Hydrometallurgy_2009.pdf

The Southern African Institute of Mining and Metallurgy Founded in 1894

The SAIMM Southern African

Hydrometallurgy Conference 2009 SYMPOSIUM SERIES S54

CONTENTS 24–26 February 2009 Misty Hills, Muldersdrift, Gauteng

i-xii:PRELIMS

2/16/09

12:08 PM

Page 1

THE SOUTHERN AFRICAN INSTITUTE OF MINING AND METALLURGY JOHANNESBURG 2008

SYMPOSIUM SERIES S54

The SAIMM Southern African

HYDROMETALLURGY CONFERENCE 2009

24–26 February 2009

MISTY HILLS, MULDERSDRIFT, GAUTENG

i-xii:PRELIMS

2/16/09

12:08 PM

Page 2

SAIMM PUBLICATIONS THE MONOGRAPH SERIES Ml

Lognormal-De Wijsian Geostatistics for Ore Evaluation (2nd ed 1981) D.G. Krige

M2

An Introduction to Geostatistical Methods of Mineral Evaluation (2nd ed 1981) J.-M.M. Rendu

M3

Principles of Flotation (1982) (3rd imp. 1986) Edited by R.P. King

M4

Increased Underground Extraction of Coal (1982) Edited by C.J. Fauconnier and R.W.O. Kersten

M5

Rock Mechanics in Mining Practice (1983) (3rd imp. 1986) Edited by S. Budavari

M6

Assay and Analytical Practice in the South African Mining Industry (1986) W.C. Lenahan and R. de L. Murray-Smith

M7

The Extractive Metallurgy of Gold in South Africa, 2 volumes (1987) Edited by G.G Stanley

M8

Mineral and Metal Extraction — An Overview (1994) L.C. Woollacott and R.H. Eric

M9

Rock Fracture and Rockbursts—an illustrative study (1997) Edited by W.D. Ortlepp

THE SPECIAL PUBLICATIONS SERIES SPI

Proceedings, Underground Transport Symposium (1986) Edited by R.C.R. Edgar

SP2

Backfill in South African Mines (1988)

SP3

Treatment and Re-use of Water in the Minerals Industry (1989)

SP4

COREX Symposium 1990 (1990) Edited by H.M.W. Delport and P.J. Holaschke

SP5

Measurement, Control, and Optimization in Mineral Processing (1994) Edited by H.W. Glen

SP6

Handbook on Hard-rock Strata Control (1994) A.J.S. Spearing

SP7

Rock Engineering for underground coal mining (2002) J. Nielen van der Merwe and Bernard J. Madden

SUPPLEMENT TO THE SAIMM JOURNAL The Metals and Minerals Industry in South Africa - Part 1 (1989) Edited by H.W. Glen

i-xii:PRELIMS

2/16/09

12:08 PM

Page 3

THE SYMPOSIUM SERIES S1 S2 S3 S4 S5 S6 S7 S8

Mathematical Statistics and Computer Applications in Ore Valuation (1966) Planning Open Pit Mines (1970) (4th imp.) Edited by P.W.J. van Rensburg Application of Computer Methods in the Mineral Industry (APCOM 1973) Edited by M.D.G. Salamon Infacon 1974 Edited by H.W. Glen Proceedings of the 12th CMMI Congress 2 volumes (1982) Edited by H.W. Glen Rockbursts and Seismicity in Mines (1984) Edited by N.C. Gay and E.H. Wainwright The Planning and Operation of Open Pit and Strip Mines (1986) Edited by J.P. Deetlefs GOLD 100: Proceedings of the International Conference on Gold (1986) Volume 1: Gold Mining Technology Edited by H. Wagner and R.P. King Volume 2: Extractive Metallurgy of Gold Edited by C.E. Fivaz and R.P. King Volume 3: Industrial Uses of Gold Edited by G. Gafner and R.P. King S9 APCOM 87: Proceedings of the Twentieth International Symposium on the Application of Computers and Mathematics in the Mineral Industries (1987) Volume 1: Mining Edited by L. Wade, R.W.O. Kersten and J.R. Cutland Volume 2: Metallurgy Edited by R.P. King and I.J. Barker Volume 3: Geostatistics Edited by I.C. Lemmer, H. Schaum and F.A.G.M. Camisani-Calzolari S10 International Deep Mining Conference (1990) Volume 1: Innovations in Metallurgical Plant Edited by G.A. Brown and P. Smith and Application of Materials Engineering in the Mining Industry Edited by B. Metcalfe Volume 2: Technical Challenges in Deep Level Mining Edited by D.A.J. Ross-Watt and P.D.K. Robinson S11 Infacon 6 ( Incorporating Incsac) (1992) Edited by H.W. Glen S12 Massmin 92 Edited by H.W. Glen (out of print) S13 Minefill 93 Edited by H.W. Glen S14 XVth CMMI Congress Publications (1994) Volume 1: Mining Edited by H.W. Glen Volume 2: Metals Technology and Extractive Metallurgy (1994) Edited by H.W. Glen S15 Surface Mining 1996 Edited by H.W. Glen S16 Hidden Wealth (1996) Edited by H.W. Glen S17 Heavy Minerals 1997 Edited by R.E. Robinson S18 The 8th International Platinum Symposium (1998) S19 Mining in Africa ’98 S20 Extraction Metallurgy Africa ’98 S21 Sixth International Symposium for Rock Fragmentation by Blasting (1999) S22 Metallurgy Africa ’99 S23 Heavy Minerals 1999 Edited by R.G. Stimson S24 Tunnels under Pressure Technical Editor T.R. Stacey S25 Mine Hoisting 2000 S26 Coal–The Future (2000) S27 The Fifth International Symposium on Rockburst and Seismicity in Mines (RaSim 5) (2001) S28 6th International Symposium on Mine Mechanization and Automation (2001) S29 XIV International Coal Preparation Congress and Exhibition S30 Surface Mining 2002—Modern Developments for the New Millennium S31 IFSA 2002, Industrial Fluidization South Africa S31A APCOM 2003—Application of Computers and Operations Research in the Minerals Industries S32 ISSA/Chamber of Mines Conference—Mines and Quarries: Prevention of Occupational Injury and Disease S33 ISRM—Technology Roadmap for Rock Mechanics S34 Heavy Minerals Conference 2003 S35 Safety in Mines Research Institutes (2003) S36 VII International Conference on Molten Slags, Fluxes & Salts (2004) Tenth International Ferroalloys Congress INFACONX 2004 S37 Deep and high stress mining 2004 S38 ‘Platinum adding value’ 2004 S39 Base Metals—Southern Africa’s response to changing global base metals market dynamics 2005 S40 Strategic versus Tactical approaches in mining 2005 S41 Best practices in rock engineering SARES 2005 S42 IFSA 2005 (Mintek) S43 Southern African Pyrometallurgy 2006 International Conference S44 Stability of rock slopes in open pit mining and civil engineering situations S45 ‘Platinum Surges Ahead’ 2006 S46 Hydraulic Transport of Solids - Hydrotransport 17 (2007) S47 Fourth Southern African Base Metals Conference ‘Africa’s base metals resurgence’ S48 Heavy Minerals Conference, 2007 S49 Cave Mining Conference, 2007 S50 International Symposium on Lead and Zinc processing – Lead & Zinc 2008 S51 Surface Mining 2008 S52 Hydrometallurgy Conference 2009 S53 IFSA 2008

i-xii:PRELIMS

2/16/09

12:08 PM

Page 4

Published by The Southern African Institute of Mining and Metallurgy Fifth Floor, Chamber of Mines Building, 5 Hollard Street, Johannesburg, 2107 Republic of South Africa © The Southern African Institute of Mining and Metallurgy, 2008

ISBN 978-1-920211-22-6 The papers in this volume have been for the most part prepared from files supplied by the authors, with additional typesetting and formatting by The Southern African Institute of Mining and Metallurgy. All papers included in this volume were subjected to review. A technical and critical review of each paper was carried out by one technical reviewer.

Printed by Camera Press, Johannesburg

i-xii:PRELIMS

2/16/09

12:08 PM

Page 5

FOREWORD

or the first time a conference is being staged that covers the true diversity of the application of hydrometallurgy in Southern Africa. This conference will bring together operating hydrometallurgical facilities, research organisations, academic organisations, project companies, technology and equipment suppliers, and reagent suppliers, encompassing a wide range of commodities and unit operations. The SAIMM Southern African Hydrometallurgy conference 2009 will cover an overview of a significant number of currently operating hydrometallurgical facilities in Southern Africa. These overviews will encompass a description of the facility, the technical challenges facing the operation, and any future process improvements, debottlenecking or expansions envisaged. The conference will highlight hydrometallurgy research being conducted by Southern African academic institutions and by mining companies. The conference will also highlight developments being undertaken by technology suppliers, equipment suppliers and reagent suppliers. Some of the countries included are South Africa, Namibia, Botswana, Zimbabwe, Zambia and the Democratic Republic of Congo. Some of the commodities covered will be copper, nickel, zinc, cobalt, uranium, gold and PGMs. A wide range of unit processes will be featured, including, atmospheric leaching, pressure leaching, biological leaching, precipitation, cementation, crystallization, solvent extraction, ion exchange and electrowinning. I believe that the papers and presentations submitted for this conference will provide a valuable future reference for hydrometallurgy, thus making the work of all the contributors to this conference highly rewarding and worthwhile.

F

Marek Dworzanowski Chairman Organising Committee

i-xii:PRELIMS

2/16/09

12:08 PM

Page 6

i-xii:PRELIMS

2/16/09

12:08 PM

Page 7

Contents Page No.

ATMOSPHERIC LEACHING Leaching of the arsenopyrite/pyrite flotation concentrates using metallic iron in a hydrochloric acid medium T. MAHLANGU, F.P. GUDYANGA, and D.J. SIMBI. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

Base metals recovery from zinc hydrometallurgical plant residues by digestion method R.B. NGENDA, L. SEGERS, and P.K. KONGOLO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

17

The progression of metallurgical testwork during heap leach design S.W. ROBERTSON and P.J. VAN STADEN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31

PRESSURE LEACHING Increasing the capacity of existing and new exothermic autoclave circuits G.M. DUNN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

43

Stress development in refractory due to the rate of temperature change: a pressure vessel refractory lining design consideration P. LAUZON, A. KONING, and I. DONOHUE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

57

Mixing system design for the Tati Activox® autoclave M. NICOLLE, G. NEL, T. PLIKAS, U. SHAH, L. ZUNTI, M. BELLINO, and H.J.H. PIETERSE . . . . . . . . .

65

Stabilization of supersonic vent gas from autoclave pressure oxidation M. FRANCESCHINI and J. WOLOSHYN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

79

BIOLOGICAL LEACHING Indigenous microorganism strains as bio-extractants of Ca, Fe, and Mg from metallurgical and mine drainages A.F. MULABA-BAFUBIANDI, E. FOSSO-KANKEU, and B.B. MAMBA . . . . . . . . . . . . . . . . . . . . . . . . . . .

93

Biosorption of cobalt and copper from hydrometallurgical solutions mediated by Pseudomona spp A.F. MULABA-BAFUBIANDI, N.P. DLAMINI, and B.B. MAMBA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

101

An investigation into fungal strains bio-extraction of metal impurities from aqueous effluents emanating from Ekhuruleni (Gauteng, South Africa) metallurgical and mining operations M. MWANZA and A.F. MULABA-BAFUBIANDI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

111

Integrated piloting of a thermophilic bioleaching process for the treatment of a low-grade nickelcopper sulphide concentrate J.W. NEALE, S.W. ROBERTSON, H.H. MULLER, and M. GERICKE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

119

The effect of elemental sulphur and pyrite on the leaching of nickel laterites using chemolithotrophic bacteria G.S. SIMATE, S. NDLOVU, and M. GERICKE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

155

Recent advances in BIOX technology J. VAN NIEKERK . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

167

Systematic characterization of the mixing process for the BIOX reactor R. VAN DEVENTER, W. KELLER, M. OOSTHUIZEN, and J. STAPELBERG . . . . . . . . . . . . . . . . . . . . . . .

177

Keynote address: Hydrometallurgical process development for complex ores and concentrates D. DREISINGER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

187

®

®

SOLVENT EXTRACTION AND ELECTROWINNING Copper electrowinning: theoretical and practical design N.T. BEUKES and J. BADENHORST . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

213

Developments in organic holding tank structure for solvent extraction processes H.T. LAITALA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

241

i-xii:PRELIMS

2/16/09

12:08 PM

Page 8

i-xii:PRELIMS

2/16/09

12:08 PM

Page 9

Contents (Continued) Page No.

SOLVENT EXTRACTION AND ELECTROWINNING (Continued) The next generation of permanent cathode and lead anode technology T. MARSDEN and J. JICKLING . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

249

Solvent extraction design consideration for the Tati Activox plant E. ROBLES, I. CRONJE, and G. NEL. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

257

The ACORGA OPT series: comparative studies against aldoxime: ketoxime reagents O. TINKLER, D. SHIELS, and M. SODERSTROM. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

273

Solvent extraction (SX) reagent selection for high temperature, acid, chloride and Cu PLS at Port Pirie and its impact on electrowinning (EW) P. CRANE, M. URBANI, K. DUDLEY, A. HORNER, and M. VIRNIG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

283

Organic degradation in uranium and cobalt solvent extraction: the case for aliphatic diluents and antioxidants D.M. VAN RENSBURG, B. MUNYUNGANO . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

295

®

®

BASE METALS REFINERIES Development of the Tati Activox® BMR ammonia recovery circuit D.A. VAN DEN BERG, P. MARÉ, and G.J. NEL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

305

pH advanced process control solution for Impala BMR first stage high pressure acid-oxygen leach A.F. KHAN, T. SPANDIEL, T. VAN SCHALKWYK, and J.A.M. RADEMAN . . . . . . . . . . . . . . . . . . . . . . . .

313

COPPER, COBALT, NICKEL, ZINC HYDROMETALLURGY A cobalt solvent extraction investigation in Africa’s Copper Belt T. KÖNIGHOFER, S.J. ARCHER, and L. BRADFORD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

329

Oxidative precipitation of Mn(II) from cobalt leach solutions using dilute SO2/air mixture N. MULAUDZI and T. MAHLANGU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

341

Soluble metal recovery improvement using high density thickeners in a CCD circuit: Ruashi II a case study M.C. MULLIGAN and L. BRADFORD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

357

Optimizing acid utilization and metal recovery in African Cu/Co flowsheets M. REOLON, T. GAZIS, and S. AMOS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

365

Processing considerations for cobalt recovery from Congolese copperbelt ores B. SWARTZ, S. DONEGAN, and S. AMOS. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

385

GENERAL HYDROMETALLURGY Kinetic modelling of chemical processes in acid solution at t ≤ 200°C i. thermodynamics and speciation in H2SO4-metal (ii) SO4-H2O system J.D.T. STEYL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The evaluation of various oxidants used in acid leaching of uranium R. VENTER and M. BOYLETT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of redox titration techniques for analysis of hydrometallurgical solutions O. BAZHKO. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

401 445 457

Investigation of granular activated carbon from peach stones for gold adsorption in acidic thiourea T.T. MASIYA and F.P. GUDYANGA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

465

Elution behaviour of metals from carbon M. JEFFREY, R. PLEYSIER, and K. BUNNEY. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

475

i-xii:PRELIMS

2/16/09

12:08 PM

Page 11

i-xii:PRELIMS

2/16/09

12:08 PM

Page 10

The organisers of the conference wish to thank the following sponsors for their support -oOo-

Premier FLSmidth Minerals

Impala Platinum

Protea Mining Chemicals

Exhibitors Mintek

MIP Process Technologies

Roymec Technologies

Air Liquide

Sponsors Anglo Operations ChemQuest Africa Csiro Minerals Cytec Industries Hatch Purolite International

i-xii:PRELIMS

2/16/09

12:08 PM

Page 12


Organising Committee M. Dworzanowski (Chairman) P. Maré

S. Ndlovu

T. Thulare

B. McGeorge

J. van Huysteen

T. Claasens

K. Manyukwi

L. van Dyk

1-16_Mahlangu:text

2/12/09

11:37 AM

Page 1

MAHLANGU, T., GUDYANGA, F.P., and SIMBI, D.J. Leaching of the arsenopyrite/pyrite flotation concentrates using metallic iron in a hydrochloric acid medium. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Leaching of the arsenopyrite/pyrite flotation concentrates using metallic iron in a hydrochloric acid medium T. MAHLANGU*, F.P. GUDYANGA†, and D.J. SIMBI† *Department of Materials Science and Metallurgical Engineering, University of Pretoria, Pretoria, South Africa †Department of Metallurgical Engineering, University of Zimbabwe, Harare, Zimbabwe

This paper describes experimental investigations into the reductive decomposition of arsenopyrite/pyrite (FeAsS/FeS2), a gold and silver bearing iron/arsenic sulphide using metallic iron powder. The main objective was to establish and experimentally confirm the thermodynamic and kinetic feasibility of reductively decomposing the sulphide matrix for precious metals recovery. In this context, cyanidation leach tests were conducted on the reductive leach residues. The arsenopyrite/pyrite flotation concentrate, mainly studied in the hydrochloric acid medium, decomposed through the non–oxidative chemical dissolution reaction and the reductive decomposition reaction for the arsenopyrite and pyrite components, respectively. Desulphurization levels below 65% were achieved at pH values below 0.15 and iron to concentrate ratios above 1, and the system was characterized by overall very slow kinetics. The reaction system had a direct linear relationship with iron to concentrate ratio and an inverse relationship with pH. An analysis of the pyrite/iron galvanic system showed that pyrite forms a partially inert cathodic surface on which the anodic dissolution of iron occurs, supported by the hydrogen evolution reaction. This phenomenon explained the low desulphurization levels and the mineral decomposition seemed to be restricted to the non–oxidative chemical dissolution reaction for the arsenopyrite component. The cyanide leach of the reductive leach residues showed very little improvement in gold recovery. The reductive leach process for arsenopyrite/pyrite has considerable limitations in terms of both desulphurization and precious metals liberation. Keywords: Pyrite, arsenopyrite reductive leaching, hydrogen sulphide, invisible gold LEACHING OF THE ARSENOPYRITE/PYRITE FLOTATION CONCENTRATES

1

1-16_Mahlangu:text

2/12/09

11:37 AM

Page 2

Introduction A significant fraction of gold in the gold ores that are mined in central and southern parts of Zimbabwe occurs in submicroscopic form as finely disseminated particles or in solid solution with the refractory sulphide host minerals, namely, arsenopyrite (FeAsS), pyrite (FeS2) and stibnite (Sb2S3) (Kadenhe and Makande, 1987; Makande, 1988; Swash, 1988; Casparini, 1983; Husein, 1981). Apart from the conventional roasting pretreatment process prior to cyanidation, hydrometallurgical pretreatment processes can be used as alternatives. The hydrogen sulphide route for the decomposition of sulphides has been widely studied in two main areas of non-oxidative dissolution (NOD) (Cservanyak, 1994; Nicol and Scott, 1979; Scott and Nicol, 1977; Awakara et al., 1980; Ingraham et al., 1972) and the reductive decomposition in acidified aqueous medium (House and Kelsall, 1985; Majima et al., 1981; Majima et al., 1985; Gudyanga et al., 1999a , Chifamba, 1996; Gudyanga et al., 1999b; House, 1986). The sulphides decompose to give either the elemental metal or metal sulphide of lower oxidation state with hydrogen sulphide evolution (Kolodzeij and Adamski, 1990; Cservanyak, 1994; Chifamba, 1996) according to the general reactions [1] and [2]: [1] [2] Iron, the proposed reductant, is a first order transitional element with an [Ar]3d64s2 outer shell electronic configuration and is able to exhibit multiple oxidation states of 0, +2 and +3 (Mackay and Mackay, 1986). The chemistry and electrochemistry of iron have been widely investigated and reported in corrosion science and engineering (Wranglen, 1985; Shreir et al., 1995a; Shreir et al., 1995b). In hydrometallurgical leaching systems, iron and its dissolved species (Fe(II) and Fe(III) ionic species) have been discussed by correlating thermodynamic predictions in the form of potential–pH and speciation diagrams, and stability constants for various complexes (Seon-Hyo et al., 1986). Seon-Hyo et al. (1986), reported two stable ferrous chloro–complexes formed according to reaction [3]: [3] where βn is the stability constant of the ferrous chloro–complexes and n = 1, 2, 3 and 4; the number of chloro–ligands in each complex. When n is equal to unity and two respectively, the o stability constants for the two complexes FeCl+ and FeCl2 were determined as 2.2908 and 1.0965 respectively (Seon-Hyo et al., 1986). It was also shown that at chloride ion concentration less than 2M, the Fe2+ species predominates over both the FeCl+ and FeCl2 complexes while the latter (FeCl2) becomes dominant at concentrations in excess of 2.5M. Metallic iron is a strong reducing agent as indicated by the redox potential [4]: [4] [5] The conditions under which reductive and chemical decomposition of sulphide minerals occur are also favourable to the chemical dissolution reaction [5] of iron. The hydrogen evolution side reaction does not take into consideration the effect of the anions present. The 2


1-16_Mahlangu:text

2/12/09

11:37 AM

Page 3

Gibb’s free energy change calculations using data from Bard et al. (1985), clearly shows the effect of the hydrochloric acid systems [6]. [6] On the basis of thermodynamic calculations, iron dissolution in chloride systems appears to be more favourable than in non-chloride systems. This has a negative effect on the leach system because the iron dissolution side reaction deprives the reductive leach reaction of any available iron. On the other hand, the Fe(II)/Fe redox system, as a leaching agent in sulphide concentrate leaching, is not expected to form insoluble compounds with the reaction products in the chloride medium. Instead, the formed chloro–complexes expectedly result in more negative potential conditions that are thermodynamically favourable to the reductive leach process. These thermodynamic predictions indicate a need for critical control of both acid concentration and the iron to concentrate ratio during reductive decomposition of sulphide minerals. The general reactions governing the reductive decomposition of arsenopyrite, pyrite and other iron sulphides have been determined by Majima and Awakura (1979). Although the pyrite decomposition reaction is often classified under the reductive decomposition system, the reaction exhibits non-oxidative dissolution (NOD) of the mineral when considered in terms of iron and sulphur oxidation states. This is primarily because neither iron nor sulphur seems to undergo any change in their original oxidation states. It would be expected that pyrite decomposition follows reduction to either stoichiometric pyrrhotite [7] or elemental iron [8] followed by the chemical dissolution of the two products [9] and [10] respectively. However, strictly speaking, in reaction [8], iron is oxidized by hydrogen ions. [7] [8] [9] [10] Reaction [7] has not been experimentally proven to occur in acidic aqueous systems (Cservanyak, 1994). Also from the potential–pH diagram (Figure 1), pyrite decomposition to elemental iron and hydrogen sulphide is feasible only at potentials below -0.41V vs SHE. The ferrous ions are the main products at any reducing potentials above -0.41 V vs SHE depending on the working pH. Elsewhere (Holdich and Broadbent, 1985; Koch, 1975; Peters and Majima, 1968), ferrous ions were detected as the principal reaction products instead of pyrrhotite and elemental iron. Czevanyak (1994) in discussing the work of Peters and Majima (1968) reported that hydrogen sulphide evolution always occurred after hydrogen evolution has already started. This scenario clearly indicates that the process efficiency of reductive decomposition will be adversely affected by any pH reduction. From the preceding discussion and available thermodynamic data, the application of elemental iron in the reductive decomposition of pyrite can be inferred. Bourgeois et al. (1979) as referenced by Czevanyak (1994), were the first to investigate elemental iron driven dissolution of pyrite. Pyrite and elemental iron, form a galvanic couple with iron being oxidized to ferrous ions and pyrite reduced to ferrous ions with the evolution of hydrogen sulphide [11]. [11] LEACHING OF THE ARSENOPYRITE/PYRITE FLOTATION CONCENTRATES

3

1-16_Mahlangu:text

2/12/09

11:37 AM

Page 4

Figure 1. Potential–pH diagram for the Fe–S–H2O system. Iron activity = 0.001, sulphur = 0.001 (Stabcal software)

The reductive decomposition of arsenopyrite, which occurs together with pyrite in typical refractory gold ores (Swash, 1988; Yannopoulos, 1990; Chifamba, 1996), can be discussed under the same principles as that of pyrite. The half reaction, representing the decomposition of arsenopyrite to elemental arsenic is shown in reactions [12] and [13]. [12] [13] The leaching system will also depend on the solution redox potential. The work of Majima and Peters (1968) gave the following findings: • Polarizing a pyrite electrode in 1M HClO4, they determined a hydrogen overpotential of 0.26V; • In constant potential experiments, monitoring current and the concentration of H2S evolved, the current efficiency of H2S evolution varied with potential from 34% (at -0.2 V vs SHE) to 20% (at -0.4 V vs SHE); • Varying the acid concentration from 0.1 to 4M at -0.3 V vs SHE, hydrogen evolution rate varied linearly with acid concentration and the reaction order for H2S > 1. The preceeding findings limit not only the working pH range but also the potential. Gudyanga et al. (1999b), using Cr(II) ionic species as the reductant, measured the solution redox potential during reductive decomposition of an arsenopyrite/pyrite concentrate and found a rapid increase in potential with time on reducing pH from 0.43 to 0.2. These results also agree with the findings of Klein and Sluvey (1978) who concluded that the hydrogen evolution reaction dominates above pH 0. The present work extends the reductive decomposition and/or dissolution of a gold bearing arsenopyrite/pyrite flotation concentrate using metallic iron in a chloride medium. The work focuses on the factors that enhance desulphurization and the recovery of precious metals, namely gold and silver from the selected sulphide concentrate. The application of iron in the reductive decomposition of arsenopyrite/pyrite flotation concentrate is envisaged to provide an alternative pretreatment method that addresses the technological and environmental limitations that are characteristic of pyrometallurgical routes while enhancing precious metals recovery. 4


1-16_Mahlangu:text

2/12/09

11:37 AM

Page 5

Experimental procedure Materials Flotation concentrate samples The reductive decomposition experiments were conducted on an arsenopyrite/pyrite flotation concentrate collected as a blend sample from the Kwekwe Roasting Plant, Zimbabwe. The mineralogical and chemical composition of the material are tabulated in Tables I and II. A bulk sample of iron shavings was collected, washed, dried, screened to 100% minus 425 μm and then stored under dry and oxygen free conditions in order to minimize the oxidation of the iron particles surfaces. All leaching solutions were prepared using analytical grade reagents of hydrochloric acid, ferrous sulphate hepta-hydrate and potassium dichromate diluted with distilled water and where necessary, deoxygenated with high purity nitrogen. The experimental setup is detailed in Mahlangu et al. (2006). Reductive leaching The flotation concentrate was pulped in 500 ml water together with the predetermined mass of iron shavings and purged with a steady stream of nitrogen gas for a period of 45–60 minutes. Nitrogen gas purging was used for the removal of oxygen prior to the addition of hydrochloric acid medium. The pH was monitored by periodic withdrawal of 10 ml samples from the reactor, cooling, recording of pH and then reintroducing the sample back into the reactor. Effluent hydrogen sulphide was scrubbed through a 1.8 litre hydrostatic column of hydrochloric (HCl) acidified potassium dichromate solution. The leaching/decomposition reaction was quantitatively followed by a redox titration of residual dichromate ions in the column with ferrous ions (Mahlangu et al. 2006). Sodium diphynlamine was used as an indicator with the colour changing from pale green to purple on reaching the endpoint. The residues and filtrates were then analysed for Au, Ag, Fe, As, Sb, Pb and S. Cyanidation leaching The reductive leach residue was thoroughly washed with distilled water and cyanide leached in a 0.23% NaCN solution for 48 hours at pH between 11 and 12. The solution pH was adjusted using technical grade sodium hydroxide. The leach residues and filtrates were then analysed for Au and Ag.

Table I Mineralogical composition of the FeAsS/FeS2 flotation concentrate FeS2 (%)

FeAsS (%)

PbS (%)

CuFeS2 (%)

ZnS (%)

Sb2S3 (%)

Other (%)

27.2

0.1

0.3

1.8

0.8

11.1

58.7

Table II Chemical composition of the FeAsS/FeS2 flotation concentrate Fe (%) 36.7

As (%)

Pb (%)

Cu (%)

Zn (%)

Sb (%)

S (%)

Au (g/t)

Ag (g/t)

Other (%)

12.5

0.1

0.1

1.2

0.6

28.7

64.7

43.1

20.2

LEACHING OF THE ARSENOPYRITE/PYRITE FLOTATION CONCENTRATES

5

1-16_Mahlangu:text

2/12/09

11:37 AM

Page 6

Results and discussion Effect of pH on the decomposition of FeAsS/FeS2 The data in Figure 2 show the variation of arsenopyrite/pyrite decomposition with pH at an iron to concentrate ratio of 0 and set leaching temperature of 105°C. The results exhibit an inverse near linear relationship between arsenopyrite/pyrite decomposition and pH. In the absence of iron, arsenopyrite decomposes according to the chemical non–oxidative dissolution reaction [13]. Under these conditions, pyrite is expected to remain inert as would be expected from the examination of the Fe–S–H2O potential–pH diagram (Figure 1). [14] In this regard, the sulphide mineral non–oxidative dissolution, observed in this work, was mainly due to the dissolution of the arsenopyrite component. The results also exhibit very low levels of desulphurization (< 16%) associated with the chemical non–oxidative or direct acid leaching of the concentrate. This is despite the relatively large negative free energy change value calculated for the system [13]. The introduction of iron, at an iron to concentrate ratio of 0.16, while working in the same pH range, gave results shown in Figure 3. An inverse linear relationship between sulphide minerals decomposition and pH was observed, with very little improvement in the extent of reaction. The results of further arsenopyrite/pyrite reductive decomposition experiments at progressively increasing iron to concentrate ratios are graphically presented in Figures 3(b), (c) and (d). These tests were conducted in the same pH range and temperature. The iron to concentrate ratio of 0.32 corresponds to the calculated stoichiometric requirement of iron in the system. In all the cases, the inverse near linear relationship between arsenopyrite/pyrite decomposition and pH continue to be exhibited. However, a further pH reduction to values below 0.25, marginal increases in arsenopyrite/pyrite decomposition are observed. The results also show a sustained increase in the sulphide minerals’ decomposition with an increase in the iron to concentrate ratio. There is a very strong influence of iron to concentrate ratio demonstrated at pH values equal to or greater than pH 0.25. In order to understand and explain these phenomena, there is need for the examination of the chemistry and electrochemistry of the two minerals in acidified chloride systems.

Figure 2. Variation of %sulphur leached arsenopyrite/pyrite with pH at different leaching times (iron/concentrate ratio = 0; temperature = 105°C)

6


1-16_Mahlangu:text

2/12/09

11:37 AM

Page 7

Figure 3. Variation of %sulphur leached from arsenopyrite/pyrite with pH at different leaching times (ron/concentrate ratio: (a) = 0.16; (b) = 0.32; (c) = 0.64; (d) = 0.96; temperature = 105°C)

In the reductive decomposition of pyrite and arsenopyrite, the reactions governing the leach systems have been calculated and their thermodynamic feasibility demonstrated in reactions [11] and [14]. Pyrite decomposes to ferrous ions with the evolution of hydrogen sulphide, whereas arsenopyrite decomposes both by chemical non–oxidative and reductive dissolution reactions, giving the same reaction products as in pyrite. [15] [16]

[17] The resultant elemental iron from the arsenopyrite decomposition reaction [17] subsequently dissolves either by supporting the sulphide mineral decomposition reactions [15 or 16] or by the hydrogen evolution side reaction [18]. [18] The involvement of the resultant iron in the reductive leach reaction has a complementary effect on the leach process, whereas the latter negatively affects the process efficiency. Thermodynamic calculations, as indicated by the free energy changes for arsenopyrite and pyrite decomposition reactions, predict a much more feasible process than illustrated by the results in Figures 2 to 3. This is in spite of the fact that the Fe–Cl–S–H2O potential–pH diagram (Figure 4) shows that in the presence of the chloride ions, ferrous ions and hydrogen, sulphide become more stable relative to pyrite. The differences are more evident when compared to the Fe–S–H2O system (Figure 1). LEACHING OF THE ARSENOPYRITE/PYRITE FLOTATION CONCENTRATES

7

1-16_Mahlangu:text

2/12/09

11:37 AM

Page 8

Figure 4. The Fe–Cl–S–H2O Eh –pH diagram ([Fe] = 0.001M; [S] = 0.01M; [Cl] = 0.01M) (Stabcal, Software)

Table III Redox reactions and Eo values arsenopyrite, pyrite and iron systems in acidic conditions [Standard temperature = 25°C and pressure] o

Redox reaction FeS2 + 4H+ + 2e- = Fe2+ + H2S E/V vs SHE = -0.108 – 0.118pH – 0.0296 log (Fe2+)(H2S)2 o

o

ΔG kJ/mol

E /V vs SHE

+20.91

-0.108

+16.44

-0.085

+78.87

-0.409

o

FeAsS + 2H+ + 2e- = Fe + As + H2S E/V vs SHE = -0.085 – 0.0591pH – 0.0296 log (H2S) Fe2+

+ 2e- = Fe E/V vs SHE = -0.409 – 0.0296 log (Fe2+)

Consideration of the galvanic interactions or the galvanic cell formed by arsenopyrite and/or pyrite with iron could possibly explain these low levels of desulphurization. In the galvanic cell, the anodic dissolution of iron can be supported by either the cathodic dissolution of the mineral sulphide or an inert sulphide mineral surface, which promote the iron dissolution reaction with the evolution of gaseous hydrogen. The standard electrode potential values of the pyrite, arsenopyrite and iron are shown in Table III. The relative potential difference between the FeS2/Fe2+, H2S and Fe/Fe2+ (301 mV) systems hugely favours the galvanic reaction since galvanic reactions are known to proceed at potential differences of 200 mV (Jackson, 1986). However, pyrite has been known to remain inert even at potential differences of more than 400 mV in the case of pyrite/sphalerite galvanic interactions. [19] [20] In the above system sphalerite dissolution is cathodically supported by the oxygen reduction reaction [21] on the pyrite mineral particles surface. [21] 8


1-16_Mahlangu:text

2/12/09

11:37 AM

Page 9

Drawing parallels with the pyrite/iron system and the pyrite/sphalerite system, it can be concluded that iron dissolution is supported by the hydrogen evolution reaction on the inert pyrite mineral particle surface. The arsenopyrite/iron redox system has a similar potential difference (324 mV) to that discussed for the pyrite/iron system (301 mV). In this regard, arsenopyrite is likely to remain partly or total inert while iron anodically dissolves on its surface supported by the hydrogen evolution reaction. It has also been demonstrated elsewhere (Peters and Majima, 1968; Klein and Shuey, 1978; Dreisinger and Abed, 2002) that the hydrogen evolution reaction will occur at same potentials at which pyrite reduction takes place. They also found out that hydrogen sulphide evolution commenced only after hydrogen evolution had begun. This possibly explains the low levels of arsenopyrite/pyrite decomposition. While thermodynamics could predict the feasibility of any reaction system, kinetic feasibility can be only experimentally confirmed. Effect of iron to concentrate ratio on the decomposition of arsenopyrite/pyrite. Figure 5 shows the variation of arsenopyrite/pyrite decomposition with iron to concentrate ratio at pH 0.25 and set working temperature of 105°C. The data exhibit a direct linear relationship between arsenopyrite/pyrite decomposition and iron to concentrate ratio. These results are typical of a reaction system directly controlled by the concentration of a single reactant (Weis, 1985). As indicated in the earlier results (Figures 2 to 3), iron to concentrate ratio seems to have a very strong influence on the arsenopyrite/pyrite decomposition relative to pH in the pH range studied. Although decomposition increases with an increase in iron to concentrate ratio, the effective desulphurization levels are significantly low (<50%). Further arsenopyrite/pyrite–iron to concentrate ratio plots of the experimental results at progressively increasing pH values (Figures 6(a) to 6(d)) show similar trends to those observed in Figure 5. However, the effective dissolution levels progressively decrease in each subsequent figure. The progressive increase in the levels of mineral sulphide decomposition at increasing iron to concentrate ratio and decreasing pH was further investigated at pH values 0.1 and 0.15 and also iron to concentrate ratios of 1.28 and 1.60. The results of these tests are plotted in Figures 7 to 8 and at pH values 0.1 and 0.15, increasing the iron to concentrate ratio from 0.96 to 1.28 improves total sulphur leached. Any further increase in iron to concentrate ratio beyond 1.28 is accompanied by marginal increases in sulphur leached. At longer leaching times of

Figure 5. Variation of %sulphur leached from arsenopyrite/pyrite with iron to concentrate ratio at different leaching times (pH = 0.25; temperature = 105°C)


9

1-16_Mahlangu:text

2/12/09

11:37 AM

Page 10

Figure 6. Variation of %sulphur leached from arsenopyrite/pyrite with iron to concentrate ratio at different leaching times ((a) pH = 0.34; (b) pH = 0.44; (c) pH = 0.54; (d) pH = 0.62; temperature = 105°C)


300 minutes, a slight decrease in total arsenopyrite/pyrite decomposition is also observed. While the marginal increases in the sulphide mineral decomposition can be appreciable, further pH reductions coupled with the high temperature (105°C) provide a highly corrosive environment that will increase both capital and operational costs of the system. Influence of arsenopyrite/pyrite decomposition on the precious metals recovery The arsenopyrite/pyrite ores generally contain significantly large fractions of gold occurring in submicroscopic form as finely disseminated particles or in solid solution (Iglesias, 1994, Swash, 1988). In these ores and concentrates of arsenopyrite/pyrite, gold often deposits along 10


1-16_Mahlangu:text

2/12/09

11:37 AM

Page 11


Table IV Sieve analysis of arsenopyrite/pyrite flotation concentrate and gold distribution (where dist = distribution) Size range (μm)

Weight (%)

Au dist (%)

Au assay (g/t)

Size (μm)

Cumulative oversize (%)

Cumulative Au dist (%)

+212

0.4

-

-

212

0.4

-

-212+150

6.4

3.4

35.2

150

6.8

3.4

-150+106

11.0

10.4

62.56

106

17.8

13.8

-106+75

9.6

6.1

41.84

75

27.4

19.9

-75+53

14.0

14.7

69.92

53

41.4

34.6

-53

58.6

65.4

121.76

Total

100

100

66.80

the arsenic rich grain boundaries and usually form irregular inclusions. These characteristics define the refractoriness of many arsenopyrite/pyrite ores. The study (Chifamba, 1996; Gudyanga et al., 1998; Mbewe, 1990; Gudyanga et al., 1999b) of flotation concentrates utilized in the present work, proved to be refractory to direct cyanide leach. Chifamba (1996) and Gudyanga et al. (1999b) further demonstrated that reductive decomposition of the same flotation concentrates with desulphurization values in excess of 95% achieved very low gold recovery. The reductive leach process dissolved iron and released sulphur as hydrogen sulphide but elemental or metallic arsenic remained in the residue. The residues were very refractory giving less than 30% gold extractions into the solution. In this regard, the sulphide matrix is not solely responsible for the refractory nature of the arsenopyrite/pyrite flotation concentrates. In fact, subsequent oxidation of the reductive leach residues realized 95% gold extraction values. These findings concurred with the earlier mineralogical results (Jha, 1987; Cook and Chryssoulis, 1990; Cabri et al., 1989). It is therefore clear that most of the gold, in the arsenopyrite/pyrite flotation concentrate exists as ‘invisible or locked’ gold. In this form, total decomposition of the host mineral is required for the complete liberation and recovery of precious metals. Reductive pretreatment does not achieve the complete decomposition of the host sulphide mineral and therefore presents serious limitations as a pretreatment process. The results presented in this present work focus on gold and silver deportment in the various sieve fractions and also the response of gold extraction to desulphurization. LEACHING OF THE ARSENOPYRITE/PYRITE FLOTATION CONCENTRATES

11

1-16_Mahlangu:text

2/12/09

11:37 AM

Page 12

Sieve analysis and precious metal deportment in the flotation concentrate The results of the sieve analysis of the arsenopyrite/pyrite flotation concentrate and gold distribution in various size fractions are tabulated in Table IV. The results show that over 50% of the material is below 53 μm and this size fraction contains more than 60% gold. Silver distribution is tabulated in Table V and over 90% silver occurs in the minus 53 μm fraction. In this regard, fine grinding is inevitable to liberate both gold and silver from the flotation concentrate. Fine to ultrafine grinding has been used to treat some refractory ores but presents two problems, namely (i) high operational costs and (ii) does not necessarily liberate ‘invisible’ or ‘locked’ gold. In this context, chemical pretreatment methods have often taken precedence over fine to unltrafine grinding (Wills, 1997; Yannopoulos, 1990). The failure by reductive decomposition (Chifamba, 1996; Gudyanga et al., 1999b) to liberate the ‘invisible or locked’ gold necessitated a subsequent oxidative pretreatment process on the residue. Also, the work of Dunn and Chamberlain (1997) revealed that the sulphide matrix is not entirely responsible of the refractory nature of the arsenopyrite/pyrite host mineral. In fact, they realized only around 33% gold recovery after the complete removal of sulphur during pyrolysis. It therefore follows that cyanidation of the reductive leach residues gives only a measure of the refractoriness caused by the sulphide matrix.

Table V Sieve analysis of arsenopyrite/pyrite flotation concentrate and silver distribution (where dist = distribution) Size range (μm)

Weight (%)

Ag dist (%)

Ag assay (g/t)

Size (μm)

Cumulative oversize (%)

Cumulative Ag dist (%)

+212

0.4

-

-

212

0.4

-

-212+150

6.4

-

-

150

6.8

-

-150+106

11.0

-

-

106

17.8

-

-106+75

9.6

1.1

5.12

75

27.4

1.1

-75+53

14.0

7.5

24.40

53

41.4

8.6

-53

58.6

91.4

71.30

Total

100

100

45.67

Figure 9. Effect of desulphurization on gold extraction during cyanidation of the arsenopyrite/pyrite reductive leach residues

12


1-16_Mahlangu:text

2/12/09

11:37 AM

Page 13

Effect of desulphurization on gold and silver extraction Figure 9 shows the variation of gold recovery from the arsenopyrite/pyrite reductive leach residue during cyanide leaching. The data at zero per cent sulphur removal represents direct cyanidation of the flotation concentrate. The desulphurization levels range between 14% and 63%. Gold extraction is limited to around 15% maximum at around 63% sulphur removal. It is evident that the reductive decomposition does not liberate significant amounts of gold. These results seem to concur with earlier work of Gudyanga et al. (1999b) and Chifamba (1996), working on similar flotation concentrates, achieved in excess of 90% desulphurization but the gold extraction was limited to around 27%. The recovery data for silver is presented in Figure 10. Similar patterns to those obtained in Figure 9 were observed. From the particle size analysis and precious metals deportment data (Tables IV and V), it was shown that over half of both gold and silver occur in the fine fraction, that is, minus 53 μm. When this data is considered together with the extraction data (Figures 9 and 10), qualitative conclusions can be made to the effect that gold exists as ‘invisible or locked’ gold and as such will require the total decomposition of the arsenopyrite/pyrite host mineral. Conclusions • The feasibility of the reductive decomposition of the arsenopyrite/pyrite flotation concentrate using elemental iron has been demonstrated in the hydrochloric acid medium with the reactions governing the leach process were: – Chemical non-oxidative dissolution reaction for arsenopyrite; – Reductive dissolution for pyrite. • The reductive leach process followed an inverse relationship with pH and also a direct relationship with iron to concentrate ratio. Desulphurization levels remained relatively low (< 65%) at very low pH (< 0.15) and high iron to concentrate ratios (> 1.5). The analysis of the pyrite/iron galvanic system revealed that pyrite acts as an inert cathode on which iron oxidation is supported by the hydrogen evolution reaction. This phenomenon satisfactorily explained the low levels of the mineral sulphide decomposition. • Cyanide leaching of the reductive leach residues showed that there was very little improvement in both gold and silver extraction. Prior to reductive decomposition, gold extraction was around 5% and after 63.5% desulphurization, only 15% gold extraction was realized. In this context, the sulphide matrix was not solely responsible for the refractoriness of the host mineral.

Figure 10. Effect of desulphurization on silver extraction during cyanidation of the arsenopyrite/pyrite reductive leach residues


13

1-16_Mahlangu:text

2/12/09

11:37 AM

Page 14

• Gold and silver exist as ‘invisible or locked gold’ and as such total dissolution of the arsenopyrite/pyrite flotation concentrate was necessary in order to access the precious metals. • The limitations of the reductive leach process both in terms of desulphurization and precious metals recovery makes the whole process less attractive. There is need for coupling it with subsequent processes that will totally dissolve the host mineral and increase the precious metals recovery. References AWAKARA, Y., KAWEI, S., and MAJIMA, H. Kinetic study of non–oxidative dissolution of galena in aqueous acid solution. Met. Trans. B, vol. 11B, 1980. pp. 377–381. BARD, A.J., PARSONS, R., and JORDAN, J. Standard potentials in aqueous solutions, Marcel Dekker, New York. 1985. BOURGEOIS, J.P., AUPAIX, N., BLOISE, R., and MILLET, J.L. Proposition D’Explication de la Formation D’Hydrogene Sulfure dans les Stockages Souterrains de Gaz Naturel Par Reduction des Sulfures Mineraux de la Roche Magasin, Revae de l’Inst. Francais du Petrole, vol. 34, no. 3, 1979, pp. 371–386. CABRI, L.J., CHRYSSOULIS, S.L., DE VILLERS, J.P.R., LAFLAMME, J.H.G., and BUSEK, P.R. The nature of ‘invisible gold’ in arsenopyrite, The Canadian Mineralogist, 27, 1990. 353 pp. CASPARINI, C. The mineralogy of gold and its significance in metal extraction. C.I.M. Bull. March (1983), 1989. 1983. pp. 145–153. CHIFAMBA, J. The reductive decomposition of refractory sulphide concentrates for the recovery of precious metals, gold and silver. MPhil Thesis, University of Zimbabwe 1996. COOK, N.J. and CHRYSSOULIS, S.L. Concentrations of ‘invisible gold’ in common sulphides, The Canadian Mineralogist, vol. 28, 1990. p. 1. CSERVNYAK, I. Electrochemical reduction of pyrite in acidic aqueous electrolytes. PhD Thesis, University of London, 1994. DREISINGER, D. and ABED, N. 2002. A fundamental study of the reductive leaching of chalcopyrite using metallic iron part I: Kinetic analysis. Hydrometallurgy 66, vol. 1–3, 2002. pp. 37–57. DUNN, J.D. and CHAMBERLAIN, A.C. The recovery of gold from refractory concentrates by pyrolysis–oxidation, Extended Abstracts, Complex Ores ’97, an International Symposium on the Processing of Complex and Refractory Ores, Bulawayo, Zimbabwe March 1997. pp. 8–10. GUDYANGA, F.P., MAHLANGU, T., CHIFAMBA, J., and SIMBI, D.J. Reductive–oxidative pre-treatment of a stibnite flotation concentrate: Thermodynamic and kinetic considerations. Minerals Engineering, vol. 11, no. 6, 1998. pp. 563–580. GUDYANGA, F.P., MAHLANGU, T., CHIFAMBA, J., and SIMBI, D.J. 1999a. Reductive decomposition of galena (PbS) using Cr(II) ionic species in an aqueous chloride medium for silver (Ag) recovery. Minerals Engineering, vol. 12, no. 71999. pp. 787–797. GUDYANGA, F.P., MAHLANGU, T., CHIFAMBA, J., and SIMBI, D.J., 1999b. Sequential–reductive pretreatment of an arsenopyrite/pyrite flotation concentrate. Paper presented at the Minerals Engineering ’99 Conference, Falmouth, England, September 1999. HOLDICH, R.S. and BROADBENT, C.P. Investigating the dissolution of pyrite in copper(II) chloride solutions, Extractive Metallurgy ’85. IMM, London, 1985. pp. 645–658. 14


1-16_Mahlangu:text

2/12/09

11:37 AM

Page 15

HOUSE, C.I. 1986. The electrogeneration of Cr(II) and V(II) solutions and the hydrometallurgical reduction of SnO2, PbS and CuFeS2. PhD Thesis, University of London. 1986. HOUSE, C.I. and KELSALL, G.H. Hydrometallurgical reduction of SnO2, CuFeS2 and PbS by electrogenerated Cr(II) and V(II) solutions, Extractive Metallurgy ’85, IMM, London, 1985. pp. 685–692. HUSEIN, D.M. Process mineralogy of auriferous pyritic ores at Carlin, Nevada. Process Mineralogy, Metal Extraction, Mineral Exploration and Energy Materials. New York: AIME. 1981. pp. 271–289. IGLESIAS, N. and CARRAZA, F. Refractory gold ores; A review of treatment methods and recent advances in Biotechnological techniques, Hydrometallurgy 34, 1994. pp. 383–395. INGRAHAM, T.R., PARSONS, H.W., and CABRI, L.J. Leaching of pyrrhotite with hydrochloric acid. Can. Met. Quarterly, vol. 11, no. 2, 1972. pp. 425–434. JACKSON, E. Hydrometallurgical Extraction and Reclamation. John Wiley and Sons, New York. 1986. JHA, M.C. Refractoriness of certain gold ores to cyanidation: Probable causes and possible solutions. Mineral Processing and Extractive Metallurgy Review, vol. 2, 1987. pp. 331–352. KADENHE, R.M. and MAKANDE, E.S. Review of the roasting plant operations and services available to small mines at Kwekwe, Zimbabwe, African Mining, IMM, London. 1987. KLEIN, J.D. and SLUVEY, R.T. Non–linear Impedence of Mineral–Electrolyte Interfaces; Part I., Pyrite, Geophysics, vol. 43, no. 6, 1978. pp. 1222–1234. KOCH, D.F.A. Electrochemistry of sulphide minerals, Modern Aspects of Electrochemistry. vol. 10, J.O.M. Brochris and B.E. Conway (eds.), 1975. pp. 211–237. KOLODZEIJ, B. and ADAMSKI, Z. Dissolution of sphalerite in aqueous hydrochloric acid solutions under reducing conditions. Hydrometallurgy, vol. 24, no. 3, 1990. pp. 393–406. MACKAY, K.M. and MACKAY, R.A. Introduction to Modern Inorganic Chemistry, 3rd Edition. International Textbook Company. 1986. MAHLANGU, T., GUDYANGA, F.P., and SIMBI, D.J. Reductive leaching of stibnite using metallic iron in a hydrochloric acid medium I: Thermodynamics. Hydrometallurgy, vol. 84, no. 3–4, 2006. pp. 192–203. MAJIMA, H. and AWAKARA, Y. Non–oxidative leaching of base metal sulphide ore, in 13th Int. Min. Proc. Congress, Warsaw. J. Laskowski (ed.), 1979. pp. 936–956. MAJIMA, H., AWAKARA, Y., and MISAKI, N. A kinetic study on non–oxidative dissolution of sphalerite in aqueous hydrochloric acid solutions. Met. Trans. B. 12B, 1981. pp 645–649. MAJIMA, H., AWAKURA, Y., and MASAKI, N. Leaching of oxides and sulphides in acidic chloride media. Extractive Metallurgy ’85, London 9–12 September, 1985. IMM, pp. 607–627. MAKANDE, E.S. Roasting and cyanide treatment of arsenical and antimonical gold concentrates and residues at the Roasting Plant, Kwekwe, Zimbabwe, Perth International Gold Conference, RANDOL. 1988. MBEWE, K. Improving the recovery of silver in the pressure oxidation leach for the refractory gold ores, BSc Eng. (Hons) Project Report, Department of Metallurgy, University of Zimbabwe. 1990. NICOL, M.J. and SCOTT, P.D. The kinetics and mechanism of the non–oxidative dissolution of iron sulphides in aqueous acidic solutions. J. South Afr. Inst. Min. Met. 1979, pp. 298–305. LEACHING OF THE ARSENOPYRITE/PYRITE FLOTATION CONCENTRATES

15

1-16_Mahlangu:text

2/12/09

11:37 AM

Page 16

PETERS, E. and MAJIMA, H. Electrochemical reactions of pyrite in acid perchlorite solutions. Can. Met. Quart, vol. 7, no. 3, 1968. pp. 111–117. SCOTT, P.D. and NICOL, M.H. The kinetics and mechanisms of non–oxidative dissolution of metal sulphides. Trends in Electrochemistry, J.O.M. Bockris, D.A.J. Rand, and B.J. Welch (eds.), Plenum Press, New York and London, 1977. pp. 303–316. SEON-HYO, K., HENEIN, H., and WARREN, G.W. An investigation of the thermodynamics and kinetics of the ferric chloride leaching of galena concentrate. Metallurgical Transactions, 17B, 1986. pp. 29–39. SHREIR, L.L., JARMAN, R.A., and BURSTEIN, G.T. Corrosion 1; Metal/Environment Reactions. vol. 1. Butterworth Heinemann. 1995a SHREIR, L.L., JARMAN, R.A., and BURSTEIN, G.T. Corrosion 2; Corrosion control. vol. 2. Butterworth Heinemann. 1995b. SWASH, P.M. A mineralogical investigation of refractory gold ores and their beneficiation, with special reference to arsenical ores. J. S. Afri. Inst. Min. Metall. vol. 88, no. 5. 1988. pp. 173–180. WEISS, N.L. SME Mineral Processing Handbook; vol. 2. Society of Mining Engineers of the Americal Institute of Mining, Metallurgical, and Petroleum Engineers, Inc., New York. 1985. pp. 13–24. WILLS, B.A. 1997. Mineral Processing Technology: An introduction to the practical aspects of ore treatment and mineral recovery. Sixth Edition. Butterworth Heinemann. 1997. pp. 11–13. Wranglen, G. An Introduction to Corrosion and Protection of Metals. Chapman and Hall 1985. 260 pp. YANNOPOULOS, J.C. The Extractive Metallurgy of Ggold, Van Nostrand Reinhold, New York, 1990. pp. 79–110.

Thamsanqa Mahlangu Senior Lecturer, University of Pretoria, South Africa Thamsanqa is a Member of the Southern Africa Institute of Mining and Metallurgy (SAIMM), he is also registered as a Professional Engineering (Pr Eng) with the Engineering Council of South Africa (ECSA). • August 2004 to date—Senior Lecturer, Department of Materials Science and Metallurgical Engineering, University of Pretoria. • June 2003 to May 2004)—Post Doctoral Fellow, Mineral Processing Research Unit, Department of Chemical Engineering, University of Cape Town • March 1995 to May 2003—Research Fellow/Lecturer, Department of Metallurgical Engineering, University of Zimbabwe • March 1994 to February 1995—Minerals Process Research Engineer, Institute of Mining Research, University of Zimbabwe. My research focus is on the solution purification by both solvent extraction and precipitation techniques. I have also been involved in the leaching of refractory gold ores as well as the general hydrometallurgical processes of leaching and metals recovery. 16


17-30_Ngenda:text

2/12/09

11:00 AM

Page 17

NGENDA, R.B., SEGERS, L., and KONGOLO, P.K. Base metals recovery from zinc hydrometallurgical plant residues by digestion method. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Base metals recovery from zinc hydrometallurgical plant residues by digestion method R.B. NGENDA*, L. SEGERS†, and P.K. KONGOLO* *University of Lubumbashi, Lubumbashi, Democratic Republic of Congo (DRC) †The Free University of Brussels, Brussels, Belgium The ‘Kolwezi Zinc Plant’, in French ‘Usines à Zinc de Kolwezi’ (UZK), has produced about 910 000 dry metric tons of residues during the hydrometallurgical treatment of calcines from sulphide zinc concentrates. The zinc hydro plant residues contain on average 19.47% Zn, 2.7% Cu, 2.1% Pb, 26.6% Fe, 0.12% Cd, 0.39% As, 157 ppm Ag, 447.8 ppm Ga, 475.4 ppm Ge, etc. This material was dumped in five ponds in the vicinity of the plant. The high Fe content in the ore concentrates induced important conversion of some of zinc and copper into ferrites during the roasting process. Those ferrites are very stable compounds, which are difficult to leach with dilute sulphuric acid solution. Therefore, they mainly report to the leaching residues. Hot leaching has been successfully applied to the treatment of zinc hydro plant residues. The method, however, presents the disadvantage of simultaneously dissolving iron (Fe). It is then necessary to remove an important iron quantity from solution prior to zinc and other metals recovery. The digestion method, which has been recently developed for the treatment of copper smelter slags, has been successfully applied to zinc hydro plant residues. It has been found in this research work to be most efficient to recover metal values from such materials. The method mainly consists of 24 h digestion with half concentrated sulphuric acid solution (48%) leading to the formation of metal sulphates. The digested material is subsequently roasted for 2 h at around 750°C to selectively convert iron sulphate into the water nonsoluble form Fe2O3 (hematite). After leaching with water at 40°C, nearly 98.7% Zn, 99.9% Cu, 100% Cd and only 6.4% Fe have been recovered into solution. Most of the Fe, Ag, Pb, Ge and Ga were concentrated in the leaching residues. Zn, Cu and Cd could be recovered from solution by the usual techniques such as solvent extraction with subsequent metal electrowinning or salts precipitation. Moreover, investigations are underway with the aim of developing an efficient method for the recovery of other metals that remain in the digestion leaching residues. Keywords: hydro plant residues, zinc, copper, cadmium, iron, digestion, sulphuric acid, roasting, water, leaching. BASE METALS RECOVERY FROM ZINC HYDROMETALLURGICAL PLANT RESIDUES

17

17-30_Ngenda:text

2/12/09

11:00 AM

Page 18

Introduction The ‘Kolwezi Zinc Plant’, in French ‘Usines à Zinc de Kolwezi’ (UZK), is located near the town of Kolwezi in the Katanga Province of the Democratic Republic of Congo (DRC). The plant belongs to GECAMINES, a state owned mining company operating in South Katanga which belongs with Zambia to the African Copper Belt. Traditionally, GECAMINES is known as one of the major copper and cobalt producers in the world. Other metal commodities such as zinc and cadmium are also produced by the company. The main source of zinc is the Kipushi underground mine, which is situated some 20 km west from Lubumbashi, the capital of the Katanga Province. Polymetalic sulphide ores are extracted from this mine, which contains important quantities of copper, germanium, gallium, silver, etc. Differential flotation is applied in Kipushi for the production of two kinds of concentrates: a copper concentrate, which is treated in the copper smelter of Lubumbashi for blister copper, and a zinc concentrate, which is processed for electrolytic zinc production. The zinc concentrate is first sent to Likasi, about 120 km north of Lubumbashi, to undergo a fluidized bed roasting in the Shituru acid plant of the Shituru hydrometallurgical plant. While the obtained SO2 gas is used for sulphuric acid production, the zinc calcines are sent to Kolwezi for further processing. The town of Kolwezi is about 380 km north west from Lubumbashi. The chain of zinc production at GECAMINES is represented in Figure 1. In the same figure, the UZK process is briefly described as essentially consisting of sulphuric acid leaching of zinc calcines, solution purification and zinc electrolysis. Solution purification is performed by

Figure 1. Simplified flowsheet of zinc production at GECAMINES, RDC

18


17-30_Ngenda:text

2/12/09

11:00 AM

Page 19

cementation of copper, cadmium and cobalt on zinc powder, with copper and cadmium being recovered as by-products. Due to the difficult situation in the country since the nineties, the collapse of the mining industry was the main reason why zinc production in Kolwezi went down. The plant, which started production in 1953, has stopped since the early nineties. UZK has been producing up to 65 000 metric tons of zinc per year, along with important quantities of leaching residues. Approximately 910 000 dry metric tons of leaching residues were produced and dumped in five ponds near the plant. The materials inside the ponds are about 5.2 m thick layers, 1.46 of density and 36.2% humidity. Some of the photographs of residues ponds are shown in Figure 2. Environmental issues are well illustrated. The residues ponds have not been stabilized, therefore crevasses have built up and the rain has done the rest by washing these materials away to the nearby Musonoie River. Figure 3 shows disastrous dusts that even invade offices. Chemical analyses of UZK residues have been carried out at the Research Department of GECAMINES in Likasi, DRC and at the Industrial Chemistry Laboratory of the Free University of Brussels in Brussels, Belgium. The results have confirmed the presence of zinc and iron, as well as valuable associated metals, namely copper, germanium, gallium, cadmium, silver, lead, etc. From XRF analyses performed on pellets, the chemical composition of UZK residues was determined as shown in Table I. This result concerns especially the so-called ‘old residues’ which were produced before the installation of the flotation unit inside the plant for the recovery of unroasted sulphide materials. These materials were recycled to the roasting plant.

Figure 2. Aerial view of the Kolwezi Plant and residues ponds (August 1962)

Table I Chemical composition of UZK residues (old residues) Elements Assay

Ag, ppm

Ga, ppm

Ge, ppm

Cu, %

Cd, %

Fe, %

Pb, %

Zn, %

S, %

80

1100

490

2.86

0.16

32.79

2.43

16.78

4.05

BASE METALS RECOVERY FROM ZINC HYDROMETALLURGICAL PLANT RESIDUES

19

17-30_Ngenda:text

2/12/09

11:00 AM

Page 20

Because of the presence of heavy metals, these materials are environmentally hazardous. Figure 3(a) shows a crevasse in pond no. 4. Part of the residues has been washed away by rain to the neighbouring Musonoie River, which is dangerously polluted. This is clearly visible by damage to vegetation around the ponds and the river. Furthermore, wind erosion generally occurs in the dry season, as shown in Figure 3(b) where a storm of residues dust invades the plant’s offices. This also happens to inhabited areas and especially to the plant, causing irritation of lung and eye to local people and workers. Table II has been calculated taking into account the average composition of UZK residues (old and new residues) and the market value of contained metal commodities on 4 September 2008. The market value of 1 metric ton of UZK residues is therefore about US$1133.8. From the foregoing it is clear that processing of UZK residues is both economically and environmentally necessary. This research work therefore aims to develop an environmentally friendly process to economically recover zinc and the associated metals. The process is essentially hydrometallurgical in combination with some useful pyrometallurgical and chemical operations which render it more competitive from a different point of view. It has been initially developed for the retreatment of copper smelter slags and is now successfully adapted to reprocessing residues from zinc hydrometallurgical plant.

(a)

(b)

Figure 3. Some views of the residues ponds: (a) Crevasse in pond no. 4, (b) Storm of residues dust to the plant’s offices (August 2008)

Table II Contained metal value of one metric ton of UZK residues (September 2008)

Assay, (average) Quantity in 1 t residues, kg Price, US$/kg Recovery rate, % Value, US$ Value, % Economic classification

20

Zn, %

Cu, %

Cd, ppm

Ge, ppm

Ga, ppm

Ag, ppm

19.5 195 1.73 100 337.35 29.7 2

2.7 27 7.221 100 194.967 17.2 4

1 199.3 1.1993 6.393 100 7.667 0.7 6

475.4 0.4754 1500 50 356.55 31.4 1

447.8 0.4478 450 100 201.51 17.8 3

157.1 0.1571 455.2 50 35.756 3.2 5

Total

1 133.8 100


17-30_Ngenda:text

2/12/09

11:00 AM

Page 21

Experimental Generalities The experimental method used in this work essentially consists of sulphatation of oxide compounds which are present in residues followed by selective thermal decomposition of the sulphates. Sulphatation is performed by digestion of the materials with sulphuric acid. The products are roasted to selectively convert some sulphate compounds into water nonsoluble oxide forms. Thereafter, leaching in water is conducted in order to dissolve metals from the remaining sulphates. Unlike previous studies where sulphatation of different kind of materials was performed in a large temperature range 1–3, 5, 6, in this work sulphatation has been conducted at room temperature in accordance with the conditions used by Banza for the recycling of copper smelter slag 4. While the digestion product from copper slag was dry porous material, digestion of UZK residues lead to a wet and pasty one, which necessitates drying at 100°C 24 h. After roasting of the digested materials, most of base metals could be leached in water, while iron preferentially went as Fe2O3 into leaching residues. In this first phase, the research work aims to develop a process for the recovery of base metals, especially zinc and copper, by a selective method towards iron. Purification of the obtained metal solutions and recovery of associated metal values will be addressed later on. Nevertheless, the behavior of these metals has also been investigated. Equipment and reagents • Roasting—the roasting equipment essentially consists of a vertical resistance electric furnace. It has two half shells in series with a cylindrical alumina crucible in between. A stainless steel rod ended with four arms has helped stirring the material to roast. Roast gas was allowed to escape into a water container prior to going to the atmosphere. The temperatures in the material and inside the furnace were measured by thermocouples and continuously recorded. Temperature control was performed with a PID regulator. • Leaching—leaching experiments were conducted in a vessel that was externally heated by thermostatically controlled water circulating from a water-bath. An alcohol thermometer and an electronic pH-meter have been used to measure both temperature and pH of the leaching pulp. • Analytic—in addition to different measurements that were taken during the experiments, numerous analyses have been performed for Zn, Fe, Cu, Ge, Ga, Cd, Pb, Ag, etc. with the following appropriate equipment: – Crystalline phase identification on solid by X-ray diffraction (XRD) with a SIEMENS D 500 apparatus – Semi quantitative determination of different elements in solid by X-ray fluorescence (XRF) with a SIEMENS SRS 300 Analyzer – Chemical elemental analysis of liquid by optical emission spectrometry—inductively coupled plasma (OES-ICP) using a VISTA MPX Varian type apparatus – Grain morphology determination by JEOL JMS-6100 scanning electron microscope (SEM) – Granulometric analysis by laser diffraction on solid using an instrument of SCIROCCO MASTER SIZER 2000 type. • Reagents—sulphuric, hydrochloric and nitric acids were used in the experiments along with distilled water. All reagents were of analytical quality. Half concentrated sulphuric acid solution has been identified from preliminary tests as the best digestion agent. Therefore 48% H2SO4 solutions were used all over the test series. BASE METALS RECOVERY FROM ZINC HYDROMETALLURGICAL PLANT RESIDUES

21

17-30_Ngenda:text

2/12/09

11:00 AM

Page 22

Experimental procedure Digestion experiments were conducted by adding a given volume of 48% H2SO4 solution to a given quantity of UZK residues, without any agitation, allowing the digestion reaction to take place for some hours. The resulting dough was then dried overnight in an oven prior to grinding. This product was then roasted and thereafter leached in water under agitation. After leaching, the leach pulp was filtered and the solid cake washed until a colourless filtrate was obtained. The filtrate was adjusted to 1 000 ml while the solid was dried overnight and weighted thereafter. Sulphuric acid consumptions are expressed in metric ton (t) of acid per dry metric ton of treated residues. The solids and liquids from different process operations were analysed using the appropriate methods and apparatus. Results Sample characterization Granulometric analysis of the UZK residues has shown that nearly 100% of this material is under 106 μm and about 80% of particles are smaller than 38 μm. A semi quantitative mineralogical analysis by XRD has identified the following crystalline phases that are present in UZK residues. The proportions are expressed in % by weight. • Zinc is present as franklinite ZnFe2O4 (80.4%), willemite Zn2SiO4 (3.9%) and sphalerite (Zn, Fe)S (1.3%) • Lead occurs as beudantite Pb(Fe2,54Al0,46)(As1,07 O4)(S0,9304)(OH)6 at 10.2% • Quartz SiO2 is also present at about 1.6% • Gypsum CaSO4.2H2O is present at about 2.6%. The SEM-EDX studies confirmed the important presence of zinc ferrite together with small quantities of zinc silicate, sphalerite and quartz. The presence of large quantities of precipitated iron hydroxide or iron oxide-hydroxide phases in intimate association with SiO2 is significant. This phase is not detected by XRD analysis; it is assumed to be an amorphous iron-silica gel. An estimate by XRD, with 5% TiO2, shows that the amorphous phase occurs at approximately 42% in UZK residues. Due to their very small quantities, cadmium, germanium, gallium and silver phases could not be identified by XRD analysis, the detection limit of the used analyzer being 1%, and also because they are probably blocked in the hydroxide phase. Sulphatation of zinc ferrites by sulphuric acid digestion Generalities Metallic oxides that are present in UZK residues are essentially ferrites with the general formula MeFe2O4 where Me represents the metallic cation Zn, Cu, Cd, etc. Digestion with sulphuric acid is conducted by allowing the mixture of acid with the material to stand for a long time, without agitation. The acid enters the matrix of ferrite to transform its structure by liberating the metals in the form of soluble sulphates according to the general Reaction [1]. [1] Iron is also converted to the soluble sulphates during this process. Digestion products that have been dried for 24 h at 100°C in oven became hard. Microscopic images of grain morphology for the raw material and the dried and ground digested can be seen in Figure 4. Grain attack by acid is clearly visible in the figures. The white areas are dominated by acid presence. Iron sulphate precipitate7 is also present. 22


17-30_Ngenda:text

2/12/09

11:00 AM

Page 23

(a)

(b)

(c)

Figure 4. Morphology of UZK residues grains before (a) and after digestion (b), (c) a white zone of high sulfuric acid quantity. Digestion conditions: 1 t H2SO4/t residues, S/L = 0.9 [g/ml], 24 h

Figure 5. Solubilization of base metals from UZK residues in function of the digestion time. Test conditions: digestion (1 t H2SO4/t residues, S/L = 0.9 (g/ml)), if dry sample (100°C, 24 h), leaching (water, 2 h, 40°C)

Impact of the digestion time In this series of experiments, digestion time has been varied from 2 to 72 hours, while acid consumption remained constant at 1 t H2SO4/t residues, which is theoretical quantity for complete sulphatation of all the contained metals. The ratio solid/liquid remained at 1/1.1 or 0.9 g/ml unchanged in all test series. XRD analyses performed on materials from different digestion times have confirmed sulphatation of zinc and iron into compounds such as gunningite ZnSO4.H2O, rhomboclase FeH(SO4)2.4H2O and hohmannite Fe(H2(H2O)4((SO4)2O).4H2O. These investigations also revealed that after 2 to 4 h digestion the treated material contained rhomboclase to 30% and gunningite to 40%. These proportions changed to 60% and 25% respectively when the digestion time was increased beyond 4 h. Leaching in water of digestion cake at different times could solubilize zinc, copper and iron to approximately 70% irrespective of the digestion time. Metal solubilization, which is the proportion of the dissolved metal, generally increased with digestion time, as shown in Figure 5. The digested materials were analysed before drying on wet samples and after drying at 100°C for 24 h. From 2 to 4 h a sharp increase in the yield of leaching was observed for all metals, except for lead. This tendency slowed after 4 h. It was observed that metal leaching was less on wet samples compared to the same samples after they have been dried, suggesting that the sulphatation process still continues during sample drying in an oven. Therefore, drying of the digested material renders the subsequent leaching operation more effective. BASE METALS RECOVERY FROM ZINC HYDROMETALLURGICAL PLANT RESIDUES

23

17-30_Ngenda:text

2/12/09

11:00 AM

Page 24

From the above, it has also been observed that zinc and iron sulphatation has not been completed, due to the presence of some quantities of zinc ferrite and sulphide which could not react with the acid. Franklinite ZnFe2O4 and sphalerite (Zn,Fe)S have been identified by XRD analysis in all samples which were obtained as leaching tests residues. Leaching of lead remained relatively very low at about 5%, due to its low solubility limit in sulphuric acid solutions. Thus, anglesite PbSO4 was also present in the new leaching residues. Semi quantitative XRD analysis unambiguously identified franklinite as the main compound of the leaching residues (75%), followed by anglesite (15%), quartz (7%) and sphalerite (4%). Jarosite (K,H 3O)Fe 3(SO 4) 2(OH) 6 has been rarely identified in leaching residues from materials which have been digested for more than 4 h. Impact of sulphuric acid consumption The impact of change in sulphuric acid consumption during the digestion process has been investigated in the range 0.6 to 2.0 t H2SO4 / t UZK residues (Figure 6). Increase of acid consumption generally induced an increase in leaching efficiency. At high acid consumption from 1.7 to 2.0 t /t, a pasty product was obtained, which was in turn difficult to dry. This product was probably silica gel, which covered the solid material particles with the effect of slowing down the sulphatation kinetics. From both metallurgical and economic reasons, a specific consumption of 1 t H2SO4/t UZK residues seemed to be a good compromise for the practice. Sulphatation efficiency was high and the resulting product easy to manipulate. In these conditions, metal leaching was performed to 79.5% Zn, 87.4% Cu and 82.8% Fe. Similar phases as those observed in the first series of experiments have been identified, but the proportion of soluble zinc and iron increased with increasing acid consumption. Investigation into thermal decomposition Thermogravimetric, differential thermal analyses and mass spectrometry Iron conversion from the soluble sulphate compound to the nonsoluble oxide one occurs at high temperature according to Reaction [2]7, 8:

Figure 6. Solubilization of base metals from UZK residues in function of sulphuric acid consumption. Test conditions: Digestion (1 t H2SO4/t residues, S/L = 0.9 (g/ml), if dry sample (100°C, 24 h), leaching (water, 2 h, 40°C)

24


17-30_Ngenda:text

2/12/09

11:00 AM

Page 25

[2] Sulphates of other metals also undergo thermal conversion in a similar way. Thermogravimetric and differential thermal analyses have been conducted in previous works together with mass spectrometry in order to determine the optimum temperature range where selective iron sulphate decomposition is most effective7, 8. The selected range of roast temperature was between 650°C and 850°C. Although different from the results in some other works10–14, this temperature interval corresponds to that given elsewhere9. In fact, UZK residues contain a number of complex sulphates which have been determined by XRD7, 8, while the literature deals with pure sulphates. Roasting experiments The dried and ground digested materials were roasted under smooth agitation in the roasting installation described above. The impact of roasting temperature and time has been investigated in this series of experiments. a) Influence of roasting time—this test series has been conducted at 750°C on UZK residues after digestion (1 t H2SO4/t residues, 24 h), drying in oven (100°C, 24 h) and grinding. The roasting time was varied in the range 1 to 4 h. The influence of the roasting time on metal leaching in water is shown in Figure 7. It has been observed that within 1 h roasting iron dissolution was still important. This was essentially due to insufficient conversion of iron sulphate into oxide compound. After 1 h roasting, iron leaching continuously decreased, as result of increasing formation of the nonsoluble hematite, while copper, cadmium and zinc solubilization increased. XRD analyses have revealed that the proportion of mikasaite (Fe2(SO4)3), which is about 57% of crystalline phase in the digested materials, continuously decrease to 37% after 1 h roasting, 14% after 2 h to reach about 2% after 4 h, in favour of hematite (Fe2O3) increasing formation. According to literature6, there are two main phenomena which occur during the roast process. In the first place, dehydration is happening thus creating micropores in the

Figure 7. Leaching recovery of base metals as function of the roast time. Operating conditions: digestion (1 t H2SO4/t UZK residues, S/L = 0.9 (g/ml), 24 h), drying (100°C, 24 h), grinding, roasting (750°C), leaching (water, 40°C, 2 h)

BASE METALS RECOVERY FROM ZINC HYDROMETALLURGICAL PLANT RESIDUES

25

17-30_Ngenda:text

2/12/09

11:00 AM

Page 26

material. The other is a sulphatation by SO3 which could come into contact with the rest of the minerals through the micropores created by dehydration. This phenomenon is observed at a temperature of about 600°C. b) Influence of roasting temperature—this test series has been conducted for 2 h in similar way to the previous one by varying the roast temperature from 650 to 850°C. The roasted materials have then been used in leaching tests. The influence of the roasting temperature on leaching efficiency is shown in Figure 8. Leaching recovery of copper, zinc and cadmium remained constant until 770°C for copper and zinc and until 800°C for cadmium. This result clearly indicated that copper and zinc were stable until the temperature of 770°C was reached. Thereafter, their sulphates are progressively converted nonsoluble oxide compounds. On the other hand, leaching recovery of iron sharply decreased from 79% at 650°C to reach 34% at 700°C, 6% at 770°C and became zero at 800°C, which clearly indicated that iron sulphates were completely converted. The best conditions for selective iron conversion were reached between 750°C and 770°C. Beyond this range, some stable iron compounds of ferrite type were formed with zinc and copper, as confirmed by XRD analysis. Summary and conclusions The method used to investigate the treatment of residues from the Zinc Hydro Plant of Kolwezi essentially consists of digestion of the materials with a 48% H2SO4 solution without agitation or heating. The test material has been used as received (100% de – 106 μm). After digestion, drying and grinding of the obtained compact product, roasting was performed prior to leaching with water. Zinc, copper and cadmium were leached into solution, while iron preferentially remained in the leach residues. The process is summarized in the simplified flow sheet shown in Figure 9. It is clear that the SO2 roast gas can be used for the manufacture of sulphuric acid. Digestion of UZK residues gave a wet product, unlike the result obtained in previous works on copper smelter slags from Lubumbashi4, where dry porous products have been obtained. In

Figure 8. Leaching recovery of base metals as a function of the roast temperature. Operating conditions: digestion (1 t H2SO4/t UZK residues, S/L = 0.9 (g/ml), 24 h), drying (100°C, 24 h), grinding, roasting (750°C), leaching (water, 40°C, 2 h)

26


17-30_Ngenda:text

2/12/09

11:00 AM

Page 27

Figure 9. Proposed flowsheet for the retreatment of UZK residues by digestion method

the present study, it was therefore necessary to dry and grind the material after digestion. Furthermore, an increase in sulphatation occurs during the drying process. Thus, sulphatation is a thermally activated reaction1–3, 5–6. Total sulphatation by digestion with sulphuric acid was difficult to achieve in the prevailing test conditions. Fortunately, it was discovered that sulphatation continues under the direct action of SO3 during the drying operation. Complete sulphatation is therefore possible. Fluidized bed roasting can hardly be conducted with such fine materials like those obtained after digestion which have been previously subject to drying and grinding. Roasting in a fixed bed with smooth agitation of the material might be the best industrial practice. In this case, using a Wedge type roaster along with careful handling of fine materials would be recommended. Following the above described outline, leaching in water could be performed to 98.7% Zn, 99.9% Cu, almost 100% Cd and only 6.4% Fe under the best test conditions as given in the flowsheet. Iron preferentially remained as hematite (Fe2O3) in leaching residues, which assayed in average 0.4% Zn, 17 ppm Cu, 0% Cd, 3.8% Pb and 49.3% Fe. These materials may be considered as Ge (800 ppm) and Ga (1 660 ppm) concentrates containing most of the silver. Subsequent studies will enable the development of a process for the recovery of these valuable metals. The leach solutions on average contained 15 g/l Zn, 2.5 g/l Cu, 1.7 g/l Fe and 134 mg/l Cd. After solutions purification and concentration by appropriate methods, the contained valuable metals could be recovered by salt precipitation or metal electrowinning. BASE METALS RECOVERY FROM ZINC HYDROMETALLURGICAL PLANT RESIDUES

27

17-30_Ngenda:text

2/12/09

11:00 AM

Page 28

From the above, it can be concluded that the described ‘digestion method’ has a significant advantage, since elaborate iron precipitation from solution was no longer necessary, this metal being selectively discarded into residues before leaching. Compared to the ‘hot leaching method’, this technique has the additional advantage of the opportunity to recycle the SO2 roast gas for sulphuric acid production. Recycling of the generated acid would substantially reduce fresh acid consumption, thereby improving the process viability. Acknowledgements The authors appreciate the support from the Gecamines Research Centre in Likasi, DRC. The fruitful cooperation between the Free University of Brussels, Belgium, and The University of Lubumbashi, DRC, is gratefully acknowledged. References 1. SULKA, L.B., PANDA, S.C., and JENA, P.K. Recovery of cobalt, nickel and copper from converter slag trough roasting with ammonium sulphate and sulphuric acid, Hydrometallurgy, vol. 16, no. 2, June 1986, pp. 153–165. 2. FREEMAN, G.M. and NIGHTINGALE, D.E. Treatment of zinc plant leach residues for recovery of contained metals values, European Patent Application, 25 09 1979. 3. STEVEN, J.A. Extraction method for non ferrous metals, United States Patent, May 25, 1976. 4. BANZA, A.N. Processes for the recycling of copper smelter slag from Lubumbashi/D.R. Congo, (Doctorale Thesis in German), Cuvillier Verlag Goettingen, Goettingen 2001 (ISBN 3 – 89873 – 151 – 0) 5. CUNEY, A. and FATMA, A. Recovery copper, cobalt and zinc from copper smelter and converter slags, Hydrometallurgy, vol. 67, 2002, pp. 1–7. 6. TURAN, M.D., ALTUNDOGAN, H.S., and TUMEN, F. Recovery of zinc and lead from zinc plant residue, Hydrometallurgy, vol. 75, 2004, pp. 169–176. 7. NGENDA, R.B. Etude de mise en solution sélective du zinc et du cuivre contenus dans les rejets des Usines à Zinc de Kolwezi par la technique de digestion acide, Mémoire de spécialisation, Services Matières et Matériaux, Facultés des Sciences Appliquées, Université Libre de Bruxelles, Janvier 2006. 8. NGENDA, R.B., SEGERS, L., and P. KONGOLO K, Décomposition thermique sélective des sulfates formés lors de la digestion acide des rejets des Usines à Zinc de Kolwezi, Annales du pôle Mines-Géologie-Université de Lubumbahi 1, 2007. pp. 97–104. 9. SIRIWARDANE, R.V., POSTON JR., J.A., FISHER, E.P., SHEN, M.S., and MILTZ, A.L. Decomposition of sulphates of copper, iron (II), iron (III), nickel and zinc: XPS, SEM, DRIFTS, XRD and TGA study, Applied Surface Science, vol. 159, 1999, pp. 219–236. 10. KOLTA, G.A. and ASKAR, M.H. Thermal decomposition of some sulphate, Thermochimica Acta, vol. 11, 1975, pp. 65–72. 28


17-30_Ngenda:text

2/12/09

11:00 AM

Page 29

11. HILDENBRAND, D.L., LAU, K.H., and BRITTAN L.D. Mechanistic aspects of metal sulphate decomposition process, High Temperature Science, vol. 26, 1990, pp. 427–440. 12. NARAYAN, R.N., TABATABAIE-RAISSI, A., and ANTAL, M.J. A study of zinc sulphate decomposition at low heating rates, Ind. Eng. Chem. Res., vol. 27, 1988, pp. 1050–1058. 13. MU, JJ. and PERIMUTTER, D.D. Thermal decomposition of inorganic sulphates and their hydrates, Ind. Eng. Chem. Process, Des. 20, 1981, pp. 640–646. 14. TAGAWA, H.H. and SAIJO, H. Kinetics of thermal decomposition of some transition metals sulphates, Thermochimica Acta, vol. 91, 1985, pp. 65–67. Kitala Pierre Kongolo University of Lubumbashi, Polytechnic Faculty, Lubumbashi, Democratic Republic of Congo (DRC) Doctoral Research, 1983—1989 • Technical University of Clausthal, Germany, Institute for Mineral Processing, Department Flotation and Chemical Processes, Research in Hydrometallurgy. • Research cooperation with the Technical University of Munich, Physic Department (Laboratory of Mossbauer Spectroscopy). • 1989—Qualification as ‘Doctor of Engineering’ (Hydrometallurgy) at the Technical University of Clausthal, Germany. Mining industry, since 1990 Gecamines (1990–2006): • Research Centre in Likasi, DRC, Senior Research Engineer (1990–2000, Mineral Processing), Research Director (2001–2006, Hydrometallurgy). • Plant manager of Kambove and Kakanda Concentrators (2000–2001). Chemaf: Director of Operations (2006–2007). Groupe Bazano: Consulting Engineer (since April 2008). Academic positions, since 1993 University of Lubumbashi, Polytechnic Faculty, Department of Metallurgy • Associate Professor (1995–2001) • Professor (2001–2007) • Ordinary Professor (2007–to date) • Deputy Faculty Dean for Research (since 1997). International activities, since 1995 • Cooperative research with the Technical University of Clausthal, Germany (since 1995), the Free University of Brussels, Belgium (since 2003). • External examiner at the University of Dar Es Salaam, Tanzania (2004–2007). • Invited lecturer at International Symposium on the Industrial Application of the Mossbauer Effect, ISIAME ’96, November 1996, Johannesburg, South Africa, ‘The extractive metallurgy of gold’. • Chairman of session on ‘solvent extraction/electrowinning’ at XX. International Mineral Processing Congress, September 1997, Aachen, Germany. • Reviewer (2006), journal Hydrometallurgy, Elsevier Publisher. Publications Numerous publications in international journals (Metallurgical Transactions, Hyperfine Interactions, Hydrometallurgy, Minerals Engineering, etc.) in the fields of mineral processing, hydrometallurgy, Mossbauer spectroscopy and surface chemistry. BASE METALS RECOVERY FROM ZINC HYDROMETALLURGICAL PLANT RESIDUES

29

17-30_Ngenda:text

30

2/12/09

11:00 AM

Page 30


31-42_Robertson:text

2/12/09

10:56 AM

Page 31

ROBERTSON, S.W. and VAN STADEN, P.J. The progression of metallurgical testwork during heap leach design. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

The progression of metallurgical testwork during heap leach design S.W. ROBERTSON and P.J. VAN STADEN Mintek, Randburg, South Africa

The depletion of favourably located and higher-grade mineral deposits, combined with more favourable metal prices and higher capital costs, has stimulated research into heap leaching of lower grade ores and wastes. This has resulted in renewed interest in heap leaching of uranium ores, and in the development of processes for the bacterial heap leaching of low-grade chalcopyrite ores and the acid heap leaching of nickel laterites. A complete programme of heap leach testwork typically involves a sequence of bottle roll tests and column tests, followed by piloting on test heaps. The initial testwork is aimed at determining the amenability of the ore to heap leaching, which depends on the characteristic of the ore with respect to porosity and permeability to leach liquor, acid consumption, metal recovery and percolation. Subsequent phases of the testwork focus on the development of the process design criteria for the treatment of the ore. Basic techno-economic models have also been developed as a tool for the initial estimation of heap leach viability. Mintek has also developed in-heap monitoring probes, coupled to an operator guidance system, in order to assist the operator in achieving optimal conditions for bacterial growth and heat preservation during pilotplant operation.

Introduction Heap leaching is used extensively for the processing of copper and gold ores. Low capital and operating costs and simple atmospheric leach processes make heap leaching suitable for lowgrade ores and small deposits. Disadvantages include low recoveries, long ramp-up times, large footprint and acid-mine drainage of wastes3. Factors that may make ores unsuitable for heap leaching are poor percolation due to the presence of swelling clays, and high gangue acid consumptions. Percolation may be improved by agglomerating with acid or binder, and by minimizing ore compaction during stacking, for example by using conveyors instead of trucks. For ores that contain silicate gangue minerals, it may be possible to reduce gangue acid consumption by introducing the lixiviant more slowly, for example, by using higher irrigation rates at lower lixiviant acid strengths. However, it is usually not possible to achieve an THE PROGRESSION OF METALLURGICAL TESTWORK DURING HEAP LEACH DESIGN

31


2/12/09

10:56 AM

Page 32

economically viable optimum between metal extraction and acid consumption if the ore contains highly reactive gangue such as carbonates. For high acid-consuming uranium ores, it is more economical to use sodium carbonate as lixiviant. Nickel sulphide and laterite ores usually have high acid consumptions compared with copper ores, but this is offset by the higher nickel value. The cost of acid typically comprises about 40% of the total operating cost for nickel laterite ores, so acid needs to be produced on site and heaps need to be operated counter-currently to optimize acid utilization. Bacterially assisted heap leaching has been applied extensively for secondary copper sulphide ores, where the bacteria assist in the oxidation of ferrous iron and sulphur species. If the ore contains pyrite, bacterial oxidation can be utilized to generate acid and exothermic heat, which may be preserved within the heap through the appropriate manipulation of aeration and irrigation rates9. Higher temperatures will improve kinetics for more refractory species such as chalcopyrite, which do not give economic leach rates at ambient temperatures. This paper discusses the metallurgical testwork sequence for heap leach design, with specific reference to chalcopyrite, uranium and nickel laterite heap leaching. Progression of metallurgical testwork Figure 1 shows the typical progression of metallurgical heap leach testwork through subsequent phases of roll bottles, column tests and pilot heaps. Since the duration and cost of testwork increases progressively, it is beneficial to optimize as many parameters as possible earlier on in the testwork programme. Table I gives a summary of outputs for various testwork stages. Roll bottle or shake-flask tests are normally performed first, to obtain an initial indication of maximum achievable extraction and acid consumptions. However, acid consumptions are normally overestimated in rolling bottles, hence the acid consumption results must be treated as semi-quantitative. The 1 metre columns provide more accurate extraction and acid consumption data under trickle bed

Figure 1. Staged approach to heap leach testwork and design

32



2/12/09

10:56 AM

Page 33

Table I Metallurgical testwork programme Stage

Outputs

Approximate duration days

Roll bottle, shake-flask or stirred tank

• benchmark recoveries and reagent consumptions • effect of mineralogy (laterites) • effect of crush size, lixiviant strength, redox and temperature

• • • •

Cu oxides — 7 D Cu sec. sulphides – 14 Cu prim. sulphides – 30 Ni laterites – 14

1 m columns

• identify severe percolation problems • maximum extraction and reagent consumptions under percolation-contact mode • optimal agglomerate moisture content • isothermal or temperature adjustment with heat model • aeration, inoculation and bacterial activity

• • • •

Cu oxides – 50 d Cu sec. sulphides – 80 Cu prim. sulphides – 150 Ni laterites – 80–150

6 m columns

• extraction kinetics, reagent consumptions under typical lift height of heap • impurity build-up in recycled solutions • more reliable simulation of compaction and percolation • neutralizing potential of ore for counter-current operation (laterites) • effect of bacterial activity on temperature behaviour predicted by heat model

• • • •

Cu oxides – 150–200 Cu sec. sulphides – 250-300 Cu prim. sulphides – 300–365 Ni laterites – 250–500

Test heaps

• best judgement on heap permeability in the absence of column side support • dynamic temperature variation as a result of bacterial activity demonstrated • net reagent consumption and impurity build-up in closed circuit with recovery step • counter-current operation to maximize copper recovery and minimize gangue reaction

• • • •

Cu oxides – 150-200 Cu sec. sulphides – 250–300 Cu prim. sulphides – 300–365 Ni laterites – 250–500

conditions, as well as an initial indication of possible severe percolation problems. But for a given irrigation rate (in terms of l/h/m2), the extraction kinetics from the column of ore (in terms of % extraction per day) are accelerated compared to the extraction kinetics that will be obtained from a taller heap of ore. Since the kinetics and pregnant leach solution composition vary with lift height in commercial size heaps, a more realistic indication of leach kinetics and impurity build-up is obtained in taller columns, at a lift height equivalent of an actual heap (typically 6 metres). Although the taller columns provide an indication of slumping and permeability, the side support from the column may result in less compaction than will be observed on an actual pilot or commercial heap. Therefore, whereas solution build-up in columns provide a definite indication that percolation problems will be experienced at larger scale, trouble-free percolation in columns unfortunately does not guarantee successful percolation for the given irrigation rate on the commercial heaps. As shown in Table I, the leach cycle increases from 150–200 days for oxides to 250–300 days for secondary sulphides and 300–365 days for primary copper sulphides (chalcopyrite). In order to demonstrate the effect of leach cycle on the pad area and mining rate, a production calculation is shown in Table II for a heap leach plant producing 20000 tonnes per annum copper, with a head grade of 0.6% copper. The chalcopyrite material has a longer leach cycle and lower recovery, resulting in bigger pad area and ore processing rate. This will result in larger capital costs for the mining, crushing and leaching sections. THE PROGRESSION OF METALLURGICAL TESTWORK DURING HEAP LEACH DESIGN

33


2/12/09

10:56 AM

Page 34

Table II Effect of leach cycle on pad area and mining rate

Cu production Cu grade Leach cycle Cu recovery Mining/stacking rate Ore mass under leach Stacked bulk density Ore volume under leach Lift height Surface area under leach

Oxide

Sec. sulphide

Prim. sulphide

Units

20 000 0.60 200 80.0 4 166 667 2 283 105 1.8 1 268 392 6.0 211 399

20 000 0.60 280 80.0 4 166 667 3 196 347 1.8 1 775 748 6.0 295 958

20 000 0.60 365 60.0 5 555 556 5 555 556 1.8 3 086 420 6.0 514 403

tpa Cu % Cu days % Cu leached tpa ore tonnes t/m3 m3 m m2

Table III Operating stages during heap leach cycle Stage

Conditions

Purpose

High irrigation rate, moderate acid strength (ILS), GL: GA ~ 10

Create suitable pH gradient for bacterial activity

Early leach

Moderate irrigation rate, moderate acid strength (ILS), GL: GA ~ 5

Maximize copper recovery, limit gangue reaction, promote bacterial activity, preserve heat

Late leach

Low irrigation rate, strong acid (raffinate)

Maintain the copper tenor while the leach rate is slowing

High irrigation rate, low copper tenor (raffinate)

Recover entrained copper prior to heap closure

pH reduction

Rinse

Low grade chalcopyrite heap bioleaching Chalcopyrite is the most abundant copper mineral, comprising about 70% of the world’s known copper reserves. Many low-grade reserves are untreatable with conventional heap leaching due to the slow heap leach kinetics at ambient temperatures. Since the leach kinetics improve with temperature, it is possible to improve kinetics by generating exothermic heat through the bacterial-assisted conversion of sulphide species (most importantly pyrite), and to preserve the heat within the heap through the appropriate manipulation of aeration and irrigation rates. The following approach is followed during metallurgical testwork for chalcopyrite ores: • Column tests are performed in conjunction with a heat balance, which simulates the axial temperature profile in the core of a heap in the absence of side effects. Daily measured oxygen consumption data are input into the heat balance, and the temperatures in each jacketed column section is adjusted to simulate dynamic heap temperature behaviour. Columns are inoculated and aerated to promote bacterial activity • The process is demonstrated on test heaps, typically 25000 tonnes, 40 m x 60 m x 6 m. Crushed ore is mixed with acid and inoculum in an agglomeration drum. The base of the pile is fitted with aeration pipes connected to a low pressure blower. The pilot heap is operated in closed circuit with a solvent extraction/electrowinning plant • The heap life cycle is divided into distinct stages, in order to meet the combined objectives of maximizing copper recovery, minimizing gangue reaction and creating conditions for bacterial growth and heat preservation. Each stage has its own set of criteria for the control of the irrigation rate (GL, kg/m2/h), aeration rate (GA, kg/m2/h) and lixiviant acid strength as described in the Table III. ‘New’ heaps are irrigated with intermediate leach solution (ILS) whereas ‘old’ heaps are irrigated with return raffinate from the solvent extraction plant. 34



2/12/09

10:56 AM

Page 35

Mintek has developed in-heap probes (lysimeters), which allow solution and gas-phase samples to be taken for daily analysis at various depths in the heap. The lysimeters are placed in augured holes, and are also fitted with thermocouples. An example of the development of pH, redox and temperature profiles in the core of the heap as measured by the in-heap probes is shown in Figure 2. The reduction in pH gradient in the heap creates suitable conditions for bacterial growth. The increase in bacterial activity is evidenced by a corresponding increase in redox potential and temperature. Temperatures of about 40°C were maintained in the bottom half of the heap. The lysimeter measurements assist the site engineer in making decisions concerning changes in irrigation and aeration rates, and changeover between subsequent stages in the leach cycle. Mintek has developed software for logging the daily measurements and assay results for multiple heaps. Each day, the software produces a report of the valve and flow adjustments required on the plant, as well as a log sheet for the measurements and assays required for the day. This operates in conjunction with a particluar irrigation manifold and valve arrangement, which simplifies the task of visualizing the daily irrigation valve adjustments required, which is illustrated on the daily report. The heap software can also be applied to oxide and secondary sulphide ores. Uranium ore The rise in uranium prices has resulted in renewed exploration, expansion and construction of new projects such as Langer Heinrich (Namibia), Keyelekera (Malawi), Ezulwini (South Africa), and Buffelsfontein (South Africa). There is also renewed focus on exploitation of previously uneconomical ores and wastes such as Rossing (heap leaching of low grade ore) and Buffelsfontein (treatment of pyrite-gold tailings). Expansion through heap leaching is being investigated at Ranger (Australia) and construction of a heap leach plant at Arlit (Niger)4, amongst others.

Figure 2. Progression of Eh, pH and temperature in bioheap core

THE PROGRESSION OF METALLURGICAL TESTWORK DURING HEAP LEACH DESIGN

35


2/12/09

10:56 AM

Page 36

Figure 3. Heap leach control software

Table IV Common uranium minerals

Leachable oxides

Mineral

Formula

Operation

UraniniteTL PitchblendeTL

+4 +6 U 1-xU xO2+x

Rossing, Dominion Reefs, Ezulwini Narbalek, Kintyre

UO2 to UO2.25

Leachable silicates

CoffiniteTL

U(SiO4)1-x(OH)4x

Rystkuil

Refractory complex oxides

BranneriteTr DaviditeTr

(U,Ca,Fe,Th,Y)(Ti,Fe)2O6 3+ (La, Ce, Ca)(Y, U)(Ti, Fe )20O38

Elliot Lake Radium Hill

Hydrated oxides

BecquereliteHL GummiteHL

7UO2.11H2O UO3.nH2O

Silicates

UranophaneHL UranothoriteHL SklodowskiteHL

Ca(UO2)2Si2O7.6H2O (UTh)SiO4 (H3O2)Mg(UO2)2(SiO4)22H2O

Rossing Dominion Reefs

Vanadates

CarnotiteHL TyuyamuniteHL

K2(UO2)2(VO4)2.3H2O Ca(UO2)2(VO4)2.8H2O

Langer Heinrich

Phosphates

TorberniteHL AutuniteHL

Cu(UO2)2(PO4)2.10H2O Ca(UO2)2(PO4)2.11H2O

Rum Jungle Rum Jungle

Carbonates

SchroekingeriteHL

NaCa3(UO)2(CO3)3(SO4)F.10H2O

Arsenates

ZeunariteHL

Cu(UO2)2(AsO4)2.10-12H2O

Hydrocarbons

ThucholiteTL

HL—hexavalent readily acid leachable without oxidation TL—tetravalent readily acid leachable with oxidation TR—tetravalent refractory

Uranium occurs in primary (tetravalent) and secondary (hexavalent) forms. Primary minerals require oxidation during hydrometallurgical extraction. The Table IV provides a summary of common uranium minerals. Uranium ores are usually processed by atmospheric tank leaching, pressure leaching, in situ leaching or heap leaching. Pressure leaching and bacterial heap leaching are applied where oxidation of pyrite is required, such as the 36



2/12/09

10:56 AM

Page 37

Buffelsfontein pyritic gold tailings project or Urgeirica heap leach (Portugal). Whereas tank leaching requires crushing and milling steps, heap leaching is performed on crushed or run of mine ore. The most common leaching medium is sulphuric acid with ferric sulphate as oxidant (e.g. Rossing). Ferric sulphate is regenerated by addition of an oxidizing agent such as MnO2, H2O2, or NaClO3. Where the ore has high gangue acid consumption (>90 kg/t), leaching is performed in carbonate medium (e.g. Langer Heinrich tank leach, Trekkopje heap leach). Uranium recovery from the leach solution is usually performed by solvent extraction, ion exhange, a combination of solvent extraction and ion exhange (Eluex/Bufflex), or direct precipitation. Uranium is precipitated from the eluate or strip liquor and calcined to produce U3O8. Figure 4 shows the normal sequence for hydrometallurgical processing of uranium ore. Most uranium heap leach plants have been satellites to existing operations, for example for treatment of low-grade waste with acidified ferric iron bleed streams (e.g. Rossing, Narbalek), and have been applied to ores with grades of below 0.1% U3O83. However, current high uranium prices and plant capital costs makes uranium heap leaching attractive as a primary (greenfield) process (e.g. Trekkoppje, Arlit, Chirundu), provided reagent costs can be kept low. For a standalone operation, ferric sulphate may be added as reagent (typically 0. g/L Fe, ORP 475–425 mV), or sufficient iron may be leached from the gangue minerals or by leaching pyrite acid plant calcine. A suitable oxidizing agent such as hydrogen peroxide, sodium chlorate, pyrolusite, or caro’s acid is added to regenerate the ferric iron. Bacterial heap leaching requires minimal additional capital cost, namely the installation of aeration piping and low pressure blowers. If the ore already contains pyrite, it may be possible to reduce the overall acid consumption by bacterial oxidation of the pyrite, which also results in exothermic temperature increase in the heap which may improve kinetics. At the same time ferric iron is regenerated by bacterial oxidation in the pile, resulting in reduced cost of oxidizing agent. The bacterial action throughout the heap maintains the iron in the ferric form, and since ferric iron precipitates above pH 2 (effectively yielding acid during precipitation), the pH throughout the heap is more rapidly buffered at a lower pH value when there is

Figure 4. Uranium processing sequence

THE PROGRESSION OF METALLURGICAL TESTWORK DURING HEAP LEACH DESIGN

37


2/12/09

10:56 AM

Page 38

bacterial activity in the heap. Minerals indicated in Table IV as ‘readily leachable’ are acid leached at pH 1.5–2.0 and 35–60°C, which are suitable conditions for bioleaching. Minerals indicated in Table IV as ‘refractory’ require higher temperature (60–80°C) and stronger acid (up to 50g/L). Figure 5 illustrates the effect of bacterial leaching versus chemical leaching for a uraninite/uranophane ore. Tests were performed in 1 metre columns, irrigated with 0.5 g/L Fe at pH 1.6. For the chemical leach column, the redox potential in the feed was maintained at 470 mV with hydrogen peroxide. For the bacterial leach, 2% pyrite was mixed into the ore, and the column was aerated and inoculated. The redox potential of the feed solution was initially adjusted to 450 mV, whereafter the redox in the recirculating solution was allowed to increase. Whereas the leach kinetics and uranium recoveries were similar, the acid consumption (measured as the acid in the feed solution minus the acid in the drainage) was about 35% lower for the bacterial column. Nickel laterite ores Laterites are usually processed with ammonia or high pressure acid leaching (suitable for limonites with high Fe, low Mg, low Si), or with smelting for ferronickel production (suitable for saprolites with high Mg, high Si). Whereas laterites were previously thought to be unsuitable for heap leaching due to the high clay content and acid consumption, the demonstration of a nickel laterite heap leach pilot plant in Caldag, Turkey, has resulted in a number of subsequent projects, e.g. Jump-Up dam, Canegrass and Murrin-Murrin. However, to date only one brownfield project is operational at Murrin Murrin. Laterites normally have very high acid consumptions (typically 500 kg/t), so a low-cost acid source or on-site acid plant is required. The leach solution also contains high levels of impurities such as iron that need to be precipitated prior to nickel recovery. The limonite fraction normally has poorer leach kinetics, so heap leaching may be suitable as an ‘add-on’ to an existing PAL process, for example heap leaching of a saprolite fraction to neutralize the pressure acid leach liquor. Figure 6 shows the linear behaviour that was observed between nickel recovery and acid consumption in column tests performed on nickel laterite ore. This means that faster nickel recoveries can be achieved by increasing the rate at which acid is added, which may be achieved by speeding up the irrigation rate, or by increasing the acid strength in the lixiviant, or by admixing a large acid dosage to the ore before stacking of the heaps.

Figure 5. Bacterial versus chemical leaching of uranium ore

38



2/12/09

10:56 AM

Page 39

Figure 6. Acid consumption vs. nickel recovery

Figure 7. Proposed heap leach plant schematic

Figure 7 shows a proposed nickel laterite heap leach arrangement. The plant is divided into a series of ‘old heaps’ and ‘fresh heaps’. In order to maximize leach kinetics, a high acid strength (75–120 g/L) is used over the ‘old heaps’. New heaps coming into production are irrigated with intermediate leach solution draining from the ‘old heaps’ and the acid is neutralized over the fresh heaps. Any residual acid remaining in the drainage from the ‘fresh heaps’ has to be neutralized prior to nickel hydroxide precipitation, resulting in increased neutralizing agent and acid costs. An important part of column testwork is therefore to determine the neutralizing potential of the fresh heaps, so that the acid strength can be maximized, while at the same time ensuring complete neutralization of residual acid over the fresh heaps. Figure 8 shows the experimental determination of the acid neutralizing potential of the laterite material in 6 metre columns. In order to simulate the arrangement in Figure 7, the column testwork was divided into ‘fresh heap’ and ‘old heap’ stages. Initially the heap was irrigated at 50 g/L acid, to simulate neutralization of intermediate leach solution (ILS) over ‘fresh heaps’. As indicated, this acid concentration was completely neutralized over the fresh ore. After 80 days, the feed acid concentration was increased to 75g/L, and later to 100 g/L and 150 g/L, to optimize the acid concentration for the ‘old heaps’. The aim was to produce drainage with an average acid content of 50 g/L over the entire ‘old-heap’ part of the leach cycle of the ore. THE PROGRESSION OF METALLURGICAL TESTWORK DURING HEAP LEACH DESIGN

39


2/12/09

10:56 AM

Page 40

Figure 8. Determination of neutralizing potential of ore in 6 metre columns

From the date in Figure 8, it appears that an average of 50 g/L acid in the drainage of ‘old’ nontronite and saprolite ores is obtained by irrigating it with 100 g/L acid solution, and to obtain the same on composite material (consisting of limonite, nontronite and saprolites) irrigant acid strength of 100 to 150 g/L is required. These figures provided the first iterations of the liquor and acid balances over the process. In addition to chemical acid leaching, there are also two potential routes for bioleaching of laterites: • In order to save on the capital cost of an acid plant, elemental sulphur may be mixed in with the ore during agglomeration, and sulphuric acid may be generated bacterially within the heaps. Evidence from past heap leaching testwork has shown that bacteria can generate and tolerate up to around 70 g/Ll acid produced from commercial-grade peletized sulphur added to the heap. This approach has, however, not been demonstrated or optimized in large-scale piloting or commercial-scale applications. There may be a capital cost saving from not building an acid plant; on the other hand, the cost of sulphur could be the same or even higher if the sulphur utilization is poor. • Fungi have been used to generate organic acids (citric, oxalic) from a carbon source such as molasses, and nickel recovery from laterites of up to 60% has been reported7. There are, however, serious disadvantages associated with the use of fungal organisms on industrial scale. The cost of the organic carbon used as energy source makes their application very expensive and unless it can be replaced by inexpensive waste organic products, it would not be a viable process. Selective growth of the fungi would also be impossible under commercial conditions due to contamination with other undesirable microorganisms. Summary and conclusions Longer leach cycles and lower recoveries associated with chalcopyrite heap leaching will typically result in larger pad capital costs and larger ore processing rates for a given copper recovery. Chalcopyrite heap leaching may therefore be more suitable to brownfield applications, for example where existing solvent extraction and electrowinning infrastructure on brownfield installations is used. It is also necessary that the ore contains sufficient pyrite, in order to generate exothermic heat through bacterial oxidation. 40



2/12/09

10:56 AM

Page 41

The application of bio-heap leaching to copper ores provides flexibility to treat a blend of secondary copper sulphides and oxides with minimal additional capital cost, and an existing operation could also be adapted to process primary copper sulphides once the oxides and secondary sulphides become depleted. However, not all oxide/secondary sulphide copper ores are suitable to heap leaching, especially if the ore contains highly reactive gangue minerals. Most uranium heap leach plants have been satellites to existing operations, for example for treatment of low-grade waste with ferric iron bleed streams. However, current high uranium prices and plant capital costs has resulted in standalone uranium heap leaching processes being developed (e.g. Trekkopje, Arlit), provided reagent costs can be kept low. Bioleaching offers a possible route with little additional capital expenditure for reducing oxidizing agent costs and also acid addition if the ore contains pyrite. Due to the high costs of sulphuric acid and poor leaching characteristics of the limonitic fraction of the laterite mineralogy, laterite heap leaching may also find application as an addon process, for example neutralizing acid liquor from a pressure leach plant over saprolite ore. Since the capital cost of the leaching section may be a small percentage of the overall capital cost of a greenfield project with large infrastructure requirements, the overall savings on capital cost may not be significant. As a result, laterite heap leaching will be more suitable to small projects with low infrastructure requirements, producing an intermediate product (e.g. Caldag), or where existing infrastructure is used, such as Murrin Murrin. References 1. LONGWORTH, M. Jump-up dam nickel laterite heap leach project, ALTA Nickel/Cobalt Conference, Perth, 2007. 2. TAN, I. Canegrass Nickel-Cobalt Heap Leaching project, ALTA Nickel/Cobalt Conference, Perth, 2008. 3. TAYLOR, A. Heap leaching and its application to copper, uranium and nickel ores, short course, ALTA, Perth, 2008. 4. TAYLOR, A. Review of Uranium industry developments in 2007–2008, ALTA Uranium Conference, Perth, 2008. 5. PEACY, J., XIAN-JIAN, G., and ROBLES, E. Copper hydrometallurgy: current status, preliminary economics, future direction and positioning versus smelting, Trans. Nonferrous Met. Soc. China, vol. 14, no. 3, June 2004, pp. 560–568. 6. VAN STADEN, P.J. May 2007. Progress at Mintek in Heap Bioleaching. Presented at IV International Copper Hydrometallurgy Workshop, 16–18 May 2007. Department of Mining Engineering, University of Chile. 7. ALIBHAI, K.A.K., DUDENEY, A.W.L., LEAK, D.J., AGATZINI, S., and TZEFENIS, P. Bioleaching and bioprecipitation of nickel and iron from laterites, FEMS microbiology reviews 11, 1993. pp. 87–96. 8. VALIX, et al., Fungal bioleaching of low-grade laterite ores, Minerals engineering, vol. 14, no. 2, 2001. 9. VAN STADEN, P.J. Heap leach research at Mintek, ALTA Copper Conference, Perth, 2008 THE PROGRESSION OF METALLURGICAL TESTWORK DURING HEAP LEACH DESIGN

41


2/12/09

10:56 AM

Page 42

Stefan Walters Robertson Chemical Engineering, Mintek, Randburg, South Africa Stefan obtained a research-based MSc at UCT in chemical engineering in catalysis (1997). He joined Mintek’s hydrometallurgy division (1997) and biotechnology divisionsion (2003), working in the fields of hydrometallurgical process development, pilot plants, development of engineering and costing models for tank and heap leach processes, design of heap leach pilot plants and prefeasibility studies. Publications include: (1) ‘A bacterial Heap Leaching Approach for Primary Copper Sulphide Ore’, presented at the 3rd Southern African Conference on Base Metals, hosted by the South African Institute of Mining and Metallurgy, Kitwe, Zambia, 26–29 June 2005, (2) ‘Heap bioleaching of a low-grade chalcopyrite ore from the Darehzare deposit’, presented at ALTA Copper Conference, 2007, Perth, Australia

42


43-56_Dunn:text

2/12/09

10:58 AM

Page 43

DUNN, G.M. Increasing the capacity of existing and new exothermic autoclave circuits. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Increasing the capacity of existing and new exothermic autoclave circuits G.M. DUNN Hydromet (Pty) Ltd, South Africa

Pressure leach autoclave circuits are employed in the leaching of ores, concentrates, mattes, alloys and intermediates for the recovery of metals into solution. Once the metals are extracted into solution, the value metals can be recovered by hydrometallurgical means such as by purification followed by electrowinning, hydrogen reduction, pyrohydrolysis, crystallization and other unit operations. In many of these integrated flowsheets the pressure leach step is pivotal to the recovery of the value metals from the host material. Furthermore, the autoclave circuit invariably is a high capital cost component of the plant and an area that is carefully scrutinized when debottlenecking and capacity increases are being considered. This paper identifies a unique proven way of increasing the capacity of existing or new exothermic pressure oxidative leach autoclave circuits by as much as two or three times. Introduction Pressure leach autoclave circuits are employed in the leaching of ores, concentrates, mattes, alloys and intermediates for the recovery of metals into solution. Once the metals are extracted into solution, the value metals can be recovered by hydrometallurgical means such as by purification followed by electrowinning, hydrogen reduction, pyrohydrolysis, crystallization and other unit operations. In many of these integrated flowsheets (refer Figure 1) the pressure leach step is pivotal to the recovery of the value metals from the host material. The exothermic leaching process The leaching of sulphide concentrates and intermediates as well as alloys is often accompanied by the release of energy, which has to be removed from the autoclave slurries in order to avoid the slurry temperatures within the vessel exceeding the design operating values. For example, in the pressure leach of chalcopyrite the reaction could be represented by the following relationship:

The energy release, as calculated from heats of formation (25°C), is approximately 1 690 kilojoules per gram mole. This exothermic heat release can limit the capacity of an autoclave INCREASING THE CAPACITY OF EXISTING AND NEW EXOTHERMIC AUTOCLAVE CIRCUITS

43

43-56_Dunn:text

2/12/09

10:58 AM

Page 44

Figure 1. Example of hydrometallurgical flowsheet for the recovery of copper from a flotation concentrate

to 1.1 to 1.3 tonnes of concentrate per active cubic metre per day where a retention time of approximately 1.2 hours has been prescribed. There are chalcopyrite concentrate autoclave circuits that have been subjected to feasibility studies and designed to operate in this capacity range (approximately 8–12% (w/w) feed solids). At higher feed concentrations the energy release is such that the autoclave temperature will exceed the 210–230°C typical operating range being considered for POX autoclaves treating chalcopyrite. This condition triggers the need to remove energy from the autoclave in order to maintain temperature control and the desired oxygen partial pressure in the vessel. There are a variety of heat removal processes that are available to the autoclave designer, and these will be examined individually. Quenching The quenching processes appear to be favoured by most engineers involved in autoclave design. The general concept is captured in Figure 2. It is simple to configure and has relatively simple control features. Quench liquor, in the leaching of copper concentrates, can be a solvent extraction raffinate or spent electrolyte. The CCD 2 liquor can be considered for the quench in cases where there is a counter current decantation solid-liquid separation circuit. The cool aqueous liquor is normally fed to the autoclave compartments on temperature control. In cases where there are little solids present in the quench fluid, multistage centrifugal pumps can be employed and, where this is not the case, an appropriate positive displacement pump may be suitable. The quench circuit of Figure 2 is also employed in the leaching of nickel and nickeliferous sulphides and mixed nickel and copper sulphides. The use of a quench liquor does require an increase in the autoclave active volume and can result in dilution of the pregnant leach solution (PLS) if water or some diluted liquor is employed in this duty. Internal cooling coils In some autoclaves, internal cooling coils have been fitted into the compartments to abstract heat. These cooling coils are often fixed to the dividing walls of the autoclave and are serviced by cooling water or other process fluids. There are numerous limitations associated with the use of coils, which include: 44


43-56_Dunn:text

2/12/09

10:58 AM

Page 45

Figure 2. Flowsheet of autoclave with quench cooling

Figure 3. Flowsheet of autoclave with internal cooling coils

• • • •

Difficulties in fixing the coils to the vessels internals Designing the coils and their supports for the thermal stresses Deprivation of reactor volume by the coils A low degree of flexibility in being able to accommodate an increase in reaction extent and its concommitant impact on where the heat is generated within the autoclave, and • Scaling of heat transfer surfaces by precipitates or gangue from the feed materials. Cooling coils are employed in the mixed mineral millerite, chalcocite and covelite leach autoclaves of Southern Africa and the USA1. They are also used extensively in the Norilsk pyrrhotite-pentlandite leach autoclaves2. Figure 3 provides the cooling system concept employed in some Southern African autoclaves. External coolers An enhancement of the internal cooling coil concept is to be found in the use of external coolers. The slurry is removed from the autoclave and pumped through a heat exchanger before being returned to the same compartment of the autoclave. INCREASING THE CAPACITY OF EXISTING AND NEW EXOTHERMIC AUTOCLAVE CIRCUITS

45

43-56_Dunn:text

2/12/09

10:58 AM

Page 46

Figure 4. Flowsheet of autoclave with external coolers

Some of the disadvantages of the internal cooling coils are addressed in the use of these external heat exchangers, however: • Scaling of the heat transfer surfaces can still occur, albeit to a lesser extent. This reduced scale propensity is as a consequence of the higher tube side velocities that can be achieved within the heat exchanger • The pumps required to circulate the autoclave contents through the external heat exchangers have certain temperature and pressure limitations, resulting in this concept being limited to low and medium pressure autoclave circuits, and • Significant wear to isolation valves can be experienced on the pump suction and heat exchanger discharge lines. One acceptable feature of external pumped coolers is that additional cooling surface can be added, thereby improving their flexibility with respect to varying or unpredictable reaction extents during the operation phase. Figure 4 provides a simple flowsheet of the external cooler employed in a copper-cobalt leach autoclave in Zambia3. Flash-recycle In most exothermic autoclave circuits a majority of the energy is liberated in the first compartment, with only smaller quantities being generated in the downstream compartments. An alternate approach to the removal of heat from exothermic autoclaves was introduced in South Africa in the mid 1980s in a pressure leach step for a blend of chalcocite, covelite and small quantities of millerite. Since then it has been installed in two other operations in South Africa1, one in Zimbabwe4, the USA6 and Australia. The flowsheet for some Southern African operations is given in Figure 5. Any energy release in the first compartment that will result in the slurry exceeding the design temperature is abstracted by removing slurry from the first compartment to the feed tank via a flashing process and returning it to the autoclave. The feed tank becomes an extension to the first compartment in this circuit and while a recycle loop is established between this tank and the autoclave, the overall retention time of the autoclave is related to the net new feed rate entering the circuit from the concentrate tank and is not influenced by the recycle loop flow rates. 46


43-56_Dunn:text

2/12/09

10:58 AM

Page 47

Figure 5. Flash-recycle flowsheet

There are several control methods available to the autoclave engineer for the flash-recycle circuit and these serve, in a well designed system, to maintain the temperature in the first compartment at the set point with a tolerance of +3°C. This temperature control is normally acceptable for most brick lined, alloy or alloy clad autoclaves. Some of the advantages of this system compared to the three options above are that it provides: • For optimal utilization of the installed reactor volume • For enhanced water balance control • The ability to concentrate the reactor contents, if required, through the evaporation of water flashed as steam • The potential to use the flashed steam as an energy source elsewhere in the operation, and • For an opportunity to introduce first compartment reagent addition into the feed tank. The flashed slurry recycle loop can provide a lower cost vehicle for reagent additions to the first compartment of an autoclave. The flash recycle (FR) circuit can be retrofitted to autoclaves employing any of the alternate heat removal systems listed above. Invariably, any exothermic heat carried over into the second and ensuing compartments can be removed, for example, by the use of coils as this normally constitutes a small portion of the overall exothermic heat load. When the above heat removal interventions are considered holistically only the FR circuit provides the autoclave designer with the maximum degrees of freedom. A disadvantage, for example, of the quench process is that the fluid introduced into the first compartment for temperature control can often contain the products of the autoclave discharge, which may disturb the desired equilibrium conditions in that first compartment. It could also, for example, as in some matte leach circuits, be a spent electrolyte, which introduces an acidic liquor into the first compartment. The acid concentration in this and ensuing compartments is therefore influenced by the thermal quench and may result in a condition where the operator is not able to control this parameter. He may in fact require a different acid concentration for the optimal leach conditions over that which is delivered by the quench temperature control. A further disadvantage in the case of the internal and external cooler system is that any feed density fluctuations above the set point may result in the introduction of a larger feed mass, INCREASING THE CAPACITY OF EXISTING AND NEW EXOTHERMIC AUTOCLAVE CIRCUITS

47

43-56_Dunn:text

2/12/09

10:58 AM

Page 48

for example, of the slurry to the autoclave. The temperature rise in the autoclave first compartment may be excessive and the energy release greater than what the coolers can handle. In continuous operations the reaction extent in the first compartment could be reduced and consequently the second and ensuing compartments’ heat removal circuits may not be able to abstract the extra heat load thrust on them by the feed density fluctuation in this case. Unless design flexibility is provided, the temperatures can ‘run away’ and an incomplete leached product can be discharged from the autoclave. However, in the flash-recycle circuit: • Temperature control is achieved without disturbing or changing the process variables within the first compartment by the return of another process fluid • The leachate can be concentrated, if required, by not replacing all the water that is flashed as steam in the flash step • If the feed mass is higher to the autoclave as a consequence, for example, of poor density control then the automatic level adjustment of the feed tank can be invoked to ensure the design reaction extent is achieved in the first compartment, thereby preventing conditions developing in which unreacted product is discharged from the autoclave, and • The mean autoclave retention time is the highest of all the known methods of abstracting heat from the autoclave, thereby increasing the autoclave productivity (tonnes per day per cubic metre of autoclave volume). Autoclaves in matte leach circuits today have been increased in capacity by in excess of 45 per cent employing the flash recycle system whereas others have been derated in pressure and temperature without impacting the original treatment rate. In autoclaves operating at temperatures close to the atmospheric boiling point of the aqueous fraction, Norilsk have proposed their patented vacuum flash. This circuit was in operation at its Tati Nickel demonstration plant in Botswana6. The first compartment flash vessel is operating at partial vacuum, which allows the flash underflow slurry to be reduced in temperature below the atmospheric boiling point. This generates a higher sensible heat loss than would otherwise occur in an flash-down to an atmospheric boiling condition. Additional mechanical equipment in the form of condensers, vacuum pumps and sealing fluid cooling circuits, etc. are required in this variant of the flash-recycle circuit. Flash-thickener-recycle (FTR) circuit An extension of the flash-recycle circuit is to be found in the incorporation of a solid-liquid separation step after the flash. This solid-liquid separation step can be on the first compartment flash but it can also be on the flash from any compartment of the autoclave. This solid-liquid separation is often best achieved with a thickener but it can be effected with other equipment such as a classifier or a filtration step. This flow sheet modification will be referred to as the flash-thickener-recycle or FTR circuit. The FTR circuit provides a means of increasing the retention time of the solids fraction within the autoclave over that of the liquid fraction. A further embellishment of the flowsheet concept of Figure 5 incorporates a thickener in the flash-recycle loop. In circumstances where there is a mass reduction of the feed concentrate in the leach, the incorporation of a thickener in the flash cooling circuit permits the removal of a fraction of the leachate. The partially leached solids are returned from the thickener to the autoclave via the feed tank. The thickener thus provides a means of increasing the solids’ retention time in the autoclave over that of the liquid phase. Alternately, for a fixed retention time, the capacity of an autoclave can be increased subject to the impellers delivering sufficient oxygen mass transfer. 48


43-56_Dunn:text

2/12/09

10:58 AM

Page 49

Figure 6. Flash−thickener−recycle flow sheet

Figure 6 depicts this concept of the FTR circuit. The FTR circuit not only permits the abstraction of leachate from the first compartment, but also provides a means of restoring some water to the feed tank to drive the equilibrium in favour of the leach extraction. Alternately, in the absence of water addition, salt hydrolysis can occur if this is desired. In the design of or a modification of an existing autoclave, the first compartment is sized to achieve in excess of 50 per cent but typically 85 to 95 per cent of the overall reaction extent. This can and often results in the thickener overflow being very similar in composition to the autoclave discharge stream after flash. The mean mass flow rate in the first and ensuing compartments of the autoclave is: In the FTR circuit the net feed rate5 through the autoclave can be adjusted to suit the required mass flow and extraction simply by adjusting the volumetric flow at stream4 subject to thickener underflow viscosity and density constraints. By diverting the thickener overflow to the discharge tank it has been possible to increase the retention time by as much as 200 per cent before the autoclave slurry density increases to a point where it is not economical to increase it any further. Other factors such as oxygen mass transfer at the autoclave impellers may become rate limiting. The flash-thickener-recycle (FTR) concept has been patented by Hydromet7. Applications Covellite-chalcocite pox leach The flowsheet of Figure 7 has been considered and tested by an operation processing nickelcopper sulphide mattes. The autoclave was operated as a bulk leach to provide an approximately 90% mass reduction in the feed. It was followed by a polishing leach autoclave circuit. This modification was made to an existing three compartment autoclave and: • Increased the retention time of the solids fraction from 3.0 hours to an excess of 7.5 hours, and • Increased the capacity of the autoclave from 0.71 to 1.32 tonnes of feed per cubic metre (active volume) per day at an operating temperature of 150°C. INCREASING THE CAPACITY OF EXISTING AND NEW EXOTHERMIC AUTOCLAVE CIRCUITS

49

43-56_Dunn:text

2/12/09

10:58 AM

Page 50

Figure 7. Covellite-chalcocite POX leach

Figure 8. Extraction of copper in the FTR circuit for a covellite-chalcocite leach

The impellers in the autoclave were not upsized for the FTR duty. These agitators became ‘rate limiting’ in their oxygen transfer rate and were responsible for a capacity increase significantly below what would have been expected for the additional retention time that was achieved with this modification. The extraction of copper and oxidation of sulphide sulphur at the higher capacity in the FTR circuit is given in Figure 8 and Figure 9. It needs to be noted that this was a three compartment autoclave possessing the normal deficiencies of such a vessel. In this case the overall reaction extent in the first compartment was in excess of 80%. When the circuit was operated as a standard flash recycle cooling circuit (without the thickener in service) and with a mean solids retention time of 3 hours, the copper extraction profile was typically that shown in Figure 10 with a first compartment copper extraction of 97%. The lower copper extraction in the FTR circuit is attributed to the rate limiting features of the agitation systems, which was unchanged in both cases. 50


43-56_Dunn:text

2/12/09

10:58 AM

Page 51

Figure 9. Oxidation of sulphide sulphur in the FTR circuit for a covellite-chalcocite leach

Figure 10. Kansanshi Mining’s chalcopyrite POX circuit with quench cooling

Chalcopyrite leach The pressure oxidative leach of chalcopyrite is a strongly exothermic process. The classical approach with quench cooling can result in the feed density being reduced to between 8 and 12% feed solids equivalent on a weight basis. In Zambia, Kansanshi Mining plc (a company 80% owned by First Quantum Minerals Limited) installed a chalcopyrite pressure oxidation (POX) process to the flowsheet in Figure 10. An atmospheric leach of oxidized copper ore is being undertaken on the present site and the plan is to integrate the sulphide POX leach into the atmospheric leach circuit to provide sulphuric acid, energy and some ferric iron as a lixiviant for some secondary sulphides in the oxide ore. The PLS from the atmospheric leach CCD circuit is fed to a copper SX circuit. The raffinate from this SX circuit is returned, in part, to the POX circuit for the repulp of concentrate and the quench. The POX circuit has been designed to operate: • At 210°C and approximately 3.0 MPa(g) • With a first compartment mean solids residence time of just under one hour and an overall residence time of approximately 1.7 hours, and • With two autoclaves for a concentrate treatment rate of 105 kt/annum. INCREASING THE CAPACITY OF EXISTING AND NEW EXOTHERMIC AUTOCLAVE CIRCUITS

51

43-56_Dunn:text

2/12/09

10:58 AM

Page 52

Figure 11. Kansanshi Mining’s chalcopyrite POX Circuit with FTR

Table I Process factors of the circuit

Specific autoclave feedrate at 1.7 hours Overall retention time Sulphuric acid requirements First compartment mean solids retention time First compartment slurry feed rate First compartment dilution feed rate First compartment oxygen transfer rate

m3)

(t.concentrate/day (t.Cu/day m3) (t.S=/day m3) (t/tonne concentrate) (h) (m3/h) (m3/h) (kg/m3h)

Quench concept

FTR concept

1.14 0.33 0.37 0.43 0.94 7.2 65 62

3.04 0.89 0.97 ~nil 0.94 78 110–140 147

Kansanshi Mining are considering the FTR circuit on a single POX autoclave employing the flowsheet of Figure 11. The FTR circuit will: • Operate at 210°C and approximately 3.0 MPa(g), i.e. at identical conditions as those in the quench concept • With the same mean solids retention time distribution as the quench circuit, and • With one autoclave designed to treat approximately 130–150 kt/annum. The specific process factors of the circuit are summarized in Table I. In the FTR circuit, in excess 75% of the leached copper in the autoclave will be removed in the thickener overflow and diverted directly to the discharge tank. This is approximately 80% of the total copper leached in the first compartment. The remaining leached copper is in the slurry overflowing from the first compartment to the second and ensuing compartments of the autoclave. The leaching of nickel sulphides Nickel sulphides mattes have been leached in Southern African matte POX circuits for over 35 years8,9. 52


43-56_Dunn:text

2/12/09

10:58 AM

Page 53

Figure 12. Pressure oxidation of nickeliferous sulphides with FTR

Figure 13. FTR circuit for minerals displaying different leach kinetics

The leaching of nickeliferous sulphide concentrates has been practiSed at Outokumpu using the Hitura concentrate10 and is currently practiSed at Norilsk on pyrhhotite rich concentrates. Inco are considering a pressure leach circuit for its Voisey Bay concentrate. The flow sheet in Figure 12 has merits where there is a significant mass loss from the feed concentrate in the leach. It is, for example, suited to the acid pressure oxidative leaching of mixed nickel and cobalt sulphides generated from a hydrogen sulphide precipitation process of a laterite leachate. The autoclave capacity can be reduced by over 200% of that employed in the classical quench cooling process. Kinetically different leach systems The platinum group metals are known to respond slower than the base metals in the acid leach of a mix of base metal sulphides and PGM sulphides, selenides, tellurides, arsenides, bismuthides, etc. INCREASING THE CAPACITY OF EXISTING AND NEW EXOTHERMIC AUTOCLAVE CIRCUITS

53

43-56_Dunn:text

2/12/09

10:58 AM

Page 54

Where it is the intent to dissolve both the base metals and the precious metals, it may be appropriate to separate a majority of the base metal leachate in a first compartment FTR circuit. The extraction of the precious metals can then be made in the downstream compartments of the autoclave in a considerably smaller autoclave. The circuit of the type shown in Figure 13 has some merits in that: • A large part of the base metals are removed in the leachate overflow from the FTR thickener • Any PGE that do dissolve in the first compartment are precipitated on the concentrate, which is blended with the first compartment leach slurry prior to thickening. The thickener overflow will therefore be low in precious metals concentration • The lixiviant that is employed for PGE dissolution e.g. chloride ions in the Platsol process may need to be only added into the second compartment of the autoclave. This could reduce the material and maintenance costs of the autoclave significantly, and • The volume of leachate containing PGE is much smaller as is the autoclave itself. Conclusions The introduction of the thickener in the flash front end cooling circuit increases the retention time of the solids fraction over that of the aqueous stream. The FTR circuit can be retrofitted to existing POX circuits to increase the capacity of these circuits at minimal capital and operating cost. The FTR concept can be designed into new POX circuits to reduce the size of the autoclave or reduce the operating complement of autoclaves where parallel units are being considered for a plant. References 1. STEENEKAMP, N. and DUNN, G.M. Operations of and Improvements to the Lonrho Platinum Base Metal Refinery, EPD Congress, 1999, pp. 356–390. 2. BORBAT, V.F. and VORONOV, A.B. Autoclave Technology for Processing of NickelPyrrhotite Concentrates, Metallurgy, Moscow, 1980, 185. 3. MUNNIK, E. SINGH, H. UYS, T. BELLINO, M. DU PLESSIS, J. FRASER, K., and HARRIS, G.B. Development and Implementation of a Novel Pressure Leach Process for the Recovery of Cobalt and Copper at Chambishi, Zambia, The Journal of the South African Institute of Mining and Metallurgy, Jan.–Feb. 2003. 4. MONTGOMERY, G.W.G. and HOLOHAN, T.N. Development and Design of the Hartley Platinum Base Metal Refinery Flow sheet, South African Institute of Mining and Metallurgy, April. 1997. 5. NEWMAN, L. and WYRICK, R. Modification of the Stillwater Base Metal Refinery Process Pressure, Hydrometallurgy, 2004, pp. 499–525. 6. SWARTS, A. HOLLIDAY, H. DONEGAN, S., and NEL, G.J.G. The Tati Activox® Demonstration Plant: Changing the Future of Nickel Production in Southern Africa, Cape Conference, South Africa, 2005. 7. DUNN, G.M. Exothermic Pressure Leach Autoclave Circuits, PCT/ZA 2005/000002. 8. PLASKET, R.P. and ROMANCHUK, S. Recovery of Nickel and Copper from HighGrade Matte at Impala Platinum by the Sherritt Process, Hydrometallurgy, 1978, pp. 135–151. 9. PLASKET, R.P. and DUNN, G.M. Commissioning Experiences in the Cobalt plant at Impala Platinum Ltd, Minerals and Metallurgical Processing, 1986, Feb Issue. 10. NYMAN, B., AALTONEN, A., HULTHHOLM, S.E., and KARPALE, K. Application of new hydrometallurgical developments in the Outokumpu HIKO process, Hydrometallurgy, 1992, pp. 471–478. 54


43-56_Dunn:text

2/12/09

10:58 AM

Page 55

G.M. Dunn Director, Hydromet (Pty) Ltd, South Africa Mr Dunn has held various positions over 22 years at Impala Platinum Refineries in pressure hydrometallurgical operations from shift operator to plant manager to general manager. He then spent brief periods at SRK and Hatch in consulting roles. In 2000 he established his own consulting company in South Africa, Hydromet (Pty) Limited, working in nickel, cobalt, manganese, uranium and PGE. He has spent 18 years in pressure hydrometallurgical operations and base metal and PGM recovery. He has spent 20 years in project engineering and project management. Mr. Dunn is a Chartered Engineer (UK) and Professional Engineer (South Africa) having graduated at the University of Cape Town in 1970 with an honours degree in Chemical Engineering. He is a Fellow of the Institute of Chemical Engineers in the UK, a member of the SA Institute of Mining and Metallurgy and the SA Institute of Chemical Engineers. He is a member of American Institute of Mining, Metallurgical and Petroleum Engineers (AIME,TMS).

