the columns. In total, the use of gas scattering, along with concurrent bombardment, increases the surface covering ability and decreases the microscopic porosity of the deposited film material as long as gas incorporation does not generate voids. 8.10 Other Properties Many other properties of the film material may be changed and improved by bombardment during deposition. They include: (a) electrical resistivity of metal films,(b) hardness of hard-coatings,(c) chemical etch rate,(d) corrosion resistance,(e) pinhole density,(f) index of refraction of dielectric coatings, (g) color of TiN films, etc.

9.0

ION PLATING SYSTEM REQUIREMENTS

Generally the equipment used for ion plating is the same as that used for sputter deposition except that now the substrate is the sputtering target and another vaporization source has been added. Figure 6.8 shows some of the possible configurations. 9.1

Vacuum System

This system is similar to sputter deposition equipment. A good base pressure and little contaminant desorption during processing is desirable in order to keep the contaminant level in the plasma low. 9.2

High Voltage Components

This is also similar to sputter deposition equipment with more attention paid to substrate cooling and the means of providing a high voltage connection to the substrate. Because of the high “throwing power” condi–

382


Figure 6.8. Some configurations for bombarding a surface from a plasma by using accelerated or reflected high energy particles: (a) diode, (b) grid to allow bombardment of complex surfaces or insulators, (c) thermoelectron sustained plasma with magnetic enhancement/confinement, (d) e-beam evaporation with a differentially pumped vacuum chamber, (e) utilizing reflected high energy neutrals and sputtering, (f) magnetron sputtering source, and (g) moving magnetron plasma to allow uniform bombardment of substrate surface.

Ion Plating

383

tions often used in ion plating systems, the insulators of electrical feedthroughs must be carefully shielded from deposition or else they will become shorted. These conditions may also lead to gas-phase nucleation of particles which will deposit on the system walls (called “black sooty crap”, BSC, by the operators). This material has a very low density and if the material is pyrophoric (Ti, Zr etc.) the BSC may ignite if disturbed in air. In such a case system cleanup should be done wet. Power supplies must be capable of withstanding electrical arcs in the deposition chamber and their attendant electrical transients. Substrate potentials may be DC or RF and with or without magnetron enhancement. In some configurations (beam, grounded cathode) the substrate may be at ground potential. Very high voltage (to 1-00 keV) pulsing of substrates immersed in plasmas is being studied as a way to modify surfaces by ion bombardment.[199] This technique could be used in ion plating to allow periodic bombardment of the depositing film material and might be termed pulsed ion plating. 9.3

Gas Handling System

This equipment is also similar to sputter deposition. Inert gases should be purified in order to decrease the contaminants in the plasma. Reactive gases may be injected directly toward the deposition region and some concern must be given to having a uniform flux of reactive gases to the deposition region. 9.4

Evaporation/Sublimation Sources

In electron beam evaporation into a plasma, the filament region of the gun must be differentially pumped in order to prevent sputter erosion of the high voltage filament. Evaporation/sublimation in a gas environment takes more power than in a vacuum because of heat loses by convection and backscattering of the vaporized material. 9.5

Sputtering Sources

Sputter target fixturing will be coated by the sputtered material. Target fixtures and nearby material will be sputtered by the bombarding ions or high energy neutrals. This may lead to contamination of the deposited material. In order to get around this problem, the target fixtures may first be coated by the material to be deposited.

384 9.6

Deposition Technologies for Films and Coatings Plasma Uniformity

Plasma uniformity over the substrate surface is desirable when using the plasma as the source of bombarding particles or the source of reactive species. Often this plasma is non-uniform due to the geometry of the surface, non-uniformity of the electric field, or non-uniformity of the magnetic field. The electric field uniformity can be improved in some cases by enclosing the surface in a high transmission grid at the same potential as the part. The magnetic field uniformity can be improved by using multiple polepieces or by using moving magnetic fields. 9.7

Plasma Generation Near the Substrate Surface

In many cases the substrate potential may be used to create a plasma near the substrate if the gas pressure and magnetic field are of the appropriate nature. In some sputtering-source configurations, i.e., DC magnetron sputtering, the plasma is held away from the substrate and it may be desirable to sustain a plasma near the substrate surface. This may be done by having an auxiliary plasma generating technique such as a hot filament or hollow cathode triode configuration near the substrates. For reactive deposition, it may be desirable to inject the reactive gases directly into this region. 9.8

Substrate Fixturing

This is similar to sputter deposition equipment with more attention paid to substrate cooling and means of providing a high voltage connection to the substrate. Moveable fixturing may be necessary in order to not leave points of electrical contact uncoated. The barrel-plating fixturing may also be used to give 100% coverage of the substrate. [200] Heating of the substrate holder may be done by having poor thermal contact to a heat sink, by using embedded heaters, or by radiant heating. Since appreciable heat in introduced into the substrate surface, substrate cooling may be an important concern. Coolants used to actively cool the substrate holder must be isolated from ground by using insulating tubing. Some leakage to ground can be expected though ionic conduction. A cold finger substrate holder may be used to allow direct cooling of the substrate.

Ion Plating

385

Holding-fixtures and nearby material will be sputtered by the bombarding ions or high energy neutrals. This may lead to contamination of the deposited material. In order to get around this problem the fixtures may first be covered by the material to be deposited.

10.0 PROCESS MONITORING AND CONTROL The ion plating process is very complex and monitoring what goes on is usually done in a comparative manner rather than an absolute manner. This means careful control of process variables. 10.1 Plasma Typically the plasma is monitored by the gas pressure, gas flow through the system, and power input. Calibrated flow meters are useful for process control. The species in the plasma may be monitored by the optical emission from the plasma, and this analysis may be correlated to the composition and properties of the resulting film.[201] In some cases, mass spectrometry may be used to determine gas species but if a plasma is used in the process, the mass spectrometer must be differentially pumped and the calibration is confused by the initial presence of ionized species. 10.2 Substrate Temperature Substrate surface temperature monitoring is complicated by the presence of the plasma and high voltage. Thermocouple readouts must be isolated from ground and float at the substrate potential. Infrared temperature monitoring techniques may be used.[202] 10.3 Specifications Process specifications and reproducible processing are the keys to good ion plating processing. Because of the many process variables and the problems of determining some of the possible variables absolutely, specifications and reliable process monitors are particularly important in the ion plating process.

386


11.0 PROBLEM AREAS Problem areas which exist with ion plating arise because of: “Activated” reactive contaminants in the plasma Redeposition of sputtered contaminates Plasma uniformity High field regions Points Focusing fields High plasma density region Trapped secondary electrons High secondary emission Unwanted shielding of substrate areas Wasted power Unshielded leads/areas Substrate heating Coating (shorting) high voltage insulators Non-uniform availability of reactive species (reactive deposition) Electrical contact points are not coated Gas incorporation These problem areas can generally be avoided by design of the fixturing and other “tricks” such as: Rotating fixturing Use of grids to smooth out electric fields Periodic plating (on-off) mode to reduce substrate heating Heating of substrate to outgas material as it deposits A major problem area in using plasmas for thin film deposition is that of obtaining a uniform plasma density over a surface so that uniform bombardment and reactive gas availability can be attained. Plasma non-uniformity can arise from a number of sources including: (a) geometrical arrangement of power input electrodes and substrate fixturing,(b) substrate geometry,(c) the presence of surfaces that allow recombination and loss of species in the nearby plasma, and (d) in the case of reactive deposition, reactive surfaces that deplete the supply of reactive gas at the growing film surface.

Ion Plating

387

As a general rule, the best system design is one that is geometrically symmetric. However, in many instances a symmetric geometry is difficult to attain. The use of magnetron configurations is an example. The use of a magnetic field to confine electrons and increase the local plasma density in one region leads to a decrease in plasma density in some other region. Figure 6.9 shows an example of how two independently sustained plasmas may be used to allow magnetron sputtering of a source and the use of a hot-filamentsustained plasma in the vicinity of the substrate to provide a plasma from which ions can be extracted to bombard the substrate and film. If the part has a very complex configuration, the electric field around points and corners focus the bombardment giving high erosion rates and heating in these areas. A thin region gives poor thermal conductance and results in heating. Holes and re-entrant features give low field gradients. In these regions heating will be high and erosion will be low, giving poor cleaning and allowing reaction with contamination. Excessive heating can sometimes be alleviated by pulse processing where the substrate bias is periodically turned on and off.

Figure 6.9. Example of using an auxiliary plasma near the substrate to allow ions to be extracted and bombard the growing film.

388


In some cases high transparency grids at the substrate potential may be used to surround the substrate giving a more uniform bombardment over a complex surface. This is the basis of the equipment used in the ion vapor deposition (IVD) process and in the barrel-plating ion plating configuration. Figure 6.10 shows a barrel-plating configuration used to coat small parts which are tumbled in the rotating cage. A grid configuration may also be useful in coating dielectric materials where charge buildup may be a problem, or in coating moving substrates where electrical contact may be a problem.

Figure 6.10. Ion plating “barrel plating” configuration using a rotating cage to contain the parts.[200]

Ion Plating

389

As with any plasma process, wall effects enhance the desorption of contamination. This contamination, when introduced into the plasma, is “activated” and can be an important source of contamination which must be controlled.

12.0 APPLICATIONS There are many applications of the ion plating process some of which are: ! Obtaining good adhesion: Ag on steel for mirrors, soft metals on surfaces for space lubrication, Ag on Be for diffusion bonding, Cu and Au on Ta and Mo for subsequent brazing, Cu-on-ceramic metallization. ! Metallization: Al, Ag, Au on plastics and semiconductors. ! Good surface coverage on complex surfaces: TiN on tool bits, molds and jewelry items; semiconductor metallization. ! Good reaction and stoichiometry: TiN on tool bits, molds (hardness, wear); jewelry items. ! Corrosion protection: Al on U, steel and Ti (galvanic); C and Ta on biological implants. ! Abrasion resistance: MgF2 coatings on plastics. ! Deposition of diffusion barriers: HfN and TiN on semiconductor devices.

13.0 SUMMARY Like any deposition technique the ion plating process has its advantages and disadvantages. Advantages: ! Excellent surface covering ability(throwing power) under the proper conditions. ! Ability to have in situ cleaning of the substrate surface.

390

Deposition Technologies for Films and Coatings ! Ability to obtain good adhesion in many otherwise difficult systems. ! A great deal of flexibility in tailoring film properties by controlling bombardment conditions. ! Equipment requirements are roughly equivalent to those of sputter deposition. Disadvantages: ! Many processing parameter that must be controlled. ! Processing may be very dependent on substrate geometry and fixturing. ! Obtaining uniform bombardment and reactive species availability over a complex surface may be difficult. ! Gas incorporation may be excessive. ! Substrate heating may be excessive. ! Contamination is desorbed from surfaces and activated in the discharge and can contaminate deposited material.

In order to achieve the desired film property modification, there must be an appreciable ratio of bombarding particles to depositing species. This ratio must be much higher to disrupt the columnar morphology than is necessary to change the film stress. The necessary bombardment conditions for each application are usually determined empirically and controlled by controlling the processing geometry and parameters. A typical condition to control film stress might be a substrate bias of -50 to -100 volts DC, a current density of 1 mA/cm2 and a deposition rate of 10 nanometers per second. For columnar structure disruption and maximum covering ability, a resputtering rate might be 30%. The ion plating process provides an alternative film deposition technique which should be evaluated for specific applications.

Ion Plating

391

REFERENCES 1. Mattox, D. M., J. Vac. Sci. Tech., 10:47 (1973) 2. Ahmed, N. A. G., Ion Plating Tech., - Dev. and Appl., J. Wiley (1987) 3. Mattox, D. M.,Electrochem. Tech., 2:295 (1964);Sandia Laboratories Report SC-DR -63-285 (Nov 1963) 4. Mattox, D. M., US Patent # 3,329,601 (1974) 5. Berghaus, B., UK Patent # 510 993 (1938) 6. Maissel, L. I. and Schaible, P. M., J. Appl. Phys., 36:237 (1965) 7. Mattox, D. M., J. Electrochem. Soc., 115:1255 (1968) 8. Muehlberger, D. E.,Ion Plating and Implantation, (R. F. Hochman, ed.) p. 75, Conference Proceedings American Society for Metals (1986) 9. Steube, K. E. and McCrary, L. E., J. Vac. Sci. Tech., 11:362 (1974) 10. Ahmed, N. A. G., Surf. Eng., 2:95 (1986) 11. Perry, A. J., Thin Solid Films, 146:165 (1987) 12. Martin, P. J., [rev] Vacuum, 36:585 (1986) 13. Bunshah, R. F., Ch. 4, “Evaporation,” this book 14. Culbertson, R. and Mattox, D. M., 8th Conf. on Tube Tech., pp., 101107, IEEE Conf. Record 1966; Culbertson US Patent 3,604,970 (1971) 15. Sterling, H. F. and Swann, R. C. G.,Solid-State Electron, 8:653 (1965) 16. Martin, P. J., Netterfield, R. P., Sainty, W. G. and Pacey, C. G., J. Vac. Sci. Tech., A2:341 (1984); Martin, P. J., J. Mater. Sci., 21:1 (1986) 17. Harper, J. M. E., Cuomo, J. J., and Kaufman, H. R., J. Vac. Sci. Tech., 21:737 (1982) 18. Greene, J. E., Motooka, T., Sundgren, J. E., Lubbens, D., Gorbotkin, S., and Barnett, S. A., Nucl. Instrum. Methods Physics Res., B27:226 (1987); Greene, J. E., J. Vac. Sci. Tech., B1:229 (1983) 19. Aisenberg, S. and Chabot, R. W., J. Vac. Sci. Tech., 10:104 (1973) 20. Howson, R. P., Avaritsiotis, J. N., Ridge, M. I., and Bishop, C. A., Thin Solid Films, 63:163 (1979) 21. Mattox, D. M. and Bland, R. D., J. Nucl. Mater., 21:349 (1967) 22. Mattox, D. M., Rev. Sci. Instrum., 37:1609 (1966) 23. Mattox, D. M. and Kominiak, G. J., J. Vac. Sci. Tech., 9:528 (1972)

392


24. Kominiak, G. J. and Mattox, D. M., J. Electrochem. Soc., 120:1535 (1973) 25. Bland, R. D., Kominiak, G. J. and Mattox, D. M., J. Vac. Sci. Tech., 11:671 (1974) 26. Mattox, D. M. and Kominiak, G. J., J. Vac. Sci. Tech., 9:528 (1972) 27. Hoffman, D. W. and Gaerttner, M. R.,J. Vac. Sci. Tech., 17:425 (1980) 28. Harper, J. M. E., Cuomo, J. J., and Henzell, H. T. G., Appl. Phys. Lett., 43:547 (1983) 29. Cuomo, J. J.,Ion Plating and Implantation, (R. F. Hochman, ed.) p. 25, Conference Proceedings American Society for Metals (1986) 30. Wie, C. R., Tang, J. Y., Tombrello, T. A., Grant, R. W., and Housley, R. M., Vacuum, 38:157 (1988) 31. Baglin, J. E. E.,Ion Beam Modification of Insulators, (P. Mazzolsdi and G. Arnold, eds.), Ch.15, Elsevier (1987) 32. Forrester, A. T., Large Area Ion Beams: Fundamentals of Generation and Propagation, J. Wiley (1988) 33. Goebel, D. M., Campbell, G., and Conn, R. W.,J. Nucl. Mater., 121:277 (1984) 34. Williams, D. G., J. Vac. Sci. Tech., 11:374 (1974) 35. Bergman, C.,Ion Plating and Implantation, (R. F. Hochman, ed.) p. 115, Conference Proceedings American Society for Metals (1986) 36. Valyi, L., Atom and Ion Sources, Wiley (1977) 37. Mattox, D. M., Adhesion Measurement of Thin Films, Thick Films and Bulk Coatings, ASTM STP 640, (K. L. Mittal, ed.), p. 54, American Society for Testing and Materials (1978) 38. Vossen, J. L., Thomas, J. H., III, Maa, J. S., and O’Neill, J. J., J. Vac. Sci. Tech., A2:212 (1984) 39. Brillson, L. J., Thin Solid Films, 89:461 (1982) 40. Kaufman, H. R., J. Vac. Sci. Tech., 15:272 (1978) 41. Lee, W. W. Y. and Oblas, D., J. Appl. Phys., 46:1728 (1975) 42. Hagstrum, H. D., Inelastic Ion Surface Collisions, (N. H. Tolk, J. C. Tully, W. Heiland and C. W. White, eds.), pp. 1 - 25, Academic Press, (1977) 43. Cuthrell, R. E., Mattox, D. M., Peeples, C. R., Dreike, P. L., and Lamppa, K. P., J. Vac. Sci. Tech., A6:2914 (1988) 44. Shubaly, M. R., Nucl. Instrum. Method Phys. Res., B26:195 (1987)

Ion Plating

393

45. Romesser, T. E., Vanek, V., Tang, J., Dixon, D., Bayless, J., Musetto, M., Strawitch, C., and Higgins, L., A Large Area Plasma Source, IEEE International Conference of Plasma Science, 83CH1847-3, US DOE Contract No. DE-ACO3-77ET33006 (1983) 46. Letokhov, V. S., Sov. At Energy, (Translation) 62(4):297 (1987) 47. Cuomo, J. J., Gambino, R. J., Harper, J. M. E., Kuptsis, J. D., and Webber, J. C., J. Vac. Sci. Tech., 15:281 (1978) 48. Boxman, R. L. and Goldsmith, S., Surf. Coat. Tech., 33:153 (1987) 49. Melngailis, J., J. Vac. Sci. Tech., B5:469 (1987) 50. Chapman, B., Glow Discharge Processes, J. Wiley (1980) 51. Thornton, J. A., Thin Solid Films, 107:3 (1983) 52. Mattox, D. M., Plasma-Surface Interactions and Processing of Materials, (O. Auciello, A. Gras-Marti, J. A. Valles-Abarca, and D. Flamm, eds.), 176:377, NATO ASI Series, Kluwer Academic Publishers (1988) 53. Mattox, D. M., Plasma-surface Engineering, (E. Broszeit, W. D. Munz, H. Oechsner, K.-T. Rie, and G. K. Wolf, eds.), 1:15, Informationsgesellschaft, Verlag (1988) 54. Gras-Marti, A., Abril, I. and Valles-Abarca, J. A., Thin Solid Films, 124:59 (1985) 55. Davis, W. D. and Vanderslice, J. A., Phys. Rev., 131:219 (1963) 56. Savlnier, J. P., Fzguerra, J., and Guille, J., Vacuum, 33:279 (983) 57. Somekh, R. R., J. Vac. Sci. Tech., A2:1285 (1984) 58. Bessaudou, A. and Weissmantel, C.,Thin Solid Films, 149:225 (1987) 59. Logan, J. S., Mazza, N. M., and Davidse, P. D., J. Vac. Sci. Tech., 6:120 (1969) 60. Yatsuzuka, M., Morishita, K., Satoh, K., and Nobuhara, S., Jpn. J. Appl. Phys., 24:1724 (1985) 61. Kohler, K., Coburn, J. W., Horne, D. E., Kay, E., and Keller, J. H., J. Appl. Phys., 57:59 (1985) 62. Thompson, B. E., Allen, K. D., Richards, A. D., and Sawin, H. H., J. Appl. Phys., 59:1890 (1986) 63. Coburn, J. W., Ann. Rev. Mat. Sci., 13:91 (1983) 64. Kaufman, H. R. and Rossnagel, S. M., J. Vac. Sci. Tech., A6:2572 (1988) 65. Kushner, M. J., J. Appl. Phys., 58:4024 (1985)

394


66. Horwitz, C. M., J. Vac. Sci. Tech., A1:1795 (1983) 67. Lamont, L. T., Jr., and Turner, F. T., J. Vac. Sci. Tech., 11:47 (1974) 68. de Vries, C. A. M. and van den Hoek, W. G. M., J. Appl. Phys., 58:2074 (1985) 69. Suzuki, M. K., Ninomiya, K., Nishimatsu, S., Thoman, J. W., Jr., and Steinfeld, J. I., Jpn. J. Appl. Phys., 25:1569 (1986) 70. Hull, US Patent # 4,431,901 (Feb. 14, 1984) 71. Vella, M. C., Ehlers, K. W., Kippenhan, D., Pincosy, P. A., Pyle, R. V., DiVergilioi, W. F., and Fosnight, V. V., J. Vac. Sci. Tech., A3:1218 (1985) 72. Kashiwagi, K., Kobayashi, K., Masuyama, A., and Murayama, Y., J. Vac. Sci. Tech., A4:210 (1986) 73. Dahimene, M. and Asmussen, J., J. Vac. Sci. Tech., B4:126 (1986) 74. Kaufman, H. R. and Robinson, R. S.,J. Vac. Sci. Tech., A3:1774 (1985) 75. Goebel, D. M., Hirooka, Y. and Sketchy, T. A., Rev. Sci. Instrum., 56:1717 (1985) 76. Kaufmann, H., US Patent # 4,346,123 (Aug 24, 1982) 77. Morley, J. R. and Smith, H. R., J. Vac. Sci. Tech., 9:1377 (1972) 78. Schalansky, C., Munir, Z. A. and Walmsley, D. L., J. Mat. Sci., 22:745 (1987) 79. Larson, D. T. and Draper, H. L., Thin Solid Films, 107:327 (1983) 80. Mah, G., Mcleod, P. S. and Williams, D. G., J. Vac. Sci. Tech., 11:663 (1974) 81. Komiya, S. and Tsuruoka, K., J. Vac. Sci. Tech., 12:589 (1975) 82. Murayama, Y. and Takao, T., Thin Solid Films, 40:309 (1977) 83. Kaufman, H. R. and Robinson, R. S.,J. Vac. Sci. Tech., A3:1774 (1985) 84. Kuo, Y. S., Bunshah, R. F., and Okrent, D., J. Vac. Sci. Tech., A4:397 (1986) 85. Thornton, J. A., Surf Eng, 2:283 (1986) 86. Brown, I. G., Feinberg, B., and Galvin, J. E., J. Appl. Phys., 63:4889 (1988) 87. Dorodnov, A. M., Sov. Phys. Tech. Phys., 23:1058 (1978) 88. Zanner, F. J., and Bertram, L. A., Proc. 8th Int'l. Conf. on Vacuum Metallurgy,1:512 (1985)

Ion Plating

395

89. Zanner, F. J. and Bertram, L. A., IEEE Trans. on Plasma Science, Vol PS-11:223 (1983) 90. Boelens, S. and Veltrop, H., Surf. Coat. Tech., 33:63 (1987) 91. McDaniel, E. W.,Collision Phenomena in Ionized Gases, Wiley (1964) 92. Hollahan, J. R. and Bell, A. T.,Techniques and Applications of Plasma Chemistry, J. Wiley (1972) 93. McTaggert, F. K.,Plasma Chemistry in Electrical Discharges, Elsevier (1967) 94. Kushner, M. J., J. Appl. Phys., 53:2923 (1982) 95. Veprek, S., Thin Solid Films, 130:135 (1985) 96. Thornton, J. A., J. Vac. Sci. Tech., 15:188 (1978) 97. Coburn, J. W. and Chen, M., J. Appl. Phys., 51:3134 (1980) 98. Dreyfus, R. W., Jasinski, J. M., Walkup, R. E., and Selwyn, G. S., Pure Appl. Chem., 57:1265 (1985) 99. Wormhoudt, J., Stanton, A. C., Richards, A. D., and Sawin, H. H., J. Appl. Phys., 61:142 (1987) 100. Murakami, A. T., Namura, T., and Itatani, R.,Jpn. J. Appl. Phys., 24:512 (1985) 101. Plas, F., Guille, J., and Machet, J., Vide, Suppl., 196:45 (1979) 102. Hoffman, D. W. and Badgley, J. S.,J. Vac. Sci. Tech., A6:1691 (1988); Theoretical treatment to be published - private communication with D. W. Hoffman 103. Hoffman, D. W. and Thornton, J. A., J. Vac. Sci. Tech., 20:355 (1982) 104. Hoffman, D. W. and Thornton, J. A., J. Vac. Sci. Tech., 16:134 (1979) 105. Thornton, J. A. and Hoffman, D. W., J. Vac. Sci. Tech., A3:576 (1985) 106. Mathews, A. and Gethin, D. T., Thin Solid Films, 117:261 (1987) 107. Roth, J., Sputtering by Particle Bombardment ll, (R. Behrisch, ed.), p. 91, Springer-Verlag (1983) 108. Itoh, N. and Hasebe, Y., J. Nucl. Mat., 139:77 (1986) 109. Winters, H. F., Coburn, J. W., and Chuang, T. J., J. Vac. Sci. Tech., B1:469 (1983) 110. Topics in Applied Physics, Vol. 47 (1981); Vol. 52, (R. Behrisch, ed.), Springer-Verlag (1983) 111. Szymonsaki, M., Huang, W., and Onsgaard, J., Nucl. Instrum. Met. Phys. Res., B14:263 (1986)

396


112. Matsuda, Y., Yamamura, Y., Ueda, Y., Uchino, K., Muraoka, K., Maeda, M., and Akazaki, M., Jpn. J. Appl. Phys., 25:8 (1986) 113. Wehner, G. K., J. Vac. Sci. Tech., A1:487 (1983) 114. Betz, G., Surf. Sci., 92:283 (1980) 115. Baritzky, B. and Taglauer, E., Surf. Sci., 162:996 (1985) 116. Cuomo, J. J. and Gambino, R. J., J. Vac. Sci. Tech., 12:79 (1975) 117. Mundy, J. N. and Rothman, S. J., J. Vac. Sci. Tech., A1:74 (1983) 118. Navinsek, G., Prog. Surf Sci., 7:49 (1976) 119. Wehner, G. K., J. Vac. Sci. Tech., A3:1821 (1985) 120. Panitz, J. K. and Sharp, D. J., J. Vac. Sci. Tech., 17:282 (1980) 121. Berg, R. S. and Kominiak, G. J., J. Vac. Sci. Tech., 13:403 (1976) 122. Kowalski, Z. W., J. Mat. Sci., 20:15211 (985 ) 123. Winters, H. F. and Sigmund, P. S., J. Appl. Phys., 45:4760 (1974) 124. Kominiak, G. J. and Mattox, D. M., Thin Solid Films, 40:141 (1977) 125. Jacobs, M. H., in Surface Modification and Coatings, (R. D. Sisson, Jr. ed.), p. 291, ASM Conference Proceedings (1986) 126. Geis, M. W., Lincoln, G. A., Efremow, N., and Piacentini, W. J., J. Vac. Sci. Tech., 19:1390 (1981) 127. Tsou, L. Y., J. Electrochem. Soc., 132:2010 (1985) 128. Lincoln, G. A., Geis, M. W., Pang, S., and Efremow, N., J. Vac. Sci. Tech., B1:1043 (1983) 129. Winters, H. F., Nucl. Instrum. Met. Phys. Res., B27:243 (1987) 130. Harper, J. M. E., Cuomo, J. J., and Henzell, H. T. G., Appl. Phys. Lett., 36:456 (1980); also Appl. Phys. Lett., 37:540 (1980) 131. Ashok, S., Krautle, H., Beneking, H., and Mogro-Campero, A., Thin Solid Films, 126:251 (1985) 132. Winters, H. F., J. Appl. Phys., 38:3928 (1967) 133. Kornelsen, E. V., Rad. Effects, 13:227 (1972) 134. Kornelsen, E. V. and Van Gorkum, A. A., Rad. Effects, 42:93, 112 (1979) 135. Mattox, D. M. and Kominiak, G. J., J. Vac. Sci. Tech., 11:52 (1974) 136. Mattox, D. M. and Kominiak, G. J., J. Vac. Sci. Tech., 8:194 (1971) 137. Cuomo, J. J. and Gambino, R. J., J. Vac. Sci. Tech., 14:152 (1977) 138. Chleck, D., Maehl, R., Cucchiara, O. and Carnevale, E., Int. J. Appl. Radiation and Isotopes, 14:581 (1963)

Ion Plating

397

139. Cheleck, D. and Maehl, R.,Int. J. Appl. Radiation and Isotopes, 14:593 (1963) 140. Chleck, D. and Cucchiara, O., Int. J. Appl. Radiation and Isotopes, 14:599 (1963) 141. Carden, J. E., Isotopes and Radiation Tech., 3(3):206 (1964) 142. Carden, J. E., Isotopes and Radiation Tech., 3(4):318 (1964) 143. Sharp, D. J. and Panitz, J. K., Surf. Sci., 118:429 (1982) 144. Corbett, J. W., Surf. Sci., 90:205 (1979) 145. Miyoshi, K., Spalvins, T., and Buckley, D. H.,Thin Solid Films, 127:115 (1985) 146. Mah, G., Mcleod, P. S., and Williams, D. G., J. Vac. Sci. Tech., 11:663 (1974) 147. Gibson, V. J. and Kennemore, C. M., III,Thin Solid Films, 124:27 (1985) 148. Satou, M., Andoh, Y., Ogata, K., Suzuki, Y., Matsuda, K., and Fujimoto, F., Jpn. J. Appl. Phys., 24:656 (1985) 149. Jacobs, M. H., Surf. Coat. Tech., 29:221 (1986) 150. Cuomo, J. J., Ion Plating and Implantation: Applications to Materials, (R. F. Hochman, ed.) ASM Conference Proceedings (1986) 151. Kimblin, C. W., J. Appl. Phys., 45:5235 (1974) 152. Daalder, J. E., J. Phys. D, Appl. Phys., 9:2379 (1976) 153. Daalder, J. E., Physica., 104:91 (1981) 154. Zanner, F. J. and Bertram, L. A., IEEE Trans. on Plasma Sci., Vol. PS11:223 (1983) 155. Mori, T. and Namba, Y., J. Vac. Sci. Tech., A1:23 (1983) 156. Bubenzer, A., Dischler, B., and Nyaiesh, A., Thin Solid Films, 91:81 (1982) 157. Weissmantel, C., Thin Films from Free Atoms and Particles, (K. J. Klabunde, ed.) p. 153, Academic Press (1985) 158. Shanfield, S. and Wolfson, R., J. Vac. Sci. Tech., A1:323 (1983) 159. Harper, J. M. E., Cuomo, J. J., and Henzell, H. T. G., J. Appl. Phys., 58:550 (1985) 160. Mattox, D. M., J. Vac. Sci. Tech., A7(3):1105 (1989) 161. Shawki, G. S. A., El-Sherbiny, M. G., and Salem, F. B., Thin Solid Films, 75:29 (1981)

398


162. Hibbs, M. K., Johansson, B. O., Sundgren, J. E., and Helmersson, U., Thin Solid Films, 122:115 (1984) 163. Miranda, R. and Rojo, J. M., Vacuum, 34:1069 (1984) 164. Rossnagel, S. M., Surf. Sci., 123:89 (1982) 165. Chopra, K. L., J. Appl. Phys., 37:2249 (1969) 166. Bovey, P. E., Vacuum, 19:497 (1969) 167. Mattox, D. M., J. Appl. Phys., 37:3613 (1966) 168. Muller, K. H., J. Vac. Sci. Tech., A4:184 (1986) 169. Narusawa, T., Shimizu, S., and Komiya, S.,J. Vac. Sci. Tech., 16:366 (1979) 170. Harper, J. M. E., Cuomo, J. J., Gambino, R. J., and Kaufman, H. R.,Ion Bombardment Modification of Surfaces Fundamentals and Applications, (O. Aucello and R. Kelly, eds.), Ch. 4, Elsevier (1984) 171. Mattox, D. M. and Cuthrell, R. E., MRS Symposium Proceedings, Adhesion in Solids, (D. M. Mattox, J. E. E. Baglin, R. E. Gottschall, and C. D. Batich, eds.), 119:141 (1988) 172. Oh, R. S., Cannon, R. M., and Richie, R. O., J. Cer. Soc., 70:C352 (1987) 173. Thornton, J. A., Thin Solid Films, 40:335 (1977) 174. Thornton, J. A., Ann. Rev. Mat. Sci., 7:239 (1977) 175. Thornton, J. A., J. Vac. Sci. Tech., A4:3059 (1986) 176. Messier, R., Giri, A. P., and Roy, R. A., J. Vac. Sci. Tech., A2:500 (1984) 177. Meissier, R. and Yehoda, J. E., J. Appl. Phys., 58:3739 (1985) 178. Patten, J. W., Thin Solid Films, 63:1211(979 ) 179. Panitz, J. K. G., Draper, B. L., and Curlee, R. M., Thin Solid Films, 166:45, (1988) 180. Klokholm, E. and Berry, B. S., J. Electrochem. Soc., 115:823 (1968) 181. Blachman, A. G.. J. Vac. Sci. Tech., 10:299 (1973) 182. Thornton, J. A., Tabcock, J., and Hoffman, D. W., Thin Solid Films, 64:111 (1979) 183. Thornton, J. A. and Hoffman, D. W., J. Vac. Sci. Tech., 18:203 (1981) 184. Thornton, J. A. and Hoffman, D. W., J. Vac. Sci. Tech., A3:576 (1985) 185. Brighton, D. R. and Hubler, G. K., Nucl. Instrum. Method Phys. Res., B28:527 (1987)

Ion Plating

399

186. Patten, J. W., McClanahan, E. D. and Johnston, J. W., J. Appl. Phys., 42:4371 (1971) 187. Yost, F. G., Romig, A. D., Jr., and Bourcier, R. J., Sandia Laboratories Report SAND88-0946 available from NTIS 188. Dobrev, D., Thin Solid Films, 92:41 (1982) 189. Kay, E., Parmigiani, F., and Parrish, W., J. Vac. Sci. Tech., A5:44 (1987) 190. Weaver, H. T., J. Appl. Phys., 42:2356 (1971) 191. Fancey, K. S. and Beynon, J., Vacuum, 34:591 (1984) 192. Fancey, K. S. and Mathews, A., Surf. Coating Tech., 36:233, (1988) 193. Chester, A. N., Phys. Rev., 169:172 (1968) 194. Hoffman, D. W., J. Vac. Sci. Tech., A3:561 (1985) 195. Rossnagel, S. M., J. Vac. Sci. Tech., A6:19 (1988) 196. Homma, Y. and Tsunekawa, S.,J. Electrochem. Soc., 132:1466 (1985) 197. Skelly, D. W. and Gruenke, L. A., J. Vac. Sci. Tech., A4:457 (1986) 198. Bader, H. P. and Lardon, M. A., J. Vac. Sci. Tech., A3:2167 (1985) 199. Conrad, J. R., Radtke, J. L., Dodd, R. A., Worzala, F. J., and Tran, N. C., J. Appl. Phys., 62:4591 (1987) 200. Mattox, D. M. and Rebarchik, F. N., J. Electrochem. Tech., 6:3741 (1968) 201. Ricard, A., Michel, H., Jauquot, P., and Gantois, M., Thin Solid Films, 124:167 (1985) 202. Korotchenko, V. and Matthews, A., Vacuum, 36:61 (1986)

7 Chemical Vapor Deposition Jan-Otto Carlsson

1.0

INTRODUCTION

Chemical vapor deposition (CVD) is a process whereby a solid material is deposited from a vapor by a chemical reaction occurring on or in the vicinity of a normally heated substrate surface. The solid material is obtained as a coating, a powder, or as single crystals. By varying the experimental conditions—substrate material, substrate temperature, composition of the reaction gas mixture, total pressure gas flows, etc.—materials with different properties can be grown. A characteristic feature of the CVD technique is its excellent throwing power, enabling the production of coatings of uniform thickness and properties with a low porosity even on substrates of complicated shape. Another characteristic feature is the possibility of localized, orselectivedeposition, on patterned substrates. CVD is employed in many thin film applications. It is, for instance, used in the microelectronics industry to make films serving as dielectrics, conductors, passivation layers, oxidation barriers, and epitaxial layers. The production of optical fibers as well as wear-, corrosion-, and heatresistant coatings with this technique is well known. Other CVD applications are the preparation of high temperature materials (tungsten, ceramics, etc.) and the production of solar cells, of high temperature fiber composites, and of particles of well-defined sizes. Recently, high-Tc superconductors have also been made by this technique. Since oxygen activity in the vapor can be precisely controlled during the deposition, no annealing in oxygen is needed to achieve superconductivity.

400

Chemical Vapor Deposition

401

There exist several types of CVD processes. In thermally activated CVD (TACVD), the deposition is initiated and maintained by heat. However, photons, electrons, and ions, as well as a combination of these (plasmaactivated CVD), may induce and maintain CVD reactions. In this chapter, the underlying principles of TACVD are introduced. In addition to large-area deposition, selective CVD on patterned substrates is discussed.

2.0

IMPORTANT REACTION ZONES IN CVD

In CVD, gaseous reactants are admitted into a reactor (see Fig. 7.1). Near or on a heated substrate surface, a chemical reaction of the following type occurs: Gaseous reactants → Solid material + Gaseous products Because of the gas flows as well as the temperature used in CVD, five important reaction zones are developed during the CVD process (see Fig. 7.2). The properties of CVD materials are affected by the interacting processes occurring in these reaction zones. In a CVD process, a main gas flow (the reaction gas mixture) passes over the substrate/coating surface. For fluid dynamical reasons, a more or less stagnant boundary layer occurs in the vapor adjacent to the substrate/coating. During the deposition process, the gaseous reactants and the gaseous reaction products are transported across this boundary layer. In reaction zone 1 (see Fig. 7.2) as well as in the main gas stream, homogeneous reactions in the vapor may occur. These reactions may lead to an undesirable homogeneous nucleation resulting in a flaky and non-adherent coating. In some cases however, these reactions, when not accompanied by homogeneous nucleation, are favorable to the CVD process (for instance CVD of Al2 O3,[1] of B13C2,[2] and of Si,[3] respectively). The heterogeneous reactions occur in the phase boundary vapor/coating (zone 2). These reactions determine, in many systems, the deposition rate and the properties of the coating. Relatively high temperatures may be used during CVD. This means that various solid state reactions (phase transformations, precipitation, recrystallization, grain growth, for example) may occur during the process (the zones 3 - 5). In zone 4, which is a diffusion zone, various intermediate phases may be formed. The reactions in this zone are important for the adhesion of the coating to the substrate.

402


Figure 7.1. The principle of CVD.

Figure 7.2. Important reaction zones in CVD.

3.0

DESIGN OF CVD EXPERIMENTS

Every CVD experiment is unique. However, some general aspects in designing CVD experiments can be given. The design is usually an iterative procedure. For instance, the choice of the reaction gas mixture affects the design of the CVD system, the cleaning procedure, the adhesion of the coating, etc.


3.1

403

Classification of CVD Reactions

CVD processes frequently proceed by complicated chemical reaction schemes. However, use of overall CVD reactions enables a classification to be made. Thermal decomposition reactions or pyrolytic reactions mean, in this case, that a gaseous compound AX is thermally dissociated into A (a solid material) and X (a gaseous reaction product). AX(g) → A(s) + X(g) Use of thermal decomposition reactions normally results in relatively pure coatings. Examples of some thermal decomposition reactions are given below: SiH4(g) → Si(s) + 2 H2(g) B2 H6(g) → B(s) + 3 H2(g) Ni(CO)4 (g) → Ni(s) + 4 CO(g) Si(CH3 )Cl3(g) → SiC(s) + 3 HCl(g) Processes like carburizing and nitriding may also be classified in this category of reaction. In carburizing, for instance, a carbon-carrying vapor species, e.g., methane, is allowed to react at/on a heated surface. Methane then decomposes in principle according to CH4(g) → C(s) + 2 H2(g) The deposited carbon reacts immediately with the substrate yielding a solid solution of carbon in the substrate and/or—if they exist—carbides of the substrate material. Reduction reactions, where hydrogen acts a reducing agent, are frequently used (see alsoCoupled reactions below). 2 AX(g) + H2(g) → 2 A(s) + 2 HX(g) Straightforward reduction reactions have been almost exclusively used in the CVD of elements.

404

Deposition Technologies for Films and Coatings WF6(g) + 3 H2 (g) → W(s) + 6 HF(g) 2 BCl3(g) + 3 H2(g) → 2 B(s) + 6 HCl(g) SiCl4 (g) + 2 H2(g) → Si(s) + 4 HCl(g)

Exchange reactions mean that an element E replaces another element, for instance X, in the molecule AX according to AX(g) + E(g) → AE(s) + X(g) Examples of exchange reactions are: Zn(g) + H2S(g) → ZnS(s) + H2(g) SiCl4 (g) + CH4(g) → SiC(s) + 4 HCl(g) SnCl4(g) + O 2(g) → SnO2(g) + 2 Cl2(g) Disproportionation reactions are rarely used in CVD. Disproportionation means a reaction where the oxidation number of an element both increases and decreases through the formation of two new species. CVD of A from AX can be obtained in disproportionations like 2 AX(g) → A(s) + AX2(g) 3 AX(g) → 2 A(s) + AX 3(g) 4 AX(g) → 3 A(s) + AX 4(g) Examples of disproportionation reactions are 2 GeI2 (g) → Ge(s) + GeI4(g) 2 TiCl2 (g) → Ti(s) + TiCl4(g) 2 SiI2 (g) → Si(s) + SiI4(g) Coupled reactions are often used in CVD. For instance, CVD of Al2O3 from AlCl3, CO2 and H2 can be described in an overall reaction:


405

2 AlCl3(g) + 3 CO2(g) + 3 H2(g) → Al2O3(s) + 3 CO(g) + 6 HCl(g) where the reaction in which water is formed CO2(g) + H2 (g) → CO(g) + H2O(g) is coupled to the hydrolysis reaction AlCl3(g) + 3 H2 O(g) → Al2O3(s) + 6 HCl(g) i.e., a reduction reaction is coupled to an exchange reaction (for example, see Ref. 7.1). Other examples of overall coupled CVD reactions are TiCl4 (g) + NH3(g) + ½ H2 (g) → TiN(s) + 4 HCl(g) Ga(CH3 )3 (g) + xPH3 (g) + (l-x)AsH3(g) → GaAs1-xP x(s) + 3 CH4(g) In general, several possibilities of preparing a substance by CVD exist. For practical reasons however, relatively few alternatives will remain after a critical evaluation of the requirements of the process (temperature, total pressure, compatibility with the substrate and the reactor, the reactions gas mixture, costs, toxicity of the substances, etc). 3.2

Thermodynamics

Thermodynamic calculations are a useful tool when choosing the experimental conditions (temperature, total pressure, reaction gas composition) for the deposition of a certain substance, and also serve as a guide when changing the experimental conditions in a CVD process. For the calculations, different computer programs are in use and there are now practically no limitations in the number of substances that can be included in the calculations. For reviews of computational methods, the reader is referred to Refs. 7.4 - 7.7. Usually the computer programs are based on the so-calledfree energy minimization technique. The free energy G is given by the following equation: G = ∑ n iµ i i

406


where ni is the number of moles of a substance i and µi is the chemical potential of the substance. The chemical potential is defined as µi = µi0 + RTlnai where µi is the reference chemical potential, and ai is the activity. Assuming ideal gas conditions, the activity of the gaseous species may be expressed as its partial pressure ai = pi = (ni /n)P where n is the number of moles in the gas phase, and P the total pressure. For pure condensed substances, the activities are equal to unity. Eriksson[8] developed a computer program (SOLGAS) based on the minimization of the free energy. This program became a prototype for many other equilibrium calculation programs. From the basic equations given above, dimensionless quantity G/RT was defined (see eq. below) and used in the calculations. G = RT

g

 0 n ig g µ ni  +  + ln P + ln n  RT  i i =1 m

∑

n

∑ i =0

s

 0 s µ ni    RT  i

The superscripts g and s refer to the gas phase and the solid phase, respectively. The value of µ0/RT for a specific substance is calculated from

(

)

0 ∆H f0,298 G 0 − H 298 µ0 = + RT RT RT

where:

(Go - Ho)/RT = free energy function ∆H0f,298 = heat of formation at 298.15 K.

By minimizing the quantity G/RT (or G) and using mass balance equations as subsidiary conditions, the equilibrium composition of a system can be calculated. The input data in the calculations are the number of moles of the different reactants, the total pressure, the substrate temperature, the different substances, and their thermochemical data. From the calculations, various quantities like the partial pressures of the vapor species, the amounts of the different substances available for CVD, i.e., the yield, thermodynamic functions (supersaturation, reaction enthalpies, driving force of different processes, etc.), are obtained.


407

Figures 7.3 - 7.5 illustrates results from equilibrium calculations. Figure 7.3 shows the change in equilibrium composition when SiH4 is added to an H2/ WF6 gas mixture. For an overview of the experimental conditions for depositing a certain substance, CVD phase diagrams are constructed.[10] Figs. 7.4 and 7.5 are examples of calculated CVD phase diagrams. The number of variables required to construct a complete CVD phase diagram is given by the phase rule. Normally various sections (constant temperature, constant total pressure, constant molar ratio between two of the reactants, while varying the number of moles of a third reactant) are used. Finally, for more theoretical work,predominance diagrams with element chemical potentials as variables are employed. In these diagrams, the phase stability ranges are limited by straight lines.

Figure 7.3. Partial pressures of vapor species in the homogeneous reaction between H 2, WF6 and SiH4. Total pressure 0.1 Torr, temperature 300°C, H2/WF6 = 39.[9]

408


Figure 7.4. Calculated CVD phase diagram for the W - Si system. Reactants WF6, SiH4, and H2, total pressure 0.1 Torr, H2/WF6 = 39.[9]

Figure 7.5. Calculated diagram for deposition of YBa2Cu3O7-x. The grey-shaded stability regions contain the superconducting phase. The contour lines represent the yield of YBa2Cu3O7-x. Precursors YCl3, BaI 2, and CuCl, O2, and H2O, molar ratios YCl3:BaI2:CuCl = 1:2:3, O2:H2O = 1:1, total pressure 1 kPa.[11]


409

The reliability of the equilibrium computations described above is dependent on the availability and accuracy of thermochemical data as well as the identification of all the substances—vapor species and condensed phases—that are of importance in the system. Examples of sources of thermochemical data are given in Refs. 12 - 14. In cases were data do not exist or the data are unsatisfactory or unreliable, estimation procedures can be used (see, for instance, Ref. 12). Finally, a few references illustrating the use of thermodynamic computations in CVD have been selected.[15]-[24] 3.3

Adhesion

Production of well-adhering coatings with desired properties is the ultimate aim of all CVD work. There are, however, several factors which reduce the adhesion between the coating and the substrate. Stresses introduced as deposition stresses or originating from a mismatch in the thermal expansion coefficients between the substrate and the coating when cooling down after the deposition process. These stresses can be reduced by depositing a substance prior to the final CVD process. The predeposited substance forms an intermediate layer. The stresses can also be reduced by decreasing the thickness of the coating as well as by changing the grain size and morphology of the coating. Homogeneous nucleation in the vapor produces a flaky/powdery deposit. By reducing the degree of supersaturation or the driving force of the process, the homogeneous nucleation can be eliminated. Intermetallic compounds formed in the coating/substrate interface may be brittle, leading to the initiation of cracks there. The risk of crack initiation increases with increasing thickness of the layer containing the intermetallic compounds. The technique of predeposition of a substance—forming an intermediate layer later—may be usable to improve the adhesion in this case. Hydriding of the substrate may cause bad adhesion. Hydrogen is frequently used in the cleaning procedure prior to the deposition stage. Some metals/alloys can dissolve a considerable amount of hydrogen. If the deposition process is then run at a temperature where the hydrogen is liberated, cracking of the coating occurs. Hydriding can be eliminated by using another cleaning procedure or heating the substrate in vacuum or an inert gas after cleaning in hydrogen. Pores in the coating/substrate interface reduce the adhesion not only because of the fewer bonds in the interface but also because the pores act as crack initiators. The pores can originate from the coalescence step at the

410


beginning of the process as well as from Kirkendal diffusion (differences in the diffusion fluxes of the atoms over the coating/substrate interface). Oxide films or other surface contaminants reduce the adhesion as a rule. A proper cleaning procedure usually can solve this problem. Chemical attack on the substrate by the reaction products formed during the CVD process may cause bad adhesion. The chemical attack on the substrate can occur as long as the substrate is exposed to the vapor and is described in the following reactions: 2 AX(g) + H2 (g) → 2 A(s) + 2 HX(g) The volatile reaction product HX formed reacts with the substrate S according to the reaction 2 S(s) + 2 HX(g) → 2 SX(s) + H2(g) The solid substance SX formed may result in poor adhesion. The reaction above can be predicted from thermodyamics. 3.4

Substrate Cleaning Procedures

A clean substrate surface free from oxides and other contaminants is a prerequisite condition for good adhesion. The cleaning procedure depends on the substrate used, the material to be deposited, the CVD equipment available, etc. Examples of some cleaning techniques are given below. Before the substrates are placed in the reactor, pickling, grit blasting, etching, degreasing, etc., are carried out. In the CVD reactor surfaces containing hydrogen-reducible oxides, e.g., tungsten oxides, are heated in a hydrogen gas flow at temperatures above the deposition temperature. Metals forming volatile oxides are cleaned by heating in an inert atmosphere. Finally, the heating operations in the cleaning step remove dust particles from the surface, in some cases by the formation of carbides in the surface. After the cleaning procedure the reactor is purged with an inert gas/hydrogen before the deposition process (interlayers or final coating). 3.5

The CVD system

The choice of the CVD system is affected by a number of factors: the reactants used in the process, the maximum acceptable leak rate for air into the system, purity of the deposit, size and shape of the substrate, process


411

economy, etc. In the following, some general comments on the design of CVD systems are given. A CVD system is advantageously constructed in three modules: 1. The reaction gas dispensing system. 2. The reactor, including components for defining the gas flows. 3. The exhaust system containing a total pressure controller, vacuum pump, scrubber and/or reactant recycle system. 3.6

The Gas Dispensing System

Reactants, which are gases at room temperature, are stored in gas bottles. After pressure regulation, their flows are measured with, for instance, mass flow meters. Use of mass flow meters yields high accuracy and allows microprocessor control of gas flows. Those reactants that are liquids or solids at room temperature have to be fed to the system in other ways (see Fig. 7.6). They can be admitted to the system by simply heating them above the boiling or sublimation point. The evaporation rate can be varied by varying the source temperature and/or the dimensions of the capillary from the sources. Another way of introducing these substances is to use an evaporator or sublimator and a carrier gas. When the evaporator is used, the carrier gas is bubbled through the liquid to be evaporated or flowed above its surface. The carrier gas picks up the liquid substance and transports it into the reactor. The evaporation rate depends on the temperature of the liquid, the liquid level in the container, and the flow rate of the carrier gas. For the highest reproducibility it is important to have a constant level of the liquid in the container. However, some alternatives to these evaporators exist which use carrier gases and are independent of the liquid level. In one alternative, the liquid is evaporated from a vessel, cooled and condensed in a cooler, leaving the carrier gas saturated at the temperature of the cooler. If two or more reactant liquids have to be used in the process, it is seldom possible to vaporize them in the same evaporator while maintaining the predetermined molar ratio since they normally have different vapor pressures. The principle of the sublimator is similar to that of the evaporator. In a sublimator the substance is transferred to the vapor by sublimation (solid→ gas) and then transported to the reactor by the carrier gas.

412


Figure 7.6. Sketch of a CVD system.

Non-gaseous reactants at room temperature can also be admitted into the reactor by generating them in situ in the gas dispensing system. If, for instance, the halide AlCl3 is to be used in a process, the generator is filled with aluminium sponge. Aluminium chlorides are then obtained by leading hydrogen chloride through the generator. Generator variables are temperature, flow rate, and concentration of the hydrogen chloride (varied by dilution with an inert gas).


413

Direct metering of liquids/solids followed by immediate vaporization in a vessel can also be used. For metering of liquids, flow meters and various dispensing pumps are available. The final vaporization takes place in, for instance, a flash vaporizer[25]—a vessel containing pieces of porcelain of high temperature. Many CVD processes are strongly affected by contaminants in the vapor. The contaminants originate from the reactants themselves and from various chemical reactions between the gases and the materials in the gas dispensing system (in the tubes, evaporators, sublimators) and from air leakage. The contamination level can be reduced by: ! Purifying the reactants. Hydrogen and argon can be purified to a level of 1 ppm in commercially available purifiers ! Having a low leak rate ! Using carrier gases which are non-reactive against the materials to be vaporized (in evaporators and sublimators) ! Using materials in the tubes, vaporizers, reactors, etc., which are compatible with the gases used ! Using degassed O-rings, where they are used for vacuum seals ! Installing purge line, which is important when reactive gases, e.g., halides, are used Finally, in CVD, explosive, flammable and toxic gases (hydrogen, silane, phosphine, arsine) are frequently employed. Correct handling of the gases is, therefore, necessary. Every precaution should be taken. Effective ventilation systems and gas detectors (commercially available) should be used. 3.7

The Reactor

The process selected and the size, shape, and number of substrates define the type of reactor and its geometry. Two main reactor types can be distinguished: 1. In the hot wall reactor (see Fig. 7.7), the reactor tube is surrounded by a tube furnace. This means that the substrates and the wall of the reactor have the same temperature. In addition to the film growth occurring on the substrates, film growth might thus take place on theinside of the reactor walls. With thicker films on the reactor walls,

414

Deposition Technologies for Films and Coatings there is a risk that particles will break loose from reactor walls, fall down on the surface of the growing film, and introduce pinholes in it. There might also be a source of contamination in this reactor type because of the reaction between the material of the reactor wall and the vapor. In the hot wall reactor, homogeneous reactions, affecting the deposition reactions and hence the structure of the films, may take place in the vapor. There is a successive depletion with respect to the reactants as they are transported through the reactor. Such a depletion may yield different deposition conditions within the reactor. Finally, in a hot wall reactor, many substrates can be deposited simultaneously.

Figure 7.7. A hot wall CVD reactor.

2. In the cold wall reactor (Fig. 7.8), the walls of the reactor are cold and usually no depositionoccurs on the walls, eliminating the risk of particles breaking loose from the walls. Furthermore, a low wall-temperature reduces the risk of contaminating vapor/wall reactions. In the cold wall reactor, the homogenous reactions in the vapor are suppressed and the importance of the surface reactions is increased. The steep temperature gradients near the substrate surface may introduce severe natural convection resulting in a non-uniform film thickness and microstructure. However, with the higher flexibility of the cold wall reactor, high cleanliness, high deposition rates (yielding high wafer throughput), high cooling rates combined with the needs of thickness uniformity, automatic wafer handling and use of increasing wafer diameter, there is tendency to more frequently use cold wall reactors in the microelectronics.


415

Figure 7.8. A cold wall CVD reactor.

Various techniques of heating the substrates exist.[26] Conductive substrates can be heated resistively or by radio frequency induction. Nonconductive substrates are normally heated by applying optical techniques (tungsten filament lamps, lasers), thermal radiation techniques, or by using susceptors and radio frequency induction heating. Examples of some reactors are shown in Fig. 7.9. Finally for coating a large number of small pieces, fluidized bed techniques can be applied.[25] To illustrate how the choice of reactor is dependent on the substrate to be coated, an example of applying a coating inside a tube is given. In this case the tube itself is the reactor. The reactants are introduced in the tube and transported to the heated zone where the deposition occurs. Induction heating as well as tubular furnace heating can be employed. By moving the tube or the heating sources continuously, a coating of uniform thickness can be produced (see, for instance, Ref. 27). The arrangement of the gas flows as well as the gas flow rate are of highest importance for obtaining good coatings. Gas flow dynamics are discussed in Sec. 4. 3.8

The Exhaust System

The exhaust system contains a vacuum pump, total pressure control, scrubbers, and a recycling system, if used. Processes working at atmospheric pressure do not require vacuum pumps and total pressure control. At reduced pressures, however, pumps as well as some kind of total pressure control have to be used.

416


Figure 7.9. Examples of some CVD reactors. (a) and (b) RF heated cold wall reactors, (c) vertical hot wall reactor, (d) barrel reactor.

The choice of the vacuum pump depends on the process (pumping capacity required, pressure range to be used, gases to be pumped). At higher process pressures (>30 Torr), water ring pumps and different mechanical chemical pumps are used. The chemical pumps are also employed at lower pressures (1 Torr), and at the lowest pressures in combination with, for instance, mechanical boosters. When mechanical pumps are used in CVD processes, the pump oil can polymerize or be damaged in other ways by certain gaseous species. The pump oilshould


417

be chosen with respect to its compatibility with the specific gaseous species. The polymerization of the oil can easily be followed by measuring its viscosity at different times. Mechanical pumps also produce back-diffusion of oil molecules into the system. The back-diffusion can be stopped in a trap (zeolite trap, liquid nitrogen cold trap) just before the pump. With the current trend of using lower pressures to create abrupt interfaces and superlattices, diffusion (to pump hydrogen) and turbo pumps are also utilized. Finally, external oil filtering systems reduce the wear of the mechanical pumps in processes where solid particles are formed and transported in the vapor to the pump. In a CVD process, more or less toxic, explosive, and corrosive gases are used/formed. To remove them before exhaust, scrubbers are used. The scrubber type is appropriate to the CVD process used. Halides can easily be neutralized in a water scrubber. Carbon monoxide and hydrogen can be burnt in a flame. Arsine can be removed by simply heating the reactor gas in a furnace especially arranged for this purpose (i.e., with a high efficiency for stripping arsenic from the gas stream). Recycling is frequently used to improve process economy. It becomes necessary in large scale processes, where expensive reactants are utilized and the conversion efficiency of the reactants is low. The technique of recycling varies from process to process. A simple recycling can be achieved in some processes by selective condensation. It can easily be applied in systems where the component to be recycled has the highest boiling point. In the production of boron fibers for instance—where hydrogen and boron trichloride are used—the unconverted boron trichloride is condensed in the exit stream from the reactor, while the hydrogen and the hydrogen chloride (formed in the process) are not condensed. 3.9

Analysis of the Vapor in a CVD Reactor

Various spectroscopic techniques have been used to analyze the vapor in a CVD reactor. The purpose of these analyses is to achieve a better understanding of the processes. Spectroscopic techniques are also used for process control. Mass-, Raman, and IR-spectroscopy are in use.[28]-[30]

4.0

GAS FLOW DYNAMICS

The rate and arrangements of the gas flows in a CVD reactor influence the deposition conditions considerably. In the following, some fundamentals

418


of gas flow dynamics are given. For further details the reader is recommended textbooks in chemical engineering or other books treating transport processes. In a gas, different states exist. In the molecular state, the mean free path of the molecules is much longer than the dimensions of the vessel. In the viscous state, the mean free path is much shorter than the vessel dimensions. The viscous state can be divided into two flow regimes. The laminar flow regime, where the flowing gas layers are parallel, is appropriate to low gas velocities. At higher velocities, the flow becomes turbulent. The limit between the laminar and the turbulent flow is defined by the value of Reynold’s number, Re:

where:

Re =

ρ • V •D η

ρ V η D

the density of the gas the velocity the viscosity the diameter of the tube

= = = =

At Re < 1100 the flow is laminar, while at Re > 2100 the flow is turbulent. The range 1100 to 2100 is a mixed flow regime. The Reynold’s number given characterizes the flow in an isothermal environment. In the non-isothermal environment existing in a cold wall reactor, natural convection induces a turbulence even at low flow rates. Consider the situation above a heated surface (Fig. 7.10). At small temperature gradients dT/dx, the varying density of the gas along the coordinate X is compensated by the gravitational field and no movement of the gas occurs. At larger gradients, the gas starts to move and the laminar flow can no longer be retained. From Fig. 7.10 it can be understood that turbulence at a heated substrate surface may be obtained at different parts of it. For instance, when the temperature gradient is perpendicular to the gravitational field turbulence occurs at smaller temperature gradients than in the antiparallel case. Different dimensionless quantities are used for identifying conditions of laminar and turbulent flows at different geometries. For instance, the Rayleigh number, Ra, and the Grashof number, Gr, are employed.[31] Ra and Gr are related to each other. Gr multiplied by the Prandtl number (nearly equal to one for gases) yields Ra.


419

Figure 7.10. Forces at a heated substrate surface. The value, g, the gravitational force, dT/dx the temperature gradient.

To summarize the flow situation, diagrams depicting flow stability regions like that in Fig. 7.11 are constructed for different geometries and reaction gas mixtures. In an isothermal environment, Gr is equal to zero and Re describes the situation completely. In a non-isothermal environment Gr is larger than zero (increases with increasingDT). Turbulence occurs at a certain Gr value, depending on the flow rate of the specific gas mixture and the temperature difference between the hot and cold part in the reactor.

Figure 7.11. Flow diagram showing flow stability regions.

420


In many CVD processe,s the laminar flow region is normally used. High flow rates (turbulence) usually decrease the conversion efficiency of the reactants to the coating and very large gas volumes have to be handled. The flow situation around the object to be coated can be visualized in smoke experiments where the smoke is generated inside the reactor from, for instance, titanium tetrachloride and water. 4.1

Gas Flow Patterns

For growth of films of uniform thicknesses and compositions the gas flow patterns are of greatest importance. This is particularly the case when “high” pressures (about 1 atm) are used. At reduced pressures the diffusivity of the vapor species increases, which results in a better mixing of the process gases, and hence the flow fields become less important. The gas flow patterns are very complicated in many CVD reactors because the flow is driven by both the pressure differences (forced convection) and gravity (free convection) in mostly complex reactor geometries. Free convections contributes to the gas flow pattern not only in cold-wall reactors with their steep temperature gradients but also in hot wall reactors with small axial temperature gradients. These are employed for correction of the successive depletion of the vapor with respect to the reactants as they flow through the reactor. Fluid flow phenomena characteristic of various CVD reactors have been reviewed by, for instance, Westphal[32] and Jensen.[33] In gas flow calculations, the continuity equation for the total mass, for the single components, for the energy, and for the momentum must be solved. For a suitable choice of experimental conditions (flow regimes and reactor geometries) simplifying equations and boundary conditions—resulting in reasonable computer times—are obtained. As an introduction to this field results from detailed flow calculations for two main reactor types are summarized. Wahl[34] has calculated the flow fields in some cold wall reactors for the laminar flow region (atmospheric pressure) for the CVD of silicon nitride from SiH4 and N2. The reactor geometry investigated can be seen in Fig. 7.12. The flow patterns calculated for this geometry and the inverted geometry (difference in the buoyance-driven convection) are shown in Fig. 7.12. The flow pattern becomes more complicated in the inverted geometry, i.e., when the forced convection and the gravity interact. The flow pattern, including generation of loops and rolls,was strongly dependent on the ratio between the free convection and the forced convection.


421

Figure 7.12. Gas flow pattern in a cold wall reactor, where the forced and buoyancedriven convection (a) interact, and (b) counteract, substrate temperature 900 K, Re = 50.[34]

To show the influence of the reactor geometry on the flow pattern, a calculation of Wahl and Hoffman[35] will be taken as an example. The reactor geometry considered as well as the results from the calculations are shown in Fig. 7.13. As can be seen, the flow pattern in this geometry is not as complicated as that obtained in the previous geometry (Fig. 7.12), where the diameter of the inlet gas tube was half the diameter of the hot plate.

422


Figure 7.13. Gas flow pattern in a cold wall reactor with a geometry different from that in Fig. 12.[35]

A technique frequently used for correction of the successive depletion of the reactants as they are transported through a hot wall reactor is the application of a temperature gradient in the axial (flow) direction of the reactor. Even small temperature gradients, however, can induce buoyancy-driven convection. The flow pattern in a hot wall reactor with a temperature gradient for the atmospheric CVD of GaAs in the Ga-AsCl3-H2 system has been calculated for different temperature gradients, different gas flow velocities, and different reactor heights by Westphal et al.[32] A typical result from their calculations is shown in Fig. 7.14. It can be seen that a convection roll, induced by free convection, is generated. The effect of free convection on the gas flow pattern decreased with decreasing temperature gradients, increasing gas flow velocities and decreasing reactor heights. No extreme conditions were required to generate convection rolls. They were, for instance, obtained at a temperature gradient of 6 K cm-1, a gas flow velocity of 2 cm s-1 and a reactor height of 5 cm.


423

Figure 7.14. Gas flow pattern in a hot wall reactor with a temperature gradient of 6 K cm-1, linear gas flow velocity of 2 cm s-1, and channel height 5 cm, deposition system Ga-AsCl3, H2.[32]

Convection rolls are frequently generated in CVD. These rolls cause dilution of the reaction gas with reaction products, resulting in an alteration of the deposition conditions. Developed rolls may yield problems for multilayer growth with well-defined phase boundaries and for the creation of sharp doping profiles. By using extreme low total pressures (in the 10-3 Torr range), these problems can be solved. 4.2

Boundary Layers

In CVD the substrates are immersed in a gas stream. From fluid mechanics it is known (see, for instance, Ref. 36) that so-called boundary layers are developed near the substrate surface. The boundary layers are defined as the region near the substrate surface where the gas stream velocity, the concentration of the vapor species and the temperature are not equal to those in the main gas stream. Thus a velocity boundary layer, a concentration boundary layer and a thermal boundary layer exist. The development of a velocity boundary layer in a laminar flow region is sketched in Fig. 7.15. The gas velocity is zero at the substrate surface and increases to a constant value (the bulk gas flow velocity). The layer over which the gas flow velocity varies is the boundary layer. The thicknessδ of a boundary layer (laminar flow) at a position X on the substrate or susceptor[36] is given by δ = a(ηX/ρν)1/2

424


where a is a proportionality constant, η is the viscosity of the gas, ν is the velocity of the gas and ρ is the density of the gas. From a knowledge of the temperature and pressure dependence of η, ρ and ν[31][36] it is deduced that the thickness of the boundary layer increases with increasing temperature and decreasing total pressure. Moreover, the thickness also increases with increasing transport distance of the gases along the substrate surface.

Figure 7.15. Definition of the velocity boundary layer.

The development of boundary layers in CVD situations has been investigated both experimentally and theoretically. Eversteijn et al.,[37] used smoke experiments to visualize the flow pattern in a horizontal epitaxial reactor. The smoke was generated from TiCl4 and water. They observed an immobile layer of gas, called the stagnant boundary layer, above the susceptor. It was shown later, however, that in steep temperature gradients (near the susceptor) fine particles are driven away from the susceptor by thermophoretic forces.[38] This shows that smoke experiments can only be used to map the flow at a greater distances from a heated susceptor. Ban and Gilbert[28] investigated the heat transport in a cold wall reactor by heating a susceptor in helium and measuring the temperature at different locations above the susceptor with a thermocouple with a small diameter. The very steep temperature gradient can be seen in Fig. 7.16. Ban and Gilbert also investigated the concentration profiles of various vapor species in silicon CVD from an H2 /SiCl4 gas mixture. They introduced a fine capillary probe for a mass spectrometer at different locations above the susceptor. The concentration profile of SiCl4 and the reaction product HCl can be seen in Fig. 7.17. The thickness of the concentration boundary layer in this case is more than 2 cm. The successive depletion with respect to the reactants as they were transported through the reactor is shown in Fig. 7.18. At a height of 7 mm above the susceptor and 15 cm downstream of


425

Figure 7.16. Temperature profile in helium. Linear gas flow velocity: 24.9 cm s-1.[28]

Figure 7.17. Concentration profiles of SiCl4 and HCl in the CVD of silicon from SiCl4 and Hc. Transport distance along the susceptor: 12.5 cm, linear gas flow velocity: 24.9 cm s-1, - - - 1000°C, —— 1140°C. [28]

426


the susceptor, the partial pressure of SiCl4 was reduced to about 50% of the initial value. Sedgwick et al.[29] measured temperature and concentration profiles in an air-cooled horizontal cold wall reactor using Raman scattering. They observed a steep temperature gradient near the susceptor. The temperature profile developed was dependent on the position along the susceptor.

Figure 7.18. Partial pressure profile of SiCl 4 as a function of the transport distance along the susceptor at a height of 7 mm above the susceptor.[28]

Giling[39] investigated the gas flow patterns and temperature profiles at atmospheric pressure in air-cooled as well as water-cooled horizontal epitaxial reactors by means of interference holography. The gases used were H2, He, N2 and Ar. H2 and He yielded stable laminar flows through both the watercooled and the air-cooled reactor. At flow velocities higher than 40 cm s-1 a cold gas finger, indicating incompletely developed flow and temperature profiles, was observed in the air-cooled reactor. N2 and Ar behaved quite differently from H2 and He and different convective effects were obtained. At flow velocities higher than 4 cm s-1 for instance, a laminar layer about 8 mm thick was developed near the susceptor, while the gas above this layer appeared to be in turbulence.


427

Giling also pointed out the importance of entrance effects, i.e., that it will take some distance (the entrance length) from the susceptor edge for full velocity and temperature profiles to develop. According to Schlichting[36] the entrance length for the development of the full velocity profile is given by the equation X = 0.04hRe where h is the height of the channel and Re is the Reynolds number. Hwang and Cheng[40] predicted that the thermal entrance length was seven times longer than the flow entrance length. Giling confirmed this for H2 in his measurements. Coltrin et al.[41] have developed a mathematical model of silicon CVD from silane in a cold wall reactor. The model includes gas phase chemistry as well as fluid mechanics, and predicts temperature, velocity, and concentration profiles for many vapor species. Figure 7.19 depicts the temperature contour in a typical calculation. The thickness of the boundary layer is in the centimeter range and increases with increasing gas transport distance along the susceptor.

Figure 7.19. Calculated temperature contours for silicon CVD from silane (0.6 Torr) and helium as a carrier gas (600 Torr). Temperature: 1018 K, gas flow velocity: 15.3 cm s-1.[41]

428


4.3

Mass Transport Processes Across a Boundary Layer

Different mass transport processes across a boundary layer can be distinguished: 1. Fickian diffusion occurs because of the concentration gradient across the boundary layer. 2. Thermal diffusion or Soret diffusion is induced by a temperature gradient, in for instance, a cold wall reactor.[31] This diffusion is of greatest importance in systems having large differences in molecular weights and molecular sizes between vapor species. 3. A concentration gradient implies a density gradient, resulting in a buoyancy-drivenadvective flux.[42] 4. In the overall CVD reaction, the number of moles of gas may be changed. This induces a flux (Stefan flux) towards or away from the substrate surface. In, for instance, the CVD of boron from BCl3 and H2 according to the reaction 2 BCl3(g) + 3 H2(g) → 2 B(s) + 6 HCl(g) the number of moles in the vapor is changed from 5 to 6, causing a flux from the substrate.[43]

5.0

RATE-LIMITING STEPS DURING CVD

In a CVD process various sequential steps occur. Each of these steps may be rate-limiting in the absence of thermodynamic limitations. Plausible rate-limiting steps are as follows (see also Fig. 7.20): (a) transport of the gaseous reactants to the boundary layer surrounding the substrate (free and forced convection);(b) transport of the gaseous reactants across the boundary layer to the surface of the substrate (diffusion and convections flows); (c) adsorption of the reactants on the surface of the substrate; (d) chemical reactions (surface reactions between adsorbed species, between adsorbed species and reactants in the vapor and or between reactants in the vapor),(e) nucleation (at least at the initial stage); (f) desorption of some of the reaction products from the surface of the substrate; (g) transport of the reaction products across the boundary layer to the bulk gas mixture; (h) transport of the reaction products away from the boundary layer. In each of these steps several processes may proceed simultaneously.


429

Figure 7.20. The various steps in a CVD process.

Even though several rate-limiting steps can be identified in a CVD process, only five main categories of control are normally discussed: 1. Thermodynamic control. Thermodynamic control means that the deposition rate is equal to the mass input rate into the reactor (corrected for the yield of the process). This occurs at extreme deposition conditions (very low flow rates, high temperatures, etc.). The temperature dependence of the deposition rate is obtained from thermodynamic calculations. 2. Surface kinetics control. If the deposition rate is lower than the mass input rate into the reactor and the mass transport rate in the vapor in the reactor to or from the substrate, a surface kinetics control or nucleation control exist. The surface kinetics control is favorable for obtaining coatings of uniform thicknesses on more complicated shaped substrates. The mechanisms of surface reactions are discussed in Sec. 6. 3. Mass transport control. A process may also be controlled by the mass transport in the vapor in the reactor to or from the substrate surface. This occurs frequently at high pressures and high temperatures. 4. Nucleation control. At low supersaturations the deposition rate may be controlled by the nucleation. 5. Homogeneous reaction control. In some processes the formation rate of key species in the vapor may control the deposition rate.

430


Since the mass transport in the vapor or the surface kinetics usually controls the deposition rate, the following discussions are limited to just these two cases. Surface kinetics control is normally desirable and means a maximum in throwing power or step coverage. Figure 7.21 shows conditions of complete mass transport control, complete surface kinetics control and mixed control. In the surface kinetics control, a fast diffusion in the vapor is combined with a slow surface reaction. For a mass transport control, the surface kinetics is fast while the mass transport in the vapor is slow.

Figure 7.21. Diagrams illustrating situations of complete mass transport control in the vapor (a), and surface kinetics control (b), respectively, and (c) shows conditions of mixed control.


431

Reaction resistances are often used to predict rate-limiting steps or control in CVD. To illustrate their principle use, reaction resistances are employed to define the surface reaction control and the mass transport control, respectively. The diffusion flux JD across the boundary layer is given by JD = where:

D Pb − Ps R• T δ

D = diffusion coefficient R = gas constant T = absolute temperature δ = boundary layer thickness Pb and Ps (see Fig. 7.21)

The mass flux JM towards the surface is expressed as JM =

km ( Ps − Peq) RT

where km is the mass transfer coefficient. Pb and Peq are known from the reaction gas composition and from thermodynamic calculations, respectively. Ps can be eliminated by assuming steady-state conditions (JM = JD)

Ps =

km • δ • Peq D km • δ + 1 D

Pb +

Km is thus given by Pb − Peq JM = 1 RT δ + 1 D km Pb - P eq is the driving force of the process and δ/D and 1/km are reaction resistances. If δ/D >> 1/km the process is controlled by the mass transport in the vapor, while surface reaction control is achieved at 1/km << δ.

432


As said before, surface kinetics control is a condition prerequisite to the obtaining of coatings of uniform thickness on substrates with a complicated shape. How can surface reaction control be achieved? To answer this question, the temperature and pressure dependences of the reaction resistances have to be analyzed. The thickness δ of the boundary layer (laminar flow) at a position x on the substrate is 1

η • x δ = a   ρ •υ

2  

where a = proportionality constant;η = viscosity of the gas; ν = velocity of the gas; andρ =density of the gas. The value ofρ depends on both the temperature and pressure while η and ν depend on the temperature. ρ=

M• p R•T

where M = molecular weight and p = total pressure;  η = ηo  T  To

  

m

where To = reference temperature, ηo = reference value,m = constant (0.6 < m < 1.0); and υ = υo T To where T0 = reference temperature, and ν0 = reference velocity. From the equations for ρ, η and ν, the pressure and temperature dependence of δ is expressed as δ = const • T p1/ 2

m/ 2

The pressure and temperature dependence of the diffusion coefficient D is

D = Di ,o •

Pi P

T •   To

  

1.75


433

where Di,o is the reference value of the diffusion coefficient andPi is the partial pressure of species i. The reaction resistance is then δ = const • P1/ 2 . −m/ 2 D T175 Hence δ/D increases with increasing pressure and decreasing temperature. The value of km follows the Arrhenius equation km = A • e

E − a RT

Thus the surface reaction resistance increases with decreasing temperature. This increase is more rapid than theδ/D increase with decreasing temperature. Hence surface reaction control can be reached at lower temperatures. Since the surface reaction control regime is normally the most attractive experimental conditions to reach (highest throwing power) this regime should be chosen. From Arrhenius plots (logarithm of the deposition rate versus the reciprocal temperature) conditions of surface kinetics control can be identified. For a surface kinetics control, the slope of the Arrhenius plot has a high negative value, often in the range 100 - 300 kJ mol-1. For mass transport control, the slope of the Arrhenius plot can either be positive (exothermic processes) or negative (endothermic processes) (Fig. 7.22). When the total pressure decreases, the diffusion rate of the species in the vapor increases, which means that surface kinetics control is readily achieved at low pressures. Figure 7.23 illustrates that the temperature region of surface kinetics control expands at lower pressures. Surface kinetics can also be attained by increasing the gas flow velocity (see Fig. 7.24). At low gas flow velocities, the thermodynamics control the deposition. Increasing the gas flow means entering the mass transport controlled regime. The surface kinetics control is reached at even higher gas flow velocities. The fourth possibility to reach the surface kinetically controlled region is to use another precursor with a higher thermochemical stability. As can be seen in Fig. 7.25, use of SiCl4 instead of SiH4 results in surface kinetics control at higher temperatures.

434


Figure 7.22. cesses.

Schematic Arrhenius plots for endothermic and exothermic pro-


435

Figure 7.23. Regions of mass transport and surface kinetics control at different total pressures (P1

Figure 7.24. Influence of gas flow velocity on the control of a CVD process.

436


Figure 7.25. Influence of the thermochemical stability of the precursor on the process control at silicon CVD.

6.0

REACTION MECHANISMS

The reaction mechanisms in CVD processes are very complicated and only a few are known. In the deposition process, the reactants are transported to the substrate surface. Molecules and/or atoms are adsorbed on specific surface sites. After surface diffusion, the molecules/atoms are incorporated in a step and finally, after diffusion along the step, incorporation in a stable crystallographic site takes place. The investigation by Bloem and Claassen[44] of the rate-determining reactions in CVD of silicon from SiH2Cl2 in the temperature range 800 - 1000°C is a good illustration of the various steps in a CVD process. A list of the reactions considered is given below. 1. Transport of SiH2Cl2 across the boundary layer: SiH2Cl2(b) → SiH2Cl2(g) where (b) and (g) refer to the bulk gas and the gas near the substrate surface, respectively.

Chemical Vapor Deposition 2. Homogeneous reactions in the vapor: SiH2Cl2(g) → SiCl2(g) + H2(g) SiCl2(g) + HCl(g) → SiHCl3(g) 3. Adsorption at free surface sites *. SiH2Cl2(g) + * → SiH2Cl2* SiCl2(g) + * → SiCl 2* HCl(g) + * → Cl* + l/2 H 2(g) l/2 H2(g) + * → H* 4. Surface reactions: SiH2Cl2* → Si* + 2 HCl(g) SiH2Cl2* → SiCl2* + H2(g) SiCl2* + H2(g) → Si* + 2 HCl(g) SiCl2* + HCl(g) → SiHCl3(g) SiCl2* + SiCl2(g) → SiCl 4(g) + Si(cryst) Si(cryst) means a stable crystallographic site in the crystal grown. 5. Growth reactions. Surface step sites are denoted (st). Si* → Si(st) Si(st) → Si(cryst) SiCl2* → SiCl2(st) SiCl2(st) + H2(g) → Si(cryst) + 2 HCl(g) In CVD of silicon from SiH2Cl2, the last reaction given was the rate-determining.

437

438 7.0

Deposition Technologies for Films and Coatings NUCLEATION

Since the properties of a material are influenced by grain size, defects, inclusions, etc., the nucleation is the most important process in the deposition of materials. At the initial stages of growth, the nucleation on the foreign substrate determines the grain size in the “first layer,” the defects in it and, to a large extent, the adhesion. In the subsequent growth secondary nucleation may occur with a generation of new grains, defects, inclusion of vapor species in pores, etc.. The various steps during the heterogeneous nucleation of an element A on a foreign substrate is schematically shown in Fig. 7.26. Hydrogen and AX react. The A atoms formed are adsorbed on the surface of the substrate. Subsequently the adsorbed atoms may desorb from the substrate, diffuse into the substrate, possibly with the formation of intermediate phases, or react with HX with the formation of AX. Unstable aggregates of A atoms, embryos, are formed after surface diffusion and direct impingement of A atoms from the vapor. Some of these embryos will grow at the expense of others and attain the status of stable A nuclei (supercritical A nuclei). An intact layer is formed after lateral growth and coalescence. The growth rate of the nuclei is determined by the concentration of the adatoms. Finally the coalescence generates, in general, defects, i.e., grain boundaries. Usually three-dimensional nucleation occurs on foreign substrates. However when nucleation takes place on native substrates—nucleus and substrate of the same material—two-dimensional nuclei may be formed. To describe 2-D nucleation the TLK model (Terrace,Ledge,Kink) of a surface is applied (Fig. 7.27). Besides the terraces, ledges, and kinks, atoms adsorbed on the surface—adatoms—exist. The deviation from the equilibrium concentration of the adatoms is a measure on the driving force of the growth process (positive deviation) or of the etching process (negative deviation). Surfaces grow by incorporating surface-diffusing atoms into the steps. This corresponds to a lateral movement of the steps. The probability of generating new nuclei between the surface steps depends on the surface diffusion and the deposition rate (the impingement flux of the atoms). At a high temperature and a low deposition rate, the adatoms have time enough for diffusion to reach the surface steps and be captured by them. A lower temperature and/or higher deposition rate results in shorter diffusion distances facilitating clustering of adatoms between the


439

steps (two-dimensional nucleation). At even lower temperatures and/or higher deposition rates—shorter diffusion distance—amorphous growth is obtained.[45] Finally defects are introduced into the layers when advancing steps meet each other or nuclei.

Figure 7.26. Schematic representation of nucleation of A on a substrate during hydrogen reduction of AX (a), and various mechanistic pathways that can be followed by A (b).

440


Figure 7.27. The Terrace, Ledge, and Kink (TLK) model of a surface.

The surface diffusion is strongly affected by the access to free surface sites. In a CVD process, it is likely that most of the surface sites are occupied by strongly adsorbed molecules. During the CVD of, for instance, silicon from Si-H-Cl gas mixtures, the surface sites are occupied to about 99% by hydrogen and chlorine atoms.[46] Moreover impurity adsorption on surface steps can effectively prevent diffusing adatoms to be captured. This means that a supersaturation high enough for nucleation can be built up between surface steps.[47] In summary, layer growth (no nucleation) can only be expected at high temperatures, low deposition rates, and low adsorption. This means long diffusion distances and the free incorporation of diffusing adatoms at the steps. After an incubation time, the nucleation rate is frequently high (~1010 -2 cm •s-1). A saturation value of the nucleus density, which remains constant during a relatively long period of time, is achieved (see Fig. 7.28).[48] The saturation value is obtained at a stage when the nuclei are so dense that a supersaturation high enough for nucleation can not be built up between the nuclei, i.e., when the mean diffusion distance is longer than half the mean nucleus distance. Subsequently the nuclei grow laterally and the nucleus density is constant until coalescence occurs. The saturation nucleus density, Ns, is strongly dependent on the experimental conditions. Figure 7.29 shows the influence of temperature on Ns for different silanes at siliconCVD.


441

Figure 7.28. The nucleus density as a function of process time.

Figure 7.29. Influence of temperature on the saturation nucleus density at silicon CVD from various silanes.[49]

442


Because of the high supersaturation in CVD, nuclei of critical size consist only of a few atoms. This means that the thermodynamical treatment of the nucleation on the basis of microscopic aggregates[50] is not justified. Instead statistical mechanical methods have to be applied.[51] The highest nucleation rate is attained at locations where the required supersaturation for nucleation is built up most rapidly. This is assumed to occur at sites of long residence times for the adatoms and/or at sites of high supply rate of the adatoms. Owing to the long residence time, nucleation on surface steps is highly probable at low deposition rates. Grain boundaries can be favorable diffusion paths, resulting in a high supply rate of adatoms and hence nucleation in the grain boundaries. The nucleation is strongly affected by the surface roughness. To illustrate this fact an example from an investigation of the preferential nucleation of boron on tungsten filaments is taken. The tungsten filaments used had a rough surface (Fig. 7.30a), which originates from the filament drawing process. The ridges of the filament serve as nucleation sites (Fig. 7.30b). The preferential nucleation on the ridges of the filament is explained as follows.[52] At the onset of the deposition reaction, boron atoms are added to the substrate surface. Simultaneously, boron is lost from the surface by diffusion into the substrate with the formation of tungsten borides. The diffusion flux, which initially is equal to the deposition rate, later decreases with increasing boride layer thickness (increased diffusion resistance). For geometrical reasons it is obvious that the thickness of the boride layer increases at a higher rate under a ridge than under a groove. Consequently the critical surface concentration for nucleation of boron is reached earlier on a ridge than in a groove.

8.0 SURFACE MORPHOLOGY AND MICROSTRUCTURE OF CVD MATERIALS The surface morphology and the microstructure of CVD materials are controlled by many factors that are often interrelated, such as the substrate, temperature, supersaturation, deposition rate, impurities, temperature gradients, and gas flows. In the following some theories and classifications of CVD morphologies and microstructures are introduced. Van den Brekel and Jansen have developed and applied a stability theory for single phase vapor growth. [53] If an arbitrary perturbation at the interface vapor/solid is reduced with increased time, the interface is


443

(a)

(b)

Figure 7.30. The surface of a tungsten filament (a), and preferentially nucleated boron on ridges of the filament (b).[52]

considered as stable. However, the interface during CVD in an isothermal environment is unstable. On the other hand, because of the fact that the relaxation times of films are much longer than the deposition times (a few minutes) smooth layers can be grown even in unstable processes. The instability of the interface in a vapor growth process can also be discussed in the same terms as those used to explain the dendritic growth from a melt in a negative temperature gradient. Random surface irregularities are frequently formed in growth processes. The surface irregularities have a higher rate of growth if they reach out into regions of higher supersaturation. In a CVD process, the surface irregularities have better

444


access to fresh reaction gas, which means a higher supersaturation and hence a higher deposition rate. Also, a negative temperature gradient, as in the cold wall reactor, may result in a higher supersaturation for the outgrowths. Blocher has related the various microstructures formed in CVD to the process conditions of temperature and supersaturation.[54] At a high temperature/low supersaturation (see Fig. 7.31) epitaxial growth occurs. Decreasing the temperature/increasing the supersaturation results in the formation of platelets, whiskers, etc. At a high supersaturation, a powder— due to the homogeneous nucleation in the vapor—is obtained. In the following, only comments on the growth of a few microstructures are given.

Figure 7.31. Microstructure sequence of CVD materials. [54]


445

Epitaxial growth, which is frequently used in the microelectronics industry, is obtained at relatively low growth rates. It is affected by the depositsubstrate crystallographic misfit, the substrate surface conditions, thermal stresses over the substrate and polycrystalline areas in the substrate. For epitaxial growth, a high surface mobility of adsorbed species is required, i.e., usually a high temperature has to be used. Columnar grains are common in CVD and exhibit a high degree of texture. During the primary nucleation, nuclei of different crystallographic orientations are formed. Depending on the anisotropy in the growth rate of various crystal surfaces, the nuclei will grow at different rates. This preferential growth results in a characteristic columnar growth. Numerous examples of columnar growth in CVD can be found inProceedings of the International CVD Conference series published by the Electrochemical Society. Surfaces grow by incorporating surface-diffusing adatoms into surface steps. However preferential adsorption of molecules at the surface steps prevents the surface-diffusing adatoms from being captured. Thus a new growth mechanism is required. Throughout the years, the structure of CVD materials has been modified by adding small amounts of foreign substances (growth modifiers) to the reaction gas mixture.

9.0

SELECTIVE DEPOSITION

Chemical vapor deposition (CVD) is known to be a large-area deposition technique. However, CVD can also be used for local deposition or selective deposition, i.e., the deposition occurs only on some areas of the substrate surface. Selectivity may be attained by using different focused beams (photons,[55] electrons,[56]-[58] or ions.[59][60] The beams induce local CVD reactions on those areas they hit. Another possibility is to irradiate the substrate surface through a mask with, for instance, a laser.[61] The openings in the mask define the substrate areas where the deposition may take place. Selective chemical vapor deposition may also be achieved on patterned substrates. The selectivity in this case is based on differences in the initial interfacial reactions between the different substrate materials and the vapor. The interfacial reactions on one substrate material should be inhibited completely to avoid nucleation, while the deposition reactions should be stimulated on those substrate areas where the deposition shall occur.