INCREASING THE CAPACITY OF EXISTING AND NEW EXOTHERMIC AUTOCLAVE CIRCUITS

55

43-56_Dunn:text

56

2/12/09

10:58 AM

Page 56


57-64_Lauzon:text

2/12/09

11:34 AM

Page 57

LAUZON, P., KONING, A., and DONOHUE, I. Stress development in refractory due to the rate of temperature change: a pressure vessel refractory lining design consideration. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Stress development in refractory due to the rate of temperature change: a pressure vessel refractory lining design consideration P. LAUZON*, A. KONING*, and I. DONOHUE* *Hatch, Ontario, Canada

Vessels that are used for pressure hydrometallurgical operations require an impermeable membrane that provides corrosion protection and one or more courses of refractory lining or ceramic brick. A significant amount of work goes into the design of a refractory lining to ensure that it is mechanically stable at steady state process conditions. Steady state analysis techniques are covered in detail in the paper ‘Design fundamentals for hydrometallurgy pressure vessel refractory linings’ by A. Koning and P. Lauzon. Since new hydrometallurgical processes are pushing the pressure and temperature envelope with each new generation of plants it becomes equally important to perform transient analysis. The purpose of this paper is to build on the design fundamentals of refractory linings by demonstrating the importance of transient thermal and stress analysis. A transient analysis is necessary because heating or cooling a vessel too quickly can result in lining failure as a result of exceeding the stress limits of the refractory. To demonstrate the effects that the rate of temperature change has on the stresses in a refractory lining, a transient thermal and stress analysis was conducted on a typical refractory lining design for an autoclave. The results obtained from the analysis shows a significant increase in peak stress when the rate of heating or cooling is increased by 5°C/h. Peak stresses were increased by approximately 2 MPa. A stress increase of this magnitude is significant because tensile failure of refractory brick occurs in a range of 6 to 10 MPa. The magnitude of stress development is highly dependent on geometry, materials, and the thermal boundary conditions. Since stress development is affected by multiple factors it is important to analyse the transient effects of heating or cooling a vessel. STRESS DEVELOPMENT IN REFRACTORY DUE TO THE RATE OF TEMPERATURE CHANGE

57

57-64_Lauzon:text

2/12/09

11:34 AM

Page 58

Introduction Pressure hydrometallurgy operations require vessels to be lined with an impermeable membrane for corrosion protection and one or more courses of refractory or ceramic brick. Examples of unit operations that utilize composite lining systems include pressure oxidation autoclaves, sulphide precipitation autoclaves, chloride leach reactors, flash vessels, cyclone separators, and direct contact condensers (heater vessels and quench vessels). The refractory lining must satisfy multiple requirements: it must thermally insulate the membrane from process fluid, be structurally stable, provide erosion resistance, be chemically compatible with process fluid, and provide an economic service life. Refractory lining design begins with selecting lining materials based on chemical stability in the process environment. The thickness of the different refractory layers is initially determined using 1D heat transfer calculations. If the membrane or shell temperature is greater than what is permitted then the refractory thickness is increased. Following the thermal analysis is a 1D mechanical stability analysis, which is used to examine various loading conditions and the effects of material properties. Mechanical stability is determined by ensuring that the hot face refractory layer is in compression through all operating conditions and that the stresses do not exceed the material’s failure strength. An equally important consideration is the calculated overlap between the brick lining and the membrane. This stability factor requires that the overlap should be positive, that is, that the steel shell should be in contact with the refractory lining throughout all operating conditions. The main drawback of 1D calculations, is the assumption of an infinitely long cylindrical vessel that does not take into account the effects of additional attachments: supports, nozzles and hemi-heads. A designer must resort to the use of finite element analysis (FEA) because the various attachments require complex mathematical analysis for which exact formulae are difficult or impossible to obtain. With the use of 2D axisymmetric and full 3D models the designer will perform thermal and stress analysis for various load conditions. FEA will highlight hot spots in the vessel or stress concentrations, which would lead to failure. The designer can then concentrate on the hot spots or stress concentrations and modify parameters such as gaps, geometry, and material selection in order to eliminate these hot spots or stress concentrations. Once a refractory lining design passes the mechanical stability and thermal requirements at steady state process conditions it becomes necessary to examine the effects of start-up and shutdown procedures. Steady state analysis techniques are covered in detail in the paper ‘Design fundamentals for hydrometallurgy pressure vessel refractory linings’ by A. Koning and P. Lauzon. Analysis of start-up and shutdown procedures for a vessel is required because a refractory lining is dramatically affected by the rate of temperature change. The purpose of this paper is to demonstrate the effect that the rate of temperature change has on refractory stresses during start-up and shutdown. If the vessel is heated or cooled too quickly stresses can be generated that exceed the stress limits of the refractory, leading to failure of the lining. It is the hot face of the process brick that will experience the greatest effects of thermal shock and will crack or spall if the rate of temperature change is too high. To assist in prolonging the life of the refractory lining it is important that appropriate heating and cooling rates be determined and followed. The damaging effects of a large rate of temperature change will be discussed and examined using FEA. 58


57-64_Lauzon:text

2/12/09

11:34 AM

Page 59

Transient thermal stresses Refractory material is selected for its excellent insulating material properties: low conductivity, high density, and high specific heat capacity. The insulating ability of the refractory is required for the thermal protection of the vessel membrane. The downside is the length of time required to heat or cool the refractory lined vessel. When the refractory in a pressure vessel is being heated it resists the transfer of heat from the hot face to the cold face, generating steep thermal gradients through the material. Stresses are generated within the refractory as a result of uneven thermal expansion due to the nonuniform temperature distribution. If the refractory is heated quickly, large enough compressive stresses can be generated, which cause the lining to fail due to thermal shock. The same type of failure can also occur when the vessel is cooled too quickly. In this case the hot face becomes cooler than the remainder of the refractory and develops a tensile stress. The results of a transient thermal and stress analysis, using FEA, effectively shows the magnitude of stresses generated as a result of rapid temperature change. Transient analysis using FEA The effects of rapid temperature change on refractory will be shown using the results from a two-dimensional FEA model. Figure 1(a) shows the general arrangement of the model used for the analysis, which approximates a cross-section of a quarter of the pressure vessel’s cylindrical body. It is assumed that this geometry represents an infinitely long cylinder into the page. In the model, plane strain applies. This model geometry will behave identically to a complete annulus and reduces the computational time because the model is smaller. The model is made up of 8 different layers as shown in Figure 1(b): steel shell, lead membrane, 3 mortar layers, and 3 brick layers. The brick layers are modelled as monolithic layers that do not account for the mortar that is between the bricks circumferentially or axially. These mortar joints could be modelled but are not necessary to show the effects of the rate of temperature change.

Figure 1. (a) General arrangement of the model used for the transient analysis and (b) the different material layers

STRESS DEVELOPMENT IN REFRACTORY DUE TO THE RATE OF TEMPERATURE CHANGE

59

57-64_Lauzon:text

2/12/09

11:34 AM

Page 60

The model geometry is a representation of the cylindrical portion of an autoclave. Thicknesses used for each layer in the model are from the design of an existing autoclave. The autoclave is being used for the processing of an orebody that contains gold. Transient thermal and stress analysis was conducted using the 210°C process design temperature for this particular autoclave. Autoclaves used to process an orebody containing gold can have process temperatures that range from about 200°C to 250°C. Process conditions will vary depending on the orebody being processed, the chemical reaction taking place, and the type of vessel being used. The effect of heating rate on refractory The refractory used to line a vessel is selected for its excellent insulating properties to keep the membrane below a particular temperature at process conditions. At steady state operating conditions the thermal gradient through the refractory generates compressive hot face stress and tensile cold face stress. Larger stresses will be developed in the refractory as the vessel is being heated, as shown in Figure 2. The peak stress occurs when the vessel reaches the 210°C process temperature and then stabilizes to a steady state position. For the given analysis a 5°C/h increase in the heating rate increased the peak stress by approximately 3 MPa. A peak stress is formed due to the thermal gradient through the refractory. The steepest thermal gradient occurs when the vessel reaches the desired process temperature. This is illustrated by the thermal gradient at the time of 19 h for Case 2 in Figure 3. This occurs because the hot face has achieved its steady state temperature but the remainder of the refractory is still being heated. The faster a vessel is heated the steeper the thermal gradient through the refractory when the vessel reaches the process temperature, as shown in Figure 4 for Case 1 and Case 2. This accounts for the largest peak stress for Case 2. The slope of the thermal gradient through the refractory and stresses are reduced over time as the refractory reaches steady state. The 12 h hold shown for Case 3 and 4 in Figure 2 is an example of a temperature hold typically included in a vessel heating schedule during commissioning or after prolonged shutdown. The hold period allows the lining to settle into place and helps prevent damage to the

Figure 2. Hot face stress of the process brick

60


57-64_Lauzon:text

2/12/09

11:34 AM

Page 61

lining. It provides time for the acid to completely soak through the refractory to prevent flashing. A hold period allows the vessel to reach an equilibrium state so the operator has an opportunity to check that nothing is out of the ordinary (i.e. hot spots) before continuing the heating process.

Figure 3. Process brick thermal gradient for Case 2

Figure 4. Process brick thermal gradient

STRESS DEVELOPMENT IN REFRACTORY DUE TO THE RATE OF TEMPERATURE CHANGE

61

57-64_Lauzon:text

2/12/09

11:34 AM

Page 62

The effect of cooling rate on refractory The tensile strength of refractory typically falls within a 6–10 MPa range. When a vessel is cooled there is greater potential for lining failure due to the possibility of exceeding the low tensile strength of the refractory. Looking at Cases 2, 3, and 4 in Figure 5 it can be seen that the peak stress is approaching the tensile strength of refractory as the cooling rate is increased. The stress on the hot face has become tensile because it is now cooler than the cold face. For the given analysis a 5°C/h increase in the cooling rate increased the peak stress by approximately 2 MPa. A peak stress is formed due to the thermal gradient through the refractory. For Cases 2, 3, and 4 the steepest thermal gradient occurs when the vessel temperature reaches 60°C and cooling by means of natural convection is initiated. This is illustrated by the thermal gradient at a time of 6 h for Case 4 in Figure 6. It can be seen that at the 7 h mark the hot face increased in temperature and the remainder of the refractory continued cooling. This occurred because the rate of heat removal due to convection is slower than the rate of heating by the latent heat remaining in the hotter portion of the refractory. The faster a vessel is cooled the steeper the thermal gradient through the refractory when the vessel temperature reached 60°C, as shown in Figure 7 for Cases 2, 3, and 4. This accounts for the largest peak stress for Case 4. The slope of the thermal gradient through the refractory and stresses are reduced over time as the refractory reaches steady state. The effect of cooling the vessel interior from 60°C to 20°C using convection can be seen looking at Case 1 and 5 in Figure 5. These two cases do not have a stress peak because the vessel is cooled to 60°C using a slow enough rate that the convective air passing though the vessel has a larger influence on the tensile stress development. The tensile stress is caused by the 40°C temperature difference between the convective air and the hot face of the process brick. Case 5 has a larger tensile stress because forced convection has a larger effect on the thermal gradient through the brick because it removes more heat than natural convection. The forced convection rapidly decreased the hot face temperature of the process brick causing the generation of the larger tensile stress on the hot face.

Figure 5. Hot face stress of the process brick

62


57-64_Lauzon:text

2/12/09

11:34 AM

Page 63

Figure 6. Process brick thermal gradient for Case 4

Figure 7. Process brick thermal gradient

Conclusions The rate of temperature change has a noteworthy effect on the stresses that are developed in a refractory lining. The stress observed, during heating or cooling, is significantly higher than steady state stress, which demonstrates the need for transient thermal and stress analysis when designing a lining system. The stresses are highly dependent on geometry considerations, material properties, and thermal boundary conditions. So there are many variables that can be changed to reduce the stresses within the refractory. Performing a transient thermal and stress analysis during the design phase will result in a better engineered refractory lining with an improved life expectancy. STRESS DEVELOPMENT IN REFRACTORY DUE TO THE RATE OF TEMPERATURE CHANGE

63

57-64_Lauzon:text

2/12/09

11:34 AM

Page 64

Al Koning Mechanical Engineer, Hatch, Ontario, Canada Mechanical engineer for the design and construction of chemical and metallurgical processing plants, encompassing detailed mechanical design and project engineering for the extraction of non-ferrous metals such as gold, nickel, cobalt and copper. Extensive experience using finite element analysis (FEA) and other analytical methods in the design, diagnoses and retrofitting of pressure vessels and piping components.

64


65-78_Nicolle:text

2/12/09

11:35 AM

Page 65

NICOLLE, M., NEL, G., PLIKAS, T., SHAH, U., ZUNTI, L., BELLINO, M., and PIETERSE, H.J.H. Mixing system design for the Tati Activox® autoclave. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Mixing system design for the Tati Activox® autoclave M. NICOLLE*, G. NEL†, T. PLIKAS‡, U. SHAH‡, L. ZUNTI‡, M. BELLINO*, and H.J.H. PIETERSE§ *Hatch Africa (Pty) Ltd, South Africa, †Norilsk Nickel Africa (Pty) Ltd, South Africa ‡Hatch (Pty) Ltd, Ontario, Canada, §Pieterse Consulting Inc., Arizona, USA The Tati Activox® Project will be the first full-scale implementation of the patented Activox® process. The process was developed by Norilsk Process Technology and has been tested on Tati mine sulphide concentrates in laboratory, pilot and demonstration plant scales and demonstrated its viability. There are inherent risks to the final scale-up of the process from the demo plant and one of them will be investigated in this paper. Compartment 1 is designed to leach approximately 77% of the total nickel leached. For this reason agitation requirements in the first compartment of the autoclave are reviewed. An attempt is made to minimize the process and mechanical risks associated in achieving oxygen mass transfer into the slurry solution. The agitator powers for oxygen mass transfer are calculated using empirical correlations and compared to demonstration plant testwork. The resulting gassed power per unit volume (P/V) is higher than most commercial autoclaves and raises uncertainty on the viability of using such high unit power inputs. Additionally, there is concern about the ability of the autoclave shell to withstand and support the higher loading of large agitators. An alternative solution to designing for the increased P/V is assessed in which the number of compartments within the autoclave is reduced from 5 to 4 by removing the compartment wall separating compartments 1 and 2. This results in an enlarged first compartment containing 3 agitators instead of 2. Therefore the compartment 1 oxygen demand is supplied through 3 agitators, which lowers the P/V per agitator. The reduction in the number of autoclave compartments raises the potential for short-circuiting the mean flow pattern by slurry particles. Short-circuiting and low velocity zones could result in a lower recovery of metal and localized hot spots, respectively. A computational fluid dynamic (CFD) analysis was conducted to quantify these concerns and also to evaluate further design considerations. The results indicate that the proposed design change to 4 compartments affects short-circuiting. The impact of increased short-circuiting on the overall autoclave recoveries is not quantified, however; it is expected to be negligible based on testwork done in the Tati Demonstration Plant and similar modifications made to another operating autoclave. The CFD analysis also suggests that there will be no low velocity zones within the compartment. MIXING SYSTEM DESIGN FOR THE TATI ACTIVOX® AUTOCLAVE

65

65-78_Nicolle:text

2/12/09

11:35 AM

Page 66

Introduction Tati Nickel base metals refinery makes use of the patented Activox® process and will be the first full-scale commercial implementation. The ultra fine grinding (UFG) and autoclave pressure leach process conditions are the heart of the Activox® patent. The hydrometallurgical process has environmental and financial benefits over a pyrometallurigcal process in that it does not produce sulphur dioxide or acid as a by-product. The hydrometallurgical base metal refinery is designed to produce 25000 tpa nickel, 22000 tpa copper and 639 tpa cobalt as cobalt carbonate. In the first phase of project implementation, process verification was completed during which the bankable feasibility study (BFS) design was assessed. Compartment one is designed to leach approximately 77% of the total nickel mass leached across the autoclave, as determined by the Tati Demonstration Plant campaigns. As a high portion of the leach reaction occurs in compartment one, significant attention was given to the compartment design. One particular area of review was the BFS basis for the oxygen mass transfer and dispersion within the autoclave’s first compartment. The following sections of the paper outline the testwork completed and methods used to evaluate the mixer design in compartment 1 of the autoclave. This resulted in a design that was subjected to CFD modelling and analysis. The results of this analysis are used to validate aspects of the final autoclave design. PROCESS OVERVIEW The concentrate is repulped with copper raffinate and fed via attrition scrubbers to a UFG mill circuit. The mill circuit consists of primary and secondary mills in series, which produce a finely ground concentrate. The size reduction significantly increases the surface area for the oxidation reactions to occur in the Activox® autoclave. Milled slurry is fed into 2 parallel Activox® autoclaves where the sulphide concentrate is oxidized. The autoclave operating temperature, pressure and chloride catalyst concentration result in the formation of favourable chemical species for handling downstream. Copper, cobalt and nickel are leached from the concentrate as sulphates in solution while elemental sulphur is formed as a by product. Elemental sulphur, a solid, allows more cost-effective processing than sulphate ions, which would require neutralization before removal from the circuit. The oxidation of concentrate is exothermic and requires careful temperature control. Temperature is maintained by direct quenching and a flash recycle. Due to the moderate operating temperatures in the leach vessels, the flash cooling and flash discharge systems are operated at a vacuum to increase the heat removal. The flash vent gases report to a condenser where cooled water condenses vapour from the vent gases. Non-condensables and the remaining water vapour is drawn by a vacuum pump and discharged into a venturi scrubber, which removes residual solids and entrained droplets, prior to discharge to the atmosphere. The leach discharge slurry is cooled in slurry cooling towers after which it reports to the solid-liquid separation area where the leach residue solids are washed with recycled process water. The washed leach solids are then pumped to the platinum group elements (PGE) flotation circuit. The solution is clarified prior to solvent extraction, electrowinning and precipitation to produce nickel cathode, copper cathode and cobalt carbonate. Basis for mixing system design The Activox® autoclave mixing system design is based on the process conditions outlined in Table I: 66


65-78_Nicolle:text

2/12/09

11:35 AM

Page 67

Table I Process conditions in autoclave compartment one Parameter

Value

Unit

Normal operating temperature Normal operating pressure Oxygen (O2) partial pressure Compartment 1 nickel mass recovery Stoichiometric O2 required in the first compartment O2 gassing rate—total required in first compartment O2 utilization Solids weight percent Solids specific gravity Liquor specific gravity Solids volume percent Slurry volume C1A (BFS compartment 1,autoclave agitator 1) Slurry volume C1B (BFS compartment 1, autoclave agitator 2) Slurry volume C2C (BFS compartment 2, autoclave agitator 3)

105 11 9.8 77 11181 12423 90 35 2859 1304 19.7 51.8 44.7 42.7

°C bar (a) bar (a) % kg/h kg/h % % wt/wt kg/m3 kg/m3 % m3 m3 m3

Design development The autoclave mixing system serves two purposes: firstly it creates an homogenous environment for the leach reactions to occur in by suspending the solids in the slurry, and secondly to replenish oxygen in solution as it reacts and oxidizes the ore. Gaseous oxygen is fed to each autoclave compartment through nozzles located under the agitator shafts and as such at the localized point in the autoclave with the highest shear. This maximizes oxygen mass transfer into the slurry solution whereafter it reacts with the sulphide ore according to the leach reactions. Empirical correlation for the P/V The agitation duty requirement in the first compartment is the greatest as the reactions in this compartment have the greatest oxygen demand. The stoichiometric mass of oxygen transferred into solution in compartment 1 is therefore critical to the leaching efficiency of the process. The transfer rate of oxygen to solution is known as the oxygen mass transfer rate (OTR) and is given in kg/m3/h. The OTR is a function of the mass transfer coefficient (kL), the interfacial area (a), oxygen solubility in the slurry solution (CO2-solution) and the actual concentration of dissolved oxygen in solution (C). This relationship is shown in Equation [1]. [1] The dissolved oxygen concentration C cannot be negative and the leach reaction extents are not controlled (the leach reactions are mass transfer limited when C = 0). Therefore, C must be greater than zero (C > 0) to achieve the required mass transfer. As the actual concentration of oxygen in the slurry solution cannot be greater than (CO2-solution), C is expressed as a percentage of (CO2-solution) in Equation [2]. [2] From test work (which gives values for O2 gassing rate and O2 utilization) it is possible to determine the OTR per agitator using Equation [3]: MIXING SYSTEM DESIGN FOR THE TATI ACTIVOX® AUTOCLAVE

67

65-78_Nicolle:text

2/12/09

11:35 AM

Page 68

[3] where O2-gassing rate = 12 423/number of agitators in compartment 1 [kg/h] O2-utlization = 0.90 Liquid volume = Slurry volume (1 – volume fraction of solids in slurry) In order to calculate the mass transfer coefficient using Equation [1] the solubility of oxygen in slurry solution (C O2-solution) must first be determined. (C O2-solution) is lower than the solubility of oxygen in pure water (CO2-water) due to the presence of dissolved salts. The relationship for (CO2-water) is shown in Equation [4] (Tromans 1998). [4] where H = Henry’s coefficient [kg.atm.mol-1] = Partial pressure of oxygen gas [atm] po CO2-water = Molal concentration of dissolved oxygen in water [mol O2. kg water-1] Henry’s coefficient is calculated from the relationship given in Equation [5]. [5]

Where T = temperature in K. At the process temperature of 105°C (378K)

g To convert to units of L·atm the formula must be multiplied by the density of water and the molecular weight of oxygen:

The oxygen solubility is lowered due to the presence of dissolved salts (mainly nickel, copper and cobalt sulfate) by a derating factor, ϕ (Tromans 1998). The relationship between solution oxygen solubility and water oxygen solubility is shown in Equation [6]. [6] The value of ϕ from Equation [6] is determined (using Figure 6 in Tromans 1998) with the expected molal concentration of nickel, copper and cobalt sulphate in the solution (2.047 mol/kg). ϕ is approximately 0.70 to 0.75. It is now possible to calculate C from Equation [2] as α varies between 0.2 and 0.3 according to industry norms. A value of 0.3 was used in this case and this assumption was validated at a later stage through testwork. It is anticipated that high actual concentrations of oxygen in solution (C) are unlikely due to the high sulphide concentration in the Tati concentrate slurry. With C known it is possible to calculate kLa from Equation [1], and so from Equation [7] below the gassed power per unit volume (P/V). The P/V required is calculated using the empirical Equation [7] (based on a recommendation made by Pieterse 2004) and adjusted using data the authors subsequently obtained from several commercial sized autoclaves. 68


65-78_Nicolle:text

2/12/09

11:35 AM

Page 69

[7] where

P/V = gassed power per unit volume in kW.m-3 And kLa = the mass transfer coefficient in h-1

P/V calculation for two agitators in compartment 1 The resulting power demand (3.89 kW/m3 – 4.73 kW/m3) on the agitators is very high in comparison to commercial autoclaves. This high power demand results in a very large agitator and raises concerns about the ability for the autoclave shell design to support the large weight and forces without the need for significant localized strengthening. By removing the weir wall separating compartment 1 from 2 the new compartment 1 will have 3 agitators under which the required oxygen can be added. This configuration is assessed in the same way as the scenario with 2 agitators in compartment 1. (Table II.) P/V calculation for three agitators in compartment 1 Table III summarizes the empirical calculation outlined in the section ‘Empirical correlation for the P/V’ of this paper. The same conditions as Table II were used with the exception that the oxygen demand was split over 3 agitators and not 2. The reduction in oxygen demand per agitator and subsequent reduced P/V (from between 3.89 kW/m3–4.73 kW/m3 to between 2.28 kW/m3 to 2.94 kW/m3) reduces concern about the agitator support on the autoclave shell as there are a limited number of commercial autoclaves operating under similar agitator duties and supports to those proposed for this application. Testwork in Tati demonstration plant Testwork was carried out on the Tati demonstration plant in Botswana in order to further clarify the agitation requirements of the autoclave. These test results were compared to the

Table II A summary of the empirical calculation as applied to 2 agitators in compartment 1 O2 gassing rate per agitator Weight percent solids Specific gravity solids Specific gravity liquor Volume percent solids Slurry volume—C1A Slurry volume—C1B Oxygen solubility in water—Equation [4] Derating factor—Equation [6] O2 solubility in solution—Equation [6] Dissolved O2 as % of soluble O2 in solution Dissolved O2 concentration in solution—Equation [2] Oxygen transfer rate in C1A—Equation [3] Oxygen transfer rate in C1B—Equation [3] kLa in C1A—Equation [1] kLa in C1B—Equation [1] P/V required in C1A [7] P/V required in C1B [7]

MIXING SYSTEM DESIGN FOR THE TATI ACTIVOX® AUTOCLAVE

6212 kg/h 35% wt/wt 2859 kg/m3 1304 kg/m3 19.7% vol./vol. 51.82 m3 44.65 m3 0.2430 kg/m3 0.735 0.1786 kg/m3 30% 0.0536 kg/m3 134 kg/h.m3 155 kg/h.m3 1069 h-1 1241 h-1 3.89 kW/m3 4.73 kW/m3

69

65-78_Nicolle:text

2/12/09

11:35 AM

Page 70

Table III A summary of the empirical calculation as applied to 3 agitators in compartment 1 O2 gassing rate per agitator Oxygen solubility in water—Equation [4] O2 solubility in solution—Equation [6] Dissolved O2 concentration in solution—Equation [2] Oxygen transfer rate in C1A—Equation [3] Oxygen transfer rate in C1B—Equation [3] Oxygen transfer rate in C1C—Equation [3] kLa in C1A—equation [1] kLa in C1B—equation [1] kLa in C1C—equation [1] P/V required in C1A [7] P/V required in C1B [7] P/V required in C1C [7]

4141 kg/h 0.2430 kg/m3 0.1786 kg/m3 0.0536 kg/m3 89 kg/h.m3 103 kg/h.m3 108 kg/h.m3 713 h-1 827 h-1 865 h-1 2.28 kW/m3 2.77 kW/m3 2.94 kW/m3

Table IV A summary of the empirical calculation as applied to 8th March 2007 testwork results O2 gassing rate per agitator Weight percent solids Specific gravity solids Specific gravity liquor Volume percent solids Slurry volume—C1A Slurry volume—C1B Oxygen solubility in water—Equation [4] O2 solubility in solution—Equation [6] Dissolved O2 concentration in solution—Equation [2] Oxygen transfer rate in C1A—Equation [3] Oxygen transfer rate in C1B—Equation [3] kLa in C1A—Equation [1] kLa in C1B—Equation [1] P/V required in C1A [7] P/V required in C1B [7] Average P/V measured in C1A Average P/V measured in C1B

36 kg/h 35% wt/wt 2859 kg/m3 1200 kg/m3 18.4% vol./vol. 0.64 m3 0.58 m3 0.2430 kg/m3 0.1786 kg/m3 0.0536 kg/m3 62 kg/h.m3 69 kg/h.m3 497 h-1 552 h-1 1.42 kW/m3 1.63 kW/m3 2.64 kW/m3 2.93 kW/m3

empirical results to develop confidence in the proposed commercial design. The results of the testwork are summarized in Table IV. The same conditions as Table II were used with the exceptions listed below. Most importantly the oxygen demand is lower than the commercial plant. By comparing the average measured P/V from the testwork and that calculated using Equation [7] in Table IV it is possible that during the testwork either the agitator power was higher than the power required for oxygen mass transfer or that too low a value for α (from Equation [2], α = 0.3) had been assumed. It was suspected that the agitator power was higher than the power required for oxygen mass transfer. This suspicion was confirmed by additional tests (summarized in Figure 1 in which the P/V was reduced but no appreciable change in Ni 70


65-78_Nicolle:text

2/12/09

11:35 AM

Page 71

metal recovery was recorded until the P/V dropped below well 1.4 kW/m3. The P/V for compartment 1A, as calculated in Table IV and based on the assumption that α = 0.3, is 1.42 kW/m3. Hence the original assumption that α = 0.3 is validated. The demo plant OTR is 62–69 kg/m 3h while the commercial plant’s design OTR is 89–108 kg/m3h. The commercial design takes into account the potential for an ore with a high sulphur content that will require additional oxygen per unit mass of feed solids to leach. In order to compare the same oxygen inputs the Tati demonstration plant conditions are recalculated in Table V to indicate what the P/V would be if the OTR were in the region of 89–108 kg/m3h. The resulting P/V is between 2.39–2.74kW/m3, which is similar to the commercial plant requirements in Table III. The same conditions as Table IV were used with the exception that the OTR was increased to what the commercial plants design will be. It should be noted that the Tati demonstration plant oxygen transfer rates (OTR) are only a guideline and cannot be scaled up linearly. Design considerations Agitator design Table VI shows the proposed design of the agitator to meet the mass transfer requirements outlined in Table III. A custom 8-bladed Rushton turbine with increased blade height connected to a 186 kW variable speed drive motor was selected. Increasing the blade height is a common means of increasing the power number of a Rushton turbine. The recommended design shaft speed is 92 RPM, which produces a gassed power input of 129 kW (69% of motor load) per agitator and a volume averaged P/V of 2.8 kW/m3 in the compartment. The design shaft speed results Table V Demo plant P/V with OTR in the region of 89–108 kg/m3h 53.3 kg/h 0.2430 kg/m3 0.1786 kg/m3 0.0536 kg/m3 92 kg/h.m3 102 kg/h.m3 739 h-1 819 h-1 2.39 kW/m3 2.74 kW/m3

i% Recovery

O2 gassing rate per agitator Oxygen solubility in water—Equation [4] O2 solubility in solution—Equation [6] Dissolved O2 concentration in solution—Equation [2] Oxygen transfer rate in C1A—Equation [3] Oxygen transfer rate in C1B—Equation [3] kLa in C1A—Equation [1] kLa in C1B—Equation [1] P/V required in C1A [7] P/V required in C1B [7]

Figure 1. Effects of P/V on nickel recovery recorded in the Tati demonstration plant

MIXING SYSTEM DESIGN FOR THE TATI ACTIVOX® AUTOCLAVE

71

65-78_Nicolle:text

2/12/09

11:35 AM

Page 72

Table VI P/V from the proposed design Agitator

P/V required Equation [7]

P/V measured during 8 March 2007 testwork

P/V at 69% of motor capacity

P/V at 85% of motor capacity

Unit

2.28 2.77 2.94

2.64 2.93

2.49 2.89 3.02

3.05 3.54 3.70

kW/m3 kW/m3 kW/m3

Compartment 1 A Compartment 1 B Compartment 2 C

Table VII P/V sensitivity analysis on assumed α values

α

CO2-solution

( ) kg

m3

C

( ) kg

kLa (h-1)

m3

Required

P V

( )

0

0.1822

0

C1A: 49 C1B: 579 C2C: 605

C1A: 1.43 C1B: 1.73 C2C: 1.84

0.1

0.1822

0.0179

C1A: 554 C1B: 643 C2C: 673

C1A: 1.64 C1B: 1.99 C2C: 2.11

0.2

0.1822

0.0357

C1A: 624 C1B: 724 C2C: 757

C1A: 1.91 C1B: 2.33 C2C: 2.47

0.3

0.1822

0.0536

C1A: 713 C1B: 827 C2C: 865

C1A: 2.28 C1B: 2.77 C2C: 2.94

0.4

0.1822

0.0714

C1A: 831 C1B: 965 C2C: 1009

C1A: 2.79 C1B: 3.40 C2C: 3.60

0.5

0.1822

0.0893

C1A: 998 C1B: 1158 C2C: 1211

C1A: 3.55 C1B: 4.32 C2C: 4.58

kW

m3

in an impeller tip speed of 6.5 m/s, which is below the design maximum of 7 m/s, reducing impeller blade erosion. At the maximum motor load (~85%) the volume averaged power per unit volume is 3.4 kW/m3. From Table VI the comparative P/Vs indicate that the proposed 3 agitator first compartment arrangement adequately provides the required P/V range obtained from empirical calculations. Agitator design sensitivity The sensitivity of the power per unit volume to the assumed value of α for the proposed commercial autoclave is summarized in Table VII. Table VII indicates the high sensitivity that the required P/V has to the assumed value of α. For this reason the additional flexibility offered by the proposed agitators is important. Calculations indicate that at the maximum motor load (~85%) α = 0.42 and provides an additional design margin on agitator power. 72


65-78_Nicolle:text

2/12/09

11:35 AM

Page 73

Short-circuiting By reducing the number of compartments from 5 to 4 the potential for slurry particles to short-circuit in the autoclave and exit unreacted is increased. This would lead to lower metal recoveries. A factor that will help minimize the affect of short-circuiting on overall recovery is the small contribution of compartment 5 to the overall recovery. This is supported by data collected during test runs on the pilot plant in which compartment 5 increased the overall recovery of nickel, copper and cobalt by an average of 0.3%, 0.9% and 0.2% respectively. Short circuiting was analysed by performing a CFD analysis and is outlined in the next section. This analysis provided insight into the 3 agitator configuration in sufficient detail to determine if the proposed arrangement will experience short-circuiting. Objectives of CFD modelling A CFD model was used to evaluate the following design objectives: • Determine overall mixing and flow patterns in compartment 1 to determine the effect on process performance • Determine the optimum impeller and baffle configuration that maximizes back mixing in compartment 1 while minimizing swirling flows beneath the impeller that lead to vessel lining wear • Determine impeller rotation direction • Determine if the required slurry particle residence time in compartment 1 is achieved given the large recirculation flow of slurry through the flash cooling and autoclave feed system. CFD simulation results A three dimensional CFD model of the first compartment in the autoclave was constructed to its actual dimensions. Figure 2 shows the model. The model was set up with the following boundary conditions (see Figure 2 for location of each): • Slurry feed inlet—velocity inlet boundary • The overflow from the weir—pressure outlet boundary • The slurry free surface—zero shear (slip boundary) • Vessel walls—no slip boundary. Agitator rotation direction The mixing was analysed with each of the agitators operating at a 92 rpm. Two different cases were studied. 1. All three agitators rotating in the same direction 2. Middle agitator rotating in the opposite direction. Figure 3 shows the velocity profile of the slurry on a horizontal plane 1.863 m from the slurry free surface. The following observations were made: • There is weak interaction between the flow from adjacent agitators and the overall flow pattern was found to be similar for both cases. • The dotted lines represent the net slurry flow from the feed end to the discharge end. It indicates a different flow pattern for the net slurry given the agitator rotation direction. However, it was observed that the net slurry flow path length was approximately the same for both cases. The residence time distribution (RTD) of the slurry explains this quantitatively in a later section. Figure 4 shows that this agitator and baffle configuration produces two distinct flow loops – one the upper half of the vessel and one in the lower half. This is required for suspension of solids and effective top to bottom blending of solids with dissolved oxygen. MIXING SYSTEM DESIGN FOR THE TATI ACTIVOX® AUTOCLAVE

73

65-78_Nicolle:text

2/12/09

11:35 AM

Page 74

Figure 2. Geometric model of autoclave first compartment illustrating major operating conditions

Figure 3. Slurry velocity profile, coloured by velocity magnitude (m/s) on a horizontal plane across impellers. Lines represent the net slurry flow direction

Figure 4. Slurry velocity profile, colored by velocity magnitude (m/s) on a vertical plane across impellers. Lines represent the net slurry flow direction

Figure 5 and Figure 6 show the velocity profile on a vertical plane. Similar flow patterns were observed for both cases (same rotation and middle agitator—reverse rotation). It was also observed that the slurry flow to the discharge dip-pipes is predominantly from the middle agitator. The flash discharge pipes are positioned at the location shown to allow a compartment wall to be inserted between the second and third agitator. A previous CFD analysis showed this as the optimum location for the dip-pipes for an autoclave with two agitators in C1 and therefore provides the option to revert back to a two agitator compartment if desired. 74


65-78_Nicolle:text

2/12/09

11:35 AM

Page 75

Figure 5. Velocity profile, on a vertical plane cutting through the left-side (looking downstream) feed and discharge pipes illustrating the slurry flow pattern

Figure 6. Velocity profile, coloured by velocity magnitude (m/s), on a vertical plane cutting through the right-side (looking downstream) feed and discharge pipes illustrating the slurry flow pattern

Swirling flow beneath the impeller is unavoidable as angular momentum is imparted to the slurry by the rotating impeller. Larger solids caught up in this swirling vortex can cause erosion of the vessel lining underneath the impeller. Therefore, it is important to minimize the magnitude of such swirling velocities. Extending the baffles closer to the bottom can minimize the swirling velocity significantly. The final baffle configuration is shown in Figure 2. Note that extending the baffles right to the bottom of the vessel, breaks the single swirling vortex into 2 or more smaller vortices on either side of the baffle. These smaller vortices can actually rotate at a higher velocity than the single swirling vortex beneath the impeller accelerating the erosion rate of the vessel lining over a smaller, localized region. Figure 7 shows the velocity profile on a horizontal plane at an off-bottom distance of 0.3 m illustrating swirling velocity magnitudes on a plane underneath the impellers. A maximum velocity magnitude of 2 m/s was observed for both cases. It must be noted that the velocities should not be minimized to an extent that can allow solids settling, and therefore movement of solids along the vessel bottom is required to avoid solids packing. For the Tati Activox® process, solids suspension does not appear to be a concern due to the very small particle sizes of solids in the slurry feed (P80=10 μm). The general flow field results indicate that the slurry flow inside the first compartment is well mixed and no low velocity regions are observed that could otherwise allow the formation of hot spots. The mixing is quantitatively described by the turnover rate of slurry volume for the agitator discharge flow. Turnover rates greater than 3 are required for medium to violent agitation (Bowen 1985). The CFD model predicts a flow rate of 20000 m3/h is generated by a single impeller corresponding to 7.2 turnovers per minute. MIXING SYSTEM DESIGN FOR THE TATI ACTIVOX® AUTOCLAVE

75

65-78_Nicolle:text

2/12/09

11:35 AM

Page 76

Mass fraction

Figure 7. Velocity profile, coloured by velocity magnitude (m/s), on a horizontal plane at 0.3 m off-bottom distance and cutting along the compartment illustrating the swirl velocity magnitudes

Time (min)

Figure 8. Residence time distribution comparison

Residence time distribution (RTD) The residence time distribution of the slurry in the compartment can also be used to evaluate the mixing performance and its subsequent effect on recovery. A tracer analysis was performed to determine the residence time distribution. Mass-less tracers, which follow the slurry flow closely, were injected into the compartment through the feed inlets and tracked to determine their residence time. The slurry exiting through the discharge pipe is recycled back to the front of the compartment after flash cooling. Thus, fresh slurry fed to the first compartment will flow through a number of recycles before it overflows to the second compartment. Figure 8 shows the RTD results for the tracer analysis, single continuous stirred reactor (CSTR) and for 2 CSTRs in series (calculated from Davis 2002). The RTD for two CSTRs in series represents the autoclave with the additional weir wall in place. The CFD model predicts an RTD similar to what would be expected from a single CSTR. When comparing the CFD model and the 2 CSTRs RTD the potential for short circuiting is likely. 76


65-78_Nicolle:text

2/12/09

11:35 AM

Page 77

The impact of the increased material short circuiting on the overall autoclave recoveries still needs to be quantified; however, it is expected to be negligible based on testwork done in the Tati demonstration plant and similar modifications done on other operating autoclaves. Conclusion Based on all of the calculations and CFD analysis, the following conclusions were reached: • The general flow field results indicate that the flow inside compartment 1 is well mixed and no distinct low velocity regions are seen that may form ‘hot spots’. • The impact of increased short circuiting on the overall autoclave recoveries still needs to be quantified, however; it is expected to be negligible based on testwork done in the Tati demonstration plant and similar modifications made on other operating autoclaves. • Due to the small particle sizes of the solids in the incoming slurry feed (P80=10 μm), solids suspension is not a concern, and it is expected that solids will be uniformly suspended to the full liquid height of the autoclave. • Very good mixing between solids and dissolved oxygen is expected, leading to a uniform leach reaction extent and process temperature throughout the vessel volume. • A baffle off-bottom clearance of 0.14 D was found to be optimum to minimize the swirling velocity near the tank bottom. • There is weak interaction between the discharge flow from adjacent agitators for the given operating condition and the overall flow patterns are uninfluenced by the relative rotational direction of adjacent agitators. • Different flow patterns for the net slurry flow from the feed end to the discharge end was observed for the two cases: (1) all agitators—same rotation direction and, (2) middle agitator—reverse rotation. However, it was observed that the net slurry flow path length is approximately the same for both the cases. Acknowledgements The authors acknowledge that this paper was originally published as part of the proceedings of Hydrometallurgy 2008—6th International Symposium honoring Robert Shoemaker, August 17–20, 2008, Phoenix, AZ. The authors would like to thank the management of Norilsk Nickel Africa, Norilsk Process Technology and Hatch for their encouragement and permission to publish this paper. References TROMANS, D. Temperature and Pressure Dependant Solubility of Oxygen in Water: A Thermodynamic Analysis. Hydrometallurgy, vol. 48, 1998. pp. 324–342. TROMANS, D. Oxygen Solubility Modeling in Inorganic Solutions: Concentration, Temperature and Pressure Effects. Hydrometallurgy, vol. 50, 1998. pp. 279–296. PIETERSE, H. Oxidation Autoclave Agitation Review. Pressure Hydrometallurgy 2004, 34th Annual Hydrometallurgy Meeting, Banff, Alberta, Canada. 2004. BOWEN, R.L. Agitation Intensity: Key to Scaling Up Flow-Sensitive Liquid Systems. Chemical Engineering, March 18, 1985. DAVIS, M.E. and DAVIS, R.J. Fundamentals of Chemical Reaction Engineering. McgrawHill Chemical Engineering Series, International Edition, July 2002. MIXING SYSTEM DESIGN FOR THE TATI ACTIVOX® AUTOCLAVE

77

65-78_Nicolle:text

2/12/09

11:35 AM

Page 78

Marc Nicolle Process Engineer, Hatch Africa (Pty) Ltd, South Africa Marc is a process engineer with over 3 years’ experience in the design, engineering and construction of mineral processing and hydrometallurgical processes. He is developing specialist technical expertise in ultra-fine grinding, countercurrent decantation, highpressure acid leach, slurry cooling towers and PGM flotation. In recent years Marc has been part of the process design and project development for various hydrometallurgical plants around the world. These include the Norilsk Nickel’s Tati Activox® Project, Phelps Dodge’s Tenke Fungurume feasibility study, First Quantum Minerals Kolwezi tailings project, Teal Mining and Exploration’s Kalumines project, African Rainbow Minerals and Norilsk Nickel’s Nkomati site options study and Anglo Platinum’s DC/DA acid plant at Waterval Smelter.

78


79-92_Franceschini:text

2/14/09

11:33 AM

Page 79

FRANCESCHINI, M and WOLOSHYN, J. Stabilization of supersonic vent gas from autoclave pressure oxidation. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Stabilization of supersonic vent gas from autoclave pressure oxidation M. FRANCESCHINI* and J. WOLOSHYN* *Hatch (Pty) Ltd, Ontario, Canada The ability to control the velocity distribution of a fluid flow is a fundamental problem in many industrial applications of fluid engineering. This problem is compounded in hydrometallurgy because the process fluid is typically two phase, as well as corrosive and abrasive in nature. This paper investigates the use of a perforated plate to achieve a stable and uniform flow field downstream of a shock wave in a blast shroud entering a quench vessel from pressure oxidation autoclave. Fluid mechanics analysis has been used to provide insight into the proper diffuser geometry in order to produce an adequate flow field both downstream and upstream of the plate. The addition of a perforated plate is found to reduce the maximum velocity entering the quench vessel to one-fifth of the original velocity, mitigating the risk of wear on the vessel bottom and excessive splashing of the water pool. The analysis also indicates that both the open area and the number of holes in the plate are important parameters in stabilizing upstream gas flow and minimizing flow separation downstream of the shock. The mechanical design of the shroud and perforated plate addresses concerns around thermal expansion, natural vibration frequency, strength limitations of nitride-bonded silicon carbide, and the erosive environment created by supersonic flow with entrained particulate. Prudent mechanical design is required to ensure a durable long lasting perforated plate, which will reduce wear on pressure letdown equipment. Placing the perforated plate in a metal spacer ring allows it to float between top and bottom supporting flanges. Strict control of fabrication tolerance is required to maintain clearance between the plate, spacer ring, and other ceramic components, limiting stresses induced by thermal expansion to 0.5 MPa. Soft packing and gasket materials, which are susceptible to erosion, are isolated from the flow by a tongue-and-groove joint between the ceramic shroud liner and perforated plate.

Introduction This paper presents the results of computational fluid dynamics (CFD) and finite element analyses (FEA) in the design of a novel blast shroud configuration using a perforated plate, as it applies to pressure letdown of vent gas from a pressure oxidation process. The blast shroud STABILIZATION OF SUPERSONIC VENT GAS FROM AUTOCLAVE PRESSURE OXIDATION

79


2/14/09

11:33 AM

Page 80

is designed to let down the pressure of a vent gas discharging from an autoclave at 4100 kPa(g) into a quench vessel at 14.7 kPa(g). Figure 1 shows an example of a typical autoclave circuit, where slurry is processed in the autoclave at elevated temperature and pressure using steam and oxygen gas. High-pressure vent gas composed mainly of steam and oxygen, but also containing slurry particles, is vented from the autoclave. The quench vessel condenses the steam and removes slurry carryover from the vent gas before final cleaning in a cyclonic separator, which then vents to the atmosphere. A shock wave is expected within the blast shroud because of the large pressure drop. Therefore, the design must ensure that the shock wave is stable in order to prevent the risk of flow-induced vibrations on the blast shroud structure. Moreover, the design must also ensure that the velocities entering the quench vessel are minimized to prevent wear on the bottom of the vessel or excessive splashing of the water pool. The addition of the perforated plate is proposed to address both the shock stability and improve flow uniformity entering the quench vessel. CFD analysis result and discussion The geometry of the blast shroud is shown in Figure 2. The gas, which is primarily composed of steam, passes through a conical upper section before reaching the perforated plate. The geometry of the conical section is defined based on an internal angle of α and a length of L1. The second part of the blast shroud, downstream of the cone and perforated plate, is a cylinder with a diameter of D and a length of L2. The boundary conditions used for the CFD model of the blast shroud are shown in Figure 2. The boundary conditions are set as the mass flow at the inlet and a specified gauge pressure at the outlet. The model inlet represents the throat of the diffuser and the flow at this location is STACK TO ATM

SAFETY RELIEF

CYCLONE SEPARATOR

AUTOCLAVE FEED SLURRY

VENT QUENCH VESSEL

FLASH TANK

AUTOCLAVE FEED TANK

TO THICKENING

AUTOCLAVE OXYGEN

OXYGEN BLOWBACK VESSEL HP STEAM

FLASH DISCHARGE PUMP BOX

STEAM BLOWBACK VESSEL

Figure 1. Example of an autoclave circuit

80



2/14/09

11:33 AM

Page 81

Figure 2. Geometry of the blast shroud and boundary conditions for the CFD model

Figure 3. Velocity distribution inside the blast shroud without perforated plate

choked (i.e. Ma = 1). Due to model limitations phase change has been neglected. Phase change is an important phenomenon near the shock wave where the temperature decreases dramatically. Although the temperature is expected to drop below 0°C in a small region just prior to the shock wave, the impact of neglecting phase change is considered minimal because the flow has an extremely low residence time within the ‘cold zone’. The overall flow field and trends predicted by the CFD model are expected to be valid. To understand the effect of adding a perforated plate to the blast shroud, the flow field is first modelled without a plate present. Figure 3 shows contours of gas velocity inside the blast shroud without a perforated plate. STABILIZATION OF SUPERSONIC VENT GAS FROM AUTOCLAVE PRESSURE OXIDATION

81


2/14/09

11:33 AM

Page 82

Figure 4. Initial perforated plate geometry investigated using CFD analysis. The geometry is based on a 50% open area and 1” diameter holes

The supersonic flow accelerates with increasing cross-sectional area. As the gas velocity increases, the pressure decreases until a shock wave occurs. As expected1, the flow separates immediately downstream of the shock wave and a central jet of high speed flow persists. The result is a complex pattern of shocks within a jet of mixed supersonic and subsonic flow. Furthermore, the position of this jet was found to be unstable. The flow separation downstream of the initial shock causes the jet to move toward the wall of the shroud. This behaviour sets up an unstable separation bubble, causing the jet to oscillate with time. The instability in the flow field is a concern because it may pose a risk of flow-induced vibrations in the structure of the shroud. This would occur if the frequency of the oscillations were to coincide with any of the natural frequencies of the structure. Figure 4 shows the initial perforated plate geometry that was investigated. The use of a perforated plate for this application is novel, so no standard practices are available for selection of hole size or open area. A 50% open area and 1” holes were selected as a starting point based on Hatch experience in other fields. Figure 5 shows pressure contours and a plot of the pressure and velocity along the centerline of the shroud both with and without the perforated plate. The presence of the plate dramatically affects the upstream and downstream flow fields. A pressure drop of approximately 13.5 in-H2O (3360 Pa) across the plate, which is very low relative to the total pressure let down, effectively creates a higher and more uniform pressure field upstream. The shock wave is stabilized and the flow separation and secondary shocks are prevented. The velocity distribution and flow field inside the shroud are shown in Figure 6. Comparing this result to that of Figure 5, a significant improvement in the uniformity of the velocity field across the shroud is seen. The two key improvements observed are: (1) the elimination of the unstable, high speed jet observed in the previous case, and (2) a uniform or plug flow profile exiting the shroud. Based on these improvements, the risk of flow induced vibrations and of wear and pool disturbance in the quench vessel are mitigated. Figure 7 shows the effect of the perforated plate on the exit velocity distribution from the blast shroud and entering the quench vessel. 82



2/14/09

11:33 AM

Page 83

Figure 5. Pressure and velocity variation through the blast shroud (a) without the perforated plate, and (b) with the perforated plate

Figure 6. Velocity distribution and flow field inside the blast shroud with a perforated plate

STABILIZATION OF SUPERSONIC VENT GAS FROM AUTOCLAVE PRESSURE OXIDATION

83

2/14/09

11:33 AM

Page 84

Velocity (m/s)


Distance from shroud centreline (m)

Figure 7. Velocity distribution across the shroud exit plane (i.e. entering the quench vessel)

Without a perforated plate, the separated flow upstream affects the exit velocity profile, which is non-uniform and unstable. The maximum velocity entering the quench vessel is in the range of 110 m/s to 285 m/s. By contrast, with the perforated plate in use, the velocity profile is uniform and the maximum velocity entering the quench vessel is reduced to 62 m/s. This velocity is considered acceptable for the current application. Effect of perforated plate geometry The geometry of the perforated plate has a significant effect on the flow stabilization inside the shroud. CFD modelling was used to test another design for the perforated plate to determine the effect of the opening density (as defined by Equation [1]).

[1]

With the open area maintained at 50%, the hole diameter was increased from 1” to 2”, i.e. the hole density was decreased from 987 m-2 to 247 m-2. Figure 8 compares the velocity distribution inside the blast shroud with this new perforated plate geometry to the previous two cases. With the reduced hole density the pressure drop across the plate is reduced from 13.5 in-H20 (3360 Pa) to 11.5 in-H20 (2870 Pa) and a mix of sub and supersonic flow exists upstream of the perforated plate. Comparing this result to the velocity distributions without any plate, the downstream flow has been somewhat stabilized; however, the upstream jet still exists. This penetrating jet creates normal shocks upstream of the plate. Further work is underway to determine the critical pressure drop at which the upstream flow field is fully stabilized and the best combination of percent open area and hole density to achieve this pressure drop. 84



2/14/09

11:34 AM

Page 85

Figure 8. The velocity distribution inside the blast shroud with (a) no perforated plate, (b) a perforated plate with a hole density of 987 m-2, (c) a perforated plate with a hole density of 247 m-2

FEA analysis result and discussion Geometry The perforated plate geometry is driven by the process opening of the blast shroud and the requirement for 50% open area, as defined by the CFD analysis. For practical reasons the geometry is modified slightly from that shown in Figure 4; however, the hole diameter and open area are maintained. The outside diameter of the perforated plate is oversized to allow for a clamping annulus, as depicted in Figure 9. To obtain the required flow characteristics, the disk has 187 holes arranged in triangular pitch pattern, as depicted in Figure 10.

Figure 9. Mechanical assembly of ceramic blast shroud and perforated plate. Gaskets and PTFE are shown in their hot, compressed form


85


2/14/09

11:34 AM

Page 86

Figure 10. Plan view of perforated plate showing hole pattern and pitch details. A total of 187 holes combine to create 91,793 mm2 of open area. Radii on the circumferential slot and leading/trailing edges of holes are visible

Table I Physical properties of nitride bonded silicon carbide (NBSC) provided by Blasch Precision Ceramics, Albany, NY, USA Young’s modulus Poisson’s ration Density Thermal expansion Thermal conductivity Specific heat Tensile strength

80 0.15 2800 4.68E-06 14 940 30–33

GPa kg/m3 /°C W/m2K J/Kg°C MPa

The layout is critical to provide sufficient webbing between holes to ensure structural integrity. Stress concentrations are minimized by smoothing sharp junctions with 6 mm radii on concave corners and 1 mm on convex edges, specifically at the leading and trailing edges of holes. The outer annulus has been slotted circumferentially to accommodate overlapping of the ceramic blast shroud and perforated plate to protect the PTFE packing and gasket. This is required to prevent the abrasive process fluid from eroding the soft packing and gasket material, thus compromising the pressure barrier. The diverging ceramic shroud and perforated plate are constructed from nitride bonded silicon carbide (NBSC). This material was chosen for its hardness and resistance to erosion. Physical properties of NBSC required for FEA analysis are listed in Table I. 86



2/14/09

11:34 AM

Page 87

Figure 11. Thermal profile through the perforated plate

Table II Summary of FEA results. Three cases are considered with different boundary constraints. Case 1 is free to expand radially and axially, case 2 is constrained in the radial direction, and case 3 is constrained axially Case

1

2

3

Radially

Free

Fixed

Free

Axially

Free

Free

Fixed

Front

232°C, 500 W/m2C

232°C, 500 W/m2C

232°C, 500 W/m2C

Back

230°C, 500 W/m2C

230°C, 500 W/m2C

230°C, 500 W/m2C

Boundary conditions

Heat load

Rim

150°C, 2

W/m2C

150°C, 2

W/m2C

150°C, 2 W/m2C

Stress intensity

0.521 MPa Hole edge—top surface

224 MPa Hole edge—inside

308 MPa Outside edge of lap joint

Max principal

0.358 MPa Outer rim

264 MPa Lap joint fillet

139 MPa Outside edge of lap joint

Min principal

-0.518 MPa Hole edge—top surface

-223 MPa Hole edge—top surface

-347 MPa Bottom of lap joint valley

0.64 0.05

none—fixed 0.31

0.68 none—fixed

Deflection/growth Radial (mm) Axial (mm)

Mechanical design It is critical that thermal expansion of the blast shroud and perforated plate be managed to mitigate the risk of overstressing the NBSC resulting in cracking, and ultimately failure. To simulate operating conditions the temperature profile in Figure 11 is generated by applying temperature and convection coefficients to the top, bottom and outer plate surfaces. Temperatures for the top and bottom faces are from the CFD analysis. The temperature and convection coefficient applied to the outer edge of the plate is approximated from past FEA experience. Thermal growth in the axial and radial directions is reported in Table II, along with maximum stresses due to the temperature gradient through the perforated plate. As expected, the stresses in the unconstrained case are orders of magnitude lower. Generally, stresses are STABILIZATION OF SUPERSONIC VENT GAS FROM AUTOCLAVE PRESSURE OXIDATION

87


2/14/09

11:34 AM

Page 88

developed in the plate when the hot centre portion expands and is constrained by the cooler outer clamping annulus. As shown in Figures 12 through 14, this generates compressive stresses in the centre portion and tensile stresses in the outer portion of the plate. These stresses are intensified in the thin wall sections between holes and the slotted circumferential groove. There is also a variation in stress between the hot and cold faces due to the 2°C temperature difference between them. Ceramic materials tend to fail in a brittle manner, which occurs abruptly with no yielding as the rupture strength is exceeded. The tensile strength of NBSC is given by the manufacturer as 30–33 MPa2. In the unconstrained state (case 1), peak stress levels in the plate are 0.36 MPa in tension and -0.52 MPa in compression. These values are less than 2% of the rupture strength. However, if growth of the perforated plate is restrained as seen in cases 2 and 3, stress levels rise dramatically and will certainly exceed the tensile limit causing failure. It is of absolute importance that the perforated plate be allowed unrestrained thermal growth, while being held in place to minimize vibration. Radial expansion is permitted by centring the ceramic perforated plate within the metal spacer ring. All mechanical loads from piping, bolting and pressure will be transferred through the spacer ring instead of the fragile ceramic plate. The spacer ring inside diameter is slightly larger than the perforated plate

Figure 12. Stress intensity plots in true scale from Ansys Workbench

88



2/14/09

11:34 AM

Page 89

Figure 13. Maximum principal stress plots in true scale from Ansys Workbench

outside diameter, thus allowing ample space for radial thermal expansion. Axial thermal growth is taken up by compressing the PTFE packing on the top and bottom surfaces of the clamping annulus. In the cold state, the PTFE will be uncompressed, but not loose. As the assembly is heated, the metal and ceramic components will expand compressing the PTFE up to 20%, while holding the perforated plate in position. Also, the PTFE will cushion the perforated plate from impacting other metal or ceramic components. Natural frequency There is potential for vibration to cause cracking and failure of the ceramic plate. The first and second natural frequencies are calculated by axially restraining the plate around the outside top edge. The first and second natural frequencies occur at 442 and 1129 Hz respectively. The undamped deformation pattern is shown in Figure 15. The first natural frequency is quite high (442 Hz or 26520 RPM) making it unlikely for process equipment such as pumps to vibrate at this frequency. However, it is difficult to predict all sources of vibration and it is possible for process valves to create excitation frequencies at this level. If vibration causes the failure of the perforated plate it is possible to increase or decrease the natural frequency by reducing or increasing the thickness of the perforated plate. STABILIZATION OF SUPERSONIC VENT GAS FROM AUTOCLAVE PRESSURE OXIDATION

89


2/14/09

11:34 AM

Page 90

Figure 14. Minimum principal stress plots in true scale from Ansys Workbench

Design benefits and conclusions The CFD analysis has shown that the use of a perforated plate in the blast shroud improves the flow field by eliminating the unstable, high-speed jet downstream of the shockwave and by creating a uniform flow profile exiting the shroud into the quench vessel. The reduced entry velocity eliminates the need for replaceable wear components such as impingement blocks and ceramic tiles. The absence of a high-speed jet eliminates the need for a large liquid pool for energy dissipation, while protecting the vessel lining and shell from erosion. As a result, pressure let down vessels can be made more compact and cost-effective without sacrificing availability or performance. The stability of the shockwave will mitigate the risk of flow-inducted vibration and fatigue in metallic the blast shroud, nozzle and vessels components. The ability to control large pressure drops has further economic potential in that multistage letdown might be achieved in fewer stages, reducing the required number of vessels and capital cost. The analysis shows that the performance of the perforated plate is related to the pressure drop across it. The hole density and percent open area are two key design parameters that can influence the pressure drop. Two bounding cases were studied, but further investigation is required to determine the optimal design case. 90



2/14/09

11:34 AM

Page 91

Figure 15. Deformation pattern at first and second natural frequencies

Thermal stresses can be minimized through generous filleting and contouring of corners, and most importantly, by allowing unrestricted thermal growth in both radial and axial directions. The spacer ring is designed to carry all mechanical loads and provide enough support to prevent chatter, thus ensuring the longevity of the ceramic perforated plate. Acknowledgements The authors would like to acknowledge Dave Warnica, Hatch Ltd., for suggesting the use of a perforated plate flow diffuser for this application. References 1. PAPAMOSCHOU, D. and JOHNSON, A. Unsteady Phenomena in Supersonic Nozzle Flow Separation, 36th AIAA Fluid Dynamics Conference and Exhibit, June 2006, San Francisco, CA. 2. CONNORS, T. Blasch Precision Ceramics, September 2007, Albany, NY, USA.


91


2/14/09

11:34 AM

Page 92

Michael Franceschini Mechanical Engineering Autoclave Technology Non-Ferrous, Hatch (Pty) Ltd, Ontario, Canada Michael is experienced in mechanical engineering in the development of high pressure acid/oxygen leach process plants and in pressure vessel design and knowledge of the ASME Boiler and Pressure Vessel Code Section VIII Division I & II and the associated ASME/ANSI/ASTM Standards. He is proficient in the use of calculation spreadsheets to perform detailed mechanical equipment design. He has worked with exotic metals such as titanium, duplex and super duplex stainless steels, tantalum, and nickel alloys. His expertise in various other computer programs including Codeware Compress, AFT Arrow and Fathom. Michael has the ability to conduct finite element analysis (FEA) using Ansys Workbench, Design Modeller and ICEM. He has completed Div II thermal, structural, modal and bucking analysis of refractory lined pressure vessels, large bore ducting and various vessel internals. • Barrick Gold Corporation, Pueblo Viejo, Dominican Republic – Detailed Engineering— responsibilities to date include pump sizing and the preparation of a pump specification for budget quotation, as well as finite element analysis of pressure equipment, ducting and vessel internals. • Barrick Gold U.S. INC., Donlin Creek, Alaska, USA, Bankable Feasibility Study— responsibilities included sizing of mechanical equipment for pressure oxidation circuit using Codeware Compress. He assembled specifications and drawing for tender packages, and completed vendor bid evaluations. He completed bankable capital cost estimate for autoclave area. • Metals Enterprise, Moa Bay, Cuba, 16K Expansions—modelled mechanical equipment using Codeware Compress and checked vendor calculation to ensure code compliance. • Sargold Resource Corporation, Sardinia, Italy, Class Four Estimate—carried out equipment sizing for a pressure oxidation autoclave circuit and generated factored capital cost and operating cost estimates

Jennifer Woloshyn Specialist in the application of computational fluid dynamics, Hatch (Pty) Ltd, Ontario, Canada Jennifer received her MA Sc in the area of fluid in 2004 from the University of Waterloo in Canada. She has been working at Hatch since 2004, specializing in the application of computational/fluid dynamics (CFD) modelling techniques to solve problems for which traditional analysis trials are not sufficient. Jennifer contributed to the design of a variety of purposes vessels as well as capture hoods and ventilation systems.

92


93-100_F-Kankeu:text

2/14/09

11:41 AM

Page 93

MULABA-BAFUBIANDI, FOSSO-KANKEU, E., and MAMBA, B.B. Indigenous microorganism strains as bio-extractants of Ca, Fe, and Mg from metallurgical and mine drainages. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Indigenous microorganism strains as bio-extractants of Ca, Fe, and Mg from metallurgical and mine drainages A.F. MULABA-BAFUBIANDI*, E. FOSSO-KANKEU*, and B.B. MAMBA† *Department of Extraction Metallurgy, Faculty of Engineering and the Built Environment, University of Johannesburg, South Africa †Department of Chemical Technology, Faculty of Science, University of Johannesburg, South Africa Attempts to remediate the high concentration of calcium, iron and magnesium in surface waters from metallurgical areas by experimenting at laboratory scale on the removal of these metals in synthetic solutions (30 and 50 ppm), using indigenous strains of Shewanella sp, Bacillus subtilis sp and Brevundimonas sp, revealed variable abilities of these microorganisms in the removal process. Bacillus subtilis sp and Shewanella sp absorbed the higher amount of each of the three metals from solution, and calcium was the metal most easily removed. Metals removal from solution decreased when their concentrations were at 50 ppm. It was found that when metals were combined in the same solution they contribute to inhibit microorganisms, change the microbial affinity for metal and affect the removal efficiency. Depending on the metal, there was a tendency for microorganisms to release the absorbed metal into the solution after a certain time, likely due to an efflux transport system previously demonstrated by some authors. Use of non-living biomass did not improve the removal efficiency. Keywords: indigenous strains, biosorption, bioaccumulation, living biomass, non-living biomass, metal removal

Introduction For decades concerned companies and local authorities have demonstrated interest in managing water from metallurgical and mining areas. A general tendency is to shift from the ineffective and costly physico-chemical techniques to the biological techniques found to be cheap and eco-friendly (Alluri et al., 2007; Cohen, 2006; Kefala et al., 1999). Although some of the bench-scale experiments have been successful, the challenge remains to implement the technique at industrial level. This results from the fact that the concept is not entirely understood and therefore requires more effort in improving the removal process on operational level (Wang and Chen, 2006). INDIGENOUS MICROORGANISM STRAINS AS BIO-EXTRACTANTS

93


2/14/09

11:41 AM

Page 94

The presence of residual metals in effluent water represents primarily a substantial loss in revenue for metallurgical companies. Secondly, it could be the cause of legal liability vis a vis environmental safety as these metals are potential pollutants of the water system. According to the South African National Standard (2005) an excess of calcium and iron in water could cause aesthetic or operational problems while excess of magnesium in water could cause esthetic and health problems. Microorganisms possess inherent abilities suitable for the removal of metals from solutions (Nies, 1999; Langley and Beveridge, 1999; Beveridge and Murray, 1976). These abilities have been identified as passive or active for accumulation and biosorption respectively (Brandl and Faramarzi, 2006). However, there are always challenges that contribute to undermine the benefit of this process: the high concentration of metals could be toxic for microorganisms and acidic conditions are unfavourable for the development of most of microorganisms which are neutrophiles (Kim et al., 2006; Sandrin and Maier, 2003). To overcome these challenges, one needs to use indigenous microorganisms adapted to conditions in situ, or alternatively use either genetically modified microorganisms or non-living biomass that could cope with existing conditions. Bacillus strains have been widely reported in literature to be effective in the removal of metals (Pb, Cd, Cu, Ni, Co, Mn, Cr, Zn) from waste waters (Kim et al., 2007; Srinath et al., 2003; Philip and Venkobachr, 2001), but other strains (Brevundimonas and Shewanella) also identified during this study at mining sites, have not been tested for the treatment of metallurgical and mine drainages. In this study, living and non-living biomass of all these strains will be tested for their effectiveness to clean out Ca, Mg and Fe predominant in surface water around mining areas in Nigel. Methodology Isolation and identification of microorganisms Water and soil samples collected in sterile glass bottles around mining areas in Nigel were preserved at 4°C during transportation to the laboratory. Soil samples were suspended in sterile distilled water vortexed, and the supernatant as well as the water samples diluted in sterile phosphate buffer prior to inoculation in chromogenic media for coliforms and Escherichia coli, and in nutrient agar. After overnight incubation of cultured media, colonies of microorganisms were isolated from the plate and subcultured in fresh media. Unknown microorganisms were subsequently identified by gene sequencing at Inqaba Biotechnical Industries (Pty) Ltd—South Africa. Preparation of synthetic solutions Synthetic metal sulphates obtained in powder or crystal form were weighed and diluted in sterile distilled water to make a stock solution of 1 000 ppm. From the stock solution various volumes were removed into the final solution to obtain concentrations between 30 and 50 ppm. Microorganisms growth and biomass preparation Microorganisms were inoculated into the nutrient broth (‘Lab-Lemco’ powder 0.1%, yeast extract 0.2% and NaCl 0.5%), incubated at 37ºC in the incubator with shaker (150 rpm) for twenty-four hours. The culture was centrifuged for 15 minutes at a speed of 8 000 rpm, the supernatant was discarded and the pellet washed several times with sterile distilled water then suspended in a sterile flask. Concentrated cells were lyophilized and autoclaved at 121ºC for 15 minutes to prepare non-living biomass. 94



2/14/09

11:41 AM

Page 95

Metal removal The ability of microorganisms to remove metal from solution depends on two mechanisms called biosorption and bioaccumulation, which are passive or active respectively: Biosorption is described as a chemical interaction between an anionic group (amino acids, hydroxyl, phosphate, etc.) on the bacterial cell membrane and the positively charged metal in solution. There is no net energy required during this process. (Figure 1.) Bioaccumulation is a reaction whereby the metal is sequestered through the bacterial membrane into the cytoplasm of the cell; during this process microorganisms use energy (ATP hydrolysis) to catalyse the reaction. This process could also contribute to the supply of cofactor for enzymatic reactions; however, an excess of metal could be excreted from the cell by an efflux transport system (Nies, 1999). (Figure 2.) For the experimentation of metal removal in this study, synthetic metal solution was mixed with microorganisms (100 mg wet cell) in a 250 ml Erlenmeyer flask, and sterile distilled water added up to a final volume of 100 ml. Microorganisms were exposed to 30 ppm and 50

Figure 1. Adsorption of copper on the cell surface of bacteria (from Kim et al., 2007)

Figure 2. Active transport of arsenate into bacterial cytoplasm (from Nies, 1999)

INDIGENOUS MICROORGANISM STRAINS AS BIO-EXTRACTANTS

95


2/14/09

11:41 AM

Page 96

ppm concentrations of each metal separately as well as with a mixture of metal ions. The mixture was then incubated at 37ºC in the incubator with shaker (150 rpm) and 5 ml of sample solution was taken after one, two and twenty-four hours, as previous authors obtained maximum metal removal at these times (Kefala et al., 1999; Kim et al., 2007). The solution was then centrifuged at 8 000 rpm for five minutes and the supernatant collected for quantification of the amount of metal. Metal quantification and experimental procedure Determination of the amount of metal in solution was done using inductively coupled plasma optical emission spectrometry (ICP-OES). All the experiments were done in triplicate with a control; the difference between the replicate was less than 10%. The average value of the triplicate was considered when drawing the graph. Strains of Bacillus subtilis, Shewanella sp and Brevundimonas sp were identified in soil and water samples from mining areas by gene sequencing. These strains were tested for their abilities to remove Fe, Ca and Mg from synthetic solution containing 30 and 50 ppm of the above metal. Metal tolerance To determine whether microorganisms were alive during the whole removal experiment, at each time sample was collected for analysis; one ml was simultaneously inoculated in phosphate buffer, diluted five times (10-1, 10-2, 10-3, 10-4 and 10-5) then plated on agar media (chromogenic media for coliforms and Escherichia coli) and incubated at 37ºC overnight. Microorganism survival was then assessed as the occurrence and number of typical colonies in the agar media. Results and discussions Removal of individual metal

% metal removed

The ability of Bacillus subtilis to remove metal ions varied depending on the metal specie; there was a greater affinity for calcium than for other metals (Figure 3). In fact calcium removal was around 14% while iron and magnesium removal was less than 10%. It was also noticed that the removal efficiency was lower at 50 ppm concentration of metal after 24 hours.

Figure 3. Separate removal of Fe, Ca and Mg by Bacillus subtilis sp

96



2/14/09

11:41 AM

Page 97

% metal removed

Shewanella sp also showed greater affinity for Ca but there was a tendency to release the metal into solution from the second hour of the experiment (Figure 4). A higher removal rate was recorded at 30 ppm concentration of metal. Among all the bacteria, Brevundimonas sp has the lower performance as far as metal removal is concerned. However, it also has greater affinity for calcium and the highest removal was approximately 9% of Ca at 50 ppm concentration (Figure 5). It is observed that metal removal in this experiment is greatly influenced by the affinity of the microorganism for the metal, as during the first hour (contact time) metal binding rate to the microbial cell wall differed from one metal to the other. According to work done by Kefala et al. (1999) the composition of the cell wall (reactive groups) of microorganism play an important role in determining the attraction of the metal. However, other factors such as surface availability on the cell wall could affect the removal efficiency, knowing that saturation of the cell is quickly reached in solution containing high concentration of metal, decreasing the ratio of metal removed to total metal in solution. This explains why generally after the first hour the percentage of metal removed is higher in solution with 30 ppm of metal.

% metal removed

Figure 4. Separate removal of Fe, Ca and Mg by Shewanella sp

Figure 5. Separate removal of Fe, Ca and Mg by Brevundimonas sp

INDIGENOUS MICROORGANISM STRAINS AS BIO-EXTRACTANTS

97


2/14/09

11:41 AM

Page 98

During the second and twenty-fourth hours it is likely that a bioaccumulation mechanism dominates over biosorption, since the second mechanism reaches the maximum after 5 minutes contact time (Sadowski et al., 1991). The removal rate could therefore increase, meaning that the metal is progressively transported into the cell or decrease when there is efflux transport or rupture of bounds between metal and reactive groups on the microorganism cell wall. The low rate of metal removed during this study is partly due to the fact that a smaller mass of cells was used compared to other studies. It is reported (Schiewer and Volesky, 1995a; Kefala et al., 1999; Wang, 2002a) that higher cell concentration could increase metal removal. Removal of mixed metals To simulate the conditions in mine areas, living biomass was exposed to solution containing all three metals, each present at concentration of 30 or 50 ppm, depending on the set of experiment. Results (not shown) revealed that the removal efficiency of the microorganisms decreased by almost half and the affinity shifted to iron. To enhance the removal efficiency by overcoming the limitations associated with living cells such as metal toxicity and inefficiency under adverse operating conditions (Tobin et al., 1994), it was decided to use non-living cells, which are reported to have higher metal uptake capacities (Gadd, 1990). No difference was found (results not shown) in the removal efficiency of mixed metal as compared to the use of living cells but the specificity of metal binding was affected as the heating process could contribute to destabilize reactive groups. Metal tolerance Evaluation of the survival of microorganisms after one, two and twenty-four hours of metal exposure showed little decrease of microbial number in solution of 50 ppm of nickel. In solution containing mixed metals, cells of Shewanella sp were all dead after the first hour while Brevundimonas sp biomass was seriously affected. However, Bacillus subtilis is not affected by the presence of metal, showing no significant change in the growth rate. This implies that in the presence of high concentrations of metals, microbial biomass can be reduced, therefore reducing the removal efficiency of metal from solution. Conclusion Attempts to remove metal from solution using indigenous microorganism isolated around mine areas, showed that this process is largely dependent on the affinity between the microbial cell wall and the metal. It was found that when using living cells, removal efficiency could be affected by factors such as inhibition and efflux transport system. However, use of non-living biomass did not bring any change in removal efficiency but has the advantage that it is simple. Further studies to determine optimal biomass efficiency through a maximum metal removal rate in the shortest possible time (the kinetics of the reaction), coupled with analysis of cost involved, will improve the understanding of the approach to the recovery of excess residual metals in process water discharged from metallurgical activities. Acknowledgements Dr Barnard of the Water and Health Research Unit (UJ) has opened the doors to their research facilities and the South African National Research Foundation (NRF) and the University of Johannesburg provided the research funds and the scholarship. 98



2/14/09

11:41 AM

Page 99

References ALLURI, H.K. et al. 2007. Biosorption: An eco-friendly alternative for heavy metal removal. African Journal of Biotechnology, vol. 6, no. 25, 2007. pp. 2924–2931. BEVERIDGE, T. and MURRAY, R.G.E. 1976. Uptake and retention of metals by cell walls of Bacillus subtilis. Journal of Bacteriology, vol. 127, no. 3, 1976. pp. 1502–1518. BRANDL, H. and FARAMARZI, M.A. Microbe-metal-interactions for the biotechnological treatment of metal-containing solid waste. China Particuology, vol. 4, no. 2, 2006. pp. 93–97. COHEN, R.R.H. 2004. Use of microbes for cost reduction of metal removal from metals and mining industry waste streams. Journal of Cleaner Production. vol. 14, no. 12–13, 2004. pp. 1146–1157. GADD, G.M. Biosorption. Chemistry and Industry, vol. 2, 1990, pp. 421–426. KEFALA, M.I., ZOUBOULIS, A.I., and MATIS, KA. Biosorption of cadmium ions by Actinomycetes and separation by flotation. Environmental Pollution, vol. 104, no. 2, 1999. pp. 283–293. KIM, S.U., CHEONG, Y.H., and SEO, D.C. et al. 2007. Characterisation of heavy metal tolerance and biosorption capacity of bacterium strain CPB4 (Bacillus spp). Water Science and Technology, vol. 55, no. 3, 2007. pp. 105–111. LANGLEY, S. and BEVERIDGE, T.J. Effect of O-side-chain-Lipopolysaccharide chemistry on metal binding. Applied and Environmental Microbiology, vol. 65, no. 2, 1999. pp. 489–498. NIES, D.H. Microbial heavy metal resistance: Molecular biology and utilisation for biotechnological processes, 1999. pp. 1–45. PHILIP, L. and VENKOBACHR, C. An insight into mechanism of biosorption of Cu by B. Polymyxa. Indian Journal of Environmental Pollution, vol. 15, 2001. pp. 448–460. SADOWSKI, Z., GOLAB, Z., and SMITH, R.W. Flotation of Streptomyces pilosus after lead accumulation. Biotechnology and Bioengineering, vol. 37, 1991. pp. 955–959. SANDRIN, T.R. and MAIER, R.M. Impact of metals on the biodegradation of organic pollutants. Environmental Health Perspectives, vol. 111, no. 8, 2003. pp. 1093–1101. SCHIEWER, S. and VOLESKY, B. Mathematical evaluation of the experimental and modeling errors in biosorption. Biotechnology Technology, vol. 9, 1995a. pp. 843–848. SOUTH AFRICAN NATIONAL STANDARD. 2005. Drinking Water. SANS 241, edition 6; ISBN 0-626-17752-9. SRINATH, T., GARG, S.K., and RAMTEKE, P.W. Biosorption and elusion of Cr from immobilized Bacillus coagulens biomass. Indian J Exp. Biol. vol. 41, 2003. pp. 986–990. TOBIN, J.M., WHITE, E., and GADD, G.M. Metal accumulation by fungi: applications in environmental biotechnology. Journal of Industrial Microbiology, vol. 13 1994. pp 126-130; WANG, J. and CHEN, C. Biosorption of heavy metals by Saccharomyces cereviceae: A review. Biotechnology Advances, vol. 24, 2006. pp. 427–451; WANG, J.L. Immobilization techniques for biocatalysts and water pollution control. Beijing: Science Press. 2002a. INDIGENOUS MICROORGANISM STRAINS AS BIO-EXTRACTANTS

99


2/14/09

11:41 AM

Page 100

Elvis Fosso-Kankeu University of Johannesburg, South Africa Master in Biotechnology completed in 2006 at the University of Johannesburg; Research Associate at the University of Johannesburg (2005-2008); Assistant consultant in the monitoring and improvement of plant water treatments (2007). One full article published in 2008 and two manuscripts submitted for publication in 2008. Registered for a doctorate in bioprocessing at the University of Johannesburg (Department of Extraction Metallurgy).

100


101-110_M-Bafubiandi:text

2/14/09

11:42 AM

Page 101

MULABA-BAFUBIANDI, A.F., DLAMINI, N.P., AND MAMBA, B.B. Biosorption of cobalt and copper from hydrometallurgical solutions mediated by Pseudomona spp. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Biosorption of cobalt and copper from hydrometallurgical solutions mediated by Pseudomona spp A.F. MULABA-BAFUBIANDI*, N.P. DLAMINI*†, and B.B. MAMBA† *Department of Extraction Metallurgy, Faculty of Engineering and the Built Environment, University of Johannesburg, South Africa †Department of Chemical Technology, Faculty of Science, University of Johannesburg, South Africa

Pseudomonas spp was isolated from mine water and tailings samples collected from mining effluents and mine dumps in the Gauteng and Northern provinces of South Africa respectively. Adsorption properties of Pseudomonas spp were then tested for the removal of Cu and Co initially from their synthetic sulphate solutions, subsequently from effluents derived from metallurgical operations. The effects of experimental conditions such as pH, temperature, time, volume and metal concentration on the efficiency of the biosorption process were studied. Absorption of 45% Cu (pH 6, 37°C, 24 hours) and 40% Co (pH 6, 37°C, 24 hours) was observed from solutions with low concentrations (0.07 M). A maximum of 73% and 65% of Cu and Co respectively were recovered from the mine water samples derived from effluents. Introduction Metal accumulation in the environment as a result of mining and metallurgical activities is arising as a matter of serious concern. The removal and recovery of metals from hydrometallurgical solutions and waste water prior to their release to the environment are important in preserving the ecosystem as well as from an economical perspective. There are several technologies used to recover metals such as Cu, Co, Zn, Hg and many others. These include among others, ion exchange, precipitation, filtration, electrochemical treatment, electrowinning, reverse osmosis and reduction. These technologies, however, tend to be excessively expensive when the metal concentrations are less than 100 mg/l (Bueno et al., 2008). Since microorganisms have developed survival strategies in metal polluted habitats, their different microbial detoxifying mechanisms such as bioaccumulation, biotransformation, biomineralization or biosorption can be applied either in situ or ex situ. Bacteria also make excellent biosorbents because of their high surface to volume ratios and a high content of potentially active chermosorption sites such as teichoic acid in their cell walls (Beveridge, 1989). BIOSORPTION OF COBALT AND COPPER FROM HYDROMETALLURGICAL SOLUTIONS

101


2/14/09

11:42 AM

Page 102

Biosorption of metals is one of the most promising technologies involved in the removal of toxic substances from industrial wastes and has been receiving a great deal of attention in recent years, not only because of its scientific novelty but also because of its potential application in industry (Wase and Forster, 1997; Bailey et al., 1999; Kumar et al., 2006). It is also a cost-effective technology since it does not require high starting costs. Metal ions uptake by biosorption is complex and may involve the contribution of diffusion, adsorption, chelation, complexaton, coordination or microprecipitation mechanisms, depending on the specific biomass or substrate (Veglio and Beolchini, 1997). Therefore this biological phenomenon can be affected by many chemical and physical variables such as pH, ionic strength, biomass concentration and presence of different heavy metals in solution. All these factors have to be investigated in order to understand how this phenomenon takes place and to optimize operating conditions. Pseudomonas spp, on the other hand, is a widespread bacteria and was observed in the early history of microbiology. The name Pseudomonas meaning ‘false unit’ was derived from Greek pseudo (ψευδο 'false') and monas (μονα´ ς / μονα´ δα 'a single unit'). The term ‘monad’ was used in the early history of microbiology to denote single-celled organisms. This genus was first defined in 1894 as a genus of Gram negative, rod-shaped and polar-flagellated bacteria ( Wikipedia, 2008-09-06) It has been used in bioremediation processes because of its ability to bind to metal ions. The extent of binding is dependent on the nature of binding, metal chemistry and metal affinity of binding sites (Brady and Tobin 1994).Copper and cobalt were chosen for biosorption studies to their wide use in industry and potential pollution impact. Mining companies have become increasingly aware of the potential of microbiological approaches for recovering base and precious metals from low-grade ores, and for bioremediating acid mine drainage and mine tailings. There are two strategies: biomining and bosorption where microorganisms are used to recover metals in solution (e.g. acid mine drainage) by precipitation of the metal, or complexing or absorption with cellular molecules. Biopyrometallurgy then enables recovery of these metals with the microbial biomass contributing as fuel. Materials and methods Isolation of microorganisms Samples of water and soil were collected from mining effluents and mine dumps in the Gauteng and Northern provinces of South Africa respectively. These samples were kept in cool conditions for about 12 hours before isolation could be done. Media preparation Pseudomonas agar base powder (24.20g) was weighed and suspended in 500 ml of distilled water. 5 ml of glycerol was added and the mixture was boiled for about 2 minutes in a microwave to dissolve the mixture completely. The media was sterilized by autoclaving using the HUXLEY HL 341 speedy autoclave at 121°C for 15 minutes. The medium was allowed to cool to about 50°C and a vial of Pseudomonas supplement (SR 103) was aseptically added. The mixture was mixed well and poured into Petri dishes and then allowed to cool for 3 hours. Dilutions (sample preparation) Each sample was mixed thoroughly using a vortex mixer and diluted with a phosphate buffer to 10-1(9 ml of phosphate buffer to 1 ml sample) and 10-2 (9 ml of phosphate buffer to 1 ml of 102



2/14/09

11:42 AM

Page 103

10-1 sample). These were then spread on the media and incubated in the Scientific Series incubator. After 24 hours, some colonies were observed in the media and the number of colony forming units in each sample was calculated using the formula below.

After notable growth was observed, the microorganisms were transferred into nutrient broth and incubated for 24 hours then concentrated by centrifugation and the obtained solid material was washed with deionized water. From this a glycerol stock culture was made for further experiments. Test solutions The metal solutions were divided into three categories i.e. single metal system (Co or Cu), binary metal (Co and Cu mixed system and multi element system (Co and/ Cu plus impurities in varying ratios). The concentrations of Cu (II) and Co (II) were 0.07 M, 0.33 M and 0.66 M and their ratios were varied depending on the category in testing. Batch biosorption experiments The factors that affect the adsorption rate were examined in a batch system. All experiments were carried out with microorganism suspention in Erlenmeyer flasks in an incubator at 37±0.5°C with constant shaking to elucidate optimum conditions (contact time, pH, initial metal concentration and bacterial dose). The contents of the flasks were filtered and the filtrates were diluted (10 ml was pipetted into 100 ml volumetric flask and filled to the mark with distilled water.) This was done in order to get lower concentrations, which could be analysed by flame absorption spectrometry to obtain residual metal concentrations. The metal adsorbed by biomass was calculated as:

where Ci is the initial metal concentration, Cf is the final metal concentration and MAS is the total amount of metal adsorbed by other co-existing possible biomass contaminants or metal precipitated from the solution. Bacterial interaction with metal species Interaction of the bacteria with the different categories of metal species was studied using the scanning electron microscope (SEM), model SEM Joel JSM 5600. Recovery from mine water samples Mine water samples were collected from mining effluents and mine dumps in the Gauteng province of South Africa. These samples were assayed to determine the presence of copper and cobalt ions. Pseudomonas spp isolated from metallurgical operation effluents was then used to recover these metals from their solutions. This was done by spiking 100 ml of the water with 20 ml of Pseudomonas spp. Results and discussion Isolation of bacteria The isolation of bacteria using Pseudomonas agar base yielded positive results meaning that there blue-green colonies of Pseudomonas spp were observed. BIOSORPTION OF COBALT AND COPPER FROM HYDROMETALLURGICAL SOLUTIONS

103


2/14/09

11:42 AM

Page 104

Effect of biomass (bacteria) and metal ion concentration In the first stage of batch biosorption experiments on Pseudomonas spp., the combined effects of the amount of biomass (bacteria) and metal ion concentration were examined. Three different amounts were used (50 × 104 CFU/ml , 100 × 104 CFU/ml and 150 × 104 CFU/ml) i.e. 1:1, 1: 3 and 3:1 ratios and 0.07, 0.33 and 0.66 M solutions of both Cu and Co sulphate solutions. These results are illustrated in Figure 1. Low concentrations of both copper and cobalt sulphate solutions are absorbed more than higher concentrations and the 3:1 (bacteria: metal ion) ratio absorbs more than 1:1 and 1:3 ratios. The effect of the initial concentration of Co (II) and Cu (II) The effect of the initial concentration of Co (II) and Cu (II) ions is illustrated in Figure 1, i.e. metallic uptake as a function of the equilibrium concentration. The results demonstrate a decrease in metallic uptake with an increase in the equilibrium concentration. This observation is linked to the metallic concentration gradient with respect to the active and available sites of the Pseudomonas sp, which means that metal uptake decreases with an increase in initial ion concentration. For example, the highest removal percentages are observed in the 0.07 M concentration solutions for both Cu2+ and Co2+ i.e. 45% and 40% respectively. This could be attributed to the microbial tolerance of the bacterial species. For example, the bacteria thrive better in the case of low concentration levels of cobalt than in high concentrations. As time progresses the die off is not as bad as in the highly concentrated solutions. Different biosorbents used on the uptaking of various heavy metals yielded similar observations (Kumar et al., 2006; Fiol et al., 2006; Abu Al-Rub et al., 2006, Pan et al., 2006). The effect of biomass concentration As illustrated in Table I and Figure 1, an increase in the amount of adsorbent, which in this case are bacteria (biomass), results in the increase in metal absorption efficiency (removal efficiency). This behaviour is due to an increase in binding sites as the bacterial species increase (surface area), (Brady and Tobin 1994).

Table I The effect of biomass (bacteria) and metal ion concentration on metal extraction Solution concentration Amount of bacteria (× 104 CFU/ ml) Amount of solution (ml)

Ratio

Removal efficiency (%) Cu

Co

0.07 M

50 100 150

150 100 50

1:3 1:1 3:1

26 38 45

21 31 40

0.33 M

50 100 150

150 100 50

1:3 1:1 3:1

22 33 38

19 22 29

0.66M

50 100 150

150 100 50

1:3 1:1 3:1

19 24 31

18 21 25

104



2/14/09

11:42 AM

Page 105

Figure 1. The combined effect of biomass (bacteria) and metal ion concentration on metal extraction

Table II Showing the effect of co-cations on metal extraction (3:1 biomass:solution ratio) Solution

Cu/Co SO4 Cu/Co SO4 Co/CuSO4 Cu/Co SO4 Co/CuSO4

Concentration of solution (M)

0.07:0.07 0.07:0.33 0.33:0.07 0.07:0.66 0.66: 0.07

Ratio CO:CU

1:1 1:5 5:1 1:9 9:1

% Metal extracted CU

CO

43 41 36 44 32

31 26 38 21 40

The effect of co-cations on metal extraction Many researchers have explored the feasibility of this approach (Tsezos, 1983). However, the majority of published work on biosorption is concerned with one metal. Very little information is available for binary and multimetal biosorption systems (Gonzalez Davila, 1990). Since biosorption is foreseen to be implemented in the fields of water treatment and hydrometallurgy where complex multicomponent metal systems are common, more work is required. The copper and cobalt sulphate solutions were further mixed to try and mimic or be as close to reality as possible and to see the effect each cation has on the other. This was done by varying the concentrations of the two cation solutions and the observations were as follows: Mixing of the two cations does affect the rate at which biosorption occurs. Table II depicts the effect the two cations have on each other. It is notable especially when comparing the single element matrices in Table I with binary element system in Table II that mixing of the two cations results in an enhancement in the amount of copper absorbed by the bacteria. The more dilute combinations showed better absorption than concentrated solutions. These results are illustrated graphically in Figure 2. The 5:1 and 9:1 copper: cobalt ratio (0.33:0.07 and 0.66: 0.07 respectively) in Table II and Figure 2 illustrates an almost comparable recovery of copper and cobalt, with cobalt being favoured. This could be due to the fact that the cobalt is more dilute than the copper and thus binds easily to the surfaces of the bacteria. BIOSORPTION OF COBALT AND COPPER FROM HYDROMETALLURGICAL SOLUTIONS

105


2/14/09

11:42 AM

Page 106

Figure 2. The effect of co-cations on metal extraction

Figure 3. Langmuir isotherm

The data fitted well on both Langmuir and Fruendlich equations but the correlation was much higher on the Fruendlich as expected, since Freundlich approach gives an equation that includes the heterogeneity of the surface of the ion exchanger and the exponential distribution of active sites and their energies. As seen from Figure 3, when 1/ (x/m) is plotted against 1/c, the Langmuir model fits the data quite well; (1/ab) is the intersection of the line with the y-axis. As observed in Figure 3 and 4 both models fit the data well. The Freundlich equation gave a plot with higher correlation as compared to the Langmuir equation. The Freundlich gave an R 2 value of 0.9622 whereas the Langmuir had an R 2 value of 0.8615. Therefore the Fruendlich plot is favoured over the Langmuir. The effect of contamination on metal extraction (for a 3:1 biomass: solution ratio) The 3: 1 bacteria: solution ratio was further examined for contaminant tolerance abilities. The contaminants used were silicon dioxide and iron chloride. 106



2/14/09

11:42 AM

Page 107

Figure 4. Freundlich Plot

It is observed that the presence of any contaminant decreases the copper and cobalt biosorption by Pseudomonas spp. From the results in Table III it is observed that contamination in single element species does not result in major decreases in extraction of metal ions. SiO2, however, decreases ion extraction more when compared to FeCl2. The same trend is observed in the binary element category. When both contaminants exist further reduction in the extracted metal ions is observed, especially in the case where the ratio of the contaminants to solution is higher (2:1). This is the same case in both single metal and binary metal categories. It is observed that in all conditions that there is better adsorption of Cu (II) than that of Co (II). This may be accounted for by the nature of the two metal ions. The charges on both Cu (II) and Co(II) ions are the same, but they have different hydrated radii; Cu(II) has a lower hydrated radius than Co(II). Adsorption of an ion by an ion exchanger largely depends on the hydrated radius. Similar observations on an adsorbent called clinoptilolite were observed by Erdem, who concluded that the larger the diameter, the slower its mobility and thus the less likely its exchange will be (Erdem et al; 2004). It appears that for ease in adsorption the water molecules surrounding the cation should be fewer. Interaction of bacteria with metal ions Interactions of the micro-organisms with the metal species were examined using scanning electron microscopy (SEM), model SEM Joel JSM 5600) and transmission electron microscopy, as illustrated in Figure 5 and 6. The SEM micrograph revealed attachment of metal ions on the cell wall of the Pseudomonad and the same observation is confirmed by the darkish accumulations observed in the cell wall of the bacterium in the TEM image. The interaction of the Pseudomonas rods with the multi-element matrix is illustrated in Figures 5 and 6. Figure 6 shows the bioaccumulation of the metal ions on the cell walls of the bacteria. Recovery from mine water The metal removal concept was then applied on mine water sampled from Nigel town and Table IV shows the recovery of copper and cobalt from the mine water. BIOSORPTION OF COBALT AND COPPER FROM HYDROMETALLURGICAL SOLUTIONS

107


2/14/09

11:42 AM

Page 108

Table III The effect of contamination using silicon dioxide and iron chloride on metal extraction (for a 3:1 biomass: solution ratio) Solution

Cu : Si

Co : Si

Cu : Fe

Co : Fe

Cu : (Si : Fe)

Co : (Si : Fe)

(Cu/Co) : Si

(Cu/Co) : Fe

(Cu/Co) : (Fe: Si)

Solution: contaminants ratios Cu SiO2 contaminant only, on single element species 1:1 29 1:2 22 2:1 35 1:1 1:2 N/A 2:1 FeCl2 contaminant only, on single element species 1:1 33 1:2 26 2:1 37 1:1 1:2 N/A 2:1 FeCl2 and SiO2 contaminants on single element species 1:1 23 1:2 19 2:1 26 1:1 1:2 N/A 2:1 SiO2 contaminant only, on binary element speices 1:1 28 1:2 24 2:1 36 FeCl3 contaminant only, on binary element species 1:1 33 1:2 28 2:1 40 FeCl3 and SiO2 contaminant on binary element species 1:1 26 1:2 19 2:1 31

% Metal extracted Co

N/A 22 19 28

N/A 24 20 28

N/A 19 14 22 23 16 28 24 19 34 18 12 24

Metallic species

Bacterium with metal ions accumulated on the cell wall

Figure 5. SEM micrograph showing bacterial interaction with metallic species (1:3 ratio)

108



2/14/09

11:42 AM

Page 109

Possible accumulation of metal ions on the cell

Pseudomonas spp

Figure 6. TEM image of metal ion accumulation on the cell wall of Pseudomonas spp.

Table IV Metal recovery from mine water Sampled site 1. Big dam Nigel 2. River Nigel A 3. River Nigel B 4. Stream Nigel 5. Nigel 169 (Artisanal) 6. Small dam (HVH) 7. Stream (HVH) 8. Shaft C 9. Piet Farm A 10.Piet Farm B 11. Piet Farm C 12. Piet Farm D

Copper % recovery

Cobalt % recovery

50 42 37 73 61 29 24 36 37 67 38 40

45 35 65 46 11 22 40 20 52 30 0 0

The results in Table IV illustrate an increased recovery efficiency of copper and cobalt. A maximum of 73% and 65% of copper and cobalt respectively was recovered. This could be attributed to the fact that the mine water samples were very dilute with concentrations ranging from 2–20 ppm of copper and from 1–8 ppm of cobalt . Other possible minor contaminants (whose effects were not considered in this study) include Fe, Ca and Mg. Conclusion From the results obtained thus far, it can be concluded that Pseudomonas spp reclaims both copper and cobalt from their sulphate synthetic solutions. It tends to remove or extract more metal ions at low concentrations and the ratio of bacteria population to solution volume being 3:1 works best. The bacteria have successfully removed up to 45% of copper from 0.07 M and up 40% from 0.07 M cobalt from cobalt sulphate solutions and 73% and 65% of copper and cobalt respectively from mine water samples. The absorption favours the removal of copper over that of cobalt, which may be accounted for by the structural differences of the two cations. BIOSORPTION OF COBALT AND COPPER FROM HYDROMETALLURGICAL SOLUTIONS

109


2/14/09

11:42 AM

Page 110

References ABU AL-RUB, F.A., EL-NAAS, M.H., ASHOUR, I., and AL-MARZOUQI, M. Biosorption of copper on Chlorella vulgaris from single binary and ternary metal aqueous solutions, Process Biochem. vol. 41, 2006. pp. 457–464. BAILEY, S.E., OLIN, T.J., BRICKA, R.M., and ADRIAN, D.D. A review of potentially lowcost sorbents for heavy metals, Water Res. vol. 33, 1999. pp. 2469–2479. BEVERIDGE, I., PULLMAN, A.L., MARTIN, R.R., and BARELDS, A. Effects of temperature and relative humidity on development and survival of the free-living stages of Trichostrongylus colubriformis, T. Rugatus and T. Vitrinus ). Veterinary Parasitology, vol. 33, no. 2, 1989. pp. 143–153. BRADY, J.M., and TOBIN, J.M. Adsorption of metal ions by Rhizopus arrhizus biomass: Characterization studies. Enzyme and Microbial Technology, vol. 16, no. 8, 1994. pp. 671–675. BUENO, B.Y.M., TOREM, M.L., MOLINA, F., and DE MESQUITA, L.M.S. Biosorption of lead(II), chromium(III) and copper(II) by R. Opacus. Equilibrium and kinetic studies, Minerals Engineering, vol. 21, 2008, pp. 65–75 ERDEM, E. KARAPINAR, N., and DONAT, R. The removal of heavy metal cations by natural zeolites, Journal of Colloid and Interface Science, vol. 280, no. 2, 2004. pp. 309–314. FIOL, N., VILLAESCUSA, I., and MARTINEZ, M. Sorption of Pb(II), Ni(II), Cu(II) and Cd(II) from aqueous solution by olive stone waste, Sep. Purif. Technol. vol. 50, 2006. pp. 132–140. GONZALEZ, D.M., SANTANA-CASIANO, J.M., and MILLERO, F.J. The adsorption of Cd (II) and Pb (II) to chitin in seawater. J. Coll. Interf. Sci. vol. 137, 1990. p. 102. KUMAR, Y.P., KING, P., and PRASAD, V.S. Equilibrium and kinetic studies for the biosorption system of copper(II) ion from aqueous solution using Tectona grandis L.f. leaves powder, J. Hazardous Mater. B137, 2006. pp. 1211–1217. PAN, J., GE, X., LIU, R., and TANG, G.H. Characteristic features of Bacillus cereus cell surfaces with biosorption of Pb(II) ions by AFM and FT-IR, Colloids Surfaces B: Biointerfaces, vol. 52, 2006. pp. 89–90. TSEZOS, M. The role of chitin in uranium adsorption by R. Arrhizus. Biotech. Bioeng. vol. 25, 1983. p. 2025. VEGLIO, F. and BEOLCHINI, F. Removal of metals by biosorption—a review, Hydrometallurgy, vol. 44, 1997. pp. 301–316. WASE, J. and FORSTER, C. Biosorbents for Metal Ions, Taylor and Francis Ltd, 1997. http://en.wikipedia.org/wiki/Pseudomonas_aeruginosa , accessed , 2008-09-06

Nonjabulo Prudence Dlamini Student, University of Johannesburg Nonjabulo Prudence Dlamini received a Bachelor of Science degree from the University of Swaziland in 2006 where she has majored in Chemistry and Biological Sciences. She has just completed her MSc (chemistry) on the use of micro-organisms in removing metals from hydrometallurgical solutions. She will be graduating in May and is looking forward to pursuing a PhD in mineral processing. 110


111-118_Mwanza:text

2/13/09

11:02 AM

Page 111

MWANZA, M. and MULABA-BAFUBIANDI, A.F. An investigation into fungal strains bio-extraction of metal impurities from aqueous effluents emanating from Ekhuruleni (Gauteng, South Africa) metallurgical and mining operations. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

An investigation into fungal strains bio-extraction of metal impurities from aqueous effluents emanating from Ekhuruleni (Gauteng, South Africa) metallurgical and mining operations M. MWANZA* and A.F. MULABA-BAFUBIANDI* Mineral Processing and Technology Research Group, Faculty of Engineering and the Built Environment, Wits, South Africa

This paper discusses an attempt to use indigenous fungal strains for the fixation and immobilization of identified metal impurities from aqueous effluents collected in Nigel, Gauteng (South Africa). Ca, Fe and Mg were predominantly found in aqueous effluents in Nigel areas. Fungal species obtained by culturing soil samples from Nigel areas on agar medium and identified by gene sequencing were tested for their abilities to absorb (biosorption/bioaccumulation). The dominant metals in situ were exposed to known concentrations of fungal cultures and assed for fungal absorption. Fungal cultures (Aspergillus niger and Aspergillus fumigatus) were washed from their ‘home’ medium and cultured in sterile water for 24 hours. Two different concentrations (30 and 50 ppm) of Ca, Fe, and Mg were spiked separately in triplicate. Concentration of metals measurements were taken after 1, 2, and 24 hours. The results obtained thus far show that Ca, Fe and Mg absorption by Aspergillus niger and Aspergillus fumigatus is reduced with time of exposure. Additionally, it was observed that both metal concentrations did not affect the fungi as these were still alive even after 24 hours’ exposure, suggesting that probably higher concentrations of metals could be absorbed without destroying the surrounding fungal ecosystem.

Introduction Environmental constraints have become an important issue and most mining industries are asked to get involved in the reduction of waste as it is known that mining wastes and drainage are the major contributors in polluting the surroundings. Several techniques as well as microorganisms such as bacteria are also being used in this regard. Barros et al. (2003) and Bhainsa and D’Souza (1999) used Aspergillus niger and Aspergillus fumigatus strains for uranium and cadminium absorption. In addition, Blanquez, 2008, Blaudezi et al., 2000, AN INVESTIGATION INTO FUNGAL STRAINS BIO-EXTRACTION OF METAL IMPURITIES

111

111-118_Mwanza:text

2/13/09

11:02 AM

Page 112

Satofuka et al., 1999, and Dupre de Boulois et al., 2007 found that some fungal species such as white fungi could be used in water detoxification and also some other fungi such as micorrhiza can be used in soil detoxification from minerals and discharging the environment from heavy metals. In addition, the presence of these fungi is a good as they may play an important role in soil by precipitating of carbonates, phosphates, and hydroxides, which increases soil aggregation and also in stimulating the precipitation of compounds that act as bonding agents for soil particles (Arnott, 1995; Burford et al., 2006; Gorbushina, et al., 2002). This study looked at the possible use of fungal strains’ (Aspergillus niger and Aspergillus fumigatus) capacity to reduce selected metal concentration in water and their possible use in the detoxification of the environment. The results obtained thus far show that Ca, Fe and Mg absorption by Aspergillus niger and Aspergillus fumigatus increased with time of exposure. Methodology Soil samples (21) from Nigel gold mines were collected in sterile caped bottles and were transported to the laboratory in a cooler box. In the laboratory samples were stored in a fridge, samples were thoroughly mixed, and 1 g was transferred into a McCartney bottle containing 9 ml of sterile ringer solution (1 ringer tablet in 500 ml of purified water). One milliliter was then transferred to a further 9 ml of ringer solution and then diluted five times using the same dilution method under aseptic conditions. One milliliter from each McCartney bottle was added to a petri dish containing 20 ml of Ohio Agricultural Experimental Station agar (OAESA) (Kaufman et al., 1963) or potato dextrose agar (PDA) and mixed. After solidification, the plates were incubated for 4–7 days at 28°C. On the fourth to the seventh day, plates were screened for different types of fungal colonies. These isolated fungal colonies were sub-cultured on PDA, Czapek 20 (CY20S), Czapek (CZ) and malt extract agar (MEA) (Klich, 2002) under aseptic conditions and incubated at 28°C for 3 to 4 days. On the fourth day, pure fungal colonies were stained with Lactophenol blue solution on microscope slides for identification. The macro scopic and microscopic identifications of fungi were determined according to Pitt and Hocking (1997) and Klich (2002) for Aspergillus and Penicillium spp. and Nelson et al. (1983) for Fusarium spp. The isolates were sent to Inqaba Biotechnology in Pretoria for DNA sequencing. From selected cultures, 200 mg of fungal cultures were inoculated in sterile medium composed of 97 ml of distilled water and 30 ppm of metal or 95 ml of distilled water and 50 ppm of metal. All samples and controls were run in triplicate. The mixture was incubated in a shaker incubator at 25ºC and a speed of 150–200 rpm. Five ml of the solution were collected every 1 hour, 2 hours and 24 hours for analysis. The collected samples were filtered using filter paper and analysed for metal quantification using the ICP-MS. Results and discussion The macroscopical and sequenced results, shown in Figure 1, revealed the presence of fungi in these samples with a predominance of Aspergillus fumigatus (100%), Aspergillus niger (100) both seen in Figure 2, Trichoderma harzianum (37%), Hypocrea lixii (6%), Mucoromycote spp (3%). Penicillium janthinellum (4%), Cochliobolus lunatus (7%), Trichoderma koningiopsis (7%), Gibberella moniliformis figure (5%), Bipolaris spicifera (7%) and Cochliobolus spicifer (1%) of analysed soil samples. In addition it was recorded that the samples were contaminated with several fungal species at one. The growth of fungi in these areas might also be explained by their biological ability to resist in lower pH environments (2–8) as compared to other bacteria (Pitt and Hocking, 1997) 112


111-118_Mwanza:text

2/13/09

11:02 AM

Page 113

tus niger anum a lixii pp ellum natus iopsis igure icifer ifer f iga i pic p e u te s in g s fum rgillu a harz ypocr myco janth olus l konin ormis laris s olus s s b o H a ilif Bipo illu Aspe erm m iob cor lliu ochlio derm mon erg chl Mu enici hod o C Co Asp lla h e c P i r Tric Tr ibbe G

Figure 1. Graph illustrating fungal contamination from Nigel soil samples

Figure 2. Aspergillus niger (left) and Aspersgillus fumigatus (right)

but also by the presence of these minerals, valuable nutrients for the growth and survival of fungi (Pitt and Hocking, 1997). Aspergillus niger, characterized by its black spherical conidia, can grow down at pH 2 at high aw (Pitt and Hocking, 1997) while Aspergillus fumigatus is characterized by overgrowth white to greenish mycelium, it is has a thermophilic nature and can grow at optimum temperature 12–42°C and required up to o.90 aw (Pitt and Hocking, 1997). The results obtained from the ICP-MS on the two major fungi found present in soil, Aspergillus niger (Figure 3) and Aspergillus fumigatus (Figure 4) show progressive increased absorption of metal over the time of exposure. It was observed that Ca and Fe were highly absorbed as compared to Mg. This might be explained by the fact that Ca and Fe are required for many vital processes in plants and animals (Moreau, 1987, and Pitt and Hocking, 1997). There was not much significant statistical difference in absorption between the three metals and between the two fungi although Aspergillus fumigatus showed a slightly higher abortion of metals as compared to Aspergillus niger. AN INVESTIGATION INTO FUNGAL STRAINS BIO-EXTRACTION OF METAL IMPURITIES

113

111-118_Mwanza:text

2/13/09

11:02 AM

Page 114

Figure 3. 30 ppm of percentage of absorption vs. exposure time of metal by Aspergillus niger

Figure 4. 30 ppm of percentage of absorption vs. exposure time of metal by Aspergillus fumigatus

Table I Percentage means concentrations of metal absorption by fungal exposure on ICP-MS Conc. ppm

Exposure time

Aspergillus niger

Aspergillus fumigatus Metal

Fe

Ca

Mg Fe % absorption

Ca

Mg

30

1 hour 2 hours 24 hours

21.50 27.3.9 81.80

21.05 29.50 85.50

06.33 17.63 46.60

22.21 33.50 85.23

27.98 35.29 88.65

26.36 30.25 42.26

50

1 hour 2 hours 24 hours

5.50 16.56 65.34

7.45 15.36 75.24

05.89 13.52 57.97

06.91 21.06 70.50

07.31 29.05 84.06

07.02 40.25 61.13

The results obtained and summarized in Table I revealed that the metal absorption capacity of Aspergillus niger and Aspergillus fumigatus showed no significant difference. However, Aspergillus fumigatus revealed a slightly higher absorption capacity as compared to Aspergillus niger. It was observed that calcium was highly absorbed by both fungi as compared to iron and magnesium sulphate with high absorption level with Aspergillus fumigatus with 84% absorption and only 76% absorption for Aspergillus niger. This might be explained by the fact that calcium is required for many vital processes in plants and animals (Moreau, 1987, and Pitt and Hocking, 1997). In addition, it was found that at the end of the experiment all fungal cultures were alive. This confirms that the absorption process is 114


111-118_Mwanza:text

2/13/09

11:02 AM

Page 115

biological and does not lead to the death of the fungal ecosystem (Moreau, 1987, and Pitt and Hocking, 1997). These findings agree with those of Strandberg et al., 1981, who demonstrated the rapid uptake of uranium and postulated that polyphosphate groups and carboxyl groups in Sarcomyces. cerevisiae cell walls are active in metal complexation. The results obtained in this study also agree with Barros et al. (2003) and Bhainsa and D’Souza (1999) who used Aspergillus niger and Aspergillus fumigatus strains for uranium and cadminium absorption. This also confirms the findings of Blanquez (2008), Kefala et al., (1999), Blaudezi et al., (2000), Satofuka et al., (1999) and Dupre de Boulois et al., (2007), who found that fungal species such as white fungi can be used in water detoxification and also some other fungi such as micorrhiza can be used in soil detoxification from minerals by discharging the environment from heavy metals. In addition, the presence of these fungi is good as they may play an important role in soil by precipitating of carbonates, phosphates, and hydroxides which increase soil aggregation and also in stimulating the precipitation of compounds that act as bonding agents for soil particles (Arnott 1995; Burford et al., 2006, Gorbushina, et al., 2002). Conclusion The current global environmental challenges are mainly due to pollution are caused, to a large extend, by modern industries and mines. It is thus important to look at environment friendly methods to reduce water and soil pollution. There is a need for metal detoxification of soil and probably of water from mines. The use of fungal strains such Aspergillus niger and Aspergillus fumigatus commonly found surviving in these soils would be of importance as they contribute to environmental detoxification by absorbing these metals as nutrient sources. Acknowledgments This work was supported by Professor Mike Dutton: Food Environmental and Health Research Group, Faculty of Health Sciences, University of Johannesburg and Dr Herman Van Niekerk: RAUSPECTRA, University of Johannesburg. This work was funded by The University of Johannesburg Research Council and National Research Foundation for financial assistance. References ALLURI, H.K., et al. Biosorption: An eco-friendly alternative for heavy metal removal. African Journal of Biotechnology. vol. 6, no. 25, 2007. pp. 2924–2931. ARNOTT, H.J. Calcium oxalate in fungi. Khan S.R. (ed.), Calcium Oxalate in Biological Systems. CRC Press, Boca Raton, 1995. pp. 73–111. BHAINSA, K.C., and D'SOUZA, S.F. Biosorption of uranium (VI) by Aspergillus fumigatus. Biotechnol. Tech. vol. 13, 1999. pp. 695–699. BARROS, L.M., MACEDO, G.R., DUARTE, M.L., SILVA, E.P., and LOBATO, A.K.C.L. Biosorption of Cd using the fungus A. niger. Braz J. Chem. Eng. vol. 20, 2003. pp. 229–239. BLANQUEZ, P., SARRA, MM., and VICENT, T. Development of a continuous process to adapt the textile wastewater treatment by fungi to industrial conditions. Process Biochemistry, vol. 43, 2008. pp. 1–7. AN INVESTIGATION INTO FUNGAL STRAINS BIO-EXTRACTION OF METAL IMPURITIES

115

111-118_Mwanza:text

2/13/09

11:02 AM

Page 116

BLAUDEZI, D., JACOBI, C., TURNAU, K., COLPEART, J., AHONEN, V., FINLAY, U., BOTTONI, B., and CHALOTTI., M. Differential responses of ectomycorrhizal fungi to heavy metals in vitro. Mycol. Res. vol. 104, no. 11, 2000. pp. 1366–1371. BURFORD, E.P., HILLIER, S., and GADD, G.M. Biomineralization of fungal hyphae with calcite (CaCO3) and calcium oxalate mono- and dihydrate in Carboniferous limestone microcosms. Geomicrobiology Journal, vol. 23, 2006. pp. 599–611. DUPRE DE BOULOIS, H., JONER, H.J., LEYVAL, C., JAKOBSEN, L., CHEN, B.D., ROOS, P., THIRY, Y., RUFYIKIRI, P., DELVAUX, B., and DECLERCK, S. Impact of arbuscular mycorrhizal fungi accumulation by plants. Journal of Environmental Radioactivity, vol. xx: 2007. pp. 1–10. IBEANUSI and WILDE. Bioremediation of coal Pile off waters using an integrated microbial ecosystem. Biotechnology letters, vol. 20, no. 11, 1998. pp. 1077–1079. GORBUSHINA, A.A, KRUMBEIN, W.E., and VOLKMANN, M. Rock surfaces as life indicators: new ways to demonstrate life and traces of former life. Astrobiology, vol. 2, 2002. pp. 203–213. KAUFMAN, D.D., WILLIAMS, L.E., and SUMNER, C.B. Effect of plating medium and incubation temperature on growth of fungi in soil-dilution plates. Canadian Journal of Microbiology, vol. 9, 1963. pp. 741–751. KEFALA, M.I., ZOUBOULIS, A.I., and MATIS, K.A. Biosorption of cadmium ions by Actinomycetes and separation by flotation. Environmental pollution, vol. 104, no. 2, 1999. pp. 283–293. KLICH, M.A. Identification of common Aspergillus species. Ponson and Looijen, Wageningen. The Netherlands, 2002. 1-107. MOREAU, R.A. Calcium-building proteins in Fungi and Higher Plants. J. Dairy Sci., vol. 70, 1987. pp. 1504–1512. NELSON, P.E., TOUSSIN, T.A., and MARASSAS, W.F.O. Fusarium Species. An illustrated Manual for Identification. University Park Pennsylvania: Pennsylvania State University Press. 1983. PITT, J.I. and HOCKING, A.D. Primary keys and miscellaneous fungi. Fungi and Food Spoilage. 2nd Edition. Blackie Academic and Professional. London, Weinheim, New York, Tokyo, Melbourne, Madras, 1997. pp. 59–171. SATOFUKA, H., SATOSHI, A., ATOMI, H., TAGAGI, M., KAZAMUSA, H., KAZUHISA, M., and TADAYUKI, I. Rapid Method for Detection and Detoxification of Heavy Metal. 1999. Ions in Water Environments Using Phytochelatin. J. of Bioscience Engineering, vol. 88, no. 3, pp. 287–292. STRANDBERG, G.W., SHUMATE II, S.E., and PARROT, J.R. Microbial cells as biosorbents of heavy metals: Accumulation of uranium by Saccharomyces cerevisiae and Pseudomonas aeruginosa. Appl Environ Microbiol, vol. 41, 1981. pp. 237–245. 116


111-118_Mwanza:text

2/13/09

11:02 AM

Page 117

Mulunda Mwanza PhD Fellow, University of Johannesburg, South Africa Qualified Vet from the University of Lubumbashi/DRC, I joined the University of Johannesburg for a master degree in Biotechnology (obtained 2007). I am Actually working on a PhD project at the University of Johannesburg on fungal and their mycotoxin contamination in animal feed, animal products and tissues with particular reference to synergistic effects of mycotoxins on animal and human system. In addition, I am interested on environmental questions such as soil and water contamination and their detoxifications.

AN INVESTIGATION INTO FUNGAL STRAINS BIO-EXTRACTION OF METAL IMPURITIES

117

111-118_Mwanza:text

118

2/13/09

11:02 AM

Page 118


119-154_Neale:text

2/14/09

11:48 AM

Page 119

NEALE, J.W., ROBERTSON, S.W., MULLER, H.H. and GERICKE, M. Integrated piloting of a thermophilic bioleaching process for the treatment of a low-grade nickel-copper sulphide concentrate. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Integrated piloting of a thermophilic bioleaching process for the treatment of a low-grade nickelcopper sulphide concentrate J.W. NEALE, S.W. ROBERTSON, H.H. MULLER and M. GERICKE Mintek Biotechnology Division, South Africa

Mintek was a leading participant in the BioMinE project between 2004 and 2008. This project, which was funded in part by the European Commission, was aimed at the development of biotechnology for the minerals industry in Europe. Mintek’s research programme focused mainly on the development of integrated bioleach-based processes for the recovery of base metals from complex, low-grade sulphide concentrates. Specific European mineral resources were targeted and used in integrated piloting campaigns involving bioleaching, solution purification, and metals recovery. This paper describes the use of thermophilic bioleaching for the recovery of nickel and copper from a low-grade nickel-copper concentrate produced at the Aguablanca Mine in southern Spain. Currently, the Aguablanca Mine produces a bulk nickel-copper concentrate for sale to a smelter, and the proposition is to increase the profitability of the operation by the on-site production of metal or metal intermediate. Initially, bench-scale bioleach tests were conducted to determine the bioleach operating conditions. These tests included an evaluation of mesophilic, moderately thermophilic and thermophilic microorganisms. In order to achieve sufficiently high levels of both copper and nickel extraction, a thermophilic process was selected— this was necessary for leaching of the refractory chalcopyrite that occurs in this concentrate. Additional bench-scale test work was carried out to derive a conceptual process flowsheet for the solution purification and metals recovery circuit. The results of the bench-scale tests were used to design, construct and commission an integrated pilot plant, which was subsequently operated at Mintek for over seven months. During this time, the solution purification and metals recovery processes were optimized, and all recycle loops were closed. The final process flowsheet included the following unit operations: concentrate regrinding, thermophilic bioleaching at 70°C, primary iron removal using limestone, copper solvent extraction and electrowinning, secondary INTEGRATED PILOTING OF A THERMOPHILIC BIOLEACHING PROCESS

119

119-154_Neale:text

2/14/09

11:48 AM

Page 120

iron removal, nickel hydroxide precipitation using magnesia, and final solution purification using lime. Where applicable, process solutions were recycled to preserve water. The process design data derived from this pilot-plant campaign formed the basis for a conceptual engineering study for the developed process. In the study, mass and energy balances were derived, and a process flowsheet was developed and used as the basis for estimating the capital and operating costs of the process. This enabled a preliminary economic analysis of the process to be undertaken. The findings of this study are discussed.

Introduction The BioMinE project The BioMinE project1,2 was a four-year-long integrated project under the Sixth Framework Programme of research supported by the European Commission, which ran from November 2004 to October 2008. The project was aimed at the evaluation of biohydrometallurgy to improve the exploitation of the European non-ferrous metal resources in a sustainable way. At the time of its completion, the project had 37 participating partners covering a diversity of interests: research organisations, academic institutions, and industrial and mining companies. South Africa was well represented: besides Mintek, two academic institutions and one mining company from South Africa were members of the BioMinE consortium. Mintek’s role in BioMinE Mintek’s Biotechnology Division played a leading role in the project, both as coordinator of the bioleaching ‘Work Package’ and by active participation in key aspects of the research and technology development (RTD). One of the main focuses of Mintek’s research programme was the development of integrated bioleach-based processes for the recovery of base metals from complex, low-grade sulphide concentrates. Specific European mineral resource types were targeted and used in integrated piloting campaigns involving bioleaching, solution purification, and metals recovery. One such resource is copper polymetallic concentrates, and a nickel-copper concentrate, which is currently produced at the Aguablanca Mine in southern Spain, was selected as one of the target resources in this project. Mintek conducted extensive bench-scale bioleach testing on this material, the results of which formed the basis for an integrated bioleaching pilot-plant campaign that was conducted at Mintek. Bioleaching of base metal concentrates The bioleaching of concentrates of base metal sulphides has been the subject of extensive research and development over several decades3. The use of mesophilic bacteria to oxidize sulphide minerals such as cobaltiferous pyrite, pentlandite and millerite (nickel), sphalerite (zinc), galena (lead), chalcocite, covellite, cubanite and bornite (all copper) has been demonstrated, and in some cases the bioleaching process has successfully been integrated with downstream metals recovery processes4,5,6. In the early 1990s, BHP Billiton demonstrated the bioleaching of a pentlandite concentrate using mesophiles and moderate thermophiles, in a demonstration plant with a design capacity 120


119-154_Neale:text

2/14/09

11:48 AM

Page 121

of 300 kg/d of concentrate. This plant produced more than 700 kg of Class 1 nickel in a sixmonth-long campaign. Subsequently, an integrated pilot plant situated at the Yabula Nickel Refinery in Queensland, Australia, was successfully operated as part of a feasibility study4. In the late 1990s, the world’s first commercial bioleaching plant treating a base metal concentrate was established at the site of the defunct Kilembe Copper Mine in Uganda. The plant, owned an operated by the Kasese Cobalt Company, was designed to produce about 1000 t/a of cobalt cathode from a cobaltiferous pyrite concentrate. The technology for this plant was supplied by the French company, BRGM5. Bioleaching of the primary copper sulphide, chalcopyrite, has presented more of a challenge, since the surface of the chalcopyrite tends to passivate at the temperatures and redox potentials found in typical mesophilic bioleach processes, effectively stopping the leaching of the mineral. This behaviour can be overcome by a variety of approaches, often applied in combination, which include increasing the operating temperature (requiring the use of thermophilic micro-organisms), finer grinding of the concentrate, the addition of catalysts (such as silver), and control of the redox potential3,6,7,8,9. In 2001, Mintek and its partners, Industrias Peñoles S.A. de C.V. of Mexico and BacTech, successfully demonstrated the bioleaching of chalcopyrite using moderate thermophiles at an operating temperature of 45°C, in an integrated demonstration plant that was commissioned and operated in Monterrey, Mexico. The plant, with a design capacity of 1 t/d of copper, produced more than 40 t of LME A-grade cathode copper from a complex polymetallic concentrate in which the main copper mineral was chalcopyrite6. In 2003, Alliance Copper, a joint venture between BHP Billiton and Codelco, demonstrated the thermophilic bioleaching of chalcopyrite concentrate, at an operating temperature of 78 °C in a prototype plant situated at Chuquicamata in Chile, which was designed to produce 20 000 t/a of copper cathode9. A unique feature of this prototype plant is that the air supply is supplemented with pure oxygen, which necessitates the use of an automated system to control the dissolved oxygen concentration in the process. In the past 20 years, the bioleaching of a wide variety of base metal concentrates has been demonstrated. In the past decade, the development has focused mainly on the bioleaching of chalcopyrite, and the investigations have been aimed either at complex polymetallic concentrates containing several sulphide minerals (the Mintek/BacTech/Peñoles project), or copper concentrates that contain impurities such as arsenic (the Alliance Copper operation). The Aguablanca Mine The Aguablanca Mine is an open-pit nickel-copper sulphide mine, located in the province of Badajoz, about 80 km north of Seville in southern Spain (Figure 1a). The mine currently produces a bulk, low-grade copper-nickel concentrate for sale to a smelter. Aguablanca was the first nickel sulphide mine in production in Western Europe. Construction of the on-site treatment plant (Figure 1b) was completed in December 2004, commissioning and production commenced early in 2005, and production parameters were reached in early 2006. The project has an initial projected mine life of 10½ years. The on-site flotation plant was designed to treat 1.5 million tonnes of ore per year, with an additional milling capacity of 0.3 million tonnes per year. The plant produces a bulk coppernickel-PGM concentrate, but also has the flexibility to produce separate copper and nickel concentrates. The process flowsheet comprises primary jaw crushing, primary semiautogenous grinding, secondary ball milling, gangue pre-flotation, flotation, concentrate thickening and filtration. Production data for the first two full years of operation are summarized in Table I. INTEGRATED PILOTING OF A THERMOPHILIC BIOLEACHING PROCESS

121

119-154_Neale:text

2/14/09

11:48 AM

Page 122

Figure 1a. Location of the Aguablanca Mine in southern Spain

Figure 1b. The Aguablanca treatment plant Table I Aguablanca Mine production data

Ore mined (tonnes) Ore milled (tonnes) Ore grade Nickel (%) Copper (%) Flotation recovery Nickel (%) Copper (%) Concentrate grade Nickel (%) Copper (%) Production (metal contained) Nickel (tonnes) Copper (tonnes)

122

2006 1 550 437 1 486 800

2007 1 707 330 1 668 959

0.6 0.5

0.5 0.4

72 90

76 92

6.6 6.8

7.3 6.9

6 398 6 616

6 630 6 281


119-154_Neale:text

2/14/09

11:48 AM

Page 123

The concentrate is transported either to Glencore’s Huelva smelter, located 140 km away on the Spanish coast, or to other smelters, for treatment. Cash costs (including by-product credits) in 2007 amounted to US$7.23 per pound of nickel sold10. The mineralization of the Aguablanca deposit is similar in type to both the Voisey’s Bay deposit in eastern Canada and the Norilsk deposit in Russia. Nickel, copper, platinum and palladium mineralization occurs within magmatic breccia bodies that form gossans at the surface. Pyrrhotite, pentlandite and chalcopyrite comprise the dominant sulphide mineralization. Following the acquisition of the operation by Lundin Mining Corporation (Lundin) in 2007, encouraging exploration results were announced, indicating the existence of a deep orebody below the existing one. The most recent exploration results have confirmed the potential for a significant increase in the mine’s resource base and for an expansion of the current open pit production to include underground reserves. In March 2008, Lundin announced that a feasibility study was being conducted on undertaking underground mining, which was expected to be completed towards the end of 2008. If approved, a plant expansion project will be finalized to cater for the processing of underground ore. Objectives and scope of work The proposition underpinning this work is to extend the Aguablanca treatment plant by the addition of a bioleaching and metals-refining facility, allowing the on-site production of metal or metal intermediate products, and thus substantially increasing the realized metal value. The selection of bioleaching for treatment of Aguablanca concentrate may have certain advantages over alternative hydrometallurgical processes such as pressure leaching: • The Aguablanca operation is relatively small, and bioleaching may be more suitable for this scale of operation than pressure leaching, which has relatively high capital and operating costs and has been shown to be more cost-effective at larger scales11. • Bioleaching is also readily able to treat low-grade concentrates, and one option that could be considered if bioleaching was to be implemented is to produce an even lower-grade concentrate than is presently the case, in order to improve the overall recovery of metal (particularly nickel) in the concentrate. • If plant expansions were implemented at Aguablanca in the future, the modular nature of bioleach tanks would make it possible to expand the plant with relative ease. Extensive bench-scale amenability and pilot-plant testing of the Aguablanca concentrate was performed at Mintek. Three phases of testing were completed. In phase one, the concentrate sample used in the test work was subjected to chemical, physical and mineralogical characterization. Amenability testing comprised open-circuit bioleach testing in small-scale, continuously operated bioleach reactors, with process volumes of a few litres. The results of these tests were used to define the bioleach operating parameters for the next two phases of test work. In the second phase, a continuous bioleach miniplant was operated in open circuit, using the bioleach operating parameters defined in phase one. The product from this miniplant was used to carry out bench-scale tests aimed at assessing the operating parameters for solution purification and metals recovery. The results of these tests were used to define the initial operating parameters for the final phase of test work. In the third and final phase, an integrated pilot plant comprising bioleaching, solution purification and metals recovery was operated for a period of over seven months. During this integrated piloting campaign, seven cycles of downstream processing were completed, and integration was achieved by recycling of the purified process solution. The primary aims of INTEGRATED PILOTING OF A THERMOPHILIC BIOLEACHING PROCESS

123

119-154_Neale:text

2/14/09

11:48 AM

Page 124

the integrated piloting campaign were to assess the impact, if any, of recycling of the purified process solution on the bioleach process performance, to optimize the operating parameters for the purification and metals recovery circuit, and to define the process design criteria to be used as the basis for a conceptual engineering study and preliminary economic evaluation of the process. In this phase, additional optimization test work was also conducted to assess the effect of bioleach process parameters such as redox potential, concentrate grind size and feed solids concentration on copper and nickel extractions and leach kinetics. The results of these additional optimization tests were included in the engineering and economic study. The outcomes of this comprehensive bench-scale and pilot-plant test work programme are discussed here, together with the results of the engineering and economic study. Conceptual flowsheet development At the outset, a conceptual process flowsheet for bioleaching and the recovery of metals from the Aguablanca concentrate was envisaged. The conceptual flowsheet aimed to address the following key issues: • Solid-liquid separation and washing of the bioresidue, wastes and metal products • Removal of iron from the leach liquor • Purification of the leach liquor • Recovery of copper from the leach liquor • Recovery of nickel from the leach liquor • Waste products and their disposal • Re-use of water • The acid balance. Solid-liquid separation and washing of the residues, wastes and products is usually achieved either by countercurrent decantation (CCD) using a series of thickeners, or by filtration. The choice depends on many factors, including the settling and filtration characteristics of the solid materials, the required washing efficiency, and the availability of water. The removal of iron from leach liquors such as these is usually achieved by precipitation using limestone, at a pH level of around 3.0. The aim of this process step is to remove iron prior to copper solvent extraction, without coprecipitating or entraining the valuable metals (nickel and especially copper), and in so doing to produce a waste product which has good filtration or settling characteristics. The factors that govern this process include the operating temperature, the overall residence time, the number of precipitation stages and the rate at which the pH level is raised, and the final pH level. The recovery of copper from bioleach liquors is usually achieved by the well-established method of solvent extraction (SX) using a copper-selective organic reagent. The SX process selectively removes copper from the pregnant solution, and produces a purified copper solution from which copper can be extracted as a cathode using electrowinning (EW). Important factors that govern this process are the number of extraction and stripping stages, the organic reagent that is applied, and the pH level and copper tenor of the pregnant solution. A few variations for the residue washing, iron removal and copper SX-EW operations were considered, including: • Precipitation of the iron directly after bioleaching, without first separating the pregnant leach liquor from the bioresidue. This would remove one solid-liquid separation and washing step from the process, but it would have to be established that the resultant solids, comprising the bioleach residue, iron precipitates and gypsum, could be thickened or filtered easily. 124


119-154_Neale:text

2/14/09

11:48 AM

Page 125

• Performing the copper SX step prior to removing the iron from the leach liquor. This would eliminate the possibility of encapsulation or coprecipitation of copper in the iron precipitate, but it would have to be established that the iron in the SX feed did not affect the extraction of copper. Purification of the SX raffinate is usually required prior to nickel recovery. The SX process reduces the pH level of the liquor, and this purification is usually achieved by further neutralization of the liquor by limestone, to a pH level of about 5.0. The objective is to remove all impurities that may contaminate the nickel product. There are several options that could be considered for the recovery of nickel from the leach liquor. In the nickel laterite industry, which has seen the implementation of hydrometallurgical nickel extraction and recovery processes in a number of projects in the past decade, several different approaches have been followed. These include precipitation to form a metal salt, precipitation to produce a metal sulphide, and SX-EW, usually after the formation and redissolution of an intermediate product (either a salt or a sulphide), to produce metal. After a careful evaluation of the various alternatives, it was decided that the integrated pilot plant would incorporate the metal salt precipitation route, for the following reasons: • It is conceptually the simplest route to follow. • Nickel SX-EW is not as well established as copper SX, and the process circuits are complex, difficult to operate, with high capital and operating costs. • The production of an intermediate metal salt does not preclude the option of producing metal by SX-EW. The metal salt could be redissolved and the solution used as a feed to a refinery including nickel SX-EW. Such a refinery could be added at a later stage; this would reduce the initial technical risk associated with the project.

Figure 2. Conceptual flowsheet for the Aguablanca nickel-copper concentrate

INTEGRATED PILOTING OF A THERMOPHILIC BIOLEACHING PROCESS

125

119-154_Neale:text

2/14/09

11:48 AM

Page 126

One of the objectives of the bench-scale test work programme conducted prior to the integrated piloting campaign was to evaluate which metal salt should be produced as a nickel intermediate. The final step in this process (in which a nickel salt is produced) would involve a partial or total neutralization of the barren liquor to remove any residual impurities that remain or that may have been added during the nickel precipitation process, prior to recycling of the liquor to the front end of the circuit. This conceptual flowsheet is shown in Figure 2. In this case, magnesia is used as the alkali in the nickel precipitation process, and the product is nickel hydroxide. For this process flowsheet, the soluble magnesium sulphate produced in the nickel hydroxide precipitation process would need to be removed by the addition of lime in the final neutralization step. Concentrate characterization and bioleach amenability testing Characterization of the Aguablanca concentrate The Aguablanca concentrate was characterized both physically and chemically. The dry density of the Aguablanca concentrate was determined to be 3 292 kg/m3. In all of the test work, the concentrate was reground prior to bioleach testing, and a particlesize distribution with a d90 of 10–12 μm was targeted. The size distributions of the as-received and reground concentrates are illustrated in Figure 3. The typical chemical analysis of the Aguablanca concentrate is summarized in Table II. A scanning electron microscope was used to perform a quantitative modal analysis of the Aguablanca concentrate, and the results are summarized in Table III. This analysis shows that the bulk of the concentrate comprises the sulphide minerals pyrite, pyrrhotite, chalcopyrite and pentlandite, with a variety of silicates such as amphibole, feldspar, talc and chlorite being the predominant gangue minerals.

Figure 3. Typical particle-size distributions for the as-received and reground Aguablanca concentrates

126


119-154_Neale:text

2/14/09

11:48 AM

Page 127

Table II Typical chemical analysis of the Aguablanca concentrate Constituent

Concentration (%) 6.29 5.24 24.4 24.36 28.9 6.28 2.74 1.16 1.00 0.18

Cu Ni STOT S2Fe Si Mg Al Ca Co

Table III Typical modal analysis of the Aguablanca concentrate Mineral groups

Mineral name

Chemical formula

Sulphides

Pyrite

FeS2

23.7

Fe1-x(Ni)S

20.7

Chalcopyrite

CuFeS2

18.5

Pentlandite

(Fe,Ni)9S8

14.5

NiAsS

0.04

CuS

0.03

Sphalerite

Cu2S

0.02

Galena

ZnS

0.02

Amphibole

NaCa2(Mg,Fe)4Al(Si6Al2)O22(OH)2

9.8

Feldspar

CaAl2Si2O8; KAlSi3O8

4.2

Talc

Mg3Si4O10(OH)2

3.2

Chlorite

(Mg,Fe,Al)6(Si,Al)4O10(OH)8

2.2

Ni-bearing silicates

Fe(K,Al,Mg,Ni)-silicate

1.8

Quartz

SiO2

0.3

Fe-oxides

Fe2O3; α-FeOOH; Fe3O4

0.6

Ilmenite

FeTiO3

0.04

Cu-bearing sulphate

Ca(Fe,Cu)-sulphate

0.2

Gypsum

CaSO4.2H2O

0.1

Phosphate

Apatite

Ca5(PO4)3(OH,F,Cl)

0.02

<0.5

PGM species

-

PtPd-BiTe and PtRh-S

<0.01

<0.5

Pyrrhotite

1

Gersdorffite Cu-sulphides

Silicates

Oxides

Sulphates

2

Mass % * 77.4

21.6

0.6

0.4

Including pyrrhotite with and without detectable nickel Mostly covellite but includes chalcocite * Where relevant, two decimal figures are given merely to demonstrate proportional differences between trace minerals 1 2


127

119-154_Neale:text

2/14/09

11:48 AM

Page 128

Bioleach amenability test work The results obtained from amenability test work performed in continuously operated benchscale reactors using mesophile, moderate thermophile and thermophile cultures indicated that nickel extractions of over 98% could be obtained using moderately thermophilic and thermophilic cultures operating at 45 and 70°C, respectively. For copper, which occurs as chalcopyrite, extractions of over 95% could only be achieved using the thermophile culture at 70°C (Table IV). Based on these results, the thermophile culture was selected for further optimization test work. Further optimization of the process focused on determining the effects of key process parameters such as grind size, residence time and feed solids concentrations on the leach kinetics, metal extractions and performance of the thermophile culture. It was shown that the performance of the thermophile culture was negatively affected when operating at feed solids concentrations above 12%, and so a conservative feed solids concentration of 10% was selected. Bioleach tests performed at 70°C at grind sizes ranging between 10 and 35 μm indicated that a nickel extraction of over 97% could be obtained over the range of grind sizes. There was, however, a reduction in copper extraction with increasing grind size, and copper extractions of 95% could only be achieved at a particle size of d90 = 10 μm and an overall sixday residence time (Table V). Based on the results of the amenability test work, a set of bioleach operating parameters was set for the integrated piloting campaign. These are summarized in Table VI. Bench-scale purification and metals recovery test work Phase two of the overall metallurgical test work programme comprised bench-scale testing that evaluated the downstream unit operations: iron removal, copper SX, nickel precipitation, and solution purification. The results of these tests were used to set the initial operating conditions for the third phase of testing—the integrated pilot plant run. Table IV Metal extractions achieved in three-stage continuously operated reactor systems, 6-day residence time, 10 % feed solids concentration, particle size of d90 = 10 μm

Mesophiles (35 °C)* Moderate thermophiles (45 °C) Thermophiles (70 °C) * Single-stage reactor at 3-day residence time

Metal extraction (%) Cu Ni 30 76 65 99 95 99

Table V Summary of bioleach results showing the effect of grind size on metal extractions Operating conditions 3-stage, 70°C, 10 μm, 6 days 3-stage, 70°C, 20 μm, 6 days 1-stage, 70°C, 35 μm, 3 days Batch, 70°C, 35 μm, 6 days R1/2/3 = Reactor 1/2/3

128

Cu extraction (%) R1 R2 R3 83.5 92.0 95.1 72.4 82.0 90.7 52.9 72.8

Ni extraction (%) R1 R2 R3 98.7 99.3 99.4 96.3 97.4 97.4 87.7 98.1


119-154_Neale:text

2/14/09

11:48 AM

Page 129

Table VI Bioleach operating parameters for Aguablanca nickel-copper concentrate Operating conditions Microbial culture Operating temperature (°C) Feed solids concentration (%) Feed grind size (d90) (μm) Overall residence time (d) Operating Eh level (mV vs. Ag|AgCl) Operating pH level Oxidation level and metal extractions Sulphide oxidation (%) Copper extraction (%) Nickel extraction (%) Solution metal tenors Iron concentration (g/l) Copper concentration (g/l) Nickel concentration (g/l)

Thermophiles 70.0 10.0 10 6.0 600 1.3–1.1 99.0 95.1 99.4 26.0 8.0 6.5

In order to provide sufficient samples for the bench-scale test work, a continuous bioleach miniplant was operated in open circuit treating a sample of Aguablanca concentrate. The product from this plant was used to undertake a series of batch bench-scale tests. These tests aimed to assess each of the proposed unit operations, and in particular to evaluate various options for the precipitation of iron and nickel, and to carry out an initial assessment of the solid-liquid separation characteristics of the products from each process. It was recognized that optimization of the precipitation processes can only be undertaken during continuous testing, where the impact of steady process operating conditions, and the use of recycle streams for seeding of the precipitation reactions, can be assessed. These factors are likely to have a direct and beneficial impact on the rate and extent of metal precipitation that can be achieved in practice, and on the settling and/or filtration characteristics of the resulting precipitates. The aim of these tests, therefore, was merely to provide the initial operating parameters for the integrated piloting campaign. Table VII contains a summary of the initial operating parameters that were selected for each unit operation. Integrated piloting test work Integrated pilot plant The pilot plant that was used for the integrated piloting tests is comprised of three main units: a bioleach plant, a precipitation plant, and a solvent extraction plant. The bioleach plant was operated continuously for the duration of the campaign. However, it was not possible to operate the precipitation and solvent extraction plants continuously throughout the campaign, because of the wide disparity between the retention time in the bioleach plant and in the subsequent purification and metals recovery processes. To overcome this, the product from the bioleach plant was collected over a period of several weeks, and then used to feed the precipitation and solvent extraction units—which were then operated continuously, but for short periods. The same precipitation plant, comprising five equal-volume reactors arranged in series, was used for each of the successive precipitation processes. INTEGRATED PILOTING OF A THERMOPHILIC BIOLEACHING PROCESS

129

119-154_Neale:text

2/14/09

11:48 AM

Page 130

Table VII Initial operating parameters for integrated piloting of Aguablanca concentrate Primary iron precipitation Temperature (°C) Final pH level

60.0 3.0

Final iron tenor (g/l)

0.45

Iron removal (%)

99.4

Limestone consumption (kg/t liquor) Precipitate settling rate (m/h)

25.0 0.0382

Precipitate underflow density (%)

27.6

Calculated thickener area (m2/(t/h))

69.1

Copper solvent extraction Organic extractant

LIX®984N-C

Diluent

ShellSol 2325

Extractant concentration (%)

25

Organic-to-aqueous ratio

0.8:1

Final copper tenor (g/l)

0.163

Copper extraction (%)

97.5

Secondary iron precipitation Temperature (°C) Final pH level

60.0 5.0

Final iron tenor (g/l)

0.05

Iron removal (%)

98.3

Limestone consumption (kg/t liquor) Precipitate settling rate (m/h) Precipitate underflow density (%) Calculated thickener area (m2/(t/h))

13.7 6.0-7.8 20-24 2.0-4.3

Nickel precipitation Temperature (°C) Final pH level Final nickel tenor (g/l) Nickel removal (%)

60.0 7.8 ~0.2 ~97.1

Product nickel grade (%)

24.2

Magnesia consumption (kg/t liquor)

18.4

Precipitate settling rate (m/h) Precipitate underflow density (%) Calculated thickener area (m2/(t/h))

9.3 12.8 4.6

Final (magnesium) precipitation Temperature (°C) Final pH level Final magnesium tenor (g/l)

60.0 10.0 0.0062

Magnesium removal (%)

99.5

Product magnesium grade (%)

12.7

Lime consumption (kg/t liquor)

4.65

Precipitate settling rate (m/h) Precipitate underflow density (%) Calculated thickener area (m2/(t/h))

130

3.0 7.0 22.7


119-154_Neale:text

2/14/09

11:48 AM

Page 131

The integrated pilot plant is illustrated in Figures 4 and 5. The integrated pilot plant was operated for a period of 220 days, during which seven downstream solution purification and metals recovery cycles were completed. Each downstream cycle consisted of five additional unit operations, carried out in sequence. The product from each unit operation was stored, and used as the feed for the next unit operation in the sequence. The five unit operations were: • Primary iron precipitation, in which the bioleach pulp was neutralized with limestone • Copper SX, treating the pregnant leach solution after primary iron removal • Secondary iron precipitation, in which the SX raffinate was neutralized with limestone

Figure 4. The integrated pilot plant: the solvent extraction plant is on the left, the bioleach pilot plant is in the centre, and the precipitation plant is on the right

Figure 5a. The integrated plant’s bioleach reactors, with the solvent extraction plant in the background: on the right is the feed tank, the dark grey vessel is the 60litre primary reactor, with three 20-litre reactors arranged in series to the left of that

Figure 5b. The integrated plant’s neutralization reactors, comprising a feed tank (hidden) and five 20litre reactors arranged in series; in the foreground are various settlers that were used during the operation of this plant


131

119-154_Neale:text

2/14/09

11:48 AM

Page 132

• Nickel precipitation, in which nickel hydroxide was precipitated from the purified solution • Magnesium precipitation, to remove excess magnesium from the circuit. At the completion of each cycle, the barren solution was recycled to the bioleach plant, by using it as the make-up water in the bioleach feed tank. In this way, full integration of the process was achieved. The use of recycled barren process solution for bioleach feed dilution was the only connection between the bioleach process and the downstream solution purification and metals recovery cycles. Bioleach plant performance The operation of the bioleach pilot plant is illustrated in Figures 6 to 12, which contain graphs of the routine data for several of the measured operating parameters. In the early stages of the pilot-plant campaign, it was observed that the measured feed solids concentration was below the target value of 10%, as shown in Figure 6. This was despite the fact that the feed constituents (concentrate, nutrients and dilution water) were carefully metered to target this feed solids concentration. At first, it was suspected that this may be due to inadequate agitation in the feed tank, but even after this was addressed (by installing a larger impeller in the feed tank), the measured feed solids concentration remained below the

Figure 5c. The integrated pilot plant’s solvent extraction units; this versatile plant has eight mixer-settlers that can easily be configured to provide the required number of extraction, washing and stripping stages

Figure 6. Bioleach feed solids concentrations during the integrated piloting campaign

132


119-154_Neale:text

2/14/09

11:48 AM

Page 133

target value. Inspection of the feed material revealed that, during the drying process after regrinding, the concentrate had formed persistent agglomerates that that did not break up in the feed tank. It is possible that residual flotation chemicals associated with the concentrate may have caused this. These large particles remained in the lower region of the feed tank, and were preferentially pumped into the primary bioleach reactor. Thus, when the solids concentration in the feed tank was measured just before refilling the tank, the measured value was below the target value. To overcome this, the concentrate was given a light regrind prior to use, in order to break up the coarse agglomerates.

Figure 7. Measured pH levels during the integrated piloting campaign

Figure 8. Measured Eh levels during the integrated piloting campaign

Figure 9. Measured oxygen uptake rates levels during the integrated piloting campaign


133

119-154_Neale:text

2/14/09

11:48 AM

Page 134

The rate of evaporation from the bioleach reactors was high, since they were being operated at a temperature of 70°C. At this scale of operation, the ratios of surface area to tank volume and of aeration rate to tank volume are very high. This results in much higher rates of evaporation than would be experienced in a production-scale process. To counter this, dilution water was pumped into each reactor to replace the water lost through evaporation. The rate of evaporation was not constant, and was affected by seasonal variations in the ambient humidity. Consequently, step changes were occasionally made to the rate of dilution water addition, which explains why the soluble metal concentrations shown in Figures 10 to 12

Figure 10. Measured soluble iron concentrations during the integrated piloting campaign

Figure 11. Measured soluble copper concentrations during the integrated piloting campaign

Figure 12. Measured soluble nickel concentrations during the integrated piloting campaign

134


119-154_Neale:text

2/14/09

11:48 AM

Page 135

display several sudden changes during the course of the campaign. This had no material impact on the process performance, as evidenced by the stable redox potential levels (Figure 8) and oxygen uptake rates (Figure 9) for most of the campaign. The bioleach plant operation was found to be stable and robust, and capable of withstanding upsets in plant operation. At various times, power, heater and compressed air failures were experienced, but the process recovered swiftly from these disturbances. Four mass balance sampling campaigns were performed over the bioleach section of the plant during the integrated pilot plant trial, allowing a detailed set of performance parameters to be calculated. The results of these mass balances were broadly similar. The results of one of these, performed on day 92, are summarised in Table VIII. The results of the mass balances indicated that the target nickel and copper extractions of 99.4% and 95.1% were consistently achieved or exceeded. The target sulphide oxidation level of 99.0% was also consistently achieved. Microbial identification The microbial populations present in the bioleach reactors were identified and quantified using Q-PCR and T-RFLP techniques, respectively (Tables IX and X). Acidianus sp. was identified as the most abundant organism in all four stages, with Metallosphaera sp. and Sulfolobus sp. present in much lower numbers. Additional bioleach optimization test work Additional bench-scale bioleach test work was conducted to study the effect of redox potential, grind size and feed solids concentration on copper and nickel extractions and bioleach kinetics. Table VIII Results of bioleach mass balance conducted on day 92 Reactor number Solids concentration % Feed flow rate (l/d) Product flow rate (l/d) Residence time (d) Temperature (°C) pH level Redox (mV, Ag/AgCl) [Fe] (g/l) [Ni] (g/l) [Cu] (g/l) Fe extraction (%) Fe precipitation (%) Ni extraction (%) Ni precipitation (%) Cu extraction (%) Cu precipitation (%) S2- extraction (%) [S°] (%)

1 8.6 20.2 19.8 3.03 70 1.35 599 19.4 6.2 5.2 65.6

Soluble analysis 2 3

4

4.04 70 1.33 667 19.8 6.3 5.6 64.7

5.06 70 1.34 685 19.3 6.2 5.8 66.7

6.07 70 1.35 687 17.9 5.8 5.6 65.9

97.0

97.7

99.6

99.3

78.4

86.7

92.7

95.0

93.8 0.2

98.2 0.3

98.9 0.26

99.4 0.23

1

Total (HCI-washed) analysis 2 3

92.5 29.1 97.5 0.46 81.7 3.94


92.6 30.1 99.4 1.69 87.8 1.24

94.3 29.3 99.7 0.17 93.0 0.31

4

95.8 30.4 99.8 0.21 96.5 3.92

135

119-154_Neale:text

2/14/09

11:48 AM

Page 136

Although the most common means of overcoming the slow and incomplete extraction of copper from chalcopyrite is to operate the process at high temperature using thermophiles, an alternative approach is to operate at controlled redox potential levels. Previous test work has indicated that faster copper leach kinetics can be achieved by maintaining the redox potential at levels between 410 and 440 mV vs. Ag|AgCl, although a reduction in the rate of nickel extraction was observed at redox potential levels below 460 mV vs Ag|AgCl. Based on these results, a three-stage continuously operated reactor system was operated at 70°C and an overall 6-day residence time, using Aguablanca concentrate milled to a particle size of d90 = 10 μm. The redox potential in the first-stage reactor was maintained at 430 mV vs. Ag|AgCl, while the redox levels in the secondary stages were not controlled. Redox potentials of 550 and 580 mV vs. Ag|AgCl were measured in the second- and third-stage reactors, respectively. The results of this test are summarized in Table XI. This test confirms that, by controlling the redox potential in the first-stage reactor, considerably faster copper leach kinetics can be achieved. Around 96% of the copper was extracted in the first-stage reactor, compared to an extraction of 78% without redox control. The overall copper extraction improved from 95 to 98%. Furthermore, the high redox potential levels maintained in the secondary reactors ensured that a nickel extraction of over 98% was achieved in the second-stage reactor. This result indicates that operating the first-stage reactor at a low redox potential level allows a reduction in the overall bioleach residence time from six to four days. In a second test, the effect of concentrate grind size on the copper and nickel extractions was evaluated using concentrate milled to a d90 of 20 μm. As before, the redox potential in the first-stage reactor was maintained at a level of 430 mV vs. Ag|AgCl, while the redox potential levels in the secondary reactors were not controlled. As before, redox potential levels of 550 and 570 mV vs. Ag|AgCl were measured in the second- and third-stage reactors, respectively. The results of the second test are summarized in Table XII. Despite the coarser feed material, a copper extraction of around 92% was achieved in the first-stage reactor, and an overall copper extraction of 96% was achieved. The high redox potential levels in the Table IX Quantification of the microorganisms using Q-PCR Microorganism Acidianus sp. Metallosphaera sp. Sulfolobus sp. ND: Not detected

Reactor 1 2.7 x 109 5.4 x 107 7.6 x 104

Cell concentrations (cells/ml) Reactor 2 Reactor 3 6.1 x 109 4.5 x 109 8 1.3 x 10 8.4 x 107 4 9.9 x 10 6.4 x 104

Reactor 4 3.2 x 109 5.4 x 107 ND

Table X Relative abundance of the microbial populations occurring in the bioleach reactors Sample Reactor 1 Reactor 2 Reactor 3 Reactor 4 *ND: not detected

136

Acidianus sp. (%) Q-PCR T-RFLP 98.0 78 97.9 85 98.2 89 98.3 92

Metallosphaera sp. (%) Q-PCR T-RFLP 1.9 22 2.0 15 1.8 11 1.7 8

Sulfolobus sp. (%) Q-PCR T-RFLP 0.003 ND 0.002 ND 0.001 ND 0 ND


119-154_Neale:text

2/14/09

11:48 AM

Page 137

secondary reactors ensured that a nickel extraction of almost 98% was achieved. This result suggests that, with a coarser feed material and by controlling the redox potential in the firststage reactor, the overall bioleach residence time can be reduced from six to five days. Subsequent testing was conducted in which the feed solids concentration was raised to 12%, and it was found that stable operation could be maintained at this solids concentration. Integrated purification and metals recovery tests As mentioned previously, seven cycles of operation of the integrated purification and metals recovery circuit were completed during the integrated pilot plant campaign. Table XI Effect of controlled low redox potential on bioleach process performance

Microbial culture Temperature (°C) pH level Redox (mV, Ag|AgCl)) Cumulative residence time (d) Feed solids concentration (%) Grind—d90 (μm)

Feed Thermophiles

Reactor 1

Reactor 2

Reactor 3

70 1.6 430 3

70 1.3 550 4.5

70 1.25 580 6

14.5 7.1 6.7 48.4 96.2 93.8 67.3

17.4 7.0 6.9 68.6 97.8 98.4 94.2

18.0 7.0 7.0 75.6 98.1 98.7 98.8

9.8 10

[Fe] (g/l) [Cu] (g/l) [Ni] (g/l) Fe extraction (%) Cu extraction (%) Ni extraction (%) S2- extraction (%)

Table XII Effect of grind size and controlled low redox potential on bioleach process performance

Microbial culture Temperature (°C) pH level Redox (mV, Ag|AgCl)) Cumulative residence time (d) Feed solids concentration (%) Grind—d90 (μm) [Fe] (g/l) [Cu] (g/l) [Ni] (g/l) Fe extraction (%) Cu extraction (%) Ni extraction (%) S2- extraction (%)

Feed Thermophiles

Reactor 1

Reactor 2

Reactor 3

70 1.7 430 3

70 1.4 550 4.5

70 1.3 570 6

13.6 7.3 6.7 40.7 92.7 85.1 63.7

16.0 7.4 7.0 63.1 95.8 96.9 84.7

19.2 7.6 7.5 70.1 96.0 97.9 94.2

9.4 20


137

119-154_Neale:text

2/14/09

11:48 AM

Page 138

Primary iron precipitation During the commissioning of the plant, it was established that it was possible to conduct the primary iron precipitation on the bioleach product pulp, without first having to separate and wash the bioresidue. Since this removed a solid-liquid separation and washing operation from the flowsheet, it was decided to proceed on this basis for the remainder of the integrated piloting campaign. The primary iron precipitation process was operated at a variety of temperatures between 45 and 70°C. The pH level in the first-stage reactor was maintained at a level of between 1.8 and 2.9, and the final pH was maintained at a level of between 3.0 and 3.5. The neutralizing agent was limestone. Recycling of the product was used to promote seeding in the first reactor. Various recycle ratios were used. It was found that the product pulp from the primary iron precipitation process was very difficult to settle, irrespective of the operating temperature and the recycle ratio. It was also found that, as the operating temperature varied, the settling characteristics changed to such an extent that different flocculants were required to achieve the best result. Although some flocculants achieved good settling rates, it was observed that the product liquor was turbid and unsuitable as a feed to the copper solvent extraction plant. The clarity of the settler overflow was improved considerably by the addition of a coagulant. It was shown that the coagulant could be used on its own or in conjunction with a neutrally-charged flocculant. However, large coagulant doses of over 1 kg/t were required, compared with around 3–30 g/t for flocculants. The best set of results resulted in a batch settling rate of 282 mm/h. In this run, the settler overflow was clear and relatively small doses of coagulant and flocculant were required. Copper solvent extraction The copper SX plant was a mixer-settler unit operated using a mixture of 25% LIX®984N-C in ShellSol 2325 as the organic extractant. LIX®984N-C, which is manufactured by Cognis, is a reagent that has been developed especially for use with pregnant liquors containing high copper concentrations, and ShellSol (made by Shell Chemicals) is a paraffin-based hydrocarbon solvent. The sulphuric acid concentration in the strip liquor was maintained at 180 g/l. The flow rate through the SX plant is determined by a required minimum residence time of three minutes in the mixer, taking account of both the organic and aqueous phases. The second parameter is the loading capacity of the organic, specified by the manufacturer as being 0.3 g/% or 7.5 g/l copper for a 25% solution of LIX®984N-C. This, together with the copper concentration in the pregnant leach solution (PLS) determines the organic-to-aqueous (O:A) ratio. The strip liquor flow rate is determined by the electrowinning section, where a reduction of 15 g/l copper is required. In the stripping section of the SX plant, this quantity of copper must be replaced, thus setting the organic-to-strip (O:S) ratio. In this instance an O:S ratio of two is typical. Provided that the O:A ratio was maintained above about 0.7:1, copper extractions between 98.2 and 98.9% were achieved using three extraction stages. Copper electrowinning At the completion of the pilot plant campaign, a sample of the strip liquor (which had built up to a copper concentration of around 50 g/l) was used to perform an electrowinning test, using a standard set of operating conditions. Two small copper cathodes (each with an area of 138


119-154_Neale:text

2/14/09

11:48 AM

Page 139

0.0315 m2) were produced in this test. They were characterized as being smooth, compact and bright. One cathode was drilled and submitted for chemical analysis, to determine whether the copper cathode meets the requirements for LMA A-grade copper. Table XIII summarizes the results. The detection limits for the analyses of the elements P and S are higher than the threshold values. The measured values of the other elements were all well below the LME A-grade threshold values, and the sum (excluding P and S) of 16.07 ppm for these elements is well below the threshold value of 65 ppm. The LME A-grade specification also provides upper limits for several elements in combination, and once again the measured values (excluding P) are well below the threshold limits. These results indicate that, unless the cathode contains over 15 ppm of either S or P, it will easily satisfy the requirements for LME A-grade classification. Secondary iron precipitation The purpose of the secondary iron precipitation process was to reduce the concentration of iron and other metals in the pregnant liquor as far as possible prior to the precipitation of nickel hydroxide, in order to reduce the level of contamination of the nickel hydroxide precipitate by these elements. This was achieved by raising the pH of the liquor to a level of around 5.0 using limestone, at a temperature of 60°C. Recycling of the product was used to promote seeding in the first reactor. Various recycle ratios were used. Table XIII Aguablanca copper cathode analysis

Element(s) Ag As Bi Cd Co Cr Fe Mn Ni P Pb S Sb Se Si Sn Te Zn Sum As+Cd+Cr+Mn+P+Sb Bi+Se+Te Se+Te Co+Fe+Ni+Si+Sn+Zn

LME A grade threshold (ppm) 25 5 2 10 5 15 4 2 2 65 15 3 3 20


Measured value (ppm) 0.17 0.00 0.53 0.00 0.00 0.00 0.00 0.50 0.00 <20 0.00 <100 0.00 1.17 12.67 0.50 nd 0.53 16.07 0.50 1.70 1.17 13.70

139

119-154_Neale:text

2/14/09

11:48 AM

Page 140

The product from this process was mostly gypsum, and the product pulp had a very low solids concentration (of between about 2 and 4 %). The gypsum particles in the product solids were large, and were easily settled. Copper and nickel scavenging There was some loss of both copper and nickel to the precipitated solids in the secondary iron removal process. Gravity separation of the product was investigated as a means of upgrading the copper and nickel. Three options were evaluated: a shaking table, a Falcon concentrator, and a cyclone. The shaking table was ineffectual, with around 90% of the copper and nickel reporting to the slimes, which comprised over 60% of the mass. The same was true of the Falcon concentrator, where over 99% of the copper and nickel reported to the tailings, which comprised over 90% of the mass. The cyclone proved effective in upgrading the copper and nickel. Over 75% of the copper and over 80% of the nickel could be recovered in the cyclone overflow, at a mass recovery of 18.4%. It is likely that further optimisation of this process, with a slightly coarser cut and with underflow recycle to scavenge the tails, would further improve this result. The cyclone overflow would be recycled back to the bioleach process to redissolve the copper and nickel. Nickel precipitation Nickel precipitation to produce nickel hydroxide was achieved by raising the pH of the liquor to a level of around 7.8 using magnesia, at a temperature of 60°C. Recycling of the product was used to promote seeding in the first reactor. Various recycle ratios were used. The objective of this process was to produce a nickel hydroxide product of the highest possible purity, with reasonable filtration characteristics, and to ensure a high recovery of nickel. This presents a challenge: a good nickel recovery is obtained at pH levels in the region of 7.7 to 8.0, but as the pH increases above a level of about 7.5, the product becomes increasingly contaminated with magnesium—either in the form of unreacted magnesia or precipitated magnesium hydroxide. One approach that has been adopted is to conduct a two-stage nickel hydroxide precipitation. The first-stage process is operated at a target pH level of around 7.2, and aims to recover around 90% of the nickel and produce a high-grade nickel hydroxide product. The second-stage process is operated at a pH level of between 7.5 and 8.0, and aims to recover the remainder of the nickel in a lower-grade nickel hydroxide precipitate. This product can either be recycled to the leach process or sold at a lower cost. A second approach is to conduct the precipitation in multiple stages with careful control of the pH level throughout the process, to ensure that both targets—recovery and product purity —are met. This approach, if achievable, has the advantage of reducing the number of unit operations required, and producing a single high-grade nickel hydroxide product. It was therefore decided to aim for a single-stage nickel hydroxide precipitation process. In the piloting campaign, the solids were settled in order to provide a recycle stream for seeding purposes. In practice, the nickel hydroxide product will be filtered to minimise the water content. It was observed that nickel hydroxide precipitation required a considerable amount of seeding to initiate the precipitation process—more so than the iron removal steps. The product was observed to flocculate well, although the overflow remained slightly turbid due to the presence of very small particles in suspension. When part of the settler underflow was recycled for seeding, nickel recoveries of well over 99% were achieved. 140


119-154_Neale:text

2/14/09

11:48 AM

Page 141

It was also noted that a fresh magnesia slurry was required for the best performance—it was found that when an older magnesia slurry was used, the product nickel grade was lower. The nickel content of the nickel hydroxide precipitate varied between 31.6 and 47.5%. This variance reflects the fact that the operating conditions in the nickel recovery process were changed in an effort to understand the impact of the operating conditions on the process. Based on the results, it is considered that a target nickel grade of between 40 and 45% can be achieved using a single-stage precipitation process. Magnesium precipitation The final step in the process flowsheet is magnesium precipitation, which was achieved by raising the pH of the liquor to a level of around 9.5 using slaked lime, at a temperature of 60°C. Recycling of the product was used to promote seeding in the first reactor. Various recycle ratios were used. In some cases, the product was settled prior to recycling, whereas in others the recycle stream comprised the product pulp. The objective of this process step was to remove magnesium and most other contaminants from the stream, to enable it to be recycled to the front of the process. The treatment of a bleed stream could be considered, provided that the presence of some magnesium could be tolerated in the bioleach process. Bioleach amenability tests had shown that up to 10 g/l of magnesium could be tolerated, but there were concerns that the presence of high concentrations of magnesium in the iron removal, copper solvent-extraction and nickel hydroxide processes would be undesirable. Therefore, it was decided to process the entire stream through the magnesium precipitation process. It was observed that the magnesium product flocculated relatively easily, but that the overflow contained some fine particles that did not settle. Upgrading of magnesium precipitate An attempt was made to upgrade the magnesium precipitate using a cyclone, in order to assess whether the upgraded product might be used in the nickel hydroxide precipitation process. This test was only partially successful, in that around 40% of the magnesium was recovered in the cyclone overflow, which had a mass recovery of 11.5%. The upgraded product had a magnesium grade of 22.2%, but also contained 12.5% calcium, indicating that gypsum had also reported to the overflow. If this was to be used in the nickel precipitation process, the gypsum would contaminate the nickel hydroxide product, and it is therefore not proposed to utilize this upgraded product in the nickel precipitation process. Summary of integrated piloting tests The results of the integrated piloting tests were used to establish the basic process design criteria for the proposed bioleaching and metals recovery plant for the treatment of the Aguablanca concentrate. The design criteria are summarized in Table XIV. The piloting programme demonstrated that integration of the bioleaching process with the downstream purification and metals recovery processes had no impact on the bioleach process performance. Therefore, the design criteria for the bioleaching process remained unchanged from those obtained in the amenability tests. These basic process design criteria were used as the basis for the process design and costing of the commercial-scale plant. Process design and costing The flowsheet and data derived from the integrated piloting of the Aguablanca nickel-copper concentrate were used to develop a process design and costing model for a hydrometallurgical INTEGRATED PILOTING OF A THERMOPHILIC BIOLEACHING PROCESS

141

119-154_Neale:text

2/14/09

11:48 AM

Page 142

Table XIV Basic process design criteria based on integrated piloting results Bioleaching Bacterial culture Operating temperature (°C) Feed solids concentration (%) Feed grind size (d90) (μm) Overall residence time (d) Operating Eh level (mV vs. Ag|AgCl) Operating pH level Sulphide oxidation (%) Copper extraction (%) Nickel extraction (%) Primary iron precipitation Temperature (°C) Number of stages Residence time per stage (h) Final pH level Final iron tenor (g/l) Iron removal (%) Limestone consumption (kg/m3 slurry) Thickener underflow density (%) Calculated thickener area (m2/(t/h)) Copper solvent extraction Temperature (°C) Organic extractant Diluent Extractant concentration (%) Organic-to-aqueous ratio Number of extraction stages Number of scrubbing stages Number of stripping stages Final pH level Final copper tenor (g/l) Copper extraction (%) Secondary iron precipitation Temperature (°C) Number of stages Residence time per stage (h) Final pH level Final iron tenor (g/l) Iron removal (%) Limestone consumption (kg/m3 slurry) Filter cake moisture (%) Filtration rate (m2/(t/h)) Nickel precipitation Temperature (°C) Number of stages Residence time per stage (h) Final pH level

142

Thermophiles 70.0 10.0 12 6.0 650 1.4-1.2 99.0 95.1 99.4 60.0 5 1 3.0 0.09 99.5 60 30-40 10-20 35 LIX®984N-C ShellSol 2325 25 1:1 3 2 3 1.1-1.2 0.1 97.0 60.0 5 1 5.0 0.00 95.0 12.5 46.5 62.4 60.0 5 1 7.8


119-154_Neale:text

2/14/09

11:48 AM

Page 143

Table XIV (continued) Basic process design criteria based on integrated piloting results Nickel precipitation Final nickel tenor (g/l) Nickel removal (%) Product nickel grade (%) Magnesia consumption (kg/m3 slurry) Thickener underflow density (%) Calculated thickener area (m2/(t/h)) Filter cake moisture (%) Filtration rate (m2/(t/h)) Final (magnesium) precipitation Temperature (°C) Number of stages Residence time per stage (h) Final pH level Final magnesium tenor (g/l) Magnesium removal (%) Product magnesium grade (%) Lime consumption (kg/m3 slurry) Filter cake moisture (%) Filtration rate (m2/(t/h))

0.02 99.5 40.0–45.0 4.0 20–25 2.0–3.0 55.2 126 60.0 5 1 9.5 0.00 98.8 10.2 10.0 64.0 166

plant consisting of bioleaching and associated purification unit operations. The data from the integrated piloting test work were used as inputs for the model. In some instances, most notably in sizing the settling and filtration equipment, in-house database information was used to obtain realistic settling and filtration rates. The settling and filtration tests that were performed during the integrated pilot plant campaign were limited in scope and extent, and were not optimized. It is therefore considered that the in-house database information is more reliable than the test work data. Process description The proposed hydrometallurgical plant will be situated at the site of the existing Aguablanca concentrator in southern Spain. The plant will be designed to treat 96 000 t/a of Aguablanca concentrate, producing 4 857 t/a of nickel in an intermediate precipitate and 5 400 t/a of copper cathode. The design criteria for the plant are summarized in Table XV. Milling The concentrate is repulped to between 50 and 60% solids, and fed to a vertical bead mill, where it is reground to achieve a particle size distribution with a d90 of 10 μm. Regrinding of the concentrate to this extent is required to achieve suitable copper recoveries in the bioleaching process. For the design capacity, it is considered that the regrinding will be achieved in a mill similar to a Deswik 1000 vertical bead mill. The Deswik 1000 is fitted with up to 24 polyurethane impellers on a vertical shaft. The milling medium comprises 1 mm zirconia beads. This mill is similar in concept to the IsaMill, with the principal difference being the orientation of the mill: the IsaMill is a horizontal bead mill, whereas the Deswik has a vertical orientation. Bead mills have been found to be more energy efficient than stirredINTEGRATED PILOTING OF A THERMOPHILIC BIOLEACHING PROCESS

143

119-154_Neale:text

2/14/09

11:48 AM

Page 144

Table XV Design criteria used in the costing model

144


119-154_Neale:text

2/14/09

11:48 AM

Page 145

media mills such as the Svedala VertiMill and the Metso Minerals Stirred Media Detritor (SMD), particularly for ultra-fine grinding to particle sizes below a d90 of around 20 μm. Based on information obtained from the Deswik mill supplier in South Africa, it is estimated that the power consumption required to achieve the target particle-size distribution in a Deswik 1000 vertical bead mill is about 14.3 kWh/t. The reground pulp is delivered to a pulp storage tank, with a sufficient capacity to allow the bioleach plant to be fed for two days. Bioleaching The bioleach feed pulp is diluted to the required solids concentration of 10% (m/m), using fresh water or water recycled from the downstream process. Appropriate control systems are utilized to maintain the desired pulp feed rate and solids concentration. The bioleach reactors are operated at a temperature of 70°C, and will consist of two modules, each comprising six reactors. Each module comprises three primary reactors operated in parallel, and three secondary reactors operated in series. The feed pulp is evenly distributed to the primary reactors to ensure a uniform feed to each primary reactor. The overall residence time of the bioleaching process is maintained at six days, and each individual reactor has a volume of approximately 1 620 m3. The bioleach reactors are mechanically-agitated, aerated vessels, made from an appropriate material—either a duplex stainless steel or a ceramic-lined concrete—to enable them to withstand the corrosive, acidic process conditions. The reactors are agitated using specialized impellers that promote the dispersion of large volumes of air and maintain the solids in suspension. Aeration is via specially-designed aerators that are positioned in the proximity of the impellers. The oxidation reactions taking place in the reactors are exothermic, and the temperature of the reactors are controlled at 70°C using cooling water that is circulated through a cooling tower. In some of the secondary reactors, where the reaction rates may be lower, some heating may be required, which is achieved by live steam injection. The product from the two bioleaching modules, comprising a pulp with a pH level of around 1.6, containing unleached gangue minerals, some iron precipitates, and a liquor containing sulphuric acid, ferric sulphate, copper sulphate and nickel sulphate, is combined and delivered to the primary iron removal plant. Primary iron removal The primary iron precipitation takes place in a series of five neutralization reactors, in which the temperature is maintained at 60°C. Limestone slurry is used as the neutralizing agent. The pH is raised to a level of 3.0, and the overall residence time is 5 hours. The main products from the neutralization reaction are gypsum (from acid neutralization) and ferric hydroxide (from iron precipitation). The product pulp from the primary iron removal reactors is fed to a continuous countercurrent decantation (CCD) circuit, comprising 4 thickeners. In this circuit, wash water is used to wash the product solids. The liquor from the CCD circuit, containing copper sulphate and nickel sulphate, is fed to the copper SX plant. Copper solvent extraction and electrowinning The SX plant receives the pregnant leach solution (PLS) from the primary iron removal circuit, and produces a rich electrolyte that is sent to the electrowinning plant. Copper SX is performed in two extraction, one scrubbing, and two stripping stages. In the extraction stages, INTEGRATED PILOTING OF A THERMOPHILIC BIOLEACHING PROCESS

145

119-154_Neale:text

2/14/09

11:48 AM

Page 146

copper is extracted into the organic phase. The organic phase comprises a mixture of 11% LIX®984N-C in ShellSol 2325 as the organic extractant (based on a specification of 0.29 g/l per volume % in the organic phase). Although a concentration of 25% organic was used in the pilot plant test work, the process model takes into account dilution from the CCD wash, which results in a lower copper tenor and organic concentration. An organic-to-aqueous ratio of 1:1 is maintained in the extraction stages. The tailing (raffinate) from the extraction stages, which contains nickel sulphate, forms the feed to the secondary iron removal circuit. The loaded organic is fed to the stripping section, where it is contacted with spent electrolyte from the electrowinning section, typically containing around 180 g/l of sulphuric acid. The copper is stripped from the organic phase into the electrolyte, which forms the feed to the electrowinning plant. The barren organic phase is returned to the extraction stages. In the electrowinning section, the rich electrolyte from the copper solvent extraction plant is fed to a series of conventional electrowinning cells, where the copper is plated out on stainless steel cathodes to produce LME A-grade copper. The electrowinning plant comprises a series of cells, with non-soluble lead anodes and 316L stainless steel cathodes, a rectifier, an overhead crane to handle the cathodes, an automatic cathode stripping machine and accessory equipment. The spent electrolyte is returned to the stripping section of the copper SX plant. Secondary iron removal The copper SX raffinate is fed to the secondary iron removal circuit, where the pH is raised to a level of 5.0 using a limestone slurry in order to purify the leach solution prior to nickel recovery. The plant comprises five reactors in series, operated at a temperature of 60°C. In this process, the acid produced in the copper SX process is neutralized, and any residual iron is precipitated. Additional elements that are removed in this process are aluminium and chromium, with partial removal of silicon and zinc. Any copper remaining in the raffinate is also precipitated, and there may be a small loss of nickel sulphate. The main products from the neutralization reaction are gypsum (from acid neutralization) and ferric hydroxide (from iron precipitation). The product pulp from the secondary iron removal reactors is fed to a continuous filter, where wash water is added. The filtrate, containing nickel sulphate, is fed to the nickel precipitation plant. The filter cake is repulped and fed to a cyclone, with the objective of concentrating and recovering most of the precipitated copper and nickel. The cyclone overflow, containing 20% of the mass and about 75–80 % of the copper and nickel, is recycled to the bioleach circuit. The cyclone underflow is disposed of. Nickel precipitation Three downstream nickel recovery options were investigated. • Hydroxide precipitation. This is the base case, and is the one that was piloted. Nickel is precipitated as an hydroxide by the addition of magnesia. This the least sophisticated option, as it generates a lower-quality product, which is difficult to dewater and therefore more expensive to transport. It is environmentally more benign than the other routes, provided inexpensive supplies of lime and magnesia can be obtained. • Sulphide precipitation through the addition of hydrogen sulphide. This process involves a higher degree of complexity, requiring pressurised reactors and a low cost-source of hydrocarbon or H2S. This route also has environmental concerns, owing to the use of pressurized reactors and the production of toxic and flammable gases, and it requires a more sophisticated workforce. However, the product has a lower moisture content than the hydroxide product, and is therefore less expensive to transport. 146


119-154_Neale:text

2/14/09

11:48 AM

Page 147

• Solvent extraction and electrowinning of nickel cathode is the most sophisticated and complex route, requiring a skilled workforce. A reliable power supply is required. This route is suitable if refining charges are high and the additional capital expenditure can be justified. Nickel hydroxide precipitation In this route, the filtrate from the secondary iron removal circuit is fed to the nickel precipitation plant, which comprises five reactors operated in series, each at a temperature of 60°C. The pH is raised to a level of 7.8 using a magnesia slurry, and the overall residence time is five hours. The nickel precipitates as nickel hydroxide, and some unreacted magnesia and magnesium hydroxide also reports to the solids. The aim is to produce a nickel hydroxide product containing at least 45% nickel and less than 10% magnesium (measured on a dry basis). The product pulp is fed to a settler, and the settler underflow is further dewatered with a continuous Larox-type filter, where the aim is to reduce the moisture content of the nickel hydroxide product to below 80%. The barren filtrate, containing magnesium hydroxide that was formed in the nickel precipitation process, is fed to the magnesium removal circuit in order to purify the process solution prior to recycling. Magnesium precipitation is performed in a plant comprising five reactors operated in series, each at a temperature of 60°C. The pH is raised to a level of about 9.5 using a slaked lime slurry, and the overall residence time is five hours. The product pulp from the magnesium removal reactors is filtered, and the barren filtrate is recycled, and the filter cake, containing mainly magnesium hydroxide and gypsum, is disposed of. Nickel sulphide precipitation Nickel sulphide precipitation is carried out in two autoclaves operated in parallel, with a 30minute residence time each. The feed solution is heated to 90°C, and H2S gas is introduced at a pressure of 600 kPa. The product is fed to a settler, and the underflow is filtered to produce a dewatered nickel sulphide product. The settler overflow is fed to a neutralization plant, consisting of five reactors in series, with a total residence time of five hours. The solution is neutralised with slaked lime to a pH level of 9.5 in order to neutralize the free acid and precipitate impurities. The gypsum product is fed to a settler, and the settler overflow is recycled to the bioleach process. Nickel solvent extraction and electrowinning Nickel SX consists of four extraction, three scrubbing, and three stripping stages. The organic extractant is 7% Versatic acid in a paraffin diluent. The O:A ratio is 1.25:1 for extraction, 5.5:1 for scrubbing and 5.5:1 for stripping. Extraction is carried out at a pH level of 6.5, using sodium hydroxide (NaOH) as the neutralizing agent. For the purposes of this study, it was assumed that nickel SX is performed in closed circuit with the leaching circuit. However, this approach may not be technically feasible, owing to the build-up of the SX neutralizing agent (NaOH or NH4OH). It may be necessary to precipitate and redissolve the nickel, followed by SX in a separate purification circuit, as is done in most laterite pressure leaching circuits. Such an option would add significantly to the capital and operating costs, and was not considered in this study. INTEGRATED PILOTING OF A THERMOPHILIC BIOLEACHING PROCESS

147

119-154_Neale:text

2/14/09

11:48 AM

Page 148

Process costing and financial analyses Mass and energy balances were developed for the overall hydrometallurgical circuit, followed by the design, sizing and costing of major equipment items, using in-house equipment cost indices. The equipment costs were factored to estimate the installed capital costs. Capital costs were calculated from installed equipment costs and factored indirect costs. Operating costs were calculated from stoichiometric reagent consumptions, estimated labour rates, and calculated power consumptions. Capital cost estimate Capital costs (basis: 2007/2008) were calculated for the three proposed process flowsheets, covering the three different nickel recovery options. Based on the mass balance and design specifications, major equipment items such as tanks, thickeners, filters and compressors were sized and costed using an in-house Mintek equipment cost database. Installation and auxiliary items were estimated by multiplying the equipment costs by the appropriate factors. Indirect capital costs were calculated by multiplying the total installed capital cost by appropriate factors. The total project cost is the sum of the total installed capital cost and the indirect costs, plus a contingency. The breakdown of capital costs is given in Table XVI for the three purification options. Capital costs were approximately US$5.7/lb annual nickel for the hydroxide and sulphide precipitation options, and approximately US$6.1/lb annual nickel for the SX-EW option. Operating cost estimate Operating costs were calculated from the stoichiometric reagent consumptions, estimated reagent costs, calculated power consumption and labour rates. The cost for maintenance is included in the labour element, and an additional 6% of the total installed capital cost was provided for maintenance materials and consumables. The breakdown of operating costs for the various components is shown in Figure 13. Figure 14 expresses the operating cost elements for four different refining options: concentrate sale, nickel hydroxide production, nickel sulphide production and nickel cathode production. In each case, the value of refining charges and by-product credits was included in the operating cost breakdown. The cost of mining and concentrate production of US$3.8/lb nickel was obtained from the most recent published data for the Aguablanca operation (2007 data). Financial analysis For the base case financial analysis, it is assumed that capital is spent in the first two years. A mining rate of 1.5 million tonnes per annum is assumed, and a concentrate production rate of 96,000 t/a. Depreciation of 33% per annum and a taxation rate of 35% is assumed. No provision is made for loan financing. The project is assumed to be fully equity funded. The capital and operating costs of the mine and concentrator are included, and these costs were taken from the published Aguablanca prefeasibility study12 and other published costs, with appropriate escalation. The economic analysis therefore comprises the costs of mining, concentrate production, bioleaching and purification. In Table XVII, the major outputs from the economic calculations, including a comparison of the IRR for each option, are summarized. In this analysis, metal prices prevailing in June/July 2008 were used: these were US$10.00/lb nickel and US$3.50/lb copper. It is noted that these 148


119-154_Neale:text

2/14/09

11:48 AM

Page 149

Table XVI Capital cost breakdown

Item Fine grinding Bioleach utilities Bioleach CCD/residue filter Fe removal I Fe removal II Ni-OH precipitation Mg precipitation Ni-S2- precipitation Neutral’sn - S2- ppt NiSX NiEW CuSX Cu EW Reagents and services Tailings neutralization Infrastructure Total directs (TIC) EPCM Site temp facilities Technology fees First Fill Owner’s cost Total Indirects Contingency Total project cost

OH ppt $ ‘000 1 920 1 937 11 497 4 702 620 2 238 3 231 2 368

S2- ppt Ni SXEW $ ‘000 $ ‘000 1 920 1 920 1 937 1 937 11 497 11 497 4 702 4 702 620 620 2 238 2 238

OH ppt $/a lb Ni 0.18 0.18 1.07 0.44 0.06 0.21 0.30 0.22

4,029 1 515

1 778 5 903 960 960 4 235 42 349 6 352 847 423 1 133 4 235 12 991 5 534 60 874

1 778 5 903 960 960 4 235 42 287 6 343 846 423 1 133 4 229 12 974 5 526 60 787

3 005 5 666 1 778 5 903 960 960 4 235 45 761 6 864 915 458 1 133 4 576 13 946 5 971 65 678

0.17 0.55 0.09 0.09 0.40 3.95 0.59 0.08 0.04 0.11 0.40 1.21 0.52 5.68

S2- ppt Ni SXEW OH ppt $/a lb Ni $/a lb Ni $/tpa conc. 0.18 0.18 20 0.18 0.18 20 1.07 1.07 120 0.44 0.44 49 0.06 0.06 6 0.21 0.21 23 34 25 0.38 0.14 0.28 0.53 0.17 0.17 19 0.55 0.55 61 0.09 0.09 10 0.09 0.09 10 0.40 0.40 44 3.95 4.27 441 0.59 0.64 66 0/08 0.09 9 0.04 0.04 4 0.11 0.11 12 0.39 0.43 44 1.21 1.30 135 0.52 0.56 58 5.68 6.13 634

S2- ppt $/tpa conc. 20 20 120 49 6 23

42 16

19 61 10 10 44 440 66 9 4 12 44 135 58 633

Figure 13. Operating cost breakdown (expressed in USc/lb Ni)


149

119-154_Neale:text

2/14/09

11:48 AM

Page 150

Figure 14. Operating cost breakdown

Table XVII Capital and operating cost summaries and financial calculations

Concentrate production Mass of ore mined (t/a) Concentrate production (t/a) Nickel price (US$/lb) Copper price (US$/lb) Capital cost (US$ million) Operating cost (US$ million/a) Metal value paid out (%) NPV, @ 8 % (US$ million) IRR (%)

1 500 000 96 000 10 3.5 119 42.0 65 145.7 31

Bioleach + hydroxide precipitate 1 500 000 96 000 10 3.5 61 10.3 90 203.5 30

Bioleach + sulphide precipitate 1 500 000 96 000 10 3.5 61 12.4 90 196.4 29

Bioleach + SX-EW 1 500 000 96 000 10 3.5 66 11.2 100 239.7 33

prices have fallen substantially since then. Based on nickel refining charges reported by Lundin in a fourth quarter 2007 press release, an average nickel refining charge of 35% of nickel value was used in this model for direct concentrate sale. Nickel refining charges of between 5 and 15% of value have been quoted13,14. An average refining charge of 10% of metal value was used in this model. An internal rate of return (IRR) over 10 years of 31% was calculated for concentrate sale, 30% for nickel hydroxide production, 29% for nickel sulphide production, and 33% for nickel cathode production. At a discount rate of 8%, the NPV is US$145.7 million for concentrate sale, US$203.5 million for hydroxide production, US$196.4 million for nickel sulphide production, and US$239.7 million for nickel cathode production. The effect of the low redox bioleach option on the process economics was calculated for the hydroxide precipitation route, and a comparison is summarized in Table XVIII. 150


119-154_Neale:text

2/14/09

11:48 AM

Page 151

Table XVIII Effect of low-redox bioleach operation on project economics

Solids concentration in leach Leach residence time days Tank sizes Capex bioleach Capex bioleach + purification plant Opex bioleach + purification plant Opex bioleach Opex bioleach utilities Aeration rate IRR, % 10 years

High redox case 10.0% 6.0 1.625 11.5 60.9 10.3 1.6 0.8 51 916 29.9%

% M3 M$ M$ M$ M$ M$ Nm3/h %

Low redox case 12.0% 4.0 922 6.9 53.3 9.9 1.3 0.7 52 004 31.7%

Oxidation levels—low redox Redox potential Pentlandite Chalcopyrite Pyrite Pyrrhotite

mV % % % %

Stage 1 430 93.0 96.0 50.0 90.2

Stage 2 550 98.0 97.0 70.0 95.1

Stage 3 560 98.0 98.0 80.0 99.0

Stage 4 570 99.5 98.0 87.0 99.5

Oxidation levels—high redox Redox potential Pentlandite Chalcopyrite Pyrite Pyrrhotite

mV % % % %

Stage 1 580 97.0 78.0 78.5 90.2

Stage 2 600 97.7 87.0 82.7 95.1

Stage 3 640 99.5 93.0 86.1 99.0

Stage 4 700 99.5 95.0 87.0 99.5

As indicated in Table XVIII, the low redox case yielded a higher overall copper extraction of 98% with a 4-day residence time, compared with 95% with a 6-day residence time for the high redox case. The low redox case also tolerated a higher feed solids concentration of 12% (against 10% for the high redox case). The decrease in residence time and the increase in solids concentration resulted in a decrease in the bioleach tank volume from about 1 600 m3 to 920 m3. This reduced the capital cost of the bioleach unit operation by 40%, and the overall hydrometallurgical plant capital cost by 12%. The impact on the overall operating cost for the hydrometallurgical plant was smaller; it decreased by about 4% for the low redox case. The net effect was that the IRR of the project increased from 29.9% to 31.7%. Discussion and conclusion Despite extensive research and development work spanning a period of at least 20 years, the uptake by industry of bioleaching technology to treat base metal sulphide concentrates has been slow. This is despite the fact that several projects have proceeded to an advanced commercial demonstration level, and despite a sustained boom in commodity prices. The main reason for this slow uptake is that smelting and refining charges have been on a longterm downward trend for over a decade, since peaking in 1998. There was a brief period (between 2004 and 2006) where these charges rose steeply, but they subsequently fell back to even lower levels. New hydrometallurgical technologies find it difficult to compete with the existing smelting industry, where the capital investment has already been made. Nevertheless, the indications early in 2009 are that smelting and refining charges are rising INTEGRATED PILOTING OF A THERMOPHILIC BIOLEACHING PROCESS

151

119-154_Neale:text

2/14/09

11:48 AM

Page 152

steeply (by between 60 and 70% for long-term contracts), and a sustained increase in these charges could result in hydrometallurgical technologies, including bioleaching, being considered for new projects in the future. Further impetus for selecting hydrometallurgical processes such as bioleaching will be gained as environmental considerations become increasingly important, and as the technologies themselves are improved, reducing costs and improving efficiencies. The present study, using Aguablanca nickel-copper concentrate as an example, has achieved several objectives: • It has demonstrated the technical feasibility of the technology. The use of thermophilic microorganisms to oxidize chalcopyrite has been demonstrated in an integrated pilot plant over an extended period of time. • A feature which distinguishes this technology with a previously developed thermophilic bioleach technology for chalcopyrite oxidation is that it is based on the use of air, as opposed to oxygen-enriched air. This obviates the need for sophisticated process control systems to regulate the dissolved-oxygen concentration in the bioleach reactors. • It has been shown that the economics of this technology is comparable with that of concentrate production and sale. • Significant advances have been made in applying a redox control strategy to improve chalcopyrite oxidation rates, thereby reducing the capital cost of the bioleach process and improving the overall process economics. This study has demonstrated that thermophilic bioleach technology is ready for commercial implementation. In Europe, where environmental factors and clean production are strong commercial drivers, and where sulphide deposits often have complex mineralogies and low grades, it could well become the technology of choice. In Africa and the rest of the resourcerich developing world, where large, higher-grade resources are still to be found in abundance, the timeline for acceptance may be longer. However, the first world is consuming most of these resources, and it can be expected that clean production practices will in future be dictated by the consumers and not the producers of raw materials—and then the same drivers currently at play in places like Europe will begin to determine technology choices in the producing countries. Bioleaching stands ready to meet this challenge. Acknowledgements This paper is a contribution to the research activities of the BioMinE project (BioMinE – NMP2-CT-2005-500329 – Contract Number 500329), which was co-financed by the European Commission (EC). Mintek received additional financial support for its research activities from the South African Department of Science and Technology (DST). The authors gratefully acknowledge the EC and the DST for their support which made this work possible, and thank all the BioMinE partners who contributed to the joint research effort. In particular, the contributions made by Bioclear and the University of Bangor to the microbial identification and quantification work are noted. References 1. MORIN, D., LIPS, A., PINCHES, T., HUISMAN, J., FRIAS, C., NORBERG, A., and FORSSBERG, E. BioMinE – Integrated project for the development of biotechnology for metal-bearing materials in Europe. Hydrometallurgy, vol. 83, 2006. pp. 69–76.

152


119-154_Neale:text

2/14/09

11:48 AM

Page 153

2. MORIN, D., PINCHES, T., HUISMAN, J., FRIAS, C., NORBERG, A., and FORSSBERG, E. Progress after three years of BioMinE—Research and Technological Development project for a global assessment of biohydrometallurgical processes applied to European non-ferrous metal resources. Hydrometallurgy, vol. 94, 2008. pp. 58–68. 3. WATLING, H.R. The bioleaching of sulphide minerals with emphasis on copper sulphides—A review. Hydrometallurgy, vol. 84, 2006. pp. 81–108. 4. WATLING, H.R. The bioleaching of nickel-copper sulfides. Hydrometallurgy, vol. 91, 2008. pp. 70–88. 5. PAVLIDES, A.G. and FISHER, K.G. The Kasese cobalt project. Extraction Metallurgy Africa ’98. Johannesburg, The South African Institute of Mining and Metallurgy, 1998. 20 pp. 6. VAN STADEN, P.J. The Mintek/Bactech copper bioleach process. ALTA Copper Hydrometallurgy Forum. Brisbane, 19–21 Oct., 1998. 7. CLARK, M.E., BATTY, J., VAN BUUREN, C.B., DEW, D.W., and EAMON, M.A. Biotechnology in minerals processing: Technological breakthroughs creating value. Hydrometallurgy, vol. 83, 2006. pp. 3–9. 8. PINCHES, A., MYBURGH, P.J., and VAN DER MERWE, C. Process for the rapid leaching of chalcopyrite in the absence of catalysts. US Patent 6,277,341: Appl.: 3 March 1997: Acc. 21 August 2001. 9. BATTY, J.D. and RORKE, G.V. Development and commercial demonstration of the BioCOP™ thermophile process. Hydrometallurgy, vol. 83, 2006. pp. 83–89. 10. ANONYMOUS. Lundin Mining Corporation Annual Report 2007. Lundin Mining Corporation, 2008. 78 pp. 11. CARTER, A.J. Economic comparison of the alternative methods for the recovery of gold from refractory ores. Colloquium: Bacterial Oxidation. Johannesburg, South African Institute of Mining and Metallurgy, June 1991. 12. ANONYMOUS. Aguablanca nickel-copper feasibility report. vol. 1, Metallurgical Design and Management (Pty) Limited. 13. ANONYMOUS. Fenix hydrometallurgical expansion preliminary assessment report. Rev. 1, Hatch Limited, 2006. 176 pp. 14. TAYLOR, A., FAIRLEY, H., and WINBY, R. Re-opening of the Radio Hill Nickel project, Western Australia using bacterial leaching of nickel sulphide concentrates. ALTA 1997 Nickel/Cobalt Pressure Leaching and Hydrometallurgy Forum, Perth. ALTA Metallurgical Services, Melbourne, 1997. 34 pp. INTEGRATED PILOTING OF A THERMOPHILIC BIOLEACHING PROCESS

153

119-154_Neale:text

2/14/09

11:48 AM

Page 154

John William Neale Specialist Engineer, Biotechnology Division, Mintek, South Africa • Metallurgical and biohydrometallurgical assessment of refractory gold ores and concentrates. • Mathematical modelling and computer-based simulation of bacterialoxidation processes. • Design, construction, commissioning, and operation of 1 t/d bacterial-oxidation pilot plant. • Project engineer for 1 t/d pilot-plant campaigns. • Mintek project engineer during commissioning and operation of 20 t/d bacterial-oxidation demonstration plant at Vaal Reefs gold mine. • Development, optimisation, and scale-up of agitation, aeration, and heat transfer systems for bacterial-oxidation reactors. • Development of downstream processes for solid-liquid separation, gold recovery, and ironarsenic precipitation. • Waste treatment and environmental management of bacterial-oxidation waste streams. • Feasibility studies involving commercial-scale bacterial-oxidation process design and costing for refractory gold and base metal ores and concentrates. • Project engineer for the development of bacterial-oxidation processes for copper and copper-nickel sulphide concentrates, including integrated flowsheets involving solid-liquid separation, solvent extraction, electrowinning, and iron precipitation. • Heap leaching of copper-sulphide ores. • Bacterial oxidation of copper, nickel, and PGM-bearing concentrate. • Bacterial oxidation of gold-copper concentrates. • Mintek process engineer responsible for process design of the Beaconsfield and Laizhou bacterial-oxidation plants. • Commissioning and trouble-shooting of the Beaconsfield bacterial-oxidation plant in Tasmania, Australia. • Project leader responsible for R4-million BioPAD-funded research project. • Development of thermophilic bioleaching processes. • Deliverable coordinator in the European Commission-funded BioMinE project, responsible for integrated piloting and techno-economic evaluation of two bioleach process flowsheets. • Co-author of numerous papers in the field of bacterial oxidation.

154


155-166_Simate:text

2/13/09

11:03 AM

Page 155

SIMATE, G.S. and NDLOVU, S., and GERICKE, M. The effect of elemental sulphur and pyrite on the leaching of nickel laterites using chemolithotrophic bacteria. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

The effect of elemental sulphur and pyrite on the leaching of nickel laterites using chemolithotrophic bacteria G.S. SIMATE* and S. NDLOVU* and M. GERICKE† *School of Chemical and Metallurgical Engineering, University of the Witwatersrand, South Africa †Biotechnology Division, Mintek, Randburg, South Africa In this study the efficiency of using either elemental sulphur or pyrite in the bacterial leaching of nickel laterites was investigated. Mixed cultures of Acidithiobacillus ferrooxidans, Acidithiobacillus caldus and Leptospirillum ferrooxidans were used. By measuring released nickel and pH changes, it was found that elemental sulphur yielded considerably better results than pyrite over a duration of one month. In the initial leaching the nickel recovered in the presence of the pyrite substrate was somewhat higher than that produced in the presence of the sulphur substrate. It should be noted, however, that although there was a definite observed trend for all the pHs studied, the difference was not that very significant in terms of statistically quantification. These observed results are attributed to the fact that bacteria are not very active in the initial leaching stages. At this stage of the process, abiotic oxidation of the substrates is the dominating factor. Ferrous ions in the leach media solution abiotically oxidize to ferric ions in the presence of oxygen. At acidic pH levels, elemental sulphur is inert to abiotic oxidation although other species such as thiosulphate and tetrathionate are oxidized abiotical to sulphate. Pyrite is oxidized by ferric ions via the thiosulphate route, producing sulphuric acid. It is thus expected that pyrite will result in slightly higher pH drops and thus the observed slight differences in nickel recoveries. Keywords: nickel laterites; chemolithotrophic microorganisms; leaching Introduction With an ever increasing demand for nickel and daily depletion of high grade nickel sulphide ore reserves, research on how to process the more abundant nickel laterite ore reserves is necessary. Nickel laterite ores have complex mineralogy, low nickel content, and are difficult to treat by conventional methods. However, new processes such as biotechnological leaching THE EFFECT OF ELEMENTAL SULPHUR AND PYRITE ON THE LEACHING OF NICKEL

155

155-166_Simate:text

2/13/09

11:03 AM

Page 156

play an ever increasingly important role in the extraction of metals from low grade ores (Nemati et al., 2000). These processes offer potential economic, environmental and technical advantages over conventional methods (Nemati and Harrison, 2000; Acevedo, 2000). Although bacterial leaching has found widespread application in the minerals industry, the use of chemolithotrophic bacteria in recovery of nickel from lateritic ores is relatively unexplored. The chemolithotrophs, which use sulphur and sulphide minerals such as pyrite as their energy sources in the bioleaching, ultimately produce sulphuric acid, which is subsequently used for the leaching of the metal values. Although nickel laterite contains metal values, it is not capable of participating in the primary chemolithotrophic bacterial oxidation process, probably because of the lack of the sulphidic content of the ore needed to produce the required sulphuric acid. Previous studies have shown that the metal value in the nickel laterites can be recovered by allowing the primary bio-oxidation of substrates such as pyrite or elemental sulphur to provide sulphuric acid solutions, which solubilize the metal content (Simate and Ndlovu, 2007; Simate and Ndlovu, 2008). These substrates are biogenically oxidized by acidophilic chemolithotrophic microorganisms for their energy supply, producing sulphuric acid (Ross, 1990), as shown in Equations [1] and [2]. The rate of recovery of nickel laterites by bacterial leaching is, however, affected by the type of substrate used to generate the acidic leaching reagent. [1] [2] [3] The sulphuric acid produced lowers the pH of the reaction media (Equations [1] and [2]). The pH or hydrogen ion concentration, [H+], is an important parameter that can be utilized to determine the extent to which nickel laterites dissolve (Equation [3]). It also determines the diversity of the microorganisms in a colony, e.g., obligate acidophiles such as those used in this study require low pH for growth since their membranes dissolve and cells lyse at neutrality (Todar, 2000). The solution pH in a given bioleaching operation is determined by the balance between the acid-producing and acid-consuming reactions. In the construction and use of culture media, one must always consider the optimum pH for growth. Therefore, the primary objective of this particular work was to investigate the pH requirements for the bacterial leaching of nickel laterites using a mixed culture of chemolithotrophic microorganisms in the presence of externally added substrates of elemental sulphur and pyrite. Materials and methods Ore samples and preparation The nickel ore was crushed and was classified into -75+63 μm size fractions using standard sieve plates. This size range was used because it was the size class in which most of the nickel laterite material fell in terms of mass. In addition, previous studies have shown that this particular size range gave higher nickel recoveries (Valix et al., 2001; Tang and Valix, 2006) not only due to the large percentage of nickel in the ore but also due to the higher survival of the microorganisms during the bioleaching process at this particular size range. The chemical composition of nickel was determined prior to chemical leaching experiments. The typical chemical composition of various oxides in the laterite ore used is given in Table I. 156


155-166_Simate:text

2/13/09

11:03 AM

Page 157

Table I Chemical composition of nickel laterite ore sample Mineral

SiO2

Fe2O3

Cr2O3

Al2O3

MgO

NiO

CoO

CaO

MnO

CuO

C

Mass %

52.79

21.86

0.99

2.5

7.46

1.87

0.25

0.68

1.28

0.01

0.119

S

P

0.034 0.064

Microbes A mixed culture of chemolithotrophic microorganisms (Acidithiobacillus Ferrooxidans, Acidithiobacillus caldus and Leptospirillum Ferrooxidans) used in the experiments was provided by Mintek, South Africa. The bacteria were cultured in standard 9K nutrient medium (Silverman and Lundgren, 1959). Experimentation Bioleaching experiments were carried out in 250 ml Erlenmeyer flasks with 100 m l of slurry. The slurry comprised a mixture of 5 g nickel laterite ore and 100 m l of medium together with an appropriate amount of sterilized energy sources (30% w/w elemental sulphur and 56% w/w pyrite to that of nickel laterite). These quantities of energy sources were chosen so as to have the same sulphur content in both elemental sulphur and pyrite. The slurries were inoculated with 10% (v/v) mixed bacterial culture. The pH of the mixtures was adjusted to 1.0, 1.5, 2.0 and 2.5, respectively. These pH ranges (i.e. pH <3) were chosen because they are known to optimize the growth of acidophilic microorganisms (Norris and Johnson, 1998). Some experiments were also run at initial pH of 1.5 using fresh liquid medium with an appropriate amount of sterilized energy source but without the inoculation of bacteria for comparison with inoculated experiments. Uninoculated distilled water and nutrient media were run as sterile at an initial pH of 1.5. The flasks were covered with pieces of aluminium foil to reduce evaporation and prevent contamination, but allow free supply of air, and then incubated in a platform shaking incubator at 30°C and 190 rpm. The pH profiles and redox potentials of the leach solution were measured but not controlled throughout the leach period using 744 pH meter Metrohm. The redox potentials readings were obtained using the Ag/AgCl, 3M KCl reference electrode and subsequently converted to the standard hydrogen electrode (SHE) (Friis et al., 1998). Samples (1.5-m l ) were drawn from flasks every three days to determine the concentration of nickel dissolved using the Varian SpectrAA-55B atomic absorption spectrophotometer. Solution loss through sampling was compensated by the addition of distilled water or 9K medium. The nickel recovery during the leaching of nickel laterites at any sampling time was calculated as a percentage of nickel in the liquid phase (concentration) to that in the original nickel laterite ore. The bacterial population was determined by measuring turbidity or optical density of the bacterial suspension using a UV-Visible double beam spectrophotometer (Model 4802). Since turbidity is directly proportional to the number of cells, this property was used as an indicator for bacterial concentration. The cells suspended in the suspension interrupt the passage of light, allowing less light to reach the photoelectric cell, and the amount of light transmitted through the suspension was measured as percentage transmission (or %T). The turbidity for cell suspension was measured at 550 nm (Plumb et al., 2008) against sterile 9K media as a reference. The wavelength of 550 nm is chosen because at this wavelength changes in both size of the cells as well as changes in the total nucleotide concentrations are reflected (Alupoaei and García-Rubio, 2004). THE EFFECT OF ELEMENTAL SULPHUR AND PYRITE ON THE LEACHING OF NICKEL

157

155-166_Simate:text

2/13/09

11:03 AM

Page 158

Note that the relationships among absorbance (A), transmittance (T) and optical density I (OD) are as follows: T = I0, where I is the light passing through the sample and I0 is the light hitting the sample. A = log10T. Optical density is a measure of absorbance and is related to transmittance by the following expression; OD = 2 – log (%T). Other available methods based upon measurement of nitrogen or protein content or ammonia consumption, and that of using the phase contrast microscope were less convenient and slower than the UV-Visible double beam spectrophotometer. On the other hand, the UVVisible spectrophotometer is a versatile, quantitative, rapid, and reliable analytical tool (Alupoaei and García-Rubio, 2004). It must be noted, however, that this method measures both the dead and active bacteria as it does not differentiate between the two states. However, the results obtained used in conjunction with pH and redox potential changes would reasonably outweigh the disadvantages of counting the dead bacteria. Prior to reading, the samples were filtered through Whatman filter number 1 to remove any solid particles. Results and discussion Effect of substrate type on nickel recovery In the first two weeks there were no significant differences in nickel recoveries, though slightly higher recoveries of nickel were observed in the presence of the pyrite substrate than sulphur substrate (Figure 1a). Initially ferrous ions in the 9K media oxidize to ferric ions according to reaction [4] below. [4] The ferric ions in turn attack pyrite producing ferrous ions and sulphuric acid through the thiosulphate mechanism (Schippers et al., 1996; Schippers and Sand, 1999). In the case of elemental sulphur under acidic conditions, oxidation is exclusively carried out by bacteria (Friedrich et al., 2001; Rohwerder et al., 2003; Rohwerder and Sand, 2007) although other sulphur species such as thiosulphate and tetrathionate are oxidized abiotically to sulphate (Schippers and Sand, 1999; Rohwerder et al., 2003). It is also expected, in the initial stages, that the bacteria are not very active; so abiotic oxidation of the substrates is supposedly dominant. However, higher recoveries were observed for the sulphur substrate than for the pyrite substrate at all pH levels after two weeks of leaching (Figure 1b). This showed that there was more acid produced by the leaching bacteria from the sulphur substrate than from pyrite substrate after two weeks. Effect of initial pH on nickel recovery Figures 2 and 3 show that the dissolution of nickel laterites is largely dependent on the initial solution pH. Although there was a general increase in the rate of nickel dissolution with time for all pH levels, higher recoveries of nickel were observed at lower pH with about 50% nickel being extracted within the first two weeks at initial pH level of 1.0. The higher recoveries at lower pHs indicate the possible dominance of abiotic leaching during this dissolution stage due to the high levels of acid in solution. Figures 2 and 3 also show that a rise in the nickel recovery was more for the pH of 2.5 after two weeks. This was attributed to the higher rate of acidification at this pH level (Figure 4 and 5). Previous studies have shown that higher rates of acidification imply that there were higher microbial activities (Rossi 1990; Hanford and Vargas, 2001; Schippers and Sand, 1999). Acidification in this context means production of H+ ions. The production of H+ ions is depicted by the pH drop, and that pH is a measure of concentration of H+ ions (i.e. pH = -log10[H+]). 158


155-166_Simate:text

2/13/09

11:03 AM

Page 159

(a)

(b) Figure 1. Dissolution rates of nickel laterites as a function of substrate type at different pH. (a) is recoveries within two weeks; (b) is recoveries in a month

Figure 2. Effect of pH on the leachability of nickel laterite with sulphur substrate

THE EFFECT OF ELEMENTAL SULPHUR AND PYRITE ON THE LEACHING OF NICKEL

159

155-166_Simate:text

2/13/09

11:03 AM

Page 160

Figure 3. Effect of pH on the leachability of nickel laterites with pyrite substrate

Figure 4. pH drop for sulphur and pyrite substrates within two weeks

Figure 5. pH drop for sulphur and pyrite substrates in a month

160


155-166_Simate:text

2/13/09

11:03 AM

Page 161

Effect of substrate on pH drop The pH rise for all the pH levels tested (i.e. pH 1.0, 1.5, 2.0, 2.5) was initially higher for sulphur than for pyrite within two weeks (Figure 4). Bacteria are expected not to be very active in the early stages of leaching. Initially, the pH rise was due to the consumption of the added acid since there was low production of acid through either abiotic oxidation or biotic oxidation of the substrates. However, in the case of the pyrite substrate, the abiotic oxidation of ferrous ions in the 9K medium to ferric ions leads to the subsequent oxidation of pyrite to produce ferrous ions and sulphuric acid. As a result there is a slight balance between the acid consuming and producing reactions and thus the pH rise is less than that for the sulphur substrate. In the subsequent leaching period pH drop was higher for the sulphur substrate than for pyrite at the same initial pH, implying that there was higher rate of acidification with the sulphur substrate as a result of bacterial activity (Figure 5). The results in Figures 4 and 5 also show that the pH decreased rapidly at higher initial pH, 2.5>2.0>1.5>1.0. This can be attributed to either; (1) the bacteria were naturally inactive at the low pH level or, (2) metal toxicity due to higher metal recoveries (due to initial rapid chemical leaching) at low pH. Furthermore, at higher pH both elemental sulphur and ferrous iron are oxidized abiotically at significant rates (Rohwerder et al., 2003). The trend in pH changes, arising from the substrates, was also observed during the identification of influential factors in the previous studies (Simate and Ndlovu, 2008). The results of the differences in pH changes can also be seen in the recoveries, being slightly higher in the presence of the pyrite substrate in the early stages, and subsequently higher with the sulphur substrates (Figures 1). Effect of substrate on bacterial growth Figure 6 shows the bacterial growth curve that was taken during the leaching period. The figure shows that bacterial concentration was higher for sulphur substrate than pyrite throughout the leaching period. The higher microbial growth with sulphur is as result of more energy gained from the oxidation of sulphur compared to the oxidation of ferrous ions (Rawlings et al., 1999; Yu et al., 2001). At the solution pH of 2, the Gibbs free energy released from the oxidation of ferrous iron is -9.49 kcal/mol, and that released from elemental sulphur is -185.9 kcal/mol (Yu et al., 2001). The trends in the nickel recoveries (Figure 1a) and pH drops (Figures 4) compared to bacterial concentration in the first two weeks (Figure 6) indicate that the rate of abiotic oxidation of pyrite was higher than the biotic oxidation of sulphur in the beginning. The higher amount of nickel recovered in the presence of the sulphur substrate after the two weeks tally with the peak of bacteria population as seen in Figure 6. These observations, therefore, show that a period of more than two weeks is an effective duration for the bacterial leaching of nickel laterites. After two weeks the high acid levels from the biotic oxidation of the sulphur substrates result in the higher nickel recoveries in the presence of the sulphur substrate than pyrite substrate. It seems, therefore, that the effect of bacteria on the nickel laterite dissolution depends mostly on the relationship between the bacteria and the substrate added to the media. Effect of media composition on pH and oxidation reduction potential Figures 7 and 8 show the effects of different media composition on pH and oxidation reduction potential (ORP), respectively. It is observed in Figure 7 that for inoculated experiments, there was a slight increase of pH initially and a subsequent decrease of pH thereafter. This was most likely due to the predominance of abiotic oxidation process initially. THE EFFECT OF ELEMENTAL SULPHUR AND PYRITE ON THE LEACHING OF NICKEL

161

155-166_Simate:text

2/13/09

11:03 AM

Page 162

Figure 6. Optical density at 550 nm as a function of time at pH of 1.5

Figure 7. Evolution of pH with time at initial pH of 1.5 for different media compositions

Figure 8. Evolution of ORP with time at initial pH of 1.5 for different media composition

162


155-166_Simate:text

2/13/09

11:03 AM

Page 163

Even though acid production took place in the long-term through the subsequent bioxidation of sulphur or pyrite, the initial net leaching process was likely acid consuming and hence the initially pH rise. In summary, the overall decrease in pH observed in the inoculated media is attributed to the oxidation of elemental or reduced sulphur to sulphuric acid by the leaching bacteria (see Equations [1] and [2]). In sterile controls (acidified distilled water, and sterilized media with no cells), the pH decrease was not observed. An initial rise and final decrease in pH was observed in experiments where uninoculated sulphur and pyrite media were used (results not shown here). However, the pH decrease for these uninoculated media was still lower than the inoculated media. This shows the efficacy of the presence of bacteria in the production of acid leading to pH reduction. The pH drop observed with uninoculated pyrite media can be attributed to the slow pyrite oxidation by oxygen in the presence of water forming ferrous iron and sulphate (Larsson et al., 1990) according to Equation [5] below. Sulphur was assumed to be slowly oxidized by oxygen in the presence of water according to Reaction [6] below (Hanford and Vargas, 2001). [5] [6] Figure 8 shows that higher ORPs were obtained with inoculated media than sterile media. The metal ion recoveries are higher with inoculated media than sterile media, thus implying higher ionic activities in the inoculated media. The ORP, which in the framework of this study, is a measure of the activities or strength of metal ions in relation to their concentration, is thus higher in the inoculated media. Conclusions This study investigated the bacterial leaching of nickel laterites using a mixed culture of chemolithotrophic microorganisms in the presence of externally added sulphur containing material (elemental sulphur and pyrite). The bacterial oxidation of sulphur and pyrite produces sulphuric acid, which dissolves nickel laterite to yield the required nickel metal. The results presented showed that the dissolution rates of nickel laterite were high in low pH media and high ORP, and in the presence of bacteria. However, the study showed that microbial activities, depicted by acidification, were lower at lower initial pH levels. Although uninoculated media with energy sources appeared to induce acidification, this was less than that in active bacterial cultures. This shows that under similar conditions, an inoculated media is more effective than an uninoculated media. This study has also shown the relationships between bacterial activities, depicted by acidification and the type of substrate. The recoveries were slightly higher for pyrite in the early stages of leaching and subsequently significantly higher for sulphur at all the initial pH levels studied. The test results and the high nickel recovery yield demonstrated that sulphur was more effective as a substrate than pyrite. It seems, therefore, that the effect of bacteria on the dissolution of minerals that are not capable of taking part in the primary chemolithotrophic bacterial oxidation process depends mostly on the relationship between the bacteria and the substrate energy source added to the media. Previous studies have shown that mixed cultures tend to be superior to those containing only iron- or sulphur oxidizing bacteria. However, little is known about how microorganisms interacted in this study, and how such interactions may have benefited or detracted the mineral dissolution process. It is, therefore, recommended that further studies should detect and differentiate the various species used during the bioleaching experiments. THE EFFECT OF ELEMENTAL SULPHUR AND PYRITE ON THE LEACHING OF NICKEL

163

155-166_Simate:text

2/13/09

11:03 AM

Page 164

Overall, this study has opened up a new era for the potential application of chemolithotrophic microorganisms for the commercial processing of the difficult-to-process low grade nickel laterite ores. In regions where the nickel laterite orebody exists the sustainability of the process will depend on the supply of sulphur-containing material (commercial sulphur or metal sulphide) for energy requirements of the bacteria. This is likely to form the dominant operating cost component of the bacterial leaching of nickel laterite ores using chemolithotrophic microorganisms. In addition, this process is promising because sulphuric acid is produced in situ whereas in processes such as high pressure acid leaching, sulphuric acid is produced in external facilities. With the addition of sulphur-containing material, the use of chemolithotrophic bacteria can be extended to the leaching of other low grade non-sulphide containing ores. Some examples where this process may be applied include silicate ores, oxidic converter furnace slags and refractory oxides. Acknowledgements The authors wish to thank The Council of Mineral Technology (Mintek), South Africa for the supply of the nickel laterite sample and microorganisms used in the study. Financial support to Geoffrey S. Simate by Mintek, and by the NRF South Africa through THRIP and the African Scholarships Flagship programmes is also gratefully acknowledged. The work is part of the MSc (Eng) dissertation submitted by Geoffrey S. Simate in 2008 to the University of the Witwatersrand, Johannesburg. References ALUPOAEI, C. E. and GARCIA-RUBIO, L.H. Growth behavior of microorganisms using UV-Vis Spectroscopy: Escherichia Coli. Biotechnology and Bioengineering, vol. 86, no. 2, 2004. pp. 163–167. ACEVEDO, F. The use of reactors in biomining processes. Electronic Journal of Biotechnology, vol. 3, no. 3, 2000. pp. 184–194. FRIEDRICH, C.G., ROTHER, D., BARDISCHEWSKY, F., QUENTMEIER, A., and FISCHER, J. Oxidation of reduced inorganic sulphur compounds by bacteria: Emergence of a common mechanism? Applied and Environmental Microbiology, vol. 67, no. 7, 2001. pp. 2873–2882. FRIIS, E.P., ANDERSEN, J.E., T., MADSEN, L.L., BONANDER, N., MØILER, P., and ULSTRUP, J. Dynamics of Pseudomonas aeruginosa azurin and its Cys3Ser mutant at single-crystal gold surfaces investigated by cyclic voltammetry and atomic force microscopy. Electrochimica Acta, vol. 43, no. 9, 1998. pp. 1114–1122. HANFORD, G.S. and VARGAS, T. Chemical and electrochemical basis of bioleaching processes, Hydrometallurgy, vol. 59, 2001. pp. 135–145. LARSSON, L., OLSSON, G., HOLST, O., and KARLSSON, H.T. Pyrite oxidation by thermophilic archaebacteria. Applied and Environmental Microbiology, vol. 56, no. 3, 1990. pp. 697–701. NEMATI, M., LOWENADLER, J., and HARRISON, S.T.L. Particle size effects in bioleaching of pyrite by acidophilic thermophile Sulfolobus metallicus (BC). Applied Microbiology and Biotechnology, vol. 53, 2000. pp. 173–179. NEMATI, M. and HARRISON, S.T.L. Effect of solid loading on thermophilic bioleaching of sulphide minerals. Journal of Chemical Technology and Biotechnology, vol. 75, 2000. pp. 526–532. 164


155-166_Simate:text

2/13/09

11:03 AM

Page 165

NORRIS, P.R. and JOHNSON, D.B. Acidophilic microorganisms. Horikoshi, K., Grant, W. D (eds.), Extremophiles: Microbial life in extreme environments. Wiley, New York, 1998. pp. 133–154. PLUMB, J.J., MUDDLE, R., and FRANZMANN, P.D. Effect of pH on rates of iron and sulphur oxidation by bioleaching organisms. Minerals Engineering, vol. 21, 2008. pp. 76–82. RAWLINGS, D.E., TRIBUTSCH, H., and HANFORD, G.S. Reasons why ‘Leptospirillum’like species rather than Thiobacillus ferrooxidans are the dominant iron-oxidising bacteria in many commercial processes for the bioxidation of pyrite and related ores. Microbiology, vol. 145, 1999. pp. 5–13. ROHWERDER, T., GEHRKE, T., KINZLER, K., and SAND, W. Bioleaching review part A: Progress in bioleaching: fundamentals and mechanisms of bacterial metal sulphide oxidation. Applied Microbiology and Biotechnology, vol. 63, 2003. pp. 239–248. ROHWERDER, T. and SAND, W. Oxidation of inorganic sulphur compounds in acidophilic prokaryotes. Engineering in Life Sciences, vol. 7, no. 4, 2007. pp. 301–309. ROSSI, G. Biohydrometallurgy. McGraw-Hill Book Company, New York. 1990. SCHIPPERS, A., JOZSA, P-G., and SAND, W. Sulphur chemistry in bacterial leaching of pyrite. Applied and Environmental Microbiology, vol. 62, 1996. pp. 3424–3431. SCHIPPERS, A. and SAND, W. Bacterial leaching of metal sulphides proceeds by two indirect mechanism via thiosulphate or via polysulphides and sulphur. Applied and Environmental Microbiology, vol. 65, 1999. pp. 319–321. SILVERMAN, M.P. and LUNDGREN, D.G. Studies on the chemoautotrophic iron bacterium Ferrobacillus ferrooxidans: I. An improved medium and a harvesting procedure for securing high cell yields. Journal of Bacteriology, vol. 77, no. 5, 1959. pp. 642–647. SIMATE, G.S. and NDLOVU, S. Characterisation of factors in the bacterial leaching of nickel laterites using statistical design of experiments. Advanced Materials Research 20–21, 2007. pp. 66–69. SIMATE, G.S. and NDLOVU, S. Bacterial leaching of nickel laterites using chemolithotrophic microorganisms: Identifying influential factors using statistical design of experiments. International Journal of Mineral Processing, vol. 88, 2008. pp. 31–36. VALIX, M., TANG, J.Y., and CHEUNG, W.H. The effects of mineralogy on the biological leaching of nickel laterite ores. Minerals Engineering, vol. 14, no. 12, 2001. pp. 1629–1635. TANG, J.A. and VALIX, M. Leaching of low grade limonite and nontronite ore by fungi metabolic acids. Minerals Engineering, vol. 19, 2006. pp. 1274–1279. TODAR, K. Physical and Environmental Requirements for Microbial Growth. Website: http://lecturer.ukdw.ac.id/dhira/NutritionGrowth/physicalandenv.html. Accessed September 2008. 2000. YU, J.-Y., MCGENITY, T.J., and COLEMAN, M.L. Solution chemistry during the lag phase and exponential phase of pyrite oxidation by Thiobacillus ferrooxidans. Chemical Geology, vol. 175, 2001. pp. 307–317. THE EFFECT OF ELEMENTAL SULPHUR AND PYRITE ON THE LEACHING OF NICKEL

165

155-166_Simate:text

2/13/09

11:03 AM

Page 166

Geoffrey S. Simate Masters Student, University of Witwatersrand, South Africa Three years experience in metallurgical operations for Zambia’s major mining conglomerate (Zambia Consolidated Copper Mines Ltd). Worked for one year with Konkola Copper Mines Plc after privatisation of Zambia Consolidated Copper Mines Ltd. Worked for one year with Anglovaal Mining Zambia Ltd Research Laboratory as a Process Engineer. Worked for four years with Chambishi Metals Plc as a Process Controller. Worked for one year with Maamba Collieries Ltd as a Plants Manager with the responsibility for ensuring efficient, safe and cost effective operations of the entire metallurgical operations. In my career I have been involved with plant commissioning, trouble shooting, pilot plant tests, plant efficiency tests and have provided technical expertise to the plant processes in order to achieve maximum efficiency. Have experience with leaching, solvent extraction, ion exchange, thickening, and electrowinning commercial plant operations. Fully qualified Quality Assurance Internal Auditor (BSI and RT&A). Scheduled to graduate with an MSc (Eng) in Metallurgy and Materials Engineering from the University of Witwatersrand in May, 2009.

166


167-176_vNiekerk:text

2/13/09

11:04 AM

Page 167

VAN NIEKERK, J. Recent advances in BIOX® technology. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Recent advances in BIOX® technology J. VAN NIEKERK Gold Fields Ltd, South Africa The past number of years has seen a sharp rise in the gold price. This has led to renewed activity in the gold sector and especially in refractory orebodies that are not economically feasible at lower gold prices. There have, historically, been a total of eleven BIOX ® plants commissioned worldwide, with eight currently in operation. The most recent plants to be commissioned were Bogoso and Jinfeng in 2007 and Kokpatas in Uzbekistan in 2008. Design of the Bogoso BIOX® plant was started by Minproc in 2005. The design throughput of the sulphide BIOX® circuit is 820 tpd concentrate at a feed sulphide sulphur grade of 20%. The Bogoso BIOX® plant has the largest biooxidation reactors with a live capacity of 1 500 m3 each. Plant design on the Jinfeng project started in 2005 and the engineering was performed jointly by Ausenco and Nerin. The BIOX® plant has a design capacity of 790 tpd concentrate at 9.4% sulphide sulphur. The Kokpatas plant will be the largest BIOX® plant in the world with a phase 1 design concentrate treatment capacity of 1 069 tpd at 20% sulphide sulphur. The throughput will be doubled during phase 2. Interestingly, unlike other BIOX® plants, this plant will use a resin-in-pulp circuit for gold recovery, but the design will allow the plant to be converted to carbon-in-pulp if required. The future growth of the technology is certainly also very good, with two new projects currently in the engineering phase, the Amantaytau project in Uzbekistan and the Mayskoye project in Russia. There are a number of major R&D projects currently under investigation with some very positive results. Most R&D is focused on the main capital and operating cost items in the process, but there are also projects focused purely on process optimization and improvement. Introduction The BIOX ® process for the pretreatment of refractory gold concentrates has been in commercial operation for the last 20 years. It has grown from a small-scale start at the Fairview Gold Mine to a mature, worldwide recognized technology with a total installed concentrate treatment capacity of 4 507 tons per day of concentrate. At current production rates a total of 1.5 million ounces of gold will be produced in 2009 using the BIOX® process. RECENT ADVANCES IN BIOX® TECHNOLOGY

167


2/13/09

11:04 AM

Page 168

Growth of the technology The BIOX® technology started off as a 10 ton per day pilot plant operating in parallel with old Edwards roasters at the Fairview plant in South Africa. The process proved to be robust and the capacity of the BIOX® section was increased in 1991 to treat the full 35 tpd concentrate. The capacity of the plant was again increased in 1994 and 1999 to the current design capacity of 62 tpd. The second BIOX® plant was the Sao Bento plant in Brazil in 1990. The first BIOX® licence was sold in 1991 for the application of the technology at the Harbour Lights project in Western Australia. This was quickly followed by Wiluna in 1993 and Ashanti in 1994/1995. Ashanti was a major breakthrough for the technology, confirming the ability of the process to be scaled up to any throughput rate using the modular design. Ashanti also proved the robustness of the technology and the suitability of the technology to remote locations. A total of eleven BIOX® plants was commissioned over the last 20 years with five new plants commissioned in the last three years. Table I gives a summary of the current and historical BIOX® plants and the concentrate treatment capacities. Figure 1 shows the installed concentrate treatment capacity, in ton per day concentrate, for all the BIOX® plants worldwide. After commercialization of the technology in 1991 the installed concentrate treatment capacity increased quickly with the commissioning of the five plants in quick succession. However, with the drop in the gold price in the middle 1990s, there was a marked slowdown in the interest in the BIOX® technology with only one low capacity plant commissioned between 1995 and 2005. Interest in the technology started to increase again from ~ 2003 with the increase in the gold price. This led to the commissioning of five new BIOX® plants between 2005 and 2008. During 2009 there will be a total of eight BIOX® plants in operation with a combined installed concentrate treatment capacity of 4 500 tpd. The total estimated gold production from BIOX® operations for 2009 will be over 1.5 million ounces (> 47 tons of gold). Table I A summary of the commercial BIOX® operations Mine

Fairview Saõ Bentob Harbour Lightsc Wiluna Ashanti Coricanchad Fosterville Suzdal Bogoso Jinfeng Kokpatas

Country

Concentrate treatment capacity (tpd)

Reactor size (m³)

Date of commissioning

Current status

S.Africa Brazil Australia Australia Ghana Peru Australia Kazakhstan Ghana China Uzbekistan

62 150 40 158 960 60 211 196 820 790 1069

340a 550 160 480 900 262 900 650 1 500 1 000 900

1986 1990 1991 1993 1994 1998 2005 2005 2007 2007 2008

Operating C&Me Decommissioned Operating Operating C&Me Operating Operating Operating Operating Operating

aThe

volume of the two primary reactors at Fairview Mine is under care and maintenance cMining operations were completed and 1994 and the plant decommissioned dOperations were temporarily stopped in 2008 eCare and maintenance bThe

168



2/13/09

11:04 AM

Page 169

Figure 1. Installed BIOX® concentrate treatment capacity

There are also currently four projects under engineering and construction, two expansions to existing facilities and two new projects that will increase the installed BIOX® treatment capacity to over 6 500 tons per day concentrate by 2012. The capacity can potentially continue to grow to over 8 500 tons per day concentrate by 2015 based on projects currently in the various stages of testwork and development. Detailed description of the new BIOX® plants This section will give a brief introduction to the three new BIOX® plants, Bogoso, Jinfeng and Kokpatas. The Bogoso BIOX® plant The Bogoso Gold Mine, owned by Golden Star Resources, is located in south-west Ghana approximately 200 km west of the capital city Accra. The first set of batch amenability tests on Bogoso concentrate was performed at Gencor Process Research in 1996. The results from the initial batch tests were very encouraging and continuous BIOX® pilot-plant testwork was performed in 1998 at SGS—Lakefield Research Africa on two bulk concentrate samples from the Bogoso deposit. The Bogoso BIOX® plant has a design capacity of 820 tons per day of concentrate at a sulphide sulphur grade of 20%. The BIOX® plant consists of two modules, each module consisting of seven BIOX ® reactors, followed by the CCD washing circuit, solution neutralisation and the carbon-in-leach circuit for gold recovery. Construction of the Bogoso Sulphide Expansion Project (BSEP) began on 13 June 2005 with GRD Minproc awarded the contract for the design and construction of the plant. The inoculum build-up and commissioning of the BIOX® plant started late in 2006 and was completed mid 2007 with the production of the first BIOX® treated gold. The BIOX ® technology achieved a new milestone at Bogoso with the successful commissioning of the largest BIOX® reactors. The plant also uses a closed circuit evaporative cooling system. Although the capital cost of this option is higher than the conventional open circuit system, it is expected that the operating cost (including maintenance) will be lower for the closed circuit due to better control of the water quality flowing through the cooling coils. Inoculum build-up at Bogoso was expedited by receiving 24 m3 of active inoculum from AngloGold Ashanti’s Sansu BIOX® plant, enabling the inoculum build-up process to start directly in a 100 m3 reactor. RECENT ADVANCES IN BIOX® TECHNOLOGY

169


2/13/09

11:04 AM

Page 170

The Jinfeng BIOX® Plant The Jinfeng gold deposit, owned by Sino Gold Limited, is located in the Guizhou Province, approximately 240 km south-west of the provincial capital Guiyang, in the Peoples Republic of China. Batch BIOX® testwork on Jinfeng concentrate samples started in 2002 at SGS Lakefield Research Africa. The concentrate samples proved to be amenable to biooxidation pretreatment, achieving high sulphide oxidation and gold recovery results. The batch testwork was followed up by a continuous BIOX® pilot-plant test programme in 2003. The first process design package was delivered to Sino Gold Limited in 2003 following the successful completion of a BIOX® licence agreement for the use of the BIOX® technology for the Jinfeng deposit. The plant was designed to treat up to 790 tons per day of concentrate at a sulphide sulphur grade of 9.37% in two BIOX® modules, each module comprising eight 1 000 m3 reactors. The retention time across the BIOX® reactors is four days at the design feed rate. The biooxidation section is followed by a standard CCD washing circuit, solution neutralization and a carbon-in-leach circuit for gold recovery. The Jinfeng Bankable Feasibility Study was completed in March 2004 and construction of the plant began in February 2005. Inoculum build-up started in April 2006 with the first gold poured in March 2007. Jinfeng is currently one of the largest gold mines in China. Optimization of the Jinfeng mine is ongoing to increase gold production to optimal levels. Jinfeng mine is also faced with major logistical challenges, especially during the wet season with an average rainfall of 1.2 m annually, accompanied by landslides and wind storms. Currently the Jinfeng project is consistently achieving operational performance at or above design parameters. The Jinfeng BIOX® plant is performing exceptionally well with sulphide sulphur oxidation well above design as well as BIOX®/CIL gold recovery above design parameters. The Kokpatas BIOX® plant The Kokpatas Mine is located in the Kyzylkum desert, 32 km north-east of the town of Uchkuduk in Central Uzbekistan. The deposit is owned and operated by the Navoi Mining and Metallurgical Combinat, the largest gold producer in Uzbekistan. Ore will be supplied from various open pits, including the Daugystau deposit some 140 km from Kokpatas. Laboratory test work conducted by GENCOR Process Research (GPR), on samples from the Daugystau and Kokpatas pits, indicated that the sulphide ore is amenable to upgrading by sulphide flotation. Pretreatment of the flotation concentrate by the BIOX® Process liberated the gold sufficiently to improve gold dissolution during cyanidation to over 90%. Pilot-scale flotation and BIOX® test work conducted in South Africa in 1994 confirmed the positive results of the laboratory test work and provided sufficient data for the design of a full scale sulphide treatment plant. In 1997, similar test work was performed at GPR’s facilities in South Africa on a 15 ton ROM ore sample from the Kokpatas sulphide deposit, and excellent results were achieved. Following the conclusion of a licence agreement between NMMC and Biomin Technologies S.A. for the supply of the BIOX® technology, a BIOX® plant was designed to treat flotation concentrate from the Kokpatas deposit during phase 1 of the project. The design concentrate treatment capacity of phase 1 of the project is 1 069 tpd. During phase 2 the plant capacity will be increased to 2 137 tpd to incorporate flotation concentrate from the Daugystau deposit. This is made possible by the modular design of the biooxidation section of the BIOX® plant. 170



2/13/09

11:04 AM

Page 171

The phase 1 BIOX® plant consists of four modules. Each module has three primary reactors and three secondary reactors. The counter current decantation thickeners (CCDs) were sized to accommodate the Phase 2 capacity of the plant. During Phase 2, four additional BIOX® modules will be added and the neutralization plant capacity increased. The supply capacity of cooling water and air will also be increased accordingly. The detailed design of the phase 1 BIOX® section was performed by Institute O’zGeotexliti in Uzbekistan with equipment fabrication, construction and the electrical installation performed by local contractors. Civil works on the BIOX® reactor foundations for phase 1 started during July 2005. Commissioning of phase 1 began mid-2008 and is expected to be completed early 2009 after the delivery of the final equipment. New BIOX® projects under development There are currently four BIOX® projects in various stages of design or construction. Two of these are expansions to existing BIOX® plants while the other two are new BIOX® projects. The Suzdal BIOX® plant in Kazakhstan is expanding the capacity of the treatment plant from 196 tpd concentrate to over 520 tpd concentrate. This will be achieved by duplicating the existing BIOX® circuit and adding additional cooling and blower air capacity. The solids concentration in the feed will also be increased slightly to ensure that sufficient residence time is maintained in the BIOX® reactors. A new CCD circuit will be constructed and the number of neutralization reactors will be increased to cater for the higher concentrate throughput. The Kokpatas plant in Uzbekistan has started with the planning for phase 2 of the project, to double the treatment capacity of the plant to 2 163 tpd concentrate. This will be achieved by duplicating the phase 1 BIOX® circuit. The CCD circuit was sized for phase 2 during phase 1 while additional neutralization reactors will be added. The blower and cooling tower capacity will also be increased as required to cater for the increased aeration and heat loads. The Amantaytau project is also located in central Uzbekistan near the town of Zarafshan. The project will also be implemented in two phases. The phase 1 BIOX® plant will have a design capacity of 376 tpd concentrate, increasing to 564 tpd in phase 2. The sulphide sulphur grade in the concentrate can be as high as 36% due to the high in situ ore sulphide grades. The Mayskoye project is located in the Chukotka region in Russia. The BIOX® plant will have a design capacity of 350 tpd concentrate. The extreme climatic conditions in this region will pose a number of engineering challenges, starting with a relatively short shipping window in the summer months. The ambient temperature also only exceeds freezing point for two months of the year. BIOX® at sub-zero temperatures The Suzdal BIOX® plant is located near the town of Semey in Kazakhstan. The BIOX® plant has a design capacity of 196 tpd concentrate at a sulphide sulphur grade of 12%. The BIOX® section consists of 6 x 650 m³ reactors, configured in the standard three primary and three secondary reactor configuration. The extreme climatic conditions experienced at Suzdal made the design of the plant very interesting. The design had to take hot, humid conditions during summer into consideration and ensure that the cooling towers can deliver the required duty under these conditions. The bigger problem, however, was to design for the winter conditions when the temperatures could go as low as—45°C, without taking any wind chill into consideration. For this reason the bulk of the plant is located inside a building with only the BIOX® reactors and the cooling towers outside the building. The walkways on top of the BIOX® reactors were covered to protect the operators from the extreme conditions when taking samples. RECENT ADVANCES IN BIOX® TECHNOLOGY

171


2/13/09

11:04 AM

Page 172

The BIOX® process is exothermic and extensive modelling was performed to ensure that sufficient heat is produced by the process to maintain the temperatures in the BIOX® reactors even at the coldest conditions. The modelling indicated that the BIOX® reactors would still have to be cooled even at the coldest conditions. Significant planning also went into planning for power failures and unplanned stoppages to ensure that the reactors do not freeze if standing for too long. The Suzdal BIOX® plant has now operated through three winters and has performed very well. So far no problems were experienced with maintaining the temperature in the primary reactors even at reduced sulphur feed rates in the first year than initially anticipated. Suzdal has certainly proven that BIOX® is not only a viable technology to consider for sub-zero temperature applications, but that it holds certain advantages over other processes. The advantage lies in the modular design of the plant and the number of reactor stages. Even if one or more of the reactors is off-line the rest of the process plant does not have to be stopped and can continue to operate as normal. For other refractory processes, if the treatment process is stopped for any length of time, the rest of the processing plant must also stop, a situation that can be catastrophic at sub-zero conditions. The performance of the Suzdal BIOX® plant has certainly proven beyond a doubt that the BIOX® technology can be used successfully at sub-zero conditions. Research and development The BIOX® process has been commercially in operation for nearly 20 years. Throughout this period Gold Fields has maintained a strong focus on research to improve the efficiency of the process and the design of the commercial reactors. Maintaining an active research and development program is critical to ensure the long-term viability of the technology. For this reason Gold Fields has developed a strategic research and development programme to address the main capital and operating cost items in the BIOX® process. The programme is being implemented in stages, focusing on the highest priority items first. Figure 2 shows a breakdown of the capital cost for a typical BIOX® plant. The graph indicates that the stainless steel for the BIOX® and neutralization reactors will make up approximately 35% of the installed equipment cost, followed by the agitators at approximately 27% and the blowers at 13%. The operating cost breakdown for four operating BIOX® plants is shown in Figure 3 and the reagent cost breakdown in Figure 4. It can be seen that between 40% and 50% of the opex cost is for power and between 30% and 45% for reagents. The reagent cost breakdown for the same four plants is shown in Figure 4. It can be seen the bulk of the cost is for pH control, accounting for ~ 70% of the reagent cost, but can vary from as low as 40% to as high as 85% depending on the relative reagent cost and consumption rates. This includes both BIOX® pH control and neutralization of the acid solution. Nutrient cost is usually fairly low at less than 20% and can be as low as 4% after a dedicated nutrient optimization programme is completed. The exception is plant B where the nutrient cost is exceptionally high, but this is a factor of low neutralization cost and the fact that the plant was fairly recently commissioned and must still undergo nutrient optimization. The remainder of the reagents is usually less than 10% of the overall reagent cost. The cost for cyanide leaching of the BIOX® product is not included in the operating cost given above. Analyses of the CIL operating cost for the same four BIOX® plants indicate that reagents make up between 70% and 90% of the operating cost, with cyanide accounting for 65% of the reagent cost. 172



2/13/09

11:04 AM

Page 173

Figure 2. Breakdown of the installed equipment cost for a BIOX® plant

Figure 3. Operating cost breakdown for four BIOX® plants

Figure 4. Reagent cost breakdown for four BIOX® plants

Four research and development areas were identified based on the capital and operating cost structures described above: • Development of an improved agitation system • BIOX® process optimization, focused on BIOX® retention time reduction and reagent cost savings RECENT ADVANCES IN BIOX® TECHNOLOGY

173


2/13/09

11:04 AM

Page 174

• Evaluation of alternative materials of construction • Cyanidation optimization. A number of projects were identified based on these main R&D areas and are progressively being implemented. The main projects currently underway are described in more detail below. Development of an improved agitation system Axial flow impeller technology in biooxidation reactors is conventionally based on the concept of down pumping. The impeller circulates the slurry in a downward motion with the objective of increasing the gas retention time in the reactor. Alternative agitation systems are being developed in an effort to reduce the overall power input required for aeration and air dispersion. One of the options under investigation is the use of an up-pumping axial flow impeller in biooxidation reactors. An extensive laboratory-scale test programme was conducted where a range of different impeller combinations was tested in a 500 l reactor using the Lightnin A310, A315 and A340 impellers. Different combinations of these impellers were compared with each other and the A315 was used as a bench mark with the main objective to reduce the power per unit volume and reducing the air demand while still maintaining high levels of oxygen mass transfer. The configuration that performed the best in the laboratory test programme was scaled up to a 20 m3 test reactor at Fairview BIOX® plant. The selected configuration was tested under the same conditions as in the laboratory determining the power numbers, oxygen mass transfer and power per unit volume, using the A315 as a benchmark. The final stage was to test the impeller under BIOX® conditions by operating the test reactor as a primary BIOX® reactor. The Fairview BIOX® plant personnel have been a tower of strength in making this test work possible. Results to date are looking very promising in achieving the main objectives. BIOX® process optimization There are a number of factors that determines the rate of sulphide oxidation. These include both BIOX® operating parameters and concentrate characteristics. The BIOX® process has been in operation for 20 years and the operating conditions are well defined. However, we have seen from results of the operating BIOX® plants that the BIOX® bacteria are capable of handling conditions outside the standard operating parameters. It was decided to start a project to investigate the effect of changing certain operating conditions or concentrate characteristics on the performance of the process. The objectives of the programme were to reduce the BIOX ® retention time, and thereby capital cost and to reduce the reagent consumption rates. Due to the number of potential parameters to investigate, a structured programme was set up. The initial set of tests focused on the effect of following parameters: • BIOX® feed density variation • Effect of fine grinding of the concentrate • Effect of BIOX® liquor removal to control the ferric iron concentration in the primary reactors and • Effect of pH in the final biooxidation stages. The test work was performed in a 120 l continuous pilot plant located at the Fairview BIOX® plant. The pilot plant was operated for a total of 6 months using a bulk Fairview concentrate sample collected over a seven-day period as feed. Batch amenability tests were also performed on the concentrate samples using similar conditions as in the pilot plant. In future the test programme will be expanded to include other parameters. 174



2/13/09

11:04 AM

Page 175

Materials of construction Stainless steel for the BIOX® reactors is one of the main capital cost items for any new BIOX® plant. For this reason Gold Fields is continually evaluating the performance of various and potentially cheaper stainless steel grades under BIOX® conditions. Various coatings for mild steel are also tested as a coated mild steel reactor may have cost benefits, provided the coating can withstand the operating conditions in a BIOX® reactor. The testwork includes both anaerobic conditions, typically found in the CCD thickeners, as well as aerobic conditions found in the BIOX® reactors. Three sets of test racks were prepared and installed at the Fosterville BIOX® plant in Australia. Test samples from a number of steel grades were included in the racks, including 304, 316, 317, 2205 and LDX2101. A number of mild steel samples, coated with different coatings were also installed. Samples will be removed after 6, 12 and 24 months for evaluation. Fosterville was selected as the test site due to the relatively high chloride content of the water, averaging approximately 600 ppm. Combination mesophile and thermophile biooxidation process Test work has shown that the use of high temperature or thermophillic micro-organisms can allow almost complete oxidation of all the intermediate sulphur species, thereby significantly reducing thiocyanate formation during the cyanidation of the biooxidation product. This will result in significantly lower cyanide consumption during leaching with potentially no loss in gold recovery. The use of thermophillic micro-organisms does, however, come at a price. • Increased capital cost—the micro-organisms (thermo-acidophillic archaea) operate in a temperature range of 65–80°C, and this increases the corrosive nature of the acidic bioleaching slurry. Standard austenitic stainless steel grades cannot be used for this application and more exotic materials of construction such as duplex stainless steels or acid lined concrete tanks are required for the manufacturing of the reactors. The process operates at reduced solids concentrations, but this is traded off with faster reaction kinetics. • Increased operating cost—the equilibrium solubility of oxygen in water is very low at these high temperatures and oxygen enrichment of the air introduced into the slurry is required to achieve the required mass transfer rates in the reactors. Water consumption is also increased due to increased evaporation rates at the higher operating temperatures, and oxygen enrichment may be required to reduce the volume of air flowing through the reactors. A combination mesophile and thermophile process was proposed to make the best use of the advantages of the different microbial strains. In this process the mesophilic BIOX® bacteria are used for the primary oxidation stage to achieve approximately 70% sulphide oxidation. This is then followed by a thermophile oxidation stage to complete the oxidation, targeting the intermediate sulphur species. A number of batch and continuous test work programmes using this process were completed over the last five years. A large-scale thermophile pilot-plant test work program was completed at the Fairview BIOX® plant to test the process under conditions expected to be encountered during full-scale operation and to generate data for the design of a commercial scale thermophile plant at Fairview. This included using daily overflow samples from the primary reactors at Fairview as feed to the thermophile plant. The pilot plant was operated for 6 months testing various parameters to optimize the process. Daily oxidation values were monitored as well as subsequent gold recovery and cyanide consumption during leaching of the oxidized product. RECENT ADVANCES IN BIOX® TECHNOLOGY

175


2/13/09

11:05 AM

Page 176

The results from the test programme were very encouraging, indicating that a very short retention time is required in the thermophile stage to maintain bacterial activity and achieve the required sulphide oxidation. The plant was able to operate at higher solids concentrations than initially expected, and it was found that no solution removal stage is required between the BIOX® and thermophile stages. Sulphide oxidations similar or better than the BIOX® plant were maintained and similar gold recoveries were achieved at substantially lower cyanide consumption rates. Conclusions The BIOX® technology has been commercially in operation for more than 20 years. The technology has proven itself to be robust and ideally suited for remote locations. The recent interest in the technology, as evidenced by the number of BIOX ® projects recently commissioned or in development, confirms that the technology offers significant advantages over other refractory processes. The combined installed treatment capacity of the eight operating BIOX® plants across the world is currently 4 500 tpd with an expected 1.5 million ounces of gold to be produced using the BIOX® technology in 2009. Recent operating experience at the Suzdal plant has confirmed the applicability of the BIOX ® technology to projects located in cold climates. BIOX ® again offers certain advantages over other treatment processes, including high plant availabilities and the ability to operate continuously even with a number of reactors offline. Gold Fields also manages a structured research and development programme to ensure continuous development of the technology. The programme is focused on the main capital and operating cost items, namely power for aeration and agitation, BIOX® retention time and cyanide consumption.