446


Principally there exist several categories of selective deposition systems. In the system described above, the deposition takes place on one substrate material while no deposition is obtained on the other and areaselective deposition is achieved. However, different phases can also be deposited simultaneously and selectively on the different materials, resulting in phase-selective deposition. In analogy to the phase-selective deposition, films of different microstructures or different chemical compositions may be deposited on the different substrate materials and hence selectivity in microstructure or chemical composition is attained. Selective deposition is an emerging field and there is a great demand for these processes in many application areas. With the steady reduction of IC feature sizes there is a need for self-aligned processes. Selective tungsten for metallization in VLSI and selective GaAs epitaxy for monolithic integration of optoelectronic devices are well-known examples. Other application areas may be in micronics, heterogeneous catalysis, engineering of film/substrate interfaces, and in growth of artificial 2-D and 3-D materials. Since selective deposition on patterned substrates is based on interfacial chemistry, there are practically no restrictions in the dimensions of the deposited materials islands. This opens a fascinating perspective to build up materials with microstructures without any thermodynamic or kinetics limitations. The underlying principles of selective deposition are briefly discussed below. 9.1

Area-Selective Growth

Epitaxial Growth Conditions. There is a considerable technological interest today in area-selective epitaxy of both silicon and gallium arsenide. A brief discussion of area-selective growth with reference to silicon and gallium arsenide, respectively is given below. Epitaxial films can be grown at relatively high rates near equilibrium conditions, i.e., at a low driving force (low supersaturation) of the deposition process. For heterogeneous nucleation, a higher supersaturation is generally required. This means that conditions of selective growth are prevailing at a supersaturation lower than that for heterogeneous nucleation. This was used by Joyce and Baldrey for growth of silicon from SiCl4 at 1200°C and atmospheric pressure in openings etched in the SiO2 mask.[62] A historical review of selective epitaxial growth, SEG, has recently been published by Borland.[63] In SEG, the growth is stopped when the surface of the growing film reaches the mask surface. A continued growth results in an overgrowth


447

over the mask. The process is then called ELO (Epitaxial Lateral Overgrowth). For a review of the ELO process, the reader is referred to Ref. 64. A key point in the SEG is the suppression of the nucleation on the mask (usually silicon oxide or silicon nitride). As mentioned in Sec. 7, the incubation time for nucleation varies with the substrate material and the deposition conditions. In an ideal case, this incubation time is longer than the deposition time required to prepare the desired structures. However, by using an alternating growth and etching process, SEG can be attained even for conditions of short incubation times for nucleation.[65] The growth conditions are then prevailing for about the incubation time. After the growth cycle, the process is switched over to etching with, for instance, HCl. A minor etch is also obtained of the monocrystalline silicon in the SiO2 openings. The GaAs SEG/ELO is nearly as old as the silicon SEG/ELO. Tausch and Lapierre reported in 1965 on a GaAs ELO process based on a chloride vapor transport system.[66] With the development of the purification techniques of metal-organic compounds like trimethyl gallium (TMG) and triethyl gallium (TEG), CVD as well as MBE, based on the use of these compounds together with AsH3, are highly attractive for GaAs SEG. MBE and elemental sources yields monocrystalline growth in etched openings as well as polycrystalline GaAs on the mask (microstructure-selective deposition).[70] GaAs SEG has received much attention during the last few years as a technique for achieving monolithic integration of electronic and optoelectronic devices. Growth of GaAs from AsH3 and TMG by MBE or CVD is usually considered to be a non-equilibrium process. The perfection of the crystals grown, their morphology, and the correlation between the growth rate and thermodynamic parameters indicates that near-equilibrium conditions exist at the interface between the vapor and the solid. Hence thermodynamics can be utilized to analyze selective growth as well heterogeneous nucleation conditions in GaAs CVD. According to nucleation theory, a certain supersaturation is needed for heterogeneous nucleation on the mask. From experimental selectivity data, the maximum supersaturation for maintaining selectivity can be calculated. The supersaturation is favorably expressed in terms of chemical potentials. The influence of temperature on the chemical potential of GaAs (expressed in elemental chemical potentials of Ga and As2 ) at equilibrium with solid GaAs is shown in Fig. 7.32. Growth will occur if the chemical potential of GaAs for the homogeneous equilibrium in the vapor is higher than that for the heterogeneous equilibrium.

448


Figure 7.32. Chemical potential of GaAs for the heterogeneous equilibrium (full line), and for two homogeneous equilibria at different total pressures (dashed lines). H2/AsH3/TMG = 500/10/1.[68]

The experimental technique used to determine, for instance, the temperature required to achieve SEG is to raise the temperature successively until no nucleation on the mask can be observed. Since the chemical potential of GaAs for the homogeneous equilibrium in the vapor has only a slight temperature dependence (see Fig. 7.2), the driving force for the deposition (or supersaturation) will decrease upon a temperature increase and a driving force value, yielding no heterogeneous nucleation, will be reached. Thermodynamics and MBE and CVD experimental SEG data were used in an effort to put experimental selectivity observations on a common basis.[68][69] In MBE a much lower pressure is used than in CVD. However, irrespective of the growth technique used, the experimental SEG data fall in the supersaturation region indicated in Fig. 7.33. By using thermodynamics, the selectivity data from CVD can be converted to MBE and vice versa.


449

Figure 7.33. Selective growth regime for GaAs. Precursors: Ga(CH3)3, AsH 3.[69]

Substrate-Activated Selective Growth. When a substrate of different materials is exposed to the vapor in a CVD process, the materials represent areas of different activities or reactivities towards the vapor. One material may, for instance, act as an effective reducing agent or as a catalyst of dissociative adsorption of gaseous reactants, which may favor the deposition. The other material may be relatively inert towards the vapor and growth may be inhibited. The inertness may be increased purposely by using gas additives which are preferentially adsorbed to one of the materials. Strongly adsorbed molecules may passivate a substrate surface considerably and suppress the deposition process completely. A tendency to substrateactivated area-selective growth is frequently seen during the initial growth stage in CVD on polycrystalline, multi-phase substrates. The different phases and the different crystallographic orientations of the grains exposed to the vapor, represent surface areas of different activities/reactivities and initial growth conditions. Taken to its extreme, this means that the deposition is inhibited on some substrate areas, while other areas are open for deposition. Area-selective deposition of refractory metals is of highest interest for metallization in VLSI and ULSI. Selective deposition of refractory metals for metallization has been reviewed by several authors (see Refs. 70 - 73). The

450


substrate-activated area-selective growth is well illustrated by the selective tungsten deposition from WF6 and H 2 on Si/SiO2 substrates. This process is described principally below. Tungsten can be deposited by CVD at low temperatures (300°C) from H2 and WF6 according to the reaction 3 H2(g) + WF6 (g) → W(s) + 6 HF(g) The deposition occurs on all substrate surfaces exposed to the vapor, since both the source material (WF6) and the reducing agent (H2) are gases. However, if the reducing agent was replaced by a solid reducing agent (like elemental silicon), the deposition should only occur on those substrate regions having a reducing agent. So if a wafer exposes areas of elemental silicon and silicon dioxide to WF6, tungsten deposition takes place only on the silicon areas and not on the adjacent silicon dioxide areas. Silicon in silicon dioxide can not act as a reducing agent, since this silicon has its maximum oxidation number. This is the basis of the initial stage of selective tungsten CVD. The selective tungsten deposition may proceed according to the scheme described above as long as elemental silicon is exposed to the vapor. After a while, however, the tungsten deposited onto elemental silicon will separate silicon from the vapor: hence a self-limiting growth process has been obtained. The mechanism of self-limitation is under discussion and might be also be due to a polymerization reaction involving lower tungsten fluorides.[74] The polymer formed may also separate silicon from the vapor, hence inhibiting the growth process. For growth of thicker tungsten layers, a reducing agent, H2 , has to be added to the reaction gas. If the WF6 concentration is low and the H2 concentration is high, i.e., conditions of low supersaturation are prevailing, deposition of tungsten will occur where tungsten already has been deposited (on elemental silicon) and not on the silicon dioxide. For deposition on the silicon dioxide, resulting in a loss of selectivity, tungsten nucleation must take place. The nucleation step requires a much higher supersaturation than growth. Hence a deposition window, ranging from the supersaturation corresponding to equilibrium conditions up to the supersaturation value needed for heterogeneous tungsten nucleation on silicon dioxide, exists. Finally, the selective deposition of tungsten to substrate areas where tungsten already has been deposited, is favored by the dissociation of hydrogen molecules on these areas.[75]


451

In summary, two main reaction steps can be distinguished in tungsten CVD (see Fig. 7.34): 1. In the first step, elemental silicon will act as the predominating reducing agent even if a large amount of hydrogen is used in the reaction gas. This results in tungsten deposition on those substrate regions where elemental silicon is exposed to the vapor. The reaction step includes an etching of elemental silicon, i.e., silicon is consumed. 2 WF6(g) + 3 Si(s) → 2 W(s) + 3 SiF4(g) Considering the stoichiometry of this reaction, about 200 Å silicon is consumed for 100 Å tungsten deposited. The topography of the Si/W interface is affected by this reaction. The etching and hence the topography can be reduced by, for instance, addition of SiF4 to the reaction gas mixture.[76] 2. In the second step another reducing agent, H2, has to take over, since the tungsten film, and probably the tungsten fluoro polymer obtained, prevent the reactions between the vapor and the silicon.

Figure 7.34. The two reaction steps in selective tungsten CVD.

452


The chemical reactions in the first step are usually extremely fast and a thermodynamically controlled CVD process is obtained. In the second process step, the deposition process was operated at a low supersaturation to avoid nucleation on the mask material (SiO2 ). The growth conditions in the second step are close to those existing in the area-selective epitaxy discussed above and can be analyzed from thermodynamics. Thermodynamics has been used as a guide for prediction of trends in selectivity and substrate etching when the deposition conditions are changed. It has also been used for identification of plausible (and often undesired) side reactions as well as of gaseous selectivity modifiers, improving selectivity.[9] Adsorption-Induced Selective Growth. As discussed above, the heterogeneous nucleation on one of the substrate materials must be suppressed during a relatively long time in an ideal selective growth system. The incubation time for nucleation is influenced by many factors: temperature, substrate reactivity, adsorption, etc. Adsorbed molecules may reduce the rate of surface reactions and, in extremes, inhibit the nucleation completely. A concept of strongly adsorbed molecules to one of the substrate materials was used to achieve area-selective growth of boron carbide on a patterned substrate exposing areas of titanium and molybdenum to the vapor.[77] Boron trichloride, ethylene, and hydrogen were used as reactants and the deposition temperature was 1400 K. The ethylene molecules (or fragments of them) were preferentially and strongly adsorbed on molybdenum and no nucleation of boron carbide was observed. On titanium, however, fast nucleation kinetics was obtained. The deposition was located only to those substrate areas having titanium. The boron carbide was amorphous and contained about 21 at% carbon. This illustrates that adsorbed molecules may act as masks and can be used to inhibit the deposition on desired substrate areas. 9.2

Phase-Selective Deposition

A new dimension in the field of selective growth is created in phaseselective growth. Phase-selective growth means that several phases are selectively and simultaneously deposited on desired substrate materials/ areas. This might result in growth of, for instance, a semiconductor together with an insulator, i.e., selectivity in properties is also obtained. Phaseselective deposition may be achieved in different ways. In this chapter, two principles of phase-selective growth are discussed: phase-selective deposition attained by differential nucleation behavior and by secondary processes in or on the growing film, respectively.


453

Phase-Selective Deposition by Differential Nucleation Behavior. The initial substrate/vapor reactions and the nucleation kinetics are usually dependent on the substrate material. This may result in nucleation of different phases on different substrate materials. Provided that no secondary processes like phase-transformation in the solid state occur in the film or that no new phase is nucleated on top of the growing film, the originally nucleated phases will continue to grow and a phase-selective deposition is obtained. This principle was used for phase selective-growth of two boron carbides: T1-BCx and B13C2. The substrate used was that obtained after the area-selective growth of boron carbide described above, i.e., the substrate exposes molybdenum and amorphous boron carbide to the vapor. The vapor contained boron trichloride, methane, and hydrogen and the growth temperature was 1400 K. T1-BCx was obtained on the amorphous boron carbide while B13C2 was grown on molybdenum. This phase-selective growth was attributed to differential nucleation kinetics since no secondary processes were observed in or on the films.[78] Phase-Selective Deposition Achieved by using Secondary Processes. Elemental boron has several crystalline polymorphs, and in addition to that, amorphous boron also exists. Phase-selective growth was studied in this system by using the Ti/Mo patterned substrates described above. Fast nucleation was observed on both Ti and Mo. Amorphous boron was obtained on Ti, while a-rhombohedral boron was grown on Mo. The boron grown on Ti contained a small amount of Ti (about 500 ppm) throughout the layers, while no traces of Mo was detected in a-B (detection limit 1 ppm). The Ti stabilizes the originally nucleated amorphous boron.[79] To obtain crystalline boron, a phase-transformation in the solid state is needed and such a transformation can be assumed to have a high energy barrier to overcome. Amorphous boron may also be obtained initially on Mo. However, an immediate phase transformation is expected because of the deposition temperature used. Moreover, the film did not contain any substrate contaminants contributing to a stabilization of the amorphous boron. The morphology of the phaseselectively deposited boron is shown in Fig. 7.35. The amorphous boron is characterized by the rounded nodules.

10.0 SOME APPLICATIONS OF THE CVD TECHNIQUE The CVD technique is known for its versatility in producing materials of greatly varying properties. This is illustrated by the examples given in the application list below.

454


Figure 7.35. Phase-selective growth of amorphous boron (rounded nodules) and a-rhombohedral boron on a substrate exposing titanium and molybdenum.[79]

! Microelectronics industries use CVD for growth of epitaxial layers (vapor phase epitaxy, VPE) and for making films serving as dielectrics, conductors, passivation layers, oxidation barriers, etc. An emerging field is selective deposition of refractory metals and silicides for metallization in VLSI. ! Semiconductor lasers of GaAs/(Ga,Al)As and InP/(In,Ga)As. These substances are also used in microwave devices and solar cells. ! Optical fibers for telecommunication. Optical fibers are produced by coating the inside of a fused silica tube with oxides of silicon, germanium, boron, etc., for obtaining the correct refractive index profile. After the deposition the fused silica tube is collapsed to a rod and the rod is then drawn into a fiber. ! Solar energy conversion by the utilization of selective absorbers and of dry solar cells of silicon and gallium arsenide. ! Wear resistant coatings have wide industrial applications. Coatings of TiC, TiN and Al2O3 on cemented carbide cuttingtool inserts and of TiC on steels (punches, nozzles, free wheels, etc.) are well-known.


455

! Friction reducing coatings for use in sliding and rolling contacts, for example. ! Corrosion resistant coatings (Ta, Nb, Cr, etc.). ! Erosion resistant coatings (TiC, Cr7C3, B 4C, etc.). ! Heat-resistant coatings (Al2O3, SiC, Si 3N4, etc.). ! Fibers for use in fiber-reinforced materials (fibers of boron, silicon carbide, boron carbide, etc.). ! Structural shapes (tubes, crucibles, heating elements, etc.) of, for instance, tungsten and silicon carbide. ! Decorative coatings of, for instance, TiN (gold color) on watches.

11.0 OUTLOOK CVD offers many advantages in thin film deposition. With the use of new precursors, the deposition temperature can usually be lowered considerably. By lowering the total pressure, extremely sharp interfaces with respect to chemical composition and topography can be obtained. The atmospheric pressure CVD is attractive in many applications with its high deposition rates and hence short process times. Since CVD processes are based on interfacial chemistry, they are sensitive to contamination and load-lock systems must be used to keep the contamination level low. The selective deposition opens fascinating prospects for the future, not only for microelectronic applications but also for materials science in general, and for engineering of interfaces and artificial materials.

456


REFERENCES 1.

Lindström, J. N. and Schachner, H.,Proc. of the 3rd European Conf. on CVD, (H. E. Hintermann, ed.), p. 208 (1980)

2.

Jansson, U., Carlsson, J. O., Stridh, B., Söderberg, S., and Olsson, M., Thin Solid Films, 172:81 (1989)

3.

Claassen, W. A. P. and Bloem, J., J. Cryst. Growth, 51:443 (1981)

4.

Zeleznik, F. J. and Goedon, S., Ind. Eng. Chem., 60:27 (1968)

5.

van Zeggeren, F. and Storey, S. H., The Computation of Chemical Equilibria, Cambridge University, Cambridge (1970)

6.

Smith, W. R., Ind. Eng. Chem. Fundam., 19:1 (1980)

7.

Smith, W. R. and Missen, R. W., Chemical Reaction Analysis, Wiley Interscience, New York (1982)

8.

Eriksson, G., Acta Chem. Scand., 25:2651 (1971)

9.

Carlsson, J. O. and Hårsta, A., Thin Solid Films, 158:107 (1988)

10.

Spear, K. E., Proc. of the 7th Int. Conference on CVD, (T. O. Sedgwick and H. Lydtin, eds.), p. 6, The Electrochemical Society, Pennington, NJ (1979)

11.

Ottoson, M., Hårsta, A., and Carlsson, J. O., J. Electrochem. Soc., 138(3):778 (1991)

12.

Stull, D. R., et al., JANAF Thermochemical Tables, Nat. Stand. Ref. Data Ser., NBS, 37 (1971)

13.

Barin, I. and Knacke, O., Thermochemical Properties of Inorganic Substances, Springer, Berlin (1973)

14.

Kubaschewski, O., Evans, E. Ll., and Alcock, C. B., Metallurgical Thermochemistry, Pergamon, Oxford (1967)

15.

Ducarroir, M. and Bernard, C., J. Electrochem. Soc., 123:136 (1976)

16.

Bessman, T. M. and Spear, K. E., ibid., 124:786 (1977)

17.

Bernard, C., Proc. of the 8th Int'l. Conf. on CVD, (J. M. Blocher, Jr., G. E. Vuillard, and G. Wahl, eds.), p. 3, The Electrochemical Society, Pennington, NJ (1981)

18.

Colmet, R., Naslain, R., Hagenmuller, P., and Bernard, C., ibid., p. 17 (1981)

19.

Vandenbulcke, L., ibid., p. 32 (1981)

20.

Wang, C. F. and Spear, K. E., ibid., p. 54 (1981)

21.

Boman, M. and Carlsson, J. O., Surf. Technol, 24:173 (1985)


457

22.

Boman, M. and Carlsson, J. O., in Proc. of the 9th Int'l. Conf. on CVD, (M. Robinsson, C. H. J. van den Brekel, G. W. Cullen, J. M. Blocher, Jr., and P. Rai-Choudhury, eds.), p. 150, The Electrochemical Society, Pennington, NJ (1984)

23.

Spear, K. E., ibid., p. 81 (1984)

24.

Carlsson, J. O., J. Vac. Sci. Technol., A6:1656 (1988)

25.

Powell, C. F., Oxley, J. H., and Blocher, J. M., Jr., Vapor Deposition, John Wiley & Sons, Inc., New York (1966)

26.

Yee, K. K., Int. Met. Reviews, 1:19 (1978)

27.

Perry, A. J., Beguin, C., and Hintermann, H. E., in Proc. of the 3rd European Conf. on CVD, (H. E. Hintermann, ed.), p. 131 (1980)

28.

Ban, V. S. and Gilbert, S. L., J. Cryst. Growth, 31:284 (1975)

29.

Sedgwick, T. O., Smith, J. E., Ghez, Rh., and Cowher, M. E., J. Cryst. Growth, 31:264 (1975)

30.

J. Nishizawa, and M., Saito,in Proc. of the 8th Int'l. Conf. on CVD, (J. M. Blocher, Jr., G. E. Vuillard, and G. Wahl, eds.), p. 113, The Electrochemical Society, Pennington, NJ (1981)

31.

Bird, R. B., Stewart, W. E., and Lightfoot, E. N., Transport Phenomena, John Wiley & Sons, Inc., New York (1960)

32.

Westphal, G., J. Cryst. Growth, 65:105 (1983)

33.

Jensen, K. F., Proc. of the 9th Int. Conf. on CVD, (M. Robinsson, C. H. J. van den Brekel, G. W. Cullen, J. M. Blocher, Jr., and P. RaiChoudhury, eds.), p. 3, The Electrochemical Society, Pennington, NJ (1984)

34.

Wahl, G., Thin Solid Films, 40:13 (1977)

35.

Wahl, G. and Hoffman, R., Rev. Int. Hautes Temp. Refract., 17:7 (1980)

36.

Schlichting, H., Boundary Layer Theory, 6th ed., McGraw-Hill, New York (1968)

37.

Eversteijn, F. C., Severin, P. J. W., van den Brekel, C. H. J., and Peek, H. L., J. Electrochem. Soc., 117:925 (1970)

38.

Talbot, L., Cheng, R. K., Schefer, R. W., and Willis, D. R., J. Fluid Mech., 101:737 (1980)

39.

Giling, L. J., J. Electrochem. Soc., 129:634 (1982)

40.

Hwang, G. J. and Cheng, K. C., J. Heat Transfer, Trans ASME, 95:72 (1973)

458


41.

Coltrin, M. E., Kee, R. J., and Miller, J. A., Proc. of the 9th Int'l. Conf. on CVD, (M. Robinsson, C. H. J. van den Brekel, G. W. Cullen, J. M. Blocher, Jr., and P. Rai-Choudhury, eds.), p. 31, The Electrochemical Society, Pennington, NJ (1984)

42.

Rosenberger, F.,Fundamentals of Crystal Growth, Springer, New York (1979)

43.

Carlsson, J. O., J. Less-Common Metals, 70:97 (1980)

44.

Claassen, W. A. P. and Bloem, J., J. Electrochem. Soc., 50, 807, (1980)

45.

Venables, J. A. and Price, C. R., Epitaxial Growth, (J. W. Matthews, ed.), p. 381, Academic Press, New York (1975)

46.

Chernov, A. A., J. Cryst Growth, 42:55 (1977)

47.

Carlsson, J. O., Proc. of the 3rd European CVD Conference, (H. E. Hintermann, ed.), p. 10 (1980)

48.

Venables, J. A., Phil. Mag., 27:698 (1973)

49.

Claassen, W. A. P. and Bloem, J., J. Electrochem. Soc., 127:1836 (1980)

50.

Hirth, J. P. and Pound, G. H.,Condensation and Evaporation, MacMillan, New York (1963)

51.

Walton, D., J. Chem. Phys., 37:2182 (1962)

52.

Vega-Boggio, J., Vingsbo, O., and Carlsson, J. O., J. Mater. Sci., 12:1750 (1977)

53.

van den Brekel, C. H. J. and Jansen, A. K., J. Cryst. Growth, 43:364 (1978)

54.

Blocher, Jr., J. M., J. Vac. Sci. Technol, 11:680 (1974)

55.

Eherlich, D. J. and Tsao, J. Y., ibid., B1:969 (1983)

56.

Chin, B. H. and Ehrlich, G., Appl. Phys. Lett., 15:253 (1981)

57.

Matsui, S. and Mori, K., J. Vac. Sci. Technol., B4:299 (1986)

58.

Kunz, R. R., Allen, T. E., and Mayer, T. M., ibid., B5:1427 (1987)

59.

Melngailis, J., ibid., B5:469 (1987)

60.

Kato, T., Morimoto, H., Saitoh, K., and Nakata, H., ibid., B3:50 (1985)

61.

Higashi, G. S., Blonder, G. E., and Fleming, C. G.,“Photon, beam, and plasma stimulated chemical processes at surfaces”, (V. M. Donelly, I. P. Herman, and M. Hirose, eds.), 75:117, Materials Research Society, Pittsburgh, PA (1987)

62.

Joyce, B. D. and Baldrey, J. A., Nature, 195:485 (1962)


459

63.

Borland, J. O., Proc. of the 10th Int'l. Conf. on CVD, (G. W. Cullen and J. M. Blocher, Jr., eds.), p. 307, The Electrochem. Soc., Pennington, NJ (1987)

64.

Jastrzebski, L., J. Cryst. Growth, 63:493 (1983)

65.

Jastrzebski, L., Corboy, J. F., McGinn, J., Pagliaro, R., Jr., J. Electrochem Soc., 130:1571 (1983)

66.

Tausch, F. W. Jr. and Lapierre, A. G., III, ibid., 112:706 (1965)

67.

Furumura, Y., Mieno, F., Nishizawa, T., Maeda, M., ibid., 133:379 (1986)

68.

Carlsson, J. O., Proc. of 13th Nordic Semiconductor Meeting, (M. Östling, ed.), p. 298, The Royal Institute of Technology, Stockholm (1988)

69.

Carlsson, J. O., Unpublished

70.

Green, M. L., Borland, J. O., Proc. of the 10th Int'l. Conf. on CVD, (G. W. Cullen and J. M. Blocher, Jr., eds.), p. 603, The Electrochem. Soc., Pennington, NJ (1987)

71.

Lami, P. and Pauleau, Y., J. Electrochem. Soc., 135:980 (1988)

72.

Pauleau, Y., Lami., P., Tissier, A., Panyel, R., and Oberlin, J. C., Thin Solid Films, 143:259 (1986)

73.

Green, M. L., Ali, Y. S., Boone, T., Davidsson, B. A., Feldman, L. C., and Nakahara, S., J. Electrochem. Soc., 134:2285 (1987)

74.

Lifshitz, N., Appl. Phys. Lett., 51:967 (1987)

75.

McConica, C. M. and Krisnamani, K., J. Electrochem. Soc., 133:2542 (1986)

76.

Levy, R. A., Greene, M. L., Gallagher, P. K., and Ali, Y. S., J. Electrochem. Soc., 133:1905 (1986)

77.

Jansson, U. and Carlsson, J. O.,J. Vac. Sci. Technol., A6:1733 (1988)

78.

Jansson, U., Carlsson, J. O., Markert, L. M., and Greene, J. E., Submitted to J. Vac. Sci. Technol., A7(6):3172 (1989)

79.

Jansson, U., Carlsson, J. O., Markert, L. M., and Greene, J. E.,J. Vac. Sci. Technol., A9(2):266 (1991)

8 Plasma-Enhanced Chemical Vapor Deposition Arthur Sherman

1.0 INTRODUCTION Chemical vapor deposition is the process of thin film formation on a solid surface by the heterogeneous reaction between a reacting gas and a hot surface.[1] In some instances, the temperatures necessary to achieve acceptable deposition rates can be too high to be useful. For example, titanium nitride deposition at acceptable rates from a gas mixture of TiCl4 , N2 and H2 requires temperatures on the order of 1000ºC. This is a disadvantage when the film is being used to provide a hard surface on tool steel, since this temperature is higher than the steel’s softening temperature, and control of critical dimensions cannot be maintained. Similarly, silicon nitride is an excellent passivation and barrier layer for integrated circuits. However, the process of deposition from SiCl2H2 and NH3, again at acceptable rates, requires temperatures as high as 900ºC. This clearly cannot be used over aluminum metallization with a melting point of 660ºC. Finally, it is also important to lower deposition temperatures because of the reduction in critical dimensions in VLSI integrated circuits (very large-scale ICs). In this instance, diffused layers become quite thin (1000 - 2000 Å), and they cannot be maintained when the wafer has to be heated to a temperature which is too high.

460


461

The requirement for lower deposition temperatures can sometimes be met by using different gaseous reactants such as TiCl4 and NH3 to deposit titanium nitride at ~600ºC.[2] However, the number of such choices is limited. An alternative would be to create an electrical discharge in the reactant gases.[3] This will produce a significant number of free radicals (i.e., SiH4 → SiH2, SiH and NH3 → NH, NH2, etc.), and these will be much more reactive at lower surface temperature. All plasma-enhanced CVD reactors operate with low pressure discharges (glow discharges) sustained by RF. This is the preferred approach since a high-volume, uniform discharge with high electron energy can be created that can dissociate a significant fraction of the reactive gases. Since reactor walls will be in contact with the plasma, it will be necessary to evaluate their influence on the discharge behavior. In addition to the deposition of PECVD silicon nitride at moderate temperatures (~350ºC), many other films can be deposited by this technique. Silicon dioxide is often deposited by PECVD, in spite of the fact that it can be deposited thermally at 400ºC. This points to the other reason for using PECVD—that it is possible to vary the stoichiometry of the resulting film. This is much more difficult with a strictly thermal process. Similarly, PECVD amorphous silicon has a substantial amount of hydrogen in the film (30 - 40 %), and as a result, it can be used to fabricate solar cells. Other materials that can be deposited by this technique include refractory metals, refractory metal silicides, and aluminum.

2.0 REACTOR INFLUENCE ON PLASMA BEHAVIOR Since we are always dealing with plasmas confined within a reactor chamber, we must study the nature of plasma-surface interactions. Initially we discuss the characteristics of AC and DC discharges. Then what happens when the two electrodes are of unequal size is covered. Finally, we analyze the effects of different frequencies on the operation of an AC discharge. 2.1 DC/AC Glow Discharges Using an applied DC voltage, a nonuniform glow discharge is created in a low pressure gas (~1 Torr), as shown in Fig. 8.1. [4] A sheath is formed

462


next to the cathode where few collisions occur and charge neutrality is not obeyed. In this region, ions are accelerated toward the cold cathode, and upon striking emit secondary electrons. These secondary electrons sustain the discharge. An alternative would be to use a hot cathode which serves as a thermionic emitter. The sheath includes the Aston, Crookes and Faraday dark spaces as well as the cathode and negative glows. This region has a net positive charge because of the excess of ions there. The positive column shows no net space charge, so this is the true plasma. It’s electrical resistivity is low, so only a weak electrical field is necessary to establish a current flow. Ions and electrons recombine to neutral atoms in this region, either by gas phase recombination or diffusion to the tube walls. They are regenerated by electron impact ionization. A potential difference is established between the positive column and the tube wall, because the highly mobile electrons tend to flow rapidly out to any surface, while the heavy ions remain immobile. This creates a negative potential on the wall which hinders further electron outflows. A sheath forms next to the wall which has a deficit of electrons and therefore a positive net charge. The ions in the plasma see the negative wall potential, however, and are attracted to the wall. This ion diffusion to the walls is referred to as ambipolar diffusion. When the glow discharge of Fig. 8.1 is operated under AC conditions (below 10 kHz), two dark spaces are observed. We have essentially created DC discharges of alternating polarity, since there is time between cycles for the discharge to extinguish at low frequencies. Depending on geometry and gas, the starting of an AC discharge depends on frequency and pressure.[5] Finally, when an AC discharge is set up with a blocking capacitor between the power supply and one of the electrodes, that electrode has a negative self-bias. This causes ions to accelerate toward this electrode at high energies, and plasma etching for ICs uses this phenomena to create anisotropic etches.[6] Figure 8.2 provides an excellent explanation of why a negative self-bias forms.[7] In the figure on the left, a conducting probe is placed in a plasma where a positive voltage causes a large current flow. A negative voltage produces only a small current because the ions cannot move readily. This produces a zero self-bias. When the probe is attached to a capacitor, the figure on the right shows what happens. The average applied voltage must be negative to satisfy the condition of no average current flow. Therefore, a negative bias forms.


Figure 8.1. A DC glow discharge at low pressure.[4]

463

464


Figure 8.2. Creation of negative self-bias in AC discharge.[7]

2.2 AC Discharges with Unequal Area Electrodes Next, we can evaluate the situation that occurs when one electrode is larger or smaller than the other, as shown in Fig. 8.3. If it is assumed that the ion current density to all internal surfaces is the same, then a relationship between V1/V2 and A 1/A2 can be derived.[8]

Eq. (1)

V1  A2   = V2  A1 

4

Some experimental work has been done to examine the validity of Eq. (1).[9] Here peak-to-peak and AC bias voltages were measured in a 13.56-MHz glow discharge. Results showed that voltage ratios depended on the electrode area ratios as well as the electrode material, the gas, the pressure and the peakto-peak voltages. Some of these results are shown in Fig. 8.4 for argon in a stainless steel chamber operated at 50 mTorr with a peak-to-peak voltage of 600 V. Under these conditions, V1/V2 = (A1/A2)n is a fair representation of the data, but n≅ 4 for only 0.6 < A1 /A2 < 1.0 Smaller ratios give n ≅ 1. From these results, we conclude that the DC bias that will appear on the electrode carrying wafers will depend on the electrode configuration. It can also be a function of the frequency used, as discussed next.


Figure 8.3. Reactor with unequal size electrodes.

Figure 8.4. Voltage ratio versus area ratio for argon plasma.[9]

465

466


2.3 Frequency Effects on RF Plasma Reactor Behavior The prevailing wisdom is that lower frequency discharges (i.e., 50 kHz) will yield films with greater compressive stress because ion bombardment is more intense. Recent experiments have confirmed this effect.[10] They were done in a parallel-plate reactor configuration with one electrode having a hole in it. Then ions passing through this hole were electrostatically retarded, and the cutoff voltage measured. This voltage as a function of power level and frequency is shown in Fig. 8.5.[10]

Figure 8.5. Cl2 plasma beam maximum ion energy. Circles = 27 MHz; triangles = 100 kHz; solid = Cl2*; open = Cl*.[10] (Reprinted by permission of the publisher, The Electrochemical Society, Inc.)


467

One explanation of why the lower frequency produces more intense ion bombardment is that the sheath potential is higher. Then, since the electrons are lighter, they diffuse out of the plasma first and the electrode assumes a negative bias. For higher frequencies, however, there is less time for electrons to diffuse out between cycles, so there is less need for a strong negative bias to form. A weaker negative bias means lower energy ions at the electrode surface. Alternately, we recognize that the plasma potential varies with time.[12] When an ion can cross the entire sheath before a reversal of the applied field, it sees the maximum sheath potential. For higher frequencies, the ion does not make it all the way across before the applied field reverses, so it sees only the average sheath potential or 1/3 the maximum. 2.4 Adjusting DC Bias for Fixed Electrode Geometry Since changing the electrode geometry of a PECVD reactor is a major hardware modification, it would be desirable to change the DC bias on the electrode holding the wafers by altering the RF power to the system. This can be done either by inserting a variable LC circuit between this electrode and ground[13] or by powering each electrode with separate power supplies.[14] This latter arrangement is shown in Fig. 8.6. By using a low frequency (50 400 kHz) on the lower electrode and a high frequency on the upper one (13.56 MHz), a stable discharge is created along with the ability to control the DC bias on the lower electrode. As discussed in Sec 3.1, this arrangement permits control of film stress, density, step coverage and stoichiometry. 2.5 Plasma-Enhanced CVD (PECVD) Reactors There are three well-known methods of creating plasmas for thin film deposition. In one, a pair of electrodes are placed in a low pressure gas, and either an AC or a DC voltage applied to create a glow discharge. If the film being deposited is a nonconductor, only AC will work, so it is generally used. A second approach uses a coil wound around a nonconducting tube containing the gas. Then an AC field excites strong fields inside the tube and a discharge can be created. Finally, a pair of electrodes can be placed, one on each side of a dielectric tube filled with gas, and again an AC voltage applied. This also produces a strong field inside the tube, and a discharge can be created.

468


Figure 8.6. Dual frequency RF configuration.[14]

Cold-Wall, Parallel-Plate PECVD Reactors. The original reactor of this type is shown in Fig. 8.7. It has circular symmetry and the wafers sit on a heated platen. Reactants are introduced at the outer edge and exhausted down the center. The theory was that there would be a stronger discharge in the center, and that this would be offset by a shorter residence time as gases flowed over the platen, leading to a more uniform deposition. Another version that introduced gases at the center was developed by Applied Materials[16][17] and is shown in Fig. 8.8. To improve uniformity of deposition, the platen is rotated so it must be heated by radiation. For this design, the platen is grounded and typically operates at ~325°C, and a 50 kHz power supply is used at 500 - 1000 watts. For 4 inch wafers, a load size of 22 wafers is practical so that good throughput is achieved. Hot-Wall, Parallel-Plate PECVD Reactors. The PECVD reactors just described operate with cold walls to minimize deposition on the reactor. However, this configuration limits the number of wafers in a single load, and therefore the throughput. Another approach would be to arrange long parallel and narrow electrodes so that they could fit into a hot tube, such as a diffusion furnace; then batch size could be much larger. Such an arrangement is shown in Fig. 8.9.


Figure 8.7. Radial-flow, plasma-enhanced CVD reactor after Reinberg.[15]

469

470


Figure 8.8. Radial-flow, plasma-enhanced CVD reactor.[16]


471

Figure 8.9. Hot-wall, parallel-plate reactor for plasma-enhanced CVD. (Courtesy of Pacific Western Systems, Inc.)

Five rectangular electrodes are positioned down the length of a quartz tube and held equidistant from each other. Wafers sit in pockets in these electrodes, which are alternately powered by 400 kHz power. A load of 84 4inch wafers can be run in this system compared to 22 in the cold-wall system, so throughput is higher. Electron Cyclotron Resonance (ECR) CVD Reactor. Finally, electron cyclotron resonance[18] has been used with 2.45 GHz power to operate a coldwall, single-wafer PECVD system such as that shown in Fig. 8.10. The system operates at low pressures (~1 mTorr) with a solenoidal magnetic field used to create the resonance condition. In this case, a very high degree of ionization is created away from the wafer, and since the pressure is low, the high number density of free radicals created persist until the plasma reaches the wafer. This prevents undesirable wafer bombardment during deposition.

472


Figure 8.10. ECR (Electron Cyclotron Discharge) reactor for plasma-enhanced CVD (after Matsuo[18]).

3.0 FILMS DEPOSITED BY CVD In the present section, we review some of the films that can be deposited by PECVD, with particular emphasis on those that are commercially important for fabrication of integrated circuits. Therefore, we concentrate on silicon nitride and silicon dioxide films. Amorphous silicon films are also deposited by PECVD, but they are used commercially in the manufacture of solar cells (not in the IC industry). Many other films can be deposited by PECVD, including conducting films, and these are reviewed briefly. More details are available elsewhere.[1] 3.1 Silicon Nitride PECVD of silicon nitride has generally been done using SiH4 as the siliconbearing reactant and some combination of NH3, N2 and H2 with one of several inert gases as diluents. The reactors used are all capacitively coupled and run at RF frequencies. For a particular reactor configuration, some of the parameters governing the film are:(i) operating pressure,(ii) operating temperature,(iii) discharge frequency, and (iv) reactant gas mixture.


473

The film quality obtained will be related to:(i)stoichiometry,(ii)H2 content,(iii) impurities, (iv) density, and (v) stress. The first three items relate to the film chemistry, and it is an interesting feature of such films that their composition can be controlled far more easily than is possible with a thermal process. The film density and stress relate to the mechanical behavior of the film, and therefore to its effectiveness as a diffusion barrier. Most of the development of PECVD silicon nitride has been done in systems such as those shown in Figs. 8.8 and 8.9 operated at a single low frequency. For the cold-wall system of Fig. 8.8, a frequency of 50 kHz is generally used. The hot-wall system of Fig. 8.9 has usually been operated at 400 kHz. More recently, silicon nitride films have been deposited in cold-wall systems operated at dual frequencies. When the reactor of Fig. 8.7 is operated at 50 kHz, 200 mTorr, gas flows of SiH4 /NH3/N2 = 140/270/800 sccm, at 500 watts, useful films are deposited. In Table 8.1, we compare these films to ones deposited thermally. Similar experiments carried out in a hot tube version reactor (see Fig. 8.8) yielded data on film quality as a function of several operating parameters.[20] For example, stress depends on wafer temperature, RF frequency, and gas pressure as shown in Fig. 8.11. For each of the three curves, the conditions were as specified below: Pressure:

T = 300°C f = 310 kHz SiH4/N2 /NH3 = 100/300/1100 sccm

Frequency:

T = 300°C P = 130 Pa SiH4/N2 /NH3 = 100/700/700 sccm

Temperature:

p = 130 Pa f = 310 kHz SiH4/N2 /NH3 = 100/200/1200 sccm

In general, a slight compressive stress in the film is the preferred condition. A film with high tensile stress will tend to crack, and one with high compressive stress will tend to delaminate. For this film and this system, it is best to stay below 4 MHz in frequency and 600°C in temperature. Since these silicon nitride films are used to passivate over aluminum, temperatures well below 500°C are preferred; and this can lead to compressive stress which is too high. Increasing the pressure can also lower the compressive stress, but this may lead to unacceptable film uniformity in a batch reactor environment.

474


Table 8.1. Physical and chemical properties of CVD and PECVD silicon nitride films.[19] Property

High Temp. Nitride 900°C

Composition Si 3N4 Si/N Ratio 0.75 Solution Etch Rate Buffered HF 20-25°C 10 - 15 Å/min 49% HF 23°C 80 Å/min 15 Å/min 85% H3PO4 155°C 120 Å/min 85% H3PO4 180°C Plasma Etch Rate 82% CF4-8% O2, 700 W 600 Å/min <100 Å Na+ Penetration IR Absorption Si-N max. ~830 cm-1 SiH minor Density 2.8 - 3.1 g/cm3 Refractive Index 2.0 - 2.1 Dielectric Constant 6-7 Dielectric Strength 1 x 107 V/cm Bulk Resistivity 1015 - 10 17Ω-cm Surface Resistivity >1013 Ω-cm Intrinsic Stress 1.2 - 1.8 x 1010 dyn/cm2 Tensile Thermal Expansion 4 x 10-6/°C Color, Transmitted None Step Coverage Good H2O Permeability Zero

Plasma Dep. Nitride 300°C SiN x 0.8 - 1.0 200 - 300 Å/min 1500 - 3000 Å/min 100 - 200 Å/min 600 - 1000 Å/min 1000 Å/min <100 Å ~830 cm-1 2,200 cm-1 2.5 - 2.8 g/cm3 2.0 - 2.1 6-9 6 x 106 V/cm 10 15 Ω-cm 1 x 1013 Ω-cm 1 - 8 x 109 dyn/cm2 Compressive Yellow Conformal Low - None

Another very important feature of PECVD silicon nitride films is the hydrogen content, which can be as high as 40%. For those systems operated at a single frequency, the hydrogen content of the film varies strongly with temperature and gas mixture, as shown in Fig. 8.12.[21] As can be seen, removing the NH3 from the gas mixture produces a lower hydrogen content. Similarly, increasing the temperature of deposition sharply reduces the H2.


475

Figure 8.11. Stress in silicon nitride films as functions of pressure, frequency and temperature.[20] (Reprinted by permission of the publisher, The Electrochemical Society, Inc.)

As circuit parameters have become more extreme with the advent of VLSI and ULSI circuits, the demands on PECVD silicon nitride films have increased. An excellent review of problems in passivation, such as its influence on aluminum voiding, electromigration, hillock formation, hot carriers, etc. has been published recently.[22] One way to deal with these new requirements is to use a PECVD reactor operated with dual frequency, such as the one shown in Fig. 8.6.[14] The use of this feature allows considerably more flexibility in film deposition

476


andis, in effect, another available system parameter to be adjusted. The effect of the high frequency is to produce a more stable discharge, while the low frequency increases the ion bombardment. For example, the stress can be changed from tensile to compressive simply by adjusting the ratio of low frequency/high frequency power. For a constant power level of 0.4 watts/cm2, the film stress is shown in Fig. 8.13. Also, even though the total hydrogen content of the film may be the same, it is possible to vary the Si-H vs. N-H bonds. Figure 8.14 shows that the Si-H bonds can be dramatically reduced at higher percentages of low frequency power. Apparently, the Si-H bond is much weaker than the N-H bond, so that in films where there is a large amount of Si-H, it is more likely that the hydrogen atoms will be mobile in the circuit. This can cause hot carrier problems as noted earlier.[22]

Figure 8.12. Hydrogen % versus deposition temperature for films deposited with SiH4 + NH3, SiH4 + N2 and SiH4 + NH3 + N2 (one point).[21]


477

Figure 8.13. Film stress as a function of percent of low frequency power.[14]

Figure 8.14. Si-H and N-H content of PECVD silicon nitride as a function of percent of low frequency power.[14]

478


3.2 Silicon Dioxide PECVD oxide can be grown from a number of reactants. Most work has been done with SiH4 plus any one of a number of oxidizers such as O 2, CO2, N2 O. Because of the potential for gas phase nucleation, O2 is generally avoided; the most commonly studied system is SiH4 + N2 O.[17] More recently, there has been considerable interest in the use of TEOS (tetraethoxysilane) plus O2, because of the improved step coverage possible with this system.[23] To illustrate the process, we can review the SiH4 + N2O PECVD process carried out in a cold-wall parallel-plate reactor (Fig. 8.8) at 57 kHz, 400 mTorr and 300°C. Typical results are listed in Table 8.2, where they are compared to similar results for PECVD silicon nitride. Observe that although the power level used for oxide is one-third that for nitride, the deposition rate is twice as high. Probably, the most significant finding is that there can be nitrogen as well as hydrogen in this film. In fact, adjusting the film stoichiometry and creating only nitrides is one of the most useful features of such a system.

Table 8.2. Plasma oxide and nitride characteristics.[17] Silicon Dioxide Gases SiH4 + N2O % SiH4 2% % N2O, NH3 resp. 98% RF Power Density 0.05 W/cm2 RF Frequency 57 kHz Operating Pressure 53 Pa Substrate Temperature 300°C Deposition Rate 60 nm/min Film Uniformity ± 5% Film Composition SiO1-9N0-15 Refractive Index 1.54 Film Density 2.38 g/cm3 Etch Rate (B.O.E.) 130 nm/min Etch Rate (CF4 + O2 plasma) 10 nm/min

Silicon Nitride SiH 4 + NH3 + N2 9% 45% 0.17 W/cm2 57 kHz 33 Pa 300°C 38 nm/min ± 4% Si3-2N4(H) 2.02 2.75 g/cm3 20 nm/min 150 nm/min


479

For example, we can adjust the dielectric constant over a range of 3:1 by changing the gas phase ratio of N2O to silane, as shown in Fig. 8.15. These experiments were carried out again in a parallel-plate cold-wallreactor at 600 mTorr and 350°C at a high frequency, 13.56 MHz.[24] Increasing the silicon content of the film appears to increase the dielectric constant.

Figure 8.15. Dielectric constant as a function of flow ratio (N 2O/SiH4).[24]

As noted earlier, one of the most perplexing problems associated with PECVD of silicon oxide films is the fact that the films tend to be nonconformal. Therefore, when covering high aspect ratio holes, the tendency is for the film to close over, leaving a hole behind as shown in Fig. 8.16.[23] The far more conformal coverage of the TEOS-based film is either due to a larger mean-free path for diffusion[23] or a much lower reactive sticking coefficient. [25]

480


Figure 8.16. SEM cross section micrographs of plasma oxide deposited from (a) silane and (b) TEOS for an aspect ratio of 0.74 (aspect ratio = metal height/metal space)[23]

The process conditions for the PECVD TEOS films are shown in Table 8.3, and the film properties are reported in Table 8.4. It is interesting to note that these depositions were done at 13.56 MHz in a cold-wall reactor.[23] Also, in spite of the film deposited from TEOS, there is very little carbon contamination of the film.


481

Table 8.3. Plasma TEOS Processing Conditions Gas Composition TEOS Flow O2 /TEOS Ratio Operating Pressure RF Frequency RF Power Substrate Temperature Deposition Rate

TEOS + O2 35 sccm 20:1 2 Torr 13.56 MHz 0.5 watts/cm2 400°C 2050 Å/min

Table 8.4. Dielectric Film Properties Thickness Uniformity Stoichiometry Carbon Content Refractive Index Stress Etch Rate (7:1 BOE) Breakdown Voltage Fixed Charge, QF Mobile Ions, QM

±1.5% Si:O = 1.0:2.0 (RBS) <0.2 atom % (SIMS) 1.45 ± 0.01 1.5 x 109 dynes/cm2 2200 Å/min 6 - 7 MV/cm <2.5 x 1011 cm-2 <6 x 1010 cm-2

3.3 Conducting Films The silicon nitride and oxide films just discussed are the only PECVD films currently being used commercially in integrated circuit manufacture. There are, however, many materials that can be deposited by this technique. These include other dielectrics, semiconductors (polysilicon, epi-silicon) as well as conductors. The latter may eventually be of commercial significance, so it is of value to summarize some of the more interesting studies that have been done. Table 8.5 lists some of the more interesting PECVD conducting films.

482


Table 8.5. Conducting films deposited by PECVD Material Reactants

W WSi2 Mo MoSi2 TaSi2 TiSi2 Al TiN

Temperature Pressure (°C) (mTorr)

WF6,H2 WF6,SiH4 MoCl5,H2 MoCl5,H2 TaCl5,SiH2Cl2 TiCl4,SiH 4 Al(CH3)3 TiCl4,N2,H2 TiCl 4,NH 3

350 230 350 400 580 450 400 400

200 600 1000 1000 1500 750 200 3000 1000

Frequency References (MHz) 4.5 13.56 3.5 0.05 13.56 13.56 12.0

26 27 28 28 29 30 31 32 33

In general, if temperature of deposition is a critical issue, then PECVD may be a viable approach to metal CVD. However, the films are likely to have a higher level of impurities compared to the thermal CVD ones, and this may prove a significant limitation.

REFERENCES 1.

Sherman, A., Chem. Vapor Deposition for Microelectronics, Noyes Publications, Park Ridge, NJ (1987)

2.

Sherman, A., J. Electrochem. Soc., 137:1892 (1990)

3.

Sherman, A., Thin Solid Films, 113:135 (1984)

4.

Brown, S. C., Basic Data of Plasma Physics, John Wiley & Sons, New York (1959)

5.

Brown, S. C.,Handbuch der Physik., Vol. 22, (S. Flugge, ed.), SpringerVerlag (1956)

6.

Thornton, J. A.,Deposition Technologies for Films and Coatings, (R. F. Bunshah, ed.), Noyes Publications, Park Ridge, NJ (1982)

7.

Butler, H. S., and Kino, G. S., Phys. Fluids, 6:1346 (1963)

8.

Koenig, H. R., and Maissel, L. I., IBM J. Res. Develop., 14:168 (1970)

9.

Horwitz, C. M., J. Vac. Sci. Technol., A1:60 (1983)


483

10. Smith, D. L. and Bruce, R. H., J. Electrochem. Soc., 129:2045 (1982) 11. Bruce, R. H., J. Appl. Phys., 52:7064 (1981) 12. Bruce, R. H.,Proc. of the Symp. on Plasma Etching and Deposition,811:243 (1981) 13. Logan, J. S., IBM J. Res. Develop., 14:172 (1970) 14. Van de Ven, E. P., Connick, I-W. and Harrus, A. S., Proceeding of the 7th Internat. VLSI Multilevel Interconnection Conf., IEEE, New York, (1990) 15. Reinberg, A. R., Radial Flow Reactor, U.S. Patent 3,757,733, (Sept. 11, 1973) 16. Rosler, R. S., Benzing, W. C., and Baldo, J., Solid State Technology, 19(6):45 (1976) 17. Van de Ven, E. P. G. T., Solid State Technology, 24(1):167 (1981) 18. Matsuo, S. and Kiuchi, M.,Proc. Symp. on Very Large-Scale Integration Science and Technology, p. 83, Electrochem. Soc., Pennington, NJ (1982) 19. Hollahan, J. R., Wauk, M. T., and Rosler, R. S., Proceedings of the 6th International Conf. on Chemical Vapor Deposition,(L. F. Donaghey, P. Rai-Choudhury and R. N. Tauber, eds.), p. 224, Electrochem. Soc., Pennington, NJ (1977) 20. Claasen, W. A. P., Valkenburg, W. G. J. N., Willemsen, M. F. C., and v.d. Wijgert, W. M., J. Electrochem. Soc., 132:893 (1985) 21. Chow, R., Lanford, W. A., Ke-Ming, W., and Rosler, R. S.,J. Appl. Phys., 53:5630 (1982) 22. Harrus, A. S., and Van de Ven, E. P., Semiconductor International, p. 124, (May 1990) 23. Chin, B. L., and Van de Ven, E. P.,Solid State Technology,p. 119 (April 1988) 24. Yokoyama, S., Dong, D. W., DiMaria, D. J., and Lai, S. K., J. Appl. Phys.,54:7058 (1983) 25. Cheng, L-Y, McVittie, J. P., and Seraswat, K. C., ULSI Science and Technology, (C. M. Osburn, and J. M. Andrews, eds.), 89-9:586, Electrochem. Soc., Pennington, NJ (1989) 26. Hess, D. W., Proc. of the Matl. Res. Soc. Symp., Vol. 38 (1985) 27. Akitomoto, K., and Watanabe, K., Appl. Phys. Lett., 39:445 (1981) 28. Tabuchi, A., Inoue, S., Maeda, M., and Takagi, M.,Proc. 23rd Symp. on Semicond. and IC Tech. of Japan, p. 60 (1982)

484


29. Hieber, K., Stoltz, M., and Wieczorek, C.,Proceedings of 9th International Conf. on Chem. Vapor Dep., (McD. Robinson, G. W. Cullen, eds.), 846:205, Electrochemical Society, Pennington, NJ, (1984). This figure was originally presented at the Spring 1984 Meeting of the Electrochemical Society, Inc. held in Cincinnati, OH. 30. Rosler, R. S., and Engle, G. M., J. Vac. Sci. and Technol., 82(4):733 (1984) 31. Ito, T., Sugii, T. and Nakamura, T., Digest of Papers of 1982 Symp. on VLSI Technol., IEEE, New York (1982) 32. Jang, D. H., and Chun, J. S., J. Vac. Sci Technol., A7(1):31 (1989) 33. Hilton, M. R. et al., Thin Solid Films, 139:247 (1986)

9 Plasma-Assisted Vapor Deposition Processes: Overview Rointan F. Bunshah

1.0 INTRODUCTION The previous chapters on Evaporation Deposition, Sputter Deposition, Ion Plating, Chemical Vapor Deposition, and Plasma-Assisted Chemical Vapor Deposition have covered all the major vapor deposition technologies. A perusal of these chapters shows that plasma-assisted vapor deposition is very important and extensively utilized for the deposition of compounds and novel technological materials as illustrated below. Compounds Oxides - TiO2, ZrO2 , Al2O3, SiO2, Y2 O3 , etc. Nitrides - TiN, ZrN, HfN, (Ti, Al)N, Ti, Zn)N, etc. Carbides - TiC, ZrC, HfC, TaC, WC, W2C, etc. Carbo-nitrides - TixCyNz, ZrxCyNz, etc. Sulfides - TiS2 , MoS2, MoS3 Novel Technological Materials Low Tc Superconductors - NbN, Nb3 Ge, CuMo6S 8 High Tc Superconductors - YBa2Cu3O7-d Photovoltaic Materials - aSi-H, CuInSe2, CuInS2 Optoelectronic Materials - ZnO, In(Sn)O2 Superhard Materials - Diamond, Cubic Boron Nitride, Amorphous (Diamond-like Carbon), C3N4 485

486


This chapter reviews the currently used plasma-assisted vapor deposition processes. They are analyzed in terms of the three steps in deposition processes, i.e., generation of the depositing species, transport from source to substrate, and film growth on the substrate. The role of the plasma in each of the steps for the various processes is discussed. All processes involve two sets of parameters, the plasma parameters and the process parameters. These parameters couple to a greater or lesser degree in each of the basic processes which reflects on their versatility. The role of plasma volume chemistry and plasma diagnostics is discussed. It is clear that a deeper basic understanding of plasma-assisted deposition processes necessitates a much greater volume of work on plasma diagnostics coupled with theoretical estimates. The role of ion bombardment on the structure, composition, and properties of the films is given. Hybrid processes which attempt to circumvent the somewhat deleterious intercoupling of the plasma and process parameters are briefly discussed. There are a large number of processes used to deposit thin and thick films of metals, alloys, ceramics, composites, etc., from solutions, gases and in a vacuum environment. They can be classified based on the size of the depositing species.[1] For example: Atomic deposition processes: electrodeposition, electroless deposition, evaporation, sputtering, chemical vapor deposition, etc. Droplet deposition processes: flame spray, wire spray, plasma spray, detonation gun, enameling, electrophoretic coating. Bulk deposition processes: painting, dip coating, printing, spin coating, explosive compaction, roll bonding, weld coating, etc. Current and future applications in the high technology areas require the deposition of simple and multiple layers of various materials in thin film form. The materials being deposited are metals, alloys ceramics, polymers, and composites on a variety of metallic and non-metallic substrates. The processes used are broadly classified into physical (PVD) and chemical vapor deposition (CVD) processes as illustrated in Table 9.1. In PVD processes, the deposition of compounds can be carried out either by direct evaporation/sputtering or reactive evaporation/sputtering. The term direct connotes that the target is the same compound as the film. Reactive implies that a metal/alloy is evaporated/sputtered in the presence of a reactive gas

Plasma-Assisted Vapor Deposition Processes

487

to deposit a compound. For example titanium is sputtered in nitrogen to deposit titanium nitride. The properties of the compounds are strongly influenced by their stoichiometry, i.e., anion:cation ratio. Control of stoichiometry in a deposited film is therefore very important and can be more readily achieved with reactive processes as compared to direct processes, as discussed in Ref. 2. This chapter will therefore concentrate on critical issues of plasma assistance in thin film deposition processes, with the emphasis on reactive deposition for compound films.

Table 9.1. Classification of PVD and CVD Processes

488


2.0 PLASMA-ASSISTED DEPOSITION PROCESSES In reactive deposition processes, the introduction of a plasma can be very beneficial or even essential. It should be noted that a plasma of a working gas (such as argon) is an inherent part of the sputtering process. On the other hand, for evaporation and CVD processes, plasmas are options that may be used if needed. In that sense, they are more versatile, i.e., offer an extra degree of freedom to the process. Plasmas are used not only in deposition processes, but also for etching and polymerization. Whenever a plasma is employed in a process, there are two sets of variables. Plasma variables: electron density, electron energy, and its distribution function Process variables: evaporation / sputtering rates, gas composition, pressure and flow rate, substrate bias, substrate bombardment, etc. Unfortunately, these sets of variables are not independent. Changes in a process variable can effect other process variables as well as the plasma parameters. The degree of this interdependency varies with the type of process as discussed below.

3.0 MODEL OF A DEPOSITION PROCESS All deposition processes consist of three basic steps as illustrated in Fig. 9.1, i.e., generation of the depositing species, transport from source to substrate, and film growth on the substrate. By way of illustration, consider these three steps for PVD processes. Step 1: Generation of the Vapor Species. Vapor species can be generated by: (a) evaporation using resistance, induction, electron beam, or laser beam heating. (b) sputtering using DC or RF plasma generation. Step 2: Transport of the Species From Source to Substrate. Various flow regimes can apply. 1. In molecular flow, the mean free path is larger than the source-to-substrate distance. This occurs at low partial pressures of the depositing specie and residual gas in the system, and is responsible for the line-of-sight feature typical of evaporation-deposition processes and low pressure magnetron-type sputtered deposition processes.


489

Figure 9.1. The three steps in film deposition.

2. Viscous flow occurs at higher partial pressures, 20 to 120 millitorr, typical of diode sputter deposition. It also is intentionally added in the evaporation deposition process to cause gas-scattering of the depositing specie and increase the throwing power of the process. An additional feature in Step 2 is the absence or presence of a plasma in the source-to-substrate region and the mode by which the plasma is excited, e.g., DC, RF, or microwave. The latter is an important aspect since it controls the electron energy and distribution function, and thereby, the plasma volume chemistry that takes place. Processes which involve low electron energies, such as plasma-assisted evaporation where the electron energies generating the plasma can be independently controlled, offer a more versatile and richer plasma volume chemistry than processes such as sputtering, where the electron energies are dictated by other considerations such as target voltage (50 - 100 eV) which controls the rate of sputtering; in sputter deposition, the electron energies thus cannot be controlled independently of other process parameters. It should be pointed out that the presence of a plasma is optional in the evaporation process but is an integral part of the sputtering process.

490


Step 3: Film Growth on the Substrate. The process depends on the energy of the incident species (which is typically 0.5 eV for evaporation and 10 to 100 eV for sputtering) and the substrate temperature. The structure, composition, and residual stress in the film can be substantially changed by bombardment of the growing film by energetic ions or neutrals. They can be generated by a separate ion source or be attracted to the film from the plasma by electrical biasing of the substrate/film. Thus, the location of the substrate inside or outside the plasma can substantially change the nature and amount of ion bombardment. In magnetron sputtering, the plasma is confined to a narrow zone near the target. Therefore, if the film is to be bombarded, a second plasma has to be created near the substrate by a suitable method. In CVD processes, the same general model applies with the differences being that the source is usually a gas or a vapor incorporated into a carrier gas. The transport phase occurs under vicious flow conditions only. This three-step model as applied to plasma-assisted deposition processes is illustrated in Fig. 9.2. Several items can be noted: 1. The plasma is optional for reactive evaporation processes but is an integral part of the reactive sputtering process.

Figure 9.2. General schematic of plasma-assisted vapor deposition process.


491

2. Plasmas can be localized around the source and/or the substrate. Thus one can independently influence the reaction occurring in each location. An example of this is given in Ref. 18 for the deposition of indium tin oxide films by reactive sputtering. 3. The substrate acquires a negative floating potential when it is immersed in a plasma thus causing ion bombardment of the substrate/film and the resultant changes in structure, composition, residual stress, and properties.

4.0 MATERIALS DEPOSITED BY REACTIVE VAPOR DEPOSITION PROCESSES Examples of ceramic compounds and novel materials that are deposited by the various vapor deposition processes were given at the beginning of this chapter. Many of the materials in this table are thermodynamically stable, i.e., ∆G is a large negative valve. The reactive deposition process is then controlled by the kinetics of the reaction between the constituent species. Conceptually, the kinetic or activation energy barrier is overcome by supplying energy into the reacting system via a plasma.[3] There are three materials listed, namely diamond, diamond like-carbon (or i-C, a-C) and cubic boron nitride that are metastable at ambient temperatures and pressures, i.e., they are not in a thermodynamically stable state. Here the problem of deposition of these materials is considerably more complex. The qualitative picture is to “persuade” the depositing species that they are locally in a “different” environmental regime where they are stable (i.e., at high temperatures and high pressures). This can be achieved by bombarding the film by ions, as well as by the associated chemical activity occurring in a plasma environment which permits the nucleation and growth of these metastable phases, which are then quenched in. There is another important difference between the deposition of the thermodynamically stable phases and the metastable phases. More energetic bombardment can actually cause etching away or removal of the deposited film or preclude the nucleation and growth of the film for the metastable phases whereas it can only change the composition, imperfection content, structure, and properties of the thermodynamically stable films.

492


An in-between case is the deposition of the high Tc superconductor films where the oxide phase may be thermodynamically stable but its oxygen content, and hence the properties, are drastically influenced by the process conditions.

5.0 KEY ISSUES IN PLASMA-ASSISTED REACTIVE VAPOR DEPOSITION PROCESSES The objective of any deposition process is to end up with a film with the desired composition, structure and properties. In plasma-assisted reactive process, there are some key issues. 5.1 Plasma Volume Chemistry Reactions taking place in the plasma volume to form molecular fragments, free radicals, etc., can then (a) condense directly on the substrate, migrate, and react to form a compound film, or (b) form precursor species in the plasma volume which deposit on the substrate and dissociate to form a compound film. Plasma diagnostic techniques, such as optical emission spectroscopy (OES), mass spectroscopy (MS), langmuir probe, and laser-induced fluorescence (LIF) spectroscopy, can be used to ascertain the species present in the plasma volume. While there is considerable published work on end point detection in plasma etching processes, there is relatively little work on deposition processes. A classic example is the deposition of TiC by reaction of evaporated titanium atoms in a hydrocarbon gas plasma to form TiC films by the activated reactive evaporation (ARE) process.[4] It was observed that if CH4 is used, no TiC films form. On the other hand, using C2 H2, TiC films readily form. Furthermore, the C/Ti ratio in the TiC film can be controlled by varying the evaporation rate of Ti or the partial pressure of C2 H2.[5a] Plasma diagnostic studies revealed that C2H2 dissociated into C, CH, CH2, CH3, fragments which reacted with Ti+ ions to form TiC. On the other hand, (CH4) did not fragment; instead it polymerized into (CH4 )n species which did not react with Ti+ and no TiC film was formed.[5b] This still does not answer the question raised above, i.e., did the Ti+ and the CH type fragments deposit on the substrate and react to form TiC directly or did a precursor species form in the plasma which then deposited on the substrate and yielded TiC film.


493

5.2 Type and Nature of the Bombardment of the Growing Film Any surface immersed in a plasma acquires a small negative potential which results in ion bombardment of the film and hence changes in its composition, structure, properties, and residual stress. The bombardment can be enhanced by intentionally applying a negative bias to the substrate. The plasma parameters, i.e., the nature and concentration of the charged species in the plasma, will therefore determine the resultant bombardment of the film. It should be mentioned that the substrate can be bombarded prior to deposition to remove surface species and an atomically-mixed interface between the substrate and film can be produced. In any plasma-assisted vapor deposition process, there are two sets of parameters to be considered, i.e., the plasma parameters (electron density, electron energy, and electron energy distribution function) and the process parameters (evaporation/sputtering rate, reactive/inert gas pressure, flow rate, substrate temperature, substrate bias, etc.). The model for film growth by PAVD process can therefore be schematically represented as shown in Fig. 9.2. One might picture, in a plasmaassisted deposition process, that the depositing species undergo various types of reactions in the plasma leading to the formation of excited neutral species, ions, free radicals, etc., which may react to form a precursor species that, in turn, deposits on the substrate, migrate on the surface, react, and form the film. The reactions forming the above species, i.e., the plasma volume chemistry, in turn are controlled by both the process and the plasma parameters, as discussed below. The rate of any chemical reaction in a plasma is primarily dependent on electron density (ne), electron energy (E), and distribution function f(E), as shown below: RA = N

∫

E =x

E =0

(E/ 2me ) 12 n e σ (E )f (E )dE

where N is the number density of colliding species,σ(E) is the collision crosssection, and f(E) is electron energy distribution function. The electrons are assumed to have a Maxwellian velocity distribution at a temperature Te and the cross-section for a given reaction is approximated by a step function of magnitude σo and threshold energy Eo. Then, RA = ne Nσo ve(1 + Eo /kTe) exp(-Eo /kTe)

494


Apart from the plasma parameters, the deposition parameters also influence the growth and properties of the films produced by any vapor deposition process. The most important deposition parameters are: 1. Rate of generation of vapor species which determine the deposition rate and stoichiometry of the films. 2. Partial pressure of all species in the gas phase which determines the mean free path of these species and hence affect the growth rate. In reactive deposition processes, partial pressures also determine the probability of the collisional reactions between various atomic and molecular species during transit from source to substrate, and hence influence the formation of precursor molecular species which in turn affects the growth and properties of the films. 3. Gas flow rate is an important process parameter, particularly in reactive deposition processes, since along with the metal species in the vapor phase, it controls the stoichiometry of the films. 4. Substrate temperature controls the composition, structure, and morphology of the films by affecting the atom mobility on the substrate as well as the rate of any chemical reaction occurring on the substrate. 5. Substrate bias, together with substrate temperature, also influences the structure and morphology since it controls the intensity of the ion bombardment of the growing film. Moreover, ion bombardment of the growing film can also lead to reduction of absorbed impurities and trapped gases in the films. In order to achieve better control of film properties, it is desirable to independently control the above parameters; however, it is not always possible in all deposition processes to achieve the process parameter flexibility conferred by the ability to vary them independently of each other. The nature and degree of intercoupling of the variables controlling the above parameters determines the advantages and limitations of a given deposition process. The presence of a plasma introduces additional constraints as some of the variables controlling the process parameters also affect the plasma parameters. To understand and optimize plasma-assisted deposition processes, it is necessary to evaluate this interrelationship between the process parameters and plasma parameters.


495

6.0 PLASMA-ASSISTED DEPOSITION TECHNIQUES IN CURRENT USAGE The most commonly used plasma assisted techniques for the deposition of compounds can be classified under the following two categories. 1. Plasma-assisted chemical vapor deposition (PACVD) processes. 2. Plasma-assisted physical vapor deposition (PAPVD) processes such as(a) reactive sputtering (RS), using DC or RF magnetron geometries and ion beams,and (b) activated reactive evaporation (ARE). Although they have been treated in other chapters, a summary is given here for each process with relevance to the main theme of the chapter. 6.1 Plasma-Assisted Chemical Vapor Deposition Plasma assisted chemical vapor deposition involves forming solid deposits by initiating chemical reactions in a gaseous discharge.[6] The discharge can be excited by using either RF, microwave, or photonic excitation. It produces a wide variety of chemical species in ionized and excited states, free radicals as well as ions and electrons. The nature, type, concentration, and energy of these species determine the growth and properties of the films. The important parameters controlling film growth by PACVD are as follows: (a) reactant partial pressure and flow rate; (b) RF power; and (c) substrate temperature and substrate bias. The above variables affect process parameters such as deposition rate on the one hand, and plasma parameters such as electron density, electron energy, and distribution function on the other. For example, the partial pressure of the reactant gas together with RF power determines the rate of dissociation of the reactive gas and hence the deposition rate. These same process variables also determine the electron energy, and electron density; moreover, the substrate floating potential depends on the average electron energy, so pressure and RF power also control the substrate bombardment. The substrate bombardment of the growing film and deposition rate both are therefore dependent on the same set of process variables, i.e., pressure and RF power. This interdependence of process and plasma parameters makes it difficult to obtain high deposition rates by PACVD processes. A variety of reactor designs have been used for carrying out PACVD in the laboratory; however, only parallel plate reactors have been used for

496


production applications. Detailed information on theory and practice of PACVD processes can be found in excellent reviews by Reinberg,[6] Hollahan and Rosler,[7] Rand,[8] Yasuda, [9] and Hollahan and Bell.[10] 6.2 Sputter Deposition The sputter deposition process involves a target and a plasma of a neutral working gas such as argon. The target material is transferred to the vapor phase by positive ion bombardment from the plasma via momentum transfer from the ions to the target atoms. The most important parameters controlling the growth and properties of the films by sputter deposition processes are: 1. target voltage and current, 2. reactant partial pressure and flow rate, and 3. substrate temperature and substrate bias. Similar to PACVD processes, these variables affect both process parameters as well as plasma parameters. For example, in conventional diode sputtering using either DC or RF, the deposition rate is dependent on target voltage and current as well as on pressure. However, these same parameters also determine the average energy of the secondary electrons, which in turn influences the floating potential and hence the bombardment of the growing film. The target voltage determines the energy of the secondary electrons ejected at the target. These are accelerated across the cathode sheath by a potential equal to target potential. The partial pressure on the other hand determines the mean free path and hence the collision frequency (number of collisions per unit length) of the electrons. As electrons lose energy in each collision, the average electron energy functionally depends on pressure. Thus the target voltage in conjunction with the operating pressure determines the average electron energy. Due to the relatively high voltage levels involved in diode sputtering, the energy of the secondary electrons is very high. Bombardment of the substrate by such high energy electrons leads to substrate heating, and radiation damage, and is thus a limiting factor in conventional DC and RF sputtering processes using the diode geometry. The target voltage/current and reactive gas flow rate exhibit a complex relationship in reactive sputter deposition processes due to target poisoning effects. This issue has been discussed in detail by Bunshah et al.[11]


497

There are many variants of the sputter deposition processes: 1. Diode geometry using DC or RF excitation. 2. Magnetron geometries with DC or RF excitation. 3. Reactive sputtering using diode and/or magnetron geometries with DC or RF excitation. In this process, a working gas (argon) is used in combination with a reactive gas. For a detailed review of physics and applications of sputter deposition processes, refer to review articles by Vossen and Cuomo,[12] and Thornton.[13] 6.3 Activated Reactive Evaporation (ARE) The activated reactive evaporation (ARE) process developed by Bunshah[14] involves evaporation of metal in the presence of a plasma of the reactive gas alone. There is no working gas in the ARE process. The two basic variants of the ARE process are:(i) the activated reactive evaporation process with an electron beam evaporation source, and (ii) the ARE process with a resistance heated source. Both of these processes are illustrated in Fig. 9.3. In ARE using an e-beam source, the metal is evaporated by an electron beam in presence of a reactive gas. The plasma is generated by accelerating the secondary electrons from the plasma sheath above the molten pool towards a probe biased to a low AC, or positive DC, potential. Nath and Bunshah[15] modified the ARE process for use with resistance heated sources. The metal is evaporated from a resistance heated source in the presence of the reactive gas. The plasma is generated by accelerating thermionically emitted electrons from a heated filament towards a positively biased anode. A transverse magnetic field is applied to cause the electrons to travel in spiral paths thereby increasing the probability of ionization. Apart from the above two basic geometries, many other variants of the ARE process have been developed. For further details, the reader is referred to a review by Bunshah.[16] The important process parameters controlling the growth and properties of films by the ARE process are: 1. evaporation rate; 2. plasma parameters such as electron density, electron energy and distribution function; and 3. substrate temperature and bias.

498


Figure 9.3. Schematic of the activated reactive evaporation system: (a) using an electron beam evaporation source, and (b) using a resistance-heated evaporation source.


499

Unlike PACVD and sputter deposition processes, the above variables can be controlledindependently. For example, one can control the deposition rate via the evaporation rate by controlling the e-beam current or the heating current passing through the boat source. This does not significantly affect the plasma parameters, which are controlled through an auxiliary anode potential. This ability to control plasma and deposition parameters relatively independently offers the ARE processes much greater flexibility to deposit films with varying stoichiometry, structure, and properties at high rates, as compared to PACVD and RS processes.

7.0 LIMITATIONS OF CURRENT PLASMA-ASSISTED TECHNIQUES As discussed earlier, the presence of the plasma in the source-substrate space significantly affects the processes occurring at each of the three steps in film deposition, which are: (a) generation of species, (b) transport from source to substrate, and(c) film growth on the substrate. Moreover, the effect of the plasma on the above three steps differs significantly between various processes. Such differences are manifest in terms of the types and concentration of the metastable species, ionized species, and energetic neutrals which, in turn, influence the reaction paths or steps involved in the overall reaction for film formation and the physical location of these reaction sites. Deshpandey et al.[11] have discussed in detail the role of plasma in plasma-assisted deposition processes. They have shown that the advantages and limitations of various plasma-assisted deposition techniques can be addressed in terms of the differences in plasma interactions at the source, during transport, and at the substrate in the respective processes. Comparisons between the three currently used plasma-assisted deposition techniques via reactive sputtering (RS), activated reactive evaporation (ARE) and plasma-assisted chemical vapor deposition (PACVD) in terms of plasma/ source–plasma/volume and plasma/substrate interactions is shown in Table 9.2. Also indicated in this table are the limitations/advantages inherent to each process. As can be seen from this table, each of the above processes suffers from limitations in terms of one or more of the following: 1. Control over the supply of the source material in vapor form. 2. Control of the number density and energy distribution of electrons and hence the associated plasma volume chemistry.

500

Table 9.2. Comparison of Plasma-Assisted Deposition Proceses



501

Most of the above limitations are due to the interdependency of the three reactions, i.e., plasma-source, plasma-volume, and plasma-substrate reactions. For an ideal plasma-assisted process, one should be able to control each of the above reactions independently of each other. In view of the above, attempts have been made to develop hybrid processes by combining various features of the plasma-assisted deposition techniques to extend the processing capabilities and to overcome the limitations of the individual techniques. Many modifications of the PACVD have been developed and have been discussed by Deshpandey and Bunshah.[17]

8.0 HYBRID PROCESSES In view of the facts discussed in the previous section, attempts have been made to develop hybrid processes by combining different plasma-assisted deposition techniques to extend the processing capabilities and to overcome the limitations of the individual techniques. The general thrust is directed towards: 1. Separation of the various parts of the process so as to exert independent control over each part and avoid complications due to overlap between the parts. 2. Use substrate/film bombardment with different species of controlled energy as contrasted to a spectrum of energies. A representative list of such hybrid techniques together with their advantages in processing is given in Table 9.3 and illustrated in Fig. 9.4

9.0 CONCLUSIONS It is clear that we have barely scratched the surface in our understanding of the detailed mechanisms of plasma assisted deposition processes. Much work remains to be done on experimental plasma diagnostics as well as the relevant theoretical modeling. This becomes particularly true when complex molecules instead of sample atomic shears are used as the reactants. It is unfortunate that very little systematic long-range support on studies leading to a basic understanding of these processes is available either from government or industrial sources. Until such a detailed understanding is developed, plasma-assisted deposition processes are still somewhat in the realm of enlightened witchcraft.

502

Table 9.3. Hybrid Processes Currently Used for Materials Synthesis


Table 9.3. (Cont'd)


503

504 Deposition Technologies for Films and Coatings

Figure 9.4. Schematics of hybrid deposition processes.


505

REFERENCES 1. Bunshah, R. F. and Mattox, D. M., Physics Today, 33:50 (1980) 2. Bunshah, R. F., Films and Coatings for Technology, (R. F. Bunshah, ed.) pp. 122-127, Noyes Publications (1982) 3. Bunshah, R. F., ibid., p. 128 4. Bunshah, R. F. and Raghuram, A. C., J. Vac. Sci. Technol., 9:1385-88 (1972) 5a. Raghuram, A. C. and Bunshah, R. F., J. Vac. Sci. Technol., 9:1389-94 (1972) 5b. Deshpandey, C. V., O’Brien, B. P., Doerr, H. J., and Bunshah, R. F., Surface and Coatings Technology, 33:1-16 (1987) 6. Reinberg, A. R., Ann. Rev. Mater. Sci., 9:341-372 (1979) 7. Hollahan, J. R. and Rosler, R. S., Thin Film Processes, (J. L. Vossen and W. Kern, eds.), pp. 335-360, Academic Press, New York (1978) 8. Rand, M. J., J. Vac. Sci. Technol., 16;420-427 (1979) 9. Yasuda, H.,Thin Film Processes, (J. L. Vossen and W. Kern, eds.), pp. 361-400, Academic Press, New York (1978) 10. Techniques and Application of Plasma Chemistry, (J. R. Hollahan and A. T. Bell, eds.), Wiley, New York (1974) 11. Deshpandey, C. and Bunshah, R. F., Surf. Coat. Technol., 27 (1986) 12. Vossen, J. L. and Cuomo, J. J., Thin Film Processes, (J. L. Vossen and W. Kern, eds.), pp. 12-75, Academic Press, New York (1978) 13. Thornton, J. A. and Penfold, A. S., ibid, pp. 75-114 14. Bunshah, R. F., "The Activated Reactive Evaporation Process", U.S. Patent #3,791,852 (Feb. 1974) 15. Nath, P. and Bunshah, R. F., Thin Solid Films, 69:63-68 (1980) 16. Bunshah, R. F., Thin Solid Films, 107:21(1983) 17. Deshpandey, C. V. and Bunshah, R. F., Thin Solid Films, 16:131-147 (1988) 18. Karim, A. A., Deshpandey, C., Doerr, H. J., and Bunshah, R. F., Thin Solid Films, 172:111-121 (1989)

10 Deposition from Aqueous Solutions: An Overview Morton Schwartz

1.0 INTRODUCTION Electrodeposition, also called electroplating or simply plating, is an economical technology to protect and enhance the functionality of parts used in many diverse industries including home appliances, jewelry, automotive, −in both decorative and engineering aircraft/aerospace, and electronics− applications. Although decorative applications have diminished somewhat primarily due to added expenses and problems associated with plant effluent (pollution) control and waste treatment, its applications in engineering, electroforming, and electronics have increased. The emphasis is on the latter applications, and the structures and properties of deposits. The purpose of decorative plating is to provide a durable, pleasing finish to the surfaces of manufactured articles, so the corrosion characteristics of the deposits and their ability to protect the substrate are important factors. These and other deposit properties involved in the selection and performance of decorative coatings including hardness, wear resistance, ductility, and stress are also important to the engineering applications of plated coatings. Engineering applications of plated coatings involve imparting special or improved properties to significant surfaces of a part or assembly and/or protecting or enhancing their function in their operating environment. Other applications include salvage of mismachined or worn parts and other types of reworking as well as material savings, use of less expensive materials, and substitution of materials more easily fabricated, Special technologies

506

Deposition from Aqueous Solutions

507

such as electroless deposition, electroforming, anodizing, thin films, magnetic coatings, and printed wiring (circuit) boards have been selected for discussion as representing specific engineering applications. Although electroplating and vacuum deposition processes are generally considered competitive processes, there are increasing applications in which they are (or can be) highly complementary. These involve utilizing the advantages of both deposition technologies: Vacuum Deposition ! Close tolerances ! Wide choice of substrates ! Wide choice of coatings

Aqueous Deposition ! ! ! !

Lower costs Thicker coatings Coating complex shapes Control and modification of deposit properties ! Control of residual stress

A combined process permits almost any substrate to be coated with a much wider range of deposits than either used alone. It extends the application of aqueous deposition to substrates which are difficult to coat, particularly non-metallics (ceramics) or active metals not readily or satisfactorily processed. Examples of combined processing include the electrodeposition of such metal substrates as Mo, Ti, and Be by initially sputtering a thin Cu or Au deposit to provide substantially improved adhesion.[1] The plating of plastics using chemical preparation is expensive, requires rigorous control, and presents formidable waste treatment problems. A pre-plate treatment using vacuum techniques to condition the surface by RF glow discharge followed by sputtering or electron beam evaporation of 1000 angstrom Ni and 1000 angstrom Cu deposits permits direct electroplating to final thicknesses.[2] In printed wiring board (PWB) fabrication, plasma processing is being employed to clean drilled holes and to remove drill smear. However, in some cases, this leaves undesirable interfering ash and decomposed residues requiring further chemical clean-up. Sputtering processes for depositing the initial Cu deposit on PWB’s and through-hole deposition have been developed to replace the electroless Cu deposition processes.[3] It has been predicted that such processes may replace the presently used electroless Cu systems completely.[4]

508


Electrodeposition has been used to fabricate magnetron sputtering targets of well defined shapes. The advantages include deposition precisely where needed or desired, eliminating waste. Such electrodeposited targets are quite pure with a minimum of oxygen or other gases. Sputtering targets of Ag, Cr, Au, Fe, Ni, Co and alloys have been prepared by electroplating.[5]

2.0 GENERAL PRINCIPLES Figure 10.1 represents a simplified plating cell. A DC source, usually a rectifier or motor generator, supplies current flowing in one direction through the external portion of the circuit when a potential difference is imposed across the system. The current flow is that of electrons in the external conductors. The mechanism of electrical transfer in the solution is by means of electrically charged “particles” called ions. Positive ions (cations) travel toward the negative electrode (cathode) and negative ions (anions) travel toward the positive electrode (anode) when the potential is applied, thus completing the electrical circuit. The electrolyte usually contains other components which influence the process (see Fig. 10.7 later in this chapter).

Figure 10.1. Plating Cell


509

The cathode or deposition reactions are characterized as reduction reactions since electrons are “consumed,” and the valence states of the ions involved are reduced. The anodic reactions are oxidation reactions wherein electrons are liberated, and the valence states are increased. Each set of reactions represents half-cell reactions and proceeds independently of the other, limited by a condition of material balance, i.e., electrons liberated in the anode reactions must equal the number of electrons “consumed” in the cathode reactions. The above describes systems such as nickel or copper deposition from acidic solutions* of their simple ions. Since these are divalent ions (Ni2+, Cu2+), the equations shown in Fig. 10.1 would involve two electrons. Deposition from solutions in which the metallic ions are combined with other ions or ligands as complex ions involves more complicated mechanisms. The cyanide-containing electrolytes represent the largest group of such systems. Some of these complex ions are so tightly constituted, i.e., the ionization constant of the metal ion is so small, that reduction or deposition of the metal atoms at the cathode occurs directly from the complex ions. This appears to be the mechanism involved with copper, silver and gold cyanide complex ions: Eq. (1)

[Cu(CN)3 ]= + e- " Cu o + 3(CN)-

Eq. (2)

[Ag(CN)2 ]- + e- " Ag o + 2(CN)-

Eq. (3)

[Au(CN)2 ]- + e- " Au o + 2(CN)-

The stability of the gold cyanide complex ion is such that it exists in mildly acidic gold plating solutions. It should be noted that the complex ions described above and other types are anionic and would migrate to the anode during electrolysis. Yet, deposition still takes place at the cathode, indicating that mechanisms other than simple electron reactions are involved in the cathode film. These complex anions approach the cathode by convection and/or diffusion where specific adsorption effects can occur in the double layer as discussed by Wagner, citing Frumkin and Florianovich.[6] The influence of simple cations present in the film are also involved in the reduction process. Faraday’s Laws of electrolysis (1833) are basic to electrodeposition.

* Most acidic plating solutions fall into this category, involving the simple ions.

510


They relate the current flow, time, and the equivalent weight of the metal with the weight of deposit and may be stated as follows: 1. The amount of chemical change at an electrode is directly proportional to the quantity of electricity passing through the solution. 2. The amounts of different substances liberated at an electrode by a given quantity of electricity are proportional to their chemical equivalent weights. Faraday’s Laws may be expressed quantitatively:

Eq. (4)

W =

I · t Eq ____________ F

where:

W I t Eq F

= = = = =

weight of deposit in grams current flow in amperes time in seconds Equivalent weight of deposited element Faraday, a constant, = 96,500 coulombs (approx.)

I · t is the quantity of electricity used (coulombs = ampere-seconds) and Eq, the equivalent weight of the element, is the atomic weight divided by the valence change, i.e., the number of electrons involved. If the current is not constant, then I · t must be integrated: t2

I dt

t1 From a practical standpoint, the weight of the deposit is converted to the more meaningful thickness of the deposit using the relationship, W (gms) = volume (cm3)/deposit density, with the volume of the deposit equal to the thickness (in µm)* times the area (in m2). The Faraday**, F, can be experimentally determined by rearranging Eq. (4): * Thickness in ∝m ξ 0.0394 = thickness in mils. ** The Faraday can be derived from the fact that 1 gram-atomic weight of an element contains 6.023 x 1023 atoms (Avogadro’s Number, N). If the charge on the ion is A, then Z x N electrons are required to deposit 1 gram-atomic weight, and Z x N/A = 6.023 electrons are required to deposit (or dissolve) 1 equivalent weight of an element. Since the charge on an electron is 1.602 x 10-19 coulombs, 6.023 x 1023 electrons x 1.602 x 10-19 coulombs = 96,496 coulombs.)


Eq. (5)

F =

511

I · t Eq ____________ W

Rearranging Eq. (4) to Eq. (6)

It =

WF

_________

Eq

permits the determination of the charge passing through a circuit by the known deposition or dissolution of an element, usually silver. Devices which utilize this application of Faraday’s Laws are known as coulometers. Coulometers are used to determine the efficiency of a deposition process. They are also employed as either timers or integrators possessing “electrochemical memory” and producing sharp potential “end-points,” i.e., significant changes in electrode potentials which activate electronic circuits. Figure 10.2 schematically illustrates such a device—an electrochemical cell called an E-cell as part of an electronic circuit.

Figure 10.2. Microcoulometer − E-Cell (Courtesy Plessey Electro-Products)

Faraday’s Laws are absolute laws, and no deviations or exceptions have been found. Apparent exceptions can be shown to be incorrect or explained by failure to take into account all the chemical or electrochemical reactions involved at the electrode. Thus, the efficiency of an electrochemical reaction can be determined:

512

Eq. (7)


% Electrode Efficiency = 100 ×

actual weight of deposit theoretical wt. of deposit

Table 10.1 indicates typical cathode current efficiencies for some common deposits from various electrolytes. With knowledge of the actual efficiency, predicted (average) thickness of deposit can be obtained, limited by the control of the current distribution. Table 10.1. Cathode Current Efficiencies of Various Plating Solutions _____________________________________________________

Deposit

Electrolyte

Range, %

_____________________________________________________

Ag

CN

100

Au

Acid Neutral CN

50 - 100

Cd

CN

85 - 95

Cr

CrO3 /H2SO4 CrO3 /SO4-F

10 - 15 18 - 25

Cu

Acid SO4 CN (low eff.) CN (high eff.) P2O7

97 - 100 30 - 45 90 - 95 ~100

Fe

Acid

90 - 98

In

Acid or CN

30 - 50

Ni

Acid

93 - 98

Pb

Acid

95 - 100

Rh

Acid

10 - 50

Sn

Acid Alkaline

90 - 95 70 - 95

Zn

Acid CN

~95 50 - 80

___________________________________________________


513

The current flowing through a conductor is driven by a potential difference or voltage, the magnitude of which is determined by the relationship expressed as Ohm’s Law (1826-27): Eq. (8)

E = IR

where E = volts, I = amps, and R = ohms. This law regulates both the current flow and its paths in an electrodeposition cell. A commercial electroplating installation and operation involves a multiplicity of series and parallel electrical circuits with only the total current and applied voltage controlled. The current distribution on each individual part or portion of a part (and resulting deposit thickness and properties) depends on the electrode potentials and resistances involved in the “mini-circuits” as well as the geometry and spacing of parts. Since the resistances of the solid, metallic conductors in the circuit are several orders of magnitude lower than the electrolytic (solution) resistances, they can usually be neglected. The potentials within the electrolyte and, more importantly, the electrode-electrolyte interfaces, are fundamental controlling factors and are not as straight-forward as suggested by Ohm’s Law. When a metal is immersed in a solution containing its ions, an equilibrium condition is set up between the tendency for the metal to go into solution and the tendency for the metallic ions in solution to deposit on the metal: M0 ← → Mn+ + ne-. However, before this equilibrium is established (i.e., the exchange currents or current densities are equal |i+| = |i-| = i0 ), one of the reactions may be faster than the other, resulting in a “charge separation.” If the reaction proceeding to the right is faster than to the left, the metal surface would be negatively charged. If the deposition reaction (to the left) is faster, then the surface would be positively charged. This resulting potential between the metal and the solution (at unit activity) is called the single or standard electrode potential. Since this is a half-cell reaction, a reference electrode, the saturated hydrogen electrode (SHE) is used to complete the circuit and is given the arbitrary value of zero potential. In many instances other reference electrodes such as the calomel electrode are substituted with appropriate corrections. Potential measurements made in this manner (or values derived thermodynamically) result in a series known as the Electromotive Force (EMF) Series. This origin of the electrode potential was first formulated by W. Nernst (1889). The magnitude of the potential difference between the metal and its ionic solution is given by the Nernst equation:

514

Eq. (9)


E = Eo +

RT _____ nF

where

E = Eo = R = T = n = F = a =

ax (products) ln ______________ ay (reactants)*

observed EMF, potential difference (volts) standard EMF gas constant, 8.314 (j · o K-1 mol -1) absolute temperature, o K valence change (electron transfer) Faraday, 96,500 coulombs (A · sec mol-1) activity (apparent degree of dissociation)

If the natural logarithm is converted to logarithm base 10, and T is 298o K (25oC), then Eq. (9) becomes: Eq. (10)

E = Eo + (0.059/n)** log a (or log c approx.)

Thus, a tenfold change in ion concentration changes the electrode potential by 59 mV/n (a negative change makes the electrode potential less positive). This is significant when complexing agents are present since the ionic concentration can be reduced drastically with the accompanying change in electrode potential.*** For example, Eo = -0.76 volts for zinc. But, when zinc is complexed with cyanide: Eq. (13)

Zn2+ + 4(CN)- → [Zn(CN)4]=

the electrode potential shifts to approximately -1.1volt. The standard electrode potential for the Cu1+ /CuM half cell is +0.52 volts which shifts to ~1.1 volts when complexed with cyanide:

* Since the metal (solid) is the reactant in a plating cell, its activity is considered = 1 for all practical purposes and can be neglected. Also, as a practical approximation, the concentration in moles/L can be substituted for activities.

** 0.059 = 2.303 x 8.316 x 298.1 / 96,496. *** When complex ion reactions are involved: Eq. (11)

Mn+ + qX p-

→ ← [MXq]n-pq

where q is the coordination number, then the Nernst equation is modified:

Eq. (12)

E=

Eo

a nM− pq RT RT q - nF ln Kf + nF ln q a p− X

Kf is the stability constant of the complex ion. Since Kf will be quite large for very stable complexes, the potential can shift substantially negatively.


Eq. (14)

515

Cu1+ + 3(CN) - → [Cu(CN)3 ]=

The practical significance is that a copper cyanide strike provides the best undercoat on a zinc surface since the potentials are essentially the same. Attempts to use an acid copper (Cu2+ ) (Eo = +0.34 volts) solution would provide a potential difference of 1.1 volts, resulting in an immersion or displacement deposit with poor adhesion. The closeness of the electrode potentials for the [Zn(CN)4]= and [Cu(CN)3 ]= complexes also permits these metals to be deposited simultaneously as a brass alloy deposit from cyanide solutions. For electrodeposition reactions to occur, an additional potential is required to overcome the equilibrium conditions discussed above, i.e., to provide a non-equilibrium, irreversible condition. Referring to Fig. 10.1, the total plating voltage is the sum of three components. E2 represents the potential required to overcome the resistance of the electrolyte and obeys Ohm’s Law; it would be the only potential required if only the single electrode potentials were involved in the electrodeposition process. E1 and E3 are the potentials at the electrodes required to sustain the electrolysis process when the current is flowing and exceed the single electrode potentials. The additional voltage is called polarization which usually increases as the current increases. The electrical energy is converted to heat according to Joules Law: Eq. (15)

Eheat = I E t = I 2R t

resulting in increased temperatures of the electrolytic solutions. Polarization, also called overpotential or overvoltage*, is an important controlling factor in electrodeposition processes. A minimum energy which the reactants must possess is a requisite for any chemical reaction to occur. For an electrochemical reaction to proceed, an overpotential is required to overcome the potential barrier at the electrode/solution interface; this is called the activation overpotential. It is the overpotential required for the charge-transfer reaction itself and is kinetically controlled. Cathodic activation overpotential shifts the energy level of the ions in the inner electrical double layer nearer to the potential barrier, so that more ions can cross it in * In more rigorous treatments, the term overvoltage is restricted to the excess potential required for a single reaction (usually irreversible) to proceed at a specified electrode whereas the term polarization is more general and includes all reactions at the electrode.

516


a given time, producing a deposit on the surface. Activation overpotential also exists at the anode but in the opposite direction. Changes in the ion concentrations at the electrodes are major contributions to polarization. Figure 10.3 depicts the increased metallic ionic concentration at the anode and the decreased concentration at the cathode as a result of the dissolution and deposition processes. This results in corresponding changes in the equilibrium potentials per the Nernst equation (Eq. 9) since it changes the value of log CE/CS, CE being the ionic concentration at the electrode and CS the concentration in the bulk of the solution (see Eq. 10). This effect due to the concentration changes is called concentration polarization and is mass transport controlled.

Figure 10.3. Concentration polarization

Figure 10.4 illustrates a typical current/potential curve indicating the regions of activation polarization (ηa ), concentration polarization (ηC), limiting current followed by a post limiting region (with gas evolution). The open circuit (rest) potential is indicated by (EM)R. Where anodic and cathodic polarization curves intersect, |io | is indicated. Increased anode concentration polarization ultimately results in the evolution of oxygen which reacts with the electrode to produce oxide insulating films increasing the ohmic resistance. The oxygen may also react with various solution constituents such as organic compounds or cyanides,


517

thereby consuming them and/or converting them into other compounds which may be detrimental to the electrodeposition process. In some processes, such as anodizing of aluminum or where insoluble anodes are involved such as in chromium plating, anode polarization is desirable. Cathodic concentration polarization may result in the evolution of hydrogen as the competing reaction. The pH of the cathode film increases and hydrates or hydroxides may precipitate and be occluded in the deposit. The co-deposition of hydrogen may result in brittleness of the deposit and, by migration and diffusion into the substrate, result in hydrogen embrittlement.

Figure 10.4. Typical polarization curve

Hydrogen overvoltage which is the polarization for the specific reaction discharging hydrogen at a specified electrode surface involves at least two steps: Eq. (16)

2H+ + 2e → 2Hadsorbed → H2 (gas molecule)

The latter step is usually the slower, rate-determining step, and a higher potential is required to discharge the gas. The factors influencing hydrogen or oxygen overvoltage include: a. Electrolyte composition b. Type of metal electrode c. Nature of electrode surface d. Current density e. Temperature

518


Agitation and increased operating temperature of the solution help minimize concentration polarization, permitting higher current densities and faster plating rates. The "throwing power" (TP) of a plating solution or, more properly, macrothrowing power since it indicates the degree of plate uniformity (thickness distribution) over the substrate surface contours, is an important characteristic because the deposit properties and overall quality are affected. Factors influencing deposit (plate) distribution are shown in Table 10.2. Generally, electrolytes containing free metallic ions exhibit poorer throwing power than those in which the ions are complexed or contain supporting, non-depositing, ions, the latter improving solution conductivity. The overall “geometry” of the plating system influences current distribution. Significantly increased cathode polarization at higher current densities results in decreased current efficiency, improving throwing power. Thus, cathode current efficiency-current density curves are useful in predicting the throwing power of a plating solution. If the cathode efficiency decreases with increased current density, the throwing power improves proportionately. The shape and slope of the curves are indicative of the throwing power. However, if the cathode efficiency remains high over a wide range of current densities, the throwing power is usually poor. Examples of desirable cathodic polarization are complex ion-containing solutions such as alkaline stannate (tin plating) and alkaline-cyanide (zinc plating) solutions. Based on Haring-Blum %TP values, Schaefer and Pochapsky[7] reported that conventional plating solutions generally fall into four classes: (i) alkaline stannate and zincate (%TP > 50), (ii) most other cyanide solutions (%TP = 25 to 50), (iii) most acid solutions (%TP > 0 to 25), (iv) chromium plating (%TP = -100 to 0). The wide range of “negative” macrothrowing power for chromium (from chromic acid solutions) is due to the fact that, within limits, the cathode current efficiency increases with increasing current density, thereby greatly exaggerating the non-uniformity of the deposits. Rothschild[8] showed substantial improvement in the throwing power of the so-called high throw acid copper sulfate plating solutions containing low metal/high acid concentrations employed for through-hole plating of printed circuits boards, 87% vs. 14% for a conventional high metal/low acid concentration solution. Foulke and Johnson[9] investigated the throwing power characteristics of various precious metal plating solution formulations. Percent throwing power values for both macro- and microthrowing power characteristics were reported.


519

The term throwing power is sometimes mistakenly applied to another property of plating solutions: namely, covering power. Covering power relates to the lowest applied current density at which a plating solution produces a deposit, i.e., it is a measure of a solution’s ability to deposit into geometric recesses of an article to be plated. At very low current densities in some plating solutions, the potential required for metal deposition (sometimes referred to as the decomposition potential) may not be reached, and some other electrode reactions support the passage of current; these may include hydrogen evolution or the reduction of addition agents or other reducible species or ions. Poor covering power or the inability to deposit metal into areas of low current density can sometimes be overcome or avoided by using a high current density (“strike”) to initiate plating into the recess and then reducing the current density to the normal operating range. Table 10.2: Factors Influencing Current/Plate Distribution Type of Electrolyte Simple “Free” Ions Complex Ions Supporting Ions Polarization Conductivity Cathode EfficiencyCurrent Density Curves Geometry of Plating System Other Factors Substrate Composition and Structure Surface Preparation and Pre-treatment

Current distribution over the electrode surfaces influence plate distribution and is differentiated as primary, secondary, and tertiary. Primary current distribution involves the plating system geometry with the potential constant over the electrode surface and negligible polarization influences. Secondary current distribution involves activation overpotential (ηa), electrode kinetics, and solution conductivity. Tertiary current distribution involves concentration overpotential (ηC), the diffusion (boundary) layer, and solution agitation, i.e., mass transport is a factor. Current and plate distributions, and methods and calculations for determining throwing power, such as the Hull Cell,[10] the Haring-Blum

520


Cell,[11] and Wagner number are reviewed and analyzed by Ibl.[12] Shawki et al[13] devised a method for measuring throwing power into recesses and holes using cathodes with varying tube diameters and determining plate thickness along the depth of the tubes. The concept of microthrowing power is discussed later. 3.0 ELECTRODEPOSITION 3.1 Mechanism of Deposition Metal deposition differs from other electrochemical processes in that a new solid phase is produced. This dynamic process complicates and introduces new factors in elucidation of the mechanisms involved in the discharge of ions at the electrode surfaces. Factors determining deposition processes include: 1. The electrical double layer (~10 angstroms thick) and adsorption of ions at the surface-some 2 - 3 angstroms away. At any electrode immersed in an electrolyte, a double layer of charges is set up in the metal and the solution ions adjacent to the surface. At solid electrode surfaces, which are usually heterogeneous, the character and constitution of this double layer may exhibit local variations, resulting in variations in the kinetics of the deposition process. This could affect the electrocrystallization processes involved in the overall growth process. −especially those 2. The energy and geometry of solvated ions− involving complex ions. All metal ions are associated with either the solvent (water) molecules or complexed with other solution constituents either electrostatically or by coordinated covalent bonding. Desolvation energy is required in transferring the metal ion out of solution to the growing crystal lattice. 3. Polarization effects. A symposium[14] on Electrode Processes was held by the Faraday Society in 1947. The excellent papers pioneered the concepts upon which the modern concepts of the deposition mechanism are based. Schaefer and King[15] compiled a chronological annotated bibliography on polarization covering the period 1875 - 1951. Thus, the condition of the metal surface to be plated is a basic determining factor in the kinetics of the deposition process and the morphol-


521

ogy and properties of the final deposit. The presence of other inorganic ions and organic additives in the double layer or adsorbed on to the surface can greatly modify the electrocrystallization and growth process (Fig. 10.5).

Figure 10.5. (a) The distribution of ions and dipoles in the electrical double-layer. (1) Cations, (2) anions, (3) specifically adsorbed anions, (4) adsorbed additives, (5) adsorbed water dipoles, (6) electrons. (b) The potential as a function of distance in the double-layer [corresponding to (a)] measured from the metal surface.[16]

522


Based on these considerations, several deposition mechanisms have been proposed.[16]-[21] The basic or essential steps as shown in Fig. 10.6 include: 1. The aquo- or complexed metal ion is transferred or deposited as an adion (still partially bound) to a surface site. Such sites include the plane surface, edges, corners, crevices or holes with the plane surface providing the primary sites. 2. The adion diffuses across the surface until it meets a growing edge or step where further dehydration or desorption occurs. 3. Continued transfer or diffusion steps may occur into a kink or vacancy or coordinate with other adions, accompanied by more dehydration until it is finally fully coordinated with other ions (and electrons) and becomes part of the metal being incorporated into the lattice.

Figure 10.6. Diagram of the crystallization process according to the theory of Kossel and Stranski. Different atom positions: (a) another phase (gas phase, melt, electrolyte), (b) in the lattice plane [ad-atom (ad-ion)], (c) edge (step) site, (d) growth (kink) site.[18]

Deposition of metal ions results in depletion in the solution adjacent to the surface. These ions must be replenished if the deposition process is to continue. This replenishment or mass transport of the ions can be accomplished in three ways: 1. Ionic migration is least effective. The mobility of the metal ion is very low, its migration rate being dependent on the current and the transport number which is usually less than 0.5. When other conducting salts, are added, these conduct most of the current,


523

reducing the metal ion migration approaching zero. In the case of complex ions where the total charge is negative (complexed as anions). the migration is actually in the reverse direction. 2. Convection is the most effective, involving substantial movement of the solution. This is accomplished by mechanical stirring, circulation or air agitation of the solution or moving the electrodes (parts) through the solution. Any one or combination may be employed. 3. Diffusion is the effective mechanism for ionic migration in the vicinity of the electrode surface itself where convection becomes negligible. The region near the electrode surface where the concentration of the ions differs from that of the bulk of solution is called the diffusion or boundary layer. It is defined somewhat arbitrarily as the region where the concentrations differ by 1% or more.[22] The diffusion layer is much thicker than the electrical double layer (approximately 15,000 to 200,000 times thicker, depending on agitation and temperature. Figure 10.7 illustrates the diffusion/boundary layer, differentiating the Nernst diffusion layer (δN) from the actual diffusion layer. The diffusion rate of the reacting species is given by: R = D (Cs - Ce) / δN

Eq. (17) where:

R D Cs Ce δN

= = = = =

diffusion rate (moles cm-2 s-1) Diffusion coefficient (cm2 s-1) solution concentration (bulk concentration) concentration at electrode the Nernst diffusion thickness

The diffusion rate increases as δN decreases. On flat, smooth electrode areas, the diffusion layer is fairly uniform. At rough surfaces or irregularities which have a roughness profile with dimensions about equal to the diffusion layer thickness, the diffusion layer cannot follow the surface profile, being thinner at the micropeaks than in the microvalleys. The deposit may be thicker at the peaks than in the valleys, a condition characterized as poor microthrowing power. A reverse condition may also exist resulting in good microthrowing power or leveling, i.e., making the surface smoother after plating than before plating. Figure 10.8 represents the three types of microthrowing power. The plating solution composition, especially organic additives, greatly influences the character of microthrowing power and brightening.

524


Figure 10.7. Relation of Nernst diffusion layer (δN) to actual diffusion layer (δ).

The limiting current (or current density) (see Fig. 10.4) can be defined* as: Eq. (18)

iL = nFD (Cs - Ce ) /,δN

where n = electron transfer, F = Faraday constant, D = diffusion coefficient. When the overpotential is sufficiently high, Ce can be neglected and Eq. (18) becomes: Eq. (19)

iL = nFDCs / δN

and D/ δN equals the mass transfer coefficient. Increasing the current density increases the plating rate. However, the deposit deteriorates when the current density exceeds some value depending on the solution composition and operating variables. Rough, burnt, dendritic, or powdery deposits maybe obtained when the limiting current density region is approached. Landau[23] observed that it is not the absolute value of the current density which determines the quality of the deposit, but

* Electroplaters consider the limiting current density as the maximum current density which will still produce acceptable deposits. This value is generally lower than calculated values using Eqs. 18 and 19.


525

rather the ratio of actual plating current density to the limiting current density,i/iL . He determined that when the ratio exceeded 0.6, rough deposits generally resulted, with the concentration overpotential, ηC, becoming appreciable. The limiting current density is higher as agitation is increased, especially in the turbulent regime.

Figure 10.8. Types of microthrowing power.

526


3.2 Parameters The parameters generally controlling the composition, structure and properties of the deposit are shown in Fig. 10.9. These are briefly reviewed. Basic Electrolyte Composition. This includes the compounds supplying the metal ions (to be deposited) and the supporting ions. The basic functions of the supporting ions or compounds are to stabilize the electrolyte, to improve solution conductivity, to prevent excessive polarization and passivation (especially anodic), and to provide compatibility with the desired plating conditions. Supporting ions or conducting salts reduce the current shared by the metallic ions or complexes, making convection (agitation) a more significant factor. Additives. Additives, commonly called addition agents (A.A.), are frequently added to plating solutions to alter desirably the character of the deposit. Read[24] discussed the effects of A.A. on the physical and mechani−intentional or accidental. They are usually organic cal properties of deposits− or colloidal in nature although some are soluble inorganic compounds. When additives produce a specific effect, they are descriptively called brighteners, levelers, grain-refiners, stress-relievers, anti-pitters, etc. Profound effects are produced with small concentrations, ranging from a few mg/L to a few percent. In general, the effective concentration range is of the order of 10-4 to 10 -2 moles/liter. The mechanisms by which these effects are achieved are not clear in spite of a considerable amount of research and published literature (including a voluminous patent literature, since most commercial additives are proprietary). However, the additive must be adsorbed or included in the deposit in order to exert its effect, and thus appears related to its role in the diffusion layer. Kardos[25] reviewed comprehensively the “diffusional” theory of leveling, which, experimentally derived, provides a scientific basis to explain the phenomenon. To date, no generally acceptable mechanism has been devised to explain satisfactorily the brightening action of addition agents. Brightness, of course, is related to the absence of roughness on a very small scale. The diffusion-controlled leveling theory may be involved for rough surfaces but becomes inapplicable on smooth surfaces or on the sustained growth of bright deposits. Kardos[25] recognized this limitation, and for the latter favored the selective adsorption of inhibitor on certain growth sites without being diffusion-controlled. The selective adsorption of brightening agents on active sites (lattice kinks, crystal projections, growth steps) or random


Basic Electrolyte Composition

Controlled Plating Variables

Impurities

Additives

pH C.D. Temperature Agitation Time

Composition and Structure of Deposit

Leveling Brightening Anti-pitting Grain refining Stress relieving

Current Characteristics

Thickness

Physical Properties Density Coefficient of thermal expansion Electrical resistance

527

DC-% ripple PR IC Superimposed AC Asymmetrical AC Pulsed Mechanical Properties Hardness Ductility Tensile Strength Stress Modulus of elasticity

Corrosion resistance Magnetic properties

Figure 10.9. Factors influencing the properties of deposition

528


adsorption to suppress crystallographic differences in the deposit represent other proposals of the mechanism, but these are still highly conjectural. Other reviews,[26]-[29] with extensive references, discuss the problems associated with elucidating a brightening mechanism. It seems likely that the “trial and error” method of selecting brightening agents will remain the really −at least for the near future. effective approach− While neither leveling nor brightening may be considered “properties” of the deposit, the resulting grain size can greatly influence the physical and mechanical properties due to the inclusion of these additives or the decom−especially sulfur and/or carbon− −in the deposit. The position products− corrosion characteristics of these deposits are also affected, usually adversely. The functions of the other types of addition agents are evident. In many instances, the same addition agent performs several of these functions or acts synergistically with other solution constituents or other addition agents. Controlled Plating Variables. The influence and effects of the operating variables are somewhat dependent upon the solution composition. They are also interdependent. All exert an influence on the structure and properties of the deposit. They are not always predictable, and establishment of optimum ranges is usually determined empirically. The use of ultrasonic energy agitation in electroplating solutions, i.e., its effect on the polarization, the diffusion layer and properties of deposits has received considerable interest since the 1950’s. Rich[30] determined that low frequency vibrations (16 - 30 kHz) produced more uniform results, and Roll[31] obtained best results in the frequency range 20 - 50 kHz with intensity (power) range at 0.3 - 0.5 watt cm-2. However, Hickman[32] found that results based only on reported frequencies and intensities provide an inadequate description and suggested the use of the limiting current method with characterization of the ultrasonic agitation intensity in terms of diffusion layer thickness. The considerable research by Russian investigators is reported by Kapustin and Trofimov.[33] Walker and Walker[34] reviewed the effects of ultrasonic agitation on properties of deposits and noted that the conflicting results reported in the literature may be due to the differences in frequency, intensity and methods of application. Forbes and Ricks[35] were able to reduce the number of preparatory steps required to silver-plate aluminum bus bars from 11 to 4, using ultrasonic agitation in key operations. In this connection, ultrasonic agitation has been widely employed in degreasing, cleaning and pickling pre-plating operations.


529

Some of the advantages attributed to the use of ultrasonics in electrodeposition are; 1. Higher permissible current densities resulting in higher rates of deposition. 2. Suppression of hydrogen evolution in favor of metal deposition, i.e., a shift in limiting current density. 3. Improved adhesion 4. Reduced porosity 5. Reduced stress 6. Increased brightness 7. Increased hardness (especially in chromium deposits) The influence of ultrasonic agitation on grain size appears to be the most important factor, controlling most of the other property changes. However, no specific effects or trends can be attributed to ultrasonics.[34] Walker and Holt[36] applied hydrosonic agitation as an alternative to ultrasonic agitation. The plating solution is circulated under pressure through an hydrosonic generator, converting solution velocity to acoustical energy with a pulse waveform from 5 kHz to the ultrasonic range solely by mechanical action. Results similar to plating with ultrasonic agitation were reported without some of the disadvantages associated with ultrasonics. Impurities. It is practically impossible to maintain a plating solution free of impurities. Common or potential sources include: 1. Chemicals used for make-up and maintenance. 2. 3. 4. 5. 6.

Impure anodes. Improperly cleaned anodes, anode bags and filters. Rubber or plastic linings and hoses. Rack coatings or maskants. Decomposition of addition agents.

7. 8. 9. 10.

Improper rinsing and drag-in of solution from the previous step Accumulated dissolution of parts during plating. Corrosion of electrical bus bars suspended above the solution. Improper or insufficient cleaning or preparation of parts prior to plating. 11. Fall-in of airborne dirt and oil particles. 12. Chemicals in water used for volume replenishment (e.g., hard water). 13. Generally poor housekeeping.

530


Particles suspended in the solution may become attached to the surface, resulting in rough, nodular deposits, or leave pits if they fall off; either result produces adverse effects on the integrity and corrosion resistance of the deposit. (A notable exception is the dispersion of controlled particles to be included into the deposit; see Dispersion Coatings in Sec. 5.2.) Organic impurities generally contribute to pitting, poor covering power, poor adhesion, and harder, more brittle, stressed, darker deposits. Metallic impurities contribute to pitting, poor throwing power, poor adhesion, lower cathode efficiency, stress and cracking, brittleness, burning and off-color deposits. These cations may co-deposit or become entrapped in the deposit, altering its structure and properties. Furthermore, the distribution of the impurity in the deposit may be current density −usually more concentrated in the low C.D. areas. (Dummying, dependent− i.e., removal of impurities from solution by electrolysis, is a common practice −especially nickel− −at C.D.’s between 0.2 and 0.5 in certain plating solutions− amps/dm2.) The effects and removal of metallic impurities (copper, zinc and iron) in nickel deposition were studied in detail by D. T. Ewing et al.[37] −with very few excepCurrent Characteristics. All plating processes− −require unidirectional or direct current (DC). Current sources are tions− motor generators or rectifiers which convert alternating current (AC) to DC, with the latter almost completely supplanting the former. At present, silicon rectifiers are the most widely used. Depending on the number of rectifying elements, the type of AC (single or three phase), and the circuitry, the output wave form can be half wave or (usually) full wave with varying percentages of ripple, ranging from 48% to less than 4%. In most plating processes, especially from complex ion type solutions, ripple may not be too significant. However, it can be a significant factor in some plating operations, notably chromium where the ripple should be low (5 - 10%), since higher ripple may co-deposit excessive oxides and adversely affect the deposit’s structure and result in dull deposits. DC rectifiers used in gold and other precious metal plating require ripple to be as low as 1% for optimum deposit characteristics. Figure 10.10 represents examples of the modulated current forms employed in attempts to reduce the magnitude and effects of polarization and to alter the structure and properties of deposits. Superimposed AC on DC, the earliest approach, has not had extensive application. Zentner[38] employed this technique to raise the coercive force (Hc) and decrease remanence (BR) in cobalt-nickel alloys developed for hard magnetics.


Figure 10.10. Examples of pulsed wave forms[47]

531

532


Periodic reverse, or PR,[39] under the proper conditions, produces dense fine-grain, striated, leveled and bright deposits. It has its greatest effect and applications on deposits from cyanide solutions, notably copper, permitting smooth, heavy deposits. [Copper deposits produced with DC from cyanide solutions generally become nodular when thicknesses exceed 0.075 - 0.1 mm (3 - 4 mils).] A typical PR cycle is 15 seconds plating and 3 seconds deplating; the longer the deplating (reversal) cycle, the smoother is the deposit. The extended plating time or increased current density required by PR to deposit a given thickness led to the use of interrupted DC employing similar cycles. The interrupted or off duty segment’s function is to permit the diffusion layer to be replenished. Asymmetric AC can be considered a variation of PR. In an interesting application, Rehrig[40] used high frequencies (500 Hz), and very high current densities [cathodic C.D. ~82.5 A/dm2 - 110 A/dm2 (~750 - 1000 A/ft2) and anodic C.D. at 25% of cathodic C.D.] for high speed spot plating of gold on lead frames to obtain good bonding properties. In contrast, DC current densities in excess of 33 A/dm2 (300 A/ft2) produced deposits with poor bonding characteristics. Co-deposited metallic impurities were removed during the anodic phase; the degree of effectiveness was proportional to the anodic C.D. with a minimum of 22 A/dm2 (200 A/ft2 ) required. The deposit hardness decreased and the bond pull strength increased as anodic C.D. increased. Considerable work is being done applying pulsed current modification in plating, especially in electronic plating applications. Wan et al.[41] and Puippe et al.[42] reviewed the literature to 1979. Two symposia[43] and a monograph, “Theory and Practice of Pulse Plating,”[44] present current practices and applications involving pulse plating. Pulse plating may be defined as on/off DC as is interrupted DC mentioned above. The primary differences are that the on pulses are of very short duration, generally 5 - 15 milliseconds, and the off time is approximately ten times longer; much higher current densities are applied. The so-called duty cycle is the ratio of on time (Ton) to off time (Toff): Eq. (20)

Duty Cycle = Ton / (Ton + Toff)

The time interval (on + off) is the reciprocal of the frequency, the on time being the product of the time interval and duty cycle. The average current density is calculated as the peak current density times the duty cycle. The average current density in pulse plating cannot


533

exceed the diffusion limiting current density determined for DC plating[45][46] The size of a pulse rectifier, i.e., the peak current required, is determined by the ratio of the average current to the duty cycle. The duty cycle is usually reported as percent. Each variable influences the properties and quality of the deposit and the optimum conditions are usually determined experimentally. Osero[47] evaluated the equipment associated with pulse current modification. Avila and Brown[48] detail the circuitry and power requirements. They indicated that the off time is critical since it is based on and determines the requirements of the diffusion layer returning to equilibrium. Cheh et al.[49] indicated that the cathode current efficiency (CCE) dropped from 100% to a 93.7 - 80.4% range due to pulse plating, shorter pulses (0.5 msec) resulting in lower CCE than longer ones (2 - 10 msec). They hypothesized that this may be due to the 2-step reduction mechanism as advanced by Mattson and Bockris:[50] Eq. 21

Cu2+ + e- → Cu 1+

Eq. 22

Cu1+ + e- → Cu0

where Eq. 21 is faster than Eq. 22 during the first interval of the pulse. Thus, the cuprous ion (Cu1+) accumulates and during the relaxation (off) period disproportionates: Eq. 23

2Cu1+ ← → Cu2+ + Cu0

The throwing power of copper, as measured with the Haring Cell, was somewhat reduced by pulse plating while that of gold from a citratephosphate solution was improved; however, the improvement diminished rapidly with increasing peak current densities. Using a rotating disc electrode, they found a slight improvement in the microthrowing power of the gold solution due to pulsing; however, the current densities and especially the agitation used had more significant effects. Reid[51] found that pulsed plating in cobalt-hardened gold deposits virtually eliminated polymer formation under low C.D. (~5 mA/cm2 ) and high off-to-on ratios (100 ms/10 ms). Other effects in the properties noted were: 1. Improved ductility without any significant decrease in hardness. −even in the presence of polymer− −from 17.1 2. Increased density− gm/cm3 for DC plating (1 mA/cm 2) to 19.2 gm/cm3.

534

Deposition Technologies for Films and Coatings 3. Significant reduction in electrical resistance for Co-hardened gold, from 14 µΩ-cm to 6 µΩ-cm, but an insignificant reduction in pure gold deposits, from 3 µΩ-cm to 2.4 µΩ-cm.

Effects of pulsed plating on the deposit compositions and properties of gold and gold alloys are reviewed by Raub and Knödler.[52] They show increased alloying element content (Ni or Co) and a decrease in carbon content as a function of off-time. The tensile stresses are reduced in alloy deposits, while the hardness is about 10% higher than that of comparable DC plated alloys. The gas content (H2, N2, and O2 ) of pulsed plated deposits is also substantially reduced. Knödler[53] reviewed the use and effects of pulse plating of the precious metals. Hosokawa et al.[54] found that desirable properties of gold and rhenium deposits sometimes lie within a narrow range of pulse parameters. They found that the CCE was five times greater than with DC plating when the duty cycle exceeded 50% with a pulse duration of 3 - 5 µsec. Puippe and Ibl[55] studied the influence of Ton and Toff on the morphology of cadmium, copper, and gold deposits. The influence of Toff proved to be important with regard to electro-crystallization; it also strongly influenced other properties unrelated to the morphology of the deposits. The effect of pulse plating on current distribution and throwing power was reviewed by Dossenbach.[56] He indicated that pulse plating does not affect primary current distribution and compared to DC plating provides a less uniform secondary current distribution whereas the tertiary distribution can be improved, especially for short duration high current density pulses. Avila[57] reviewed pulse plating of alloys. The pulse plating of other (individual) metals have been discussed in Refs. 43 and 44. Fundamental aspects of pulse plating were presented by Ibl.[43a][58] The influence of pulsing and the effect on the double layer at the electrode surface are discussed by Puippe and Ibl.[59] Some of the advantages claimed for pulse plating are: 1. Faster plating rates due to increased permissible current densities. 2. Denser deposits (less porosity). 3. Higher purity of deposits, less tendency for impurities to deposit. 4. Smoother, finer-grained deposits. 5. Reduced need or elimination of addition agents. 6 Less hydrogen evolution, providing sharper, finer lines at masking interfaces and possibly less hydrogen embiittlement. 7. Decreased stress in deposits. 8. Increased Ni or Co contents in alloy-hardened gold deposits with less polymer formation.


535

Some of the effects and advantages attributed to pulse plating and other modulated current wave forms are very similar to those for ultrasonic agitation. Both attempt to reduce the adverse polarization effects by decreasing the Nernst diffusion layer thickness while increasing mass transfer of the reacting species, permitting the use of higher current densities. In many instances, pulse plating has an effect similar to organic addition agents, especially as related to grain size. A more recent approach to modification of plating processes and resulting deposits involving the simultaneous application of laser energy was first reported by von Gutfeld et al.[60] The impingement of a laser beam on the cathode surface resulted in increased plating rates by as much as a factor of 1000. The mechanisms responsible for this deposition rate increase were investigated by Puippe et al.[61] The absorption of laser energy resulted in localized increase in temperature at the cathode/solution interface which produced vigorous agitation (microstirring), a shift in the rest potential (open circuit potential), and an increase in both the charge transfer and mass transfer rates. Gutfeld and Romankiw[62] described the application of laser-enhanced plating to gold patterning, i.e., the selective deposition on spots and patterns or tracks with the ability to “write directly” without the use of masks as well as potential use in repair of electronic circuitry. Bocking[63] described plating equipment and set-up combining laser-enhanced plating with high speed jet selective plating with plating rates as high as 16 ∝m/s. Pure gold was deposited on both metallic and metallized ceramic substrates without the need for any masking. Gutfeld et al.[64] developed a method for selective pattern plating by applying a dielectric coating and utilizing a laser (Nd-YAG laser) to produce the desired pattern leaving a clean surface which could then be plated by conventional means. Gelchinski et al.[65] found that laserenhanced jet plating of gold increased deposit smoothness and decreased nodularity and voids with increasing laser power density; hardness of the deposits was in the range characteristic of soft gold. Kuiken et al.[66] indicated that laser-enhanced plating was not very effective on good heat-conducting substrates since only a limited temperature rise results. They suggested depositing an undercoat of a relatively poor heat-conducting material such as a nickel-phosphorus alloy (which is almost 1/20 that of nickel), significantly improving the effectiveness of laserenhanced plating and reducing the need for high laser power densities.

536


Zahavi et al.,[67] using a Nd/YAG laser, deposited Au and Pd-Ni alloys directly on semiconductor and polymeric substrates with conventional electroplating solutions but without external current. The deposition was highly selective and accomplished without masking or any surface preparation. The deposits exhibited Schottky contact behavior on n-type silicon and GaAs substrates.

4.0 PROCESSING TECHNIQUES The preparation of metal surfaces for plating involves the modification or replacement of interfering films to provide a surface upon which deposits can be produced with satisfactory adhesion. The type and composition of the soils present as well as the composition and metallurgical condition of the substrate determine the “preparation cycle” and the materials used. The operations involved are designed to accomplish these objectives: 1. Clean the surface. 2. Pickle or condition the surface. 3. Etch or “activate” the surface. 4. Stabilize the surface. Strike In some cycles, several objectives are combined in the same operation. Rinsing steps follow each treatment step. Each of these steps is examined separately. Cleaning. The cleaning steps serve two functions: (i) Removal of bulk soils (oils, grease, dirt). This may involve mechanical operations such as wet or dry blasting with abrasive media, brushing or scrubbing or chemical cleaning with solvents (degreasing) or emulsions. (ii) Removal of last “trace” residues. Usually chemical soak (or spray) and electrochemical cleaners are employed. These can affect the substrate and therefore should be compatible with it. Such cleaners may contain alkaline chemicals, surfactants, emulsifying or dispersing agents, water softeners, inhibitors, and chelating agents. Acidic formulated cleaners are also available. Pickling or Conditioning. These are acid dips which neutralize and solubilize the residual alkaline films and micro-etch the surface. The common acid dips are either sulfuric acid (~5 - 15% v/v) or hydrochloric acid (~5% to full strength) and are satisfactory for most alloys. Where undesirable reactions or effects may occur, the acid dip should be formulated to be compatible with the substrate composition.


537

Etching or Activating. Undesirable (from the plating viewpoint) metallurgical micro-constituents are removed or rendered non-interfering; e.g., silicides in aluminum alloys or nickel, or chromium in stainless steels or super alloys, or these steps remove or reduce oxides or other passive conditions prevalent to some surfaces. High nickel and/or chromium containing alloys usually have a tenacious oxide or passive films which must be destroyed with strong acids or anodic etching in strong acids. Solutions containing 15 - 25% v/v or more sulfuric acid are usually employed at low current densities, 2.2 A/dm2 - 5.5 A/dm2 (20 - 50 A/ft2) for metal removal, or at high current densities, 10 - 30 A/dm2 (100 - 300 A/ft2) for smut removal or oxide alteration. Both current density ranges may be employed to maximize adhesion of thick deposits. In special cases, activation may be accomplished by cathodic treatment in acid or alkaline (cyanide) solutions. Hydrogen is deposited at the surface to reduce superficial oxide films. Solution contamination must be avoided or minimized since such contamination especially heavy metal ions may be codeposited as smut. Stabilizing. Very active materials alloys of aluminum, magnesium or titanium tend to oxidize or adsorb gases readily, even during rinsing and transfer. These continue to interfere with adhesion of deposits. Therefore, a necessary step involving an immersion deposit of zinc or tin, electroless coating, or modified porous oxides is required to make the surface receptive to an adherent electrodeposit. The electrodeposition of thin coatings from specially formulated solutions called strikes are considered stabilizing steps since they provide new, homogeneous, virgin surfaces upon which subsequent deposits are plated. These strike solutions and plating conditions are usually designed to be highly inefficient electrochemically. The considerable hydrogen gas evolution assists any final cleaning, reduction of oxides, and activation of the surface while the thin deposit covers surface defects and remaining soils (smut). The most widely used strike is the cyanide copper strike. The pH and “free” cyanide content are varied depending on the alloy being plated. A typical formulation range is: Copper cyanide, CuCN 15 - 25 g/L (2 - 3.5 oz/gal) Sodium cyanide, NaCN 22 - 40 g/L (3 - 5.3 oz/gal) “Free cyanide”, NaCN 2.5 - 11 g/L (0.3 - 1.5 oz/gal) Sodium carbonate, Na2CO3 15 - 60 g/L (2 - 8 oz/gal) pH Temperature C.D.

10.5 - 13.0 RT or slightly elevated, 38 - 45oC 0.55 - 1.1 A/dm2 (5 - 10 A/ft2) 1.1 - 2.2 A/dm2 (10 - 20 A/ft2)

538


The lower pH’s and free cyanide are used for sensitive metals such as aluminum or zinc alloys; the higher pH is used for steels. The cleaning or activating ability is increased with increasing pH, free cyanide, C.D. and temperature. The deposit thickness is approximately 0.25 - 0.50 ∝m (0.01 - 0.02 mils); thicker deposits—1.25 - 2.5 ∝m (0.05 - 0.1 mils)—are applied at slightly higher temperatures. The second most common strike is the Woods nickel strike or its modifications. This strike is effective (and preferred) on high nickel or chromium containing alloys. A typical formulation is: Nickel Chloride, NiCl2· 6H2 O 240 g/L (32 oz/gal) Hydrochloric Acid (Conc.): 125 ml/L (16 fl. oz/gal) Temperature RT (20 - 30oC) C.D. 5 - 20 A/dm2 (50 - 200 A/ft2) Time 0.5 - 3 minutes Silver or gold strikes are used prior to plating thicker deposits of these metals. Either a gold strike or the strike of the particular precious metal is used prior to plating the specific metal. These are generally formulated similar to the plating solution except that they contain approximately onetenth the metal ion concentration. The strikes may be applied directly to the substrate or, more commonly, on the copper or nickel strikes discussed. The use of these strikes minimizes the possible contamination of precious metal plating solutions. Unusual strikes are sometimes employed in special procedures. For example, a chromium strike appears to be most effective for plating on molybdenum alloys or an acid copper or electroless nickel strike on titanium alloys followed by a thermal diffusion treatment to obtain adhesion of subsequent deposits. Properly designed preparation cycles and the establishment of a stable receptive surface are prime requisites for good quality deposits. However, the condition and integrity (or lack of it) of the surface prior to plating also affect the quality of the deposit; this is becoming more evident as quality and functional requirements of electrodeposits are increased. Some plating processes require post-plating treatments. To improve the corrosion resistance of zinc or cadmium deposits or the tarnish resistance of silver, chromate conversion coatings are applied by chemical or electrochemical treatments; these gel-like films also improve adhesion of paint films. Since most preparation and plating processes generate hydrogen which can be occluded and can migrate into the substrate, possibly causing


539

hydrogen embrittlement, stressed articles or high-strength materials are usually given a stress relief bake in air at 190o C (350 - 400oF) for 3 - 24 hours within 3 - 4 hours after plating. Procedures for the preparation of difficult-to-plate substrates have been prepared as “Standard Recommended Practices” by ASTM. These are listed in Appendix A. The “Standards” reference the literature upon which they are based. Included in the Appendix is a discussion of preparation of less common metals.

5.0 SELECTION OF DEPOSIT 5.1 Individual Metals Only nineteen or so of all the known individual (single) metals are presently of practical interest in aqueous electrodeposition. Of these, only ten have been reduced to large scale commercial practice. These are indicated in Table 10.3 with the most widely used ones underlined. Holt[68] reviews the electrodeposition of “uncommon” elements from aqueous, organic and fused salt media. Alloy deposition, electroless deposition and deposition with dispersed particles (inclusion plating) extend the practical use of aqueous coating systems considerably. These are discussed separately. In order to make a proper selection of a deposited coating, one must be cognizant of the fact that these coatings can vary widely in structure and physical and chemical properties, depending on the electrolyte composition and operating conditions as discussed above. For example, the hardness of as-plated chromium deposits can be varied from 350 to 1100 DHN, and nickel from about 150 to 650 DHN. The corrosion protection afforded by a coating depends upon its electrochemical relationship to the substrate, its thickness, continuity (porosity), and the environment as well as its overall quality. The important factors to be considered in the selection of a deposit are the purpose of the deposit and the use (function) of the finished article. Other factors which must be considered are the size, shape, and expected useful life of the article and the costs and environment involved. Table 10.4 comprises a list of various engineering functions of deposited coatings and the deposits usually employed. Table 10.5 gives “representative” hardness values for various deposits in relation to some common materials and hardness scales. Spencer[69] discusses selection factors for coatings, their properties and characteristics, and uses.

540

Table 10.3. The Periodic Table



541

Table 10.4. Selection of Deposits ________________________________________________________________________

Primary Function of Coating

Most Widely Used Coating

Representative Application

_________________________________________________________________________

Corrosion Resistance

Zn, Cd Sn Ni, Cr

Decorative

Sacrificial coatings, fasteners, hardware fittings Food Containers Food processing equipment (wear resistance required)

Cu/Ni/Cr composite, Brass (Cu-Zn) Ag, Au, Rh

Household appliances, automotive trim Jewelry

Dielectrics

Anodized oxide coatings of Al & Ti, Ta

Condensers Capacitors Coatings

Electroforms

Ni, Cu, Fe, (Cr) Co, composites

Radar “plumbing,” screens, bellows, containers, molds

High temp. oxidation Cr, Rh, Pd, Pt, resistance Au, Ni Diffusion Barrier

Air and space craft Electronic devices

Maskant

Cu, Sn Bronze Sn, Pb-Sn

Selective carburizing, nitriding Etch Resists

Reflectors

Ag, Rh, Cr Au

Visible light reflectors Infra-red reflectors

Salvage

Cu, Ni, Cr, Fe

Mismachined, worn parts

Soldering, Bonding

Pb, Sn, Sn-Pb Cu, Ag, Au Sn-Ni, Cd, Ni

Containers, printed circuit and other electronic assemblies and chassis

Wear Resistance

Ni, Cr, E-Ni, Hard Anodizing Rh, Au, Au alloys

Air and space craft, hydraulics Electronic contacts

_________________________________________________________________________

542


Table 10.5. Comparison of "Normal Hardness" of Commonly Deposited Coatings in Relation to Hardness Scales

(Modified and based on Metal Progress, p. 131, Sept. 1959)


543

5.2 Alloy Deposition Alloy deposition extends the availability and applicability of coatings from aqueous solutions. It is an area of increasing research and development, although most of the systems have not attained commercial application. An extensive literature has developed. Brenner’s two-volume comprehensive, definitive monograph[70] details compositions, operating conditions, structures and properties of the deposits, covering developments up to 1960. A Russian monograph[71] details their extensive research in this area. Brenner[72] updated the state of the art to 1964. Krohn and Bohn[73] reviewed the literature to 1973 with a count of more than two hundred binary alloys; Fig. 10.11 summarizes the binary alloy combinations reported to June, 1972. Over one thousand abstracts on alloy deposition were reported in Chemical Abstracts between 1964 and 1972. Sadana et al.[74] annually review developments in alloy plating.

Figure 10.11. Binary alloys which have been electrodeposited form aqueous solution: % indicates alloys reported up to 1960, $ indicates alloys electrodeposited for the first time between 1961 and 1964, and & indicates alloys reported since 1964.[73]

544

Deposition Technologies for Films and Coatings The most widely used plated alloys are: Cu-Zn

brasses ranging from red brass to white brass, primarily decorative and for rubber bonding.

Cu-Sn

bronzes, decorative, antiquing and as corrosion resistant undercoats substituting for a copper strike.

Sn-Pb

compositions ranging from 5% Sn to 65% Sn. Applications include bearings, corrosion resistant coatings, solderable coatings and etch-resists in electronic assemblies.

Au-Co, Au-Ni

hardened gold alloy deposits used for electronic contacts and wearing surfaces.

Sn-Ni

for corrosion resistance and solderability.

Ni-Fe

as substitute for nickel plating (decorative), soft magnetics on computer heads (Permalloy).

Ni-P

deposited either electrolytically or (more prevalently) electrolessly for its hardness, wearability and corrosion resistance and as non-magnetic undercoat on computer hard disks.

Co-Ni

for decorative plating, magnetic applications electroforming (molds for plastics).

Co-P

for hard magnetics, sometimes as ternary alloys containing Ni, Fe, Zn, W, Mo, etc.

The electrodeposition of tungsten alloys[75]-[78] of Fe, Ni and especially Co is commercially feasible but has remained largely experimental although their properties should be of sufficient interest for engineering applications. While the as-deposited hardness is lower than chromium or Ni-P, these alloys can be precipitation hardened. One drawback is the high optimum temperature (600oC) for the Co-W alloys, which can be detrimental to the substrate. The deposits retain hot hardness similar to the Stellites. Binary and ternary alloys of Fe, Ni and Cu have been produced almost as stainless steel coatings[79]-[83] other studies[22][73][84] include reviews. Machu[85] investigated the problems with anodes especially oxidation to higher valence states and the use of insoluble anodes, alone and in combination with soluble anodes. Other work with ternary alloys has been


545

with gold alloys to increase hardness; Au-Ag-Sb alloys[86] reportedly showed wear resistances 25 - 33 times greater than pure Au. Srivastava[87] reviewed the electrodeposition of ternary alloys with special reference to solution compositions and characteristics and applications. Amorphous coatings, i.e., coatings exhibiting no x-ray diffraction patterns, have been produced by electrodeposition and electroless deposition. Aqueous deposition possibly is the best means for producing amorphous metals and alloys since low operating temperatures are involved and rapid solidification (as with metallurgically produced alloys) is not involved. Iron, nickel and cobalt-based alloys containing sufficient phosphorus or boron are generally amorphous in the as-plated condition. Thick amorphous electroforms of Ni-P have been produced.[88] The deposition of amorphous alloys is not restricted to alloys containing these light non-metallic elements. Amorphous single metal electrodeposits have also been produced, e.g., amorphous chromium deposits.[89] Amorphous deposits are generally hard, and corrosion and wear resistant. Methods other than co-deposition have been developed to produce alloy coatings. These include diffusion of sequential deposits, dispersion of particles or fibers in deposits (electro-composites), electrophoretic phenomena, and mechanical plating. Diffusion coatings. These processes involve the deposition of coatings sequentially similar to the composite; Cu under Ni under Cr for decorative finishes, followed by a thermal diffusion treatment. Such techniques have been applied to improve the adhesion of deposits on difficult-toplate substrates (diffusion bonding). They have not, however, been extensively applied to producing alloy coatings by deposition possibly due to temperature requirements and the formation of intermediate diffusion zones with undesirable properties (brittleness, etc.). A proprietary alloy of Ni-Zn called “Corronizing”[90] was used commercially as an improved corrosion resistant coating. (Subsequently, codeposited Ni-Zn alloys were developed.) The substitution of Cd for Zn by Moeller and Snell[91] produced a corrosion preventive coating for jet engine parts, permitting the use of low alloy steels operating at temperatures up to 535o C (1000oF). The coating consisted of 5 - 10.2∝m (0.2 - 0.4 mils) Ni plus 2.54 - 5 ∝m (0.1 - 0.2 mil) Cd diffused at 332oC (630 oF) (M.P. of Cd = 321o C (611oF)). The satisfactory function of this diffused alloy coating is dependent on the quality and characteristics of the Ni component.[92] Sequentially deposited coatings of Co-W alloy and Cr diffused in air and in a carburizing atmosphere are shown in Fig. 10.12 to illustrate the potential of producing unique alloy coatings by controlled heat treatments.

546


(a)

(b)

(c)

Figure 10.12. Diffused Co-W/Cr/Co-W composite coatings. (a) H.T. in air, 1680°F, 10 hrs (500x) (unetched), (b) H.T. in carburizing atmosphere, 1680°F, 10 hrs (500x) (unetched), (c) H.T. in atmosphere, 1680°F, 10 hrs (500x) (etchant: hot Murakami).


547

Dispersion Coatings. One of the common problems in electroplating is roughness of the deposit, the primary cause of which is the presence and suspension of discrete particles in the solution with subsequent entrapment in the deposit. To overcome this problem, continuous or periodic filtration is part of the operation for many types of plating solutions; so it is not difficult to include foreign material into a deposit. The purposeful addition of a second, dispersed phase of controlled particle size into a plating solution, is referred to variously as: dispersion, inclusion, occlusion, composite or electrophoretic plating, deposition or coating. The requirements are simple: 1. The particles must be insoluble (or only slightly soluble) in the solution. 2. The particles must be compatible with the solution, i.e., not produce any detrimental effects. 3. The particles must be dispersed either “naturally” (as colloidal size particles) or mechanically (stirring, agitation) in order to contact physically the surface being coated. 4. The particle size is usually in the colloidal range (~0.005 - 0.2 ∝m) or slightly larger, usually less than 0.5 - 1.0 ∝m although there are exceptions for certain applications. The possibilities are numerous. Satin nickel deposits[93] were developed to reduce glare on automotive trim, also providing improved corrosion characteristics. Kilgore[94] described various applications including: (a) non-galling Ni deposits containing 1000 mesh silicon carbide for pistons and cylinder walls on internal combustion engines, (b) inclusion of Cr in Ni deposits producing nichrome by subsequent heat-treatment, (c) 120 grit diamond dust in nickel to produce permanent abrasive grinding wheels. The hardness and wearability of Cd deposits from acid baths were improved by inclusion of corundum or boron carbide particles.[95] An important, desired result of dispersion plating is the improved strength, hardness, creep and other properties of the deposit, including the retention of strength after thermal treatments. Sautter[96] reported increased yield strength from 8 kg/mm2 (11,375 psi) for pure Ni deposits to 35 kg/mm2 (50,000 psi) for dispersed alloys containing 3.5 - 6.0 volume percent (v/o) Al2 O3 ; the particle size ranged from 0.01 - 0.04 ∝m to 0.3 ∝m and the plating parameters other than agitation had little or no effect. Electroformed lead and lead alloys were strengthened only by additions of

548


TiO 2 (0.01 - 0.03 ∝m) although Al 2O3, BaSO4, Pb3 O4 and W additions were also studied,[97] indicating the possibility of specificity with respect to the dispersoid. Greco and Baldauf[98] found 2 - 15% of Al2O3 to be the effective range for dispersion-hardening of Ni deposits from a sulfamate bath. The increase in hardness appeared to be linear to the square root of the volume fraction of the dispersoid with Al2 O3 showing a higher slope than TiO2 . The deposits contained three times (v/o) more TiO2 than Al2O3 at the same solution concentrations and plating conditions; the particle size averaged 0.074 ∝m Al 2O3 (0.013 - 0.339 ∝m range) and 0.2 ∝m TiO2 (0.037 - 0.313 ∝m range). Table 10.6 indicates the variations in mechanical properties of particledispersed nickel alloys due to the dispersoid material and the plating solution composition. Electrophoresis is the term used to describe the migration, by virtue of the electric charge on their surfaces, of colloidal or near-colloidal particles in a suspending medium when a potential is applied. This migration is analogous to ionic migration through a solution. The electrical double layer of charges discussed above is involved. The process has been applied to the deposition of a variety of materials including metal powders, oxides, cermets and other particles to metal substrates. Usually the particles ranging in size between 0.5 and 45 ∝m are suspended in a non-conducting (or poorly conducting) medium and a high potential (50 - 1000 V) is applied to the electrodes. High rates of deposition are obtained and coating thicknesses can be varied by controlling voltage, electrode spacing, suspension concentration and time. The coating is air dried and baked to remove the solvent medium. The coating is nonadherent and must be processed further by compression and/or sintering or by subsequent electrodeposition to bond it to the substrate. Electrophoretic deposition has been applied to produce Ni, Ni-Cr, Ni-Cr-Fe coatings to base metals as well as inclusion of such dispersoids as molybdenum disulfide or silicon carbide.[99] Ortner[100] applied electrophoretic deposition of TaC-Fe-Ni coatings onto graphite, sintered at 2300oC (4170oF) for the protection of rocket nozzle inserts and oxidation resistant coatings for refractory alloys. A mechanical method of applying a coating involves peening soft metals (Cd, Zn) and alloys onto a substrate with glass beads in an aqueous medium in a tumbling operation. The equipment is similar to a cement mixer. “Alloys” of Cd-Sn deposited in this manner exceeded two thousand hours in salt fog corrosion tests.

Table 10.6. Mechanical Property Data for Nickel-Particle Composites[122]

Deposition from Aqueous Solutions 549

550


Alloy deposits, however produced, offer certain advantages over single metal deposits: 1. Increased corrosion resistance due to greater density and finer grain structure. 2. Combination of properties of the individual constituents. 3. New properties, unlike the individual constituents. 4. “Tailor-made” properties by proper selection of the constituents. The limitations include the greater control required, the difficulty of reproducing the alloy composition, the greater attention to the anode systems used and their effects on the solution constituents and complexes.

6.0 SELECTED SPECIAL PROCESSES 6.1 Electroless Deposition Electroless plating processes differ from electroplating processes in that no external current source is required. Metal coatings are produced by chemical reduction with electrons supplied by a reducing agent (R.A.) present in the solution: catalytic Eq. (24)

M+n

+

ne-

(supplied by R.A.)

→

Mo (+reaction products)

surface The uniqueness of the process is that the reduction is catalyzed by certain metals immersed in the solution and proceeds in a controlled manner on the substrate’s surface. The deposit itself continues to catalyze the reduction reaction so that the deposition process becomes self-sustaining or autocatalytic. These features permit the deposition of relatively thick deposits. Thus the process is differentiated from other types of chemical reduction: (a) simple immersion or displacement reactions in which deposition ceases when equilibrium between the coating and the solution is established (e.g. copper immersion on steel from copper sulfate solutions), and (b) homogeneous reduction where deposition occurs over all surfaces in contact with the solution (e.g. silvering-mirroring). To prevent spontaneous reduction (decomposition), other chemicals are present; these are generally organic complexing agents and buffering agents. Other additives provide special functions as in electroplating solutions: additional stabilizers, brighteners, stress relievers.


551

The reducing agents most widely used are: Sodium hypophosphite (for Ni, Co) Sodium borohydride (for Ni, Au) Dimethylamineborane (or other substituted amine boranes) for Ni, Co, Au, Cu, Ag) Hydrazine (for Ni, Au, Pd) Formaldehyde (for Cu) The process was reported by Brenner and Riddell[101] in 1946 for nickel and cobalt coatings and has enjoyed very active interest since, resulting in extension* to electroless plating of copper, gold, palladium, platinum, silver and a variety of alloys involving one or more of these metals. Comprehensive reviews[102]-[108][175] with extensive bibliographies cover the considerable technology, solution composition and operating conditions, and literature (including patent) which have accumulated. Representative solution formulations are given in Appendix B. Nickel deposits produced with hypophosphite or the boron-containing reducing agents are alloys containing the element P or B. They are very fine polycrystalline supersaturated solid solutions or amorphous metastable alloys[109]-[111] with hardness ranging approximately 500 - 650 VPN and can be precipitation hardened, being converted to crystalline nickel and nickel phosphide (Ni3P) or boride (Ni3 B). Maximum hardness ranging from 900 1100 VPN is obtained at 400o C (750oF) for 1 hour (Fig. 10.13). The effects of heat treatment at various times and temperature on the hardness of electroless Ni-P have been extensively investigated.[109][112]-[114] Johnson and Ogburn[115] supplement more fully previous work, showing the influence of phosphorus contents and the specific heat treatments on the range of hardness obtained (Fig. 10.14). Higgs[116] investigated the effects of heat treatments on the hardness and structure of the deposits reporting the presence of several NixPy compounds present other than the usually reported Ni3P. Alloys containing more than 7 wt. % P do not exhibit ferromagnetism in the as-plated condition. Schwartz and Mallory[117] found differences in the increasing ferromagnetism of alloys from various solutions as a result of heat treatments. The phosphorus content of the deposit increases as the hypophosphite concentration increases and the pH decreases in the solution. The boron * These are commercially available. Other electroless processes for iron, chromium, cadmium, and tin have been reported but either not confirmed or commercially applied. Undoubtedly, new developments will continue to be reported.

552


Figure 10.13. Hardness of electroless Ni alloys as a result of heat treatments (1 hr).

Figure 10.14. Hardness of Ni-P alloy: $ as-plated, X after 8 hrs at 200°C, O after ½ hrs at 400°C.[115]


553

content in Ni-B systems is generally similar. The complexing agents in the solution influence deposition rate[118] (along with pH) and may also have an effect on the as-plated deposit; Mallory[119] related differences in salt fog corrosion tests to this factor. It appears that the properties of the deposit may vary considerably depending on the phosphorus content which, in turn, is determined by the solution used and its operating pH. Graham et al.[110] observed abrupt changes in structure, strength and ductility of deposits at a phosphorus content of about 7 w/o with both strength and ductility increasing with increasing phosphorus content. They also observed that the lamellar banded structure was 10 times broader (5 ∝m ≡ 0.2 mils) in deposits from alkaline solutions than in acid solution deposits (0.5 ∝m ≡ 0.02 mils). Parker and Shah[120] determined that the stress in electroless Ni-P alloys varies from tensile to compressive as the phosphorus content of the deposit increases. They also observed variations in stress depending on the thermal expansion coefficient of the substrate. However, increased thickness reduced the stress on most substrates. Baldwin and Such[121] indicated that zero stress can be obtained by adjustment of solution pH and that any desired value between 11.25 kg/mm2 (16,000 psi) (tensile) and 5.6 kg/mm2 (8,000 psi) (compressive) is achievable; maximum ductility was obtained with a 5.5 w/o P alloy from a solution at pH 5.6 ± 0.2. The least wear of hardened electroless NiP vs. quenched annealed steel was obtained with deposits containing 8 - 12 w/ o P and the maximum and minimum values of average friction coefficient were 0.43 and 0.57, respectively, compared to 0.63 - 0.64 for pure nickel.[114] Thus, it is evident that the compositions, structures and properties of electroless deposits can vary widely and are dependent on many factors. Safranek[122] reviewed those for electroless nickel and cobalt, and Okinaka[123] those for electroless gold. Saubestre [124] studied various reducing agents for electroless copper, concluding that formaldehyde was the most suitable. He also studied the effects of inhibitors or stabilizers to extend the useful life of the solution. [125] The costs of the complexing and reducing agents used in electroless plating solutions make them non-competitive with electroplating processes. The application of electroless plating is usually based on one or more of the following advantages over electroplating: 1. Deposits are very uniform without excessive build-up on corners or projections or insufficient thickness in recessed areas. Internal surfaces are also evenly coated. The uniformity is limited only by the ability of the solution to contact the surface and be replenished at the surface.

554

Deposition Technologies for Films and Coatings 2. Deposits are usually less porous and more corrosion resistant than electroplated deposits (of equal thickness). 3. Almost any metallic or non-metallic, non-conducting surfaces, including polymers (plastics), ceramics, glasses can be plated. Those materials which are not catalytic (to the reaction) can be made catalytic by suitable sensitizing and nucleation treatments (see Sec. 6.4, Plating on Plastics). 4. Electrical contacts are not required. 5. The deposits have unique chemical, mechanical, physical and magnetic properties.

The disadvantages of electroless plating compared to electroplating include; 1. 2 3. 4. 5.

Solution instability More expensive Slower deposition rates Frequent replacement of tanks or liners Greater and more frequent control for reproducible deposits.

Properties and Trends of Electroless Nickel Deposits. Tensile strength increases from 40 to 60 kg/mm2 for deposits containing 5 - 7% P to as high as 85 kg/mm2 for deposits containing more than 9% P. Ductility also increases with increasing P content but decreases with increasing hardness. Ductility is reduced ~75% by heat treatment at 400oC (750oF). Severe strain or impact results in cracking with no plastic deformation. Heat treatment above 250oC (480oF) causes recrystallization and precipitation of Ni3 P or Ni3B and other phases in a Ni matrix, resulting in increased hardness. Figures 10.13 and 10.14 show the effect of heat treatment on hardness of the deposits. Heating at 400oC (750o F) for 1 hour produces maximum hardness for most compositions. ≤1500 angstroms) in High internal stress is reported for thin deposits (≤ the range of 28 - 35 kg/mm2 , tensile. Stress values for thicker deposits vary considerably, from -10.8 (compressive) to +15 kg/mm2 (tensile). High tensively stressed deposits usually contain ≤7% P. Generally, stress values vary inversely with the phosphorus content. Deposits containing ≥9% P are usually compressively stressed. Heat treatment increases stress tensively, compressively stressed deposits becoming tensively stressed even after heating at 200oC (390oF) or less for several hours. Wear test data vary greatly since many factors are involved, making comparisons and trends difficult. Some of the variables include: type of


555

solution and operating conditions, % P and thickness of deposit, heat treatment temperature and time, and the kind of wear test used. Figures 10.15 and 10.16 indicate some trends in the wear resistance of electroless Ni coatings. Abrasion resistance as measured by the Taber Abraser Test indicates that heat treatment improves resistance. However, the hardest deposits do not necessarily provide the greatest abrasion resistance. Generally, NiB deposits are superior to NiP deposits. The excellent wear resistance of electroless nickel coatings may be due, in part, to the presence (and amount) of P which may improve the (dry) lubricity of the coating and prevent seizure or galling except at high loads or sharp impact conditions. However, the deposits do not break off as discrete particles under heavy loads as do chromium deposits, the latter causing excessive scoring. Although electroless nickel and chromium deposits have similar hardness ranges, they perform well as a wearing combination. The coefficient of (dry) friction for NiP varies from approximately 0.3 against grey iron, to 0.38 against steel, to 0.43 against chromium, with only slight differences due to phosphorus content or heat treatment. NiB deposits generally show higher values than NiP deposits.

Figure 10.15. Wear of electroless Ni in Taber Wear Test. Adapted from Ma and Gawne.[126]

556


Figure 10.16. Wear of electroless Ni (8.5% P, 40 ∝m) Falex Test: EN Plated Pin Unplated V-grooved Blocks. Adapted from Ma and Gawne.[126]


557

Composite Electroless Nickel Coatings. Electroless nickel coatings are readily produced containing dispersed inert particles from highly stabilized solutions (to minimize solution decomposition) in which the particles are mechanically dispersed. Parts are usually rotated to obtain uniform particle distribution. The particles are physically entrapped and not co-deposited. Particle size range from 0.5 to 10∝m. Hard particles such as diamond, boron carbide, silicon carbide, tungsten carbide, titanium carbide, aluminum oxide, and chromium have been used to produce composite coatings with the particles constituting up to 30 volume % of the coatings. Applications include metal forming dies, molds for plastics, oil well equipment, textile (yarn) spinning equipment, and friction disks. The asplated coatings are rough and dull but can be polished and lapped to provide smooth, semi-bright finishes. Heat treatment increases hardness and wear resistance as it does for deposits without particles. However, such treatments should not exceed 400o C (750o F) for composites containing carbides since nickel carbide is produced and hardness and wear resistance are greatly reduced. Other applications involve the incorporation of soft or polymeric particles such as PTFE (polytetrafluoroethylene) into the electroless nickel deposits. These provide excellent lubricating characteristics, wear resistance and corrosion resistance. PTFE composites containing between 18 and 25 volume % are commercially produced for many wear, mold release, and corrosion resistance applications.[127] Figure 10.17 shows the reduced wear resistance of PTFE-composite coating over conventional electroless nickel with extended testing. Parker[128] has compiled literature data on hardness, wear resistance, coefficient of friction, stress and other properties and characteristics of electroless nickel coatings with and without dispersed particles. 6.2 Electroforming Electroforming is defined[129] as the “production or reproduction of articles by electrodeposition upon a mandrel or mold that is subsequently separated from the deposit.” (Occasionally the mandrel may remain in whole or in part as an integral functional part of the electroform.) The mandrels used are classified as permanent or expendable. The choice, composition, design considerations, preparation cycles, and methods of removal of mandrels are probably the most vital aspects of electroforming.[129]-[132] Various types of mandrels are given in Table 10.7.

558


Figure 10.17. Taber wear test on as-plated PTFE-electroless nickel composite coatings.[127]

Since the electrodeposits, called electroforms, are used as separate structures, they are usually substantially thicker than plated coatings. The fixturing or tooling of the mandrel and the anode positioning are quite critical. These determine the current distribution and resulting thicknesses of the deposit. A wide range of current densities produces changes in the structure, concentration of impurities and properties of the deposit which, in view of the function as an electroform, now are of paramount interest. Braddock and Harris[133] reported increases in carbon content of nickel deposits from 0.004 w/o to 0.008 w/o and sulfur contents from 0.0002 w/o to 0.0014 w/o when the C.D. was reduced from “normal” (537 A/m2 ≡ 50 A/ ft2 ) to very low C.D. (21.5 A/m 2 ≡ 2 A/ft 2 ). Dini et al.[134][135] discussed the effects of variations in carbon and sulfur contents of nickel electroforms from sulfamate solutions; these are shown in Figs. 10.18 and 10.19. Sulfur in nickel deposits causes embrittlement and cracking, limiting high temperature applications (370oC ≡ 700o F max.) (Fig. 10.20). Nickel, copper and iron are the most widely used electroforming deposits. Knowledge of the solution compositions, operating conditions and resulting structures and properties of deposits makes it possible to specify a given solution and the desired results. These are tabulated by DiBari.[130]


559

Table 10.7. Comparison of Mandrel Materials (from Spencer, Ref. 131) Type Permanent

Material

Advantages

Carbon steel

Availability, low cost

Attacked by some plating solutions, such as acid copper and hot ferrous chloride.

Carbon steel, chromium or silver plated

Improved hardness and/ or corrosion resistance. Coating may be stripped and renewed.

Chromium coatings may be pitted by hot chloride type baths.

Stainless steel

Inert to most plating solutions.

Costly. Soft surface of non-hardenable types is easily scratched.

Inconel

Natural oxide film prevents adhesion of most deposits.

Invar Kovar

Low temperature coefficient of expansion facilitates removal from electroform. Nonadherent.

Brass, Ni Cr, Ag Good machinability, low cost. plated

Soluble

Costly, poor machinability.

Surface easily scratched.

Glass, Quartz

Close tolerance, high finish.

Costly, fragile and requires a conductive coating.

Wood, plaster, plastic, etc.

Low cost. Moldable. Flexible types can be withdrawn from undercuts.

Large tolerances. Requires a conductive coating and/or sealing.

Aluminum

Good machinability. Good finish. Close tolerances can be held in complex non-withdrawable shapes. Soluble in sodium hydroxide.

Costly. Surface easily scratched.

Zinc and zinc base Can be die-cast alloys Plastics

Fusible

Disadvantages

Moldable. Low cost. Fairly close tolerances.

Low melting alloys Can be cast at low cost. (Pb-Sn-Bi types) Waxes

Can be cast or molded at low cost.

Acid stripping solution more likely to attack electroform than caustic solution used for dissolving aluminum. Cannot be used in hot plating baths. May swell in some baths. Requires conductive coating. Difficult to remove from electroform completely. Easily scratched. May deform by creep. Requires a conductive coating.

560


Electroforming with fiber re-inforced composites and methods used are described by Withers and Abrams[136] and Wallace and Greco.[137] Representative fibers or filaments included tungsten, boron, carbides and borides. Greco et al.[138] investigated the bond strength characteristics of electrodeposited nickel on boron and silicon carbide filaments. Electroforming is very costly and is a very slow method for producing parts. It finds application when: 1. Producing parts by mechanical or other means is unusually difficult or costly. 2. Extremely close dimensions and tolerances must be held, especially on internal dimensions or surfaces with irregular contours. 3. Very fine reproduction of surface details is required. 4. Thin walls are required. 5. Unusual physical, chemical and/or mechanical properties are required in the part. 6.3 Anodizing Anodizing is an electrochemical process in which the part is made the anodic (positive) electrode in a suitable electrolyte. Sufficiently high voltage is deliberately applied to establish the desired polarization to deposit oxygen at the surface (O2 overvoltage). The metal surfaces or ions react with the oxygen to produce adherent, oxide coatings, distinguishing the process from electrobrightening or electropolishing processes. Industrial anodizing processes are confined mainly to aluminum and to a much lesser extent to magnesium and titanium alloys. Anodized tantalum is used in capacitors. Anodic coating applications include: 1. Protection corrosion, wear and abrasion resistance. 2. Decorative clear coatings on polished or brightened surfaces, dyed (color) coatings. 3. Base for subsequent paint or organic coating. 4. Base for plating an aluminum. 5. Special based on some specific property or the coating, e.g., thermal barrier films, refractory films, electrolytic condensers, capacitors (dielectric films). Anodizing of aluminum has been investigated intensively. Wernick and Pinner[139] definitively discuss the various processes and the nature and properties of the oxide coatings produced.


561

Figure 10.18. Influence of carbon on tensile strength: full curve, all data[134]

Figure 10.19. Influence of sulfur content on impact strength of electroformed nickel.[135]

562


Figure 10.20. Fracture surface of part with sulfur content varying from 88 to 210 ppm (x 1000).[135]

The anodic films are classified according to the solvent action of the electrolyte. The films produced in sulfuric or chromic acids are porous type films. Phosphoric acid has even greater solvent action, resulting in oxides with a greater degree of porosity; these coatings are used for adhesive bonding and for plating on aluminum processes to provide deposit adhesion by mechanical locking in the enlarged pores. On the other hand, less aggressive mild electrolytes such as tartaric acid, ammonium tartrate, boric acid, borate compounds, citric acid, etc., have little or no ability to attack the anodic oxide. These films are essentially non-porous and thin (approximately 0.5 ∝m = 20 ∝in) and are considered barrier type coatings. Due to their unique electrical characteristics, the barrier type films are used for such applications as electrical capacitors; they arealso applied as protective


563

coatings (“overlays”) for vacuum deposited aluminum on precision mirrors for optical equipment. Specification MIL-A-8625 (see latest revision), used for both military and non-military applications, describes the most widely used processes and the expected requirements and tests for quality coatings. Three types of anodized coatings are called out: Type I

from chromic acid solutions

Type II

from sulfuric acid solutions

Type III

from cold sulfuric acid processes (plus additives), producing thicker deposits (12.7 - 127 µm) (0.5 - 5.0 mils). primarily for wear and abrasion resistance. (Table 10.8 presents the most widely used processes in the U. S. A.)

Types I and II are usually sealed with a 5% (w/v) sodium dichromate solution (Class 1) or after absorption of a dye (Class 2) with a nickel (or cobalt) acetate solution. Typical processing cycles are illustrated in Figure 10.21. The advantages and limitations of these three types of anodizing processes are analyzed in Appendix C. ASTM Specification B 580-73 designates seven types of anodizing: Type A. B. C. D. E. F. G.

Description Hard Coat Architectural, Class I Architectural, Class II Automotive - exterior Interior - Moderate Abrasion Interior - Limited Abrasion Chromic Acid

Minimum Thickness µm mils 50 2.0 17.5 0.7 10 0.4 7.5 0.3 5.0 0.2 2.5 0.1 1.2 0.05

The chemical composition of unsealed sulfuric acid anodized films is approximately: 80% aluminum oxide 18% aluminum sulfate 2% water* + traces of alloying elements The coatings can probably be considered as approximating 2Al2O3·H2O and after sealing convert to Al2O3·H2 O with accompanying increased volume, providing enhanced corrosion resistance. Hot sealing reduces the hardness of the coating as much as 40%. * The water content may vary between 1 - 6 %, probably entrapped.

564


Rack

Clean Etch

Brightening Mechanical

Alkaline Acid

Type I

Cold Rinse

Chemical Electrochemical

Anodize

CrO3 — 3 - 10 w/o 0.15 - 0.5 A/dm2 (1.5 - 5 A/ft2) 40 (-50) v 32 - 50oC (90 - 120 oF)

Type II H2SO4 — 10 - 20 w/o 1 - 2 A/dm2 (10 - 20 A/ft2) 12 - 20 v 21 - 30oC (70 - 85oF)

Cold Rinse Special Sealing Dye Cold Rinse

Hot Rinse Seal

Dye Sealing Ni or Co acetate (4 - 12 g/l)

Hot Rinse

Dry Unrack

Dichromate — 5 w/o or Silicate or Other (98 - 100 oC, 208 - 212 oF)

Hot Rinse

Figure 10.21. Anodizing Aluminum – Sequence of Operations


565

Table 10.8: Aluminum Hard Anodizing Processes[141] ________________________________________________________________________________________________________

Conditions

Alumilite

Martin

Hardas

Sanford

________________________________________________________________________________________________________

Solution Composition

12 w/o Sulfuric Acid + 1/w/o Oxalic Acid

Temperature (oC)

15 w/o Sulfuric Acid, saturated with CO2

Sulfuric, Oxalic Acids

Sulfuric, Organic Acids

9 - 11

-3.9 - 0

0

4

2.7 - 3.2

10.8 - 32.4

1.3 - 1.6

10 - 60 (or higher)

10 - 75

DC or DC/AC in various proportions

15 - 150

Film Growth Rate

25 µm/hr

25.4 µm /40 min

Alloy Limitations

4% Cu 7% Si 7 - 9% Cu+Si

5% Cu

C.D. (A/dm2) Voltage*

25.4 µm /5-10 min

-17.8 to -9.5

25.4 µm /10-20 min

?

?

* At a film thickness of approximately 50 µm, voltage requirement is approximately 40 - 45 volts. __________

__________

Typical Properties of Hard Anodize Coatings [140] Hardness: usually ranges between 350 - 450 DPH (35 - 55 Rc). Abrasion resistance (Taber): 30,000 - 40,000 cycles/µm. Porosity: 5 - 15% Heat resistance: to approximately 400 oC (750oF). Break-through voltage: 7 - 10 v/µm. ____________________________________________________________________________________________________________

The oxide coating consists of two different structures: an inner (nonporous) barrier or dense structure, and an outer, thicker, porous cell-like hexagonal structure.[142] The barrier layer is approximately 250 angstroms thick and constitutes about 1 to 2% of the total anodic film thickness. The pore diameter may range from 100 angstroms to 300 angstroms, depending on the electrolyte, operating temperature, and voltage. The porosity of the coating is very high; a coating which exhibits 15% porosity contains approximately 62 x 109 pores/cm2 (400 x 10 9 pores/in2 ). Spooner[143] emphasized the importance of sealing methods, operating conditions (temperature and time), water quality, and the detrimental effects of contaminants in the water seal on the quality of the sealed coating,

566


especially corrosion resistance. The suggested maximum contaminant levels in the sealing solution are: Sulfate (SO4)= 250 ppm Chloride (Cl)100 ppm = Silicates (SiO3 ) 10 ppm Phosphates (PO4)= 5 ppm Fluorides (F) 5 ppm In the 1980’s, cold sealing (temp. 20 - 30oC) processes, claimed to be equivalent to conventional hot sealing, were developed. These processes are based on heavy metal salts (e.g., Ni), fluorides or silicates in water/ various alcohol mixtures. Apparently, the pores are sealed by “plugging” with precipitated compounds; the term “impregnation” is considered more appropriate by some. Wernick[144] reviewed the development of cold sealing processes; Short and Morita[145] discussed the mechanism(s) involved. Since the oxide film is a growth film at the expense of the aluminum substrate (and not simply an add-on-film as in electrodeposition) the dimensional changes depend on the equilibrium set up between film growth and the dissolving action of the electrolyte. For Type I and II films, it may be assumed that the dimensional increase per surface is about one-third the actual thickness of the film. For Type III, Hard Anodized coatings, the dimensional increase per surface is about one-half the actual oxide thickness. Thus, stripping and re-anodizing would require approximately twice the original film thickness to meet the same dimensional requirements. This could present serious problems in salvaging rejected parts. The wear resistance of Hard Anodized coatings may vary significantly with coating thickness and alloy composition. George and Powers[146] proposed a more concentrated modified Alumilite (Alcoa’s Hard Anodizing process) solution which appeared to provide improved wear characteristics for difficult-to-coat alloys. Some of the trends of the effects of operating conditions on the properties of the coatings are summarized in Table 10.9. The following observations are noted: 1. Recesses of parts receive lower current densities (at least initially) resulting in softer coatings. 2. Conversely, projecting surfaces, especially sharp corners, receive higher current densities which produce harder coatings, resulting in cracking. 3. Cracking can occur at either concave or convex corners due to stresses.


567

Table 10.9. Effect of Operating Conditions on Anodic Film Characteristics (from Wernick and Pinner, Ref. 147) __________________________________________________________________________________________________________________

Condition

Limiting Film Thickness

Hardness*

Corrosion Resistance

Adhesion/ Porosity Dye Absorption

_________________________________________________________________________________________________________

Temperature increased

'

'

"

(

(

Current Density increased

(

(

"

'

'

Anodizing time increased

'

'

(

'

(

Acid concentration increased

'

'

"

(

(

Use of less aggressive electrolyte

(

(

"

'

'

Alloy homogeneity increased

(

(

'

'

'

( = increases, '

= decreases,

"

= passes through a maximum

* Hardness of sealed coatings is approximately 60% of unsealed coatings. Sealing time also affects hardness, inversely; increased sealing time results in decreased hardness.

Notes: − Effects on hardness and dye absorption ability of coating are usually in opposite directions. − Voltage requirements increase for all above conditions.

568

Deposition Technologies for Films and Coatings 4. Coatings which grow laterally as the dielectric film spreads are softer than coatings formed rapidly. 5. Properties of the coatings are influenced by the geometry of the parts as well as the alloying constituents or the electrolyte and its operating conditions.

Additions of certain organic acids* to sulfuric acid anodizing solutions produce integral colored anodized coatings, ranging from a light bronze or gold to black. These have been used in architectural applications.[148] Another approach to coloring anodized coatings involves a 2-step process.[149] After the anodizing step, the parts are immersed in a solution containing nickel or tin salts and after one minute immersion, current is applied at 10 to 18 volts. The desired colors are produced by varying either the time (voltage constant) or the voltage (time constant). The colors produced range from light bronze (in 10 - 15 sec) to black (in 15 min). The advantages over the more widely used organic dyed coatings include better light-fastness and better protection since the precipitated inorganic deposits are at the base of the pores prior to subsequent hot sealing. Pulsed current modifications have been applied to both conventional and hard anodizing.[150] Superior coatings produced at slightly lower voltages in shorter times are claimed for more alloys. Konno[151] reviewed these processes for aluminum, magnesium, and zinc. The anodizing of magnesium alloys has not found extensive use, possibly because it is somewhat more difficult than anodizing aluminum. Magnesium oxide (MgO) is more water-soluble and considerably softer than aluminum oxide (Al2 O3 ). The anodizing processes are similar and sealing is also required. The primary purpose is as a preparatory coating for painting or for corrosion and abrasion resistance. The older processes are referred to as Dow 12, Dow 14, and Manodyzing; these are AC or DC low voltage processes. The “newer” processes are fluoride-containing solutions and include Dow 17, CR 22 and HAE (Hardcoat). These are high voltage (from 80 V up to 320 V) processes. CR22 and HAE processes require alternating current. Solution formulations and operating conditions can be found in the referenced Handbooks.[107][130][152] Titanium and its alloys are anodized to provide: 1. Protection from galvanic corrosion when assembled or in contact with dissimilar metals by reducing or minimizing potential differences. * Sulfosalicylic acid, Kaiser Aluminum & Chemical Company, U.S. Patent 3,031,387 (April 24, 1962) and Sulfophthallic acid, Aluminum Company of America, U.S. Patent 3,277,639 (June 4, 1966).


569

2. Anti-galling, anti-fretting properties to the surfaces of parts in moving assemblies. 3. Part identification using a range of integral colors produced by the particular anodizing process. Both acid and alkali solutions have been used. Table 10.10 indicates typical solution formulations and operating conditions. The colors produced due to variations in current densities and voltages are also indicated.

Table 10.10. Representative Titanium Anodizing Formulations and Operating Conditions (153) ____________________________________________________________________________________________________________

Composition g/L

H2SO4 150 - 180

H2SO4 - 100 H3PO4 - 800

NaOH 50

Temperature, oC

18 - 24

20

90 - 95

Current Density (A/dm2)

0.2 - 0.4

3.0 - 5.0

Voltage

20 - 25

30 - 110

35 - 40

Color Range

Blue to Blue-Violet

Blue to Opaque Grey

Dull Grey

5

____________ Colors Produced on Pure Titanium Anodized in 15% H2SO4 Color

C.D. (A/dm2)

Yellow

0.15

Violet to 0.3 Blue-Violet Dull Blue

Volts 5 - 12 13 - 22

0.75

23 - 30

____________________________________________________________________________________________________________

570


6.4 Plating on Plastics Commercial plating on plastics became feasible with the development −especially the low temperature electroless of electroless plating processes− nickel and copper processes. Large scale, high production automatic decorative (Cu/Ni/Cr) plating on plastics is increasing on automotive trim, houseware and other articles. The technology for manufacturing printed circuit (PC) boards is another development of electroless plating. This discussion is limited to these developments. The plastics most widely plated today for decorative applications are (in decreasing order and increasing difficulty): Acrylonitrile-butadiene-styrene (ABS), polyphenylene-butadiene-stryene (Noryl), polysulfones, polypropylenes, nylons and polytetrafluoroethylene. It should be noted that most of these plastics are “filled”, i.e., they contain mineral fillers, additives, modifiers, or are co-polymers or mixtures of co-polymers. ABS is a mixture of acrylonitrile-styrene and butadiene-styrene; polysulfones generally contain ABS; nylons are mineral-filled. In the etching step, one or more of the components is selectively etched, providing a non-uniformly roughened surface for improved mechanical bonding of the deposits (with possible chemical bonding). The plating cycle[154]-[156] for decorative coatings (may) include: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

−mechanically or chemically Surface deglazing− Etch Neutralize Activation (catalyze) “Accelerate” (remove residual tin hydroxide) Electroless Deposit (Cu or Ni) Racking (if above steps done in bulk) Clean (if required) Strike (if required) Plate: Bright acid copper Bright nickel or dual nickel Chromium, microcracked preferred (Note: Rinses are critical between various steps.)

The etch step is a critical one. Usually chromic acid, either supersaturated*, or mixed with sulfuric acid or with sulfuric-phosphoric acids[156] is employed. An alternative etching technique involves the use of plasmas.[157] * L. Kadison, U.S. Patent 3,668,130 (June 6, 1972), (assigned Crown City Plating Company).


571

The adhesion of the deposits is also related to the activation of the surface. The most widely used system is the stannous chloride (SnCl2)/palladium chloride (PdCl2) 1- or 2-step treatment based on the redox reaction: Eq. 25

Sn+2 + Pd+2 → Sn+4 + Pd0

The 2-step activation involves first adsorption of SnCl2 on the etched surface, followed by the redox reaction in a solution of PdCl2. The 1-step or mixed catalyst system includes both components and is considered either a complexed chloride of Sn and Pd or a colloidal mixture. There is considerable controversy regarding the nature of the system.[158] Perrins[156] determined that the adhesion of electroless nickel and electroless copper (on polypropylene co-polymers) was dependent on the amount of palladium deposited. Low palladium gave high copper adhesion and low nickel adhesion. High palladium initially gave low adhesion to both which improved, peaking, with 3 - 5 week aging. Accelerated aging at 70oC for 1.5 hours gave a 70% improvement over control values. (Improved adhesion by heating is also found with other plastics.) An oxidation mechanism at the deposit/polymer interface is suggested as being responsible for increased adhesion. Selective plating of plastics[159] can be accomplished by applying an organic stop-off which remains on the surface as a finish coat. Deposition is prevented on the stop-off film by use of a chromating treatment after etching. 6.5 Plating Printed Circuit Boards The printed circuit board (PCB), also called printed wiring board, has made rapid advances since its development in the late 1930’s. It is a pre-determined electrical conducting design or path, on a non-conductive base, whose primary function is to carry an electronic impulse or signal. The non-conductive base or board can be made of a wide variety of materials including wood, masonite, or resins such as epoxy, epoxy-glass, phenolics (flame-retardant or paperreinforced), polybutadienes, polyimides, and ceramics. Presently, the most popular materials seem to be phenolics, epoxies (and glass), and polyimides. The types of PC boards fabricated today include: 1. Print-and-Etch*

* Print-and-Etch involves no plating. A (photo) resist is applied exposing unwanted copper (on a copper-clad board) which is etched away. Holes are drilled and eyelets inserted for connecting circuitry.

572


2. Plate-and-Etch 3. Plated-Through-Hole (PTH) a. Panel Plate b. Pattern Plate 4. Multi-Layered (MLB)* 5. Additive Circuits 6. Integrated Circuits 7. Flexible Circuits The pre-plating preparation steps involve alkaline cleaning, acid etching as do other plating cycles. Additional steps such as abrasive cleaning or honing to remove smeared polymer in the drilled holes and “etch-back” (of polymer) to expose the intermediate layers of copper in MLB’s (Fig. 10.22) are required. Also required for through-hole plating is the SnCl2/PdCl2 activation treatment discussed above. After activation, electroless copper is deposited over the exposed outer circuits and through the hole. This is followed by electrodeposited copper. Rothschild and Schwartz[160] and Smith[161] describe fabrication and plating operations. A trouble-shooting chart[162] and manual[163] identify possible sources of trouble and their rectification or suggested cures. Rothschild and Kilgore[164] discuss the problems of plate distribution(throwing power, T.P.) in MLB’s and relate T.P. to the ratio of surface to minimum hole thickness (S/H) and the ratio of total board thickness (hole length) to hole diameter (B/D). They also discuss fabrication and plating steps, the choice of deposits (Au, Sn-Pb, Sn-Ni), solderability and diffusion and/or migration problems. Copper plating is used for through-hole plating. Acid sulfate and acid fluoborate plating solutions possessing high throwing power have been developed. These are low-metal ion, high acid concentration formulations (to promote desirable polarization at high current density surfaces) with grain refiners to eliminate columnar structures which may develop cleavage planes at corners, resulting in cracking. The pyrophosphate copper solution is the other type of solution employed. (Cyanide copper formulations damage the board due to the high alkalinity and cyanide content.) Which solution is the preferred plating solution is a moot question and invokes considerable controversy among the “practitioners of the art.” However, the “high throw” bright acid copper sulfate solutions are the most widely used; they are easier to control and are more economical.

* This is similar to Plated-Through-Hole except two or more PCB's are bonded together using an epoxy/glass pre-preg. Interconnections are made by drilling holes after laminating layers. The individual layers are referred to as details or inner layers.


573

Figure 10.22. Through-Hole solderplate on multilayer printed circuit board (x 50) (Courtesy of B. F. Rothschild.)

574


A high-throw Sn-Pb, solder plate, has also been developed for throughhole plating.[165] Other electrodeposited coatings used on PCB’s include: Sn-Pb, Sn-Ni, Sn, Ni, Au with various functions as etch resists and to provide solderability, corrosion resistance, wear resistance, or low contact resistance. 7.0 STRUCTURES AND PROPERTIES OF DEPOSITS The structure and properties of a deposit are related to the deposition parameters and deposit thickness (Fig. 10.9). Changes in these parameters may produce significant differences in a deposit, making generalizations difficult, if not misleading. Some investigators have omitted indicating important plating parameters or deposit thickness or testing conditions when reporting property measurements, making these data at least suspect. Further, extrapolation from a narrow set of conditions and data could also be misleading. Figures 10.18, 10.19, and 10.23 emphasize the influence of impurities and thickness on properties of nickel from sulfamate solutions. They also illustrate the importance of taking measurements in a thickness range related to the intended application of the deposit. There appears to be a certain degree of specificity, yet trends have been established that correlate structure with deposition parameters and properties. Weil[166] reviewed how nucleation and growth, including epitaxy, twinning, and dislocations determine deposit structure and properties. A few examples and data for copper, nickel and chromium deposits are used in an attempt to illustrate these points. Since approximately 1947, the American Electroplaters’ and Surface Finishers’ Society (AESF) has initiated and supported research programs at various institutions on structure and properties of electrodeposits. These studies and other published data (about 1500 references) have been “compiled and systematized” into a single source book by Safranek.[122] Dini provides a comprehensive materials science approach relating deposition parameters to structure, texture, properties of deposits, and the interrelationship between deposits and substrates. Test methods and data to evaluate deposit properties and performance are presented.[176] Most of the data presented here are based on these sources. The structures of electrodeposits are classified as: Columnar Fibrous Fine-grained (usually equiaxed) Banded (or striated or lamellar)


575

Figure 10.23. Influence of thickness on mechanical properties of deposits. [167]

576


Columnar structures are characteristic of deposits from solutions (especially acid solutions) containing no additives, high metal ion concentration solutions at low deposition rates. They usually exhibit lower tensile strength, percent elongation and hardness than other structures; they are generally more ductile. Such deposits are usually of highest purity (high density) and low electrical resistivity. Fibrous structures represent a grain refinement of columnar structure. Stress relieving additives (such as saccharin or coumarin) promote such refinement as do high deposition rates. These may be considered intermediate in properties between columnar and fine-grained structures. Fine-grained deposits are usually obtained from complex-ion solutions (such as cyanide) or with certain addition agents. These deposits are less pure, less dense and exhibit higher electrical resistivities due to presence of foreign material. Banded structures are characteristic of bright deposits (as a result of brightening addition agents—usually S-containing organic compounds which result in small amounts of S and C in the deposit) and some alloy deposits. These deposits generally possess higher tensile strength, hardness, and internal stress and decreased ductility than other structures. The use of plating current modifications (PR, IC, pulse) favors the conversion of structure from a solution to a banded structure. Electro- and electroless deposits generally conform to the Hall-Petch relationship. Eq. 26

H (or YS) = σo + kd-1/2

where

H = YS = d = σo , k =

hardness of the deposit yield strength of the deposit grain size constants

That is, factors which decrease grain size increase hardness, yield and tensile strength of the deposit. In aqueous deposition, grain size of the deposit decreases as: ) ) ) ) ) )

Current Density increases Cathode potential increases Solution agitation increases Solution temperature decreases Metal ion concentration decreases Addition agents are added

) Complexing agents are present


577

Also, the brightness and smoothness of the deposit increase as grain size decreases. Grain size can vary widely from 100 to 50,000 angstroms; the grain size of fine-grained or banded deposits is usually between 100 and 1000 angstroms. Read[168] observed that frequently the grain size of electrodeposits is much larger than indicated by etched specimens (the metallographic procedures usually used) and that x-ray techniques are more reliable, especially for measuring larger grain sizes. As indicated previously, certain deposits, especially alloys, show no grain structure, i.e., are amorphous. Some metals (notably Cu, Ni, Co, and Au) can be deposited in all four types of grain structures depending on the solution composition and plating conditions. This is shown in Figure 10.24 for copper deposits. Typical properties of these structures are given in Table 10.11. Zentner, Brenner, and Jennings[169] (AESF Research Project No. 9) studied the structure-property relationships of nickel electrodeposits to plating solution composition and operating variables. The effect of current density, pH, temperature, and chloride content on deposit structure are shown in Figs. 10.25 - 10.28. The trends appear to be: 1. Grain structure changed from fine-grain to coarse-grain as temperature increased. 2. Significant structural changes occurred at both low and high current densities. Typical columnar structure is obtained between 2 and 25 A/dm2 (20 - 250 A/ft2) in Watts-type solutions. The structural changes at low C.D. may be explained by the increased sulfur and carbon contents of the deposit as shown in Table 10.13. Thus, low C.D. produced a banded structure similar to bright nickel deposits. 3. There is essentially no structural change in Watts-type deposits in the pH range 1 - 5. At pH’s above 5 there is a distinct change from columnar to fibrous or fine-grained which is probably due to inclusion of basic material (Ni(OH)2?). 4. Deposits from Watts solutions produced the coarsest, columnar deposits. Increasing the chloride content of the solution results in finer-grained deposits. All-sulfate (no chloride) solution showed a somewhat finer columnar structure than a Watts deposit, with some evidence of a banded structure. A good correlation was found to exist between structure and properties as shown in Fig. 10.29. Typical values of the mechanical properties of nickel deposited from various engineering electroplating solutions are given in Table 10.12.