Jan van Niekerk Senior Consultant: Metallurgy (Refractory Gold), Gold Fields Ltd, South Afirca Jan van Niekerk started his career in 1995 at Gencor Process Research. He spent the early part of his career on various BIOX® projects, including a number of BIOX® continuous pilot plant tests. He spent a short period of time at Beatrix Gold Mine to gain operational experience until the forming of Gold Fields Ltd from the gold assets of Gencor and Gold Fields of South Africa. He then returned to the BIOX® department at Gold Fields’s corporate office. There he was still involved in testwork, but also started to get involved in the design of BIOX ® plants and giving assistance to operating plants. He was involved in the commissioning of the Tamboraque plant in Peru and was the BIOX® representative for the testing, design and commissioning of the Fosterville BIOX® plant in Australia. In 2006 he was promoted to the head of the BIOX® department within Gold Fields overlooking all aspects of managing thetechnology. 176


177-186_vDeventer:text

2/13/09

11:06 AM

Page 177

VAN DEVENTER, R., KELLER, W., OOSTHUIZEN, M., and STAPELBERG, J. Systematic characterization of the mixing process for the BIOX® reactor. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Systematic characterization of the mixing process for the BIOX® reactor R. VAN DEVENTER*, W. KELLER†, M. OOSTHUIZEN‡, and J. STAPELBERG§ *Gold Fields Mining Services, Johannesburg, South Africa, †Ekato RMT GmbH, Schopfheim, Germany, ‡Ekato Corporation, Oakland, NJ USA, §Ekato Africa, Johannesburg, South Africa

The BIOX ® process was developed for the biological pre-treatment of refractory gold ores and concentrates ahead of conventional cyanide leach gold recovery. Rights to the process are currently with BIOMIN Technologies SA, a subsidiary of Gold Fields Limited. There are currently eight operating installations worldwide employing the process on a commercial scale. EKATO is a company specializing in developing new or optimizing existing mixing applications. EKATO has acquired a vast amount of expertise in agitation over its 75 year history. In a joint research project the reactors of the BIOX ® process were characterized from the aspect of their mixing performance. BIOMIN is constantly testing new mixing solutions and reviewing potential vendors to increase the efficiency of the BIOX® process. EKATO’s intention was to qualify as an approved vendor for supplying agitation equipment to the BIOX® process. A two step approach was planned: In phase one of the project EKATO would develop an alternative agitation solution to the established option, exhibiting at least similar performance levels. In phase two potential possibilities to improve the overall design would be developed. This paper focuses on phase one and addresses the systematic approach applied by EKATO in characterizing the mixing processes. Mixing tasks were assessed individually applying basic theory and scale-up and -down rules to the individual tasks. Testing with model systems was done at the EKATO technology center in Schopfheim, Germany, on scales of 50 liters and 1 m³. The results showed that the ISOJET-B impeller design showed promising performance under BIOX® reactor conditions. Scale-up with the ISOJET-B was done in order to apply it in the 21 m³ live volume test reactor at Fairview mine in South Africa. At Fairview tests were performed using model systems and eventually under real plant conditions with BIOX® slurries. The results were a success and confirmed that the scale-up factors determined during the testing at EKATO’s technology center could be substantiated in the ‘model scale’ 25 m³ reactor. Testing under the actual process conditions showed that all design parameters were within the desired range to be considered an excellent leaching performance. This successful joint development shows the importance of a close cooperation of technology licensors or end-users and equipment suppliers. It is vital for understanding all factors influencing the mixing performance of processes. Exercises such as this will provide equipment suppliers and plant operators more confidence in managing further scale ups at ever larger engineered plants. SYSTEMATIC CHARACTERIZATION OF THE MIXING PROCESS FOR THE BIOX® REACTOR

177


2/13/09

11:06 AM

Page 178

Introduction The BIOX® process was developed for the biological pretreatment of refractory gold ores and concentrates ahead of conventional cyanide leach gold recovery. Rights to the process are currently with BIOMIN Technologies SA, a subsidiary of Gold Fields Limited. There are currently eight operating installations worldwide employing the process on a commercial scale. EKATO is a company specializing in developing new or optimizing existing mixing applications. EKATO has acquired a vast amount of expertise in agitation over its 75 year history. In a joint research project the reactors of the BIOX® process were characterized from the aspect of their mixing performance. BIOMIN is constantly testing new mixing solutions and reviewing potential vendors to increase the efficiency of the BIOX® process. EKATO’s intention was to qualify as an approved vendor for supplying agitation equipment to the BIOX® process. A two-step approach was planned: in phase one of the project EKATO would develop an alternative agitation solution to the established option, exhibiting at least similar performance levels. In phase two potential possibilities to improve the overall reactor design would be developed. This paper focuses on phase one and addresses the systematic approach applied by EKATO in characterizing the mixing processes. The BIOX process The BIOX® plant typically consists of six equidimensional reactors configured as three primary reactors operating in parallel followed by three secondary reactors operating in series. The feed concentrate from the stock tank is diluted to 20% solids by mass before being fed to the primary BIOX® reactors. The operating slurry solids content is determined primarily by the sulfide content of the concentrate feed and the rate of oxygen transfer necessary to maintain the required rate of oxidation. The slurry solids content is also determined by the amount of toxic elements present in the slurry fed to the BIOX® reactors. The pulp residence time in the bio-oxidation reactors is typically four to six days depending on the oxidation rates achieved, which is a function of the sulphide/sulphur content and mineralogical composition of the concentrate. Generally, half of the retention time is spent in the primary reactors to allow a stable bacterial population to be established and to prevent bacterial washout. Once a stable bacterial population has been established, a shorter retention time can be tolerated in the secondary reactors where sulphide sulphur oxidation is completed. Nutrients in the form of nitrogen, phosphorous and potassium salts are also added to the primary reactors to promote bacterial growth. The mixed culture of mesophilic bacteria used in the BIOX® process can operate at temperatures ranging from 30°C to 45°C. The pulp temperature in commercial reactors is controlled between 40°C and 45°C. This temperature allows maximum sulphide oxidation rates to be achieved while minimizing cooling requirements. The oxidation of sulphide minerals is an exothermic process and the reactors must be cooled continuously by circulating cold water through a series of cooling coils installed inside the reactors. Low-pressure air is injected into the BIOX® reactors to supply oxygen for the oxidation reactions. It is extremely important that a dissolved oxygen concentration of > 2 ppm be maintained at all times in the slurry. Axial flow fluid foil impellers are used in the BIOX® reactors they offer improved efficiency over radial-flow turbines. Figure 1 shows the typical BIOX® process flowsheet. 178



2/13/09

11:06 AM

Page 179

Concentrate from flotation/regrind

Stock tank

Water

Nutrients Nutrient make-up tank Feed splitter

Blower Secondary BIOX reactors

Cooling tower

CCD Thickeners

Primary BIOX reactors Wash water

To leach Neutralization To tails

Figure 1. Typical BIOX® process flowsheet

Objectives and developments A systematic approach was applied by EKATO in characterizing the mixing processes. Mixing tasks were assessed individually applying basic theory and scale-up and -down rules to the individual tasks. Testing with model systems was done at the EKATO technology centre in Schopfheim, Germany, on scales of 50 litres and 1 000 litres. Different types of impellers which were an option for the applied process conditions were tested and their performance compared. The results showed that the ISOJET-B impeller (Figure 2) design yielded the best performance at BIOX® reactor conditions. Scale-up with the ISOJET-B was done in order to apply it in the 21 m³ live volume test reactor at Fairview mine in South Africa. At Fairview tests were performed using model systems and eventually under real plant conditions with BIOX® slurries. Mixing: characterization of the BIOX® process The basic mixing tasks encountered in practice can be classified into four main categories: • Blending - mixing miscible liquids; eliminating differences in concentration, temperature, etc. • Suspension - uniform distribution of solids - off-bottom suspension of solids • Dispersion gas/liquid and liquid/liquid - mass transfer between the gas/liquid phases and liquid/liquid phases - stable emulsions • Heat transfer Every mixing task is governed by natural laws, which must be known for the scientific design of the agitator. Complex mixing tasks involve two or more of these mixing tasks simultaneously. In these cases, special attention must be given to the dominant mixing category. (Figure 3.) SYSTEMATIC CHARACTERIZATION OF THE MIXING PROCESS FOR THE BIOX® REACTOR

179


2/13/09

11:06 AM

Page 180

Figure 2. The EKATO ISOJET-B impeller

Figure 3. Basic mixing tasks

For the continuous BIOX® process all of these mixing tasks have to be fulfilled. Homogeneous process conditions, i.e. of oxygen, pH, temperature, nutrients, are very important to be able to establish and keep a stable bacterial population. All of these parameters are related to blending. High pumping rates and therefore short mixing times will ensure a high degree of homogeneity. 180



2/13/09

11:06 AM

Page 181

Since ore slurries are processed solids, these have to be suspended. A high degree of homogeneity for all particle sizes has to be achieved to prevent an accumulation of particles with time. The physical properties of the treated solids and the pure liquid have to be taken into account when designing the agitator. These properties will affect the hindered settling velocity of the particles which release a settling power. In order to maintain a defined degree of uniformity the agitator must provide a power input to the liquid to counteract this settling power. Air has to be dispersed so that enough surface is created between the air and the slurry. This will ensure that enough oxygen is transferred to the bacteria. The specific power input by agitation is one of the main parameters to ensure high mass transfer rates. Therefore it is necessary to have sufficient knowledge of the power number Ne of an impeller at gassed conditions. With gas being added to an impeller the friction between the impeller and the fluid will be reduced, so that the ungassed power number Ne0 is greatly diminished, as shown in Figure 4. The amount of gas being added to an impeller is described by the dimensionless gassing number Q,

where q is the gas feed rate, n the shaft speed and d2 the impeller diameter. Besides the air dispersion duty of the impeller it has to be confirmed that the impeller will not flood, meaning running completely in a gas cavity. The oxidation of sulphide minerals is an exothermic process and the reactors must be cooled continuously, meaning heat transfer is of importance as well. Cooling is done by circulating cold water through a series of cooling coils installed inside the reactors. Using the experience of an agitator manufacturer, the required power inputs to solve the separate mixing tasks can be evaluated. Calculations showed that the gas/liquid dispersion duty requires the highest level of power input and therefore is the dominant mixing task.

Figure 4. Reduced power input due to gassing

SYSTEMATIC CHARACTERIZATION OF THE MIXING PROCESS FOR THE BIOX® REACTOR

181


2/13/09

11:06 AM

Page 182

Test work EKATO laboratory As described earlier, the gas dispersion duty is the main mixing task and therefore the decisive design parameter. Therefore tests were mainly focused on the gas dispersion duty. Since the power input is of utmost importance, power number measurements have been done for different impeller types and gassing conditions. Measurements at the EKATO laboratory were carried out at two different test scales, 50 litres (Figure 5) and 1 000 litres. Air was, as in the production scale the BIOX® reactors, added via a sparge ring located below the impeller. Tests were run at different gas feed rates and shaft speeds and the torque measured. From the torque measurements the power number could be derived. Figure 6 shows the comparison of two impeller types for BIOX® conditions.

Figure 5. Test set-up 50 litres

Figure 6. Bioleaching—power input due to gassing

182



2/13/09

11:06 AM

Page 183

As can be seen, the power did not drop at the very low Q numbers as described before but increased. Two curves for different impeller types are given. This increase of the power number is a known characteristic of impellers at very low gassing rates, although so far no explanation can be found in literature. Anyhow, this increase in power number has to be known and considered to properly design the agitator. By choosing the appropriate test conditions in the 50 litre and 1000 litre scale a wide set of dimensionless data was obtained. This enables the design of the agitator for a required gassed power input for any production scale and therefore to fulfil the mass transfer mixing task. To compare different impeller types regarding their suspending performance, model particles were added to the test vessel while measuring the gassed power numbers. This allowed EKATO to compare different impeller types easily. Not surprisingly when tested, the radial flow types of impellers showed a poor solids suspension performance. Although these types of impellers in general are able to handle more gas than axial flow types, this disqualifies them for the BIOX® process. The EKATO ISOJET-B impeller showed the best suspending performance at gassed conditions. To compare the mass transfer performance of the different tested impeller types, a model reaction was applied at the 50 l scale. At comparable test conditions, i.e. the agitator power input, all impeller set-ups showed more or less the same mass transfer capability. All tests at the EKATO lab so far were carried out using model systems. To get additional information how the different impeller types perform in original slurry, a 50 l BIOX® sample was shipped from the Fairview mine site to EKATO. Slurry samples were taken from the vessel bottom and liquid surface and the particle concentrations compared. The ISOJET-B showed a very homogeneous distribution of the solids over the filling height; no differences in solids contents were measurable. The main advantage of the ISOJET-B was its fast resuspending capability for complete settled out conditions (after 16 h), see Figure 7.

Figure 7. Testing with original slurry—restart after 16 h


183


2/13/09

11:06 AM

Page 184

Fairview test reactor After the successful tests with the ISOJET-B at the EKATO technology centre a test reactor of 21 m³ live volume was installed at the Fairview BIOX® plant in South Africa. The test reactor is shown in Figure 8. A scale-up for the ISOJET-B was done using the data derived from model testing. An impeller with a diameter of 1 250 mm was machined and sent to the mine site. Again model tests were performed before finally testing the impeller at real BIOX® conditions. First the power number was measured for different gassing rates in water. Test results from the EKATO laboratory could be confirmed. After these tests mass transfer measurements were carried out to be able to predict the mass transfer performance for bigger production reactors. After these tests with model liquids, the ISOJET-B was tested at real plant conditions. Ore slurry was added from the production plant and the same data derived/analysed as for the production-scale reactors. • Mass transfer/chemical results—continuous testing over several weeks was carried out. Achieved oxidation rates and process specific parameters such as pH, dissolved oxygen, Fe2+ to Fe3+ conversion values, OUR etc., were measured during a daily routine sampling procedure. All measured data were within specifications and constant. The average amount of oxidized sulphur was higher than achieved in the adjacent production plant. • Solids suspension—Two litre slurry samples were taken at four sample nozzles at different filling heights. Solids’ homogeneity over the filling height was excellent. No difference in solids concentration from vessel bottom to filling level was observed. Summary and conclusions Extensive water testing and continuous slurry testing was performed at lab and pilot scale. Based on the favorable results generated from these test, the EKATO ISOJET-B impeller was classified as suitable impeller to be used in commercial BIOX® installations. Additional

Figure 8. Fairview test reactor 21 m³

184



2/13/09

11:06 AM

Page 185

information on the mixing characteristics in this specific process could be derived and helped to even better understand the process. Design values were confirmed by testing, and will be applied to design the production scale reactors for upcoming projects. General design parameters were agreed upon between BIOMIN and EKATO. This successful joint development shows the importance of close cooperation between technology licensors or end-users and equipment suppliers. It is vital for understanding all the factors influencing the mixing performance of processes. Exercises such as this will provide equipment suppliers and plant operators more with confidence in managing further scale-ups at ever larger engineered plants.

Wolfgang Keller Senior Process Engineer, EKATO RMT GmbH, Schopfheim, Germany Education: MSc Process Engineering University of Karlsruhe, Germany. Professional Memberships: VDI Current position: Senior Process Engineer working for EKATO Mixing Technology since 9 years. Previous affiliations: Assistant production manager and research and development at LONZA, Germany. Experience: • polymer plant operation • engineering / construction • process development • R&D mixing technology • EKATO expert for minerals processing.


185


186

2/13/09

11:06 AM

Page 186


187-212_Dreisinger:text

2/14/09

11:57 AM

Page 187

DREISINGER, D. Keynote address: Hydrometallurgical process development for complex ores and concentrates. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Keynote address: Hydrometallurgical process development for complex ores and concentrates D. DREISINGER University of British Columbia, Department of Materials Engineering, Canada

Hydrometallurgical processing of complex ores and concentrates is becoming increasingly important as the mining and metallurgical industry seeks to exploit mineral deposits that are difficult to treat using conventional mineral processing and pyrometallurgical technologies. Mineral processing is often challenged by the difficulty or inability to separate valuable minerals into marketable concentrates. Hydrometallurgical processing, using selective leaching technology, can often ‘chemically beneficiate’ such difficult deposits. Pyrometallurgical treatment of base metal concentrates is capital intensive and subject to ever more stringent environmental control. Hydrometallurgical processing is generally lower in capital cost for an equivalent metal production rate and avoids the gas and dust issues associated with pyrometallurgical processing. The possibility of byproduct recovery may also increase with hydrometallurgical treatment. Two examples of new technology or flowsheet development for treatment of complex ores and concentrates are used as illustration. These include the El Boleo process of Baja Mining for recovery of copper, cobalt, zinc and manganese from a complex clayey ore and the PLATSOL™ process for recovery of copper, nickel, cobalt, platinum, palladium and gold from a bulk sulphide concentrate. Introduction The treatment of base and precious metal deposits has as a final goal the recovery of final metal products of sufficient purity and in a suitable commercial form to meet customer specifications. Historically for metals like copper, nickel, zinc and lead recovery from sulphide deposits, processing has included mining, flotation of concentrates and finally smelting/refining of the concentrates through to final products (generally high purity metals). However, a number of factors are causing a gradual but steady shift away from the traditional processing routes. Mineral processing technologies may be constrained in the processing of complex ores (fine grained) or ores that contain multiple valuable metals. Hydrometallurgical leaching technology may actually benefit from the fine grained character of the raw material and offers the possibility of recovery of multiple metals from a single feed source. Similarly, pyrometallurgical processing of sulphide concentrates is becoming increasingly expensive, KEYNOTE ADDRESS: HYDROMETALLURGICAL PROCESS DEVELOPMENT

187


2/14/09

11:57 AM

Page 188

environmental constraints due to emission of gases and production of dusts are increasing and pyrometallurgical processing often results in the losses of valuable metal byproducts into slags and residues. Hydrometallurgical technologies have lower unit capital costs, avoid production of gaseous or dust emissions, and can manage recovery of multiple metals from a single feed source. In this way, hydrometallurgical processing is growing in importance to the global metal industry as shown in Figure 1. The success of hydrometallurgy as a technological field lies in the ability to extract valuable metals into solution, purify the metal containing solutions to concentrate valuable metals and reject impurities, and finally to recover a pure metal product by electrolysis, metal precipitation or other means. Advances in hydrometallurgy are rooted in improvements in the science and technology of leaching, solid-liquid separation, solution purification (solvent extraction, ion exchange, cementation) and metal recovery. The following sections illustrate two new process developments that highlight the potential of hydrometallurgical treatment to solve ‘difficult’ metallurgy and extract base and precious metals into marketable form. The Boleo process1 is applied to the recovery of copper, cobalt, zinc and manganese from a mixed sulphide/oxide deposit hosted in clay. The keys to process development for Boleo were to incorporate novel seawater based leaching, high rate thickening for solid/liquid separation, the CSIRO DSX technology2 for cobalt and zinc recovery away from manganese, and finally the use of manganese precipitation for recovery of a manganese carbonate by-product. The PLATSOL™ process 3 was developed to process mixed base and precious metal sulphide concentrates containing copper, nickel, cobalt, platinum, palladium, gold and silver where the grade and nature of the concentrates produced by flotation were not amenable to toll smelting. The PLATSOL™ process uses chloride assisted total pressure oxidation of bulk sulphide concentrates to extract base and precious metals into an autoclave solution. The ability to directly extract precious metals as chloro-complexes is the key to the PLATSOL™ process development. The Boleo process The Boleo Copper-Cobalt-Zinc-Manganese Project of Baja Mining Corp. is situated adjacent to Santa Rosalia on the Baja Peninsula of Mexico (Figure 2). The Boleo deposit has a large geological resource with 277 million tonnes @1.77% Cu. Eq grade of measured and indicated and 253 million tonnes @ 1.29% Cu Eq. grade of inferred material (www.bajamining.com).

Figure 1. Growing importance of hydrometallurgy

188



2/14/09

11:57 AM

Page 189

Figure 2. The Boleo mine site close to Santa Rosalia, Baja California

The ore will be treated by a hydrometallurgical process involving acid—seawater leaching with recovery of copper and cobalt metal cathode, zinc sulphate crystal and eventually manganese carbonate precipitate. Process development of the metallurgical flowsheet for Boleo has gone through bench and integrated pilot-plant testing. An initial ‘proof of principle’ pilot plant was executed at SGS Lakefield Research in Canada in November 2004. The major focus of the proof of principle pilot plant was to confirm that the clayey Boleo mineralization could be thickened and washed in a conventional CCD train using high rate thickeners and that the CSIRO ‘DSX’ solvent extraction system could be used to recover cobalt and zinc. The DSX system involves mixing LIX 63 with Versatic 10 extractant to improve selectivity for (Co+Zn)/Mn. Since the initial proof of principle pilot plant in 2004, further work has been undertaken with CSIRO (Australia) to optimize the DSX solvent extractant composition and to add a manganese recovery process to the Boleo flowsheet. The simple addition of sodium carbonate (soda ash) to the DSX raffinate was found to precipitate a manganese carbonate product of high purity. This manganese carbonate material may form a feedstock to the production of manganese chemicals, manganese metal or electrolytic manganese dioxide. The improved metallurgical flowsheet for Boleo has now been evaluated in a fully integrated demonstration pilot plant at SGS Lakefield Research in Canada. Metallurgical treatment of Boleo ore After a quick review of the major elements of the process, the focus will be on the demonstration pilot plant results. Figures 3–5 show the overall flowsheet in simplified format. Milling and leaching circuit The Boleo ores are clayey and generally fine grained and easily broken. The milling circuit design takes material from the ROM pad into a static grizzly with the oversize to a primary crusher. The combined product is then scrubbed in copper solvent extraction raffinate. The KEYNOTE ADDRESS: HYDROMETALLURGICAL PROCESS DEVELOPMENT

189


2/14/09

11:57 AM

Page 190

Figure 3. Flowsheet for Boleo process—Part 1: feed preparation, leach, CCD, copper SX/EW, iron removal, acid/power plant

scrubber product is screened with the oversize sent to a secondary crusher. The combined product goes to the ball mill sump before cycloning. The coarse cyclone product is directed to the ball mill. Milling is conducted in raffinate. (Figure 3.) The milled product is heated to a target temperature of 80ºC for atmospheric leaching. Leaching proceeds in two steps: acid, oxidative leaching with manganese dioxide oxidation for leaching of Cu, Zn and Co sulphides and: acid reductive leaching with addition of sulphur dioxide gas to reduce residual manganese dioxides. The reductive leaching of surplus manganese dioxide is essential for maximizing the extraction of cobalt. Oxidation leaching (acid leaching with manganese dioxide in the ore) Cu2S + 2MnO2 + 4H2SO4 = 2CuSO4 + 2MnSO4 + S + 4H2O ZnS + MnO2 + 2H2SO4 = ZnSO4 + MnSO4 + S + 2H2O CoS + MnO2 + 2H2SO4 = CoSO4 + MnSO4 + S + 2H2O Reduction leaching (addition of sulphur dioxide to the ore slurry) MnO2 + SO2 = MnSO4 The slurry from oxidation and reduction leaching is then partially neutralized using local limestone that is available on the Boleo mine lease. The purity of the limestone is about 65–70% and it is highly reactive for acid neutralization. A small amount of air is added during partial neutralization to ensure that there is no residual cuprous ion (that may form if reductive leaching goes to too low a redox value) in solution. Cuprous ion is not extracted by conventional oxime solvent extraction reagents. Solid-liquid separation and CCD washing The first proof of principle pilot plant confirmed that the Boleo leach pulp could be thickened and washed using high rate thickening. High rate thickening involves the dilution of the 190



2/14/09

11:57 AM

Page 191

incoming slurry (by recycle of overflow from the same thickener) so as to create a dilute slurry (3–5% solids) for flocculation and settling into the thickener bed. This approach was verified by testing performed by Pocock Industrial and by Outokumpu in lab and pilot-scale testing at SGS—Lakefield. The wash solution to be used at Boleo consists of a combination of barren solution (after cobalt, zinc and manganese removal) and fresh brine solution. Copper solvent extraction and electrowinning The recovery of copper from complex chloride containing solutions is feasible using modern selective copper solvent extractants combined with a ‘wash stage’ during SX recovery of copper to prevent transfer of chloride from leaching through to electrowinning. This approach has been widely reported at plants in Chile with saline PLS solutions and has been adopted for Boleo as well. High levels of copper extraction are expected due to the low free acid level in solution (after partial neutralization) and the use of a strong oxime formulation. (Figure 4.) Copper extraction CuSO4 + 2HR(org) = CuR2(org) + H2SO4 Copper stripping CuR2(org) + H2SO4 = CuSO4 + 2HR(org) Copper electrowinning CuSO4 + H2O = Cu + O2(gas) + H2SO4

Figure 4. Flowsheet for Boleo process – Part 2: cobalt and zinc SX using DSX, zinc sulphate evaporation, manganese carbonate precipitation and limestone milling

KEYNOTE ADDRESS: HYDROMETALLURGICAL PROCESS DEVELOPMENT

191


2/14/09

11:57 AM

Page 192

Copper is electrowon conventionally using permanent cathode blank technology. LME Grade A cathode quality was produced in the proof of principle pilot plant and is also expected to be produced in the commercial plant. The copper solvent extraction raffinate contains free acid (from copper solvent extraction) and so it is advantageous to return a portion of the raffinate to the milling and leaching circuit, to reduce overall use of fresh acid and to concentrate Co, Zn and Mn in solution. The balance of the raffinate is directed to the Co, Zn, Mn recovery circuit. Iron removal The removal of iron from the raffinate advancing to Co, Zn, Mn recovery is accomplished with pH adjustment and aeration. The bulk of the iron in solution will be present as ferrous. Ferrous is oxidized and precipitated as ferric hydroxide precipitates. Aluminum will coprecipitate with iron in this step. Two stages of iron removal were allowed in the pilot plant design. The first stage is used to precipitate more than 95% of the iron with the second stage precipitation reducing iron to less than 10 ppm residual. The second stage precipitate is returned to oxidative leaching to ensure that any co-precipitated ‘pay-metals’ are recovered. The first stage cake is filtered, washed and then repulped and sent to tailings. The iron removal solution is then advanced to cobalt and zinc solvent extraction. Cobalt and zinc solvent extraction CSIRO (Australia) has developed a range of synergistic extractants, tailored to specific metal separations. In the case of Boleo, the major challenge is the extraction of cobalt and zinc without extraction of manganese. The particular synergistic system selected for Boleo is the Versatic 10 – LIX 63 mixture. Cobalt is loaded in preference to zinc, which in turn is loaded in preference to manganese with this system.

Figure 5. Flowsheet for Boleo process – Part 3: zinc and cobalt solvent extraction and cobalt electrowinning

192



2/14/09

11:57 AM

Page 193

The cobalt and zinc circuit is designed for bulk extraction and bulk stripping of zinc and cobalt. Some of the zinc strip solution is also employed as a manganese scrub solution makeup. Zinc and cobalt loading must be done with pH control. The most convenient alkali for this duty is sodium carbonate. Bulk stripping of zinc and cobalt is performed using sulphuric acid and pH control. Cobalt and zinc solvent extraction (with small amount of Mn co-extraction) CoSO4 + 2HR(org) + Na2CO3 = CoR2(org) + Na2SO4 + CO2(g) + H2O ZnSO4 + 2HR(org) + Na2CO3 = ZnR2(org) + Na2SO4 + CO2(g) + H2O MnSO4 + 2HR(org) + Na2CO3 = MnR2(org) + Na2SO4 + CO2(g) + H2O Manganese scrubbing MnR2(org) + ZnSO4 = ZnR2(org) + MnSO4 Zinc and cobalt stripping ZnR2(org) + H2SO4 = ZnSO4 + 2HR(org) CoR2(org) + H2SO4 = CoSO4 + 2HR(org) The fully stripped organic is then recycled back to loading. The DSX raffinate advances to manganese recovery. Cadmium control Cadmium is a minor impurity in the Boleo ore and must be removed from both the zinc/cobalt strip solution prior to zinc sulphate extraction and crystallization and cobalt solvent extraction and electrowinning. The method chosen for cadmium control is zinc dust cementation. Zinc dust is widely used for cadmium control in conventional roast-leach-electrowin (RLE) plants for zinc recovery from concentrates. In this application, it is important to cement only cadmium and not cobalt. This can be accomplished using lower temperature and not ‘activating’ the cementation. Activation with Cu/As or Cu/Sb is the method used in the removal of cobalt by cementation in conventional RLE plants. Cadmium cementation CdSO4 + Zn = Cd + ZnSO4 Cobalt metal electrowinning The zinc/cobalt (bulk) strip solution from DSX will be treated sequentially by zinc solvent extraction and stripping and then cobalt solvent extraction, stripping and electrowinning. The zinc and cobalt solvent extraction circuits will both use Cyanex 272 extractant. The zinc sulphate strip solution will be sent to zinc sulphate crystallization. The cobalt strip solution is purified and electrowon. There is provision in the cobalt electrolysis circuit for ion exchange polishing of minor elements prior to sulphate electrolysis. The cobalt cathode deposit is harvested, crushed and marketed. Zinc extraction (Cyanex 272) ZnSO4 + 2HR(org) + Na2CO3 = ZnR2(org) + Na2SO4 + CO2(g) + H2O Zinc stripping ZnR2(org) + H2SO4 = ZnSO4 + 2HR(org) Cobalt extraction (Cyanex 272) CoSO4 + 2HR(org) + Na2CO3 = CoR2(org) + Na2SO4 + CO2(g) + H2O KEYNOTE ADDRESS: HYDROMETALLURGICAL PROCESS DEVELOPMENT

193


2/14/09

11:57 AM

Page 194

Cobalt stripping CoR2(org) + H2SO4 = CoSO4 + 2HR(org) Cobalt electrowinning CoSO4 + H2O = Co + O2(gas) + H2SO4 Zinc sulphate crystallization The recovery of zinc as a metal product by electrolysis is technically challenging and probably not cost-effective at a low production rate (notwithstanding the current prices of zinc metal). Hence, the zinc product selected for the Boleo process is zinc sulphate crystals. Zinc sulphate is used in a variety of agricultural formulations (fertilizers and feeds). Zinc sulphate crystallization ZnSO4 (aqueous) + H2O = ZnSO4.H2O (crystals) Manganese carbonate recovery The raffinate from the DSX circuit has been subjected to iron/aluminum removal and the use of DSX to recover cobalt and zinc. The DSX is also effective at removing other heavy metals from solution. The remaining cations in solution are largely manganese, magnesium, calcium and sodium. The manganese rich nature of the Boleo ore provides a perfect opportunity for recovery of a manganese by-product from this solution. The simplest and most cost-effective method to precipitate manganese from solution involves addition of sub-stoichiometric amounts of sodium carbonate to the DSX raffinate. The reason for sub-stoichiometric addition is to prevent co-precipitation of impurities such as calcium and magnesium and various minor elements. Manganese precipitation MnSO4 + Na2CO3 = MnCO3 + Na2SO4 Manganese carbonate is not hydrated and hence is lower in weight (relative to hydroxide precipitates) for shipping off site. Initial proof-of-principle testing indicated that the precipitate graded 46–48% Mn after washing and drying of free moisture. Demonstration pilot-plant results Feed preparation The blended feed material was sampled during each milling campaign. The average assay was 2.18% Cu, 0.135% Co, 0.49% Zn, 5.01% Mn, 8.26% Fe, 1.08% Ca, 2.85% Mg, 5.18% Al, 0.014% Ni and 19.7% Si. The average particle size for the milled feed was 80% passing 38 microns. The feed rate to leaching was set at 14 kg/h. Over the 16 day pilot plant, approximately 5 tonnes of Boleo feed material were processed. Leaching, partial neutralization and counter-current decantation The leaching conditions were varied during the 16-day pilot run. Specifically the pH in oxidative leaching was varied from around 1.2 to 1.7. The extractions of Cu, Co, Zn were generally increased when the pH was lowered. However, the addition of acid to maintain a low pH leach environment can be excessive. The partial neutralization was performed with Boleo limestone addition. This worked well and produced a pH 2 product slurry to CCD. The CCD itself generally performed very well. The use of high rate thickeners with overflow recycle for dilution of the feed to ~ 3% solids worked well. The solids settled rapidly, producing a well clarified, low-solids overflow as feed to copper SX. (Table I.) 194



2/14/09

11:57 AM

Page 195

Table I Summary of metal extraction and reagent addition as a function of leach pH Oxid Leach pH

Red. Leach ORP (mV vs Ag/AgCl)

H2SO4 (kg/t)

SO2 (kg/t)

Limestone (kg/t)

Cu

Co

399 397 399 427

225 235 315 513

80 73 124 152

68 75 113 368

90.9 90.9 92.7 94.4

82.6 81.4 83.8 90.3

53.9 55.4 61.0 72.3

97.1 96.4 98.0 96.7

Ni

Sn

Ag

S

<1 <1 <1

<1 <1 <1

<25 <25 <25

<15 17.4 <15

1.7 1.5 1.4 1.2

Extraction (%) Zn Mn

Table II Analysis of three copper cathode samples Sample

Cathode 1 Cathode 2 Cathode 3

Se

Te

Bi

Sb

Pb

<1 <1 <1

<1 <1 <1

<1 <1 <1

<1 <1 <1

<1 1.2 <1

Analysis (g/t) As Fe <1 <1 <1

2.4 1.4 1.3

Copper solvent extraction and electrowinning The CCD overflow solution advanced to a conventional solvent extraction and electrowinning circuit utilizing 2E – 1Sc – 2S configuration. An organic solution of 20% LIX 664N (Cognis) in ORFOM SX80 CT (Chevron Phillips) was used for copper recovery. Copper was electrolysed on stainless steel blanks at ~ 250 A/m2 current density. The copper concentration in the feed to copper SX was typically between 2 000 and 3 000 mg/l (2–3 g/l) with raffinates ranging between 10 and 50 mg/l. This represents 95–99% copper extraction through most of the pilot-plant run. The raffinate was largely recycled to leach during the pilot plant allowing for secondary recovery of a significant portion of the copper remaining in the first raffinate. A total of 47.4 kg of copper was plated in 4 cycles during the pilot plant at a current efficiency of 97%. Table II shows the assays for three cathode samples from the pilot plant. With the exception of Cathode 2 with a slightly high level of sulphur in cathode, the copper quality was excellent. It is expected that copper produced commercially at Boleo would exceed requirements for LME Grade A or COMEX Grade 1 cathode. The cathode product was smooth, except for slight nodulation at the edges. Iron and aluminum removal The iron and aluminum removal circuit operated successfully, typically producing less than 10 mg/l of residual iron in the feed to the DSX circuit. The circuit was started as a two-stage circuit but after commissioning was changed to one-stage as the second stage circuit was deemed to be superfluous. For the purpose of the pilot plant, supplemental oxidation with hydrogen peroxide addition was required to ensure complete iron oxidation and removal. The iron precipitate thickened well but demonstrated variable filterability, depending on the operating temperature of the circuit (T > 45°C produced a more easily filtered product). Cobalt and zinc recovery using the DSX circuit Cobalt and zinc were recovered using the CSIRO DSX system. The extractant used was KEYNOTE ADDRESS: HYDROMETALLURGICAL PROCESS DEVELOPMENT

195


2/14/09

11:57 AM

Page 196

13.2% LIX 63 and 6.25% Versatic 10 in Orfom SX80CT. The DSX circuit consisted of 3 extract stages (pH ~ 4.5), 2 scrub stages (with zinc sulphate solution), 2 zinc strip stages at pH 3.2 and 2 bulk strip stages at pH 1. The results were excellent with > 99% recovery of both cobalt and zinc with near perfect rejection of manganese. Selective stripping was trialed in the — – i.e. too much cobalt was stripped in the zinc selective strip. A single bulk strip to zinc and cobalt recovery circuits has been selected for the commercial plant. Zinc solvent extraction The zinc solvent extraction circuit used 30% Cyanex 272 in Orfom SX 80CT. The circuit consisted of 4 extraction stages at pH 2.6–2.9, 2 scrub stages (using zinc strip solution) and 2 strip stages at pH 1. The zinc concentration in the raffinate was controlled to around 100 mg/l Zn. The zinc strip solution was gradually increased to nearly 80 g/l (80 000 mg/l Zn). The cobalt concentration of the zinc strip solution was less than 20 mg/l Co (Zn:Co ratio of 4 000:1), confirming an excellent separation of zinc and cobalt. Cobalt solvent extraction and electrowinning The raffinate from zinc solvent extraction was processed via cobalt solvent extraction and electrowinning. Cobalt was extracted in 4 stages at pH 5.2 to 5.5 using 30% Cyanex 272 in Orfom SX 80CT followed by 2 stages of scrubbing, 3 stages of stripping (with spent electrolyte) and 1 stage of conditioning before organic recycle. The cobalt strip solution was polished by using DOWEX M4195 and PUROLITE S-950 resins for minor element capture prior to cobalt electrowinning in a divided cell. Cobalt was deposited at 250 A/m2 at 70ºC. The residual zinc entering the cobalt circuit is extracted and stripped in the conditioning step. The conditioning solution is recycled back to zinc solvent extraction, permitting recovery of this zinc. The average feed solution concentration was about 6 000 mg/l of Co with raffinates of less than 1 mg/l indicating nearly 100% cobalt recovery in this process. The cobalt strip solutions rose to nearly 90 g/l (again due to optimization) allowing effective cobalt electrolysis. Table III shows the analysis of two cobalt samples from the production. The cobalt is of very high purity with respect to key elements. There may be opportunity to further reduce the lead levels in cathode by strontium carbonate treatment of Co EW cell feed. Similarly, even lower levels of Cd in cathode may be achieved by more efficient Cd cementation ahead of the zinc and cobalt solvent extraction circuits. Manganese carbonate precipitation The DSX raffinate was treated with sodium carbonate solution to selectively precipitate manganese carbonate as a potential by-product from the Boleo process. The manganese

Table III Analysis of two cobalt cathode samples Sample

Cathode 1 Cathode 2

196

Analysis (ppm) Cd

Cu

Fe

Mn

Ni

Pb

Zn

66 43

<2 <2

<4 <4

<0.3 <0.3

55 56

40 41

<10 <10



2/14/09

11:57 AM

Page 197

carbonate formed as a fine pinkish precipitate that settled, filtered and washed very well. The typical composition of the manganese carbonate is shown below. Nearly 200 kg of wet MnCO3 was recovered. (Table IV.) The Boleo project of Baja Mining Corp continues to move forward to production. The process selected for Boleo involves oxidative and reductive leaching of the milled ore, conventional countercurrent decantation (CCD) washing of the leached ore in high rate thickeners, copper solvent extraction and electrowinning, cobalt and zinc solvent extraction and recovery as zinc sulphate crystals and cobalt metal and finally manganese carbonate production as a by-product. The demonstration pilot plant for the Boleo project was built and operated at SGS Lakefield Research in Canada in June and July of 2006. A total of about 5 tonnes of blended feed were processed. The optimized extraction of metals yielded 91% Cu, 82% Co, 55% Zn and 97% Mn extraction. These values represent the overall extraction after oxidation and reduction leaching, partial neutralization and CCD 6 washing. The oxidation leach condition was set at pH 1.5 to 1.7 to control the addition of acid. The leach residue was effectively settled and washed in a 6 stage countercurrent decantation circuit using ‘high–rate’ thickener design. Copper was very efficiently recovered by SX/EW and electrowon as smooth, high purity copper cathode. A portion of the copper raffinate was treated to remove iron and aluminum by oxidation and pH adjustment followed by DSX recovery of cobalt and zinc. The CSIRO DSX system performed extremely well with >99% recovery of cobalt and zinc from the DSX feed with nearly complete manganese rejection. Zinc was recovered as high purity zinc sulphate solution and cobalt was recovered as cathode metal of high purity. Finally, manganese carbonate grading up to 47% Mn was recovered by precipitation with sodium carbonate. The Platsol™ process PolyMet Mining Corp. (PolyMet) is advancing the development of the NorthMet Project. The NorthMet deposit is located in northern Minnesota, adjacent to the historic Iron Range. The deposit was discovered in the 1960s and consists of a large, magmatic, disseminated sulphide, polymetallic deposit with values in Cu, Ni, Co, Zn, Au, Ag and platinum group metals (PGM). The shallow, tabular orebody may be mined by open-pit methods with minimal prestrip and a low waste:ore strip ratio. The NorthMet deposit is large with measured and indicated resources of 638.2 million tonnes and an inferred mineral resource of a further 251.6 million tonnes. Within this resource, there is a proven and probably reserve of 274.7 million tonnes and the operating plan is to mine 224 million tonnes of this reserve over the 20 year life of the operation (32 000 tpd). A definitive feasibility study (DFS) has been completed by Bateman Engineering (Australia) in 2006. The project is currently in the environmental permitting phase of the project. The heart of the processing strategy revolves around the acquisition of the Erie Plant from

Table IV Typical analyses of MnCO3 precipitate Analysis (%)

Analysis (g/t)

Species

Mn

Ca

Mg

Na

Al

CO3

SO4

Ni

Zn

Fe

Cu

Co

Cd

Cl

Precipitate

47

1

0.3

0.6

0.5

45

2

350

300

200

<5

50

<5

100


197


2/14/09

11:57 AM

Page 198

Cleveland Cliffs. The Erie plant acquisition allows for PolyMet to proceed with a ‘brownfields’ processing strategy with an existing 100 000 tpd crushing and grinding plant and associated infrastructure (roads, power, railway, water supply) including waste disposal. The Erie plant was built in the 1950s for approximately $355 million USD and operated until 2001 processing taconite ore. The acquisition of this plant and related infrastructure has provided most of the heavy equipment needed for production. Initial plans are to use approximately one-third of the existing plant capacity. Figure 6 below shows the NorthMet site in northern Minnesota, adjacent to the Mesabi Iron Range. Figure 7 provides an overview of the Erie plant site with the tailings area in the background. Figure 8 shows the (existing) indoor view of the Erie plant milling circuit with parallel rod and ball mill lines.

Figure 6. Location of the NorthMet property in northern Minnesota

Figure 7. The Erie plant site

198



2/14/09

11:57 AM

Page 199

Figure 8. The Erie plant milling circuit with parallel rod and ball mill lines

The current development strategy for NorthMet is to produce a saleable concentrate product upon plant start-up and then move to implementation of a hydrometallurgical process plant for recovery of the valuable metals. The PLATSOL™ process has been selected. The DFS programme has included several pilot-scale test programmes at SGS Lakefield Research Limited in Ontario, Canada. Pilot-plant tests have been performed on ore flotation to produce bulk concentrate samples and on hydrometallurgical treatment of the bulk concentrate materials. The ‘base case’ flowsheet involves high temperature chloride-assisted leaching (PLATSOL™), solid-liquid separation and washing, PGM/Au precipitation, neutralization and Cu SX/EW to produce copper cathode. A portion of the Cu SX raffinate is treated for nickel and cobalt recovery (the remainder is recycled to the autoclave) sequentially by neutralization, first and second stage iron/aluminum removal, residual copper removal by precipitation as a sulphide, first stage Ni/Co/Zn precipitation with magnesia, second stage Ni/Co/Zn precipitation with lime, and then magnesium removal with lime. The product from this treatment is a ‘mixed’ hydroxide of nickel, cobalt and zinc. NorthMet process flowsheet and pilot-plant studies The overall NorthMet process flowsheet involves two distinct circuits: a classical mineral processing flowsheet to produce a bulk concentrate and a hydrometallurgy flowsheet for extracting Cu-Ni-Co-Zn-Au-PGM from the bulk concentrate. The further separation of the bulk concentrate into copper and nickel concentrates for sale is a further refinement of the basic flowsheet. The hydrometallurgy flowsheet involves chloride-assisted leaching of base and precious metals followed by a series of metal recovery steps for the base and precious metals. Figures 9, 10 show the process flowsheets in pictorial form. Figure 9 is the mineral processing flowsheet (to produce a bulk concentrate) with Figure 10 showing the hydrometallurgy circuit.The pilot-scale tests for the DFS at SGS-Lakefield were performed in KEYNOTE ADDRESS: HYDROMETALLURGICAL PROCESS DEVELOPMENT

199


2/14/09

11:57 AM

Page 200

Figure 9. NorthMet process flowsheet: crushing, grinding and flotation of a bulk Cu-Ni-Co-Zn-Au-PGM concentrate

Figure 10. NorthMet process flowsheet: hydrometallurgical treatment of bulk Cu-Ni-Co-Zn-Au-PGM concentrate with production of copper cathode, Au/PGM precipitate and mixed Ni/Co hydroxide

200



2/14/09

11:57 AM

Page 201

2005 and 2006. The ore samples sent to SGS-Lakefield comprised four composites and a sample of assay reject material that was used as ‘start-up’ material for the flotation pilot plant. The ore samples were processed through crushing, grinding and flotation to produce a series of bulk concentrates for hydrometallurgical pilot-plant testing. Table V below summarizes the pilot plant feed assays and weights processed. The following is a general description of each part of the proposed plant along with selected results from the 2005 and 2006 continuous pilot testing of the process at SGS Lakefield. The pilot-scale tests at SGS Lakefield form the test work basis for the final feasibility study design for the NorthMet plant. Crushing, grinding and flotation The crushing, grinding and flotation circuit (Figure 9) receives ore from the mine into a coarse crusher dump pocket. The ore then proceeds through primary, secondary, tertiary and quaternary crushing followed by rod and ball milling to a size of 100–125 μm P80. The crushing and grinding circuits are existing circuits within the ‘Cliff’s Erie’ plant and will utilize only a third of the available capacity. The flotation circuit is designed to produce a bulk concentrate carrying pay metals and a tailing with low residual sulphur content. The float circuit has the following elements: • Rougher conditioning with potassium amyl xanthate (PAX) • Rougher flotation with MIBC/DF250 addition • Scavenger flotation conditioning with copper sulphate (to activate sulphides) • Scavenger flotation with application of PAX, MIBC and DF250 • The rougher float concentrate (and the first scavenger float concentrate—not shown on Figure 9) is sent to conditioning followed by three stages of cleaning to produce a final concentrate • The scavenger flotation concentrate and the first cleaner flotation tailings are reground to 25–30 μm particle size (P80) and directed back to the rougher flotation circuit • The third cleaner concentrate is thickened and reground to approximately 15 μm particle size (P80) and sent to the autoclave feed tanks. A series of batch and continuous pilot-plant crushing/grinding/flotation tests was performed at SGS-Lakefield as part of the metallurgical development programme. The results of a series of six test periods are summarized in Table VI. The first two sample periods were treating assay reject ‘start-up’ material followed by treatment of a composite (C4). The C4 composite sample was obtained from NorthMet deposit in the area of the first five years of the NorthMet mine life. On average for the four periods in which C4 sample was processed, the results indicate: Table V Pilot plant ore samples processed at SGS-Lakefield Research in 2005 and 2006 Sample*

Composite 1 Composite 2 Composite 3 Start-up material Composite 4

Weight (t)

9.5 19.2 9.6 3.6 4.4

Assay (% or g/t) Cu

Ni

S

Pt

Pd

Au

0.31 0.35 0.40 0.31 0.31

0.095 0.10 0.11 0.09 0.10

0.73 0.77 0.87 0.87 0.91

0.06 0.07 0.09 0.04 0.05

0.23 0.33 0.30 0.31 0.28

0.03 0.05 0.05 0.05 0.08

*Composites 1–3 were processed in 2005 and the start-up material and composite 4 were processed in 2006


201


2/14/09

11:58 AM

Page 202

Table VI Selected flotation pilot-plant test results obtained at SGS Lakefield in 2006 Test Comp Product

Wt

No.

%

1

2

Start-up

Start-up

C4

C4

C4

C4

0.05

Au

Pt

Pd

96.8 0.036 0.035 0.13 0.007 0.008 0.006 0.044

9.9

Conc

3.17

10.1

2.20 20.8 0.110

1.28

1.46

6.71

90.1 67.6 83.8 34.0 84.0 88.8 83.3

Feed

100.0 0.38

0.11 0.85 0.008

0.05

0.05

0.25 100.0 100.0 100.0 100.0 100.0 100.0 100.0

0.26 100.0 100.0 100.0 100.0 100.0 100.0 100.0 32.4 16.2 66.0 16.0 11.2 16.7

96.2 0.032 0.030 0.12 0.005 0.009 0.009 0.043

8.1

Conc

3.80

9.22

2.04 19.3 0.095

1.17

1.17

6.47

91.9 73.0 85.9 42.8 83.7 83.7 85.6

Feed

100.0 0.34

27.0 14.1 57.2 16.3 16.3 3.4

0.10 0.87 0.008

0.04

0.06

0.29 100.0 100.0 100.0 100.0 100.0 100.0 100.0

96.4 0.034 0.030 0.12 0.005 0.014 0.008 0.053

9.6

Conc

3.58

8.57

1.97 21.1 0.095

0.83

1.41

6.63

90.4 71.3 86.4 41.4 68.8 86.7 82.3

Feed

100.0 0.34

0.10 0.93 0.008

0.04

0.07

0.29 100.0 100.0 100.0 100.0 100.0 100.0 100.0

28.7 13.6 58.6 31.2 13.3 17.7

95.8 0.037 0.029 0.11 0.005 0.006 0.005 0.051 10.5 28.7 11.0 62.7 13.6 6.7 16.8

Conc

4.20

7.16

1.66 19.8 0.068

0.87

1.59

5.76

Feed

100.0 0.33

0.10 0.85 0.009

0.04

0.06

0.28 100.0 100.0 100.0 100.0 100.0 100.0 100.0

Tailing

6

0.05

Co

Tail

Tailing

5

0.10 0.79 0.010

S

100.0 0.35

Tailing

4

Distribution, % Ni

Feed

Tailing

3

Grade Cu, % Ni, % S, % Co, % Au, g/t Pt, g/t Pd, g/t Cu

89.5 71.3 89.0 37.3 86.4 93.3 83.2

96.3 0.032 0.028 0.10 0.006 0.010 0.008 0.044

9.3

Conc

3.74

8.05

1.91 20.0 0.088

0.74

1.30

6.25

90.7 72.7 88.5 36.3 74.2 86.3 84.7

Feed

100.0 0.32

0.10 0.86 0.010

0.03

0.06

0.26 100.0 100.0 100.0 100.0 100.0 100.0 100.0

Tailing Conc

27.3 11.5 63.7 25.8 13.7 15.3

96.6 0.039 0.034 0.13 0.007 0.013 0.018 0.052 11.6 33.1 15.1 70.2 36.2 27.6 19.4 3.41

8.4

1.97 21.5 0.084

0.65

1.34

6.12

88.4 66.9 84.9 29.8 63.8 72.4 80.6

• 3.73% mass pull and float recoveries of 89.8% for Cu, 70.5% for Ni, 87.2% for S, 36.2% for Co, 73.3% for Au, 84.7% for Pt and 82.7% for Pd. A recent series of flotation tests has been performed using two further samples of PolyMet ore at 0.30 and 0.25% Cu head grade with the goal of producing a higher grade bulk concentrate product. The results of the recent series of tests are encouraging. The mass pull was reduced by cleaning more aggressively and use of a gangue depressant in the cleaning stages. The higher grade bulk concentrate will be separated into copper and nickel rich concentrates using conventional separation methods. Autoclave leaching The autoclave leaching process for NorthMet utilizes a small amount of chloride (approximately 7–10 g/l) in solution under ‘total pressure oxidation’ conditions to extract Cu, Ni, Co, Zn, Au, Pt, Pd from the bulk concentrate. The principles of the PLATSOL® process for NorthMet ore processing have been described elsewhere 3 and will be only briefly reviewed here. The process is regarded as a nominal ‘step-out’ from commercial practice in total pressure oxidation for base and precious metal ore/concentrate treatment. The autoclave oxidation process converts metal sulphide minerals into metal sulphates and iron hydrolysis products (primarily hematite but some basic ferric sulphate may also form under high acid conditions), while the precious metals are converted to chloro-complexes. 202



2/14/09

11:58 AM

Page 203

The chemical reactions believed to occur in the autoclave are shown below. (Note that the mineralogy of the PGMs may be very complex, but for simplicity only the metallic species are considered.) Chalcopyrite oxidation/iron hydrolysis CuFeS2 + 4.25O2 + H2O = CuSO4 + 0.5Fe2O3 + H2SO4 Pyrite oxidation FeS2 + 3.75O2 + 2H2O = 0.5Fe2O3 + 2H2SO4 Pyrrhotite oxidation Fe7S8 + 16.25O2 + 8H2O = 3.5Fe2O3 + 8H2SO4 Nickel sulphide oxidation NiS + 2O2 = NiSO4 Basic ferric sulphate formation Fe2O3 + 2H2SO4 = 2Fe(OH)SO4 + H2O Gold oxidation/chlorocomplex formation Au + 0.75O2 + 4HCl = HAuCl4 + 1.5H2O Platinum oxidation/chlorocomplex formation Pt + O2 + 6HCl = H2PtCl6 + 2H2O Palladium oxidation/chlorocomplex formation Pd + 0.5O2 + 4HCl = H2PdCl4 + H2O The temperature range for total pressure oxidation is typically 220 to 230°C. The autoclave discharge from the leach process contains dissolved Cu, Ni, Co, Zn, Au, Pt and Pd. The solids are relatively ‘barren’ of value, consisting of iron precipitates, unreacted gangue, and minor amounts of residual base and precious metal minerals. The original PLATSOL® pilot plant utilized a ‘straight through’ design in which fresh concentrate was introduced into the first compartment of a six-compartment pilot autoclave with raffinate (for controlled cooling of the autoclave slurry) and oxygen (for oxidation). The results of the ‘straight through’ pilot plant indicated that virtually all the copper and nickel were extracted in the first compartment of the autoclave due to the rapidity of the sulphide oxidation process. However, the platinum and palladium minerals continued to react through the entire autoclave volume, reaching ultimate extractions of about 95%. This situation was revisited in the more recent bench and pilot-scale testing. It was concluded that a ‘recycle’ design would be more efficient with respect to autoclave design. The ‘recycle’ design involves thickening the autoclave discharge and recycling a portion of the underflow to the autoclave feed. This recycle allows for any unreacted mineral to have a chance at second-pass extraction. The greater the recycle, the longer the ‘average’ residence time of solids in the autoclave. The limit of the recycle will be when the solids density in the autoclave becomes unmanageable. Figures 11 and 12 show the impact of solids recycle on base and precious metal extraction. While there is some ‘noise’ in the results, there is a strong indication of improvement in both base and precious metal extraction in applying the recycle system. Nickel extraction improved from 97% extraction to over 98%, whereas platinum and palladium extraction improved from less than 90% to between 90% and 95% overall extraction. Regardless of the recycle ratio, the copper extraction was excellent at over 99%. Gold extraction was highly variable, probably due to the small amount of gold in the feed sample (0.8 to 1.3 g/t), but always around 90%. The method of reporting recycle is mass of recycle solids per mass of feed solids (t/t), expressed as a per cent. KEYNOTE ADDRESS: HYDROMETALLURGICAL PROCESS DEVELOPMENT

203


2/14/09

11:58 AM

Page 204

Figure 11. Copper and nickel extraction as a function of recycle of autoclave discharge solids. Standard conditions: 225°C, 1.1 hour single pass residence time, 100 psig of oxygen overpressure, 10 g/l Cl, six-compartment autoclave

Figure 12. Au, Pt and Pd extraction as a function of recycle of autoclave discharge solids. Standard conditions: 225°C, 1.1 hour single pass residence time, 100 psig of oxygen overpressure, 10 g/l Cl, six-compartment autoclave

Inspection of the graphs indicates that at least 100% recycle ratio will improve overall metal extractions. Further insight into the kinetics of the pressure oxidation process can be obtained by taking compartment samples from the autoclave at steady-state. Figure 13 shows the precious metal solution and solids assays in the compartment samples obtained at 74% recycle ratio. Figure 14 gives the comparable information for the base metals. Note that the feed grades in each case have been diluted with the barren solids recycled from the autoclave discharge. The results show clearly that even with recycle and a relatively short autoclave residence time of 1.1 hours, the oxidation and extraction of the bulk of the base and precious metals is largely complete within the first two compartments of the autoclave (approximately 22 minutes of the 66 minute residence time). This result is consistent with all other reported information on total pressure oxidation of sulphide concentrates including the developments of Placer Dome and Phelps Dodge on copper concentrate total oxidation4,5. 204



2/14/09

11:58 AM

Page 205

Figure 12. Compartment sample assays for 74% recycle ratio for Au, Pt, Pd at steady state. (a) solution assays and (b) solid assays

Figure 14. Compartment sample assays for 74% recycle ratio for Ni, Cu, Fe at steady state. (a) solution assays and (b) solid assays


205


2/14/09

11:58 AM

Page 206

Au and PGM precipitation The autoclave discharge slurry is partly recycled (after thickening) and partly filtered to advance the metal bearing solution to the downstream recovery steps in the process. The gold and platinum group metals in solution are the first target for recovery using reductive precipitation. In the first PLATSOL® pilot plant reported in 2001, a combination of sulphur dioxide reduction of ferric ion followed by sulphide precipitation was used to recover the precious metals from the pregnant solution. In the recent pilot campaign an improved method of precious metal recovery was tested. Copper sulphides produced by precipitation of copper from the bleed stream advancing to nickel and cobalt recovery were utilized for precious metal recovery. The copper sulphides ‘cement’ the precious metals onto the solid surface, resulting in an enriched product to advance to further processing. The improved process was proven to be robust and flexible. It is still advantageous to reduce the ferric species in the advancing solution using sulphur dioxide gas to minimize the CuS demand for Au and PGM precipitation. Ferric reduction Fe2(SO4)3 + SO2 + 2H2O = 2FeSO4 + 2H2SO4 Gold precipitation 2HAuCl4 + 3CuS + 3H2SO4 = 2Au + 3CuSO4 + 8HCl + 3S Platinum precipitation H2PtCl6 + 2CuS + 2H2SO4 = Pt + 2CuSO4 + 6HCl + 2S Palladium precipitation H2PdCl4 + CuS + H2SO4 = Pd + CuSO4 + 4HCl + S The recovery of gold, platinum and palladium from the autoclave solution (after filtering and washing the solids) was accomplished by precipitation with CuS in the pilot plant. The precipitate solids are collected in a thickener/clarifier arrangement and then filtered. Recoveries of gold, platinum and palladium into this precipitate concentrate were excellent, in excess of 99.5% in each case. Base metal losses from solution into the PGM and gold residue were negligible. Approximately 4 kg of precipitate were collected during the 2005 pilot plant analysing 56 g/t Au, 211 g/t Pt and 907 g/t Pd. Most of the precipitate mass was copper (35.7%) and sulphur (49%). Batch releaching of the precipitate to remove copper and sulphur was tested and resulted in an upgraded material analysing approximately 1.6% or 16 000 g/t total contained gold, platinum and palladium. The solution after gold, platinum and palladium recovery advances to base metal recovery. The solution is still acidic and must be neutralized prior to copper solvent extraction. Solution neutralization The extraction of copper by solvent extraction is inhibited by acid in solution. It is therefore important to neutralize the excess acid from the autoclave process prior to advancing to copper recovery. The neutralization process was piloted in a three-stage neutralization circuit with limestone slurry for neutralization. The gypsum product is thickened. A 300% recycle of gypsum as ‘seed’ for precipitation was used to grow coarse, clean, crystals of gypsum. The chemistry of neutralization is shown below. Gypsum precipitation H2SO4 + CaCO3 + 2H2O = CaSO4.2H2O 206



2/14/09

11:58 AM

Page 207

The utilization of carbonate exceeded 99% in the pilot tests due to the use of the 300% recycle. The gypsum product assay is shown below. The chemical quality, physical nature and colour of this synthetic gypsum produced in the pilot plant appear to be suitable for entry into the gypsum wallboard market in the US. If an ‘off-take’ for synthetic gypsum can be finalized, the gypsum residue will not fill space in the lined tailings facility planned for ‘hydromet’ tailings. The value of the gypsum will partially offset the cost of oxygen for oxidation of sulphur in the concentrate to sulphate and the cost of purchasing limestone for neutralization. (Table VII.) Copper SX/EW The recovery of copper by SX/EW is conventional technology. The presence of chloride in the feed solution to copper solvent extraction necessitates the application of a wash or scrub step to displace any entrained chloride from the loaded organic solution. Copper extraction CuSO4 + 2HR(org) = CuR2(org) + H2SO4 Copper stripping CuR2(org) + H2SO4 = CuSO4 + 2HR(org) Copper electrowinning CuSO4 + H2O = Cu + O2(gas) + H2SO4 In the pilot plant, copper was extracted in three countercurrent stages, scrubbed in one and stripped in two stages with spent electrolyte from the electrowinning cell. Two extractants were evaluated, 35% Acorga M5640 from Cytec in Orfom SX 80CT and 35% LIX 973NS LV from Cognis in the same diluent. Both extractants performed well in the pilot-plant operation. Extractions averaged about 95% from a starting solution concentration of 17.4 g/l of Cu (average). A total of 68 kg of copper were extracted and electrolyzed during the 2005 pilot plant run at a current density of +270 A/m2. The copper metal cathode was analysed and found to meet LME grade A copper purity specifications. This product can be sold directly to copper consumers. Table VIII summarizes the assay for copper for two samples taken from the second strip cycle in the pilot plant. Raffinate neutralization The purpose of raffinate neutralization is to trim the acid level in the copper raffinate prior to splitting the raffinate flow between the nickel and cobalt recovery circuit and returning the raffinate to the autoclave circuit as a cooling solution. Acid must be neutralized prior to the Table VII Gypsum solids assay from neutralization of autoclave acid Ni % 0.005

Co % <0.02

Cu % 0.005

Fe % <0.05

Zn % 0.003

SiO2 % 0.19

Al2O3 % <0.01

Fe2O3 % 0.06

MgO % 0.2

CaO % 31.7

Na2O % <0.05

K 2O % <0.01

TiO2 % <0.01

P2O5 % <0.01

MnO % <0.01

Cr2O3 % <0.01

V2O5 % <0.01

LOI % 21.3

SO4 % 51

CO3 % 0.07

Cl g/t 24

Dry product P80 ~105 microns, wet PSD determinations using Malvern Mastersizer 2000.


207


2/14/09

11:58 AM

Page 208

Table VIII Copper assays for samples taken from the second pilot plant strip cycle* Analysis (g/t) Sample Cathode 1 Cathode 2 LME Grade A

Fe 1.02 1.31 10

S 10.50 6.00 15

Pb 0.52 3.63 5

Ag <1 <1 25

nickel and cobalt recovery process steps and excess acid is undesirable in the feed to the autoclave as excess acid will increase the formation of basic ferric sulphate in the autoclave solids. The chemistry of raffinate neutralization is the same as the primary neutralization discussed above. Fe and Al removal The removal of iron and aluminum prior to recovery of nickel and cobalt is necessary to prevent product contamination. The process of iron oxy-hydrolysis is well known. Iron is oxidized from the ferrous to the ferric state with oxygen or air with limestone addition for neutralization and pH control. The temperature for iron removal was set to 60°C and the pH to 3.0. The iron was removed to less than 5 mg/l by this process with negligible losses of Ni/Co to the iron precipitate residue. Similar results were obtained in the pilot plant using limestone to pH 4.6–4.7 and 65°C. Terminal aluminum levels were 38 mg/l on average. The iron precipitate is washed and disposed to tailings. The aluminum precipitate is thickened and recycled to the iron removal step. Copper removal Residual copper is removed before nickel, cobalt and zinc recovery using NaSH precipitation. The copper sulphide product (synthetic covellite) is then recycled internally to precipitate gold and PGM from the autoclave discharge solution. The chemistry is straightforward CuSO4 + NaSH = CuS + 0.5Na2SO4 + 0.5H2SO4 In the 2005 pilot plant, the copper level in solution advancing to copper removal was reduced to less than 50 mg/l. Nickel, cobalt and zinc recovery There were two options for nickel, cobalt and zinc recovery that were continuously piloted at SGS—Lakefield in 2005. The first option utilized a classic ‘mixed hydroxide’ precipitation

*The

first strip cathode had higher levels of contaminants. This was attributed to the ‘start-up’ conditions used and the fact that the lead anodes in electrowinning were not conditioned prior to commencement of the pilot plant. The purity of the second strip cathodes is believed to be representative of the full-scale chemistry of the copper cathodes to be produced at NorthMet.

208



2/14/09

11:58 AM

Page 209

route for recovery of the pay metals in a single product. The second option was to separate cobalt and zinc away from nickel using solvent extraction, followed by selective stripping of cobalt and zinc. The purified nickel, cobalt and zinc streams can then all be treated with magnesia to recover purified hydroxides (or in the case of zinc, be sold directly into the zinc chemicals industry). Mixed hydroxide results The mixed hydroxide process uses two-stage precipitation for nickel and cobalt recovery. In the first stage, approximately 85% of the nickel and cobalt are precipitated with magnesia (MgO). This precipitate is then thickened, filtered and washed and sent to a nickel off-take party. The balance of the nickel and cobalt are then precipitated with lime to form mixed gypsum metal hydroxide precipitate. This precipitate is recycled to solution neutralization to redissolve the precipitated nickel and cobalt. Zinc in solution will report quantitatively to the final mixed hydroxide precipitate. Nickel precipitation with magnesia NiSO4 + MgO + H2O = Ni(OH)2 + MgSO4 Cobalt precipitation with magnesia CoSO4 + MgO + H2O = Co(OH)2 + MgSO4 Zinc precipitation with magnesia ZnSO4 + MgO + H2O = Zn(OH)2 + MgSO4 Residual nickel precipitation with lime NiSO4 + CaO + 3H2O = Ni(OH)2 + CaSO4.2H2O Residual cobalt precipitation with lime CoSO4 + CaO + 3H2O = Co(OH)2 + CaSO4.2H2O Residual nickel precipitation with lime ZnSO4 + CaO + 3H2O = Zn(OH)2 + CaSO4.2H2O The mixed hydroxide material is high quality and may be placed with an off-take/refinery partner for final separation into pure nickel and cobalt and by-products. (Table IX.) Magnesium removal The NorthMet ore and concentrate contains some magnesium silicate mineralization. During high temperature autoclave leaching of the concentrate, some magnesium is leached. In addition, magnesia is used around the circuit for nickel and cobalt precipitation, which also results in an increase in the magnesium content of recirculating solutions. In order to control the build-up of magnesium in solution, a magnesium removal step (with lime) was introduced into the circuit. Magnesium sulphate reacts to form magnesium hydroxide and gypsum with lime. Table IX Analysis of the mixed hydroxide product from the pilot plant Sample

Moist. %

Ni %

Co %

Cu %

Fe %

Zn %

Al %

Mg %

Ca %

Si %

Mn %

1

51.2

36.3

1.92

0.37

0.59

4.84

0.07

1.04

0.02

0.05

0.03

2

41.2

31.5 31.3

1.67 1.67

0.31 0.32

0.51 0.54

4.31 4.27

0.04 0.04

0.62 0.62

0.04 0.04

0.03 0.03

0.02 0.02

3

56.3

40.6

2.17

0.41

0.68

0.56

0.05

0.76

<0.08

0.04

0.03


209


2/14/09

11:58 AM

Page 210

Magnesium precipitation with lime MgSO4 + CaO + 3H2O = Mg(OH)2 + CaSO4.2H2O The removal of magnesium is a control for the build-up of magnesium across the circuit. In the pilot plant approximately 50% of the magnesium was precipitated per ‘pass’ with the balance of the magnesium allowed to recirculate in the overall circuit. The NorthMet project of PolyMet Mining is well advanced and ready to proceed to construction upon receipt of key permits. The NorthMet project will begin with an initial phase of separate copper and nickel concentrate production for sale followed by commissioning of the hydrometallurgical circuit for bulk concentrate leaching. The PLATSOL™ process will be used for the treatment of the concentrate. The PLATSOL™ process will dissolve all the metals of value to high levels of extraction in an autoclave, followed by a series of precipitation and SX/EW processes for recovery of final metal products. The final products from the commercial plant will include: • Copper cathode of LME Grade A quality • Au and PGM precipitate for toll processing • Mixed hydroxide product containing Ni-Co-Zn • Synthetic gypsum. The current design basis is for 32,000 short tons of ore per day to be processed through the refurbished Erie plant and a new hydrometallurgical refinery for metal extraction. Figure 14 shows how the hydrometallurgical plant will be integrated with the existing Erie facility. The existing Erie plant has capacity to operate at approximately three times the planned production rate. Expanded production could be supported by existing resources at NorthMet, exploration success or treatment of other known deposits in the area. Expansion beyond the current permit application would involve additional environmental review and permitting.

Figure 14. Plant layout for NorthMet plant with hydrometallurgical concentrate treatment

210



2/14/09

11:58 AM

Page 211

Conclusions Hydrometallurgical processing of complex ores and concentrates offers the possibility of unlocking new and valuable mineral deposits for production of metals. Two such developments have been reviewed as examples of what is possible by developing or integrating new science and technology into a process flowsheet. The Boleo process offers the promise of being able to unlock the value in a complex, clayey ore containing significant amounts of copper, cobalt, zinc and manganese. The use of high rate thickener technology permits the separation and washing of a clayey leach residue. The integration of the CSIRO DSX technology for cobalt and zinc separation away from manganese permits recovery of these important by-product metals, vastly improving the economics of the metallurgical treatment of the Boleo ore. The PLATSOL™ process has opened the way to treat the NorthMet ore of PolyMet mining. Small additions of chloride to total pressure oxidation leaching conditions allows for direct extraction of small concentrations (but economically significant) of platinum, palladium and gold. This novel leaching process, when combined with known methods of copper, nickel/cobalt and precious metal recovery have permitted the development of a hydrometallurgical process for treatment of the NorthMet ore. Acknowledgements The author’s education in metallurgical process development began many years ago with W. Charles Cooper (PhD supervisor) and E. Peters (PDF supervisor). In the case of the Boleo and PLATSOL™ developments, extensive collaborative work has occurred with colleagues at SGS Lakefield Research, Bateman Engineering, CSIRO and others. The technical work presented in this paper is a reflection of the combined efforts of many people in various roles moving from early testing and concepts through to final feasibility for each of the process developments. References 1. DREISINGER, D., MURRAY, W., BAXTER, K., HOLMES, M., JACOBS, H., and MOLNAR R. The metallurgical development of the El Boleo copper-cobalt-zinc project. Proceedings of ALTA Copper 2005, Perth, ALTA Metallurgical Services, Melbourne, Australia, 2005. 2. CHENG, C.Y. and URBANI, M.D. Solvent extraction process for separation cobalt and/or manganese from impurities in leach solutions. Patent Publication No. WO 2005/073415 A1. 3. FERRON, C.J., FLEMING, C.A., O’KANE, P.T., and DREISINGER, D. Pilot plant demonstration of the PLATSOL™ process for the treatment of the NorthMet coppernickel-PGM deposit, Mining Engineering. Littleton, CO, United States, 2002, vol. 54, no. 12, pp. 33–39. 4. MARSDEN, J.O., BREWER, R.E., and HAZEN, N. Copper Concentrate Leaching Developments by Phelps Dodge Corporation, Hydrometallurgy 2003, C.A. Young, A.M. Alfantazi, C.G. Anderson, D.B. Dreisinger, B. Harris and A. James (eds.), TMS, Warrendale, USA, 2003, pp. 1429–1446. 5. BREWER, R.E. Copper Concentrate Pressure Leaching – Plant Scale-Up from Continuous Laboratory Testing, Minerals and Metallurgical Processing, November, 2004, vol. 21, no. 4, pp. 202–208. KEYNOTE ADDRESS: HYDROMETALLURGICAL PROCESS DEVELOPMENT

211


2/14/09

11:58 AM

Page 212

David Dreisinger Professor, University of British Columbia Canada and Chairholder, Industrial Research Chair in Hydrometallurgy David Dreisinger holds the position of Professor and Chairholder of the Industrial Research Chair in Hydrometallurgy at the University of British Columbia (UBC). Dr Dreisinger received BSc and PhD degrees in Metallurgical Engineering from Queen’s University in Kingston before beginning his career at UBC in 1984. At UBC, Dr Dreisinger supervises a wide ranging programme of research and development in atmospheric and pressure leaching of ores and concentrates, solution purification and the use of electrochemical methods for metal recovery. Dr Dreisinger (with co-workers) has been actively involved in commercializing the Mt. Gordon and Sepon copper processes in Australia and Laos as well as a number of novel ion exchange technologies. Through the Industrial Research Chair in Hydrometallurgy, Dr Dreisinger is engaged in teaching technical short courses to the global metallurgical industry. Within the METSOC of CIM, David has participated in the organizing or coorganizing of a number of technical conferences. Dr Dreisinger has received awards including the Sherritt Hydrometallurgy Award (1993), the Extraction and Processing Division Science Award from the TMS (2005) and the Alcan Award from the METSOC of CIM (2005). Recently, David has been named a Fellow of the Canadian Academy of Engineering and has received a Meritorious Achievement Award from the Professional Engineers and Geoscientists of BC. Dr Dreisinger serves as a director or officer of a number of TSX listed mining companies.

212


213-240_Beukes:text

2/17/09

10:52 AM

Page 213

BEUKES, N.T. and BADENHORST, J. Copper electrowinning: theoretical and practical design. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Copper electrowinning: theoretical and practical design N.T. BEUKES* and J. BADENHORST* *TWP Matomo Process Plant, South Africa

An engineering house’s perspective of required inputs in designing a copper electrowinning tank house and ancillary equipment calls for both understanding of the key fundamental controlling mechanisms and the practical requirements to optimize cost, schedule and product quality. For direct or post solvent extraction copper electrowinning design, key theoretical considerations include current density and efficiency, electrolyte ion concentrations, cell voltages and electrode overpotentials, physical cell dimensions, cell flow rates and electrode face velocities, and electrolyte temperature. Practical considerations for optimal project goals are location of plant, layout of tank house and ancillary equipment, elevations, type of cell furniture, required cathode quality, number and type of cells, material of construction of cells, structure and interconnecting equipment, production cycles, anode and cathode material of construction and dimensions, cathode stripping philosophy, plating aids, acid mist management, piping layouts, standard electrical equipment sizes, electrolyte filtration, impurity concentrations, bus bar and rectifier/transformer design, electrical isolation protection, crane management, sampling and quality control management, staffing skills and client expectations,. All of the above are required to produce an engineered product that can be designed easily, constructed quickly and operated with flexibility.

Introduction The electrowinning of copper ions derived from leaching, or solvent extraction is a significant contributor to the global copper commodity supply. The process of electrolysis for copper was first developed in the late 19th century and despite numerous advancements in technology the principles and basic equipment remain the same. The first part of this paper deals with the theoretical requirements and fundamental equations and principles that govern copper electrowinning. The second part discusses the practical requirements for designing a copper electrowinning plant. COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

213

213-240_Beukes:text

2/17/09

10:52 AM

Page 214

The hardware used is simple in nature; for electrowinning an acid resistant bath with anodes and cathodes submersed in solution with current passing through the electrodes is the fundamental process unit. The fundamental concepts lie in reaction kinetics, mass transfer phenomenon, thermodynamics and other electrochemical specific models, the application of which leads us to a deeper and more appreciative knowledge of the ‘simple’ electrowinning reactor. The first part of the paper goes through the fundamentals and culminates in an example reactor being developed. Note that not all design procedures are named as this would compromise TMPs intellectual property. However the reader will be able to get a very good understanding of what is required to design and build a copper electrowinning plant. Part 1: Copper electrolysis theoretical considerations Faraday’s law For the winning of copper by the addition of electrons [1] Cations go towards the cathode, and anions go to the anode. The working electrode is where reduction takes place and the counter electrode is where oxidation occurs. The working electrode is the cathode and the counter electrode the anode. For the general oxidation/reduction reaction: [2] Faraday’s Law gives the total amount of charge spent to reduce M mols of Ox (Q) is: [3] The charge spent per unit time is defined as the current (I): [4] Normalizing with unit area gives Faraday’s Law expressed in Current Density (i) : [5] Faraday’s law then is: the current flowing in an external circuit is proportional to the rate of the reaction at the electrode. Nernst equation The standard electrode potential is the potential difference between energy states of product and reactant and is a manipulation of the Gibbs free energy reaction (G) Reaction thermodynamics gives the following relationship for Gibbs Free energy: [6] For a single electrode [7] Since the electrode potential regulates the energy of electron exchange, it also controls the current and thus the rate of exchange. Current and Potential (E) are dependant variables of one another. Where the work done (W) is related to the Potential difference by: 214


213-240_Beukes:text

2/17/09

10:52 AM

Page 215

[8] Substituting for (Q) and work (W) for (G), where (W) and (G) are in joules then Gibbs Free Energy can be re-written as: [9] Substituting (7) and (8) with (9) gives the Nernst Equation for an electrode (or half cell): [10] Mechanism of electron transfer For elementary reactions at an electrode the following two mechanisms are primarily responsible for electron transfer. Mass transfer controlled 1) Diffusion of copper cations from the bulk phase to where the reaction occurs at the surface. [11] Reaction kinetics controlled 2) Heterogeneous transfer of electrons from the solid electrode to the copper cation at the surface of the electrode. [12] Further phenomena, coupled chemical reactions, adsorption and formation of phases are reported to also have a role in the electron transfer mechanism. The formation of phases is relevant to the plating of copper on the cathode and involves nucleation and crystal growth steps. Copper atoms diffuse through the solid phase to a location in an appropriate site of the crystal lattice. Adsorption and nucleation steps are considered to be included in the Heterogeneous Electron Transfer reaction rate mechanism. The overall rate is controlled by the slowest step which can be either mass transfer or reaction kinetics. For the purposes of copper electrowinning reactor design it is necessary to determine the rate limiting step to optimize conditions so that capital costs and operating ability is optimized. Heterogeneous electron transfer By analogy with chemical kinetics for a simple first order reaction: [13] [14] Using (5) the current for the forward reaction is given by: [15] And for the reverse reaction: COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

215

213-240_Beukes:text

2/17/09

10:52 AM

Page 216

[16] The total current density for the electrode is: [17] Using Arrhenius and the Activated Complex Theory it can be shown that rate of reaction kf and krev takes the forms: [18a] and [18b] E is the applied potential to the electrode and E°′ the formal electrode potential that differs from the standard electrode potential by the activity coefficients. Recalling the Nernst Equation: [10] The activity is equal to activity coefficient multiplied by the concentration in the bulk phase. Therefore: [19] It then follows that: [20] Equation [17] can be written as: [21] Substituting for kf and krev from equation [18] gives the Bulter-Volmer Equation (B-V): [22]

This current-potential relationship governs all fast and single step heterogeneous electron transfer reactions. At equilibrium the exchange current density is: [23] The overvoltage (η) can be defined as: [24] 216


213-240_Beukes:text

2/17/09

10:52 AM

Page 217

o

o

o

where Eeq is the equilibrium voltage, Eeq=E ′when Cox(bulk) = Cred(bulk) and E ′=E when activity coefficients are equal to one, see Equation [20]. An expression for the equilibrium potential is derived and shows Eeq to be close to the standard electrode potential and to vary according to changes in temperature and bulk concentrations. [25] The Butler-Volmer equation can then be written as follows: [26] This relationship shows that exponential changes to the current can result from changes to the potential. Furthermore current is constrained by the surface to bulk concentration ratios of oxidant and reductant species. The reaction rates do not grow indefinitely as potential is increased and are thus limited by the transport of species to the electrode. A system that is moved from equilibrium for Ox species to be reduced and Red species to be oxidized is described by the B-V equation. This is achieved by setting the potential different to the equilibrium potential, increasing the voltage thus increases the equilibrium difference which increases the current hence speeding up the Faradaic process. The maximum current that can be applied to maintain a reaction is known as the Diffusion Limited Current. No matter what the standard rate constant is if the applied potential is sufficiently large the maximum current will be reached. Assuming and adequate supply of reactants to the reaction surface (the electrode) the rate of reaction is described by the ButlerVolmer Equation. If the applied potential is adequate to maximize the Heterogeneous Electron Transfer reactions the rate of reaction is then limited by the supply of reactants to the electrode surface and is said to be mass transfer limiting (or controlled). Assuming that the surface and bulk concentrations are equal (condition of non mass transfer limited), for only large negative or positive overpotentials (only forward or reverse reaction dominant) the B-V equation can be manipulated by taking a Log of both sides of the equation then resolving for overvoltage gives: [27] And has the general form: [28] Recognized as the well know Tafel Equation and is derived from the B-V equation for specific condition of non mass transfer limiting, equal surface to bulk concentrations and dominant forward or reverse reactions. The procedure provides a means of linearizing the relationship between overpotential and current or rate of reaction. Mass transport The movement of species from the bulk solution to the electrode surface occurs via three possible mechanisms: • Convection (or conveyance), forced or natural described by hydrodynamics or density/temperature differences • Diffusion described by a gradient in concentrations • Migration described by a gradient in electrical potential. COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

217

213-240_Beukes:text

2/17/09

10:52 AM

Page 218

Experimental conditions of the electrode reactions are generally chosen to minimize the effects of migration. This is done by providing a large quantity of inert electrolyte that does not interfere with the electrode reaction, leaving only diffusion and convection mechanisms for consideration. Fick’s First Law for one-dimensional diffusion [29] Expanded to include convection (conveyance) and migration is the Nernst-Planck Equation8: [30] Minimizing the potential gradient using an inert electrolyte reduces the equation to: [31] This equation describes the one dimensional flux of species across the bulk solution to the electrode interface due to the mechanism of conveyance and diffusion. For a three dimensional volume it can be extended to: [32] Assuming that at steady state there is no change in concentration with time and that the conveyance inside the diffusion layer is significantly smaller than the diffusion component, Equation [32] becomes: [33] and by definition of rate of consumption [34] [35] The rate of consumption (or generation) is equal to the rate change of concentration difference across the diffusion layer. Using a circulation tank to minimize the change in concentration over two of the three spatial axes we find that. [36] and [37] The use of the circulation tank to minimize the change in concentration over the two axes parallel to the electrode surface significantly simplifies the mathematical mass transfer relationship. Also sufficient concentration of ions across the face of the electrode is provided to ensure that the mass transfer and supply of species to the surface of the electrode remains sufficient. If the concentration gradients across the two parallel axes were not minimized a 218


213-240_Beukes:text

2/17/09

10:52 AM

Page 219

varying current over the electrode surface would be required to maintain the rate of reaction. The mathematical and practical difficulties of this are self evident. The resulting mass transfer relationships are the following, and are relevant to one dimension normal to the electrode surface. [38] And: [39] Faraday’s law: [5] Where: [40] Gives: [41] Substituting Equation [39] into equation [41] gives: [42] Defining the mass transfer coefficient (kd) as: [43] then [44] Equation [44] is different to Equation [42] in that using the mass transfer coefficient forces an adequate hydrodynamic treatment of the flow reactor. Equation [42] is widely regarded in the technical literature as the only mass-transfer equation for electrochemical systems. While the difference may be trivial to the diffusion layer mechanism advocate not so from the perspective of good chemical engineering practice, the universal application of heat and mass transfer coefficients provides a more rigorous means of solving a hydrodynamic problem. Mass transfer coefficients and diffusion limited current Two equations that represent the same thing, namely, the reaction rate have been derived. The B-V equation which describes Heterogeneous Electron Transfer and Ficks Law substituted into Faradys Law that describes Mass Transfer. Increasing the overpotential up to a point increases the reaction rate. When all species that reach the electrode are oxidized or reduced the rate of mass transfer of the species to the electrode surface is the rate limiting step for the production of copper. The diffusion limited current is the current above which an increase in potential will not increase the rate of reaction. For that reason the Diffusion Limited Current COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

219

213-240_Beukes:text

2/17/09

10:52 AM

Page 220

(DLC) should be determined to optimize the reactor design. In the field, operating reactors are normally run at levels well below the DLC to achieve a good adherent product. This should be considered when determining the actual applied current to the cell. At the Diffusion Limited Current the surface concentration of species is zero meaning that all surface species are consumed as quickly as they are supplied to the electrode. The DLC equation then becomes: [45] Two important observations can be made at this point: 1) Increasing the concentration of the bulk reactant increases the DLC. This is a function of the extraction process that was used to remove the copper from the host body and other upgrade processes used before the electrowinning of the copper 2) Increasing the mass transfer coefficient increases the DLC. This is a function of the hydrodynamics of the reactor cell and physical properties of the solution in which the electrolyte is present such as temp, viscosity and other competing ions. The most efficient supply of fresh solution is provided by enhancing the bulk motion. This is easily achieved by stirring or by flowing the solution past the electrode or using other methods to enhance the mass transfer coefficient. Mass transfer coefficients are usually determined using empirical correlations that are based on test work and made up of dimensionless parameters such as: Reynolds Number: [46] Schmidt number: [47] Sherwood number: [48] Grashof number: [49] Prandtl number: [50] Rayleigh number: [51] A number of correlations appear in the literature and for convective mass transfer have the general form: [52]

220


213-240_Beukes:text

2/17/09

10:52 AM

Page 221

For parallel plate electrodes of finite width and fully developed laminar flow. (Only applicable if the maximum electrode length is less than 35 times the equivalent diameter. Re <2000)3: [53] For parallel plate electrodes of infinite width and fully developed laminar flow. (Only applicable if the maximum electrode length is less than 35 times the equivalent diameter. Re <2000)3, 15: [54] Natural convection nearly always occurs at an electrode even when forced convection takes place. Natural convection dominates forced convection at very low flow cell flow rates. The upward flow component of natural convection may oppose forced convection and results in very low mass transfer coefficients. The following correlations are concerned with comparing natural convection to forced convection. In stationary solutions the effects of natural convection will be the dominant mode of mass transport. By analogy with Heat transfer4: [55] [56] [57] [58] For mass transfer by natural convection1: [59] Equations [53] [54] [55] and [59] are solved for the mass transfer coefficient (kd). Using data from the world EW copper survey16 values for kd and the DLC were calculated for 22 operating plants. The plants chosen represent a wide variety of plant operating conditions. Table I compares the four different correlations’ mass transfer coefficients for copper electrowinning plants. An unveiled warning to all chemical engineers is that any mass transfer coefficient correlation must be used with caution as the accuracies could easily be 50% off, as practical as possible mass transfer coefficients should try to be determined by test work. The mass transfer coefficients compare well with each other for the different calculation methods over a wide variety of plant operating conditions. Table II compares the DLC’s calculated using the four different methods to each other and to the Faradaic current requirement. The table shows that the DLC’s calculated by the four different methods compare well with the Faradaic current density. The Faradaic Current Density includes the current efficiencies reported for the process plants. COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

221

213-240_Beukes:text

2/17/09

10:52 AM

Page 222

Table I Calculated kd (m/s) for 22 operating Cu EW plants

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

Nullabar M Pasimco M Girilambone Cu CM Cerro Colorado CM Quetrada Blanca Minera El Abra Codelco Chile Div Empresa Minera CM Zaldivar Mantos Blancos CM Carmen Andacolla Hellenic Copper Mines Nicico sarchesh Mexicana de corbe Miccl main plant Southern Peru limited Silver Bell Mining LLC BHPB Copper San Manuel Phelps Dodge Morenci S-side Phelps Dodge Morenci Central Phelps Dodge Morenci Stargo First Quantum Bwana Mkubwa

kd3 m/s

kd15,3 m/s

kd4 m/s

kd1 m/s

2.09E-06 1.73E-06 2.08E-06 2.31E-06 2.03E-06 2.20E-06 2.12E-06 2.05E-06 2.18E-06 2.36E-06 2.24E-06 2.13E-06 2.02E-06 2.10E-06 2.05E-06 2.22E-06 2.12E-06 1.95E-06 1.79E-06 1.67E-06 1.85E-06 1.94E-06




Table II Calculated DLC (A/m2) for 22 operating Cu EW plants

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22

222

Nullabar M Pasimco M Girilambone Cu CM Cerro Colorado CM Quetrada Blanca Minera El Abra Codelco Chile Div Empresa Minera CM Zaldivar Mantos Blancos CM Carmen Andacolla Hellenic Copper Mines Nicico Sarchesh Mexicana de corbe Miccl main plant Southern Peru limited Silver Bell Mining LLC BHPB Copper San Manuel Phelps Dodge Morenci S-side Phelps Dodge Morenci Central Phelps Dodge Morenci Stargo First Quantum Bwana Mkubwa

i3 A/m2

i15,3 A/m2

i4 A/m2

i1 A/m2

i Faradaic A/m2

240 193 246 266 278 263 260 246 285 254 275 260 213 207 236 253 268 258 205 197 210 205

235 188 241 263 271 259 257 242 283 253 271 256 209 204 232 250 264 251 199 190 205 199

203 197 209 203 241 211 216 211 230 189 215 214 185 173 202 201 223 233 201 208 200 186

250 236 260 251 312 264 267 261 287 227 271 265 224 202 249 247 281 299 245 255 243 226

226 239 321 256 246 233 234 236 244 262 286 272 231 246 240 266 316 231 234 251 239 196


213-240_Beukes:text

2/17/09

10:52 AM

Page 223

The DLC should always be greater than the Faradaic current for a plant to ensure that a good adherent copper cathode is produced and that the hydrodynamics of the system are sufficient to ensure production is not limited by inadequate flow conditions. It is important to note that the DLC’s reported here are based entirely on the correlations used and not on any test work done by the authors. This method is used to demonstrate the phenomenon of determining the DLC with common literature information, as well as the reasonably comparative results obtained from the different correlations. It can be seen that some Faradaic currents are greater than the DLC’s calculated this does not mean that the plant is incapable of adequate production but only means that the correlation may not be representative of the exact conditions in that plant. The correlations do not take into account gas evolving electrodes of which all copper electrowining anodes are, this omission in the correlations may underestimate the DLC significantly. This gas evolution can have a strong influence on the mass transfer of ions due to the forced convection effects inside the parallel plate arrangement. Overall electrode process The earlier sections dealt with the two mechanisms of electron transfer for an electrode. This section deals with an overall process of both heterogeneous electron transfer and mass transfer that is relevant to copper electrowinning in practice. The B-V equation that describes the Heterogeneous Electron transfer in terms of current and overpotential is: [26]

The current density is described by Fick’s and Faraday’s law for mass transfer is: [44] And [45] Substituting and resolving for concentrations gives: [60] For copper reduction on an electrode the last term of the B-V equation can be neglected to give: [61] Taking the log of both sides of the equation and separating for overpotential gives: [62]

Substituting for the concentration ratio gives: COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

223

213-240_Beukes:text

2/17/09

10:52 AM

Page 224

Table III Operating conditions for the four scenarios compared

Fictional cell design Number of cathodes Number of anodes Cell spacing A-C Cell inlet concentration Cell outlet concentration Cell volumetric flow rate Copper production rate Temperature Density Diffusion coefficient Current efficiency iL3 iL15,3 iL4 iL1 Faradaic current density

units no off no off m g/l Cu2+ g/l Cu2+ m3/h kg/h °C kg/m3 m2/s % A/m2 A/m2 A/m2 A/m2 A/m2

Cell 1

Cell 2

Cell 3

Cell 4

Lo MT 48 49 0.045 40.00 35 2.50 12.5 65 1216 1.95E-09 85 191 182 273 272 129

Hi MT 48 49 0.045 35.77 35 16.25 12.5 65 1216 1.95E-09 85 300 301 244 237 129

Lo MT Hi Cu 48 49 0.045 38.75 35 5.00 18.75 65 1216 1.95E-09 85 228 221 264 262 194

Hi MT Hi Cu 48 49 0.045 36.07 35 17.50 18.75 65 1216 1.95E-09 85 309 311 246 239 194

[63] This equation describes overpotential, applied current and DLC in a single equation. Example of fictional cell reactor A fictional cell reactor is developed to show the relationship of overpotential to current for varying conditions of concentration and flow in the cell. High level differences between the scenarios are: • Cell 1, low flow rate • Cell 2, high flow rate • Cell 3, low flow rate high copper concentration • Cell 4, high flow rate high copper concentration. Each cell has the same physical characteristics and features a cathode plating area of 1 m x 1 m. The physical properties of the electrolyte are kept constant except for the concentrations and flow rates in the cells. Table 3 also gives the DLC results for various correlations. • The Forced Convection correlation of Equation [53] and Equations [54] described by the hydrodynamic characteristics of the cell are strongly influenced by the change in flow rate. Compared to the Natural convection correlations of Equations [55] and Equations [59] that, as expected, are not influenced by the changing flow rate as these correlations do not take into account the flow system but only density difference along the plate surface • The graph of overpotential versus current is called a Voltammogram. For the development of the Voltammogram the correlation from Equations [53] is used as this DLC changes with concentration and flow rate which is phenomenologically correct. Although it may not fully account for all mass transfer driving forces such as the gas evolution at the anode this correlation demonstrates that reactor design is strongly 224


213-240_Beukes:text

2/17/09

10:52 AM

Page 225

Figure 1. Voltammograms for fictional cells

influenced by flow rate and concentrations. The DLC3 for each cell is used to plot the voltammogram on overpotential versus current density graph. The exchange current density io is taken from17 as [64] Figure 1 compares the Voltammograms for the four fictional cells: This graph shows the following conclusions to be drawn: • Increasing the flow rate at the same copper production rate results in an increase of DLC from 190 A/m2 to 300 A/m2, ‘Cu Lo MT’ to ‘Cu Hi MT’ • Increasing the copper concentration and flow rate slightly results in increasing the DLC from 190 A/m2 to 230 A/m2, ‘Cu Lo MT’ to ‘Cu Lo MT Hi Conc’ • Increasing the copper concentration and flow rate significantly has the largest effect of increasing the DLC, from 190 A/m2 to 310 A/m2, ‘Cu Lo MT’ to ‘Cu Hi MT Hi Conc’. Although the DLC equations used to determine this voltammogram are based on literature work, the functionality remains correct. In order to optimize the cell design in terms of capital cost for the same production rate, the DLC must be as high as possible so that the Faradaic current density is as high as possible (limited to producing a good product). These conditions will allow for the same transformer/rectifier arrangement to: • Minimize installed electrowinning total plating area • Minimize installed electrowinning cell size • Minimize installed electrowinning cell house structure and civil footprint. Methods to increase current density Increased DLCs can be achieved by a number of different methods described in an excellent paper on the subject10. Three main ways to increase the current density are suggested: • Optimizing the cell design - EMEW Cell13 - Continental copper and steel (CCS) cell11 - Increased flow rate pattern and distribution in standard cell • Employing various types of forced convection - Air sparging - Ultrasonic agitation • Periodic current reversal. COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

225

213-240_Beukes:text

2/17/09

10:52 AM

Page 226

Table IV Comparison of current density (A/m2) operating conditions of different cell types Reported current densities A/m2 Type

Low

High

Traditional cell3,6,10,15,16 EMEW cell12,13 CSS cell11 Air sparging10 Ultrasonic agitation10 Periodic current reversal10

110 250 430 1000 3300 430

350 600 1000 3000 10500 ---

The EMEW cell is a recent design used to produce copper powder and circular plate at high flow rates and high current densities. The CCS cell is a redundant design that used nozzles placed at the cathode face to improve flow patterns. The present paper does not discuss the various types of improvements that can be made but does note that significant increases in current density can be obtained by using the methods listed above. Table III compares operating conditions that produce a stable adherent pure copper product at vastly increased current densities. The table is based on information from reference10 as well as other sources. The reported figures for the various techniques are significantly greater than standard practice and design currently allows for. There are negative effects at these current densities such as: • Increased anode wear • Acid mist control • Corrosion of suspension bars Cell potential The total voltage across the cell can be divided into three components: • The reversible decomposition potentials (Vmin) • The activation and concentration overpotentials of the electrodes (ηa) (ηc). • The Potential drop due to Ohmic resistance of the electrolyte and the electrical contacts (Vohm). [65] The reversible decomposition potential is: [66] This is the difference between the two standard reduction potentials for the species being oxidized and reduced and is the minimum voltage required at standard conditions The overpotential due to potential and concentration has been discussed previously and are represented by the modified B-V equation as: [63] And for no concentration overpotential influences then [28] 226


213-240_Beukes:text

2/17/09

10:52 AM

Page 227

The ohmic resistance due to the solution and the contacts are: [67] By Ohm’s law for a current flowing between the solution and an electrode. The relationship between current drop and potential drop in an electrical conductor is5: [68] For current density in one direction only (parallel plates), Ohm’s law reduces to: [69] where the specific conductivity is a function of concentration and ionic mobility of the form5 [70] The solution resistance and required potential difference is then a function of the interelectrode gap, the ionic mobility, the concentration of the species in solution and the current density applied to the solution. Then for the electrowinning of copper from acid sulphate matrix: [1] [71] [72] Operated at a DLC of 300 A/m2 and an applied current density of 200 A/m2 (Cell 4 Conditions) the cathodic overpotential is: [73] For the production of oxygen at the anode, excluding the acid concentration overpotential, the Tafel equation is3 [74] This gives: [75] The solution and contact resistance can be taken as <0.6 V combined6. Then the required cell voltage to obtain a current flow to produce copper cathode from solution is [76] Copper electrowinning cell voltages are well standardized at ~2-2.3 V per cell. Current efficiencies The current efficiency is the ratio of the current producing the copper to the total current applied to the cell. It is normally quoted in percentages. Practically it is determined by the amount of copper produced divided by the amount of current applied to the system for a theoretical amount of copper. COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

227

213-240_Beukes:text

2/17/09

10:52 AM

Page 228

[77]

The efficiency is strongly dependant on additional side reactions that occur, for the electrowinning of copper the decomposition of water lowers the current efficiency. However the single greatest effect is the presence of Iron in the solution. The cyclic oxidation and reduction of dissolved iron can significantly reduce the efficiency of an operation. Iron can be removed by a number of methods that include. • Solution precipitation • Prior reduction of iron (III) by SO2 or copper metal • Increasing the bleed stream volume • Use of a diaphragm cell. Current efficiencies for direct electrowinning operations can be as low as 65% while efficiencies for post SX electrowinning can be as high as 95%. Part 2: Practical considerations The following is based on a case study of the Ruashi BMR designed by TWP Matomo Process Plant Production rates Production rates are usually specified by the client, depending on the mass balance and the environment in which the electrowinning is to be designed for. Design conditions will normally be in the region of specified rate plus 10%, this excludes availability. Availability Mechanical availability depends on the following factors: • Site conditions and location. The Ruashi BMR is located in Lubumbashi, DRC, and as such, consideration was given to have stand-by units for all key mechanical equipment. Operation within South Africa would not necessarily require the same design. Dual availability design was included for cellhouse circulation pumps, feed to and return from pumps servicing the cellhouse, overhead cranes, stripping machines (for which an additional allowance was made to allow manual stripping of cathodes in a specifically designated area in case of equipment breakdown). Both the loaded electrolyte pond feeding the cellhouse as well as the spent electrolyte pond returning spent electrolyte solution to the solvent extraction circuit is of split design, providing for maintenance on the pond linings without having to shut down the plant • Maintenance programmes proposed or in place at the operation depend on site management in terms of pro-active maintenance, service schedules, etc. This cannot always be quantified, especially for new operations; however, in general, larger established companies normally have a better maintenance system in place than small companies • Education and experience level of both maintenance and operational personnel. Consider geographic location of the area, whether there are existing operations from where personnel could be sourced, language barriers, culture of work, political stability. Current density Depending on the type of operation, current densities can vary between 200 and up to about 228


213-240_Beukes:text

2/17/09

10:52 AM

Page 229

375 A/m2. Operations where Solvent Extraction precedes Electrowinning normally produce cathodes with a good surface quality at higher current densities than direct electrowinning operations. Furthermore, the clarity of the feed solution needs to be taken into account. The Ruashi design includes multimedia coalescing filters for the removal of both solids as well as organics from the Cu Feed solution before the solution reports to the Cellhouse. Process Engineers are in general quite comfortable operating in the region of 250 to 300 A/m2 when SX and multimedia filters are employed. It is also advisable to consult the vendors of the rectiformers. There are standard designs for the transformer-rectifier pairs, ranging in size from small up to 30 kA. For the Ruashi design, two sets of 30 kA were purchased to satisfy the design requirement of 60 kA total DC supply to the electrowinning cells. Depending on the rectiformer supply, the cell design can be modified to suit. Cathode centre to centre spacing Design current density is also dependant on electrode spacing. The minimum spacing recommended for Cu Electrowinning operations for cathode centre to cathode centre is 95 mm. Below this the risk of shorts due to electrode alignment and nodular growths is considered to be unacceptably high. Cathode quality This is normally specified by the client. Within the Cellhouse, there are very few variables that influence cathode quality; it is normally a function of the preceding purification steps. Sampling specification and procedure Sampling on site is normally done on site for in-house quality control purposes only. The client usually sells his product through a third party, who will be responsible for sample analysis. Some mining houses have accredited laboratories, in this case, the laboratory analyze samples received from the plant. Two commonly used methods to obtain a sample are to either drill holes in randomly selected samples on randomly selected locations of the cathode. The drillings are then sent for analysis in the accredited laboratory. The second method follows the same procedure, however instead of drillings; a manually operated punching machine is used. It has to be taken into consideration that normal drill bits are not used for the drilling of samples, since iron contamination of the samples frequently occurs. Regarding the selection of cathodes in a group to be analyzed and the location on each cathode, a standard procedure was published by the ISO organization. The standard title is ‘ISO 7156 – Refined Nickel – Sampling’. No procedure for Copper sampling could be found for Copper Sampling, however the same principles for Cu sampling as for Ni sampling is assumed to be applicable. Number of cells This is dependant on a number of factors, including cell dimensions, real estate available for the Cellhouse building, applied current density, production rate, DLC, current efficiency. There is normally an equal amount of cells, making the busbar arrangement practical. Cells in a Leach-Solvent Extraction-Electrowinning design should be split into scavenger and production cells to limit organic contamination of cathodes in case of organic carry-overs occurring at the Solvent Extraction Purification Plant. Scavenger Cells are then designed to maintain the same cathode face velocity as for the Production/Commercial Cells of approximately 0.08m3/hr/m2 cathode surface, the difference being that there is a once-off pass COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

229

213-240_Beukes:text

2/17/09

10:52 AM

Page 230

of electrolyte through these cells. The scavenger cells then produce the potentially poor surface quality cathode by burning off any organic substances, before the bulk production of cathodes in the commercial cells. For Ruashi, the design accommodated 24 scavenger cells, which constituted exactly one bank of cells. This design facilitated easy piping arrangement of feed and overflow to the cells. Cellhouse layout Practical considerations to be considered include optimizing availability and maintaining flexibility inside the Cellhouse. For Ruashi, the Cellhouse was split into two sides, each containing two banks of 24 cells. This configuration allowed the two stripping machines to be placed in the middle of the Cellhouse, minimizing travel time for the cranes harvesting the cathodes and returning blank cathodes to the cells. Furthermore, it allows one half of the cells to be shorted out with relatively little effort, which came in handy during the commissioning of the Cellhouse. The connecting busbar from the end of the Cellhouse was connected to the centre of the Cellhouse, allowing a ramp-up stage of electrowinning on the one side, while construction was still ongoing on the other side of the Cellhouse. The distance between the end of the cell and the supporting structure of the Cellhouse has to be taken into consideration in order to minimize the possibility of electrical accidents. A distance of at least 2 to 3 metres is recommended. The same applies to the distance between the cells comprising one bank and the opposite bank of cells (the ‘middle’ of the Cellhouse walkway). Feed valves to the individual cells should be practically located in order to allow easy operation of these valves, without interfering too much with the harvesting procedure. For Ruashi, the feed valves were located on top of the cells in the middle of the Cellhouse walkway. Walkways need to be constructed from non-conductive material. Two options are available, the one being wooden walkways, of which a high grade knot-free pine is considered the most feasible option. Wood is then also treated with CCA (Copper/Chrome/Arsenic). The second option is the use of FRP grating. This is much more expensive, but operating cost is much lower. Certainly, for the areas close to the liquid, FRP will have to take preference, since the timber will not last in this environment. For Ruashi, the majority of the walkways consist of CCA treated timber, with some FRP inserts at the two ends of the cell to allow for visual inspection of overflows.

Figure 2. Cell house layouts showing access platform, feed pipes

230


213-240_Beukes:text

2/17/09

10:52 AM

Page 231

Figure 3. Showing Intercell busbar and lead anode

Figure 4. Showing internal feed pipe arrangement

Figure 5. Showing Cathode bail removing four cathodes

COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

231

213-240_Beukes:text

2/17/09

10:52 AM

Page 232

For practicality, the Cell Feed inlet pipes all run along the centre of the Cellhouse, allowing for only two feed manifolds to feed the entire Cellhouse, that is, both the East and the West Banks. The feed manifolds enter the Cellhouse at the electrical neutral point in case of the production cells, and enters in the middle of the Cellhouse for the scavenger cells. This was to accommodate piping where limited space was available between the two banks of cells. Ideally, the scavenger cells’ feed manifold would also have entered the Cellhouse at the electrical neutral point to minimize stray currents. Cell drain points were also located in the centre of the building, thus allowing the civil slope of the floor to act as a run-off trench for spillages. Spillages are thus mainly confined to only the centre of the Cellhouse, minimizing acidic attack of the civil structures to only the centre of the building. Piping layout For practicality, the cell feed inlet pipes all run along the centre of the Cellhouse, allowing for only two feed manifolds to feed the entire Cellhouse, that is, both the East and the West Banks. The feed manifolds enter the Cellhouse at the electrical neutral point in case of the production cells, and enters in the middle of the Cellhouse for the scavenger cells. This was to accommodate piping where limited space was available between the two banks of cells. Ideally, the scavenger cells’ feed manifold would also have entered the Cellhouse at the electrical neutral point to minimize stray currents. Cell drain points were also located in the centre of the building, thus allowing the civil slope of the floor to act as a run-off trench for spillages. Spillages are thus mainly confined to only the centre of the Cellhouse, minimizing acidic attack of the civil structures to only the centre of the building. Cell elevation In order to facilitate the gravity overflow of cells to the Circulation Tank, cells need to be elevated of the ground. When elevation is already a necessity, it is then useful to elevate the cells high enough to accommodate access for personnel to inspect the civil structures from time to time, especially since the Cellhouse is such a corrosive environment. Additionally, floor space underneath the cells can then also be used as storage space for cell furniture. The Ruashi Cells were lifted approximately 1,6 m of the ground. Even so, the terrace level of the Tankfarm had to be lowered to allow overflow into the Circulation Tank. Additionally, the Circulation Tank was designed low with a high floor space (3,5 m high by 12,5 m diameter) to further accommodate gravity feed into the tank. Cell potential and current efficiency Cell potential is a function of the respective half reaction potentials, losses through electrolyte, busbars and competing reactions. It is normally in the range of 1,9V to 2,3V, lower for the purer Leach-SX-EW operations than for the direct electrowinning operations. Current efficiency for direct electrowinning operations can be as low as 65%, and for LeachSX-EW operations up to about 93%. It depends on the concentration of Fe in the feed solution, as well as proper housekeeping to ensure good electrical contact between electrode hanger bars and the triangular busbars. Cathode face velocity A number ranging between 0,05 and 0.1m3/hr/m2 of cathode surface area available for plating. It influences the surface quality of the cathode by breaking down the diffusion layer, but too high a face velocity can cause ‘flushing’ of the cathode surface area, creating an area of no plating where the feed is introduced into the cells. For Cu Electrowinning operations, a face 232


213-240_Beukes:text

2/17/09

10:52 AM

Page 233

Figure 6. Overflow launders between cells

Figure 7. Showing Copper busbars

Figure 8. Showing cathode stripping machine

COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

233

213-240_Beukes:text

2/17/09

10:52 AM

Page 234

velocity of 0.08 is normally employed. The cathode face velocity will then determine the flow rate per cell, and thus the internal recycle required inside the Cellhouse. Feed and spent tenors The Leach and Solvent Extraction processes will determine the copper and acid tenors of the feed to the Cellhouse. For Leach-SX-EW operations, feed tenors of Cu can go as low as 25 g/l Cu, depending on the applied current density, which will have to be reduced as the Cu tenor is reduced. The sulphuric acid tenor of the feed solution is a function of the stripping process inside SX, and that of the spent solution is a function of the rate of copper production, where a stoichiometric balance exists of 1,98 g of acid produced for every gram of copper plated. The Ruashi EW operation was started up with unpurified Primary Leach Solution at a pH of ~1,8 and a Copper tenor of 24 g/l, the resultant cathode had a very good surface quality, albeit at an applied current density of only 75 A/m2. Raising the current density to 110A/m2 resulted in the formation of nodules, but this was attributed to no control over the flow rate (cathode face velocity) to the cells. Since the flow rate was increased, good surface quality cathodes are achieved at current densities of 100A/m2 and Cu tenors of ~27 g/l. Ruashi design caters for a Cu Feed tenor of 50 gpL and a Cellhouse bite of 5 g/l Cu, at these operating conditions; the design applied current density of 275A/m2 is expected to deliver good surface quality cathodes. This has not yet been achieved at Ruashi, due to the SX section still being constructed and the front end of the plant still being commissioned. Cathode harvesting The cathode harvesting cycle was designed so as to accommodate a deposit thickness of ~5mm. The thickness is dependant on the capability of the stripping machines to separate the copper from the blanks, for which the stripping machine suppliers have to be consulted. Furthermore, the longer a cathode is allowed to plate in a cell, the higher the risk of shorting becomes due to the exponential nature of nodular growths inside the cell. A harvesting cycle of more than eight days is not recommended, but this is again dependant on the applied current density (rate of plating). Busbar design In order to minimize heat losses through the busbars, the current density through the busbars have to be limited. Conservative clients prefer not to go above 1A/mm2, while design companies are normally quite comfortable designing busbars at 1,2A/mm2. Of importance is to run parallel busbars with a gap between them so as to allow heat reduction through air cooling. The Ruashi busbar designs for the main busbars were 6 busbars, each of size 450 mm x 20 mm, separated by a 20 mm gap between each busbar. This fulfilled the 1,2A/mm2 requirement for the total design current of 61kA. Intercell busbar design Two options are available for the design of intercell busbars. Conventional dogbone type of designs was previously employed and is still preferred by some clients. It is however much more expensive than triangular busbars, due to the larger amount of copper needed. Triangular busbars are a more viable option, provided that proper consideration is given to the design and layout of contact points between hanger bar (cathode and anode) and triangular busbar. It 234


213-240_Beukes:text

2/17/09

10:52 AM

Page 235

becomes even more critical when the design had to cater for cathodes that are symmetrical to cater for the 180o rotation during the stripping operation. Furthermore, due consideration needs to be given to the different elevations between cathode and anode hanger bars on the cell to allow for shorting out of the cell with a shorting frame. Although this can be avoided if a bullhorn anode hanger bar design is employed, it is unnecessary and a straight anode hanger bar design allows for the maintenance of anodes using the same bailer as is used for cathode harvesting. Cathode design to cater for hooking by the bailer can be through either cut-outs from the cathode surface above the liquid level, or through the attachment of stainless steel fish-eyes to the hanger bars. Fish eyes are economically more feasible since it comprises less work for the suppliers. It then implies that cut-outs are made from the anode surfaces to accommodate the use of one bailer size to both harvest cathodes as well as do anode maintenance. Dimension and number of electrodes per cell Cathode size is selected to accommodate industry standards, where shipping of cathodes in containers as well as furnace openings in the production industry has to be taken into consideration. Client liaison will normally give a good guideline as to cathode sizing. Anodes are designed at least 30 mm bigger than cathodes on both the width and the height (wetted dimensions). This is to ensure that copper plating does not occur around the sides of the cathode. There will always be one more anode than cathode per cell, and depending on the harvesting design, either every second or every third cathode is harvested per pull. Using the standard design of every third cathode, cathodes per cell then have to equal multiples of three. Anode composition and type New electrowinning operations at the moment almost exclusively employ cold-rolled lead anodes, since it yields better dimensional stability. Standard additions to the lead anode includes a fraction of Calcium (0,05 to 0,08%) and Tin (1,2 to 1,5%). Anode life expectancy is in the order of 7 to 9 years. Electrode furniture Anode buttons from non-conductive PVC material is readily available on the market. Three to five anode buttons are located to the anode, that being in the two bottom corners of the anode, one in the centre of the anode, and an optional two at the two top corners of the anode. The anode buttons prevent shorts between cathode and anode. Edge strips are located to the edges of the stainless steel cathodes to ensure that the two sides of the cathode are not intergrown. Various types of edge strips are available. Ruashi and various EW operations in Zambia employ the Rehau ‘cross-slot’ configuration of edge strips. It is important to design the edge strips long enough so that it ends well above the liquid level of the cell, since liquid between the edge strip and the cathode will cause metal plating and thus damage to the edge strips. Electrowinning cells Continuous improvement to the polymer concrete industry has led to the almost exclusive use of polymer concrete as material of construction for electrowinning cells. There are two main suppliers in Southern Africa, they being CSI and PCI. Product quality from both suppliers is acceptable and the Copper industry in Zambia is split almost equally between the two COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

235

213-240_Beukes:text

2/17/09

10:52 AM

Page 236

suppliers. The polymer concrete cells contain a fiberglass lining, which is part of the moulding/casting process. This further protects the concrete from acid attack. Other operations use lead linings or normal concrete with fiberglass lining moulds to house the electrowinning process. This is not feasible for new operations. Cathode stripping Various suppliers can custom design stripping machines for the electrowinning operation. Options include fully automatic machines, semi automatic machines and options for water spray rinsing booths. Design considerations should include simplicity if the machine is used in remote locations, availability of spares, stripping rate and capital expenditure. Ruashi employs two semi-automatic Styria stripping machines; this gives flexibility in case of a breakdown as well as ease of maintenance. Plating agents Various smoothing agents are commercially available, ranging from glues to guars to synthetic products. It is best to test these in research facilities such as at Mintek to determine the effect of the smoothing agent on current efficiency and surface quality of the cathode. Dosage rates range from 150 to 400g of smoothing agent per ton of Cu produced. Chloride levels inside the Cellhouse should be kept below 30ppm to ensure the stainless steel cathodes are not corroded. Addition points can be at the Circulation Tank for the Commercial Cells and inline at the Feed manifold for the Scavenger Cells. Care should be taken to design the Scavenger system so as not to have reverse flow of electrolyte back into the smoothing agent supply tank, as such a non-return valve is recommended in-line for the scavenger cells smoothing agent supply. Acid mist management Hollow polypropylene balls are commonly employed in Cellhouses. Additionally, synthetic foaming agents are available, the effect of these on the Solvent Extraction should be considered. Designing a Cellhouse with open sides helps with natural ventilation, and if CAPEX considerations are not of primary importance, extraction fans can be employed to minimize acid mist. Temperature management Feed to the Cellhouse could be heated using heat exchangers, the optimum temperature is considered to be in the range of 45 to 55°C for optimum electrolyte conductivity. Due to heat generation inside the Cellhouse, a cooling heat exchanger is then required for both circulating electrolyte and spent electrolyte. Solvent Extraction plants which use Spent electrolyte for stripping need to be maintained below 40°C to prevent the degeneration of the organic phase. Inventories For optimum plant availability, sufficient storage capacity is required. Ruashi employs split ponds with double liners and leak detection systems. In designing the storage pond systems, consideration need to be given to the type of pumps to be used. Options available include selfpriming pumps (Sulzer) or conventional pumps with priming tanks. The self-priming pumps employed at Ruashi have proved to be difficult to commission, and to date little support from the supplier was received. Additionally, self-priming pumps are considerably more expensive than the combination conventional pump and priming tank system. 236


213-240_Beukes:text

2/17/09

10:52 AM

Page 237

Cathode washing Heated dip tanks are allowed for at the Ruashi BMR, however other systems available include the use of heated water spray systems or, as previously mentioned, spray systems as part of a stripping machine. Electrolyte filtration Scheibler filters are employed in industry to aid in the filtration of circulating electrolyte, there-by reducing the need to clean cells. Capital expenditure availability has to be considered, and if a filter is not part of the initial design, real estate allowance should be made to later locate the filter when client finances allows. Conclusions The above paper has two parts, the first part describes the theoretical requirements for copper electrowinning and the second part describes the practical requirements based on the Ruashi case study. The first part is based on the fundamental equations of reaction rate kinetics and thermodynamics as well as hydrodynamics to describe the baseline equations that can be used in designing a copper electrowinning process. The important considerations for the designer are determination of the Diffusion Limited Current and the appropriate selection of the operating current. These numbers which are based on numerous variables such as flow rate to cell, cell bite, plating area, copper, acid and iron concentrations, temperature of electrolyte and current efficiency will determine the structural size and footprint of the copper electrowinning process building. They have the largest effect on overall capital cost and consequently the timeline of building the process plant. It has been discussed that the use of high current densities requires some modifications to the conventional 100 year old electrowinning cell design. Conventional cells are operated at maximums of approximately ~350-400A/m2 but most designers will not exceed these values in conventional designs, with good reason. With adequate hydrodynamics and increase of the Diffusion Limited Current the capital cost of a copper electrowinning building can be reduced. There are problems associated with this, particularly acid mist generation. Adequate ventilation systems or enclosed hoods similar to those used in the nickel industry can be used. Any designer straying from the path of tried and tested technology should confirm the new design with appropriate test work before committing the clients’ project and their career to the annals of technical success or failure. The second part of the paper discusses practical requirements around the electrowinning building and process areas based on the Ruashi case study. A significant number of issues need to be addressed during the design phase to ensure a successful project and an operable plant. Key aspects such as plant location and skill set are critical. Physical aspects such as electrolyte storage, operating temperatures, cell sizes, busbar design, circulation tank and piping layout are very important. Plating and leveling agents need to be considered. Cathode product handling and acid mist control are vital. Acknowledgement The authors would like to thank Metorex the owners of Ruashi BMR for the opportunity to present their findings. COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

237

213-240_Beukes:text

2/17/09

10:52 AM

Page 238

Nonmenclature Symbol Q I n F M i A dQ/dt dM/dt G o G R T n aP m aR aRed aOx V W E kd kr Ox Red e kf Cox ic ia krev Cred ko α o’ E o E γ io Eeq η Nox” D (x,t) 238

Definition Charge Current Valance Faraday’s constant Number of Mols Current density Area Charge per time Mols per time Gibbs free energy Standard Gibbs free energy Universal gas constant Temperature Activity of products Activity of reactants Activity of oxidized species Activity of reduced species Potential Work Potential Mass transfer coefficient Reaction rate constant Oxidized species Reduced species Electrons Rate constant of forward reaction Concentration of oxidized species Current density of cathodic reaction Current density of anodic reaction Rate constant of reverse reaction Concentration of reduced species Rate constant at equilibrium Separation factor Formal electrode potential Standard electrode potential Activity coefficient Exchange current density Equilibrium potential Overpotential (overvoltage) Mass flux of oxidized species Diffusion coefficient of species Species vector at time t and position x

SI Units Coulombs Amps, C/s gEq/gmol C/gEq gmol A/m2 m2 C/s gmol/s J/gmol J/gmol J/(gmol.K) Kelvin Dimensionless Dimensionless Dimensionless Dimensionless Volts Joule Volts m/s 1/s Dimensionless Dimensionless Dimensionless 1/s gmol/m3 A/m2 A/m2 1/s gmol/m3 1/s Dimensionless Volts Volts Dimensionless A/m2 Volts Volts gmol/(m2.s) m2/s -------


213-240_Beukes:text

2/17/09

v dΦ/dx ∇ • ra • rox • rred dx,dy,dz δ iL Re L ρ μ Sc Sh de Gr g β Ts T∞ Pr Cp k Ra Nu Le Va Vc Vmin Vohm Vir Vcontacts κ ui ε

10:52 AM

Page 239

Velocity of species Potential gradient across electrolyte Laplace operator Reaction rate for species a Reaction rate for oxidized species Reaction rate for reduced species Distance in x,y and z directions Boundary layer thickness Diffusion limited current Reynolds number Characteristic length Density Viscosity Schmidt number Sherwood number Characteristic length/hydraulic diameter Grashof number Gravitational acceleration Thermal expansion coefficient Temperature at the surface Temperature at bulk Prandtl number Specific heat capacity Thermal conductivity Rayleigh number Nusselt number Lewis number Anodic standard electrode potential Cathodic standard electrode potential Decomposition potential Ohmic resistance Solution potential Contacts potential Resistivity Ionic mobility of species i Current efficiency

m/s Volts/m ------gmol/(m3.s) gmol/(m3.s) gmol/(m3.s) m m A/m2 Dimensionless m kg/m3 kg/(m.s) Dimensionless Dimensionless m Dimensionless m/s2 1/K K K Dimensionless J/(kg.K) W/(m.K) Dimensionless Dimensionless Dimensionless Volts Volts Volts Volts Volts Volts S/cm gmol.cm2/(J.s) percentage

References 1. HAYES, P.C. Process Principles in Mineral and Materials Production, Hayes Publishing Co., 1983. 2. PETRUCCI, R.H. AND HARWOOD, W.S. General Chemistry Principles and Modern Applications, Prentice Hall International, Inc., 1993. 3. LORENZEN, L. Mineral Processing Class Notes, Electrochemical Reactor Design, University of Stellenbosch, 2000. COPPER ELECTROWINNING: THEORETICAL AND PRACTICAL DESIGN

239

213-240_Beukes:text

2/17/09

10:52 AM

Page 240

4. INCOPERA, F.P. and DEWITT, D.P. Fundamentals of Heat and Mass Transfer, John Wiley and Sons, Inc., 1996, (4th Edition), Chap 6, Chap 9. 5. PERRY, R.H. and GREEN, D.W. Perry’s Chemical Engineers’ Handbook, McGraw-Hill, 1997. 6. JACKSON, E. Hydrometallurgical Extraction and Reclamation, Ellis Horwood Limited, 1986, Chap 5. 7. WANG, J. Analytical Electrochemistry, John Wiley and Sons, Inc., 2001, (2nd Edition). 8. ZANELLO, P. Inorganic Electrochemistry: Theory, Practice and Application, The Royal Society of Chemistry, 2003. 9. BRETT, C.M.A. and BRETT, A.M.O. Electrochemistry Principles, Methods, and Applications, Oxford University Press, Inc., 1993. 10. MACKINNON, D.J. and LAKSHMANAN, V.I. Recent Advances in Copper Electrowinning, Mineral Research Program, Mineral Research Laboratories, CANMET Report 76–10, 1976. 11. ANDERSEN, A.K. and BALBERYSZSKI, T. Electrowinning of Copper at High Current Densities with the CCS Cell, The Metallurgical Society of AIME, TMS Paper Selection, Paper A68-17, 1968. 12. ESCOBAR, V., TREASURE, T., and DIXON, R.E. High Current Density EMEW® Copper Electrowinning, Electrometals Technologies Ltd. Official Website. http://www.electrometals.co.au. 13. ROUX, E., GNOINSKI, J., EWART, I., and DREISINGER, D., Cu-Removal From the Skorpion Circuit Using EMEW® Technology, The South African Institute of Mining and Metallurgy, The Fourth Southern African Conference on Base Metals, 2007. 14. Short course on Electrochemistry, Electrochemical Engineering and Electrometallurgy: Module on Thermodaynamics and Kinetics, University of the Witwatersrand, Johannesburg, 2005. 15. SANDENBERGH, R. University of Pretoria Electrochemistry Class Notes, 2007. 16. HOULACHI, G.E., EDWARDS, J.D., ROBINSON, T.G. Cu 2007, vol. V, Copper Electrowinning and Elecrorefining, Toronto, Metsoc Publication, August 2007. 17. Short course on Electrochemistry, Electrochemical Engineering and Electrometallurgy: Module on Applications of Fundamentals to Electrowinning and Electrorefining of Metals. University of the Witwatersrand, Johannesburg, 2005.

Nicholas Terence Beukes Process Engineer, TWP Matoma Process Plant, South Africa Nick has worked as a process engineer for seven years since graduating as a chemical engineer. Experience includes projects commissioning for Anglo Platinum copper-nickel matte smelting and converting. Following that he has worked as a process design engineer involved primarily in mass and energy balance, pipe and instrumentation, process flow diagrams and equipment selection and sizing, mechanical layout and functional specification design and development. The above includes ore beneficiation, hydrometallurgy and pyrometallurgy applications, such as chrome recovery, gold, uranium and base metal refining. 240


241-248_Laitala:text

2/13/09

9:20 AM

Page 241

LAITALA, H.T. Developments in organic holding tank structure for solvent extraction processes. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Developments in organic holding tank structure for solvent extraction processes H.T. LAITALA Outotec Oyj, Finland

Traditionally every solvent extraction process contains an organic tank. The purpose of such a tank is to provide constant feed to the organic pumps and to offer the buffer volume needed for the dynamic variations in the organic phase’s inventory. A conventional organic tank is a large piece of equipment, requiring a large footprint in the SX process area, and it contains a high inventory of costly organic phase. The latest development in organic holding tanks is a combination of the organic tank and a mixer-settler unit, and thus no separate organic tank is needed in the solvent extraction process. This new generation organic tank has several additional important process tasks to perform. These include removal of the aqueous entrainment coming from the extraction stages, washing of the organic phase before organic is sent to the stripping stage(s), and the organic inventory in the whole SX process can be minimized. Organic inventory reduction can vary from 10 to 40% of the conventional total organic inventory. The current development in the design of the organic tank makes the investment in the SX process cheaper, with the added benefits of higher electrolyte quality and lower operating cost.

Introduction A conventional organic holding tank in a solvent extraction process is a tank that provides a constant feed to the organic pumps. The physical size of the new, innovative organic tank design is comparable to a settler unit, while a conventional organic tank can contain up to 40% of SX process organic inventory. In this paper different organic tank modifications are presented. All these organic tank modifications are in use or will be taken into service in the near future. The aim of these modifications is to reduce the total investment and operating costs of a solvent extraction-electrowinning (SX-EW) plant, while maintaining product quality and quantity. Conventional loaded organic tank A conventional loaded organic tank is primarily an organic holding tank that provides DEVELOPMENTS IN ORGANIC HOLDING TANK STRUCTURE

241


2/13/09

9:20 AM

Page 242

constant feed to the organic pumps in the solvent extraction process, hence providing buffer capacity for any organic phase inventory changes. A schematic drawing of a conventional organic tank is shown in Figure 1 below. The conventional organic tank can be located after the stripping stages (barren organic tank = BOT) or after the extraction stages (loaded organic tank = LOT). For copper and zinc solvent extraction processes, a LOT is usually preferred, but for other metals such as Ni and Co a barren organic tank can also be installed for process reasons. In Cu and Zn SX processes LOT can be used to wash out physically and chemically entrained impurities. In Ni and Co SX processes BOT can be used for organic phase pretreatment purposes before the extraction stages. The shape of the conventional organic tank is unimportant and can be round or rectangular. The effective residence time of a conventional organic tank is usually between 30 and 60 minutes. Construction material for the organic tank generally is concrete with a liner. Commonly used liners are for example FRP, HDPE or stainless steel. Organic tanks can also be made from FRP or stainless steel with external supports. In a conventionally designed organic tank no reduction of physically or chemically entrained impurities can be expected and no specific flow patterns inside the organic tank are used. For the conventional organic tank arrangement, equipment investment cost is low, organic inventory cost is high, and operational cost for the SX-EW process is high. Loaded organic tank with entrainment removal and washing functions The modern organic tank in a solvent extraction process has multiple tasks including provision of constant flow to the organic pumps, and to remove and wash out impurities from the incoming organic phase (Figure 2). The design of this modern organic tank is usually rectangular and the tank’s effective retention time is 15 to 30 minutes, which is significantly shorter when compared to that of a conventional organic tank. The organic tank’s shape is important because a certain flow profile has to be achieved inside the tank. It is equipped with coalescing zones, where physically entrained aqueous impurities can be coalesced and separated from the organic phase. In these zones 50–90% of the physical entrainment can be removed. This generation of organic tank can also be equipped with a washing/scrubbing function, where aqueous phase is circulated inside the organic tank. Washing is done with pure water and scrubbing is done with acidic, metal containing aqueous solution (depending on the specific process under consideration).

Figure 1. Conventional organic tank

242



2/13/09

9:20 AM

Page 243

Figure 2. Outotec’s organic tank with entrainment removal and washing functions

Figure 3. Loaded organic tank with entrainment removal and washing/scrubbing functions. Total organic flow 2 500 m3/h

Because separate dispersion-creating pumps and mixers are not used in this design, removal of impurities that were chemically transferred to the organic phase during extraction is limited to 20–50%, depending on the impurity type and the washing/scrubbing conditions being employed. This type of LOT is used, for example, at Escondida, Cobre Las Cruces and Sepon Copper. An alternative LOT design includes a washing/scrubbing function (Figure 3) that can be used in some cases as the primary stage of organic phase washing, generally where a separate washing stage is not justified, but some impurity removal from the loaded organic is still needed. A LOT with a washing/scrubbing function can also be installed to assist a separate washing stage in cases where PLS (pregnant leach solution) impurity levels are very high or difficult to handle. The capital investment required cost for a LOT with a washing/scrubbing function is lower, organic inventory cost is lower, and operational cost for the SX-EW process is lower when compared to an SX-EW process employing a conventional organic tank. Combined lot and mixer-settler stage It is also possible to combine a mixer-settler stage with the organic tank. This is illustrated in Figure 4. DEVELOPMENTS IN ORGANIC HOLDING TANK STRUCTURE

243


2/13/09

9:20 AM

Page 244

Figure 4. Combined mixer-settler and organic tank

Figure 5. Design of the Franke-project combined washing mixer-settler and organic tank

The combination of a mixer-settler unit and an organic tank exhibits all the main features of individual mixer-settler and LOT/BOT units. This combination can be done to any solvent extraction stage, but the combination is most suited to the first extraction stage, last stripping stage or washing stage. This combination can reduce the solvent extraction organic inventory by 10–40% and the total SX process investment cost from 10–25%. This SX process upgrade will be taken into service by Franke-project in Chile in year 2009 (Figure 5). This equipment and process is proprietary technology of Outotec. From Figure 6 it can be seen that the combination of a mixer-settler unit and an organic tank reduces the SX plant area around 10–20%. 244



2/13/09

9:20 AM

Page 245

(a)

(b)

Figure 6. Mixer-settler unit’s combination with an organic tank and its effect on the SX process layout (a = SX layout with configuration 3E+1W+1S+LOT, b = same SX layout but with combined washing stage and loaded organic tank)

Figure 7. Process connection of the conventional organic tank and mixer-settler units, and that of the new approach where the mixer-settler unit and organic tank are combined

Impurity removal in a loaded organic tank The conventional link between the organic tank and the mixer-settler units and that of the more modern arrangement where the mixer-settler unit and organic tank are combined are presented in Figure 7. DEVELOPMENTS IN ORGANIC HOLDING TANK STRUCTURE

245


2/13/09

9:20 AM

Page 246

Figure 8. Impurity removal in a Cu SX-EW process

Table I Effect of impurity removal on Cu EW’s bleed stream Conventional organic tank, no separation, no washing Income organic A/O entrainment from SX CI concentration in aqueous Electrolyte CI max. conc. Electrolyte bleed

1 000 m3/h 300 ppm 10 000 ppm 30 ppm 100 m3/h

Outotec organic tank, separation, no washing Income organic A/O entrainment from SX CI concentration in aqueous Electrolyte CI max. conc. Electrolyte bleed

1 000 m3/h 300 ppm 10 000 ppm 30 ppm 25 m3/h

Outotec organic tank, separation, and washing Income organic flow A/O entrainment from SX CI concentration in aqueous Electrolyte CI max. conc. Electrolyte bleed

1 000 m3/h 300 ppm 10 000 ppm 30 ppm 2.5 m3/h

In Figure 8 and Table I an example is calculated of how the different process connections affect the EW bleed stream. In this example, the main impurity has been the physically entrained chloride ion in the copper electrowinning circuit. Similar calculations can be made for example, to iron, aluminum, manganese and calcium. Conclusions In the new generation organic tank designs, several process tasks can be combined. Conventionally the organic tank acts as a dynamic volume for the SX process organic phase, but it can also be employed to reduce the amount of physically and chemically entrained impurities, 246



2/13/09

9:20 AM

Page 247

and provide residence time for the SX reactions to proceed. By combining an organic tank with a mixer-settler stage, the overall SX process organic inventory will be reduced, thus reducing the associated fire risk and the first fill cost of the SX process. The total plant size and the number of main process equipment pieces will also be reduced, thus lowering the project investment cost while not jeopardizing the cathode quality and quantity.

Hannu Tapani Laitala Process Manager, SX Technology Manager, Outotec Oyj, Finland 2005

Outotec Oyj, Base Metals, Hydrometallurgy, SX Technology Manager 2002–2005 Ashland Specialty Chemicals, Porvoo Polyester Plant, Finland, Plant Engineer 1999–2002 JP-Engineering, Process Design Department, Vantaa, Finland, Senior Process Designer 1997–1999 Outokumpu Engineering Contractors, Project Implementation Department, Espoo, Project Engineer 1994–1997—Outokumpu Harjavalta Metals, Ni Refinery, Harjavalta, Finland, Process development engineer at Harjavalta Nickel plant (leaching, purification, Co SX, hydrogen reduction and EW) Relevant project experience 2008–2010 Tia Maria, Peru, SPCC Cu SX-EW Plant, Process Manager, SX technology manager, (responsible for the technical solutions used in the project). 2008–2009 Chambishi, Zambia, Chambishi Metals Plc Cu SX Plant, Process Manager, SX technology manager (responsible for the technical solutions used in the project). 2008–2009 Harjavalta, Finland, Norilsk Nickel Ca SX Plant, Process Manager, SX technology manager (responsible for the technical solutions used in the project). 2007–2009 Franke, Chile, Centenario Copper Cu SX-EW Plant, Process Manager, SX technology manager (responsible for the technical solutions used in the project). 2007–2009 Miheevsky, Russia, RCC Cu SX-EW Plant, Process Manager, SX technology manager, (responsible for the technical solutions used in the project). 2007–2009 Assarel, Bulgaria, Assarel Medet Cu SX-EW Plant, Process Manager, SX technology manager, (responsible for the technical solutions used in the project). 2008 EdZ, Spain, Espanola del Zinc Hydrometallurgical Plant, Process Manager, SX technology manager, (responsible for the technical solutions used in the project and pilot run). 2007–2008 Pueblo Viejo, Dominican Republic, Barrick Gold Hydrometallurgical Base Metal Plant, Process Manager, SX technology manager, (responsible for the technical solutions use in the project and pilot run). 2005–2006 Outokumpu Technology, Finland, New Generation SX Cell, SX Technology Manager, design and development of a new generation SX cell. 2005–2006 Outokumpu Technology, Finland, New Generation SX Mixers and Separation Fences, SX Technology Manager, Design and development of a new generation SX equipment. 2006 Mopani Copper Mines Plc, Zambia, Co Plant Audit, Project Manager, Technical audit of the Co plant’s leaching, solution purification and EW departments. DEVELOPMENTS IN ORGANIC HOLDING TANK STRUCTURE

247


2/13/09

9:20 AM

Page 248

1994–2005 Many implementation projects for different customers mainly in Finland and Europe for metallurgical and chemical industry, Senior Process Designer, Project Manager and Engineering Manager. Extension courses 2005

1999 1999

248

AEL—Centre for technical training, Examination for pressure equipment supervisors, 2003 TUKES, Safety Technology Authority in Finland, examination for supervisors of usage and storage of flammable and explosive chemicals. Helsinki University of Technology, Metallurgy, 48 CEUs. Helsinki University of Technology, Plant Engineering, 18 CEUs, Patent no 107236: Method to reduce the SX cells physical size and a SX cell to be used in a SX process.


249-256_Marsden:text

2/13/09

9:22 AM

Page 249

MARSDEN, T. and JICKLING, J The next generation of permanent cathode and lead anode technology. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

The next generation of permanent cathode and lead anode technology T. MARSDEN* and J. JICKLING† *Thuthuka Group Ltd, Johannesburg, South Africa †EPCM Services Ltd, Toronto, Canada The demand for continuous improvement in the industrial world is an ongoing process. Some sectors, where the competitive forces are well established, have led the development of new and reliable technological innovations. Although the metallurgical industry can rightfully be proud of its process improvements, there is still work to be done in materials and in material handling. In some cases, the improvements will be as simple as adopting technologies that have been perfected in other processing or manufacturing industries. This paper discusses the next design generation for copper tankhouses which will yield: • Improved copper quality • Improved equipment availability • Lower life cycle costs • Safer operations. These benefits will be realized through new developments and design improvements in: • Permanent cathodes • Cathode washing • Cathode stripping machines • Anode preparation machines • Improved lead anode design • Recycling of lead anodes and refurbishment of cathodes. The demand for base metals, and in particular copper, has led to more research into improving efficiencies in electrowinning and refinery plants around the world. One of the areas previously taken as a matter of course was the production of lead anodes and stainless steel cathodes. In South America alone there are over 600 000 cathodes and anodes in circulation—a very significant figure.

Lead anodes Much has changed in the manufacture of lead anodes over the past 50 years. Alloys Similar to the automotive battery, antimony was the first metal to be used as an additive to THE NEXT GENERATION OF PERMANENT CATHODE AND LEAD ANODE TECHNOLOGY

249


2/13/09

9:22 AM

Page 250

improve the grain structure and give the anode more mechanical strength. The additional strength was mainly required to give the anode blade more rigidly and reduce shorting caused by the anode buckling due to deformation occurring during normal handling in the cleaning or stripping process. Six% to 8% antimony was added to the refined lead. Antimonial lead was used in the recovery of copper, cobalt and nickel. One% silver alloy was used for the recovery of zinc and manganese. The automotive battery manufactures had in the interim reduced the antimony content in their alloys with the addition of arsenic and selenium, and subsequently introduced calcium as an alloy, particularly in relation to the maintenance free battery. Calcium, for the same reason, was first introduced worldwide as an additive to a silver lead alloy in a patent developed by Michael Thom in South Africa in 1982. The addition of calcium enabled the end user to reduce the silver content first from 1% to 0.75% and then eventually to 0.5%. Prior to this the silver lead alloy anode was easily bent, causing shorting on contact with the cathode. The addition of the calcium gave the blade more rigidity and increased the life of the anode by more than double. Calcium, with tin and aluminium, was then introduced into anodes for the recovery of copper, nickel and cobalt. Whereas lead antimonial and silver lead alloys cast relatively easily, problems were encountered in casting calcium lead alloys, which necessitated a change in manufacturing techniques. Method of manufacture Casting Initially the anodes were cast in book-type moulds. Casting was from the top with the copper hanger bar in position. To get sufficient flow and give sufficient strength the anode, blade was cast up to 16 mm thick. Casting from the top resulted in air entrapment, dross inclusion and hot spots in the centre of the anode. The latter items all increased corrosion of the anode, reduced the life of the anode, and in some cases increased the migration of lead into the electrolyte. Another South African first was the introduction of a semi-pressure die casting method by casting the anode through a valve and runner at the bottom of the mould. The runner (later cut off from the blade bottom) absorbed the heat and eliminated the hot spot. The lead pushed the air upwards and escaped through air vents above the copper hanger bar. The moulds are hydraulically operated and water cooled. It must be emphasized that temperature control is critical in the process. Rolled anode At the time South Africa was a leader in the improvement of casting lead anodes. The same success was not achieved elsewhere, particularly with calcium lead alloys, and in some cases this accelerated the introduction of the anode with a rolled blade. The automotive battery manufactures had moved to a process involving a continuous cast rolling mill producing thin lead plates. The electrical flux density (the capacity of a conductor to convey electrical current) is directly proportional to the grain size and grain structure in the metal conveyer. The smaller the grain size, the better the electrical current properties. It was found by rolling the blades, particularly in the case of calcium lead, the size of grain structures decreased. Lead, similar to copper, also work hardens. The rolling also increased the mechanical strength of the blade. The anode manufacturers cast the header bar section (encapsulating the copper in position) and rolled lead plates on a conventional rolling mill. The lead blade was then lead burnt (welded) to the header bar section. 250



2/13/09

9:22 AM

Page 251

In the case of calcium lead, it is extremely difficult and in fact not practical to weld calcium lead sheet to calcium lead sheet. The calcium lead oxidizes immediately on the introduction of the welding heat. It is virtually impossible to weld once it is oxidized. As a result it was found that the hanger bar had to be cast in antimonial lead (normally 6% antimony) and then welded to the calcium rolled plate using a flux and antimonial rod. It is a well-known fact that even in steel the weld is more susceptible to corrosion than the rest of the plate. Lead and the other base metals are not the exception. Another method of overcoming the above welding problem was slotting the lead blade into the pre-lead plated copper hanger bar and soldering the blade in place. In 2005 John Turner and Thomas Meyer developed a patent for the manufacture of a rolled anode with no weld or joint between the lead and copper hanger bar. This was a seamless rolled anode and was described by a leading international consultant as the first meaningful innovation in lead anodes in 20 years. In principle, the patent is held on the process whereby a complete anode with its busbar cast in position and the blade portion of the anode then put through a rolling mill and the blade thickness reduced with resulting benefits including the reduction of grain size, added mechanical strength and weight reduction. The seamless rolled anode is manufactureds as follows: Plant and equipment • A casting facility incorporating water cooled hydraulically operated mould. • A patented rolling mill, which enables the cast anode to be rolled with the copper busbar in position with a minimum of 5 mm lead encapsulated around the copper and eliminating all soldered or welded joints. Method of manufacture After casting the anode is rolled with the copper busbar section in position and reduced by 30% to 40%. This work hardens the metal hence giving the anode more rigidity, and the lead has a finer grain structure, giving the anode a longer life than a cast anode. Conventional rolled anodes as described are being supplied to base metal electrowinning plants around the world. While these conventional rolled anodes are an improvement on a cast anode, the conventional rolling method has the following disadvantages when compared to the patented anode. • The conventional rolled anode has a welded or soldered joint. The weld is dependent on manual welding (lead burning) which is subject to human error. The same applies to a soldered joint, which could be prone to corrosion from mechanical damage to the plated lead layer. The corrosion attack on the weld is at times greater than that on the rest of the anode and as a result the solution level in the cell may have to be kept lower than the level of the weld of the anode. The weld leads to the common corrosion found in the liquid gas phase corrosion area, which is often the most severely attacked area in a reaction vessel. The soldered joint is also subject to splashing and any mechanical damage to the plated copper busbar will expose the copper to a corrosion attack. The lead covering on the cast header bar is at least 5mm thick, which will protect the copper busbar. • The conventional rolled calcium anode has a calcium lead blade but the lead used covering the header bar is often antimonial lead (As mentioned, the antimonial lead facilitates the welding of the joint.) When recycling, the two different alloys (antimonial and calcium) are often not separated before the time, and the alloys are then mixed. Costly alloy adjustments have to be done to adjust the alloys to the original compositions. The seamless rolled anode has calcium lead throughout with the distinct advantage that on recycling, the calcium lead alloy is not contaminated by the antimonial lead THE NEXT GENERATION OF PERMANENT CATHODE AND LEAD ANODE TECHNOLOGY

251


2/13/09

9:22 AM

Page 252

• The method of manufacture of the new anode eliminates the weld and this makes the anode more competitively priced in comparison to the conventional rolled anode. It must be emphasized that the anode in question is not a new anode but an improved anode. • Anodes from this method of manufacture have performed to expectation in the DRC (Democratic Republic of Congo) for over two years. Resultant samples from these anodes passed on to Mintek (Mineral technology of South Africa) showed no visible corrosion. It has been estimated that a service life of at least five years can be achieved, provided the electrolyte remains constant. In particular chlorides must be kept to a minimum. Replacement of the anodes will mostly be due to mechanical damage and not due to catastrophical failure of the anode. Cathodes This section describes the cathode blanks available from the vendors, outlines the relative advantages and disadvantages of these cathodes and provides a ‘qualitative level’ valuation of several common parameters of the cathodes. Cathode design alternatives Outotec cathode This cathode blank is manufactured by welding the stainless steel sheet to a stainless steel hanger bar, which has an internal solid copper core. We have limited knowledge of the service experience of this Outotec cathode. To the best of our knowledge the Outotec cathode is in at least five plants that are in commercial operation or in construction. The plants include Las Cruces SX-EW in Spain, the Milpilas SX-EW plant in Mexico, the Frambros refinery in China, a small SX-EW plant in Laos, and a refinery in Finland. Cobra unsheathed cathode This cathode blank is manufactured by welding the stainless steel sheet to a copper hanger bar, using a dissimilar metal weld. This cathode is based on cathode designs that have been used for over 30 years. The number in service exceeds five hundred thousand (500 000) cathodes. EPCM sheathed cathode This cathode blank is similar in design to the Cobra unsheathed cathode described above, except that it has a stainless steel sheath covering the hanger bar above the sheet. The sheath provides protection for the copper and the Cu/SS weld against corrosion attack by acid mist. The ends of the cathode sheath can be sealed to the hanger bar using either a seal weld or by installation of a copper sleeve, with a chemical resistant sealant. Xstrata ISA cathode This cathode blank is manufactured by welding a stainless steel hanger bar to the stainless steel sheet. To provide electrical conductivity, a 2.5 mm copper layer is plated over the hanger bar and 15 mm down the sheet. This cathode is based on cathode designs that have been used for over 30 years. The number in service exceeds one million five hundred thousand (1 500 000) cathodes. 252



2/13/09

9:22 AM

Page 253

Xstrata Kidd cathode This cathode blank is similar in construction to the EPCM sheathed cathode described above, except that the cavity between the sheath and the hanger bar is filled with a chemical resistant filler to isolate weld from contact with electrolyte. It is important to note that all of the available cathodes are now supplied with a 90 degree Vgroove along the bottom edge of the sheet. The V-groove aids in separation of the copper deposit from the cathode blank. Electrical resistance The electrical resistance is an important parameter of the cathodes as it affects the cell voltage drop and the overall power consumption of the tankhouse. For the purpose of this paper, the resistance is calculated by adding the theoretical resistance in the hanger bar copper from one end of the bar to the centre, and the calculated resistance in the stainless steel sheet from the hanger bar attachment point to the electrolyte level. This is an approximation only as several factors will affect the resistance of the cathode in service. However, it provides a close approximation to the relative resistance between the different cathode types and is appropriate for this paper. The estimate of the cathode resistance also determines the estimated annual power cost per cathode, for this cathode resistance, based on the nominal current density and operating time of the refinery. Another important aspect to the electrical resistance is the increase in resistance over time. As discussed any exposed copper on the hanger bar is subject to dissolution over time due to contact with sulphuric acid in the atmosphere above the cells. While dissolution rates as high as 1 mm per year have been reported in some SX-EW plants, the slow dissolution of the copper can significantly affect the electrical resistance over time. Corrosion resistance Generally, three types of corrosion occur with stainless steel cathodes, as follows: • Chloride pitting of stainless steel—corrosion of the stainless steel sheet can occur, generally in the form of pitting near the solution line. The cause of this corrosion is normally high chloride levels in the electrolyte and it generally occurs in SX-EW plants, and is a rare occurrence in refineries. All of the cathodes being considered would be similarly affected by chloride pitting, as they all use the specified 316L stainless steel for the sheets. New materials are being substituted for 316L stainless steel and this will be discussed further in this paper. • Dissolution of copper on hanger bars—due to the solubility of copper in sulphuric acid and the presence of acid mist in the air above the cells, some dissolution of any exposed copper on the hanger bar will occur. This is very prevalent in some SX-EW plants due to the high levels of acid mist but is generally less prevalent in refineries. Nevertheless, some copper dissolution will occur and cathodes with exposed areas of copper will require repair or refurbishment at some point. • Galvanic corrosion—any area where dissimilar metals are in contact in the hanger bar or at the hanger bar/sheet connection will be subject to galvanic corrosion. Again this is most prevalent in SX-EW plants as condition is aggravated by acid mist, but it does occur in refineries. Experience has shown that some refineries have inherently more corrosive environments than others and there is no obvious technical reason for this. Durability The durability of a cathode blank relates to how it will withstand the normal operations of a copper refinery, excluding corrosion that is discussed earlier. Generally, cathodes face possible damage due to the following: THE NEXT GENERATION OF PERMANENT CATHODE AND LEAD ANODE TECHNOLOGY

253


2/13/09

9:22 AM

Page 254

• Overloading of hanger bar—cathode hanger bars can sometimes be overloaded by overplating on the cathode, impact loading on the cathode during handling, or excessive force being applied in the CSM, normally during copper stripping. Over-plating is not normally an issue as the weight of the copper is well within the load carrying capacity of the hanger bar. However, impact loading during handling and excessive forces in the CSM can sometimes lead to large forces being applied to the hanger bars • Abrasion of the hanger bar—abrasion of the hanger bar occurs during handling of the cathode, mainly by the CSM. During processing through the CSM the cathode is necessarily supported by the hanger bar on transfer cars, conveyors and in flexing/stripping stations. Additionally, abrasive media brushing normally cleans the end of the hanger bar where it contacts the busbar. • Bending of sheet—bending of the sheet may occur due to overflexing of the sheet or otherwise overstressing the sheet with a compressive force, most frequently in the CSM. Generally, this affects the verticality of the cathode so that it is not acceptable for use in the process, but does not lead to structural damage to the cathode. Maintainability Maintainability refers to the ease at which minor and major repairs can be performed on a cathode. Minor repairs include: • Replace edgestrips • Repair minor scratches on sheet • Straighten slightly bent sheet • Straighten slightly bent hanger bar • Repair cathode verticality • Clean sheet. Major repairs are normally required only as a result of gross damage to the cathodes from handling or CSM mishaps and include the items listed below. Major repairs include: • Replace entire hanger bar • Replace entire sheet • Recuperation of damaged sheet • Replate stainless steel header bar. Alternative materials of construction The sharp increase in nickel and molybdenum during the period of 2004 through 2008, led both permanent cathode suppliers and customers to look for alternative and substitutes for the traditional 316L stainless steel cathode sheet. The substitutes currently being offered are lean duplex stainless steels. Duplex stainless steels contain approximately 50% ferrite and 50% austenite, which result in a material that has properties representative of both classes of stainless steels. Like nickel free ferrite grades, duplex grads are magnetic and more resistant to chloride stress corrosion cracking than austenitic series grades. Duplex alloys have a good ductility and toughness, approaching that of the austenitic grades. Also duplex alloys are stronger than comparable austenitic and ferritic stainless steels. Lean duplex stainless steels have been designed to have the properties of a duplex material, but contain less nickel and molybdenum than standard duplex grades such as 2005, which results in a cost savings. 254



2/13/09

9:22 AM

Page 255

Tom Marsden Marketing Manager, Thutuka Group Limited, South Africa Tom Marsden joined the family business in 1961 to eventually become sole owner and Managing Director in 1969. Introduced Lead anodes as a product line in the latter part of the nineteen sixties and has been involved in the manufacture of anode plant and manufacture ever since. With Michale Thom patent. Introduced calcium as a beneficial addition to the silver lead alloy. This was a worldwide first for lead anodes as used in the zinc industry. Introduced new casting techniques for the casting of all lead alloys and in particular calcium and strontium lead alloys. Since involved in various methods of rolled anodes including plant and equipment for the manufacture of seamless rolled anodes. Besides the anode industry, has been involved in development of continuous cast lead sheet programs in Australia, the UK and the USA and lead radiation shielding with various international companies.

John Jickling Director, EPCM John Jickling studied Mechanical Engineering at Queen’s University in Canada. After completion of his Bachelor’s Degree, he joined EPCM Services Ltd. A consulting and technology company located in Ontario Canada. Afrter working for two (2) years at EPCM, he was prompted to start-up Technologies Cobra S.A., an EPCM Technology Group Company located in Antofagasta, Chile. While at Tecnologias Cobra, EPCM developed our services to provide after market support to the Permanent cathode Technology users in the Copper Refining Industry. John has been active in Permanent Cathode Product development, Manufacturing processes and the development of Permanent Cathode Refurbishment Technologies.

THE NEXT GENERATION OF PERMANENT CATHODE AND LEAD ANODE TECHNOLOGY

255


256

2/13/09

9:22 AM

Page 256


257-272_Robles:text

2/13/09

9:24 AM

Page 257

ROBLES, E., CRONJE, I., and NEL, G. Solvent extraction design consideration for the Tati Activox® plant. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2008.

Solvent extraction design consideration for the Tati Activox® plant E. ROBLES*, I. CRONJE*, and G. NEL† *Hatch Africa (Pty) Ltd, South Africa †Norilsk Nickel Africa (Pty) Ltd, South Africa

The Tati Activox® Project (TA ®P) is the first step towards the implementation of the Norilsk Nickel Activox® process. The ultra fine grinding (UFG) and autoclave pressure leach process conditions are the heart of the Activox® patent. Low grade base metal sulphide concentrates are leached in the process to recover the base metals and to produce LME grade nickel and copper cathodes and cobalt carbonates. The downstream processing of the metal-containing liquor produced by this process includes solvent extraction circuits to extract and concentrate metal-rich solutions, before the various final products are obtained. This paper gives an overview of many of the relevant design aspects taken into consideration in the design of the Tati Activox® solvent extraction circuits. It addresses key design elements related to leach process conditions such as scaling, materials selection for corrosive solutions, pH control, crud formation and treatment, multiple SX trains designed to prevent organic cross-contamination, and design aspects related to the mitigation of fire risks.

Introduction The Activox® process is a hydrometallurgical process route developed by Norilsk Process Technology to treat a wide variety of metal sulphide concentrates1. A combination of ultra fine grinding, which activates the sulphide minerals and oxidation at a relatively low temperature and pressure, form the core of the patented technology, which is expected to provide a competitive alternative to the traditional pyrometallurgical process routes. Work on the Tati Activox® Project (TA®P) started in 1998 with laboratory testwork. The development towards commercialization has advanced systematically from laboratory-scale testing in Australia through various studies to the design, construction and successful operation of a 1:170 scale demonstration plant on the Phoenix mine site in the North Eastern district of Botswana, Southern Africa. In 2006 Hatch was awarded the contract for the execution of the Botswana Metal Refinery (BMR) for TA®P. This hydrometallurgical refinery is designed to produce 25000 tpa nickel and 22000 tpa copper as LME grade cathodes and 630 tpa cobalt as a cobalt carbonate. SOLVENT EXTRACTION DESIGN CONSIDERATIONS FOR THE TATI ACTIVOX® PLANT

257

257-272_Robles:text

2/13/09

9:24 AM

Page 258

Figure 1. Tati Activox® flowsheet

The general flow sheet for the TA®P process is shown in Figure 1. The Tati concentrate is trucked from the Tati Phoenix mine complex to the BMR where the concentrate is repulped in attritioners to 50% solids with recycled copper raffinate. The repulped concentrate is ground to 80% passing 10 microns using two-stage vertically stirred detritor ultra fine mills. The mill product is fed to two parallel autoclaves where the slurry reacts with oxygen at moderate high pressures (10 bar gauge) and moderate temperatures (105 degrees Celsius) to produce copper, cobalt and nickel sulphates in solution for downstream recovery. The pregnant leach solution (PLS) reports to a solid-liquid separation section, which includes a multi-stage counter current decantation (CCD) washing circuit where the leach residue solids are washed with recycled process water. Washed leach residue solids are pumped to the PGE recovery circuit. The overflow solution from the CCD train is clarified prior to reporting to a copper PLS pond. Copper PLS is contacted with an organic solvent in the copper SX circuit that selectively extracts copper from the copper PLS solution. Copper is stripped from the organic using spent electrolyte from the copper EW circuit (Cu EW). The rich electrolyte is treated to remove trace organic before being pumped to the Cu EW circuit. The bulk of the copper SX raffinate is recirculated back to the concentrate repulp and the remainder is advanced to the cobalt and nickel recovery circuits. The Cu-rich electrolyte is sent to the Cu EW circuit to produce copper as LME grade copper cathode. Spent electrolyte is recirculated back to copper SX for enrichment via stripping of the copper-loaded organic phase. Iron in the copper SX raffinate is removed to less than 10 ppm in solution using limestone, in a two-stage iron removal circuit. Cobalt PLS is contacted with an organic solvent that selectively extracts cobalt from solution. Cobalt is stripped from the loaded organic phase with an acidic strip solution and the resultant loaded strip liquor (LSL) is filtered and pumped to the cobalt precipitation circuit. Organic is removed from the Cobalt SX raffinate before proceeding to the nickel PLS storage tank. Nickel PLS is contacted with an organic solvent that extracts the nickel (selectively over calcium and magnesium) from solution. Nickel is 258


257-272_Robles:text

2/13/09

9:24 AM

Page 259

stripped from the loaded organic phase with spent nickel electrolyte recirculated from nickel electrowinning circuit (Ni EW). The rich electrolyte is treated to remove trace organic before reporting to the Ni EW cell house. The nickel SX raffinate stream is transferred to the ammonia recovery circuit. The Ni EW process is designed to produce LME grade nickel cathode. Recovered spent electrolyte from Ni EW is treated to remove lead prior to being recirculated to nickel SX for stripping. Ammonia is recovered from the nickel SX raffinate stream for reuse as an aqueous ammonia solution for pH control in the cobalt and nickel SX processes. The ammonia is liberated from the nickel raffinate using the Norilsk patented ammonia recovery process. The process solution from ammonia recovery is recycled back to the CCD circuit and used as wash water. The residue from the CCD circuit is treated by flotation to recover the PGEs in the leach residue. Tails from the rougher circuit are neutralized together with other process effluent streams before reporting to the tailings dam. Description of TA®P solvent extraction plants In TA®P the process for leach liquor purification and concentration of metals consists of three SX circuits to remove Cu, Co and Ni in sequential stages by using different selective extractants. Description of the SX circuits Copper solvent extraction circuit (Cu SX) Copper is selectively transferred from the relative low grade Cu PLS into a high-purity electrolyte stream using a copper selective extractant. The extractant employed is an oximebased copper-selective chelating reagent that is dissolved in a high flashpoint hydrocarbon diluent to make up the organic phase. The copper solvent extraction circuit consists of three main process steps: • Extraction—copper PLS is contacted with the stripped organic phase in two extraction stages to transfer copper from the aqueous to the organic phase • Scrubbing—the loaded organic is scrubbed with copper electrolyte bleed to remove coextracted iron and aqueous-entrained chloride and manganese. Although highly selective for copper over most other cationic species under extraction conditions, a small quantity of Fe3+ is co-extracted. The scrubbing stage minimizes entrainment of aqueous in the loaded organic phase that otherwise will result in the undesirable build-up of species such as chloride and manganese in the electrolyte • Stripping—the scrubbed organic phase is contacted with spent electrolyte from EW in two stages, utilizing the acid generated during the copper deposition process in Cu EW. Copper is transferred from the organic into an aqueous phase. Cobalt solvent extraction circuit (Co SX) Cobalt is recovered from the Co PLS following the iron removal process, using a solvent extraction circuit. The extractant employed is the phosphinic acid-based Cyanex 272, dissolved in a high flashpoint hydrocarbon diluent. This provides selective extraction for cobalt over nickel by controlling the pH in the extraction stages. Co-extraction of iron, manganese, zinc and copper also occurs. The cobalt solvent extraction circuit consists of four main process steps: • Extraction—Co PLS is contacted with stripped organic phase and cobalt is transferred from the aqueous to the organic phase in three stages. The pH is controlled by the addition of aqueous ammonia SOLVENT EXTRACTION DESIGN CONSIDERATIONS FOR THE TATI ACTIVOX® PLANT

259

257-272_Robles:text

2/13/09

9:24 AM

Page 260

• Scrubbing—loaded organic is washed with acidified, filtered water to remove coextracted nickel from the organic phase in a single stage • Stripping—scrubbed organic is contacted with acidified cobalt strip liquor and the cobalt is transferred from the organic to the aqueous phase in two stages • Cyanex recovery—raffinate is washed with fresh diluent to minimize organic cross contamination to the nickel solvent extraction circuit in a single stage. Nickel solvent extraction circuit (Ni SX) Nickel is recovered from Co SX raffinate using solvent extraction. The extractant employed is the carboxylic-acid based Versatic 10 that is dissolved in a high flashpoint hydrocarbon diluent. By control of the extraction pH, the solvent-extraction process is selective for nickel over such species as calcium and magnesium. The nickel solvent extraction circuit consists of the following four main process steps: • Extraction—nickel pregnant leach solution (Ni PLS) is contacted with organic phase and nickel is transferred from the aqueous to the organic phase in five stages. The pH is controlled by the addition of aqueous ammonia • Versatic recovery—prior to entering the extraction circuit the stripped organic phase is contacted with nickel raffinate exiting the extraction circuit in a single stage to recover aqueous-soluble Versatic 10 from the aqueous into the organic phase. This is achieved by reducing the pH with diluted sulphuric acid solution in this stage. Recovering the soluble Versatic 10 minimizes reagent losses and downstream frothing in the ammonia recovery circuit. • Scrubbing—loaded organic is scrubbed with acidified nickel rich electrolyte bleed solution to remove co-extracted calcium from the organic phase in a single stage. • Stripping—scrubbed organic is contacted with spent electrolyte from Ni EW in two stages and nickel is transferred from the organic to the aqueous phase. Most of the sulphuric acid utilized in stripping is generated in the electrowinning process. Contactor equipment The Tati Activox ® SX plant design is based on the use of mixer-settlers as contactor equipment. It makes use of the countercurrent Side-FeedTM mixer-settler concept of Miller Metallurgical Services (MMS)2. Each mixer-settler consists of a two-stage mixer and a settler. A dual mixing system provides high overall stage efficiency by minimizing back-mixing, minimizing short circuiting, and by ensuring sufficient retention time in the mixers. The primary mixing vessel is fitted with a pump-mix type of impeller, which provides the head for the inter-stage pumping of solutions as well as mixing of the two phases. The primary mixing vessel incorporates a false bottom into which the organic, aqueous and recycle streams are introduced. Mixed phases enter the secondary mixing vessel via a connecting pipe. This vessel is fitted with a mixing impeller designed for maximum mixing efficiency with minimum shear. The dispersion overflows into a shallow gravity settler where sufficient area is provided for disengagement of the aqueous and organic phases. Cross-sectional picket fences maintain a band of dispersed phase through which finer droplet coalescence is promoted. A weir arrangement at the discharge end of the settler allows for separate discharge of the organic and aqueous phases. The aqueous/organic interface level in the settler is set during commissioning by manual adjustment of the physical height of the aqueous overflow weir. The design allows all the mixer-settlers to run with organic or aqueous continuity by adjusting the internal recycles to the primary mixer from the settler. The general philosophy followed in all three circuits is that the extraction stages be operated in organic-continuous 260


257-272_Robles:text

2/13/09

9:24 AM

Page 261

mode. This minimizes the entrained organic loss in the raffinate and assists in the compaction of crud solids. Organic continuity in the first extraction stage could result in high aqueous entrainment to the loaded organic phase. The deleterious effects of this are curtailed to some extent by the inclusion of the loaded-organic scrub stage that is present at all the circuits. Configuration of the TA®P solvent extraction circuits The Table I below shows a comparison between the three SX circuits for TA®P. Parameters compared are type of extractant and diluent, number of mixer-settlers per process step, existence of pH control and configuration for organic removal and recovery. The number of extraction stages differ from circuit to circuit depending on the extraction time, type of extractant employed and the concentration of the metal present in the PLS. The propensity for scaling in the circuit may require the addition of an extra extraction stage as in the case of the nickel circuit. The likelihood of downtime is reduced by the addition of an extra extraction stage, which allows extraction operations to carry on as planned while cleaning can be carried out simultaneously. Stripping in all three circuits comprise two stages where loaded organic is contacted with spent electrolyte or acidic strip liquor. As for extraction, the aqueous and organic phases flow countercurrently with loaded organic entering the first stripping stage and spent electrolyte or strip liquor entering the second stripping stage. The aqueous solution exiting the first stripping stage is enriched in metal, and after organic removal, becomes the feed to the metal winning circuit that follows. Table I Comparison of the TA®P SX circuits

Extractant Extractant family Diluent Total flow rate per settler Operating temperature PLS feed storage Nr of extraction mixer-settlers

Copper SX

Cobalt SX

LIX 984N Oxime based Sasol SSX 210 1002 m3/h 30–40°C Pond (4 days) 2

Cyanex 272 Phosphinic acid Sasol SSX 210 309 m3/h 40°C Pond (2 days) 3

Nickel SX

Versatic 10 Carboxylic acid Sasol SSX 210 383 m3/h 40°C Tank (4 hours) 5 (plus one spare stage for potential descaling of stage 1 or 2) Nr of scrub mixer-settlers 1 1 1 Nr of stripping mixer-settlers 2 2 2 Organic recovery stage No Yes Yes pH control in extraction stages No Yes (ammonium hydroxide) Yes (ammonium hydroxide) pH control in scrub stage No Yes (sulphuric acid) Yes (sulphuric acid) pH control in stripping stages No Yes (sulphuric acid) Yes (sulphuric acid) pH control in organic recovery No No Yes (sulphuric acid) Potential scaling No Yes Yes Raffinate after-settler Yes Yes Yes Rich electrolyte/LSL after settler No No Yes Raffinate coalescing filters No Yes No Electrolyte/LSL coalescing filters Yes Yes Yes Electrolyte/LSL carbon columns No No Yes Crud treatment system Tri-phase centrifuge Tri-phase centrifuge Tri-phase centrifuge

SOLVENT EXTRACTION DESIGN CONSIDERATIONS FOR THE TATI ACTIVOX® PLANT

261

257-272_Robles:text

2/13/09

9:24 AM

Page 262

Sufficient surge capacity is provided upstream of the SX circuits. In the case of copper, fourday surge capacity is provided in the PLS storage pond. This pond features two pre-settlement compartments to minimize solids ingress to the solvent extraction process. The Co PLS pond has two days of surge capacity. Cobalt raffinate is stored in a four-hours surge tank with sufficient capacity to buffer against operational surges upstream of Ni SX. All three SX circuits can therefore be operated independently without being affected by upstream bottlenecks. This provides the system with flexibility and continuity avoiding unnecessary shutdowns and start-ups. The surge capacities between process steps were sized based on a dynamic model. In all three circuits loaded organic flows via gravity from the respective extraction stages into a series of two organic surge tanks. Both tanks are equipped with internal weirs for the separation of entrained aqueous. Loaded organic is pumped to a single scrub stage where it is contacted with an acidified aqueous solution. The scrub raffinate flows by gravity into the raffinate after settler to minimize the entrainment of organic in the aqueous phase. The other significant difference between these SX circuits is the strategy to remove or recover entrained organic from the aqueous products (raffinate and rich electrolyte/loaded strip liquor). The strategy for organic removal or recovery is essentially dictated by the requirements of the downstream process stage, as explained below. For TA®P it is very important to minimize the loss of extractant from the circuit, to reduce the impact on the downstream operations. Also, organic extractant entrainment losses must be minimized from an operating-cost perspective. The Cu raffinate reports to an after settler followed by the iron removal process, which forms a buffer between Cu SX and Co SX. The Co raffinate stream first passes through an after settler, then a diluent wash stage and finally mixed-media coalescing filters before advancing to Ni SX. In the nickel circuit, the aqueous solubility of Versatic 10 increases significantly with an increase in pH. Acidification of the raffinate is required to recover the Versatic from the aqueous phase. This is achieved in a single stage where raffinate is contacted with stripped organic phase at a reduced pH ensuring that soluble Versatic is recovered into the organic phase. Similarly to Cu SX and Co SX, the raffinate is then transferred to an after settler. Similarly to the raffinate, the rich electrolyte or loaded strip liquor needs to be treated to remove or recover entrained organic. The organic removal strategy for both copper rich electrolyte and cobalt loaded strip liquor is similar. In these two cases the metal-rich liquor is advanced to a series of mixed media coalescing filters. The nickel EW process is particularly sensitive to the presence of SX organic phase components. Before advancing to the EW circuit the Ni rich electrolyte undergoes two organic removal steps. The first is a system of mixed media coalescing filters where the bulk of the entrained organic phase is removed. The solution exiting this stage is passed through activated carbon absorption columns to further reduce the organic content. Crud that collects in the settlers is removed on a regular basis by suction with a peristaltic pump from just ahead of the overflow weir via a manifold collection system into the crud holding tank. The emulsion is fed into a tri-phase centrifuge that separates the solution phases (organic and aqueous) for recycle to the process and discharges a solid for disposal. Removal of degradation products and solids from the organic phase is achieved in a similar way in the copper, cobalt and nickel SX circuits. An agitated crud treatment tank is provided where a portion of the organic phase is contacted with diatomaceous earth in the case of cobalt and nickel, while clay for the copper circuit to help remove solids and degraded organic products. On each SX plant a remote emergency pond can receive overflows from all the sumps located in the bunded areas for the SX process and its associated tank farm. 262


257-272_Robles:text

2/13/09

9:24 AM

Page 263

Design issues related to TA®P process conditions The conditions of the leaching process that are dictated by the type of ore to be processed, the nature of the leach reagents, and the type and sequence of processes used to obtain the metal products, have an important effect in the design of the multiple SX circuits. In TA®P the most important design issues in the SX circuits that are related to the leach process conditions and operating parameters are scaling, the presence of a highly corrosive environment, and crud formation and control. Scaling The Tati concentrate is primarily a pyrrhotite (iron sulphide) type ore with some calcium, magnesium and manganese minerals. The dissolved iron is removed in a two-stage atmospheric precipitation process using limestone. Ground limestone slurry is introduced into the reaction tanks to increase the pH to precipitate ferric hydroxide. A small amount of nickel and cobalt will co-precipitate at the operating pH. The most common cause of scaling in SX is the saturation of CaSO4.2H2O due to the extraction and loading of calcium in the last few extraction stages and stripping of the calcium off the organic into the aqueous phase in the first extraction stages. Excessive scaling results in blocked pipes and reduced processing capacity. This in turn may lead to reduced production capacity, increased energy requirements (e.g. pumping and mixing) and increased downtime and maintenance cost. The approach for the design of the areas most likely to be affected by scaling has been to make adjustment in the process parameters and to take special care to design details that could decrease downtime and maintenance cost. Adjustment of process parameters The TA®P design has catered for the dilution of the Co PLS liquor with 5% filtered raw water to maintain the calcium concentration in the Co PLS below Ca saturation. Although this reduces the risk of scaling, the increased flow rate requires larger processing capacity with incremental capital cost. Solution temperature is also an important parameter that affects the scaling rate. The temperature in Co SX is maintained around the range of temperature that allows the maximum solubility of calcium sulphate. The change of pH of the solutions feeding Co SX and Ni SX, which is essentially dictated by the needs of the selective extraction of these metals, is controlled to minimize co-extraction of Ca in the back end of extraction and stripping in the front end of the extraction stages. Piping design Conductive fibre reinforced plastic (FRP) piping materials are selected for the TA®P SX plants to prevent static electricity build-up in the pipes. FRP piping also minimizes the adhesion of scale to the material surfaces, due to its smoother surface finish compared to stainless steel piping. The pipe spools were designed for easy removal and descaling. The plant layout was optimized to improve access to the pipes for periodical de-scaling. Design for easy cleaning and maintenance The Ni SX plant, which is particularly prone to scaling, is designed for regular descaling. One extraction mixer-settler is included as a spare unit with dedicated piping to allow descaling of any of the first two extraction mixer-settlers without operation interruption. The SX train SOLVENT EXTRACTION DESIGN CONSIDERATIONS FOR THE TATI ACTIVOX® PLANT

263

257-272_Robles:text

2/13/09

9:24 AM

Page 264

layout is optimized to improve access to mixer tanks, settler and pipes and facilitate the removal of internal parts and pipe spools. The mixer tanks have removable lids and the settlers’ roofs are designed with removable sections that include inspection hatches to monitor the accumulation of gypsum in the settler internals. Special nozzles were included in the mixer tanks to facilitate hydro-blasting of scale from tank internals. Use of synergists The use of synergists to improve the selectivity of certain metals is a relatively novel idea not widely implemented in the industry. A study conducted by du Preez et al.3 which evaluates the use of a synergist system for nickel-calcium separation on Tati Demonstration Plant solutions shows that selectivity for nickel over calcium can be improved by adding a synergist to the Ni SX organic. This performance versus the increased operating cost has to be weighed up to establish the overall added value of the synergist. Corrosive solutions The TA®P design relies on the addition of chlorides to improve copper extraction in the autoclave. This creates a corrosive leach solution requiring the use of higher grade metallic materials for certain specified applications. Based on the experience at the Tati Demonstration Plant, duplex stainless steel SAF 2205 was selected for metal components in contact with all process liquors that contain chlorides. For the SX circuits, where the working temperature is below 45°C, the TA®P has chosen to use FRP as an alternative to SAF 2205, as the maximum temperature recommended for the use of FRP is 60°C. Crud formation and control In SX the formation of crud is generally caused by a variety of substances present in the solutions being processed. This includes fines from ultra fine milling, entrained solids/precipitates from leaching; foreign material such as dust, insects; or silica in the TA®P filtered water. The strategy for crud control and treatment for the TA®P includes the following: • PLS clarification—pin bed clarifiers are used in the TA®P design to reduce the solids content of the Cu PLS and Co PLS. The uncontrolled separation of solids from the process liquor is usually a significant contributor to crud formation • Settling ponds prior to SX—settling ponds provide sufficient surface area and reduce flow rates so that ultra fine material has an opportunity to settle out prior to SX • Removal of degradation products—degradation products are produced by reactions on the organic phase caused by biological, chemical and environmental factors. These degradation products tend to be surface active and promote crud formation. Degradation products are continually removed in the Cu SX circuit by clay treatment and in Ni SX by the use of carbon columns • Isolation of SX mixer-settlers from external environment—roofing of the SX mixersettlers, which provides isolation from the external environment significantly reduced the collection of dust and insects • Water purification—water treatment and purification is used in the TA®P design to minimize silica concentration in the plant filtered water that is used as SX dilution water. • Phase continuity—the TA®P design utilizes preferably organic phase continuity to promote compaction of crud solids • pH control—the TA®P design optimizes pH control to prevent saturation of calcium in Ni SX and subsequent crud formation 264


257-272_Robles:text

2/13/09

9:24 AM

Page 265

• Crud treatment—the TA ® P design includes the installation of a permanent crud collection manifold system, which removes the crud that accumulates in the settlers, using a peristaltic pump for crud suction. The crud collection manifold consists of short sections of perforated pipes, located adjacent to the organic weir, that are connected to a header. The collected crud is treated by means of a tri-phase decanter centrifuge for organic recovery. Each SX circuit has its own crud treatment system to prevent organic cross-contamination. pH control The TA®P design requires very tight pH control in Co SX and Ni SX to prevent co-extraction of metals. Impurities are scrubbed of the loaded organic using acidified scrub liquor in the scrubbed organic phase. The most important challenge for a pH control system is the reliability of the pH measurement in the solution feeding the settler. Such a measurement could be hampered by organic coating or scaling of the pH electrode probe. Hence, in theTA®P design the electrode is located in a dedicated pH pot that receives aqueous solution from the feed end of the settler, through a pipe that connects the lower part of the settler’s sidewall with the pot. This location allows pH measurement in clean aqueous solution and facilitates access for cleaning and calibration of the electrode. The TA®P design allowed for a cascading control system where the flow rate of neutralizing or acidification reagent is the primary control measure and the pH of the aqueous solution is the secondary measure. The reagent flow set point is a calculated ratio of the main incoming flow (PLS, strip liquor or loaded organic) with a multiplier that is based on the output provided by the pH controller. The control logic minimizes reagent flowrate fluctuations and reliability on accurate pH measurements. The control logic allows for timely identification of pH deviations. Low and high alarms alert the operator to pH measurement faults or dosing ratios outside the set ranges. The pH electrodes in the first three extraction stages of Ni SX were duplicated, with the additional electrodes located in the aqueous launders of these settlers to provide back-up for the pH control loops. Design issues related to multiple SX trains The multiple SX trains in the TA®P design present certain challenges in terms of optimizing plant layout, contactor design and minimizing possible organic cross-contamination. Organic cross-contamination occurs when organic from one SX circuit is carried over to the next SX circuit causing process upsets, reduction of efficiency, metal loss, and off spec product. Plant layout One of the major challenges that the SX design for the TA®P faced was to accommodate the various SX circuits in an optimal manner. The following had to be taken into account: minimization of the SX train footprint; location of storage ponds and/or connecting pipelines with upstream and downstream process facilities; and safe separation distance between SX circuits to minimize fire risks. Norilsk selected the side-feed design of MMS for the SX settlers. This allows a smaller footprint for the SX train of mixer-settlers and better access for operational control, cleaning and maintenance. All operational and maintenance work can be carried out on one side of the settler. SOLVENT EXTRACTION DESIGN CONSIDERATIONS FOR THE TATI ACTIVOX® PLANT

265

257-272_Robles:text

2/13/09

9:24 AM

Page 266

Each SX plant is laid out with all equipment containing large quantities of organic solution grouped together in the SX area (SX train, organic storage tanks, after-settler(s), crud collection tank) and associated equipment including organic recovery and crud treatment systems are grouped into the SX tank farm area. A safe separation distance is allowed between these two areas to ensure asset protection in the case of a fire. For the same reason, the three SX trains are separated by a safe distance. These distances were defined after a fire plume analysis was performed for each SX plant. Optimization of flow patterns inside the SX settler The optimal design of the main phase separation equipment in the SX plants is critically important since excessive turbulence and insufficient coalescence and phase separation result in increased entrainment that may lead to cross-contamination. Hatch used computational fluid dynamics (CFD) models for settler optimisation in theTA®P. CFD is an important design tool that allows the designer to visualize and test innovative design concepts quickly and economically4–7. Single and multi-phase CFD models were used for theTA®P to predict the flow behaviour and flow distribution in the entrance and early flow development regions of the side-feed settler vessels. The crossover launder that connects the secondary mixer tank with the settler and its guide vanes were modified to provide an optimum low-speed transition into the settler vessel, which is critical to its overall performance. Multiphase modelling has also provided significant guidance in the design of the TA®P mixer interconnecting pipe, picket fences and collection launders. In reality all settlers deviate from an ideal ‘plug-flow’ distribution and results from visual inspection or physical modelling can be difficult to interpret and are therefore of limited use in design. Computation of the integrated volumetric-flow distributions across the width of the settler, as shown in Figure 2, has been used to evaluate and rank a number of different designs. The settler internals were modified to achieve a uniformly distributed flow across the width of the settler.

Figure 2. Post processing is used to evaluate the mass flow across each quarter of the width

266


257-272_Robles:text

2/13/09

9:24 AM

Page 267

Table II Distribution of mass flow across the width of the settler for the original design and after using CFD analysis Integrated mass flow through lateral quarters of the settler cross-section (kg/s) Inner

Inner-mid

Outer-mid

Outer

106 91

46 92

66 90

105 87

Original design Optimized design

The computed mass flow rates through each of the lateral quarters of the settler for the original design and after using CFD analysis are listed in Table II. The optimized design provided a significantly more uniform flow distribution with a lower overall pressure drop. Organic removal equipment The TA®P design defined various strategies and equipment for the removal of organic phase from aqueous streams. These include physical and chemical processes depending on the state of the organic for recovery (dissolved and/or entrained). Organic recovery processes fall into four process categories: • Primary—reducing organic entrainment from >200 ppm or more down to <50 ppm and recovering the organic phase • Secondary—reducing organic entrainment from 100 ppm to <2 ppm and recovering the organic phase • Dissolved organic recovery—reducing the soluble organic content from >200 ppm to <10 ppm • Tertiary—reducing the total organic content, both entrained and soluble, from 10 ppm to <1 ppm. Primary organic removal equipment After-settlers are utilized in the three SX circuits as the preferred primary organic removal equipment in the TA®P design. Essentially they provide additional area for phase separation and are the first barrier to stop organic entrainment peaks should a major upset happen in the mixer-settlers. The organic collected in the after-settlers is periodically removed and returned to the extraction stages via the corresponding crud treatment system. The only exception to this is in the Co loaded strip liquor stream, where this aqueous stream is passed though a tank with coalescing media before downstream processing. The coalescing media generate large organic droplets from small entrained droplets by creating random collision opportunities. Larger droplets are separated more readily. The coalesced organic is periodically removed from the tank and returned to the extraction stages via the Co SX crud treatment system. Non-mechanically agitated flotation cells, which are also widely used as primary organic removal equipment, were not included in the design because of the induction of high volumes of air (at ratios of about 0.8:1 v/v) may result in evaporative loss of significant proportions of the valuable organic phase or cause other process problems including oxidation. Secondary organic removal equipment As a secondary organic removal equipment downstream of the after-settlers, the TA®P design allowed for pressure filtration using dual-media filters, with a combination of anthracite and garnet media, to reduce entrained organic level to <2 ppm. Organic removed from the filters is returned to the extraction stages via the corresponding crud treatment system. SOLVENT EXTRACTION DESIGN CONSIDERATIONS FOR THE TATI ACTIVOX® PLANT

267

257-272_Robles:text

2/13/09

9:24 AM

Page 268

The only two product streams that do not have downstream processing by pressure filtration are the raffinate streams in Cu SX and Ni SX. As they are fed into storage ponds, floating organic is removed from the ponds’ surfaces using rope mop-type organic skimmers. Organic removed from the ponds is also returned to the extraction stages via the corresponding crud treatment system. Soluble organic recovery Highly soluble extractants need to be recovered through physicochemical and chemical means as opposed to simple physical removal. The TA®P design has two methods for bulk removal of soluble organic from the aqueous phase, depending on the nature of the extractant. One is for removal of phosphinic acid-based extractant solubilized in Co SX raffinate that is simply washed with diluent in a single stage. On the other hand, carboxylic acid-based extractant solubilized in Ni SX raffinate has to be recovered through pH adjustment with an acid solution in a single stage to reduce organic solubility in the aqueous phase8. Tertiary organic removal equipment In the TA®P design, tertiary recovery equipment is employed to ensure the achievement of <1 ppm organic in the rich electrolyte going to Ni EW. The equipment selected is a package of columns filled with granular activated carbon. The carbon columns package consists of three columns, which are connected as a lead-lag system. At any given time two of the columns are in service and one is out of service. Feed electrolyte flows into the lead column, from there to the lag column, and then leaves the system. When the granular activated carbon media in the lead column is saturated with organics, that column is taken out of service. The lag column then takes lead status and the previously out-of service column takes lag status. The process is sequenced so that over a given period each of the columns will have been in lead status, followed by being out-of service, followed by being in lag status. Mitigation of fire risks Recent fires in SX plants have highlighted the importance of safety in design in order to eliminate or mitigate the risk of fire and subsequent damage to facilities, injuries and fatalities. Although SX plants normally operate below the flashpoint of the organic solvent rendering the environment inherently safe, the creation of aerosols or static build-up create conditions suitable for ignition by an electrostatic charge9. The TA®P SX plants are designed to reduce the risk of fire initiation with the moderate operating temperature in SX, the minimization of aerosols formation in all the feeds to tanks and launders and the prevention of static charge build-up by the use of conductive FRP for all the piping systems and tanks and appropriate earthing devices along the piping systems and on tanks. These plants are also designed under an integrated fire management strategy that allows early and reliable fire detection, effective fire suppression, and fire containment and isolation. Fire detection and alarm The fire detection system is designed to be reliable, robust and fast. This system has detection in multiple points to prevent false alarms and subsequent damage caused by the unnecessary release of water and or foam into the processing units. The fire detectors in the TA®P SX plants, which are of the infrared flame type, are placed under the removable settler roofs, across groups of mixer-settlers and storage tanks and along pipe routes. 268


257-272_Robles:text

2/13/09

9:24 AM

Page 269

The purpose of the alarm system is to provide an audible warning that enables safe and timely evacuation of the plant. Manual call points and evacuation speakers are positioned around the SX areas. Fire alarm panels located in the field centralize monitoring of all the devices, which form part of the fire detection and alarm system of the area and are connected to the main fire alarm panel in the central control room. Fire suppression system Two options of fire suppression were analysed for the TA®P design: a water deluge system with dumping of the settler content during fire and a fixed foam-water deluge system. The two options were assessed based on minimum statutory requirements, international recognized codes and standards, insurance requirements, best industry practice, occupant life safety, asset protection, potential business interruption, site operational procedures, and capital cost. The initial design for the fire suppression system was a water deluge with a dump system. This system was designed to prevent escalation of fire by cooling the equipment and structures adjacent to a fire to protect them from heat radiation. The system included the automatic water deluge on tank externals, structures, pipe racks and bunds and manual water hydrants along plant perimeter. The dump system was designed to allow the dumping of the contents of the mixer-settler on fire by means of actuated valves, which are activated by the fire detection and alarm systems. Sumps within the SX areas were also designed in such a way to drain the area fast and effectively to the remote fire dump pond. This water deluge system was found to be insufficient after evaluating against the design requirements. Four main drawbacks were identified in the system. Firstly, the need to extinguish a fire in the SX mixer settler as quickly as possible was not met by this design since deluge water does not offer the extinguishing properties that foam does. Secondly, the location of the TA®P is fairly remote and the availability of trained fire crews are limited. Thirdly, the need to minimize loss of plant production leading to minimizing down time and minimizing damage to equipment, highlighted the need for a more robust fire suppression system with lower water demand. The final drawback was the reliability of the dumping system. Regular testing of the dump valves was required to maintain the integrity of the system. False alarms would have led to the dumping of the SX inventory. The water deluge system was then replaced by a fixed automatic foam-water deluge system. The fixed foam system consists of permanently installed piping and deluge/foam sprays tied into a fire water supply and dedicated foam concentrate supply and proportioning equipment. The deluge system provides external protection to vessels, tanks and pumps, located in each SX plant area and in the diluent storage area. Manual yard hydrants/monitors with foam making equipment are placed along the plant perimeter. The spray foam/deluge system is activated with a pilot operated water fusible plugs network. Special dedicated foam lines are provided for the internal organic tank/vessel protection. These lines are manually operated. The SX plants are divided into three to five areas with separately activated deluge foam systems. All deluge-foam nozzles are located outside the mixer settler tanks to prevent process contamination. The foam recommended for use on the TA®P design is a low viscosity AFFF/A4P 3%. This conventional foam is suitable for extinguishing hydrocarbon liquid and polar fire. Fire containment and isolation A fire plume analysis was used as basis for the layout of the TA®P SX areas. The analysis takes into account the area of the fire, the volume and type of fuel that could burn, climatic and wind conditions, and the presence of other assets and public access. From the fire plume SOLVENT EXTRACTION DESIGN CONSIDERATIONS FOR THE TATI ACTIVOX® PLANT

269

257-272_Robles:text

2/13/09

9:24 AM

Page 270

analysis, the physical layout of the assets within each SX plant is arranged in such a way to reduce impact of fire escalation and increase ease of isolation and containment. This analysis also provides information to arrange each SX plant in relation to the others. Each SX area containing all mixer-settlers, after-settlers and crud collection and organic tanks, is separated into individual fire suppression zones. These zones facilitate fire containment within the SX area. On each one of these zones there is a bund with a sump that allows fluids to overflow into a remote emergency pond to prevent foam/water or process liquor spillage spreading to other zones during a fire event. Fire retardant, conductive FRP piping were used inside the SX plants to prevent the spreading of a SX fire through the piping. Conclusions The three SX circuits of the TA ® P were designed with careful consideration to the characteristics of the leach solution, the operation of multiple SX circuits in sequential mode, and the mitigation of fire risks. The TA ®P design utilizes several process steps (clarifiers, settling ponds, filters) to minimize solids entrainment into the PLS. The design allows for a full standby mixer settler, PLS dilution and tight pH control to minimize scaling during operation. Primary, secondary and tertiary organic removal steps are utilized to prevent organic cross contamination and rich electrolyte contamination. A foam-water deluge system was selected for all three TA®P SX plants with emergency ponds to collect overflow form the SX bunded areas. Acknowledgements The authors would like to thank Norilsk Nickel Africa (Pty) Ltd. and Hatch Africa (Pty) Ltd. for their permission and support in writing this paper. We also like to acknowledge the consultancy work provided by Graeme Miller and the good work of the multidiscipline SX design team that worked out of the Hatch’s Brisbane office. References 1. PALMER, C.M. and JOHNSON, G.D. The Activox® Process: Growing significance in the nickel industry. Journal of Minerals, Metals and Materials Society, vol. 57, no. 7, July 2005, pp. 40–47. 2. MILLER, G. Design of Mixer Settlers to Maximise Performance, Proceedings from ALTA 2006 Copper Conference, 10th Copper Event, ALTA Metallurgical Services, Melbourne, Australia. 3. DU PREEZ, R., NEL, G., KOTZE, M., DONEGAN, S., and MASSINA, H. Solvent Extraction test work to evaluate a Versatic 10/NicksynTM Synergistic system for Nickel Calcium separation, Proceedings of The Fourth South African Base Metals Conference, Africa’s base metals resurgence. Symposium Series S47, South African Institute of Mining and Metallurgy, 2007, pp. 193–210. 4. GUNNEWIEK, L.H., OSHINOWO, L., PLIKAS, T., and HAYWOOD, R. Using CFD in the Design and Scale-up of Hydrometallurgical Processes, 3rd International Conference on CFD in the Minerals and Process Industries, CSIRO, Melbourne, Australia, 2003 270


257-272_Robles:text

2/13/09

9:24 AM

Page 271

5. SCHWARZ, M.P., LANE, G., NIKOLOV, J., YANG, W., BUJALSKI, J.M., SOLNORDAL, C.B., and HOUCHIN, M. Development of a Design Capability for SX Contactors Using Multi-Phase CFD and Experimental Modelling, Proceedings ALTA 2007 Copper Conference, 11th Copper Event, ALTA Metallurgical Services, Melbourne, Australia. 6. ROBLES, E., CRONJE, I., POULTER, S., and HAYWOOD, R. Aspects of the Design of SX Plants for the Modern Hydrometallurgical Refinery, Proceedings of the International Solvent Extraction Conference ISEC 2008, Tucson, Arizona. 7. GIRALICO, M., GIGAS, B., and PRESTON, M. Advanced Mixer Settler Designs that will Optimize Tomorrow’s Large Flow Production Requirements, Proceedings Hydrometallurgy 2003—Fifth International Conference in Honour of Professor Ian Ritchie—vol. 1, Leaching and Solution Purification. TMS, 2003. 8. BACON, G. and MIHAYLOV, I. Solvent Extraction as an Enabling Technology in the Nickel Industry, Proceedings of the International Solvent Extraction Conference ISEC 2002, K.C. Sole, P.M. Cole, J.S. Preston, D.J. Robinson, (eds.), South African Institute of Mining and Metallurgy, Johannesburg. pp 853–863. 9. HAIG, P., KOENEN, T., and MAXWELL, J. Electrostatic Hazards in Solvent Extraction Plants, ALTA SX/EW World Summit, Perth, ALTA Metallurgical Services, Melbourne, Australia, 2003

Ilne Cronje Process Engineer, Hatch, South Africa Recent experience in the design of hydrometallurgical plants. Involvement in the area of Solvent Extraction consisted of flowsheet development, Process and Instrumentation Diagrams, equipment sizing, completion of Mechanical Data Sheets and hydraulic calculations for sizing of Mixer Settler units, developing control philosophies, commissioning and operating manuals. More recently involved in a project management role as a Study Manager on a Smelter Pre-Feasiblity Study and a Waste Heat Utilization Concept Study. Involvement included setting up of the project, project controls, management of the engineering disciplines and assisting with Process Deliverables.


271

257-272_Robles:text

2/13/09

9:24 AM

Page 272


272

273-282_Tinkler:text

2/13/09

9:25 AM

Page 273

TINKLER, O., SHIELS, D., and SODERSTROM, M. The ACORGA® OPT series: comparative studies against aldoxime: ketoxime reagents. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

The ACORGA® OPT series: comparative studies against aldoxime: ketoxime reagents O. TINKLER*, D. SHIELS†, and M. SODERSTROM‡ *Cytec Industries Inc., South Africa Australia Holdings Pty Ltd, Melbourne, Australia ‡Cytec Industries Inc., Phoenix, United States of America †Cytec

In 2006, Cytec Industries Inc. announced the development of a new series of solvent extraction reagents that optimize copper transfer through the use of formulations containing aldoxime and ketoxime as well as selected modifiers. The patented technology results in increased copper transfer due to the interaction between the modifier and oximes within the blend. These new formulations also show a significant improvement in copper iron selectivity, which can result in lower bleed and copper reprocessing costs. The name given to the new series is ‘OPT’ which falls under the Cytec Industries ACORGA® brand of copper solvent extractants. These new copper extractant formulations have been successfully commercialized and show a significant metallurgical advantage over the unmodified aldoxime: ketoxime formulations that are in common use in Zambia. This paper discusses results from extended side-by-side testing of ACORGA OPT5510 against an aldoxime:ketoxime blend, at two agitated leach operations in Zambia and the full conversion to ACORGA OPT5510 from an aldoxime:ketoxime blend, at a heapleach operation in Arizona.

Introduction In 2006 Cytec Industries inc. announced the development of a new series of solvent extraction reagents that optimize copper transfer through the use of formulations containing aldoxime and ketoxime as well as selected modifiers. The incorporation of selected modifiers results in increased copper transfer, due to the interaction between the modifier and oximes within the blend. These new formulations also show a significant improvement in copper iron selectivity, which can result in lower bleed and copper reprocessing costs. The name given to the new series is ‘OPT’ which falls under the ACORGA® brand of extractants. The first part of this paper discusses results from extended side-by-side testing of ACORGA OPT5510 against LIX 984N* (supplied as LIX984N-C*), an aldoxime:ketoxime blend, at Kansanshi Mining Plc, *LIX is a registered trademark of Cognis Corporation

THE ACORGA® OPT SERIES: COMPARATIVE STUDIES

273


2/13/09

9:25 AM

Page 274

Solwezi, Zambia. The objective of the trial was to compare copper transfer, iron transfer, phase disengagement time, rate of crud formation under the same operating conditions—i.e. equal reagent concentration, equal flow rates, equal settler rating, equal bed depths and equal mixer speeds. The second part of the paper reviews results of a high concentration application as well as the conversion from LIX 984N to ACORGA OPT5510 at a heap-leach operation in Arizona. Study 1: Kansanshi Mining Plc, Solwezi, Zambia Experimental A side-by-side pilot plant comparison of ACORGA OPT5510 and LIX 984N was conducted on site at Kansanshi Mining Plc. during September 2007 (total run-time 30 days). Kansanshi Mining Plc. operates an agitated leach-SX-EW circuit, which produces 120 000 tpa of copper cathode. The pilot plant circuits were configured with two extract stages and one strip stage, with separate loaded organic tanks but common PLS and spent electrolyte feeds. The physical and metallurgical behaviour of the two reagents was evaluated under identical operating conditions (i.e. equal reagent concentration, equal flow rates, equal settler rating, equal bed depths and equal mixer speeds). Operating parameters are shown in Table IV. The evaluation was conducted at three different O/A ratios: • Low recovery conditions Extract O/A ratio of 1.2 • Medium recovery conditions Extract O/A ratio of 1.5 • High recovery conditions Extract O/A ratio of 1.8 PLS and spent electrolyte ranges are presented in Table I. Temperature profiles were taken during the trial for record purposes. Table II shows the average daytime mixerbox temperatures for 28 and 30 September. The primary consequence of the warm operating temperature was the high level of diluent evaporation, which resulted in a steady increase in max. loads (Table III) through the course of the trial. Table I PLS min. and max. range during the trial PLS

Max Min Average

Spent electrolyte

Fe3+

Mn

Si

H2SO4

Cu

Fe

Fe2+

(gpl)

(gpl)

(gpl)

(gpl)

(ppm)

(ppm)

(gpl)

7.96 5.27 6.77

2.63 0.59 1.16

0.16 0.33 0.23

2.47 0.26 0.93

614 408 522

360*

8.8 3.4 5.7

pH

1.87 1.18 1.51

TSS

Cu

H2SO4

Fe

Fe3+

Mn

(ppm)

(gpl)

(gpl)

(gpl)

(gpl)

(ppm)

180 36 97

40.8 29.2 35.3

215 174 193

2.03 0.6 1.28

1.33 0.12 0.77

140 190 160

Table II Mixer box temperature Data

28/9/2007 30/2/2007

274

ACORGA OPT 5510

LIX 984N

E1

E2

S1

E1

E2

S1

35.2 39.1

37.2 40

39.8 41.5

36.5 39

37.3 40.2

39.7 41.2



2/13/09

9:25 AM

Page 275

Table III Organic maximum loads and reagent concentrations ACORGA OPT 5510 Date 6 September 10 September 13 September 15 September 26 September 1 October Average

LIX 984N

ML - Cu (gpl)

Vol %

ML - Cu (gpl)

Vol %

11.9 11.0 11.8 12.9 13.8 14.1 12.6

21.8 20.2 21.6 23.7 25.3 25.9 23.1

11.6 10.3 11.2 13.6 13.4 14.3 12.4

22.3 19.8 21.5 26.1 25.8 27.5 23.8

Table IV Pilot-plant parameters Working volume of mixer Mixer retention time

O/A 1.2 extract O/A 1.2 strip O/A 1.5 extract O/A 1.5 strip O/A 1.8 extract O/A 1.8 strip

Impellor tip speed

0.029

m3

2.64 3.95 2.32 3.16 2.33 2.96

min min min min min min

250–300

m/min

Settler rating

O/A 1.2 extract O/A 1.2 strip O/A 1.5 extract O/A 1.5 strip O/A 1.8 extract O/A 1.8 strip

3.0 2.0 3.4 2.5 3.4 2.7

m3/m2/hr m3/m2/hr m3/m2/hr m3/m2/hr m3/m2/hr m3/m2/hr

Linear velocity

O/A 1.2 extract and strip O/A 1.5 extract and strip O/A 1.8 extract and strip

0.28 0.35 0.37

cm/s cm/s cm/s

Despite the relatively high operating temperature in the pilot plant, neither organic showed any signs of hydrolytic degradation (aldehyde and ketone levels were <2 g/l). Throughout the period of the trial, the physical behaviour of the two circuits was closely monitored and data carefully recorded. Mixer continuity, phase break times and crud build-up were measured and recorded several times per shift while organic entrainment was monitored visually throughout. PLS and spent electrolyte were bled-off from the feed lines to SX2-HG. For the low, medium and high recovery tests (Ext O/A 1.2, 1.5 and 1.8) the PLS flow was 0.3; 0.3 and 0.267m 3/h, respectively; the organic flow was 0.36, 0.45 and 0.48 m3/h and the spent electrolyte 0.08, 0.1 and 0.107m3/h. The resulting operating parameters are recorded in Table IV. The strip O/A was maintained at 4.5:1 throughout the trial. With the exception of the last three days, all mixers were operated organic continuous and no aqueous recycle was used in extract or strip at any time during the trial. THE ACORGA® OPT SERIES: COMPARATIVE STUDIES

275


2/13/09

9:25 AM

Page 276

Results Metallurgical O/A Ratios and Copper Recovery Comparison The metallurgical O/A ratio is the change in analysed copper in the aqueous phases divided by the change in analysed copper in the organic phases. It is an important value as it verifies the accuracy of the aqueous and organic copper analysis. Figure 1 shows the extract and strip metallurgical O/A ratios for the three target extract O/A ratios. The average copper recoveries for the two reagents at the three different O/A ratios are presented in Figure 2. Although ACORGA OPT5510 outperformed LIX 984N under all three conditions, the difference was greatest under the ‘low recovery’ conditions where it was over 10%.

Figure 1. Metallurgical O/A ratios

Figure 2. Average copper recoveries

276



2/13/09

9:25 AM

Page 277

Selectivity of copper over iron (III) In addition to significantly improving copper transfer, the addition of ester-modifier to the aldoxime-ketoxime blend has a beneficial impact on selectivity of copper over iron. Maximizing Cu/Fe selectivity will become increasingly important at Kansanshi as the availability of the pressure leach circuit increases, which will result in a steady increase in the total Fe concentration in the PLS. Figure 3 shows the average E1 copper loading vs. the average E1 iron loading for two circuits. As expected, the iron loading increases sharply as the copper loading decreases. The difference between the two reagents, under the high recovery condition, is around 2 ppm. Although this might not seem a lot, by reducing the chemical iron transfer into the electrolyte by just 1 ppm per pass, the annual Co savings would be in the range of $100 000–$150 000 under current conditions. Rate of crud formation The rate of crud build-up both at the interface and on the bottom of the settlers was very similar for the two reagents (Figures 4 and 5).

Figure 3. Comparison of selectivity of copper over iron (III)

Figure 4. Comparison of crud build-up in extract stages


277


2/13/09

9:25 AM

Page 278

Figure 5. Comparison of crud build-up in strip stages

Figure 6. E 1 - phase break times

Phase disengagement times Phase break times were monitored on a bi-hourly basis. The daily averages for the 30 day trial are shown graphically in Figures 6–8. The trend line in Figure 6 shows a gradual trend upwards, relative to the fresh reagents though from about the 12th of September onwards the phase break times appear to level out for both reagents. The range of 90–150 seconds is normal for this reagent concentration. The phase break times in the E2 stage (Figure 7) stayed fairly constant through the trial period. The S1 phase break decreased slightly as the trial progressed (Figure 8). This is not particularly significant as there was no aqueous recycle and the O/A was 4.5:1. Organic in aqueous entrainment was monitored visually throughout the trial. No difference was seen in behaviour between the two circuits in any of the stages. Visual indications were that there was no difference in the aqueous or in the organic entrainment either. 278



2/13/09

9:25 AM

Page 279

Figure 7. E2 Phase break times

Figure 8. S1 Phase break times

Study 2: Silver Bell Mining LLC Conversion from LIX 984N to OPT5510 In April 2007, Silver Bell Mining LLC decided to convert from LIX 984N, a blend of aldoxime and ketoxime to ACORGA OPT5510 extractant. The initial approach was to let the composition change over time as the new formulation was added during normal monthly make-up additions. However, after approximately seven months of make-up, only approximately 20–25% of the plant inventory was OPT5510. Although some of the expected improvements were seen, the full benefits were not being realized because the plant organic was only partly converted. Additional testwork was completed which highlighted the benefits of carrying-out a ‘one-step’ conversion using a booster formulation. THE ACORGA® OPT SERIES: COMPARATIVE STUDIES

279


2/13/09

9:25 AM

Page 280

In December 2007 a programme was initiated to optimize the full plant organic inventory at one time by the addition of a carefully calculated volume of ACORGA OPT Booster 1. Detailed physical and metallurgical plant audits were conducted before and after the addition of ACORGA OPT Booster 1. Approximately 1 200 gallons (4.5 m3) was added into the loaded organic weir from intermediate bulk containers in less than one hour. Even at this addition rate, there was no shock to the system or noticeable change in physical performance. Post-conversion plant performance After six months of operation after the full conversion, the plant operators and supervisors report continued strong performance. Phase separation is good and entrainments are low. There has been no change in the rate of crud formation. Cu recovery performance has remained high and Cu:Fe selectivity has improved. In addition, copper recovery remains higher during times when the PLS Cu grade spikes (converted formulation gives better Cu recovery). Study 3: high Cu concentration PLS An isotherm-based metallurgical comparison of ACORGA OPT5540 against a 2.6:1 blend of aldoxime and ketoxime (LIX 973N) was carried out using 287 g of each extractant made up to 1 000 ml with ORFOM SX-7**. The maximum loads, uptake factors and concentrations are reported in Table V. Extraction and strip isotherms were prepared at ambient temperature by mixing the two organic phases with synthetic pregnant leach solutions (PLS) and spent electrolyte (compositions in Table VI) at different O/A ratios. Comparative models were generated using Cytec’s MINCHEM® SX modeling software. In all cases extract and strip stage efficiencies of 95% were used. The circuit modelled contains three primary extract stages (O/A 3.4), two secondary extract stages (O/A 1) and two strip stages (O/A 1.4), with the organic split 86/14 between the primary and secondary extracts. Results are reported in Table VII.

Table V Organic phase composition Reagent

Mass of reagent (g) made up to 1 000 ml with diluent

Max load (g/l)

Uptake (g/l/vol %)

Conc. (vol %)

287.1 287.1

17.12 17.57

0.545 0.54

31.4 32.5

ACORGA OPT5540 LIX 973N

Table VI PLS and spent electrolyte composition

Primary PLS Secondary PLS Spent electrolyte

Cu (g/l)

pH

Sulphuric acid (g/l)

Total Fe (g/l)

Fe (II) (g/l)

Fe (III) (g/l)

43.3 7.4 34.0

0.80 1.55 -

176.4

45.88 11.7 -

23.6 6.5 -

22.27 5.2 -

**ORFOM is a registered trademark of Chevron Phillips

280



2/13/09

9:25 AM

Page 281

Table VII Modelled recoveries Reagent

ACORGA OPT5540 LIX 973N

HG raff (g/l)

LG raff (g/l)

HG Cu Rec. (%)

13.1 14.9

0.4 0.6

69.5 65.4

LG Cu rec. Org. Cu loading (%) (g/l) 94.4 92.1

93.8 99.1

Strip efficiency Net Cu t/fer (%) (g/lvol %) 68.7 55.2

0.32 0.29

The OPT formulation performs better than the unmodified aldoxime-ketoxime blend because of the addition of Cytec’s ester-modifier, which dramatically improves stripability of the C9-aldoxime component. The result is a shift in the strip isotherm to the left, which results in the generation of a lower barren organic copper concentration and hence better overall extraction. Conclusions The copper transfer and copper:iron selectivity improvements achieved by optimal modification of aldoxime:ketoxime blends has been demonstrated over a wide range of conditions. Although the metallurgical advantages will vary dependent on specific circuit conditions within individual plants, on-site pilot testing can quickly determine the optimum ACORGA OPT formulation and quantify production increases and/or cost savings that can be expected. Conversion of aldoxime:ketoxime circuits to ACORGA OPT can be done over time with normal make-up or in a single step with the use of an ACORGA OPT Booster formulation. References 1. SODERSTROM, M. New Reagent Developments in Cu SX, ALTA 2006 Copper Session, Perth, Australia. 2006. 2. PHILLIPS, T., MANG, W., SODERSTROM, M., and CRAMER, K. Optimizing Metallurgical Performance at Silver Bell mining LLC, Hydro 2008, Phoenix, AZ.

Owen Tinkler Regional Manager, Cytec Industries 15 years experience in copper & Ni/Co solvent extraction 3 years experience in gold processing Owen graduated from the University of the Witatersrand with a BSc(Hons) in Applied Chemistry in 1989. After spending 10 years in Phoenix, AZ in a variety of positions for Cytec Industries, he moved back to South Africa in 2007 and currently holds the position of Regional Manager for Cytec's Metal Extraction Group.


281


282

2/13/09

9:25 AM

Page 282


283-294_Crane:text

2/13/09

10:52 AM

Page 283

CRANE, P., URBANI, M., DUDLEY, K., HORNER, A., and VIRNIG, M. Solvent extraction (SX) reagent selection for high temperature, acid, chloride and Cu PLS at Port Pirie and its impact on electrowinning (EW). Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Solvent extraction (SX) reagent selection for high temperature, acid, chloride and Cu PLS at Port Pirie and its impact on electrowinning (EW) P. CRANE*, M. URBANI*, K. DUDLEY†, A. HORNER‡, and M. VIRNIG§ *Cognis Australia Pty Ltd., Australia †KD Hydromet Consulting Pty, Ltd., Australia ‡Nyrstar Port Pirie Smelter, Australia §Cognis Corporation, Tucson, USA

Nyrstar’s Port Pirie Copper Plant produces copper cathode from the lead smelting process copper/lead matte by-product. More than 98% Cu is recovered from the matte by a unique mixed chloride-sulphate leach technology (formerly known as the BHAS process) followed by conventional copper solvent extraction and electrowinning processes. The highly effective agitated chloride-sulphate leach process presents several challenges for the operation of the downstream Cu SX-EW plant, namely it: • Produces a high temperature PLS with high concentrations of acid, chloride and copper • Requires effective solids liquid separation and soluble silica control • Requires high O/A ratios and high concentrations of extractant • Requires careful monitoring of organic health and regular, effective organic treatment. The strategy that was undertaken to examine the historical performance of the Port Pirie Cu SX plant and the development of a planned approach on implementing effective measures to combat the high chloride transfer to the EW process is outlined in this paper. This involved determining the plant organic constituents and understanding the characteristics that presented the underlying causes to many of the Cu SX-EW plant problems being encountered. Operational data collected during the Cu SX plant reagent transition phase is presented. The data illustrate the changes in organic chemistry and organic health that ultimately produced a very different EW electrolyte to that previously experienced in this operation. A key result of this reagent change and attention to detail in Cu SX was the significant reduction of the chloride concentration in the Port Pirie EW plant electrolyte. From historical levels of >200 ppm, which required the use of titanium cathodes, electrolyte chloride levels were reduced to less than 30 ppm. SOLVENT EXTRACTION (SX) REAGENT SELECTION

283

283-294_Crane:text

2/13/09

10:52 AM

Page 284

Introduction The Port Pirie smelter is located on the eastern side of the Spencer Gulf in South Australia, a little over 200 kilometres north of Adelaide. It has a rich history, being the nearest seaport to Broken Hill, where the rich ore field containing lead and zinc was discovered in 1883. The smelter received its first load of ore in 1889 and in due course Port Pirie evolved to be the centre of a lead smelting and refining industry. Over 100 years later what started out as a minor smelter grew to become one of the world’s largest. Originally built in 1889 with a capacity of 80 000 tonnes of lead per year the smelter was progressively expanded in the 1950s and 1960s. The lead refinery itself was largely rebuilt in 1998 and has a demonstrated production capacity of approximately 270 000 tonnes per year of lead metal and alloys, although the lead blast furnace and sinter plant capacity limits annual production to approximately 235 000 tonnes. The precious metals refinery in which gold and refined silver are recovered after extraction in the lead refinery was also largely rebuilt in 1998. The current zinc production facilities were commissioned in 1967 and have a capacity of 45 000 tonnes per year. Copper production facilities were commissioned in 1984 and have a capacity of approximately 4 500 tonnes per year of copper cathode. The copper units enter the site through numerous concentrates and are largely ‘free’. This makes the manufacture of copper at the smelter very economical at this small scale. In addition to lead, zinc and copper metal, Port Pirie also produces approximately 80 000 tonnes of sulphuric acid, approximately 11.5 million troy ounces of refined silver and approximately 16 000 troy ounces of gold. Port Pirie Cu plant process description The feed material for the Cu plant is primarily a copper-lead matte by-product from the lead blast furnace typically containing about 35% copper, 38% lead and 12% sulphur with the balance being minor elements such as arsenic, zinc, tin, silicon, antimony, iron, cobalt, cadmium and nickel. In the 1970s the BHAS Research Department developed a process for leaching the copper matte using an acidic chloride/sulphate solution and oxygen. The successful laboratory work led to the construction and operation of a pilot plant during 1980 and the BHAS Board approved a commercial plant in September 1982. This plant was designed to have a nominal capacity of 4 000 tonnes EW Cu cathode per annum and was commissioned in 1984. The copper plant consists of five sections: grinding, leaching/solids liquid separation, SX, EW and residue neutralization. A brief description of each section follows.1,4 Grinding The grinding mill treats over 12 000 tonnes of crushed matte per year in a conventional ball mill charged with 90 mm diameter forged steel balls, operating in closed circuit with a spiral classifier. Thickened matte slurry, at 100%–75 micron, is pumped to a 24-hour capacity surge tank, where it is continuously agitated to prevent settling. Leaching chemistry2,3 The BHAS leaching process centres on an oxidative chloride/sulphate leach of the matte to extract copper into solution, leaving a solid lead and silver rich residue. Leaching is conducted at atmospheric pressure at 85°C. The overall chemistry of the process is summarized by the Equation: 284


283-294_Crane:text

2/13/09

10:52 AM

Page 285

[1] The copper dissolution reaction can be described in two stages: [2] [3] Equation [1] implies the dependency of the reaction on the availability of acid and oxygen. However, what is not shown is the role of the chloride in the system. The conversion of chalcocite to covellite, Equation [2], is rate-controlled by the amount of oxygen that can be dissolved into the system and available to the sulphide surfaces. Distribution of oxygen is manipulated so as to provide maximum possible oxidant in the initial leaching reactors, and ensures that approximately 50% Cu dissolution and 100% lead sulphate conversion is achieved within the first few hours of leaching. The rate limiting parameter for the oxidation of covellite to Cu2+(aq), Equation [3], is the concentration of chloride in the system. The role of chloride in sulphide copper leaching is important to minimize the re-precipitation of covellite from Cu+(aq) if contacted with elemental sulphur: [4] High levels of arsenic also promote Cu re-precipitation in the BHAS process to form a Cu/As compound (2). It is thought that Cu+(aq) is unstable in low chloride concentrations, therefore chloride concentrations at the Nernst boundary layer must be sufficient to complex the Cu+(aq) prior to transport into the bulk solution where the oxidation of Cu+(aq) to Cu2+(aq) occurs, as: [5] Thus, the level of chloride in the BHAS leach process is maintained at around 20 gpl in the leach liquor to enable sufficient Cu+(aq) to be complexed and minimize the contact potential with S0 and As during leaching3. Primary leach Matte slurry is fed at approximately 35 tonnes per day to six first stage leach reactors in series, each with a nominal residence time of around 40 minutes, with the leached slurry gravitating through the leach circuit via covered launders. The slurry is diluted with reheated SX raffinate and concentrated sulphuric acid, to 8% solids. Compressed oxygen is dispersed through each of the first five reactors. Leaching is maintained at an optimum temperature of 85°C with the addition of steam. The overflow from the final leach reactor containing 7% solids gravitates to the primary leach thickener where flocculant, Magnafloc E10, is added to assist with solids liquid (S/L) separation. The hot thickener overflow, pregnant leach solution (PLS), is pumped to a void cooling tower where it is cooled so that the temperature in the PLS Tank is around 40°C. The cooled PLS solution is clarified to reduce the final level of suspended solids to SX. Second stage leach—high acid repulp The primary leach thickener underflow at 35–40% solids is pumped to the acid repulp reactor, similar to the primary leach reactors, but with a larger residence time of around 4–5 hours. All SOLVENT EXTRACTION (SX) REAGENT SELECTION

285

283-294_Crane:text

2/13/09

10:52 AM

Page 286

of the chloride, as salt, and additional acid is added at this point, to ensure a high acid, high density, high chloride repulp of the leach slurry. Oxygen is also dispersed through this reactor. The repulped slurry is then washed with all the returning SX raffinate by mixing in a wash mixing tank (repulp mixer) before gravitating to the repulp thickener where Magnafloc E10 is again added. To aid in silica control, Polysil® is also added to the repulp mixer along with the returning SX raffinate and the repulp reactor overflow. The repulp thickener overflow is directed to the first leach reactor in the primary leach. The underflow is pumped to an agitated neutralization tank. Copper SX The solvent extraction circuit is configured as a conventional 2E x 1W x 2S circuit. The PLS containing around 40 g/l Cu, 20 g/l acid and 20 g/l chloride is pumped to two extract stages in series. The loaded organic from E1 is washed with acidified water in order to reduce impurity transfer (particularly chloride) then stripped in two stages with spent electrolyte returning from EW. Due to the high copper content of the PLS, a higher than ‘typical’ extractant concentration of approximatley 30–35 volume % (in Shellsol 2046 diluent) is employed in conjunction with a high O/A ratio of 4:1. This provides a nominal Cu recovery of approximately 70%. While this may be considered low by industry standards, the high PLS acidity makes further copper recovery very challenging without making significant circuit modifications or without neutralization. Copper electrowinning (EW) The Cu from SX is recovered by EW process employing the ISA process using 30 cells each containing 40 stainless steel or (later) titanium cathode plates and 41 anodes of 6% antimonial lead. More recently a Pb/Ca/Sn composition cast anode has been trialled. FC-1100 is added to control acid mist. Residue treatment The repulp thickener underflow contains recoverable lead and silver. To render the thickener underflow solids suitable for recirculation through the smelter, it is necessary that all of the free acid and most of the Cu process solution associated with it be removed. It is also desirable that prior to discharging the solution associated with the thickener underflow most of the dissolved metals be precipitated out. Neutralization of repulp thickener underflow is conducted continuously by controlled addition of milk of lime and the neutralized slurry is filtered on an Andritz belt filter. The residue discharges from the filter at 60–70% solids and is recycled to the lead plant. The metal free filtrate collects in a filtrate tank and is returned to the Andritz filter with any excess being discharged to on-site waste collection. Summary of process difficulties The use of the highly effective chloride/sulphate leach process produced a warm pregnant leach solution (PLS) high in acid, chloride and copper content, which resulted in several challenges for the operation of the Cu SX-EW plant requiring, for example, effective solids liquid separation and soluble silica control; high concentrations of SX extractant and high organic/aqueous (O/A) ratios in extraction; and careful monitoring of organic health with regular, effective, organic treatment. 286


283-294_Crane:text

2/13/09

10:52 AM

Page 287

Following commissioning of the SX ciruit in 1984, the process experienced a gradual deterioration in the quality of the organic phase, so causing an increase in the transfer of chloride to the EW circuit. As chloride levels in the EW electrolyte surpassed 100 ppm, the original 316L cathodes were replaced by titanium cathode mother plates. In 1990 the original Acorga P5100 extractant was replaced by Acorga M5640. Over the following years, the plant struggled with low nett copper transfer rates and high EW chloride levels in the range of 200– 500 ppm, which resulted in high chlorine gas concentrations in the surrounding environment causing corrosive damage to the EW infrastructure. In August 2003 a sample of Port Pirie plant organic was analysed by Cognis in Tucson to determine reasons for the poor performance by analysing the composition of the circuit organic. The main findings were as follows: • The organic phase was dark orange in colour. It was extremely viscous and exhibited very poor organic continuous phase separation behaviour, >15 minutes at 25°C. Aqueous continuous phase separation time was reasonable • 16.2% of the total oxime in the circuit was present as the nitro-C9 aldoxime—a very high level—causing the organic phase to exhibit a very high viscosity (and a dark orange colour) • The TXIB:C9 aldoxime ratio was significantly higher (0.38) in the plant organic compared to that in Acorga M5640 (0.28) as a result of the differential degradation rate of C9 aldoxime compared to the TXIB modifier. This had caused the viscosity of the plant organic to increase with time and assisted in promoting nitration • Despite the reoximation process, the concentration of C9 aldehyde in the circuit was quite high (2.3%) compared to other circuits elsewhere. The same conditions that favour nitration also promote hydrolysis of the aldoxime to the aldehyde (warm temperature and high acid concentrations in the aqueous phase). Given the operating conditions at Port Pirie higher hydrolysis levels than conventional circuits may be expected, but the measured levels seemed excessive, especially as Port Pirie practised continuous reoximation of the plant organic • The organic also contained significant amounts of the 2-cyano-4-nonylphenol and 5nonylsalicylamide, which is unusual. They are typically the result of thermal degradation of the C9 aldoxime at temperatures higher than those said to be experienced in the Port Pirie operation. They were estimated to be present at about 10–15% of the level of the C9 aldoxime, thus making a significant contribution to the viscosity of the organic phase • Copper strip kinetics were surprisingly slow, 62% @ 30 seconds. Typical circuit organics would be expected to be in the range of 95% @ 30 seconds • The overall net transfer capability of the organic was low. It did not strip well due to the presence of the nitro-C9 aldoxime, which is an extremely strong copper extractant. This resulted in high stripped organic values. The net Cu transfer was further exacerbated by the low PLS pH, which in conjunction with a highly modified circuit organic provided a very ‘shallow’ equilibrium extraction isotherm • The organic did not respond well to clay treatment probably due to the presence of unfilterable entrained aqueous solution which deactivates clay. The entrained aqueous may have existed as micro-emulsions formed due to the presence of the degradation products or contaminants. Reagent selection and Isocalc® TM modelling– proposal to optimize the plant operation Cognis has experience with the effects of nitration at other Cu SX plants and has conducted laboratory degradation studies using different extractant compositions to determine which extractants were more resistant to nitration5,6. SOLVENT EXTRACTION (SX) REAGENT SELECTION

287

283-294_Crane:text

2/13/09

10:52 AM

Page 288

Figure 1. Effects of modifiers on the stability of oximes in the presence of nitrate

Clearly, the modified aldoximes, LIX ®622N and Acorga ®M5640, and the modified ketoxime, LIX®84-I plus TDA, were much more sensitive to degradation/nitration than the non-modified reagents, LIX®984N and LIX®84-I. Thus, Cognis recommended that (a) Port Pirie cease addition of the modified reagent Acorga M5640 and employ the more stable nonmodified reagent LIX® 984, and (b) Port Pirie cease re-oximation. Cognis have much experience in re-oximation, and attempts to re-oximate heavily degraded Cu circuit organics typically result in an organic that demonstrates very poor physical performance. LIX 984 was selected for several reasons: • LIX 984 is a 50/50 blend of ketoxime and C12 aldoxime, both of which are more stable than the C9 aldoxime. LIX 984 does not contain a modifier • Acorga M5640 contains both a C9 aldoxime and modifier, TXIB • TXIB (and all modifiers in general) promote nitration where nitrates are present in the aqueous solution • LIX 984 showed good performance in the screening tests in Tucson. In order to support the reagent conversion, a technical service package was negotiated. Port Pirie accepted the Cognis offer and LIX® reagent addition began in May 2004. Pre-conversion plant survey In April 2004 Cognis visited the Port Pirie smelter to conduct a pre LIX reagent addition plant survey and to collect a second plant organic sample to provide a baseline for the conversion.7 Analysis of the plant organic revealed that the Acorga M5640 concentration was approximately 43 vol% (24.3 g/l Cu). However, the concentration of Cu in the E1 organic was only 9.05g/l, which calculates to a copper loading of only 37%. The Cu net transfer was very poor, at approximately 0.14 g/l Cu per % oxime due in part to the presence of the nitro-C9 aldoxime and the weak reagent blend (due to excess TXIB modifier). All mixers were operated aqueous continuous due to the extremely slow phase separation under organic continuous mixing. The phase disengagement times (PDTs) of the plant loaded organic with both the plant PLS and a synthetic PLS were measured in the laboratory under both aqueous and organic continuous conditions at room temperature (23°C). For comparative purposes, fresh samples of LIX984 and Acorga M5640 were also tested. The results are provided in Table I. 288


283-294_Crane:text

2/13/09

10:52 AM

Page 289

Table I Laboratory PDT results pre-LIX reagent addition PLS/continuity Plant organic Plant O/C Plant A/C Synthetic O/C Synthetic A/C

Sample PDT (sec) Acorga M5640

Did not separate 126 Did not separate 71

103 119 149 49

LIX 984 104 107 139 35

Table II Organic viscosity results pre-LIX reagent addition Sample SG (@ 43 vol%) Viscosity (cSt)

Plant LO

Acorga M5640

LIX 984

0.93 26.48

0.86 5.01

0.83 3.93

Under aqueous continuous mixing conditions, the PDT was measured at similar rates to that as experienced in the operating plant, whereas under organic continuous conditions the emulsion demonstrated very poor phase separation behaviour taking several minutes for the emulsion to indicate any signs that the phases would separate; and after 24 hours the aqueous phase contained a considerable amount of ‘rag’ and an indefinable interface. This indicated high levels of organic contaminants were interfering with the coalescence of aqueous droplets within the organic continuous emulsion. Clay treatment—even at high dosage (10 wt% clay, typically 2–5 wt% is sufficient for heavily contaminated organics)—failed to improve the PDT. Viscosity and specific gravity (SG) measurements were conducted on the plant loaded organic and compared to fresh samples of LIX 984 and Acorga M5640 at 43 vol%, with all sample temperatures equilibrated in the laboratory (23°C). The resultant viscosities are listed in Table II. The viscosity and SG of the plant organic were much higher than that for fresh LIX 984 and fresh Acorga M5640. Both the high viscosity and SG resulted from an accumulation of degradation products that no longer participated in the extraction and stripping chemistry but which had a profound impact on phase separation, aqueous entrainment and plant performance. The physical appearance of the plant organic was unusual in that it had a distinct dark orange/red coloration when in a thin film, and it tended to coat all surfaces with which it came in contact. This orange/red colouration is consistent with the presence of nitration products. It had a very syrup-like consistency and a distinct smell. When filtering plant organic, the samples required extensive time to filter through Whatman 1PS filter paper and there was a reasonable amount of aqueous remaining in the filter paper. This would tend to indicate that the aqueous formed a very fine micro-emulsion in the organic phase that was not readily released. After the initial plant survey it was concluded that the plant organic was in extremely poor shape. Reagent changeover period and resultant plant operation Following the addition of LIX 984 to the circuit in April 2004, plant monitoring and organic sampling were conducted on a regular basis to track the chemical composition as well as the SOLVENT EXTRACTION (SX) REAGENT SELECTION

289

283-294_Crane:text

2/13/09

10:52 AM

Page 290

Table III Reagent concentrations in the Port Pirie plant organic post-LIX reagent addition Vol% or Wt/Vol%* Aug 03 (pre-LIX) Jan 05 Jul 05 Oct 05 Mar 06 Jun 06 Dec 06 Feb 07 Sep 08

LIX 84-I

M5640

LIX 860-I

Nitro Oxime^

TXIB*

7.6 11.0 12.3 14.0 12.8 8.2 7.0 6.9

34.9 10.9 6.5 4.3 3.1 2.6 1.8 1.6 0.2

9.8 11.9 12.4 12.4 12.9 16.4 18.1 20.5

3.1 3.0 2.8 2.2 2.1 2.0 2.4 1.8 1.2

13.12 7.78 5.76 3.88 1.96 1.64 1.75 1.27 0.32

Note. ^ Concentration is calculated as an M5640 equivalent since the majority of the nitrated material is nitro C9 aldoxime. The LIX reagent addition blend changed from LIX 984 to LIX 973 in April 2006

chemical and physical performance of the plant organic. Plant samples were assayed for reagent composition, including C9 and C12 aldoxime, ketoxime, TXIB, hydrolysis products and nitration products using a variety of analytical techniques refined by Cognis’ Tucson lab. The organic composition monitoring was compared with the corresponding chemical and physical performance monitoring. Analytical results are presented in Table III. The cessation of re-oximation and the conversion to LIX reagent has gradually improved the ‘health’ of the plant organic. This is evident by the lower concentrations of Acorga M5640 (C9 aldoxime) and TXIB with time. The TXIB and M5640 concentrations had decreased from 13.12% and 34.9% in August 2003 to 0.32% and 0.2% in September 2008 respectively, indicating a significant change from modified to unmodified plant organic. The level of C9 nitro aldoxime decreased at a lesser rate than the Acorga M5640 and TXIB. This is consistent with the level of residual copper on the stripped organic. Nitration appeared to occur on an infrequent basis, suggesting that Port Pirie was getting nitrate into the system via ‘specific events’ throughout this period. The chemical performance of the Port Pirie Cu SX plant was monitored regularly and compared with the health of the plant solvent. Plant surveys and chemical analysis were used to monitor the important parameters such as Cu recovery, reagent concentration, E1 organic loading as % of max load and net Cu transfer. Survey data are listed in Table IV. The reagent net transfer has improved as the LIX reagent conversion has progressed due to the reduction in the concentration of nitro oxime species present in the plant organic and an increase in the transfer capacity. The E1 organic loading has also increased from ~60% to above 80%. (The low Oct 07 levels were due to a dip in PLS Cu.) Although organic loading is highly dependent on the Cu and acid concentration in the PLS, by reducing the TXIB content in the organic the E1 organic loading was increased, both of which reduce the potential for crud formation. The extraction and stripping kinetics of the Port Pirie plant organic have improved significantly since the initial addition of the LIX reagent, as shown in Table V. Surface active impurities crowd and compete with the oximes at the interface, adversely affecting extraction and stripping kinetics. The reduction in the concentration of these surface active impurities in the plant organic has improved the extraction and stripping kinetics. 290


283-294_Crane:text

2/13/09

10:52 AM

Page 291

Table IV Survey data for the Port Pirie Cu SX plant Parameter

May 04

Nov 04

Jan 05

Jun 05

Nov 06

Apr 07

Oct 07

Sep 08

PLS flow (m3/h) PLS Cu (g/l) Raffinate Cu (g/l) Cu recovery (%) Organic flow (m3/h) Loaded organic Cu (g/l) Strip organic Cu (g/L) Reagent conc. (vol%) Net Transfer (gCu/vol%) % max load

18.0 41.0 12.9 68.5 78.8 14.94 8.20 45.0 0.150 59.0

19.8 45.8 15.4 66.4 80.4 12.19 4.70 34.0 0.220 66.4

19.2 44.6 13.9 68.8 84.0 10.62 3.60 30.0 0.234 65.5

20.2 42.3 15.0 64.5 83.9 11.37 4.80 25.8 0.255 83.9

20.4 45.0 15.5 65.6 80.4 13.95 6.46 32.6 0.229 81.5

20.0 51.9 14.8 70.7 80.0 15.52 6.34 33.5 0.273 88.2

18.3 32.0 8.1 65.5 77.0 10.59 4.91 33.3 0.174 65.5

13.3 45.2 10.9 75.8 72.0 11.23 4.98 30.0 0.211 69.3

Table V Extraction and stripping kinetics data for the Port Pirie Cu SX plant organic Kinetics Extraction @ 30 sec Strip @ 30 sec

Aug 03

Jul 05

Mar 06

Jun 06

Dec 06

Feb 07

Sept 08

62%

71%

86% 78%

90% 90%

90% 92%

96% 96%

92.2% 97%

Table VI Results from Laboratory PDTs for Port Pirie plant organic and plant PLS Date

Apr 04 Jan 05 July 05 Oct 05 Mar 06 Jun 06 Dec 06 Feb 07 Apr 07 Oct 07 (45°C) Sep 08

PDT (seconds) OC

AC

No break No break No break >600 >600 600 600 474 114 49 152

126 130 110 90 72 100 93 109 201 26 450

Clay treatment on the Sept 2008 sample increased kinetics to 99.6% and 98.9% respectively. The plant physical performance was monitored by measuring operating parameters such as emulsion temperature, mixer continuity, mixer O/A ratio, emulsion PDT under both AC and OC mixing (before and after clay treatment), and organic viscosity. Laboratory tests were conducted at room temperature (22 ± 2°C). Plant physical parameters such as temperature, mixer continuity and mixer O/A ratio remained relatively constant for all sampling dates. The temperature was 40 ± 2°C, the mixer continuity was always aqueous continuous, and the O/A ratio in the mixers was in the range 1.0 to 1.3. The plant PDT trends were consistent with that for the laboratory PDT. (Table VI) SOLVENT EXTRACTION (SX) REAGENT SELECTION

291

283-294_Crane:text

2/13/09

10:52 AM

Page 292

As the LIX reagent conversion progressed, the PDT of the Port Pirie plant organic decreased significantly. This reduction in PDT corresponded with the reduction in degradation products and impurities in the plant organic. The Port Pirie Cu SX plant is now at a stage where it can convert all mixers to OC, which is the preferred operation for Cu SX, especially operations that have high levels of silica and solids present in the PLS. Initial clay treatment tests conducted in August 2003 proved unsuccessful, possibly due to the presence of micro-emulsions in the plant organic, as witnessed previously with nitrated oximes. Later clay treatment proved successful as the levels of nitrated oxime and other impurities declined, as shown in Table VII. After 2006, the clay treatment process became a part of normal operation at Port Pirie, which also aided in improving the physical performance of the plant organic. As the LIX reagent conversion progressed, the concentration of degradation products and contaminants decreased, resulting in a subsequent reduction in plant organic viscosity, as shown in Table VIII. The chloride concentration in the electrolyte is compared against the % LIX reagent in the organic below (Figure 2). Table VII Clay treatment results Date Pre-CT April 04 March 06 June 06

Extraction (OC) Post-CT

No break 920 sec 627 sec

Strip (AC) Pre-CT

Post-CT

72 100

20 20

440 22 22

Table VIII Viscosity of Port Pirie plant organic in cSt at 25°C Apr 04

Sep 04

Jan 05

Jul 05

Oct 05

Mar 06

Jun 06

Dec 06

Feb 07

Sep 08

26.48

19.43

15.06

15.13

13.15

13.90

12.20

9.50

8.93

6.66

Figure 2. Chloride concentration in the electrolyte

292


283-294_Crane:text

2/13/09

10:53 AM

Page 293

The chloride concentration in the PLS remained consistent throughout the LIX conversion period in the range 15–25 gpl. It is evident that as the LIX reagent conversion progressed, the chloride concentration in the electrolyte has decreased significantly from unacceptable levels of ~200 ppm, to approximately 100 ppm with less than 50% LIX reagent in the circuit, to less than 30 ppm when operating with more than 90% LIX reagent in the circuit. This reduction in electrolyte chloride corresponded exceptionally well with the reduction in nitration products, TXIB and other impurities in the plant organic. Conclusions The conversion of the Port Pirie plant organic from a modified C9 aldoxime to a mainly unmodified blend of C12 aldoxime and ketoxime, plus the decommissioning of the reoximation stage, has directly resulted in significant improvements in the health of the plant organic. Since the modified Acorga reagent M5640 was replaced by the non-modified LIX reagent the health of the SX organic and the performance of the circuit have both significantly improved: • The strip point has decreased as the level of nitrated oxime has declined • Extraction and strip kinetics have improved • Net transfer has increased to greater than 0.2 gpl Cu/vol% • Reagent strength has decreased to less than 35 vol% • Wash stage acid has been reduced from 40 gpl to 15 gpl, resulting in savings on neutralization and reduced soluble Cu losses • Organic phase viscosity has declined to a quarter of the viscosity of the pre-LIX period • Organic continuous phase separation time has very significantly improved • Clay treatment is now used to maintain organic health • And most importantly, aqueous entrainments through to EW have significantly declined, so reducing EW chloride from more than 200 ppm to less than 30 ppm, and improving the EW atmosphere and environment. References 1. MEADOWS, N.E. and VALENTI, M. The BHAS Copper–Lead Matte Treatment Plant, The AusIMM Non Ferrous Smelting Symposium, Port Pirie, September 1989. 2. JOHNSON R.D., MILLER, I.B., MEADOWS, N.E., and RICKETS, N.J. Oxygen treatment of Sulphuric Materials at Atmospheric Pressure in and Acid Chloride-Sulphate Lixiviant, The AusIMM Non Ferrous Smelting Symposium, Port Pirie, September 1989. 3. MEADOWS, N.E. and POLLARD, D.M. Oxidative Leaching of Chalcopyrite in a Chloride-Sulphate Lixiviant, The AusIMM Adelaide Branch, Research and Development in Extractive Metallurgy, May 1987. 4. TYSON R.K., MEADOWS, N.E., and PAVLICH, A.D. Copper production from matte at Pasminco Metals—BHAS, Port Pirie, SA, The AusIMM, pp. 732–734. 5. VIRNIG, M.J., EYZAGUIRRE, D., JO, M., and CALDERON, J. Effects of Nitrate on Copper SX Circuits: A Case Study, Copper 2003, Santiago Chile, Hydrometallurgy of Copper, vol. VI (2), Riveros, P.A., Dixon, D.G, Dreisinger, D.B., and Menacho, J.M. (eds.). Canadian Institute of Mining, Metallurgy and Petroleum, 2003. pp. 795–810. 6. BART, H., MARR, R., BAUER, A., SCHEIN, R., and MARAGETER, E. Copper Extraction in Nitrate Media, Hydrometallurgy, vol. 23, 1990. pp. 281–295. 7. DUDLEY K., HORNER, A., VIRNIG, M., CRANE, P., and URBANI, M. SX-EW process optimisation at Nyrstar Port Pirie—the influence of employing the correct reagent type, ALTA 2008, Perth. SOLVENT EXTRACTION (SX) REAGENT SELECTION

293

283-294_Crane:text

2/13/09

10:53 AM

Page 294

Phil Crane Business Manager, Cognis Australia, Australia Phil graduated from Birmingham University with BSc Hons in Minerals Engineering in 1975; with final year studies on copper solvent extraction. He worked in metallurgical plants and in research and development in Zambia and Iran; then in the design and supply of solids-liquid separation and flotation process equipment in both England and later South Africa. Moved to Cognis, then Henkel in 1988; working in the PGM flotation field and supporting Henkel’s copper and uranium solvent extraction business activities in southern Africa. He then transferred to Australia in 1996 and from 2002 Phil has been responsible for Cognis Mining Chemical Technology’s business and technical support in the Asia-Pacific region, specifically in business development and customer’s plant troubleshooting.