578


Figure 10.24. Structure of copper deposits (x 500) (etchant: ferric chloride). Structures are typical for: a) acid sulfate (no A.A.); b) acid sulfate with A.A. (gelatin + phenolsulfonic acid; c) acid sulfate with brighteners or pyrophosphate solution; d) cyanide solution with PR. [22]


Figure 10.24. (Cont'd)

579

580


Figure 10.25. Effect of temperature of the plating solution on the structure of nickel deposited at 5 A/dm2 (46 A/ft2). Cross section x 250. Etchant: glacial acetic and nitric acid.[169]



581

582


Figure 10.26. Effect of current density on the structure of nickel deposited from the SIIICI solution at 55oC (131oF), and a pH of 3.0. Cross section x 250. Etchant: glacial acetic and nitric acid.[169]



583

584


Figure 10.27. Effect of the pH of the plating bath on the structure of nickel deposited from the SIIICI solution at 5 A/dm2 (46 A/ft2)and 55oC (131oF). Cross section x 250. Etchant: glacial acetic and nitric acid.[169]


585

Figure 10.28. Effect of increasing chloride content of the solution on the structure of nickel deposited at 55oC (131oF), 5 A/dm 2 (46 A/ft2)and a pH of 3.0. Cross section x 250. Etchant: glacial acetic and nitric acid.[169]

586 Deposition Technologies for Films and Coatings Table 10.11. Comparison of Structure and Properties of Copper Deposited at 4 A/dm 2 in Several Different Copper Solutions.[122] (From The Properties of Electrodeposited Metals & Alloys by W.H. Safranek, published by AESFS, 1986. Reprinted with permission.)

Table 10.12. Nickel Solutions for Heavy Plating[22]


588


Table 10.13. Results of Elemental Analysis of Nickel Electrodeposits[133]

Figure 10.29. Range and trend of physical properties of nickel deposited from 5 different types of solutions, each point is the average of the properties of 5 or more deposits obtained under various conditions of plating.[169] 1 equals a, b, c, d, e, f, OB, and OBT solutions. Bright nickel. 2 equals SICIII, C, Ac, C (-4N) solutions. Chloride nickel. 3 equals SICI solution. 4 equals S and oS solutions. 5 equals SIIICI, oSIIICI, S IIICI (-1N), NH4, Na and F solutions. Watts nickel.


589

Table 10.14. Recommended Basis Metal Hardness and Chromium-Plate Thickness for Various Applications[22]

Table 10.15. Coefficient of Friction for Various Metal Combinations[22]

590


Properties of chromium deposited under a wide variety of plating conditions and solution compositions were extensively covered by Brenner, Burkhead and Jennings.[170] The deposits especially the bright deposits are very fine-grained, as small as 10 angstroms on the basis of x-ray data. They concluded that the oxide content of the deposit had far greater influence on the properties than crystal orientation or structure. Increased plating temperature from 10oC to 100 oC caused reduction of oxygen content from ~1 w/o to~0.1 w/o. The hydrogen content of the deposit also decreases with increasing plating temperatures. The hardness of chromium is probably its most important engineering property. The oxygen content to the deposit is one of the most important factors affecting its hardness. Above 0.12 w/o O2, the hardness ranges between 850 - 1000 KHN (Knoop Hardness Number) and when below 0.12 w/o O2 , the hardness ranges from 625 to 325 KHN. However, it was noted that hardness values may fluctuate as much as 200 points KHN for the same oxygen content. It also appears that bright deposits are hardest. The hardness of chromium deposits, therefore, is probably the result of oxide inclusion, small grain size and internal stress. The hardness of the substrate along with that of the deposit is an important factor in the application for improved wear resistance of various tools (Table 10.14). In other wear applications the coefficient of friction is a factor; Table 10.15 gives values for various combinations. Different etching techniques reveal interesting structural characteristics in chromium deposits.[171] In fact, no single etchant reveals all possible features and it is advisable to use several techniques. Structures which have been observed include: fibrous texture, banded or striations associated with the crack pattern (and not found in crack-free deposits), bands delineating changes in plating variables (C.D. and temperature) during deposition. The internal stress, negative coefficient of thermal expansion (initial shrinkage) and the effect on fatigue strength of the substrate are properties (besides hardness) of interest in engineering applications. These are adequately covered in references already cited.[22][122][170] The reported stress values for chromium deposits cover a very broad range, from highly tensile to compressive in microcracked deposits (>1000 cracks/linear inch). It is influenced by the solution composition and concentration, C.D., temperature, deposit thickness and probably other factors. The high tensile stress and resulting cracking lower the fatigue* limits of substrates (primarily steel, but possibly also aluminum and titanium). * The higher the stress (in tension) of the deposit, the greater the reduction in fatigue strength.


591

No ductility was found for chromium deposits from aqueous solutions. In general, the physical properties of electrodeposits approach those of metallurgical wrought metals as the purity increases. Observations regarding the physical properties are: 1. The density is related to pores, voids and impurities in the deposit. Corrosion and high temperature characteristics can be significantly affected by low density. 2. The coefficient of thermal expansion is also affected by impurities in the deposit. Thermal properties are not too well established for electrodeposits. Most deposits expand with thermal cycling, notable exceptions being chromium and cobalt-tungsten alloys. Deposits which expand appreciably develop voids on thermal cycling and could not be considered for high temperature service since they would exhibit decreased corrosion and oxidation resistance. 3. Electrical resistivity is quite sensitive to the presence of small concentrations of impurities. Most deposits, therefore, exhibit higher values than wrought counterparts. Impurities such as oxides, sulfides, hydrates or inclusions tend to concentrate at grain boundaries especially after a thermal treatment or annealing. With respect to mechanical properties, the relationship of hardness to strength is not always similar to wrought metals where a constant relationship exists. Although the generalization that the strength of a deposit increases with hardness and ductility varies inversely with strength and hardness holds in many cases, the exceptions are too numerous to make it reliable. Other observations regarding mechanical properties of deposits are described below. Hardness. Hardness (microhardness) of the deposit is the most widely measured property (probably due to the ease of measurement). It may also be the most abused. The literature is replete with inconsistencies and contradictions. This may be due, in part, to techniques of specimen preparation, methods of measurement, differences in deposit thickness, plating solution differences, quality of deposit, inadequacy in reporting data, neglect to indicate load applied*, type and condition (hardness) of substrate, and other factors. * Hardness values should be reported with designated loads, e.g., VHN100, or KHN25, where 100 and 25 (as subscripts) represent the load in grams. Loads less than 25 grams are subject to serious errors and are undesirable due to poor reproducibility.

592

Deposition Technologies for Films and Coatings To obtain reasonably reliable microhardness measurements: 1. The deposit thickness should be at least ten times the depth of the indent. For the same load, the depth of a Knoop indent is approximately 1/7 that of a Vickers indent. 2. The distance of the indent from the substrate interface should be at least 1/2 the diagonal of the indent (the short diagonal for the Knoop indent) to minimize the “anvil” effect. 3. When taking multiple measurements on the same specimen, a transverse track should be followed with the distance between indents as in 2.

Vickers microhardness measurements are less sensitive to errors arising from elastic properties than are Knoop measurements and result in less serious errors as loads are increased. It appears that too much value is sometimes placed on hardness measurements. The assumed relationship between hardness and strength was discussed above. The same may be said to some degree for the correlation of hardness to wear resistance. The excellent wear resistant characteristics of chromium deposits are related to the low coefficient of friction (Table 10.15) as much as to hardness. The wear resistant characteristics of electroless nickel alloys is related to the presence of phosphorus (or boron) as well as to the hardness. Despite these comments, hardness measurements are useful in evaluating deposits and predicting their usefulness. They are especially useful in evaluating alloy deposits since changes in hardness reflect (possibly) changes in structure or composition of the alloy deposit. It is not unexpected that the hardness values of deposits (of the same metal) vary greatly (Fig. 10.30). Noteworthy are the great ranges reported for chromium and iron deposits and the ability of some alloy deposits to undergo precipitation hardening. Tensile strength. In many instances, the tensile strengths of deposits exceed those of annealed metallurgical counterparts (Table 10.16). The primary reason is the finer-grain structure of electrodeposits. Coarser grained or columnar structures may exhibit lower strengths. Ductility. The ductility of electrodeposits may equal metallurgical counterparts but is usually lower in the as-plated condition. Stress. The mechanism of internal (residual) stress in electrodeposits is not completely understood, but undoubtedly a distorted atomic lattice is involved. If the deposited atoms are closer together than normal lattice spacing, the tendency is for the atoms to “push” further apart, pulling on the substrate and resulting in tensile stresses. Conversely, if the depositing


593

atoms are farther apart than they should be in a normal lattice spacing, they tend to pull closer together, exerting a compressive stress on the substrate. Stress measurements are subject to variations in testing procedures and conditions and are generally not reproducible. A particular stress measuring instrument or technique is useful in controlling a plating solution and its operating conditions as well as in predicting the quality of the deposit within parameters experimentally established and observed. Weil[172] reviewed the various methods used to measure internal stress of electrodeposits and discussed the reasons for possible variances between measured values and those actually present in plated parts. He also made a comprehensive analysis of the various types of stresses encountered in electrodeposits.[173] The various mechanical methods and calculations (formulas) used to measure macrostress were examined critically (microstresses can be measured only from broadening of x-ray diffraction lines). It contained an extensive review of the literature dealing with stress, including the various theories proposed on the origin of stresses in electrodeposits. Stress Corrosion. The tensile strength and ductility and internal stress of the deposit are interrelated in determining the degree of resistance to stress corrosion cracking when deformation may be involved or anticipated. Magnetic Properties. Magnetic properties of deposits are usually restricted to ferromagnetism and characterized by B-H hysteresis loops, where H is the applied field and B the induced magnetic flux density. Magnetic materials are classified as soft or hard, depending on the value of the coercive force, Hc, which is the magnitude of H when B = O, i.e. the force required to cause random orientation to the domains. If Hc is small, the magnetic material is considered “soft”. These are generally materials which are mechanically soft, i.e., they have a low yield strength. Permalloy (80 Ni, 20 Fe) is such an alloy. If Hc is large, usually>200 oe, the material is considered a hard magnetic material, useful in fast switching computer memory components. Alloys of Co with P and other constituents are usually of this type. The saturation flux density (BS) is a physical property determined by the chemical composition of the material. The remanent flux density or retentivity (BR) and Hc are structure-sensitive properties. The composition, microstructure (grain size and orientation and defects), stress, thickness and impurities of the deposit affect these properties. Romankiw and Thompson[174] reviewed the magnetic properties and applications of plated magnetic films as well as methods of measurements.

594


(a)

(b)

Figure 10.30. Microhardness ranges.[122]

Table 10.16. Strength and Ductility Data for Electrodeposited Metals[122]


596


Epitaxial Growth. If the lattices of the substrate and deposit are similar, the substrate structure can be extended into the deposit. This is called epitaxial growth. High rates of deposition, the presence of addition agents and impurities tend to break down epitaxy. If the lattices differ, then the initial epitaxial growth shifts toward the structure of the deposit. The thickness of the epitaxial transition zone may vary from 0 to>5 µm before the deposition variables control growth. Also, certain crystal faces grow more rapidly than others, resulting in grain orientation. These factors may be significant for thin film applications such as semiconductor or magnetic applications.

8.0 SUMMARY Aqueous deposition is a complex process; the structure and properties of the resulting deposits depend on many factors (see Fig. 10.9). It is the oldest deposition technology and is receiving renewed and increasing interest. Research and new applications are providing increased understanding of electrode processes and solution chemistry with the development of new alloy and multilayered coatings and films. Electro- and electroless deposition are much more suitable than other deposition technologies for depositing films on complex geometric surfaces and into through-holes and blind recesses (vias). It has a wide and varied range of applications (see Table 10.4). Continued development of the newer techniques will undoubtedly result in further engineering and electronic applications of strip line, very high speed plating, improved selective and maskless plating. These include current modifications; laser, ultrasonic, and jet enhanced deposition; new cell designs; computer-controlled processes, solution analyses, and chemical additions. Improved and new processes and techniques to control, treat, minimize, and recycle plating solutions, wastes, and effluents are being studied and developed. These may result in near-zero discharge from plating processes and installations.


597

APPENDIX A - Preparation of Substrates for Electroplating ASTM Recommended Practices* Number

Title

B 177-68 (73)

Rec. Practice for Chromium Plating on Steel for Engineering Use

B 183-72

Rec. Practice for Preparation of Low-Carbon Steel for Electroplating

B 242-54 (71)

Rec. Practice for Preparation of High-Carbon Steel for Electroplating

B 253-73

Rec. Practice for Preparation of and Electroplating on Aluminum Alloys by the Zincate Process

B254-70

Rec. Practice for Preparation of and Electroplating on Stainless Steel

B 281-58 (72)

Rec. Practice for Preparation of Copper and Copper-Base Alloys for Electroplating

B 322-68 (73)

Rec. Practice for Cleaning Metals Prior to Electroplating

B 343-67 (72)

Rec. Practice for Preparation of Nickel for Electroplating with Nickel

B 431-69

Rec. Practice for Processing of Mandrels for Electroplating

B 450-67 (72)

Rec. Practice for Engineering Design of Electroformed Articles

B 503-69

Rec. Practice for Use of Copper and Nickel Electroplating Solutions for Electroforming

B 480-68

Rec. Practice for Preparation of Magnesium and Magnesium Alloys for Electroplating

B 481-68 (73)

Rec. Practice for Preparation of Titanium and Titanium Alloys for Electroplating

* Book of ASTM Standards, Vol. 2.05, Sec. 2, revised annually. Also approved by the American National Standards Institute.

598


B 482-68 (73)

Rec. Practice for Preparation of Tungsten and Tungsten Alloys for Electroplating

B 488-71

Spec. for Electrodeposited Coatings of Gold for Engineering Uses

B 558-72

Rec. Practice for Preparation of Nickel Alloys for Plating

B 580-73

Spec. for Anodic Oxide Coatings on Aluminum

Preparation for electroplating of less common substrates including those used in nuclear, electronic or high temperature alloys of Fe, Co, Ni or Cr usually requires activation treatments* in order to obtain satisfactory adhesion. Other techniques involve diffusion bonding with thermal treatments. Beach and Faust** and Friedman*** review procedures for light metals − and for high temperature applications including plating on refractory metals− U, Mo, W, Th, Zr, Nb and Si. For plating Cr on previously plated Cr, the following procedure has been satisfactory: 1. If Cr is oiled (due to grinding), degrease and polish lightly. Then clean in alkaline cleaner by immersion or scrubbing, or clean cathodically. 2. Provide a light etch anodically in alkaline, sulfuric or chromic acid solutions. 3. Immerse in Cr plating solution and allow parts to reach solution temperature. 4. Plate at low C.D. (77.5 mA/cm2, ~0.5 A/in2 ) to deposit only hydrogen to activate the surface, for 0.5 - 3 minutes approximately. 5. Slowly increase C.D. to (0.5 - 1.0 A/cm2, ~3 - 6 A/in2 ) for 15 - 30 seconds to guarantee coverage, then reduce to normal plating C.D. (0.15 - 0.5 A/cm2, ~1 - 3 A/in2). * See C. Levy, Proc. AES, 43, 219 (1956) for activation for Cr plating and W. W. Sellers and C. B. Sanborn, Ibid., 44, 36 (1957) for Ni and Ni alloys prior to Ni plating for detailed formulations. ** Modern Electroplating, 3rd ed., Ch. 27, 618, (F. Lowenheim, ed.), John Wiley & Sons (1974) *** Plating, 54 (No. 9), 1035 (Sept., 1967)


599

APPENDIX B - Representative Electroless Plating Solution Formulation 1.

Nickel-Phosphorus (See reference below) Nickel sulfate Sodium hypophosphite Sodium hydroxyacetate Sodium acetate Sodium citrate Sodium pyrophosphate Ammonium chloride pH Temp. oC w/o P in deposit

2.

Nickel-Boron Nickel chloride Dimethylamine borane Malonic acid pH Temp. oC

3.

Cobalt-Phosphorus Cobalt sulfate Sodium hypophosphite Socium citrate Ammonium sulfate Sodium laurylsulfate pH Temp. oC

(a)

(a)

(a)

(b)

35 g/L 10 10

35 g/L 10

30 g/L 10

25 g/L 25

10 100 50 50 4.5 - 5.5 90 - 95 7-9 (c) 30 g/L 3.5 34 5.5 77 (d) 24 g/L 20 70 40 0.1 8.5 92

4.5 - 5.5 90 - 95 7-9

9.0 - 9.5 10 - 10.5 90 - 95 25 - 75 5-7 4-6

600 4.

Deposition Technologies for Films and Coatings Copper

(e)

Copper sulfate Sodium potassium tartrate Versene T Sodium hydroxide Sodium carbonate Formaldehyde (37%)

29 g/L

Temp. oC 5.

Palladium Palladium chloride (as ammino complex) EDTA Na2 Ammonium hydroxide Hydrazine Temp. oC

6.

Gold Potassium cyanoaurate Potassium cyanide Potassium hydroxide Potassium borohydride Dimethylamine borane Temp. oC

7.

Silver Sodium silver cyanide Sodium cyanide Sodium hydroxide Dimethylamine borane (thiourea Temp. o C

142 17 42 25 167 ml/L 25 (f) 5.4 g/L 33.6 350 0.3 80 (g) 5.8 g/L 13 11.2 21.6

(g) 0.86 g/L 6.5 11.2 10.8

(g) 5.8 g/L 1.3 45 23.6

75 (h) 1.83 g/L 1.0 0.75 2.0 0.25) 55 - 65

75

85

Deposition from Aqueous Solutions 8.

Platinum Sodium platinate (Na2 Pt(OH)6) Ethylamine 10 Hydrazine (as sulfate) Sodium hydroxide Temp. oC

601

(i) 10 g/L

as required for reduction as required for pH 10 30

REFERENCES (for Appendix B) (a) (b) (c) (d)

Brenner, A. and Riddell, G., Surf. Technol., 10:81 (1980) Schwartz, M., Proc. AES, 176 (1960) Mallory, G. O., Plating, 58:319 (1971) Ransom, L. D. and Zentner, V., J. Electrochem. Soc., 111:1423 (1964) (e) Saubestre, E. B., Proc. AES, 46:264 (1959) (f) Rhoda, R. N., Tans. Inst, Met Finish, 36:82 (1959) (g) Okinaka, Y., Plating, 57:914 (1970) (h) Pearlstein, F. and Wightman, R. F., Plating, 58:1014 (1971) (i) Rhoda and Vines, U.S. Patent 3,486,928 (1969) Note: (1) Some of the above formulations are protected by U.S. Patents. Their listing here does not imply any right to infringe. (2) See Ref. 175 for additional solution formulations.

602


APPENDIX C - Comparison of Aluminum Anodizing Processes (Types I, II and III) Advantages of Type I coatings 1.

Corrosion resistance of coatings are as high (if not higher) than Type II coatings.

2.

Provide excellent bond for organic coatings.

3.

Chromic acid is a corrosion inhibitor, therefore it is not essential to assume (or provide for) complete removal from crevices, joints or recesses due to spot welding, riveting, bolting or blind holes.

4.

It has practically no effect on the fatigue strength of the part.

5.

Although thinner, less porous, and somewhat opaque due to pick up of chromate ion and alloying constituents, the coating is capable of absorbing dark dyes for Class 2 requirements.

6.

It is preferred as a maskant for selective Hard Anodize since it is less porous than Type II films, especially for assemblies with joints or recesses.

Limitations of Type I coatings 1.

A smaller increase in abrasion resistance is obtained as compared to Type II coatings due to lower thickness and structure differences.

2.

Limited to alloys containing less than 5% copper or 7% silicon.

3.

Higher voltage is required with extended time as compared to Type II coatings.

4.

Under conditions used for wrought alloys, casting alloys tend to use excessive current and “burning” may occur. In such cases, conditions might require changes to 30 - 35 volts at 90oF with compensating increase in time to obtain adequate coating thickness.

5.

Alloys in the annealed condition do not anodize satisfactorily, Heat treatable alloys should be tempered by solution heat treatment and approved aging.

Deposition from Aqueous Solutions 6.

603

Wrought and cast alloys with high alloy content (such as 7075) tend to develop thinner coatings and may behave erratically or poorer in salt spray tests.

Advantages of Type II coatings 1.

Less expensive (compared to Type I coatings) with respect to chemicals involved (and waste treatment thereof), heating and power costs, length of time to obtain required coating.

2.

More alloys can be treated

3.

Coatings are harder than Type I coatings.

4.

Coatings may be slightly more corrosion resistantafter sealing than Type I coatings (due to thicker and more porous coating).

5.

Clear coating permits dyeing with greater variety of colors.

Limitations of Type II coatings 1.

Cannot be used where possibility of solution entrapment exists, especially joints, laps or recesses since any sulfuric acid residue may be corrosive.

2.

Reduces the fatigue characteristics of the alloy.

3.

Difficult to control where small dimensional changes are desired or required since coatings grow faster and are thicker for corrosion resistant requirements as compared to Type I coatings. (Thus, Type I coatings should be considered on close tolerance parts such as threads.)

Characteristics of Hard Anodize Coatings 1.

Corrosion resistance is excellent, several thousand hours in salt spray tests have been reported (after proper sealing).

2.

Abrasion and wear resistance excellent.

3.

Chemical resistance is poor as compared to calcined aluminum oxides; will not resist alkalies or acids as well.

604


4.

Coefficient of thermal expansion is different from that of the aluminum alloys and spalling may result at temperatures above 200 - 300oC.

5.

Film crazing - As part temperature increases from formation temperature (-4 - 0°C = 25 - 32oF) to room temperature or the higher sealing temperatures (93 - 99°C = 200 - 210oF) or post honing temperatures, the coating may craze or fracture since it is tensively stressed; this phenomenon becomes aggravated as film thickness increases. Sometimes this crazing seems to disappear after aging.

6.

“Chalking”−This refers to a white film which sometimes appears on the surface after drying. It is not considered detrimental and is usually not noticed unless (or until) surface is wiped. The mechanism is not understood; it may possibly be a bleed-out phenomenon.

Effect of Alloying Elements on the Hard Coating. 1.

Thicker coatings are obtained with the purer or higher conducting alloys containing magnesium or zinc: Purer alloys EC, 1100, 3003 Al-Mg alloys 5005, 5050, 5052, 5252 Al-Mg-Si alloys 6061, 6063 Al-Zn alloys 7075

2.

Copper-containing alloys produce intermetallic compounds (after HT) which increase the ohmic resistance resulting in thinner coatings. Type III Hard Anodize is restricted to those alloys containing less than 5% Cu.

3.

High silicon-containing alloys also produce intermetallic compounds and do not anodize readily. These involve most castings which depend on reduction of the alloy’s melting point by the eutectics formed with the silicon (even less than 7% Si). The Si or silicides do not anodize, being “inert” and acting as inclusions, depending on “bridging” for continuity of coating.

4.

Since copper and silicon constituents may result in poorer coatings, a total of 7 - 9% of the combination of these two elements is usually considered as a maximum in an alloy to be hard anodized.

5.

The color of the Hard Anodize Coating reflects the alloying constituents.


605

REFERENCES 1. Dini, J. W., Plat. and Surf. Fin., 72(7):48 (July 1985); 1st Int’l SAMPE Metals Conf., 1:91 (August 1987) 2. Lindsay, J. H. and LaSala, J., Plat. and Surf. Fin., 72(2):54 (1985) 3. Lee, R. A., Proc. 29th Ann. Conf., IPC, Boston (1986) 4. Poskanzer, A. M., Plat. and Surf. Fin., 74(3):20 (1987) 5. Best, T. G. et al., Int’l Conf. on Met. Ctgs, AVS (April 1985) 6. Wagner, C., J. Electrochem. Soc., 101(4):181 (1954); Frumkin, A. N. and FLorianovich, G. M., Doklady Akad, Nauk SSSR, 86:907 (1951) 7. Schaefer, R. A. and Pochapsky, H., Proc. American Electroplaters’ Soc., 38:155 (1951) 8. Rothschild, B. F., Plat. and Surf. Fin., 66(5):70 (May 1979) 9. Foulke, D. G. and Johnson, D. C., Proc. American Electroplaters’ Soc., 50:107 (1963) 10. Hull, R. O., Proc. American Electroplaters’ Soc., 27 (1939) 11. Blum, W., Trans. Electrochem. Soc., 44:313 (1923) 12. Ibl, N.,Electrodics:Transport, (E. Yeager et al., ed.), 4:6, Plenum Press (1983) 13. Shawki, S., Hanna, F. and Hamid, Z. A., Metal Finishing, 85(12):59 (December 1987) 14. Electrode Processes, Discussions of The Faraday Society, No. 1:1947, Butterworths (1961) 15. Schaefer, R. A. and King, W., Plating, 39:487,627,769 (1952) (American Electroplaters’ Soc., Res. Proj. No. 8) 16. Conway, B. E. and Bockris, J. O’M., Plating, 46(4):371 (April 1959) 17. Damjanovic, A., Plating, 52(10):1017 (October 1965) 18. Vetter, K. J., Electrochemical Kinetics, p. 282 ff., Academic Press (1967) 19. Fischer, H., Plating, 56(11):1229 (November 1969) 20. Fischer, H., Electrodep. and Surf. Treatment, 1:239 (January 1973) 21. Nageswar, S., Electrodep. and Surf. Treatment, 3:417 (Sept/Nov 1975) 22. Modern Electroplating, 3rd edition, (F. A. Lowenheim, ed.), p. 17ff, (Sponsored by the Electrochemical Society) John Wiley & Sons (1973)

606


23. Landau, U., Electrochemistry in Industry, (U. Landau, E. Yeager, D. Dortan, eds.), p. 215, Plenum Press, NY (1982); Also, LBL-2702, PhD thesis, University of Calif., Berkeley (January 1976) 24. Read, H. J., Plating, 49(6):602 (1962) 25. Kardos, O., Plating, 61(1):61 (1974); (2):129 (1974); (3):229 (1974); (4):316 (1974) 26. Theory and Practice of Bright Electroplating, (Yu Matulis et al., eds.), Proceedings of All-Union Conference (December 1962); Akademiya of Sciences of the Lithuanian SSR. Translated from Russian by the Israel Scientific Translations, Jerusalem (1965) TT 65-50000, U.S. Dept of Commerce. 27. Vagramyan, A. T., and Solov’eva, Z. A., Technology of Electrodeposition, pp. 151-175, Robert Draper, Ltd (1961) 28. Raub, E. and Müller, K., Fundamentals of Metal Deposition, pp. 105 135, Elsevier Publ. Co. (1967) 29. Bato, K., Electrodep. and Surf. Treatment, 3(2):77 (March 1975) 30. Rich, S. R., Plating, 42(11):1407 (November 1955) 31. Roll, A., Metal Finishing, 55(9):55 (September 1957) 32. Hickman, R. G., Plating, 52(5):407 (May 1965) 33. Kapustin, A. P. and Trofimov, A. N. Electrocrystallization of Metals in an Ultrasonic Field., Translated from the Russian by the Israel Scientific Translation, (1970) TT-70-50036, U.S. Dept of Commerce 34. Walker, C. T. and Walker, R., Electrodep. and Surf. Treatment, 1(6):457 (July 1973) 35. Forbes, C. A. and Ricks, H. E., Plating, 49(2):279 (1962) 36. Walker, R. and Holt, N. S., Plat. and Surf. Fin., 67(5):92 (May 1980) 37. Ewing, D. T. and Associates, Plat. and Surf. Fin., 36:1137 (1949); 37:1157 (1950); 39:1033 (1952); 39:1342 (1952), Sponsored by the American Electroplaters’ Society, Research Project No. 5 38. Zentner, V., Proc. American Electroplaters’ Soc., 47:166 (1960) 39. Jernstedt, G. W., Proc. American Electroplaters’ Soc., 36:63 (1949); ibid., 37:151 (1950) 40. Rehrig, D. L., paper presented at American Electroplaters' Society 65th Annual Technical Conference, Washington, D.C. (June 1978) 41. Wan, C. C., Cheh, H. Y. and Linford, H. B., Plating, 61(6):559 (June 1974)


607

42. Puippe, J. C., Ibl, N., Angerer, H. and Schenk, H. J., OberflächeSurface, 20(4):77 (1979) 43. Proc. Pulse Plating Symposia., American Electroplaters and Surf. Fin. Soc., (a) Boston, MA (April 19 - 20, 1979) (b) Rosemont, IL (October 6 - 7, 1981) 44. Theory and Practice of Pulse Plating, (J. C. Puippe, F. Leamon, eds.), published by American Electroplaters and Surf. Fin. Soc., Orlando, FL (1986) 45. Cheh, H. Y., J. Electrochem. Soc., 118:551 (1971) 46. Ibl, N., Surf. Technol., 10:81 (1980) 47. Osero, N., Theory and Practice of Pulse Plating, (J. C. Puippe, F. Leamon, eds.), Ch. 13, published by American Electroplaters and Surf. Fin. Soc., Orlando, FL (1986) 48. Avila, A. J. and Brown, M. J., Plating, 57(11):1105 (1970) 49. Cheh, H. Y., Linford, H. B. and Wan, C. C.,Plat. and Surf. Fin., 64(5):66 (1977); Chey, H. Y., Andricacos, P. C. and Linford, H. B., ibid. 64(7):42 (1977); Chey, H. Y., Andricacos, P. C. and Linford, H. B., ibid. 64(9):44 (1977) 50. Mattson, E. and Bockris, J. O’M.,Trans. Faraday Soc., 55:1586 (1959) 51. Reid, F. H., Metalloberfläche, 30(10):453 (1976) 52. Raub, C. J. and Knödler, A., Gold Bulletin, 10(2):38 (April 1977) 53. Knödler, A., Theory and Practice of Pulse Plating, (J. C. Puippe, F. Leamon, eds.), Ch. 9, published by American Electroplaters and Surf. Fin. Soc., Orlando, FL (1986) 54. Hosokawa, K., Angerer, H., Puippe, J. C. and Ibl, N., Plat. and Surf. Fin., 67(10):52 (October 1980) 55. Puippe, J. C. and Ibl, N., Plat. and Surf. Fin., 67(6):68 (June 1980) 56. Dossenbach, O., Theory and Practice of Pulse Plating, (J. C. Puippe, F. Leamon, eds.), Ch. 6, published by American Electroplaters and Surf. Fin. Soc., Orlando, FL (1986) 57. Avila, A. J., Theory and Practice of Pulse Plating, (J. C. Puippe, F. Leamon, eds.), Ch. 11, published by American Electroplaters and Surf. Fin. Soc., Orlando, FL (1986) 58. Ibl, N., Surf. Technol., 10:81 (1980) 59. Puippe, J. C. and Ibl, N., J. Appl. Electrochem., 10:775 (1980)

608


60. von Gutfeld, R. J., Tynan, E. E., Melcher, R. L. and Blum, S. E., Appl. Phys. Lett., 35:651 (1979) 61. Puippe, J. C., Acosta, R. E. and von Gutfeld, R. J., J. Electrochem. Soc., 128(12):2539 (1981) 62. von Gutfeld, R. J. and Romankiw, L. T., Gold Bull., 15(4):120 (October 1982) 63. Bocking, C., Trans. Inst. of Metal Finish., 66:50 (1988) 64. von Gutfeld, R. J., Gelchinski, M. H. and Romankiw, L. T.,J. Electrochem. Soc., 130(9):1840 (1983) 65. Gelchinski, M. H., Romankiw, L. T., Vigliotti, D. R. and von Gutfeld, R. J., J. Electrochem. Soc., 132(11):2575 (1985) 66. Kuiken, H. K., Mikkers, F. E. P. and Wierenga, P. E., J. Electrochem. Soc., 130(3):554 (1983) 67. Zahavi, S., Tamir, S. and Halliwell, M. J., Plat. and Surf. Fin., 73(2):56 (February 1986) 68. Holt, M. L., Modern Electroplating, 3rd edition, (F. A. Lowenheim, ed.), pp. 461-485, John Wiley & Sons (1973) 69. Spencer, L. F., Metal Finishing, 69(10):69 (October 1971) 70. Brenner, A., Electrodeposition of Alloys, Principles and Practice, Academic Press (1963) 71. Electrodeposition of Alloys (V. A. Averkin, ed.), (1961), Translated from Russian by the Israel Program for Scientific Translations (1964) OTS 64-11015, U.S. Dept of Commerce 72. Brenner, A., Plating, 52(12):1249 (1965) 73. Krohn, A. and Bohn, C. W., Electrodep. and Surf. Treatment, 1(3):199 (January 1973); Plating, 58(3):237 (March 1971) 74. Sadana, Y. N., Metal Finishing, 82,83,84 (1984 - 1986) 75. Brenner, A., Burkhead, P. S. and Seigmiller, E., J. Res. Nat’l. Bureau of Standards, 39:351 (1947) 76. Clark, W. E., and Holt, M. L., J. Electrochem. Soc., 94:244 (1948) 77. Hoar, T. P. and Brickley, I. A., Trans. Inst. of Metal Finish., 32:186 (1955) 78. Frantsevich, T. F. and Zayats, A. I.,Zhur. Priklad Khim., 31:234 (1958); English translation, p. 224, Ukrain. Khim Zhur. 24:585 (1958) 79. Machu, W. and El Ghandour, Werkstoffe u. Korrosion, 11(7):420 and 11(8):481 (1960) 80. Machu, W. and Fathi, M., U.S. Patent 3,093,556 (June 11, 1963)


609

81. Domnikov, L., Metal Finishing, 62(3):61 (March 1964) 82. Hayashi, T. and Ishihama, A.,Plat. and Surf. Fin., 66(9):36 (September 1979) 83. Lashmore, D. S., Weisshaus, I. and Pratt, K., Plat. and Surf. Fin., 73(3):48 (March 1986) 84. Chisholm, C. V. and Carnegie, R. J. G., Plating, 59(1):28 (1972) 85. Machu, W., Metalloberfläche, 30(10):460 (1976) 86. Domnikov, L., Metal Finishing, 68(12):54 (1970) 87. Srivastava, S. C., Surf. Technol., 10:237 (1980) 88. Mayer, A., Standhammer, K. and Johnson, K., Plat. and Surf. Fin., 72(11):76 (November 1985) 89. Hashino, S., Laitinen, H. A. and Heflund, G. B., J. Electrochem. Soc., 133(4):681 (April 1986) 90. U.S. Patent 2,315,740, also Black, G., Metal Finishing, 44:207 (1946) 91. Moeller, R. W. and Snell, W. A., Proc. American Electroplaters’ Soc., 42:189 (1955) 92. Moeller, R. W. and Snell, W. A., Proc. American Electroplaters’ Soc., 43:230 (1956) 93. Tomaszewski, T. W., Clauss, R. J. and Brown, H., Proc. American Electroplaters’ Soc., 50:169 (1963) 94. Kilgore, C. R., Products Finish., 34 (May 1963) 95. Sayfullin, R. S. and Safina, R. A., Zashchita Metal (USSR), 3(2):215 (1967); See also, ASM Rev. of Met. Lit., 24(7):99 (July 1967) 96. Sautter, F. K., J. Electrochem. Soc., 110:557 (1963) 97. Weisner, H. J., Frey, W. P., Vanderwoort, R. R. and Raymond, E. L., Plating, 57(4):358, 362 (April 1970) 98. Greco, V. P. and Baldauf, W., Plating, 55(3):250 (March 1968) 99. Shyne, J. J., Barr, H. N., Fletcher, W. D. and Scheible, H. G., Plating, 42(10):1255 (October 1955) 100. Ortner, M. Plating, 51(9):885 (September 1964) 101. Brenner, A. and Riddell, G., J. Res. Nat’l. Bureau of Standards, 39, (November 1947), Res. Paper R.P. 1835,Proc. American Electroplaters’ Soc., 33:23 (1946) and 34:156 (1947) 102. Brenner, A., Metal Finishing, 52(11):68 (November 1954); 52(12):61 (December 1954)

610


103. “Symposium on Electroless Nickel Plating,” ASTM Special Technical Publication No. 265, American Soc. for Testing and Materials, Phila., PA (1959) 104. Gorbunova, K. M. and Nikiforova, A. A., Physiochemical Principles of Nickel Plating, translated from Russian by Israel Program for Translation, OTS 63-11003, U.S. Dept of Commerce, (1960) 105 Saubestre, E. B., Metal Finishing, 60(6):67; (7):49; (8):45; (9):59 (1962) 106. Gawrilov, G., Metalloberfläche, 25(4):118 (1971); 25(8):277 (1971); 26(4):139 (1972) 107. Gutzeit, G., Saubestre, E. B. and Turner, D. R., Electroplating Engineering Handbook, 3rd edition, (A. K. Graham, ed.), pp. 486 - 502, Reinhold Publ. Co. (1971) 108. Pearlstein, F., Modern Electroplating, 3rd edition, (F. A. Lowenheim, ed.), Ch. 31, John Wiley & Sons (1974) 109. Goldenstein, A. W., Rostocker, W., Schossberger, F. and Gutzeit, G., J. Electrochem. Soc., 112:104 (1957) 110. Graham, A. H., Lindsay, R. W. and Read, H. J., J. Electrochem. Soc., 112:401 (1965) 111. (a) Morton, J. P. and Schlessinger, M., J. Electrochem. Soc., 115:16 (1968); (b) Chow, S. L., Hedgecock, N. E., Schlessinger, M. and Resek, J., ibid., 119:1614 (1970) 112. Ziehlke, K. T., Dritt, W. S. and Mahoney, C. H., Metal Progress, 77:84 (1960) 113. Lee, W. G., Plating, 47:288 (1960) 114. Randin, J. P. and Hintermann, H. E., Plating, 54:523 (1967) 115. Johnson, C. E. and Ogburn, F., Surf. Technol., 4(2):161 (March 1976) 116. Higgs, C. E., Surf. Technol., 2(3):315 (1973/74) 117. Schwartz, M. and Mallory, G. O., J. Electrochem. Soc., 123 (5):606 (May 1967) 118. deMinjer, C. H. and Brenner, A., Plating, 44(12):1297 (1957) 119. Mallory, G. O., Plating, 61(11):1005 (1974) 120. Parker, K. and Shah, H., Plating, 58(3):230 (March 1971) 121. Baldwin, C. and Such, T. E., Trans. Inst. of Metal Finish., 46:73 (1968) 122. Safranek, W. H., The Properties of Electrodeposited Metals and Alloys, A Handbook, 2nd edition, published by The American Electroplaters’ and Surface Finishers’ Soc. (1986)


611

123. Okinaka, Y.,Gold Plating Technology, (F. H. Reid and W. Goldie, eds.), Ch. 11, Electrochemical Publications, Ltd. (1974) 124. Saubestre, E. B., Proc. American Electroplaters’ Soc., 46:264 (1959) 125. Saubestre, E. B., Plating, 59(6):563 (June 1972) 126. Ma and Gawne,Trans. Inst. of Metal Finish., 65, (part 2) (August 1985) 127. Hadley J. S. and Harland, L. E.,Metal Finishing, 85(12):51 (December 1987) 128. Parker, K., The Properties of Electrodeposited Metals and Alloys, A Handbook, 2nd edition, 23:497, and 24:531, published by The American Electroplaters’ and Surface Finishers’ Soc. (1968) 129. “Recommended Practice for Processing of Mandrels for Electroforming,” ASTM Standard B431-65, Also see, Plating, 51(11):1075 (November 1964) 130. DiBari, G. A., 64th Metal Finish. Guidebook and Directory, p. 435, Metals & Plastics Publ., Inc. (1978) 131. Spencer, L. F., Metal Finishing, 57(5):48 (May 1959) 132. Spiro, P.,Electroforming, 2nd edition, International Publ. Services, NY (1971) 133. Braddock, D. M. and Harris, S. J., Electrodep. and Surf. Treatment, 2(2):123 (1973/74) 134. Dini, J. W. and Johnson, H. R., Surf. Technol., 4(3):217 (May 1976) 135. Dini, J. W., Johnson, H. R. and Saxton, H. J., Electrodep. and Surf. Treatment, 2(2):165 (1973/74) 136. Withers, J. C. and Abrams, E. F., Plating, 55(6):605 (June 1968) 137. Wallace, W. A. and Greco, V. P., Plating, 57(4):342 (April 1970) 138. Greco, V. P., Wallace, W. A., and Cesaro, J. N. L., Plating, 56(3):262 (March 1969) 139. Wernick, S. and Pinner, R., The Surface Treatment and Finishing of Aluminum, 3rd edition, 1 vol. (1964); 4th edition, 2 vols. (1972) Robert Draper, Ltd. 140. Wernick, S., Metal Finishing, 53(6):92 (1955) 141. Sweet, A. W., Plating, 44(11):1191 (November 1957) 142. Keller, F., Hunter, M. S. and Robinson, D. L., J. Electrochem. Soc., 100(9):411 (1953) 143. Spooner, R. C., Paper No. AN-10, Aluminum Finishing Seminar, Detroit, Michigan, sponsored by Aluminum Assoc. (1968) 144. Wernick, S., Plat. and Surf. Fin., 75(6):51 (1988)

612


145. Short, E. P. and Morita, A., Plat. and Surf. Fin., 75(6):102 (1988) 146. George, D. J. and Powers, J. H., Plating, 56(11):1240 (1969) 147. Wernick, S. and Pinner, R., Metal Finishing, 53(11):(1955) 148. Coulston, E. L., Paper No. AN-6, Aluminum Finishing Seminar, Detroit, Michigan, sponsored by Aluminum Association (1968) 149. Tin and Its Uses, 133, Tin Research Institute (1982) 150. Woods, J. L., U.S. Patent 3,857,766 (December 31, 1974); Newman, F. S., Hartman, J. T., and Dedona, F. A., U.S. Patent 3,983,014 (September 28, 1976); Knodo, M. and Shizouka, T. (Japan) U.S. Patent 3,996,125 (December 7, 1976) 151. Konno, H., Theory and Practice of Pulse Plating, (J. C. Puippe, F. Leamon, eds.), 12:209, published by American Electroplaters and Surf. Fin. Soc., Orlando, FL (1986) 152. Metals Handbook, 9th edition, 5:632, ASM (1982) 153. Geduld, H., Metal Finishing, 65(4):62 (April 1967) 154. Saubestre, E. B., Durney, L. J., and Washburn, E. B., Metal Finishing, 62(11):52 (1964) 155. Saubestre, E. B., Trans. Inst. of Metal Finish., 47:228 (1969) 156. Perrins, L. E., Trans. Inst. of Metal Finish., 50:38 (1972) 157. Courduvelis, C. L., “Applications of Plasmas in the Electroplating of Plastics,” paper presented at 65th Annual Technical Conf., American Electroplaters’ Soc. (1978), pre-print, American Electroplaters’ Soc., Orlando, FL 158. Shipley, C. R., U.S. Patent 3,011,920; Matijevic, E.,Plating, 63(11):1051 (1974); Cohen, R. L.. and West, K. W., J. Electrochem. Soc.,120(4):502 (1973); Plating, 63(5):52 (May 1974), (colloid hypothesis); Zeblinsky, R. J., U.S. Patent 3,672,938; Rantell, A. and Holtzman, A., Trans. Inst. of Metal Finish., 51,62 (1973) and Plating, 63(11):1052, 1054 (1974) (complex hypothesis) 159. Martin, J. J., Plating, 58(9):888 (1971) 160. Rothschild, B. F. and Schwartz, M., “Plating and Finishing of Printed Circuit Boards,” American Electroplaters’ Soc., Illustrated Lecture #41, American Electroplaters’ Soc. 161. Smith, C. M., Plating, 56(4) (April, 1969) 162. Rothschild, B. F., Farmer, M. E. and Brewer, T. W.,Plating, 49(12):1269 (December 1962) 163. Jawitz, M. W., Insulation/Circuits, p. 5 (April 1976)


613

164 Rothschild, B. F. and Kilgore, L. C., “Electroplating: Cornerstone of Multilayer Board Fabrication,” presented at Western Regional Technical Session, American Electroplaters’ Soc., March, 1966, (Available as pre-print from Autonetics Div., Rockwell International, Paper X6-362/ 3111.) 165. Rothschild, B. F. and Sanders, D., Plating, 56(12):1363 (December 1969) 166. Weil, R., Plat. and Surf. Fin., 69(12):46 (1982) 167. Johnson, H. R., Dini, J. W., and Stoltz, R. E., “On the Mechanical Properties of Sulfamate Nickel Electrodeposits.” Presented at 65th Annual Technical Conference, American Electroplaters’ Soc., (Preprint) (June, 1978) 168. Read, H. J., Plating, 49(6):602 (1962) 169. Zentner, V., Brenner, A. and Jennings, C. W., Plating, 39:865 (1952) 170. Brenner, A., Burkhead, P., and Jennings, C., J. Res. Nat’l. Bureau of Standards, 40:31, R.P. 1854 (January 1948) 171. Jones, M. H., Kenez, M. G., and Saiddington, J., Plating, 52(1):39 (1965) 172. Weil, R., Properties of Electrodeposits, Their Measurements and Significance, (R. Sard, H. Leidheiser, Jr., and F. Ogburn, eds.), 19:319, The Electrochemical Soc. (1975) 173. Weil, R., Plating, 57(12):1231 (December 1970); 58(1):50 (January 1971); 58(2):137 (February 1971) 174. Romankiw, L. T., and Thomposn, D. A.,Properties of Electrodeposits, Their Measurements and Significance, (R. Sard, H. Leidheiser, Jr., and F. Ogburn, eds.), 23:389, The Electrochemical Soc. (1975) 175. "Electroless Plating-Fundamentals and Applications", (O. Mallory and J. B., Hajdu, eds.), American Electroplaters and Surface Finishers Soc., (1990) 176. Dini, J. W., Electrodeposition, The Materials Science of Coatings and Substrates, Noyes Publications, Park Ridge, NJ (1993) Supplementary References–Journals Electrochim. Acta Electronic Packaging and Production*, Milton S. Kiner, Publ., 222 W. Adams, Chicago, IL 60606

614


Gold Bulletin*, World Gold Council, 1, rue de la Rôtisserie, Ch. 1204, Geneva, Switzerland Printed Circuit Fabrication*, Gary W. Smith, Publ., 174 Hembase Rd., Alpharetta, GA 30210 Product Finish.*, Gardner Publ. Co., 600 Main St., Cincinnati, OH 45202 Trans. Faraday Soc. * Trade journals (complimentary on controlled circulation). Supplementary References–Books AESF, “Symposium on Electroforming,” American Electroplaters’ and Surf. Fin. Soc., Orlando, FL (1967) AESF, “Symposia on Plating in the Electronics Industry,” (Proceedings), 1st - (1966), 2nd - (1969), 3rd - (1971), 4th - (1974), 5th - (1975), 6th - (1977), 7th - (1979), 8th - (1981) AESF, “Illustrated Lecture Series,” (Slides and text, 41 subjects available.) ASM,Metals Handbook 9th edition, Vol. 5; “Heat Treating, Cleaning and Finishing,” American Society for Metals, Metals Park, OH ASTM, “Anodizing Aluminum,” STP 388, American Society for Testing and Materials, Philadelphia, PA (1965) ASTM, “Electroforming-Applications, Uses and Properties of Electroformed Metals,” (1962) ASTM, “Hydrogen Embrittlement Testing,” STP 543 (1974) Bard, A. J. and Faulkner, L. R., Electrochemical Methods, John Wiley & Sons (1980) Bennington, H. and Draper, R. “Tables and Operating Data for Electroplaters,” Robert Draper Ltd., Teddington, Eng Bernstein, I. M. and Thompson, A. W., eds., Hydrogen in Metals, Amer. Soc. for Metals (1974) Blum, W. and Hogaboom, G. B.,Principles of Electroplating and Electroforming, 3rd edition McGraw-Hill Publ. Co., NY (1949) Bogenschutz, A. F., Surface Technology and Electroplating in the Electronics Industry, Porticullis Press, Ltd., London, Eng. (1974) Bockris, J. O’M. and Reddy, A. K. N., Modern Electrochemistry, (2 vols) Plenum Press, NY (1970)


615

Brugger, R., Nickel Plating, International Publ. Services, Porticullis, NY (1970) Burns, R. M. and Bradley, W. W., Protective Coatings for Metals, 3rd edition, (ACS Monograph Series), Reinhold Publ. Corp., NY (1967) Cobalt Monograph, (prepared by staff, Batelle Memorial Inst.) edited by Centre D’Information du Cobalt, Brussels, Belgium (1960) Coombs, C. F. Jr., ed., Printed Circuits Handbook, McGraw-Hill Book Co. (1967) Dubpernell, G., (1977)

Electrodeposition of Chromium, Pergamon Press, NY

Fischer, J. and Weiner, D. E., Precious Metal Plating, Robert Draper, Ltd., Teddington, Eng. (1964) Gileadi, E., Kirowa-Eisner, E. and Penciner, J., Interfacial Electrochemistry, Addison-Wesley Publ. (1975) Goldie, W., Metallic Coating of Plastics, (2 vols), Electrochemical Publ., Ltd., Middlesex, Eng. (1968) Graham, A. K., ed., Electroplating Engineering Handbook, 3rd edition, Van Nostrand, Reinhold Co., NY (1971); Durney, L., ed., 4th edition (1975) Greenwood, J. D., Hard Chromium Plating, 2nd edition, International Publ. Services, Porticullis, NY (1971) Greenwood, J. D., Heavy Deposition, Robert Draper, Ltd., Teddington, Eng. (1970) Hall, H. ed., Metal Finishing Guidebook-Directory, Metals and Plastics Publ., NJ (issued annually) Hampel, C. A., ed., Encyclopedia of Electrochemistry, Reinhold Publ. Corp., NY (1964) Jarrett, G. D. R., Draper, C. R., Muller, G., and Baudrand, D. W., Plating on Plastics, 2nd edition, International Publ. Services, Porticullis, NY (1971) Kutzelnegg, A., Testing Metallic Coatings, Robert Draper, Ltd., Teddington, Eng. (1963) Lowenheim, F. A., Electroplating, (Sponsored by American Electroplaters’ Soc.) McGraw-Hill Book Co., NY (1978) Murphy, J. A., ed.,Surface Preparation and Finishes for Metals, McGraw-Hill Book Co., NY (1971) Narcus, H., Metallizing of Plastics, Reinhold Publ. Co., NY (1960)

616


Read, H. J., ed., Hydrogen Embrittlement in Metal Finishing, (Sponsored by American Electroplaters’ Soc.) Reinhold Publ. Co. (1961) Reid, F. H. and Goldie, W., Gold Plating Technology, Electrochemical Publ., Ltd. (1974) Raub, E. and Muller, K., Fundamentals of Metal Deposition, Elsevier Publ. Co., NY (1967) Riedel, W.,Electroless Nickel Plating, ASM International (1991) (Trans from German, Kuhn, A. T.) Sard, R., Leidheiser, H. Jr. and Ogburn, F., eds., Properties of Electrodeposits, Their Measurements and Significance, The Electrochemical Soc., Princeton, NJ (1975) Ulhig, H. H., Corrosion and Corrosion Control, 2nd edition, John Wiley & Sons, NY (1971) Uhlig, H. H., ed., Corrosion Handbook, (Sponsored by Electrochemical Society) John Wiley & Sons, NY (1948) Van Horn, K. R., ed., Aluminum, 3 vols., American Society for Metals, Metals Park, OH (1967) West, J. M.,Electrodeposition and Corrosion Processes, Van Nostrand Co., NY (1965)

11 Advanced Thermal Spray Deposition Techniques Robert C. Tucker, Jr.

1.0

INTRODUCTION

Advanced thermal spray coatings, including plasma, detonation gun* and high velocity oxy-fuel (HVOF) coatings have been used in industry for over thirty-five years. They are line-of-sight processes in which powder is heated to near or above its melting point and accelerated (by either a detonation wave, or plasma or high velocity combustion gas stream). The powder is directed at a substrate (surface to be coated) and, on impact, forms a coating consisting of many layers of overlapping thin lamellar particles or splats. Almost any material that can be melted without decomposing can be used to form the coating. The substrate, for most applications, is not heated above 150°C, so its metallurgical properties (strength, etc.) remain unchanged. Typical coating thicknesses range from 0.05 to 0.5 mm (0.002 to 0.020 inches), but in a few applications may exceed 5 mm (0.2 inches). The description of the processes and coatings that follows is divided into three sections: equipment and processes, coating structure, and properties.

* The detonation gun process was developed by Union Carbide Corporation[1] and detonation gun coatings are currently available in the United States, Japan, Singapore, and Western Europe through Praxair Surface Technologies, Inc. (formerly Union Carbide Corp.). Plasma spray and HVOF coatings (also developed by Union Carbide[2]) are available from a number of coatings service organizations and the equipment is available from several sources for in-house use.

617

618 2.0

Deposition Technologies for Films and Coatings EQUIPMENT AND PROCESSES

In this section, plasma torches, detonation guns, HVOF torches, auxiliary equipment, and equipment-related coating limitations are discussed. A description of the physics of plasma, detonation, or combustion generation would be too lengthy to be included here and is unnecessary to an understanding of the utilization of the processes. 2.1

Plasma Spray Process

The essential elements of a plasma torch are shown in Fig. 11.1. The anode is usually copper and the cathode tungsten. A gas, usually argon or nitrogen or a mixture of these with hydrogen or helium, flows around the cathode and through the anode which serves as a constricting nozzle. A direct current arc, usually initiated with a high frequency discharge, is maintained between the electrodes. The current and voltage vary with the anode/cathode design, gas flow, and gas composition. The power varies from about 5 to 120 kilowatts depending on the type of torch and the operating parameters. In one variant of a coating torch, a partially transferred arc is used; i.e., part of the arc goes to the anode and part to the substrate being coated. This causes substantial heating of the substrate and is used only in special situations. Fully transferred arc surfacing torches will not be discussed here, since their use constitutes a form of welding rather than coating. The gas plasma generated by the arc consists of free electrons, ionized atoms, and some neutral atoms and undissociated diatomic molecules if nitrogen or hydrogen are used. The specific anode/cathode configuration, gas density, mass flow rate, and electrical power determine the plasma temperature and velocity. Plasma gas velocities with most conventional torches are subsonic, but supersonic velocities can be generated by using convergingdiverging nozzles with critical exit angles. The temperature of the core of the plasma may exceed 30,000°C (50,000°F). A schematic of a typical distribution of temperature in the plasma is shown in Fig. 11.2. The enthalpy of the plasma and efficiency of heat transfer to the powder particles can be increased substantially with the inclusion of diatomic gases, such as hydrogen or nitrogen, Fig. 11.3.

Advanced Thermal Spray Deposition Techniques

Figure 11.1. Schematic of a plasma spray torch.

Figure 11.2. Distribution of temperature in a plasma flame.

619

620


Figure 11.3. Enthalpy of gases commonly used in plasma spraying.

The velocity that powder achieves in a plasma stream depends on the integrated effect of mass flow rate of the plasma and the distance the powder is carried in the stream. Similarly, the temperature the powder achieves is a function of the integrated effect of the plasma temperature, plasma composition and the transit time in the plasma stream. (Both powder velocity and temperature are functions of other factors as well, such as particle size, powder composition, heat capacity, density, emissivity, etc., as discussed in subsequent sections.) It follows, therefore, that the point of entry of the powder into the plasma stream is very important. The ideal location would be in a uniform pattern upstream of the anode throat since this would probably allow the best distribution of the powder in the plasma stream, expose the powder to the highest plasma temperature, and provide the longest path or time in transit before the plasma temperature and velocity decrease. Most torch manufacturers, however, have been unable to prevent powder adherence to the entry or throat of the nozzle and excessive superheating using this approach.


621

As a result, powder entry is usually in the diverging portion of the nozzle or just beyond the exit as shown in Fig. 11.1. Attempts[3][4] have also been made to adjust the point and angle of entry of the powder into the plasma stream for the melting point of the powder. The goal in some cases was to heat the powder close to, but not over, the melting point. In one high velocity torch design,[5] in which shock diamonds are generated, the powder is introduced a short distance beyond the exit in a region of rarefaction in the plasma stream. In another plasma torch design,[6][7] a cylindrical extension is placed on the nozzle encompassing the entry for the powder. Additional inert gas is also introduced. The result is a cooler, but more uniform and higher velocity gas effluent with higher powder velocities. The most important parameters relative to the powder particles at impact on the substrate are their temperature, velocity, and extent of reaction with the gaseous environment. The velocity of the powder, as previously mentioned, is a function of the mass flow rate of the plasma, the density, mass and shape of the powder, and the distance the powder travels in the plasma. With most of the conventional commercial torches available up to the mid 70’s, velocities varied from about 400 to 1000 ft/s. Higher velocity torches have since become available[3]-[5][8] with powder velocities claimed [9] to be in excess of 1800 ft/s (measured by a rotating mirror), but velocities measured with a more sophisticated technique (Doppler laser) for similar torches were reported to be 1200 ft/s.[10] It has often been stated that any material that can be melted without decomposition can be used as a plasma coating. There appear, however, to be two schools of thought on whether or not the powder should be molten on impact. Certainly the combination of particle plasticity or fluidity and velocity must be high enough to allow the particle to flow into a thin, lenticular shape that molds itself to the topology of the substrate or previously-deposited material. The extent to which this is achieved determines the density and strength of the coating. With a relatively low-velocity torch, reasonably high densities can only be achieved if the particles are substantially molten. As noted previously, the intent of at least some high velocity torch designs is to achieve highly plastic, but not molten, particles. Excess fluidity (superheating) can lead to undue shattering and bounce of the particles, resulting in a poor microstructure and low deposition efficiency. Care should be exercised in developing the coating parameters to not heat the powder to an excessive temperature for other reasons as well. The most obvious hazard is vaporization of all or part of the powder. This is most likely when,(a) the difference between the melting and boiling point of a single-

622


phase powder is too small,(b) one or more of the components in a multiphase powder has a substantially lower boiling point than the others,(c) one or more of the components in a mixture of powder has a substantially lower boiling point than the others, or (d) the powder size distribution is too wide with a single component or not adjusted for heating rates with a mixture. In addition to vaporization through boiling, there may be some loss of a component in an alloy or compound that has a particularly high vapor pressure. This is generally not a significant problem because transit times are so short. The temperature in a plasma is high enough to melt (or decompose) any material, given enough time. Comparison of the relative heating rates of powders is not as simple as comparing their melting points, however. Heat transfer in the plasma jet is primarily the result of the recombination of the ions and re-association of atoms in diatomic gases on the powder particle surfaces and absorption of radiation.[11] The ultimate temperature of the powder particles, therefore, is a function of the catalytic activity of their surface, their emissivity (particularly in the ultraviolet range), their heat capacity (including any heats of phase transformations and heat of melting), their thermal conductivity, and their surface to volume ratio (shape). Many metals, having high absorption in the ultraviolet range, high surface activity, and high thermal conductivity, tend to heat much more rapidly than most oxides. Specific tables of heating rates are not available, but tables of the pertinent physical properties can be used as guidelines in selecting appropriate coating parameters. The extent of reaction of the powder with its gaseous environment during transit depends both on the composition of the plasma gas and the amount of intermixing of the plasma gas with the ambient gas between the nozzle and the substrate. It is generally assumed that argon and helium are inert and no degradation of the powder occurs in the torch when they are used as the only plasma gases. Obviously for this to be true, the gas source must be free of oxygen and other contaminants, and the torch and other equipment must be gas-tight. Substantial adsorption of argon and, presumably, helium on the coating surfaces, both external and internal, however, can occur as evidenced by the evolution of relatively large quantities of these gases during vacuum heat-treatment. Whether or not hydrogen or nitrogen, when used in the plasma gas, are effectively inert relative to the powder depends on the composition of the powder. The transit time and temperature of the powder in the plasma determine the extent of reaction and/or solution of the gas in the powder in


623

those cases where the gas is not thermodynamically inert. The use of hydrogen to reduce the amount of oxidation during spraying may be somewhat effective, but the effect may be due as much to shielding (by reacting with oxygen from the air inspirated into the plasma stream) as to actual reduction of oxide formed on metallic powder. On the other hand, oxide powders or oxide films on metallic powders may be decomposed in the plasma spray; e.g., zirconia coatings sprayed with an argon plasma are slightly oxygen deficient, and the amount of oxygen in copper can be lowered simply by thermal decomposition of its oxide. Usually of greater concern than reaction with the plasma gas is the extent of reaction of the powder with oxygen or nitrogen from the air inspirated into the plasma stream after it exits the nozzle. This effect is strongly a function of the type of torch used, as illustrated in Fig. 11.4. None of these coatings were shielded from the atmosphere, yet the differences in extent of oxidation is dramatic. Nitration of some materials may also occur, but has not been extensively studied. If coatings with even less oxide than that shown in Fig. 11.4 are desired, several means of shielding the plasma stream are available. One of the best, and certainly the most adaptable to production, is a patented inert gas shroud that surrounds the effluent with argon.[12] A comparison of the results obtained with this shield compared to those obtained with the same torch that produced the relatively clean microstructure of Fig. 11.4 is shown in Table 11.1. Note that the oxygen contents of molybdenum, copper, and nickel are all lower in the coating than in the starting powder when using the inert gas shroud, while that of titanium, a very reactive metal, is only slightly higher. Alternative methods of excluding air include spraying in a low partial pressure of inert gas in a vacuum chamber[13][14] or in an enclosure filled with argon. An extreme example used for coating large parts is an entire room or cubicle filled with argon in which the operators wear life support suits.[15] Both argon-shrouded and low pressure, inert-gas-chamber spray coating methods are used in the commercial production of the very reactive “MCrAlY” coatings on gas turbine components (described more fully in Sec. 4.8). Some of the relative advantages and disadvantages of the two methods are listed in Table 11.2 and the process steps used with both are shown in Table 11.3.

624


Figure 11.4. Microstructures of aluminum bronze coatings made with three types of standard plasma spray torches illustrating varying degrees of oxidation during deposition. As-polished.


Table 11.1.

Oxygen Content of Plasma Deposited Coatings Oxygen Content (%) Conventional Coaxial Gas Coating Shielded Coating

Coating Material

Starting Powder

Copper

0.126

0.302

0.092

Nickel

0.172

0.456

0.151

Tungsten

0.027

0.274

0.030

Titanium

0.655

2.0

0.730

Molybdenum

0.419

0.710

0.160

Table 11.2.

625

Advantages and Disadvantages of Inert Gas Shroud and Low Pressure Inert Gas Plasma Deposition Plasma Spray with Inert Gas Shroud Advantages: Clean Deposition Low Capital Cost Low Operating Cost High Production Rate Disadvantages: Difficult to Preheat Parts Plasma Spray in Low Pressure Inert Gas Advantages: Clean Deposition Longer Stand-Off or Higher Velocity Preheat Parts to Reduce Stress Sputter Cleaning Disadvantages: High Capital Cost High Operating Cost Low Production Rate

626


Table 11.3(a). Typical Coating Sequence for Plasma Spray with Inert Gas Shroud 1. Clean and grit blast part 2. Load in fixture 3. Coat 4. Unload 5. Heat-treat and peen part

Table 11.3(b). Typical Coating Sequence for Plasma Spray in Reduced Pressure Inert Gas 1. Clean and grit blast part 2. Load in vacuum chamber 3. Pump down and back-fill chamber to reduced pressure 4. Preheat part 5. Sputter clean part (optional) 6. Coat 7. Cool part 8. Back-fill to atmospheric pressure 9. Unload 10. Heat-treat and peen part

2.2

Detonation Gun Deposition Process

The detonation gun, shown schematically in Fig. 11.5, consists of a water-cooled barrel several feet (about one meter) long with an inside diameter of about one inch (25 mm), and associated gas and powder metering equipment. In operation, a mixture of oxygen and acetylene is fed into the barrel along with a charge of powder. The gas is then ignited and the detonation wave accelerates the powder to about 2400 ft/s (760 m/s) while heating it close to, or above, its melting point. The maximum free burning temperature of oxygen/acetylene mixtures occurs with 45% acetylene and is about 3140°C, but under detonation conditions probably exceeds 4200°C,


627

so most materials can be melted. The distance that the powder is entrained in the high velocity gun is longer than in a plasma device which accounts, in part, for the much higher particle velocity. After the powder has exited the barrel, a pulse of nitrogen purges the barrel. The cycle is repeated about four to eight times a second.

Figure 11.5. Schematic of a detonation gun.

Each pulse of powder results in the deposition of a circle of coating about 25 mm in diameter and a few microns thick. This circle of coating is, of course, composed of many overlapping thin lenticular particles or splats corresponding to the individual powder particles. The total coating is, in turn, produced by many overlapping circles of coating. This pattern of overlapping is closely controlled to produce a smooth coating and minimize substrate heating and residual stress. Because of the gases used in the detonation gun, the powder may be exposed to either an oxidizing or carburizing environment, although an essentially inert mixture can be achieved with precise control. Carburizing conditions, in particular, can be used to advantage,[16] as illustrated in Sec. 4.2 on microstructures. Recently, a significant advance in detonation gun technology has been made with the introduction of the Super D-Gun™. This device uses a mixture of fuel gases rather than just acetylene. As a result, the volume of gaseous detonation products is substantially increased with concommitant increases in gas pressure and gas velocity. The higher gas velocity, in turn, results in

628


higher powder particle velocities, to about 1000 m/s or more. Thus the kinetic energy of the particles are about double those of the standard detonation gun particle energies. This yields coatings with higher densities, better bonding, and improved mechanical and other properties. 2.3

High Velocity Oxy-Fuel Deposition

High velocity oxy-fuel (HVOF) deposited coatings are produced by heating and accelerating powder in a high velocity gas stream generated by the combustion of a fuel gas and oxygen. The powder is heated to near or above its melting point and projected against the substrate to be coated forming a dense, lamellar coating. The tungsten carbide-cobalt group of materials are probably the most widely used HVOF coatings, but other cermets, metals, and some oxides can be used. For most applications, coatings range from 0.002" to 0.020" (0.05 mm to 0.5 mm) in thickness, but substantially thicker coatings of some materials can be used if necessary. Although a variety of high velocity combustion spray devices have been developed, most have in common a combustion chamber with ports leading to a nozzle, shown schematically in Fig. 11.6. Continuous combustion of oxygen and fuel gas occurs in the chamber and the resulting hot, high pressure gas is allowed to expand and accelerate in the nozzle. The fuel gas is usually propane or propylene; however, acetylene can be used in some devices. In at least one device, liquid fuels such as kerosene can be used to allow a higher effective mass flow of fuel to the combustion chamber than is possible with most gases. Powder is introduced axially into the nozzle, allowing relatively efficient heating and acceleration of the powder particles. The powder is heated and accelerated by the products of combustion, usually to temperatures above its melting point and to velocities that may exceed 1800 ft/s (550 m/s). Since the powder particles are being heated and accelerated in a stream of combustion products, the surrounding atmosphere may be either oxidizing or carburizing. In addition, air may be inspirated into the gas stream as it exits the nozzle leading to oxidation of the powder. The degree to which these gaspowder reactions occur depends, of course, on the specific device, the operating parameters, and the material being deposited. It is probably more significant with metallic materials and carbides than with oxides. High velocity combustion spray, like all other thermal spray processes, is a line-of-sight process. Thus, it should be expected that the properties of the coatings will vary with the angle of deposition. In addition, stand-off may be in important parameter—a distance which is too short allows too little time for heating and acceleration of the powder particles or overheating of the substrate, while a


629

stand-off which is too long may allow the powder particle velocities to diminish and their temperature to drop too far. The actual powder temperature and velocity distributions are strongly a function of the design of the high velocity combustion spray device as well as the operating parameters, morphology, and composition.

Figure 11.6. High velocity oxy-fuel coating process.

2.4

Thermal Control

Control of the temperature of the substrate during deposition is essential. Usually it is desirable to slightly warm the surface of the substrate before coating and then maintain the temperature no higher than about 150°C (300°F) while coating. This control is achieved by limiting the deposition rate (mass per unit area per unit time) and using auxiliary cooling such as CO2 or air. On rare occasions it might be advantageous to apply the coating at an elevated substrate temperature to reduce residual stress in the coating. This allows the coating and the substrate to cool together thus minimizing the mismatch in temperature and concommitant stress. Using this technique may change the quench rate of the powder and hence the structure and properties of the coating.

630 2.5

Deposition Technologies for Films and Coatings Auxiliary Equipment

In addition to the plasma torch, detonation gun, or HVOF device itself, gas controls, power supplies, and powder feeders are required. Most of these are supplied with the basic unit. A detailed discussion of their characteristics is not appropriate here, but there are several general criteria that all such equipment should meet. Excellent gas control can be achieved with either rotameters or critical flow orifice columns, but, in either case, attention should be paid to both upstream and downstream pressures to insure that the control device is capable of accurately measuring flow. Mass flow meters may, of course, also be used. All guages, meters, rotameters and orifices should be calibrated periodically. Electrical power supplies should be reasonably ripplefree and, again, all meters should be periodically calibrated. For optimum plasma or HVOF spraying, powder must be distributed uniformly in the plasma stream at a constant rate. There are a variety of powder dispensers designed to do this including those based on an auger, aspirated flow, or fluidized bed.[17]-[19] Continuous measurement of powder feed-rate with closed-loop adjustment provides the best control. Conversely, a pulsed flow of powder is required for a detonation gun. Again, however, uniform distribution of the powder in the barrel is important, as is the constancy of the amount of powder in each pulse. The highest quality thermal spray coatings can only be achieved with automated or semi-automated torch or gun and part handling. Hand-held torches lead to varying stand-off, poor thermal control, and nonuniform thickness—all of which result in varying coating properties across the part. The most commonly used method of part and torch motion control utilizes a modified lathe concept with the torch mounted on what would be the tool post and the parts to be coated either rotated as a cylinder or mounted on an annulus plate. Predetermined torch-to-part surface speeds and overlap can then be maintained by varying the rotation and torch speeds. A variety of cam actions can be used to maintain a uniform deposition rate from the center to the outside of an annulus plate. Another method of controlling relative motion, particularly suitable for the detonation gun because lower surface speeds can be used, is that of traversing and indexing in a raster pattern. Using this technique, very large flat surfaces can be coated. More extensive automation has been developed for all types of advanced thermal spray deposition including part transfer handling and robotic torch and/ or part manipulation. Computer control of the torch motion and/or part motion


631

tremendously increases the productivity of the equipment. Computers may also be used to monitor and control the complete process, interlocking part and torch or detonation gun motion, powder and gas flow, and power level. 2.6

Equipment-Related Coating Limitations

All types of thermal spray deposition are line-of-sight processes, and the structure of the coatings is a function of the angle of deposition, i.e., the angle between the axis of the plasma, HVOF, or detonation gun effluent and the surface of the substrate being coated. Normally coatings with the highest density and bond strength are achieved at a 90° angle of deposition. The extent of changes in plasma coating structure is a function of the type of plasma torch and the operating parameters. With some low velocity torches, angles less than 75° may cause significant degradation of properties,[18] with some higher velocity torches, angles as low as 60° can be tolerated. This limitation may cause some problems in coating complex parts, particularly those with narrow grooves or sharp angles, and may require several set-ups to adequately coat the different faces or surfaces of a part. The detonation gun, with its higher particle velocity, can usually tolerate a wider deviation from 90° (down to about 45° in many cases). The sensitivity of HVOF coatings to angle of deposition is probably intermediate between plasma and detonation gun. Another limitation, of course, is the size of the torch or gun and the required stand-off (distance from the nozzle or front face of the torch to the workpiece) when an inside diameter must be coated. One of the smaller torches can apply a metallic coating to about a 30 mm (1.2 in), or a ceramic coating to about a 50 mm (2 in), inside diameter cylinder at 90°. Another torch with an effluent at 45° to the torch axis can apply a coating to the inside of a blind cylinder about 50 mm (2 in) in diameter. An HVOF device is quite bulky and requires a long stand-off. It can therefore be used to coat the inside surfaces of only very large cylinders. The detonation gun, of course, cannot fit into a cylinder or other cavity. It can be used, however, to coat the inside surface of a cylinder to a depth about equal to the diameter, i.e., to an angle of deposition of about 45°. While there is some change in microstructure as the angle decreases, the inherently high density and bond strength of detonation gun coatings, as previously mentioned, still allow very good coatings to be deposited at the lower angles.

632 3.0

Deposition Technologies for Films and Coatings TOTAL COATING PROCESS

The total coating process includes specification and procurement of powder, substrate preparation, masking, and finishing, in addition to the coating operation itself. Each of these is discussed in the following subsections. 3.1

Powder

Most of the powder used for advanced thermal spray deposition falls between 5 and 60 microns in size. To achieve uniform heating and acceleration of a single component powder, it is advisable to have the size distribution as narrow as possible. The additional cost of sizing is, at least partially, recovered in higher deposition efficiency and better coating quality. The specific powder size range to be used is a function of the torch or detonation gun design and the heating characteristics of the powder discussed earlier. Generally speaking, fine powders are accelerated and heated more rapidly, but they also tend to lose momentum more rapidly when spraying at longer distances (greater stand-offs). They generally result in denser, but more highly stressed coatings. Finer powders also tend to create more torch operating problems and have higher oxide contamination levels. Good quality control of powder is essential, not only during manufacture, but during storage and handling. Powder specifications and quality control should include, as a minimum, chemical analysis (including interstitials for metallic powders), shape characterization, size distribution, and flowability. A wide variety of equipment is available for analyses, and selection of a specific technique or type of test will vary with the type of powder. It is obvious that the powder should be kept clean and dry; too little attention paid to this will result in dispensing problems, torch clogging, and lumps in the coating. 3.2

Substrate Preparation

It seems quite obvious that any part to be coated (substrate) must be clean, yet this step in the total coating process is frequently given too little attention. Not only must all oxide scale or other solid foreign matter be removed, but all oils, machining lubricants, etc. must be eliminated. It is therefore usually good practice to degrease a part after any descaling, machining, or grinding is done. Grit blasting, discussed below, should not be relied upon to remove heavy scale, since it may simply embed it in the surface, leading subsequently to a weakly bonded area or a site forcorrosion.


633

Most plasma coatings require a roughened substrate surface. Although machining, chemical etching, and other techniques are sometimes used, the most frequently used method is grit blasting. The type of grit and grit blasting pressure used should be determined by the composition and heat-treat condition of the substrate. For many relatively soft substrates, chilled steel grit is satisfactory. It does not shatter and does not embed excessively in the surface. For harder substrates, alumina or silicon carbide grit has better cutting action. For some applications, special grit may be used to achieve unusually low levels of grit inclusions.[20] Regardless of the method used, the surface roughness should normally exceed 4 micrometers (150 microinches) Ra . In addition, the surface topology should be sharply peaked, not smoothly undulating. Excessive grit blasting can be detrimental due to work hardening, blunting of the peaks, and increased grit entrapment. For detonation gun coatings, and perhaps some high velocity plasma or HVOF coatings, grit blasting may not be necessary if the substrate is not excessively hard. The unusually high particle velocity in itself results in some surface roughening, particularly with some carbide-based coatings. This is generally true for titanium substrates, for example. Grit blasting, of course, increases the surface area significantly, so whether bonding, discussed in Sec. 4.3, is due to a mechanical interlocking, to interdiffusion, surface reaction, or a combination of these, it is advantageous in increasing bond strength. In any case, the coating should be applied as soon after grit blasting as possible to ensure a clean surface. 3.3

Masking

A wide variety of masking techniques are used to limit the deposition to the required area on the part. In most cases masking is less expensive than subsequent removal by grinding. Many types of tape and oxide-loaded paints or stop-off lacquers are satisfactory for low velocity, long stand-off plasma torches. For high velocity, short stand-off torches, more substantial masking is required, e.g., glass-fiber reinforced high-temperature tape, adhesivebacked steel or aluminum foil, or sheet metal masking. For detonation gun coatings, metal masking is used most frequently. Efficiently designed masking can significantly reduce the total cost of a coating and deserves careful consideration. 3.4

Coating

The coating process parameters that must be selected to apply a coating of a given powder composition and size distribution include the types

634


of gases to be used and their flow rates, the torch or gun design (e.g., anode design), the power level to be used, and for some plasma torches, the point of powder entry. All of these vary with the specific torch or gun model used. The torch or gun manufacturer should be able to provide specific instructions, or at least detailed guidelines. It is always advisable to coat a quality control specimen to verify the coating deposition rate and coating microstructure before coating any parts. Metallographic examination of this specimen should include, as a minimum, general phase content, the amount of oxidation occurring during deposition, apparent porosity, and microhardness. It is also advisable to check the grit inclusion level and/or amount of substrate surface contamination, but this is only meaningful if the quality control specimen is made of the same material, is in the same heat-treat condition, and has undergone the same surface preparation as the parts to be coated. Metallographic examination is only meaningful if well-standardized mounting and polishing techniques have been developed as well as appropriate visual and numerical standards based on significant statistical analyses. It should be noted that the microstructure, hardness, etc. of the coating on a special quality control specimen may not (in fact, usually will not) be the same as on the part because of differences in angle of deposition, relative part/ torch surface velocity, cooling, mass of the part, etc. This should not be of concern relative to quality, since the objective of examining the quality control specimen is to ensure that the torch or detonation gun is producing the right coating (process control). With the process in control, a coating with consistent properties will be applied to the part (assuming all the deposition parameters then remain constant while coating the parts). Whether or not the properties of coating produced meet the objectives of using it is a separate issue. One of the major advantages of plasma and detonation gun coatings is that they may be applied to substrates without significantly heating them above room temperature. As a result, a part can be fabricated and fully heattreated without changing the substrate microstructure or strength. This also avoids any possibility of distortion or volumetric change during any postcoating heat-treatment that is common to many other coating methods. It is, however, advisable to warm the surface slightly, usually with a pass of the torch without powder flowing, to remove most adsorbed gases from the surface before applying the coating. The surface temperature usually does not exceed 125 to 150°C during this warming pass. During coating deposition, a substantial amount of heat is transmitted to the part through the plasma gas


635

and the molten powder. To ensure uniform coating thickness and minimize residual stress within the coating, it is necessary to carefully control the areal rate of deposition. This can only be accomplished satisfactorily by using automated part and torch handling equipment with the selection of appropriate surface speed, overlap pattern and deposition rate. Cooling air or CO2 jets may be used as well. Under normal circumstances, the part temperature does not exceed about 150°C during coating. 3.5

Finishing

For many applications, plasma and detonation gun coatings can be used as-coated. In fact, in at least one application, a detonation gun tungsten carbide-cobalt coating is grit blasted to further roughen the surface for better gripping action. Probably in the majority of applications, however, the coatings are finished before being placed in service. Finishing techniques vary from brush finishing to produce a nodular surface, to machining, grinding, and lapping to produce surfaces with surface roughnesses down to less than 0.05 micrometers (2 x 10-6 inches) Ra. Machining can be used on some metallic coatings, but most coatings are ground with silicon carbide or diamond (diamond is usually preferred for detonation gun coatings). The best surface finish that can be obtained is a function not only of the finishing technique, but of the coating composition, the deposition parameters, and the part geometry. Recommendations for the machining, grinding, and lapping techniques for specific coatings can be obtained from coatings service organizations or coating equipment manufacturers. Great care should be exercised in finishing operations to avoid damaging the coating through heat checking, pull-out, or edge chipping. A typical check list[21] for diamond grinding follows: 1. Check the diamond wheel specifications. 2. Make sure the grinding equipment is in good mechanical condition. 3. Balance and true the diamond wheel on its own mount. 4. Check the peripheral wheel speed. 5. Use a flood coolant. 6. Before grinding each part, clean the wheel with minimum use of a silicon carbide stick. 7. Maintain proper infeeds and crossfeeds. 8. Never spark out—stop grinding after last pass.

636

Deposition Technologies for Films and Coatings 9. Maintain a free-cutting wheel by frequent cleaning with a silicon carbide stick. 10. Clean parts after grinding. 11. Visually compare the part at 50x with a known quality control sample.

Similarly, a typical checklist[15] for diamond lapping follows: 1. Use a hard, usually serrated, lap such as GA Meehanite or equivalent. 2. Use recommended diamond abrasives. 3. Embed the diamond firmly into the lap. 4. Use a thin lubricant such as mineral spirits. 5. Maintain appropriate lapping pressures. 6. Maintain low lapping speeds. 7. Recharge the lap only when necessary. 8. Clean the parts after grinding and between changes to different grade diamond laps. 9. Visually compare the part at 50x with a known quality control.

4.0

COATING STRUCTURE AND PROPERTIES

In this section, the macro- and microstructure of advanced thermal spray coatings are discussed as well as several important characteristics in coating design, bond strength, residual stress, and density. In the balance of the section the mechanical, wear, thermal, and electrical properties of the coatings are discussed including a few illustrations taken from service experience. 4.1

Surface Macrostructure and Microstructure

The surface roughness of most advanced thermal spray coatings is greater than 100 x 10-6 inches Ra . Most of the metallic and cermet coatings are a dull grey, but some, sprayed with an argon shroud, may be a fairly bright metallic, light grey. The oxide coatings vary from black to white with the color frequently differing from the powder or a conventional ceramic part of the same composition. This is usually due to some dissociation and/or oxygen deficiency of the coating. Very slight deficiencies, in some cases, can produce substantial color changes. Exposure to air at high temperatures often


637

returns the oxide to stoichiometry and its normal color without any other noticeable changes in the coating other than, perhaps, its electronic characteristics. The surface topography of as-deposited coatings is quite complex. Not only do molten or highly plastic drops flow and conform to the rough surface (grit-blasted substrate or previously-deposited coating), but some particles rupture with subdroplets “skittering” along the surface for some distance before sticking. An excessive amount of this behavior leads to higher porosity, poor intracoating fracture strength, higher roughness, and lower deposition efficiency. An occasional microcrack can be observed in some coatings, particularly cermet and oxide coatings, due to residual stresses developed within an individual particle during freezing. While usually undesirable, they may not be significantly detrimental to performance (e.g., wear resistance) if they are limited in number and do not propagate from one particle to another. For some thermal barrier applications, they are intentionally induced during deposition to increase thermal shock resistance (in part, by lowering the in-plane elastic modulus). 4.2

Microstructure

Both plasma and detonation gun coatings consist of many layers of thin lamellar particles, the result of the impact of molten or semimolten powder particles as illustrated in Fig. 11.7. The major microstructural difference between the two types of coatings is that detonation gun coatings have a higher density. The impacting particles may split with some small droplets branching out or separating from the central particle. Thus, the average splat volume may be smaller than the average starting powder size, and the total surface area much larger in the coating. Typically, a splat may be a few microns thick and 10 to 50 microns in diameter. The cooling rate of the impacting particles has been estimated[22] to be 4 10 to 106 C/s for oxides and 106 to 108 C/s for metals. It is evident, however, that rates may vary significantly with the substrate material and thickness of the coating. As a result of the rapid cooling, some coatings have been found to have no crystallographic structure by x-ray[23] or neutron diffraction, [24] or low temperature heat capacity measurements.[25] Others may have a thin amorphous layer next to the substrate followed by crystalline layers.[22] Many coatings form columnar grains within the splat in one or two layers perpendicular to the surface of the substrate, Fig. 11.8.

638


Figure 11.7(a). Cross-section micrographs of detonation gun WC-9Co (top), detonation gun WC-15Co (center), and plasma WC-12Co (bottom). As-polished, DIC.


639

Figure 11.7(b). Cross-section of a detonation gun alumina-titania coating. Aspolished.

Figure 11.7(c). Cross-section of a detonation gun tungsten-cobalt coating. Aspolished, DIC.

640


Figure 11.8. Scanning electron micrographs of a fractured plasma-deposited tungsten coating (top) and a cross-section of the same coating polished and etched showing the columnar grain structure within the lamellar particles.


641

In most cases where crystalline structure can be determined by xray diffraction, the peaks are quite broad, indicative of high local residual stresses due to the rapid quenching. Also as a result of the rapid quenching, non-equilibrium phases may be present; e.g., alumina coatings[27] usually consist of a high volume fraction of gamma and other phases in addition to the equilibrium alpha. In particular, when the particles are highly superheated and impact on a substrate with high thermal conductivity, delta and theta may be formed in addition to gamma, with alpha suppressed. Similar effects may occur in detonation gun coatings, as illustrated in Fig. 11.9 for a Laves phase coating. In addition to phase shifts due to the rapid quench, some changes in composition may occur due to selective evaporation of one component in an alloy, to decomposition to a gas, or to reaction with the atmosphere as previously mentioned. If the loss of a component with a high vapor pressure can be predicted, it can obviously be compensated for in the powder manufacture. It must be kept in mind, however, that such a loss will be more rapid from a fine powder than a coarse powder, and it becomes even more imperative to use a narrow powder-particle-size distribution to ensure a homogeneous coating composition. The slight decomposition or loss of oxygen in oxide coatings has already been noted relative to color changes. Zirconia coatings are an example of this effect. The reaction of the powder particles with their local environment in transit, particularly the extent of their oxidation, is very important to the properties of the coatings. The loss of carbon from tungsten carbide plasma coatings through oxidation of WC to form gaseous CO, W2 C and free tungsten has been reported.[26]-[28] Metallic or cermet coatings may also react with air inspirated into the plasma stream, as previously noted, forming oxide scales on the particles, or dissolving the gases in the molten droplet. The effects on the properties of the coating can be extensive as shown in Sec. 4.6. The extent of these reactions varies greatly with the type of plasma torch used as shown earlier in Fig. 11.4. None of these torches used an inert gas shroud or low pressure chamber and none were made in an inert gas chamber, yet the extent of oxidation is extremely different. Similar effects can be obtained with detonation gun and HVOF coatings, both by reaction with the combustion/ detonation gas mixture and with air after the powder leaves the barrel. An example of reaction with the gas mixture is the carburization of a Laves phase alloy for added wear resistance, Fig. 11.9.

642


Figure 11.9. Cross-section of a Laves phase d-gun coating (LDT-400), as-coated (top) and after 4 hrs at 1080°C in vacuum (bottom) illustrating the metastability of the as-coated structure. The arrow identifies a carbide formed by reaction with the detonation gases during deposition. Etched, DIC.