294


295-304_vRensburg:text

2/13/09

10:54 AM

Page 295

VAN RENSBURG, D.M. MUNYUNGANO, B., HAIG, P., LOUIS, P., and STOLTZ, J. Organic degradation in uranium and cobalt solvent extraction: the case for aliphatic diluents and antioxidants. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Organic degradation in uranium and cobalt solvent extraction: the case for aliphatic diluents and antioxidants D.M. VAN RENSBURG*, B. MUNYUNGANO† *ChemQuest (Pty) Ltd, South Africa †Rössing Uranium Limited, South Africa

In the solvent extraction of cobalt, nickel, uranium and zinc, various parties 1,2 have reported over the years that a certain amount of degradation of the organic phase has taken place, affecting kinetics, loading and phase separation times. When the degradation products of the organic phase were detected at a uranium operation in Namibia3, a cobalt refinery in South Africa, and a cobalt pilot plant in the DRC4, a number of steps were taken to combat the problem, including the use of an aliphatic diluent and the addition of butyl hydroxy toluene as an antioxidant. Through a combination of laboratory testing and plant observation, it was concluded that plant management to identify, isolate and control the source of the oxidant material was sufficient to combat the problem, and that neither the diluent change nor the use of the antioxidant was proven to have made any significant difference to the degree of organic degradation

Introduction In 2002, Rössing Uranium mine in Namibia reported that organic phase breakdown products may be the cause of problems in the strip section of the solvent extraction plant, which seemed to relate to the presence of nitrosamines3. Extensive crud formation, poor stripping efficiency and excessive organic entrainment were noted. ChemQuest was approached as the supplier of the isodecanol in use as a phase modifier, and as link to the supplier of the diluents, which at that time was Shell Chemicals. At the same time, a small cobalt extraction facility in South Africa reported very poor performance on the nickel/cobalt separation stage of their process4, which also appeared to be linked to breakdown products of the organic extractant and diluents. ORGANIC DEGRADATION IN URANIUM AND COBALT SOLVENT EXTRACTION

295


2/13/09

10:54 AM

Page 296

As ChemQuest was extensively involved in the supply of organic reagents, and a number of projects in the DRC were beginning to develop, we decided to invest some time in researching the subject of organic degradation in certain solvent extraction operations, and the measures taken to combat the problem. Organic degradation in uranium solvent extraction The organic phase at Rössing is made up as follows: • Alamine 336, a tertiary amine: 7% by volume • Isodecanol phase modifier: 3% by volume • Diluent, initially Shellsol 2325: 90% by volume The tertiary amine extractant R3N functions as follows: First it protonates in sulphuric acid 2R3N + H2SO4 ⇔ (R3NH)2SO4 Then it exchanges/extracts uranyl sulphate: (3(R 3 NH) 2 SO 4 + UO 2 (SO 4 ) 2 2- ⇔ (R 3 NH) 6UO2(SO4) 4 +SO42This is a process of ion association and differs from copper extraction (chelation, or complex formation) and systems using acidic extractants such as D2EHPA and Cyanex 272 (solvation). It is neither highly selective nor as pH dependent as the chelation and solvation processes, and in practice ion exchange with many anionic species takes place, including coextraction of nitrates to form nitrosamines:

The presence of the nitrosamines and other organic degradation products, detected by gas chromatographic analysis, was correlated directly with the upset conditions on the SX plant. In turn, the formation of the nitrosamines was initially linked to two factors: the ingress of nitrates with process water originating from the explosives in the open pit and high redox potentials from the leach circuit. The presence of Mo as a catalyst was also identified as a contributor. Nitrates and nitrosamine formation Laboratory tests indicated that the presence of nitrous acid at a concentration of 1 g/l and at a redox potential of 550 mV led to the formation of nitrosamines. Water is a scarce resource in the Namib Desert and the recycling of water is unavoidable— there simply is not enough water to ‘bleed and top-up’ to keep species like nitrates under control cost-effectively. The use of ‘pit’ water in the elution of the ion exchange resin (upstream from the SX, to produce the pregnant leach feed solution to the SX) is inevitable and thus also the ingress and build-up of nitrates. However, the reduction plant managers have found that controlling the nitrates in the PLS to SX at 0.5–0.7 g/l and loadings on the organic to below 1 g/l controls the nitrosamine formation to acceptable levels. It was found that the nitrosamines were scrubbed from the organic phase during the solvent regeneration step using sodium carbonate. Routine monitoring of nitrates in the concentrated eluate (effectively the SX PLS), the ammonium sulphate, the strip solvent and at other points has been implemented (see Figures 1 and 2). Control of the pH in the strip and scrub sections was integral to the removal of the nitrosamines, and better management measures were also introduced. Obviously, the reduction of the amount of nitrates in the incoming water sources, although nearly impossibly expensive, is an ongoing exercise. 296



2/13/09

10:54 AM

Page 297

Conc. chlorides and nitrates Cl-

NO3-

4

1.2

3.5 1

0.8 2.5 2

0.6

1.5

Nitrates (g / l )

Chlorides (g / l )

3

0.4 1 0.2 0.5

2007/01/28

2006/12/31

2007/01/14

2006/12/17

2006/11/19

2006/12/03

2006/11/05

2006/10/08

2006/10/22

2006/09/10

2006/09/24

2006/08/27

2006/07/30

2006/08/13

2006/07/02

2006/07/16

2006/06/18

2006/06/04

2006/05/21

2006/05/07

2006/04/09

2006/04/23

2006/03/12

2006/03/26

2006/02/12

2006/02/26

2006/01/29

2006/01/15

0 2006/01/01

0

Figure 1. Rössing plant monitoring

Amm. sulph and strip solv nitrates Amm. sulph

Target

Strip solvent

8 7 6

g/l

5 4 3 2 1

2007/01/28

2007/01/14

2008/12/31

2008/12/17

2008/12/03

2008/11/19

2008/11/01

2008/10/22

2008/10/03

2008/09/24

2008/09/10

2008/08/27

2008/08/13

2008/07/30

2008/07/16

2008/07/02

2008/06/18

2008/06/04

2008/05/21

2008/05/07

2008/04/23

2008/04/09

2008/03/25

2008/03/12

2008/02/25

2008/02/12

2008/01/29

2008/01/01

2008/01/15

0

Figure 2. Rössing plant monitoring

Redox potential The link between high redox potential and the degradation of the organic phase was investigated, both on the plant and by a survey of similar circumstances reported in the literature. The acid leaching of the Rössing uranium-bearing ore requires the oxidation of tetravalent uranium oxide to the acid soluble hexavalent form. This is achieved by indirect oxidation via ferric ions. The ferric is reduced to ferrous, then re-oxidized to ferric by the addition of pyrolusite, manganese dioxide. The process is controlled by the maintenance of the ferric to ferrous ratio. The carry-over of high redox potentials from the leach to the IX and in turn to the SX is controlled by the addition of iron metal (wire) to the pregnant solution. ORGANIC DEGRADATION IN URANIUM AND COBALT SOLVENT EXTRACTION

297


2/13/09

10:54 AM

Page 298

Despite the link found in research and in the laboratory between high redox potentials and organic degradation/formation of nitrosamines, no significant correlation was found between the redox potential in the leach and the redox potential in the SX circuit. The poor milling of pyrolusite and associated carry-over of MnO2 to the IX did not influence the redox potential in the SX. Varying redox potentials in the leach and in the concentrated eluate from IX did not reflect in the measured redox potential anywhere in the SX circuit, nor did it correlate with nitrosamine formation or SX performance upsets. Other organic degradation factors/products The literature survey and the experience of some applications (see below) led to Rössing investigating the possibility of the degradation of the other organic constituents5,6, namely the isodecanol and the paraffinic diluent. Dr. Gordon Ritcey7 noted from personal experience as follows: ‘Isodecanol (CH3(CH2)9OH) was oxidized in the presence of air and high redox potential to aldehyde, carboxcyclic acid and eventually primary alcohol which will result in poor coalescence. Iso-tridecanol (CH3(CH2)12OH) which is inherently more stable and has lower solubility in both the aqueous phases and in many organics, had been used in applications where degradation was noted and possible.’ Isodecanol is an essential phase modifier in uranium solvent extraction systems as it improves the solubility of the tertiary amine in the diluent5. Rössing have a standard method for testing the kinetics, loading and phase separation characteristics of their fresh organic constituents, and tested a sample of isotridecanol. They found it offered no significant difference to isodecanol, and as it was considerably more expensive, they did not pursue its possible use. Furthermore, in the examinations of the both the fresh organic and degraded organic chromatographic scans, no difference in the peaks of the isodecanol could be detected. Without further evidence, it was decided that no isodecanol-linked degradation products could be detected, and that no change would be made to the isodecanol use or regime. The diluent presented another scenario entirely. Following the reported experience of cobaltcatalysed degradation of diluents (discussed below), and laboratory testing under accelerated oxidative conditions, it was decided that the aromatic diluent in use, Shellsol 2325 containing 16–23% aromatics, would be replaced by Sasol SSX 210, with < 0.1% aromatic fraction. Initial testing was performed by ChemQuest in the laboratory, using a jacketed glass reaction vessel, similar to the Acorga apparatus used for testing copper solvent extraction reagents, together with a variable speed stirrer. A number of extraction and strip tests were carried out using a synthetic solution containing uranyl sulphate, and in the case of the degradation tests, some permanganate ions as the oxidizing agent (5 mg/l as KMnO4) and butyl hydroxy toluene (BHT) as an antioxidant (0.2% m/v). The starting pH was maintained exactly by adjustment with NH4OH. The purpose of the test was to determine, inter alia, the effect of organic oxidation on extraction and phase separation times with the type of diluent and the presence or absence of BHT as the variables. The accelerated oxidation was done at an elevated temperature (45°C) using the same synthetic solution used by Rössing to perform their own kinetic tests, but with the addition of the permanganate ions. Organic to aqueous ratios were kept at exactly 1:1 and mixing times and stirrer speed kept constant (180 minutes, 220 rpm). Air was bubbled through uniformly during the oxidation phase using a constant pressure air pump. 298



2/13/09

10:54 AM

Page 299

The organic phase was removed after the oxidation and tested for extraction and phase disengagement using the same synthetic PLS solution but without the permanganate ion, at ambient temperature, and using the same apparatus. (See Figure 3.) The complete test methods, solution concentrations, test parameters and all results are available from the author in a separate document8 which includes work on copper, zinc and cobalt/nickel solvent extraction systems. The key results related to uranium were as follows: (Table I) The most significant indication was that, with or without BHT anti-oxidant, the aliphatic diluent showed far fewer deleterious effects after oxidation than was seen by the diluent containing a portion of aromatics. As there was no funding or incentive available for ChemQuest to continue these tests, they were limited to single results and the single set of oxidizing conditions. Rössing, however, accepted the trends indicated and decided to continue with a plant change-over to the aliphatic diluent and to test the use of the BHT.

B 24 Sockets

Water out

1 cm Baffle (4 in all)

Thermometer pocket 1 litre Beaker

Graduated scale (mm)

3 cm

Water in

10.2 cm

Figure 3. Apparatus used in accelerated degradation tests phase disengagement and kinetic apparatus

Table I Results related to uranium Extraction of uranium, %

Shellsol 2325 Sasol SSX 210 Shellsol 2325 with 0.2% BHT Sasol SSX 210 with 0.2% BHT

Phase disengagement time, secs

Before oxidation

After oxidation

Before oxidation

After oxidation

73.1 72.9 72.2 72.9

33.8 54.9 63.8 71.8

35 38 35 36

95 78 51 40

ORGANIC DEGRADATION IN URANIUM AND COBALT SOLVENT EXTRACTION

299


2/13/09

10:54 AM

Page 300

To date, the aliphatic diluent has not been shown to provide any significant changes in the operation of the solvent extraction facility. A number of factors, such as strip efficiency, crud formation at low temperature, and bacterial growth may or may not be linked to the diluent change, especially as other influencing parameters have also changed. The BHT has not yet been tested on the plant. As Rossing have simultaneously instituted corrective measures and tests to control the ingress of nitrates, and therefore the formation of nitrosamines, the influence of the diluent change cannot be measured or quantified, and no conclusion drawn. What is clear from this exercise, however, is that it remains better to tackle the source of the problem, rather than treat the symptoms and effects—i.e. controlling the formation of nitrosamines is probably far more beneficial and effective than trying to adapt the organic reagents to handle the oxidative conditions. Organic degradation in cobalt/nickel separation solvent extraction ChemQuest was alerted to a problem encountered in a very small solvent extraction facility in South Africa. Though not much more than a pilot plant, the facility was part of a plant used to produce a pure cobalt carbonate product, which in turn was calcined to cobalt oxide for sale in Europe. The solvent extraction part of the facility used di-ethyl hexyl phosphoric acid (Bayer D2EHPA) to extract zinc and manganese from cobalt-rich leachate. A cobalt nickel separation was then performed using bis (2,4,4-trimethylpentyl) phosphinic acid (Cyanex 272). The D2EHPA circuit showed increased viscosity, poor phase separation, and an increase in organic entrainment. Although the zinc and manganese extraction was not significantly different, it was seen that co-extraction of nickel and increased extraction of cobalt was taking place, both of which are not expected from the extraction curves for D2EHPA at the pH levels used in the plant. The Cyanex 272 circuit was performing within normal parameters initially, but started demonstrating poorer extraction kinetics. Even though the above plant was shut down by Umicore before resolution of the problem, samples were obtained for laboratory testwork. At the same time, Kasese Cobalt in Uganda was reporting crud runs that occurred without obvious reason in both their D2EHPA and Cyanex 272 (later replaced by Ionquest 290) circuits9, Chambishi Metals in Zambia reported crud problems in their D2EHPA circuit strip section10 and Namzinc Skorpion in Namibia was experiencing silicaceous gels in the Zincex D2EHPA SX and downstream coalescers and filters linked to high acidity in the leach plant11. As ChemQuest was in the process of introducing the product Chemorex D2EHPA to the market, and was assisting on a pilot-plant testing Ionquest 290, it was decided to do a study on organic degradation in these SX circuits as part of the technical service. Figures 4 and 5 give extraction curves generated with actual leached ore samples from the Congo Cobalt Corporation (now Boss Mining) Kakanda Cobalt Facility, which produces a heterogenite based cobalt concentrate in an oxide flotation circuit. The literature1,2, as well as a study done by Barnard12 indicated that problems could be generated by aromatic portions of the diluent breaking down. This could not be the case at either the Umicore circuit or at Kasese Cobalt, as they were both using aliphatic diluents (Shellsol D70 and Sasol SSX 210 at Umicore, Escaid 110 at Kasese). However, a common factor observed was the presence of manganese in high oxidation states at both plants. At Kasese, the manganese was oxidized in the electrowinning plant and returned with spent 300



2/13/09

10:54 AM

Page 301

100 90 80

% Extraction

70 60 50 40 30 20 10 0 0

1

2

3

4

5

6

pH

Figure 4. Chemorex D2EHPA extraction curves. 30% v/v in Escaid 100 at 35°C. All aqueous solutions soluble metal salts, pH adjusted by H2SO4. All points generated by changing O: A ratios

CYANEX 272 17.5% in ESCAID Using actual PLS solutions ex-Congo + NH4CH 120

100

% Extraction

80

60

40

20

0 0

1

2

3

4

5

6

7

8

9

Equilibrium pH

Figure 5. Cyanex 272 extraction curves

electrolyte as the strip solution in the cobalt/nickel SX. At Umicore, the leach section used strong oxidants, including peroxide, which were probably carried into the PLS entering the D2EHPA circuit. At Kasese, however, parallel testing and plant observations linked the crud formation to biological growth and high levels of acid-soluble silica. No evidence of permanganateinduced degradation was proven. The Umicore D2EHPA (18% by volume in Shellsol D70/Sasol SSX 210 blend) was sent for GCMS scans—both fresh solutions and the sample of degraded organic. Initial observation indicated peaks similar to those found on samples from a Versatic 10 (neo-decanoic acid) nickel pilot circuit at Tati Nickel (now Norilsk) in Botswana. In the degraded plant organic, ORGANIC DEGRADATION IN URANIUM AND COBALT SOLVENT EXTRACTION

301


2/13/09

10:54 AM

Page 302

Table II The effect of BHT as an antioxidant Extraction of zinc, %

Shellsol D70 Sasol SSX 210 Shellsol D70 with 0.2% BHT Sasol SSX 210 with 0.2% BHT

Phase disengagement time, secs

Before oxidation

After oxidation

Before oxidation

After oxidation

88.2 87.0 88.1 87.4

81.3 80.1 88.2 87.9

27 29 28 28

45 49 33 34

carboxylic acid type peaks were detected at a similar wavelength and retention time as detected by Barnard. These were not present in the freshly prepared organic. We drew an interim conclusion that the organic breakdown products contained carboxylic acids, which led to the co-extraction of nickel that had been observed. No accidental cross-contamination of the D2EHPA by Cyanex 272 was noted. Unfortunately, the Umicore plant was shut down before the work could be completed, and as there was no particular financial advantage for ChemQuest to continue the research, the work was not completed. However, during the course of the study, a more detailed study was done on the effect of BHT as an antioxidant, as mentioned above and in reference8, but with a D2EHPA system instead of the amine, and using only the aliphatic diluents. Using the same methods and apparatus described above, and similar artificial oxidation conditions to degrade the organic, on samples without BHT, the major results of the laboratory study were as follows: (Table II) The above exercise was aimed at proving firstly that there was no significant difference between the two diluents, and secondly that there was a measurable influence on extent of possible degradation by using BHT antioxidant. These objectives were partly achieved, but the project was not completed for the reasons indicated above. If organic degradation is found or suspected in SX circuits using solvation-type extractants such as Cyanex 272, Ionquest 290, D2EHPA or Versatic 10, then the use of an antioxidant is probably indicated. There is some contradictory evidence on whether aromatic diluents should or should not be used, but this becomes largely irrelevant when the antioxidant is used. Acknowledgement We would like to thank P. Haig, P. Louis and J. Stoltz for their contribution to this paper. References 1. FLETT, D.S. and WEST, D.W. The Cobalt Catalysed Oxidation of Solvent Extraction Diluents Proceedings ISEC ’86, Munich, 1986, vol. II. pp. 3–10. 2. RICKLETON, W.A., ROBERTSON, A.J., and HILLHOUSE, J.H. The Significance of Diluent Oxidation in Cobalt-Nickel Separation, Solvent Extraction and Ion Exchange, vol. 9, no. 1, 1991. pp. 73–84. 3. Personal correspondence between the author and Brodrick Munyangano and others, Rössing Uranium, 2002–2008. 302



2/13/09

10:54 AM

Page 303

4. Personal correspondence between the author and various parties at Umicore, Waste Products Utilisation, Chambishi Mines, SGS Lakefield and Mintek. All written. 5. MACKENZIE, J.M.W. Uranium Solvent Extraction using Tertiary Amines, Presentation Uranium Ore Yellowcake Seminar February 1997, Melbourne, Australia 6. MAXWELL, B., RASDELL, S., and CARLIN, P. Oxidative Stability of Diluents in Co/Ni solvent Extraction Presentation ALTA 1999. 7. Personal correspondence between the author, Peter Haig of Shell Chemicals and Dr. Gordon Ritcey. All written correspondence on record. 8. VAN RENSBURG, D.M. Accelerated Degradation Tests on Uranium, Copper, Zinc and Cobalt/Nickel Solvent Extraction Solutions Currently under preparation and refereeing for presentation. 9. Personal correspondence between the author and Moses Mugabe, Amos Silungwe and Stanford Saungweme of Kasese Cobalt. 10. Personal correspondence between the author and Hira Singh, Robert Minango and Kennedy Mwanza of Chambishi Metals. 11. Personal correspondence between the author and Johan van Rooyen (now with TWM), Herman Fuls and Jurgen Gnoinski of Namzinc and Dr. Kathy Sole of Anglo Research. 12. BARNARD, K.R. (AJ Parker Co-operative Research Centre) Tools for Diagnosis of Crud and Organic Degradation Problems in SX Circuits ALTA 2001.

Deon van Rensburg Product Manager, ChemQuest (Pty) Ltd, South Africa • 1995 to date: Manager of Solvent Extraction, Ion Exchange and Adsorbents Division at ChemQuest (Pty) Ltd; – Responsible for the development of Chemorex range of extractants; – Solvent Extraction plant technical service and design. • 1988–1995: NCP ACIX Division, manager activated carbon division • 1983–1988: Allied Colloids, technical representative • 1980–1982: Student • 1975–1979: Anglo American Research, learner official/student

ORGANIC DEGRADATION IN URANIUM AND COBALT SOLVENT EXTRACTION

303


304

2/13/09

10:54 AM

Page 304


305-312_vdBerg:text

2/13/09

10:56 AM

Page 305

VAN DEN BERG, D.A., MARÉ, P., and NEL, G.J. Development of the Tati Activox® BMR ammonia recovery circuit. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Development of the Tati Activox® BMR ammonia recovery circuit D.A. VAN DEN BERG*, P. MARÉ*, and G.J. NEL† *Hatch, South Africa †Norilsk Nickel, South Africa

The TATI Base Metal Refinery (BMR) was designed to be a hydrometallurgical processing plant employing Norilsk Nickel Process Technology’s Activox ® leaching technology, solvent extraction and electrowinning to annually extract and recover 25 kt of LME grade nickel, 22 kt of LME grade copper and 640 tonnes of cobalt as a carbonate. The BMR requires acid neutralization in the cobalt and nickel solvent extraction (SX) circuits where pH control is crucial for selective extraction of the base metal sulphides from the pregnant leach liquors. The choice of neutralizing reagent and treatment of the resultant neutralization product affect the overall plant design. This work discusses the criteria used to select ammonia as the neutralizing reagent for the BMR design, as well as the development of an ammonia recovery circuit. The ammonia forms a soluble ammonium sulphate neutralization product, which reports to the SX raffinate stream. The BMR ammonia recovery circuit uses vibrating mills to contact the ammonium sulphate with calcium oxide (fine quicklime). The ammonium sulphate is converted to CaSO4 (fine gypsum precipitate) and aqueous ammonia. A series of seeded reaction tanks are used to further encourage the gypsum precipitation and crystal growth. Aqueous ammonia is then stripped from the slurry in steam stripping column, and condensed to a 10% solution for reuse. The barren gypsum slurry reports to the tailings dam, where supernatant solution is recovered for use as process water. Anhydrous ammonia is used as top-up to account for losses due to the ammonia recovery circuit. Introduction The Tati Base Metal Refinery (BMR) has been designed as a greenfield project close to Francistown in Botswana. The remote site location in Sub-Saharan Africa introduces several challenges specific to the BMR design including logistics, as well as water and power constraints. An ammonia recovery circuit was designed for the BMR specifically to ease these constraints. DEVELOPMENT OF THE TATI ACTIVOX® BMR AMMONIA RECOVERY CIRCUIT

305

305-312_vdBerg:text

2/13/09

10:56 AM

Page 306

The BMR project is a hydrometallurgical processing plant employing Norilsk Nickel Process Technology’s proprietary Activox® leaching technology, solvent extraction and electrowinning to extract and recover annually 25 kt of LME grade nickel, 22 kt of LME grade copper and 640 tonnes of cobalt as a carbonate from the Tati Nickel Mine concentrates. The process consisted of ultra fine milling, low pressure and temperature leach, solid liquid separation, solvent extraction, electrowinning, iron removal, cobalt precipitation and ammonia recovery. The BMR requires acid neutralization in the cobalt and nickel solvent extraction (SX) circuits to selectively extract Co and Ni from the pregnant leach liquors into a high grade, pure loaded liquor for final metal production. Ammonium sulphate is produced during the SX neutralization process and remains in the SX raffinate solution. The ammonia recovery circuit is designed to recover aqueous ammonia from the SX raffinate solution for reuse in the SX as a neutralization agent. Recovering the ammonia reduces reagent make-up requirements and reduces the plant fresh water consumption by producing recyclable process water. The process solution from the ammonia recovery circuit can be reused as wash water in the solid liquid separation step. This paper will outline the decision to use ammonia as the SX neutralizing agent, the decision to use an ammonia recovery circuit, and the overall role played by ammonia recovery in the BMR design. The Tati Activox plant Tati Plant process description The Tati process is outlined in Figure 1, including the ammonia recovery and ammonia dosing. Tati mine concentrates are repulped from 90 to 50% solids using copper raffinate from the CuSX and fed to an ultra-fine grinding mill circuit where the concentrate is ground to 80% less than 10 microns. Milled slurry is then fed into the Activox® autoclaves where it is reacted in the multi-compartment horizontal autoclave, using oxygen at elevated pressures of 11 bar (g) and moderate temperatures of 105°C to produce copper, cobalt and nickel sulphates in

Figure 1. Tati Activox BMR process flow

306


305-312_vdBerg:text

2/13/09

10:56 AM

Page 307

solution for downstream recovery. The leach discharge slurry proceeds to the solid-liquid separation section that includes a 7-stage CCD washing circuit, where the leach residue solids are washed with recycled process water. The washed leach solids are then pumped to the PGE flotation circuit to recover the PGEs from the wash leach residue. The pregnant leach solution (PLS) from the CCD train is clarified prior to reporting to the copper PLS pond. The PLS is contacted with an organic solvent that selectively extracts copper from the aqueous solution in copper solvent extraction. Copper is stripped from the organic using spent electrolyte from the copper EW circuit. A portion of copper raffinate is recirculated to the concentrate preparation area for repulping the concentrate and for temperature control inside the autoclaves. The remainder of the copper raffinate is advanced to the iron removal circuit where the iron is precipitated using limestone. The thickener overflow is passed through clarifiers before it reports to the Co PLS pond. Cobalt PLS is contacted with an organic solvent that selectively extracts cobalt from solution in cobalt solvent extraction. Cobalt is stripped from the loaded organic phase with an acidic strip solution and the resultant loaded strip liquor (LSL) is filtered and pumped to the cobalt precipitation circuit. The pH in the cobalt SX circuit is maintained using aqueous ammonia. The LSL from the cobalt SX circuit is treated with sodium carbonate (soda ash) to recover cobalt as a carbonate product. The cobalt raffinate is then contacted with an organic solvent that selectively extracts nickel from the process solution in nickel solvent extraction. Nickel is stripped from the loaded organic phase with spent nickel electrolyte recirculated from nickel EW. The pH in the nickel SX circuit is maintained using aqueous ammonia addition. The nickel raffinate stream is transferred to the ammonia recovery circuit. Ammonia is dosed for pH control in the cobalt and nickel SX circuits. The ammonia forms a soluble ammonium sulphate neutralization product, which reports to the SX Raffinate stream. The TATI ammonia recovery circuit uses vibrating mills to contact the ammonium sulphate with calcium oxide (fine quicklime). The ammonium sulphate is converted to CaSO4 (fine gypsum precipitate) and aqueous ammonia. A series of seeded reaction tanks is used in the BMR design to further encourage crystal growth. The slurry stream is then pumped into a steam stripping column, where aqueous ammonia is stripped to the column overheads and condensed to a 10% solution for reuse. The recovered aqueous ammonia is topped up with anhydrous ammonia to account for losses and recycled for pH control in the SX circuits. The barren gypsum slurry reports to the tailings dam, where supernatant solution is recovered for use as process water in the solid liquid separation step. The Tati reagent selection Reagent types The acid neutralization in the SX circuits of hydrometallurgical facilities generally utilize ammonia, sodium hydroxide or sodium carbonate. The choice of pH control reagent is largely driven by economic factors, logistics and availability. Ammonium hydroxide Ammonia is supplied as liquid anhydrous ammonia, which is stored under pressure. The anhydrous ammonia is diluted in water to form aqueous ammonium hydroxide; the solution is easier to handle and simplifies dosing and pH control. The ammonium hydroxide neutralizes the sulphuric acid in the SX circuit and produces a soluble ammonium sulphate product, as shown in Equation [1]. DEVELOPMENT OF THE TATI ACTIVOX® BMR AMMONIA RECOVERY CIRCUIT

307

305-312_vdBerg:text

2/13/09

10:56 AM

Page 308

[1] The ammonium sulphate can be crystallized and sold as fertilizer, or reacted with quicklime to recover the ammonia for reuse while forming a gypsum by-product. Sodium hydroxide Sodium hydroxide is generally provided as a 50% solution and stored in stainless steel tanks. It is similarly diluted to lower concentrations to simplify usage. The sodium hydroxide reacts with sulphuric acid to produce a soluble sodium sulphate product, as shown in Equation [2]. [2] The sodium sulphate is typically crystallized, as is the case of Rustenburg Base Metals Refinery, and sold for use in detergents, Kraft sulphate pulping, and glass manufacturing. Sodium carbonate Sodium carbonate is generally delivered as an anhydrous solid and dissolved in tanks to around 30% solution. These tanks are mild steel, and must be heated above 36°C to prevent heptahydrates and decahydrates from precipitating. The sodium carbonate reacts similarly to sodium hydroxide to produce a soluble sodium sulphate neutralization product, as shown in Equation [3]. [3] The sodium sulphate is similarly sold for use in detergents, Kraft sulphate pulping, and glass manufacturing. Ammonia as the reagent of choice for Tati The remote plant location in Francistown Botswana resulted in high transportation costs for all reagents—ammonia, sodium carbonate and sodium hydroxide. The use of ammonia provided the lowest cost option, provided that 90% of the ammonium sulphate product was recovered and reused to reduce the fresh reagent make-up requirements. The option to produce ammonium sulphate crystals, which could be upgraded to a saleable product, was not feasible for the BMR project due to the lack of market in Botswana. Therefore the choice of neutralizing reagent for the Tati plant was selected due to financial drivers, and largely to reduce logistics costs6. The ammonia recovery process The traditional lime boil process The recovery of ammonia from ammonium sulphate is historically referred to as a lime boil. The lime boil consists of two operations, preparing the slaked lime and reacting to recover ammonia, as discussed by Regan (1999)7. Typically quicklime is reacted with water in a slaker to form milk of lime. The reaction is exothermic and the solution temperature increases. The milk of lime is then pumped to a stirred reactor tank containing the ammonium sulphate solution. The ammonium sulphate and lime mixture reacts and is brought to the boil by direct steam injection. Ammonia has a lower partial pressure than the solution and will concentrate in the vapour phase. The vapours are condensed and ammonia recovered. 308


305-312_vdBerg:text

2/13/09

10:56 AM

Page 309

The traditional lime boil process equipment is subject to poor operability due to severe scaling and the risks associated with descaling the reactor tanks. Furthermore, the gypsum tends to coat any unreacted lime particles leading to reagent utilizations as low as 50%, resulting in high reagent consumptions and costs (Regan, 1999)7. The Tati ammonia recovery circuit The Tati ammonia recovery developments The ammonia recovery process utilized in the BMR design was developed by Norilsk Process Technology to overcome the main drawbacks of the traditional lime boil process, particularly lime utilization and gypsum scaling, while minimizing steam consumption as discussed by Johnson and Zhuang (1999)4 and later by Harrison (2008)3. The BMR ammonia recovery circuit is presented in the Figure 2 as a block flow diagram. It includes the vibrating mills, reaction tanks, stripping columns, vapour condensers and off-gas scrubbers. The detailed design of the ammonia recovery circuit called for materials of construction that could handle the high chloride levels in the nickel raffinate stream. This necessitated the use of SAF2205 as the primary material of construction for the mill reactors, reaction tanks and stripping columns in contact with nickel raffinate. The stripping column vapours did not contain chlorides and therefore the condensing systems were constructed from 304L stainless steel. The innovations in the BMR ammonia recovery process are intended to manage gypsum scaling. The high scaling probability and process risk demanded that the circuit be designed with 100% redundancy to allow for descaling on line and 95% plant availability. Vibrating mills The barren nickel raffinate solution, rich in ammonium sulphate, reports to the ammonia recovery circuit. The raffinate is preheated to 90°C and fed into a vibrating mill with dry quicklime (CaO) powder. The quicklime reacts exothermically with the aqueous solution to form soluble calcium hydroxide, as shown in Equation [4].

Figure 2. Block flow diagram of the Tati ammonia recovery circuit

DEVELOPMENT OF THE TATI ACTIVOX® BMR AMMONIA RECOVERY CIRCUIT

309

305-312_vdBerg:text

2/13/09

10:56 AM

Page 310

[4] This exothermic reaction maintains the mill temperature and reduces the mill steam requirement. The fine quicklime particles further react with the ammonium sulphate to release 95% of the ammonia into solution while producing gypsum precipitate, as shown in Equation [5]. [5] The use of vibrating mills is unique for this process (Western Minerals Technology, 2000)5 and is intended to reduce gypsum scale from forming on the milling media or the vibrating mill walls. Furthermore, the agitation of the milling media serves to abrade the gypsum coating off any unreacted quicklime particles, improving the reagent utilization to as high as 91% (Harrison, 2008)3. The direct addition of quicklime to the ammonium sulphate solution is a second significant departure from the traditional lime boil process, where the quicklime is first slaked with water. This reduces the amount of water dilution and utilizes the significant lime hydration energy. This results in lower steam consumption, smaller equipment requirements, and improved stripping column ammonia purity. The mills are maintained at a temperature close to the solution boiling point by the reaction energy (and steam as required), thus favouring the liberation of an ammonia rich vapour. The mill off-gas is vacuum extracted to draw any ammonia gas into the ammonia stripping column. Reaction tanks Analysis of the pilot plant stripping column showed scaling on the slurry feed plate during start-up and upset conditions, suggesting that complete gypsum precipitation had not occurred in the vibrating mills during upset conditions. A series of two reaction tanks was included in the design to provide additional gypsum precipitation residence time and to limit scaling to the vibrating mill and reaction tanks. The first reaction tank is seeded with recycled gypsum particles. The seeding serves to discourage gypsum precipitation on the reaction tank walls, agitators, pipes and pumps by improving crystal growth. The seeding of precipitation tanks is a recommended method for reducing gypsum scale on equipment as discussed by Cooper and Slabbert (2005)2. The reaction tank vapours contain some ammonia and are combined with the vibrating mill vapours before reporting to the stripping column. Stripping column The gypsum slurry, containing ammonia in solution, is pumped from the reaction tanks into the countercurrent steam stripping column. The ammonia in solution is preferentially stripped, as shown by Equation [6]. [6] The ammonia and water vapours exit the top of the stripping column and are condensed in a series of condensers. A portion of the condensed ammonia solution is removed as a 10% ammonia product while the remainder is refluxed back to the top of the stripping column to improve the final ammonia concentration. The product ammonia solution is combined with fresh anhydrous ammonia to account for the 10% ammonia losses in the plant. 310


305-312_vdBerg:text

2/13/09

10:56 AM

Page 311

The stripping column, condensers and scrubber were designed by Process Plant Technology PTY LTD1. The stripping column was designed to achieve a 95% stripping efficiency and combined with the 95% reaction efficiency in the vibrating mills, results in an overall ammonia recovery of 90% in the ammonia recovery circuit. The stripping column design also minimized steam requirements and catered for the highly scaling gypsum slurry. The Tati BMR stripping column design was optimized to reduce the plant steam consumption. This was achieved by using a higher liquid to vapour (L/V) ratio than in the Tati Hydrometallurgical Demonstration Plant (HDP) plant. The BMR design tray efficiency was higher than the HDP tray efficiency, due to larger tray spacing, column diameter and improved liquor flows. The final tray efficiency selected for the Tati BMR design was 15%. The theoretical number of stages was initially predicted using a McCabe Thiele diagram, as shown in Figure 3, and later confirmed with ASPEN™ modelling1. The theoretical number of stages was corrected using the predicted tray efficiency to calculate the actual number of stages and the design column height. The BMR column discharged slurry into a vacuum flash cooler. The flash cooler incorporated a thermo-compressor, which could recover steam from the discharge slurry and reduce the live steam requirements by almost 30%. The discharged gypsum slurry is thickened and sent to the tailings pond. The gypsum is settled and the solution recovered for reuse as gypsum saturated process water in the solid liquid separation step. Condensers and scrubber The ammonia vapours are drawn from the top of the stripping column, through a primary cooler and secondary condenser system. The resultant ammonia rich solution proceeds to a product tank, with the non-condensable gas and entrained ammonia vapour reporting to a multi-stage scrubber. The primary condenser was designed to cool the ammonia vapours and preheat the nickel raffinate feed to the vibrating mills. The secondary condenser was designed to condense ammonia vapours while preheating cooling water to provide a hot water utility for use elsewhere. These design constraints complicated the condenser and scrubber design. The BMR design utilized a three-stage scrubber, where the first two stages utilized dilute ammonia

Figure 3. McCabe thiele diagram for ammonia-water system with Tati operating parameters

DEVELOPMENT OF THE TATI ACTIVOX® BMR AMMONIA RECOVERY CIRCUIT

311

305-312_vdBerg:text

2/13/09

10:56 AM

Page 312

condensate to improve the overall ammonia recovery while meeting the cooling constraints. The final scrubbing stage used sulphuric acid to reduce the ammonia concentration in the off gas to allow venting to atmosphere. Conclusion The choice of neutralizing reagent for the SX circuits affects the overall plant design. The logistic and economic drivers for BMR favoured the use of ammonia integrated with an ammonia recovery circuit. The ammonia recovery circuit incorporated several new technologies over the traditional lime boil operations, resulting in higher quicklime utilizations, improve plant availabilities, smaller equipment, lower energy requirements, and improved management of gypsum scale. References 1. BRADBURY, S. Interview by author, Johannesburg, 27 August 2008. 2. COOPER, R.M.G. and SLABBERT, W. Designing for Scale Control in Hydrometallurgy. CHEMECA 2005, Brisbane, 25–28 September, 2005. 3. HARRISON, R. Norilsk Process Technology’s ammonia recovery. Alta 2008 NickelCobalt Sessions, Perth, 16–18 June, 2008. 4. JOHNSON, G.D. and ZHUANG, Y. Western Minerals Technology’s ammonia regeneration process. Alta 1999 Nickel-Cobalt Pressure Leaching & Hydrometallurgy Forum, Perth, 11–12 May, 1999. 5. JOHNSON, G.D., ZHUANG, Y., and WESTERN MINERALS TECHNOLOGY PTY LTD. Ammonia Recovery. International publication number WO 00/41967. 2000. 6. NEL, G. Confirmed by e-mail correspondence with author, Johannesburg, 18 August 2008. 7. REGAN, T.A. An investigation into ammonia recovery by lime boiling. M.S. thesis. Department of Chemical Engineering, Univ. of Queensland. 1999.

Dylan van den Berg Process Engineer, Hatch, South Africa Dylan is a process engineer with two years of experience in the design and engineering of mineral processing, pyrometallurgical and hydrometallurgical facilities. He has developed specialist technical expertise in areas that include METSIM process modelling and managing mass balance tables. Dylan has been involved as a graduate process engineer on two significant hydrometallurgical execution projects in Africa, namely the Kamoto Redevelopment Project in the DRC (Copper and Cobalt) and the TATI Activox Project in Botswana (nickel, copper, cobalt). Recently Dylan has been involved in several concept level studies for both pyrometallurgical and hydrometallurgical projects, including the Eastplats furnace expansion Project in Marikana (PGM smelter/converter) and Akanani Project (PGM and base metal refinery/roaster/smelter/converter). 312


313-328_Khan:text

2/13/09

3:42 PM

Page 313

KHAN, A.F., T. SPANDIEL, VAN SCHALKWYK, T., and RADEMAN, J.A.M. pH advanced process control solution for Impala BMR first stage high pressure acid-oxygen leach. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

pH advanced process control solution for Impala BMR first stage high pressure acid-oxygen leach A.F. KHAN*, T. SPANDIEL*, T. VAN SCHALKWYK†, and J.A.M. RADEMAN‡ *Impala Platinum Ltd Nickel Solutions CC ‡CSence Systems (Pty) Ltd †Blue

The CSense advanced process control (APC) solution’s main objective was to improve the stability of the pH in the first stage leach process thereby improving nickel and iron extraction efficiencies and reducing the base metal (BM) content in the platinum group metals (PGM) concentrate. By improving the stability of the control of the pH on the first stage leach it had the corresponding effect of improving the Ni extraction efficiency by 0.5% and the Fe extraction efficiency by 3.3%. The system relieved the operators of many decisions that were virtually impossible to make given the complex, variable and real-time nature of the processes in their charge. On the operational side, the operators understand the APC system and they trust it. Another benefit is the reduction in pH peaks in the autoclave, which can oxidize certain elements whereby they become difficult, if not impossible, to leach. These elements go right through the process and end up contaminating the PGM solids, with the result that the entire batch has to be recycled through a lengthy and costly processing pipeline. Introduction Impala Platinum’s Base Metals Refinery in Springs gets its raw material from the company’s mining, concentrating, smelting and converting facilities in Rustenburg. The BMR then removes as much of the base metals as possible and sends the PGM (Platinum Group Metals) concentrate to the PMR plant for further processing. The base metals are refined and sold separately to maximize the conversion of raw material into revenue (Figure 1). A background to first stage leach The aim of the first stage leach process is to maximize the dissolution of nickel, cobalt and impurities and to leave copper and PGMs in the solids for treatment in the second stage processes. The subsections that follow describe the first stage leach process together with process chemistry and the process challenges. pH ADVANCED PROCESS CONTROL SOLUTION FOR IMPALA BMR

313

313-328_Khan:text

2/13/09

3:42 PM

Page 314

Figure 1. A simplified flow diagram of Impala’s processes

Figure 2. First stage leach process flow diagram

Process description Figure 2 shows a flow diagram of the first stage leach process. The matte from the mineral processing plants is fed with demineralized water to the ball mill to increase the surface area of the particles for leaching. Tank TK2100 serves as a buffer tank for the milling operation and the feed tank to the first stage circuit. Pulp density is adjusted in tank TK2102 with spent electrolyte solution. The spent electrolyte solution is return solution from the copper electrowinning section. The slurry from tank TK2102 is pumped to the first compartment of the autoclave. Currently two autoclaves operate in parallel. Steam is added to the first compartment to maintain the required temperature and oxygen is added to maintain pressure as well as to oxidize the sulphides to sulphates in the presence of sulphuric acid from the spent electrolyte. The spent electrolyte also serves to control pH in the autoclave. The overall reaction mechanism could roughly be divided into three stages (as determined by a batch leaching experiments by Rademan, 1995) where the major reactions occurring in each stage of the leaching process differ, i.e.: • Stage I (10–40 minutes)—the cementation of copper and the leaching of nickel from the alloy phase and out of the Ni 3 S 2 phase. Refer to reactions Equations [1]–[7] as determined by Rademan (1995) and Rademan, et al. (1999). [1] [2] [3] 314


313-328_Khan:text

2/13/09

3:42 PM

Page 315

[4] [5] [6] [7] • Stage II (40–160 minutes)—the selective leaching of nickel to form various Ni-S mineral phases and the simultaneous leaching and cementation of copper to form the various CuS mineral phases. Refer to reactions Equations [8]–[13] as determined by Rademan (1995) and Rademan, et al. (1999). [8] [9] [10] [11] [12] [13] • Stage III (160–300 minutes)—the simultaneous leaching of nickel (to form NiS and Ni 3S 4) and copper (to form and CuS). Refer to reactions Equations [14]–[18] as determined by Rademan (1995) and Rademan, et al. (1999). [14] [15] [16] [17] CuS is oxidized by O2 (Equation [18]) to form Cu2+ and SO42-. [18] The principal reactions occurring in the initial stages of the first stage leach are the reactions of nickel alloy (Ni) and heazlewoodite (Ni3S2) with sulphuric acid (H2SO4) and Cu2+ in solution in the presence of oxygen (O2) to form copper sulphide and nickel sulphate. For control purposes the process needs to be controlled as close to the end of Stage II as possible at all times. Process challenges The required iron (Fe), an impurity, in the matte is below 1%. However, matte batches delivered to the BMR sometimes have more than 1% Fe. The Fe needs to be leached in the first stage process and is then removed in the jarosite circuit. When this does not happen, the Fe finds its way to the PGM circuit where it is very difficult to leach out and therefore constrains the downstream processes. pH ADVANCED PROCESS CONTROL SOLUTION FOR IMPALA BMR

315

313-328_Khan:text

2/13/09

3:42 PM

Page 316

The problem experienced in the first stage leach process is that the process controllers are not always able to cope efficiently with the apparently capricious behaviour of the process. This could be due to: • a lack of sufficient understanding of the critical control elements of the process • having to cope with plant emergencies and breakdowns and not being able to closely monitor the process (specifically pH), and • the fact that they have, most of the time, two processes to monitor continuously, i.e. two autoclaves. Because most of the process controllers control the process by experience and feeling, which inherently means that different operators will control the process differently. Therefore, this causes an unstable process from the one shift to the next. Nickel (Ni) in the matte is in the region of 47% and is primarily leached in the first stage process. Therefore it is crucial to have a high Fe and Ni extraction efficiency in the first stage leach process. It is known that pH in the region of 1.8 and 2.2 serves to offer good extraction efficiencies. pH was manually controlled by process operators with some shifts performing better than others. Considering the myriad reactions taking place in the first stage leach process as well as the other control variables, the challenge was to control pH within the specification limits in real time. Samples are taken hourly from compartment nos. 1 and 4 of the autoclave for analysis of the metals content (Ni, Cu and Fe). pH samples are taken at shorter intervals to determine the pH of the pulp in these two compartments because it is the primary control variable. Depending on the pH, the operator will vary the spent electrolyte flow rate to the autoclave or adjust the pulp flow rate to the autoclave. In certain instances the operator will also increase the pulp density in the feed to try and make up for lost production to the detriment of the efficiency of the process. The performance improvement design A feasibility study was conducted to define the boundaries of the problem, assess current status and to present a solution design before starting with the implementation. Feasibility study The problem experienced in the first stage leach process is that the process controllers are not always able to cope efficiently with the apparently capricious behaviour of the process. To provide the context of the complex behaviour each parameter used as part of the solution and its effect on the process are briefly discussed below: Process disturbance variables • Matte composition—the variation in the composition (amounts of the different elements) in the feed matte will have an effect on the leaching process to a varying degree, depending on the actual increase or decrease of a specific element. For example, if the feed matte contains a higher concentration of iron (Fe) it will result in a higher concentration of iron, in either the leach discharge solution or in the solids (depending on the control efficiency of fist stage leach). • Spent electrolyte solution composition—variations occur in the acid concentration of the spent electrolyte solution coming from the copper electrowinning section, as well as in the concentration of other ionic species (Ni, Cu, Fe, Co, etc.). The variation in the acid concentration will influence the rate of chemical reactions taking place in the autoclave. 316


313-328_Khan:text

2/13/09

3:42 PM

Page 317

The variation in the ionic species will have the same effect on the leaching process as does the variation in the matte composition. Furthermore, it is possible that ammonium ions (NH4+) in the spent electrolyte solution will form jarosite precipitates under high pH conditions in the first stage leach, making it almost impossible to leach out in the latter stages of the process, and it will eventually result in high Fe in PGM (platinum group metals) concentrate product to PMR (platinum metals refinery). • Pulp feed rate—the feed rate to the autoclave should be set at a setting that is believed to be the optimum for the maximum throughput while obtaining the desired degree of leaching. Currently the feed rate is sometimes varied by the operator, to help to get the pH within the required control range more quickly, but mostly when downstream bottleneck conditions occur. Process state variables • Pulp density—the pulp density is a very important factor in the reaction kinetics of the pulp, because for lower or higher pulp densities the leaching will either be more, or less efficient. More importantly, this will affect the pH in the autoclave. Ultimately the pulp density in the first compartment in the autoclave needs to be controlled. The design requirement was for a pulp density of minimum 1.35 kg/m3 to a maximum of 1.50kg/m3 in the autoclave. A too high pulp density will result in high pH values and too low pulp density will result in too low pH values. • Cu in solution—the Cu in leach solution is necessary for the cementation reaction with metallic Ni, but an oversupply of Cu in solution negatively affects downstream processes. The Cu in solution is primarily supplied via the spent electrolyte that can contain high concentrations of Cu. Adjustable variable • Spent electrolyte flow rate—the spent electrolyte flow rate to the autoclave is the primary variable to control the pulp density and indirectly the pH in the autoclave. The spent electrolyte flow rate is also adjusted by the operator to achieve the desired pH in the first compartment of the autoclave that ultimately influences the performance of the process. Target variable pH in 1st compartment of autoclave—the pH in the first compartment of the autoclave is controlled by the addition of spent electrolyte solution to the autoclave. Furthermore, it is assumed that if the pH (and pulp density) in the first compartment is correct and the standard process conditions exist, the pH in the fourth compartment will be correct. Therefore, the primary parameters, from the above list, that lead to variation of the pH in the first stage leach are: • Pulp density • Spent electrolyte flow rate to the autoclave • Pulp feed rate to the autoclave. PID control loops Proportional-integral-derivative (PID) control loops are single-input single-output controllers based on first order Laplace transform calculations. PID loops are the norm for forming the base layer control of any industrial plant. Therefore, it is important to ensure that PID control is used as far as possible, and that the current PID control is maintained properly. pH ADVANCED PROCESS CONTROL SOLUTION FOR IMPALA BMR

317

313-328_Khan:text

2/13/09

3:42 PM

Page 318

Sampling and measurement A key requirement for implementation of automated advanced process control (APC) solution is that the target measurement is available online and timeously. An online pH sampling system has been developed by Impala in-house, as shown in Figure 3. pH sampling sequence: • Sample and the drain valves open for 3 seconds • The drain valve closes but the sample valve remains open for a further 4 seconds • Sample valve closes • All valves remain close for 50 seconds giving the solution in the pot time to settle and the pH transmitter opportunity to read the sample value • The reading from the transmitter is stored into the correct PLC address location for the time of sample taken. (Each time interval sample for the 10 minute sequence has its own address allocated and displayed on Scada) • The drain and flush valves open simultaneously for 6 seconds, flushing out the sapling pot • Drain valve closes but the flush valve remains open for another second before closing • The pot should now be filled with water waiting for the next sample to be taken. The sampler sequence is triggered from the PLC clock every 10 minutes, starting on the hour. Performance benchmark The average pH and standard deviation values for the autoclaves 2110C and 2110D are shown in Table I. The average pH in the first compartment of both autoclaves is acceptable, but the problem is rather the high standard deviations, i.e. > 1. The high standard deviations are also an indication of the reactive control that operators have to use to bring the pH back to the target value, which is roughly around 2.0. The average pH in the fourth compartment of autoclave 2110D poses a problem as this average of 3.4 is excessively high and will lead to insufficient leaching (Rademan, 2005). This ‘unleached’ material is then propagated through the circuit to the final PGM concentrate that is sent to the PMR.

Figure 3. Picture of actual pH sample pot arrangement on the autoclave

318


313-328_Khan:text

2/13/09

3:42 PM

Page 319

Table I pH averages and standard deviations for autoclave 2110C and 2110D

Average pH #1 Standard deviation pH #1 Average pH #4 Standard deviation pH #4

2210C

2210D

2.2 1.16 2.3 0.82

2.4 1.23 3.4 1.49

Figure 4. pH in #1 distribution histogram for autoclaves 2110C and 2110D

Figure 4 is an indication of the distribution of the pH values in the first compartment of autoclave 2110C (top) and 2110D (bottom). The objective of the proposed pH control system would be to reduce the tailing of the pH to the high side (as marked on graph). The reduced variation in pH in the first compartment should lead to reduced pH variation in the fourth compartment. The pH in the fourth compartment of the autoclave is a direct result of the pH in the first compartment. If one analyses the average pH values with the daily extraction efficiencies for Ni, Cu and Fe, certain trends can be recognized: • The lower the average pH value the higher the Ni extraction • A large difference between the pH average on the #1 and #4 compartments result in a low Fe extraction • A low Fe extraction gives rise to a high Cu cementation rate (high negative value). The lower the average pH value, the higher the Ni extraction. This confirm the findings by Rademan (1995) that for higher acid concentrations, the higher the Ni extraction. Therefore, Ni extraction is purely a function of acid concentration or pH. A negative outcome of too high acid concentration is that Cu might start leaching, i.e. the kinetics for Cu extraction will increase. A large difference between the pH average on the #1 and #4 compartments result in a low Fe extraction. This is an effect caused by unstable pHs were the operator has to decrease the pH in the first compartment significantly to quickly get the pH in the fourth compartment under pH ADVANCED PROCESS CONTROL SOLUTION FOR IMPALA BMR

319

313-328_Khan:text

2/13/09

3:42 PM

Page 320

control. This leads to an excessive amount of spent electrolyte being added in the first compartment, where the Cu in solution has to be cemented first before any significant Fe leaching can occur. The end result is that only Cu cementation occurs with the associated Ni leaching, but not much Fe is leached in the process. This then results in the phenomenon where a low Fe extraction gives rise to a high Cu cementation rate (high negative value). The objective is to improve the pH control on the first stage leach process. The first step is to stabilize the control as far as possible by trying to avoid significant pH variations. The second step is to optimize the process to increase throughput and improve quality. Control philosophy Based on the results and discussions above, a practical and low risk control solution can be developed. However, in certain areas it will also require a change in the current operating philosophy. In principle the philosophy is to automatically manipulate the spent electrolyte flow rate via the controller, leaving the pulp feed rate to the autoclave as a variable that the process controller can adjust. The proposed control philosophy is to vary as few parameters as practically possible. • The temperature and pressure in the autoclave should be set to 145°C and 450 kPa, respectively and should not be changed • The pulp density in tank 2102 should be kept constant as far as possible • For increasing or slowing down production, the pulp feed rate to the autoclave should be adjusted • The spent electrolyte flow rate should be used as the parameter for primarily controlling the pH in the autoclave. In summary, change the feed rate setpoint for increasing or slowing down production and change spent electrolyte flow rate to control the pH. By varying any of the parameters unnecessarily it becomes exponentially more difficult for the process controller to maintain a constant pH in the first stage leach process. The controller will be able to adapt for variations in the pulp feed rate to the autoclave. The controller will also be able to adapt for any change in the variables mentioned above, but it will result in less stable pH control. The control philosophy for the advanced controller is schematically shown in Figure 5. The controller will consist of two components, i.e. a feedback control component and a feed forward control component. The feedback control component will make use of the latest available pH measurement and implement set point changes on the spent electrolyte flow rate through a fuzzy controller. The feed forward control component will calculate the required amount of spent electrolyte given the current pulp density in tank 2102 and the pulp feed rate to the autoclave. A rule based method will be followed to combine the feed forward and feedback suggested changes to the spent electrolyte flow rate to achieve optimal pH control in the first compartment of the autoclave. Controller design and simulation Process and instrumentation data were collected from the plant historian. The data were analysed to establish the following: • Scope and opportunities for process control and automation improvements • Base layer control stability • Equipment health and constraints • Control element operating range • Current operating and control philosophy • Sufficient step test data to perform system identification. 320


313-328_Khan:text

2/13/09

3:42 PM

Page 321

Figure 5. Proposed advanced pH process control solution for the first stage leach process

Figure 6. CSense simulation—structure layout

Plant data analysis revealed that there was a real need for higher level automation of the autoclave since regular operator action was required to maintain plant stability. Due to regular process step changes by the operators, sufficient step test data were available to perform system identification. System identification was performed (Ogunnaike and Ray, 1994; Seborg et al. 1989) and dynamic process models for the process were obtained. This was used to construct a simulation of the process that in turn was used for control system design (Ogunnaike and Ray, 1994; Juuso, 2007). The initial modelling and control system design was done in Matlab after which it was migrated to the implementation platform, CSense, where further simulation testing was performed before commissioning of the control system. Figure 6 shows the simulation configuration that was used in CSense. Commissioning of the control system was rapid since control system architecture and controller parameter optimization was done upfront. Figure 7 describes the relationship between the pH setpoint, feed flow rate and copper concentration. For higher copper concentrations a higher pH is preferred with a maximum pH of 2.5. pH setpoint is adjusted based on autoclave throughput and discharge copper content according to Figure 7. Figure 7 was formalized based on operator experience. The control system consists of the following elements: • Feed forward control on spent electrolyte proportional to the autoclave feed rate • Fuzzy controller feedback control (Reznik, et al. 2000) on the pH measurement by manipulating spent electrolyte flow rate • Feed forward and feedback controller speed adjustment based on the spent electrolyte acid concentration • pH setpoint selection based on feed rate and Cu concentration • Logic for that caters for exception process conditions. pH ADVANCED PROCESS CONTROL SOLUTION FOR IMPALA BMR

321

313-328_Khan:text

2/13/09

3:42 PM

Page 322

Figure 7. pH setpoint as function of feed flow rate and Cu concentration in first compartment

Figure 8. First stage summary page on SCADA

SCADA interface The SCADA pages were modified to give the process controller feedback from the APC. The buttons on the top right-hand corner of the page indicate the status of the controller. From the first stage summary page one can navigate to the APC page by clicking on the header text for pH values. Figure 8 shows the first stage summary page from which the APC pages can be accessed. Figure 9 is an illustration of the SCADA mimics that the operator can view the samples, stop and start the pH controller and do some basic troubleshooting. On the APC page the various parameters used by the APC can be viewed. The pH values obtained for the previous hour can also be seen. Commissioning of the APC began in mid-August and lasted four days. This reduced commissioning period was achieved through good simulation and presentation exercises. The commissioning did not interrupt the daily operation of the process. Minor changes were required for better performance of the control system. A value for the concentration of acid in the spent electrolyte (SE) was added as the concentration of the acid in the SE affects the leaching kinetics. 322


313-328_Khan:text

2/13/09

3:42 PM

Page 323

Figure 9. SCADA mimic of pH sampler and advanced process control (APC) solution

On these mimics a suggested setpoint value, target pH value, APC communications status, spent control faceplate, matte feed value and Cu value can be obtained. • SE flow suggested setpoint—this setpoint value is fed back from the advanced controller and will replace the setpoint of the spent flow control should the APC controller be turned on. The value is determined by calculation in the APC controller • APC Target pH value—this value is determined by the APC controller taking into account the feed rate of matte into the autoclave and also the Cu value of solution in first compartment of the autoclave • SCADA-APC communications status—this is also called the heartbeat of the APC controller. Should communication be lost between APC controller and the plant PLC, this status will indicate an error. If this error or error indication between RUN/STP buttons (Bad Data) activates, the controller will automatically be turned off and an audible alarm will activate to inform the process controller of this action. When this happens, the last known setpoint will remain active in normal auto mode and can be changed by the process controller if required • Actual SE flow control—this is the loop faceplate to control the flow rate of spent into the autoclave. The setpoint will automatically be adjusted if the APC controller is turned on. The setpoint will then be the same as the suggested setpoint value from the APC controller. The process controller must enter the setpoint value for flow when the APC controller is turned off • Matte feed to autoclave—this value is the total matte feed to the autoclave from the air pump feed settings • Cu value in first compartment—this value represents the Cu as sampled in the first compartment of the autoclave. The value must be entered by the process controller to adjust spent setpoint accordingly. This value is used in the calculation of pH setpoint and needs to be updated when significant changes occur in copper values • TK0313 FA—this value gives spent electrolyte acid concentration. It is to increase or decrease the response speed of the APC feedback controller. When spent acid concentration is high, the APC controller is slowed down and vice a versa pH ADVANCED PROCESS CONTROL SOLUTION FOR IMPALA BMR

323

313-328_Khan:text

2/13/09

3:42 PM

Page 324

Figure 10. Typical control infrastructure for leaches domain

• Trends—trends are also added to view historical data of APC controller outputs. These trends can be viewed by selecting the pop-up trend buttons located on the mimic. Controller limitations are its dependency on accurate measurements and on the operator to enter Cu assays as they become available. Server infrastructure layout and OPC connection A server is located in the process control server room and connected on the process control network is used as the CSense server. It communicates with the Citect SCADA servers via OPC to allow control via SCADA mimics (see Figure 10). The server is supported by a Dell maintenance agreement providing 4 hours’ turnaround time and a cold desktop standby is available in the event of failure. A backup copy of the control blue print is made every time an audit is conducted on the system. Although the CSense APC server has not failed since installation two years ago, Impala BMR is considering a hot server standby as more advanced process controllers are being installed in the plant to control critical process parameters. Project management The project was managed using the guidelines of the project management body of knowledge (PMBOK) by the PMI (Project Management Institute) and managed by a PMP (project management professional). A well developed capital application backed by a good feasibility study, choosing the best project team, having regular project meetings, managing change as the project progressed, involving stakeholders, good communication, and rewarding team efforts contributed towards a successful project that was completed on time and within budget. Informal hands on training were provided for the first stage process operators and the supervisor during the commissioning period. The functioning and the operation of the APC were explained. Formal training was provided on CSense Architect for engineers and instrumentation technicians. Work procedures have been developed for the use of the advanced controller via the Citect SCADA. The work procedures are a quality document and registered on the SAP system. The work procedure is used by operational personnel in conjunction with the leach domain operational work procedures. 324


313-328_Khan:text

2/13/09

3:42 PM

Page 325

All stakeholders were involved in the project from the onset. The presence of a control engineer on site during the commissioning period also helped with change management. The success of the implementation of the controller could be seen immediately with improved pH values. The plant manager—leach as well as the leach supervisor were kept fully abreast of the progress of the project and also participated in project progress meetings. Buy-in from the leach supervisor was instrumental in getting the process controllers to accept the new controller. The immediate improvements in the pH values in both first stage autoclaves resulted in happy customers. The extent of the satisfaction of the end user is remarkable as they would not allow for the controller to be switched off when a request was made to conduct a before/after analysis. Performance results Data gathered during the scope study (termed ‘old data’) were used to compare with data obtained after controller implementation. The analysis focused on proving the following: • Better pH values • Improvement in Ni and Fe extraction efficiency • Reduction of base metal (BM) content in the PGM concentrate. The control solution’s main objective was to improve the stability of the pH in the first stage leach process thereby improving nickel and iron extraction efficiencies and reducing the base metal (BM) content in the PGM concentrate. All data refer to corresponding periods in October-November 2005 and October-November 2006. The standard deviation for pH decreased from 1.16 to 0.82 and 1.23 to 0.74 in Autoclaves AC2110C and AC2110D respectively. By improving the stability of the control of the pH on the first stage leach it had the corresponding effect of improving the Ni extraction efficiency by 0.5% and the Fe extraction efficiency by 3.3%, as shown in Table II (Rademan, 2007). The BM content in the PGM concentrate was lower for the period examined. Nickel in the concentrate was reduced from 1.93% to 1.46% and iron from 5.58% to 3.75%. The standard deviation for BM in the PGM concentrate was lower for the period. A net improvement of 1.4% in the total PGM concentrate despatched to the PMR was achieved by better control of the pH in the first stage leach autoclaves. These results indicate conclusively that the CSense pH controllers have improved the operation of the first stage leach significantly. Controller performance monitoring For support and ongoing monitoring an e-mailed report using the CSense platform is made available to key stakeholders of the first stage process. The report is emailed at 03:00 every morning and provides the graphical performance of the pH control on both first stage autoclaves for the previous day’s 24 hours. The report offers assistance with detection of deviations and observance of good performance. Figure 11 shows an example of the daily report. Table II Extraction efficiencies for Ni and Fe in corresponding periods in 2005 and 2006 Extraction efficiency

Ni Fe

2005

2006

Average

Std deviation

Average

Std deviation

91.2% 85.2%

1.70 5.69

91.7% 88.5%

1.35 6.97

pH ADVANCED PROCESS CONTROL SOLUTION FOR IMPALA BMR

325

313-328_Khan:text

2/13/09

3:42 PM

Page 326

Real-time monitoring of the controller performance is done via SCADA. The haphazard adjustment of spent electrolyte flow rate has been reduced. In the past, process controllers used to change the pH setpoint every 20 minutes, provided that they were present in the control room. Now, the pH setpoint is changed automatically, every 10 minutes, by the APC. Figure 12 shows an impressive trend of the pH in autoclave AC2110D. The green line represents the pH in the first compartment and the red line is the pH setpoint. The yellow line represents the pH in the fourth compartment. The blue line displays the changes made on the spent electrolyte flow rate. Change management The manner in which the project was implemented enhanced personnel confidence in the control system. A control engineer spent time on site, not only to gain an understanding of the process but also to explain the new control system to those who were directly involved. Any questions or uncertainty experienced by the operators were quickly resolved.

Figure 11. Daily controller performance report

Figure 12. pH and spent electrolyte SCADA trend

326


313-328_Khan:text

2/13/09

3:42 PM

Page 327

Before the controller was implemented, a setpoint for the spent electrolyte flow was suggested to the process controller. The process controller then manually adjusted the spent electrolyte flow. In this manner the process controller could assess the performance of the APC and also indicate it was achieving the correct result. On the success of this process the APC then assumed automatic control of the spent electrolyte flow rate. The leach process supervisor and the first stage process controllers have verbally expressed confidence in the system. The project was also primarily motivated on the basis of improving the PGM concentrate quality and this was achieved, making the project a resounding success. ‘For me, the CSense-based APC is like an extra process controller on that section of the plant. Previously, if a process controller was to leave the plant for any reason, there was no one left to control the pH. Today the APC takes care of that. From a manual system, this has evolved to a system where pH samples are evaluated in real time (instead of through a lab) and you can see the result of these evaluations changing the process as it happens.’ Arnoldus (Vossie) Vosloo, Leach Plant Supervisor, Impala Base Metals Refineries (Khan, et al. 2008). Conclusion The system relieved the operators of many decisions that were virtually impossible to make given the complex, variable and real-time nature of the processes in their charge. On the operational side, the operators understand the process and the APC system. pH is the control parameter and the CSense-based APC currently in use has resulted in operators trusting it and using it to their advantage. Another benefit is that the system allows the operator time to look after the multitude of critical process variables and equipment. Realized benefits: • Through less iron and nickel content, the PGM concentrate grade is much improved • From the performance analysis: – The pH variation has been reduced from 1.2 to 0.7 (an improvement of 40% in pH stability) – The nickel extraction efficiency has increased by 0.5% – The iron extraction efficiency has increased by 3.3% – The PGM grade has increased by 1.4% These numbers may seem small, but they are extremely significant to the financial benefit of all our refining processes. Other benefits include: • A reduction in pH peaks in the autoclave, which can oxidize certain elements whereby they become difficult, if not impossible, to leach. These elements go right through the process and end up contaminating the PGM solids, with the result that the entire batch has to be recycled through a lengthy and costly processing pipeline • Acceptance of the system by operating staff who now also trust it to do the right thing • Stable process control in spite of variations caused by disturbance variables • Indirectly limiting environmental emissions • Real savings affecting Impala’s business bottom line. Acknowledgements We want to acknowledge Dauw Venter and Dennis Lee for the assistance in commissioning the APC solution as well as for the write up and picture of the sampling system. We would also like to acknowledge the operational staff that assisted and contributed in making this a very successful project, i.e. Selilo Semosa and Vossie Vosloo, as well as Impala Platinum Ltd for allowing the publication. pH ADVANCED PROCESS CONTROL SOLUTION FOR IMPALA BMR

327

313-328_Khan:text

2/13/09

3:42 PM

Page 328

References JUUSO, E.K. Modelling and Simulation in Advanced Process Control, Control Engineering Laboratory, Department of Environmental and Process Engineering, University of Oulu, Finland. 2007. KHAN, A.F., SPANDIEL, T., VOSLOO, A., and SEMOSA, S. Advanced pH controller helps Impala BMR. MMS Mag, June 2008, pp. 20–22. OGUNNAIKE, B.A. and RAY, W.H. Process Dynamics, Modeling and Control. Oxford University Press, New York and Oxford. 1994. RADEMAN, J.A.M. The simulation of a transient leaching circuit. PhD thesis, University of Stellenbosch, South Africa. 1995. 261 pp. RADEMAN, J.A.M. Impala BMR First Stage Leach pH control, Scoping Study CSKM05100263-1.1, 15 December 2005 (Internal document), 2005. 49 pp. RADEMAN, J.A.M. Impala BMR First Stage Leach pH control, Performance Evaluation. 2007. CSKM0510-0263-3.3, 12 March 2007 (Internal document), 10 pp. RADEMAN, J.A.M., LORENZEN, L., and VAN DEVENTER, J.S.J. The leaching characteristics of Ni-Cu matte in the acid-oxygen pressure leach process at Impala Platinum. Hydrometallurgy, vol. 52, 1999. pp. 231–252. REZNIK, L., GHANAYEM, O., and BOURMISTROV, A. PID plus fuzzy controller structures as a design base for industrial applications. Engineering Applications of Artificial Intelligence, vol. 13, 2000. pp. 419–430. SEBORG, D.E., EDGAR, T.F., and MELLICHAMP, D.A. Process Dynamics and Control. Johnson Wiley & Sons, New York. 1989.

Abdullah F. Khan Plant Manager - Systems Abdullah Khan is equipped with a chemical engineering degree from The University of Natal. He started his career doing chemical engineering research and then moved on to consulting work in the automation industry. He was thereafter employed by Impala Platinum to lead the implementation of information and control systems across the Refineries. He has served Impala Platinum Refineries for almost 7 years and currently holds the Systems Manager position. Over the years, Abdullah has successfully managed many information and control systems projects that provided significant value to stakeholders. He serves to implement and support these systems as well as to provide strategic direction in the functionalities that are required from such systems to meet business goals. He has gained insightful experience in the fields of APC, MES, LIMS, SCADA, project management, business intelligence, IT infrastructure and systems integration whilst at the same time being exposed to operations management. Impala Platinum has afforded Abdullah the opportunity to gain experience and exposure to the wider management functions within the Refineries which ably equips him to deliver on key performance areas. He is also a project management professional (PMP) and is currently completing his MBA studies. 328


329-340_Konighofer:text

2/13/09

3:43 PM

Page 329

KÖNIGHOFER, T., ARCHER, S.J., and BRADFORD, L. A cobalt solvent extraction investigation in Africa’s Copper Belt. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

A cobalt solvent extraction investigation in Africa’s Copper Belt T. KÖNIGHOFER*, S.J. ARCHER†, and L. BRADFORD‡ *TWP Matomo Process Plant (Pty) Ltd, South Africa †MDM Technical Africa (Pty) Ltd, South Africa ‡Metorex Ltd. South Africa

Processing of copper-cobalt orebodies in Africa’s Copper Belt has received much recent attention with various flowsheet options being developed specifically for the refining of cobalt. The upfront leach and copper refining routes are well understood whereas the refining of cobalt is more complex, requiring numerous impurity removal steps to produce high purity metal. This paper describes an investigation into one of the possible processing steps in the refining of cobalt, namely, cobalt solvent extraction. The paper emphasizes the importance of upstream impurity removal to achieve the required cobalt solvent extraction feed composition. 18% (v/v) Cyanex 272 in an aliphatic diluents is used as the organic phase. The pH profiles in the various stages are evaluated in order to obtain a raffinate containing <10mg/l cobalt and with a low magnesium content reporting to the stripping section. The pH profile is also used to minimize the impurity deportment to the cobalt electrolyte. The effect of adding a pre-neutralization stage before feeding the organic phase to the extraction circuit is investigated. Zinc build-up in the stripping stages is also looked at. It is recommended for the purpose of this study that four extraction stages at an O:A ratio of 1 with a pH profile from 4.9 to 5.7 be employed. The pH is controlled with 40g/l NaOH. A preneutralization stage is required, where the organic phase is contacted with 10 M NaOH. Two scrubbing stages are recommended at an O:A ratio of 40, using the cobalt electrowinning advance electrolyte as the scrub liquor. Three strip stages are to be employed at an O:A ratio of 0.67. Introduction The high commodity prices in recent years have spawned many investment opportunities in this market. The Sub-Saharan Copper Belt, extending from Zambia to the Democratic Republic of the Congo, is rich in copper and cobalt. Recovery of copper from these deposits results in dissolution of cobalt as well as impurities, and in most operations the dissolution of A COBALT SOLVENT EXTRACTION INVESTIGATION IN AFRICA’S COPPER BELT

329


2/13/09

3:43 PM

Page 330

cobalt is promoted by altering the copper leach conditions. Producing copper from the leach liquor is relatively easy and well understood. However, the purification and production of premium grade cobalt metal from a copper-circuit bleed stream is more complex and a variety of flowsheet options can be considered. The interest shown in producing cobalt metal is attributed to the high price that can be obtained for premium grade cobalt metal, which was $34/lb in August 2008 (audited by Pricewaterhouse Coopers LLP)1. The composition of premium and standard grade cobalt metal is shown in Table 1. The difficulty in producing premium grade cobalt metal is the ability to maintain low levels of impurities in the cobalt electrowinning electrolyte in order to achieve a metal that contains >99.83% cobalt. The metal impurities in a cobalt electrowinning process have high upgrade ratios and as such a solution composition, as shown in Table II, is required to meet the metal specification.

Table I Composition of premium and standard grade cobalt metal Element Ni Fe Cu Mn Si, P, Sn, Sb, As, N Al Pb, S Zn Cd Ag Se, H Mg Bi C O

Premium grade (99.83% Co), mg/kg

Standard grade (99.73% Co), mg/kg

<1000 <50 <30 <20 <10 <10 <30 <50 <50 <5 <5 <10 <1 <100 <200

<1500 <100 <50 <50 <20 <10 <60 <100 <50 <5 <10 <30 <5 <200 <250

Table II A typical cobalt electrowinning solution composition Element Zn Cu Fe Ni Element Mn Mg Co Na

330

Concentration, mg/l <3 <2 0 46 Concentration, g/l 2.5 7 55 16



2/13/09

3:43 PM

Page 331

In a copper-processing circuit a bleed stream is required to avoid a build-up of impurities. This bleed stream is the feed to the cobalt circuit and contains various metals, such as iron, aluminium, magnesium, zinc, nickel, manganese and trace amounts of copper. Typical process unit operations to purify the copper bleed in order to achieve the cobalt electrolyte compositions include precipitation of iron and aluminium, copper ion exchange, impurity solvent extraction, zinc ion exchange and cobalt solvent extraction. Figure 1 shows a simplified flowsheet option incorporating cobalt solvent extraction as one of the process steps in the production of cobalt metal2, 7. Cobalt solvent extraction is used to selectively extract cobalt from a purified solution, which typically contains less than 15g/l cobalt and transfers it into a concentrated electrolyte containing 50g/l cobalt. A typical copper-circuit bleed stream composition feeding the cobalt purification circuit is compared to the purified cobalt solvent-extraction feed stream in Table III and shows the extent to which the bleed-stream cation is required. The flowsheet considerations for each element are indicated.

Figure 1. A flowsheet option for Co metal production

Table III Typical copper circuit bleed and cobalt solvent extraction feed compositions Element Copper circuit Cobalt solvent extraction bleed mg/l feed mg/l Co Zn Cu Fe Ni Mg Al Mn

2000 <50 175 1200 <30 2000 700 1400

8000 <2 <2 <2 <25 1100 10 30

Flowsheet considerations for undivided-cell cobalt electrowinning Cobalt upgrade reduces purification equipment size Preferential plating resulting in a stressed deposit Preferential plating resulting in a stressed deposit. Difficulty in stripping from organic phase. Cobalt metal contamination Increases electrolyte viscosity Crystal-lattice inclusion in cobalt cathode Manganese dioxide formation at anode—requires electrolyte bleed. At high concentrations permanganate forms in cobalt electrowinning circuit that oxides the organic phase.

A COBALT SOLVENT EXTRACTION INVESTIGATION IN AFRICA’S COPPER BELT

331


2/13/09

3:43 PM

Page 332

The focus of the paper is on the testwork programme and considerations for the cobalt solvent-extraction component of the process flowsheet and uses Ruashi Phase II as a case study performed at MINTEK research facility. The objectives of the testwork are to: • Produce a raffinate stream containing <10 mg/l cobalt in order to minimize cobalt losses • Limit the amount of magnesium, reporting to the cobalt electrowinning to less than 5% of that contained in the feed to the cobalt solvent-extraction circuit • Test pre-neutralization with 10 M NaOH3 • Determine the zinc build-up within the cobalt solvent extraction strip circuit. Theory In the cobalt solvent-extraction circuit Cyanex 272 in an aliphatic diluent is primarily employed to recover cobalt from solution. Cyanex 272 is a phosphinic acid extractant and metals are extracted through a cation exchange mechanism. It should be noted that a variety of metals can be extracted by this reagent depending on the pH of the aqueous solution with which it is in contact. The overall reaction occurring in solvent extraction is: [1] -

where A represents R2P(O)O and the overbar represents a species in the organic phase. During extraction the forward reaction is predominant, whereas for striping the reverse reaction is favoured due to LeChatelier’s principle. Occasionally to achieve better pH control and to alleviate water balance issues, the stripped organic phase is contacted with NaOH at high concentration before being introduced into the extraction circuit6: [2]5 Testwork Batch testwork The order of metal extraction for Cyanex 272 with increasing pH is well established4 and is as follows: Fe3+>Zn2+>Al3+>Cu2+>Mn2+>Co2+>Mg2+>Ca2+>Ni2+. It is, however, necessary to obtain case specific data using the actual solution for verification purposes and as a basis for further test work. The metal extraction from a feed solution using 18% (v/v) Cyanex 272 in an aliphatic diluent at ambient temperature is shown in Figure 2. It can be seen that the pH for optimum separation of cobalt from magnesium and nickel, while still extracting a high proportion of cobalt, is between 5 and 5.5. Zinc is quantitatively extracted, even at low pH values. Manganese is preferentially extracted, accentuating the need to purify the feed to the cobalt solvent-extraction circuit. Once the pH for extraction is selected, distribution isotherms can be produced in which the feed solution is contacted with organic phase at varying phase ratios at a constant temperature and pH. Distribution isotherm for the extraction of cobalt at pH 5.5 and an operating line are shown in Figure 3. An organic-to-aqueous (O:A) phase ratio of 1 is assumed to produce an operating line that achieves a maximum cobalt loading of about 7.5 g/l in the organic phase at pH 5.5 from a feed liquor containing 7.5 g/lL cobalt. The McCabe-Thiele plot method is used to predict the number of stages required in the extraction circuit. Figure 3 shows the McCabe-Thiele plot at an O:A phase ratio of 1 with pH controlled at 5.5. It can be seen that a cobalt content of less that 10 mg/l can be achieved in 4 stages. 332



2/13/09

3:43 PM

Page 333

Figure 2. pH dependence of metal extraction at ambient temperature (with 18% (v/v) Cyanex 272)

Figure 3. Distribution isotherm for the extraction of cobalt by 18% (v/v) Cyanex 272 at pH 5.5 and ambient temperature

The 18% (v/v) Cyanex 272 is then ‘max-loaded’ to produce a loaded organic that would represent that obtained on the pilot and full-scale plant. This loaded organic is used for conducting stripping distribution isotherms in the laboratory. The loaded organic phase and strip solution (8g/l sulphuric acid) is contacted at various O:A phase ratios at ambient temperature. At high O:A phase ratios contacts are maintained at a pH of <3.5 (as expected on the full-scale plant) with addition of 50g/l H2SO4. The McCabe-Thiele plot method is used to predict the number of stages required for stripping. Figure 4 shows the McCabe-Thiele plot at an O:A phase ratio of 0.67. It can be seen that three stages are required to strip the loaded organic of cobalt. Continuous testwork Based on the batch test work, the start-up conditions for the piloting campaign are chosen. The continuous test work utilizes 11 stages in the cobalt solvent-extraction circuit, namely 5 extraction stages, 3 scrub stages and 3 stripping stages as shown in Figure 5. 40g/l NaOH solution is employed for pH control in the extraction stages and a pH profile as shown in Table IV is recommended for start-up. A COBALT SOLVENT EXTRACTION INVESTIGATION IN AFRICA’S COPPER BELT

333


2/13/09

3:43 PM

Page 334

Figure 4. Distribution isotherm for the stripping of cobalt from 18% (v/v) Cyanex 272 using 8g/l H2SO4

Figure 5. Schematic of cobalt solvent-extraction circuit for continuous test work

Table IV pH Profile used for start up of pilot-plant campaign Stage E1 E2 E3 E4 E5 Sc1 Sc2 Sc3 S S2 S3

pH 4.8 (controlled) 4.5 (controlled) 5.2 (controlled) 5.4 (controlled) 5.5 (controlled) 4.5 4.5(monitored) 4.5(monitored) 3.5 to 4.0 (controlled) (not controlled) (not controlled)

The pH profile in the extraction stages increases from pH 4.8 in the aqueous phase of E1 to a value of 5.5 in the raffinate (E5). This profile ensures that a raffinate of <10 mg/l cobalt is achievable by choosing extraction conditions that are favourable for cobalt (see Figure 2). The co-extraction of magnesium is also enhanced but as the organic phase approaches E1 the pH is lowered, thus forcing a portion of the magnesium from the organic phase before it exits the extraction circuit. The scrubbing stages operate at a high O:A ratio of approximately 40:1 and a pH slightly lower than the extraction circuit using scrub liquor that contains 55g/l cobalt (the advance electrolyte). These conditions provide a large driving force to scrub impurities from the 334



2/13/09

3:43 PM

Page 335

loaded organic phase by displacing them with cobalt in the scrub liquor. The cobalt in the scrub liquor is not intended to load the organic phase to a maximum, but rather to replace impurities. The cobalt-depleted scrub liquor is normally directed to the extraction circuit but can be removed from the circuit for separate treatment if water balance issues are of concern. pH control in these stages is not required as scrub liquor is pre-adjusted to the desired pH value. The extracted cobalt does not exchange hydrogen ions, but rather other metal impurities, and the pH therefore remains constant. The stripping circuit receives spent electrolyte containing approximately twice the amount of hydrogen ions (produced in the electrowinning cells) required for stripping cobalt from the organic. Thus pH adjustment is required only in S1 to make up for any deficit in hydrogen ions. S2 and S3 are not controlled providing that the organic is completely stripped of metal species before being recycled to the extraction circuit. Extraction circuit The extraction efficiencies for cobalt and magnesium and the pH in the final extraction stage (E5) are shown in Figure 6. A reproducible extraction efficiency of 100% for cobalt is achieved for the second half of the campaign. The co-extraction of magnesium averaged 13%, hence substantiating the need for scrubbing in order to achieve less than 5% magnesium reporting to the stripping circuit. Initially cobalt is lost to the raffinate, i.e. low cobalt extraction efficiencies. The initial low cobalt extraction efficiency is attributed to low pH in the final extraction stage (E5). The low pH is due to the release of hydrogen ions (see Equation [1]) from the organic phase during loading and 40g/l NaOH is insufficient for neutralization in this stage. A pre-neutralization stage, where the organic phase is contacted with 10 M NaOH, is employed to achieve better pH control in the final stages of the extraction circuit (see Equation [2]). Once the preneutralization stage is incorporated in the cobalt-solvent extraction circuit, the required raffinate tenor of < 10 mg/l cobalt is consistently achieved. The pH profile automatically increases to 4.9 in E1, to 5.7 in E4. Cobalt slippage occurred even when utilizing 5 extraction stages. The extraction circuit can be reduced to 4 stages if a pre-neutralization stage is introduced.

Figure 6. Extraction efficiencies of Co and Mg and pH values in the final stage of extraction


335


2/13/09

3:43 PM

Page 336

The behaviour of magnesium in the aqueous solution in the extraction stages is shown in Figure 7. Results show that the magnesium concentration increases from E1 to E3. The concentration of magnesium falls from E3 to E4. At high pH values magnesium loads more readily onto the organic phase and is scrubbed off at lower pH values and higher cobalt concentrations present in stage E1. This results in a aqueous-phase recycle that builds up magnesium in the extraction circuit until it eventually leaks out of the circuit via the raffinate stream. Figure 8 shows the magnesium concentration in the organic phase in the circuit. The concentration of magnesium in E4 is high as it is co-extracted into the organic phase, hence the low concentration of magnesium in the aqueous solution in E4. The loaded scrub liquor exists at Sc1 in the aqueous solution. The concentration of magnesium in the organic phase in the scrub circuit is lowered as the magnesium is displaced by the cobalt in the scrub liquor. The scrub stage operates at a slightly lower pH than the extraction stages, thereby allowing magnesium to be scrubbed. Scrubbing circuit The cobalt solvent-extraction circuit is started up utilizing 3 scrub stages and in the later phase of the test work is reduced to one stage. As a result of the high phase ratio in the scrub

Figure 7. Magnesium concentration in the aqueous phase in the extraction circuit

Figure 8. Magnesium concentration in the organic phase

336



2/13/09

3:43 PM

Page 337

section, and due to the limited duration of the pilot campaign, it is not possible to fully optimize the scrubbing performance. This should be conducted during the full-scale plant operation. The objective of achieving less than 5% of magnesium in the feed to the cobalt solvent extraction circuit from reporting to the strip section is achieved with three scrub stages; the results are summarized in Table V. Figure 8 shows that the magnesium concentration in the organic phase on Sc1 is 50% of that at E1, thus showing the effectiveness of the scrubbing stage. Figure 9 shows the cobalt concentration of the organic phase in E1 and the final scrub stage (Sc1). The results show that the cobalt concentration in the organic phase is generally slightly lower in E1 than in Sc1. This is attributed to cobalt from the scrub liquor displacing magnesium on the organic phase in the scrub circuit. Stripping circuit A phase ratio of 0.67 is employed in the strip circuit, which ensures that all loaded metal species are completely stripped from the organic in S3 and that a Δ 5g/l of cobalt is achieved across the strip stages. The pH in S1 is maintained at 3.5 with the addition of 30g/l sulphuric acid. This pH value is ideal for cobalt electrowinning. Figure 10 shows the build-up of zinc in the stripping circuit. The concentration of zinc in the feed to the cobalt solvent extraction is maintained below the analytical detection limit. Figure 11 shows that zinc does not leak into the cobalt electrowinning circuit via the loaded strip liquor but rather that it remains in the strip circuit. A pH value of 3.5 in S1 is conducive to maintain the zinc loaded on the organic phase, thus preventing it from leaking into the electrowinning circuit. Zinc is stripped in S2 and S3 due to the lower pH values in these Table V Percentage magnesium extracted, scrubbed and reporting to the strip liquor Number of scrub stages 3 1

% Magnesium extracted

% Magnesium scrubbed

% Magnesium reporting to strip circuit

18 15

68 39

5 6

Figure 9. Comparison of cobalt concentration in organic phase of E1 and final scrub stage (Sc1)


337


2/13/09

3:43 PM

Page 338

Figure 10. Zinc concentrations in aqueous and organic phase in S1

Figure 11. Zinc concentration in aqueous and organic phase in S3

stages, thus preventing zinc from exiting the strip circuit via the stripped organic, as seen in Figure 9. The zinc in the strip liquor exiting S2 is re-extracted in S1. This internal recycle of zinc gradually builds up, resulting in it eventually leaking into the cobalt electrowinning circuit or remaining loaded on the stripped organic phase, leading to a low extraction capacity of the organic phase and cobalt losses. A major concern is a spike of zinc existing in the loaded strip liquor due to pH deviations in S1. If the pH in S1 decreases to less than 3 then zinc will report to the cobalt electrowinning circuit, contaminating the electrolyte leading to the production of off-specification cobalt cathode. Impurity species build-up in the loaded strip liquor during the pilot-plant campaign. It is thus recommended for a full-scale plant to incorporate a small bleed stream from the strip circuit to remove these impurities in order to avoid accumulation to levels that would compromise the metal product quality. The strip-liquor bleed can be recycled upstream to remove zinc and recover cobalt. 338



2/13/09

3:43 PM

Page 339

Table VI Optimum operating conditions for cobalt solvent extraction Extraction Number of stages Extractant Diluent O/A pH profile pH control Pre-neutralization

4 18% (v/v) Cyanex 272 Aliphatic 1 4.9 to 5.7 40g/l NaOH 10 M NaOH

Scrubbing Number of stages Scrub liquor O/A pH profile

2 Advance electrolyte (~55g/l Co) 40 4.5 (monitored)

Stripping Number of stages Strip liquor O/A pH control in S1

3 (possibly a 4th) Spent electrolyte (50g/l Co) 0.67 30g/l H2SO4

Conclusions and recommendations Extraction circuit It is shown that by increasing the pH profile in the extraction stages results in a marked improvement in the recovery of cobalt resulting in less than 10 mg/l of cobalt in the raffinate. The pH profile is inherently raised by pre-neutralization of the stripped organic with 10 M NaOH. Greater than 99% extraction for cobalt is achieved in the extraction circuit with the high pH profile. A pH profile of 4.9 in E1 to 5.7 in E4 and a phase ratio of 1.0 are recommended for optimum results for the feed solution tested. Magnesium scrubbing is also enhanced due to the decreasing pH profile from E4 to E1. Scrubbing circuit Scrubbing of magnesium is established during the pilot campaign although it is not optimized due to the high phase ratio in the scrub circuit and the short duration of the pilot campaign. The objective of <5% magnesium in the feed reporting to the loaded strip liquor is achieved with three scrub stages and is shown to be possible with one stage, provided that the coextraction of magnesium is limited within the extraction circuit. 68% magnesium scrubbing is achieved within three stages across the scrubbing circuit while only 39% is achieved with one stage. It is therefore recommended to employ a minimum of two scrub stages that can be optimised during the operation of the full-scale plant. Stripping circuit 100% stripping of cobalt in the scrubbed organic phase is achieved in 3 strip stages providing that the pH in S1 is maintained between 3.5 and 4.0 to achieve a delta of 5g/l cobalt across the strip circuit. The pH in stage S1 is controlled by the addition of 30 g/l H2SO4. It is shown that zinc accumulates within the strip circuit. Although zinc in the feed was kept below the A COBALT SOLVENT EXTRACTION INVESTIGATION IN AFRICA’S COPPER BELT

339


2/13/09

3:43 PM

Page 340

analytical detection limit to prevent contamination of the cobalt cathode, the zinc does not exit the strip circuit and over time builds up to concentrations that, with a slip in pH control in the strip circuit, will contaminate the loaded strip liquor and ultimately the cobalt metal. The build-up of zinc in the strip circuit will occur on the full-scale plant and provision must be made for controlled periodic stripping of the organic phase at a low pH. A fourth strip stage can be used for selective stripping of impurities. The strip liquor generated in a fourth stage can be recycled to the cobalt upgrade step (see Figure 1) to minimize cobalt losses. The optimum operating conditions for cobalt solvent-extraction determined in the piloting campaign are given in Table VI. REFERENCES 1. http://cobalt.bhpbilliton.com/ 31 August 2008. 2. SOLE, K. C., FEATHER, A. M., and COLE, P. M. Solvent Extraction in Southern Africa: An update of some recent hydrometallurgical developments, Hydrometallurgy 78, Elsevier, 2005. pp. 52–78. 3. ROUX, L.M., MINNAAR, E., CILLIERS, P.J., BELLINO, M. and DYE, R. Comparison of Solvent Extraction and Selective Precipitation for the Purification of Cobalt Electrolytes at the Luilu Refinery, DRC, The South African Institute of Mining and Metallurgy Base Metal Conference, Symposium Series S47, 2007. pp. 343–364. 4. Cyanex 272® Handbook, Solvent Extraction Reagent, Cytec Industries, pp. 1–37. 5. DAUDINOT, A.M.M. and LIRANZA, E. G. Proceeding of the International Solvent Extraction Conference, ISEC 2002, Johannesburg, pp. 964–969. 6. BOURGET, C. and JAKOVLJEVIC, B. Operational Practices for Cyanex 272 Extractant Circuits, Proceedings of ISEC 2008, vol. I, pp. 447–452. 7. COLE, P.M. and FEATHER, A.M. Processing of African Copper-Belt Copper-Cobalt Ores: Flowsheet Alternatives and Options, Proceedings of ISEC 2008, vol. I, pp. 131–138.

Tanja Könighofer Process Engineer, TWP Matomo Process Plant, South Africa After completing her Masters at the University of the Witwatersrand, Tanja joined Mintek’s Hydrometallurgy Division where she worked on various laboratory and pilot plant studies. Thereafter, she joined TWP Matomo in the capacity of a Process Engineer on the Ruashi Cu/Co project in the DRC. Currently she is involved in the MC plant expansion project for Anglo.

340


341-356_Mulaudzi:text

2/14/09

11:27 AM

Page 341

MULAUDZI, N. and MAHLANGU, T. Oxidative precipitation of Mn(II) from cobalt leach solutions using dilute SO2/air mixtures. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Oxidative precipitation of Mn(II) from cobalt leach solutions using dilute SO2/air mixture N. MULAUDZI*† and T. MAHLANGU† *Hydrometallurgy Division, Mintek, South Africa †Department of Materials Science and Metallurgy, University of Pretoria, Pretoria, South Africa The use of SO2/air gas mixtures as an oxidant to precipitate Mn from Co(II) leach liquors was investigated. The effects of SO2/air ratio, pH and temperature on Mn precipitation were evaluated. It was found that the use of SO2/air gas mixtures resulted in significantly higher Mn precipitation kinetics compared to using air or pure O2 alone. The SO2/air ratio was varied from 0% to 6% SO2 (v/v) in air and similar Mn removals were achieved at 0.75% to 3% SO2 at pH 3. The solution pH was varied from pH 2 to pH 4; Mn precipitation did not increase considerably from pH 2 to pH 3, but increased significantly at pH values higher than pH 3. Cobalt co-precipitation also increased as pH increased, with 1% Co co-precipitation at pH 3. An increase in temperature from 30°C to 60°C also increased Mn precipitation and 100% Mn precipitated at 50°C. Cobalt co-precipitation also increased significantly with an increase in temperature. An activation energy of 25 kJ/mol was calculated from the Arrhenius plot, which is an indication that the precipitation reactions were both chemically and diffusion controlled. XRD analysis showed that Mn precipitated in the form of Mn2O3 instead of MnO2 that was predicted from thermodynamic data. SEM and XRD analysis also revealed that the precipitate consisted mainly of gypsum or bassanite (99%), with the Mn containing phase (< 1%) distributed within the gypsum phase. The co-precipitated Co reported to the Mn phase. Keywords: manganese precipitation; SO2/O2 gas mixtures; cobalt solutions

Introduction Manganese occurs in most cobalt, copper, nickel and zinc ores as an impurity, which reports to the leach solution when the valuable metal is leached from the ore. The leach liquor obtained from these ores often must be purified to an extent before the valuable metal is recovered either by electrowinning or precipitation. OXIDATIVE PRECIPITATION OF MN(II) FROM COBALT LEACH SOLUTIONS

341


2/14/09

11:28 AM

Page 342

Some Mn(II), approximately 2–5 g/l, is desired in Co(II) electrowinning in order to form a protective MnO2 layer on the lead anode surface, which limits corrosion of the anode. Consequently, control of the manganese concentration in the advance electrolyte is not as stringent if present at low concentrations. However, if the valuable metal has to be recovered by precipitation of a relatively high quality cobalt salt, the level of Mn(II) in the purified leach solution should be extremely low, in the range of 10 mg/l or less. There are various methods that are used for the removal of Mn from leach solutions, namely: ion exchange, solvent extraction and precipitation. Precipitation of manganese is a cheaper and simpler alternative compared to solvent extraction and ion exchange. Several research studies have been undertaken to investigate the use of SO2/O2 gas mixtures as an oxidant to precipitate Mn from Zn(II), Co(II) and Ni(II) leach liquors (Menard et al., 2007; Zhang et al. 2002; Zhang et al., 2007). The use of SO2/O2 or SO2/air gas mixtures as an oxidant is attractive because it is cheaper compared to using other strong oxidants such as ozone and hydrogen peroxide. The running costs for precipitating metals with SO2/O2 or SO2/air gas mixtures can be reduced by using scrubbed SO2 from the sulphide ore roasting operations if the plant is processing a sulphide ore. The subject of this paper was to investigate ways to optimize the precipitation of manganese from cobalt leach solutions to produce a less contaminated cobalt solution from which cobalt can be recovered as a high quality precipitate. The effects of the SO2/air ratio, pH and temperature were investigated to determine the optimum conditions for maximum Mn precipitation with reasonably low cobalt losses (<1%). The use of pure oxygen instead of air in the gas mixture was also compared. Theory Thermodynamic considerations The possible reactions of Mn and Co can be predicted from the Pourbaix diagram, Figure 1, which illustrates the regions where the various species that may form predominate. Oxidative precipitation of Mn(II) to MnO2 becomes thermodynamically feasible if the redox potential of

Figure 1. The Pourbaix diagram for 2 g/l Mn(II), 6.5 g/l Co(II) and 4.8 g/l S(IV) at 30°C and 1 atm as predicted by the HSC thermodynamic model

342



2/14/09

11:28 AM

Page 343

the system rises above that of the MnO2/Mn(II) line, represented by the long red line from pH 0 to pH 7. Precipitation of Mn(II) as Mn2O3 becomes thermodynamically possible at pH > 5, represented by the region between the two parallel red lines from pH 5 to pH 7. In the pH range investigated, namely pH 2 to pH 4, Co(II) is expected to remain in solution at redox potentials below the Co3O4/Co(II) line as indicated by the purple line in Figure 1. If the redox of the system rises above this line, oxidative precipitation of Co(II) as Co3O4 becomes thermodynamically feasible. The experimental conditions should therefore be controlled such that the redox potential does not exceed Co(II) oxidation potentials at all pH values. Solubility of sulphur dioxide and oxygen Sulphur dioxide is more soluble in water compared to oxygen. The solubility of and O2 at 25°C has been reported to be 94.1 g/l and 0.04 g/l respectively (Brandt et al. 1994). Therefore in the SO2/air system one would expect the mass transfer of O2 to be limiting due to its low solubility compared to gas. Several authors have reported that O2 mass transfer is limiting for oxidative precipitation of Mn(II) and Fe(II) using SO2/O2 gas mixtures (Menard et al. 2007; Zhang et al. 2000). Zhang et al. (2000) also reported that an increase in temperature resulted in a decrease in the optimum SO2/O2 ratio due to reduced O2 solubility as temperature increases. The solubility of SO2 is known to be pH dependent; the SO2 uptake in water droplets was found to decrease with decreasing pH. As a result SO2 is less soluble in acidic water (Brandt et al. 1994). The speciation of S(IV) oxides in water also varies with pH as illustrated in Figures 2 and 3. Different S(IV) species are known to have different reactivities, with SO32reported to be 20–40 times more reactive than HSO3 (Brandt et al. 1994). In addition Brandt et al. (1994) cited Ali et al. who reported that for the oxidation of S(IV) by a Co(III) complex, the reactivity of the S(IV) species was in the following order; SO32- > HSO3 > .H2O; where SO32- was reported to be 16 times more reactive than HSO3 and HSO3 was 53 times more reactive than SO2.H2O. It has been reported that pH affects the stability of the produced transition metal complexes (Brandt et al. 1994). Zhang et al. (2000) reported that the optimum SO 2/O 2 ratio was dependent on pH and it has also been shown that an increase in pH

-2

Figure 2. Distribution of sulphur(IV) oxide species as a function of pH (5 x 10 M Na2S2O5 , 25°C)(Brandt et al. 1994)

OXIDATIVE PRECIPITATION OF MN(II) FROM COBALT LEACH SOLUTIONS

343


2/14/09

11:28 AM

Page 344

increased Mn precipitation kinetics (Zhang et al. 2002). In the pH range investigated, pH 2 to pH 4, the dissolved is expected to be mainly in the form of HSO3 as indicated in Figure 2. The Pourbaix diagram in Figure 3 also shows that HSO3 is the predominant S(IV) species at pH 2 to pH 4 at relatively low potentials. Proposed oxidation mechanism A radical mechanism has been proposed for the autoxidation of S(IV) by O2 in the presence of certain transition metal ions, namely: Cu(II), Fe(III), Ni(III), Co(III) and Mn(IV) (Brandt et al. 1994; Berglund et al. 1994; Das et al. 1999). The proposed mechanism results in the net oxidation of the transition metal ions as well as the oxidation of S(IV) species. The mechanism proposed by Berglund et al. requires tetravalent metal ions to react with the S(IV) species in order to form a complex, which is known to initiate the radical chain reactions. Brandt et al. suggested that it was possible to form a similar complex with the divalent metal ion and the S(IV) species, which also gives rise to radical formation. Huie et al. (1987) reported that ultraviolet rays were required to initiate radical formation. The mechanisms proposed by the various authors are similar; Brandt et al. studied the mechanism using computer simulations and derived it as indicated by Equations [1] to [6]. Initiation reactions [1a] [1b] In the presence of O2, the SO3*- radical reacts with O2 to form the peroxymonosulphate radical, SO5*-, which is understood to control the redox cycle. [2] Propagation reactions SO5*- is a more reactive oxidant than O2 and is believed to oxidize the divalent transition metal ions and the S(IV) as follows: [3a]

Figure 3. Pourbaix diagram for 4.8 g/l S(IV) at 30°C and 1 atm as predicted by the HSC thermodynamic model

344



2/14/09

11:28 AM

Page 345

[3b] [3c] The SO4*- radical is known to be a very strong oxidant with a standard reduction potential of 2.43 V vs. SHE, and is therefore expected to rapidly oxidize sulphite and transition metal ions (Das et al. 1999). The produced SO4*- radical can participate in oxidation reactions as follows: [4a] [4b] -

The HSO5 anion formed by the oxidation of S(IV) and M2+ by the SO5*- radical, together with the SO 5*- radical open various reaction pathways. These reaction pathways may influence both the decomposition process and the product formation as indicated by Equations [5] to [6] [5a] [5b] [5c] [5d] [5e] Termination reactions [6a] [6b] The recombination of SO4*- was proposed to be unlikely due to excess S(IV) and M2+ and was therefore not included in the reaction scheme (Brandt et al. 1994). Studies conducted by Das et al. (1999) on the reduction potentials of the SO3*- and SO5*radicals by pulse radiolysis revealed that the potential of the radicals was dependent on pH as indicated in Figure 4. This indicates that redox reactions that involve the radicals shown in Figure 4 are dependent on the pH of the system. Reaction stoichiometry The overall reactions for oxidative precipitation of Mn with SO 2/O 2 gas mixtures are illustrated by Equations [7] to [10]. Mn is expected to precipitate as MnO2 at pH values below pH 7 and high redox potential; and as Mn2O3 in the pH range of 5 to 7 at relatively high redox potential values below those required for MnO2 precipitation (see Figure 1). [7] [8] SO2 can also react with O2 to form sulphuric acid in the following side reaction: [9] OXIDATIVE PRECIPITATION OF MN(II) FROM COBALT LEACH SOLUTIONS

345


2/14/09

11:28 AM

Page 346

Figure 4. The redox potential of the SO3*-/SO32- and SO5*-/SO52- couples as a function of pH (Das et al. 1999)

Figure 5. Reactor set-up

Sulphuric acid produced by reactions 7 to 9 can be neutralized with hydrated lime to form gypsum: [10] Experimental Materials and apparatus Synthetic solutions were prepared using sulphate salts of AR grade and 98% sulphuric acid. The solutions prepared resembled the solution obtained from sulphuric acid cobalt leach operations containing approximately 6.5 g/l Co(II) and 2 g/l Mn(II). Laboratory tests were carried out in a 10 l glass reactor fitted with baffles and a flat-blade impeller as shown in Figure 5; the solution volume for these tests was 4 l. The impeller and baffles were designed to ensure efficient mixing and consequently efficient mass transfer of the gas mixture into the liquid phase. The gases were mixed in the pipeline using a ‘venturi effect’, where the smaller diameter SO2 line was inserted such that it bends into the bigger 346



2/14/09

11:28 AM

Page 347

diameter air delivery line. The SO2 was mixed with the air in the T-piece before the gas mixture was delivered to the reactor. The gas mixture was introduced into the reactor through a glass frit located below the impeller. The temperature was controlled using a water bath, with the thermocouple positioned inside the reactor. (Figure 5.) All tests were performed at atmospheric pressure at an impeller speed of 500 rpm. The pH and redox potential were monitored using Hamilton pH and Eh (Ag/AgCl) probes. Rotameters were used to control the SO2 and air flow rates into the reactor while the pH was controlled by addition of 17% (w/w) hydrated lime slurry using an autotitrator connected to a peristaltic pump. Experimental design The experiments were designed to investigate the effects of the SO2/air ratio, pH, temperature and the effect of using pure O2 versus air, as indicated in Table I. One parameter was varied at a time. The same SO2 flowrate of 11ml/min was used in all tests. This flowrate was calculated from the stoichiometric amount of SO2 required to precipitate 2 g/l Mn(II) according to Equation [7] within a period of 5 hours. However, each test was run for 6 hours, hence some excess reagents were introduced. All tests were conducted at 30°C unless otherwise stated. (see Table I.) Solution analysis was performed using two instruments: Varian atomic absorption spectrometer (AAS) and Varian inductively coupled plasma optical emission spectrometer (ICP-OES). Solid samples were analysed with ICP-OES, X-ray Fluorescence (XRF), X-ray diffraction (XRD) and scanning electron microscopy (SEM). Results and discussion Effect of SO2/air ratio The concentration of SO2 in the gas mixture was varied by changing the air flow rate while keeping a constant SO2 flow rate of 11 ml/min. The various air flow rates used were; 170 ml/min, 231 ml/min, 352 ml/min, 714 ml/min, 1439 ml/min, which were equivalent to 6%, 4.5%, 3%, 1.5% and 0.75% SO2 (v/v) in air respectively. The use of SO2/air gas mixtures resulted in faster kinetics for manganese precipitation compared to using air alone, as indicated in Figure 6. Mn precipitation increased significantly in the first 3 hours and thereafter little or no Mn precipitated further.

Table I Experimental matrix Parameter pH Temperature, °C SO2 in air, % (v/v) SO2 flow rate, ml/min Air flow rate, ml/min SO2 in O2, % (v/v) O2 flow rate, ml/min

Range 2–4 30–60 0–6 11 170–1439 3.5–23 36–301


347


2/14/09

11:28 AM

Page 348

Figure 6. Effect of % SO2 in air on Mn precipitation (2 g/l Mn(II); 6.5 g/l Co(II); pH 3; 30°C)

Figure 7. Effect of % SO2 in air on Mn and Co precipitation at pH 3 (2 g/l Mn(II); 6.5 g/l Co(II); 30°C; 6 hours)

The extent of Mn precipitation was similar for the dilute SO2 concentrations up to 3% SO2 in air, which is equivalent to 13% SO2 in O2, as indicated in Figures 6 and 7. This SO2/O2 ratio is similar to the optimum of 12% SO2 in O2 obtained by Schulze-Messing et al. (2006). Menard et al. (2007) has, however, shown that at pH 4, Mn precipitation could still be carried out at SO2/O2 ratios as high as 50% if adequate mixing is provided. The high SO2/O2 ratios, however, result in slow kinetics compared to the low SO2/O2 ratios and therefore would result in longer residence times. Zhang et al. (2000) reported that the optimum SO2/O2 ratio is O2 mass transfer dependent, with Mn precipitation increasing with the increase in O2 content in the gas mixture. This O2 mass transfer dependency of the system is likely because of the large difference in the solubility of SO2 and O2. The slower kinetics observed at SO2/O2 ratios higher than the optimum can be attributed to insufficient O2 supply in solution. Cobalt losses of 1–2% were recorded for all the SO2/air ratios (Figure 7). The relationship between SO2 efficiency and lime consumption versus the SO2/air ratio is shown in Figure 8. The efficiency of SO2 was calculated by dividing the predicted moles of SO2 used for Mn oxidation by the total moles of SO2 input. The efficiency followed the same trend as Mn precipitation with maximum SO2 efficiency recorded at 3% SO2 in air. The total lime consumption, however, remained relatively constant, which indicated that for less oxidizing gas mixtures, the side reaction (Equation [9]) becomes dominant and consumes the 348



2/14/09

11:28 AM

Page 349

Figure 8. Effect of % SO2 in air on SO2 efficiency and lime consumption at pH 3 (2 g/l Mn(II); 6.5 g/l Co(II); 30°C; 6 hours)

Figure 9. Effect of % in air on the redox potential at pH 3 (2 g/l Mn(II); 6.5 g/l Co(II); 30°C; 6 hours)

SO2 rather than Mn precipitation reaction (Equation [7]). The relationship between the redox potential and %SO2 in air at pH 3 is illustrated in Figure 9. The redox potential increased sharply on addition of SO2 in air and reached a maximum potential of 1.1V vs. SHE at 3% SO2 and decreased thereafter. It can be deduced from these results that the /air ratio controls the redox potential and therefore the oxidizing strength of the solution, and consequently the extent of Mn precipitation. The comparative oxidizing ability of SO2/air and SO2/O2 gas mixtures was investigated. The same molar quantity of O2 was used in both instances. The extent of Mn precipitation achieved using SO2/O2 gas mixtures was higher than that achieved with SO2/air gas mixtures (Figure 10). This can be attributed to the higher partial pressure of O2 in the gas mixture and consequently higher dissolved oxygen concentration when using pure O2 compared to air. Effect of pH The solution pH was varied from pH 2 to pH 4 at 30°C and 3% SO2 in air. An increase in pH resulted in increased Mn precipitation (Figure 11). At pH 4, all Mn was precipitated from the solution after 5 hours, which represented a stoichiometric consumption of SO2 according to Equation [7]. OXIDATIVE PRECIPITATION OF MN(II) FROM COBALT LEACH SOLUTIONS

349


2/14/09

11:28 AM

Page 350

Figure 10. Effect of % SO2 in O2 when using air and pure O2 on Mn precipitation (2 g/l Mn(II); 6.5 g/l Co(II); pH 3; 30°C; 6 hours)

Figure 11. Effect of pH on Mn precipitation (2 g/l Mn(II); 6.5 g/l Co(II); 3% SO2 in air; 30°C)

[7] An increase in pH also resulted in increased Mn and Co precipitation, as illustrated in Figure 12. Mn and Co precipitation was relatively constant at pH values less than 3, but a sharp increase in precipitation of both metals was observed at pH values above pH 3, as indicated in Figure 12. Zhang et al. (2002) also found the extent of Mn precipitation at pH values less than 3 to be slow; but as pH increased above pH 4, the rate of Mn precipitation became fast. It appears from these results that Mn and Co precipitation is relatively independent of pH at pH values less than 3. A cobalt loss of 1% was recorded at pH 3 and the cobalt loss increased linearly as pH was increased from pH 3 to pH 4. The higher cobalt losses recorded at higher pH values are possibly due to higher Mn removal at those pH values, which results in the presence of excess oxidant in solution, which consequently oxidizes the Co(II). Therefore Co losses at higher pH values can be limited by stopping the process as soon as sufficient Mn has been removed from solution. The marked increase in Mn and Co precipitation at pH values above pH 3 shows the strong influence of pH on the SO2/O2 system, as already reported by other authors (Menard et al. 2007; Zhang et al. 2000). The increase in Mn precipitation at pH values above pH 3 could be attributed to a change in the speciation of the S(IV) species (Figure 2) as both Co(II) and 350



2/14/09

11:28 AM

Page 351

Figure 12. Effect of pH on Mn and Co precipitation for 3% SO2 in air (2 g/l Mn(II); 6.5 g/l Co(II); 30°C; 6 hours)

Figure 13. Effect of pH on Mn precipitation and redox potential at 3% SO2 (2 g/l Mn(II); 6.5 g/l Co(II); 30°C; 6 hours)

Mn(II) are expected to remain in solution in the pH range of pH 2 to pH 4 (Figure 1). From the S(IV) speciation diagram (Figure 2), it is observed that at pH 2 the SO2.H2O and HSO3 species are equally dominant, but as pH increases from pH 2 to pH 4, HSO3 becomes the dominant S(IV) species. It has been reported that the HSO3 species is 53 times more reactive than the SO 2 .H 2 O species for the oxidation of S(IV) (Brandt et al. 1994). Consequently, an increase in Mn and Co precipitation at pH values above 3 was possibly due to improved kinetics caused by an increase in dominance of the more reactive HSO3 species over the SO2.H2O species as pH was increased. Figure 13 shows the effect of pH on Mn precipitation as well as redox potential for 3% SO2 in air. The redox potential decreased linearly as the pH increased but Mn precipitation increased significantly, at pH values greater than 3. The redox potential followed the same trend observed for the sulphite radical couples indicated in Figure 4. From these results it appears that the redox potential alone is not responsible for the oxidative precipitation of Mn since only half of Mn(II) was removed at the high redox potentials measured at pH 2 to pH 3, whereas 100% Mn(II) removal was achieved at the lower redox potential measured at pH 4. Figure 14 shows the redox potential measured for the 3% SO2/air gas mixture and the redox potential measured when using O2 alone. The MnO2/Mn(II) line and the Co3O4/Co(II) lines predicted by the HSC model in Figure 1 were also superimposed on the diagram. Although the redox potentials measured for O2 followed a similar trend as the SO2/air gas mixture, the extent of Mn precipitation by O2 alone was negligible (0.2%), compared to the OXIDATIVE PRECIPITATION OF MN(II) FROM COBALT LEACH SOLUTIONS

351


2/14/09

11:28 AM

Page 352

Figure 14. Effect of pH on the measured redox potential for O2 at 700 ml/min and 3% SO2 in air at 362 ml/min (2 g/l Mn(II); 6.5 g/l Co(II); pH 3, 30°C, 6 hours)

Temperature, °C Figure 15. Effect of temperature on Mn precipitation (6.5 g/l Co(II); 2 g/l Mn(II); pH 3; 3% SO2; 6 hours)

extent of Mn precipitation by the 3% SO2/air gas mixture (50%). It can also be observed from Figure 14 that the redox potential values measured for the 3% SO2/air gas mixture and O2 alone at various pH values lie within the (MnO2 + Co(II)) region in the Pourbaix diagram. Hence, thermodynamically, MnO2 is expected to precipitate from pH 2 to pH 4, as it was the case for the 3% SO2/air gas mixture; however, it was not the case for the O2 gas. These results further highlight that high redox potential values alone are not responsible for the precipitation of Mn, but the presence of reactive S(IV) species, which are responsible for the radical mechanism, is probably required for Mn precipitation. The fact that Mn precipitation increased significantly at pH values above pH 3 even though the measured redox for all the pH values lies within the MnO2 predominance region, further emphasizses the dependence of the kinetics on the S(IV) speciation. From Figure 14 the measured redox potential crossed into the Co3O4 predominance region only at pH 3.5, which indicates that cobalt oxidative co-precipitation is thermodynamically feasible at pH 3.5 and higher. Only 1% cobalt was lost at pH 3. Effect of temperature The effect of temperature was investigated by varying the solution temperature from 30°C to 352



2/14/09

11:28 AM

Page 353

60°C at pH 3 and 3% SO2 as indicated in Figure 15. The extent of Mn and Co precipitation increased linearly with increasing temperature. The increase in the rate of reactions upon increasing the temperature was expected due to the temperature dependence of reaction kinetics. Complete Mn precipitation was achieved at 50°C but at a cobalt loss of 17%. The high cobalt loss could be due to the fact that all the Mn had been precipitated within this period, leaving the oxidant free to oxidize the Co(II) in solution. An activation energy of 25 kJ/mol was calculated from the Arrhenius plot at pH 3 and 3% SO2. This activation energy is similar to the value of 23.5 kJ/mol obtained by Schulze Messing et al. (2007). The activation energy calculated for these conditions lies between 20 and 40 kJ/mol, which indicates that Mn precipitation is under mixed reaction control and that O2 mass transfer is most likely the dominant limiting factor compared to slow chemical reaction steps. The dominance of mass transfer reaction control can be supported by the large difference in the solubility of SO2 compared to O2 as well as the increase in Mn precipitation observed when the quantity of O2 in the SO2/air and SO2/O2 gas mixture was increased. Precipitation product The precipitate was analysed using ICP-OES, SEM, XRF and XRD techniques. The XRD results indicated that the precipitate consisted mostly of gypsum (CaSO 4), bassanite (CaSO4.xH2O), calcite (CaO) and bixbyite (Mn3+,Fe3+)2O3. According to the XRD analysis, Mn precipitated in the form of Mn2O3 and not MnO2 as predicted from the thermodynamic data. The reason for this is still under investigation. The SEM micrographs in Figures 16 and 17 indicate the two phases that were present in the precipitate. The major phase in the precipitate contained CaSO4 (99%), which is represented by the white areas in Figure 16. The manganese precipitate phase (< 1%) is the bright white spot indicated by the arrow in Figure 17. The cobalt that co-precipitated reported to the manganese precipitate phase as indicated by the X-ray scan in Figure 17. Conclusions Precipitation of Mn with SO2/air gas mixtures resulted in higher Mn removal compared to using air or oxygen alone. The use of pure O2 instead of air in the SO2 gas mixtures also resulted in improved Mn precipitation. Solution pH was found to play a major role in the precipitation reactions and the rate of Mn precipitation increased significantly only at pH

Figure 16. The SEM micrograph for the CaSO4 phase


353


2/14/09

11:28 AM

Page 354

Figure 17. The SEM micrograph for the Mn precipitate phase

values above 3. The SO2 concentration in the gas mixture was found to control the redox potential of the system and therefore the oxidizing strength of the solution to oxidatively precipitate Mn. The oxidizing strength of the solution decreased at SO2/air ratios higher than 3% SO 2, which is equivalent to 13% SO 2 in O 2. The efficiency of SO 2 use for Mn(II) precipitation also followed the same trend as the extent of Mn precipitation; however, lime consumption did not decrease significantly at SO2/air ratios > 3%, which indicates that the SO2 was mainly consumed by the side reaction of sulphuric acid production. An increase in temperature resulted in both increased Mn precipitation and cobalt coprecipitation. An activation energy of 25 kJ/mol was calculated from the Arrhenius plot at 3% SO2 and pH 3; which indicates that at these conditions, the reactions are both diffusion and chemically controlled. Diffusion control could be more limiting given the possible O2 mass transfer limitation. Although pure O2 resulted in similar measured redox potentials compared to the 3% SO2/air gas mixtures at pH 3, the extent of Mn precipitation with pure O2 was insignificant compared to the extent of Mn precipitation by the 3% SO2/air gas mixture. It was evident from these results that high redox potential values alone are not responsible for Mn precipitation and that S(IV) species play an important role in the mechanism of Mn precipitation with SO2/air gas mixtures. Therefore the reaction mechanism is possibly governed by both pH and redox potential, depending on the oxidizing ability of the gas mixture. However, pH and consequently the reactivity of the S(IV) species, had the most pronounced effect on the Mn precipitation reactions. XRD and SEM analysis revealed that the precipitate produced consisted mainly of gypsum or bassanite (99%), with the Mn containing phase (< 1%) distributed within the gypsum phase. The co-precipitated cobalt reported to the manganese phase. The manganese precipitate was found to be in the form of Mn2O3 rather than MnO2. Acknowledgements I would like to acknowledge and thank my mentors from Mintek, Marthie Kotze and Johanna van Deventer, who supported and advised me at different stages of the project. I also thank Dr Brian Green for his inputs in the editing of this paper and all the other Mintek staff who contributed to this work. 354



2/14/09

11:28 AM

Page 355

References 1. BERGLUND, J. and ELDING, L.I. Reaction of peroxomonosulphate radical with manganese(II) in acidic aqueous solution, Journal of Chemical Society Faraday Transactions, vol. 90, no. 21, 1994. pp. 3309–3313. 2. BRANDT, C. and VAN ELDIK, R. Transition metal-catalysed oxidation of sulphur(IV) oxides. Atmospheric relevant processes and mechanisms, Chemical Review, vol. 95, 1995. 1994. pp. 119–190. 3. DAS, N.T., HUIE, R.E., and NETA, P. Reduction potentials of SO3*-, SO5*- and S4O6*3radicals in aqueous solution, Journal of Physical chemistry, vol. 103, 1999. pp. 358–3588. 4. HUIE, R.E. and NETA, P. Journal of Atmospheric Environment, vol. 21, 1987. p. 1743. 5. MENARD, V. and DEMOPOULOS, G.P. Gas transfer kinetics and redox potential considerations in oxidative precipitation of manganese from an industrial zinc sulphate solution with SO2/O2, Hydrometallurgy, vol. 89, 2007. pp. 357–368. 6. SCHULZE-MESSING, J. ALEXANDER, D.C., SOLE, K.C., STEYL, J.D.T., NICOL, M.J., and GAYLARD, P. An Empirical Rate Equation for the Partial Removal of Manganese from Solution using a Gas Mixture of Sulphur Dioxide and Oxygen, Hydrometallurgy, vol. 86, 2007. pp, 37–43. 7. ZHANG, W., SINGH, P., and MUIR, D.M. SO2/O2 as an oxidant in hydrometallurgy, Minerals Engineering, vol. 13, 2000. pp. 1319–1328. 8. ZHANG, W., SINGH, P., and MUIR, D.M. Oxidative precipitation of manganese with SO2/O2 and separation from cobalt and nickel, Hydrometallurgy, vol. 63, 2002. pp. 127–135. 9. ZHANG, W. and CHENG, C.Y. Manganese metallurgy review. Part III: Manganese control in zinc and copper electrolytes, Hydrometallurgy, vol. 89, 2007. pp. 178–188.

Ndinanwi Mulaudzi Metallurgical/Process Engineer in Training, Mintek The author is a recent engineering post graduate; she obtained her undergraduate degree, B.Eng Metallurgy, from the Department of Materials Science and Metallurgical Engineering at the University of Pretoria in 2005. In 2006 she obtained an honours degree in the same discipline. She spent 2007 at Mintek conducting testwork for an M.Eng Metallurgy thesis and became a Mintek employee in the Hydrometallurgy division in March 2008. The thesis involved investigating the effects of pH, SO2/air ratios and temperature on precipitation of Mn using SO2/air gas mixtures. The work presented in the current paper is based on the thesis which is still


355


356

2/14/09

11:28 AM

Page 356


357-364_Mulligan:text

2/13/09

3:49 PM

Page 357

MULLIGAN, M.C. and BRADFORD, L. Soluble metal recovery improvement using high density thickeners in a CCD circuit: Ruashi II as case study. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Soluble metal recovery improvement using high density thickeners in a CCD circuit: Ruashi II a case study M.C. MULLIGAN* and L. BRADFORD† *FLSmidth Minerals (Pty) Ltd, South Africa †Metorex Ltd, South Africa

Counter current decantation (CCD) thickener circuits are used to recover soluble metal as pregnant liquor solution from ore leach residue. The basis of CCD operation is to concentrate suspended solids thereby minimizing liquor content in underflow slurry that flows in one direction. Then the underflow slurry liquor is diluted with wash liquor, that flows in the opposite direction, and the suspended solids are concentrated again and again. The amount of liquor in the thickener underflow contributes to determining the number of CCD stages required to recover the desired amount of soluble metal. High density thickeners (HDT) are designed to use gravity and compression, and minimize the amount of liquor in the underflow, thus minimizing the number of CCD stages. This paper reviews the process used to select HDTs versus high rate thickeners (HRT) for the Ruashi II copper-cobalt hydrometallurgical process in the Democratic Republic of Congo (DRC), from lab HDT simulations, to CCD simulations and thickener design details.