4.3

643

Bond Strength

Bond strength is, quite naturally, an important property of a coating. It is most frequently measured in a tensile test (ASTM-C633) in which the coating is applied to the face of a one inch diameter round bar, and a mating bar is attached to it, usually with an epoxy. The limit of the test is the strength of the epoxy, currently about 10,000 to 12,000 psi (69 - 82 MPa). Most plasma coatings have bond strengths below this, but almost all detonation gun coatings and some plasma and HVOF coatings have strengths that exceed it, with the test serving only as a “proof” test. The ASTM test procedure specifies a coating thickness of at least 0.020 inches (0.51 mm). This was established to prevent penetration of the epoxy through porous coatings, such as oxy-acetylene flame spray coatings, and is usually reduced for use with the denser advanced thermal spray coatings to more closely measure their strength at a thickness more typically used in service, e.g., 0.010 inches (0.25 mm). A variety of other tensile bond strength and shear strength tests have been used, but most introduce undue stress risers. In recent modified Ollard tests of some detonation gun coating, the tensile strengths exceeded 25,000 psi and some Super D-gun coatings exhibited strengths exceeding 45,000 psi. An epoxied lap shear test is still used for some quality control purposes. It is unfortunate that more satisfactory shear tests have not been developed, since the coatings are more often loaded in shear than in tension. The mechanism of bonding of plasma-deposited coatings in many respects is still in dispute.[29]-[31] Mechanical interlocking has been considered the most important mechanism by most investigators.[30] Grit blasting, as previously discussed, provides a good surface topography for interlocking, and it has been shown in several studies[32]-[34] that bond strength increases with increasing surface roughness in both shear and tensile tests, although it may diminish again above 250 - 300 x 10-6 inches RMS. It is also interesting to note that grit blasting significantly increases the total surface area available for “chemical” bonding. Only a few published reports have taken into account the detrimental effects of surface embrittlement, peak blunting, and grit inclusions[26][35][36] when excessive grit blasting is used. Other mechanisms have been suggested as contributing to the bond strength including Van der Waals forces, interdiffusion or alloying, epitaxy, oxide cementation or spinel formation, and surface reaction. There is some microstructural evidence that when the refractory metals, tungsten or molybdenum (with high melting points and heat capacities), are sprayed on steel or nickel or chromium on aluminum, there is some interdiffusion; i.e., a

644


metallurgical bond is formed.[37] Similar results are reported for the so-called exothermically reacting nickel aluminide coatings.[38] Interdiffusion and/or a more nebulous “surface reaction” may depend in part on added surface energy in the substrate due to grit blasting, as evidenced by recrystallization of the surface.[39] The bond strength of ceramic coatings is generally attributed to interlocking, but some degree of spinel formation or similar reaction has been reported for Al2O3 on steel[23] and Al2 O3/TiO2 on aluminum.[33] Oxide cementation was cited as important in bonding metals or cermets to metals[4l][42] in earlier work, but is not considered desirable in modern practice. In general, oxides on the surface of the substrate or oxidation occurring during spraying[43] decreases bond strength. Most of the factors that affect the bond strength of “conventional” plasma deposited coatings also apply to detonation gun coatings and some HVOF and “high velocity” plasma coatings, but the situation may be somewhat more complex. Because of the unusually high velocity of the particles, some, particularly carbide or oxide, particles are actually driven into the surface of most metallic substrates. As previously noted, some substrates require no grit blasting to achieve adequate bonding, since the coating itself roughens the interface. This embedding/roughening process creates atomically clean interfaces between the coating and substrate over most of the coating area, which facilitates chemical bonding and can be likened to the explosive bonding of sheets of metal. This undoubtedly plays a role in forming the unusually high bond strengths of such coatings. A more theoretical approach to the impact of thermal spray particles on a substrate was undertaken by Houben.[44] His thermodynamic and mechanical considerations provide, perhaps, some insight into the conversion of the kinetic energy, predicting that coarse grains at high velocity may explode on impact, the temporary inversion of liquid into an amorphous solid. He also provides a method to calculate the shock, nonequilibrium, and final temperature of the material. Only a qualitative discussion of the lateral spreading or flow of the material is given along with illustration of the flow of both wetting and nonwetting material. Wetting almost always leads to cracking of the adhering material, while nonwetting leads to a weak interface. 4.4

Residual Stress

Residual stress has already been discussed to some extent in Secs. 4.1 and 4.2, but a few additional remarks may be in order. It occurs as a result of cooling individual powder particles or splats from above their melting point


645

to the temperature of the part and is usually tensile. The magnitude of the residual stress is a function of torch parameters, deposition rate, the relative torch to part surface speed, the thermal properties of both the coating and the substrate, and the amount of auxiliary cooling used. The use of finer powders frequently leads to higher residual stresses, but this can generally be controlled by adjusting the coating parameters. If the part temperature is allowed to rise above room temperature, there will be a secondary change in the state of stress of the coating as both the part and the coating cool to room temperature. Residual stress frequently increases linearly with coating thickness above some minimal initial thickness.[45]-[47] The rate of increase, however, is a function of the parameters of deposition already listed and the coating material. While the residual stress in most thermal spray coatings is tensile, the stress in some detonation gun and perhaps a few HVOF coatings is moderately compressive. This is thought to be due to the relatively high kinetic energies carried by the impacting powder particles, particularly some of the cermets. With the extraordinary velocities and kinetic energies of the powder particles in Super D-Gun deposition, very high compressive stresses can be developed if it is desirable. Residual stress may have a significant effect on bond strength, as already noted, and must be considered when the coating is placed in service, since it may detract from its inherent mechanical strength. For example, coatings are frequently in tension as a result of the residual stress, and this stress must be subtracted from the allowable fracture stress calculated from mechanical property tests of free-standing specimens. Residual stress is, however, reproducible and can be controlled with adequate knowledge of the stress and adequate control of the coating parameters. 4.5

Density

As with most properties of coatings, density is a function of the angle of deposition and substrate geometry. At high angles of deposition, the density of detonation gun coatings is greater than 95% of theoretical, usually greater than 98%. This high density is due, as with other properties, to the unusually high kinetic energy of the particles on impact. Plasma coatings have densities varying from less than 80 to 95% of theoretical with some of the “high velocity” plasma and HVOF coatings being reported with densities greater than 95%. The density of a plasma sprayed coating is, of course, a function of the deposition parameters. In addition, it is a function of the powder

646


size, as illustrated in Table 11.4[44] for tungsten carbide-cobalt, with finer powders producing denser coatings. The same effect has been noted in many other systems, e.g., Ni-Cr-Al,[49][50] and chromium carbide-nickel aluminum.[23] It has also been shown that oxidation during deposition can decrease coating density as shown by comparing the densities (and tensile strength) of stainless steel and aluminum coatings sprayed in argon with those sprayed in air, Table 11.5.[51] The combined effects of powder size and oxidation during deposition are shown for tungsten in Table 11.6.[51] Similar results were found for nickel.[33]

Table 11.4. Effect of Powder Size on the Structure of Plasma Deposited Tungsten Carbide

Coarse (10 - 105)

Coating Property

Apparent Density (g/cc) Bulk Density (g/cc) % Theoretical Apparent Hardness (Kn5OO)

Powder Size Medium (10 - 74)

Fine (10 - 44)

10.5

13.0

14.2

8.7

11.1

13.0

60

77

89

538

684

741

Table 11.5. Properties of Plasma-Deposited Coatings Sprayed in Argon and Air

Coating

Stainless Steel

Aluminum

Atmosphere

Density (%)

Tensile Strength (psi)

Argon

91

33,900

Air

84

19,200

Argon

86

5,600

Air

76

4,000


647

Table 11.6. Properties of Plasma-Deposited Tungsten

Powder Size

Deposition Atmosphere

Density (%)

Modulus of Rupture (psi)

Average Grain Diameter (µ)

200+325 Mesh

Argon Argon* Air

90 70* 86

31,900 21,700 17,000

3.5 3.5 3.0

400 Mesh + 10

Argon Air

91 85

51,000 29,000

2.2 1.5

*Intentionally produced with low density.

The porosity in plasma and detonation gun coatings is partially interconnected and, hence, may have a strong influence on the corrosion rates of the coatings in some environments. Some detonation gun coatings have been shown to have sufficiently small pores as to be unimportant in oxidation in air at high temperatures.[52] Electrochemical corrosion studies have shown that several Super D-Gun™ coatings are impervious to aqueous media when more than about 100 microns thick. It should be noted that porosity levels are frequently inferred from the observation of metallographic specimens, but that there can be (and usually is) a significant difference between these measurements and the true porosity. Therefore, such measurements should always be identified as metallographic apparent porosity. True porosity can be more accurately measured through density measurements, albeit there is sometimes a problem in determining the theoretical density because of a lack of knowledge of the relative amounts of various phases present, or porosimetry measurements (gas or mercury). Metallographic apparent porosity can, nonetheless, be useful for quality control, if reproducible metallographic techniques are employed. 4.6

Mechanical Properties

The mechanical properties of advanced thermal spray coatings are sensitive to the angle of deposition, other deposition parameters used, the substrate, cooling, etc. Therefore, any general tabulation of properties based only on coating composition would be meaningless. Moreover, most of the data has been generated on specimens coated under ideal conditions of angle and standoff while in many service applications both of these variables may be

648


less than ideal on part or all of the coated area. Nonetheless, a good deal of data has been compiled on a wide range of coatings to serve as very useful guidelines to equipment designers and other users. For purposes of illustration in subsequent discussion and to provide a general feeling of comparison with other types of materials, the mechanical properties of a few specific coatings are listed in Table 11.7. The moduli of elasticity and rupture and strain-to-failure were measured on free-standing rings of coatings 1 inch in diameter, 0.5 inch wide and 0.010 inch thick. The mechanical (as well as other) properties of advanced thermal spray coatings are anisotropic because of their splat structure and directional solidification. This anisotropy is probably more pronounced for cermets and metallic coatings with somewhat oxidized splat boundaries than it is for either pure ceramic or pure metallic coatings. An example of this anisotropy is given in Table 11.8.[53] Although most coatings are used with loading normal to the surface, measurement of mechanical properties normal to the surface is particularly difficult because of the limited thickness of most coatings and is seldom done. Properties parallel to the surface are also important, however, particularly if the substrate expands or contracts thermally or under mechanical loading. The most frequently quoted mechanical property is hardness. The hardness of the higher-velocity coatings is generally higher than that of conventional plasma coatings of the same composition as shown in Table 11.3. This is primarily due to their higher density and greater cohesive strength. For a plasma coating with a given composition, the hardness usually increases with an increase in density. Thus, for example, hardness generally increases with the use of a finer powder, as already shown in Table 11.4. Hardness may be reduced for a given material if the coating is applied in an inert atmosphere as compared to spraying in air, as has been noted for WC-Co,[43] for Mo,[54] and for Ti, Nb, and Zr.[55] Although it may increase the hardness of the coating, excessive oxidation will weaken its internal cohesive strength and may be detrimental to the coating’s performance. Hardness is used not only as a guideline for wear resistance, but for the strength of the coating. In both cases it may be quite misleading. The measurements of hardness are usually made on metallographic crosssections of the surface, even though loading is usually perpendicular to the surface, and the hardness in the two directions may be different due to the anisotropic microstructure of the coatings. Hardness measurements made on test specimens may differ from those on actual parts due to differences in angle of deposition and stand-off and, in some cases, residual stress.

Table 11.7. Properties of D-Gun and Plasma Coatings

Advanced Thermal Spray Deposition Techniques 649

650


Table 11.8. Mechanical Properties of Plasma Cu-2Be* in Compression Parallel to Surface Elastic Modulus 0.2% Yield Strength Ultimate Strength** Strain to Fracture**

13 x 106 psi 82 x 103 psi 97 x 103 psi 3.3%

Perpendicular to Surface 10 x 106 psi 73 x 103 psi 164 x 103 psi 26%

* Union Carbide, UCAR LCU-3 ** Function of specimen geometry

The following is an example of a situation in which hardness, used as a guide to wear resistance, was the initial criteria for coating selection and too little weight given to impact resistance or toughness: Most midspan shrouds on gas turbine engine compressor blades have a detonation gun tungsten carbide-cobalt coating. In the initial development of this application, the most wear-resistant grade of tungsten carbide-cobalt with a hardness of 1300 HV300 and nine percent cobalt was tried. This coating was found to fail, however, not from typical wear, but because of surface fatigue which resulted in spallation of the coating. Success was achieved when a more impact-resistant grade of tungsten carbide with a hardness of 1075 HV300 and 14% cobalt was tried. The greater “toughness” of this coating, combined with a wear-resistance that is still excellent, solved the problem. The modulus of rupture, elastic modulus, and strain-to-fracture in bending of plasma and detonation gun coatings has been measured more often than conventional uniaxial tensile and compressive properties. The former measurements can be made on free-standing rings of coatings as thin as 0.010 inches. On the other hand, it is often difficult to produce coatings thick enough for conventional specimens. This difficulty arises from the thickness limitations of some coatings due to residual stress and the inherent brittleness of the coatings. Even most metallic coatings have a strain-to-failure of less than one percent. Some typical values from ring tests in Table 11.7 show that the detonation gun coatings have a higher modulus of rupture than comparable plasma coatings; compare, for example, the tungsten carbide-


651

cobalt coatings. Also note that, as expected, increasing the cobalt content increases the strain-to-fracture (either measured directly or calculated from the ratio of modulus of rupture to the elastic modulus). An example of the use of this kind of test data is as follows: A plasma chromium oxide coating was specified on the interior surface of an aluminum hydraulic cylinder in an aircraft landing gear, because of its earlier success on another landing gear and cyclic pressure bench testing on prototype cylinders without pistons. Even though visual examination of the bench tested cylinders revealed no irregularities, the coating failed when the complete assembly was placed in service. The cylinder expansion under pressure clearly exceeded the strain-to-failure of the coating and it cracked. The additional stress of the piston caused spalling. Re-examination of the bench-tested cylinder revealed microscopic cracks. The designer had not adequately taken into account the difference in elastic moduli between the coating and the substrate and the limited strain-to-failure of the coating. Evaluation of data from ring tests indicated both aluminum bronze and nickel coatings had adequate strain-to-failure. Subsequent tests verified this as well as the fact that they had sufficient wear resistance. It is obvious, of course, that all the coating process variables and the resulting microstructures strongly affect the mechanical properties of the coating. For example, tungsten coatings made with fine powder have a higher modulus of rupture than those made with coarse powder when both are protected from oxidation by spraying in an inert atmosphere,[51] as shown in Table 11.6. Referring to Tables 11.5 and 11.6, it is apparent also that oxidation during deposition can seriously weaken a coating. In a study of the effect of oxidation on aluminum bronze, it was found that even minor oxidation during deposition was detrimental to compressional strength, both parallel and perpendicular to the surface, Table 11.9.[53] Additions of discrete oxide particles, on the other hand, not only strengthened the coatings, but added wear resistance (discussed in Sec. 4.7). Before leaving the subject of mechanical properties, it might be well to mention that the properties of the substrate cannot be ignored in considering a coating application. One of the first considerations is that the substrate must be able to support the coating without yielding beyond the coating’s strain-tofailure as shown in the following:

652

Mechanical Properties of Plasma-Deposited Aluminum Bronze (Cu-IOAl)

Type of Deposition

Alumina Addition

Vol.% Al2 O3

Hardnessa HV 300

Compressional Properties Perpendicular to Surface Parallel to Surface E (10 6 psi)b YS (106 psi)b E(10 6 psi) b YS (103 psi)b

Standard

No

2.36

246

7.1

58

8.4

62

Oxidizing

No

3.20

200

2.9

47

7.0

42

Standard

Yes

7.26

170

—

—

—

—

Standard

Yes

12.4

202

8.5

73

9.7

89

Standard

Yes

21.0

246

—

—

—

—

Standard

Yes

40.5

186

—

—

—

—

Oxidizing

Yes

10.8

142

7.6

57

10.9

70

—

252

—

—

—

—

Wrought (AMS 4640)

—

a Hardness perpendicular to surface. b

E is the elastic modulus, YS is the 0.2% yield strength.


Table 11.9.


653

A detonation gun tungsten carbide-cobalt coating has extended the life, by a factor of ten, of roller guides used in steel mill pickle lines. In the first trials of this coating, however, the coating occasionally cracked and spalled because the substrate yielded under the heavy impact of the steel sheet. This deformation exceeded the strain-to-failure of the coating. When a change was made to ensure that all substrates had a hardness greater than 55HRC no failures were experienced. In some applications, the coating affects the fatigue life of the substrate. Some coatings, particularly detonation gun coatings, are so well bonded that a crack generated in the coating may propagate into the substrate under sufficient cyclic stress. The results of a number of studies, especially by airframe and gas turbine engine manufacturers, suggest that as long as the strain-to-failure of the coating is not exceeded, the coating has no measurable effect on the fatigue strength of the substrate. More work needs to be done, however, before the effects of a specific coating on a given substrate can be predicted without experimental verification. In those cases where stresses are very high and the component is particularly susceptible to fatigue, care should be taken to prevent both direct coating and overspray. For example: The midspan shrouds or stiffeners used on many titanium compressor blades must be coated with a detonation gun tungsten carbide-cobalt coating, as previously mentioned. The root area of the midspans is extremely sensitive to fatigue and all coating and overspray must be excluded. This is successfully achieved by either very careful masking or directing the coating away from the radii during deposition. 4.7

Wear and Friction

The major use of advanced thermal spray coatings today is for wear resistance, particularly for adhesive and abrasive wear resistance. Their use in erosive situations is growing steadily as well, particularly for detonation gun coatings. No attempt will be made here to tabulate the wear resistance of coatings or, conversely, to recommend specific coatings for the various types of wear. To do so, it would be necessary to assume that all coatings of a given composition are the same (while, in fact, they are a function of the specific coating device and operating parameters used), and to assume that all wear situations can be fit into a relatively few, well defined categories (which is definitely not the case). The situation is far from hopeless, however, and experienced coatings service engineers or equipment manufacturers

654


can, after analysis of a specific situation, usually make reasonably accurate recommendations of one or two coatings that will solve the problem. Some of the considerations that are involved are listed in Table 11.10. Table 11.10.

Considerations in Coating Selection for Wear Resistance

I. Wear System A. Adhesive or Abrasive 1. Type of relative motion—unidirectional, oscillating, impact 2. Surface speed—velocity and frequency, if cyclic 3. Load or impact energy 4. Abrasive particles or wear debris—trapped or removed, size, shape, and composition 5. Conformability requirements 6. Embeddability requirements B. Erosive 1. Gas, liquid, or solid particle erosive material 2. Media—gas or liquid 3. Gas or particle velocity and angle of impingement 4. Particle size, shape, mass, and composition II. Environment A. Temperature—maximum, minimum, and rate of change B. Media—gas or liquid C. Contaminants D. Corrosive characteristics—chemical, galvanic E. Lubricant III. Mating Material A. Composition B. Heat-treatment condition C. Hardness D. Surface roughness and topology IV. Substrate Material A. Composition B. Heat-treatment condition C. Dimensional changes after coating 1. During assembly due to press fit, shrink fit, etc. 2. In service due to thermal expansion/contraction or mechanical loading V. Coating Requirements A. Cost limitations B. Required life, time or maximum wear C. Compositional limitations D. Thickness limitations E. Coefficient of friction requirement F. Surface finish G. Geometric constraints H. Overspray limitations


655

Although no specific recommendations will be made here, a few general comments may be in order. Hardness is a useful first approximation to abrasive and adhesive wear resistance as long as materials of the same type and general composition are compared. For example, a detonation gun WC-9Co coating is harder and more wear resistant than detonation gun WC15Co which, in turn is more wear resistant than a plasma WC-13Co coating, Table 11.11. (An example, compressor midspans, has already been cited as an application where hardness and wear resistance had to be tempered with toughness for success, however.) Hardness can be misleading, however, when comparing coatings with wrought materials of the same composition. For example: Plasma-deposited aluminum bronze or beryllium copper coatings are softer than their wrought counterparts. In an adhesive wear test under boundary lubrication conditions simulating many bearing applications, the plasma coatings were far more wear resistant, as shown in Fig. 11.10.[56]

Table 11.11.

Wear Tests

Material

Dry Rubbing Wear Rate (10-6 in/1000 ft. of sliding)

LFW-1 450 lb/load in Hydraulic Fluid vs. Steel for 5409 rev. (10-6 cm3 )

Detonation Gun Tungsten Carbide-Cobalt

35

10

Plasma Tungsten Carbide-Cobalt

80

23

52100 Steel (Wrought)

2,000

—

Hard Chrome Electroplate

3,600

44

656


Figure 11.10. Alpha block-on-ring wear test of aluminum bronze, wrought (Al-Cu) and plasma-sprayed (PD Al-Cu), and beryllium copper, wrought (Be-Cu) and plasma-sprayed (PD Be-Cu) vs. SAE 4640 steel (Rc 60) in hydraulic fluid at 65 ft/ min for 1950 ft.

There are also situations in the comparison of coatings where hardness can be misleading, particularly in adhesive wear with coatings of somewhat different compositions. For example: As shown in Table 11.9, the addition of an oxide dispersion to an aluminum bronze coating slightly reduces its hardness, yet in an adhesive wear test its wear resistance is increased significantly, as shown in Fig. 11.11.[57] Similar


657

results were obtained with carbide additions. It should be noted that this increase in wear resistance should not affect the conformability and embeddability of the basic aluminum bronze. It is also important to note again that an intentional oxide dispersion deposited under conditions that do not significantly oxidize the metal matrix is far superior to a coating heavily oxidized during deposition, both in wear resistance and mechanical properties.

Figure 11.11. Alpha block-on-ring wear test of aluminum bronze with Al2O3 addition (block) vs. SAE 4640 steel (Rc 60) in hydraulic fluid at 65 ft/min for 1950 ft under a 180 lb load. (#) Wrought alloy; (∆) standard plasma; (◊) oxidizing plasma; ($, *) alumina additions to plasma; (U,X) plasma pure alumina.

658


Physical and chemical compatibility of the mating surfaces are, of course, important in selecting a coating. Laboratory testing can be an excellent guide in this aspect of selection as long as the other considerations (temperature, load, lubricant, etc.) are similar to those in service. A few examples of satisfactory and unsatisfactory combinations are shown in Table 11.12.[45][53] Table 11.12.

Mating Materials Selection for Dry Rubbing Wear

Materials D-Gun WC-CO* vs.: GA Meehanite

Rating

Wear**

Coefficient of Friction

Excellent

33

0.08 at Room Temp. 0.11 at 400°F

440 Stainless

Good

35

0.34 at Room Temp. 0.25 at 1000°F

Inconel X

Poor

562

0.53 at Room Temp. 0.42 at 1000°F

D-Gun WC-CO

Good

39

0.46 at Room Temp. 0.33 at 100°F

Haynes 25

Excellent

16

0.25 at Room Temp. 0.17 at 1400°F

Hastelloy C

Good

35

0.32 at Room Temp. 0.10 at 1400 F

D-Gun Al2 O3

Poor

245

D-Gun Al2O3 vs.:

0.24 at Room Temp. 0.27 at 1400°F

* Rotating member ** Total system wear, 10-6 in/1000 ft.

In some applications, both surfaces can be coated. Combinations of very hard coatings such as chromium oxide provide excellent self-mating characteristics when no conformability or embeddability of either surface is required. When these are required, a combination such as plasma-deposited aluminum bronze with an oxide dispersion versus several types of hard detonation gun coatings offer both mechanical compliance and greatly increased wear resistance.


659

The erosion resistance of conventional plasma deposited coatings is not very high, probably because of their porosity and relatively low cohesive strength. Detonation gun and some HVOF and high velocity plasma coatings, on the other hand, have shown exceptional erosion resistance in some applications. For example: Some compressor blades of gas turbine engines may suffer from severe particle erosion. Tests and service experience have shown that detonation gun coatings of (tungsten, titanium) carbide-nickel and tungsten carbide-cobalt coatings significantly increase the lives of the blades. The coatings are only applied to the outer portion of the blades where the erosion problem is most severe to reduce their cost and avoid any potential fatigue effects on the blades. Coatings can be used to adjust the frictional characteristics of a system whether or not wear is a problem. Thus, for example, coatings might be used to reduce power losses through frictional heating. Reduction in frictional heating can also extend the life of mating organic materials. For example: A manufacturer of bearing and sealing systems for the power trains of large ships was experiencing difficulties with the forward seal for the tailshaft. This seal is located well inside the ship and is not cooled by the outside water. The seal is formed by mating a rotating shaft liner with stationary rubber seals. The liner, made of a special alloy, was sufficiently wear resistant without a coating; however, the heat generated by friction caused a rapid deterioration of the rubber seals. The problem was solved by (a) incorporation of a cooling device for the oil in the system and (b) reducing the friction at the seal by the adoption of a specially finished plasma-deposited chrome oxide coating on the rotating shaft liner. The special finish minimizes contact with the mating rubber material while still maintaining the necessary seal. The rubber seal now operates at a lower temperature, and its life is significantly extended. Needless to say, there is no sacrifice in the wear life of the liner with the addition of the coating. Occasionally it is necessary to prevent self-welding between essentially static components and ensure that the static coefficient of friction will be low enough to prevent equipment start-up failure. For example: The sodium-cooled breeder reactor requires that both of these criteria be met by the load pads on the fuel ducts.

660

Deposition Technologies for Films and Coatings Uncoated stainless steel, stripped of its oxide film by the sodium, is self-welding. Extensive testing[14] has shown that a solution to the problem is a detonation gun coating of chromium carbide-nichrome.

It is evident from the preceeding that the surface finish of a coated surface is extremely important. The smoothest finish that can be obtained on a given coating is a function of its composition and the method of deposition. However, it should be borne in mind that the lowest coefficients of friction are not always obtained with the smoothest surface. A nodular brush finish, for example, provides the best frictional behavior in liquid sodium. In other applications, specific, intermediate-range coefficients of friction are used. For example: Textile machinery components, such as snick plates, tension gates, and draw rolls are in contact with fast-moving fibers being processed and are, of course, subject to high rates of wear. Hard plasma and detonation gun coatings are used to resist this wear. Equally important, however, the coating finish must provide rather precise intermediate frictional properties to hold the fiber in constant tension. The tension control is essential in order to prevent slack or breakage of the fiber. High friction forces are required in many types of drive mechanisms. Coatings can often meet this requirement and provide longer life than most other solutions. For example: Many large rolls in sheet steel production rely on high surface friction to move the steel sheet through the line without slippage. Others require this gripping action in order to tightly wrap the steel sheet into non-telescoping coils. Experience has shown that a detonation gun tungsten carbide-cobalt coating, used as-coated or slightly roughened by controlled grit blasting, resists wear, grooving, and gouging 6 to 40 times longer than the previously used hardened steels or chrome plate. An additional benefit of the coating is its resistance to the transfer of material from the steel sheet to the roll, which is a rather common problem with other materials. 4.8

Corrosion Properties

Obviously the use of a coating in a corrosive environment requires that the coating itself resist the corrodant, but it should be kept in mind that the


661

corrosion resistance of a wrought, cast or sintered composition may change when deposited by thermal spray. For example, alpha alumina is very corrosion resistant, but plasma sprayed alumina is a mixture of phases, all of which are not corrosion resistant. Virtually all advanced thermal spray coatings have varying degrees of interconnected porosity that allow attack of the substrate in corrosive environments. In most cases, at temperatures up to about 350°F, this may be at least partially overcome by the proper selection and application of a sealant. For example: Bronze shaft sleeves running in centrifugal pumps handling saturated brine in a chlorine processing plant were being rapidly worn beyond tolerance. They ran against asbestosfilled Teflon™ with no lubrication at 250°F. The solution was a machineable metallic plasma undercoat to restore the sleeves to size followed by plasma-deposited chromium oxide coating that was sealed with epoxy to inhibit substrate and undercoat corrosion. The coating was ground to a 4 to 6 x 10-6 inches RMS surface. The coated sleeves not only salvage worn parts, but outlast the original sleeves several times and reduce downtime. Galvanic corrosion can occur in some environments, most commonly salt water, when an improper selection of coating composition is made. For example: An aluminum bronze coating on an aluminum substrate creates a galvanic cell in the presence of an electrolyte. In aircraft landing gear cylinders with this coating/substrate combination, galvanic corrosion of the substrate was observed when the hydraulic fluid became contaminated with salt water. The problem was solved by sealing the coating. The problem might also have been avoided by selecting a modified aluminum coating with an electrostatic potential virtually identical to the substrate. Recently, several Super D-Gun coatings have been shown to be effectively impervious to aqueous media when they are more than a few microns thick. These coatings include both corrosion resistant metal alloys which can be used singly or as an undercoat to form a corrosion barrier and tungsten carbide-based coatings which can be used without an undercoat or sealant in applications requiring both wear and corrosion resistance. For corrosion resistance at elevated temperatures, plasma-deposited coatings must be sealed by sintering, sometimes combined with mechanical

662


surface treatment. For example: Some gas turbine blades and vanes, depending on the type of fuel and operating environments, are subject to hot corrosion. The best solution to this problem at the present time is an MCrAlY (where M is Ni, Co, and/or Fe) type of coating. (Other elements may also be present; e.g., Pt, Hf, or Si.) These coatings were first applied commercially by physical vapor deposition, an expensive method with some elemental limitations. Plasma deposition offers significant economic advantages and has no elemental limitations. To be effective, however, the coatings must be deposited without oxidation and then sealed to prevent rapid internal oxidation of the coating and oxidation of the substrate. Methods have been developed to achieve this using inert gas shrouding during deposition and post-coating heat-treatment and peening to effectively sinter the coating. Alternative methods using deposition in a low pressure, inert gas chamber followed by peening and heat-treatment have also been developed. Since the substrates are superalloys, the coating heat-treatment can be combined with or precede the alloy heat-treatment and not interfere with the structural properties of the component. Detonation gun coatings, because of their high density, often do not need to be sealed with a high temperature sintering to prevent internal oxidation or oxidation of the substrate. For example: For many years detonation gun coatings of chromium carbide-nickel chromium have been used on the shroud edges and lacing wire of gas turbine engines to prevent fretting and impact wear. More recently a new family of cobalt based alloys with oxide additions has been developed[52] to provide better performance for more advanced engines. These coatings do not require heat-treatment to prevent internal oxidation, although heat-treatment is used to further improve the already superior bond strength in particularly severe impact situations. 4.9

Thermal Properties

The thermal properties of coatings are important both during their formation and in elevated temperature applications. The effects of thermal contraction from their freezing point during coating formation have already been mentioned. Conversely, the relative thermal expansion of the coating and substrate if heating occurs during service is important. As a rough estimate


663

of the strain that may be placed on a coating as a result of such heating, handbook values of coefficients of thermal expansion may be used. Care must be taken to ensure that the values used are those for the phases actually present in the coating. Because of their lamellar microstructure, the thermal conductivity of coatings is lower than that of solid, fully dense materials of the same composition. Their absorption characteristics may be very different, because of their surface topology and, in some cases, slight shifts in composition (already mentioned for some oxides). One of the most common uses of coatings for their thermal properties is as thermal barriers. For purposes of discussion, thermal barriers may be divided into two categories—relatively thin ones, less than about 0.020 inches thick, and thicker ones, up to about 0.25 inches thick. The thinner thermal barriers have been used for years on gas turbine engine combustion chambers and are currently used to a limited extent for turbine blade and vane airfoil surfaces, thrust reversers, diesel and combustion engine piston heads and valves, and many other applications. In addition to having a low thermal conductivity, these coatings must be resistant to corrosion, thermal shock, gas erosion and, sometimes, particle erosion. They usually consist of a metallic undercoat such as nickel-chromium, nickel aluminum, or an MCrAlY alloy (where M is Ni, Co or Fe) and an outer layer of an oxide, usually zirconia or magnesium zirconate. Occasionally one or more intermediate layers of mixtures of metal and oxide or a continuous gradation from pure metal to pure oxide is used. This approach improves thermal shock resistance, but if the temperature in service at the first zone of mixed metal and oxide is too high, the metal will rapidly oxidize (since the oxide layer is permeable to air) and cause spallation of the outer portion of the coating.[59] The same thing will happen to the metallic undercoat in a two layer coating if it has inadequate oxidation resistance because of its composition or because it is too porous. These thick thermal barriers are being investigated on outer air seals in advanced gas turbine engines. In this case, the already complex task of increasing thermal shock resistance without sacrificing oxidation resistance and erosion resistance is complicated by the need for abradability. Two layer systems appear to be best for most outer air seal applications. In thick thermal barrier systems it may be advantageous to use an essentially continuous gradation from metal to oxide or multiple layers with increasing oxide content to have adequate thermal shock resistance. As in the case of thin thermal barriers with intermediate layers, the temperature at which the outermost

664


metallic component is exposed must be low enough to prevent any significant oxidation of the metal. 4.10 Electrical Characteristics The microstructure of metallic coatings has an effect on the electrical conductivity similar to that on the thermal conductivity. Thus the resistance is higher than that for wrought alloys of the same composition, and it is somewhat higher perpendicular to the surface than parallel. The conductivity of coatings deposited with an inert gas shroud or in an inert gas chamber with very little oxidation during deposition is much higher than conventional coatings, since the conductivity is particularly sensitive to oxide films in the splat boundaries. Coatings are used as both conductors and insulators. The use of oxides as an insulator is fairly obvious, but the flexibility that this type of coating offers the designer is often overlooked. For example: Aluminum oxide coatings applied to the tips of pliers, screwdrivers, and diagonal cutters for electrical insulation are especially useful in work on confined electrical circuit installations. The coating guards against short-circuiting which would otherwise be possible during accidental contact with adjacent terminals. In a steel mill ferrostan tin line, where sheet steel is tinplated, wringer rolls are used to remove water from the stock. These rolls are usually rubber coated for electrical insulation. An insulative aluminum oxide coating instead of rubber resists the wear and grooving which, in the rubber coated rolls, eventually allow arcing and subsequent “arc burns” on the sheet steel. Coatings are usually used as conductors when the application simultaneously requires wear resistance and/or corrosion resistance. For example: Also operating in steel mill ferrostan tin lines are rollers designed to conduct electricity to the sheet stock during plating. Typically, the conductive surface of the roller has been clad copper. Experience has shown a plasma-deposited tungsten coating to be a better material selection. The conductivity of the tungsten coating is more than adequate, and it is far more resistant than copper to wear, grooving, and gouging.

Advanced Thermal Spray Deposition Techniques 5.0

665

SUMMARY

Thermal spray coating technologies are capable of depositing a very wide range of compositions without significantly heating the substrate. The range of surfaces or components which can be coated is, however, limited by the line-of-sight nature of the processes. The process technology is fairly mature, but incremental improvements will continue. On the other hand, substantially improved materials should be expected and new applications, sometimes in completely new fields, are constantly being developed. While a few suppliers of coatings have developed and implemented adequate quality control measures, this is not the norm in the industry. With proper attention to this issue, to the proper selection of a coating material and process, and its specification, however, thermal spray coatings can be used for a wide variety of proposes in virtually every industry, from submarines to space shuttles and steel mill rolls to computers, solving problems in wear resistance, corrosion resistance, thermal or electrical resistance or conductance, radiation reflectance or absorption, etc.

REFERENCES 1.

Poorman, R. M., Sargent, H. B., and Lamprey, H., “Method and Apparatus Utilizing Detonation Waves for Spraying and Other Purposes," U.S. Patent 2,714, 563 (August 2, 1955)

2.

Gage, R. M., Nestor, O. H., and Yenni, D. M., “Collimated Electric Arc Powder Deposition Process,” U.S. Patent 3,016,447 (January 9, 1962)

3.

Muehlberger, E., “Coating Heat Softened Particles in a Plasma Stream of Mach 1 to Mach 3 Velocity,” U.S. Patent 3,914,573 (October 21, 1975)

4.

Nicoll, A. R., Gruner, H., Prince, R., and Wuest, G., Surf. Eng. 1:59 (1985)

5.

Fabel, A. J. and Ingham, H. S., “Plasma Flame-spraying Process Employing Supersonic Gaseous Streams,” U.S. Patent 3,958,097 (May 18, 1976)

6.

Sokol, L. S., McComas, C. C., Hanna, E. M., U.S. Patent 4,256,779

7.

Metals Products Div., United Technologies Corp., Lantana, Florida, 1982.

666


8.

Muehlberger, E. and Kremith, R. “New Sonic and Supersonic 80 kW Plasma Spray Systems,” presented at Ninth Airlines Plating Forum, Montreal, Canada (1973)

9.

“Selected Coating Properties - The 7M High Energy Plasma System,” Metco, Inc. (1975)

10.

Wallace, F. J., “High-Energy Plasma-Sprayed Tungsten Carbide Cobalt Development for Turbine Applications,” presented at the 14th Annual Airline Plating Forum, Tulsa, Oklahoma (25-27 April 1978)

11.

Meyer, H., Ber. Dtsch. Keram. Ger. 39(H2):115-124 (1963)

12.

Jackson, J. E., “Method for Shielding a Gas Effluent,” U.S. Patent 3,470, 347 (1969)

13.

Shanker, S., Koenig, D. E., and Dardi, L. E., J. Metals, 33:13-20 (Oct., 1981)

14.

Gruner, H., Thin Solid Films, 118:409-420 (1984)

15.

Kayser, H., Thin Solid Films, 39:243-250 (1976)

16.

Price, M. O., Wolfla, T. A., and Tucker, R. C., Jr., Thin Solid Films, 45:309-319 (1977)

17.

Fabel, A. J., “Powder Feed Device for Flame Spray Guns,” U.S. Patent 3,976, 332 (24 August 1976)

18.

Smart, R. F. and Catherall, J. A., Plasma Spraying, Mills and Boon, Ltd., London (1972)

19.

Crammer, D. E., Bartoe, R. L., and Kramer, J., “Improved Universal Powder Mass Flow control for Thermal Spray Applications,” presented at International Conference on Metallurgical Coatings, San Diego, CA. (March 1987)

20.

Wolfla, T. A. and Johnson, R. N., J. Vac. Sci. Technol. 12:777-783 (1975)

21.

“Finishing - UCAR Metal and Ceramic Coatings,” Union Carbide Corp.

22.

Wilms, V. and Herman, H., Thin Solid films, 39:251-262 (1976)

23.

Taylor, T. A., unpublished data.

24.

Alperin, H. and Taylor, T. A., unpublished data.

25.

Reynolds, H. and Taylor, T. A., unpublished data.

26.

Levinstein, M. A., Eisenlohr, A., and Kramer, B. E., “Properties of Plasma Sprayed Materials,”Weld. J.; Weld. Res. Suppl. 40:8s (1961)

27.

Levy, M., Sklover, G. N., and Sellers, D. J., “Adhesion and Thermal Properties of Refractory Coating-Metal Substrate Systems,” U. S. Army Materials Research Agency, AMRA TR 66-01 (1968)


667

28.

Milewski, W., “Sonic Phenomena Occurring During Plasma Spraying WC+CO Compositions,” presented at the 7th International Metal Spraying Conference, London (1973)

29.

Tucker, R. C., Jr., J. Vac. Sci. Technol. 11:725-734 (1974)

30.

Matting, H. A. and Steffens, H. D., Metall.17(6):583 (1963); 17(9):905 (1963)

31.

Van Vlack, L. H., “The Metal-ceramic Boundary,” presented at the 1964 Metals/ Materials Congress, Philadelphia, PA., Technical Report No. P (10-1-64)

32.

Grisaffe, S. J., “Analysis of Shear Bond Strength of Plasma-Sprayed Alumina Coatings on Stainless Steel,” NASA Technical Note, NASA TN D-3113 (1965)

33.

Union carbide Corp., unpublished data.

34.

Marchandise, H., “The Plasma Torch and its Applications,” European Atomic Energy Community,EUR 2439.f (1965)

35.

Wolfla, T. A., unpublished data.

36.

Leeds, D. H., “Some Observations on the Interface Between PlasmaSprayed Tungsten and 1020 Steel,” Defense Documentation Center, AD-803286 (1966)

37.

Kitahara, S. and Hasui, A., J. Vac. Sci. Technol. 11:747-754 (1974)

38.

Longo, F. N., Weld. J. 45(2):66s (1966)

39.

Matting, H. A. and Steffens, H. D., Metall. 17(12):1213 (1963)

40.

Durmann, G. and Longo, F. N., Ceram. Bull. 48(2):221 (1969)

41.

Ingham, H. S., Jr., in: Composite Engineering Laminates, (A. G. H. Dietz, ed.), MIT Press, Cambridge (1966)

42.

Ingham, H. S. and Sheepard, A. P., Metco Flame Spray Handbook, Metco, Inc., Westbury, NY (1965)

43.

Okada, M. and Maruo, H., Brit. Weld. J., 15:371 (1968)

44.

Houben, J. M., Proc. Conf. on Thermal Spray Coatings, pp. 1-19, October 1984, Long Beach CA., (F. N. Longs, ed.), American Society for Metals (1985)

45.

Poquette, G. E., Linde Division, Union Carbide Corp., private communication

46.

Yu, S., Sharivker, Poroshk. Metall. 54(6):70 (1967)

47.

Marynowski, C. W., Halden, F. A., and Farley, E. P., Electrochem. Technol.3(3-4):109 (1965)

48.

Donovan, M., Brit. Weld. J. 13:490 (1966)

668


49.

Tucker, R. C., Jr., Linde Division, Union Carbide Corp., private communication

50.

Yenni, D. M., Linde Division, Union Carbide Corp., private communication

51.

Mash, D. R. and Brown, I. M., Met. Eng. Quarterly 18 (1964)

52.

Wolfla, T. A. and Tucker, R. C., Jr., “High Temperature Wear Resistant Coatings,” presented at Int'l Conf. on Metallurgical Coatings, San Francisco, CA (3-7 April, 1978)

53.

Tucker, R. C., Jr. and Bishop, T. N., “The Utilization of Plasma and Detonation Gun Coatings in Design,” presented to AIME Symp. on Interaction of Design and Materials II (1973)

54.

Elyutin, V. P., et al., Svar. Proizvod, 6:72 (1969)

55.

Muller, K. N., “Structure and Properties of Arc-Sprayed Titanium Coatings,” presented at the 7th Int’l Metal Spraying Conf. (1973)

56.

Tucker, R. C., Jr., and P. W. Traub, “Wear Behavior of Wrought and Plasma-Deposited Aluminum Bronze and Beryllium Copper,” presented to the Metallurgical Soc. of AIME (1971)

57.

Tucker, R. C., Jr., “Wear Characteristics of Modified Plasma-Deposited Aluminum Bronze,” presented to the Am. Soc. for Testing and Materials Symp. on Erosion, Wear and Interfaces with Corrosion (1973)

58.

Taylor, T. A., Overs, M. P., Gill, B. J., and Tucker, R. C., Jr.,J. Vac. Sci. Technol. A3:2526-2531 (Nov/Dec 1985)

59.

Tucker, R. C., Jr., Taylor, T. A., Weatherly, M. H., “Plasma Deposited MCrAlY Airfoil and Zirconia/MCrAlY Thermal Barrier Coatings,” presented at the Third Conf. on Gas Turbine Materials in a Marine Environment, Bath, England (20-23 September 1976)

A substantial amount of research and development of thermal spray coatings has occurred since the first edition of this book. No attempt has been made here to provide an extensive list of references to such work. The interested reader, however, may wish to refer to the Proceedings of the International Conference on Metallurgical Coatings, published by Elsevier Sequoia; and theJournal of the Thermal Spray Technology published by ASM International.

12 Non-Elemental Characterization of Films and Coatings Donald M. Mattox

1.0 INTRODUCTION A coating may be defined as a near-surface region having properties differing from the bulk of the material which is prepared by adding a material to the surface (overlay coating). A modified surface is a near-surface region whose properties differ from the bulk of the material and which is formed from the bulk material by changing the composition, phase, or properties; the substrate material is detectable in this region. Generally a modified surface is also referred to as a coating. These definitions imply no thickness limitation but usually involve a functional or property difference between the coating and substrate. Thus a coating allows the dissociation of the surface properties from the bulk properties and allows engineering, fabrication, and design flexibility which can be obtained by separating the surface properties from the structural requirements. Disadvantages of coatings are associated with: 1. Presence of an interface and the need for adhesion 2. A sharp discontinuity in material properties at the interface 3. Need for fabrication methods, some of which are expensive 4. Need for process control for a reproducible product 5. Properties of the coating material may differ significantly from the material in bulk form and the properties may be very process dependent

669

670


Films are thin coatings, and in some instances the film properties are influenced by the substrate properties. In this chapter, a film is defined as a coating with a thickness less than 1 micron (103 nanometers or 40 microinches). Films and coatings may be fabricated in a variety of compositional, morphological, and microstructural configurations. These include: 1. Monolithic—one composition throughout 2. Alloyed or mixed and not reacted 3. Compound 4. Graded composition 5. Layered structures—few to many, alternating 6. Composite (dispersed phases) 7. Dispersed impurities—possibly to greater than solubility limits 8. Special configuration, e.g., fine line metallization 9. On surfaces with properties that influence the film properties, e.g., roughness, hardness Films, coatings, and modified surfaces are often unique materials with properties that differ from those normally encountered in the same materials prepared in other ways, and these unique characteristics should be considered when making property, stability/degradation or compositional measurements. In many instances, these unique properties are derived from the fabrication techniques and parameters as well as the limited size and thicknesses that are encountered in film structures. Unique conditions, characteristics, and properties of films and coatings include: 1. Substrate influence on properties 2. Presence of the interface and interfacial (interphase)material 3. Graded composition and properties with thickness 4. Dispersed impurities 5. Non-stoichiometric compositions 6. Unique microstructures (bulk, surface), e.g., columnar morphology 7. High surface/volume ratio 8. Local property variations, e.g., pinholes, nodules 9. Non-equilibrium conditions (defects, stress, crystallographic phase, structures, composition, impurities, etc.)

Non-Elemental Characterization of Films and Coatings

671

2.0 CHARACTERIZATION There are many reasons to characterize a film or coating. These include: 1. In development: determining the effect of processing variables on properties of the material (process sensitivity). Determining degradation modes. 2. Determining functionality and establishing performance limits for a specific application. 3. Establishing product acceptance specifications (functionality, stability). 4. Establishing a baseline for satisfactory composition, structure, or performance so that subsequent materials may be compared to this “standard.” 5. Monitoring reproducibility of processing. 6. Determining the stability of the material under service and degradation conditions. 7. Assisting in failure analysis. 8. Avoiding surprises. Note: characterization is essentially meaningless unless the formation conditions are reproducible. This means that the process must be reproducible and this is generally insured by using process controls and specifications. In many cases, property measurements are used to establish processing reproducibility. For instance, in the deposition of a metallization film, one might make: 1. A thickness measurement to insure that the right amount of material has been deposited and that the deposition conditions (contamination in a plasma when sputtering, for instance) have not changed when using the specified deposition parameters 2. An adhesion measurement to insure that the surface preparation was adequate and that the surface was not recontaminated during processing 3. An electrical resistivity (or resistance) measurement to insure functionality of the material

672

Deposition Technologies for Films and Coatings 4. Environmental aging to insure stability of adhesion and electrical resistivity during subsequent processing, storage, and service 5. Pinhole density measurements to insure that the likelihood of developing “opens” in patterned metallization is small

Characterization may be categorized as: (i) absolute, (ii) relative, (iii) functional,(iv) behavioral, and (v) stability. Absolute characterization means obtaining a specific value such as:(i) specific elemental composition (weight percent),(ii) resistivity (ohm-centimeters), (iii) geometrical thickness (microns, angstroms), (iv) density (grams/ cm3), etc. In order to get absolute values it is often necessary to use accurate measuring techniques and to compare the measured values to standards for the parameter of interest. Relative characterization means a comparison to an acceptable value (or known variation thereto) such as: an Auger peak height, x-ray fluorescence intensity, color, relative hardness, etc. Often precise, but not necessarily accurate, measurement techniques are used. Relative evaluations are generally more easily obtained and are less costly than are absolute values. Functional characterization relates to the final use of the material and include such properties as: adhesion, electrical resistivity, hardness, wear behavior, optical absorption, etc. Behavioral properties are not directly related to functionality but are a function of processing. These properties may be important in use or to indicate possible changes in film properties. An example is adsorption of gases or contaminants. Stability properties refer to the property changes in the product during subsequent processing, handling/storage, and service. Stability measurements are usually done as a function of environment (temperature, chemical species, fatigue, etc.). These environments must be carefully defined and specified. Properties may be general, such as film thickness, or may vary locally such as the presence of pinholes, nodules in the film, or small areas of high film stress. The general properties may not be uniform over a large surface area or may not be constant from one area to another on the deposition fixtures (position equivalency). Often variations may be due to substrate conditions, deposition parameters, etc. This means that some care must be taken in selecting the samples (or areas) to be characterized and the sampling statistics must take into consideration the possibility of such variations.


673

The importance of the property also determines the type of statistics used in property measurements. For example, one may measure the meantime-to-failure of a conductor due to electromigration, but since one failure can cause failure of a circuit, it may be more important to know the time-to-firstfailure for reliability calculations. It is often helpful (or necessary) to interact with a statistician in order to develop a meaningful statistical evaluation program. In some cases, special substrates (witness plates or monitor plates) may be used to give properties or conditions that are not generally available on the product to be used. Examples are: (i) the use of thin substrates that can be deformed by film stress, and(ii)smooth surfaces that may be masked to give “steps” for stylus or interferometric thickness measurements. In some measurements such as those used for adhesion tests or stress measurements, it is very important that the witness plates be of the same material as the substrates and processed in the same manner. In cases where different materials, surface conditions (e.g., smooth vs. rough) or different processing (e.g., cleaning) is used for the witness plates, the effects of these differences on the measured parameters must be known. Some film properties may be measured during the deposition process (in situ) and may be used to control the deposition process. This may be called in situ characterization and includes such measurements as: 1. Mass deposited (using deposition rate monitors, weight gain measurements) 2. Optical transmission, reflectance, and extinction (used with optical coating processes) 3. Film resistivity (using masks and conductor patterns) Upon opening a deposition system, some characteristics may be determined before the parts are removed from their fixtures. These characteristics may be called the first check characterization and include: 1. Uniformity of appearance and color over the deposition fixture, i.e., from sample-to-sample or over a large area. 2. Color (e.g., TiN[1]) and reflectivity—is it like other deposition runs? 3. Optical texturing—when viewed from different angles does the reflectance look different from different areas? This is an indication of morphological variation. If there are a number of samples in the run, or if the area is large, one should determine if all the positions in the deposition system are equivalent

674


(i.e., position equivalency). It may be helpful to identify each sample and its position in the fixture for future reference—variations in properties may be position dependent for reasons such as: angle-of-incidence of the depositing flux, plasma density variations, heating variations, presence of nearby virtual leaks, etc. After the samples have been removed from the fixturing they may be subjected to further testing. Simple and easy tests may be used to determine functionality and relative comparison. These simple tests include: (i) adhesion,(ii) residual stress,(iii) reflectivity,(iv) electrical resistivity,(v) thickness, (vi) optical transmission of films on optically transparent substrates for thickness or pinholes, (vii) some types of elemental composition, (viii) chemical etch rate, and(ix) oblique lighting to see bumps on smooth surfaces. Often these simple tests can give the first indication of problems in processing or functionality. In many cases one characterization technique will give results that depend on several properties of the material. For example: a chemical etch rate test will depend on film density, pinholes, surface area, thickness, and chemical composition. After the films have been exposed to the ambient, do they change with time? Changes may be evident in color, adhesion, chemical composition of the surface, wetting angle, or bondability. After the simple and easy tests, the films may be subjected to more complex and comprehensive tests which generally take a much longer time and require special techniques and configurations. In many cases, the functionality of the system must be determined in context of the intended use of the film. The best test is the operational lifetest where the film is used as it would be in service and samples are tested periodically to determine any degradation. Since this means a long test period, it is often desirable to used accelerated life-tests where the degradation mechanisms are accelerated by increasing the temperature (corrosion, diffusion processes), chemical concentration (corrosion), cyclic rate (fatigue failures), etc. A comparison between the accelerated tests and the operational tests gives an acceleration factor. A major concern in accelerated life tests is to be sure the right degradation mechanisms are being accelerated. Most often, both types of tests (operational and accelerated) are run, and in addition, control samples (archival or shelf samples) are kept in pristine condition so that operational or accelerated aged samples can be compared to the original materials. All of this assumes that the samples were reproducible when fabricated.


675

Many characterization techniques require destruction of the sample. Examples are: many adhesion tests, some compositional profiling techniques, mechanical properties testing. In some cases, evaluations may be made by non-destructive evaluation (NDE) tests and the tested sample can be used for further processing. Examples are: electrical resistivity (four-point probe resistivity), adhesion (tensile pull to value, “Mattox bad breath test”), and composition (x-ray fluorescence). Characterization may be at all levels of sophistication and expense. In this chapter, we discuss some of the most common characterization techniques, but before a characterization strategy is developed, the following questions should be asked. • Most important—is the processing and product reproducible? • How will the information be used? • How varible is the product from lot-to-lot, and from various positions in the deposition system? • Are the statistics correct? Should a statistician be consulted? • In development work: are the experiments properly designed to give the information needed to establish limits on the processing variables and the product properties? • What is important? Who determines what is important and the acceptable limits? • How quickly is the information (feedback) needed? • Who will do the characterization? Are the right questions asked, and is the necessary background information provided? • Does the testing program include subsequent processing, operational, and environmental considerations? • Is needless characterization being done, or can simpler and less expensive characterization methods be used? • Can the characterization be done effectively in the necessary environment (development, production, quality assurance) and by the required workers (Ph.D.'s, hourly workers)? • How will the specifications for the characterization methods be written? Specifications must be written for characterizations that must be done repeatability. Methods of characterizing the sample should be carefully

676


specified. Often professional organizations have specifications and standards for determining specific properties, tests, procedures, or processing.[2] These specifications and standards may be classed as:(i) industrial(ii) military, and (iii) international. Some of the organizations that have specifications and standards are: Mil. Specs - Military Specifications ISO - International Standards Organization ASTM - American Society for Testing and Materials ANSI - American National Standards Institute API - American Petroleum Institute ASME - American Society of Mechanical Engineers ASQC - American Society for Quality Control AWS - American Welding Society AVS - American Vacuum Society (recommended practices) EIA - Electronic Industries Association IEEE - Institute of Electrical and Electronic Engineers IES - Illuminating Engineering Society ISA - Instrument Society of America NEMA - National Electrical Manufacturers Association SAE - Society of Automotive Engineers UL - Underwriters Laboratories NBS - National Bureau of Standards IPC - Institute for Interconnecting and Packaging Electronic Circuits Others There are several ways of retrieving the standards and specifications. One is the VSMF microfilm system with a subject index and microfilmed standards and specifications. ASTM (American Society for Testing and Materials) has a series of publications of their standards. In many cases, published standards and specifications have to be modified for a specific application.


677

3.0 FILM FORMATION In order to understand how some of the unique film properties come about, we need to understand how the film fabrication is performed. Other chapters in this volume treat the subject of film fabrication methods. In addition, we need to understand how a film or coating is formed with a specific fabrication process. For example: let us summarize how atomistically deposited films are formed. The stages of atomistically deposited film formation are:[3] (i) surface preparation, (ii) condensation and nucleation of the adatoms, (iii) interface formation, (iv) film growth and, in some cases(v) post-deposition treatments. The characteristics of these stages may be very dependent on processing parameters. For example: (i) substrate heating is normally a very important process variable,(ii) angle-of-incidence of the depositing material flux may be important in developing the film morphology, and (iii) concurrent energetic particle bombardment during deposition may be used to modify all stages of film growth.[4][5] Surface preparation may be defined as the treatment of a surface in order to obtain satisfactory processing, function, or stability.[6] Surface preparation may be in the form of: (i) cleaning, (ii) modification of surface chemistry, (iii) modification of the physical or morphological properties of the surface, (iv) formation of nucleation sites or addition of nucleating agents (sensitization), and (v) activation of the surface to make it more chemically reactive. When adatoms impinge on a surface they may have a degree of mobility on the surface before they nucleate and condense.[5][7][8] The nucleation density of adatoms on a substrate surface (and mode of growth) determines the interfacial contact area and the development of interfacial voids—generally a high nucleation density is desirable for good film adhesion. The nucleation density depends on the kinetic energy and surface mobility of the adatoms, chemical reaction, and diffusion of the adatoms with the surface, adsorbed surface species,[9] and the nucleation sites available.[10] The nucleation stage of film formation may be studied by: (i) Transmission Electron Microscopy (TEM), (ii) Scanning Tunneling Microscopy (STM),[11] (iii) electrical conductivity and temperature coefficient of electrical conductivity (conductors on insulating substrates), (iv) optical transmission as a function of mass deposited (film on transparent substrate), (v)

678


extinction of the XPS (x-ray photoelectron spectroscopy) signal from the substrate as a function of deposited mass, or(vi) changes in the work function with deposited mass. Interface formation will begin during nucleation of the adatoms on the surface and may proceed throughout the deposition process and even during post-deposition processing, subsequent processing, and in-service usage, depending on conditions. The interfacial types may be categorized as:[3] (i) abrupt,(ii) mechanical, (iii) diffusion,(iv) compound, and(v) pseudodiffusion. The abrupt interface is formed when there is no diffusion and thus the interface is a sharp transition from one material to another in the space of a lattice parameter (e.g., Au on NaCl). In this case the gradient of materials properties is large. Due to the lack of reaction and the method of film growth, interfacial voids may be formed at the abrupt interface giving poor adhesion. The mechanical interface is an abrupt interface with mechanical interlocking. This type of interface may provide good adhesion if the surface roughness is “filled-in” and interfacial voids are avoided. The diffusion type interface is formed when there is interdiffusion of the film and substrate materials. A problem with this type of interface may be the development of voids in the interfacial(interphase)material if the diffusion rates of the materials are different (Kirkendall voids).[12] In the compound interface, diffusion is accompanied by reaction to form a compound material. The interphase material formed may be brittle, have Kirkendall voids, and develop microcracks due to the stresses developed in forming the compound material[13]—all of which reduce the fracture strength of the interface region and hence lower the film adhesion.[14] The pseudodiffusion type of interface may be formed under low-temperature deposition conditions or when the materials are insoluble, by physically mixing the depositing materials during multilayer film deposition, or by implantation or recoil implantation of atoms into the substrate surface. Figure 12.1 schematically depicts the types of interfaces and problems that can be associated with each type of interface. Heating during deposition may enhance diffusion of chemical reactions of the depositing atoms with the surface. Energetic particle bombardment may alter the interface formation by affecting the nucleation processes (cleaning, changes in surface chemistry, nucleation sites), by increasing the contact area, decreasing the interfacial voids, generating surface defects, enhancing chemical reaction, and by providing a high thermal input into the surface region.


679

Figure 12.1. Types of interfaces and problems that can be associated with each type of interace.

Generally interfaces and interphase materials are difficult to characterize because of their small extent and the dependence of the materials properties on interfacial flaws and the properties of the surrounding materials. Interfaces may be studied by TEM techniques. Fracture-related studies may also be informative. Film growth occurs by nucleation on a “like-material” and the same considerations as for nucleation on a foreign surface apply. In addition, largerscale effects must be considered. In particular, at low deposition temperatures, geometrical effects may lead to the development of a columnar growth morphology[15] that often leads to undesirable film properties such as microporosity, low film density, high chemical etch rates, contamination retention and others. The addition of energetic particle bombardment during deposition can change the growth morphology giving a more dense film.[16]-[20]

680


For reactive film deposition processes, two general cases exist. In the first case, there is a condensible species and a gaseous reactive species (e.g., Ti + N). In the second case, both species are condensible and reactive under the proper conditions but may only form a mixture under other conditions (e.g., Ti + C). In reactive film deposition processes, the rate and degree of reaction is dependent on the chemical reactivity of the reactive species, the temperature, the extent of the reaction, and the availability of the reactive species to the depositing species which, in turn, may be very dependent on system geometry and relative surface areas.[21] When a reactive species is present, concurrent energetic particle bombardment enhances chemical reactions. The nature of this enhancement is poorly understood since heating, physical collisions, molecular fragmentation, intermediate species, and the presence of energetic electrons (secondary electrons) may each play a role in the chemistry of the reaction. The existence of bombardment-enhanced chemical reactions iswell established in etching studies where the reaction products are volatile[22][23] and bombardment effects are found in reactive film deposition processes where the reaction products are non-volatile (reactive deposition).[24] In the condensation of atoms, there is developed a residual lattice strain which is usually evident as a residual tensile stress in the film. Where there is concurrent bombardment during deposition, this strain may be compressive in nature due to theatomic peening (stuffing) of atoms into the lattice by recoil implantation. These growth stresses are very important to some film properties such as adhesion and stability.[14][25][26] It should be realized that very few surfaces are chemically and physically homogeneous. Inhomogeneity in the substrate surface leads to variations in film nucleation, growth, and properties. Processing which leads to greater surface and growth homogeneity will lead to greater film homogeneity. Substrate morphology, surface chemistry, and physical properties may have a important affect on film growth and thus on the subsequent film properties. Substrate specification and characterization are important parts of process development and production reproducibility. Specifications for the in-coming substrate material and the surface preparation must be included as part of the process development. Post-deposition processing may be used to change the film or interface properties. Such processing includes: (i) burnishing or shot peening (soft metal films), (ii) rapid thermal processing,[27] (iii) annealing, (iv) ion beam mixing.[28]


681

4.0 ELEMENTAL AND STRUCTURAL ANALYSIS Some elemental and structural analysis techniques are covered in other chapters of this book (others are not, see Ref. 29) but since they are closely tied to the non-elemental characterization techniques that are covered in this chapter, a few points about this type of characterization from that point of view are included here. Elemental and structural analysis is typically done by someone separate from the processing activity. Often the analysts are very parochial, so careful consideration is required to determine which analytical technique is best suited to the question/problem at hand. At the least, several people and techniques and several sources of information may be required to make a decision. It may be necessary to work with the analyst to develop a program which will answer the questions that need to be addressed. Each analytical person/technique must be given the background necessary obtain the needed information. If the person using the scanning electron microscope looks for unusual features and takes pictures of strange things on the surface, a very distorted view of the product will emerge. If an Auger analysis of a sample that was carried in a week ago doesn’t have carbon on the surface, then the Auger system is not working right. If hydrogen in the film ia a concern, then Auger analysis is not appropriate; Nuclear Reaction Analysis is needed. Many analytical techniques are very dependent on the sample preparation. Generally, one must determine if the findings of elemental and structural analysis are important or not. The product or process engineer must work with the analyst to obtain meaningful results. In compositional analysis there is a big difference in time and effort between detection with relative values and absolute numbers. Usually, to get absolute values, it is necessary to obtain or make standards for comparison and this may be very difficult. Often it is just as informative to get relative values. It is also very important to have archival samples with which to compare the analytical results. Surface analysts take great pride in getting the highest resolution possible, but often variations (e.g., pinholes) in a large area and high resolution of a small area (such as a SEM analysis) is not appropriate, and other characterization techniques should be used. Elemental and structural analysis is a great way to run up characterization costs. Typically the turn-around time and feedback times are long.

682


Pretty pictures of microstructures and absolute numbers from compositional analysis may not be necessary but they might look good in a report or paper.

5.0 SOME PROPERTY MEASUREMENTS Many different property measurements may be made on a film or coating. In many cases, the property measurements are highly specific. This part of the chapter will concentrate on a few of the most common property measurements. 5.1 Adhesion Good adhesion, as defined by the fabrication, testing, and service conditions, is a fundamental requirement of any film-substrate system. Good adhesion is determined by a large number of factors, many of which are difficult to control without careful processing. Process development is often done in an empirical manner, aided by some basic considerations of the factors most likely to give good adhesion and properties which are detrimental to good adhesion. From these considerations one can decide what must be done to obtain good adhesion and the proper procedures for testing the adhesion The American Society for Testing and Materials definesadhesion as the “condition in which two surfaces are held together either by valence forces or by mechanical anchoring or by both together,” (ASTM Definition D 907 - 70).[30] In engineering applications, adhesion is the physical strength of an interface between regions of a material system. Such interfaces are found in grain boundaries, solids in contact (friction, wear), and in film-substrate systems. Adhesion failure is the separation of the materials at or near the interface over a large area, usually under stress. Adhesion failure is the end result of fracture and/or deformation of material and may depend on the properties of the substrate material near the interface(nearby material), the interphase material or the film material. Adhesion strength is an irreversible macroscopic property of the system and is amenable to specification and testing. “Good adhesion” is when the interfacial region (or nearby material) does not fail under service conditions nor at unacceptably low stress levels under fabrication and test conditions.[3]


683

The mechanical properties of the interfacial (interphase) material are crucial to good adhesion. This interfacial material may have a composition, microstructure, and properties which differ from either of the bulk materials. An interface may be sharp or diffuse, rough or smooth. The formation or presence of the interfacial region may affect the composition, stress, or microstructure of the nearby material (film or substrate) which may be weakened by a number of processes such as diffusion or flaw formation. The stresses which may cause adhesion failure include: mechanical (tensile, shear, compressive, shock, fatigue), chemical (corrosion, solution) and thermal/time (diffusion, reaction). Internal (residual) film stresses may contribute to the failure.[3] The principal methods of attaining adhesion are by: surface energy reduction,[32]-[34] high fracture energy of the interfacial region,[3][14] or the use of bonding agents to provide a “new” surface. The deposition process and process variables may have an important bearing on the resultant adhesion by changing the nucleation, growth and properties of the deposit. Energetic processes, such as high temperature or high particle kinetic energy processes,[35] promote the formation of diffusion, compound, or pseudodiffusion type interfaces. High temperature processing imposes constraints such as matching the coefficient of expansion of the deposited material to the substrate materials so that thermally-induced residual stresses are not produced during cool-down. Post-deposition treatments may be used to increase adhesion. These treatments include: heating, ion mixing (e.g., see Refs. 28, 36) and the diffusion of reactive species to the interface.[37][38] Heating of the filmsubstrate couple allows stress relief in some systems.[39][40] Even time alone under ambient conditions can give changes in the adhesion. This may occur by allowing stress relief (grain growth) and diffusion of species to the interface. For example, plasma cleaning of glass surfaces prior to silver deposition has been shown to give a time-dependent improvement in the adhesion of the silver films after deposition.[41] Of course some post-deposition treatments, such as may occur during subsequent processing, can cause loss of adhesion. The loss of adhesion under mechanical stress (tensile, compressive, shear) occurs by deformation and fracture of material at or near the interface. The fracture mode (brittle or ductile) depends on the properties of the material and the presence of flaws which may create easy fracture

684


paths and can act as stress concentrators to initiate and propagate the fracture. Another important factor in fracture propagation is the presence of stress and stress gradients in the material. These stresses may enhance fracture propagation (tensile stress) or retard fracture propagation (compressive stress), to some degree. The stress may be intrinsic to the system— arising from the deposition process, or may be extrinsic—arising from applied stresses. In either case, the nature of the stresses which appear at the interface depend on the properties of the materials involved. Localized regions of high intrinsic stress may be found in films due to growth discontinuities or defects such as pinholes[42] or nodules, or near features such as edges. These stressed areas may lead to localized adhesion failure under applied stress. The fracture path is determined by the properties of the film and substrate materials, the presence and distribution of flaws, stress distribution, and the presence of features which may blunt or change the fracture propagation direction. Conceptually, the energy needed to propagate the fracture and create the new surfaces can be measured and this fracture energy would be a good measure of the adhesion of the system.[43][44] Energy may be absorbed by:(i) plastic deformation (slip, atom motion),(ii) elastic deformation (heat),(iii)generation of free surfaces. The fracture of a brittle material is often accompanied by acoustic emission which results from the release of energy. This acoustic emission has both an energy and a frequency spectrum.[46] In one method of the detection of adhesion failure by acoustic emission, the coated surface is scratched by a rounded diamond point and the load on the point is increased while monitoring the acoustic emission using a piezoelectric accelerometer. This mode of detection is often more sensitive than the normally-used optical detection techniques The fracture of an insulator interface is often accompanied by the emission of electrons, photons and/or ions (fractoemission). This fractoemission is probably due to microdischarges resulting from charge separation during fracture.[46] Adhesion is determined by the nature of the stresses that appear at the interface and the fracture energy needed to propagate a fracture. Good adhesion is promoted by: high fracture toughness of the materials, low concentration of flaws, non-planar defects, presence of fracture blunting features, interfacial roughness that necessitates the change of direction of a propagating fracture, low stresses and stress gradients, and the absence of operational degradation mechanisms.


685

Poor adhesion may be attributable to: low degree of chemical bonding (as evidenced by a low nucleation density), poor interfacial contact , low fracture toughness (brittle materials, flaws, stresses), high residual film stresses, and operational degradation mechanisms. Poor adhesion may be localized, giving local failure on stress (pinholes, nodules, spallation). Loss of adhesion may also occur due to non-mechanical stresses such as: corrosion or solution of interfacial material, generation of flaws, diffusion to or away from the interface of species which can influence adhesion, precipitation of diffusing species (e.g., H, He) at the interface, or static fatigue processes that propagate existing flaws in brittle materials. These degradation processes are often time, temperature, and environment dependent. An example of the loss of adhesion due to corrosion effects is the degradation of some Ti-Au metallizations in an HCl environment.[47] This electrochemical degradation may be eliminated by the addition of a thin intermediate layer of palladium between the titanium and the gold. An example of the loss of adhesion due to diffusion is the diffusion of chromium from the interface of an oxide-Cr-Au metallization through the gold to the surface on heating to >200°C in air. At the surface, the chromium oxidizes and creates a non-bondable surface and the loss of chromium at the interface results in loss of adhesion. This out-diffusion of the interfacial material is dependent on the composition of the gaseous ambient, and a nonoxidizing ambient reduces the diffusion.[48] The addition of a small amount of oxygen in the chromium and/or the gold during deposition reduces the chromium diffusion rate and gives a more thermally stable metallization.[49] The adhesion of the Ti-Au metallization can be degraded by the diffusion of Ti to the surface and by chlorine impurities in the film material (chemicallyinduced segregation).[50] The diffusion of hydrogen through a film to an interface where it precipitates has been used by the electroplating community as an adhesion test.[51] Gases incorporated into a surface or film during surface preparation or film deposition may diffuse to the interface on heating, giving a loss of adhesion. Diffusion of water vapor through a polymer film to the interface can lead to the degradation of metal-polymer adhesion.[53] Interfacial mixing can improve the moisture degradation properties of polymer-metal film systems.[53]

686


Film properties may influence the apparent adhesion of a film-substrate couple. The deformation, microstructural, and morphological properties of the film material determine the ability of the material to transmit mechanical stress and to sustain residual stresses. The objective of adhesion testing is to duplicate the stresses to which the interface will be subjected in subsequent processing, testing, and service. Adhesion testing is used to monitor process and product reproducibility. A part of the adhesion testing program should include possible time, environment, or stress dependent degradation mechanisms. Generally adhesion tests subject rather large areas to stresses and often do not detect localized areas of poor adhesion. The use of acoustic emission with some adhesion tests may give an indication of the onset of failure. Adhesion tests are generally very difficult to analyze and are most often used as comparative tests in product acceptance specifications. The best test of adhesion is functionality under service conditions! Typically adhesion testing is done by lot sampling on product or witness samples that are characteristic of the product. It should be remembered that the properties of the substrate material and surface preparation procedures may have an important effect on the measured adhesion, so the witness sample material and its preparation should be carefully controlled. Stressing a film to test for adhesion may result in other failure modes such as cracking of the film, even though the film does not separate from the surface.[54] Methods of accelerating the degradation modes for accelerated adhesion testing should reflect the same degradation modes as are to be found in service. Acceleration may be accomplished by increased temperature, mechanical fatigue, thermal fatigue, concentrated chemical environment, or by the introduction of interfacial flaws by some technique. Non-destructive adhesion testing techniques would be highly desirable but are of limited availability at the present time. Possibly thermal-wave techniques, which have been used to monitor ion implantation damage,[55] can be used to detect interfacial flaws. Testing-to-a-limit may be used and some use of acoustic emission to detect onset of failure has been attempted. Since adhesion is a macroscopic property of the system, the adhesion test methods generally involve testing over an appreciable area. In some cases, the testing may be over a much larger area than we are really interested in. Adhesion testing should test the coating under stresses similar to those encountered in production and service.


687

There are a large number of potential adhesion tests[56]-[60] and each investigator or technology community favors different tests. Adhesion tests may be categorized by the method that the stress is applied to the film/ coating. The following are some types of adhesion tests: Functionality Tensile (pull) tests: Wire bond Thermocompression (TC) bond Soldered bond Epoxy bonded stud Electroplated stud Rotor tests Peel tests; “Tape test” Topple tests Shear tests: TC ball bond (push-off test) Ring shear Lap shear tests Deformation (of substrate ) tests: Bend Pull Indentation test Scratch test (with acoustic emission) Stress-wave test: Flyer plate/foil Laser pulse Abrasion tests Thermal stressing Diffusion test—diffusion of hydrogen to interface (electroplaters) Weird tests: Mattox bad breath test (unpublished)

688


The best test of adhesion is “does it work?” under subsequent processing, testing, and service. This may be called a functionality test. The peel test is a common test for adhesive bonding[61][62] and a variation of the peel test is the tape test, where an adhesive tape is stuck on the film surface, then a peel test is performed. This test is good for detecting poor adhesion (up to about 1000 psi) but is very sensitive to the method used—type of tape, method of application of the tape, pull angle, pull rate etc. Often the film is scribed (cut, cross-hatched) beneath the tape to provide an edge on which the tape pulls. Measurement is in grams/mm. The scratch test is an old adhesion test method[63] (more sophisticated than the scrape test) where a complex deformation is introduced into the surface and then the failure mode is observed and a critical load for failure is assigned.[64] This test has been the subject of numerous investigations. The stresses associated with a moving stylus have been analyzed.[65] The loaded stylus used for the scratch test may fracture a brittle substrate material giving erroneous results.[66] The use of an SEM with an in situ scratch testing capability allows the observation of the failure and material transfer without time or environmental effects.[67] The scratch test can be combined with acoustic emission to give an indication of the onset and magnitude of failure.[63][68] The hardness of the substrate material may have a significant affect on the scratch resistance (cracking) of thin coatings.[69] A commercial unit is available to perform the scratch test along with acoustic emission. The tensile test generally utilizes a wire or stud bonded to the surface and a tensile tester. Bonding of the wire is usually done by thermocompression bonding, ultrasonic bonding, or soldering. Bonding of a stud to the surface is usually done by thermosetting epoxy bonding. Tensile strengths to about 10,000 psi can be measured, but the analysis of the result can be difficult.[70] Care must be taken to avoid bonding stresses which will reduce the apparent adhesion. Commercial testers are available for the stud-bond test. One interesting variation of the tensile test is used to study the fracture energy of the interface. This test involves bonding a surface to the film, then performing a notch tensile test.[43] The shear tester[71] uses a bump bonded to the surface and a shearing (actually peeling) motion to determine the strength of the bond or of the adhesion of the film. Commercial units are available to perform this test. In stress wave adhesion tests, a stress wave is propagated through the system and the reflection of the stress wave at the interface results in a tensile stress.[72]-[74] The stress can be injected into the solid from a flyer plate, a flyer foil or a pulse of radiation (laser). Conceptually, this technique could be used to initiate, then stop, an interfacial fracture so the fracture


689

mode could be studied. The onset of the fracture could be detected by acoustic emission. The most recent advance in adhesion tests is the monitoring of acoustic emission during adhesion testing. As fractures form and propagate, there is a release of acoustic energy which may be monitored. The onset of acoustic emission correlates with the onset of adhesion failure in deformation tests. Acoustic emission can be correlated to the fracture of the interface of films on plastic surfaces,[75] plasma sprayed coatings,[76] and hard coatings on tools.[63][68] Thermal stress adhesion testing is used on coatings intended for high temperature applications and are often combined with mechanical stresses such as found in thermal barrier coatings[77] and coatings for fusion reactor applications.[78] The Mattox Bad Breath test consists of breathing on the film (best on a brittle substrate material) so that moisture condenses on the film. If the film has a high residual stress, the moisture will tend to accelerate fracture propagation, and blistering (compressive stress) or cracking (tensile stress) will be enhanced. This is an easy “first test” and the test is non-destructive if the film adhesion and adhesion stability are good. 5.2 Film Thickness A film or coating thickness may be defined in three ways:(i) geometrical thickness—separation between surfaces; (ii) mass thickness; and (ii) property thickness. The geometrical thickness is the separation between surfaces and is measured in mils, microinches, nanometers, angstroms, or microns, and does not take into account the composition, density, microstructure, etc. A general problem with this measurement is the definition of the surfaces. Mass thickness is measured in micrograms/cm2 which can be converted to a geometrical thickness if one knows (or assumes) the density of the material. Property thickness measures some property such as x-ray absorption, beta (electron) backscatter, ion backscattering, optical adsorption or electrical conductivity which may be sensitive to density, composition, microstructure, crystallographic orientation of the film, etc. Property thickness measuring techniques often require calibration standards. Different thickness measuring techniques may give differing values for the thickness. Thickness measuring techniques may also be categorized as contacting and non-contacting. The following are some of the most commonly used.

690

Deposition Technologies for Films and Coatings Contacting techniques: • Surface profilometer (stylus technique). Measures the height of a step from the substrate surface to the film surface. Step is formed by masking during deposition or by masking and etching. Stylus scans length of several centimeters with a resolution of <0.2 mictrons and measures height of greater than 100 Å.[79] Sensitivity is dependent on surface roughness, flatness, and abruptness of the step. Commercial units are available that scan over a surface and present the surface topography on a screen. Non-contacting techniques: • Michelson interferometry - Measures the height of a step using a split beam of light. The differing optical path lengths give constructive and destructive interference patterns. By knowing the wavelength of the light and the number of fringes, the step height can be calculated. Measures step heights of 300 - 20,000 Å ± 150 - 300 Å.[80][81] • X-ray fluorescence (XRF) - Measures the mass per unit area of a material. By assuming the density (or calibrating the instrument) the measurement can be presented as a thickness. Measures thicknesses from 100 nm to 40 microns, depending on the material.[82] • X-ray absorption - Measured by x-ray attenuation. Thickness by knowing the absorption coefficient or by calibration. Measures thicknesses from 0.1 to >1000 µm ±5%. • Ellipsometry - Measures dielectric film thickness by the rotation of polarization axis as the beam passes through the film. Thickness is determined by knowing the index of refraction of the dielectric or calibration.[83] • Beta backscatter - Energetic electrons from a radioactive source are backscattered from the film and underliying substrate. Thickness is measured by calibration. Thickness range depends on the electron source and the scattering properties of the material. For example, using a C14 source, 1.25 to 1.9 microns of gold can be measured; using a Ru106 source, 15 to 38 microns gold can be measured ± 5%. [83] • Other techniques - Scanning Tunneling Microscopy (STM) (step height) - Atomic Force Microscopy (AFM) (step height) - Photon Tunneling Microscopy (PTM) (step height) - Magnetic eddy current techniques - Multiple beam interferometry (step height, 10 - 10,000 Å)


691

The determination of which technique is best for a particular application depends on a number of factors.[84]-[86] 5.3 Film Stress Films and coatings on substrates may have a residual stress that is either compressive—as if the material were being compressed, or tensile— due to the differences of coefficient of expansion of the film and substrate (high temperature deposition), or from strains grown-in during the growth process. These stresses contribute to adhesion failure (immediate or long-term) or may affect mass transport properties such as void growth,[87] low temperature recrystallization (crystalline materials),[88] or a low strain point temperature (glasses).[89] Films under compression will try to expand and if the substrate is thin the film will bow the substrate with the film being on the convex side. If the film has a tensile stress, the film will try to contract, bowing the substrate so the film is on the concave side. Tensile stress may relieve itself by microcracking the film. Compressive stress may relieve itself by buckling, giving wrinkled spots (usually associated with contamination on the surface), or a wavy pattern (clean surface) if the stress is isotropic.[9] The residual stresses may be anisotropic with direction in the film.[81] A great deal can be learned about the film stress by observing the stress relief patterns.[14] The film stress may not be uniform through the film thickness, i.e., there may be a stress gradient in the deposit. (If the stress is not uniform, the film will curl up when separated from the substrate; if uniform, the separated film will lie flat). The total film stress is a function of film thickness. By knowing the mechanical properties of the substrate and film material, the film thickness and the substrate deflection the film stress can be calculated. There are a number of ways that the deflection of a beam can be measured and the stress calculated.[91]-[97] Figure 12.2 shows a commercially available attachment for use with a microscope to generate an interference pattern that can give the radius of curvature.[81] If the beam is long and narrow so that there is no “angle-iron” stiffening effect, and the beam was clamped flat during the deposition, the film stress (σf) can be calculated from:[98]

Eq. (1)

t f E s  t s  σf= ( 6 ρ)   t f 

2  t E y   − s + 6 f f  tf t s tf   

692


where Tf and Ts are the thicknesses of the film and substrate, Ef and Es are the elastic moduli of the film and substrate material,ρ is the radius of curvature and the term yf /tf is the relative position in the film for which the stress is calculated and is measured from the midplane of the film (yf = 0) and is positive toward the film-substrate interface where the film stress is maximum. Figure 12.3 shows a sample calculation.

Figure 12.2. Michelson interferometer attachment for optical microscope.


693

Figure 12.3. Sample stress calculation for a molybdenum film on a thin glass substrate.

A major uncertainty in measuring film stress is the elastic modulus (and Poisson's ratio) of the film material which has to be assumed in most cases. If the last term in Eq. (1) can be made small in comparison with the other terms, the stress determination can be made without knowing or assuming Ef. This can be done by making ts/tf very large, which also means measuring a small Rs. This can be done with a sensitive, large optics interferometer.[98] The system shown in Fig. 12.4 is capable of detecting the radius of curvature of more than 1 km over an area 2.5 cm in diameter.

694


Figure 12.4. Large area Michelson interferometer with associated illumination and data treatment system.[98] The setup shown is for measuring the mechanical properties of a coated substrate by four-point loading of the sample.

Film lattice strain (stress) may also be measured by x-ray diffraction and lattice parameter measurements.[99] However, this technique may not give the same value of stress as measured by the deflection techniques since it does not sum over all the stresses (those associated with the grain boundaries for instance) and is influenced by other factors such as grain size and film morphology. Strain in the surface lattice (few atom layers) can be measured by LEED techniques.[100]


695

5.4 Coefficient of Thermal Expansion The coefficient of thermal expansion of residual growth-stress-free films (annealed) can be determined using the same techniques used for determining the stress in the films by making the measurements at various temperatures. Again, one must know (or assume) the properties of the film and substrate materials. One often finds low temperature mass transport in as-deposited films (driven by high residual stresses and high defect concentrations/ mobilities) giving low temperature annealing (strain point for glasses)[89] and grain growth[88] during testing at elevated temperatures. These changing properties will affect the expansion measurements until the film is annealed. 5.5 Mechanical Properties The hardness of a material is usually defined as the resistance to deformation and is usually measured as the permanent deformation of a surface by a specifically shaped indenter under a given load.[101]-[102] This does not give an indication of the plastic deformation associated with loading. The hardness of a material may be influenced by the grain size, dispersed phases, defect structure, microstructure, density, temperature, deformation rate, etc. For films and coatings, there may be substrate influences on the deformation which affect the measurements. As a rule, the coating thickness should be 10X the indentation depth to obtain meaningful results. Surface effects may also influence the indentation measurements for thin films, particularly those with oxide layers. Techniques to measure the microhardness of films and modified surfaces (particularly ion implanted ones) usually use microindentation techniques.[103]-[111] In addition to hardness, the elastic properties of the material can be determined from the maximum penetration depth compared to the residual depth of the indentation after the indenter has been removed. The impact of microspheres with a surface may be used to measure microhardness and its variation over a surface.[112] An advanced microindentation hardness testing system is commercially available. It is a computer-controlled machine capable of performing indentation tests with load and depth resolutions of 2.5 millinewtons and 0.4 nanometers up to a maximum load of 10 grams. It detects penetration movement by changes in capacitance between stationary and moving plates.

696


5.6 Electrical Resistivity The bulk resistivity of a material is given in micro-ohm cm and the resistance of a path is calculated from: R = ρL/A where ρ is the resistivity, L the length, and A the area. For a thin film, the resistivity may be a strong function of the film properties such as morphology, composition, etc.[113] The film resistivity is often given as the sheet resistivity (sheetρ) in ohms per square since the resistance of any square is the same no matter what the size of the square, as long as the thickness is uniform and other properties are the same: The sheet resistance is measured using a four-point probe technique where the current[1] is injected through two probes and the voltage drop (V) between two other probes is measured.[114][115] This technique avoids contact resistance problems.[116][117] For a linear probe arrangement, the resistivity is given by: Rs = 4.532 V/I Probe separation of commercial units may be as low as 0.025 inches. For layered structures of materials having a nonuniform resistivity, the measurement is more complicated.[118][119] Resistivity (conductivity) can also be measured by induction without contacting the surface of the film.[120] 5.7 Temperature Coefficient of Resistivity (TCR) The TCR of metals is positive, i.e., increasing resistance with increasing temperature while that of tunneling-type conductors (insulators) is negative, i.e., decreasing resistance with increasing temperature. To measure the TCR, one only needs to combine a resistance measuring device with a temperaturecontrolled environment. The measured TCR combines effects found in the film, i.e., metallic conduction in the grains (columns), with tunneling through oxides at columnar boundaries. Often the film TCR is much less than that of the bulk material and may be of an opposite sign altogether. TCR measurements can give an indication of the perfection of the film material.


697

5.8 Electromigration Electromigration is the movement of film atoms in theelectron wind when metallizations are used to carry high current densities (Al: 106 A/cm2 - steady, 107 A/cm2 - pulse). The origin of the electromigration effect is poorly understood but it is probably dependent on local temperature, film stress, and lattice defects in the film material. The electromigration may result in hillock formation or void formation. Electromigration is measured by subjecting the conductor to the high current density, detecting failure (often defined as 50% reduction in cross-section area) and evaluating many samples prepared in the same way. Elevated temperatures, and processing techniques that introduce lattice defects, broaden the statistical failure curve and bring the early failures to a shorter time. Electromigration seems to be a statistical problem with some failures occurring far below the mean value. This can cause early failure of the electrical circuits if there is no circuit redundancy. Electromigration can be minimized and the statistical spread can be lessened by process control, burn-in of the circuits to eliminate the metallizations most prone to failure (infant mortality), addition of dispersed particles (1 - 4% Cu in Al), multilayering of the metallization (e.g., 3000 Å Al alternated with 50 - 100 Å Ti) or the use of cap (passivating) material. [121] Electromigration kinetics are dependent on the composition and structure of the film.[122] 5.9 Density The density of a material depends on:(i) composition,(ii) closed porosity (void) volume, and (iii) definition of the surfaces. Densities are given in g/cm3. A deposited material may easily have several atomic percent of foreign material incorporated into the lattice or may easily be off-stoichiometry by an appreciable amount, and this composition variation affects the density of the deposit. Voids affect the density in an obvious way. Density may be measured by several general techniques:(i) geometryproperty relationships, and(ii)displacement-flotation techniques In the geometry techniques, the size or thickness of the sample is determined as well as some property such as mass or x-ray absorption. For example, Rutherford Backscattering may be used to give the areal atom density and a profilometer can be used to give the geometrical thickness.[123]

698


Properties other than mass must be related to the mass by known properties or calibration. A principle difficulty with this technique is in defining the surface in order to make accurate thickness measurements. The displacement techniques include:(i) micropycnometry, (ii) density gradient column, (iii) hydrostatic weighing, and (iv) volume displacement. The most accurate techniques have been developed to study radiationinduced void formation in metals and utilizes hydrostatic weighing of small samples (30 mg) with an electrobalance to a precision of 0.04%.[124] Density gradient columns utilize a thermal gradient in liquids of varying density (liquid densities to 3.2 gr/cm3 ). The sample will float at a level of the same density fluid (watch out for buoyant air bubbles attached to the sample). Calibration floats are used to determine the fluid density. Pycnometry involves the displacement of a liquid or gas from a container of accurately known volume and the weight of the sample. Density is often related to other film properties such as chemical etch rate, corrosion, compressive strength, index of refraction, etc. 5.10 Porosity The porosity in a deposit may consist of:(i) open porosity where the pores are interconnected, (ii) closed porosity where the pores are isolated and not interconnected, and(iii) through-porosity where the pore extends through the deposit from the surface to the interface. Typically, a deposit will contain both open and closed porosity to some extent. A material with closed porosity will show a decrease in density while a material with open porosity will not (as measured by many of the techniques described under density measurements). Voids is another term used for isolated pores, whilemicrovoids is the term used for very small voids down to clusters of lattice vacancies (few angstroms in diameter). Voids in the bulk of the material form by the growth processes or by agglomeration of defects during or after deposition. Voids in the bulk affect density of the material, the deformation and fracture properties of the material, and the thermal and acoustic transmission of the material. Closed voids in materials are typically measured and studied by: (i) density measurements or (ii) transmission electron microscopy (TEM). In TEM, the sample is thinned and the voids are observed directly by using the underfocus-overfocus technique. Voids as small as 7Å in diameter may be resolved using this technique.


699

Open porosity may be of several forms. Uniformly distributed interconnected pores develop as the pore volume becomes large (greater than about 5%). Oriented porosity develops due to the growth mode of the deposited material such as forming a columnar morphology. Through-porosity often develops because a substrate-surface discontinuity gives geometrical shadowing of the depositing flux. Open porosity can affect material properties in both desirable and undesirable ways. Generally undesirable effects include: (i) high surface areas, (ii) easy access to the interfacial region, and (iii) easily deformed material. A high surface area results in: (i) high chemical etch rates, (ii) high corrosion rates,(iii) easy contamination/difficult cleaning, and(iv) dependence of properties (e.g., resistivity) on surface effects (e.g., oxidation). Easy access to the interfacial region may result in; (i) interfacial corrosion (loss of adhesion) and(ii) rapid diffusion paths (surface diffusion). Desirable effects of porosity include: (i) less residual film stress, (ii) low thermal conductivity (thermal barriers),(iii) higher resistance to thermal shock, and (iv) reduced mass transport effects such as grain growth. Open porosity in thick deposits may be measured by: 1. Mercury porosimetry 2. Gas absorption/desorption (BET [Brunauer-EmmettTeller],[125] Surface Acoustic Wave attenuation 3. Dye penetrants - fluorescence, radioactive In mercury porosimetry the sample is immersed in mercury and pressure (0.5 to 30,000 psia ) is used to force mercury into the pores (3.2 nm to 213 microns) (mercury intrusion).[126] The smaller the pores, the greater the pressure that is needed to force the mercury into the pores. Force vs. volumechange is then a measure of the pore volume distribution. Because of the “bottle-neck” effect, the measurement is often biased toward the small pore size. Calculations of pore size depend on the contact angle between mercury and the material being measured. Through-porosity, or cracks through metal films, on metal substrates may be measured by electrographic printing[127] where a chemical solution in a paper or gel is placed in contact with the film and a copper electrode is placed behind the paper. The electrode is made the cathode and the substrate is made the anode and a current is passed through the system (typically 200 mA, 30 sec). The paper is then observed for spots which indicate that some of the dissolved substrate material has reacted with the chemical solution.

700

Deposition Technologies for Films and Coatings Some electrographic porosity tests are listed below: Deposit

Au on Cu

Reagent solution

(Indication)

Potassium ferricyanide

(brown spots)

Ag on Cu


(brown spots)

Sn on Fe


(blue spots)

Au on Ni

Ammoniacal dimethylglyoxime and sodium chloride

(red spots)

Cr on Ni

Dimethylglyoxime

(pink spots)

Cu on Fe

Dimethylglyoxime

(deep cherry red spots)

Ni on steel

Sodium chloride + hydrogen peroxide

(rust spots)

Zn or Cd on steel

Sodium hydrosulfide

(black spots)

Porosity through thin dielectric films on metallic substrates may be measured by:[128]-[131] 1. Corrosion (liquid, gas) 2. Selective chemical dissolution (electrographic printing, solution analysis) 3. Electrochemical decoration 4. Anodic current measurement 5. Gas bubble generation (electrolytic) 6. Absorption (dyes, radioactive materials, liquids, gases) Porosity through metal films on metallic substrates may be measured by: 1. 2. 3. 4.

Corrosion - selective of substrate materials Selective chemical dissolution Anodic currents (controlled potential)[132] Corrosion potentials (anodic polarization)

Figure 12.5 shows the corrosion products that have built up in a pore in a gold film on a Kovar® surface. A special case of porosity measurement is the use of a Surface Acoustic Wave (SAW) device where a film deposited on a piezoelectric crystal adsorbs gases, changes the mass, and thus the acoustic dampening.[133] From the adsorption/desorption curves, a pore size distribution can be calculated. The SAW configuration is capable of detecting mass changes of 100 picograms/cm2.


701

Figure 12.5. SEM photograph of corrosion products emerging from a pore in a gold coating on Kovar®. Corrosion was performed in a moist UV/O3 atmosphere.

5.11 Chemical Etch Rate (Dissolution) The chemical etch rate of a material depends on density, surface area, intrinsic stress, stoichiometry, solution strength, and dissolution rate of the bulk, fully-dense stoichiometric material. Chemical etch rates are primarily used as a comparative technique.[134] Reactive Plasma Etching (RPE) and Reactive Ion Etching (RIE) are versions of chemical etching which use a plasma to activate the reactive species, which react with the surface, giving a volatile reaction product. The film microstructure and phase distribution can have an important effect on the RIE etch rate and uniformity.[135]

6.0 SUMMARY This discussion has given the reader an indication of the factors in film deposition technologies and film growth that affect the properties of the resulting material. Some simple characterization techniques have been discussed but there are many more to be found in the literature. The need for reproducible samples and characterization specifications has been strongly emphasized.

702


REFERENCES 1. 2. 3.

4.

5.

6. 7.

8. 9. 10. 11. 12. 13. 14.

15. 16. 17. 18. 19. 20. 21. 22.

Perry, A. J., Thin Solid Films, 135:73 (1986) Walters, S., Mechanical Engineering, p. 38 (April 1984) Mattox, D. M., Adhesion Measurement of Thin Films, Thick Films and Bulk Coatings, (K. L. Mittal, ed.), p. 54, ASTM STP 640, American Society for Testing and Materials (1978) Mattox, D. M., Plasma Surface Engineering,(E. Broszeit, W. D. Munz, H. Oechsner, K.-T. Rie, and G. K. Wolf, eds.) 1:15, Informationsgesellschaft, Verlag (1989) Greene, J. E., Proc. NATO Advanced Studies Institute on Plasmasurface Interactions and Processing of Materials, Alicante, Spain NATO ASI Series, (Sept 4-16, 1988) to be published Mattox, D. M., “Surface Preparation” Ch. 6, this volume Surface Mobilities on Solid Materials - Fundamental Concepts and Applications, (V. T. Binh, ed.), NATO ASI Series, Series B, Physics Vol. 86, Plenum Press (1983) Lewis, B. and Anderson, J. C., Nucleation and Growth of Thin Films, Academic Press (1978) Mattox, D. M., J. Appl. Phys., 37:3613 (1966) Miranda, R. and Rojo, J. M., Vacuum, 34:1069 (1984) Chidsey, C. E. D., Loiacono, D. N., Sleaton, T., and Nakahara, S.,Surf. Sci., 200:45 (1988) Olumura, K., J. Electrochem. Soc., 128:571 (1981) Philofsky, E., Solid State Electronics, 13:1391 (1970) Mattox, D. M. and Cuthrell, R .E., MRS Proc., Vol. 119, (D. M. Mattox, J. E. E. Baglin, R. E. Gottschall, and C. D. Batich, eds.) Materials Research Society (1988) Movchan, B. A. and Demchishin, A. V., Fiz Met Metalloved, 28:653 (1969) Mattox, D. M. and Kominiak, G. J., J. Vac. Sci. Technol., 9:528 (1972) Bland, R. D., Kominiak, G. J., and Mattox, D. M., J. Vac. Sci. Technol., 11:671 (1974) Thornton, J. A., J. Vac. Sci. Technol., A4:3059 (1986) Messier, R., Giri, A. P., and Roy, R. A., J. Vac. Sci. Technol., A2:500 (1984) Meissier, R. and Yehoda, J. E., J. Appl. Phys., 58:3739 (1985) Berg, S., Blom, H-O., Larsson, T., and Nender, C., J. Vac. Sci. Technol., A5:202 (1987) Geis, M. W., Lincoln, G. A., Efremow, N., and Piacentini, W .J., J. Vac. Sci. Technol., 19:1390 (1981)

Non-Elemental Characterization of Films and Coatings 23. 24. 25. 26. 27. 28. 29. 30. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43.

44. 45. 46. 47. 48. 49. 50.

703

Winters, H. F., Coburn, J. W., and Chuang, T. J.,J. Vac. Sci. Technol., B1:469 (1983) Harper, J. M. E., Cuomo, J. J., and Henzell, H. T. G., Appl. Phys. Lett., 36:456 (1980); also Appl. Phys. Lett., 37:540 (1980) Hoffman, R. W., Thin Solid Films, 89:155 (1982) Pulker, H. K., Thin Solid Films, 89:191 (1982) Singh, R., J. Appl. Phys., 63(8):R59 (1988) Wie, C. R., Yang, J. Y., Tombrell, T. A., Grant, R. W., and Housley, R. M., Vacuum, 38:157 (1988) Metals Handbook, 9th. ed., Vol. 10, (R. Wahn, ed.), American Society for Metals, Metals Park, OH 44073 (1986) Good, R. J., J. Adhesion, 8:1 (1976) Pulker, H. K., Perry, A. J., and Berger, R., Surf. Technol., 14:25 (1981) Kinloch, A. J., [polymer] J. Mat. Sci., 15:2141 (1980) Adhesion Aspects of Polymeric Coatings, (K. L. Mittal, ed.), Plenum (1981) Mattox, D. M., J. Vac. Sci. Technol., 10:47 (1973) Baglin, J. E. E., Ion Beam Modification of Insulators, (P. Mazzolsdi, and G. Arnold, eds.), Ch. 15, Elsevier (1987) Benjamin, P. and Weaver, C., Proc. Royal Soc., 261A:516 (1961) Laugier, M., Thin Solid Films, 75:L19 (1981) Kominiak, G. J. and Mattox, D. M., J. Electrochem. Soc., 120:1535 (1973) Hershkovitz, M., Blech, I. A., and Komem, Y.,Thin Solid Films, 130:87 (1985) Kikuchi, A., Baba, S., and Kinbara, A.,Thin Solid Films, 124:343 (1985) Zito, R. R., Thin Solid Films, 87:87 (1982) Bascom, W. D., Becher, P. F., Bitner, J. L., and Murday, J. S., Adhesion Measurement of Thin Film, Thick Film and Bulk Coatings, (K. L. Mittal, ed.), ASTM STP 640, pp. 63-82 (1977) Oh, T. S., Cannon, R. M., and Richie, R. O.,J. Am. Cer. Soc., 70:C352 (1987) Hintermann, H. E., J. Vac. Sci. Technol., B2:816 (1984) K’Singam, L. A., Dickenson, J. T., and Jensen, L. C., J. Am. Cer. Soc., 68:510 (1985) Speight J.D . and Bill, M. J. Thin Solid Films, 15:325 (1973) Ray, S. K. and Lewis, R. K., Thin Solid Films, 131:197 (1985) Mattox, D. M., unpublished results Krzyzanowski, S., Sylwestrowicz, W. D.,J. Mat. Sci. Lett., 1:35 (1982)

704 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.

Deposition Technologies for Films and Coatings Hothersall, A. W. and Leadbeater, C. J., J. Electrodepositers Tech. Soc., 14:207 (1938) Venables, J. D., J. Mat. Sci., 19:2431 (1984) Yasuda, H. K., Sharma, A. K., Hale, E. B., and James, W. J., J. Adhesion, 13:269 (1982) Grosskreutz, J. C. and McNeil, M. B., J. Appl. Phys., 40:355 (1969) Smith, W. L., Rosecwaig, A., Willenborg, D. L., Opsal, J., and Taylor, M. W., Solid State Technol, 29:85 (1986) Mittal, K. L., J. Adhesion Sci. Technol., 1:247 (1987) Mittal, K. L., Electrocomponent Sci. Technol., 3:21 (1976) Davies, D. and Whittaker, J. A.,Metallurgical Rev., 12:15 (1967);Metals and Materials, 1 (1967) Valli, J., Makela, U., and Matthews, A., Surf. Eng., 2:49 (1986) Chapman, B. N., J. Vac. Sci. Technol., 11:106 (1974) Brown, D. J., Windle, A. H., Gilbert, D. G., and Beaumont, P. W. R., J. Mat. Sci., 21:314 (1986) Yoon, Il-B., Jpn. J. Appl. Phys. Suppl., 2 Pt 1:849 (1974) Perry, A. J., Thin Solid Films, 107:167 (1983) Perry, A. J., Thin Solid Films, 78:77 (1981) Laugier, M. T., J. Vac. Sci. Technol., A5:67 (1987) Laugier, M. T., J. Mat. Sci. Lett., 5:253 (1986) Prasad, S. V. and Kosel, T. H., J. Matl. Sci. Lett., 3:133 (1984) Hintermann, H. E., J. Vac. Sci. Technol., B2:816 (1984) Je, J. H., Gyarmati, E., and Naoumidis, A., Thin Solid Films 135, 57, 86 Jankowski, A. F., J. Mat. Sci., 22:346 (1987) Jellison, J. L., IEEE PHP-11:206 (1975) Anderholm, N. C. and Goodman, A., Patent # 3,605,486 (Sept 20, 1971) Vossen, J. L., Adhesion Measurements of Thin Films, Thick Films and Bulk Coatings, (K. L. Mittal, ed.), ASTM STP-640, pp. 122-131, ASTM Publications (1978) Dini, J. W. and Johnson, H. R., Rev. Sci. Instrum., 46:1705 (1975) Van de Leest, R. E., Thin Solid Films, 124:335 (1985) Aithal, S., Rousset, G., Bertrand, L., Cielo, P.,and Dallaire, S., Thin Solid Films, 119:153 (1984) Berndt, C. C. and Miller, R. A., Thin Solid Films, 119, 173 (1984) Mattox , D. M., Mullendore, A. W., Whitley, J. B., and Pierson, H. O., Thin Solid Films, 73:101 (1980) Sherrington, I. and Smith, E. H., Wear, 125:241 (1988)


705

80. Sherrington, I. and Smith, E. H., Wear, 125:289, 1988 81. Cuthrell, R. E., Mattox, D. M., Peeples, C. R., Dreike, P. L., and Lamppa, K. P., J. Vac. Sci. Technol., A6(5):2914 (1988) 82. Ferrandino, F., Metal Finish, 84(5):29 (1986) 83. Yaghmour, S. and Neal, W. E. J., Surf. Technol., 25:297 (1985) 84. Rajora, O. S. and Curzon, A. E., Thin Solid Films, 123:235 (1985) 85. Piegari, A., and Masetti, E., Thin Solid Films, 124:249 (1985) 86. Pliskin, W. A. and Zanin, S. J., Handbook of Thin Film Technology, (L. I. Maissel, and R. Glang, eds.), Ch. 11, McGraw-Hill (1970) 87. Li, C. Y., Black, R. D., and LaFontaine, W. R., Appl. Phys. Lett., 53:31 (1988) 88. Patten, J. W., McClanahan, E. D., and Johnson, J. W., J. Appl. Phys., 42:4371 (1971) 89. Kominiak, G. J. and Mattox, D. M., J. Electrochem. Soc., 120:1535 (1973) 90. Ogawa, K., Ohkoshi, T., Takeuchi, T., Mizoguchi, T., and Matsumoto, T., Jpn. J. Appl. Phys., 25:695 (1986) 91. Brenner, A. and Senderoff, S.,J. of Research of the National Bureau of Standards, Research Paper RP1954, 42:105-123 (Feb 1949) 92. Kouyumdjiev, C. N., Surf. Technol., 26:35 (1985) 93. Kouyumdjiev, C. N., Surf. Technol., 26:45 (1985) 94. Kouyumdjiev, C. N., Surf. Technol., 26:57 (1985) 95. Sotirova, G. and Armyanov, S., Surf. Coat. Technol., 28:33 (1986) 96. Kouyumdjiev, C. N., Surf. Coat. Technol., 28:39 (1986) 97. Pulker, H. K., Thin Solid Films, 89:191 (1982) 98. Cuthrell, R. E., Gerstile, F. P., and Mattox, D. M., Rev. Sci. Instrum., 60(6):1018 (1989) 99. Hauk, V. M. and Macherauch, E., Adv. X-ray Anal., 27:81 (1983) 100. McRae, E. G. and Malic, R. A., Surf. Sci., 163:L702 (1985) 101. Angus, H. T., Wear, 54:33 (1979) 102. Microindentation Techniques in Material Science, (Blau and Lawn, eds.), ASTM Special Publication No. 889 (1986) 103. Blau, P. J., Metallography, 16:1 (1983) 104. Dirks, A. G., van den Broek, J. J., and Wierenga, P. E., JAP 55:4248 (1984) 105. Wierenga, P. E. and Franken, A. J. J., JAP 55, 4244, 84 106. Jonsson, B. and Hogmark, S., Thin Solid Films, 114:257 (1984) 107. Oliver, W. C., Mat. Res. Soc. Bull., 11(5):15 (1986) 108. Doerner, M. F. and Nix, W. D., J. Mat. Res., 1:601 (1986) 109. Bourcier, R. J., Stone, C. M., and Yost, F. G., Sandia Report SAND850486, (Sept 1985)

706


110. Bourcier, R. J., Nelson, G. C., Hayes, A. K., and Romig, A. D., Jr, J. Vac. Sci. Technol., A4:2943 (1986) 111. Schmale, D. T., Bourcier, R. J., Martinez, E., Sandia Laboratory Report SAND86-0509 (April 1986), available from NTIS 112. Cook, S. and Latham, R. V., Surf. Coat. Technol., 27:379 (1986) 113. Angadi, M. A., J. Mat. Sci., 20:761 (1985) 114. Valdes, L. B., Proc. IEEE (IRE), 42:420 (1954) 115. Keenan, W. A., Johnson, W. H., and Smith, A. K., Solid State Technol, 28(6):143 (1985) 116. Cohen, S. S., Thin Solid Films, 104:361 (1983) 117. Cohen, S. S., Gildenblat, G., Ghezzo, M., and Brown, D. M., J. Electrochem. Soc., 129:1335 (1982) 118. Albers, J. and Berkowitz, H. L.,J. Electrochem. Soc., 132:2453 (1985) 119. Albers, J. and Berkowitz, H. L., J. Electrochem. Soc., 131:392 (1984) 120. Esqueda, P., Octavio, M., and Callarotti, R. C.,Thin Solid Films, 89:33 (1982) 121. Teal, V., Vaidya, S., and Fraiser, D. B.,Thin Solid Films, 136:21 (1986) 122. Felton, L. E., Schwartz, J. A., Pasco, R. W., and Norbury, D. A., J. Appl. Phys., 58:723 (1985) 123. Antilla, A., Koskinen, J., Bister, M., and Hirvonen, J., Thin Solid Films, 136:29 (1986) 124. Pratten, N. A., J. Mat. Sci., 16:1737 (1981) 125. Young, D. M. and Cromwell, A. D., The Physical Absorption of Gases, Butterworth Pub. (1962) 126. Rootare, H. M.,Advanced Experimental Techniques in Powder Metallurgy, (J. S. Horschorn and K. H. Roll, eds.), 5:225, Plenum (1970) 127. Tvarusko, A. and Hinterman, H. E., Surf. Technol., 9:209 (1979) 128. Kern, W., RCA Rev, 34:655 (1973) 129. Kern, W. and Comizzoli, R. B., J. Vac. Sci. Technol., 14:32 (1977) 130. Kern, W., Solid State Technol., 17:78 (1974) 131. Kern, W., Solid State Technol., 17:35 (1974) 132. Morrissey, R. J., J. Electrochem. Soc., 119:446 (1972) 133. Martin, S. J., Frye, G. C., Ricco, A. J., and Zipperian, T. E.,Proc. IEEE, (1987); Ultrasonics Symposium, p. 563, (1987) 134. Pliskin, W. A., Physical Measurement and Analysis of Thin Films, (E. M. Murt and W. G. Guldner, eds.), Ch. 8, Plenum Press (1969) 135. Adachi, S., Susa, N., J. Electrochem. Soc., 132:2980 (1985)

13 Nucleation, Film Growth, and Microstructural Evolution Joseph E. Greene

1.0 INTRODUCTION The primary deposition variables which determine the nucleation and growth kinetics, microstructural evolution, and, hence, physical properties of films grown by physical vapor deposition (evaporation and sputter deposition in all of their various forms, Chs. 4, 5, and 6) are: the film material, the incident film flux, the kinetic energy E of species incident at the film growth surface, the film growth temperature Ts, the flux of contaminants, and the substrate material, surface cleanliness, crystallinity, and orientation. These represent the control variables that the crystal grower has at his disposal to tailor the properties of as-deposited materials. Note that the flux of contaminants which competes with the flux of film material for incorporation during deposition is strongly dependent upon the base pressure, pumping speed, and the design of the vacuum system (e.g., whether a substrate loadlock is used to circumvent repeated air-exposures) while substrate surface cleanliness depends also upon pre-deposition processing. The kinetic-energy of the incident film flux during film growth by thermal evaporation, for which E is of the order of 0.1 eV, is determined by the temperature of the evaporant source. However, in plasma or ion-beam deposition techniques, E can be increased up to several hundred eV. Lowenergy (often < 100 eV) ion irradiation during vapor-phase film growth has been shown to be useful in controllably altering the physical properties

707

708


of as-deposited layers through trapping, preferential sputtering, enhanced adatom diffusion, and dynamic collisional mixing.[1]-[4] This chapter is organized in the following manner. Section 2.0 deals with nucleation and the early stages of film growth. Microstructure development in both the ballistic-aggregation, low-adatom-mobility, limit and in the adatom migration regime are discussed in Sec. 3.0 while Sec. 4.0 reviews microstructural evolution and structure-zone diagrams. The important role of low-energy ion/surface interactions in all stages of film growth is treated in Sec. 5.0. Atomic-level mechanisms, the generation of point and extended defects, as well as the relationship of film properties such as stress to growth mechanisms is discussed throughout the chapter using experimental as well as computer-simulated results.

2.0 NUCLEATION AND THE EARLY STAGES OF FILM GROWTH There are three primary modes of film growth on substrates[5] as illustrated schematically in Fig. 13.1. During three-dimensional (3-D) island, or Volmer-Weber, growth, small clusters are nucleated directly on the substrate surface. The clusters then grow into islands which in turn coalesce to from a continuous film as shown schematically in Fig. 13.2. This type of growth occurs when the film atoms are more strongly bound to each other than to the substrate as is often the case for metal films on insulators or contaminated substrates. Two-dimensional (2-D) layer-by-layer, or Frankvan der Merwe, growth occurs when the binding between film atoms is equal to or less than that between the film atoms and the substrate. In addition to the obvious example of homoepitaxial growth on a clean substrate, there are numerous other examples of 2-D growth in metal-metal (e.g., Cd on W) and semiconductor-semiconductor (e.g., Ga1-xAlxAs on GaAs) systems. The third growth mode, often referred to as Stranski-Krastanov, is a combination of the first two. In this case, after first forming one or more monolayers, further layer growth becomes unfavorable and 3-D islands form. The transition from 2-D to 3-D growth is not completely understood but can be driven in some cases by the release of elastic energy stored in the film due to film/substrate lattice mismatch. This growth mode occurs much more frequently in metal-metal and metal-semiconductor systems than was originally believed. A typical example of such a system is In on Si(100)2x1.[6]

Nucleation, Film Growth, and Microstructural Evolution

Figure 13.1. Schematic representation of three film growth modes. overlayer coverage in monolayers (ML).

709

θ is the

Figure 13.2. Schematic representation of the island density n as a function of the coverage θ during three-dimensional growth. The value nmax is the maximum island number density. The island formation rate dn/dt decreases and becomes negative with increasing θ due to island coalescence. The remaining channels and voids are filled by “secondary nucleation,” island growth, and coalescence.

710


2.1 Three-Dimensional Nucleation and Growth Figure 13.3 schematically illustrates the essential features involved in 3-D nucleation and growth. An impinging flux of film species must first be thermally accommodated with the substrate. This typically occurs within a few vibrational periods. The adatoms can then diffuse on the surface to interact with other adatoms or re-evaporate. A fraction of the adatom clusters continue to grow in size and become islands which in turn coalesce to form a continuous film. Note that at sufficiently high deposition rates R or low deposition temperatures such that R > NoDs, where No is the substrate surface site number density (of the order of 1015 cm-2 depending upon material and orientation) and Ds is the adatom surface diffusivity, the film is amorphous. This occurs since the adatoms do not have enough time to diffuse across the surface and find low energy sites before they are buried by subsequently deposited adatoms. Covalently and ionically bonded materials have low packing densities and strong bonding directionality, and are thus easily deposited in the amorphous state. Metals, on the other hand, exhibit much higher diffusivities and are considerably more difficult to obtain in the amorphous state.

Figure 13.3. Schematic representation of processes leading to three-dimensional nucleation and film growth.

From Fig. 13.3, the minimum thermodynamic requirements to obtain net deposition would appear to be that the condensate pressure P in the gas phase equal its equilibrium vapor pressure Pe over the solid. Actually


711

however, the supersaturation ratio S = P/P e must be larger than one since small particles such as nuclei have a larger vapor pressure than that of bulk material due to their high surface-to-volume ratio. A large surface-to-volume ratio also leads directly to the requirement that clusters must be greater than a certain minimum critical size in order for growth to occur. This is easiest to understand in the case of homogeneous nucleation such as the formation of an embryonic ice particle in water cooled below its freezing point Tm. At temperatures less than Tm, solidification lowers the volume free energy ∆Gv of the system and is, hence, favorable. However, the formation of the solid particle increases the total Gibbs free energy ∆G since it introduces new interfacial surface area. The difference between the decrease in ∆Gv and the increase in the free energy Γ results in a free energy activation barrier for nucleation. That is, the incipient clusters must reach a critical size before the volume term overcomes the surface term and stable nuclei are formed. Most theoretical treatments of heterogeneous three-dimensional nucleation from the vapor phase are an extension of homogeneous nucleation theory using the capillarity, or droplet model.[7]-[9] More sophisticated treatments are also available.[5][10][11] However, the capillarity model has the virtue of being simple while still retaining all the essential physical features observed in experiments. In any case, all models, even the more sophisticated ones, suffer from several shortcomings. For example, macroscopic thermodynamic values, which are not strictly appropriate for small clusters, are used for surface energies, free energies of formation, contact angles, etc. In addition, convenient geometries, e.g., a spherical cap, are used to represent nuclei which are often anisotropic or crystallographic in shape.[6] Thus, for clarity of discussion, the capillarity model will be used in this chapter. Assume that a cluster of mean dimension r forms on a solid surface. The cluster has a surface area a1r2 exposed to the vapor phase, a contact area a2r2 with the substrate, and a volume a3 r3 where the ai terms are constants of geometry. The total free energy of the cluster with respect to dissociation into the vapor phase is Eq. (1)

∆G = a1r2Γc-v + a2r2Γs-c - a2 r2 Γs-v + a3r3∆Gv

Γc-v is the positive free energy associated with the formation of a new surface between the condensate and the vapor phase; Γs-c, which may be either positive or negative, is the surface free energy between the substrate and the

712


condensate; and the term a2 r2 Γs-v accounts for the disappearance of free substrate area. The problem can be simplified if we consider an isotropic geometry such as a spherical cluster. Equation 1 then becomes Eq. (2)

∆G = 4/3(π r3) ∆Gv + 4π r2 Γc-v

Equation 2 is plotted in Fig. 13.4 showing that there is an activation freeenergy barrier ∆G* to nucleation. Clusters larger than the critical size r* can lower their free energy by continuing to grow while clusters with r < r* will dissolve. An expression for the critical cluster size r* can be obtained by maximizing ∆G in Eq. 2, i.e., by setting δ (∆G) /δr = 0, and solving to yield Eq. (3)

r * = − 2Γ ∆Gv

where, for simplicity of notation, Γ is used to represent Γc- v. The free energy barrier is then obtained by substituting Eq. 3 into Eq. 2 to give Eq. (4)

∆G ∗ =

16πΓ 3

3(∆ Gv )

2

The term ∆Gv in Eqs. 3 and 4 can be evaluated using a general expression of the combined first and second laws of thermodynamics Eq. (5)

d(∆G) = VdP - SdT

where V and S are the system volume and entropy, respectively. Substituting the ideal gas law PV = kT at constant temperature, Eq. (6)

d( ∆G) = kT dP P

Since ∆Gv is just ∆G/Ω where Ω is the volume of an adatom, then Eq. (7)

∆Gv =

kT P kT ln = ln(S) Ω Pe Ω

Substituting Eq. 7 into 3 gives Eq. (8)

r* = −

2ΩΓ kT ln(S)


713

Figure 13.4. Schematic diagram showing free energy vs. the radius r of a spherical nuclei.

Equation 8 shows that r* decreases as the supersaturation S increases. S can be increased either by increasing P, i.e. increasing the incidentflux of condensing species, or by decreasing Pe which depends exponentially on the growth temperature Ts. That is, r* decreases as Ts decreases. An order of magnitude estimate of r* can be obtained from Eq. 8. Γ for a noble metal is of the order of 1000 ergs/cm2. Assuming a supersaturation of 2 x 104 at 300 K gives r* ≈ 0.5 nm. This is in reasonable agreement with experimental results[7] obtained by the use of field electron microscopy (FEM) and transmission electron microscopy (TEM) observations fitted to nucleation kinetics models.[5][10] Such models, originally developed by Zinmeister,[11a] are based upon rate theory in which a set of differential equations of the form

Eq. (9a)

dn1 n = R − 1 − 2 J1 − ∑ Ji dt τd i ≥1

Eq. (9b)

dn 1 = J i −1 − J i dt

(i ≥ 2)

714


is used to describe the nucleation process. n i is the number density of clusters containing i atoms, τ d is the mean residence time of an adatom, and Ji is the net capture rate of adatoms by clusters of size i atoms. The mean residence time τd depends exponentially on the desorption energy E d and J i is proportional to the product of a capture cross-section and the adatom surface diffusivity Ds(T s) which is, in turn, exponentially dependent upon the activation energy for surface diffusion Es. The experiments involve measuring n i as a function of R and T s to obtain critical nuclei sizes (typically one to a few atoms at low temperatures) and activation energies for adatom surface diffusion and desorption. As illustrated in Fig. 13.3, critical nuclei can grow by direct impingement and capture of vapor-phase species and by the capture of condensed adatoms diffusing across the substrate surface. At low coverages, the second mechanism dominates as nuclei capture any adatoms within a diffusion distance x = (2 Dsτs) 1/2 where τs is the mean time between surface diffusion jumps. Nucleation density measurements by TEM, carried out as a function of R and T s, thus provide an estimate of x and, hence, D s. A schematic illustration of sequential steps during the early stages of three-dimensional film growth, as determined from transmission electron microscopy (TEM) studies, is shown in Fig. 13.5. Supercritical nuclei are first observed at sizes corresponding to the resolution of the microscope. With further deposition time, the average nuclei size as well as the number density n increases. n, however, reaches a maximum at a value typically in the range from 10 10 to 1012 cm -2 corresponding to an average island size of 10 to 100 nm. Continued deposition then leads to a decrease in n due to coalescence of adjacent clusters. Coalescence is often described as occurring in a “liquid-like” manner although electron diffraction results show that is generally a solid-state reaction. (Note that liquid clusters have been reported for low melting point, Tm, films grown at high Ts/Tm ratios[12] due to the depression in the freezing point associated with small clusters.[13] As large islands continue to grow by capture of mobile adatoms and small clusters as well as by coalescence with nearby smaller islands and occasionally with other large islands, the film becomes semi-continuous with a network of channels and holes. Secondary nucleation, nuclei growth, and island coalescence can also occur in the voids. The secondary islands are then incorporated into the growing film as it becomes continuous.


715

Figure 13.5. Schematic illustration of sequential steps during nucleation and the early stages of film growth. (From Ref. 8.)

Figure 13.6, taken from the work of Donohoe and Robins,[14] is a plot of the number density of evaporated Au nuclei on NaCl(100) as determined by replication TEM studies carried out as a function of deposition time.[15] The experiments were performed in ultrahigh vacuum using a deposition rate of 0.1 nm/min onto vacuum-cleaved substrates maintained at 250oC. The number density n reached a maximum of≈1.5 x 1011 cm-2 after depositing an equivalent thickness of ≈1 nm. A series of electron micrographs which provided some of the data for Fig. 13.6 are shown in Fig. 13.7. Both n and the average island size are increasing in the sequence Figs. 13.7a (0.5 min of deposition) through 13.7d (8 min of deposition) with a significant amount of coalescence clearly observable in Fig. 13.7d. A series of micrographs illustrating the nature of the morphological changes which occur during coalescence are shown in Fig. 13.8 from the work of Pashley et al.[15] on the growth of Au on MoS2 at 400oC. The clusters labeled A and B in Fig. 13.8a, observed at time t, have well-defined crystallographic shapes. As the islands touch in Fig. 13.8b, taken 1 to 2 s later, there is a very rapid mass transport between them. The driving force for coalescence is a reduction in surface energy causing the islands to become taller and more rounded. Their total projected area on the substrate

716


is decreased allowing further secondary nucleation. In the early stages of deposition when the fractional coverage is still relatively low, composite islands after coalescing can once more assume a crystallographic shape as shown in Fig. 13.8c which was taken at time (t + 60) s. Recrystallization, especially for the smaller islands, can also occur during coalescence. This tends to eliminate mutual misorientation and provide an eventual average grain size which is much larger than the average nuclei size prior to coalescence.

Figure 13.6. Nuclei number density as a function of deposition time for Au evaporated onto NaCl(100). The deposition rate R and growth temperature Ts were 1013 cm-2 s -1 and 250oC, respectively. (From Ref. 14.)

As islands continue to coalesce, holes and channels are left in an otherwise continuous film. These voids are eventually filled by secondary nucleation and island growth. Figure 13.9[15] is a sequence of transmission electron micrographs showing the liquid-like bridging and filling of a channel in a Au overlayer on MoS2 . Islands resulting from secondary nuclei are also visible. Films which display a dense population of small islands during the initial stages of deposition will become continuous at a relatively low average film thickness, typically a few nm to a few tens of nm. However, films consisting of only a few large islands during the early stages of deposition will exhibit an island structure which persists up to relatively large average film thicknesses.


717

Figure 13.7. Replication transmission electron micrographs of Au islands on NaCl(100) as a function of deposition time: (a) 0.5 min, (b) 1.5 min, (c) 4 min, (d) 8 min, (e) 10 min, (f) 15 min, (g) 30 min, and (h) 85 min. The deposition rate R and growth temperature Ts were 1x10 13 cm-2 s-1 and 250 oC, respectively. (From Ref. 14.)

718


Figure 13.8. Successive transmission electron micrographs showing the coalescence of Au islands (labeled A and B) on MoS2 during deposition at 400oC. (From Ref. 15.)

Figure 13.9. Successive transmission electron micrographs showing the filling of a channel during Au deposition on MoS2 at 400oC. (From Ref. 15.)


719

The average thickness at which three-dimensionally nucleated films become continuous depends primarily upon the film and substrate materials, the supersaturation, and Ts. This can be demonstrated by re-deriving Eq. 3, this time accounting for all of the terms in Eq. 1. The critical cluster size then becomes

Eq. (10)

r* =

−2 a1Γc − v + a 2 Γs − c − a 2 Γs − v 3 a 3 ∆Gv

Thus, r* can be decreased, leading to lower average thicknesses, , required for obtaining a continuous films, by choosing a low surface-energy condensate, a high surface-energy substrate, and a condensate/substrate combination with a low interfacial energy. Equation 10 also shows that r*, and hence , varies inversely with ∆Gv. Therefore, following the earlier discussion of Eq. 8, should decrease with an increase in the deposition rate (i.e., the degree of supersaturation) at constant growth temperature, and with a decrease in Ts at constant R. Note, however, that orders-of-magnitude changes in R are required in order to significantly affect r* since ∆Gv only depends logarithmically on the supersaturation. Finally, for given values of both R and Ts, lower values of r* and are also expected for materials which have high boiling (or sublimation) temperatures such as W, Mo, Ta, Pt, and Ni. This follows from the fact that they exhibit lower equilibrium vapor pressures, and hence higher supersaturation rates, than low-boiling-temperature materials deposited under the same conditions. The average grain size of thin polycrystalline films, except in the case of deposition under conditions which result in very low adatom surface mobility, will be larger than the critical nucleus size and, generally, larger even than the average island size upon coalescence. However, will usually be less than the film thickness t and will increase with increasing t. Typical results are shown in Fig. 13.10a, for evaporated InSb films on cleaved mica substrates.[16] The films were grown at ambient temperatures and then annealed at 465oC. Similar results were obtained by Greene and Wickersham[17] for sputter deposited InSb on CaF2(111). The latter authors also found (see Fig. 13.10b) that for a given film thickness, increased with both increasing Ts and decreasing R as expected. Increasing Ts results in an increase in adatom diffusivity while decreasing R provides adatoms a longer time to find low-energy sites before they become buried by subsequent layers.

720


Figure 13.10a. Average grain size vs. film thickness t for InSb layers deposited by evaporation onto cleaved mica substrates at ambient temperature and then annealed at 465oC. (From Ref. 16.)

Figure 13.10b. Average grain size normalized to the film thickness t plotted as a function of the growth temperature Ts and the deposition rate R of InSb layers deposited on CaF2(111) substrates by rf sputtering. (From Ref. 17.)


721

In addition to large-angle grain boundaries, a high density of other mechanical defects such as dislocations, dislocation loops, twins, stacking faults, and low-angle boundaries are often observed in polycrystalline films. Electron microscopy studies[18] have shown that islands during the initial stages of epitaxial film growth are essentially perfect crystallites. However, as coalescence occurs, the observed defect density increases rapidly. Figure 13.11[19] shows dislocation densities as a function of the thickness of Au films evaporated onto cleaved MoS2(111) at 300oC. Such dislocations, as well as other line and volume defects, form for a variety of reasons. Island rotation and recrystallization to eliminate misorientation during coalescence becomes more difficult in the later stages of growth as the islands become larger. Stresses in continuous films due to film/substrate lattice constant and thermal expansion mismatch can be partially relieved by the generation of climbing dislocations. Contamination can also play an important role by inhibiting island re-orientation and recrystallization during coalescence.[20]

Figure 13.11. Dislocation density as a function of film thickness t for Au layers deposited by evaporation onto cleaved MoS2(111) substrates at 300oC. (From Ref. 19.)

2.2 Two-Dimensional Nucleation and Growth Much of the discussion of three dimensional nucleation in the above section evolved from the implicit assumption in Eq. 1 that

722


Eq. (11)

(

2 2 2 a2 r Γs −v < a2 r Γs −c + a1r Γc −v

)

In other words, the net surface free energy associated with the formation of a cluster is positive. This led directly to the establishment of a free-energy barrier (or, equivalently, a critical cluster size) to be overcome in order for embryonic clusters to grow and, in turn, required that the supersaturation necessary to obtain film growth be greater than unity. The assumption represented by Eq. 11 is generally a good one. However, there are cases, such as, for example, the growth of material A on a clean single-crystal surface of A in ultra-high vacuum (UHV), where Eq. 11 does not hold and no nucleation barrier exists. Instead, growth occurs in a quasi layer-by-layer fashion with the motion of steps, which may be as small as one monolayer, across the surface. In the presence of such steps, often insured by the use of substrates with vicinal surfaces, growth can proceed with supersaturations approaching unity. Even in the absence of an initially high step density, steps can be created under conditions of high supersaturation. A more general description of the requirement for a film/substrate system to exhibit two-dimensional nucleation is that the desorption energy of film atoms condensed on the substrate be equal to or larger than the desorption energy of film atoms condensed on other film atoms. That is, Eq. (12)

Ed,s ≥ Ed,f

where Ed,s and Ed,f are the desorption energies for film adatoms on substrate and film surfaces respectively. One obvious way to fulfill this requirement is to grow a film on a substrate of the same material, e.g., Si on Si or GaAs on GaAs, under very clean conditions. In the case of Si, it was shown by Joyce et. al.[21] that small traces of carbon, near the detection limit of Auger spectroscopy (≈ 0.01 monolayer), resulted in island formation during film growth on Si(111)7x7 surfaces. Reflection high-energy electron diffraction (RHEED) analysis showed that the carbon was in the form of βSiC particles.[22] Removing the carbon by flash heating the sample to ≥ 1200o C,[21] sputter cleaning and annealing,[23] or depositing a thick buffer layer over the contaminated surface[24] resulted in two-dimensional nucleation and growth of Si films by molecular beam epitaxy (MBE),[24] chemical vapor deposition (CVD),[23] and sputter deposition. [25] RHEED has long been used as an in situ diagnostic technique for investigating surface reconstruction and roughness during film growth. However, more recently, it has been applied to in situ nucleation studies (see, for example, Refs. 26 - 28). In one mode of operation, the time variation of


723

the intensity (RHEED oscillations) of a particular RHEED reflection is recorded. RHEED oscillations occur due to periodic changes in island and step number densities (i.e., alternate surface roughening and smoothening) during two-dimensional growth. Examples are shown in Fig. 13.12 for MBE GaAs growth at Ts = 600o C on a vicinal (100) substrate cut at an angle toward the [110] direction to provide terrace lengths of 280 nm. The deposition rate (controlled by the Ga flux) was varied such that the time to deposit one monolayer (one oscillation period) was: (a) 150 s, (b) 40 s, (c) 16 s, and (d) 5 s. Note that both the amplitude and the minimum value of the intensity oscillations decrease with increasing Ga flux indicating a rougher surface at higher deposition rates. Van Hove and Cohen[27] used the data in Fig. 13.12 together with a two-level diffraction model to extract surface-adatom diffusion lengths. The overall decrease in oscillation amplitude which occurs with increasing deposition time is an indication of increasing surface roughness. Van Hove and Cohen also showed that extremely smooth growth surfaces can be obtained by starting with vicinal substrates having very short terrace lengths such that the dominate growth mode is step propagation and island formation is minimized. In this case, RHEED oscillations are not observed since the average terrace lengths do not change with time. Rockett [29][30] has recently applied Monte-Carlo simulation techniques to model surface roughening during two-dimensional Si growth and used the results to calculate RHEED oscillations.

Figure 13.12. RHEED intensity oscillations during GaAs MBE growth on a vicinal (100) GaAs substrate at 600oC. The deposition flux was varied such that the time to deposit one bilayer was: (a) 150 s, (b) 40 s, (c) 16 s, and (d) 5 s. (From Ref. 27.)

724


Two-dimensional heterostructure growth has also been investigated using RHEED oscillations. Figure 13.13 shows that the recovery time for the intensity of a specular RHEED beam from (100)2x4 reconstructed surfaces is much longer for Ga0.79Al0.21Ga layers than for GaAs[31] indicating that the mean adatom diffusion length decreases in the presence of Al. Note that only one monolayer of GaAs on (Ga,Al)As dramatically decreases the recovery time.

Figure 13.13. RHEED intensity data showing differences in the recovery time for the intensity of a specular beam after shuttering the group-III effusion cells during the MBE growth of GaAs and (Ga,Al)As on GaAs(100). (From Ref. 31.)

The growth of rare-gas crystals on a variety of substrates also fulfills the conditions given by Eq. 11. Price and Venables[32] have used in situ transmission electron microscopy to investigate the two-dimensional nucleation of fcc Ar, Kr, and Xe crystals on graphite. Uniform epitaxial layer


725

growth of Xe was observed at substrate temperatures between 9 and 55 K and for incident fluxes from 2 x 1014 to 1017 cm-2 s-1. The orientation relationships observed were Xe(111) || C(0001) and Xe( 220 ) || C(1010). The authors noted that, as in the case of Si epitaxy, small amounts of substrate surface contamination were sufficient to cause the nucleation mode to revert to three-dimensional. An interesting metallic heterostructure system which exhibits twodimensional nucleation, Cd on W, was investigated by Wagner and Voorhoeve.[33] The authors used a combination of mass spectrometric desorption measurements as a function of Ts and overlayer coverage θ with replication TEM to study nucleation and the initial stages of crystal growth in UHV. They found that the desorption energy of Cd on W was greater than that for Cd on Cd. On clean polycrystalline W substrates, Ed decreased from an initial value of 2.2 eV at low coverages to a value approximately equal to the heat of sublimation of bulk Cd, 1.2 eV, at a coverage of a few monolayers. Within a given W grain, Cd grew epitaxially at Ts ≈ 100oC. Film growth occurred with no barrier to nucleation, regardless of coverage, and no supersaturation was required. The introduction of a small amount of oxygen (less than a monolayer) reduced the growth rate, provided a nucleation barrier which changed the nucleation mode to three dimensional, and inhibited epitaxial growth.[34] The replication electron micrographs in Fig. 13.14 show the effect of oxygen contamination very graphically. Other examples of two dimensional growth of metals on metals include Pb on Cu(111)[52] and Bi on Cu(100).[36] Heteroepitaxial films which are grown in a layer-by-layer fashion are often pseudomorphic with the substrate up to a critical thickness at which misfit dislocations are generated. That is, in thin layers the lattice constant mismatch is accommodated through elastic strain and the film/substrate interface is coherent. Figure 13.15[37] shows measured changes in film lattice constants as a function of thickness for PbSe on PbTe, PbSe on PbS, and Au on Pd. The first two films were grown at 230o C while the latter was grown at 300oC. The lattice of PbSe is ≈ 5.3% smaller than PbTe leaving the film in tension during growth while the other two systems correspond to the introduction of a compressive stress in the film. In each case, the lattice constant of very thin films, < 2 nm, was approximately equal to that of the substrate. As the films became thicker, they were less elastic and their lattice constant increased towards the bulk film value. At critical film thicknesses, the interfaces became incoherent as dislocations and other mechanical defects were generated.

726


Figure 13.14. Replication transmission electron micrographs of (a) Cd deposited epitaxially on polycrystalline W by evaporation in UHV (from Ref. 33) and (b) Cd deposited on polycrystalline W in the presence of less than a monolayer of oxygen contamination (from Ref. 34). The upper and lower micrographs illustrate “layer-bylayer” and three-dimensional nucleation and growth, respectively.


727

Figure 13.15. Film/substrate lattice-constant mismatch plotted as a function of film thickness for: (a) PbSe on PbTe, (b) PbSe on PbS, and (c) Au on Pd. (From Ref. 37.)

728


Eltoukhy and Greene[38] found that sputter-deposited single-crystal InSb/GaSb(100) interfaces (room-temperature lattice-constant mismatch ≈ 7%) became completely incoherent in a superlattice with layer thicknesses of ≥ 3.5 nm. This was in good agreement with theoretical predictions. Using expressions derived by Matthews and Blakeslee,[39] the critical thicknesses for (100) and (111) interfaces were calculated to be 2 and 5.7 nm respectively. In the (110) case, incoherence was associated with the generation of 60o dislocations with ½a<110> Burgers vectors on {111} slip planes. Another interesting example of pseudomorphism is the MBE growth of GexSi1-x alloys on Si(100) at 550o C.[40] Single-crystal films with completely coherent interfaces, as judged by cross-sectional TEM, were obtained with alloys having compositions out to at least 50 at. % Ge. The critical thickness was found to vary from ≥ 10 nm for films with x = 0.5 up to ≈250 nm for Ge0.2Si0.8. These thicknesses are quite large, certainly larger than would be predicted from an equilibrium analysis of strain energy (room temperature SiGe lattice mismatch, ≈ 4.5%) and dislocation formation.[40a][40b] 2.3 Stranski-Krastanov Nucleation and Growth Stranski-Krastanov growth refers to a mixed-mode in which the film initially nucleates two-dimensionally and then transforms to three-dimensional growth. A wide variety of systems including Cu on Ag(111),[41] Cu on Mo(100),[42] Sb on W(110),[43] K on W(100),[44] Ag on Si(111)7x7,[45] In on Si(100)2x1,[6] and InAs/GaAs(100)[46] follow this growth mode. The name derives from a calculation by Stranski and Krastanov[47] in which they showed that for a monovalent ionic crystal M+X- condensing onto a divalent M2+X2- substrate, the second M+X - layer is less strongly bound, while the first M+X - epitaxial layer is more tightly bound, than the surface layer of a bulk MX crystal. The normalized desorption energy as a function of film layer thickness is shown schematically in Fig. 13.16[48] for representative twodimensional (Xe/graphite),[49] three-dimensional (Au/KCl), and StranskiKrastanov (K/W) systems. Figure 13.17, is a plot of the peak-to-peak intensities of differentiated Auger electron spectroscopy (AES) In 404 eV MN4,5N4,5 and Si 92 eV LM2,3M2,3 lines as a function of In coverage θ In on Si(100)2x1 at Ts = 70oC.[6] The data were taken in situ during an MBE nucleation experiment and provide, together with RHEED and low-energy electron diffraction (LEED) data, a clear signature of Stranski-Krastanov growth. The solid line in Fig. 13.17 represents calculated Si AES intensities ISi based upon two-


729

dimensional growth. The escape depth for Si Auger electrons through the In overlayer was obtained by fitting data from the first In monolayer (ML).

Figure 13.16. Adatom desorption energy Ed plotted as a function of the number of adlayers for Au on KCl, Xe on graphite, and K on W. Each system is representative of a different growth mode. (From Ref. 48.)

Figure 13.17. Peak-to-peak intensities I of differentiated Si 92 eV and In 404 eV Auger lines as a function of In coverage θ on Si(100)2x1 substrates at 70 and 300oC. A calculated curve for ISi vs. θ at 70oC, assuming two-dimensional growth, is also shown. (From Ref. 6.)

730


During deposition of the first three monolayers at Ts = 70 oC, the measured Auger intensities were in good agreement with the intensities calculated for two-dimensional growth. At higher coverages, however, the rate of change of the experimentally obtained intensities was much less than that of the calculated curves, indicating the onset of three-dimensional growth on top of the two-dimensional adlayer. The local maximum in IIn and the corresponding local minimum in ISi at θ ≈ 3.5 ML in Fig. 13.17 is due to the fact that nucleation of three-dimensional islands requires a critical supersaturation of adatoms in order to obtain nuclei of critical size r* (see Sec. 2.1). However, once nuclei with r > r* were formed, further growth occurred three-dimensionally. Note that at Ts = 300oC, three-dimensional growth begins at θIn = 1 ML. Figures 13.18a and 13.18b show high-resolution scanning electron micrographs of the surfaces of clean Si(100) samples with In coverages of 6 and 200 ML, respectively. Island growth is seen to be highly oriented with the islands forming single-crystal polyhedra having major axes along the [011] and [011 ] directions. Island orientations were easily obtained from electron channeling patterns of the Si(100) substrate since the islands covered only ≈ 1% of the surface in Fig. 13.18a and ≈ 7% in Fig. 13.18b. The quasi “one-dimensional wire” shaped In islands that form along <011> directions at θIn > 3 ML exhibit a clear crystallographic relationship with the underlying substrate. However, contamination of the substrate with less than 0.01 ML of oxygen and carbon prior to deposition completely altered the nucleation mode from Stranski-Krastanov to three-dimensional growth of hemispherical islands. The formation of the elongated islands on clean substrates was explained based upon diffraction[6][50] and sychnrotron XPS core-level studies[51] of the In-stabilized Si(100)2x2-In surface which has one-dimensional channels along <011> directions. Assuming that the Si(100)2x1-In surface has similar channels, one would expect higher adatom mobility down the channels rather than across them. Enhanced mobilities along surface channels have been observed directly for fcc and bcc metal surfaces using field-ion microscopy and atom-probe techniques.[52]

3.0 COMPUTER SIMULATIONS OF MICROSTRUCTURE EVOLUTION Continued deposition past the nucleation stage eventually leads to island coalescence and the growth of a continuous layer. The film microstructure, especially during deposition at relatively low temperatures, continues to evolve from the nucleation coalescence stage often through many


731

Figure 13.18. Scanning electron micrographs of polyhedral In islands on Si(100)2x1 for coverages θ of 6 and 200 monolayers (Ts = 70°C). The islands are oriented along [011] and [ 011 ] directions. (From Ref. 6.)

732


hundreds of nanometers to micrometers of film thickness, before reaching a steady-state configuration. Recent computer simulations of microstructural evolution have been shown to qualitatively explain many of the features observed experimentally (see Sec. 4 for a discussion of experimental results). In addition, the simulations provide a very useful tool for visualizing, as well as for testing, atomistic models of film growth. 3.1 Film Growth in the Ballistic Aggregation, Low-Adatom Mobility, Limit The cross-sections of both polycrystalline and amorphous films deposited from the vapor phase at low temperatures are typically composed of open columnar structures with extended voids along the column boundaries. (“Low-temperatures” in the above sense generally corresponds to deposition temperatures Ts which are less than ~0.3 of the melting point Tm of the deposited material). Figure 13.19 is a scanning electron micrograph showing an example of such a microstructure observed in a Cr film deposited by cylindrical-magnetron sputter deposition onto a glass substrate cooled by liquid nitrogen.[53] (For reviews of early literature citing many examples of open columnar microstructures, see Refs. 54 and 55.) The porous network in such columnar films results in poor mechanical properties together with optical properties which are very sensitive to the environment, due to adsorption of water vapor and other atmospheric contaminants in the voids upon air-exposure. Other properties, such as magnetization and electrical resistivity, are extremely anisotropic.

Figure 13.19. Scanning electron micrograph of a Cr film deposited by cylindricalmagnetron sputter deposition onto a glass substrate cooled by liquid nitrogen. (From Ref. 53.)


733

Monte Carlo computer simulations[54]-[58] have shown that the open columnar structure is caused by low-adatom mobilities combined with selfshadowing by previously deposited atoms. Figure 13.20 shows the results of ballistic aggregation simulations of film growth using hard-sphere atoms which are incident at randomly chosen surface positions.[54] The adatoms are not allowed to diffuse over the surface but only to relax into the nearestlying cradle formed by at least two deposited atoms. As a consequence, extended microvoid formation leading to columnar structures occurs due to atomic self-shadowing by protruding clusters and small ledges. The simulated microstructures in Fig. 13.20 are slices, five atom diameters thick, through three-dimensional arrays. The columnar structure becomes increasing more noticeable with larger angles of incidence α, measured with respect to the substrate surface normal, of the vapor flux.

Figure 13.20. Monte-Carlo computer simulations of amorphous films deposited with incident flux angles a of (a) 45°, (b) 60° , and (c) 75 °. (From Ref. 54.) The figures show slices, five atoms thick, through three-dimensional arrays.

Films with underdense columnar microstructures such as shown in Fig. 13.19 are typically found to exhibit in-plane tensile stress.[59]-[61] This was shown by Müller,[62] using molecular-dynamic computer simulations of the growth of a Lennard-Jones metal, to be due to attractive interatomic forces in microvoided regions. Note, however, that contamination along intercon-

734


nected column and void boundaries following air-exposure of the film can result in large, and time-dependent, changes in measured stress values. 3.2 Effects of Adatom Migration Müller[58] has included adatom migration effects in microstructureevolution growth simulations. He allowed thermally-activated adatoms to jump to empty neighboring sites of maximum coordination number. The activation barrier∆E for migration on a terrace from a site i with Ni neighboring atoms to site j with Nj neighbors was assumed to be Eq. (13)

∆E = =

Q

if Ni < Nj

(Nj - N i)Φ + Q

otherwise

where Q is the activation energy for surface diffusion and Φ is the energy of a single bond. Boltzmann statistics were used to simulate fluctuations in adatom vibrational energy. Müller found that above a critical temperature range the porous columnar microstructure changes to a configuration of maximum packing density. That is, the hopping rate of adatoms to shadowed regions becomes large enough to exceed the rate of void incorporation. Figure 13.21 shows calculated results for a two-dimensional Ni lattice (Φ = - 0.74 eV) deposited at a rate of 1 nm s-1 with a vapor impingement angle α = 45o . The lower three atom layers in the figure correspond to the substrate. The typical open columnar structure characterizing low adatom mobility growth was obtained at Ts = 350 K. Increasing the deposition temperature to 420 K resulted in a film with much higher density, although still columnar. At Ts = 450 K, a fully dense film with local defects was obtained. Müller’s simulations also predict that the temperature range over which the transition occurs from a film with a columnar microstructure to one that is densely packed increases slowly with increasing deposition rate. Figure 13.22 shows results for a Ni film deposited at α = 45o in which R was increased from 0.01 to 100 mn s-1 . The four orders of magnitude increase in deposition rate raised the “transition” temperature from 340 to 525 K since higher adatom migration rates were required to overcome the larger void incorporation at higher deposition rates.

Nucleation, Film Growth, and Microstructural Evolution 735

Figure 13.21. Computer-simulated two-dimensional microstructures of Ni films deposited with an incident flux angle α = 45o, deposition rate R = 1 nm/s, and growth temperatures Ts of (a) 350 K, (b) 420 K, and (c) 450 K. The deposition time, t, is shown. (From Ref. 58.)

736


Monte Carlo simulations by Srolovitz and co-workers[63[[64] have recently demonstrated mechanisms by which “grain growth”, leading to bimodel grain size distributions, can occur during film deposition at temperatures which are too low for significant grain boundary migration in the bulk, but high enough to allow sufficient adatom diffusion for grain boundary migration to occur at the free surface. In this model, the evolution of grain size is determined by the curvature of the grain boundaries intersecting the growth surface. The introduction of grains with low surface energy, i.e. crystallographic texture, in the early stages of film deposition was found to greatly accelerate grain growth.[64]

Figure 13.22. Calculated packing densities as a function of growth temperature Ts for Ni films grown at a vapor impingement angle α = 45o and depositions rates R = 0.01, 1, and 100 nm/s. Ta, Tb, and Tc are the “transition temperatures” for the three R values. (From Ref. 58).

4.0 MICROSTRUCTURE EVOLUTION AND STRUCTURE-ZONE MODELS Movchan and Demchisin[65] were the first to categorize microstructures observed in vapor-deposited films using a structure-zone diagram (SZD) in which the general features were schematically illustrated as a function of the normalized growth temperature, Ts/T m. The SZD, reproduced in Fig. 13.23, was based on the results of their studies of the microstructure of thick (0.3x10 3 - 2x10 3 µm) Ti, Ni, W, ZrO 2, and Al 2 O3 coatings deposited by high-rate (of the order of 100 - 700 µm h-1)


737

electron-beam evaporation as well as general features reported by other researchers. Based primarily upon optical metallographic studies, Movchan and Demchisin concluded that their deposited coatings could be represented as a function of Ts/Tm in terms of three zones, each with its own characteristic microstructure and physical properties.

Figure 13.23. Structure-zone diagram for thick high-rate evaporated films. T1 and T2 are the growth transition temperatures between zone-1/zone-2 and zone-2/zone3 microstructures, respectively, as described in the text. (From Ref. 65.)

Microstructures in zone 1 (Ts/Tm < 0.2 - 0.3) consisted of tapered crystals with domed tops which are separated by voided boundaries. The internal structures of the crystals, on the scale of the resolution of optical metallography, was poorly defined. The crystallite width increased with increasing Ts/Tm following a dependence that implies an apparent activation energy, 0.1 - 0.2 eV, which is too low to be explained by grain-growth (i.e., bulk or surface diffusion) mechanisms. Zone 2 (Ts/Tm 0.3 - 0.5) microstructures consisted of columnar grains separated by dense intercrystalline boundaries. The surface structure exhibited a more smooth matte appearance. Average grain widths were typically less than the film thickness t and increased with increasing Ts/Tm. The dependence of upon Ts/Tm yielded apparent activation energies of the order of that expected for surface diffusion. Zone 3 (0.5 < Ts/Tm < 1) microstructures consisted of more equiaxed grains and a bright (for metals) smooth surface. The apparent activation energy for as a function of Ts/Tm corresponded to that of bulk

738


self-diffusion. The transition between zone 2 and zone 3 microstructure was gradual, thus the boundary was drawn with a positive slope to it. The simplicity of the Movchan and Demchisin SZD insured its popularity and many researchers showed that the general features represented in the diagram also applied to films whose thickness was of the order of micrometers rather than millimeters (see, for example, Ref. 66), to films deposited by other techniques such as sputtering (see, for example, the review by Thornton in Ref. 67), and to amorphous as well as polycrystalline materials.[68]-[73] Thornton[53][74] extended the Movchan and Demchisin SZD by adding an additional axis to account for the pressure of the sputtering gas during cylindrical-post magnetron sputter deposition of 25 - 250 µm thick Ti, Cr, Fe, Cu, Mo, and Al coatings. The effect of increasing Ar pressure in Fig. 13.24 is shown to increase the normalized temperatures at which the zone boundaries occur. However, the sputtering gas pressure P is not a fundamental parameter. Rather, the pressure affects the film microstructure through several indirect mechanisms. Increasing the

Figure 13.24. Structure-zone diagram showing schematic microstructures of films deposited by cylindrical magnetron sputtering as a function of growth temperature and Ar pressure. (From Ref. 53.)


739

pressure to values such that the mean-fee path for elastic collisions between sputtered (or evaporated) species and the fill gas becomes of the order of the source-substrate distance increases the oblique component of the deposition flux resulting in a more open zone-1 type structure. In addition, decreasing the pressure during sputter deposition results in increased energetic-particle bombardment, and hence densification, of the growing film as discussed below in Sec. 5. Finally, it has been proposed that increasing the inert gas pressure during deposition leads to a decrease in adatom mobilities.[53] While this is undoubtedly true, it is probably not a major effect since the surface residence time of physisorbed inert gas species is very short, and hence the steady-state coverage is quite low. Thornton also added an additional region, labeled zone T in Fig. 13.24, to his SZD which consisted of a “dense array of poorly defined fibrous grains” which represented the transition between zones 1 and 2. He defined the zone T structure as the limiting form of the zone 1 structure at zero Ts/Tm.[67] That is, the zone T “fibers” formed the internal structure of the zone 1 crystallites. Examples of fracture cross-sections of metal coatings exhibiting zones 1, T, and 2 structures are shown in Fig. 13.25. The coatings were deposited using cylindrical hollow-magnetron sputtering at the Ts/Tm values indicated and examined by scanning electron microscopy (SEM). Grovenor et al.[75] have further modified earlier SZDs based upon the results of their plan-view transmission electron microscopy (TEM) examinations of evaporated metal films with thicknesses of either 100 nm (Ni, Pt, Au, Cu, Al, Pb, Ti, Co, W, and Cr) or 9 - 14 µm (Ni and Ni-Al alloys thinned from both sides and examined at a thickness of ≈ 5-7 µm from the substrate). These researchers observed that the tapered columns in the zone-1 structure are not single grains but are composed of bundles of small grains (with sizes of order of tens of nm depending upon film thickness) which are relatively equiaxed. Their zone 1 corresponded to Ts/Tm < 0.1 while zone T, in which the small-grained substructure consisted of a bimodel distribution of sizes, extended to Ts/Tm = 0.3. Substructure was also observed in sputter-deposited amorphous Ge films by Messier et al.[73] using a combination of SEM, TEM, and field-ion microscopy. The microstructure of the a-Ge films consisted of columns which were observed to be composed of smaller columns spanning several different size scales, in a fractal-like behavior,[76] ranging from ≈ 2 nm to 300 nm. Messier further extended earlier SZDs by adding a film thickness axis to underscore the evolutionary nature of film microstructure as the dominant features move through different observable size scales.

740 Deposition Technologies for Films and Coatings Figure 13.25. Scanning electron microscopy cross-sections of metal coatings deposited by cylindrical magnetron sputtering in Ar illustrating (a) zone-1 microstructure, (b) zone-T microstructure, and (c) zone-2 microstructure. (From Ref. 53).


741

The researchers proposing the structure-zone diagrams described above discuss several mechanisms which play important roles in determining microstructural development. That is, the columnar structure with open voided boundaries characteristic of zone 1 is formed due to atomic selfshadowing and clustering effects which occur in the very low adatommobility, or ballistic aggregation, growth regime. Increasing adatom surface diffusion at higher growth temperatures (zones T and 2) give rise to denser, although still columnar, structures while grain growth (recrystallization during deposition) occurs at deposition temperatures above ≈ 0.5 Tm. Grovenor et al.[75] also point out that activation energies of grain boundaries are a function of their crystallography. Thus, even at low temperatures (e.g., zone T), a few boundaries could become mobile. It is important to note that while SZDs provide a useful, simple, method of qualitatively categorizing observed film microstructures, they do not provide quantitative insight into the mechanisms of film growth. Moreover, one should take care in using such diagrams to “predict” expected film microstructures and properties since film-growth kinetics are strongly dependent on other factors in addition to Ts/Tm. For example, substrate surface roughness due to poor substrate preparation,[67] or patterning as used in microelectronic device fabrication,[77] can promote zone-1 behavior at elevated temperatures by enhancing shadowing due to oblique deposition angles. Contamination can play a significant, and often controlling role, in determining film microstructure. Even very small concentrations of chemically-reactive species such as oxygen can have dramatic effects on nucleation and growth kinetics as discussed in Sec. 5.2. In terms of SZDs, contamination generally reduces adatom mobilities [78][79] and therefore acts to promote zone-1 structures. Substrate material, crystal structure, and orientation are also important (see discussion below in Sec. 5.2). This is illustrated in the cross-sectional TEM (XTEM) micrographs in Fig. 13.26 which show microstructures of polycrystalline TiN films deposited by reactive magnetron sputter deposition on a two-phase high-speedsteel substrate. [80] Even at Ts = 200 oC (T s/T m = 0.15), the microstructure of the film grown over the substrate martensetic matrix is a dense columnar structure (zone 2) with single-crystal columns (as determined by selective-area diffraction), many of which extend through the entire ≈ 4-µmthick film. Over the VC precipitates in the substrate, large TiN grains grew epitaxially and, at higher growth temperatures (Ts = 500 oC, T s/T m = 0.24) also extended throughout the thickness of the film. Low-energy particle

742 Deposition Technologies for Films and Coatings Figure 13.26. Cross-sectional transmission electron micrographs of the microstructure of TiN films deposited by reactive magnetron sputtering on the martensitic matrix and the MC carbide phase (VC) of high-speed steel substrates at (a) 200oC, (b) 450oC, and (c) 550oC. (From Ref. 80).


743

bombardment of the substrate and growing film during deposition, as often found in sputtering and plasma-assisted depositiontechniques and discussed in detail in the following section, can also have significant effects on film microstructure.

5.0 EFFECTS OF LOW-ENERGY ION IRRADIATION DURING FILM GROWTH Low-energy (often < 100 eV) ion and fast-neutral irradiation during vapor-phase film growth has been shown to be useful for controllably altering the microstructure of as-deposited layers. In PVD, ion irradiation during film growth is commonly used in glow-discharge bias sputter deposition (see Ch. 5). In addition, low-energy ion bombardment is also an important technique for modifying the properties of films deposited by primary-ion deposition (PID) and ion-assisted molecular-beam epitaxy (MBE) as well as by CVD techniques such as plasma-assisted chemical vapor deposition (PA-CVD). Examples of applications in which low-energy ion/surface interactions are used to modify film microstructure include: densification and increased oxidation resistance in optical films; minimization or elimination of columnar microstructure in microelectronic metallization layers; altering the state of stress, average grain size, and preferred orientation; increased film/substrate adhesion; enhanced conformal coverage; controlling magnetic anisotropy in recording layers; and low-temperature epitaxy. Although the focus of this chapter is on microstructural effects, it is important to be aware that ion irradiation is also used to controllably alter film microchemistry through collisionally-induced dissociative chemisorption,[81] preferential resputtering,[82]-[86] and selective deposition on patterned substrates.[87][89] Ion irradiation has been used to promote the growth of unique new single-crystal metastable semiconductors[90]-[92] and to increase elemental incorporation probabilities with a corresponding decrease in segregationinduced broadening of dopant profiles in MBE-grown Si.[93]-[95] 5.1 Effects of Low-Energy Ion/Surface Interactions on Nucleation Kinetics As discussed in recent review articles,[1]-[4] there have been many reports in the literature over the past 10 to 15 years on nucleation studies

744


involving low-energy ion irradiation during deposition. The experiments have been carried out by a variety of means including: evaporation in which a portion of the evaporant stream is ionized and accelerated to the substrate (e.g., ion plating or PID from solid-source ion guns), evaporation in the presence of a separate ion source which provides the accelerated particle bombardment, and sputter deposition or PA-CVD with the application of a substrate bias. Depending upon the deposition technique, there are, in addition to ions which are purposely accelerated to the substrate, other energetic particles such as secondary electrons, ions reflected (generally as neutrals) from a sputtering target, UV photons, and sputtered species incident at the substrate and growing film. Although the average ejection energy of sputtered particles under typical deposition conditions is ≈ 5 - 20 eV, there is a high energy tail in the sputtered atom velocity distribution extending well above this range.[96] Since much of the early work concerning ion-irradiation effects on nucleation kinetics was carried out in low to medium vacuum, at least part of the observed effects were very likely due to sputter cleaning of the substrate. Donahue and Reif[97] have argued, for example, that this effect played a major role in allowing the homoepitaxial growth of Si by PA-CVD from SiH4 at substrate temperatures as low as 650o C. It should be noted, however, that ion irradiation sometimes causes increased substrate contamination, particularly in the presence of high hydrocarbon background pressures, due to collisionally-induced dissociative chemisorption. Low-energy ion bombardment of the substrate and growing film can lead, in addition to sputter cleaning, to fundamental changes in nucleation kinetics. Examples of irradiation-induced effects include the production of defects in the substrate surface which can act as preferred adsorption sites, trapping or implanting of incident species in the near-surface region, the dissociation of small clusters during the early stages of growth, enhanced adatom diffusion, and local electric field effects due to charging. The effects which dominate in a given experiment depend upon the film/substrate combination, the energy Ei, flux Ji, and mass mi of the incident particles, and the growth temperature Ts. The number density of active ion-irradiation-induced preferential nucleation sites n will be determined by the difference between the production rate, which is a function of E i, Ji , and the ion/substrate species, and the loss rate due to annealing (which increases with increasing Ts) during deposition. Krikorian and Sneed[98] have shown, for example, that ion irradiation can be used to either increase or decrease the nucleation rate dn/


745

dt of Ge depending upon the choice of Ts and the substrate material. Figure 13.27 shows a comparison of island densities n for Ge deposited on amorphous carbon substrates at 550oC by 5 keV Ar+ ion-beam sputtering from a Ge target, and by evaporation at similar deposition rates and in similar vacuum conditions, 10-7 to 10-8 Torr. The sputter-deposited films, which were subjected to bombardment by both fast sputtered Ge species (average energy ≈ 20 eV) and Ar ions backscattered from the target as fast neutrals, exhibited much higher nucleation rates. In addition, nuclei coalescence began to occur at considerably lower nominal film thicknesses in the sputterdeposited layers than in the evaporated films.

Figure 13.27. Ge island number densities n on amorphous carbon substrates as a function of time during deposition by evaporation (deposition rate R = 0.6 nm/min) and ion-beam sputtering (R = 0.53 nm/min). In both cases, the film growth temperature Ts was 550oC. (From Ref. 98.)

One of the first studies of the role of incident energetic species during 3-D nucleation was carried out by Chapman and Campbell[99] who used very high energy, 40 - 50 keV, Ar and Xe ion-beam sputtering to deposit Au onto

746


NaCl(100) at Ts between 30 and 310oC. They reported an increase in the maximum number density nmax of Au islands and an enhancement in the degree of (100) preferred orientation compared with results for thermal evaporation. Lane and Anderson[100][101] also reported increases in nmax for Au, sputtered using a 2 keV Ar ion gun, onto cleaved NaCl at Ts = 130 - 325oC. The latter authors fit their results using classical nucleation theory with additional terms to account for the production of preferred adsorption sites. They assumed that these more strongly binding sites were due to defects produced in the NaCl substrate surface by bombardment with fast sputtered atoms. Harsdorff and Jark[102] used an analogous argument to explain increased dn/dt and nmax values obtained for RF-sputtered Au vs. evaporated Au films on cleaved NaCl substrates maintained at 270 - 360oC. Sputtering was carried out in He at 30 mTorr and the authors used magnetic fields to suppress electron and ion bombardment of the substrate. However, as in the previous experiments, this still did not eliminate bombardment by energetic neutral particles reflected from the target. In fact, the backscattered flux is expected to be particularly large in the latter experiments due to the large mass mismatch between He and Au. There are other mechanisms for producing preferred sites such as the trapping of incident particles due to low-energy “implantation” or adatom recoil processes. In addition, the disordering of a crystalline substrate surface by ion-bombardment at low temperatures prior to, or during, the early stages of deposition may lead to the production of new lower binding energy sites. Barnett et al.[103] have recently used modulated-beam mass spectrometry and thermally-stimulated desorption techniques to directly measure ion-irradiation-induced changes in Sb binding energies on Si(100) surfaces. Several authors (see, for example, Ref. 104) have reported a decrease in island number densities in the presence of ion irradiation leading to larger average island sizes , for a given nominal film thickness, and hence larger ultimate grain sizes. One mechanism which has been proposed[15] to provide increased values is the depletion of small clusters by sputtering and ion-induced dissociation.[105] As small clusters are reduced in size, they become more mobile on the surface and can diffuse to feed larger stable islands. Moreover, clusters reduced to subcritical size are energetically unfavorable and will spontaneously dissociate to form adatoms, some of which will desorb or diffuse to larger clusters. On the other hand, ion bombardment of large islands will only result in a minor loss in material by sputtering. These effects are expected to occur


747

when growth conditions leading to high nucleation rates are combined with, for example, self-ion bombardment under conditions in which there is a high incident ion-to-vapor flux ratio. Hasan et al.[105] have recently carried out experiments using both thermal In and partially-ionized In+ beams (ion-to-neutral ratio = 0.35) to deposit In islands on amorphous Si3 N4 substrates in an ultra-high vacuum (UHV) MBE system. The low-energy ion beams were provided using the UHV metal-ion source described in Ref. 106. The deposition temperature was ≈ 30oC (Ts/Tm = 0.7 where Tm is the In melting point in K), the thermal In flux was 3 x 1013 cm-2 s-1, and the total flux was 4.6 x 1013 cm-2s-1. A series of depositions were carried out with nominal film thicknesses t of 1.5 and 10 nm (the loss of material by sputtering was accounted for in determining t) with acceleration energies EIn of 0 (i.e., thermal), 150, 200, and 300 eV. Typical transmission electron microscopy (TEM) images of the resulting layers are shown in Fig. 13.28. For a given film thickness, ion irradiation clearly resulted in larger average island sizes. In fact, from island size distribution histograms obtained using lower-magnification, larger-area, micrographs, for films with t = 10 nm increased from 6.5 to 8 to 13 to 50 nm with increasing EIn. Another striking feature was the decrease in secondary nucleation at higher acceleration energies until at 300 eV essentially no islands were observed with ≤ 11 nm. (The minimum island size resolution was ≈ 1 nm). The histograms showed that the island size distributions were more uniform at higher EIn values. Increases in and decreases in the secondary nucleation rate were explained, as discussed above, as being due primarily to the loss of small clusters (incipient islands) by ion-bombardment-induced processes. The suppression of secondary nucleation in these experiments led, in turn, to significant differences in island growth kinetics. In the accelerated-beam case, island growth continued to be dominated by random surface-diffusion processes even for 10-nm-thick overlayers. However, in the thermal-beam deposits, coalescence of small islands had a much stronger influence on island growth resulting in the observed differences in island-size distribuions. Ion irradiation can also directly enhance adatom diffusivities during deposition through the initiation of shallow collision cascades and the excitation of surface phonons. However, the excess energy of incident accelerated ions, as well as the excess energy gained by adatoms involved in individual collision cascades, is lost to the lattice, i.e., the atoms become thermalized, within several vibrational periods. Thus this mechanism, while it may be important in contributing to bombardment-induced decreases in the

748


Figure 13.28. Transmission electron micrographs of In islands on Si3N4 substrates. Results are shown for two different nominal film thicknesses, t = 1.5 and 10 nm. The incident In beams were either thermal or partially ionized and accelerated to Ei = 150, 200, or 300 eV. (From Ref. 105.)


749

epitaxial temperature as discussed below, would not be expected to result in enhanced diffusion over distances of more than several lattice spacings, except in special cases. One example of such a special case was pointed out by Dodson[107] who used molecular dynamic simulations to show that low-energy (≤100 eV) ions incident at grazing angles on single crystal surfaces can exhibit translations of up to hundreds of nm via surface channeling. In summary, it is clear that energetic particle bombardment can greatly affect 3-D nucleation and growth kinetics of polycrystalline films, and hence, as will be discussed in the next section, grain size, preferred orientation, and defect concentrations. However, much work remains to be done to understand these effects in detail. Considerably less is known about the growth of epitaxial layers by accelerated beams. Several years ago, Narusawa et al.[108] reported a decrease in the epitaxial temperature Te of MBE Si on Si(111) and Al2O3 (1102) by ionizing a small fraction of the evaporant flux and accelerating it to the substrate. Te on both substrates was found to decrease by more than 100oC after ionizing ≤ 1% of the incident Si flux and accelerating it to 200 and 100 eV, respectively. Beckers and co-workers[109][110] later developed a hot cathode discharge, high-energy (10 - 15 keV), UHV, mass-filtered ion beam system followed by a deceleration lens to provide accelerated beams of, for example, Ag+ and Si+ with currents of the order of 10 µA at energies between ~ 20 and 100 eV. With this apparatus, they were able to grow epitaxial films of Ag on Si(111) at room temperature using acceleration energies between 25 and 100 eV. Epitaxial Si layers were also grown on Ge(100), Si(100) and Si(111) at Ts ≥ 230o C using 50 eV Si+. Herbots et al.[111] have recently demonstrated the low-temperature epitaxial growth of Si and Ge layers in UHV using decelerated beams from a modified Freeman ion source. Possible mechanisms, other than sputter cleaning,[112][113] for ionbombardment-induced enhancement in film epitaxy can be visualized using molecular dynamic simulations such as those by Müller (see, for example, Ref. 114). The use of an accelerated incident growth flux increases what Müller refers to as the impact mobility[115] of adatoms resulting in an increase (of the order of a few lattice spacings in these T = 0 K simulations) in the average distance between the point of first interaction with the substrate surface and the position where the atom is finally adsorbed. In addition, as discussed in more detail in the following section, atomic rearrangements of lattice atoms occur during both the first few hundred femtoseconds (fs)

750


following a bombardment event and during the subsequent relaxation, or refreezing, period. Figures 13.29a and 13.29b show calculated layer-by-layer film densities in which the first two layers are the upper substrate planes. A fully-dense layer in this two-dimensional simulation corresponds to 20 atoms per unit length. The simulated Ni overlayer in Fig. 13.29a was deposited at Ts = 0 K with no ion bombardment while that in Fig. 13.29b was grown with simultaneous 100 eV Ar ion irradiation using an ion-to-neutral flux ratio of 0.042. The thermally-deposited overlayer had a high defect concentration and only the first four layers crystallized in the same orientation as the substrate. Subsequent layers were disordered with no indication of texture. Figure 13.29b, on the other hand, shows that ion irradiation during growth, even in the absence of thermal diffusion, resulted in the deposition of well-oriented epitaxial layers. Tsao et al.[116] have recently carried out some of the first in situ measurements of ion-irradiation effects during epitaxial film growth. The authors videotaped changes in reflection high-energy electron diffraction (video-RHEED) patterns resulting from the addition of 500 eV Ar+ ion irradiation during MBE growth of Ge(100) at 550o C. They proposed that the primary effect of ion irradiation in their initial experiments was to break up 3-D clusters to provide a smoother growth surface. Both experimental results and growth simulations show that low-energy ion bombardment during deposition can provide local atomic re-arrangement allowing atoms to relax into lower energy sites. However, the question of residual ion damage still needs to be addressed in more detail. The growth of high-quality films with reduced epitaxial temperatures requires a balance between the beneficial effects of ion irradiation such as enhanced diffusion and minimizing residual damage by annealing out bombardment-induced defects during deposition. As discussed in the next section, the most favorable deposition conditions would seem to be: low Ei, relatively high ionto- neutral flux ratios, and (provided that contamination from background impurities is not a problem) low deposition rates. 5.2 Effects of Low-Energy Ion/Surface Interactions on Film Growth Kinetics Interest in the use of low-energy ion irradiation during film growth to modify the morphology, microstructure, defect concentration, preferred orientation, state of stress, and physical properties of layers deposited from the vapor phase is continuously growing. In areas such as optical and wear-


751

Figure 13.29. Molecular-dynamic simulation of the layer densities for a Ni film deposited at Ts = 0 K with (a) no ion bombardment and (b) 100 eV Ar+ ion bombardment at an ion-to-vapor flux ratio of 0.042. (K. H. Müller, unpublished).

752


protective films, the application of ion irradiation to densify and increase the internal strength of the layers as well as to increase film/substrate adhesion is already a prerequisite to commercial success. However, ion irradiation during growth can also be disadvantageous for certain applications due to, for example, the generation of very high compressive stress levels in the films which can lead to spalling. During growth at low substrate temperatures (Ts/Tm less than ~ 0.3) for which, as discussed in Sec. 4, the films are generally underdense with a columnar (zone 1) microstructure, many experiments involving both electrically conductive[117]-[121] and insulating films[122][123] have shown that the number density of voids and pores decreases dramatically with increasing ion energy and/or ion flux. Mattox and Kominiak[117] were among the first to demonstrate this effect. They found, in the case of sputter-deposited Ta films, that the microstructure changed from a pronounced columnar morphology to a more equiaxed structure and that the film density increased from ≈ 14.5 to 16.3 g cm-3 (bulk density = 16.6 g cm-3) as the negative substrate bias was increased from 0 to 500 V (see Fig. 13.30).

Figure 13.30. The density, as a function of negative substrate bias Vs, of 6-µm-thick Ta films deposited at 300oC by DC sputter deposition in Ar. (From Ref. 117.)


753

The effectiveness of ion irradiation in decreasing film porosity can be seen directly by comparing the TEM micrographs in Figs. 13.31a - d of polycrystalline Ti0.5Al0.5N films deposited by reactive magnetron sputtering on grounded stainless-steel substrates at ≈ 400oC (Ts/Tm ≈ 0.21).[121] The growth conditions were the same for all four films except for the imposition of applied substrate negative potentials Vs of 0, 75, 120, and 250 V, respectively. The ion-to-vapor flux ratio was ≈ 0.9 for the three biassputtered films. nv was estimated from planview TEM micrographs taken from regions approximately in the middle of the 4-µm-thick films. The void density nv decreased sharply for V s > 100 V until, for Vs > 120 V, no voids could be observed using under- and over-focus contrast. In addition to an increase in film density, Fig. 13.31 also shows that at low Ts, ion irradiation disrupts the columnar structure, as evidenced by the presence of Moire fringes, and increases the number density of defects such as dislocation loops.

Figure 13.31. Plan-view transmission electron micrographs of polycrystalline Ti0.5Al0.5N films deposited by reactive magnetron sputter deposition at Ts ≈400oC with applied negative biases of (a) Vs = 0, (b) Vs = 75 V, (c) Vs = 120 V, and (d) Vs = 250 V. (From Ref. 121.)

754


Monte Carlo (MC) and molecular dynamic (MD) simulations by Müller[58][62][114][115][124][125] of film growth under ion irradiation have also shown an increase in film density towards bulk values. For example, in lowtemperature growth simulations, the MC films had porous columnar structures in the absence of ion bombardment. However, if ion irradiation effects were taken into account (this was done using a modified version of the TRIM computer code), quite different growth morphologies were obtained. Ion incorporation, sputtering, and recoil implantation resulted in a film density that increased almost linearly with the ion-to-vapor flux ratio Ji /Jv. The calculations also predicted that there should be an optimum ion energy Ei* for densification. The optimization resulted from the fact that at Ei < Ei* the number of recoil events is small while at Ei > Ei * an increasing fraction of the ion energy is lost deeper in the lattice leaving vacancies which cannot be filled by arriving vapor species. In cases for which the computer simulations were compared to experimental data, good agreement was found. Examples include ZrO2 and CeO2 films deposited at ambient temperature on optically flat silica substrates by evaporating the metal in the presence of O2+ ions provided by a dual-grid Kaufmann ion source.[122][123] Figure 13.32 shows both calculated and experimental results for CeO2 film density as a function of ion energy.[125] The films were deposited with Ji /Jv = 1 and Ei* was found to be ≈ 200 eV. It should be noted, however, that the films were underdense even at Ei = Ei *. Two-dimensional MD calculations including up to 800 particles have been used to simulate Ar+-ion-assisted growth of Ni films at Ts = 0 K.[114] A typical result from a single 100 eV ion impact event during deposition is depicted in Fig. 13.33 which shows the time evolution of subsequent atomic rearrangements. Figures 13.33b, 13.33c, and 13.33d correspond to times of 0.3, 1.1, and 10.9 picoseconds (ps) after ion impact. The incident ion transfers kinetic energy to a few surface atoms which in turn transfer energy to other atoms as the collision sequence develops. The initial violent collision events occur over times of the order of a few hundred femtoseconds while the relaxation process requires several picoseconds. Figure 13.33 illustrates a collision sequence leading to the disappearance of a protruding ledge overshadowing an incipient void. The central void also decreased in size resulting in a denser structure. The forward sputtering events shown in Fig. 13.33 can be viewed as an enhancement in the apparent rate of adatom surface diffusion. Simulations such as this are not intended to correspond to a particular physical reality. However, they do provide a means for visualizing the general effects of ion irradiation on the atomic level.


755

Figure 13.32. Experimental and theoretical values of the density of CeO2 films deposited at ambient temperature by simultaneous evaporation of Ce and ion-beam acceleration of O2+ as a function of the ion energy Ei for an ion-to-vapor flux ratio Ji/ Jv of 1. The bulk density of CeO2 is 8.1 g/cm3. (From Ref. 125.)

While ion irradiation is useful for increasing the density and modifying the morphology of films deposited at low temperatures, other irradiationinduced effects occur simultaneously. For example, as the ion energy and ion flux are increased, atomic displacements produced in the collision cascades result in an increasing number of residual interstitials and vacancies. These point defects can, in turn, lead to an increased density of extended defects such as dislocation loops. Huang et al.[126] have studied the effect of Ar+ ion bombardment during the growth of Ag films at room temperature using a UHV dual-ion-beam apparatus. They found that the void density decreased with increasing ion energy in agreement with the results presented in the previous paragraphs. However, they also showed that the use of average irradiation energy densities ranging from thermal (obtained by evaporation) to 190 eV per deposited metal atom yielded a decrease in the grain size from 42 to 14.5 nm while the dislocation number density nd

756


increased from 0.7 x 1011 to 13.2 x 1011 cm-2 (see Fig. 13.34). In addition, the degree of (111) preferred orientation decreased while the plane stress reversed from 0.6 x 108 N m-2 tensile to - 4.5x108 N m-2 compressive for larger than 42 eV. Increased defect concentrations and reduced grain sizes have also been observed in a number of other polycrystalline thin film systems grown under low-energy ion irradiation.[121][127]-[130]

Figure 13.33. Molecular-dynamic simulation of the structure of a Ni film deposited at 0 K at various times t after bombardment by a 100 eV Ar+ ion. Atomic displacements (not trajectories) are indicated by straight line segments with origins at the zero-time positions of the relocated atoms. (From Ref. 114.)


757

Figure 13.34. The average grain size and dislocation number density nd in Ag films deposited at room temperature as a function of the average energy per deposited atom. (Plotted from data given in Ref. 126.)

At elevated growth temperatures, low-energy ion irradiation can, in contrast to the above low temperature results, have the opposite effect and actually reduce residual defect densities in as-deposited films. Direct evidence has recently been published by Hultman et al.[131][132] who used TEM analyses to investigate the dislocation structure in epitaxial TiN films grown on MgO(100) substrates at Ts between 550 and 850 oC by reactive magnetron sputtering in pure nitrogen discharges. (The minimum epitaxial temperature in this case is 525 - 550oC.) The primary defects in the TiN films were dislocation loops on (111) planes. The dislocation number density nd in epitaxial TiN layers was found to decrease with increasing Ts, due to higher adatom surface mobilities, for a given negative substrate bias Vs. For example, in films grown with V s = 0,

758


n d continuously decreased from ≈ 1013 cm-2 at Ts = 550oC to ≈ 1012 cm -2 at Ts = 850o C. (Note that 850oC is still a relatively low temperature for TiN and corresponds to Ts/Tm = 0.35.) However, nd decreased much more rapidly with increasing Vs at constant Ts until a minimum defect density was obtained at a specific voltage Vs*. For Vs > Vs*, nd increased rapidly and eventually the films became polycrystalline. Ji/Jv in these experiments ranged from ≈ 1 to 1.4 and the energy per incident accelerated N atom was Vs/2. Vs* was found to be ≈ 300 V at Ts = 650oC for which nd ≈ 2 x 1010 cm2 compared to ≈ 5 x 1011 cm -2 at V = 0 and T = 650o C. The minimum in n (V ) s s d s became broader and Vs* increased slightly with increasing T s. Films grown at Ts > 750 oC and Vs = Vs* were essentially free of dislocation loops. This can be seen in Fig. 13.35 which shows TEM micrographs from films grown at 850oC with Vs = 0 and at 800oC with Vs = Vs* = 400 V. XTEM micrographs of multilayer films in which sequential layers were grown with different values of Vs showed that ion bombardment effects were reversible. [132] For example, changing Vs from 0 to Vs* to 0 resulted in nd abruptly decreasing and then abruptly increasing again.

Figure 13.35. Plan-view transmission electron micrographs of epitaxial TiN films grown by reactive magnetron sputtering on MgO(100) substrates at (a) Ts = 850oC with a negative substrate bias Vs = 0 and (b) Ts = 800oC with Vs = 400 V. (From Ref. 131.)


759

Ion irradiation in the above experiments played at least two major roles. For the lower biases, the primary effect was to enhance adatom mobilities thereby accelerating the rate at which defects (both growth-related and ionirradiation-induced) were annealed out during deposition. At higher bias voltages (Vs > V s*), the increased projected range of the impinging ions resulted in a larger fraction of the irradiation-induced defects being trapped in the growing film. Eventually, nd became high enough that renucleation occurred during growth and polycrystalline films were obtained.

ACKNOWLEDGEMENTS The author gratefully acknowledges the financial assistance of the Joint Services Electronics Program, the Materials Science Division of the Department of Energy, and the Semiconductor Research Corporation during the course of this work. He also appreciates, and has gained much, from several years of scientific collaboration and close personal friendship with Prof. JanEric Sundgren (Physics Dept., Linköing University, Sweden) who proofread this manuscript.

760


REFERENCES 1.

Greene, J. E. and Barnett, S. A., J. Vac. Sci. Technol., 21:285 (1982)

2.

Greene, J. E., Motooka, T., Sundgren, J. E., Lubben, D., Gorbatkin, S., and Barnett, S. A., J. Nucl. Instr. Methods, B27:226 (1987)

3.

Greene, J. E., Sol. St. Technol., 30:115 (April 1987)

4.

Greene, J. E., Barnett, S. A., Sundgren, J. E., and Rockett, A., in IonAssisted Film Growth, (T. Itoh, ed.) p. 101, Elsevier, Amsterdam (1988)

5.

Venables, J. A., Spiller, G. D. T., and Hanbcken, M., Rept. Prog. Physics , 47:399 (1984)

6.

Knall, J., Sundgren, J. E., Hansson, G. V., and Greene, J. E.,Surf. Sci., 166:512 (1986)

7.

Hirth, J. P. and Moazed, K. L., Thin Film Physics 4, (G. Hass and R. E. Thun, eds.) p. 97., Academic Press, New York (1967)

8.

Neugebauer, C. A., in Handbook of Thin Film Physics (L. I. Maissel and R. Glang, eds.) Ch. 8., McGraw Hill, New York (1970)

9.

Wagner, R. S. and Voorhoeve, R. J. H.,J. Appl. Phys., 43:3948 (1971)

10.

Lewis, B. and Anderson, J. C., Nucleation and Growth of Thin Films, Academic Press, New York (1978)

11.

Stoyanov, S. and Kaschiev, D., inCurrent Topics in Materials Science, Vol. 7, (E. Kaldis, ed.) North Holland, Amsterdam (1981)

11a. Zinmeister, G., Vacuum, 16:529 (1966); Thin Solid Films, 2:497 (1968), 4:363 (1969); Kristall Technik 5:207 (1970); and Thin Solid Films 7:51 (1971) 12.

Pocza, J. F., Barna, A., and Barna, P. B., J. Vac. Sci. Technol., 6:472 (1969)

13.

Chopra, K. L., Thin Film Phenomena, p. 160, McGraw Hill, New York (1969)

14.

Donohoe, A. J. and Robins, J. L., J. Cryst. Growth 17:70 (1972)

15.

Pashley, D. W., Stowell, M. J., Jacobs, M. H., and Law, J. T.,Phil. Mag. 10:127 (1964)

16.

Williamson, W. J., Solid State Electron. 9:213 (1966)

17.