Introduction Copper at the Ruashi II plant is leached into solution in stirred tanks and is then recovered in a counter-current-decantation (CCD) circuit. The pregnant liquor solution (copper-rich solution) from the CCD circuit goes to a solvent extraction plant, followed by an electrowinning plant where the final copper product is produced. The washing efficiency in the CCD circuit is important to the overall copper production. The basis of CCD operation is to concentrate suspended solids thereby minimizing liquor content in underflow slurry. The underflow slurry liquor is diluted with wash liquor and the suspended solids are concentrated again. The amount of liquor in the thickener underflow is a parameter in determining the number of CCD stages required to recover the desired amount of soluble metal. Minimizing liquor in the thickener underflow leads to a higher recovery of soluble metal. This paper reviews the SOLUBLE METAL RECOVERY IMPROVEMENT USING HIGH DENSITY THICKENERS

357


2/13/09

3:49 PM

Page 358

process used to select the best thickener technology for the Ruashi II copper-cobalt hydrometallurgical process in the Democratic Republic of Congo (DRC), starting with lab simulations, then to CCD simulations, and finally thickener design details. Lab simulations The process simulations of milling and leaching the ore was performed by Mintek. Various leached samples were supplied to FLSmidth Minerals for thickening testwork. All post-leach thickening testwork was done straight after the leach was done in order to prevent ‘ageing’ of the sample. Thickening testwork The thickener simulations used a combination of bench-scale batch tests and continuous fill tests to measure the physical characteristics to be used in the size selection and design of the thickeners. Settling and flocculant flux curves Figure 1 exhibits a summary of the measured suspended solids settling flux (kg/h/m2) versus feed slurry suspended solids concentrations at different flocculant doses. A maximum settling flux is identified at a suspended solids concentration of between 7 and 8 wt%. The thickener feed slurry must be diluted to this concentration to optimize flocculation and suspended solids settling flux. Figure 2 exhibits a summary of the measured suspended solids settling versus flocculant dose at different feed slurry suspended solids concentrations. The optimal flocculant dosage is identified by a small change in settling flux at between 50 g/t and 60 g/t. Batch and continuous thickener simulations Various batch simulations were done at the optimal feed solids percent and optimal flocculant dosage conditions. The average underflow density achieved in the batch tests was 55% solids (by mass). Various continuous simulations with rakes were done to check the effect of residence time on underflow suspended solids concentration. Figure 3 exhibits the residence time required to

Figure 1. Optimal % solids

358



2/13/09

3:49 PM

Page 359

Figure 2. Optimal flocculant dosage

Figure 3. Continuous test showing mud residence time vs. solids %

concentrate the slurry. These simulations are used to determine the suitability of using a high density thickener rather than a high rate thickener. As one can see in Figure 3, the continuous simulation achieved an underflow suspended solids concentration of 59 wt% when the mud residence time is between 4 and 6 hours. This residence time is normally associated with high density thickeners. Rheology for thickener Rheology measurements were taken using a Haake VT550 vane viscometer using FLSmidth Minerals procedures. The slurry yield stress, the force required to produce movement from a stationary bed, is measured at various underflow suspended solids concentrations. The yield stress is a function of physical properties of the suspended solids (including chemical composition, particle size distribution and concentration), flocculant type, flocculant dose, and temperature. The yield stress results are used to: • Define the limit to underflow density each thickener type can achieve based on ability to discharge • Define the torque required to rotate the rakes during normal and abnormal operation. The thickener rakes must be able to restart in unsheared slurry after an unscheduled shutdown. FLSmidth Minerals thickener rake design and drive torque are designed to overcome this most severe condition. SOLUBLE METAL RECOVERY IMPROVEMENT USING HIGH DENSITY THICKENERS

359

3:49 PM

Page 360

Solids concentration

2/13/09

Yield stress (Pascals)


Underflow suspended solids concentration wt%

Figure 4. Underflow suspended solids concentration wt%

Figure 5. Underflow mud rheology

Figure 4 identifies a general relationship of thickener type to yield stress or underflow suspended solids concentration. Typically conventional thickeners (CT) and HRT are designed to consistently discharge underflow slurries exhibiting a yield stress of ~<25 Pa. FLSmidth Minerals HDT is designed to consistently create and discharge underflow slurries exhibiting a yield stress ~<100 Pa. FLSmidth Minerals deep cone paste thickeners (DCPT) are designed to consistently create and discharge underflow slurries exhibiting ~<500 Pa. The yield stresses measured at various slurry suspended solids concentrations for the Ruashi post-leach sample is shown in Figure 5. As one can see on the graph, at an underflow density of 55% the yield stress is 20 Pa, and at an underflow density of 59% the yield stress is 60 Pa. At 60 Pa yield stress a FLSmidth Mineral HDT should be used. As was seen above, the residence time to achieve 59 wt% underflow suspended solids was between 4 and 6 hours. Counter current decantation (CCD) simulations Once the laboratory thickener simulations were completed, measured data were used to simulate various counter current decantation (CCD) design options. These CCD simulations calculate soluble metal recoveries for changes in all the main variables affecting CCD wash recovery, which are: • Suspended solids concentration in underflow slurry • Soluble metal content in wash liquor 360



2/13/09

3:49 PM

Page 361

• Suspended solids concentration in leach residue of CCD feed • Wash ratio (mass wash liquor/mass suspended solids) • Interstage mixing efficiency. These different soluble metal recoveries form the basis for a sensitivity analysis of the CCD circuit. (Figure 6.) CCD wash recovery Metorex Ltd’s feasibility study used a high rate thickener performance scenario for the CCD soluble metal recovery. The underflow suspended solids concentration was 55 wt%. Metorex’s requirement was to achieve 99% soluble copper recovery in the CCD circuit. The initial CCD Circuit design variables are shown in Table I below. As shown in Figure 7, a minimum of 6 stages would have been required to obtain the wash recovery of >99% at the HRT operating conditions. Variables influencing wash recovery The five variables shown in Table I are the variables that have an influence on the overall CCD wash recovery. The wash ratio and underflow suspended solids concentration, however, are the two variables that have the greatest influence on wash recovery. Interstage Mix Wash Liquor

Flocculant

Leach Residue

Interstage Mix

Interstage Mix

Interstage Mix

CCD5

Low Grade PLS

Leach Residue

CCD4

High Grade PLS CCD3 CCD2 CCD1

Figure 6. High density thickeners Table I Variables used in CCD wash recovery calculations Wash ratio 1.6

U/F suspended solids concentration (wt%)

Interstage mixing efficiency (%)

Soluble Cu in wash liquor (g/l)

Feed suspended solids concentration (wt%)

Soluble Cu recovery (%)

55

95

0.0

30

+99

No. of stages

Figure 7. CCD high rate thickeners wash recovery

SOLUBLE METAL RECOVERY IMPROVEMENT USING HIGH DENSITY THICKENERS

361


2/13/09

3:49 PM

Page 362

Figure 8 below shows that a greater difference in wash ratio, from 1.1 to 2.1, has a significant effect on CCD wash efficiency (% soluble metal recovery). Increasing the wash ratio from 1.6 to 2.1 does not offer significant benefits at 5 stages of CCD and more. It can also be seen from Figure 8 that 5 stages at a higher wash ratio of 2.1 can achieve the 99% wash recovery rather than the 6 stages at a wash ratio of 1.6. The recommended wash ratio will be between 1.6 and 2.1. Figure 9 shows that a greater difference in underflow suspended solids concentration, from 55 wt% to 59 wt% has a significant effect on the number of stages required. An underflow suspended solids concentration of 59 wt% can be produced by a high density thickener, 5 stages are able to achieve the 99% wash recovery rather than the 6 stages at underflow densities of 55 wt%. It is evident from Figures 8 and 9 that both wash ratio and underflow suspended solids concentration have a big influence on the wash recovery (efficiency): • Increasing both the underflow (from 55 to 59 wt%) and wash ratio leads to an increase in wash efficiencies • By increasing the underflow solids concentration from 55 to 59 wt% five stages are required to achieve similar wash efficiencies as six stages at 55% as shown Figure 9 • Keeping the number of stages and underflow solids concentration the same, a 0.51% increase in wash efficiency can be realized by increasing the wash ratio from 1.6 to 2.1.

Figure 8. Sensitivity to wash ratio

Figure 9. Sensitivity to underflow suspended solids concentration

362



2/13/09

3:49 PM

Page 363

The above is based on ± 95% interstage mixing efficiencies between CCD stages. It can be seen in the above analysis, selecting high density thickeners rather than high rate thickeners, means that 5 stages can be used instead of 6 stages to achieve the required wash recovery. This means that the wash ratio of 1.6 can be used. By keeping the wash ratio at 1.6, the soluble copper concentration reporting downstream to the solvent extraction plant is greater and the plant does not need to be increased in size (i.e. if the wash ratio was 2.1 the solvent extraction plant would need to increase in size to handle the additional solution volume). High-rate vs. high-density thickeners CCD wash recovery In the previous section one can see the impact of underflow suspend solids concentration on the CCD wash recovery. The higher underflow suspended solids concentration means a greater wash recovery of soluble copper. This thus means that with all other variables the same, high density thickeners will always produce greater underflow suspended solids concentration because of a combination of gravity, compression and altering of permeability of the solids in the thickener. For Ruashi II this also means that with high density thickeners, the required wash recovery can be achieved with one fewer stage. Cost and payback implications In general there is a perception that high rate thickeners have a lower cost than high density thickeners. This is perhaps another reason for the preference for high rate thickeners in CCD circuits. However, high density thickeners are always smaller in diameter than high rate thickeners, allowing for a significant savings in installed cost. The use of high density thickeners at Ruashi has, however, meant one stage fewer in the CCD circuit. The high density thickener had an installed cost about the same as a high rate thickener, and with one fewer thickener in the CCD circuit offered a significant capital and operating cost saving. Even if the same number of stages were to be used with a high rate CCD circuit or high density CCD circuit, the increase in wash recovery has a good payback. Table II is an indicative increase in income due to a 0.61% increase in wash recovery. The extra copper recovery leads to an extra 1.5 million dollars per annum. This increase in income leads to a quick payback for the additional cost of the high density thickeners. Differences in thickener design There are some differences in the design between the high rate and high density thickeners. The following list gives an overview of the main design differences between the two types of thickeners. • Tank side wall depth—high density thickeners use a combination of gravity and compression to consolidate suspended solids. The mud residence time required to increase the underflow suspended solids concentration is achieved by increasing the sidewall height. • Tank floor slope—due to the greater rheology as a consequence of higher underflow suspended solids concentration, a steeper floor slope is used to assists the slurry movement to the centre of the thickener for discharge.. • Rake drive—since high density thickeners are always smaller than high rate thickener, the unit torque of ‘K factor’ must be greater. Typically the same rake drive required for a high rate can be used for a high density thickener and achieve a significantly greater unit torque input. • Rake mechanism—the rakes are designed to minimize the cross-sectional area of the mechanical members to minimize resistance or torque production and allow the most dense underflow suspended solids to discharge from the thickener. High density SOLUBLE METAL RECOVERY IMPROVEMENT USING HIGH DENSITY THICKENERS

363


2/13/09

3:49 PM

Page 364

Table II Increase in income due to increase in copper recovery Copper recovery for 45 000 tpa plant Copper produced Recovery

(tons/day)

130.00

(%)

98.78%

99.39%

Copper leached

(tons/day)

128.41

129.21

Production days

(days)

Copper produced

(tons/year)

350 44 945

Copper price

($/ton)

Income per annum

($/year)

242 901 639

Increase per annum

($/year)

1 500 000

45 222 5,404 244 401 639

thickeners also have pickets attached to the rake arms to alter the permeability of the compacted solids. Pickets create paths for the liquor in the compacted solids to escape, allowing solids concentration to continue • Thickener discharge—high density thickeners are designed with discharge cylinders with much more volume than cones to facilitate discharge and to minimize or prevent ratholing or underflow dilution. Conclusions The use of extensive laboratory thickening simulations to measure solid-liquid separation properties and CCD simulations enabled good information to be gathered for the design of the Ruashi II CCD circuit. The use of the CCD simulations showed that higher underflow densities in the thickeners means 5 stages rather than 6 stages in the CCD circuit can be used to achieve the same wash recovery of soluble copper. This thus led to the decision to use high density thickeners in the CCD circuit rather than High Rate thickeners. The use of high density thickeners allows Capex and Opex cost savings to be made due to the reduced number of stages. Acknowledgements The authors would like to thank Metorex who kindly gave permission for this paper to be published.

Mark Mulligan Process Engineering Manager, FLSmidth Minerals (Pty) Ltd Had a bursary from Sappi whilst studying BSc Chemical Engineering. Worked in the Pulp and Paper industry for 6 years in Process and Projects Engineering roles. Spent these years working on optimization projects and large Capital projects in the Pulp, Liquor recovery, Leach and Chlorine dioxide plants.Then spent a year working at Process Projects in a Process Engineering position, primarily working on projects in the Phosphoric acid and Fertiliser industries. Have been with FLSmidth Minerals for 3 years in the Solid Liquid separation part of the business, getting involved in all process related issues on all the projects relating to these technologies. 364


365-384_Reolon:text

2/13/09

11:10 AM

Page 365

REOLON, M., GAZIS, T., and AMOS, S. Optimizing acid utilization and metal recovery in African Cu/Co flowsheets. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Optimizing acid utilization and metal recovery in African Cu/Co flowsheets M. REOLON*, T. GAZIS*, and S. AMOS* *GRD Minproc Pty (Ltd), South Africa

Recent global demand for copper and cobalt has resulted in renewed interest in developing the copper and cobalt deposits in the DRC. In addition, recent dramatic increase in sulphur and sulphuric acid prices, as well as the copper and cobalt prices, has led to the creation of innovative ways to increase the utilization of reagents and recovery of metals. As a result of the potential in saving in operating costs, and the possibility of increased recovery of metal, a series of trade-off studies was performed to determine the cost benefits to several copper/cobalt studies for projects in the DRC. The purpose of these trade-off studies was to determine the feasibility of implementing any of the alternate flowsheet options into the next phase of these projects. This paper presents the base case and the various trade-off options and includes a brief description of the flowsheets considered. Special considerations specific to the base case are identified as these may influence the benefits of the options for individual projects. The major advantages and disadvantages are listed in order of importance to indicate where capital, operating or latent risk costs are affected by the options. High-level cost estimate ranges for capital, operating and annualized risk are given for the various options. Where applicable, comments on sensitivity to any identified parameters are discussed. In order to compare the various options to each other, both process and economic summary tables comparing the options are shown. Discussion on the interaction of more than one of the options will complete the comparisons.

Democratic Republic of the Congo Introduction The Democratic Republic of the Congo was initially established as the Belgian Congo in 1908. It subsequently gained independence in 1960. The early years of independence were marked by social and political turmoil. OPTIMIZING ACID UTILIZATION AND METAL RECOVERY IN AFRICAN CU/CO FLOWSHEETS

365

365-384_Reolon:text

2/13/09

11:10 AM

Page 366

In a coup in 1965, an army colonel, Joseph Mobutu seized power, renaming himself Mobutu Sese Seko and the country Zaire. Under his presidency, social and economic conditions in the country deteriorated further. In 1997, Mobuto was toppled by a rebel leader, Laurent Kabila. Kabila restored the name of Congo in the form of the Democratic Republic of the Congo (DRC). He, in turn faced a rebel movement backed by Rwanda and Uganda. Troops from a number of African allies were flown into the country and the rebellion was put down and a ceasefire signed in 1999. Kabila himself was assassinated in early 2001, and his son, Joseph Kabila, was installed as president of the Republic. Joseph Kabila subsequently negotiated the removal of Rwandan forces, and a peace agreement was signed in South Africa in 2002. A transitional government was formed with the objective of eventually holding democratic elections in the country. These elections were held in July 2006, and resulted in the leading two presidential contenders each receiving less than 50% of the vote, although Joseph Kabila did gain a significant majority over his leading rivals. In the terms of the election rules, a run-off between the last two took place in November 2006 with Kabila gaining an outright majority. The United Nations were invited to install a peacekeeping force in the country at the end of last century, and this force has contributed to a general improvement in the internal security of the country, although, even today, there is continuing insecurity in the north-east of the country. This area is quite remote from the main copper and cobalt mining province of Katanga, where conditions have remained calm. Economy Since 2001, when Joseph Kabila became president, the economic policy of the country has aimed at stabilization. Since about that date, there is little doubt that the mining and construction sectors are improving, albeit from a low base. Gross national income is US$120 per capita, but since 2002, the economy has shown signs of expansion with a GDP growth in 2002 of 3.5%, 2003: 5.7%, 2004: 6.8%, 2005: 6.5%. The latter two figures are provisional. The independent central bank has bought inflation under control in recent years, again indicating a normalizing of the economic and social situation within the country. The corporate tax and investment codes are being revised to continue to liberalize the domestic business environment and attract foreign investment, and an increasing number of foreign companies are establishing themselves in the country. Gecamines Prior to the departure of the Belgians from the Congo in 1960, the copper industry in the Katanga area was owned and operated by Union Miniere de Haut Katanga (UMHK), a company dating from 1906. After independence, nationalization of the copper industry was inevitable, and UMHK were eventually dispossessed of their ownership of the industry, and a state owned company, initially Generale Congolaise des Minerals and later La Generale des Carriers et des Mines (Gecamines) took over as owners and operators of all the copper mines in Katanga. Gecamines initially prospered throughout the 1970s and 1980s, but since the late 1980s, the company went through a series of increasing crises, until finally, by the late 1990s, copper production was only 35 000 t of copper and 3 940 t of cobalt, compared to 437 000 t of copper in 1987, plus 11 900 t of cobalt in the same year. As a result of this poor performance, the Government started to seek ways of improving the situation both within Gecamines and within the copper industry as a whole. As a result of these initiatives, foreign private industry has been encouraged to enter into agreements with Gecamines in an attempt to restore the industry. After some stumbling, these initiatives are now beginning to bear fruit. 366


365-384_Reolon:text

2/13/09

11:10 AM

Page 367

Code Minier A Code Minier has been introduced by the Government and provides a transparent legal framework for investors in mining. The code has been gazetted in the French and English, making it easier for foreign Anglophone companies to review the legislation. Copper cobalt process—the base case A typical hydrometallurgical copper cobalt flowsheet as used in southern Africa is shown in Figure 1. This flowsheet is used as a base case against which the various options are compared. A predominantly oxide ore is crushed and milled to expose the copper and cobalt minerals for leaching. The milling step is the main entry point for water in the process The milled slurry then undergoes a solid/liquid separation stage in a thickener in order to reduce the water entering the leach circuit. The separated water is recycled back to the mill. The thickened slurry is sent to the leach reactors, where acid rich raffinate (from solvent extraction) and makeup acid is blended with the slurry. Copper and other elements are leached by the acid. The leach reaction is also performed under reducing conditions using sulphur dioxide in order for the cobalt to be effectively leached. After leaching the slurry is washed in a 6 stage counter current decantation (CCD) wash circuit using slightly acidified process water. This step removes most of the metal containing solutions from the leached solids. The washed solids are discharged to the tailings neutralization step, and the metal containing solution is sent to the solvent extraction (SX) step as pregnant leach solution (PLS) The washed solids from the underflow of the last CCD thickener are pumped to the tailings neutralization circuit. Iron/manganese precipitate is also neutralized in this step. Slaked lime is used to control the pH at above 8. The neutralized tails are then pumped to the tailings disposal site. Some of the water is decanted from the tailings dam and recycled to the process water pond. The balance of the water remains in the tails. This is an exit point for water.

Figure 1. Base case process flowsheet

OPTIMIZING ACID UTILIZATION AND METAL RECOVERY IN AFRICAN CU/CO FLOWSHEETS

365-384_Reolon:text

2/13/09

11:10 AM

Page 368

The PLS is fed to the SX circuit where copper ions are selectively exchanged for acid (or hydrogen ions) using a liquid organic solvent extraction medium. The discharge solution (copper lean but acid rich) is discharged from the SX circuit as raffinate, which is recycled and used in the leach step (see above). In order to maintain a water balance, and to bleed cobalt and other metals out of this recycle stream, some of the raffinate is bled off and sent to the cobalt production circuit. The copper extracted by the SX organic is stripped with a strongly acidic electrolyte, where the copper ions are exchanged for acid. The copper rich electrolyte is sent to the copper tank house where the copper is electroplated, harvested and prepared for market. The electrowinning (EW) process produces acid while plating copper metal. The raffinate bleed that is sent to the cobalt production circuit is depleted of copper, but rich in acid (from leach plus SX) and other accumulated metals. The cobalt circuit consists of a series of stage-wise neutralization/precipitation reactions and solid/liquid separation steps. The first stage is to remove primarily iron, manganese, aluminium and phosphates as precipitates. This is done using limestone as a neutralizing agent controlled to pH of 3–3.5, and a mix of SO2/air as an oxidizing agent for iron and manganese. The precipitate is thickened, filtered, washed and transferred to tailings neutralization (see above). The next stage is to precipitate residual copper at pH 5–5.5 using slaked lime. The precipitate is thickened and pumped to the leach circuit for recycle of the copper and some coprecipitated cobalt. The final stage is to precipitate cobalt as an impure hydroxide at pH 8 using slaked lime. This cobalt product is thickened, filtered, dried and bagged for market. The remaining solution from the cobalt circuit is sent to the process water pond for recycle in the plant. Excess water is treated at this point for disposal. Alternative flowsheets for the optimization of reagent utilization and copper/cobalt recovery The focus of the trade-off studies has primarily been on the reduction of acid consumption, and the subsequent need to neutralize the acid with limestone, lime and other acid neutralizing reagents. This drive for acid consumption minimization was particularly powerful during the recent sulphur price increase during 2007 and 2008. Even though the sulphur price has returned to the levels prior to this increase, there remains a significant cost to transport the sulphur into the DRC, therefore one can always expect sulphur costs at substantially above coastal sulphur costs. The following list of options has been considered in recent projects, initially as trade-off studies as well as inclusions in the feasibility studies and EPCM. • Split SX • Milling in raffinate • Pre-leach filtration • Acid tailings • Post-leach filtration instead of CCD. The projects evaluated during mid 2007 to mid 2008 are located in the DRC and range in size as follows: • ROM throughput—1 300 000 to 4 500 000 t/a • Copper production—30 000 to 80 000 t/a. the base case is centred around 40 000 t/a • Cobalt production—1 500 to 7 000 t/a. 368


365-384_Reolon:text

2/13/09

11:10 AM

Page 369

Capital and operating costs are based on the costs received during the period given above. It must be noted that these costs did vary significantly during this period. For this reason the results given in this report are indicative only. An evaluation of this nature will need to be performed on specific projects to obtain relevant data specific to the project under evaluation. Split SX Description The base case flowsheets and design criteria have proposed that the leach product be washed in a series of six CCD thickeners before passing the PLS (CCD1 overflow) to copper SX/EW. At this point, this is the only place in the circuit where copper is removed. By introducing the split SX option, the first CCD thickener will assume the function of a post-leach thickener as well as a point of separation for the high grade and low grade PLS streams. The high grade PLS stream will pass directly through the high grade copper SX/EW stage where copper is removed. The resulting raffinate solution is recycled to the leach feed. The low grade PLS stream will be passed through the remaining five CCD thickeners, undergoing washing before being sent to the low grade copper SX/EW stage. The resulting low grade raffinate is sent to the cobalt removal circuit. The amount of Copper produced increases slightly in comparison with the Base Case, as well as a slight increase in the production of Cobalt, resulting too in less Cu/Co settling out in the tailings dam. (Figure 2.) Analysis for splitting the SX stream into high grade and low grade streams In comparing the mass balances for both the base case and for the split SX alternative, it is observed that both the copper and cobalt production increase slightly (having fewer losses to the tailings dam), with a significant decrease in total acid consumption. Due to splitting the PLS and redirecting a portion of it to the high grade SXEW stage, the acid content of the CCD

Figure 2. Split SX flowsheet


365-384_Reolon:text

2/13/09

11:10 AM

Page 370

underflow to the tailings neutralization decreases as well as the acid content of the cobalt bleed stream; therefore limestone required to neutralize this acid decreases. These changes are used to determine the change of operating cost and revenue earned. Advantages and disadvantages Below is a list of advantages and disadvantages of the split SX option in relation to the base case. Advantages Capital • Reduced lime plant • Reduced acid plant Operating and revenue • Reduced acid costs • Reduced Co bleed and tails neutralization • Cu production increase • Co production increase Risk • SX plant redundancy • Potential for additional SX throughput • Potential to continue partial operation on SX fire destroying one train Disadvantages Capital • Additional SX stream • Additional PLS pond • Additional raffinate pond Operating and revenue • Additional plant operation (labour) Risk • Operational complexity Capital estimate The capital cost changes for the implementation of the split SX option is based on the following. • Reduction in limestone plant size • Reduction in acid plant size • Addition of a secondary copper SX/EW circuit as well as associated ponds. Operating cost estimate The operating cost changes for the implementation of the split SX option is based on the following. • Reduction in the consumption of limestone and acid • Increase in copper production • Increase in cobalt production. Risk estimate Annualized risk equivalent costs have been determined based on the following significant risks. 370


365-384_Reolon:text

2/13/09

11:10 AM

Page 371

• SX train redundancy • Partial production after a fire event. Results The summarized results for the split SX estimated capital costs, operating costs and revenues earned, and annualized risk assessment is given in Table I. The implementation of the split SX option will negatively affect capital, where the addition of an extra PLS and raffinate pond capacity as well as supplementary SX costs will load the capital expenditure by an additional $2 000 000. Capital advantage for reduced acid and lime plant is also allowed. An operating cost advantage of $4 200 000 is realized, however, in the split SX option. This emanates greatly from the increased copper and cobalt revenue, as well as a reduction in the consumption of both acid and limestone. The costs associated with risk are advantageous in excess of $300 000 /a. Recommendation The split SX option offers both risk as well as operating and revenue advantages and is recommended to be considered in any future hydrometallurgical copper cobalt projects. Milling in raffinate Description The base case flowsheets and design criteria have proposed that the mill circuit uses process water recycled from the downstream preleach thickener overflow and makeup process water. This results in the addition of water into the leach circuit due to the slurrying of the relatively dry ROM. Any water that passes into the leach circuit will need to be removed from the process. The water trapped in the settled tailing is the only substantial outlet for water in the plant. Any further water accumulation will have to be treated in the raffinate bleed. By introducing the milling in raffinate option, recycled raffinate is used to slurry the crushed ROM, thus reducing the water input into the process. This will have the net effect of reducing or eliminating excess water accumulation in the plant, therefore reducing the raffinate bleed to the cobalt plant. Other benefits include better leaching and the elimination of the preleach thickener step, reduced acid and limestone consumption. (Figure 3.)

Table I Cost advantage summaries for the split SX option Capital cost advantage Operating/revenue cost advantage Annualized risk advantage Net annual advantage Payback

-$2 000 000 $4 200 000 $300 000 $4 500 000 0.45 years

Positive cost values imply savings or financial gain to the project. Negative cost values imply expenses or financial loss to the project. Data indicative of 40 000 t/a copper production.


371

365-384_Reolon:text

2/13/09

11:10 AM

Page 372

Figure 3. Milling in raffinate flowsheet

Analysis of milling in raffinate compared to milling in water (base case) In comparing the mass balances for both the base case and for the milling in raffinate alternative it is observed that the acid consumption decreases. The acid content of the raffinate bleed also decreases, therefore limestone to neutralize this acid decreases. Due to the reduced size of the cobalt bleed, the tenor of cobalt will increase in the raffinate returning to leach. This in turn will cause the cobalt loss to CCD underflow to increase, resulting in cobalt production decrease. These changes are used to determine the change of operating cost and revenue earned. Advantages and disadvantages Below is a list of advantages and disadvantages of the milling in raffinate option in relation to the base case. Advantages Capital • Cobalt plant feed and capital reduced • Acid plant size reduced • Limestone plant size reduced • No pre-leach thickener required. Operating and revenue • Cobalt plant operating cost reduced • Acid consumption reduced • Limestone consumption reduced • No preleach op costs required • Reduced maintenance on acid and lime plants. Risk • Co tenor therefore product purity increase. 372


365-384_Reolon:text

2/13/09

11:10 AM

Page 373

Disadvantages Capital • Acid resistant mill required • Acid resistant mill area required. Operating and revenue • Co production decrease • Additional mill maintenance. Risk • Risk of mill and area corrosion • Added mill delivery schedule of approximately 20 weeks • HSE risk. Capital estimate The capital cost changes for the implementation of the milling in raffinate facility are based on the following. • Reduction in the size of the cobalt plant • Reduction in the size of the acid plant • Complete removal of the preleach thickener area • Reduction in the size of the limestone plant • Additional cost for an acid resistant mill, stainless steel chute and plate work and protected structural steel and concrete. Operating cost estimate The operating cost changes for the implementation of the milling in raffinate facility are based on the following. • Reduction in the consumption of acid and limestone • Reduced power, maintenance and consumable in the cobalt plant • Remove preleach thickener operating costs • Reduced operating cost in the acid and limestone plants • Additional mill operating costs • Reduced cobalt production. Risk estimate Annualized risk equivalent costs have been determined based on the following significant risks: • Additional delay in delivery of the mill causing delayed revenue. This risk is highly variable, and could be mitigated by earlier procurement, if such timing is available • Downtime and replacement of corroded mill shell • Increased HSE risk in the area. Results The summarized results for the milling in raffinate estimated capital costs, operating costs and revenues earned, and annualized risk assessment is given in Table II. The implementation of the milling in raffinate option will have an overall capital advantage where the reduction of process and reagent plant capital more than offsets the additional cost of the milling area. The reduction in the cobalt plant and acid plant result in the major capital reduction. OPTIMIZING ACID UTILIZATION AND METAL RECOVERY IN AFRICAN CU/CO FLOWSHEETS

365-384_Reolon:text

2/13/09

11:10 AM

Page 374

Table II Cost advantage summaries for the milling in raffinate option Capital cost advantage Operating / revenue cost advantage Annualized risk advantage Net annual advantage Payback

$ 2 000 000 $ 5 000 000 $-5 000 000 $0 N/A

Positive cost values imply savings or financial gain to the project. Negative cost values imply expenses or financial loss to the project. Data indicative of 40,000 t/a copper production.

An operating cost advantage of $5 000 000 is realized in the milling in raffinate option. The largest contributor is from the reduction in sulphur and limestone reagent requirements. The costs associated with risk have been shown to be about $1 000 000/a if the added delay of production start-up can be avoided. If however the added delay can not be mitigated, then the total annualized risk climbs to $5 000 000 /a. Recommendation In comparing milling in raffinate to milling in water, the result established that milling in raffinate is potentially a high risk option, both technically as well as commercially. Delivery times are high, and the need for expensive spares (mill shell), or prolonged shutdown are typically unacceptable. Risk mitigation strategies may well swing this option positively. The alternative option of preleach filtration vs. thickening offers similar benefits while eliminating the high risk factor of milling in raffinate and is preferred where filtration is feasible. Preleach filter Description The base case flowsheets and design criteria have proposed that the feed to leach be dewatered in a thickener, thus allowing process water to be recycled back to the mill. Any water in the underflow of this thickener will result in the addition of water into the leach circuit. Any water that passes into the leach circuit will need to be removed from the process. The water trapped in the settled tailing is the only substantial outlet for water in the plant. Any further water accumulation will have to be treated in the raffinate bleed, resulting in the neutralization (loss) of acid using limestone By using the preleach filtration option, the water content of the slurry to leach is reduced from 50–60% to an estimated 20–25%, thus reducing the quantity of water into the process. This will have the net effect of reducing excess water accumulation in the plant, therefore reducing the raffinate bleed to the cobalt plant and increasing raffinate utilization in the leach. This will result in reduced acid and limestone consumption. (Figure 4.) Analysis of preleach filtration as compared to the base case of preleach thickening In comparing the mass balances for both the base case and for the preleach filtration alternative it is observed that the acid consumption decreases. The acid content of the raffinate bleed also decreases, therefore limestone to neutralize this acid decreases. Cobalt production decreased due to increased tenor in the raffinate, resulting in increased losses to tails. These changes are used to determine the change of operating cost and revenue earned. 374


365-384_Reolon:text

2/13/09

11:10 AM

Page 375

Figure 4. Preleach filtration flowsheet

Advantages and disadvantages Below is a list of advantages and disadvantages of the preleach filtration option in relation to the base case. Advantages Capital • Reduce cobalt plant size • Reduce acid plant size • Eliminate preleach thickener • Reduce limestone plant size. Operating and revenue • Reduce acid consumption • Reduce limestone consumption • Reduce cobalt plant operating cost • Eliminate preleach thickener operating cost. Risk advantage • Increased co tenor possibly increased product concentration. Disadvantages Capital • Require a preleach belt filter Operating and revenue • Increased complexity of preleach filter—operation and maintenance • Lower cobalt recovery. Risk • Risk of increased operational shutdown. OPTIMIZING ACID UTILIZATION AND METAL RECOVERY IN AFRICAN CU/CO FLOWSHEETS

375

365-384_Reolon:text

2/13/09

11:10 AM

Page 376

Capital estimate The capital cost changes for the implementation of the pre-leach filter facility are based on the following: • Reduction in the size of the cobalt plant • Reduction in the size of the acid plant • Complete removal of the preleach thickener area • Reduction in the size of the limestone plant • Addition of the preleach filter unit(s). Operating cost estimate The operating cost changes for the implementation of the preleach filter is are based on the following: • Reduction in the consumption of acid and limestone • Reduced power, maintenance and consumable in the cobalt plant • Reduced operating cost in the acid and limestone plants • Greater operating costs for filtration than thickener • Reduced cobalt production • Removal of the thickener operating costs. Risk estimate Annualized risk equivalent costs have been determined based on an estimated likelihood of unplanned days shutdown of the filter plant. Results The summarized results for the preleach filtration estimated capital costs, operating costs and revenues earned, and annualized risk assessment are given in Table III below. The implementation of the preleach filter option will have an overall capital advantage of about $5 000 000 where the reduction of process and reagent plant capital more than offsets the additional cost of preleach belt filters. The reduction in the cobalt plant and acid plant result in the major capital reduction. An operating cost advantage of estimated $4 500 000 is realised in the preleach filtration option. The largest contributor is from the reduction in acid and limestone reagent requirements. Unfortunately a reduction of cobalt has been observed resulting in substantial revenue losses, but this factor is outweighed by the other savings in operational costs. Table III Cost advantage summaries for the pre-leach filtration option Capital cost dvantage Operating/revenue cost advantage Annualized risk advantage Net annual advantage Payback

$5 000 000 $ 4 500 000 -$2 000 000 $2 500 000 N/A


376


365-384_Reolon:text

2/13/09

11:10 AM

Page 377

The cost associated with risk is based on an approximately 5 days of production throughput, and is calculated at $2 000 000 /a. Recommendation Where filtration is reasonably possible, the preleach filtration option offers both capital and operational cost savings, which offset any additional risk. It is therefore recommended to implement the preleach filtration option into hydrometallurgical copper cobalt projects. Recent results on a project have, however, shown that a preleach thickener underflow density in excess of 65% solids was observed, thus the thickener option was retained. Acid tails Description The base case flowsheets and design criteria have proposed that the CCD underflow slurry stream be mixed with an iron and manganese impurity slurry, and then neutralized with lime prior to being pumped to the tailings facility. The alternate proposal is to pump the CCD underflow slurry directly to a modified tails facility. This will result in a tails facility operating at approximately a pH of 3. The iron and manganese impurities stream will have to be placed in a separate compartment of the tailings facility in order to prevent releaching of the impurities into the acidic water in the acid tailings facility. The major changes to the plant equipment are the lining of the tails facility, the creation of a separate compartment in the tails facility for the impurities slurry, and the removal of the neutralization circuit. The conversion to an acid tails facility is expected to increase the utilization of acid, increase the production of copper and cobalt, and possibly increase the leaching of copper and cobalt in the acidic environment of the tails. These benefits are achieved by the recovery of the clarified acidic water that is decanted off the top of the tails facility. This water, containing the acid, copper and cobalt, is recycled to the plant as CCD wash water, where the acid and metals have an opportunity to be extracted in downstream processes. To enhance the recovery of tails leached material, it is suggested that a means of removing recycle acidic water from the bottom of the tails pond is implemented; else most of the leached copper will remain trapped in the tails sediment. (Figure 5.) Analysis for the implementation of an acid tails facility In comparing the mass balances for both the base case and for the acid tails alternative, it is observed that the copper and cobalt production increases, while the acid consumption decreased by a small amount. The acid content of the raffinate bleed also increases by a small amount; thus more limestone is needed in the cobalt plant. These changes are used to determine the change of operating cost and revenue earned. Advantages and disadvantages Below is a list of advantages and disadvantages of the acid tails facility in relation to the base case of neutralized tails. Advantages Capital • Remove tailings neutralization capital OPTIMIZING ACID UTILIZATION AND METAL RECOVERY IN AFRICAN CU/CO FLOWSHEETS

377

365-384_Reolon:text

2/13/09

11:10 AM

Page 378

Figure 5. Acid tails flowsheet

Operating/revenue • Reduced tails neutralization costs • Recover some soluble copper and cobalt from CCD underflow • Reduce acid costs • Potential to continue leaching on tails pond. Disadvantages Capital • Acid tails lining costs • Require a separate compartment for Fe/Mn tails in pond • Tailings facility closure costs Operating • Increased limestone consumption for Co plant • Storm water capacity. Risk • HSE Risk if liner is breached in tailings. Capital estimate The capital cost changes for the implementation of the acid tailings facility are based on the following: • The tails neutralization plant can be removed completely • The acid tails facility needs a lining to prevent the acid from entering the groundwater • The acid tails facility needs to be compartmentalized to keep the impurities separate from the CCD underflow tails • Mine closure costs for the acid tails pond have not been included, but need to be considered in evaluating this option. 378


365-384_Reolon:text

2/13/09

11:10 AM

Page 379

Operating cost and revenue income estimate The operating cost changes for the implementation of the acid tailings facility are based on the following: • An increase in the revenue from copper and cobalt due to better recoveries • Elimination of neutralization operating costs • Reduced acid consumption • A small increase in limestone consumption in the cobalt plant • Possibly none, but potentially some revenue from the continued leaching of copper and cobalt in the acid tailings facility. The benefit from this source would have to be tested in a laboratory or pilot plant for each project in order to confidently assign any benefits from increased revenue. Risk estimate The main risk identified is that the liner of the acid tailings facility could be compromised during the life of the mine. A significant outflow into the soil and groundwater system below the tailings facility will result in environmental discharge requiring remedial actions to be put in place. An estimate for the scenario of liner failure has been made. The estimate is based on the probability of a liner installations failing, an assumed cost to repair the failure, and determining the annualized loss of revenue for such remediation actions. Results The result for implementation of the acid tails facility is presented in Table IV. Operating cost and revenue earned values represent the annualized net gain including an estimated income for the extended leaching advantage. If the advantage of extended leaching is ignored, then the payoff of the capital occurs in more than 10 years. These conclusions are based on current cost analysis only and may differ considerably once further investigation of the cost of lining and lining maintenance of an acid pond is taken into account. Furthermore, the cost of mine closure will further reduce the performance of this option. Recommendation Further investigation into actual costs of constructing, lining and maintaining an acid proof pond is required for specific projects considered, as no clear trend has been observed.

Table IV Cost advantage summaries for acid tails option Capital cost advantage Operating/revenue cost advantage Annualized risk advantage Net annual advantage Payback

$-6 500 000 $2 900 000 -$1 500 000 $1 400 000 4.6 years

Positive cost values imply savings or financial gain to the project. Negative cost values imply expenses or financial loss to the project. Data indicative of 40,000 t/a copper production.


379

365-384_Reolon:text

2/13/09

11:10 AM

Page 380

Furthermore, the risk evaluation for tailings dam integrity is often subjective and therefore difficult to determine. The specific findings of such an investigation significantly alter the cost advantage summary. Filtration vs. CCD Description The base case flowsheets and design criteria have proposed that the solids in the leached stream be separated from the solution using a conventional CCD circuit operating with a wash ratio of 1.26. This results in the addition of large quantity of water into the PLS stream. By introducing the post-leach filtration option (which required an initial thickener to preconcentrate the filter feed), a lower wash water ratio of 0.8 is used, thus adding less water to the PLS. Furthermore, it is anticipated that the recovery of copper, cobalt and acid through the filter is improved due to more efficient washing and the production of a filter cake containing less moisture than the CCD underflow to tailings. A smaller PLS stream also implies that the raffinate bleed, hence the cobalt plant can be reduced in size, with subsequent savings on acid and limestone. (Figure 6.) Analysis of post-leach filtration compared to CCD (base case) In comparing mass balances for both the base case and for the post-leach filtration alternativeit is observed that the acid consumption has increased, together with the raffinate bleed stream. This is mainly due to increased copper production, causing an increased acid production in the raffinate. Since a significant portion of this raffinate is neutralized, acid consumption is increased. Copper and cobalt production has, however, increased due to better recoveries from the tails. These changes are used to determine the change of operating cost and revenue earned.

Figure 6. Postleach filtration flow sheet

380


365-384_Reolon:text

2/13/09

11:10 AM

Page 381

Advantages and disadvantages Below is a list of advantages and disadvantages of the post-leach filtration option in relation to the base case. Advantages Capital • Remove CCD capital. Operating and revenue • Remove CCD operating costs • Additional Cu production • Additional Co production. Disadvantages Capital • Add post-leach thickening and filter capital • Larger cobalt plant • Larger acid plant • Larger limestone plant. Operating and revenue • Add thickener operating cost • Add filter operating costs • Additional acid consumption • Additional limestone consumption. Risk • Loss of filter. Capital estimate The capital cost changes for the implementation of the post-leach filtration facility are based on the following: • Remove the CCD, adding pre-filter thickener and filters • Increasing the cobalt plant • Increasing the acid plant • Increasing the limestone plant. Operating cost estimate The operating cost changes for the implementation of the post-leach filtration facility are based on the following: • Remove CCD operating costs but adding the pre-filter thickener and filters operating costs • Adding copper and cobalt revenue • Increased acid and limestone consumption. Risk estimate • Since a spare filter has been allowed for, the risk due to filter shutdown is considered to be negligible. OPTIMIZING ACID UTILIZATION AND METAL RECOVERY IN AFRICAN CU/CO FLOWSHEETS

381

365-384_Reolon:text

2/13/09

11:10 AM

Page 382

Results The summarized results for the post-leach filtration estimated capital costs, operating costs and revenues earned, and annualized risk assessment is given in Table V. The implementation of the post-leach filtration option will have an overall capital cost of $15 000 000. The cost of the filters represents the major capital increase. An operating cost advantage of only $400 000 is insufficient to justify the extra capital expense. Recommendation In comparing post-leach filtration to conventional CCD, the result established that post-leach filtration is not a viable option as the capital costs far outweigh the operating cost advantages. Improvement in the filtration properties may reduce the capital, but to date this appears unlikely to make this option viable. Overall summary of results Table VI shows the averaged costing summaries for the various options considered in this paper. From the analysis of recent copper cobalt hydrometallurgical plants, it is clear that the split SX and pre-leach filtration options offer a definite advantage in enhancing the performance of the process plant. The milling in raffinate option can potentially offer advantage, however, the risk is high, and thus there has been a tendency not to select this option. The acid tail and post-leach filtration options are costly to implement, with limited returns, as reflected in the payback periods. The implementation of multiple options has been considered and analysed in individual studies; however, no analysis has been attempted here. Table V Cost advantage summaries for the post leach filtration option Capital cost advantage Operating/revenue cost advantage Annualized risk advantage Net annual advantage Payback

-$15 000 000 $400 000 $$400 000 37 years


Table VI Averaged costing summaries Split SX Capital million US$ Cost 2.0 Operating and revenue million US$/a Save 4.2 Annualized risk million US$/a Reduced 0.3 Payback, years 0.45

Mill in raffinate Pre leach filtration Save 2.0 Save 5.0 Increased 5.0 N/A

Save 5.0 Save 4.5 Increased 2.0 N/A

Acid tails

Post-leach filtration

Cost 6.5 Save 2.9 Increased 1.5 4.6

Cost 15.0 Save 0.4 0 37

The data presented is indicative of 40,000 t/a copper production facility

382


365-384_Reolon:text

2/13/09

11:10 AM

Page 383

Marco Reolon Principal Process Engineering, GRD Minproc, South Africa Marco has extensive hydrometallurgical experience over a broad range of metals. This includes: pressure leaching of nickel, cobalt and gold, copper leach SX/EW, PGM refining and nickel/cobalt separation. Experience Principal Process Engineer, GRD Minproc (Pty) Ltd, December 2007 to present As principal process engineer he is responsible for various projects and studies, and for the development of the process engineering group focusing on copper/cobalt projects and studies in Southern Africa. Project achievements to date include: • Teal, Kalumines Hydrometallurgical Plant Pre-Feasibility Study, Democratic Republic of the Congo, (2008) • Platmin Congo, Deziwa Copper/Cobalt Definitive Feasibility Study, Democratic Republic of the Congo, (2008 - present) • CuCo Resources, Kisanfu Scoping Study, Democratic Republic of the Congo, (2008 - present). Senior Process Engineer, Fluor Canada Ltd, January 2006 to November 2007 As a senior process engineer in the Mining and Metals business unit he is responsible for the process aspects of predominately hydro-metallurgical and mineral processes. His major project achievements as Senior Process Engineer and Process Lead are: • Barrick, Pueblo Viejo Feasibility Project, Gold Pressure Oxidation Leach and Base Metals Recovery, Dominican Republic (2007 - Present) • Phelps Dodge, Safford Leach Project, Heap leach, Solvent Extraction and Electrowinning. Safford, Arizona, USA (2006-2007) • Sociedad Contractual Minera, El Abra Sulfolix Project, Feasibility (2006). Solaris Management Consultants Inc. Canada, September 2002 to December 2005 • Coordinating process engineers and process activities within the company. • Process engineering including Studies, Flow sheets, P&ID, equipment sizing and specification. • Gas field development for energy companies. • Developing standards, procedures and forms for process engineering activities. • Developing Data Books, Operating Manuals and start-up assistance for new facilities. Major project achievements as process engineer, construction or field engineer and commissioning engineer include: • Strachan Gas Plant Flare and Relief Upgrade Study, Rocky Mountain House, Alberta (2002–2003) • Arc, Prestville Oil Battery, Manning, Alberta (2003–2004) • SESI, PCB Destruction Plant, Annacis Island, British Columbia (2004–2005) • Apache, Leafland Gas Plant Upgrade, Rocky Mountain House, Alberta (2005) • Encana, Resthaven Gas Plant Project, Cache Creek, Alberta (2004–2005) • Encana, Steeprock Gas Plant Project, Tupper, British Columbia (2005) • Minor well site, compression, gas plant and gas gathering projects. (2002–2005). Flour Daniel Wright Ltd - Vancouver BC Canada, Melbourne and Perth Australia, Calgary and Ft McMurray AB Canada, December 1995 to August 2002 As a senior process engineer in the Mining and Metals business unit he was responsible for the process aspects of predominately hydro-metallurgical and mineral processes. Major project achievements as process and Lead process engineer are: • Albian Sands, Muskeg River Oil Sands Project, Fort McMurray, Alberta, Canada (2000–2002) • Phelps Dodge, Morenci Mine for Leach, Solvent Extraction and Electrowinning. Arizona, USA (1999–2000) • Miramar Con Mine Re-commissioning, Yellowknife, NWT, Canada (1999) • Murrin Murrin Nickel Cobalt Project, Western Australia (1997–1999) • Lomas Bayas Copper Heap Leach, Solvent Extraction and Electrowinning Project, Chile (1996–1997) • Molecular Recognition Technology Development (1995–1996) • Several pre-feasibility and feasibility studies. Process engineer B.M.I Division of E.L Bateman, February 1994 to November 1995 Responsible for the process aspects of the project Enhanced Platinum Metals Refinery Study and Project, Springs, South Africa (1994–1995). The process utilized a hydrochloric acid process environment to ensure the solubility and appropriate chemistry of the platinum group metals. OPTIMIZING ACID UTILIZATION AND METAL RECOVERY IN AFRICAN CU/CO FLOW SHEETS 383

365-384_Reolon:text

2/13/09

11:10 AM

Page 384

Extraction and separation process includes Molecular recognition technologies, Ion Exchange, Distillation and solid liquid separation. Sastech—Division of Sasol Ltd, January 1991 to February 1994 As a process engineer in the petrochemical, oil and gas business he was responsible for the process and business development aspects of the new ventures group. Worked on the following projects: • Acetic Acid Pilot Plant Commissioning, Secunda, South Africa (1994) • Hydrogen Sulfide Emission Reduction Study (1993) • Alpha Olefins Pilot Plant Re-commissioning, Sasolberg, South Africa (1993) • Alpha Olefins Pilot Plant Conversion, Sasolburg, South Africa (1992) • Refinery Feed Stream Simulation (1992) • Alpha Olefins Pilot Plant Commissioning, Sasolburg, South Africa (1991) • Cresylics Acid Plant Commissioning; Sasolburg, South Africa. Impala Platinum Ltd, Projects Department, Refineries, January 1988 to December 1990 As a chemical engineer in the projects department he was responsible for the process design and implementation of new projects. Worked on the following projects. • Selenium/Tellurium Removal Plant Project (1990), as project engineer. • Chlorine Hydrochloric Matte Leach Upgrade Project (1990), as project engineer. • Bulk Chemicals Handling Facilities (1990), as project engineer. • Platinum Metals Refinery Pilot Plant Design (1990), as area engineer. • Pressurized Oxygen Nickel Leach Autoclave commissioning (1989–1990), as commissioning engineer. • Rhodium/Iridium/Base Metals Separation Pilot Plant Design (1988), as assistant project engineer. • Rhodium/Iridium/Base Metals Separation Research and Development (1988), as research engineer. Overall experience Tasks: • During his career he gained experience in the following work processes or project and operating activities. • Research and Development (literature surveys, laboratory and pilot test program, outsourcing test work, data analysis, analytical technique development, flow sheet development, new product business development) • Pre-feasibility studies (estimation of deliverables). • Proposal (for studies, including schedules and execution plan) • Feasibility studies (design criteria, flow sheet, pilot plant data evaluation, process simulation, mass balance, process description, equipment list and sizing, piping and instrumentation diagrams (P&ID), equipment and piping layouts, equipment costing, operation costing) • Detailed engineering projects (as above, operating philosophy, process control and variables, instrument and equipment data sheets, design and equipment reviews, batch operating schedule, HAZOP and PHA studies, functional specifications, operating manuals) • Pre-commissioning (including construction completion and tracking, equipment inspection, water testing system turnover,) • Commissioning and Process Optimization (pre-start-up piping and equipment checks, cold commissioning, hot commissioning, equipment performance tests, operator training, optimization, data recovery, high rate trails, equipment evaluation and plant data recovery.) • Plant Training (operations, trouble shooting and plant optimization, commissioning, production utilities and maintenance scheduling, safety and loss control, plant supervisory training) Technologies: • Platinum group metals hydrometallurgical processes • Nickel/Cobalt hydrometallurgical processes • Copper/Cobalt hydrometallurgical processes • Zinc hydrometallurgical processes • Oil Sands processes • Molecular Recognition Technology. • Synthol technology • Sulfur recovery technology • Plant Utilities (water treatment, steam and condensate, cooling water, air, vacuum, bulk storage) • Autoclave Leach Reactors (Pressure oxidation, acid and Chlorine/hydrochloric) • Heap Leach (acid) 384


385-400_Swartz:text

2/13/09

11:11 AM

Page 385

SWARTZ, B., DONEGAN, S., and AMOS, S. Processing considerations for cobalt recovery from Congolese copperbelt ores. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Processing considerations for cobalt recovery from Congolese copperbelt ores B. SWARTZ*, S. DONEGAN†, and S. AMOS* †GRD Minproc Pty (Ltd), Perth, Australia *GRD Minproc Pty (Ltd), Bryanston, South Africa

The resource market has seen a year of turmoil in 2008. During the first five months of the year a spectacular increase in cobalt price was observed peaking at around $50/lb but since July there has been an equally spectacular decrease in the cobalt price to current levels of around $16/lb. Prior to 2008, a consistent growth period, stretching over a few years which, driven mainly by an increased demand from the developing world, in particular China, resulted in an increased demand for both copper and cobalt.

Introduction The increased demand for copper and cobalt together with the increase in political stability in the region, have resulted in renewed interest in developing the copper and cobalt deposits in the Katanga Province of the Democratic Republic of the Congo (DRC). A number of projects, in various stages of development, are currently underway in the DRC. The primary product from most of these projects is copper; however, cobalt is often a significant and sometimes major contributor to the project revenue. Cobalt in the form of heterogenite is commonly associated with copper bearing oxide ores and hence it is commonly co-extracted with copper. In order to extract cobalt with copper, a reducing environment is created for reduction of Co3+ to Co2+ as divalent cobalt is soluble at the pH values at which copper dissolution is conducted. Cobalt is then generally recovered from a copper solvent extraction (SX) raffinate bleed stream. The general trend in the development of these copper/cobalt projects is to produce an intermediate cobalt product at start-up and progress to cobalt metal production once the operation has proven to be viable for a period. The decision not to go all the way to cobalt metal is generally due to the perceived risk associated with operating a complex cobalt refinery in central Africa. Shifting from production of intermediate cobalt products to pure metal is largely driven by increased revenues The decision to produce an intermediate cobalt product vs. metal significantly affects the processing routes employed subsequent to copper solvent extraction (SX). If an intermediate product is to be produced, then removal of iron, manganese and copper will occur with little if no emphasis placed on reducing the levels of other impurities. Production of cathode necessitates stringent impurity control in the cobalt electrolyte, which is critical in ensuring PROCESSING CONSIDERATIONS FOR COBALT RECOVERY

385

385-400_Swartz:text

2/13/09

11:11 AM

Page 386

that cathode metal specifications are met. Typical impurities, other than those mentioned above, are nickel, zinc and aluminum. The methods for removal of these are precipitation, SX and/or ion exchange (IX), and the choice of technology depends on impurity concentration, relative capital costs for removal and associated operational preferences. This paper discusses some of the processing options available for cobalt production and provides an evaluation of the associated costs, risks and benefits applicable to the DRC. Democratic Republic of the Congo Introduction The Democratic Republic of the Congo was initially established as the Belgian Congo in 1908. It subsequently gained independence in 1960. The early years of independence were marked by social and political turmoil. In a coup in 1965, army colonel Joseph Mobutu seized power, renaming himself Mobutu Sese Seko and the country Zaire. Under his presidency, social and economic conditions in the country deteriorated further. In 1997, Mobuto was toppled by a rebel leader, Laurent Kabila. Kabila restored the name of Congo in the form of the Democratic Republic of the Congo (DRC). He in turn faced a rebel movement backed by Rwanda and Uganda. Troops from a number of African allies were flown into the country and the rebellion was put down and a cease fire signed in 1999. Kabila himself was assassinated in early 2001, and his son, Joseph Kabila, was installed as President of the Republic. Joseph Kabila subsequently negotiated the removal of Rwandan forces, and a peace agreement was signed in South Africa in 2002. A transitional government was formed with the objective of eventually holding democratic elections in the country. These elections were held in July 2006, and resulted in the two leading presidential contenders each receiving less than 50% of the vote, although Joseph Kabila did gain a significant majority over his rivals. In the terms of the election rules, a run-off between the two preferred candidates took place in November 2006 with Kabila gaining an outright majority. Economy Since 2001, when Joseph Kabila became president, the economic policy of the country has aimed at stabilization. Since about that date, there is little doubt that the mining and construction sectors are improving, albeit from a low base. Gross national income is US$120 per capita, but since 2002, the economy has shown signs of expansion with a GDP growth in 2002 of 3.5%, 2003: 5.7%, 2004: 6.8%, 2005: 6.5%. The latter two figures are provisional. The independent central bank has bought inflation under control in recent years, again indicating a normalizing of the economic and social situation within the country. The corporate tax and investment codes are being revised to continue to liberalize the domestic business environment and attract foreign investment, and an increasing number of foreign companies are establishing themselves in the country. Gecamines Prior to the departure of the Belgians from the Congo in 1960, the copper industry in the Katanga area was owned and operated by Union Miniere de Haut Katanga (UMHK), a company dating from 1906. After independence, nationalization of the copper industry was inevitable, and UMHK was eventually dispossessed of its ownership of the industry, and a state owned company, initially Generale Congolaise des Minerals and later La Generale des 386


385-400_Swartz:text

2/13/09

11:11 AM

Page 387

Carriers et des Mines (Gecamines) took over as owners and operators of all copper mines in Katanga. Gecamines initially prospered throughout the 1970s and 1980s, but since the late 1980s, the company endured through a series of increasing crises, until finally, by the late 1990s, metal production was only 35 000 t of copper and 3 940 t of cobalt, compared to 437 000 t of copper in 1987, plus 11 900 t of cobalt in the same year. As a result of this poor performance, the Government began to seek ways of improving the situation both within Gecamines and within the copper industry as a whole. As a result of these initiatives, foreign companies have been encouraged to enter into agreements with Gecamines in an attempt to restore the industry. After some stumbling, these initiatives are now beginning to bear fruit. Code Minier A Code Minier has been introduced by the Government and provides a transparent legal framework for investors in mining. The code has been gazetted in French and English, making it easier for foreign Anglophone companies to review the legislation. Cobalt processing options In the DRC there are typically three major saleable product types produced, namely: • A cobalt concentrate via gravity separation • An intermediate cobalt salt • Cobalt metal. The value of the product increases as the level of refining increases; however, this is also proportional to the capital and operating cost associated with the facility required to produce the end product. The complexity of the operation also increases significantly from concentrate production through intermediate to pure metal. Introduction Gravity separation is by far the cheapest and lowest cost method of producing a saleable product from ore. Not all ore types are amenable to gravity separation and generally low recoveries are achieved even for those materials that are amenable. Product grade and recovery are ore specific and generally quite poor. Dense media separation is the most common technology used and is often employed as a preliminary measure to generate cash flow before proceeding to the production of a purer product. This method is mentioned for completeness and will not be discussed further. There are a number of different cobalt intermediate products that can be produced, namely: • Cobalt hydroxide • Cobalt carbonate • Cobalt sulphide • Cobalt sulphate. In addition, there are a number of refined cobalt metal products that can be produced from a copper raffinate bleed stream or any of the above intermediates: • Cobalt cathode by electrowinning (EW) • Cobalt powder/briquettes by hydrogen reduction. Cobalt refining is not only complex, but product types vary considerably in terms of purity (and hence market premium and transport costs) and by-product/effluent generation. It is important, therefore, to establish factors with the greatest importance to and impact on the project to be able to identify the optimum solution. Four main categories should be considered: PROCESSING CONSIDERATIONS FOR COBALT RECOVERY

387

385-400_Swartz:text

2/13/09

11:11 AM

Page 388

• Operationally safe, robust and simple • Technically robust and proven technology • Environmentally compliant • Commercially acceptable. Unfortunately there is no single technology, flowsheet and product that rank highly in all of the above categories. It is therefore necessary to select a flowsheet that has an ‘acceptable ranking’ in all categories while providing the best return to the project. A decision diagram (Figure 1) illustrates the options, ranked in environmental and operability terms. Options producing high levels of soluble salt form the left-hand leg of the decision chart, and those processes that minimize the production of soluble salt effluent are indicated in the right-hand leg. Those processes on the right of the decision chart reduce soluble salt generation by an order of magnitude. It should be noted that none of the options provides a zero discharge scenario. Process options with lower risk profiles are located at the top of the chart, whereas higher risk options such as sulphide precipitation with hydrogen sulphide gas are located in the bottom. For the purpose of this ranking, process risk includes operational safety, operability and robustness. Intermediate product options There are four main cobalt intermediate product forms that can be produced from either selective precipitation or evaporation/crystallization techniques:

Figure 1. Cobalt intermediate product decision tree

388


385-400_Swartz:text

2/13/09

11:11 AM

Page 389

• Cobalt hydroxide (precipitation using lime, magnesia or caustic) • Cobalt carbonate (precipitation using sodium carbonate) • Cobalt sulphide (precipitation using sodium hydrosulphide) • Cobalt sulphate (crystallization of a cobalt sulphate solution). For all of the above products the option exists to produce a crude or ‘pure’ form. A crude form is defined as the product obtained after limited impurity removal (only iron); a pure product is defined as the product obtained after removal of the major typical impurities, namely, iron, manganese, copper and aluminium. The process description for impurity removal is similar for all four intermediate products and is summarized below. Impurity removal prior to intermediate product production Fe/Mn/(Al) removal Iron and manganese (and aluminium in the presence of phosphate) are removed from the cobalt bleed stream (copper SX raffinate) using a patented process that uses an air/SO2 mixture to oxidize and precipitate Fe and Mn/Al in a single step. It must be noted that no Mn or Al removal is required when producing a CoS product as these two metals do not precipitate as sulphides. This section is usually equipped with approximately four agitated, overflow tanks in series. The feed liquor is often heated to around 50°C via indirect heating and a residence time of up to 5 hours is required for substantial (>90%) manganese removal. The tanks will be equipped with gas spargers to inject the air/SO2 mixture. The addition of the air/SO2 gas mixture is controlled by adding the air at a fixed volume and then controlling the SO 2 into the tank at an approximate SO 2 content of ~0.5% by volume. The redox controller ensures that the reaction remains in an oxidizing regime with a measured redox potential above 600 mV by controlling the flow of SO2. The pH is controlled by adding limestone to the solution to increase the pH to around 2.5 to 3.0 (a pH of ~3.5 is necessary for aluminum removal in the presence of phosphate). The pH control at this stage is important to ensure that as little Co as possible is precipitated with the Fe and Mn (Al). The slurry coming from the Fe/Mn/(Al) removal tanks flows by gravity to a high rate thickener where the supernatant liquor is decanted from the precipitated solids. The solids are concentrated in the underflow to around 30%, following which they are pumped to the Fe/Mn/(Al) removal filters. A thickener underflow recycle ratio of approximately 200% is required for crystal growth. The Fe/Mn/(Al) removal filters are used to separate the supernatant liquor left in the thickened slurry from the solids. Two filters will normally be used in this application: one operating and a standby. A counter current washing regime is used to minimize the wash solution required while still achieving high wash efficiencies. The washed cake is disposed of with the neutralized leach residue. Secondary copper solvent extraction Copper in solution can be removed via precipitation with lime (similar to Fe and Al removal) or via SX. The SX route is generally the preferred route as the capital and operating costs associated with the precipitation and thickening route are usually higher, and it imposes a significant recirculating load of cobalt. PROCESSING CONSIDERATIONS FOR COBALT RECOVERY

389

385-400_Swartz:text

2/13/09

11:11 AM

Page 390

The aqueous liquor from the Fe/Mn/(Al) thickener is contacted with a bleed of the stripped organic from the primary copper SX circuit to remove minor quantities of copper. This increases copper recovery and minimizes contamination of the cobalt product. Copper extraction is usually completed in one or two mixer-settlers. Cu-loaded organic is mixed with the loaded organic from the primary Cu SX in the primary loaded organic wash or strip mixer-settler. Aluminium removal This step is often not required as aluminium is often precipitated to a sufficient extent as the phosphate complex during iron removal. Al precipitation is not required when producing CoS as Al is not precipitated as the sulphide. Al precipitation is completed in agitated tanks with a combined residence time of around two hours. The precipitation is run at an elevated temperature of around 50ºC usually via direct steam injection, and controlled to a pH of around 5.5 using slaked lime (Ca(OH)2). The underflow is concentrated to around 40% solids in the Al removal thickener. A large underflow recycle is required to aid crystal growth and the underflow is recycled to the first tank in the Fe and Mn removal stage. The thickener O/F will still contain solids that need to be removed from the process before the solution is pumped to the cobalt precipitation area. A pinned bed type clarifier is used as it can efficiently take out these solids in a single stage. Intermediate product production Cobalt hydroxide precipitation The Co(OH)2 intermediate product is formed by contacting Fe, Mn, Al and Cu free solution in a series of agitated tanks at elevated temperature with a slurry of Ca(OH)2.

The reaction co-precipitates gypsum, which contaminates and significantly lowers the grade of the product. Magnesium and residual manganese are also partially precipitated as hydroxides, the extent of which is dependent on the terminal pH of the reaction sequence. The combined cobalt hydroxide and gypsum slurry is thickened, filtered and washed, and the filter cake dried before being loaded into bulk bags for dispatch. A preferred method for producing the hydroxide product is by using milk of magnesia as the precipitant. This route avoids gypsum contamination of the product, thereby producing a purer cobalt hydroxide product than the milk of lime precipitation. Sodium hydroxide may also be used as the precipitant with advantages similar to magnesia. However, this will result in a final effluent containing sodium sulphate, which will have to be disposed of or contained. For all three precipitants, product purity may be increased further by utilizing a two-stage precipitation process. The intent here it to precipitate ~90% of the cobalt at pH ~7.5 in the first stage as a relatively pure cobalt hydroxide product, low in manganese and magnesium. The resultant slurry is thickened, filtered, bagged and dried for dispatch. The first stage thickener overflow solution is then subjected to a second precipitation step to recover the remaining ~10% of cobalt. This is achieved with milk of lime addition at pH 8.0 to 8.5. The slurry is thickened and returned to a suitable place in the cobalt purification circuit. 390


385-400_Swartz:text

2/13/09

11:11 AM

Page 391

Cobalt carbonate precipitation A pure CoCO3 intermediate product is formed by contacting Fe, Mn, Al and Cu free solution in a series of agitated tanks at elevated temperature with a slurry of Na2CO3. The resulting CoCO3 slurry is thickened, filtered and dried before being loaded into bulk bags for dispatch. If the product is to be refined further on site the drying is generally not required. Cobalt sulphide precipitation Cobalt sulphide can be precipitated using sodium hydrosulphide (NaHS) or hydrogen sulphide (H2S). While it is possible to perform a hydrogen sulphide cobalt precipitation at atmospheric pressure, this would recover only approximately 85 to 90% of the cobalt to the sulphide. The remainder would thus need to be recovered by an additional precipitation step using, for example, milk of lime, and perhaps recycled in this instance to the Cu scavenger neutralization circuit for redissolution and subsequent reprecipitation. The precipitation of a sulphide intermediate using hydrogen sulphide is a series of relatively complex and, due to the toxic nature of hydrogen sulphide, high risk operations. The requirement for hydrogen for the generation of hydrogen sulphide necessitates the installation and operation of a hydrogen plant in addition to a hydrogen sulphide plant. The large volume of purge gas for maintenance operations may also dictate the requirement for a small nitrogen (air separation) plant. Each of these gas plants is complex and intricate, requiring a skilled labor complement for their reliable and safe operation. For these reasons the use of hydrogen sulphide will not be discussed further. Neutralized secondary SX raffinate constitutes the feed to the Co sulphide precipitation circuit. (Note that no specific Fe, Mn or Al removal step is required as Fe(II), Mn and Al will not precipitate as a sulphide; however, some Fe(III) and Al removal may occur during neutralization.) Pure CoS intermediate product is formed by contacting Fe(III)-free solution in a series of agitated tanks at elevated temperature with a solution of NaHS. The resulting CoS slurry is thickened and filtered, and the filter cake loaded into bulk bags for dispatch. Cobalt sulphate crystallization Iron, manganese, aluminium and copper free solution constitutes the feed to the CoSO4 crystallizer plant. The CoSO4 intermediate product is formed by first concentrating the feed solution in a large falling film evaporator, and then crystallizing the concentrated liquor in a smaller evaporator/crystallizer. Washing, drying and bagging facilities are provided to ensure a pure crystal product with minimum contamination and moisture. The operation of the crystallizer is complicated by the fact that a calcium saturated solution is formed, resulting in gypsum precipitation and scaling in the crystallizer. This process will also result in the co-crystallization of all impurities with the cobalt product, including magnesium and calcium resulting in a low grade product. Intermediate product comparison The four product options listed above are compared and discussed with reference to a number of key project drivers. Capital cost (excluding effluent treatment) Capital costs associated with the carbonate and hydroxide routes are similar. Both routes are only partially selective for manganese and magnesium and thus a similar degree of purification is required prior to final product precipitation. The sulphate route has the highest PROCESSING CONSIDERATIONS FOR COBALT RECOVERY

391

385-400_Swartz:text

2/13/09

11:11 AM

Page 392

capital cost due to the requirement for an evaporator and crystallizer in addition to the preceding impurity removal steps. The sulphide flowsheet is the cheapest option in terms of capital cost, despite the requirement for closed vessels and off-gas scrubbing equipment. Cobalt sulphide is selectively precipitated in the presence of manganese and aluminium and thus the removal of these elements is not required. Operating cost The sulphate route has the advantage that no reagents are required for final product production. However, a significant amount of energy is required for solution evaporation, and this cost makes it the most expensive to operate. Operating costs for hydroxide, carbonate and sulphide are similar; if lime is utilized for hydroxide precipitation that route is slightly cheaper. Metal recovery The recoveries of all four routes are similar. For the hydroxide route a two stage precipitation process is recommended to recover and recycle soluble cobalt. The disadvantage of the sulphide route is that any off specification material is difficult to redissolve without specialized equipment (autoclave). This is often the most important factor to consider when choosing a preferred route. HSE issues are more prevalent for the sulphide intermediate due to the potential for H 2 S generation. In addition to this, the use of NaHS will result in a final effluent containing sodium sulphate, which will have to be disposed of or contained. The problem associated with sodium sulphate disposal is common for the hydroxide-caustic and carbonate route as well and can often be a significant additional cost. Hydroxide (lime/magnesia) and sulphate intermediates usually have lower HSE issues. Product purity The sulphide product contains the highest percentage of cobalt followed by the hydroxide (caustic or magnesia), carbonate then the low-grade hydroxide (lime), while sulphate has the lowest contained cobalt. The major impurity associated with the sulphide intermediate is magnesium; with sulphate it is manganese, magnesium and calcium; with carbonate it is manganese and calcium; and with the hydroxide it is manganese and either calcium or magnesium, depending on the reagent used for precipitation, i.e. lime or magnesia. Apart from the rejection of Al and Fe(II) by sulphide precipitation, there is negligible selectivity in this application for the remaining minor elements (Cu, Zn, Ni) by all options. Operability and maintenance The sulphide process requires off-gas scrubbing and tight pH control during precipitation. In addition to this, the handling and corrosive nature of NaHS must be considered. For the production of sulphate, the operation of a crystallizer and evaporator is relatively complex and maintenance intensive so is rated more difficult than the carbonate and hydroxide routes. Large recycles required for the hydroxide route make the circuit cumbersome from an operability perspective, and the relatively poor settling and filtration characteristics of the hydroxide product aggravate this. Descaling will be required in all cases. Sales and marketing This will depend on product purity, but in general terms marketing information indicates that the customer base for hydroxide, carbonate and sulphate is larger than for the sulphide. The value received from sales for carbonate and sulphate is higher than that for hydroxide and sulphide. 392


385-400_Swartz:text

2/13/09

11:11 AM

Page 393

Future cobalt metal production In a number of cases clients wish to adopt a strategy of cobalt intermediate production in the initial years of project life with the conversion to metal production as revenue from copper and intermediate sales stabilizes and operational experience is gained. The sulphide intermediate would potentially require ultra-fine grinding followed by pressure oxidative leaching. Intermediate hydroxide and sulphate products would require additional manganese impurity removal steps prior to cobalt EW. The carbonate would be the ‘easiest’ to refine. Those products containing significant gypsum will obviously produce calcium saturated liquors upon dissolution, requiring additional consideration and control within the cobalt refinery. The above comparison is summarized in Table I below; the intermediate options are rated 1–4, with 4 being the preferred option. Recommended intermediate product The cobalt sulphate route is eliminated on the basis of cost and the production of a low purity product. Cobalt sulphide production is by far the most complex process, and HSE as well as operability issues eliminate this option. The two best intermediate product routes are the carbonate and the hydroxide. The choice between the two has to be made considering environmental issues associated with the disposal of sodium sulphate effluent. If the plant is close to a large water body, the disposal of the effluent via dilution may be possible. However, with increasingly stringent environmental legislation, this is becoming less likely unless the effluent can be pumped to the sea. The production of the hydroxide via magnesia (MgO) precipitation will result in a reasonably high quality product, and the liming of the effluent to a pH of around 10.5 will result in a magnesium free solution, which can be disposed of a conventional tailing dam. The precipitate from the neutralization can be stored in a relatively small lined facility. The production of cobalt hydroxide using a two-stage process is in most cases the recommended route. Pure metal product options Cobalt metal can be produced via hydrogen reduction (powder or briquettes) or EW (full plate cathode or rounds). EW is believed to be the preferred method in almost all cases in terms of lower capital and operating cost, higher metal recoveries and less stringent operability and Table I Cobalt intermediate product ratings Parameter Capex Opex Recovery HSE Purity Op and main Sales and marketing Future metal production Total

Co(OH)2

CoCO3

CoS

CoSO4

2 3 2 4 2 3 2 2 20

3 4 4 2 3 4 3 4 27

4 2 1 1 4 1 1 1 15

1 1 3 3 1 2 4 3 18

PROCESSING CONSIDERATIONS FOR COBALT RECOVERY

393

385-400_Swartz:text

2/13/09

11:11 AM

Page 394

maintenance requirements, and for the purpose of this comparison only EW will be considered. The production of cobalt rounds is a patented process, so this paper will concentrate on the production of full plate cathode. Two different processing routes will be considered; these are described below: • Cobalt hydroxide precipitation followed by redissolution, impurity removal and EW • Cobalt SX, impurity removal and EW. Precipitation and redissolution Cobalt precipitation The primary cobalt hydroxide precipitate is the feed material to the cobalt metal plant. The use of magnesia is unnecessary and a cheaper precipitant, i.e. milk of lime, is more suitable, although calcium saturation issues and larger volumes of redissolution residue must be considered. Primary precipitate, in the form of a filter cake, is advanced to the cobalt redissolution circuit; secondary precipitate (high manganese) is recycled to the iron/manganese/(aluminum) precipitation circuit feed for recovery of the contained cobalt. Cobalt redissolution Primary cobalt hydroxide filter cake is dissolved using cobalt spent electrolyte and sulphuric acid in a cascade of agitated tanks. Oxidation and partial neutralization of the leached slurry may be used to scavenge iron and manganese using milk of lime and air/SO2 gas mixture. Aluminum will redissolve and will remain largely ‘inert’ in the refinery, unable to plate with the cobalt. It will either exit as entrained liquor in the redissolution residue filter cake, or as part of the spent electrolyte bleed. Leached slurry is thickened and filtered and the filter cake is recycled upstream to recover any undissolved cobalt. Filtrate is clarified in a polishing filter for final solids removal, and is then diluted with clean water to ensure liquor advancing to the IX circuits is comfortably below calcium saturation. Zinc ion exchange (IX) Zinc and residual copper are removed from the cobalt rich liquor using a macroporous aminophosphonic acid chelating resin within fixed-bed columns operated in a typical lead-lag configuration. A split elution technique using sulphuric acid is used to maximize zinc-cobalt selectivity. Weak eluate is recycled to cobalt precipitation, and strong eluate is disposed via tails neutralization. Regeneration of the resin is achieved using a caustic solution. Nickel ion exchange (IX) Nickel is partially removed from the zinc and copper-free cobalt rich liquor using a macroporous bis-picolylamine chelating resin. As electrowinning is selective for cobalt over nickel, removal of nickel to ~0.2 g/l is generally sufficient to meet cathode specification. Similar configuration and elution/regeneration techniques to the Zn IX circuit are used; however, the inherently low nickel-cobalt selectivity and high unit cost of the resin, encourage the use of smaller bed volumes and a consequent increased number of columns. Cobalt electrowinning (EW) Electrowinning is conducted at ~70°C in standard undivided cells. Stainless steel permanent blanks are used as cathodes, with Pb-Sn-Ca anodes at 120 mm cathode-cathode spacing. A 394


385-400_Swartz:text

2/13/09

11:11 AM

Page 395

rich electrolyte tenor of at least 50 g/l Co is achieved by adjusting spent electrolyte recycle to the re-solution circuit. This permits EW operation with a cell drop of 5 g/l, producing a spent electrolyte of 45 g/l Co and ~8 g/l H2SO4. Under these conditions, a current density of 250 A/m2 will typically give a current efficiency of ~80%. The use of divided cell technology, particularly anode bags, would potentially allow higher cell drops to be achieved (>20 g/l), thereby significantly reducing the flow through the purification circuit. However, given the highly stressed nature of full-deposit cobalt cathode and the resultant potential for frequent damage to the surrounding bags, significant development is required to reduce process risk before this technology can be considered for application in the DRC. Cathodes are harvested live on a six/seven-day cycle. They are washed with hot water to remove surface electrolyte and are then manually stripped. Stripped blanks can be treated with dilute nitric acid for removal of residual metal deposits. Prior to return to the cells, stripped blanks are dipped in gelatine to create a favorable deposit-blank bond strength. Stripped cobalt cathode pieces are crushed to -70 mm, vacuum degassed to remove contained hydrogen and nitrogen, and burnished in rotating drums to achieve the desired product finish. Final product is packaged in sealed drums for dispatch. Manganese that is oxidized at the anode forms insoluble MnO2. This is periodically washed from the anodes and hosed from the cell bottoms. Spent electrolyte is returned to the redissolution circuit as required. Surplus liquor is bled to the cobalt precipitation circuit, providing an important outlet for such species as magnesium, aluminium, sodium, sulphate and chloride. Cobalt SX EW For this refining configuration, liquor exiting the secondary copper SX stage forms the feed to the cobalt SX circuit. Sufficient removal of organic from the secondary copper SX raffinate must be performed to limit the transfer of copper extractant (oxime) to the Cobalt SX organic. A diluent wash is considered to have good potential in this application due to the availability of relatively large volumes of make-up diluent required in the primary copper SX circuit. Cobalt SX Cobalt is recovered from the partially purified bleed liquor in a multiple-stage SX circuit utilizing a phosphinic acid organic extractant dissolved in a suitable hydrocarbon diluent. Zinc, iron, aluminum, copper and manganese are all co-extracted, while calcium, nickel and some magnesium are rejected. The extraction circuit usually consists of four countercurrent stages. Each stage comprises a single mixer providing a total of 1 minute mixing time and a settler for phase disengagement. To maximize cobalt-nickel selectivity, pH is controlled within each stage by direct addition of caustic solution to the mixers. Two-stage organic removal is used for the raffinate, generally comprising a diluent wash stage and Jameson flotation cell. Clean raffinate can be recycled within the process (e.g. CCD wash water), with surplus advanced to tails neutralization. The use of sodium hydroxide results in a raffinate containing sodium sulphate which, as per several of the intermediate options, will have to be disposed of or contained. Care must be used to avoid recycling cobalt SX extractant directly into the copper SX circuits. Circuits containing solids generally act as good organic adsorbents, thereby significantly reducing this risk. PROCESSING CONSIDERATIONS FOR COBALT RECOVERY

395

385-400_Swartz:text

2/13/09

11:11 AM

Page 396

Cobalt loaded organic is scrubbed in a single stage to remove co-extracted and entrained nickel (and magnesium). Demineralized water and sulfuric acid are contacted with the loaded organic at a suitably low pH to achieve the required Co:Ni ratio on the scrubbed organic. A bleed of cobalt loaded strip liquor can be added to the scrub stage to improve scrubbing efficiency if required. Scrub raffinate is returned to the first cobalt extraction stage. Scrubbed organic is stripped in two countercurrent stages to remove all loaded metals. Cobalt spent electrolyte is used as strip liquor, the flow rate of which is adjusted to achieve the required cobalt tenor for downstream electrowinning. Dilute sulphuric acid solution is used to control pH in the first strip stage to ensure required stripping efficiency is achieved. Stripped organic is returned to extraction via a stripped organic tank. Zinc SX Zinc (and ferric iron) is removed from the cobalt SX loaded strip liquor in a multiple-stage SX circuit. While Di-Ethyl Hexyl Phosphoric Acid (DEHPA) is an option in this application, the use of phosphinic acid is preferred due to the ability to reject calcium, relative ease of iron stripping, and improved compatibility with cobalt SX (no extractant cross-contamination issues, ability to use same crud treatment equipment, etc.). The phosphinic acid extracts all zinc and ferric, whilst aluminium, copper, manganese and cobalt are rejected. To ensure the target iron level is maintained in the zinc SX raffinate, all ferrous in the cobalt loaded strip liquor is oxidized to the ferric form just prior to zinc extraction using hydrogen peroxide in an in-line mixer. Zinc is extracted in three countercurrent pH-controlled stages. Thorough removal of organic is required for the downstream circuits, so a three-stage system is used generally consisting of a Jameson cell, liquid-liquid coalescer and carbon columns. Zinc loaded organic is scrubbed with water and sulphuric acid in a single stage to remove co-extracted and entrained cobalt. Scrub raffinate is returned to the first zinc extraction stage. Scrubbed organic is stripped with acidified water in two countercurrent stages to remove all loaded metals. Stripped organic is returned to extraction via a stripped organic tank. Entrained organic in the zinc loaded strip liquor is recovered using a liquid-liquid coalescer, and the cleaned liquor is disposed via tails neutralization. Copper IX Residual copper is removed from the zinc SX raffinate using a macroporous aminophosphonic acid chelating resin within fixed-bed columns operated in a typical lead-lag configuration. A single stage elution is used due to the ability to recover all eluted copper (and cobalt) upstream. Regeneration of the resin is achieved using a caustic solution. Cobalt EW Cobalt electrowinning is carried out in identical fashion to that described in the precipitation/re-dissolution process. Spent electrolyte is returned to the cobalt SX strip circuit as required. Surplus liquor is bled to the cobalt SX extraction circuit (blended with PLS) providing the only significant outlet for such species as magnesium, aluminum, sodium, sulphate and chloride. Metal production route comparison The two general production routes described above are compared and discussed with reference to a number of key project drivers. 396


385-400_Swartz:text

2/13/09

11:11 AM

Page 397

Capital cost (excluding effluent disposal) Capital costs associated with both the precipitation and solvent extraction routes are similar. For the second route, the additional cost of the cobalt SX circuit (organic inventory) is effectively offset by the elimination of the nickel IX circuit, comprising a very expensive resin inventory of its own. Being solely dependant upon fixed bed IX, the precipitation/redissolution route is considered to be more capital sensitive to the florate and composition of the cobalt rich liquor feeding the purification steps. For this reason, the adoption of divided cell technology to increase cell drop in the cobalt tank house would bring significant capital savings to the precipitation/redissolution route (with smaller savings to the SX route). Operating cost The SX route consumes much larger amounts of caustic than the precipitation/redissolution based route due to its use for primary cobalt recovery in the cobalt SX circuit. This is offset by lower consumption of lime (effectively replaced by caustic) sulphuric acid and limestone (reduced recirculating loads of cobalt and impurities). The operating cost differential between the two options is therefore highly sensitive to the prevailing price of caustic; below US$800/t (including delivery) the SX route is generally favoured. Metal recovery Due to the higher selectivity of the impurity removal steps, cobalt recovery for the SX route is somewhat higher (~10%) than that for the precipitation/redissolution route. Cobalt loss with both nickel and zinc is reduced considerably by the use of SX. Cobalt losses are also reduced due to a significant reduction in the recirculating load of cobalt (elimination of secondary cobalt precipitate, cobalt re-solution residue, weak eluates and a smaller electrolyte bleed). Product purity Both production routes have been compared on the basis of equivalent product purity. The SX route, however, provides additional flexibility with respect to impurity removal, and could tolerate higher than design levels of both zinc and nickel with greater ease than the precipitation/re-solution based route. Operability and maintenance Both processing routes may be considered complex refineries, requiring steady operation and tight control to achieve desired performance. The elimination of virtually all solid-liquid circuits and consequent batch filtration steps is expected to provide an operability advantage to the SX route. The SX route also produces cobalt rich liquor (electrolyte) very low in calcium, as opposed to the high calcium liquor produced via the precipitation/redissolution route. The need for periodic de-scaling of the cobalt precipitation and redissolution circuits, and the potential for downstream gypsum precipitation within the IX and EW circuits places the precipitation/resolution route at a further disadvantage in terms of both operability and maintenance. Health, safety and environmental issues The presence of larger volumes of organic solutions in the SX route requires increased measures to control personnel exposure and fire risks. PROCESSING CONSIDERATIONS FOR COBALT RECOVERY

397

385-400_Swartz:text

2/13/09

11:11 AM

Page 398

Environmentally, the SX route produces an effluent containing large amounts of sodium sulphate that must be adequately disposed of or contained, whereas in its place the precipitation/redissolution route produces gypsum, the disposal of which is quite straightforward. The above comparison is summarized in Table II below; the metal production options are rated 1 or 2, with 2 being the preferred option. Recommended metal production process Technically, the SX based route is equal or superior on most parameters, and excluding the cost of effluent treatment, would be expected to give a superior return to the precipitation/redissolution based route. As with the choice of intermediate, however, the selection of a refining route is driven by the costs and environmental issues associated with the disposal of a sodium sulphate containing effluent. The use of ammonia in place of caustic for the SX route would allow ammonia recovery using a lime-boil process, producing a gypsum residue for disposal. The low concentration and high volume of effluent to be treated, however, do not make this option particularly attractive, and a cheap source of low pressure steam would be essential. On this basis, the SX route is effectively eliminated, and the recommended route is precipitation of a cobalt hydroxide using lime, redissolution in spent electrolyte, purification and metal recovery via electrowinning. Comparative nett present value (NPV) between hydroxide intermediate and metal A high level NPV model has been applied to the following scenarios: • Cobalt hydroxide intermediate production • Cobalt hydroxide intermediate production followed by redissolution and cathode production (i.e. conversion to a metal production circuit after operating an intermediate production circuit) • Cobalt cathode production via intermediate production (i.e. direct metal production with no saleable intermediate). The following assumptions have been applied • 15% discount rate • 9 000 tpa cobalt metal production • US$16/lb cobalt metal • Cobalt hydroxide intermediate realizes 70% of the contained metal value. Table III summarizes the input data for the NPV calculation. Table II Cobalt metal production route ratings Parameter Capex Opex Recovery Purity Op and main HSE Total

398

Precipitation and redissolution

SX

2 1 1 1 1 2 8

1 2 2 2 2 1 10


385-400_Swartz:text

2/13/09

11:11 AM

Page 399

Table III NPV input data Product Hydroxide intermediate Cathode from intermediate Cathode via re-dissolution

Capex MUS$

Opex MUS$/a

Co product t/a

Revenue MUS$/a

28.4 99.8 87.4

60.3 81.0 81.0

9000 9000 9000

222.2 317.4 317.4

Capital cost The costs listed above are high level estimates and meant for comparative purposes. The estimate consists of costs for earthworks, civils, structural steel, plate work, mechanical equipment, piping, electrical and instrumentation and buildings. It excludes indirect costs but includes a 15% contingency. The battery limits covered are from the copper SX raffinate pond to product dispatch, all internal recycles are costed, and effluent reporting to the copper circuit is costed at a convenient point between the two plants. All reagent storage and utilities are included. The cost for the ‘cathode from intermediate’ option is US$12.4 million higher than the ‘cathode via redissolution’ option, and this additional amount covers a second stage cobalt precipitation, drying and bagging facility. Operating cost The operating cost is defined as the incremental cost (total cost being for the copper and cobalt facility) required to produce the cobalt product. This includes labour, power, operating and maintenance spares, vehicles, reagents (including SO2 for leaching), consumables (including laboratory) and other. The operating cost estimate is to the same level of accuracy as the capital cost and is meant for comparative purposes. Production and revenue A cobalt production of 9 000 tpa has been assumed at a cobalt price of US$16/lb (spot price on 21 November 2008). There will be some minor differences in cobalt recovery; however, at the level of accuracy considered above this is regarded as being negligible. NPV calculation The NPV model shows an overwhelming financial benefit for production of cobalt metal. Figure 2 summarizes the values. From Figure 2 it can be seen that the NPV for the cobalt metal options is far larger than for the intermediate option. This is due to the large increase in revenue associated with metal production, as only 70% of the cobalt value has been assigned to the intermediate product, and this is believed to be an optimistic view. The NPV for the intermediate option is MUS$647 as opposed to MUS$894 for the cathode from intermediate option and MUS$905 from the cathode via re-solution option. The relative change in NPV between the two metal production options is small. Conclusions and recommendations Numerous factors have been discussed, which affect the type of cobalt product that is produced. It is often environmental considerations and operational complexity that determine the optimum route. PROCESSING CONSIDERATIONS FOR COBALT RECOVERY

399

385-400_Swartz:text

2/13/09

11:11 AM

Page 400

It most cases copper is the major contributor to project revenue but cobalt often makes a significant contribution. In the early years of operation it is wise to follow the lowest complexity route possible, and thus the production of cobalt hydroxide intermediate is recommended. This also reduces the upfront capital required and lowers overall project risk. Once operational experience has been obtained and cash flow realized, the conversion from the intermediate product to metal should be considered as this returns greater value. The complexity of cobalt metal production should not be underestimated, particularly in the DRC where experienced operators are scarce. Cobalt metal production should not be undertaken before understanding the complexity and risk.

Figure 2. Product NPV values

Steve Amos Process Engineering Manager, GRD Minproc, South Africa Graduated from Wits University in 1988 with a BSc Hons degree in Applied Chemistry. As a JCI bursar I was assigned to Randfontein Estates Gold Mine for in service training. After completion of this training I was transferred to the JCI Minerals Processing Research Labs (later Anglo Platinum Research Centre), where I was involved in development work for the Anglo Platinum Precious Metals Refinery, Anglo Platinum Base Metals Refinery and managed the pilot plant campaign for the ACP project (Anglo Platinum Converting Process). After this I was seconded to ATD (Anglo American Technical Division) where I was the Senior Process Engineer on the Konkola High Pressure Leach Project, amongst other studies, and worked for a short period at KCM in Zambia as Metallurgical Projects Manager. I returned to Anglo Platinum Process Technology Division in early 2002 as a Senior Process Engineer and worked projects and optimisation work associated for the Anglo Platinum smelting and converting operations in Rustenburg and Polokwane, including the ACP project, Waterval smelter upgrade, Waterval Slag Cleaning Furnace and Polokwane smelter project. Joined GRD Minproc as a Senior Process Engineer in January 2006 and was promoted to Process Engineering Manager in September of that year. I am currently responsible for all process engineering activities in the South African office and have managed a number of bankable feasibility studies. 400


401-444_Steyl:text

2/16/09

3:09 PM

Page 401

STEYL, J.D.T. Kinetic modelling of chemical processes in acid solution at t ≤ 200°C. (i) thermodynamics and speciation in H2SO4-Metal (II) SO4-H2O system. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Kinetic modelling of chemical processes in acid solution at t ≤ 200°C. (i) thermodynamics and speciation in H2SO4-Metal (II) SO4-H2O system J.D.T. STEYL Anglo Research, South Africa

Although thermodynamic theory has been widely applied to hightemperature hydrometallurgical processes to describe properties such as solubility and acidity, few of those studies have attempted to develop an accurate platform to interpret process kinetics. This study utilized existing theoretical frameworks to model the ternary H2SO4MeSO4-H2O system up to 200ºC (where Me = divalent metal ions of Cu, Zn and Fe) and less than 1 mol/kg sulphate, using MgSO4 as a prototype salt. A qualitative interpretation of the chemistry at quantum level was used as a complementary tool to add more confidence to the ambiguous nature of the thermodynamic data reported in the literature and to fill the gaps where no such data was available at all. The Pitzer ion-interaction approach was used to build a phenomenological model of this system but with the explicit − recognition of four contact ion pairs, i.e. HSO4 , H2SO4°, MgSO4° and Mg(SO4)22−. This required an iterative approach around the ioninteraction framework. The interaction parameters simulated the longrange electrostatic interactions and the formation of outer-sphere complexes, while the explicit inclusion of the above complexes recognized the important covalent interactions. Once parameterization was optimized for the two binary systems (H2SO4-H2O and MeSO4H2O), three mixing parameters were derived for the ternary system. Very limited experimental information has been reported for the mixed system, especially at higher temperature. By selecting a minimum number of adjustable interaction parameters, constraining the system to the available thermodynamic data and incorporating the observed species distributions at lower temperature, predictions at higher temperature were made possible. Consistent and interesting trends have been generated, some of which are presented in this paper. The overall solution chemistry model was ultimately used as a tool to interpret the kinetics of ferrous oxidation in acidic sulphate medium, which is discussed in more detail in the second paper of the current series. KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

401

401-444_Steyl:text

2/16/09

3:09 PM

Page 402

Introduction Despite the fact that sulphuric acid is a major chemical commodity, the effort expended to describe the chemistry of sulphuric acid/base metal sulphate aqueous mixtures does not reflect its industrial importance. This is especially prevalent at higher temperature, under conditions frequently encountered in oxidative pressure leach applications. Although thermodynamic theory has been widely applied to describe high-temperature hydrometallurgical processes, focus has usually been directed, either towards calculating specific thermodynamic properties (e.g. solubility in solution) or to calculate solution chemical properties such as acidity, derived from potentiometric or conductivity measurement (examples in Liu and Papangelakis, 2005a, 2005b, Seneviratne et al., 2003 and, Baghalha and Papangelakis, 2000). This study attempts to establish a simple thermodynamic model, with emphasis on the explicit recognition of the minimum, but most important complex species in sulphuric acid/base metal sulphate solution, temperatures below 200ºC and concentrations less than 1 mol/kg total sulphate. The chemical model and peripheral arguments ultimately aims at developing a platform for interpreting hydrometallurgical process kinetics. Methodology The methodology of modelling thermodynamic properties has traditionally followed an ioninteraction approach, i.e. based on complete electrolyte dissociation. The virial type of equations of Pitzer (Pitzer, 1991), which is based on statistical mechanics, have been most widely applied. However, certain electrolyte systems, such as H2SO4-H2O, require the explicit − recognition of the strong complexes (bisulphate inner-sphere complex, HSO4 in this case) in order to accurately describe its thermodynamic properties (Pitzer et al., 1977). Although such ‘modifications’ introduce more interaction parameters, it is compatible with the fully dissociation framework. At the other extreme, many researchers studying hydrometallurgical systems frequently opt for a speciation-only type approach, for example to predict solubility (Papangelakis et al., 1994) or to support proposed reaction mechanisms (Crundwell, 1987). In cases where such speciation models also confirm independent measurement, for example spectroscopic or pH, it becomes an attractive alternative to the interaction-type approach (e.g. Casas et al., 2005a). Casas et al. (2005b) compared the performance of three different chemical models, i.e. Debye-Hückel (DH) B-dot, Pitzer and Bromley-Zemaitis at high temperature (>200ºC). Although the three models differed in complexity and recognized species (equilibria), they demonstrated almost the same ability to fit the solubility of Mg and Al in ternary systems. This illustrated that ion interaction parameters can make up for differences in the number of explicitly recognized species and vice versa. Measured information on species abundance is severely restricted in the open literature, especially at higher temperature. It is therefore natural to limit the model to explicitly contain only the complex species that are most likely to kinetic processes, the driving force behind this study. These complexes are likely to be entropy driven and would therefore become even more significant at higher temperature. An added advantage is that the interaction framework, based on thermodynamic and speciation information at room temperature, can be used to infer the speciation behaviour at higher temperature where only thermodynamic data are available. Phenomological model In view of the above discussions, the following interaction-type (some, also including speciation) models have been considered for this study (references indicate recently applied examples); Bromley-Zemaitis (Liu and Papangelakis, 2005a), Mixed Solvent Electrolyte 402


401-444_Steyl:text

2/16/09

3:09 PM

Page 403

(MSE) (Liu and Papangelakis, 2005b), Pitzer (Casas et al., 2005b) and Electrolyte NRTL (Haghtalab et al., 2004). Although these models have successfully been applied to numerous chemical systems, each having their own subtleties, the relatively low ionic strengths covered in this study makes the choice somewhat ambiguous. Although the ion-interaction model of Pitzer is more complicated, it has been widely applied to thermodynamic studies over the past 35 years and its theoretical framework was also adopted for this study. Thermodynamic framework The conventional unsymmetrical reference states of infinite dilution for the solute and pure water for the solvent have been chosen for this low to medium ionic strength application. For the same reason, the molality (temperature independent) scale was selected, treating the proton as an unhydrated species. The standard Pitzer ion-interaction model is based on an expression for the excess Gibbs energy of the solution, which consists of an extended DebyeHückel (DH) term and virial expansion terms. The relatively dilute electrolyte concentration also makes ionic strength dependences of the third virial coefficients (Clegg et al., 1994) redundant; these generalized equations can be found in numerous publications (e.g. Pitzer, 1991) and the important functions are reproduced here. The single-ion activity coefficient of a cation (M), anion (X) and neutral ion (N) may respectively be represented as follows: [1]

[2]

[3] where the primed indices refers to summation over all distinguishable pairs, and: [4]

[5] and zi is the valency of ion, i. The water activity (aw), more appropriately expressed as the osmotic coefficient, φ may then be written as: [6]

The ionic strength (I) dependence of the DH terms may be represented as follows: KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

403

401-444_Steyl:text

2/16/09

3:09 PM

Page 404

[7] [8] The universal parameter, b has a value of 1.2 (kg/mol)1/2. In this study, the DH parameter for the osmotic coefficient (A φ ) was directly calculated, following from the original DH derivation (natural logarithm scale): [9] The dielectric properties reported by Bradley and Pitzer (1979) and the density reported by Cooper and Le Fevre (1982) was used for pure water as the solvent. These properties for water and the fundamental constants yielded an Aφ value of 0.3914 (kg/mol)1/2 at 25ºC. The binary (Bca) (and its derivative) and ternary (Cca) virial coefficients are expressed in terms of φ their respective interaction parameters (βca and C ca) and their ionic strength functionalities: [10] [11] [12] [13] The values of parameters α1 and α2 are usually set to 1.4 and 12 (kg/mol)1/2, respectively, for 2–2 electrolytes but 2 and 0 (kg/mol)1/2, respectively, for other types of interactions (Pitzer, 1991). The functionalities may be represented as follows: [14] [15] In mixed electrolyte systems, binary short-range interaction between ions of like charge (Φ) and various ternary interactions (ψ) may be accounted for. In this study, the long-range electrostatic forces (denoted by E) were also included for unsymmetrical mixing. The binary mixing term, Φ, its derivative and the corresponding mixing term for the osmotic coefficient are then expressed by Equation [16], [17] and [18], respectively. The values of the electrostatic terms were calculated using Chebyshev approximations (see Pitzer, 1991). [16] [17] [18] As mention before, the Pitzer framework becomes more complicated when strong ion associations are explicitly treated as a separate species. Thermodynamically, a complex species is considered to be a new entity in solution when the mutual electronic attraction of the individual ions is considerably greater that their thermal energy (Robinson and Stokes, 1959). In aqueous solution, water molecules are chemically bonded to the ion and this form the inner coordination sphere. An inner-sphere complex is formed when water molecules are replaced from the inner coordination sphere of the metal ion by a ligand ion (such as SO42−) 404


401-444_Steyl:text

2/16/09

3:09 PM

Page 405

and then, for example, forming a direct chemical bond with the metal ion. When ligand ions do not replace any water molecules from the inner-sphere, and is attached to the positive ions only by virtue of electrostatic considerations, it forms an outer-sphere complex (Hancock, 1976). Activity coefficients for outer-sphere complexes are strikingly similar up to high concentrations, and this is reflected in very similar values of their association constants. Innersphere complexes on the other hand, always show large differences in stability (Prue, 1966). Inner- and outer-sphere complexes represent two extremes and although most ion complexes would show predominance of one type, both are likely to be present. Aqueous species are therefore present as simple and complex ions and the modelling objectives would dictate which species need to be explicitly expressed. The ionic strength defines the charge in the electrolyte solution and would obviously vary between these frameworks: [19] The ionic strength value corresponding to complete dissociation is referred to as the nominal or formal ionic strength (independent of temperature). Whatever framework is used to describe non-ideality in an electrolyte solution, the condition of electrical neutrality has to be fulfilled: [20] If equilibrium is approached from an association perspective, either a step-wise or cumulative description may be used, i.e., respectively: [21] [22] By definition, the activity of an ion is related to its molality in the following manner (the final term on the right incorporates the standard state, which is conveniently chosen at unit molality, so that fi° = mi° = 1): [23] The thermodynamic association constant (denoted by °) at the thermodynamic transcendent condition of infinite dilution for the step-wise (K°) and cumulative (β°) reaction may then be represented as follows, respectively: [24]

[25] Thermodynamic models are used to set up a framework in which thermodynamic properties can be regressed with experimental data. Colligative properties such as freezing point depression, boiling point elevation, heat capacity, molar volume, the electromotive force (e.m.f.) of cells, and isopiestic measurement (osmotic coefficient) are most often used in correlation methods. Only the mean activity coefficient of a salt can be obtained from these measurements (individual ions cannot be isolated during measurement). For a salt, Mν1Xν2 the mean activity coefficient is defined as follows (Robinson and Stokes, 1959): KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

405

401-444_Steyl:text

2/16/09

3:09 PM

Page 406

[26] The ion number ( ν ) of the salt is equal to the sum of the individual stoichiometric coefficients (ν1, ν2). Regression to solubility data is popular because its availability for both pure and mixed electrolyte systems (e.g. Linke and Seidell, 1958 and 1965). The generalized solubilization reaction for the hydrated salt may be written as follows: [27] The thermodynamic solubility product is derived by taking the activity of the solid phase as unity (pure substance). The right-hand terms were respectively obtained after mathematical substitution of Equations [23] and [26]: [28] The molalities and activity coefficients in Equation [28] obviously refer to the individual ions, whereas the water activity is directly related to the osmotic coefficient as calculated in Equation [6]. In order to maintain internal consistency, it may be necessary to convert the calculated osmotic and mean activity coefficients to their formal (also called stoichiometric or observed) values: [29]

[30]

Finally, the molality scale is preferred scale in solution thermodynamics because it has the advantage that it is independent of temperature and pressure. However, the concentration scale has more practical significance and it may be necessary to convert between the scales. The following relation is easily derived (where C refers to the molarity scale and ρ is the solution density in g/l): [31] The standard thermodynamic properties of an equilibrium reaction are constituted from the sum of the primary components taking part in the reaction. Often, the partial molar properties of a species are known only at the reference temperature, Tr (usually 25ºC) and sometimes, not at all. The equilibrium constant can be calculated if the partial molar enthalpy of formation of the constituent species, or the standard molar enthalpy of the reaction is known as a function of temperature (or at the required temperature). This relationship is known as the Van’t Hoff isochore and is derived from classical thermodynamics (the derivation is beyond the scope of this study—see Atkins, 1986): [32] Often heat capacity data is available and integration of Kirchhoff’s Law (see Atkins, 1986) may be used to calculate the enthalpy at the correct temperature: 406


401-444_Steyl:text

2/16/09

3:09 PM

Page 407

[33] The temperature dependence of the heat capacity cannot be predicted and mathematical correlation is therefore used to represent experimental data. The partial molar heat capacity of a species is often reported by the following mathematical expression (also known as the Kelley equation—original reference in Horvath, 1985): [34] The coefficients a, b, c, and d are temperature independent. The standard entropy of a reaction is related to the standard heat capacity by the following relation (see Atkins, 1986): [35] The definition of the standard Gibbs free energy stipulates: [36] When no functional expression for the heat capacity is available, it is often assumed that the heat capacity is constant, i.e. the enthalpy of reaction varies linearly with temperature (after integration of Equation [33]): [37] The entropy value may then be estimated after integration of Equation [35]: [38] When no heat capacity data is available, it is often assumed to be zero, i.e. the enthalpy of the reaction is constant. Integration of Equation [32] then yields: [39] This equation would give only an estimate over a very short temperature range. Predictive methods have been developed to estimate the average heat capacity between T and Tr, such as the method of Criss and Cobble (see Zemaitis et al., 1986). This method, is however, questionable when applied the non-simple cations (like metal sulphate ions) and metal-oxyanions (like the bisulphate ion) (see Blakey and Papangelakis, 1996). Alternatively, a close approximation of the value of Kº at higher temperature may be obtained using Helgeson extrapolation (see description and original references in Liu and Papangelakis, 2005a). However, this method may give an overestimation at higher temperatures (>100ºC) (Papangelakis, 2004). The density function (Anderson et al., 1991) is remarkably accurate in estimating equilibrium constants of reactions at higher temperatures when the molal heat capacity value is known at the reference temperature. The model is based on the observation of almost linear behaviour of lnK° with lnρ and adopts the form as presented in Equation [40]. [40] The density (ρ) and coefficient of thermal expansion (α) of water are well documented, e.g. study of Anderson et al. (1991). In another approach, the Balanced Like Charge Method (BLCM) utilizes equilibrium reactions with balanced like charges to extrapolate over KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

407

401-444_Steyl:text

2/16/09

3:09 PM

Page 408

temperature (see e.g. Papangelakis, 2004 and Oscarson et al., 1988). Lindsay (1980), referring to these reactions as isocoulombic equilibria, found that because charge is conserved, their heat capacity is usually constant with temperature and relatively small. Substitution of Equations [37] and [38] into Equation [36] and simplifying yields the following relationship and may be used to test the validity of the BLCM approach: [41] The following general equation (after substitution of Equations [33], [35] and [36] into Equation [32] and rearranging) would still yield the superior result, i.e. if the heat capacity is available as a function of temperature: [42] No ΔV° terms were included in any of the above equations since temperatures were limited to 200ºC and, hence, to relatively insignificant pressure effects (at saturation). Chemistry This section briefly considers a quantum level interpretation of the chemistry and tries to add more confidence to the ambiguous nature of the data reported in the literature. A basic understanding of ion solvation is an essential part of this analysis. The concept of the hydration shell has been verified for cations and anions by spectroscopic measurement and is also supported by ab initio calculations (see e.g. Pye and Rudolph, 2001). An ion in water may be considered to be surrounded by concentric shells of water molecules, where the successive shells become more weakly bound to the preceding shells until a bulk water structure is reached (Pye and Rudolph, 1998). Calculation methodology All calculations were carried out using the DMol3 (Delley, 1990) density function theory (DFT) code imbedded in the Materials Studio (v.4.2) quantum chemistry package from Accelrys Inc, i.e. the calculations were conducted within the Kohn-Sham formulism (Kohn and Sham, 1965). In order to increase the chemical accuracy when treating the transition metal complexes, an all electron scalar relativistic basis set was employed. The gradient correction method (GGA = gradient generalized approximation) was employed throughout utilizing two DFT Hamiltonians, i.e. the local spin density (LSD) method utilizing the Vosko et al. (1980) (VWN) parameterization of the Ceperley-Alder Monte Carlo result for a homogeneous gas; and the non-local spin density (NLSD) method, utilizing a combination of the Becke (B) gradient corrected exchange energy (Becke, 1988) and the Pedrew (P) gradient corrected correlation energy (Pedrew and Wang, 1992). Fully self-consistent field (SCF) calculations were conducted using the VWN-BP functional throughout. High quality double numerical atomic basis sets including polarization functions (DNP) were used throughout, i.e. to yield best accuracy, especially since hydrogen bonding was an important aspect of this study. Because these functions were treated numerically (rather than analytically) in DMol3, basis set superposition errors (BSSE) were minimized (Delley, 1990—after Materials Studio help file). The aim of this chemical modelling exercise was more qualitative in nature and no additional effort was therefore expended to quantify the effect of BSSE errors, or to test alternative basis sets or exchange-correlation functionals. 408


401-444_Steyl:text

2/16/09

3:09 PM

Page 409

In addition to the use of explicit water molecules as described above, the Conductor-Like Screening Model (COSMO) (Klamt and Schüürman, 1993) was used to account for long range electrostatic solvent effects, using the dielectric constant of pure water. COSMO is an example of a continuum solvent model where the solvent molecules are not treated explicitly but are expressed as a homogeneous medium characterized by a bulk dielectric constant. The effect of the solvent is modelled by imposing charges on the continuum surface, which leads to a polarization of the wave function within the solute cavity (Ziegler and Autschbach, 2005). Therefore, the electronic structure and geometry of the solute is still described by the DFT method, but the solute is placed inside a cavity which has the same shape of the solute molecule. Outside of the cavity, the solvent is represented by a homogeneous dielectric medium (Andzelm et al., 1995). The approach followed in this study is based on the default DMol3/COSMO recommendations imbedded within Materials Studio. The DMol3/COSMO can predict solvation energies for neutral solutes with an accuracy of about 2 kcal/mol (Andzelm et al., 1995) and has been tested extensively (Klamt and Schüürman, 1993, and Andzelm et al., 1995). Although continuum models are effective when the impact of the solvent is predominantly electrostatic in nature, the lack of an explicit solvent treatment is an oversimplification and neglects specific information on the intermolecular interactions, such as hydrogen-bonding in the first hydration shell. Therefore, in this study, the total coordination of the cation and the composite, polyatomic sulphate anion was restricted by explicitly using water as a coordination filling species in the first hydration shell, while treating the effect of the solvent beyond this level via the COSMO methodology. The most important simpler structures were confirmed to be minima via analytic second derivative calculation. Since neither the vibrational energy, nor the entropy of the solvated species could be obtained using the above methodology, the enthalpy rather than the total free energy formed the basis of comparison with reported literature values. It was assumed that the enthalpy could be represented by the potential energy; hence, it was assumed that the kinetic energy (translational, rotational, etc.) cancels approximately for reactions between solvated ions at constant temperature. First principles perspective The objective of this simulation exercise was to create a thermodynamic framework from a first principles perspective and consistent with reported standard state properties of ions and neutral species in aqueous solution (e.g., NBS tables, Wagman et al., 1982). The modelling was conducted at the thermodynamic condition of infinite dilution in water. Within the context of the conventionally defined zero Gibbs energy of formation of the proton at any temperature, the hydrogen electrode was the appropriate starting point. The absolute electrode potential in electrochemistry is the difference in electronic energy between the Fermi level energy of a metal electrode and a universal reference system (without any additional metalsolution interface) (see Trasatti, 1986). The standard hydrogen electrode (SHE) is a redox half cell and forms the basis of the thermodynamic scale of oxidation-reduction potentials: [43] This reaction can be achieved at a practical level with the use of a platinum electrode because it possesses the ability to catalyse the reduction of the proton to form hydrogen gas at a high exchange current density, and vice versa. The recommended value of the absolute electrode potential (Eabs, often referred to as the vacuum scale potential) of this cell in water is 4.44±0.02 V at 25ºC (Trasatti, 1986). This number is the required link between the electrochemical potential scale and the physical energy scale. Reported literature values vary KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

409

401-444_Steyl:text

2/16/09

3:09 PM

Page 410

considerably for such an important quantity, with an average around 4.6 V (see details in Bockris and Khan, 1993). The absolute value of the solution energy of a single ion is just one of the uncertainties that have to be dealt with. The exact nature of the hydrated proton at infinite dilute is clouded by the fact that the proton is geometrically very small and forms fluctuating hydrogen bonds with the surrounding water molecules. Senanayake and Muir (1988) used e.m.f. measurement (with some extra-thermodynamic assumptions) in chloride media to estimate ionic activities and hydration numbers. The hydration number of 7 obtained for the proton in dilute solution indicated further hydration beyond the first hydration cell. Based upon the acidity function and other physical evidence (Bell, 1959, and Bockris and Reddy, 1977), the existence of this well-defined species is usually assumed to predominate in strongly acid solutions. However, the more recent and insightful paper of Marx et al. (1999) suggests that both the H5O2+ and H9O4+ species are important only in the sense of ideal structures and that numerous unclassifiable situations exist inbetween. Ab initio path integral simulations were used to show that the hydrated proton actually forms a fluxional defect in the highly structured, hydrogen-bonded network of water. This ‘protonic defect can assume many different structures, so that an unambiguous distinction between H5O2+ and H9O4+ can no longer be achieved’ (Marx et al., 1999). In view of these discussions, a thermodynamic basis needed to be established using the molar enthalpy of hydration as reference. The thermochemical (Born-Haber) cycle consisted of atomizing the gas, ionizing the gas, followed by solvation in water medium and finally, reduction on a platinum electrode to again form hydrogen gas (see Equation [44]). Since neither the vibrational energy, nor the entropy of the solvated species could be explicitly calculated within the context of the DMol3/COSMO framework, the enthalpy (rather than the total free energy) formed the basis of comparison.

[44]

Comparison between the calculated and reported enthalpy of hydration (Table I) suggests that either the H3O+ or the H5O2+ species could be used as basis. A similar conclusion can be drawn by comparing ΔH°red, obtained by difference of the reported total ionization enthalpy and the hydration enthalpy. Therefore, for the purpose of this study, the following reaction was used as thermodynamic basis: [45] This approach yields the free enthalpy of hydration, ΔHºhyd of the proton as -1114 kJ/mol and the reduction enthalpy, ΔHºred as -429 kJ/mol. If an entropy change of -131 J/mol.K (NBS Tables) is used for the hydration reaction, a value of 4.65 V for the absolute potential of the hydrogen reaction is obtained. This is in good agreement with the average value reported in Bockris and Khan (1993), i.e. 4.6 V. The next step was to define a thermodynamic basis for the sulphate dianion. The most direct approach was to use the enthalpy of dilution of concentrated sulphuric acid to ‘calibrate’ the system: [46] The ions refer to the formal species in excess water. The structure of a H2SO4 molecule in concentrated sulphuric acid consists of extensive hydrogen-bonded networks. This is evident from its exceptionally high dielectric constant (ε) of 110 (Lide, 1991). Kazansky and Solkan 410


401-444_Steyl:text

2/16/09

3:09 PM

Page 411

Table I Calculated and literature values of the various energy contributions to the Born-Haber cycle for the hydrogen system Energy

Step (Born-Haber Cycle)

ΔHºat+ΔHºion

H(g) → H +

Calculated (kJ/mol)

Literature (kJ/mol)

References

+ (g)

1543

1536

Wagman et al. 1982 (NBS tables)

+

-524 -1081 -1147 -1185 -1216

1103

Marcus, 1987

-1019 -462 -396 -358 -327

a

ΔHºhyd, 25ºC

H (g) → H (aq) + + H (g) + H2O(aq) → H3O (aq) + + H (g) + 2H2O(aq) → H5O2 (aq) + + H (g) + 3H2O(aq) → H7O3 (aq) + + H (g) + 4H2O(aq) → H9O4 (aq)

ΔHºred, 25ºC

H (aq) → H(g) + H3O (aq) → H(g) + H2O(aq) + H5O2 (aq) → H(g) + 2H2O(aq) + H7O3 (aq) → H(g) + 3H2O(aq) + H9O4 (aq) → H(g) + 4H2O(aq)

+

-424

Bockris and Khan, 1993: 4.6 Vred

aCalculated

(Reaction [45]) from average reported absolute potentials and the entropy of hydration, ΔShyd = -131 J/mol.K (Wagman et al., 1982)

Table II Calculated molecular energies and average hydrogen bond lengths of the symmetrical complexes of H2SO4 in 100% sulphuric acid (bulk permittivity of 110) Molecule

Etot (hartrees)

Ediff (hartree) (per sulphur)

Distance (Å) H-Bond(O−O)

Distance (Å) H-Bond(O−H)

[H2SO4] [H2SO4]2 [H2SO4]3

-702.0775021 -1404.1582559 -2106.2385225

− -0.0016259 -0.0003795

− 2.74 (avg.) 2.74 (avg.)

− 1.72, 1.74 1.70, 1.78

(2003) used ab initio quantum chemical calculations, whereas Walrafen et al. (2000) used experimental methods, to show that these hydrogen bonds form very definite ionic clusters. These localized structures may be cationic, anionic or neutral in nature, e.g., in the case of the − dimer, [H3SO4 H2SO4]+, [HSO4 H2SO4] or [2H2SO4], respectively. The modelling of periodic structures, using a number of interacting molecules (e.g., Arrouvel et al., 2005), would obviously yield the best results. In the context of this study, only the most stable electroneutral complexes, as reported by Kazansky and Solkan (2003), were used. The average O−H distance of 1.74 Å is close to the average calculated values of Kazansky and Solkan (2003) of 1.8 Å. The total electronic energies (Table II) suggests that the neutral trimer is a sufficiently stable (23% change per mole sulphur) to be used as basis to define the sulphate dianion structure in water solvent. A description of the structure of the sulphate ion in water is important because it would affect its interaction with cations, such as the proton and divalent metal ions. The formation of a contact ion pair between magnesium and sulphate ion was postulated as a temperature dependent equilibrium with the hydrated sulphate and the hexaaquo magnesium ion (see Pye and Rudolph, 1998). In a later publication (Pye and Rudolph, 2001), water was proposed to coordinate with the sulphate ion in a bidentate fashion to form two hydrogen bonds per pair of KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

411

401-444_Steyl:text

2/16/09

3:09 PM

Page 412

sulphate oxygen atoms. Structural energetic and vibrational characteristics of isomers containing less than six waters in the primary hydration shell were considered. However, comparison between calculated and experimental S-O bond lengths suggested that more than six waters should be present in the primary hydration shell. Cannon et al. (1994) presented evidence, using ab initio calculations, that this tetrahedral dianion strongly interacts with water; roughly three water molecules were coordinated with each of the sulphate oxygen atoms. Therefore, unlike most anions, sulphate can be considered a structure-making ion with respect to the solvent around it (Cannon et al., 1994). The model developed by Cannon et al. orientated the water with one hydrogen pointing towards sulphate oxygen and the other hydrogen pointing towards the neighbouring water oxygen. The number of waters in the first hydration shell of the sulphate ion has been reported to be approximately 6 to 14, based on diffraction data (see Cannon et al., 1994). A more recent study by Vchirawongkwin et al. (2007), the hydrated sulphate ion was characterized using ab initio quantum mechanical (QM) charge field (QMCF)—molecular dynamics (MD) simulation and large angle X-ray scattering (LAXS) methods. The LAXS data showed an average coordination number (i.e. hydrogen bonded to sulphate ion) of up to 12, while the QMCF-MD simulation displayed a range of between 8 and 14. Experimental investigations in several sulphate solutions (see references in Vchirawongkwin et al., 2007) have been conducted and coordination numbers between 6 and 2− 8 were assumed. The study of Wang et al. (2000) indicated that isolated SO4 is unstable in the gas phase (strong Coulomb repulsion between the two excess electrons) and a minimum of three water molecules is necessary to stabilize the dianion. Around 12 water molecules were claimed to be present in the first hydration shell, with four water molecules, each forming two direct hydrogen bonds with the sulphate oxygen atoms. The first principles study of Gao and Liu (2004) considered a large number of isomers in two structural series, using ab initio molecular dynamics simulations. They considered up to 12 water molecules in the primary hydration shell at 100K. However, when the induced conditions approached 200 K, a ‘crowding-out’ of the first hydration shell was observed, with four molecules moving to the second shell and eight remaining the primary hydration shell. As pointed out by Cannon et al. (1994), the above wide range of values for the hydration number in the primary hydration shell exemplifies the difficulties involved in building solvation models for polyatomic ions. The more accurate approach would be use the QM-MD formulation (as suggested by Sterzel and Autschbach, 2006), but since this was outside the scope of this study, the thermodynamic reference framework defined above (for the sulphuric acid system) was expanded to define the sulphate ion and its complexation with cations. The sulphate ion is a polyatomic group with the double charge evenly distributed over the four oxygen atoms in tetrahedral symmetry. The hydrogen bond network in this first shell must adapt to the geometry imposed by the sulphate group (Gao and Liu, 2004). The Td structure of Pye and Rudolph (2001) was used as starting point and geometrically optimized using the calculation methodology described earlier. The starting structure consisted of each H2O molecule acting as a bidentate hydrogen bond donor to two sulphate oxygen atoms. Eight optimizations, each time modifying the starting structure slightly, were conducted and on each occasion the structure relaxed to form only one direct hydrogen bond to sulphate oxygen for each H 2O molecule, while the other hydrogen orientated itself towards the dielectric continuum. However, at least one and often two H2O molecules donated their ‘free’ hydrogen to another H2O oxygen in the same group for hydrogen bonding. This phenomenon suggested that water molecules are likely to form clusters in the primary hydration sphere. In fact, as 412


401-444_Steyl:text

2/16/09

3:09 PM

Page 413

illustrated by Gao and Liu (2004), the geometrical coincidence of the sulphate group with its tetrahedron structure either form cyclic rings (up to 3 molecules) to accommodate more H2O in the primary hydration shell, or alternatively a ‘crowding-out’ effect of the first hydration shell is observed at finite temperature, with some molecules moving to the second shell. With this in mind, the H2O molecules was arranged either in 2 or 3 clusters; for example, with 6 molecules in the primary hydration shell, either 2 clusters of 3 H2O molecules or 3 clusters of 2 H2O molecules could be considered. As expected, the highest combination of 3 cyclic rings always resulted in the lower overall energy for the total structure. In most cases, one of the H2O molecules in the 3 cyclic clusters orientated itself towards the dielectric continuum. This is in line with the observations of Gao and Liu (2004): ‘the hydrogen bond network in the first shell must adapt to the geometry imposed by the requirement to solvate the sulphate group, which introduces strain in the network and even repulsion among the first H2O molecules’. The calculated energies are listed in Table III. It is important to realise that these structures may not be the true minima but that further ‘modifications’ are likely to result only in marginal improvements of the calculated reaction enthalpy for the general association reaction: [47] Besides disagreement between calculated and reported reaction enthalpies, the asymmetrical nature of the lower complexes suggests at least 8 H2O molecules in the inner hydration sphere at infinite dilution. This is a simplified approach and if the reality is in fact more molecules in the primary hydration sphere, the bonding is likely to be less structured than depicted in Figure 1. This may be the case since the calculated value of the absolute standard molar enthalpy of hydration for the sulphate ion (-1213 kJ/mol) is lower than the reported value + + (-1026 kJ/mol, Marcus, 1987, based on H (g)→ H (aq): ΔH° = -1103 kJ/mol). Figure 1 presents a static snapshot of the sulphate ion in water solvent; the more accurate approach would be to consider a dynamic simulation (e.g., Vchirawongkwin et al., 2007), as mentioned above. Nevertheless, the static approach was considered sufficient for the purpose of this study. Typical calculated hydrogen bond lengths (Steyl, 2008) were in reasonable agreement with measured (Vchirawongkwin et al., 2007) and calculated (Gao and Liu, 2004) values from the literature.