Greene, J. E. and Wickersham, C. E., J. Appl. Phys. 47:3630 (1976)

18.

Bassett, G. A., Menter, J. W., and Pashley, D. W., in Structure and Properties of Thin Films, (C. A. Neugebauer, J. B. Newkirk and D. A. Vermilyea, eds.) p. 11, John Wiley and Sons, New York (1969)


761

19.

Jacobs, M. H., Pashley, D. W., and Stowell, M. J., Phil. Mag. 13:129 (1966)

20.

Barna, P. B., Proc. IX Internet. Vac. Congr., p. 382, Madrid (1983)

21.

Joyce, B. A., Neave, J. H., and Watts, B. E., Surf. Sci. 15:1 (1969)

22.

Henderson, R. C., Polito, W. J., and Simpson, J., Appl. Phys. Letters 16:15 (1970)

23.

Henderson, R. C. and Helm, R. F., Surf. Sci. 30:310 (1972)

24.

Abbink, H. C., Broudy, R. M., and McCarthy, G. P., J. Appl. Phys. 39:4673 (1968)

25.

Aleksandrov, L. N. and Lovyagin, R. N., Thin Solid Films 20:1 (1974)

26.

Van Hove, J. M., Pukite, P. R., and Cohen, P. I., J. Vac. Sci. Technol. B3:563 (1985)

27.

Van Hove, J. M. and Cohen, P. I., J. Cryst. Growth 81:67 (1987)

28.

Reflection High-Energy Electron Diffraction and Reflection Electron Imaging of Surfaces, NATO AISI Series B: Physics, (P. K. Larson and P. J. Dobson, eds.) Vol. 188, Plenum Press, New York (1988)

29.

Rockett, A., J. Vac. Sci. Technol. B6:763 (1988)

30.

Rockett, A., SPIE Proceedings, 944:63 (1988)

31.

Shtrikman, H., Heiblum, H., Seo, K., Galbi, D. E., and Osterling, L., J. Vac. Sci. Technol. B6:670 (1988)

32.

Price, G. L. and Venables, J. A., Surf. Sci. 49:264 (1975)

33.

Wagner, R. S. and Voorhoeve, R. J. H., J. Appl. Phys. 42:3948 (1971)

34.

Voorhoeve, R. H. J. and Wagner, R. S., Met. Trans. 2:3421 (1971)

35.

Henrion, J. and Rhead, G. E., Surf. Sci. 29:20 (1972)

36.

Delamare, F. and Rhead, G. E., Surf. Sci. 35:172 (1973)

37.

Yagi, Y., Takayanagi, K., Kobayashi, K., and Honjo, G., J. Cryst. Growth 9:84 (1971); Honjo, G. and Yagi, K., in Current Topics in Materials Science (E. Kaldis, ed.) Vol. 6, North Holland Publishing Co., Amsterdam (1980)

38.

Eltoukhy, A. H. and Greene, J. E., J. Appl. Phys. 50:505 (1979)

39.

Matthews, J. W. and Blakeslee, A. E., J. Cryst. Growth 27:118 (1974)

40.

Bean, J. C., Sheng, T. T., Feldman, L. C., Fiory, A. T., and Lynch, R. T., Appl. Phys. Letters 44:102 (1984)

40a. People, R. and Bean, J. C., Appl. Phys. Letters 47:322 (1985) and 49:229 (1986) 40b. Tsao, J. Y., Dodson, B. W., Picraux, S. T., and Cornelison, D. M., Phys. Rev. Letters, 59:2455 (1987)

762


41.

Horng, C. T. and Vook, R. W., J. Vac. Sci. Technol. 11:140 (1974)

42.

Soria, F. and Poppa, H., J. Vac. Sci. Technol. 17:449 (1980).

43.

Hopkins, B. J. and Watts, G. D., J. Phys. C: Solid State 7:4259 (1974)

44.

Steinhage, P. W. and Mayer, M., Thin Solid Films 28:131 (1975)

45.

Van Loenen, E. J., Iwami, M., Tromp, R. M., and van der Veen, J. F., Surf. Sci. 137:1 (1984)

46.

Honzay, F., Guille, C., Moison, J. M., Henoc, P., and Barthe, F., J. Cryst. Growth, 81:67 (1987)

47.

Stranski, I. N. and Krastanov, L., Acad. Wiss. Math-Nat. KIIIb 146:797 (1938)

48.

Venables, J. A., in Current Topics in Materials Science (E. Kaldis and H. J. Scheel, eds.) Vol. 2, p. 165, North Holland, Amsterdam (1977)

49.

Price, G. L., Surf. Sci. 46:697 (1974)

50.

Knall, J., Barnett, S. A. and Sundgren, J. E., Surf. Sci., in press

51.

Rich, D. H., Samsavar, A., Miller, T., Lin, H. F., Chiang, T. C., Sundgren, J. E., and Greene, J. E.,Phys. Rev. Letters, 58:579 (1987)

52.

Ehrlich, G. and Stolt, K., Ann. Rev. Phys. Chem. 31:603 (1980)

53.

Thornton, J. A., J. Vac. Sci. Technol. 11:666 (1974)

54.

Dirks, A. G. and Leamy, H. J., Thin Solid Films 47:219 (1977)

55.

Leamy, H. J., Gilmer, G. H., and Dirks, A. G., in Current Topics in Materials Science (E. Kaldis, ed.) 6:309 North Holland Publishing Co., Amsterdam (1980)

56.

Henderson, D., Brodsky, M. H., and Chauderi, P., Appl. Phys. Letters, 25:641 (1975)

57.

Kim, S. and Henderson, D. J., Thin Solid Films, 47:155 (1977)

58.

Müller, K. H., J. Appl. Physics, 58:2573 (1985)

59.

Holmwood, R. A. and Glang, R., J. Electrochem. Soc. 112:831 (1965)

60.

Klokholm, E. and Berry, B. S., J. Electrochem. Soc. 115:823 (1968)

61.

Huang, T. C., Lim, G., Parmiagiani, F., and Kay, E., J. Vac. Sci. Technol. A3:2161 (1985)

62.

Müller, K. H., J. Appl. Phys. 62:1796 (1987)

63.

Srolovitz, D. J., J. Vac. Sci. Technol. A4:2925 (1986)

64.

Srolovitz, D. J., Mazor, A., and Bukiet, B. G., J. Vac. Sci. Technol. A6:2371 (1988)

65.

Movchan, B. A. and Demchisin, A. V., Phys. Met. Metallogr. 28:83 (1969)


763

66.

Hentzell, H. T. G., Grovenor, C. R. M., and Smith, D. A., J. Vac. Sci. Technol. A2:218 (1984)

67.

Thornton, J. A., Ann. Rev. Mater. Sci. 7:239 (1977)

68.

Guenther, K. H., Thin Solid Films, 77:239 (1981)

69.

Guenther, K. H., Appl. Optics, 20:1034 (1981)

70.

Messier, R., Krishnaswamy, S. V., Gilbert, L. R., and Swab, P., J. Appl. Phys. 51:1611 (1980)

71.

Swab, P., Krishnaswamy, S. V., and Messier, R., J. Vac. Sci. Technol. 17:362 (1980)

72.

Ross, R. C. and Messier, R., J. Appl. Phys. 52:5329 (1981)

73.

Messier, R., Giri, A. P., and Roy, R. A., J. Vac. Sci. Technol. A2:500 (1984)

74.

Thornton, J. A., J. Vac. Sci. Technol. 12:830 (1975)

75.

Grovenor, C. R. M., Hentzell, H. T. G., and Smith, D. A., Acta. Metall. 32:773 (1984)

76.

Messier, R., J. Vac. Sci. Technol. A4:490 (1986)

77.

Thornton, J. A., J. Vac. Sci. Technol. A4:3059 (1986)

78.

Venables, J. A. and Price, G. L., in Epitaxial Growth, Part B, (J. W. Matthews, ed.) p. 381, Academic Press, New York (1975)

79.

Leamy, H. J. and Dirks, A. G., J. Appl. Phys. 49:3430 (1978)

80.

Helmersson, U., Sundgren, J. E., and Greene, J. E., J. Vac. Sci. Technol. A4:500 (1986)

81.

See, for example, Winters, H. F., J. Chem. Phys. 44:1472 (1966)

82.

Winters, H. F., Ramondi, D. L., and Horne, D. E., J. Appl. Phys. 40:2996 (1969)

83.

Tarng, M. L. and Wehner, G. K., J. Appl. Phys. 42:2449 (1971)

84.

Cuomo, J. J. and Gambino, R. J., J. Vac. Sci. Technol. 12:79 (1975)

85.

Harper, J. M. E. and Gambino, R. J., J. Vac. Sci. Technol. A4:448 (1986)

86.

Zilko, J. L. and Greene, J. E., J. Appl. Phys. 51:1549 (1980)

87.

Berg, S., Nender, C., and Gelin, B., J. Vac. Sci. Technol. A4:448 (1986)

88.

Nender, C., Berg, S., Gelin, B., and Stridh, B., J. Vac. Sci. Technol. A5:1703 (1987)

89.

Kondo, N. and Kawashima, M., GaAs and Related Compounds 1985, Inst. Phys. Conf. Series 79, p. 97 (1985)

764 90.

Deposition Technologies for Films and Coatings Greene, J. E., J. Vac. Sci. Technol. B1:229 (1983)

91. Romano, L. T., Robertson, I. M., Greene, J. E., and Sundgren, J. E. Phys. Rev. B36:7523 (1987) 92. Shah, S. I., Greene, J. E., Abels, L. L., and Raccah, P. M., J. Cryst. Growth, 91:71 (1988) 93. Fons, P., Hirashita, N., Markert, L. C., Kim, Y. W., Greene, J. E., Ni, W. X., Knall, J., Hansson, G. V., and Sundgren, J. E.,Appl. Phys. Letters, 53:1732 (1988) 94. Hasan, M. A., Knall, J., Barnett, S. A., Sundgren, J. E., Markert, L. C., Rockett, A., and Greene, J. E., J. Appl. Phys. 65:172 (1989) 95. Noel, J. P., Hirashita, N., Markert, L. C., Kim, Y. W., Greene, J. E., Knall, J., Ni, W. X., Hasan, M. A., and Sundgren, J. E., J. Appl. Phys. 65:1189 (1989) 96. Wehner, G. K. and Anderson, G. S., in Handbook of Thin Film Technology, (L. I. Massel and R. Glang, eds.) Ch. 3, McGraw-Hill, New York (1970) 97. Donahue, T. J. and Reif, R.,Semiconductor International, 142, (August, 1985) 98. Krikorian, E. and Sneed, R. J., Astrophys. Space Sci. 65:129 (1979) 99. Chapman, B. N. and Campbell, D. S., J. Phys. C2:200 (1969) 100. Lane, G. E. and Anderson, J. C., Thin Solid Films, 26:5 (1975) 101. Lane, G. E. and Anderson, J. C., Thin Solid Films, 57:277 (1979) 102. Harsdorff, M. and Jark, W., Thin Solid Films, 128:79 (1985) 103. Barnett, S. A., Winters, H. F., and Greene, J. E., Surf. Sci. 181:596 (1987) 104. Marinov, M., Thin Solid Films, 46:267 (1977) 105. Hasan, M. A., Barnett, S. A., Sundgren, J. E., and Greene, J. E.,J. Vac. Sci. Technol. A5:1883 (1987) 106. Hasan, M. A., Knall, J., Barnett, S. A., Rockett, A., Sundgren, J. E., and Greene, J. E., J. Vac. Sci. Technol. B5:1332 (1987) 107. Dodson, B. W., J. Vac. Sci. Technol. B5:1393 (1987) 108. Narusawa, T., Shimizu, S., and Komiya, S., J. Vac. Sci. Technol. 16:366 (1979) 109. Thomas, G. E., Beckers, L. J., Vrakking, J. J., and de Koning, B. R., J. Cryst. Growth, 56:257 (1982) 110. Zalm, P. C. and Beckers, L. J., Appl. Phys. Lett. 41:167 (1982)


765

111. Herbots, N., Noggle, T. S., Appleton, B. R., and Zhur, R. A., J. Vac. Sci. Technol., in press 112. Yagi, K., Tamura, S., and Tokuyama, T., Jpn. J. Appl. Phys. 16:245 (1977) 113. Tokuyama, T., Yagi, K., Miyaki, K., Tamura, M., Natsuaki, N., and Tachi, S., Nucl. Instr. Meth. 182/183:241 (1981) 114. Müller, K. H., Phys. Rev. 35:7906 (1987) 115. Müller, K. H., Surf. Sci. Lett. 184:L375 (1987) 116. Tsao, J. Y., Chason, E., Horn, K. M., Brice, D. K., and Picraux, S. T., Nucl. Instr. Meth., in press 117. Mattox, D. M. and Kominiak, G. J., J. Vac. Sci. Technol. 9:528 (1972) 118. Mizzoguchi, T. and Cargill, G. S., III, J. Appl. Phys. 50:3570 (1979) 119. Mnz, W. D. and Hofmann, D., Metalloberflche, 37:279 (1983) 120. Parmiagiani, F., Kay, E., Huang, T. C., Perrin, J., Jurich, M., and Swalin, J. D., Phys. Rev. B33:879 (1986) 121. Hakanssan, G., Sundgren, J. E., McIntyre, D., Greene, J. E., and Mnz, W. D., Thin Solid Films, 153:55 (1987) 122. Martin, P. J., Netterfield, R. P., and Sainty, W. G.,J. Appl. Phys. 55:235 (1984) 123. Netterfield, R. P., Sainty, W. G., Martin, P. J., and Sie,S. H., Appl. Opt. 24:2267 (1985) 124.

Müller, K. H., J. Appl. Phys. 58:2803 (1986)

125. Müller, K. H., Appl. Phys. A40:209 (1986) 126. Huang, T. C., Lim, G., Parmiagiani, F., and Kay, E., J. Vac. Sci. Technol. A3:2161 (1985) 127. Igasaki, Y. and Mitsuhashi, H., Thin Solid Films 70:17 (1980) 128. Poitevin, J. M., Lemperiere, G., and Tardy, J., Thin Solid Films 97:69 (1982) 129. Johansson, B. O., Sundgren, J. E., and Helmersson, U.,J. Appl. Phys. 58:3112 (1985) 130. Kay, E., Parmigiani, F., and Parrish, W., J. Vac. Sci. Technol. A5:44 (1987) 131. Hultman, L., Helmersson, U., Barnett, S. A., Sundgren, J. E., and Greene, J. E., J. Appl. Phys. 61:552 (1987) 132. Hultman, L., Barnett, S. A., Sundgren, J. E., and Greene, J. E., J. Cryst. Growth 92:639 (1988)

766


14 Metallurgical Applications Rointan F. Bunshah

1.0 INTRODUCTION Corrosion and wear are often referred to as the twin demons of materials degradation. The loss of material annually due to these causes is a significant fraction of their total production each year. Often, materials degradation results from both corrosion and wear phenomena acting simultaneously, e.g., corrosive wear, corrosion erosion, etc. The following sections discuss each of these phenomena and the role of coatings in decreasing the resultant materials degradation.

2.0 CORROSION Corrosion is the destructive attack of a metal or alloy by chemical reaction with its environment which can be aqueous solutions, molten salts, molten metals, or corrosive gases at high temperatures. In some cases, chemical attack accompanies physical degradation which is described by the termscorrosion erosion, corrosive wear, fretting corrosion, etc. Polymers and ceramics at high temperatures are also subject to corrosion. The role of coatings in protection of polymers and ceramics at high temperature has not been given much attention and hence will not be considered here. For the sake of convenience in the discussion of this multi-faceted topic, the subject is divided into two broad categories:

766

Metallurgical Applications

767

• Galvanic corrosion in aqueous solutions, molten salts, etc. • High temperature corrosion: oxidation, sulphidation, etc.

3.0 GALVANIC CORROSION Types of corrosion damage: 1. Uniform attack. Examples are rusting of iron, high temperature oxidation of metals, “fogging” of nickel etc. Units are milligrams per square/deciliter per day. 2. Pitting Corrosion. Localized attack usually confined to a small area which is anodic to the rest of the metal. 3. Fretting Corrosion. It is due to the slight relative motion of two materials in contact usually leading to a series of pits at the metal interface filled with metal oxide. 4. Corrosion-erosion. It results from formation and collapse of vapor bubbles at a dynamic metal-liquid interface causing a sequence of pits or fissures. 5. Dezincification and Parting. Dezincification occurs in zinc alloys such as brass in which zinc corrodes preferentially leaving a porous residue of copper and corrosion products. Parting is similar to dezincification in which one or more reactive components of an alloy corrode preferentially leaving a porous residue that may retain the original shape of the alloy. It is usually restricted to noble metal alloys. 6. Intergranular Corrosion. This is a localized type of attack at the grain boundaries of a metal resulting in loss of strength and ductility. Grain boundary material of limited area, acting as anode, is in contact with large areas of grain acting as cathodes. Attack is often rapid and deeply penetrating. Examples are improperly heat-treated austenitic stainless steel, Al - 4% Cu alloys, etc. 7. Cracking. Metal cracks when subjected to repeated or alternative tensile stresses in a corrosive environment leading to corrosion-fatigue. If the metal which is subjected to constant tensile stress and exposed to a specific corrosive environment cracks immediately or after a time delay, the failure is called stress-corrosion cracking.

768


3.1 Galvanic Cells A combination of two electrical conductors (electrodes) immersed in an electrolyte is called galvanic cell in honor of Luigi Galvani, a physician in Bologna, Italy, who published his studies of electrochemical action in 1791. A galvanic cell converts chemical energy into electrical energy. On shortcircuiting such a cell (attaching a low-resistance wire to each electrode), positive current flows through the metallic path from positive electrode to negative electrode. The direction of current flow follows an arbitrary convention established before anything was known about the nature of electricity, and is followed today despite present-day knowledge that only negative carriers of electrons move in a metal. Electrons, of course, go from negative to positive pole, opposite to the imaginary flow of positive carriers. Within the electrolyte, current is carried by both negative and positive carriers known as ions (electrically charged atoms or groups of atoms). The current carried by each ion depends on its respective mobility and electric charge. The total of positive and negative current in the electrolyte of a cell is always exactly equivalent to the total current carried in the metallic path by electrons alone. The electrode at whichchemical reductionoccurs (or + current enters the electrode from the electrolyte) is called the cathode. Examples of cathodic reactions are: H+ → ½H2 - eCu++ → Cu - 2eFe+3 → Fe ++ - eall of which represent reduction in the chemical sense. The electrode at whichchemical oxidationoccurs (or + electricity leaves the electrode and enters the electrolyte) is call theanode. Examples of anodic reaction are: Zn → Zn++ + 2eAl → Al +3 + 3eFe++ → Fe+3 + eThese represent oxidation in the chemical sense. For metals, it is at the anode that corrosion usually occurs.


769

In galvanic cells, the cathode is the positive pole and the anode is the negative pole. There are three main types of cells that take part in corrosion reactions. Dissimilar Electrode Cells. These are illustrated by the dry cell discussed earlier. A metal containing electrically conducting impurities on the surface as a separate phase, or a copper pipe connected to an iron pipe, or a bronze propeller in contact with the steel hull of a ship are examples of this type of corrosion cell. These cells also include cold-worked metal in contact with the same metal annealed, or grain boundary metal in contact with grains, or a single metal crystal of definite orientation in contact with another crystal of differing orientation Concentration Cells. These are cells having two identical electrodes each in contact with a solution of differing composition. There are two kinds of such cells. The first is called a salt concentration cell. For example, if one copper electrode is exposed to a concentrated copper sulfate solution, and another to a dilute copper sulfate solution, on short-circulating such a cell, copper dissolves from the electrode in contact with the dilute solution (anode) and plates out on the other electrode (cathode). Both reactions tend to bring the solutions to the came concentration. The second kind of concentration cell, which in practice is the more important, is called a differential aeration cell. This may include two iron electrodes in dilute NaCl solution, the electrolyte around one electrode being thoroughly aerated (cathode), and the other deaerated (anode), brought about, for example, by bubbling through nitrogen. The difference in oxygen concentration produces a potential difference, and causes current to flow. This type of cell accounts for pronounced damage at crevices such as are formed at the interface of two coupled pipes, or at threaded connections, because O2 concentration is lower within the crevice or at the threads than elsewhere. It also accounts for pitting damage under rust or at the water line (water-air interface). Less oxygen reaches the metal that is covered by rust or other insoluble reaction products than at other portions where the permeable coating is thinner or absent. Differential aeration cells also usually initiate pits in the stainless steels, aluminum, nickel, and other so-called passive metals when they are exposed to aqueous environments, such as water. Differential Temperature Cells. Components of these cells are electrodes of the same metal, each of which is at a different temperature, immersed in an electrolyte of the same initial composition. Less is known about the practical importance and fundamental theory of differential

770


temperature cells than for the cells previously described. They occur in heat exchangers, boilers, immersion heaters, and similar equipment. In CuSO4 solution the copper electrode at the higher temperature is cathode and the copper electrode at the lower temperature is anode. On shortcircuiting the cell, copper deposits on the hot electrode and dissolves from the cold electrode. Lead acts similarly, but for silver the polarity is reversed. For iron immersed in dilute aerated NaCl solutions, the hot electrode is anodic to colder metal of the same composition, but after a matter of hours, depending on aeration, stirring rate, and whether the two metals are shortcircuited or not, the polarity may reverse. In practice, cells responsible for corrosion may be a combination of these three types. 4.0 EMF AND GALVANIC SERIES The EMF series shown in Table 14.1 is determined by the equilibrium potential of a metal in contact with its ions at a concentration equal to unit activity of the two metals composing a cell. The reactivity of the metals shown in Table 14.1 decreases from Li at the top (the most reactive) to Au at the bottom (very noble). Thus if two metals are in contact, the one with the higher oxidation potential will dissolve with respect to the other. For example, Zn is more reactive (sacrificial) with respect to Fe—but Fe is more reactive (sacrificial) with respect to Sn. The anode is the more active metal in the EMF series provided that the ion activities in equilibrium are both unity. In some cases this corresponds to impossible concentrations of metal ions because of the restricted solubility of metal salts. Hence, the EMF series has limited utility in predicting which metal is cathodic to another. 5.0 COATINGS FOR GALVANIC CORROSION There are a large number of different types of protective coatings. They can be classified as: 1. Anodic coatings - coating dissolves, e.g., Zn, Al vs. Fe 2. Cathodic coatings - substrate attacked, e.g., Sn vs. Fe (defect free coating is necessary in this case) 3. Inert coatings 4. Inhibitive coatings

Metallurgical Applications Table 14.1. Electromotive Force Series

771

772


Let us consider the case of a coated metal article exposed to an electrolyte with discontinuities in the coating. There are several possibilities. An electric current may flow from the coating through the electrolyte to the base metal. The coating is then anodic to the base metal. If the current density at the exposed area of the base metal is of the correct magnitude, corrosion of the base metal is prevented. Thus anodic coatings tend to prevent corrosion of exposed areas of the base metal by sending electric current to them through any contacting layer or electrolyte. In contrast, cathodic coatings stimulate corrosion of exposed areas of the base metal. Metallic coatings show the most pronounced anodic or cathodic behavior. Non-metallic coatings especially oxides and sulfides act as cathodic coatings. The same metallic coating on the same base metal can behave as an anodic coating under one set of exposure conditions, and as an inhibitive or inert coating under different conditions. For example, tin is cathodic to exposed areas of steel base in sea water, natural water, or even to many food products in the presence of air. However, when exposed to nearly air-free food products, tin is definitely anodic to steel. Inorganic coatings are sometimes inert, sometimes cathodic, and sometimes inhibitive. Organic coatings are generally inert or inhibitive.

6.0 METHODS OF DEPOSITION OF METALLIC COATINGS 1. Hot-dip process for zinc, tin, aluminum and lead. 2. Metal spraying for most common metals using a wire spray gun. 3. Metal cementation for zinc, chromium, aluminum and silicon in which the protective metal is alloyed with the surface of the steel. Other names for this process are: SHERADIZING, CHROMIZING, CALORIZING, IHRIGIZING (for Sc), CORRONIZING. 4. Metal cladding: copper clad onto steel by dipping, and aluminum cladding by hot rolling a pack. 5. Fusion welding. 6. Electroplating. 7. Sputter deposition or evaporation deposition.


773

7.0 EXAMPLES OF CORROSION-RESISTANT COATINGS 7.1 Preamble At the outset, an important aspect of such coatings is noted. These coatings can be classified into sacrificial or non-sacrificial coatings. Sacrificial coatings are corroded in preference to the base metal which is protected. In this case, the microstructure of the coating does not have to be dense. On the other hand, for a non-sacrificial coating, the microstructure has to be fully dense to isolate the underlying metal from the corrosive fluid (liquid or gas). Hence microstructure control becomes very important. Nowak[1] has discussed the subject in more detail with emphasis on composition and microstructural factors. Egert[2] has presented a theory on corrosion of substrates with protective metal coatings. Assuming no galvanic interaction between the coating and the substrate, corrosion occurs at defects in the coating and is proportional to the number of defects per unit area. Multilayer coatings and amorphous alloy coatings appear to be more effective in general than single layer coatings. In the latter case, the absence of grain boundaries which are often corrosive channels is an important microstructural factor. Zinc coatings, produced by a dip coating process, have been used to protect steel products for 150 years. A modern development is the one-sided zinc coatings for the automobile industry by PVD techniques with the ability to make paint adhere much better than on a galvanized surface. Zinc coatings are outstanding in their ability to protect steel by galvanic action where the coating is mechanically damaged such as at the sheared edges of a steel sheet. On the other hand, galvanized coatings do not last as long in sulfatebearing industrial atmospheres. Zinc oxide coatings are not protective as gauged from their linear oxidation rate. Conversely, metals such as aluminum and chromium form protective oxide films and are more corrosion resistant. Aluminum likewise, does not confer galvanic protection to a steel base. As a result, aluminum coated steel shows unsightly rust staining or rust spots at points of mechanical damage such as sheared edges or scratches through the coating. An early solution in 1961 was to produce a Zn-55%Al alloy coating by the hot-dip process. This coating combines the best features of galvanizing and aluminizing. Horton[3] has discussed the physical metallurgy of the Al-Zn alloy coatings. In the middle of the 1960 - 1970 decade, several steel companies studied the possibility of depositing very thin aluminum and/or zinc coatings on steel sheet in air-to-air coating lines by high rate electron-beam evaporation of Al[4]

774


or zinc by resistance evaporation from a graphite boat.[5] For thin (sub-micron) aluminum coatings, the degree of pitting in the coating after corrosion testing was influenced by the deposition rate, and the surface preparation of the steel strip. About a decade later, in the former state of East Germany, a peculiar economic circumstance, i.e., shortage of hard currency, forced the investigation of substitution of Al for Sn in steel sheet for food and beverage container application. The development of an air-to-air coating line used in tandem with steel strip rolling in a commercial steel plant was successfully implemented in 1981 at the BEJO Steel Company in Bad Salzungen.[6] Typical deposition conditions were as follows: for Al deposition, the film thickness was 0.5 - 3 µm on a steel strip 400 mm wide by 0.2 mm thickness at a substrate temperature of 220 - 300°C, a deposition rate of 20 µm/second at a strip speed of 2 m/sec using 50 kg/hour of aluminum from a continuously-fed electron beam evaporation source. In subsequent investigations, mechanical activation of the steel strip by wire brushing prior to deposition was found to enhance the coating performance. Since the unification of Germany, this coating line is now being used to deposit copper and chromium onto steel. Zn deposition from a resistance-heated graphite source produced a 15 µm thickness coating on steel strip at a substrate temperature of 150 - 300°C at a line speed of 100 ft/min with Zn usage of 40 - 100 lb/hour.[6] According to Schiller et al.,[6] there are fifteen plants world-wide for coating metal strips continuously by PVD techniques, of which eight are in Japan. They use thermionic electron beam, hollow cathode arc, resistance-heated sources and magnetron-sputtering sources to produce the metal vapor. Another application involves the deposition of Al onto steel and titanium alloy fasteners used in air-frame construction. The problem to be solved was galvanic corrosion caused by dissimilar metal contact, i.e., steel or titanium of the fastener and aluminum in the air-frame. The process used is “ion-plating” of aluminum using a resistance-heated source in a partial pressure of 10 millitorr of Ar gas with the parts biased to a -2 kV potential.[7] The parts are hung on racks if they are large, or tumbled in a barrel for small parts such as fasteners. The Al coating thickness is ~1 µm and 100 lbs of fasteners can be coated per hour. Other examples of corrosion resistance imparted by coatings are as follows: 1. Antler in 1977[8] discusses corrosion resistance conferred by intermetallic phases, particularly those in the nickel-tin system.

Metallurgical Applications 2. Nowak and co-workers[9][10] have studied thin films of Al-Cu, Al, Mg and Al1-xZn x ion-plated onto steel substrates for enhancement of corrosion resistance. These thin, microcrystalline films can surpass bulk material in corrosion resistance. The ion-plated Al-Zn alloys have markedly superior corrosion resistance compared to the commercial hot dipped coatings Galvlume (55Al-Zn) and Galfan (5Al-Zn). 3. Microcrystalline/amorphous iron alloy films (Fe-180Cr-8Ni) with additions of about one atomic percent Al or Si were deposited onto 304 stainless steel or Erbnite substrates. Substantial improvement in corrosion resistance was produced by modification of the microstructure with further improvements through chemical and structural mechanisms. The results indicate that sufficient alloy additions to stabilize the material to an entirely amorphous state would produce even greater improvements.[12] 4. Type 304 stainless steel coated by CVD techniques with TiN produced substantial improvements in both corrosion and whirledsand abrasion resistance in sea water.[13] 5. Wialla et al.[14] deposited TiN and ZrN coatings onto high speed steel and stainless steel by reactive triode ion plating. The results showed improved corrosion resistance. 6. Beveskog et al.[15] used potentiodymanic measurements to demonstrate the excellent corrosion resistance imparted by a TiC coating deposited by the ARE process onto quartz substrates. 7. Bearings of steel in a stagnant lubricating oil environment with Cl ions present in the oil show marked corrosion at opposing mating surfaces. Such a circumstance occurs in aircraft engines used in a marine environment which are inactive over a period of time, a circumstance that can occur with military aircraft. It was found that ion-implantation of Mo and Cr solved the corrosion problem. Overlay coatings of Cr, Mo, TiC, TiN also solved the problem at a much lower processing cost.[16] 8. Type 52100 bearing steel and type 304 stainless steel were overcoated with TiN and ZrN by cathodic arc plasma deposition

775

776

Deposition Technologies for Films and Coatings technique and tested in a 0.5N NaCl solution.[17] The TiN coatings were also ion implanted with N, Ti and Au to determine the effect of ion implantation on corrosive behavior. TiN coatings did not provide corrosion protection and ionimplantation of the TiN coatings also did not improve the corrosion resistance. ZrN coatings, on the other hand, did provide corrosion protection by the formation of a passive film. 9. Platinum coatings were deposited onto Mo substrate by electrodeposition from molten cyanide electrolytes. They were then melted with a Nd-YAG laser and corrosion tested in an In HNO3 environment. Coatings of 25 µm thickness were produced and showed corrosion behavior similar to bulk Pt, thus demonstrating a defect free coating.[18] 10. Thick coatings of stainless steel were produced by the particle occlusion technique followed by heat treatment. The corrosion rates were equivalent to bulk stainless steel.[18] 11. For valve seats and stems used in nuclear power plants, Yoshioka et. al.[19] developed a unique plasma CVD technique to deposit multilayer Al2 O3/TiN/TiC, Al2 O3 /TiC, and TiN/TiC coatings on stellite and stainless steel substrates. The coatings showed good wear resistance in a high temperature water environment. The coated specimens also showed excellent corrosion resistance when tested at 548°K for 200 hours.

8.0 HIGH TEMPERATURE OXIDATION/CORROSION The focus here is to decelerate the degradation mechanisms of high temperature alloys caused by high temperature exposure to air and corrosive fuel/combustion product mixtures encountered in heat exchangers, fire-box grates, blades and vanes in the hot section of the gas turbine engine, etc. The coating degradation modes are oxidation, hot corrosion, thermal fatigue, and erosion. Their role in gas turbines and diesel engines is illustrated in Table 14.2. The basic materials science approach is to develop a protective oxide surface on the high temperature alloy. The most suitable oxides are Cr2O3, SiO2 , and Al2 O3 as shown in Fig. 14.1. Additionally, abrasive wear problems

Metallurgical Applications 777

Table 14.2. Coating Degradation Modes for Various Gas Turbine and Diesel Engines.

778


caused by particles in the gas stream can occur. The severest problem is in rotating parts such as the turbine airfoil where a combination of high temperature strength and oxidation resistance is desired. In spite of extensive superalloy development (i.e., Co, Ni, or Fe base material) over a 20 year span, a single alloy composition which has the requisite strength (produced by γ’ precipitation strengthening) and which develops a sufficiently dense protective oxide by thermal growth has not been developed. Fortunately, coating technology has enabled the development of a composite material where the high temperature strength results from the base superalloy composition and the corrosion resistance is provided by the coating which is enriched in the oxide forming elements such as Al, Cr or Si.

Figure 14.1. Plot of parabolic rate constant vs. 1/T for various metals forming CoO, NiO, SiO2 and Al2O3 layers showing the diffusional stability of Al2O3 forming system.


779

The requirements of high temperature protective metallic coating compositions are: 1. Ability to form Al2 O3 (or other suitable stable and protective oxides). 2. Ability to form Al2O3 over a wide compositional range thereby providing compositional freedom to optimize coating mechanical and physical properties. 3. Ability to promote oxide adherence; this provides increased lifetimes or comparable lifetimes at lower coating alloying content. 4. Freedom from detrimental elements and phases which can compete with the formation of the preferred protective oxide, or interact with the corroding product to promote protective oxide breakdown. 5. Freedom from detrimental elements and phases which can interact with the substrate to reduce its stability, load carrying ability, and fatigue properties. 6. Presence of sufficient levels of beneficial “backup” elements to provide resistance to oxide breakdown and penetration, and to increase primary oxide-element-former activity, e.g., Cr increases the activity of Al. Several coating methods are used. The first development (1950’s) was the chemical vapor deposition process involving aluminizing either in a pack or out of a pack.[20] It is a simple, inexpensive, non-line-of-sight process which results in good oxidation resistance from the NiAl coating. However, the process is limited to specific substrate compositions, and the coating layer has limited low temperature ductility. It became unsuitable for the advanced gas turbines and many industrial turbines developed in the 1960’s and 1970’s. The need arose to produce a coating tailored to the specific environmental and mechanical conditions, and one which is independent of the substrate composition and structure. Such an overlay coating is produced by the simple PVD evaporation process using high rate electron beam evaporation. A key development was the ability to deposit alloys from a single source. An extended rod-fed evaporation source is heated by the 150 kW electron beam guns with a sophisticated beam deflection program to obtain precise thickness distribution and coating chemistry. Eleven turbine blades are coated at one time. Up to 20,000 parts can be coated each month.[21] These alloys are known as MCrAlY where M can be Ni, Co, Fe or Ni + Co. The steps in electron

780


beam evaporation-deposition processes for MCrAlY coatings are given below. Receiving Inspection and Batching Surface Preparation: - Degreasing - Grit Blasting - Vapor Honing Weighing (if used as thickness and process control) Loading into Fixtures and Masking Coating Cycle - Pre-Heating - Coating Deposition - Cooling Removal from Fixtures Weighing Overspray Removal Peening Diffusion Heat-Treatment (often substrate alloy solution treatment) Aging Heat-Treatment (if specified) Inspection and Documentation Shipment Reworking—if applicable (stripping and re-cycling through appropriate process) Later developments include the use of an alternate deposition technique, low pressure plasma spray (LPPS), although e-beam evaporation is still the production technique used. As the source temperature of the engine has increased to the limit for the use of superalloys (even including hollow airfoils with forced air cooling) the use of stabilized zirconia as a thermal barrier outer layer is being assessed on an experimental basis. These coatings can be deposited by both e-beam evaporation and LPPS. Currently, the e-beam evaporation method is the preferred one. However, due to a tendency towards brittleness, a microlaminate composite consisting of multiple alternate layers of MCr-Al-Y and Al2O3 deposited by e-beam evaporation has been developed by Professor B. A. Movchan at the Paton Electric Welding Institute, Kiev,


781

Ukraine and is reported to have been used by him on an experimental scale. The thermal conductivity of microlaminate composites has been recently studied by Radhakrishna, Doerr, Deshpandey, and Bunshah,[22][23] who studied the microlaminate materials systems Ni-NiCoCrAlY and Ti-CoCrAlY with up to 480 alternate layers. They showed that the thermal conductivity of the microlaminate composites perpendicular to the laminate plane decreases with increasing number of interfaces (layers) but in a non-linear fashion. The drop in conductivity is associated with the interface. This would present an alternate to the stabilized zirconia coating and has the advantage that is not brittle.

9.0 FRICTION AND WEAR Friction and wear are interrelated phenomena. Tribology is the science of friction and wear. Friction may be defined as the force necessary to initiate sliding between two contacting surfaces—surfaces arenot atomically smooth. Contact is between asperities. Surface topography is thus an essential factor in the study of wear. Tribological contacts are both elastic and plastic, the latter leading to the formation of junctions or microwelds between asperities on mating surfaces. Lubrication is the means used to attempt to separate contacting surfaces by inserting a lubricant between them. The lubricant can be a solid or liquid. There are several types of wear. 9.1 Adhesive Wear The various steps in adhesive wear are: 1. Formation of microwelds between asperities which are deformed under load. 2. Shear of the two surfaces leading to fracture in or near the microweld region causing the generation of wear debris which can then cause abrasive wear. 3. Repeat steps (1) and (2). 9.2 Fretting Wear Two loaded surfaces in contact undergo relative oscillating tangential movement known as “slip” as a result of vibration or cyclic stressing. The amplitude of slip is 2 - 20 µm. The mechanism is somewhat similar to adhesive

782


wear and consists of: 1. Adhesion by microwelding causing material to be raised above the level of the original surface. 2. Shearing of the raised regions. 3. Removal of material by delamination causing the formation of wear debris. 9.3 Abrasive Wear Abrasive wear may be described as damage to a surface by a harder material. Other terms used to describe abrasive wear arescratching, scoring, gouging. There are two types of abrasion wear: 1. Two body abrasion, where the harder surface cuts into the softer surface such as in grinding, cutting, and machining operations. 2. Three body abrasion, where a hard particle like a wear debris is caught between softer surfaces, thus abrading them. 9.4 Fatigue Wear Repeated stresses or stress cycles with two surfaces in contact cause initiation of failure at subsurface levels, finally leading to delamination of the surface. 9.5 Impact Erosion Wear by Solid Particles and Fluids This type of wear occurs when solid particles impact against a target material at speed. It is measured as a weight of material removed (Ε) by a unit weight of impacting particles. When considering the performance of target materials of different densities, it is more appropriate to use volumetric erosion εν which is ε/ρ), ρ being the density. The important parameters are the size, morphology, hardness, and angle of impingement of the particles, and the hardness of the impacted surface. Brittle erosion can occur by micro-fracture with little or no deformation. Ductile erosion can occur by: (i) cutting action, (ii) extrusion and fragmentation Special circumstances lead to the following additional types of wear.


783

9.6 Corrosive Wear The type of wear is due to the dynamic interaction between the environment and the mating material surfaces. It occurs in a two-step cyclic manner: 1. The contacting surfaces react with the environment and reaction products are formed on the surface. 2. The reaction products are attrited off the surfaces by crack formation and/or abrasion in the contact interactions. 9.7 Electric Arc Induced Wear Arcing between surfaces due to the presence of a high potential causes wear by melting, corrosion, and even direct ablation of material. The large craters caused by arcing, in subsequent sliding in oscillatory motion, lead to shears and fractures, abrasion, corrosion, surface fatigue, and fretting. Thus arcing can initiate several modes of wear and can cause catastrophic failure in electrical machinery. 9.8 Solution Wear (Thermodynamic Wear) This type of wear occurs due to the relative thermodynamic instability of one material with respect to the other. It occurs predominantly at high temperatures where chemical interaction is active, e.g., at cutting tool tips contacting the work piece. Consequently, a diamond cutting tool is unstable against a steel work piece because of the large solid solubility of carbon in iron at elevated temperatures. Similarly, various ceramic coatings have different thermodynamic stability vis-à-vis the specific material being cut. The theory has been developed by Professor Bruce Kramer and appears to work well for various carbide and nitride CVD coatings on cemented carbide substrates for the machining of steel. 10.0 COATINGS TO REDUCE FRICTION AND WEAR 10.1 Friction As discussed above, friction is the force necessary to initiate sliding between two mating surfaces. It is proportional to the real area contact between the two surfaces, keeping in mind that mating surfaces are in contact along the asperities. Under load, if the real area of contact increases by

784


deformation of the asperity contacts, friction will increase. For stronger (harder) materials, the real contact area will be smaller than with weaker materials. This is illustrated in Table 14.3 where the coefficient of friction decreases when the surfaces are strong ceramics as compared to weaker metals. Also, in the case of ceramic-ceramic contacts, there is no tendency for microwelding, which occurs between metal-metal contacts, which would increase the friction.

Table 14.3. Friction Coefficient Data at Ambient Temperatures Without Lubrication

Material Pair METAL/METAL 1. SS-Type 304/SS Type 410 2. Titanium/SS Type 410 METAL/CERAMIC 1. Tool Steel/SiC 2. Tool Steel/TiC 3. Tool Steel/TiN 4. SS Type 304/TiC 5. SS Type 304/TiN 6. Ti/TiN 7. Steel/BN, TiN CERAMIC/CERAMIC 1. TiC/TiN 2. TiC/TiN 3. TiC/TiC 4. TiN/TiN 5. TiN/TiN 6. TiN/Ti2N + TiN 7. Al2O3/TiC

Friction Coefficient Low Moderate Humidity Humidity (0.5 - 20%) (~50%)

Deposition Methodology Reference

0.67 0.75

-

-

1 1

0.42 0.75 0.45 0.1 - 12

0.23 0.25 0.49 -

CVD CVD CVD ARE (PVD) ARE (PVD) ARE (PVD) -

2 2 2 1 1 1 4

0.18 0.32 -

0.05 - 0.2 0.19 0.65 0.1 - 0.3 0.19

ARE (PVD) CVD CVD CVD ARE (PVD) ARE (PVD) CVD

3 2 2 2 3 3 -

References: 1. 2. 3. 4.

Suri, A. K., Nimmagadda, R., and Bunshah, R. F., Thin Solid Films, 64:191 (1979) Hintermann, H. E., Thin Solid Films, 84:215 (1981) Jamal, R., Nimmagadda, R., and Bunshah, R. F., Thin Solid Films, 73:245 (1980) Dimigen, H. and Hübsch, H., ICMC '83


785

10.2 Lubrication Lubrication is a universally applied technique to decrease friction and wear. In most engineering applications, oil is common lubricating fluid. Thin films of soft materials are also good lubricants if they can deform easily by shear. Well-known examples are pure face-centered cubic metals like silver and gold, or hexagonal close-packed materials like graphite, molybdenum disulfide (MoS2), tungsten di-selenide (WSe2), etc., which slip easily on the basal plane. These solid lubricants are particularly useful in space-craft operations of long duration (years) where a bearing would have to be activated after a long period of time to move an instrument for example. If liquid lubricants were to be used, the danger is that they may evaporate in the high vacuum of space or be polymerized by the various radiation fluxes present. This danger does not exist with solid lubricants which are usually deposited by sputtering. It is important that the temperature of deposition be such that the film is fully crystalline to enable it to shear,[23] as contrasted to an amorphous deposit produced at lower deposition temperatures. 10.3 Wear A recent book by Bhusan and Gupta[24] gives an extensive discussion on coatings to resist wear. Most of these are hard coatings accompanied preferably by a low coefficient of friction. Hard surfaces can be created by a number of surface treatments: Mechanical - work hardening Thermal

- heating and coating to produce phase changes, e.g., martensite formation in studs, precipitation in non-ferrous alloys.

Chemical

- diffusion of various elements into the surface, e.g., carburizing, nitriding, boriding, chromizing, aluminizing, etc.

Alternately, an overlay coating of a hard layer can be produced on the surface by various deposition techniques such as hard facing (plasma spacing, wire spray, detonation gun, etc.), chemical vapor deposition (CVD), physical vapor deposition (PVD) such as evaporation and sputtering techniques, and plasma-assisted physical vapor deposition (PACVD) such as activated reactive evaporation (ARE), reactive sputtering (RS) and reactive ion plating (RIP). The materials deposited by these overlay processes are metal, alloys, cermets, ceramics, and multilayer composites on a macro and micro scale.

786


The most prominent application of hard coatings for wear is in the life improvement of cutting tools. The CVD process for deposition of hard coatings was developed in the late 1940's and commercially applied to improve cutting tool life nearly 20 years later. One of the problems that had to be overcome to make CVD processing commercially feasible was the need to reharden and temper the high-speed steel substrate after the CVD process was carried out at temperatures ranging from 1000 to 1100°C (1830 to 2010°F). This spurred the development of a low-temperature high-rate deposition process for hard coatings. Such a process (the ARE process) was developed by Bunshah and Raghuram in 1971 (Ref. 25). The application of TiC and TiN coatings onto M42 high-speed steel tools by Bunshah and Shabaik (Ref. 26) in 1975 was the first demonstration of large increases in cutting-tool life. Bhusan and Gupta[24] have compiled a table that summarizes tool wear-life improvement. Table 14.4 is an adaptation of Bhusan’s original table.

Table 14.4. Summary of Tool Wear-Life Improvement by Hard Coating Deposition Tools substrate

Coating material/ treatment

Coating Thickness, (µm)

Deposition method

Improvement in wear life

Reference

HSS (M42) cutting tool

TiC

5-8

ARE

Three to eight times

Bunshah et al. (1975,1977)

Cemented carbide cutting tool

TiN, TiC

5

IP, CVD

IP comparable to CVD

Kobayashi and Doi (1978)

HSS (M-10) drill

TiC, TiN

2

ARE

Twenty times

Nimmagadda et al. (1981)


TiN

5

MS

Several Times

Ramalingham and Winer (1980)


TiC, TiN

8 - 10

CVD

Several times

Hintermann (1981, 1984)


HfN, TiC/Al2O3, TiC/Al 2 O3 , HfN, Al2O3, TiC/TiN Al2O3, TiC/TiN

-

CVD

HfN most superior

Oakes (1983)

HSS (M-10) drill

TiN

1-2

MS

Fifty times

Sproul and Rothstein (1985)


(Ti, Al)N, TiN/TiC

-

CVD

Three times better with (Ti, Al)N

Knotek et al. (1987)

HSS (M-10) drill

(Ti, Al)N, TiN

-

IP

Three times better with (Ti, Al)N

Knotek et al. (1987)


TiN, HfN, ZrN

8 - 18

CVD, IP

Hardness of IP, MS, ARE Coatings at Room Temperature Superior to CVD

Quinto et al. (187)


787

Other important developments include diamond and diamond like carbon coatings, cubic boron nitride coatings and multilayer composites on a microscale, a development which is spurred on by the realization of large multitarget unbalanced magnetron sputtering machines.

REFERENCES 1.

Nowak, W. B., Surface and Coatings Technology, 49:71 (1989)

2.

Egert, C. M., Corrosion, 44:36 (1988)

3.

Horton, J. B., Corrosion Control by Coatings, (H. Leidheiser, ed.), p. 59, Science Press, Princeton (1978)

4.

Meyers, R. G. and Morgan, R. P., Trans. Vac. Met. Conference, p. 271 (1966)

5.

Butler, J. F., J. Vac. Sci. Tech., 1:S52 (1970)

6.

Schiller, S., Forster, H., and Jasch, G. J., Vac. Sci. Tech., 12:800 (1975); Schuller, S., Goedicke, K., and Metzner, C.,Plasma Activated High Rate Electron Beam Evaporation for Coating Metal Strips, 12th International Vacuum Congress, (Oct. 1992), to be published

7.

Fannion, E. R., Reports McAir No. 77-012 and 77-014, McDonald Douglas Corp, St. Louis (1977)

8.

Antler, M., Corrosion Control by Coatings, (H. Leidheiser, ed.) p.115, Science Press, Princeton (1978)

9.

Novak, W. B., and Wong, G., J. Vac. Sci. Tech., A5(4):2164 (1987)

10.

Novak, W. B. and Seyyedi, J., Fundamental Aspects of Corrosion Protection by Surface Modifications, p. 89, The Electrochemical Society, Pennington, NJ (1984)

11.

Novak, W. B., Burns, L. E., and Harris, V. G., J. Vac. Sci. Tech., A7(3):2350 (1989)

12.

Novak, W. B., Materials Science and Engineering, 23:301 (1976)

13.

Motojima, S. and Kohno, M., Thin Solid Films, 137:59 (1986)

14.

Wiiala, U. K., et. al.,Surface and Coatings Technology, 41:191 (1990)

15.

Beverskog, B., et. al.,Surface and Coatings Technology, 41:221 (1990)

16.

Agarwal, P., Nath, P., Doerr, H. J., Bunshah, R. F., Kuhlnam, G., and Koury, A. J., Thin Solid Films, 83:37 (1981)

788


17.

van Leaven, L., Alias, M. N., and Bronon, R.,Corrosion Behavior of Ion Plated and Ion Implanted Films, - to be published

18.

Walters, R. P., Surface and Coatings Technology, 39/40:655 (1989)

19.

Yoshioka, T., Ukegawa, H., Kawai, H., Fujita, N., and Iragashi, T., paper presented at the International Conference on Metallurgical Coatings, San Diego (1988)

20.

Goward, G. W., J. of Metals, 22:31 (1970)

21.

Stephan, M., Dietrich, W., Feurstein, A., and Hoffmann, O. H., Metallurgy, 5:2 (1981)

22.

Radhaknshna, M. C., Doerr, H. J., Deshpandey, C. V., and Bunshah, R. F., Surface and Coatings Technology, 36:143 (1988)

23.

Radhaknshna, M. C., Doerr, H. J., Deshpandey, C. V., and Bunshah, R. F., Surface and Coatings Technology, 39/40,153 (1989)

24.

Goward, G. W., J. of Metals, 22:31 (1970)

25.

Stephan, M., Dietrich, W., Feuerstein, A., and Hoffmann, O. H, Metall, 5:2 (1981)

26.

Radhakrishna, M. C., Doerr, H. J., Deshpandey, C. V., and Bunshah, R. F., Surface and Coatings Technology, 36:143 (1988)

27.

Radhakrishna, M. C., Doerr, H. J., Deshpandey, C. V., and Bunshah, R. F., Surface and Coatings Technology, 39/40:153 (1989)

28.

Spalvins, T., Thin Solid Films, 53:285 (1978)

29.

Bhusan, B. and Gupta, B. V., Handbook of Tribology, McGraw-Hill, (1991)

30.

Bunshah, R. F. and Raghuram, A. C.,J. Vac. Sci. Tech., 9:1385 (1972)

31.

Bunshah, R. F. and Shabaik, A. H.,Research and Development, 26:46 (1975)

15 Characterization of Thin Films and Coatings Gary E. McGuire

1.0 INTRODUCTION Characterization techniques for the analysis of thin films and coatings encompass a broad range of spectroscopies. Each one has a special niche which is based on the information it provides that is not available from other analytical techniques. To fully characterize a material requires a combination of these tools. There are hundreds of different characterization techniques but only a small fraction of them are widely used as general purpose analytical techniques. In this chapter some of the more widely utilized techniques for characterizing thin films and coatings will be described. The discussion will address surface analysis, microscopy and optical techniques. The basic principles of each technique will be reviewed and applications will be described which illustrate the use of the tool. Characterization strategies will be considered in light of the type of information that may be obtained.

2.0 SURFACE ANALYSIS TECHNIQUES 2.1 Auger Electron Spectroscopy Auger electron (AE) excitation is usually accomplished with an energetic electron beam. Figure 15.1 shows an energy level diagram which

789

790


depicts the AE process. Incident electrons with sufficient energy create a core hole through the excitation of an ionizing photoelectron.[1] The atom, left in an excited state, de-excites through the emission of soft x-rays or Auger electrons, both of which are characteristic of the energy levels involved. The kinetic energy of the Auger electron illustrated in Fig. 15.1 may be described as Eq. (1)

E KL1L 2,3 = E K − EL1 − EL 2,3 − ∅

where EK is the energy of the ionized core level, E L1 is the energy of the level from which the electron originates to fill the initial core hole,E L2,3 is the energy level from which the Auger electron is emitted and Ø is the work function. Except for the lighter elements, multiple characteristic Auger transitions occur due to the various core electrons that undergo electron stimulated emission and the multiple combinations of outer energy levels available for de-excitation and Auger emission. The characteristic AE is usually described by the three energy levels involved in its emission as in Eq. 1.

Figure 15.1. Energy level diagram describing the process for the emission of Auger and photoelectrons.

Characterization of Thin Films and Coatings

791

The kinetic energy of the AE is independent of the excitation source. Electron beams are the preferred excitation source because they can be focused to a small spot and deflected to, or rastered over, a region of interest on a sample. Electron beams in the 1 - 20 kV range have been utilized, however, the maximum cross-section for excitation is usually 2.5 - 3 times the energy of the core shell.[2] Higher electron beam energies are more favorable in order to focus the beam to a small spot size. Spatial resolution of 25 - 50 nm has been achieved, however spatial resolution is gained by sacrificing beam current and subsequently sensitivity. High brightness electron sources such as LaB6 or field emitters are used to minimize the loss in sensitivity. The AE transition is a small feature sitting on a large background of inelastically scattered electrons.[3] The most prominent features in the AE spectrum are the backscattered electrons from the primary beam and low energy secondary electrons. The data is presented in the N(E) versus E mode and background subtracted to enhance the weak Auger signal. Historically this was done using a lock-in amplifier and presenting the data in the dN(E)/ dE versus E format. Figure 15.2 shows a schematic diagram of an Auger spectrometer. The optics for the primary electron beam are coincident with the cylindrical mirror analyzer. The focal point of the primary electron beam is designed to be identical to that of the electron spectrometer. When a sample is positioned at the focal of the electron spectrometer, Auger electrons are excited at the surface by the primary beam and pass through the acceptance slits into the spectrometer. A negative potential applied to the outer cylinder of the analyzer deflects the electrons through the exit slit onto an electron multiplier. By sweeping the voltage on the outer cylinder, the electron energy spectrum may be scanned. A secondary electron detector is incorporated in the chamber to facilitate locating the primary beam on the area of the sample to be analyzed. The shallow attenuation length or inelastic mean free path of electrons as a function of energy is the factor which gives all of the electron spectroscopies their surface sensitivity.[4] In the range of interest, from 0 - 2000 eV, the inelastic mean free path (IMFP) is only a few monolayers, Fig. 15.3. The IMFP dependence on electron energy in this range varies considerably from material to material. These variations are associated with differences in the electron energy loss functions among the materials.[5] These variations lead to uncertainty in the quantification of the data.


Figure 15.2. Schematic diagram of a cylindrical mirror analyzer Auger spectrometer.


793

Figure 15.3. Plot of electron inelastic mean free path versus energy which illustrates the basis for the surface sensitivity of the electron spectroscopies. (Reprinted with the permission of the publisher, John Wiley & Sons, Ltd.)

Characteristic Auger transitions may be observed for all elements with three or more electrons. As a result, AES is often used to survey the surface composition of materials. For example, Fig. 15.4 shows an AES spectrum of the surface of a Si wafer coated with an Al - 4% Cu (atomic weight) after etching in a CCl4 plasma. The plasma etch removes the Al but leaves a Cu residue since Cu-plasma reaction products are not as volatile as those of Al. The Curich residue is only a few monolayers thick and, as a result, can only be detected by surface analysis techniques such as AES. Utilizing the AES surface sensitivity with ion sputtering provides a measurement of the elemental composition as a function of depth (depth profile). An ion gun is utilized to bombard the surface with a flux of inert gas

794


ions in the 1 - 5 keV range, removing controlled amounts of material due to the transfer of momentum from the impinging ions to the surface atoms. By monitoring the Auger signal intensity of selected elements as a function of sputtering time, a plot can be generated which represents the concentration as a function of depth. Figure 15.5 shows the in-depth profile of a sample consisting of multiple 50 nm layers of Ni and Cr. [6] The depth resolution between the Ni and Cr layers is excellent and is achieved by using a low primary beam energy to minimize knock-on effects and sample rotation to reduce ion induced surface roughness. A variety of ioninduced artifacts may occur, however the additional information gained as a result of an in-depth profile with a resolution of 20 - 50 Å usually outweighs the disadvantages.

Figure 15.4. Auger spectrum of a silicon surface after removing a copper-doped aluminum layer using a CCl4 plasma.


795

Figure 15.5. AES depth profile of a sample of multiple 50 nm layers of Ni and Cr. (Reprinted with the permission of the publisher, John Wiley & Sons, Ltd.)

Auger spectra usually contain features which are characteristic of the surface chemistry of the material under investigation as a result of the participation of the valence band electrons in the Auger process. These features have been studied for many systems and may be used as a means of identifying the chemical oxidation state. Figure 15.6 illustrates the change in the Ga L3M4,5M4,5 Auger electron kinetic energy and line shape in two different chemical environments.[7] The Ga Auger transition is shifted by 4.9 eV for the oxide formed on GaAs by anodic oxidation relative to the peak for the underlying GaAs substrate. A similar shift of 5.8 eV is observed for the As L3M4,5M4,5 Auger transition for As in the anodic oxide relative to the transition for As in GaAs. Chemical shifts of this magnitude have been observed for most elements. Since the spectral features are complex and the magnitude of the chemical shift relatively small, it is not simple to determine the composition of multicomponent systems. The primary electron beam can be focused to less than 200 Å. By rastering the beam over the sample surface, elemental distribution maps may be obtained of the surface composition. This is accomplished by fixing the pass energy of the spectrometer so that only one Auger transition is being monitored while the beam is scanned over the surface. If more than one element is of interest the pass energy is adjusted for each elemental map. One typically looks for inter-relationships in the maps as an indication of surface structure, compound formation, corrosion, etc.[8] When the Auger transitions exhibit features which are indicative of certain oxidation states, these transitions may be mapped to obtain the chemical state distribution.

796


Figure 15.6. Typical relationship between minimum detectable concentration and primary beam current and diameter at 10 kV for AES.

AES has a detection limit of approximately 0.1% atomic or 1018 atoms/ Several handbooks of Auger data[9] provide relative sensitivity factors for the elements which may be used for quantitative analysis. A new journal,Surface Science Spectra, archives AES and XPS spectra that have been peer reviewed. Figure 15.6 illustrates the typical relationship between minimum detectable concentration and primary beam current and diameter at 10 kV for AES.[10] At higher resolution the beam current is restricted resulting in lower sensitivity. Even though AES has relatively poor sensitivity, it is one of the more popular surface and thin film analysis techniques. cm3 with a sensitivity variation of 50 - 100 across the Periodic Table.


797

AES may be utilized on a wide variety of materials but, due to the use of an electron beam for excitation, it suffers limitations. The electron beam, and ion beam used for sputtering, may induce sample decomposition. This problem is accentuated by the high current densities that occur with small probe diameters. Insulating materials may be difficult to evaluate due to sample charging effects. The imbalance of currents from the primary beam, the secondary electrons, and the sample result in a surface potential which distorts the Auger electron energy. 2.2 Photoelectron Spectroscopy Photoelectron spectroscopy is a technique which has many similarities to AES.[11] The same energy level diagram (Fig. 15.1) may be used to describe the photoemission process. Excitation of the ionizing photoelectron may be accomplished through the use of a variety of energetic photons or charged particles. The primary focus in this section will be monochromatic x-ray excitation of photoelectrons (XPS). Use of a monochromatic excitation source is essential to this spectroscopy since the photoelectron’s kinetic energy is directly dependent on the energy of the excitation source. By knowing the energy of the x-ray (hv) with a high degree of accuracy and measuring the kinetic energy (KE) of the emitted photoelectron, the binding energy (BE) of any electron energy level less than the photon energy can be determined from the relationship: Eq. (2)

BE = hv - KE + Ø

where Ø is the work function. A variety of electrostatic electron energy analyzers have been produced commercially. The most widely used come from the family of spherical sector analyzers illustrated schematically in Fig. 15.7. The lens in this case is simply a transfer lens which transfers electrons from the analyzed area on the sample onto the entrance slit of the analyzer. Removing the sample from close proximity to the entrance slit of the analyzer in this way provides much greater working space around the sample. Most spectrometers make use of a position-sensitive, multiple-array detector to enhance the count rate. An x-ray source with either an Al or Mg anode, mounted in proximity to the sample is used for excitation. The x-rays flood a broad area of the sample since they, unlike the electron source in AES, can not be easily

798


focused. The acceptance angle of the spectrometer in combination with the transfer lens determines the area of analysis which is typically a few millimeters in area. Figure 15.7 illustrates an x-ray source with a monochromator, although x-ray sources with and without monochromators are in widespread use. The monochromator is based on the diffraction of Al x-rays off of a bent quartz crystal resulting in a narrower linewidth (0.4 eV versus 0.9 eV), focusing of the x-rays (<150 µm), and elimination of the satellite x-ray lines.

Figure 15.7. Schematic diagram of an XPS system utilizing a bent quartz crystal xray monochromator in conjunction with an electrostatic lens and spherical sector analyzer.

Characterization of Thin Films and Coatings 799 The spectra are usually plotted in the N(E) versus BE format. Each element exhibits a unique set of photoelectron (PE) transitions corresponding to its atomic energy levels. The PE transition energies are a function of atomic number so that the energy levels of adjacent elements in the Periodic Table are shifted in binding energy.[12] The spectral features are Gaussian-like, sitting on a low background. In addition to elemental identification, the strongest attribute of XPS is its ability to distinguish different oxidation states. Figure 15.8 shows the XPS spectrum of a bare Si wafer. The two peaks associated with the Si2p transition are from elemental Si and SiO2 of the 20 - 30 Å of native oxide. So, in addition to illustrating the chemically-shifted peaks, this figure also illustrates the surface sensitivity of XPS. The surface sensitivity, as in AES, is controlled by the inelastic mean free path of the electron as illustrated in Fig. 15.3, rather than the path length of the x-rays used for excitation.

Figure 15.8. XPS spectrum of a Si surface showing two Si2p peaks, one associated with elemental Si and the other associated with the SiO2 of the native oxide.

800


The ability to distinguish different oxidation states, as illustrated by the chemical shift of the Si2p transition of SiO2 relative to that of Si, is one of the strengths of XPS. The shift in the core level binding energy is due to changes in the valence electron density. Chemical shifts as large as 12 eV have been observed; however, since there are many compounds in whichthe element of interest is in similar oxidation states, the binding energies are not unique. This is illustrated in Table 15.1 for the Cr2p3/2 transitions for a series of chromium compounds. Similar chemical shifts may be observed for both cations and anions. Table 15.2 lists the relative chemical shifts of S, Se and Te compounds[13]for anionic species X−2 , XO3 −2 and XO4−2. In addition, there are other spectral features which provide additional chemical information.[14]

Table 15.1. Chemical Shifts in the Cr2p 3/2 Transition of Chromium Compounds


801

Table 15.2. XPS Chemical Shifts of Anion Compounds

XPS, like AES, can be combined with ion sputtering in order to generate in-depth profiles. Since the area of analysis for XPS is larger than for AES, the ion beam must be defocused or rastered over a larger area in order to obtain an analysis area of uniform depth. As a result, the sputter rate is much slower which, especially when coupled with the longer data acquisition times of XPS, results in longer times to generate concentration versus depth profiles. By monitoring the chemically shifted peaks it is possible to generate chemical oxidation state versus depth profiles. This is extremely valuable for multicomponent systems. For example, Table 15.3 lists the oxide formed during the thermal oxidation of various compound semiconductors. Typically one of the components of the system will oxidize more readily; based on the heats of formation, the predominant oxide may be predicted. Frequently the other component is concentrated at the interface or lost due to evaporation. The composition of the oxide, especially as a function of depth, has been found to be strongly dependent on the method and conditions of formation. Since the photoelectron spectra of many elements exhibit only small chemical shifts for a series of compounds, it is frequently necessary to examine the other features of the spectrum. One of these features which frequently exhibits useful chemical information, even when the photoelectron transitions do not, is the corresponding Auger transition. Since Auger emission is a multi-step process in which two electrons are emitted—the Auger and photoelectron—the electron shells surrounding the atom have more

802


time to undergo relaxation. This results in larger chemical shifts for the Auger transition as compared to the corresponding photoelectron transition. Table 15.4 compares the binding energies of photoelectron and Auger transitions for a series of metals and their oxides.[12] In all cases, the Auger transition exhibits a factor of two larger chemical shift over that of the photoelectron transition. However, Auger spectra are more complex and exhibit broader line widths which limits the use of the chemical shift information they contain. In addition, Auger transitions are not always excited due to the limited energy range of the typical Al or Mg x-ray source. The detection limit for XPS is approximately 0.5% atomic or 5 x 1018 atoms/cm3 with a sensitivity variation of 102. There are several data sets available which provide relative sensitivity factors based on peak area.[15] The x-rays used for excitation do less damage to the surface than the electron beams used for AES. Sample charging is minor since only the photoemission and sample return currents must be balanced.

Table 15.3. Chemical Compound Formation During Thermal Oxidation of Compound Semiconductors


803

Table 15.4. Chemical Shifts in X-ray Excited Auger Spectra

2.3 Secondary Ion Mass Spectroscopy Secondary ion mass spectroscopy (SIMS) is the mass analysis of secondary ions generated by ion sputtering. As illustrated in Fig. 15.9, bombarding the surface of a solid with an energetic ion beam generates a variety of secondary transitions, including the emission of electrons, photons and ions. Detection of any of these secondary events could serve as the basis for an analytical probe, however, SIMS is optimized for the detection of positive and negative secondary ions.

804


Figure 15.9. Schematic diagram of secondary particles generated by an incident ion beam.

There are several different types of SIMS instruments in widespread use. They are based on a magnetic sector or quadrupole mass spectrometer. Figure 15.10 is a schematic diagram of a quadrupole mass spectrometerbased instrument which contains all the essential features of a SIMS system. Many SIMS instruments have two ion sources; a duoplasmatron ion source which generates ions from a gas source such as Ar or O2 and a liquid metal ion source which generates ions using emission of low melting-point metals. The primary beam is mass analyzed to separate the positive and negative ions and the neutrals which are produced in the ion source. A condenser lens is used to focus the ions and charge deflection plates are used to position the beam or raster it over the sample surface. A simple electrostatic analyzer is used prior to the mass analyzer in order to select a narrow energy distribution of the secondary ions. A quadrupole mass spectrometer is used for low cost, low to intermediate mass resolution and high speed peak switching. Magnetic sector mass spectrometers are used for high mass resolution and high collection efficiency but are slow at peak switching. Instruments frequently make use of a secondary electron detector to position the primary ion beam and an electron gun for charge compensation.


Figure 15.10. SIMS.

805

Schematic diagram of a quadrupole mass spectrometer based

Since SIMS uses ion bombardment to generate the secondary ions, it is intrinsically a depth profiling technique. In order to insure that the ions originate from a uniform depth it is necessary to raster the Gaussian-shaped beam in order to achieve a uniform ion flux. The ion beam is rastered over an area slightly larger than the area of analysis in order to avoid accepting a signal from the sloping sidewall. The signal is gated such that the signal is only accepted when the primary beam is away from the crater wall.

806


One of the most important aspects of SIMS analysis is the successful generation of secondary ions. One aspect of this is the selection of the primary beam energy. The sputtering yield, atoms removed per incident ion, is dependent on the incident ion energy. SIMS ion sources are usually designed to operate in the 2 to 20 keV energy range. There is no benefit in going to higher accelerating voltages since the sputtering yield is flat or decreases above 20 keV. Operation at lower primary beam energies results in slower sputtering rates and decreased ion yields but results in improved depth resolution. For shallow doping profiles, cascade mixing and surface roughening may limit the depth resolution of SIMS analysis. Several studies have demonstrated a practical lower limit of approximately 3 keV before the ion beam broadens and limits the depth resolution [17] as illustrated in Fig. 15.11. The angle of incidence of the primary ion beam as well as sample rotation, which was discussed in the section on depth profiling using AES,[18] play a role in depth resolution. Only a few percent of the atoms removed by sputtering are ionized. The remainder are neutral atoms or atom clusters. Proper selection of the primary ion can enhance the ion yield. Positive primary ions enhance the yield of negative secondary ions while negative primary ions enhance the yield of positive secondary ions. The favored primary ions are O − and Cs+ for their high yield of positive and negative secondary ions, respectively. The secondary ion yield is also a function of the electronegativity of the elements in the sample. For example, when a negative primary ion beam is used, the relative positive ion yield will be greatest for those elements with the lowest electronegativity. Figure 15.12 illustrates the variation in secondary ion intensity as a function of atomic number for O − bombardment. The secondary ion yield correlates with the periodic nature of the electronegativity of the elements.[19] Conversely, when a positive primary ion beam is utilized the relative negative ion yield will be greatest for those elements with the highest electronegativity. The SIMS spectrum is a plot of the secondary ion intensity versus the mass-to-charge ratio. As can be seen from Fig. 15.13, the spectrum from even high purity elements like Si can be very complex. The spectrum results from the detection of singly and multiply-ionized atoms and multiatomic species formed during ion sputtering. The ion intensity is usually plotted on a log scale due to the large dynamic range, 10 6, of the data.


807

Figure 15.11. The energy dependence of the depth resolution of a quardrupolebased SIMS is illustrated for Cs ion beam profiling of As-implanted Si. (Reprinted with the permission of the Amer. Vac. Soc.)

808 Deposition Technologies for Films and Coatings Figure 15.12. Plot of the relative positive secondary ion yield versus atomic number for 13.5 keV oxygen ions. (Reprinted with the permission of the journal, Analytical Chemistry.)

Characterization of Thin Films and Coatings 809

Figure 15.13. Plot of the relative secondary ion intensity versus the mass-to-charge ratio resulting from oxygen ion bombardment of high purity silicon.

810


The bulk detection limit of SIMS is 1014 - 10 15 atoms/cm 3 for many elements. This is a factor of 104 - 105 more sensitive than AES or XPS. However, SIMS data is more difficult to quantify because the ion yield is matrix dependent. As a result, other techniques such as Rutherford backscattering spectroscopy (RBS) are used to normalize the SIMS data. This is most often done for implant profiles where the SIMS profile may vary from 1020 to 1015 atoms/cm3. RBS is used to calibrate the high concentration portion of the SIMS profile. The depth resolution of these profiles is less than 50 Å, depending on the analysis conditions. SIMS depth profiles up to several micrometers may be obtained with this depth resolution but, without precautions such as sample rotation, cumulative ion mixing may degrade the depth resolution. The depth scale of a SIMS profile requires secondary calibration using a mechanical stylus technique, ellipsometry, or an interferometric technique. SIMS is one of a few analytical tools capable of distinguishing isotopes. This has resulted in some well designed experiments that take advantage of this capability. For example, Coleman et al.[20] utilized SIMS to investigate the anodic oxidation of GaAs. By using isotopically labeled H2O, they were able to distinguish the mechanism by which anodic oxidation proceeds. Figure 15.14 illustrates the depth profiles that were obtained from a GaAs (001) wafer anodized in H2O18 then H2 O16 and another sample anodized in the opposite sequence. From this study the authors were able to show that oxygen is incorporated into the growing oxide at the oxide-electrolyte interface as opposed to the oxide-semiconductor interface. Mass transport occurs through the interstices of the growing oxide. SIMS is effectively used in the analysis of a broad range of materials; however, it does suffer from several limitations in addition to the matrixdependent ion yield. One of these is charge-induced migration of easily ionized elements such as Na, Li or Cl. The charge that builds up on insulators during ion bombardment may reach sufficient field strength to cause chargeinduced migration. Since both positive and negative primary ion beams are used for analysis, charge-induced migration may result in diffusion and accumulation at the surface or at an interface. This effect may be minimized by lowering the sample temperature or by neutralizing the surface charge. Charge neutralization may be accomplished by positioning a hot filament in proximity to the sample or by exposing the surface to low energy electrons from an electron gun. The degree of success in neutralizing the surface charge, however, greatly influences the secondary ion yield.


811

Figure 15.14. Depth profile of the isotopic distribution of oxygen in an anodic oxide grown on GaAs (001). Curve (A) is for GaAs (001) anodized in H2O18 then H2O16. curve (B) is for anodization first in H2O16 then H2O18. (Reprinted with the permission of the Electrochem. Soc.)

Since the secondary ion yield is so dependent on a number of different factors, it limits the application of SIMS. This has led to the development of techniques which enhance the ion yield in order to reduce the matrix dependence and improve the sensitivity. By adding a photon source with sufficient energy to ionize the material removed during the sputtering process, the ion yield can be increased significantly. This approach is referred to as sputter-assisted laser ionization

812


(SALI). [21] Another technique, resonance ionization spectroscopy (RIS), uses a laser beam which is tuned to the frequency necessary to ionize the atoms of interest.[22] This provides a means of selectively analyzing for one element at a time. Many elements require multiple photon pulses to excite the atom to an excited state then finally to ionization. It is important to choose resonate states that can be easily excited and that have large photoionization cross-sections. RIS uses ion sputtering to remove material from the surface analogous to SIMS. However, the secondary ions generated by ion bombardment are extracted without analysis. The laser beam is pulsed to excite the remaining neutral material which constitutes approximately 95% or more of the material removed by sputtering. Using this approach, detection limits of down to 1010 atoms/cm3 have been reported for Na in Si.[23] 2.4 Rutherford Backscattering Spectroscopy Rutherford backscattering spectroscopy (RBS) is the energy analysis of ions that are backscattered from a surface. Typically ions with low mass, such as H+ or He+, are accelerated toward the sample at a potential of 0.5 - 2.0 MeV. As shown in Fig. 15.15, the target, M2 , recoils while the primary ion, M1, scatters at an energy E1 at an angle Ø. The scattering energy, E1, is easily calculated from the relationship, Eq. (3)

E1 = K(M1, M2, Ø) Eo

where  M cos ∅ + M 2 − M 2 sin 2 ∅   1  2 1 K =  M1 + M2    

2

The scattering cross-section is a smoothly varying function of the target mass as shown in Fig. 15.16. From this curve it is obvious that the scattering efficiency is very low for those elements with low atomic mass. In addition, it is difficult to distinguish elements which have similar masses when the elements have high atomic masses.[24] Figure 15.17 is a schematic diagram of the equipment necessary to perform RBS. The accelerator must be capable of generating MeV ions of the light elements. Modern instruments make use of compact tandetron accelerators which allow the construction of RBS systems which are not signifi-


813

cantly larger than other surface analysis equipment. The analysis of the backscattered ions may be achieved through the use of an electrostatic analyzer or a solid state detector. The solid state detector, the preferred detection system, is positioned in front of the target at an angle of approximately 30° from the incoming primary beam. A thin mylar sheet is placed in front of the detector to attenuate low energy secondary ions and secondary electrons.

Figure 15.15. collision.

Schematic diagram illustrating a Rutherford backscattering


Figure 15.16. Plot of the Rutherford backscattering cross-section versus target atomic mass.

Characterization of Thin Films and Coatings 815

Figure 15.17. Schematic diagram of a Rutherford backscattering spectrometer.

816


The spectra are plots of scattered ion intensity versus energy. An RBS spectrum is the sum of a family of scattering curves from each atomic mass in the target. As shown in Fig. 15.18, KEo represents inelastic scattering from the front surface of the target.[24] At a depth X, the primary ion loses additional energy through electron scattering, both going into and escaping from the solid. Since Rutherford scattering occurs at all depths, a curve is generated which is the sum of all these events. Each atomic mass in the target generates a separate curve based on its scattering cross-section.[25]

Figure 15.18. A plot of the Rutherford backscattering yield versus the energy of the backscattered ion with an accompanying illustration showing the scattering location in the sampled depth. (Reprinted with the permission of the publisher, Acad. Press.)


817

Figure 15.19 illustrates the application of RBS in the analysis of the silicides formed during the interaction of Ni and Si. [26] The dashed line represents the as-deposited Ni on Si case, where He + scattering from Ni and Si results in distinct He+ scattering energies. Upon heating at 300°C for 90 minutes, Ni2 Si forms for which the scattering curve is represented by open circles. The signal from Ni has decreased in intensity and broadened while the front of the Si scattering curve has moved toward the Ni curve indicating the formation of silicide. Additional heating results in a further decrease in intensity and broadening of the Ni curve and an increase in the Si signal associated with the silicide. Since the scattering cross-sections for Si and Ni are known, the stoichiometry for the different phases of silicide can be calculated without the use of standards. There are many examples like this in the literature where a heavy metal in a matrix of a low atomic mass element lends itself to RBS analysis.

Figure 15.19. RBS spectra of the phases of nickel silicide formed following the deposition and annealing of nickel on silicon. (Reprinted with the permission of the publisher, Akademie-Verlag.)

818


Since RBS is essentially a non-destructive quantitative analysis technique, it is frequently used to calibrate other surface analysis techniques. It, however, has a sensitivity limit of about 10 18 atoms/cm 3 in a Si matrix. [27] This is comparable to the sensitivity limit for AES and XPS but is much less than that of SIMS. Of the surface analysis techniques, RBS is unique in its ability to distinguish whether a dopant occupies a substitutional or interstitial site in a crystalline lattice. When the primary ion beam is oriented along the crystalline planes, the ions penetrate long distances by channeling along the open planes. Scattering occurs at crystal imperfections and interstitial impurity sites. Figure 15.20 compares the RBS spectra from Si implanted Si(100) samples which were positioned such that random scattering and channeling occur.[28] The virgin sample exhibits minimal scattering except in the random orientation, indicating the quality of the crystal. After implantation, the crystal has undergone extensive damage which is evident in the increased scattering along the channeling direction. Subsequent heating at 550°C and 850°C anneals out much of the damage, however, the crystal quality of the virgin sample is not recovered.

Figure 15.20. RBS spectra for Si (100) in the random and < 110 > aligned direction before and after 80 keV 30Si+ implant and subsequent anneal at 550° and 850°C. (Reprinted with the permission of the Electrochem. Soc.)


819

The same equipment used to do RBS can be used for nuclear reaction analysis (NRA).[24] Instead of Rutherford scattering, the primary ion must penetrate the nucleus of the target atom and induce a nuclear reaction as depicted in Fig. 15.21. The nuclear reaction cross-section as a function of incident energy must be known in order to select an energy which will result in adequate yield. The energies required for NRA are frequently higher than those used for RBS. Table 15.5 lists some nuclear reactions that are used for thin film analysis. NRA compliments RBS in that many of the useful nuclear reactions are for low atomic number elements for which RBS has low sensitivity. Since the nuclear reaction cross-sections are well known, NRA, like RBS, is quantitative without the use of standards. This is especially beneficial for elements like H which are difficult to detect and quantify by other analytical techniques. NRA is also used to calibrate other surface analysis techniques like SIMS. For example, Fig. 15.22 shows the data from NRA and SIMS analysis of B implanted into Si at 10 keV. NRA is not as sensitive as SIMS but it is less matrix dependent and as a result can be used to calibrate the higher concentration portion of the profile.

Figure 15.21. Schematic diagram illustrating an ion-induced nuclear reaction.

820


Table 15.5. Nuclear Reactions Useful for Thin Film Analysis


821

Figure 15.22. Depth profile analysis of boron implanted into silicon at 10 keV using NRA (circle) and SIMS (solid line). (Printed with the permission of M. L. Swanson and N. R. Parikh.)

822


3.0 IMAGING ANALYSIS TECHNIQUES 3.1 Scanning Electron Microscopy Scanning electron microscopy is surface imaging of solids using electron-beam-generated secondary electrons. Figure 15.23 illustrates the electron beam interaction with a solid. The primary beam may be focused to a spot < 50 Å in diameter.[29] Upon interacting with the solid, secondary electrons are generated which are utilized to image the surface. As the high energy primary electrons penetrate the solid, they undergo scattering which increases the interaction volume. Some of the primary electrons will be backscattered toward the surface with little or no loss in energy. Energetic primary electrons ionize atoms in the solid producing x-rays which are characteristic of the elements that are present. With suitable detectors, the x-rays may be detected to provide elemental analysis. Secondary electrons are low energy even though the primary electron beam is several keV or higher. Figure 15.24 shows the average energy distribution of secondary electrons from metals. [30] The peak in the energy distribution is below 5 eV. In order to efficiently collect the secondary electrons, a high potential bias is applied to a scintillator tube which is positioned in proximity to the sample. The signal is converted to light and fed out through a light pipe to a photomultiplier tube as shown in Fig. 15.25. SEM images at less than 20 Å resolution have been obtained with several hundred thousand times magnification. SEM’s provide higher magnification with greater depth resolution than optical microscopes. SEM images may become distorted by the surface potential that builds up on insulators or edge effects at sharp contours. Insulators may be coated with a thin (~100 Å) conductive layer to dissipate the surface charge. Backscattered electrons have the same energy as the primary electrons. The electron backscattering coefficient, like that of ions in RBS, is a well known, smoothly varying function of atomic number as shown in Fig. 15.26. Since the backscattering yield varies more than the secondary electron yield across the Periodic Table, backscattered electrons yield better image contrast in many situations.[31] The high energy backscattered electrons sample much greater depths than low energy secondary electrons.


823

Figure 15.23. Diagram illustrating the interaction of the primary electron beam with a solid surface in the production of secondary and backscattered electrons, x-rays, and other secondary radiation.

824


Figure 15.24. Plot of the average intensity of secondary electrons from metals as a function of energy.

Figure 15.25. Schematic diagram of a scintillator tube used for the detection of secondary electrons.


825

Figure 15.26. Plot of the electron backscattering coefficient versus atomic number.

One of the most common analytical attachments to the SEM is the energy dispersive x-ray spectrometer (EDX). The high energy primary electron beam causes emission of a core electron which leaves the atom in an excited state. The atom undergoes de-excitation by x-ray emission, as described previously for Auger electron emission. The x-rays are characteristic of the elements from which they originate. The emitted x-rays are detected by a solid state detector which is positioned in the vicinity of the sample, Fig. 15.27. The detector is a Li-doped Si crystal which is biased at high voltage. X-rays interacting with the detector create electron-hole pairs which are swept through the detector due to the high voltage bias. The charge pulse is converted to a voltage pulse by a charge-sensitive

826


preamplifier. The useful energy range for EDS systems is from 1.0 to 220 keV which limits the analysis to elements with Z > 9. EDS detectors with thin protective layers or no protective layer, when used in ultra-high vacuum systems, permit analysis of the lighter elements down to B. The analysis depth is dependent on the path length of the x-rays, not the primary electron beam. As a result, EDS signals may originate from depths of 0.5 µm or more.

Figure 15.27. Diagram illustrating the detection of electron-beam-excited x-rays in an SEM using a solid state lithium-doped silicon detector.

The major advantage of EDX is its ability to operate in the pulsecounting mode and detect the characteristic x-rays for all elements above F in the Periodic Table. A complete spectrum may be obtained in a much shorter time with EDX than with a wavelength dispersive x-ray (WDX) analyzer. An EDX analyzer has a resolution of approximately 150 eV, whereas a WDX analyzer has a resolution of 5 eV. The Li-doped Si detector


827

used with EDX requires liquid nitrogen cooling to keep the Li from diffusing and rapidly degrading the detector’s performance.[30] A schematic diagram of a wavelength-dispersive detector is shown in Fig. 15.28. The electron beam excited x-rays interact with a crystal which disperses the x-rays. As the crystal is rotated, the different wavelength x-rays enter the detector. A variety of crystals are used in order to optimize the energy resolution and collection efficiency of the broad range of x-ray energies for elements Z≥ 6. A spectrum of x-ray intensity versus wavelength is generated from which the characteristic x-ray lines may be identified. The detector may also be operated at a fixed wavelength, so that the detector output represents an intensity map of the sample surface for one characteristic x-ray when the electron beam is rastered. The most commonly used detector for the WDX spectrometer is a gas flow proportional counter. When an x-ray enters the tube through a thin window and is absorbed, it causes a photoionization-induced cascade which gives rise to a charge pulse.

Figure 15.28. Diagram illustrating the detection of electron beam excited x-rays with a wavelength dispersive detector.

For bulk samples more than a few micrometers thick, spatial resolution for elemental analysis does not improve for beam diameters much less than one micrometer since the volume of x-ray production is determined by electron beam scattering. It is possible to obtain images of magnetic domains in an SEM. Ferromagnetic materials are composed of small subgrain-sized regions

828


called domains. The magnetic moment of these domains may be along a certain crystallographic axis. In some crystals, the magnetic moment at a surface will often have a component normal to the surface. A secondary electron ejected from the surface of a uniaxial crystal will experience a force proportional to the surface magnetic moment. Because the local magnetic flux changes sign over each domain affecting the secondary electron signal, images of domains can be obtained using a secondary electron detector (type I magnetic contrast). In most ferromagnetic crystals, there is more than one preferred axis. In these crystals, closure domains form at the surface that have their magnetic moment lying along the axis most closely parallel to the surface. Inside the metal, there is an abrupt change in the magnetization direction at a domain boundary. Therefore, a primary electron will experience a force in different directions from domain to domain. This gives rise to changes in the backscattering yield as the primary beam sweeps across a domain boundary (type II magnetic contrast). Type I magnetic contrast can be performed at resolutions to 1 µm while type I magnetic contrast is limited to 0.2 µm.[32] 3.2 Transmission Electron Microscopy The transmission electron microscopy (TEM) utilizes an electron beam much like the SEM but at higher accelerating potential. A higher accelerating potential is utilized since only electrons that are transmitted through thinned specimens are imaged. The accelerating potential required depends on the sample thickness and atomic mass but is typically 100 - 400 kV. The TEM has superior resolution (0.15 nm) to the SEM resulting from the very small wavelength of high-energy electrons and the limited sample volume for electron scattering. In the conventional TEM mode, in which the entire region of the specimen is flooded with incident electrons, the images can be viewed directly on a fluorescent screen or recorded on photographic film. The information obtained by electron microscopy is derived from either elastic or inelastic scattering processes. Electrons that do not undergo any scattering or elastic scattering with little change in trajectory will form the transmitted beam (bright field mode). Elastically scattered electrons with a significant change in trajectory form the diffracted beam (dark field mode). Modern analytical electron microscopes are often equipped with a wide variety of signal detectors. In the scanning TEM mode (STEM), any of these signals (transmitted electrons, diffracted electrons, backscattered electrons,


829

secondary electrons and characteristic x-rays) can be used to modulate the input signal to a cathode ray tube (CRT) to form an image (Fig. 15.29).

Figure 15.29. Diagram of the analytical attachments and modes of operation of a STEM for evaluating thinned specimens. (Reprinted with the permission of the publisher, Marcel Dekker, Inc., Ref. 36.)

Samples are thinned to approximately 50 - 300 nm through a combination of chemical and/or mechanical polishing and ion milling.[33] The specimen is mounted on a wire grid for ease of handling. Sample may be thinned vertically or horizontally. Thin films may be prepared by mechanically or chemically removing the substrate. Figure 15.30 is a cross-sectional TEM micrograph of an epitaxial CoSi2 grown using a bimetallic layer process. [34] The CoSi2 /Si system is of interest since it has one of the lowest resistivities of the silicides and has a CaF2 structure with a lattice parameter of about 1.2% less than that of Si, allowing CoSi2 to grow epitaxially on Si. Even though the lattice mismatch between Si and CoSi2 is 1.2%, this is considered relatively large, and relaxation of the CoSi2 lattice is expected to occur by introduction of misfit dislocations. The dislocations are associated with atomic steps at the interface. Crosssections like this may be utilized to evaluate various epitaxial growth techniques, implantation damage, deposited films, and contact formation.

830


Figure 15.30. Cross-sectional TEM micrograph of epitaxial CoSi2 grown on Si (100).

Figure 15.31 provides a different perspective of a silicide film on Si, a plane view. Polycrystalline silicides, such as CoSi2 begin to thin along grain boundaries when annealed at elevated temperatures.[35] The film tends to break up into islands, agglomerate, when annealed at sufficiently high temperatures even for relatively brief periods. The agglomeration leads to films with high resistivity which defeats the purpose of using the low-resistivity silicides as contacts. When an electron beam interacts with a thin film of a crystalline material, some electrons will undergo elastic scattering with essentially no loss in energy but significant change in trajectory. The directions in which electrons are elastically scattered is determined by the orientation between the atomic planes in the specimen and the incident beam. Coherent elastic scattering in the forward direction produces the conventional electron-diffraction patterns in TEM. The angles through which the electrons are scattered are given by Bragg’s law. Diffraction of electrons is identical to the diffraction of x-rays by a crystal except for the wavelength of the diffracting radiation. The standard method for generating diffraction patterns using conventional TEM is by selected area diffraction (SAD) where an aperture is used to limit the area of the specimen from which the diffraction pattern is obtained. Figure 15.32 illustrates the nature of the electron diffraction


831

pattern that may be observed.[36] Single crystalline samples produce ordered diffraction patterns which depend on the crystal structure of the system being studies, qualitatively analogous to the Laue technique of x-ray diffraction. From the pattern it is possible to deduce the indices of the crystal plane giving rise to the diffraction spots. As the sample becomes more disordered, the ordered diffraction pattern is accompanied by diffuse rings until only the diffuse rings appear for randomly oriented samples. The ratio of ring diameters, analogous to a Debye-Scherrer x-ray diffraction pattern, is used to identify the crystal structure. The advantage of SAD over x-ray techniques is the analysis of small specimens or individual grains.

Figure 15.31. Plane view TEM micrograph of the growth of polycrystalline CoSi2 on the open Si regions between patterned SiO2 lines. Upon high temperature annealing, the CoSi2 tends to agglomerate toward the center of the open region and away from the silicide/oxide boundary.

832


Figure 15.32. Diagram of the electron scattering that occurs from single crystalline, polycrystalline, and randomly oriented films. (Reprinted with the permission of the publisher, Marcel Dekker, Inc., Ref. 36.)

Compositional analysis in a TEM may be performed using x-ray analysis or electron energy loss spectroscopy (EELS). A schematic illustration of an EELS apparatus is shown in Fig. 15.33. Electrons in the transmitted beam may suffer only inelastic scattering. EELS involves analysis of the energy distribution of the inelastically scattered electrons


833

contained in the transmitted beam along with the unscattered electrons. A magnetic sector spectrometer is used to energy-analyze the electrons based on their radial trajectory in the magnetic field of the spectrometer. A magnetic instrument is utilized since it is the only type of electron spectrometer with the resolving power to handle the high electron energies necessary for STEM analysis. The high energy primary electrons lose energy passing through the sample due to ionization of the energy levels of atoms present. This results in loss peaks at discrete energy levels. During EELS all the inelastically scattered electrons are detected so that the signal intensity should be higher than the corresponding x-ray intensity. In addition, the spatial resolution should approach the diameter of the incident beam. The physics of energyloss favors strong EELS signal generation for light elements, which is complimentary to EDS analysis which is insensitive to light elements.

Figure 15.33. Schematic diagram of a magnetic sector electron energy loss spectrometer.

834


X-ray analysis in the STEM is accomplished with essentially the same equipment configuration as in the SEM. However, EDS in the STEM provides much higher spatial resolution than in an SEM. The higher spatial resolution is accomplished as a result of sample thinning. The electron beam interaction with the bulk samples used in SEM analysis results in electron scattering which results in secondary x-ray excitation from a sample volume much larger than would be excited by the primary beam without scattering (Fig. 15.23). When the sample is thinned to several thousand angstroms thickness, the electron beam penetrates without significant scattering. Since the volume scattering is minimized, x-ray analysis may be accomplished in a volume not much larger than that defined by the diameter of the primary electron beam.

4.0 OPTICAL ANALYSIS TECHNIQUES 4.1 Ellipsometry Ellipsometry is an optical technique which is widely used to characterize the optical properties such as refractive index, thickness, surface roughness, etc., of thin films. It is based on the interaction of linearly-polarized monochromatic light with materials. When light passes from one medium into another (in ellipsometry, typically from air into the sample) some of the light is reflected and some passes into the material as shown schematically in Fig. 15.34.[37] The angles of the incident and reflected light are equal. The portion of the light that enters the sample does not continue at the same angle but is refracted to a different angle. The angle of refraction can be determined by Snell’s Law where: Eq. (4)

N1 sin ∅1 = N2 sin ∅2

where N1 and N2 are the indices of refraction (complex numbers) of air (or medium of the incident beam) and the substrate, respectively and Ø1 and Ø 2 are the angles of incidence and angle of refraction, respectively. For dielectrics, Eq. 4 consists of only real numbers. Ellipsometry invariably involves the reflection of light from a surface. When a light beam is reflected at an interface, the reflection coefficient is defined as the ratio of the amplitude of the reflected wave to the amplitude of the incident wave for a single interface (Fresnel reflection coefficient).


835

Figure 15.34. Schematic of the planar structure assumed for ellipsometric analysis showing a collimated monochromatic beam of light interacting with a surface at the air/medium interface. The electric vectors of the plane polarized light are defined as p waves (in the plane of incidence) and s waves (perpendicular to the plane of incidence). (After Woollam et al.)[37]

Reflection coefficients are defined in terms of the electric vectors of the plane polarized light; p waves in the plane of incidence and s waves perpendicular to the plane of incidence. The Fresnel reflection coefficients are given by[38]

Eq. (5a)

p r12 =

N2 cos ∅1 − N1 cos ∅2 N2 cos ∅1 + N1 cos ∅2

s = r12

N1 cos ∅1 − N2 cos ∅2 N1 cos ∅1 + N2 cos ∅2

and Eq. (5b)

where the superscript refers to waves parallel or perpendicular to the plane of incidence and the subscripts refers to medium 1 and medium 2. In ellipsometry the ratio of rs and rp is measured. Since these are both complex numbers their ratio is complex and is expressed in terms of amplitude and phase:

836

Eq. (6)


rp s = tanΨexp ( j∆ ) r

whereΨ (Psi) and∆ (Delta) are the parameters determined by the ellipsometer from which one of the parameters of interest, the index of refraction, can be determined. Delta denotes the difference between the phase angle between the parallel and perpendicular component of the incoming wave, δ1, and the outgoing wave, δ2, respectively. Psi is the angle whose tangent is the ratio of the magnitudes of the total reflection coefficients. There are a variety of ellipsometer configurations; these include null ellipsometers, modulation ellipsometers, and rotating element ellipsometers. A typical null instrument, for example, is shown in Fig. 15.35. When the analyzer is rotated about the light beam, a sinusoidally varying intensity will be detected. When the polarizer is properly oriented so that the ellipticity is just canceled by the reflection, the light being detected will be linearly polarized and the signal intensity will be at its maximum and minimum. The proper orientation is found by adjusting alternately the polarizer and the analyzer until the true extinction is found.

Figure 15.35. Schematic diagram of a rotating-element ellipsometer in which the polarizer and analyzer rotate alternately until the null is found.


837

Ellipsometry has been used to determine the refractive index and thickness of numerous materials, in both single and multiple layers. In many instances the ellipsometry data is compared with other data or used to calibrate other techniques. For example, Fig. 15.36 shows a plot of the TiO2 thickness as determined by ellipsometry versus the corresponding values determined by AES for a thermally oxidized TiNx film.[39] To determine the thickness of the oxide film, it is necessary to determine the optical properties of the underlying TiNx; thermal annealing (as in oxidation) will change the optical properties of the TiN layer. AES analysis was used in an iterative approach to determine the ∆ and Ψ values of the oxide film.

Figure 15.36. Plot of the TiO 2 film thickness for a thermally oxidized TiN x measured by ellipsometry versus that measured using AES. (Reprinted with permission of the publisher, the Amer. Inst. of Phys.)

The basic equations developed for ellipsometry assume plane parallel surfaces. However, spectroscopic ellipsometry can be used to provide insight regarding microscopically rough surfaces. In film growth, several possibilities exist: the substrate/film interface can be rough, and/or the

838


film/ambient interface may be rough. Without the use of other ancillary techniques, the roughness can not be determined precisely. Some simplifying assumptions that may be made in order to determine film thickness include: measure the∆/Ψ values of a film-free rough surface, use these values to calculate an index for the substrate, and assume that the film growth results in a single film with plane parallel interfaces; or use the true index values of the bulk material determined by some other means and assume that the film growth yields a single film with parallel interfaces. The error in thickness determined using either set of assumptions is less than the amplitude of the roughness.[38] With the advent of the use of plasmas or ion beams in microelectronic processing, there has been concern for the damage the energetic ions or neutrals may cause to the substrate. The damage may be monitored nondestructively using spectroscopic ellipsometry[41]which provides information regarding the damage profile, thickness, and the degree of crystallinity as well as the presence of an oxide and microroughness. These studies were conducted for variable energy at constant dose and for varying dose at constant energy. For low energy ion implantation, the damage was modeled as a silicon film amorphized by ion implantation on the surface of a crystalline substrate. For higher energy implants, the amorphous Si layer became buried beneath a damaged crystalline film.[42] 4.2 Fourier Transform Infrared Spectroscopy The significant improvements in infrared spectroscopy brought about by the introduction of computerized Fourier transform infrared (FTIR) spectrometers have resulted in a dramatic expansion in the application of this technique. One version of FTIR spectrometer is shown schematically in Fig. 15.37. Radiation from the IR source passes through a beam splitter onto a fixed and movable mirror. The IR radiation is recombined in an interference pattern which is determined by the position of the movable mirror. Infrared radiation transmitted through the sample (as shown in Fig. 15.37) or reflected from the surface is detected and a plot of IR absorption versus wavelength is generated. FTIR is used for both bulk and thin film analysis. When used for bulk analysis, FTIR is one of the most sensitive analytical techniques for selected materials. Table 15.6 provides the strongest absorption lines, frequencies, and sensitivities of selected impurity elements in Si. [43] In order to achieve these sensitivities, it is necessary to cool the sample to liquid helium temperature and use a relatively thick specimen (5 mm) since


839

the sensitivity is proportional to the optical path length. Since C and O are not electrically active in Si, no significant gain in sensitivity for these impurities is obtained by lowering the sample temperature. Thinner specimens may be used with a proportionate drop in sensitivity. Quantitative analysis is based upon the measured IR absorption at a characteristic wavelength compared to standards or a known absorption coefficient. Since impurities such as O and C can occupy different sites in the Si lattice, they exhibit several IR absorption bands. The frequencies of these bands have been used to determine the presence of interstitial and precipitated O and the intensities provide a measure of the concentration.[44]

Figure 15.37. Schematic diagram of a Fourier Transform infrared spectrometer.

One unique aspect of FTIR is the capability for measuring epitaxial layer thicknesses of Si. This is a challenge for most techniques since the film and substrate are the same material. Epitaxial Si layers are typically lightly doped and will transmit in the IR range of 2 - 50 micrometers. The substrate is heavily doped and will reflect the IR radiation. Depending on the difference in doping levels between the epitaxial Si and the substrate, multiple internal reflections will occur before the radiation reaches the detector. At each reflection, the IR will undergo a phase shift which will be different for the epi-air interface and the epi-substrate interface. The resulting spectrum will show interference fringes with a period which is related to the epitaxial film thickness.[45]

840


Table 15.6. FTIR Absorption Line Frequencies and Sensitivities

Diamond-like amorphous carbon (DLC) films have a number of attractive properties such as hardness, chemical inertness, electrical insulation, and infrared transparency. DLC films are believed to consist of a mixture of sp2 and sp3 bonding structures. IR spectroscopy can easily distinguish these two bonding structures by peak positions of characteristic absorption bands. Table 15.7 shows C-H stretching absorption bands and their assignment for DLC films.[46] Numerous other applications may be found in the literature where FTIR has been used in both the transmission mode or reflection mode to determine H content on plasma-deposited films, moisture adsorption and others thin film properties. The first FTIR microscope accessory was introduced in 1983. Since then this capability has grown rapidly to provide analysis of areas as small as 5 x 5 micrometer in both the transmission and reflection mode. The FTIR microscope sampling technique has been used to determine the B and P concentrations in borosilicate[47] and phosphosilicate[48] passivation layers,


841

H concentration in silicon nitride passivation layers, and the carbon impurity concentration in GaAs.

Table 15.7. C-H Stretching Bands Observed for Diamond-like Films

4.3 Photoluminescence Spectroscopy Photoluminescence (PL) spectroscopy is a measure of the intensity of radiation versus wavelength emitted as a result of radiative recombination of electron-hole pairs or excitons from their thermal equilibrium states by optical excitation.[49] An electron-hole pair excited from the ground state can recombine radiatively through various kinds of recombination processes as shown schematically in Fig. 15.38. The most simple recombination process is a band-to-band recombination where a free electron excited in the conduction band recombines radiatively with a free hole excited in the valence band. Impurities which introduce traps, donors or acceptor levels in the band gap provide alternate paths for de-excitation. When an electron or hole is captured by a trap center and then the trapped carrier recombines radiatively with the remaining carrier, this is called band-to-impurity recombination. When both the excited electron and hole are captured by different trap centers and then the trapped electron and hole recombine radiatively, this is known as donoracceptor pair recombination.

842


Figure 15.38. Diagram of the possible photoluminescence transitions.

At low temperatures a generated electron-hole pair becomes an exciton. An exciton is a complex with an electron and hole bound together by a Coulomb attraction which can move freely as a quasiparticle in a semiconductor crystal. These free excitons decay in the ground state through free-exciton (EF) recombination accompanied by luminescence. Impurity-exciton complexes are formed when free excitons are bound to impurity centers. Bound excitons (EB) radiatively decay at just below the free-exciton energy. It is apparent that the primary application of PL is in the analysis of semiconductors impurities and defects. The most effective application of PL is the identification of shallow impurities. This is accomplished by measuring the characteristic positions of the EB luminescence lines at low temperature. The spectral positions will differ depending on the impurity, while the intensity is related to the concentration. PL has been used in the analysis of elements such as B, P, Al, As, and N in Si in the concentration range 1011 to 1015 atoms/ cm3. It has been used to study impurities such as C, Si, Mn, Mg, and Te in GaAs to 1013 atoms/cm3.[50] The PL intensity is not directly related to shallow-impurity concentration because of competing non-radiative decay processes for the EB. The intensity also depends on the excitation level. It has been found empirically that good correlation can be obtained between impurity levels deter-


843

mined by electrical measurements and the intensity ratio of the EB to EF when recorded at moderate excitation levels. Measurement of the intensity ratios minimizes the influence of variables dependent on the crystal growth and process conditions. Tajima generated the calibration curves shown in Fig. 15.39 for B and P in float-zone refined Si.[51] The concentration range between 1011 to 1015 atoms/cm3 represents the practical range over which PL may be applied to Si.

Figure 15.39. Calibration curves for the P and B concentrations in Si from analysis of the EB and E F photoluminescence intensity ratios. (Reprinted with the permission of the publisher, the Amer. Inst. of Phys.)

844


Room temperature PL due to band-to-band recombination can be used to characterize thermally induced defects in Si. Some thermally induced defects in Si act as non-radiative recombination centers which trap excess carriers. The presence of such non-radiative recombination centers leads to a reduction in the PL intensity. It is believed that the thermally induced defects are related to oxygen precipitates since a strong correlation was found between the etch-pit density and the PL intensity.[52] PL is a non-destructive technique which requires minimal sample preparation. It is restricted to analysis of single crystalline wafers or epitaxial layers. The sampling depth is approximately three micrometers, the optical attenuation length. Through the use of laser excitation spatial resolution of one micrometer can be achieved which may be used to map the PL intensity distribution over a surface. 5.0 CONCLUSION A wide variety of analytical techniques are available for the analysis of thin films. The ones described in this chapter represent some of the most widely used; however, there are many others that provide unique capabilities not described here. This chapter is intended as a brief overview so references are included which provide more detailed information about the analytical tools described here, as well as other related techniques.


845

REFERENCES 1.

McGuire, G. E. and Holloway, P. H., Electron Spectroscopy: Theory, Techniques and Applications, (C. R. Brundle and A. D. Baker, eds.), Vol. 4, Academic Press, New York (1981)

2.

Gryzinski, M., Phys. Rev., A138:336 (1965)

3.

Harris, L. A., J. Appl. Phys., 39:1419 (1968)

4.

Seah, M. P. and Dench, W. A., Surf. Interface Anal., 1:2 (1979)

5.

Tanuma, S., Powell, C. J., and Penn, D. R., J. Elec. Spect. Relat. Phenom., 52:285 (1990)

6.

Zalar, A., Surf. Interface Anal., 9:41 (1986)

7.

Schwartz, G. P., Gualtieri, G. T., Kammlott, G. W., and Schwartz, B., J. Electrochem. Soc., 126:1737 (1979)

8.

Thomas, M. T., Baer, D. R., Jones, R. H., and Bruemmer, S. M.,J. Vac. Sci. Technol., 17:25 (1980)

9.

Seah, M. P., Surf. Interface Anal., 9:85 (1986)

10.

Hofmann, S.,Proc. 6th Int. Symp. High Purity Materials in Science and Technology, (A. Drescher, ed.), 2:149-169, Akad. d. Wiss. d. DDR, Dresden (1985)

11.

Practical Surface Analysis: Auger and X-ray Photoelectron Spectroscopy, (D. Briggs and M. P. Seah, eds.), John Wiley and Sons, New York (1990)

12.

Wagner, C. D., Riggs, W. M., Davis, L. E., Molder, J. F., and Muilenberg, G. E., Handbook of Photoelectron Spectroscopy, PerkinElmer Corp., Minnesota (1978); Moulder, J. F., Stickle, W. F., Sobol, P. E., Bomben, K. D.,Handbook of X-ray Photoelectron Spectroscopy, Perkin-Elmer Corp., Minnesota (1992)

13.

Swartz, W. E., Wynne, K. J., and Hercules, D. M., Anal. Chem., 43:1884 (1971)

14.

Carlson, T. A.,Photoelectron and Auger Spectroscopy, Plenum Press, New York (1975)

15.

Wagner, C. D., Davis, L. E., Zeller, M. V., Taylor, J. A., Raymond, R. M., and Gale, L. H., Surf. Interface Anal., 3:211 (1981)

16.

Hofmann, S., J. Vac. Sci. and Technol., BIO, 316-322 (1992)

17.

Lee, J. J., Fulghum, J. E., McGuire, G. E., Ray, M. A., Osburn, C. M., and Linton, R. W., J. Vac. Sci. Technol., A8:2287 (1990)

18.

Stevie, F. A., J. Vac. Sci. and Technol., BIO, 323-328 (1992)

846


19.

Anderson, C. A. and Hinthorne, J. R., Anal. Chem., 45:1421-1438 (1973)

20.

Coleman, D. J., Shaw, D. W., and Dobrott, R. D.,J. Electrochem. Soc., 124:239-241 (1977)

21.

Becker, C. H. and Gillen, K. T., J. Vac. Sci. Technol., A3:1347 (1985)

22.

Arlinghaus, H. F., Spaar, M. T., Thonnard, N., McMahon, A. W., and Jacobson, K. B., Optical Methods for Ultrasensitive Detection and Analysis: Techniques and Applications, (B. L. Fearey, ed.), 1435:26, SPIE (1991)

23.

Mayo, S., Lucatorto, T. B., and Luther, G. G.,Anal. Chem., 54:553-556 (1982)

24.

Chu, W. K., Mayer, J. W., and Nicolet, M. A., Backscattering Spectrometry, Academic Press, New York (1978)

25.

Feldman, L. C., Mayer, J. W., and Picraux, S. T.,Materials Analysis by Ion Channeling, Academic Press, New York (1982)

26.

Lien, C. D., Nicolet, M. A., and Lau, S. S., Phys. Stat. Sol., 81:123-128 (1984)

27.

Helms, C. R., J. Vac. Sci. Technol., 20:950 (1982)

28.

Narayan, J. and Holland, O. W., J. Electrochem. Soc., 131:2651-2662 (1984)

29.

Principles of Analytical Electron Microscopy, (D. C. Joy, A. D. Romig, Jr., J. I. Goldstein, eds.), Plenum Press, New York (1986)

30.

Koshikawa, T. and Shimizu, R.,J. Phys. D: Appl. Phys., 7:1303 (1974)

31.

Heinrich, K. F. J., X-Ray Optics and Microanalysis, (R. Castaing, P. Deschamps, and J. Philbert, eds.), p. 1509, Paris (1966)

32.

Newbury, D. E. and Yakowitz, H., Practical Scanning Electron Microscopy, (S. I. Goldstein and H. Yakowitz, eds.), Plenum Press (1975)

33.

Materials Research Society Symposium Proceedings, (J. C. Bravman, R. M. Anderson, and M. L. McDonald, eds.), 115, (1988)

34.

Hsia, S., Tan, T. Y., Smith, P. L., and McGuire, G. E., J. Appl. Phys., 70(12):7579-7587 (1991)

35.

Osburn, C. M., Wang, Q. F., Kellam, M., Canovai, C., Smith, P. L., McGuire, G. E., Xiao, Z. G., and Rozgonyi, G. A., Appl. Surf. Sci., 53:291-312 (1991)

36.

Murr, L. E., Electron and Ion Mircoscopy and Microanalysis, Marcel Dekker Inc., New York (1982)


847

37.

Woollam, J. A., Snyder, P. G., and Johs, B., Characterization of Semiconductor Materials, Vol. II, Noyes Publications, Park Ridge, NJ (1993)

38.

Tompkins, H. G.,A Users Guide to Ellipsometry, Academic Press, San Diego, CA (1993)

39.

Tompkins, H. G., J. Appl. Phys., 70:3876 (1991)

40.

Aspnes, D. E., Theeten, J. B., and Hottier, F., Phys. Rev., B20:3292 (1979)

41.

Burns, T. E., Chongsawangvirod, S., Andrews, J. W., Irene, E. A., McGuire, G. E., and Chevacharoenkul, S., J. Vac. Sci. Techn., B9:4149 (1991)

42.

Hummel, R. E., Xi, W., Holloway, P. H., and Jones, K. A., J. Appl. Phys., 63:2591 (1988)

43.

Baber, C. S., Thin Solid Films, 72:1332 (1980)

44.

Chen, C. S. and Schroder, D. K., Appl. Phys., A42:257 (1987)

45.

Schumann, P. A., Jr. and Schneider, C. P., J. Appl. Phys., 45:3532 (1970)

46.

Dischler, B., Bubenzer, A., and Koidl, P.,Solid State Commun., 48:105 (1983)

47.

Krishnan, K. and Kuehl, D. T.,ASTM Spec. Tech. Publ., 850:325 (1984)

48.

Zearing, D. J. and Coates, V. J., Proc. SPIE, 276:249 (1981)

49.

Nishino, T., Nakayama, H., and Hamakawa, Y.,Japan Annual Reviews in Electronics, Computers, and Telecommunications: Semiconductor Technologies, (J. Nishizawa, ed.), 8:295, North Holland, NY (1985)

50.

Smith, K. K., Thin Solid Films, 84:171 (1981)

51.

Tajima, M., Appl. Phys. Lett., 32:719 (1978)

52.

Katsura, J., Nakayama, H., Nishino, T., and Hamakawa, Y., Jpn. J. Appl. Phys., 21:712 (1982)

16 Jet Vapor Deposition Bret L. Halpern and Jerome J. Schmitt

1.0

INTRODUCTION

Modern coating technology must accomodate demands for increasingly complex materials, more economical throughputs, and tighter environmental constraints. Issues of versatility, speed, cost and cleanliness challenge established physical and chemical vapor deposition methods, and require innovative alternatives. A powerful response to those challenges has emerged in the Jet Vapor Deposition (JVD) concept, a patented, proprietary and general approach to thin film deposition.[1]-[3] JVD can generate thin and thick films of unlimited chemical identity at high rate with negligible environmental threat.[4]-[9] The key innovation is a novel vapor source: a “sonic jet in a low-vacuum fast flow” which transports condensible atomic, molecular or cluster-laden vapor to a substrate. The potential of such a jet for efficient deposition, first discerned by Halpern[1] and Schmitt,[2] has been swiftly evolving at Jet Process Corporation, with useful implications at the levels of both manufacturing efficiency and microscopic film quality control. The jet source is the fundamental element in a “multiple jet, moving substrate” strategy[3] for depositing metals, semiconductors, dielectrics, oxides, nitrides, and organics. These can be grown in many forms: multicomponent, alloy, multilayer, “host-guest” and “cluster-embedded” films. The growth rates are high, approaching microns per minute over areas of several hundred cm2, even on room temperature substrates, with potential for scale-up. Novel coatings

848

Jet Vapor Deposition

849

are made as easily in high-throughput commercial production as in small-scale basic research, with neither toxic precursors nor effluent. At this writing, Jet Process Corporation has devised singular approaches to film property control, based on gasdynamic and energetic ion effects, which have been made possible by the unique “high” pressure conditions in JVD. In brief, JVD’s linkage of high speed gasdynamics and low cost, low vacuum technology enables synthesis of unusual, complex materials for new applications in nonlinear optics, integrated electronics, and surface protection. The “jet in low vacuum” is deceptively simple, but it has a combination of fortunate characteristics that provide the core of JVD’s versatility; we trace that relationship in this chapter. We present a semi-quantitative discussion of jet structure and behavior and of vapor transport; we show how use of multiple jets in concert with relative jet-substrate motion leads to synthesis of multicomponent films. We then summarize several applications of JVD.

2.0

PRINCIPLES AND APPARATUS OF JVD

The most important elements in a JVD system are the jet source and substrate motion mechanism; these are mounted in a low pressure deposition chamber in which a fast flow of gas is sustained by a mechanical pump.[4][5] The jet source is based on a nozzle, made of metal or glass, having an exit diameter Dn from several mm to 2 cm. Helium or other inert gas is supplied to the nozzle and exits from it as a jet. Ordinarily the nozzle pressure P n is several torr, and the downstream pressure P d is a torr or less, but the range may be wide: in a few applications both pressures can be ten times higher. When the ratio Pn /P d > 2, the flow is “critical”, and the jet emerges at its maximum velocity, the speed of sound; for He at 298°K, the exit velocity is ~105 cm/sec. The structure and operating parameters of a generalized jet source are shown in Fig. 16.1. The jet can convey any atomic, molecular or cluster vapor to a substrate for condensation as a film. The vapor source is placed in the nozzle within several nozzle-diameters of the exit, a region in which the He carrier velocity is nearing the speed of sound. Vaporization can be effected by any convenient technique, such as thermal evaporation, glow discharge sputtering, microwave or DC discharge reaction, and laser ablation. An atom of gold, for example, injected upstream into the He flow will be swiftly

850


captured and transported downstream by the He jet. If a flat substrate is placed perpendicular to the jet, a bright gold deposit quickly appears on it; virtually all the gold atoms deposit in a circular zone comparable to the nozzle exit area. The deposit is symmetric but nonuniform, being thicker toward the center.

Figure 16.1. A generalized "jet in low vacuum" JVD source showing representative operating conditions. Both jet velocity and density change at the Mach disc, but the jet remains nearly cylindrical. At the substrate the jet transforms sharply to a thin, radially flowing wall jet, but deposition remains localized. Flow is sonic at the nozle exit, subsonic after the Mach disc.


851

In order to deposit uniform films over larger areas, either the substrate, nozzle, jet, or some combination of them must be moved.[4][5] Although Jet Process Corporation has implemented all these possibilities, this chapter will consider only substrate motion. This is accomplished as in Fig. 16.2; substrates are mounted on a “carousel” which can both spin rapidly around and translate slowly along its axis; the carousel motion is computer controlled, and various motions can be programmed. The jet is aimed radially at the carousel. If the carousel is only spinning, a band of deposition appears around it; if the carousel is also vertically “scanned” at constant speed, the band is broadened to cover the entire carousel surface, and that of any substrate on it. The deposit thickness is uniform because all areas are exposed for equal times to a constant jet flux.

Figure 16.2. Different motions of the carousel yield different deposition patterns. The combination of spin and oscillation produces uniform coverage over large areas.

852


These observations suggested that several jets can be aimed at the carousel and their contributions “integrated” even at high overall growth rate. Jets operated singly in a prescribed time sequence yield multilayer structures; jets operated together give alloys or “doped” films, or, in a “reactive” mode, compounds such as oxides and nitrides. This “multiple jet, moving substrate” strategy , seen in Fig. 16.3, has proven very successful in research and production at Jet Process Corporation; below we examine its underlying basis.

Figure 16.3. A variety of multicomponent, multilayer, host-guest and clusterembedded films can be made by operating jets simultaneously or in sequence.

Jet Vapor Deposition 3.0

DISCUSSION

3.1

Jet Structure, Behavior, and Vapor Transport

853

As background to multiple jet JVD synthesis, it is useful to describe the structure of a single helium “jet in low vacuum.” The contours of the jet can be illuminated by visible light emission from gas phase glow discharges or chemiluminescent reactions. The structure of free jets has been amply described,[10] but particularly useful features arise under JVD conditions. As soon as the jet exits the nozzle, it expands into a zone of reduced density in which nearly all atoms move in the same direction at the speed of sound. This zone is collision free; it terminates at the Mach disc, located a distance x ≈ 0.67 Dn (Pn/Pd)½ downstream of the nozzle of diameter Dn .[10] In JVD the usual pressure ratios lie in the range 2 < Pn /Pd < 10, and the Mach disc is located several nozzle-diameters, or several cm, downstream. Beyond the Mach disc, the density rises sharply to the downstream background value, the jet speed drops to ~ 104 cm/sec, and collisions again occur in the jet. Despite this variation in density, our visualization experiments confirm that the jet diverges little, and remains almost cylindrical over distances of tens of centimeters downstream of the nozzle. On impact with the substrate, the “free” jet flares radially into a “wall jet,” resembling a laminar stream of water impinging on a plate. Our visualization experiments show that the transition from free jet to wall jet is sharp; the thickness of the wall jet is only a few mm for a 1 cm diameter free jet. Both free jet and wall jet are laminar, and there are no turbulent regions. The above observations show that JVD’s “jets in low vacuum” have a relatively simple form, despite variations in density and speed before and after the Mach disc. Given this background, we single out for discussion the following features which are key to multiple jet, moving substrate synthesis: 1. The jet is fast, collimated, well defined and delivers vapor efficiently; deposition is localized and film growth is fast. 2. The wall jet is thin; its radial flow does not greatly broaden the deposit. 3. Substrate transverse velocity is high even for moderate carousel rotation frequencies. The short residence time of the substrate as it passes through a jet assures accumulation of less than a monolayer and minimizes heating.

854


The strong collimation of the jet is important in JVD. It can be understood by examining the “random walk” of a helium atom at the jet boundary after it emerges from the nozzle and travels with the jet at speed v to the substrate a distance L downstream. During its transit time τ = L/v, a helium atom diffuses radially a distance x given by: x2 ≈ 2Dτ ≈ 2DL/v where D is the diffusion coefficient; we take v as the jet speed downstream of the Mach disc. For L ≈ 10 cm, v ≈ 104 cm/sec, and D≈ 600 cm2/sec (He at 1 torr), the diffusion distance is 6 mm; it diminishes at higher pressure. This is consistent with visualization experiments: axial transport is much faster than radial diffusion, and the jet remains a collimated, spatially distinct source independent of other jets. A similar conclusion holds for heavy species injected into the light carrier jet. For example, in a JVD “wirefeed” jet, Au wire can be vaporized from a “point” source lying on the nozzle axis. A gold atom injected on the jet axis will have diffused only a few millimeters from the axis by the time it arrives at the substrate. The small thickness of the wall jet is also critical. When the gold atoms arrive near the substrate, it might be expected that the wall jet would carry many of them away, as well as broaden the deposit. However, these effects are minimized because the wall jet is thin; gold atoms entrained in the wall jet flow diffuse to the substrate and deposit before they are transported far, and few are lost. The small thickness of the wall jet can be understood by the following rough argument. The jet impact zone can be regarded as a cylindrical “pillbox;” the jet enters the top, and the wall jet exits through the cylindrical side. The impact of a jet at 1 torr and 10,000 cm/sec results in negligible pressure change compared to 1 torr; therefore, by Bernoulli’s principle, the speed of the jet entering the pillbox and the speed of the wall jet leaving it must be nearly equal. Continuity then determines the relative areas of the top and side of the pillbox, and requires that the wall jet thickness be approximately one-quarter of the jet diameter. The large inertia of gold atoms entrained in the jet also favors localized deposition. When the axial He jet transforms sharply into a radial wall jet, heavy gold atoms “turn the corner” only with difficulty, and tend to move straight toward the substrate.[11] We have not yet determined the


855

relative importanceof diffusion and inertia in the wall jet region. However, we have measured the deposit profile of gold atoms injected along the jet axis and find it to be Gaussian with a half-width less than the jet radius.[12] We have also shown by microbalance measurements that 95% of the wire vaporized in the nozzle is deposited on the substrate.[12] Accordingly, jet collimation and localized deposition give JVD the “line of sight” characteristics of high vacuum vaporization. The efficiency is far higher, however: all vaporized material goes in one direction rather than many; it is confined to the jet, and most of it deposits in the impact zone. The independence of the jets is an invaluable feature of JVD. Not only are the jets spatially distinct, but the operation of one does not alter the upstream nozzle conditions of any other. As long as Pn/Pd exceeds ~2, Pn remains constant despite changes in Pd . The jets all emerge at the speed of sound, and neither “information” nor mass can propagate upstream. An example of the benefits: we vaporize many metals in a “glow discharge sputter jet” and oxidize the growing metal film with O2 or O atoms injected downstream. The metal sputtering target in the nozzle is shielded from the oxidants downstream by the sonic gas flow out of the nozzle; a comparable process in high vacuum PVD is difficult to carry out. Controlled deposition of multicomponents is a direct consequence of jet independence. The wide range of workable jet pressures in JVD is useful in syntheses involving metal atoms: we have used JVD to deposit single atoms or clusters. At the lower range (several torr), only single metal atoms deposit. Cluster formation must be initiated by three body collisions involving two metal atoms and a helium atom. While we cannot rule out contributions of heterogeneous processes in the nozzle,[13] three body collisions are highly improbable in the jet: the metal atom and He concentrations are too low, and the transit times from nozzle to substrate are too short. The time for a metal atom to undergo a three body collision, using a typical three body recombination rate constant,[14] is τ3B ≈ 1032/(M)(He) ≈ 0.1/PmPHe where (M) and (He) are gas phase concentrations (#/cm3) and Pm, PHe the pressures in torr. For Pm = 0.01 torr and PHe = 1 torr, τ3B ≈ 10 seconds; this is much longer than the transit time from nozzle to substrate. Only at much higher pressures of several tens of torr does cluster growth becomes dominant in the jet.

856 3.2

Deposition Technologies for Films and Coatings Substrate Motion

Carousel and substrate motion makes it possible to react materials from different jets. For reaction to be efficient, “micromixing” must be carried out at the sub-monolayer level. For example, to make lead zirconate titanate (PZT) using three metal jets, we require that less than one monolayer of any metal be deposited on any substrate during its time of passage through a jet. Given the diameter D of the jet, the height H of the carousel, the monolayer thickness d, and the thickness rate of change dz/dt over the entire carousel, we can calculate the approximate carousel rotation frequency fmonolayer that assures monolayer per pass coverage: fmonolayer ≈ (dz/dt) (H/D)(1/δ) In the case of PZT we have dz/dt = 5000 A°/hr, H = 10 cm, D = 1 cm, and δ = 3A°; the required rotation frequency is f ≈ 5 Hz. Micromixing is assured by this easily attained frequency. Micromixing is indispensible in JVD. It enables reaction of many components to yield complex materials, results in better film uniformity, and can reduce required substrate temperatures as well as annealing times after deposition. It assures that transport limitations in the growing solid are as absent as they are in the high speed jet. Micromixing and multiple jet, moving substrate synthesis are exploited in a number of JVD applications described later. Common to many JVD applications is an environmentally sound strategy: volatilization of the elemental metal, deposition of metal atoms from one or more jets, and conversion of the growing film to oxide or nitride, layer by layer, with a flux of O or N. Since most JVD metal sources are based on techniques such as glow discharge sputtering or direct vaporization, capable of depositing nearly every metal in the Periodic Table at high rate, no toxic metal precursors or harmful exhausts need ever be involved. The “jet in low vacuum” strategy also has a decisive manufacturing advantage: it is economical to implement. Even relatively small, inexpensive mechanical pumps will maintain critical flow conditions, providing high speed collimated jets in a small footprint, flexible apparatus. The batch process turnaround times in a JVD chamber are matters of minutes, negligible in comparison with non-load locked high vacuum systems. The carrier gas flows represent only a small material cost, and in the range of several torr, even <<1% entrained condensible vapor corresponds to an economically high deposition rate.

Jet Vapor Deposition 4.0

EXAMPLES OF JVD FILMS AND APPLICATIONS

4.1

Cu, Au Multilayer Electrodes; Al, Al2 O3 Microlaminates.

857

Cu and Au can be thermally vaporized and deposited at high rates using JVD “wirefeed” sources such as that in Fig. 16.4. Areas approaching 1000 cm2 can be uniformly coated to a depth of one micron in one minute by means of Cu and Au jet sources that dissipate only 60 watts. Structures containing alternating layers of Cu and Au are made with two jet sources, operated alternately for appropriate times.

Figure 16.4. Wirefeed jet vapor source. The wire feed rate is computer controlled; this allows the deposition rate to be determined and varied. For some metals, the tungsten filament must have a protective sheath to avoid alloying.

858


Control of Cu/Au layer thickness via the wirefeed approach is precise. We verified this by depositing one hundred alternating layers of Cu and Au, each intended to be 50 angstroms thick. Total thickness and uniformity were confirmed by stylus profilometry; the accumulated thickness was measured at 15 points over an area of 6 cm2 to be 4996 A ±10 A. Interface sharpness was verified, and layer thickness confirmed by low angle x-ray scattering.[15] Jet Process Corporation utilizes this high rate, multilayering capability in production runs for AT&T Bell Laboratories in a U.S. Navy application. The substrates are 50 mm diameter piezoceramic wafers less than 1 mm thick; these require a 1.5 micron Cu electrode, flashed with 50 nm Au, on both sides of the wafer, with a 0.25 mm border, free of metal and cleanly defined, at the wafer perimeter. Wafers are mounted on the carousel in accurately machined receptacles which serve to support the wafer, orient it toward the jets, and define the perimeter border. Wafers are processed at the rate of > 750 per week using a single JVD deposition chamber operated by one person. In this application JVD displaced an electrochemical technique both on grounds of quality and environmental concerns. Given that Au comprises a considerable part of the operating expense, the localization and efficiency of JVD is a notable advantage. We used a variation of this technique to deposit 100 micron thick “microlaminates” consisting of 50 nm layers of Al alternating with 5 nm of Al oxide.[9] This was done in ten minutes by means of a steady, high rate jet of Al (~0.1 cc/min of metal) into which oxygen was pulsed (slightly downstream of the nozzle) at appropriate intervals. Microlaminates having nanoscale component layers are expected to show enhanced strength and mechanical properties; JVD nanocomposites exhibited a hardness of ~ 2.5 GPa,[9] equal to that of microlaminates made by sputtering,[16] a far slower ultrahigh vacuum process. 4.2

PZT: Ferroelectric FRAM Nonvolatile Memories

Lead zirconate titanate (PZT) is a candidate for thin film memories. PZT can store charge at high density, and retain it in zero applied field; films of several thousand angstroms can switch states within the 5 volt range of computer power supplies. JVD is one of several processes (sol-gel, vacuum sputtering) now competing for this future market. In addition, PZT thin films are potentially useful for pyroelectric infrared detectors.


859

JVD employs four jet sources to supply Pb, Ti, Zr and oxygen to Ptcoated, heated Si wafers mounted on the spinning carousel, as seen in Fig. 16.3. The Pt barrier layer is also deposited by JVD. These jet sources build up a one micron film of PZT in less than one hour,[6] to give high quality PZT films at high rate. The ferroelectric perovskite phase appeared on deposition; after annealing, the film was entirely 100 oriented perovskite. Key parameters for effective PZT memory cells, and the value measured for JVD PZT films, are listed below: remanent polarization

6 - 20 microcoulombs/cm2

coercive field

60 kilovolts/cm2

switching endurance

> 1011 cycles

dielectric constant

> 1300

These values equal or exceed those obtained with more conventional methods. 4.3

Electronic Grade Silicon Nitride

Silicon nitride films of remarkable electronic quality were generated using a microwave discharge equipped jet source that produces Si atoms, Si bearing molecular fragments and N atoms. [7] A hydrogen atom jet, run for several minutes, removes thermal oxide from a silicon wafer; slow deposition on that substrate gave a nitride whose electrical behaviour in a metal-nitridesemiconductor (MNS) capacitor was superior to that of any previously reported.[7] In many respects, such as breakdown strength, radiation hardness, and interface trap density, these nitride films were equal orsuperior to the best thermal silicon dioxide grown at high temperatures (T > 1000°C). The etch rate in buffered oxide etch was a low 10 A/min; the index of refraction 2.03 at 632.8 nm, close to that for stoichiometric Si3N4 . But the outstanding fact is that this SiN was produced on a room temperature substrate. JVD silicon nitride shows great promise as a gate, as well as for passivation. The reason for this high quality is not fully understood; it is possible that excited species are transported at high speed from discharge to substrate where they liberate energy at the growing film surface to annihilate imperfections in it at a sufficiently low deposition rate. 4.4

Fiber Coating for Composite Materials

Coating of ultrafine fibers and multifilament tows is important for fiberreinforced composite materials; JVD exploits jet collimation in several

860


unique approaches to the problem. Jet conditions can be controlled to obtain uniform coating despite jet directionality; jet momentum is sufficient to disrupt and agitate fiber bundles and overcome shadowing of one fiber by another. In this way we have achieved uniform coating of 12.5 micron alumina fibers, singly and in bundles of several hundred, with Cu and Al metal several thousand angstroms thick, as verified by SEM. We have adapted “reel to reel” techniques to JVD for fiber coating. 4.5

Coating of Thermally Sensitive Membranes

The low substrate temperature capability of JVD is being exploited in the coating of Au and Pt as fine line electrodes on a 9-micron PVDF piezoelectric membrane. PVDF is wrapped around the carousel and covered with a 0.001" foil mask which defines the electrode pattern (0.25 mm wide line 10 cm long). The jet deposits a 2000 angstrom film through the pattern as the carousel spins. A similar line is deposited on the other side of the PVDF, perpendicular to the first electrode; the 1 mm x 1 mm intersection zone defines a capacitor whose output can be used to detect impinging sound waves.[17] PVDF is thermally sensitive, and can be depoled at T ~ 350°C; the low temperature capability of JVD is therefore critical. JVD also deposits the noble metals only on the mask/substrate area where they are needed. 4.6

“Ceramic Host–Organic Guest” Films

We have trapped complex “guest” organic molecules such as Rhodamine B and Methyl Red in a range of “host” ceramic films: SiO2, SiNx, Al2 O3 , and MgO. Co-deposition takes place at room temperature and with no degradation of the organic guest; for example, trapped Rhodamine still fluoresces under ultraviolet light. The guest concentrations are high; a one micron film of Methyl Red in silicon dioxide appears deep crimson, implying doping levels approaching a percent. Such host-guest films can have optical-electronic and thin film sensor applications. In a collaboration with Professor R. Zanoni and colleagues at Oklahoma State University, we demonstrated deposition, patterning and photobleaching of Methyl Red /ceramic host-guest films to make thin film wave guides. We have also observed that guest Methyl Red in microporous silicon dioxide changes color from red to yellow when exposed to vapors of HCl or NH3; Methyl Red is a well-known acid-base indicator, suggesting


861

application of JVD to sensors. Ceramic host–organic guest films can be made by mechanical or sol gel methods,[18] but these are multi-step processes, limited to soluble species, often including time-consuming thermal treatment. In JVD the host-guest combination is generated in minutes, at room temperature, and by a vapor deposition technique compatible with existing semiconductor microelectronic processing. 4.7

Polymer Deposition: Parylene

Parylene [poly(para-xylylene)] has properties such as high electric breakdown strength, impermeability to water, and biocompatibility which make it attractive for protective functions. Parylene can be vapor deposited by a unique mechanism in which di-para xylylene is cracked at high temperature, and convected slowly to a cold surface where polymerization then takes place; this is the “conventional” Gorham process.[19] Parylene’s main weakness is that it adheres poorly to surfaces when deposited in this way. However, we have observed that adherence and hardness were greatly improved by deposition from a sonic JVD source. In addition, a 2-micron JVD parylene film on a Pt wire survived 93 days in a soak test at 5 volt applied potential.[20] These results suggest that JVD parylene has an enhanced ability to withstand rigorous electrical and mechanical conditions.

5.0

SUMMARY

The use of single or multiple “jets in a low vacuum” coupled with “mobile substrates” makes Jet Vapor Deposition a flexible technology for a wide range of film applications. JVD links high speed gasdynamics with low cost, “low vacuum” equipment to give synthetic versatility at economic throughputs. Jets operating in the JVD pressure regime are collimated, intense sources of localized deposition. Jets are independent and noninterfering; a “multiple jet, moving substrate” strategy permits fluxes from different jets to be “micromixed” on moving substrates. Much of JVD’s versatility arises by coordinating several spatially separated, independent jets to give multicomponent, multilayer, and alloy structures, synthesized from component metals, semiconductors, dielectrics and organics. The possible material combinations are numerous and unconstrained by the identity of the components; the synthesis of known and potentially useful film materials is being systematically explored and commercially applied at

862


Jet Process Corporation. The range of applications is already wide and the technique is maturing rapidly.

REFERENCES 1.

Halpern, B. L., J. Colloid Interface Sci. 86:337 (1982)

2.

Schmitt, J. J., U.S. Patent No. 4,788,082 (11/29/1988)

3.

Schmitt, J. J. and Halpern, B. L., U.S. Patent 5, 256,205 (10/26/1993)

4.

Halpern, B. L., Schmitt, J. J., Golz, J. W., Johnson, D. L., McAvoy, D. T., Zhang, J. Z., and Di, Y., Proceedings of 35th Annual Technical Conference, Society of Vacuum Coaters (March 22-27, 1992)

5.

Halpern, B. L., Schmitt, J. J., Di, Y., Golz, J. W., Johnson, D. L., McAvoy, D. T., Wang, D., and Zhang, J.-Z.,Metal Finishing,(December 1992)

6.

Huang, C.-L., Chen, B. A., Ma, T. P., Golz, J. W., Di, Y., Halpern, B. L., and Schmitt, J. J., Ferroelectrics (March 1992)

7.

Wang, D., Ma, T. P., Golz, J. W., Halpern, B. L., and Schmitt, J. J., IEEE Electron Device Lett., 13:482 (1992)

8.

Zhang, J.-Z., McAvoy, D. T., Halpern, B. L., and Schmitt, J. J., Connecticut Symposium on Microelectronics and Optoelectronics (March 18-19, 1993)

9.

Hsiung, L. M., Zhang, J.-Z., McIntyre, D. C., Golz, J. W., Halpern, B. L., Schmitt, J. J., and Wadley, H. N. G., Scripta Metall.Mater., 29:293 (1993)

10.

Anderson, J. B., in:Molecular Beams and Low Density Gas Dynamics, (P. P. Wegener, Ed.), Chap.1, Marcel Dekker, New York (1974)

11.

Fernandez de la Mora, J., Halpern, B. L., and Wilson, J. A., J. Fluid Mech., 149:217 (1984)

12.

Golz, J., Johnson, D., Halpern, B. L., and Schmitt, J.J., in preparation

13.

Knauer, W., J. Appl. Phys., 62:841 (1987)

14.

Kerr, J. A. and Moss, S. J., CRC Handbook of Bimolecular and Termolecular Rate Constants, Vol. II, Table 197, CRC Press, Inc., Boca Raton, FL

15.

Spaepen, F., Professor, private communication

16.

Alpas, A. T., Embury, J. D., Hardwick, D. A., and Springer, R. W., J. Materials Sci.,25:1603 (1990)


863

17.

Everbach, C., Professor, private communication

18.

Avnir, D., Kaufmann, V. R., and Reisfeld, R., J. Non-Cryst. Solids, 74:395 (1985)

19.

Beach, W. F., Lee, C., Bassett, D. R., Austin, T. M., and Olson, R., Encyclopedia of Polymer Science and Engineering,2nd edition,17:990, John Wiley & Sons (1989)

20.

Edell, D., private communication

864


Index

A Abrasion resistance 555 Abrasive cleaning 121 Abrasive wear 782 Absolute characterization 672 Absorption bands C-H stretching 840 AC asymmetric 532 discharges 464 on DC 530 Accelerated adhesion testing 686 Accelerating potential in TEM 828 Acceleration factor 674 Accelerator tandetron 812 Acid hydrochloric. See HCl hydrofluoric. See HF Acid-base indicator 860 Acoustic emission 684, 686, 689 Actinometry 141 Activated Reactive Evaporation. See ARE Activated reactive evaporation 76

Activated reactive evaporation (ARE) 52 Activation energy barrier 491 of a surface 148, 537 overpotential 515 Activation barrier 711 Activity of gaseous species 406 Adatom diffusion lengths 723 diffusivities 747 migration 734 mobility 38, 741 nucleation 677 recoil 746 surface diffusivity 710 Adatoms 438 Additives 526 Adhesion 409, 682 loss of 685 of a deposited film 376 of deposits 571 testing 686 Adhesion tests 687 Adhesive wear 781 test 656 Adsorption 95

864

Index

AE transition 791 Aeration cell 769 AES detection limit of 796 spectrum of Si with AlCu 793 Agitation 528 Air and airless spraying 51 Al deposition 858 conditions 774 Al-Zn corrosion resistance 775 Alkaline cleaners 122 Alloy deposition of 201, 202, 543 Alloys 852 advantages of 550 binary and ternary 544 by JVD 848 Altered layer 293, 369 Altered region 366 Alumina fibers 860 Aluminum oxide coatings 664 Ambipolar diffusion 74, 462 Amorphous coatings 545 Analysis compositional 681 of semiconductors impurities 842 surface 681 Analysis techniques elemental and structural 681 for vapor 417 Angle of deposition 631 Anodic arc 191 Anodic coatings 770 Anodization 49 Anodized types of coatings 563 Anodizing 560 aluminum 602 magnesium 568 titanium 568 Applications of coatings 41, 44 of CVD 453 of dispersion coatings 547 of glow discharge plasmas 55 of JVD 862

Aqueous deposition 596 Ar electron energy 59 Arc 59 definition of 189 deposition 191 evaporation 189, 371 plasma spraying 51 Arcing can initiate wear 783 Arcs 189 ARE 52, 213, 497 arc evaporation 220 BARE 52, 218 ECR excitation 222 electron-beam-heated 216 enhanced 218 LPPD 218 modifications of 218 plasma electron-beam 218 process parameters 497 pulsed laser beams 221 reactive ion plating 218 resistance-heated 217 RF excitation 221 triode reactive ion plating 220 using plasma electron-beam guns 218 ARE process mechanism 222 mechanism of 222 types 189 using an arc evaporation source 220 variants of 216 As Auger transition 795 Atom transfer processes 160 Atomic peening 680 Au on NaCl 745 Au-Ag-Sb alloys 545 Auger electron kinetic energy of 790 Auger spectrometer 791 Auger transition 793, 795, 801

865

866


B Backfill using Ar 276 Backscattering 291 Backscattering yield 822 Bacteriological contamination 119 Ballistic aggregation 708, 732, 733, 741 Band-to-band recombination 841 Banded structures 576 Bearings corrosion resistance 775 Behavioral properties characterization of 672 Bernoulli's principle 854 Beta backscatter 690 Bias sputtering 332 substrate 236 Biased activated reactive evaporation (BARE) 52 Bimetallic layer process 829 Binding energy 298 of emitted photoelectrons 797 Biomedical uses of coatings 44 Blow-off 120 Bombardment by energetic species 141 concurrent 373 during deposition 679 effects 360, 364, 373, 374 sources of 360 Bombardment enhanced-chemical etching 363 Bombardment-enhanced chemical etching 678 Bond strength 148, 643, 644 Boron evaporation of 220 Bound excitons 842 Boundary layer gas stream 423 mass transport across 428 thickness of 432 Breakdown strength 859, 861 Brightness 526 Brittle erosion 782

Brush painting 51 Buffered hydrofluoric acid. See HF Bulk deposits 159 effect of bombardment 369

C c-BN synthesizing of 224 Capacitive coupling 319 Capillarity model 711 Capture cross-section 714 Carbides hydrogen ion bombardment of 368 Carbon evaporation of 220 Carburizing 49, 403 Carousel 851, 858 motion 856 rotation frequency 856 Carrier gas 411 Cascade photoionization-induced 827 Cathode current efficiency (CCE) 533 dark space 353 fall region 352 hollow 83, 370 poisoning 330 Cathodic arc 190 Cathodic coatings 770 Cd on W 725 Cells corrosion 769 Ceramic host–organic guest 860 Channeling of ions 818 Characterization of thin films and coatings 789 Charge exchange 65, 353 Charge separation 513 Charge-induced migration 810 Chelating agents 123 Chemical cause of adhesion loss 410 etch rate measurement 701

Index

potential 406 pumps 416 vapor deposition 460 Chemical conversion coating 28 Chemical ion plating 53 Chemical shifts of Si2p 800 Chemical sputtering 363 Chemical vapor deposition 160 Chemical vapor deposition (CVD) 50, 722 Chemically functional applications 41 Chemisorption 95 Child-Langmuir law 72 Chlorinated solvents 123 Chromium deposits 590 hardness of 539 Class 100 Federal Standard 114 Cleaning before plating 536 for CVD substrates 410 in situ 134 monitoring of process 133 processes 119 Cluster 712 critical size 712, 719, 722 growth 855 Cluster ion beam deposition 53 Cluster-embedded 848 Clusters liquid 714 subcritical 746 Coalescence 715, 716, 721 described 714 island 730 morphological changes during 715 Coating alloy 191 methods 779 of ultrafine fibers 859 on plastics 191 process 634 Coating processes classification of 34

867

Coatings 28. See also Films and coatings applications of 41 by atomistic deposition 35 corrosion-resistant 773 definition of 669 deposition methods 772 described 28 diffusion 28, 31 disadvantages of 669 for galvanic corrosion 770 full-density 158 high temperature protective 779 novel 848 overlay 28, 31 overlay by PVD 779 protective 770 sacrificial 773 solid lubricants 785 strength 643 structure of 636 to reduce friction 783 to reduce wear 783 to resist wear 785 uses of 29 zinc 773 Cobalt base alloys 662 Coefficient of friction for Cr 589, 592 for NiP 555 Coefficient of thermal expansion 591 Coherent elastic scattering 830 Cold finger 426 Cold wall reactor 414 Cold-cathode plasma electron beam 186 Cold-cathode discharge 82 Cold-wall PECVD reactors 468 Collective behavior 68 Collimation of the jet 854 Collision 56 cascades 284, 361, 368 cross section 56, 57, 62, 141 electron-electron 60 electron/ion 62

868


frequency 61, 64 inelastic 60 Collision free zone 853 Collisional damage 364 Collisionally-induced dissociative chemisorption 743 Collisionless ion transport 94 Color anodized coatings 568 of coatings 636 Columnar grains 445, 637 microstructure 376 structures 576, 732, 741 Columnar morphology 752 large-grained 228 Compaction of the near-surface region 369 Complex ions deposition of 509 Complexing agents 553 Composite materials 28 behavior of 29 fiber-reinforced 859 Compound semiconductor growth 192 Compounds deposition of 485 Compressive stress 691 Computer simulations 754 Concentration cells 769 Concentration polarization 516 cathodic 517 Concentration profile 424 Concurrent energetic particle bombardment 375 Conditioning of deposition systems 113 Conductivity electrical 65 Cones 294 Conservation of materials 43 Contaminants flux of 707 in vapor 413

Contamination 721 environmental 113 origins of 110 reactive gas 112 role in microstructure 741 sources 116 Control of film properties 494 Convection buoyancy-driven 422 rolls 423 Conversion coating 50 Conversion coating 50 Conversion/diffusion coating 50 Copper cyanide strike 515 Copper-nickel alloys microstructure of 233 Corona discharges 102 Corronizing 547 Corrosion damage 767 described 766 galvanic 767 high temperature 776 resistance 661, 774 salt fog c 553 Corrosion-resistant coatings 42 Corrosive wear 783 CoSi2/Si system 829 Coulomb domination 63 Coulometers determine efficiency of deposition 511 Coupled reactions 404 Covering power 519 Cracking 767 Creep rate 244, 254 strength 251 Critical cluster size 712 Critical thickness 728 Cross section 56, 57 total 65 Crystal oscillators 196 Crystalline structure 641

Index

Crystallographic orientation 380 Cubic boron nitride 45 Current density 72 defined 524 Current distribution in plating system 519 Current sources for plating 530 Cutting tools 43 Cutting-tools 786 CVD 50, 95, 400 applications 453 deposition temperatures 460 exhaust system 415 phase-selective 452 plasma-assisted 77, 92 plasma-enhanced 347 process control 429 reaction zones 401 reactions 403 reactor 413 selective deposition 445 types of processes 401 CVD processes classification of 487 Cyanide copper strike 537 Cyclotron radius 67 Cylindrical magnetron 279, 306, 311 Cylindrical-post magnetron 307

D Damage by energetic ions 838 Dangling bonds 148 Dark-space 302, 462 thicknesses 302 DC 76 discharge 322 glow discharge 301 magnetrons 315 sputtering 279 DC diode advantages and disadvantages 354 discharge 135, 351

869

DC discharge 849 Debye length 69, 70, 75 Decorative coatings 41, 570 Deep-level defects. See Defects Defects 237 flake 237, 241 fracture 684 leaders 241 mechanical 721 spit 237, 241 voids 698 weak grain boundaries 240 Density of coatings 645 of films and coatings 697 Deposit control parameters 526 formation of 32 structure and properties of 574 Deposit profile 855 Deposited coating selection of 539 Deposited materials unique features of 40 Deposition area-selective 449 chamber 849 electron beam evaporation 780 formation 160 low pressure 849 of alloys 201 of cluster-embedded 848 of dielectrics 848 of elemental semiconductors 201 of host-guest 848 of intermetallic compounds 205 of metallic coatings 772 of metals 201, 848 of multicomponent 848 of multilayer 848 of nitrides 848 of organics 848 of oxides 848 of refractory compounds 209 of semiconductors 201, 848 phase-selective 452 plasma-assisted 213

870


reactive 680 steps in 489 Deposition mechanism 522 Deposition process model 488 parameters 493, 494 variables 488 Deposition processes 520 atomic 486 atomistic 35 bulk 486 classified 486 defined 33 definitions 49 droplet 486 selection criteria 46 types of 159 Deposition rate 157, 304, 495, 856 control of 199 for various processes 163 of metal 303 of planar-diode 279 of TiC 216 techniques to increase 332 Deposition rate monitors 194 Deposition techniques arc evaporation 189 evaporation 160, 166 gas jet 37 hybrid 501 ion-plating 162 laser 192 laser evaporation 193 low pressure plasma spray (LPPS) 780 physical vapor 707 plasma-assisted 499 PLD 192 PVD 159 sputtering 163 Deposition technologies 29 definitions and distinctions 31 Depth profile 793, 801 resolution of 794 SIMS 810 Depth profiling technique 805 Desolvation energy 520

Desorption ion-induced 367 Desorption energy 714, 722, 728 Detergent cleaning 122 Detonation coating 51 Detonation gun 36, 626 Dezincification corrosion 767 Diagnostic techniques 492 langmuir probe 492 LIF 492 MS 492 OES 492 RHEED 722 Diamond 45 Diamond-like carbon 45 Diamond-like carbon (DLC) 840 Dielectric film growth 192 Differential aeration cell 769 Differential temperature cells 769 Diffusion coatings 545 flux 74 in JVD 854 rate 523 Diffusion coatings 31, 49 Diffusion coefficient 65, 66 electron 75 Diffusivity surface 710 Diode DC 301 geometry 497 parallel-plate 278 planar 278, 281, 301 sputtering systems 301 Dip coating 51 Direct evaporation 209 Discharges cold cathode 82 magnetron sputtering 75 Dislocation 721 number density 757 Dispersion coatings 547 Dispersion-strengthened alloys 233, 251

Index

Disproportionation 50 Disproportionation reactions 404 Dissimilar electrode cells 769 Dissociation degree of 359 Dissociative chemisorption 95 Domain boundary 828 magnetic 827 Donor-acceptor pair recombination 841 Drift velocity 64 x 75 Drying after fluid cleaning 132 Ductile erosion 782 Ductility 592 Duoplasmatron 327 Duty cycle 532

E ECR CVD reactor 471 EDS in STEM 834 EDX analyzer resolution of 826 Effluent 849 Ejection energy under Ar+ bonbardment 298 Ejection velocities 300 Elastic modulus 650, 693 Electret materials 115 Electric-arc induced wear 783 spraying 52 Electrical characteristics 664 resistivity 591, 696 uses of coatings 44 Electrically active defects. See Defects Electrically functional applications 41 Electrochemical reaction

871

efficiency of 511 Electrode geometry of PECVD reactor 467 Electrodeposition 160 applications 506 principles of 508 Electrodeposits physical properties of 591 structures of 574 Electroforming defined 557 Electroless deposition 50 nickel deposits 554 plating 550 plating solutions 599 Electrolyte composition 526 Electrolytic cleaning 131 deposition 49 Electromigration 697 Electromotive Force (EMF) Series 513 Electron backscattering coefficient 822 bombardment 100 cloud 321, 323, 370 collisions 309 density 59 drift speed 67 energy distribution functions 59 high brightness sources 791 interactions with molecules 88 ionization 87 irradiation 100 motion 309 spectrometer 791 temperature 60 Electron beam gun 182 cold cathode plasma 186 disc cathode 185 hot hollow cathode 187 Pierce-type 185 plasma 183 self-accelerated 182 thermionic 183

872


transverse linear cathode 185 work-accelerated 182 Electron beam heated sources 181 Electron beams for Auger excitation 791 Electron cyclotron resonance. See ECR Electron cyclotron resonance coupling 356 Electron emitter 356 advantages and disadvantages 357 plasma generation 139 Electron energy 56 distribution function 59, 60, 61 in plasma-assisted deposition 489 Electron energy loss spectroscopy (EELS) 832 Electron-electron collisions 60 Electron/atom interactions 87 Electron/ion collision frequencies 62 Electron/molecule interactions 88 Electronegative molecules 89 Electrophoresis 548 Electrophoretic coating 49 Electroplating Preparation of substrates for 597 Electrostatic charge buildup 115 contributes to contamination 116 Electrostatic deposition 49 Ellipsometer configurations 836 Ellipsometry 690, 834 EMF and galvanic series 770 EMF series 771 End point detection 492 End-confinement 312 End-point filtration 119 Energetic ions 490 neutrals 291, 353, 490, 499 particles 350 Energy densities 755 exchange 56

of depositing species 38 of incident species 490 Energy dispersive x-ray spectrometer (EDX) 825 Energy distribution 299 Energy level diagram 789 Energy transfer coefficient 290 Enhanced ARE process 218 Entrance effects 427 Environmental aging of films and coatings 672 Environmental corrosion 42 Environmentally sound 856 Epitaxial layer thickness 839 Epitaxial growth 596 conditions 446 in CVD 445 Epitaxial layer 724 by accelerated beams 749 Epitaxy 725 Equilibrium calculation by SOLGAS 406 calculation results 407 conditions 513 vapor pressure 710 Equipment coating 618 for CVD 410 for deposition 630 for ion plating 381 gas handling 383 power supplies 383 substrate fixturing 384 torch and part handling 630 Equivalent circuit for RF glow discharge 323 Erosion rate 288 resistance 659 volumetric 782 wear 782 Etch cleaning 121 Etching before plating 537 bombardment-assisted 367

Index

Evaporation 52 apparatus 169 direct 486 electron beam heated 160 flash 206 high energy electron beam 370 high rate 168 process control 199 purification by 257 rate 166 reactive 213, 486 system 169 theory 166 thermal 849 Evaporation source materials 178 Evaporation sources 172. See also Vapor sources types of 172 Exchange reactions 404 Excitation-dissociation process 89 Excitons 841 Exhaust system for CVD 415

F Faraday’s Laws of electrolysis 509 Fatigue wear 782 fcc 724 Ferromagnetic crystals magnetic contrast of 828 preferred axis in 828 Fiber coating 859 Fibrous structures 576 Film adhesion 376 density 376 material 707 morphology 376 quality 473 stress 378 Film deposition steps in 489 Film properties characterization of 671 measurement of 682 modified by ion bombardment 375

873

Films. See Films and coatings definition of 670 deposited by PECVD 472 thick 31, 158, 159, 848 thin 31, 158, 848 Films and coatings. See Coatings alloy 848 atomistically deposited 677 by JVD 848 characterization of 671, 675, 677 cluster-embedded 848 compounds 852 configurations 670 definition of 670 doped 852 formation of 677 host-guest 848 multicomponent 848 multilayer 848, 852 properties of 670 stresses in 691 testing of 674 thickness of 689 uniform 851 Filters activated carbon 119 Finishing of coatings 635 First check characterization 673 Flame spraying 51 Flash evaporation 206 Floating potential 319 negative 491 Flow viscous 489 Flux ratio 335 Fluxing 122 Footprint 856 Fourier transform infrared (FTIR) spectrometers 838 Fracture propagation 684 Frank-van der Merwe growth mode 708 Free electron kinetic energy 58 Free energy 711 barrier 722 minimization 405

874


of formation 215 volume 711 Free jet 853 Free-exciton (FE) recombination 842 Frequency effects on RF plasma 466 Fresnel reflection coefficient 834 Fretting corrosion 767 wear 781 Friction 783 and wear 781 and wear coatings 42 coefficient of 784 FTIR measuring epitaxial layer 839 Full-density coatings 158 Functional characterization 672 coatings 41

G Ga Auger transition 795 GaAs anodic oxidation of 810 growth 723 sputtering of 298 Galvanic cell 768 corrosion 661, 770 Gas charging 143 control equipment 630 dispensing system 411 in a deposited film 380 pumping 380 states 418 Gas flow calculations 420 dynamics 417 patterns 420 proportional counter 827 rate controls stoichiometry 494 Gas jet deposition 37 Gasdynamic deposition source 849 Gaseous anodization 49

Gate 859 Geometrical thickness measurement of 689 GexSi1-x alloys on Si(100) 728 Glow discharge 52, 461, 849 cleaning 100 DC 301 evaporation 52 low-pressure 301 plasma 55, 59, 70, 76 polymerization 77 sputter jet 855 sputtering 327 Gold 850, 854 Graded interface 237 Grain boundaries 721 boundary 39 growth mechanism 736 size 719 size can vary 577 Grain boundaries weak 240 Grashof number 418 Grit blasting 632 Growth adsorption-induced 452 area-selective 446 kinetics 707 mixed-mode 728 modes 708 rate 848 single phase vapor 442 substrate-activated 449 three-dimensional 708, 710 two-dimensional 708, 721 Growth mechanism of PVD films 224 Gyro radius 67

H Hall-Petch relationship 249, 253, 576 Halogen solvents 124 Handling of prepared surfaces 147

Index

Hard coating tool wear-life improvement by 786 Hardness 591 of higher velocity coatings 648 of metal and alloy deposits 244 values for various deposits 539 varies with deposition temperature 255 Hardness testing 695 Haring-Blum %TP 518 Harmonic electrical spraying 52 Heat treatment of Ni3P or Ni3B 554 Heating sources 370 the substrate 415 Hertz-Knudsen equation 167 Heteroepitaxial films are often pseudomorphic 725 High temperature protective metallic coatings 779 High temperature corrosion 42 History of evaporated thin films 158 Hollow-cathode ion sources 327 Homogeneous nucleation 409 Homogeneous reaction control 429 Host-guest 848, 860 Hot hollow cathode discharge beam 187 electron beam gun 188 Hot wall reactor 413 Hot-cathode triode sputtering systems 305 Hot-wall PECVD reactors 468 Humidity in clean room 116 Hybrid processes 501 Hydriding 409 Hydrogen charge 368 overvoltage 517 reduction cleaning 130 Hydrophilic wetting method 51 Hydrosonic agitation 529 cleaning 128

875

I Impact erosion 782 Impact mobility 749 Impurities in deposits 256 in plating solutions 529 In 747 Induction heated sources 180 Inelastic collisions 60 Inelastic mean free path 791, 799 Inert coatings 770 Inert gas entrapped 291 shroud 623 Inhibitive coatings 770 Injection rate reactive-gas 330 Integrated electronics 849 Interface characterization of 679 during CVD is unstable 443 formation 350, 678 trap density 859 types 678 Interfacial regions classified 374 Intergranular corrosion 767 Intermetallic compounds 409 deposition of 205 Interphase 678 Ion bombardment 93 bombardment during deposition 750 carburizing 49 current 324 current monitor 194 hollow-cathode source 327 implantation 53 implantation accelerators 348 irradiation 97, 333, 743 irradiation effects 744, 759 mixing 683 mobility 72 nitriding 49 replenishment 522

876


scrubbing 134, 354 source 350 Ion beam assisted deposition 53 Ion beam deposition 52 Ion implantation 53 Ion plating 53, 77 advantages and disadvantages 389 applications 389 barrel-plating 388 chemical ion plating 348 control of 385 control parameters 354 DC plasma conditions 354 defined 346 discharge 77 history 346, 348 IAD 348 IBED 348 IVD 388 monitoring of 385 problem areas 386 process 162 process parameters 358 process specifications 385 pulsed 383, 387 reactive 348, 373 rules 349 sputter 348 stages 349 vacuum 348 Ion-assisted chemical etching 367 Ion-assisted molecular-beam epitaxy 743 Ion-beam sputtering 327 Ion/surface interactions 708 low-energy 743 Ionic migration 522 Ionization 56, 69 balances 77 potential 56 rate of 80 Ionization gauge rate monitor 194 Ionized gas 55 IR absorption bands of O and C 839 Iron alloy films corrosion resistance 775

Irradiation during film growth 743 Irradiation-induced effects 755 Island coalescence 714 Island growth 224, 716 Islands In 730 secondary 714

J Jet boundary 854 conditions 860 independence 855 sonic 848 source 849 Jet in low vacuum 849, 853, 856, 861 Jet Vapor Deposition (JVD) 848 JVD 848, 853, 856, 859, 861 JVD sources 849 microwave 859 sputtering 856

K Kinetic energy (KE) emitted photoelectrons Krytonation 368

797

L LaB6 791 Lamellae thickness of 253 Lamellar microstructure 663 Laminate composites 253 structure 253 Larmor radius 67 Laser ablation 192 Laser induced evaporation 192 Lattice atom displacement 363 constant 725 defects 368 strain 378, 680, 694

Index

Laves phase 641 Li-doped Si detector 825, 826 Life-tests accelerated 674 operational 674 Limit 708 Low-contaminant materials 116 Low-energy ion irradiation 707 Low-pressure plasma deposition (LPPD) process 218 Low-temperature deposition 732, 860 Lubricant coatings 42 Lubrication coatings 785

M Mach disc 853 Macroparticle removal of 191 Magnesium alloys anodizing of 568 Magnetic contrast type I, II 828 Magnetic fields and plasma particles 67 Magnetic moment 828 Magnetic sector spectrometer 804, 833 Magnetron discharge sources 84 enhanced plasma 140 Magnetron discharge 84 Magnetrons 279, 281, 306, 326 advantages and disadvantages 358 cylindrical 311 discharge 357 double-ended RF 326 geometries 497 sputtering of a source 387 unbalanced 333 Manufacturing advantages of JVD 856 Masking techniques 633

877

Mass spectrometer quadrupole 804 Mass spectrometers magnetic sector 804 Mass transfer coefficient 524 Mass transport across a boundary layer 428 control 429 of ions 522 Material cost 856 Materials multicomponent 292 Materials conservation 43 Maxwellian velocity distribution 62 MCrAlY 663 coating process steps 780 Mean free path defined 56 of electrons 303, 791 Mean residence time 714 Measurement in situ 673 of adhesion 671, 682 of density 697 of electrical resistivity 671 of electromigration 697 of films and coatings 671 of resistivity 696 of stability 672 pressure 171 thickness of films and coatings 671 Mechanical properties 647. See also Test techniques of laminates 254 of thick condensates 244 of thin films 241 Mechanical scrubbing 125 Mechanically functional applications 41 Metal ions deposition of 522 Metal-nitride-semiconductor (MNS) capacitor 859 Metallic contaminates 117 impurities 530

878


Metallic coatings deposition of 772 Metalliding 50 Metallographic apparent porosity 647 Metallurgical properties 39 Metastable species 90 Metastable phases 491 Metastable species 90, 499 Metering of liquids 413 Methyl Red 860 Microbalances 196 Microhardness measurement 592, 695 Microlaminate composites thermal conductivity of 781 Microlaminates 858 Micromixing 856 Microstructural development mechanisms 741 difference between coatings 637 evolution 707, 708, 732 Microstructure 38 evolution 224, 730 evolution of 730 of films 375 of PVD condensates 224 of thick single phase films 226 Microstructures columnar 732, 733 Microthrowing power 523 Microwave discharge 77, 356, 849 plasma generation 139 Migration charge-induced 810 Mobilities low-adatom 733 Mobility 64 of ions 65, 302 Model capillarity 711 droplet 711 TLK 438 Models, nonlinear of the reactive sputtering 332 Modified surface definition of 669

Modulus of rupture 650 Molecular beam epitaxy 52 Molecular beam epitaxy (MBE) 722 Molecular bombarding species 286 Molecular dynamic simulations 754 Molecular flow of species 488 Molecular-dynamic simulations 733 Momentum exchange 65, 289, 291 collisions 64 cross section 58 Momentum transfer 353, 363 Monitoring of deposited mass 196 Monitoring of specific film properties 196 Monitors optical 196 resistance 196 Monochromator with x-ray source 798 Monolayer per pass 856 Monte Carlo simulations 733, 736, 754 Morphology large-grained columnar 228 of films 375 of thick single phase films 226 structural 227 Motion of charged particle 66 Movchan-Demchishin model 230 Multifilament tows 859 Multiple internal reflections 839 Multiple jet, moving substrate 852, 856, 861 Multiple sources 201

N Nano-particles 37 Negative glow 301 Negative glow region 82 Negative ion emission 297 Negative ions 89 Nernst equation 513

Index

Ni and Cr depth profile of 794 Ni on Si interactions 817 Ni-Cr-Fe electrophoretic deposition of 548 Nickel composite electroless 557 containing P or B 551 deposition of 230 electrodeposits 577 Nitridation of UHV cleaned surfaces 151 Nitriding 49, 403 NixPy compounds 551 Noble metals 860 Non-destructive evaluation (NDE) 675 Novel coatings 848 Nozzle pressure 849 Nuclear fuels 44 Nuclear reaction analysis (NRA) 819 Nucleation 438, 707, 710 3-D 710 activation barrier 711 control 429 density 374, 677 heterogeneous 452, 711 kinetics 743 kinetics model 713 secondary 716 two-dimensional 722, 724, 725 Nucleation sites preferential 744 Nuclei size 716 Number density of Au nuclei on NaCl 715 of ion-irradiation-induced 744 surface site 710

O Optical emission 304 Optical monitors 196 Optically functional applications 41

Optics nonlinear 849 Orbiting frequency 66 Organic impurities 530 Orientation crystallographic 380 Oscillations plasma 75 Outgassing after fluid cleaning 132 Overlay coating definition of 669 Overlay coatings 31 Overvoltage 515 Oxidation cleaning 128 high temperature 776 Oxidation states ability to distinguish 800 Oxide films 410 Oxygen plasma cleaning 144

P PACVD 495 Palladium adhesion dependent on 571 Partial pressure affects growth of films 494 Particle impingement rate monitor 194 Particles lenticular 617, 621 Particulate contamination 114 deposition processes 36 origin of 111 removal 120 Parting corrosion 767 Parylene 861 Paschen curves 80 relation 77, 83 Passivation 859 of clean surfaces 151 PAVD film growth by 493

879

880


PECVD 461 conditions for silicon nitride 473 dual frequency reactor 475 films 481 Peel test 688 Penning discharge 75 ionization 91 Penning ionization 141, 359 Periodic reverse 532 Phase diagrams 407 Phosphorus content of deposits 553 Photoelectron (PE) transitions 799 Photoelectron spectroscopy 797 Photoluminescence (PL) spectroscopy 841 Physical sputtering of a surface 143 Physical vapor deposition processes 160. See also Deposition techniques Pickling 121, 536 Pierce gun 185 Piezoceramic wafers 858 Pitting corrosion 767 PL practical range of 843 primary application of 842 Planar diodes 83, 301 Planar magnetron 279, 316 Plasma 55 activation 360 chemistry 140, 359 cleaning 134 defined 186, 351 density 335 diagnostic techniques 492 discharge operating conditions 81 enhancement 358 etching 77, 145 frequency 69 gas velocities 618 generation of 135, 351 glow discharge 55, 76 methods of creating 467

microwave 356 monitoring of 385 near the substrate surface 384 oscillations 75 oxidation 49 parameters 493, 494 polymerization 53, 93 processing 351 properties of 351, 360 RF 355 sheath 69, 70 spraying 36 temperature 618, 622 torch 618, 621 uniformity 384 variables 488 volume chemistry 492, 499 Plasma electron beam gun 186 Plasma excitation geometries 221 Plasma excitation modes 221 Plasma rings 312, 315 Plasma-Assisted CVD 50 Plasma-assisted CVD 743 conditions 92 Plasma-assisted deposition limitations of 499 Plasma-assisted etching 93 Plasma-enhanced CVD 461 Plating cell 508 laser-enhanced 535 on plastics 570 operations 536 variables 528 Platinum coatings 776 PLD. See Deposition techniques Point of entry of powder 620 Poisoning of sputter cathode 330 Polarization 515 effects 520 Polycrystalline silicides 830 Pores 409 Porosimetry 699 Porosity 698

Index

decreased by ion irradiation 753 in coatings 647 tests 700 Position equivalency 674 Positive column region 83 Post-deposition processing 680 Post-plating treatments 538 Potential floating 70 negative 276 Powder dispensers 630 size distribution 622 temperature 620 used for plasma 632 velocity 620 Prandtl number 418 Precursor species 492 Precursors chemical vapor 373 toxic 849 Predominance diagrams 407 Preferential sputtering 366 Preferred orientation 746, 750, 756 Presheath 72 Primary electrons 302 Primary ions for SIMS 806 Primary-ion deposition (PID) 743 Printed circuit board plating of 571 Printing process 51 Process control 194 Process parameters control of 494 Processes for MCrAlY coatings 780 wet-chemical. See Wet-chemical Production capability 159 Professional organizations 676 Properties of deposited coating 38 Property measurements of films and coatings 682 Pseudomorphism 728 Pulse plating 532

881

Pulse-counting 826 Pulsed laser deposition (PLD) 192 Pulsed plating effects of 534 Pump chemical 416 water-ring 416 Pumping systems 171 Purification of metals by evaporation 256 PVD. See Deposition techniques microstructure of condensates 224 process terminology 32 processes 160 vs. CVD 165 PVD processes classification of 487 PVDF piezoelectric membrane 860 Pycnometry 698 Pyroelectric infrared detectors 858 Pyrolysis 50 PZT 856, 858, 859

Q Quadrupole mass spectrometer

R Radiation hardness 859 Random walk of a helium atom 854 Rare-gas crystals growth of 724 Rate control 199 of chemical reaction in a plasma 493 of dissociation 495 Rate monitors crystal oscillator 196 ion current 194 ionization gauge 194 microbalances 196 of deposited mass 196 particle impingement 194 spectroscopic 195

804

882


Rate-controlling steps 216 Rate-determining reactions in CVD 436 Rate-limiting in CVD 428 Ratio of ring diameters 831 Rayleigh number 418 RBS sensitivity limit of 818 spectrum 816 used to normalize SIMS 810 Reaction CVD 403 kinetics 215 mechanisms in CVD 436 rate 63 resistance 431 Reactive evaporation 52 gas 330 ion etching 98, 145 ion plating 53 plasma cleaning 144 plasma etching 367 sputtering 328 Reactive evaporation model 215 process 213 Reactive ion plating (RIP) processes 218 Reactor cold wall 414 CVD 413 geometry 420 hot wall 413 Recoil implantation 348, 369, 754 Recombination of electron-hole pairs 841 Recombination rate constant 855 Recontamination 111 Recrystallization 716, 721, 741 during coalescence 716 Recycling of reactants 417 Redeposition of sputtered material 375 Reducing agent 550, 553

Reduction reactions 403 Reflected power 324 Reflection coefficients 835 of ions 290 Reflection high-energy electron diffraction (RHEED 722 Refraction indices of 834 Refractive index 837 Refractory compounds 209, 254 deposition of 209 mechanical properties of 254 Refractory materials evaporation of 370 Refractory metals area-selective deposition 449 Relative characterization 672 Research needed 40 Residual stress 237, 645 in deposits 237 Residue contamination 117 Resistance heated sources 175 Resistance monitors 196 Resistivity of films and coatings 696 Resonance ionization spectroscopy (RIS) 812 Resputtering rate 375, 377 Reynold’s number 418 RF planar-diode 319 plasma reactor 466 power supplies 355 sputtering 279, 318 RF activation 220 RF discharge 76, 85, 322, 355 advantages and disadvantages 356 capacitively coupled 355 plasma generation 138 RHEED 722 RHEED oscillations 723, 724 Rhodamine 860 Room temperature deposition 634

Index

Roughness can not be determined 838 of coatings 636 of surface 633 Rutherford backscattering spectroscopy (RBS) 812

S S-gun magnetron 306 Sacrificial layer 151 Safety 152 Salt concentration cell 769 Saturated hydrogen electrode 513 Saturation flux density 593 Scanning electron microscopy 822 Scattering energy 812 Scratch test 688 Scrubbers exhaust 417 Scrubbing 125 Secondary electron emission coefficient 82, 352 Secondary electrons 302, 308 energy distribution of 822 Secondary ion mass spectroscopy (SIMS) 803 Secondary ions generation of 806 Secondary nucleation 714, 716 rate 747 Selected area diffraction (SAD) 830 Selective deposition by CVD 445 Self-accelerated gun 182. See Electron beam gun Self-bias negative 462 Self-limiting growth 450 Self-mating characteristics 658 Self-shadowing 733 Self-sputtering 366 Self-supported shapes 158 Self-welding 659 SEM resolution of 822 Semiconductor microelectronic processing 861

883

Semiconductors 201 Sensitization of a surface 149 Shear strength 643 test 688 Sheath 74 capacitance 324 includes dark spaces 461 plasma 70 potential 137, 467 thickness 72 Sheath potential 354 Sheet resistance 696 Shroud inert gas 623 Si 728, 730 thermally induced defects in 844 Silicon dioxide by PECVD 478 Silicon nitride by PECVD 472 etch rate 859 films 859 interface trap density 859 radiation hardness 859 SIMS detection limit of 810 RBS is used to calibrate 810 spectrum 806 voltage range 806 SiN 859 Single rod-fed electron beam source 202 Smut 122 Snell’s Law 834 Snow scrubs 120 Sodium contamination 117 Solid state reactions during CVD 401 Solution wear 783 Solvent cleaning 123 systems 125 Sonic gas flow 855 Sonic jet in a low-vacuum fast flow 848

884


Sources. See Vapor sources for electron beam evaporation 383 hollow-cathode 327 ion 327 of depositing species 369 of energetic particles 350 of ions 350 reaction with evaporants 178 Space charge sheaths 70 Spark-hardening 50 Specifications and standards professional organizations 676 Spectroscopic ellipsometry 837 Spectroscopic methods 195 Spectroscopy 797 Splat 637 Spray cleaning 126 Spraying processes 51 Sputter cleaning 143, 349 Sputter deposition 52, 275, 722 applications of 279 flux profiles 316 forward 361 history of 283 magnetron 333 of multicomponent materials 292 parameters 496 process variants 497 universality of 276, 279 Sputter deposition modes reactive sputtering 328 Sputter deposition technology variations 276 Sputter-assisted laser ionization (SALI) 811 Sputter-deposited films composition of 292 Sputtered as dimers 297 clusters 297 molecules 297 species 296, 298, 299 Sputtering 52 bias 332 cost of 282 DC 279

defined 275 deposition rate 304 direct 486 efficiency 291 efficiency of 286 erosion rate 288 glow discharge 297, 327, 849 in N2 332 ion plating 371 ion-beam 327 mechanisms 284 mechanisms of 284 of alloys 294 of alloys and compounds 279, 293 of compound semiconductors 298 of molecular species 297 of PTFE (Teflon) 279 physical 353, 364 preferential 365 process 163. See also Deposition techniques rate 285, 289, 330 reactive 328, 486 RF 279 targets. See Targets with reactive species 295 Sputtering systems balanced 324 bias sputtering 332 configurations 278 in-line 314 ion-beam 327 load-lock 281 magnetron 325 magnetrons 281, 306 magnetrons, cylindrical 311 multisource 280 operating conditions 311 parallel-plate diode 278 planar diode 281, 301, 303, 305 planar-diode 278 RF 318, 324 RF planar-diode 319 selection of 282 single-ended 324 targets 282

Index

triode 279 triode discharge 305 with magnetrons 279 Sputtering yield 285, 286, 287, 365 defined 286, 364 dependence on angle of incidence 287 dependence on ion species 295 expression for 289 influenced by surface topography 296 SIMS 806 Stabilizers added to chlorinated solvents 124 Stabilizing for electrodeposition 537 Stainless steel corrosion resistance 775 STEM 828 Sticking coefficient 257, 329 Sticking probability 329 Storage of prepared surfaces 147 Strain-to-fracture 650 Stranski-Krastanov 728 growth mode 708 Stress 409, 592, 644, 733 calculation of 691 for chromium deposits 590 growth 680 in continuous films 721 in deposits 237 in electroless Ni-P alloys 553 in films and coatings 691 in growing film 361 in PECVD films 473 in the near-surface region 369 in-plane tensile 733 measurements 593 mechanical 683 residual growth 378 varies with phophorus 554 Stress wave adhesion tests 688 Stress-corrosion cracking 767 Strike 350

885

Strikes before plating 537 Strip processing line 170 Stripline 170 Structure zone model 39 Structure-zone diagram 736 Structure-zone diagrams 708 Sublimation 370 sources 176 Substrate bias influences structure 494 preparation 632 preparation defined 108 Substrate motion 851, 856 Superalloy 778, 780 Superconducting film growth 192 Supercritical nuclei 714 Supersaturation 444, 713, 725 degree of 719 in CVD 442 ratio 711 Surface coverage 380 diffusion 440 effects of bombardment 368 engineering 27 kinetics control 429, 432 modification 150 morphology 367, 442 preparation 677 preparation for ion plating 349 profilometer 690 protection 849 reaction control 433 Surface Acoustic Wave porosity measurement 700 Surface Charge Analysis. See SCA Surface energy 715 Surface free energy 711 Surface mobility 229 Synthesis of compounds 224 of unusual, complex materials 849 SZD 738, 739, 741

886


T TaC-Fe-Ni electrophoretic deposition of 548 Target poisoning 496 Target voltage 489 Targets 295 composite sputtering 294 for sputtering 282 hot-pressed 295 non-conducting 321 polycrystalline or amorphous 300 poorly conducting 295 semiconducting 297 sputter 383 TEM scanning 828 Temperature control 629 detonation gun 626 during CVD 401 gas 70 profiles 427 Temperature Coefficient of Resistivity (TCR) 696 Tensile bond strength 643 strength 592 stress 691 test 688 Tensile properties of metals and alloy deposits 244 of thin film 242 TEOS films 479 Terrace lengths 723 Test techniques of mechanical properties 241 Texture of evaporated deposits 236 Thermal barriers 663 decomposition 403 evaporation 370 expansion coefficient 695 properties of coatings 663 Thermal stress adhesion test 689

Thermally sensitive membranes 860 Thermionic Gun 183 Thermodynamic calculations 405 control of CVD 429 wear 783 Thermoelectron emitter system 357 Thick films 31 Thick single phase films microstructure and morphology of 226 Thickness and uniformity 858 coating 617 control 199 measurement of 689, 698 Thin films 31 Three body collisions 855 Three-dimensional growth 708 Throwing power 32, 381, 489 of a solution 518 TiC deposition of 216, 222 microhardness of 255 Time of atomic rearrangements 754 TiN 757 by reactive magnetron sputter 741 Titanium anodizing of 568 deposition of 229 Tool wear-life improvement by hard coating 786 Topographical evolution 294 Tows 859 Toxic metal precursors 856 Transition mechanism 331 Transition zone temperatures 228 Zone T 226 Transmission electron microscopy (TEM) 828 Transport of species 488 Trapping 368 Tribology 781

Index

Triode 279, 282 configuration 357 discharge devices 305 hot-cathode 305 sputtering systems 305 Triode configuration 220 Tungsten alloy electrodeposition 544 deposited by CVD 32, 450 Tungsten carbide-cobalt 659 Turnaround time 856 Two-dimensional growth 708

Ultimate electrons 310 Ultrasonic agitation 528 cleaning 127 Unbalanced magnetron 333 Uniform deposit thickness 168, 172, 851 plasma density 386 Uniformity 858 of plasma 384 Uniformly coated 857

induction heated 180 multiple 201 resistance-heated 175 rod-fed 202 sonic jet in low-vacuum fast flow 848 sublimation 177 wire-fed 202 Vaporization 849 Variables in plasma deposition processes 488 Velocity carrier 849 distribution 60, 62 drift 64 of powder 621 Video-RHEED 750 Void density 753 Volatilization of elemental metal 856 Volatilization cleaning 130 Volmer-Weber growth mode 708 Volume free energy 711 Volume reactions 92

V

W

Vacuum arc vaporization 371 chamber configurations 282 chamber pressure 276 deposition and electroplating 507 evaporation theory 166 system for ion plating 381 Vacuum chamber 169 Vacuum pump 416 Vacuum pumping system 171 Vapor contamination 116 degreasers 126 equilibrium pressure 710 species generation 488, 494 Vapor pressure equilibrium 166 Vapor sources 849 arc 189 electron beam heated 181

Wall jet 853, 854 Water ultrapure 118 Water break test 133 Wavelength dispersive x-ray (WDX) 826 Wear 781, 785 Wear resistance 648, 653 of electroless nickel 555 Wear-life improvement by hard coating Welding processes 51 Wetting angle 111, 133 Wetting processes 50 Wirefeed jet 854 sources 857 Witness plates 673 Woods nickel strike 538

U

887

786

888


Work function 797 Work-accelerated gun 182. See Electron beam gun

X X-ray analysis in STEM 834 emission 825 source 797 X-ray fluorescence (XRF) 690 XPS 797 detection limit for 802

Y YBCO film deposition 193, 194 Yield secondary ion 806, 811

Z Zinc coatings 773 complexed with cyanide 514 Zirconia as a thermal barrier 780 Zn deposition conditions 774 Zone T 226 Zones defined 737 Zone T 739

Handbook of Deposition Technologies for Films and Coatings

Recommend Documents