Table III 2− Calculated molecular energies of SO4 (H2O)n (n = 6-9) and the corresponding reaction enthalpy (Reaction [47]) Molecule

a

H2O 2− SO4 (2H2O)3 2− SO4 (3H2O)2 2− SO4 (2H2O)2(3H2O) 2− SO4 (2H2O)4 2− SO4 (2H2O)(3H2O)2 2− SO4 (2H2O)3(3H2O) 2− SO4 (3H2O)3

-76.6240464 -1160.989182 -1160.993309 -1237.624055 -1314.254551 -1314.260114 -1390.889044 -1390.894635

aCalculated

Etot (hartrees)

b

ΔHºrx,calc (kJ/mol) − 38.8 49.6 67.2 84.1 98.7 111.6 126.2

c

NBS tables (kJ/mol) − 95.3 95.3 95.3 95.3 95.3 95.3 95.3

(absolute) energies of molecular species; bCalculated (potential) energies of Reaction 47; cWagman et al. (1982)

KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

413

401-444_Steyl:text

2/16/09

3:09 PM

Page 414

2−

Figure 1: Optimized structures for the lowest energy complexes of SO4 (H2O)8

Table IV − Calculated molecular energies of HSO4 (H2O)n (n = 3-5) and H2SO4(H2O)2, reaction enthalpies and equilibrium (protonation) constants Molecule −

HSO4 (H2O)3 (a) − HSO4 (H2O)3 (b) − HSO4 (H2O)4 − HSO4 (H2O)5 (a) − HSO4 (H2O)5 (b) H2SO4(H2O)2

a

½Etot (hartrees) -931.546511 -931.547928 -1008.180816 -1084.809288 -1084.811148 -855.3544882

b

½ΔHºrx,calc (kJ/mol)

51.3 (Rx 52) 55.1 (Rx 52) 78.3 (Rx 52) 89.9 (Rx 52) 94.8 (Rx 52) 7.7 (Rx 53)

c

½ΔHºrx,lit (kJ/mol)

Log (Kº) (Calculated)

Log (Kº) (Literature)

73.35 73.35 73.35 73.35 73.35 −

− − dpK2: 2.2 − − fpK1: -1.3

− − c1.99; e1.96 − − g-4.7 to -2.0

aCalculated

(absolute) energies of molecular species; bCalculated (potential) energies of corresponding reactions; cWagman et al. (1982); dΔSºrx = 111.7 J/mol.K (Wagman et al. 1982); eDickson et al. (1990); fΔSºrx = 0 J/mol.K; gPerrin (1982) −

The bisulphate, HSO4 ion and H2SO4° molecule were the next species to consider. In view of the strong interaction between the acidic hydrogen and the oxygens around the sulphur, the smaller overall charges, and the unsymmetrical geometries, that these species would be less hydrated than the solvated ‘free’ sulphate ion. The bond distance between the proton and the sulphate oxygen is so short (~1Å) that the proton and one water is not enough to form a cyclic ring between two sulphate oxygens. However, the proton and two or three waters may form cyclic rings, as suggested in the study of Arrouvel et al. (2005) for the hydrated H2SO4° molecule. The measured Raman frequencies of Walrafen et al. (2000) in sulphuric acid − + solutions pointed to strong hydrogen bonding, resulting from direct H3O -HSO4 ion pair interaction. They suggested 3 or 4 H2O molecules in association with the bisulphate ion in more dilute acid environment. To be consistent with the way the hydration of the proton was dealt with in this study, a single H2O molecule was bound to each of the O-H groups of the + bisulphate ion and acid molecule, i.e. to form the hydronium ion, H3O . The rest of the H2O molecules were then arranged in either 2 or 3 cyclic rings, similarly to the approach followed for the hydrated sulphate ion. The acid molecule was modelled by adding one H2O molecule to each of the protons. It should be noted that these ‘unconstraint’ hydronium ions resulted in several structures which were very close in energy. The overall association reaction for the first protonation of the sulphate ion may be represented as follows: [48] Comparison between the calculated and literature data (Table IV) suggests that the sulphate − ion sheds half its water upon protonation, i.e. forms HSO4 (H2O)4. Taking into consideration the crude assumptions made thus far, the calculated pK2 of 2.2 is in reasonable agreement 414


401-444_Steyl:text

2/16/09

3:09 PM

Page 415

with values listed in the literature (~2) (the pK refers here to the negative logarithm of the ionization constant). This calculation included the reaction entropy of 111.7 J/mol.K from Wagman et al. (1982). Assuming that the H2SO4(H2O)2 molecule (as modelled above) is an accurate enough description, the following reaction may be written for the protonation of the bisulphate ion: [49] The calculated pK1 of -1.3 (assuming zero reaction entropy) is slightly higher than typical values listed in the literature, but is considered acceptable in view of the spread of the reported values (see Figure 6). The next and final step in the quantum level description of the chemistry of the system was to build a basis for the metal ion and its interaction with the sulphate ion to form contact ion pairs. Since Mg2+ fulfilled the role of surrogate metal for the all the other divalent metal ions in solution, a description of its solvation was the logical starting point. This alkaline earth element has a hexagonal crystal structure. The unit cell, consisting of two atoms, was modelled using space group P63/mmc and the experimental cell dimensions of a = 3.2088 Å and c = 5.2099 Å (ICSD Database, 2008). The optimized energy, using periodic DFT calculation, was calculated as -400.837182 hartrees, while keeping the number of k points within reasonable limits. The optimized cell dimensions were calculated as a = 3.1811 Å and c = 5.1493 Å. The optimized energy of the Mg atom in vacuo was calculated as -200.360445 hartrees, allowing the cohesive energy, Ecoh to be obtained as follows: [50] where n in this case is the number of atoms in the unit cell. The calculated value of -1.58 eV is in good agreement with the literature value (Kittel, 1996) of -1.51 eV. The work required to decompose the metal into a single atom (atomization) is thus 153 kJ/mol, which compares well with the NBS Tables value of 148 kJ/mol. Divalent transition metal cations, like Mg2+ and the other metals encountered in this study (Cu2+, Fe2+ and Zn2+), are strongly hydrated at room temperature in dilute aqueous environment and consist of 6 waters (see Horvath, 1985) in an octahedral arrangement in the primary hydration shell. The studies of Pye and Rudolph (1998) and Rudolph et al. (2003) confirmed the hexaaquo magnesium complex by Raman Spectroscopy (RS) and ab initio calculation. The calculated energy of the Mg2+ ion in the gas phase of -199.533566 hartrees may be compared to the gas phase value of the atom (listed above) to yield the calculated enthalpy of ionization of 2171 kJ/mol. This compares well with the literature value of the ionization enthalpy of 2201 kJ/mol (Wagman et al., 1982). The free enthalpy of hydration may be calculated according to the following reaction and is seen to be in close agreement with the literature (Table V): [51] In order to complete the thermodynamic cycle, the standard reduction potential may be written as follows: [52] Since the difference in the energy of the electron between the gas (vacuum) and the metal (Fermi level) of the hydrogen electrode has been calculated as 4.65 V (previous discussions), the standard reduction potential (Reaction [52]) is easily calculated as -2.57 V. This is reasonably close to the literature value of -2.36 V (Atkins and De Paula, 2006). The conventional molar entropy of hydration of Mg2+(aq) (-138.1 J/mol.K) and the entropy change KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

415

401-444_Steyl:text

2/16/09

3:09 PM

Page 416

+

for the hydration reaction, H2(g) → 2H (aq) (-131 J/mol.K) are very similar (values from Wagman et al., 1982). The fact that the entropy change for the overall reaction is small (~7 J/mol.K), explains why the calculated reduction potential agrees reasonable well with the literature value. As mentioned before, this study also neglects any kinetic energy changes; it is assumed that these energies cancel approximately, at least for reactions between solvated ions 2− and at constant temperature. The reaction between the solvated Mg2+ and SO4 ions to form the first metal contact ion pair, may now be evaluated: [53] If the Mg2+ ion bonds to a sulphate oxygen in a monodentate fashion, the removal of sulphate water may be analogous to the way the bisulphate molecule was formed, i.e. half of 2− the H2O molecules from the SO4 (H2O)8 ion were shed. The primary hydration shell around 2− the Mg2+(H2O)6 ion is stronger and more structured compared to the SO4 (H2O)8 ion. Assuming only one H 2 O molecule will be lost from the Mg 2+ (H 2 O) 6 ion due to this monodentate bond, three cases may be considered for the contact pair: (H2O)5MgSO4(H2O)3, (H2O)5MgSO4(H2O)4 and (H2O)5MgSO4(H2O)5. The calculated energies are summarized in Table V. Comparison between the estimated reaction enthalpy from Akilan et al. (2006a) and the calculated values of the complexes, suggests 8 or 9 waters of hydration for the contact ion pair. The optimized structure of the MgSO4(H2O)9 complex, is illustrated in Figure 2. Table V 2+ Calculated molecular energies of Mg(H2O)6 and monodentate MgSO4º contact ion pairs, reaction enthalpies and equilibrium constants a

Molecule

Etot (hartrees) 2+

b

ΔHºrx,calc (kJ/mol)

ΔHºrx,lit (kJ/mol)



− − e-2.06 g-0.96 e1.05 g2.15 −

− − − f1.21 − f1.21 −

Mg(H2O)6 MgSO4(H2O)8 (a) MgSO4(H2O)8 (b)

-660.024857 -1514.529050 -1514.523722

-1961 (Rx 55) − 44.55 (Rx 57)

c-1949

MgSO4(H2O)9

-1591.1545255

26.74 (Rx 57)

d32

MgSO4(H2O)10

-1667.787060

-10.08 (Rx 57)

d32

− d32

aCalculated

(absolute) energies of molecular species; bCalculated (potential) energies of corresponding reactions; + + (1987) (expt.), based on H (g) → H (aq): ΔHº = -1103 kJ/mol; dAkilan et al. (2006a); eΔSºrx = 110 J/mol.K (Akilan et al., 2006a); fAtkinson and Petrucci (1966); gΔSºrx = 131 J/mol.K (see discussion below) cMarcus

Figure 2. Optimized structure for the lowest energy complex of MgSO4(H2O)9

416


401-444_Steyl:text

2/16/09

3:09 PM

Page 417

Dielectric spectroscopy (DRS) (Akilan et al., 2006a) and ultrasonic relaxation values (Atkinson and Petrucci, 1966) have proven the existence of a stepwise mechanism, commonly referred to as the Eigen mechanism: [54] [55] [56] The above mechanism presents a general description of the association of free hydrated ions to first form a double solvated ion pair, then solvent-shared ion pair, and ultimately, the contact ion pair. The enthalpy and entropy terms for the overall inner-sphere complex reaction, as reported in Table V, was obtained by summing the thermodynamic contributions of each of the mechanistic steps. The alternative value of 131 J/mol.K for the reaction entropy was obtained from the ultrasonic relaxation data reported by Atkinson and Perucci (1966): the enthalpy for each of the mechanistic steps were fixed at the values reported by Akilan et al. (2006a), while the entropy was calculated to give the average thermodynamic equilibrium constants in Atkinson and Perucci (1966). This value of the reaction entropy gives log(Kº) = 2.15 and illustrates the sensitivity of the equilibrium constant to the reaction entropy. Various studies (e.g. Zhang et al., 2002 and Rudolph et al., 2003) speculate on the possibility of a bidentate contact ion pair. The optimized structure (9 waters of hydration) is illustrated in Figure 3, while the calculated energy values are listed in Table VI. By comparing these energies with the literature enthalpies (Akilan et al., 2006a), it becomes clear that the bidentate complex may explain the formation of a second ‘type’ of contact ion pair. Although the reported thermodynamic values of Akilan et al. (2006a) are qualitative in nature, they do provide estimates of the enthalpy and entropy contributions of the various equilibria steps. The DRS study of Akilan et al. (2006) identified the formation of a possible triple ion at higher ionic strengths, assumed to be Mg2SO42+. Although the existence of a triple ion is 2− supported by RS (Rudolph et al., 2003), the presence of the corresponding anion, Mg(SO4)2 cannot be ruled out. This is because the geometry of this anion would be close to linear and therefore, have a zero dipole moment (cannot be detected by DRS—see Akilan et al., 2006a and Buchner et al., 2004). The study of Zhang et al. (2002) even considered 2− larger Mg2+-SO4 clusters. Since modelling of the primary contact ion pair has revealed that 2− the SO4 (H2O)8 ion may lose half its waters to form the Mg(H2O)5SO4(H2O)4 monodentate or the Mg(H2O)4SO4(H2O)5 bidentate complex, no waters are likely to be directly associated with the anionic component if the Mg2SO42+ triple ion is formed, i.e.:

Figure 3. Optimized structure for the lowest energy complex of MgSO4(H2O)9 (bidentate)


417

401-444_Steyl:text

2/16/09

3:09 PM

Page 418

[57] [58] Various structures were considered for the Mg 2 SO 4 2+ (H 2 O) 10 (monodentate) and Mg2SO42+(H2O)8 (bidentate) ion pairs, but the minor electronic energy differences between the structures had almost negligible effects on the calculated equilibrium constants. The values reported in Table VI clearly suggest that the Mg2SO 42+ complex is an unlikely candidate for the second contact ion pair, even when using a larger reaction entropy change of 2− 130 J/mol.K. The more likely triple ion contact ion pair is the Mg(SO4)2 species, as can be seen from the values reported in Table VI. The corresponding optimized symmetrical structure is illustrated in Figure 4 and the overall reaction is represented by Equation [59]. [59] The structure presented in Figure 4 was obtained by adding another partially hydrated 2− SO 4 (H 2 O) 4 ion to the optimized Mg(H 2 O) 5 SO 4 (H 2 O) 4 monodentate structure in a symmetrical fashion. The presence of this species is in line with the potentiometric study of Fedorov et al. (1973), who assumed the formation of a series of anionic sulphate complexes Table VI x Calculated molecular energies of possible second Mg-SO4 contact ion pairs, reaction enthalpies and equilibrium constants a

Molecule

Etot (hartrees)

b

c

ΔHºrx,lit (kJ/mol)



43 43 43 43 43 43 43

-6.21 -0.95 e-10.40 f -5.18 e-23.32 f-18.10 e -6.98 g-3.5

-0.22 -0.22 -0.36

MgSO4(H2O)8 (bidentate) MgSO4(H2O)9 (bidentate) Mg2SO42+(H2O)10 (monodent.)

-1514.511298 -1591.1467755 -1868.033131

77.17 (Rx 57) 47.14 (Rx 57) 68.32 (Rx 61)

Mg2SO42+(H2O)8 (bidentate)

-1791.381005

142.04 (Rx 62)

-2522.275826

48.77 (Rx 63)

2–

Mg(SO4)2 (H2O)12 (monodent.)

c

ΔHºrx,calc (kJ/mol)

d d

-0.36 -0.36

aCalculated

(absolute) energies of molecular species; bCalculated (potential) energies of corresponding reactions; et al. (2006a); dΔSºrx = 140 J/mol.K for the (overall) second contact ion pair (Akilan et al., 2006a). eΔSºrx = 30 J/mol.K (Akilan et al., 2006a) for ‘triple’} ion formation: Reactions [57], [58] and [59]; fΔSºrx = 130 J/mol.K; gΔSºrx = 100 J/mol.K cAkilan

2−

Figure 4. Optimized structure for the lowest energy monodentate complex of Mg(SO4)2 (H2O)12

418


401-444_Steyl:text

2/16/09

3:09 PM

Page 419

for Zn(SO4)n{2-2n} and Cd(SO4)n{2-2n}, up to n=5 at high ionic strengths. It should, however, be mentioned that this claim of a high degree of anionic pairing is controversial and has been questioned on the basis anion exchange data (see Rudolph, 1998). On the other hand, the formation of inner ferric, Fe(SO4)n{3-2n} complexes are known to form (e.g. Magini, 1979) and although it can be partially explained by virtue of the high positive charge of the hydrated ferric ion, it does suggest that this kind of anionic pairing may, although to a lesser extent, be present in the divalent case. This may specifically be true at higher temperature, especially if the reaction entropy is higher than the estimated value of 30 J/mol.K of Akilan et al. (2006a). A reaction entropy of about 100 J/mol.K yields log(Kº) of -3.5. It is concluded that the presence of this species could be significant at higher temperature, especially in the case of a softer transition metal cation (cf. the harder Mg alkali-earth cation), e.g. Cu2+. That the entropy for the formation of the anionic triple ion may be closer to 100 J/mol.K (cf. 30 J/mol.K), stems from a comparison between the change in the solvent cavity volume of the various reactions, as computed within the continuum solvent methodology. The calculated volume increase for the first contact ion pair formation (Reaction [53]) was approximately 20Å3, while the entropy change for this reaction was estimated to be about 130 J/mol.K (Table V). The calculated volume change for the formation of the anionic triple ion 2− (Mg(SO4)2 (H2O)12), via Reaction [59], was very similar, i.e. around 23Å3. The triple anion carries a net negative charge that would result in some structuring effect in the secondary solvation shell and beyond. An estimated entropy change of 100 J/mol.K is justified on the grounds that the ‘disruptive’ effect caused by the accommodation of this large molecule in the surrounding solvent is only partially compensated for by its structuring effects. The above results should be viewed as qualitative in nature due to the static modelling approach followed. It should also have been extended to the divalent transition metals (Cu2+, Fe2+ and Zn2+). However, available thermodynamic data were largely limited to the Mg case (next section), especially for the ternary system, and such a study was not justified within the current context (a static model of the Fe2+/sulphate system is presented in Steyl, 2008). Model regression This section requires some quantitative thermodynamic values, while the chemical modelling (previous section) has generated only qualitative information. Thermodynamic data The first association of the proton (second protonation of acid, H2SO4) may be written as follows: [60] The thermodynamic values of this association constant (K1º) was obtained from the study of − Dickson et al. (1990). They measured (potentiometrically) the dissociation constant of HSO4 in NaCl background solution and fitted the data to a series of equations. The following expression (molality scale) was directly incorporated into the solution chemistry model and gives the thermodynamic association constant up to 250°C: [61] where T (kelvin) and the vector, p = [562.7097,-13273.75,-102.5154,0.2477538,-1.117033E4]. Various researchers have in the past used the thermodynamic values of Marshall and Jones (1966). They determined the equilibrium constant from solubility measurements of CaSO4 in sulphuric acid solution (0–1 mol/kg) from 25–350°C. However, their results have been criticized by various authors (e.g. Shock and Helgeson, 1988 and Dickson et al., 1990) KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

419

401-444_Steyl:text

2/16/09

3:09 PM

Page 420

because they ignored the presence of the neutral aqueous species, CaSO4° and H2SO4°, which may exist in significant amounts, especially at high temperatures. The study of Dickson et al. (1990) also acknowledged the most important results from the literature between 1952 and 1990 and incorporated some of it into their expression (Equation [61]). One of the most important studies during that period was the work of Pitzer et al. (1977), but their equations are valid only to 55°C. Figure 5 compares the performance of the Equation [61] against other thermodynamic data reported in the literature. The broken lines represent the calculated equilibrium constant after integration of the enthalpy and entropy terms (Equation [42]), using reported heat capacity parameters (Equation [34]). The data points were obtained from a random scan of published values in the open literature (Steyl, 2008). The equation of Dickson et al. (1990) performed well up to high temperatures (>200ºC) and was therefore adopted for this study. The HSC (2006) heat capacity parameters, imbedded within the software, were obtained from Helgeson extrapolation (Shock and Helgeson, 1988). The second association of the proton (first protonation of acid, H2SO4) may be represented as follows: [62] Reported thermodynamic values for this association constant (K2º) is very rare because the concentration of the neutral species, H2SO4º, can be accurately determined only at high temperature and concentration. In most studies of the dilute sulphuric acid system, even at high temperature (e.g. Marshall and Jones, 1966, Dickson et al., 1990 and Rudolph, 1996), the formation of the neutral species was ignored. Some of the older Raman studies (e.g. Young and Blatz, 1948 and Young et al., 1959) and the Nuclear Magnetic Resonance (NMR) study of Hood and Reilly (1957) quantitatively took the existence of the neutral ion pair into account, even at low temperatures (<50ºC), albeit at high acid concentrations. The RS study of Rao (1940) suggested (qualitatively) around 95% dissociation of the neutral species in 1 mol/litre acid. More recent work include the RS (structural) investigation of Walrafen et al. (2002), the spectroscopic measurements of Xiang et al. (1996) and the flow calorimetric work of Oscarson et al. (1988), all at high temperatures (>150ºC). Agreement between the studies of Xiang et al. and Oscarson et al. is good, and the published equations of latter were used as the primary source of data for this study. The isocoulombic reaction was obtained by considering the water dissociation reaction in addition to Reaction [62], i.e.: [63] Both the assumption of zero reaction heat capacity and the BLCM model for the isocoulombic reaction fit their respective linearity assumptions very well over the 150–200ºC range (Steyl, 2008). The thermodynamic data reported by Sweeton et al. (1974) was used for

Figure 5. Effect of temperature on the thermodynamic association constant (K1°)

420


401-444_Steyl:text

2/16/09

3:09 PM

Page 421

the water dissociation reaction. The Density and BLCM models predicted very similar values of log(K2º) at 25ºC, i.e. -1.7 and -1.8, respectively. The considerable spread in reported values exemplifies the difficulties in measuring this equilibrium constant in dilute sulphuric acid solution, i.e. the neutral molecule is not present in measurable quantities at room temperature. The thermodynamic values reported by Oscarson et al. (1988) at the reference temperature of 150ºC (see Table VIII) were therefore used in the phenomenological model, along with the Density model to extrapolate to other temperatures. Earlier discussions highlighted the lack of information on the structure and thermodynamic properties of contact ion pair formation of the divalent metal sulphates. Spectroscopic studies of MgSO4 solutions (Akilan et al., 2006a, Buchner et al., 2004, Rudolph et al., 2003), CuSO4 solutions (Akilan et al., 2006b), ZnSO4 solutions (Rudolph et al., 1999a and 1999b), CdSO4 solutions (Rudolph, 1998), FeSO4 solutions (Rudolph et al., 1997) and, NiSO4 and CoSO4 solutions (Chen et al., 2005) have provided a major contribution to the understanding of species complexation in the metal sulphate self-medium. However, the reported thermodynamic values undoubtedly contain uncertainties (see Akilan et al., 2006a). The RS study of Rull et al. (1994) has even gone as far as dismissing the presence of contact ion pairs altogether in solutions, up to 2.9 mol/kg MgSO4 and 80ºC. Figure 7 compares the overall (SIPs+CIPs) equilibrium data with various extrapolated lines. The data (open symbols) at high temperature, used in the models of Baghalha and Papangelakis (1998) and Casas et al. (2005), were calculated using Helgeson extrapolation, while the model of Liu and Papangelakis (2005) used the Density model. The lines in Figure 7 were calculated using the Density model and thermodynamic reference data from various spectroscopic studies. This figure exemplifies

Figure 6. Extrapolation of the thermodynamic association constant (K2°) to 25ºC

Figure 7. Experimental data vs. extrapolated values of the overall thermodynamic constant, K3°(tot)


421

401-444_Steyl:text

2/16/09

3:09 PM

Page 422

the disagreement in the literature, especially when extrapolating to higher temperature. From a phenomenological modelling perspective, Casas et al. (2005) illustrated that explicit recognition of species are not actually required. In fact, numerous studies (e.g. Holmes and Mesmer, 1983, Phutela and Pitzer, 1986 and Reardon and Beckie, 1987) used an ion interaction approach without any explicit species recognition to describe the thermodynamics of metal sulphate systems up to high temperatures. However, as pointed out by Akilan et al. (2006), models that take no note of the actual species present ‘cannot be much more than exercises in numerology, with little physical significance’. Since this study was concerned with building a framework to assist with the interpretation of kinetic processes, it was important to recognize the existence of the contact ion pairs, despite the lack of thermodynamic data. In view of the rationale followed thus far, the total ‘free’ metal ion molality refers to the sum of the actual free hydrated ions and all the solvent separated ion pairs (SSIPs): [64] where y refers to the mole fraction of apparent unassociated metal ions (m) and mf refers to the formal metal salt molality (CIP refers here to the first contact ion pair, MgSO4º and αCIP is the fraction contact ion pairs): [65] and z is the true mole fraction of unassociated ions. This equation may be combined with some of the thermodynamic relationships presented earlier, the yield the following equation of equilibrium constant (Steyl, 2008): [66] The activity coefficient of the neutral contact ion pair is not likely to be 1, especially since the previous section suggested that the ion pair may still be hydrated. Errors in species distributions would ultimately be absorbed in the rate constants of the kinetic processes. The more important issue was an accurate description of the changes in species predominance with solution concentration and temperature changes. The value of log(β 3º) of 1.5 was therefore retained for the MgSO4º contact ion pair (Table VIII). No distinction between different bonding mechanisms, i.e. monodentate or bidentate was made in the phenomenological model. This is because monodendate bonding was assumed to be only present in the case of the alkali earth element, Mg and not in the case of the divalent metal ions of Cu, Fe and Zn (Steyl, 2008). Table VII Reported thermodynamic values from the literature System MgSO4-H2O FeSO4-H2O CuSO4-H2O ZnSO4-H2O

Method RS DRS RS RS RS

a

log(β3º)

Reference

~1.5 ~1.6 b ~l.5 c 1.0 ~1.5

Rudolph et al. (2003) Akilan et al. (2006a) Rudolph et al. (1997) Akilan et al. (2006b) Rudolph et al. (1999a)

a

Average calculated value (Equation [66]), using reported data above 0.1 mol/kg (from the corresponding reference) b and assuming unit activity coefficient for the contact ion pair; Stoichiometric mean activity coefficient from c Reardon and Beckie (1987); Value taken directly from the literature

422


401-444_Steyl:text

2/17/09

10:45 AM

Page 423

Table VIII Thermodynamic values used in the regression of the phenomenological model Reaction + 2H (aq) + SO4 (aq) = HSO4 (aq) + o H (aq) + HSO4(aq) = H2SO4 (aq) 2+ 2o Mg (aq) + SO4(aq) = MgSO4 (aq) 2o 2SO4(aq) + MgSO4(aq) = Mg(SO4)2 (aq)

a

No. K1º K2º β3º K4º

log(Kº) 1.964 -1.05 1.5 -3.5

b

ΔHºrx (kJ/mol) 22.8 19.62 c 10 d 49

b

ΔSºrx (J/mol.K) 114 b 26 c 62 d 97

Reference a

Dickson et al. (1990) Oscarson et al. (1988) This study This study

b

Equation [61] is used to extrapolate over temperature; Value at 150ºC (ΔCpº = 113 J/mol.K), using Equation [40] to c extrapolate over temperature; Average enthalpy from RS measurement ~10 kJ/mol for various divalent metal sulphates d (see Rudolph, 1998, Rudolph et al., 1997, 1999a and 1999b) and ΔSº calculated (no ΔCpº estimate); ΔHº estimated from ab initio calculation, ΔSº calculated and assumed ΔCpº = 0 J/mol.K (Akilan et al., 2006a)

Ab initio calculation suggested the possibility of triple-ion pair formation. This was confirmed for the Mg system (Akilan et al., 2006a, Buchner et al., 2004 and Akilan et al., 2006a) and the Ni and Co systems (Chen et al., 2005), but not in the Cu system (Akilan et al., 2006b). However, calculations in the previous section suggested that this triple ion may be the 2− anionic complex, Mg(SO4)2 , which would not be detectable by DRS. The thermodynamic data for this stepwise reaction, as estimated by crude ab initio modelling (previous section), was therefore adopted for the phenomenological model. The positive enthalpy values of all the above reactions may be ascribed to the energy required to break the coordinated H2O molecules of the constituent species prior to bonding. However, the positive entropy values may be ascribed to the release of H2O molecules (reflecting less structure around the ion pairs), which compensates for the positive enthalpies, especially at higher temperature, i.e. these reactions can be viewed as being entropically driven. No explicit interactions between − − HSO4 and the metal cations were taken into account because the HSO4 anion has a noble − gas electronic structure (similar electronic structure to perchlorate, ClO4 ion) and is not expected to form contact ion pairs (Tremaine et al., 2004 and Rudolph et al., 1997). The following subsections discuss the binary and ternary systems up to 200ºC, i.e. the H2SO4-H2O, MgSO4-H2O and H2SO4-MgSO4-H2O systems, respectively. H2SO4-H2O system Although the thermodynamic properties of this system have been studied extensively, there still exists discrepancy in the actual species molalities in the dilute system. At higher temperatures, very little data are available at all. The MSE model of Liu and Papangelakis (2005b) suggests almost 40% H2SO4º in a 0.5 mol/kg acid solution at 200ºC. Wang et al. + (2006) adjusted this model by explicitly treating of the proton as the hydrated species, H3O , and adjusting the interaction parameters to reflect experimental measurements (e.g. Hood and Reilly, 1957, Young et al., 1959 and Walrafen et al., 2000). However, as pointed out earlier, current experimental techniques have not yet developed to a level where quantitative information about minor species abundance, such as H2SO4º, can be provided in the dilute (<1 mol/kg) range at high temperatures. Due to the relative dilute range of this study, it was not deemed necessary to follow the hydrated cation approach of Wang et al. (2006). With, (i) careful selection of the important equilibrium constants (discussed previously), (ii) selecting a minimum number of adjustable interaction parameters, (iii) constraining the system to measured thermodynamic properties (salt and water activity), and (iv) incorporating the − experimentally observed (at lower temperature) distribution of the primary species (HSO4 and SO42−), predictions at higher temperature should be realistic. KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

423

401-444_Steyl:text

2/17/09

10:45 AM

Page 424

Table IX Optimized interaction parameters for the H2SO4-H2O system (with inclusion of the long-range electrostatic terms)

β(0)HB β(1)HB θBS

p1(kg/mol)

103· p2 (kg/mol)

0.2291 0.3736 0.1151

-0.4641 0.0498 -0.1788

Table X Objective functions and their relative deviations for the H2SO4-H2O system a

Temp (ºC) 25 50 75 100 a

Objective function

1

1

1

/3〈φ〉 + /3〈γ±〉 + /3〈α1´〉 ½〈φ〉 + ½〈γ±〉 ½〈φ〉 + ½〈γ±〉 ½〈φ〉 + ½〈γ±〉

b

a

b

AARD (%) (φ, γ±, α) Temp (ºC) Objective function AARD (%) (φ, γ±, α) 0.31, 0.22, 2.49 0.49, 0.40, n/a 0.12, 0.34, n/a 0.06, 0.39, n/a

The error of property, p, defined as: 〈p〉 = ∑(|pi calc/pi exp – 1|; respectively

125 150 175 200

½〈φ〉 + ½〈γ±〉 ½〈φ〉 + ½〈γ±〉 ½〈φ〉 + ½〈γ±〉 ½〈φ〉 + ½〈γ±〉

0.13, 0.42, n/a 0.09, 0.22, n/a 0.14, 0.05, n/a 0.67, 0.74, n/a

b

Absolute average relative deviation of φ, γ± and α,

The studies of Pitzer et al. (1977), Clegg et al. (1994) and Clegg and Brimblecombe (1995) present excellent literature reviews of the important thermodynamic studies of the H2SO4H 2O system over the years, all of which were limited to below 60ºC. The only high temperature study based on own experimental measurement and where a description of the self-medium was attempted, is the isopiestic study of Holmes and Mesmer (1992). The most important experimental data were incorporated into a regression analysis, using the model described above and minimizing the objective function (the relevant regression information is presented in Table X). The symbol α1´ refers to the degree of dissociation of the bisulphate ion. Various combinations of the interaction parameters (β(0)HB, β(1)HB, CφHB, β(0)HS, β(1)HS, CφHS, θBS, ψHBS) were attempted. The activity coefficient of the neutral complex was assumed to be unity, i.e. no interaction parameters involving the neutral acid molecule was included. – The subscripts H, B and S refer to the proton (H+), bisulphate (HSO4 ) and sulphate ions 2– (SO4 ), respectively. In order to simplify this task, a scan of their relative influences was conducted, both with and without the inclusion of the long-range electrostatic parameters. The best fit with the minimum number of adjustable parameters was achieved with the parameters presented in Table IX. A power series was used to calculate the temperature dependence of the interaction parameters (after Holmes and Mesmer, 1992): [67] with the maximum value of i = 4, Td = T – 298.15K and To = 1K. Due to the lack of thermodynamic data and species abundance, especially at higher temperatures, the interaction parameters were allowed to vary only linearly (i = 2). The lower end of the regression was limited to 0.1 mol/kg H2SO4 at higher temperatures (>50ºC) because isopiestic measurements (Holmes and Mesmer, 1992) are generally not regarded as accurate at the lower molalities. A comparison between the experimental data and the model output are represented in Figures 8 and 9. Figure 10 presents the speciation diagram at 1 mol/kg H2SO4 and suggests that the neutral species (H2SO4º) plays a minor role (less that 5%), even at a temperature of 200ºC. This is in line with the predictions of Wang et al. (2006). 424


401-444_Steyl:text

2/16/09

3:09 PM

Page 425

Figure 8. Comparison between the experimental values of the osmotic coefficient and the model output (this study) for the H2SO4-H2O system at 25ºC intervals

–

Figure 9. Comparison between the experimental values of the fraction HSO4 dissociated (α´1) and the model output (this study) for the H2SO4-H2O system at 25ºC intervals

Figure 10. Sulphate species (fraction of the total sulphate molality) as a function of temperature for the H2SO4-H2O system at 1 mol/kg acid

MeSO4-H2O system Thermodynamic data of the base metal sulphates encountered in this study (CuSO4, ZnSO4, FeSO4 and MgSO4) in water medium is scattered through the literature, with significant variances. The most important sources of measured and reported osmotic and mean activity coefficient data can be found in Robinson and Stokes (1959), Pitzer (1972) and Majima et al. (1988). The more recent paper of Quendouzi et al. (2003) also gives a good summary of reported data and it compares well with their calculated values for various metal salts in water. KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

425

401-444_Steyl:text

2/16/09

3:09 PM

Page 426

Additional data for the CuSO4-H2O system was found in Harned and Owen (1958), Downes and Pitzer (1976) and Majima and Awakura (1988). The more recent measurements of Albright et al. (2000) and Miladinovic et al. (2002) provided useful information on the ZnSO4-H2O system. However, only the study of Oykova and Balarew (1974) revealed information about the FeSO4-H2O system, which exemplifies the difficulties in obtaining accurate experimental data for this system. Additional information for the MgSO4-H2O system was obtained from the measurements of Snipes et al. (1975), Archer and Wood (1985), Phutela and Pitzer (1986) and, especially, Archer and Rard (1998) for the osmotic coefficient, and Pitzer and Mayorga (1974), Rard and Miller (1981) and Holmes and Mesmer (1983) for the mean activity coefficient. Agreement between the above studies is reasonable, except for the measurements of Majima et al. (1988). Their data were not included for the ZnSO4 and MgSO4 systems. Figure 11 presents the average reported values of the mean activity coefficients (at 25ºC). The observed mean activity coefficient is relatively insensitive to the type of divalent metal salt. Although uncertainty exists to the degree of contact ion pair formation, earlier discussions emphasized that it would vary significantly between the different salts. Notwithstanding, Figure 11 suggests that contact ion pair formation is not significant at 25ºC (~10% at 1 mol/kg MgSO4; see Rudolph et al., 2003) and also reflects the similarities in the solvent-separated ion pairs between these salts. This is also evident from the very similar values of the total thermodynamic equilibrium constants of these salts at 25ºC, as reported in various publications (e.g. Nair and Nancollas, 1958 and 1959, Helgeson, 1967, Pitzer, 1972 and Högfeldt, 1982). The overall equilibrium constants are therefore dominated by electrostatic forces (see Pitzer, 1972). Previous discussions and information from the literature (e.g. Rudolph et al., 1997, 1999a, Rudolph, 1998 and Akilan et al., 2006a) suggests that contact ion pair formation increases rapidly with temperature. In an attempt to gauge if an increase in temperature would transpire into significant differences between salts, their reported thermodynamic values were compared. For example, the emf study of Nair and Nancollas (1959), conducted up to 45ºC, yielded very similar enthalpy (20.3 kJ/mol and 16.8 kJ/mol) and entropy (111 J/mol.K and 102 J/mol.K) values for MgSO 4 and ZnSO 4 , respectively. Helgeson extrapolation (Helgeson, 1967) yielded similar equilibrium constant values at 100ºC (consistent with other reported conductivity results, see e.g. Högfeldt, 1982), and log(Kº) = 4.8 and 4.6, respectively, at 200ºC. No reported data could be found for CuSO4 and FeSO4 at higher temperatures. Since there were no data available at high temperatures to suggest otherwise, this study assumed that these salts behave similarly up to 200ºC, which justifies the Mg surrogate approach, albeit artificially (e.g. Pitzer, 1972, suggested that possible triple ion pair formation is expected to be more prominent in the case of Cu2+ as compared to Mg2+). Experimental data for MgSO4 were incorporated into a regression analysis. Numerous combinations, from a minimum of

Figure 11. Average mean activity coefficients of various metal sulphates in water (data points represent averages)

426


401-444_Steyl:text

2/16/09

3:09 PM

Page 427

3 to a maximum of 6 parameters were tested, involving the main interactions between ions (β(0)MS, β(1)MS, β(2)MS, CφMS, β(0)MT, β(1)MT, CφMT, θST, ψMST). The subscripts M, C and T refer to 2– the metal, first contact ion pair (MgSO4º) and the triple ion pair (Mg(SO4)2 ), respectively. Reactions [68] and [69] represent the first and triple contact ion pair reactions, respectively: [68] [69] The above interaction parameters were later extended to include binary interactions involving the neutral contact ion pairs ( λ MC , λ SC , λ CC , λ CT ) and then also to ternary interactions (μMCC, μSCC, μCCC, ζMSC). The best regressions at 25ºC with a minimum number of 4 parameters was obtained with the parameters β (0)MS, β (1)MS, β (2)MS, with explicit 2– recognition of MgSO4º and Mg(SO4)2 , in addition to one of the following CφMS, λSC, or λMC. Although previous studies of MgSO4 at high temperature (e.g. Holmes and Mesmer, 1983, and Phutela and Pitzer, 1986) only utilized the parameters, β(0)MS, β(1)MS, β(2)MS and CφMS, the explicit inclusion of the two contact ion pairs demanded a modified approach. Excellent results at high temperatures were obtained with the inclusion of one of the neutral interaction parameters, λSC or λMC. The inclusion of these parameters may be perceived to capture the electrostatic interactions between the dipole of the first contact ion pair and the charged sulphate anion or metal cation, respectively. The inclusion of the β (2) MS term in the conventional treatment of 2–2 electrolytes (see Pitzer and Mayorga, 1974) is related to the rapid decrease in the ion activity coefficients in the dilute range (~0.03 to 0.1 mol/kg), which in turn, is related to a maximum degree of association found for typical 2-2 electrolytes in this dilute range. With optimized β (2)MS values for MgSO 4 in the range -37.23 (Pitzer and Mayorga, 1974) to -32.7 (Rard and Miller, 1981), its value needed to be adjusted because of explicit contact ion pair formation. In order to prevent the regression from getting stuck in local minima, the initial β(2)MS values were systematically varied before each optimization. Figure 12 illustrates the results of this scan, suggesting an optimum β(2)MS value in the region of -29 (kg/mol) and maintaining the 2–2 electrolyte parameters, α1 and α2, at 1.4 and 12 (kg/mol)1/2, respectively. Although the trough is relatively shallow around the minimum, it is (as expected) different than what the conventional treatment requires, i.e. the explicit recognition of ion pairs lowers the absolute value of β(2)MS. The optimized interaction parameters at 25ºC are presented in the first column of Table XI. In order to determine the temperature dependence of the various interaction parameters, Equation [67] was again utilized. However, as may be recalled from Table VIII, no heat capacity information was available for the formation of two contact ion pairs (MgSO4º and

Figure 12. Residual sum of squares versus the second virial coefficient at 25ºC


427

401-444_Steyl:text

2/16/09

3:09 PM

Page 428

Table XI Optimized interaction parameters for the MgSO4-H2O system

β(0)MS β(1)MS β(2)MS λSC Reaction [68] Reaction [69]

p1 (kg/mol)

102· p2 (kg/mol)

105· p3 (J/mol.K)

ΔCpº (kg/mol)

0.2575 3.1890 -29.2839 -0.0990 − −

0.1532 0.4970 -7.1213 -0.5775 − −

0.5653 -0.6343 -3.1555 1.3769 − −

− − − − 267.5036 −

2–

Mg(SO4)2 ), and hence, no accurate description of their enthalpy and entropy changes with temperature could be made. The heat capacities of these two reactions were therefore incorporated into the regression via Equations [34] and [42]. Experimental data at higher temperatures for the MgSO4-H2O system were limited to the heats of dilution measurements of Snipes et al. (1975) (up to 80ºC), the isopiestic vapour pressure measurements of Holmes and Mesmer (1983) (at 110ºC) and the heat capacity measurements of Phutela and Pitzer (1986) (up to 150ºC). The studies of Archer and Wood (1985) and Archer and Rard (1998) also included other data sources (most of them limited to 150ºC) and their model outputs provided excellent generalizations of all available data up to 150ºC. Although the study of Phutela and Pitzer (1986) extended up to 200ºC, this information was limited to very low molality (~0.1 mol/kg), with questionable accuracy (discrepancies with the Archer and Wood model above 140ºC). It was important for this study to include additional information in the regression at high temperatures (>150ºC). Incorporation of the solubility of the monohydrate (kieserite) was ideally suited for this purpose: [70] A major unknown was the thermodynamic equilibrium constant, Kºsp at higher temperatures and more specifically, the partial molal heat capacity of the crystal phase. This study used a Cpº25ºC value of 145 J/mol.K, while the study of Archer and Rard proposed a value of 126 J/mol.K and the study of Pabalan and Pitzer (1987) suggested 134 J/mol.K (value adopted in the study of Liu and Papangelakis, 2005a). The temperature dependence of Kºsp would obviously also depend on the specific choice of thermodynamic parameters for the other species taking part in Reaction [70]. Baghalha and Papangelakis (1998) and Casas et (2005b) used Helgeson extrapolation to calculate Kº sp at high temperatures, whereas Liu and Papangelakis (2005a) calculated slightly larger values, using the Density model. The data chosen for this study originated from the HSC (2006) database and resulted in the following thermodynamic values for Reaction [70] at 25ºC: log(Kºsp) = 0.0662, ΔCpº = -355.92 J/mol.K and ΔHº = -52.43 kJ/mol. Equation [6] was used to obtain the water activity. This value and the mean activity coefficient were then plugged into Equation [28] during the regression analysis to obtain the calculated value of the solubility at different temperatures. The experimental solubility values for the kieserite phase were obtained from Marshall and Slusher (1965), and Linke and Seidell (1965). Because the exact heat capacity values of the two metal contact ion pair reactions had a significant influence on the optimized interaction parameters, a systematic ‘manual’ scan was first conducted. Figure 13 presents a surface plot of the area close to the global minimum and supports the assumption of Akilan et al. (2006a) of zero heat capacity for triple ion formation reaction (Reaction [69]). The overall regression was subsequently conducted, initializing ΔCpº (Reaction [68]) at 270 J/mol.K. In order to prevent overparameterization, no temperature 428


401-444_Steyl:text

2/16/09

3:09 PM

Page 429

Figure 13. Residual sum of squares versus ΔCpº for the first (CIP1) and triple (CIP2) contact ion pairs

Table XII Objective functions and their relative deviations for the MgSO4-H2O system a

Temp (ºC)

25 45 65 80 100 110 120 a

Objective function

1

1

1

/3〈φ〉 + /3〈γ±〉 + /3〈α〉 1 1 /3〈φ〉 + /3〈γ±〉 + /3〈α〉 ½〈φ〉 + ½〈γ±〉 ½〈φ〉 + ½〈γ±〉 ½〈φ〉 + ½〈γ±〉 ½〈φ〉 + ½〈γ±〉 ½〈φ〉 + ½〈γ±〉

1

b

AARD (%) (φ, γ±, α)

Temp (ºC)

0.85, 1.65, 1.14 1.04, 2 .14, n/a 1.61, 1.55, n/a 1.50, 1.63, n/a 1.14, 0.96, n/a 1.37, 1.02, n/a 1.37, 2.39, n/a

130 140 150 170 180 190 200

a

b

Objective function

AARD (%) (φ, γ±, log Kºsp)

½〈φ〉 + ½〈γ±〉 ½〈φ〉 + ½〈γ±〉 ½〈φ〉 + ½〈γ±〉 102· 〈log(Kºsp)〉 102· 〈log(Kºsp)〉 102· 〈log(Kºsp)〉 102· 〈log(Kºsp)〉

1.34, 3.42, n/a 1.26, 4.32, n/a 0.35, 6.58, n/a n/a, n/a, 0.55 n/a, n/a, 0.16 n/a, n/a, 0.01 n/a, n/a, 0.16

b

The error of property, p, defined as: 〈p〉 = ∑(|pi calc/pi exp – 1|; Absolute average relative deviation of φ, γ± and α (or log Kºsp), respectively

dependence of the reaction heat capacities was deemed necessary, i.e. the enthalpy and entropy expressions reduced to Equations [37] and [38], respectively. A power series (Equation [67]) was once again used to calculate the temperature dependence of the interaction parameters. However, higher order (i = 3) terms needed to be included to represent the solubility of kieserite at the lower temperature end, i.e. 170ºC. All the optimized parameters are listed in Table XI. The objective functions and their relative deviations (at their corresponding temperatures) are summarized in Table XII. Figure 14 illustrates the goodness of the regression by comparing the model prediction with the measured solubility of MgSO4 in the kieserite region. The performance of the phenomenological model from this study is compared to the published data in Figures 15 and 16. Except for the data of Phutela and Pitzer (1986), the model gives an excellent representation of the experimental osmotic coefficient and observed mean activity coefficient of MgSO 4 up to high temperatures. Figure 17 compares the calculated fraction contact ion pairs with the available experimental data (25ºC). The model predictions at higher temperatures are also illustrated, emphasizing the preferential ion association at low molalities (discussed previously). These trends also suggest that the 2– formation of the triple ion (Mg(SO4)2 ) becomes increasingly important at high temperatures and at the expense of the first contact ion pair (MgSO4º). KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

429

401-444_Steyl:text

2/16/09

3:10 PM

Page 430

Figure 14. Comparison between the model output (this study) and the measured solubility of the kieserite crystal phase in the MgSO4-H2O system

Figure 15. Comparison between the experimental values of the osmotic coefficient and the model output (this study) for the MgSO4-H2O system at selected temperatures

Figure 16. Comparison between the experimental values of the mean stoichiometric activity coefficient of MgSO4 and the model output (this study) for the MgSO4-H2O system at selected temperatures (the broken line artificially exceeds the solubility limit)

H2SO4-MeSO4-H2O system Reliable experimental information on this ternary system is rare, which is surprising in view of its industrial importance. The most comprehensive set of experimental data was generated by Majima et al. (1988) at 25ºC, but its accuracy is questionable (elaborated upon later in this section). Valuable information was obtained from the isopiestic measurements of Rard and Clegg (1999) for the H2SO4-MgSO4-H2O system at 25ºC. The isopiestic measurements of 430


401-444_Steyl:text

2/16/09

3:10 PM

Page 431

Figure 17. Model predictions of the fraction (of total Mg) MgSO4º (α3) and Mg(SO4)22– (α4) and the experimental values for the MgSO4-H2O system. The dotted line refers to α3, while the solid line represents the fraction total contact ion pair formation (α3+4) (the broken line artificially exceeds the solubility limit)

Majima et al. (various acid-salt mixtures) and the vapour pressure and emf measurements of Tartar et al. (1941) (H2SO4-ZnSO4-H2O system) is also compared (see the data of Rard and Clegg). References to other data sources for these ternary systems can be found in Rard and Clegg, but were not considered for this study since they were all limited to 25ºC. Only one paper was found that reported thermodynamic data at higher temperature, i.e. the study of Jaskula and Hotlos′ (1992) for the H2SO4-CuSO4-H2O system at 60ºC. It was unfortunately of little use to this study because no comparative values were reported for the same system at 25ºC. Various researchers have applied the Pitzer ion interaction model to these types of ternary systems, e.g. Guerra and Bestetti (2006) (H2SO4-ZnSO4-H2O system, ≤ 45ºC), Rard and Clegg (1999) (H 2SO 4-MgSO 4-H 2O system, 25ºC) and Reardon and Beckie (1987) (H2SO4-FeSO4-H2O system, ≤ 90ºC). Due to the relatively dilute nature of the solutions and the explicit recognition of selected contact ion pairs, this study had to derive its own unique mixing parameters. The poor agreement between the experimental data of Majima et al. and the other studies mentioned above prevented an accurate derivation of the mixing parameters. In order to compare the results of the various studies, the relative mean activity coefficient, γ′± was defined as follows: [71] This equation allowed the mean activity coefficient in the ternary system to be normalized to its value in the pure binary system and thus provided a more suitable basis of comparison. Figure 18 compares the relative mean activity coefficient of sulphuric acid in ternary mixtures and suggests reasonably close agreement, even for different divalent metal salt systems. However, Figure 19 illustrates a considerable spread of the relative mean activity coefficient of metal sulphate between the different salts in ternary mixtures with sulphuric acid and water. The relative mean activity coefficient of MgSO4 from Rard and Clegg (1999) is also significantly different from the values reported by Majima et al. (1988). Only the data of Rard and Clegg provided consistency between the mean activity coefficient of MgSO4 and the osmotic coefficient (Figure 20), and was therefore the only source of data used in the regression analysis. Due to the lack of data, only three mixing parameters were selected. Because of the relative obscurity of the contact ion pairs, especially in acidic media (see Rudolph et al., 1997), their mixing parameters were not considered. Only combinations of the following mixing parameters were therefore considered: β(0)MB, β(1)MB, CφMB, θHM, ψHMB, ψ HMS and ψ MBS. The subscripts H, M, B and S refer to the proton (H+), metal (Mg 2+), KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

431

401-444_Steyl:text

2/16/09

3:10 PM

Page 432

Figure 18. Relative mean activity coefficient of sulphuric acid in ternary mixtures from different studies (the surface represents the average fit through all the data points)

Figure 19. Relative mean activity coefficient of the metal sulphate in ternary systems from different studies (the surface represents the best fit through the data points of Rard and Clegg, 1999, for the H2SO4-MgSO4-H2O system)

Figure 20. The osmotic coefficient of the H2SO4-MgSO4-H2O system. The surface represents the best fit through the data points of Rard and Clegg (1999), while the solid lines represent the model output from this study for the two binary systems

432


401-444_Steyl:text

2/16/09

3:10 PM

Page 433

–

bisulphate (HSO4 ) and sulphate (SO42–), respectively. Various combinations were tested, ultimately suggesting that at least two of the binary interaction terms β(0)MB, β(1)MB or θHM needed to be included. The study of Rudolph et al. (1997) also suggested that mixing of Me2+ − and HSO4 ions, although not forming contact ion pairs, should be approached from an ioninteraction perspective. The limited experimental data available for the mixed system made the choice of the third (ternary) interaction term somewhat arbitrary. It was therefore decided to use the same ternary interaction term as used in the work of Reardon and Beckie (1987), i.e. ψ HMB, but with the inclusion of the long-range electrostatic terms, Eθ HM and Eθ′ HM. Figure 21 illustrates the goodness of the regression by comparing the model prediction with the measured data of Rard and Clegg and its comparative offset with the data of Majima et al. (1988). The optimized parameters (at 25ºC) are summarized in Table XIII, corresponding to an α1 value of 2 (kg/mol)1/2. The objective functions and their relative deviations are summarized in Table XIV. The only thermodynamic data found for the H2SO4-MgSO4-H2O system at higher temperatures was the solubility data of kieserite at 200ºC from Marshall and Slusher (1965). This lack of data could only justify a linear dependency on temperature (i = 2), using Equation [67]. Figure 22 illustrates the excellent agreement between the measured solubility data in the ternary system and the model output, even though some areas (dotted lines) lie outside the range used in the regression at 25ºC. Summary and conclusions The theoretical framework provided by the Pitzer ion-interaction approach was used to model the ternary H2SO4-MeSO4-H2O system, using MgSO4 as the surrogate salt to represent other

Figure 21. Comparison between the experimental values of the mean stoichiometric activity coefficient of H2SO4 in ternary mixtures with MgSO4 and H2O, and the model output from this study

Table XIII Optimized interaction parameters for the H2SO4-MgSO4-H2O ternary system (with inclusion of the long-range electrostatic interaction terms between H+ and Mg2+)

β(0)MS β(1)MS ψHMB

p1 (kg/mol)

102· p2 (kg/mol)

0.2606 2.0949 0.2930

0.0121 1.3823 -0.1662


433

401-444_Steyl:text

2/16/09

3:10 PM

Page 434

Table XIV Objective functions and their relative deviations for the H2SO4-MgSO4-H2O ternary system Temp (ºC) 25 200 a

a

b

AARD (%) (φ, γ±, log Kºsp)

Objective function

½〈φ〉 + ½〈γ±, MgSO4〉 〈log(Kºsp)〉

3.09, 8.03, n/a n/a, n/a, 0.03 b

The error of property, p, defined as: 〈p〉 = ∑(|pi calc/pi exp – 1|; Absolute average relative deviation of φ, γ± and log Kºsp respectively

Figure 22. Comparison between the experimental values of kieserite solubility in the ternary system, H2SO4-MgSO4H2O at 200ºC and the model output from this study (the dotted line refers to the model prediction outside the regression range)

divalent metal sulphate electrolytes. Since this study aimed at developing a platform for interpreting hydrometallurgical process kinetics in acidic metal sulphate solutions, the most important inner-sphere complexes were explicitly recognized. Thermodynamic data was used to constrain the phenomenological model, thereby inferring the speciation behaviour at higher temperatures were no qualitative experimental information was available. Due to the relatively dilute nature of the electrolyte modelled in this study, all thermodynamic expressions referred back to the transcendent position of infinite dilution. A qualitative interpretation of the quantum level chemistry was used as a complementary tool to add more confidence to the ambiguous nature of the thermodynamic data reported in the literature. Molecular geometries were optimized within a DMol3/COSMO framework. The total coordination of aqueous species were restricted by explicitly using water as a coordination filling species in the first hydration shell, while treating the effect of the solvent beyond this level via the continuum solvent methodology. Since neither the vibrational energy, nor the entropy of the solvated species could be explicitly calculated within the context of the DMol3/COSMO framework, the enthalpy rather than the total free energy formed the basis of comparison with reported literature values. It was assumed that the enthalpy was represented by the potential energy; hence, it was assumed that the kinetic energy (translational, rotational, etc.) cancels approximately for reactions between solvated ions at constant temperature. The thermodynamic basis for the proton yielded an average hydration of between H3O+ and H5O2+, while a structured and static approach to the solvation of sulphate suggests the dianion to have 8 H2O molecules in the primary hydration shell. It was estimated to lose half its H2O molecules when reacting with the proton to form the bisulphate ion. With zero reaction entropy, the pK1 was estimated to be around -1.3, assuming 434


401-444_Steyl:text

2/16/09

3:10 PM

Page 435

the hydrated sulphuric acid molecule consists of two H2O molecules in its primary shell. The first contact ion pair of MgSO4° was likely a monodendate complex, but the presence of a second contact ion pair could also be explained by the MgSO4° bidentate complex. The presence of a monodentate Mg2SO42+ triple ion (reported in the literature) could not be 2− verified by calculation. Instead, its anionic analogue, Mg(SO4)2 seemed more probable, especially at higher temperatures (significant entropy contribution). The log(Kº) was calculated as -3.5, using an estimated reaction entropy of 100 J/mol.K. The equation of Dickson et al. (1990) gave a good description of the experimental second ionization constant of acid up to high temperatures, and was adopted for this study. The first ionization constant, estimated from a quantum level treatment, compared very well with the thermodynamic data of Oscarson et al. (1988) (the Density model was used to extrapolate the reported high-temperature data to 25ºC). The association constant for the first contact ion pair for different metals was estimated from spectroscopic data (reported in the literature) to be about 1.5, despite having different bonding mechanisms. The Density model was used to extrapolate to higher temperatures, using an average reported reaction enthalpy of 10 kJ/mol. Since no heat capacity value was available, it was considered a variable in the regression analysis. The reaction enthalpy of 49 kJ/mol for the formation of the triple ion was adopted from ab initio calculations, with a linear dependence of log(Kº) on the inverse temperature, i.e. zero reaction heat capacity. − The explicit recognition of the four contact ion pairs, i.e. HSO4 , H2SO4º, MgSO4º and 2− Mg(SO4)2 required an iterative approach around the ion-interaction framework. Using the thermodynamic values and extrapolation techniques described above, selecting a minimum number of adjustable interaction parameters, constraining the system to the available thermodynamic data and incorporating the experimentally observed lower temperature species distributions in the regression routine, speciation predictions at higher temperature were made possible. The binary H2SO4-H2O and MgSO4-H2O systems were first regressed, using the interaction parameters {β(0)HB, β(1)HB, θBS} and {β(0)MS, β(1)MS, β(2)MS, λSC}, respectively. The neutral acid species, H2SO4°, was found to play an insignificant role, even at 1 mol/kg H2SO4 and 200ºC, which is in line with recent predictions from the literature. Although the pure MgSO4-H2O system introduces an oversimplified approach to complexities of a mixed system of various divalent metal sulphates, their thermodynamic values agree to such an extent that the Mg-surrogate approach could be justified. These values and also very similar overall equilibrium constant values, reflect the fact that these salts consist predominantly of outersphere complexes at room temperature, i.e. they are dominated by electrostatic forces. The interaction parameters may be regarded as capturing the long-range electrostatic effects and the formation of outer-sphere complexes, while the explicit inclusion of equilibrium constants recognizes the important covalent interactions. The formation of the first contact ion pair, MgSO4º, was found to become more important at higher temperatures (using a regressed and constant reaction heat capacity of 268 J/mol.K), while the second contact ion pair, Mg(SO4)22−, started to become more prevalent at even higher temperatures (>150ºC) and at the expense of the first contact ion pair. The disadvantage of the Mg-prototype approach is that softer transition metal cations, such as Cu2+, may be expected to form more prominent first and second contact ion pairs with the sulphate dianion. Since this study was concerned with trends, rather than exact values, these discrepancies may be absorbed in the kinetic rate constants when modelling process reactions. Reliable experimental data was found to be rare for the mixed H2SO4-MgSO4-H2O ternary system, especially at higher temperatures. Once parameterization have been optimized for the binary systems, three adjustable mixing parameters were required for the ternary system (ignoring the role of the two higher-order contact ion pairs), i.e. β(0)MB, β(1)MB and ψHMB. KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

435

401-444_Steyl:text

2/16/09

3:10 PM

Page 436

These parameters were varied linearly with temperature because only one set of data was available in the dilute range (≤ 1 mol/kg total sulphate) at higher temperature, i.e. the solubility of kieserite at 200ºC. The model was found to behave in a consistent manner, even beyond the range used in the regression. In conclusion, the following two figures illustrate typical trends that may be expected. Figure 23 presents the variation in the unassociated proton fraction (0.2 mol/kg H2SO4) with changing metal sulphate molality. The fraction unassociated H+ is clearly more sensitive to the total metal sulphate molality at higher H2SO4:MeSO4 ratios. Put differently, the fraction unassociated H+ is more sensitive to the total acid at lower total MeSO4 molality. Figure 24 represents the total fraction metal sulphate contact ion pairs (MeSO4º+Me(SO4)22−) with varying acid molality and temperature. Contact ion pair formation is clearly favoured at higher temperature and lower acid conditions. At low metal sulphate molalities, the temperature and acid concentration has a relatively smaller effect on the total fraction contact ion pairs present. The above model provides a powerful tool that may be used to interpret the kinetics of reactions occurring in hydrometallurgical processes.

Figure 23. Predicted trends of the fraction unassociated H+ (0.2 mol/kg H2SO4) with varying metal sulphate molality in the ternary system, H2SO4-MgSO4-H2O, at various temperatures: (a) Surface plot; (b) Contour plots (the dotted line represents the 150ºC contour at 0.1 mol/kg H2SO4)

Figure 24. Predicted trends of the total fraction metal sulphate contact ion pairs with varying acid molality in the ternary system, H2SO4-MgSO4-H2O, at various temperatures: (a) Surface plot at 0.5 mol/kg MgSO4; (b) Contour plots at varying MgSO4

436


401-444_Steyl:text

2/16/09

3:10 PM

Page 437

Acknowledgements The author is grateful to Dr. Maggie Burger for helpful discussions on the quantum level chemistry aspects of this study, whilst Prof. Vladimiros Papangelakis is thanked for his objective commentary regarding the phenomenological model. Anglo American plc is acknowledged for financially supporting this work. The author would also like to thank Luke Ngubane for sourcing the numerous references so efficiently. References AHRLAND, S. How to Distinguish between Inner and Outer Sphere Complexes in Aqueous Solution. Thermodynamic and Other Criteria. Coordination Chemistry Reviews, vol. 8, 1972. pp. 21–29. AKILAN, C., ROHMAN, N., HEFTER, G., and BUCHNE, R. Temperature Effects on Ion Association and Hydration in MgSO4 by Dielectric Spectroscopy. European Journal of Chemical Physics and Physical Chemistry, vol. 7, 2006a. pp. 2319–2330. AKILAN, C., HEFTER, G., ROHMAN, N., and BUCHNER, R. Ion Association and Hydration in Aqueous Solutions of Copper(II) Sulphate from to 65ºC by Dielectric Spectroscopy. Journal of Physical Chemistry B, vol. 110, 2006b. pp. 14961–14970. ALBRIGHT, J.G., RARD., J.A., SERNA, S., SUMMERS, E.E., and YANG, M.C. Isopiestic Determination of the Osmotic and Activity Coefficients of ZnSO4(aq) at T = 298.15K, and the Standard Potential of the Electrochemical Cell ZnHgx(two phase)⎜ZnSO4(aq)⎜PbSO4(s)⎜PbHgx(two phase). Journal of Chemical Thermodynamics, vol. 32, 2000. pp. 1447–1487. ANDERSON, G.M., CASTET, S., SCHOTT, J., and MESMER, R.E. The Density Model for Estimation of Thermodynamic Parameters of Reactions at High Temperatures and Pressures. Geochimica et Cosmochimica Acta, vol. 55, pp. 1769–1779. ANDZELM, J., KOLMEL, CH., and KLAMT, A. Incorporation of Solvent Effects into the Density Functional Calculations of Molecular Energies and Geometries. Journal of Chemical Physics, vol. 103, 1995. pp. 9312–9320. ARCHER, D.G. and RARD, J.A. Isopiestic Investigation of the Osmotic and Activity Coefficients of Aqueous MgSO4 and the Solubility of MgSO4.7H2O(cr) at 298.15 K: Thermodynamic Properties of the MgSO4 + H2O System to 440 K. Journal of Chemical and Engineering Data, vol. 43, 1998. pp. 791–806. ARCHER, D.G. and WOOD, R.H. Chemical Equilibrium Model Applied to Aqueous Magnesium Sulphate Solutions. Journal of Solution Chemistry, vol. 14, no. 11, 1985. pp. 757–780. ARROUVEL, C., VIOSSAT, V., and MINOT, C. Theoretical Study of Hydrated Sulphuric Acid: Clusters and Periodic Modeling. Journal of Molecular Structure: Theochem. vol. 718, 2005. pp. 71–76. ATKINS, P.W. Physical Chemistry, 3d Edition, Oxford University Press, Oxford. 1986. ATKINS, P. and DE PAULA, J. Atkins’ Physical Chemistry. 8th Edition, Oxford University Press, Oxford. 2006. ATKINSON, G. and PETRUCCI, S. Ion Association of Magnesium Sulphate in Water at 25ºC. Journal of Physical Chemistry. vol. 70, no. 10, 1966. pp. 3122–3128. BAGHALHA, M. and PAPANGELAKIS, V.G. Pressure Acid Leaching of Laterites at 250ºC: A Solution Chemical model and Its Applications. Metallurgical and Materials Transactions B, vol. 29B, 1998. pp. 945–952. BAGHALHA, M. and PAPANGELAKIS, V.G. High-Temperature Conductivity Measurements for Industrial Applications. 2. H2SO4−Al2(SO4)3 Solutions. Industrial and Engineering Chemistry Research, vol. 39, 2000. pp. 3646–3652. BALE, C.W., PELTON, A.D., and THOMPSON, W.T. Facility for the Analysis of Chemical Thermodynamics (FACT), Thermofact Ltd./Ltée & McGill University, Montreal. 1991. BECKE, A.D. 1A Multicenter Numerical Integration Scheme for Polyatomic Molecules. Journal of Chemical Physics, vol. 88, 1988. pp. 2547–2553. BELL, R.P. The Proton in Chemistry. Methuen, London. 1959. KINETIC MODELLING OF CHEMICAL PROCESSES IN ACID SOLUTION

437

401-444_Steyl:text

2/16/09

3:10 PM

Page 438

BLAKEY, B.C. and PAPANGELAKIS, V.G. A Study of Solid-Aqueous Equilibria by the Speciation Approach in the Hydronium Alunite – Sulphuric Acid -Water System at High Temperatures. Metallurgical and Materials Transactions B, vol. 27B, 1996. pp. 555–566. BOCKRIS, J.O’M. and REDDY, A.K.N. Modern Electrochemistry, vol. 1, Plenum Press, New York. 1977. BOCKRIS, J.O’M and KHAN, S.U.M. Surface Electrochemistry: A Molecular Level Approach. Plenum Press, New York, Chapter 5.3, 1993. pp. 490–495. BRADLEY, D.J. and PITZER, K.S. Thermodynamics of Electrolytes. 12. Dielectric Properties of Water and Debye-Hückel Parameters to 350ºC and 1 kbar. Journal of vol. 83, no. 12, 1979. pp. 1599–1603. BUCHNER, R. CHEN, T., and HEFTER, G. Complexity in Simple Electrolyte Solutions: Ion Pairing in MgSO4(aq). Journal of Physical Chemistry B, vol. 108, 2004. pp. 236–2375. CANNON, W.R., PETTITT, B.M., and MCCAMMON, J.A. Sulphate Anion in Water: Model Structural, Thermodynamic, and Dynamic Properties. Journal of Physical Chemistry, vol. 98, 1994. pp. 6225–6230. CASAS, J.M., CRISOSTOMO, G., and CIFUENTES, L. Speciation of the Fe(II) – Fe(III) – H2SO4 – H2O System at 25 and 50ºC. Hydrometallurgy, vol. 80, 2005a. pp. 254–264. CASAS, J.M., PAPANGELAKIS, V.G., and LIU, HAIXIA. Performance of Three Chemical Models on the High-Temperature Aqueous Al2(SO4)3 – MgSO4 – H2SO4 – H2O System. Ind. Eng. Chem. Res., vol. 44, 2005b. pp. 2931–2941. CHEN, H. and IRISH, D.E. A Raman Spectral Study of Bisulphate-Sulphate Systems. III. Salt Effects. Journal of Physical Chemistry, vol. 75, no. 17, 1971. pp. 2681–2684. CHEN, T., HEFTER, G., and BUCHNER, R. Ion Association and Hydration in Aqueous Solutions of Nickel(II) and Cobalt (II) Sulphate. Journal of Solution Chemistry, vol. 34, no. 9, 2005. pp. 1045–1065. CLEGG, S.L., RARD. J.A., and PITZER, K.S. Thermodynamic Properties of 0-6 mol kg−1 Aqueous Sulphuric Acid from 273.15 to 328.15K. Journal of Chemical Society Faraday Transactions, vol. 90, 1994. pp. 1875–1894. CLEGG, S.L. and BRIMBLECOMBE, P. Application of a Multicomponent Thermodynamic Model to Activities and Thermal Properties of 0-40 mol kg-1 Aqueous Sulphuric Acid from <200 to 328 K. Journal of Chemical Engineering Data, vol. 40, 1995. pp. 43–64. COOPER, J.R. and LE FEVRE, E.J. Thermophysical Properties of Water Substance, Edward Arnold (Publishers) Ltd., 1982. pp.1–12. CRUNDWELL, F.K. Kinetics and Mechanism of the Oxidative Dissolution of Zinc Sulphide Concentrate in Ferric Sulphate Solutions. Hydrometallurgy, vol. 19, 1987. pp. 227–242. DAVIES, C.W., JONES, H.W., and MONK, C.B. E.M.F. Studies of Electrolyte Dissociation. Part I. Sulphuric Acid in Water. Transactions of Faraday Society, 1952. vol. 48, pp. 921–928. DELLEY, B. An All-Electron Numerical Method for Solving the Local Density Functional for Polyatomic Molecules. Journal of Chemical Physics, 1990. vol. 92, pp. 508–517. DICKSON, A.G., WESOLOWSKI, D.J., PALMER, D.A., and MESMER, R.E. Dissociation constant of bisulphate ion in aqueous sodium chloride solutions to 250°C, Journal of Physical Chemistry, vol. 94, 1990. pp. 7978–7985. DOWNES, C.J. and PITZER, K.S. Thermodynamics of Electrolytes. Binary Mixtures Formed from Aqueous NaCl, Na2SO4, CuCl2 and CuSO4 at 25ºC. Journal of Solution Chemistry, vol. 5, no. 6, 1976. pp. 389–398. EIGEN, M. and TAMM, K. Schallabsorption in Elektrolytlösungen als Folge Chemischer Relaxation. II Meßergebuisse und Relaxationsmechanismen für 2-2 –weertige Elektrolyte. Zeitschrift für Elektrochemie, vol. 66, no. 2, 1962. pp. 107–121. FEDOROV, V.A., CHERNIKOVA, G.E., KUZNECHIKHINA, M.A., and MIRONOV, V.E. Effect of the Ionic Medium and Temperature on the Formation of Sulphato-Complexes of Zinc and Cadmium and Calculations of their Thermodynamic Stability Constants. Russian Journal of Inorganic Chemistry, vol. 18, no. 3, 1973. pp. 337–340. FILIPPOU, D., PAPANGELAKIS, V.G., and DEMOPOULOS, G.P. Hydrogen Ion Activities and Species Distribution in Mixed Metal Sulphate Aqueous Systems. American Institute of Chemical Engineering Journal, vol. 41, no. 1, 1995. pp. 171–184. 438


401-444_Steyl:text

2/16/09

3:10 PM

Page 439

GAO, B. and LIU, Z.-F. A First Principles Study on the Solvation and Structure of SO42-(H2O)n, n=6-12. Journal of Chemical Physics, vol 121, no. 17, 2004. pp. 8299–8306. GUENDOUZI, M.EL., MOUNIR, A., and DINANE, A. Water Activity, Osmotic and Activity Coefficients of Aqueous Solutions of Li2SO4, Na2SO4, K2SO4, (NH4)2SO4, MgSO4, MnSO4, NiSO4, CuSO4, and ZnSO4 at T = 298.15 K. Journal of Chemical Thermodynamics, vol. 35, 2003 pp. 209–220. GUERRA, E. and BESTETTI, M. Physicochemical Properties of ZnSO4-H2SO4-H2O Electrolytes of Relevance to Zinc Electrowinning. Journal of Chemical and Engineering Data, vol. 51, 2006. pp.1491–1497. HAGHTALAB, A., PAPANGELAKIS, V.G., and ZHU, X. 2004. The Electrolyte NRTL Model and Speciation Approach as Applied to Multicomponent Aqueous Solutions of H2SO4, Fe2(SO4)3, MgSO4 and Al2(SO4)3 at 230–270ºC. Fluid Phase Equilibria, vol. 220, pp. 199–209. HAMER, W.J. The Electromotive Force Method of Determining ionization Constants: Ionization of the Bisulphate Ion. In The Structure of Electrolyte Solutions, W.J. Hamer (ed.), John Wiley & Sons Inc., New York, Ch.15. 1959. HANCOCK, R. Metal Ions in Solution. Hydrometallurgy: Theory and Practice, Vacation School, The South African Institute of Mining and Metallurgy, vol. 1, 1976. pp. 1–27. HARNED, H.S. and OWEN, B.B. The Physical Chemistry of Electrolyte Solutions, 3d Edition, Reinhold Publishing Corporation Inc., New York. 1958. HEFTER, G. When Spectroscopy Fails: The Measurement of Ion Pairing. Pure and Applied Chemistry, vol. 78, 2006. pp. 1571–1586. HELGESON, H.C. Thermodynamics of Complex Dissociation in Aqueous Solution at Elevated Temperatures. Journal of Physical Chemistry, vol. 71, no. 10, 1967. pp. 3121–3136. HOGFELDT, E. Stability Constants of Metal-Ion Complexes. Part A: Inorganic Ligands, IUPAC Chemical Data Series, no. 21, Pergamon Press, Oxford. 1982. HOLMES, H.F. and MESMER, R.E. Isopiestic Studies of Aqueous Solutions at Elevated Temperatures. VII. MgSO4 and NiSO4. Journal of Chemical Thermodynamics, vol. 15, 1983. pp. 709–719. HOLMES, H.F. and MESMER, R.E. Isopiestic Studies of H2SO4(aq) at Elevated Temperatures. Thermodynamic Properties, Journal of Chemical Thermodynamics, vol. 24, 1992. pp. 317–328. HOOD, G.C. and REILLY, C.A. Ionization of Strong Electrolytes. V. Proton Magnetic Resonance in Sulphuric Acid. Journal of Chemical Physics, vol. 27, no. 5, 1957. pp. 1126–1128. HORVATH, A.L. Aqueous Electrolyte Solutions: Physical Properties, Estimation and Correlation Methods, First Edition, Ellis Horwood Limited, West Sussex, England. 1985. HOVEY, J.K., PITZER, K.S., and RARD, J.A. Thermodynamics of Na2SO4(aq) at Temperatures from 273K to 373K and of {(1-y)H2SO4 + yNa2SO4}(aq) at T = 298.15K, Journal of Chemical Thermodynamics, vol. 25, 1993. pp.173-192. T.P., I. and NANCOLLA, G.H. Equil—A general Computational Method for the Calculation of Solution Equilibria. Analytical Chemistry, vol. 44, no. 12, 1972. pp. 1940–1950. IZATT, R.M., EATOUGH, D., CHRISTENSEN, J.J., and BARTHOLOMEW, C.H. Calorimetrically Determined Log K, Hº and Sº Values for the Interaction of Sulphate Ion with Several Bi-and Tervalent Metal Ions, Journal of Chemical Society A, 1969. pp. 45–53. JASKULA, M. and HOTLOS′. J. Mean Thermodynamic Activity Coefficient of CuSO4 in the Ternary system CuSO4-H2SO4-H2O at 60ºC. Hydrometallurgy, vol. 31, 1992. pp. 233–242. JONES, M.W., PAPANGELAKIS, V.G., and STEYL, J.D.T. Chemical Modelling of MgSO4 Solubility in Atmospheric Chloride Leaching of Laterites. Hydrometallurgy 2008, Phoenix, Az. Society for Mining, Metallurgy and Exploration, 2008. pp. 561–569. KAZANSKY, V. AND SOLKAN, V. Proton Solvation and Self-Dissociation of 100% Sulphuric Acid: The Quantum-Chemical Analysis. Physical Chemistry Chemical Physics, vol. 5, 2003. pp. 31–35. 439


401-444_Steyl:text

2/16/09

3:10 PM

Page 440

KITTEL, C. Introduction to Solid State Physics, 7th Edition, John Wiley & Sons, New York. 1996. KLAMT, A. and SCHUURMANN, G. COSMO: A New Approach to Dielectric Screening in Solvents with Explicit Expressions for the Screening Energy and its Gradient. Journal of Chemical Society, Perkin Transactions 2, 1993. pp. 799–805. KOHN, W. and SHAM, L.J. 1965. Self-Consistent Equations including Exchange and Correlation Effects. Physical Review A, vol. 140, 1965. pp. 1133–1138. LINDSAY, W.T. Estimation of Concentration Quotients for ionic Equilibria in High Temperature Water: The Model Substance Approach. Proceedings of the 41st International Water Conference, Pittsburgh, PA, 1980. 284 pp. LINDSTROM, R.E. and WIRTH, H.E. Estimation of the Bisulphate Ion Dissociation in Solutions of Sulphuric Acid and Sodium Bisulphate. The Journal of Physical Chemistry, vol. 73, no. 1, 1969. pp. 218–223. LINKE, W.F. and SEIDELL, A. Solubilities: Inorganic and Metal-Organic Compounds, A Compilation of Solubility Data from the Periodical Literature, 4th Edition, American Chemical Society, Washington, D.C. 1958. vol. 1 and 1965, vol. 2. LIDE, D.R. Handbook of Chemistry and Physics, 71st Edition, CRC Press, Boca Raton. 1991. LIDE, D.R. Handbook of Chemistry and Physics, 85th Edition, CRC Press, Boca Raton. 2004. LIU, H. and PAPANGELAKIS, V.G. Chemical Modeling of High Temperature Aqueous Processes. Hydrometallurgy, vol. 79, 2005a. pp. 48–61. LIU, H. and PAPANGELAKIS, V.G. Thermodynamic Equilibrium of the O2-ZnSO4-H2SO4-H2O System from 25 to 250ºC. Fluid Phase Equilibria, vol. 234, 2005b. pp. 122–130. LIETZKE, M.H., STOUGHTON, R.W., and YOUNG, T.F. The Bisulfide Acid Constant from 25 to 225ºC as Computed from Solubility Data. Journal of Physical Chemistry, vol. 65, 1961. pp. 2247–2249. LONG, D.T. and ANGINO, E.E. Chemical Speciation of Cd, Cu, Pb, and Zn in Mixed Freshwater, Seawater and Brine Solutions. Geochimica et Cosmochimica Acta, vol. 41, 1977. pp. 1183–1191. MAGINI, M. Solute Structuring in Aqueous Iron(III) Sulphate Solutions. Evidence for the formation of Iron(III)-Sulphate Complexes. Journal of Chemical Physics, vol. 70, no. 1, 1979. pp. 317–324. MAJIMA, H. and AWAKURA, Y. Water and Solute Activities of the Solution Systems of H2SO4-CuSO4-H2O and HCl-CuCl2-H2O. Metallurgical Transactions B, vol. 19B, 1988. pp. 347–354. MAJIMA, H., AWAKURA, Y., and KAWASAKI, Y. Activities of Water and Solutes in the Aqueous Solution Systems H2SO4–Mx(SO4)y and HCl–MClx. Agne Shofu Publishing Inc., Tokyo, 1988. 50 pp. MARCUS, Y. The Hydration Entropies of ions and their Effects on the Structure of Water. Journal of Chemical Society, Faraday Transactions, vol. 82, 1986. pp. 233–242. MARCUS, Y. The Thermodynamics of Solvation of Ions. Journal of Chemical Society, Faraday Transactions, vol. 83, 1987. pp. 339–349. MARGOLIS, E.J. Chemical Principles in Calculations of Ionic Equilibria. The MacMillean Company, New York. 1966. MARSHALL, W.L. and JONES, E.V. Second Dissociation Constant of Sulphuric Acid from 25 to 350°C. Evaluated from Solubilities of Calcium Sulphate in Sulphuric Acid Solutions. Journal of Physical Chemistry, vol. 70, 1966. pp. 4028–4040. MARSHALL, W.L. and SLUSHER, R. Aqueous Systems at High Temperature. XV. Solubility and Hydrolytic Instability of Magnesium Sulphate in Sulphuric Acid-Water and Deuterosulphuric Acid-Deuterium Oxide Solutions, 200º to 350ºC. Journal of Chemical and Engineering Data, vol. 10, no. 4, 1965. pp. 353–358. MARX, D., TUCKERMAN, M.E., HUTTE, J., and PARRINELLO, M. The Nature of the Hydrated Excess Proton in Water. Nature, vol. 397, 1999. pp. 601–604. MILADINOVIC, J., TODOROVIC, M., and NINKOVIC, R. Osmotic Coefficient of the ZnSO4(aq) at T = 298.15 K. Journal of Chemical Thermodynamics, vol. 34, 2002. pp. 1769–1776. NAIR, V.S.K. and NANCOLLAS, G.H. Thermodynamics of Ion Association. Part IV. Magnesium and Zinc Sulphates. Journal of Chemical Society, 1958. pp. 3706–3710. NAIR, V.S.K. and NANCOLLAS, G.H. Thermodynamics of Ion Association. Part VI. Some Transition-Metal Sulphates. Journal of Chemical Society, 1959. pp. 3934–3939. OBLAD, A.E. Simulation of Mine Water Neutralization, M.Sc Thesis, University of Utah. 1975. 440


401-444_Steyl:text

2/16/09

3:10 PM

Page 441

OSCARSON, J.L., IZATT, R.M., BROWN, P.R., PAWLAK, Z., GILLESPIE, S.E., and CHRISTENSEN, J.J. Thermodynamic Quantities for the Interaction of SO42− with H+ and Na+ in Aqueous Solution from 150 to 320ºC. Journal of Solution Chemistry, vol. 17, no. 9, 1988. pp. 841–863. OYKOVA, T.G. and C. BALAREW. Thermodynamic Study of Magnesium Sulphate-Ferrosulphate-Water System at 25ºC, Comptes`Rendus de l’Academie Bulgare des Sciences, vol. 27, 1974. pp. 1211–1214. PABALAN, R.T. AND K.S. PITZER. Thermodynamics of Concentrated Electrolyte Mixtures and the Prediction of Mineral Solubilities to High Temperatures for Mixtures in the System Na-K-Mg-Cl-SO4-OH-H2O. Geochimica et Cosmochimica Acta, vol. 51, 1987. pp. 2429–2443. PAPANGELAKIS, V.G. High Temperature Aqueous Chemistry. Short Course Notes – Autoclave Chemistry and Engineering. Pressure Hydrometallurgy 2004, Banff, Canada. The Metallurgical Society of CIM, Montreal, Ch. 1, 2004. pp. 1–58. PAPANGELAKIS, V.G., BLAKEY, B.C., and LIAO, H. Hematite Solubility in Sulphate Process Solutions. In Hydrometallurgy ’94 Symposium, Cambridge, England. The Institute of Mining and Metallurgy and the Society of Chemical Industry, Chapman & Hall, London, 1994. pp. 159–175. PERDEW, J.P., WANG, Y. Accurate and Simple Analytic Representation of the Electron-Gas Correlation Energy, Physical Reviews B, vol. 45, 1992. pp. 13244–13249. PERRIN, D.D. Ionisation constants of inorganic acids and basis in aqueous solution, IUPAC Chemical Data Series, No. 29, International Union of Pure and Applied chemistry, Pergamon Press. 1982. PITZER, K.S. Ion Interaction Approach: Theory and Data Correlation. Activity Coefficients in Electrolyte Solutions, 2nd Edition, K.S. Pitzer (ed.). CRC Press, Boca Raton, Ch. 3. 1991. PITZER, K.S. Thermodynamic Properties of Aqueous Solutions of Bivalent Sulphates. Journal of Chemical Society, Faraday Transactions (II), vol. 68, 1972. pp. 101–113. PITZER, K.S. and MAYORGA, G. Thermodynamics of Electrolytes. III. Activity and Osmotic Coefficients for 2-2 Electrolytes. Journal of Solution Chemistry, vol. 3, no. 7, 1974. pp. 539–546. PITZER, K.S, ROY, R.N., and SILVESTER, L.F. Thermodynamics of Electrolytes. 7. Sulphuric Acid. Journal of the American Chemical Society, vol. 99, 1977. pp. 4930–4936. POWELL, H.K.J. Entropy Titrations: A Reassessment of Data for the Reaction of the Sulphate Ions with Protons and with Bivalent Metal Ions, Journal of Chemical Society, Dalton Transactions, Part III, 1973. pp. 1947–1951. PRUE, J.E. Ion Association and Solvation. Chemical Physics of Ionic Solutions, Ed. B.E. Conway and R.G. Barradas. John Wiley & Sons, New York, Ch.9, 1966. pp. 163–173. PHUTELA, R.C. and PITZER, K.S. Heat Capacity and Other Thermodynamic Properties of Aqueous Magnesium Sulphate to 473 K. Journal of Physical Chemistry, vol. 90, 1986. pp. 895–901. PYE, C.C and RUDOLPH, W.W. An ab initio and Raman Investigation of Magnesium(II) Hydration. Journal of Physical Chemistry A, vol. 102, 1998. pp. 9933–9943. PYE, C.C and RUDOLPH, W.W. An ab initio and Raman Investigation of Sulphate ion Hydration. Journal of Physical Chemistry A, vol. 105, 2001. pp. 905–912. RAO, N.R. Electrolyte Dissociation in Sulphuric Acid as Studies by Raman Effect. Indian Journal of Physics, vol., XIV, Part II, 1940. pp. 143–151. RARD, J.A., HABENSCHUSS, A., and SPEDDING, F.H. A Review of the Osmotic Coefficients of Aqueous H2SO4 at 25ºC. Journal of Chemical and Engineering Data, vol. 21, no. 3, 1976. pp. 374–379. RARD, J.A. and MILLER, D.G. Isopiestic Determination of the Osmotic Coefficient of Aqueous Na2SO4, MgSO4, and Na2SO4-MgSO4 at 25ºC. Journal of Chemical and Engineering Data, vol. 26, 1981. pp. 33–38. RARD, J.A. and CLEGG, S.L. Isopiestic Determination of the Osmotic and Activity Coefficients of {zH2SO4 + (1-z)MgSO4}(aq) at T = 298.15K. II. Results for z = (0.43040, 0.28758, and 0.14399) and Analysis with Pitzer’s Model. Journal of Chemical Thermodynamics, vol. 31, 1999. pp. 399–429. REARDON, E.J. and BECKIE, R.D. Modelling Chemical Equilibria of Acid Mine-Drainage: The FeSO4H2SO4-H2O System. Geochimica et Cosmochimica Acta, vol. 51, 1987.pp. 2355–2368. REMSUNGNEN, T. and RODE, B.M. Molecular Dynamics Simulation of the Hydration of Transition Metal Ions: The Role of Non-Additive Effects in the Hydration Shells of Fe2+ and Fe3+ Ions. Chemical Physics Letters, vol. 385, 2004. pp. 491–497. 441


401-444_Steyl:text

2/16/09

3:10 PM

Page 442

RENON, H., FURST, W., PLANCHE, H., and BALL, F.X. Thermodynamics of Concentrated Electrolyte Solutions Applied to Liquid-Liquid Extraction of Metals and Solid-Liquid Equilibria. Hydrometallurgical Process Fundamentals, R.G. Bautista (ed.), NATO Conference Series VI, Plenum Press, New York, vol. 10, 1984. pp. 293–313. ROBINSON, R.A. and STOKES, R.H. Electrolyte Solutions, 2nd Edition, Academic Press, New York.1959. RUDOLPH, W. Structure and Dissociation of the Hydrogen Sulphate Ion in Aqueous Solution over a Broad Temperature Range: A Raman Study. Zeitschrift für Physikalische Chemie, vol. 194, 1996. pp. 73–95. RUDOLPH, W.W. Hydration and Water-Ligand Replacement in Aqueous Cadmium(II) Sulphate Solution – A Raman and Infrared Study. Journal of Chemical Society, Faraday Transactions, vol. 94, no. 4, 1998. pp. 489–499. RUDOLPH, W., BROOKER, M.H., and TREMAINE, P.R. Raman Spectroscopic Investigation of Aqueous FeSO4 in Neutral and Acidic Solutions from 25ºC to 303ºC: Inner- and Outer-Sphere Complexes. Journal of Solution Chemistry, vol. 26, no. 8, 1997. pp. 757–777. RUDOLPH, W., BROOKER, M.H., and TREMAINE, P. Raman- and Infrared Spectroscopic Investigation of Aqueous ZnSO4 Solutions from 8ºC to 165ºC: Inner- and Outer-Sphere Complexes. Zeitschrift für Physikalische Chemie, vol. 209, 1999a. pp. 181–207. RUDOLPH, W., BROOKER, M.H., and TREMAINE, P.R. Raman Spectroscopy of Aqueous ZnSO4 Solutions under Hydrothermal Conditions: Solubility, Hydralysis, and Sulphate Ion Pairing. Journal of Solution Chemistry, vol. 28, no. 5, 1999b. pp. 621–630. RUDOLPH, W.W., IRMER, G., and HEFTER, G.T. Raman Spectroscopic Investigation of Speciation in MgSO4(aq). Physical Chemistry Chemical Physics, vol. 5, 2003. pp. 5253–5261. RULL, F., BALAREW, CH., ALVAREZ, J.L., SOBRON, F., and RODRIGUEZ, A. 1Raman Spectroscopic Study of ion Association in Aqueous Magnesium Sulphate Solutions. Journal of Raman Spectroscopy, vol. 25, 1994. pp. 933–941. SHOCK, E.L. and HELGESON, H.C. Calculation of the Thermodynamic and Transport Properties of Aqueous Species at High Pressures and Temperatures: Correlation Algorithms for Ionic Species and Equation of State Predictions to 5 kb and 1000ºC. Geochimica et Cosmochimica Acta, vol. 52. 1988. pp. 2009–2036. SENANAYAKE, G. and MUIR, D.M. Studies on Liquid Junction Potentials in Concentrated Chloride Solutions and Determination of Ionic Activities and Hydration Numbers by the EMF Method. Electrochimica Acta, vol. 33, no. 1, 1988. pp. 3–9. SENEVIRATNE, D.S., PAPANGELAKIS, V.G., ZHOU, X.Y., and LVOV, S.N. Potentiometric pH Measurements in Acidic Sulphate Solutions at 250ºC Relevant to Pressure Leaching. Hydrometallurgy, vol. 68, 2003. pp. 131–139. SILLEN, L.G. and MARTEL, A.E. Stability Constants of Metal-ion Complexes. Special Publication No.17, The Chemical Society, Metcalfe & Cooper Ltd., London. 1964. SILLEN, L.G. and MARTELL, A.E. Stability Constants of Metal-ion Complexes. Special Publication No.25, The Chemical Society, Alden & Mowbray Ltd., Oxford. 1971. SMITH, R.M. and MARTELL, A.E. Critical Stability Constants, vol. 6, Plenum Press, New York. 1989. SNIPES, H.P., MANLY, C., and ENSOR, D.D. Heats of Dilution of Aqueous Electrolytes, Journal of Chemical and Engineering Data, vol. 20, no. 3, 1975. pp. 287–291. STERZEL, M. and AUTSCHBACH, J. Toward an Accurate Determination of 195Pt Chemical Shifts by Density Functional Computations. Inorganic Chemistry, vol. 45, no. 8, 2006. pp. 3316–3324. STEYL, J.D.T. Kinetic Modelling of Chemical Processes in Acid Solution at T ≤ 200°C. I. Thermodynamic and Speciation in H2SO4-Metal(II)SO4-H2O System. Anglo Research Report I200802918, Report 1 (not published). 2008. SWEETON, F.H., MESMER, R.E., and BAES, C.F. Acidity Measurements at Elevated Temperatures. VII. Dissociation of Water. Journal of Solution Chemistry, vol. 3 no. 3, 1974. pp. 191–214. TARTAR, H.V. and NESS, A.T. A Thermodynamic Study of the System Zinc Sulphate-Sulphuric Acid-Water at 25ºC. Journal of the American Chemical Society, vol. 63, 1941. pp. 28–36. TRASATTI, S. The Absolute Electrode Potential: An Explanatory Note (Recommendations 1986). International Union of Pure and Applied Chemistry, Pure and Applied Chemistry, vol. 58, no. 7, 1986. pp. 955–966. 442


401-444_Steyl:text

2/16/09

3:10 PM

Page 443

TREMAINE, P.R., TREVANI, L.N., and RUDOLPH, W.W. Acid-Base Ionization and Metal Complexation under Hydrothermal Conditions by UV-Visible and Raman Spectroscopy. Pressure Hydrometallurgy 2004, Banff, Canada. The Metallurgical Society of CIM, Montreal, 2004. pp. 545–560. TURNER, D.J. Dissociation of the Bisulphate Ion at Moderate Concentrations. Journal of Chemical Society, Faraday Transactions, vol. 70, 1974. pp. 1346–1356. VASIL’EV, V.P. Influence of Ionic Strength on the Instability Constants of Complexes. Russian Journal of Inorganic Chemistry, vol. 7, no. 8, 1962. pp. 924–927. VOSKO, S.J., WILK, L., and NUSAIR, M. Accurate Spin-Dependent Electron Liquid Correlation Energies for Local Spin Density Calculations: A Critical Analysis. Canadian Journal of Physics, vol. 58, 1980. pp. 1200–1211. WAGMAN, D.D., EVANS, W.H., PARKER,V.B., SCHUMM, R.H., HALOW, I., BAILEY, S.M., CHURNEY, K.L., and NUTTALL, R.L. The NBS Tables of Chemical Thermodynamic Properties, Selected values for inorganic and C1 and C2 organic substances in SI units. Journal of Physical and Chemical Reference Data, vol. 11, Supplement No. 2, Published by the American Chemical Society and the American Institute of Physics for the National Bureau of Standards. 1982. WANG, X.-B., NICHOLAS, J.B., and WANG, L.-S. Electronic Instability of Isolated SO42- and its Solvation Stabilization. Journal of Chemical Physics, vol. 113, no. 24, 2000. pp. 10837–10840. WALRAFEN G.E., YANG, W.-H., CHU, Y.C., and HOKMABADI, M.S. Structures of Concentrated Sulphuric Acid Determined form Density, Conductivity, Viscosity and Raman Spectroscopic Data. Journal of Solution Chemistry, vol. 29, no. 10, 2000. pp. 905–936. WALRAFEN G.E., YANG W.-H., and CHU, Y.C. High-Temperature Raman Investigation of Concentrated − Sulphuric Acid Mixtures: Measurement of H-Bond ΔH Values between H3O+ or H5O2+ and HSO4 . Journal of Physical Chemistry A, vol. 106, 2002. pp. 10162–10173. WANG, P., ANDERKO, A., SPRINGER, R.D., and YOUNG, R.D. Modeling Phase Equilibria and Speciation in Mixed-Solvent Electrolyte Systems: II. Liquid-Liquid Equilibria and Properties of Associating Electrolyte Solutions. Journal of Molecular Liquids, vol. 125, 2006. pp. 37–44. XIANG, T., JOHNSTON, K.P., WOFFORD, W.T., and GLOYNA, E.F. Spectroscopic Measurments of pH in Aqueous Sulphuric Acid and Ammonia from Sub- to Supercritical Conditions. Industrial & Engineering Chemistry Research, vol. 35, no. 12, 1996. pp. 4788–4795. YATSIMIRSKII, K.B. and VASIL’EV, V.P. Instability Constants of Complex Compounds (Translated from Russian), Consultants Bureau, New York. 1960. YOUNG, T.F. and BLATZ, L.A. The Variations of the Properties of Electrolyte Solution with Degrees of Dissociation. Chemical Reviews, vol. 44, no. 1, 1949. pp. 93–115. YOUNG, T.F., MARANVILLE, L.F., and SMITH, H.M. Raman Spectral Investigations of Ionic Equilibria in Solutions of Strong Electrolytes. The Structure of Electrolyte Solutions, W.J. Hamer (ed.), John Wiley & Sons Inc., New York, Ch.4, 1959. pp. 35–63. ZEMAITIS, J.F., CLARK, D.M., RAFAL, M., and SCRIVNER, N.C. Handbook of Aqueous Electrolytic Thermodynamics − Theory & Application. A publication of the Design Institute for Physical Property Data (DIPPR), American Institute of Chemical Engineers Inc., New York. 1986. ZHANG, X., ZHANG, Y., and LI, Q. Ab initio Studies on the Chain of Contact Ion Pairs of Magnesium Sulphate in Supersaturated State of Aqueous Solution. Journal of Molecular Structure (Theochem), vol. 594, 2002. pp.19–30. ZIEGLER, T. and AUTSCHBACH, J. Theoretical Methods of Potential Use for Studies of Inorganic Reaction Mechanisms. Chemical Reviews, vol. 105, 2005. pp. 2695–2722. Software references Accelrys Software Inc. 2008. Materials Studio, San Diego, Release 4.2. HSC Chemistry. 2006. Chemical Reaction and Equilibrium Software with Thermochemical Database, Outokumpu Research, Pori, Finland, Ver. 6.0. Inorganic Crystal Structure Database (ICSD). 2008. Fachinformationzetrum, Karlruhe, Germany and US Department of Commerce, Ver.1.4.4. Matlab Software. 2008. The MathWorks Inc., Natick, MA, Release 2008b. 443


401-444_Steyl:text

2/16/09

3:10 PM

Page 444

Johann Steyl Technical Specialist, Anglo Research, South Africa Johann started his employment at Mintek in 1994, after graduating from the University of Pretoria with a degree in Chemical Engineering. He joined Anglo Research in 2001 and is currently working as technical specialist in the Technology Division. His research focuses mainly on the development of new processes for the treatment of ores and concentrates.

445-456_Venter:text

2/14/09

12:05 PM

Page 445

VENTER, R. and BOYLETT, M. The evaluation of various oxidants used in acid leaching of uranium. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

The evaluation of various oxidants used in acid leaching of uranium R. VENTER* and M. BOYLETT* *GRD Minproc Pty (Ltd), Bryanston, South Africa Uranium in the tetravalent state has very low solubility in acid and alkaline solutions and needs to be oxidized to the hexavalent state to become soluble. In acid leaching of uranium the uranium is oxidized by ferric iron, which in turn is reduced to ferrous iron. The reduced iron needs to be oxidized to the ferric form to allow the uranium dissolution reaction to proceed. This paper considers the known and proven oxidants used in acid leaching of uranium. These include manganese dioxide, sodium chlorate, hydrogen peroxide, ferric iron, oxygen and sulphur dioxide/air (oxygen). The reaction chemistry of each oxidant as well as the practical applications are considered and discussed. All the oxidants discussed work adequately in acid leaching of uranium and it is possible to engineer solutions for the use of the oxidants in acid leaching on plant scale. The availability and supply of the oxidant as well as the cost and environmental impact of the oxidant plays a major role in the selection of a suitable oxidant. The location of the plant will also have an impact on these issues.

Introduction Dissolution of uranium is the first hydrometallurgical process in the extraction of uranium from the uranium containing minerals in the ore. Leaching procedures selected for dissolving uranium are dependant in part on the physical characteristics of the ore such as: type of uranium mineralization, ease of liberation, and the nature of other constituent minerals present. In nature uranium occurs in the tetravalent and hexavalent oxidation states. Tetravalent uranium has a low solubility in both dilute acid and carbonate (alkali) solutions. To achieve economic recovery of uranium in the tetravalent state, oxidation to the hexavalent state is essential. It is therefore important to maintain proper oxidizing conditions during leaching of these minerals to achieve high uranium extraction. A number of oxidants have historically been used and are currently used to oxidize tetravalent uranium to hexavalent uranium in acid and alkaline circuits. When selecting an oxidant to be used on commercial scale, consideration must be given to a number of factors. Apart from the effectiveness of the oxidant to maintain an oxidizing environment, the availability and cost of the oxidant must be considered. Part of this is the logistics of getting the oxidant to site and administering it to the leach. The latter could introduce a number of THE EVALUATION OF VARIOUS OXIDANTS USED IN ACID LEACHING OF URANIUM

445

445-456_Venter:text

2/14/09

12:05 PM

Page 446

engineering challenges to the flowsheet. This paper considers different oxidants typically used in acid leach uranium circuits, namely: • Manganese dioxide (MnO2) as milled pyrolusite • Sodium chlorate (NaClO3) • Hydrogen peroxide (addition as H2O2 or Caro’s acid (H2SO5)) • Ferric iron, frequently as roaster calcine from a pyrite burning acid plant • Oxygen in pressure leach circuits and • Sulphur dioxide/air(oxygen) mixture. The reaction chemistry and application of each oxidant, as well as engineering issues on plant scale and the availability and logistics are discussed. Mineralogy Uranium in typical uranium bearing minerals occur in the tetravalent (U4+) and hexavalent (U6+) oxidation states. Primary uranium ores are predominantly in veins or pegmatites but are also found in sedimentary and placer deposits. Primary ores include uraninite and pitchblende (UO2) (Kinnaird and Nex, 2008). Secondary uranium ores are found in weathered zones of primary deposits and precipitated in sediments. These include autinite (Ca(UO 2 ) 2 (PO 4 ) 2 .10H 2 O), torbernite (Cu(UO 2 ) 2 (PO 4 ) 2 .10H 2 O), uranophane (Ca(UO 2 ) 2 SiO 3 (OH) 2 .5(H 2 O)), carnotite (K2(UO2)2[VO4]2.3H2O) and schoepite ((UO2)8O2(OH)12.12(H2O)) (Yan and Connely, 2008). Multiple oxides are complex associations of uranium with rare earths, tin, tantalum, niobium and titanium with extensive crystal lattice substitutions that makes them extremely refractory. These include davidite ((La,Ce, Ca)(Y, U)(Ti, Fe3+)20O38), brannerite ((U, Ca, Ce)(Ti, Fe)2O6) and betafite ((Ca, U)2(Ti, Nb, Ta)2O6(OH)) and are more difficult to process, and products can pose environmental issues (Yan and Connely, 2008.) The silicates, uranothorite ((Th, U)SiO4) and coffinite (U(SiO4)1-x(OH)4x), having the zircon structure, would be expected to be refractory, but in practice readily dissolve in oxidizing acid or carbonate solutions (Lunt, et. al, 2007.) Uraninite and coffinite occur in the tetravalent state (Lunt, et al., 2007). In the Southern African context examples of uraninite are the Witwatersrand ores in South Africa (Ford, 1993) and Rössing in Namibia (Kinnaird and Nex, 2008). Kayelekera in Malawi is an example of coffinite (Chilumanga, 2008). Minerals occurring in the hexavalent state typically include carnotite found at Langer Heinrich and Trekkopje in Namibia (Kinnaird and Nex, 2008). In phosphate deposits uranium is found in the hexavalent form in the apatite structure. Examples of the phosphate deposits are Bakouma in the Central African Republic, Florida and other commercial phosphoric acid producers from which uranium is a possible by-product (Merrit, 1971:22). There are also minerals described as refractory, such as brannerite, which require more aggressive leach conditions, but the basic chemistry is unchanged (Yan and Connely, 2008). Simple chemistry Uranium is typically recovered in acid circuits using sulphuric acid, or in alkaline circuits using a sodium carbonate bicarbonate mix. Acid leach is the preferred method, but in cases where the acid consumption is very high, mostly caused by gangue minerals, it makes more sense to consider alkaline leach. The selection of acid or alkaline leach is based on a number 446


445-456_Venter:text

2/14/09

12:05 PM

Page 447

of key process drivers. These are discussed elsewhere (Lunt et al., 2007). In typical extraction circuits, after dissolution, uranium is usually extracted using strong base resins and/or alamine 336 in a solvent extraction plant. Solvent extraction is used only for acid circuits (Lunt et al., 2007). In these circuits the uranium is solubilized in its hexavalent form. This is true for both acid and alkaline circuits: Hexavalent uranium dissolves as the UO22+ ion and produces complex uranyl anions: Uranyl sulphate [UO2(SO4)3]4- and Uranyl carbonate [UO2(CO3)3]4The focus of this paper is on oxidation in acid circuits and to leave the alkaline circuitry as a separate issue. A typical uranium acid leach flowsheet can be seen in Figure 1. The typical flowsheet comprises comminution to reduce the particle size to a size acceptable to the leach. A pre-concentration step could also be included where the ore is upgraded. The material is then leached in sulphuric acid at atmospheric or elevated pressure and ambient or elevated temperature. Leaching is followed by liquid/solids separation in countercurrent decantation (CCD) thickeners or on belt filters, followed by solvent extraction (or ion

SULPHURIC ACID AND OXIDANT

WASH WATER TAILS TO WASTE

ELUATE

YELLOW CAKE

Figure 1. Typical acid leach flowsheet (Lunt and Holden, 2006)

THE EVALUATION OF VARIOUS OXIDANTS USED IN ACID LEACHING OF URANIUM

447

445-456_Venter:text

2/14/09

12:05 PM

Page 448

exchange or both) and finally product precipitation/recovery. Variations in ion exchange methodology include fixed bed ion exchange (FIX) and countercurrent ion exchange (CCIX). An alternative to liquid/solid separation is resin in pulp (RIP) the development of which will closely parallel the gold CIP designs. Uranium dissolution chemistry in acid Tetravalent uranium must be oxidized to the hexavalent form before dissolution occurs in sulphuric acid. Ferric iron (Fe3+) acts as the principal oxidant of tetravalent uranium in acid leaching circuits. Iron is generally present, either as a constituent in the ore, or introduced as metallic iron as a result of wear and abrasion in the grinding circuit. Iron can also be added on demand. The mechanism for the dissolution of uranium in sulphuric acid is therefore based on the following reactions: The Fe3+ oxidizes the tetravalent uranium (U4+) in the solution to the soluble hexavalent uranium (U6+). The Fe3+ is then reduced to Fe2+. The oxidant is added to regenerate the Fe2+ to Fe3+ so that the leach reaction can continue. In practice it is found that the Fe3+/Fe2+ mechanism is to be considered more than the uranium reaction. The mechanisms are described by the following reactions with MnO2 as oxidant (Merritt, 1971:64):

Thus, the Fe3+/Fe2+ couple facilitates the leach process by acting as the electron carrier during oxidation. The intention of this paper is to consider various ways of achieving the oxidation of ferrous to ferric iron in acidic conditions. Source of iron Iron needs to be present in the leach solution. If insufficient iron is present it can be added by adding ferrous or ferric sulphate or adding steel in some form. Iron can also be added via solution recycle streams from downstream processes. Oxidants Following is a discussion of each of the oxidants typically used in acid leaching of uranium. Manganese dioxide (MnO2) This is the traditional oxidant used for Witwatersrand ores and elsewhere, e.g. Ranger in Australia. The oxidation takes place according to the following equation: From the equation it is clear that each mole of MnO 2 requires 2 moles of acid. Commercially available pyrolusite typically contains in the order of 30% to 50% MnO2 with the remainder potentially being acid consuming gangue minerals. These gangue minerals could also be a source of iron needed for the leach reaction to take place. Manganese dioxide (as pyrolusite) was the first choice in the Witwatersrand as it was freely available as surface outcrops near to the mines and cheap sulphuric acid was locally produced from pyrite in the ore. Vaal Reefs South Uranium Plant operates with this exact flowsheet and has done so for almost 30 years. 448


445-456_Venter:text

2/14/09

12:05 PM

Page 449

In the 1950s through to 1990s in excess of 20 Witwatersrand uranium plants used pyrolusite as oxidant in the leach in preference to the auto oxidation process, which introduced SO2 and air into the leach to oxidize the ferrous iron to ferric iron, as a result of cheap sources of pyrolusite being available (Ford, 1993). It is easy to mill and is added as a slurry under redox control. No special agitation is therefore necessary and the MnO2 does not deactivate or dissociate in the slurry if it does not react with the iron immediately, as is the case with some of the other oxidants (e.g hydrogen peroxide). At Randfontein, Cooke Plant, the nearest supply was from a small pit at the top of the access road. Today most of this pit is surrounded by housing. The apparent lack of availability and environmental considerations weigh against the use of pyrolusite today. South Africa is a major exporter of manganese ores from the Northern and Eastern Cape and a producer of electrolytic manganese dioxide. All of these products would be suitable but with increasing distances from site transport costs increase. Outside the Witwatersrand, for example in Namibia, the most likely source of supply is from Ghana. It is practical via the port of Walvis Bay, but also expensive. The environmental aspect involves the production of Mn2+ in solution. In theory this will be deposited on the slimes dam as the hydroxide. In practice, the pH to achieve this is high and must be maintained, which equates to high neutralization costs. Pyrite oxidation will lower the dam pH in the long term, thus soluble manganese may enter water tables and water courses. This is a greater problem on the Witwatersrand than in Namibia for instance. Namibian ores are essentially pyrite free and the climate is dryer, compared to the pyrite containing ores in a wetter climate on the Witwatersrand. The higher pH needed to precipitate the manganese still holds true for Namibia. The relatively low MnO2 concentrations in the pyrolusite cause relatively high volumes of pyrolusite to be shipped to site, which introduces logistical challenges. The acid consuming gangue minerals in the pyrolusite can also significantly increase acid consumption in the leach. Sodium chlorate Sodium chlorate was traditionally the first choice oxidant in North American uranium plants. The oxidation reaction takes place according to the following equation: From the equation it is clear that each mole of sodium chlorate requires 3 moles of acid. In Southern Africa sodium chlorate was not considered as an oxidant due to availability of cheaper and more convenient alternatives (i.e. pyrolusite). Sodium chlorate is added to the leach slurry as a solution. No special agitation is therefore necessary. During the reaction with sodium chlorate in the leach, chloride ions go into solution with the possible adverse effects on ion exchange loading and materials of construction. Sodium chlorate is expensive. It is also used as oxidant in PGM refining. Partial reduction leads to intermediate products, such as chlorite and hypochlorite. Sodium chlorate powder has an auto-ignition risk when in contact with organic. It therefore needs procedures and proper engineering for safe handling. Hydrogen peroxide (H2O2) Hydrogen peroxide was used as oxidant in acid leaching in the nineteen eighties. Lucas et al. (1983) describe plant trials performed in Queensland where pyrolusite was replaced by Caro’s acid as oxidant. De Vries (1984) patented a process that utilizes hydrogen peroxide in an acid leach of uranium ore. THE EVALUATION OF VARIOUS OXIDANTS USED IN ACID LEACHING OF URANIUM

449

445-456_Venter:text

2/14/09

12:05 PM

Page 450

Hydrogen peroxide can be used as is or can be made up as Caro’s acid or peroxymonosulphuric acid (H2SO5) by mixing with concentrated sulphuric acid according to the following reaction: The oxidant regenerates the ferric iron (Fe3+) from ferrous iron (Fe2+). In the case of Caro’s acid the reaction is as follows: In the case of hydrogen peroxide the reaction is as follows:

From these equations it is clear that one mole of acid is consumed for every one mole of Caro’s acid or hydrogen peroxide. Lucas et al. (1983) reported similar uranium extractions to using pyrolusite with a reduction in acid consumption and oxidant consumption. They also reported a reduction in lime consumption for neutralization of acid effluent but this could not be fully quantified. Mention was also made that excluding manganese from the leach liquors should reduce the neutralization requirement for the manganese and pyrolusite gangue minerals. Test work performed by GRD Minproc and Mintek supports the reduction in acid and oxidant consumption. From the plant trial and other operations the main benefits of using hydrogen peroxide (or Caro’s acid) as oxidant were claimed to be: • Better process control • Oxidant savings • Sulphuric acid savings • Lime savings • Cleaner oxidant handling • Simplified effluent treatment. On the downside, hydrogen peroxide will dissociate if it does not react rapidly with the iron in solution. The dispersion of the hydrogen peroxide through the slurry is therefore very important. It is reported that difficulties with dispersion of the hydrogen peroxide in the leach slurry was experienced during scale-up to plant scale at Rössing and Western Areas (Boylett, 2008). It is also reported that using Caro’s acid will reduce the hydrogen peroxide consumption as a result of more efficient use of the hydrogen peroxide (James, 2008). Hydrogen peroxide can also be used for precipitation of uranium as uranyl peroxide (UO4.xH2O) instead of the more traditional ammonium diuranate (ADU), which eliminates the use of ammonia as a precipitation reagent. Uranyl peroxide is also reported to be easier to thicken and filter or centrifuge and requires only drying and not calcining for dispatch. It is also reported to produce a purer product. This is the subject of another paper but it is sufficient to say that this product is the more common end product of an alkaline leach circuit than from an acid circuit. It is the end product from Langer Heinrich for example. As hydrogen peroxide becomes more popular for precipitation in acid circuits this might change. Hydrogen peroxide was often discounted based on its relatively high price. It is very favourable from an environmental perspective as the reaction chemistry produces only water. Production of Caro’s acid was traditionally a challenge as a result of high heat generation by the reaction between the acid and the hydrogen peroxide. Manufacturers of hydrogen peroxide claim that they now have simple methods of producing ‘Caro’s acid’ for use in the uranium 450


445-456_Venter:text

2/14/09

12:05 PM

Page 451

leach circuit (James, 2008). It is now made in situ by feeding sulphuric acid and hydrogen peroxide into a funnel device. The high heat of reaction is immediately dissipated into the slurry below. For those gold plants using hydrogen peroxide for cyanide destruction the storage, use and availability of peroxide is an advantage. The greater use of peroxide by the gold plants enhances its applicability and availability in remote areas. Hydrogen peroxide is supplied in large volumes to a number of operations throughout Africa, for instance gold producers in Ghana. Suppliers maintain that availability and logistics of hydrogen peroxide is not a problem. Ferric iron from roaster calcine Calcine from the roasting of pyrite, which has been treated with concentrated sulphuric acid, is also used as oxidant in acid leaching of uranium. The resulting slurry is combined with the return filtrate and contacted with sulphur dioxide in flotation cells. The auto-oxidation reaction results in the conversion of ferrous iron to ferric iron with the formation of additional sulphuric acid. The process is described by the following reactions (Ford, 1993):

This process not only produces a concentrated source of ferric iron, but also liberates any uranium and gold present in the calcine for subsequent recovery. A number of plants practise milder forms of calcine digestion by adding sulphuric acid to the calcine slurry in air-agitated pachucas at between 70ºC and 90ºC. Pyrolusite is also added to maximize the concentration of ferric iron. This practice is followed by Rössing and was followed by Hartebeesfontein uranium plant amongst others. Oxygen Uranium ores containing sulphidic minerals can be leached by the addition of oxygen only at elevated temperature and pressure. The addition of sulphides to ores that do not contain sulphidic minerals has also been proposed. Sulphuric acid and ferric sulphate are generated in situ during pressure leaching by the reaction of oxygen with the sulphides. If pyrite and uraninite are present the following reactions might be expected: Pyrite is oxidized to produce soluble iron and acid: Ferrous iron is oxidized to ferric iron by oxygen:

As the temperature increases above 170ºC ferric iron is removed from the solution through iron hydrolysis reactions:


451

445-456_Venter:text

2/14/09

12:05 PM

Page 452

Jarosite also forms during the hydrolysis reaction as described by the following general reaction (Guise and Castro). M indicates the cation.

←

An additional source of dissolved iron is from steel from the crushing and grinding plants:

The following reactions represent the dissolution of uranium: Tetravalent uranium is oxidized by ferric iron: The dissolution of tetravalent uranium by oxygen: Hexavalent uranium will dissolve in the presence of sulphuric acid alone: Acid consuming gangue, including carbonates, is leached by acid: These reactions proceed more rapidly to completion at high temperature and with the increased solubility of oxygen in solution at increased pressure. Temperatures of above 170ºC are suggested. Oxygen consumption is in the order of 2 tons of oxygen per ton of sulphur. Pressure leaching is utilized in the treatment of refractory ores and is performed in autoclaves at elevated temperature and pressure. Autoclaves were trialled at the Western Deep Levels plant in the late ’70s (Bovey and Stewart, 1978). Acid leaching in autoclaves, based on these plant trials, were used most recently in the design of the autoclaves for the Dominion Reefs plant owned by Uranium One. Indicated advantages of the pressure leaching process include: improved extraction, decreased operating costs, a decreased amount of impurities and free acid in the leach solutions and improved slurry filtration properties. Pressure leaching also results in increased recovery of gold and uranium from the pyrite. Disadvantages include increased corrosion and maintenance. Another disadvantage is the production of soluble SiO2 that contaminates resin in ion exchange circuits and forms crud in solvent extraction circuits. Sulphur dioxide/air (SO2/air) It is a well known fact that a mixture of SO2 gas and air acts as an oxidant in hydrometallurgy (Ho and Quan, 2003). It is used in iron and manganese removal in cobalt circuits, although it has not been proven commercially yet. It was first suggested by workers in the USA and the operating conditions and detailed chemistry was determined by the Government Metallurgical Laboratories (GML) in South Africa in the 1950s (Ford, 1993). It has also been suggested for use in leaching of uranium in an oxidizing sulphuric acid solution as an alternative to MnO2. (Ho and Quan, 2007). In the case of SO2/air mixture as oxidant, oxygen in the air together with the SO2 is responsible for oxidizing the Fe2+ to Fe3+ according to the following reaction:

452


445-456_Venter:text

2/14/09

12:05 PM

Page 453

Both oxygen and SO2 are required in solution for the reaction to proceed, but due to the lower solubility of oxygen compared to SO2/O2 mass transfer limits the maximum reaction rate. The Fe2+ oxidation rate is controlled by the SO2/O2 ratio and the rate of oxygen mass transfer. It is independent of Fe2+ concentration. Higher oxidation rates can be obtained in solutions compared to slurries, as the solids in the slurries decreases the rate of oxygen transfer. Ho and Quan (2007) focused on producing a ferric solution that can be added to the leach slurry rather than in situ addition of SO2/air to the leach. Earlier work by Ho and Quan used sodium sulphite as the source of SO2; however, at a metallurgical plant direct sparging of SO2 would be preferred. The measured Fe2+ oxidation rates using SO2 were very similar when sodium sulphite was used compared to SO2 gas. Ho and Quan found that the SO2/O2 volumetric ratio plays a major role in the oxidation rate. At relatively low SO2/air ratios where there is an excess of oxygen in the solution, the reaction rate is proportional to the SO2 dissolution rate into the solution. However, once the SO2 flow rate exceeds the maximum rate of oxygen mass transfer, the Fe2+ oxidation rate decreases as a result of reducing conditions caused by SO2. The main contributor to the reduced SO2 efficiency is the side reaction of SO2 to produce sulphuric acid, according to the following reaction: A SO2/O2 ratio of greater than 0.36 caused the Fe2+ oxidation rate to become variable. Unlike most of the other oxidants, SO2/air is not an acid consumer. From a large-scale implementation point of view, SO2/air has a number of challenges. For slurries dispersion agitation is necessary with more installed and utilized power. If an acid plant is running on site, SO2 can be sourced from the acid plant. Care must, however, be taken to ensure that the SO2/air ratio from the burning sulphur is satisfactory, as this gas stream might have a higher nitrogen content compared to when pure SO2 gas is mixed with air. Any SO2 escaping from the leach tanks needs to be collected and passed through a scrubbing system. The relatively low SO2 concentration in the gas stream can also give rise to large gas flow rates into the leach. Introduction of the SO2/air mixture into the leach tanks also presents a challenge as it will probably need to be introduced against high slurry heads. This can influence the partial pressures of the different gases in the gas mixture. The solubility of SO2 is higher than oxygen and if the SO2/oxygen ratio in the solution gets too high it will cause reducing conditions and affect the reaction kinetics. Oxidizing iron in a solution stream added to the leach will be easier to achieve. Care should, however, be taken to ensure that sufficient iron is oxidized to the ferric state to oxidize the uranium. As far as is known to the authors, at the time of writing this paper, no commercial operation exist that uses SO2/air as oxidant in acid leaching of uranium. Conclusion Traditionally the tendency in Southern Africa was to use pyrolusite as oxidant in acid leach of uranium. Availability and logistical issues as well as environmental issues seem to be changing this tendency. SO2/air has been tested on many occasions and has proved to work satisfactorily; however, administering the SO2/air gas mixture to the leach on large scale presents a number of challenges. SO2 is an attractive option if acid is produced on site and if sulphur prices are low. It also reduces the number of reagents required to be shipped to site. THE EVALUATION OF VARIOUS OXIDANTS USED IN ACID LEACHING OF URANIUM

453

445-456_Venter:text

2/14/09

12:05 PM

Page 454

Hydrogen peroxide is also an alternative. It has no environmental effects and leads to a reduction in acid and oxidant consumption compared to pyrolusite. Handling and transport of hydrogen peroxide, as well as availability, may still be an issue. Hydrogen peroxide might also become costly compared to the other oxidants. Sodium chlorate is not really considered as a result of introduction of chlorides into the system and the adverse effect it has on ion exchange resins and corrosion of materials of construction. Sodium chlorate also has an explosion risk. All the oxidants described work adequately in acid leaching of uranium. It is also possible to engineer solutions to introduce the oxidant into the leach slurry on plant scale. The factors that play a significant role in selecting a suitable oxidant will include the availability and supply of the oxidant, the cost of the oxidant and the environmental impact of the oxidant. All three of these issues will be affected by the location of the plant and possible sources of oxidant close to the operation. Acknowledgements The authors would like to thank the management of GRD Minproc (Pty)Ltd for permission to publish this paper and to acknowledge the input of their colleagues in undertaking uranium projects and uranium feasibility studies. References BOVEY, H.J. and STEWART, L.N. Pressure Leaching of Uranium Bearing Witwatersrand Ores, SAIMM, 22 March 1978. BOYLETT, M. Personal communication. 2008. CHILUMANGA, P. An Overview of Uranium Mining Development in Malawi, Implementation of Global Best Practice in Uranium & Processing IAEA HQs – VIENNA. 2008. DEVRIES, F.W. Hydrogen Peroxide in Sulfuric Acid Extraction of Uranium Ores, United States Patent 4,425,307, January, 10, 1984. FORD, M.A. Uranium in South Africa, Journal of the South African Institute of Mining and Metallurgy, vol. 93, no. 2, 1993. pp. 37–58. GUISE, L. and CASTRO, F. Purification of Sulphuric Acid Solutions from the Leaching of Nickelliferous Laterites, University of Minho, Department of Mechanical Engineering, Azurem – 4800 Guimaraes - Portugal. HO, E.M. and QUAN, C.H. Iron(II) Oxidation by SO2/O2 in Uranium Leach Solutions, ANSTO, Australia. Hydrometallurgy 2003. HO, E.M. and QUAN, C.H. Iron(II) oxidation by SO2/O2 for use in uranium leaching, Hydrometallurgy, vol. 85, 2007. pp. 183–192. JAMES, A. Sales and Technical Marketing Manager – Exports, Solvay SBU H2O2, Email communication. 2008 KINNAIRD, J.A. and NEX, P. Assessing the Geological Occurrences of Uranium in Africa Implications for Mining and Processing, Uranium Mining and Exploration Conference, Johannesburg, March 2008. LUNT, D., BOSHOFF, P., BOYLETT, M., and EL-ANSARY, Z. Uranium Extraction The Key Process Drivers, The Southern African Institute of Mining and Metallurgy – Namibian Branch – Uranium in Namibia. 2007. 454


445-456_Venter:text

2/14/09

12:05 PM

Page 455

LUNT, D. and HOLDEN, A. Uranium Extraction The Key Issues for Extraction, ALTA Uranium Conference, Perth, Western Australia, May 2006, LUCAS, G.C., FULTON, E.J., VAUTIER, F.E., WATERS, D.J., and RING, R.J. Queensland Mines Plant Trial with Caro’s Acid, Proc. Australas. Inst. Min. Metall., 287, September 1983. MERRITT, R.C. The Extractive Metallurgy of Uranium, Colorado School of Mines. 1971. YAN, D. and CONNELLY, D. Implications of Mineralogy on Uranium Ore Processing. ALTA 2008 Uranium Conference, Perth, Western Australia, June 2008.

Riaan Venter Senior Process Engineer, GRD Minproc, South Africa Riaan has 10 years of metallurgical and process control experience in the minerals processing industry. His experience includes various positions at Anglo Platinum where he started his career in 1998 where he gained production and technical experience on precious metals refining, and research, design and optimization experience on various concentrator plants, smelters and refineries. He moved to GRD Minproc in 2007 as a senior process engineer and has been involved in projects and feasibility studies for platinum, gold and uranium extraction facilities as well as techno economical evaluations of projects. Riaan is registered as a Professional Engineer at ECSA.


455

445-456_Venter:text

456

2/14/09

12:05 PM

Page 456


457-465_Bazhko:text

2/14/09

12:07 PM

Page 457

BAZHKO, O. Application of redox titration techniques for analysis of hydrometallurgical solutions. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Application of redox titration techniques for analysis of hydrometallurgical solutions O. BAZHKO Mintek, South Africa

Regular analysis of different streams is important for the control of hydrometallurgical processes. Various physical and physical-chemical methods are applied for this purpose. Many modern methods require costly equipment and are subject to interferences. Therefore, redox titration methods remain of interest. The use of different types of oxidizing and reducing agents as well as redox indicators enables many cations and anions to be rapidly and accurately analysed. Sometimes redox titrations require the prior preparation and addition of masking agents. Selective, sensitive and low-cost methods of redox titration of copper and SO2 in plant conditions are described.

Introduction Hydrometallurgical processes involve the removal of metals from different types of ores, concentrates, and waste products by aqueous solutions containing different chemical reagents. Processes include a number of sequential steps such as ore preparation, leaching, precipitation, extraction, ion exchange, cementation and electrodeposition. Successful realization of metal recovery requires accurate analyses of feedstock composition, intermediate and auxiliary solutions, and final products. Effective analytical methods provide the opportunity to manage and improve a process. The main problems of analysis in hydrometallurgical processes are associated with the complicated compositions of solutions. Raw materials usually contain many components and can include base metals such as Fe, Co, Ni, Cu, Cr, Mn, Al, Pt and U groups of metals, and rare-earth metals. These metals as a rule are in the form of different salts or oxides. After the leaching process the solutions also contain an excess of leaching reagent. The most commonly used leaching agents are sulphuric acid (Cu, Zn), sodium carbonate (U, V, Mo, W), sodium hydroxide (W), ammonia (Cu, Ni), cyanide and thiosulphate (Au, Ag), sulphite (Sb, Hg), chlorine and chloride (platinum group metals and rare-earth metals). Selection of an appropriate method of analysis is based on criteria such as accuracy, precision, sensitivity, urgency for results, costs per analysis, number of samples to be analysed, amount of sample available, and avoidance of chemical and physical interferences1. This paper summarizes methods used to analyse hydrometallurgical solutions, with titration methods receiving particular attention. APPLICATION OF REDOX TITRATION TECHNIQUES

457

457-465_Bazhko:text

2/14/09

12:07 PM

Page 458

Physical and physical-chemical methods At present different physical and physical-chemical analytical methods are widely applied in hydrometallurgy. These include various spectral techniques based on the absorption, emission, or scattering of electromagnetic radiation. Atomic absorption spectroscopy (AAS) with flame and atomization, atomic emission spectroscopy (AES), inductively coupled plasma-mass spectroscopy (ICP-MS), and X-ray fluorescence analysis (XRF) can be considered the most sensitive detection methods for about 30 elements. Important characteristics of spectral techniques are the short time of analytical signal registration, low detection limit, and wide calibration ranges, which give the advantage of minimal sample preparation2. Along with these complicated methods, photometry and spectrophotometry are effectively and widely used by analysts. There are many selective organic and inorganic reagents that form the coloured substances that absorb light in visible or ultraviolet regions and provide the opportunity to determine almost all elements of the Periodic system.3–8 Electrochemical methods such as voltamperometry, coulometry, and potentiometry are also used in hydrometallurgy for analysis, but generally for single-component samples.9–11 Titration methods In spite of intensive development of physical-chemical methods, titration has not lost its importance for chemical analysis. This method of analysis is easy to use and fast. Titration does not require complicated and expensive devices or equipment and is based on selective reactions. It is usually used to determine medium and high concentrations of elements. Furthermore, titration gives reliable results even in field conditions. One of the titration methods used in hydrometallurgy is an oxidation-reduction (redox) titration. Redox titrimetry is used to analyse a wide range of inorganic analytes. Although many of these methods have been replaced by newer ones, a few of them continue to be listed as standard methods of analysis. Redox titrimetry is based on a redox reaction between the analyte and titrant. The oxidizing and reducing agents most commonly used in hydrometallurgy12, 13 are shown in Table I. In some cases samples require preparation before the titration. Sometimes substances analysed by titration with an oxidizing agent must be converted into a reduced state with auxiliary reducing agents. There are several ways to prereduce an analyte quantitatively. Reductors could be used: a Jones reductor is a column filled with granular Zn coated with a Zn(Hg) amalgam14, 15, and a Valden reductor contains spongy, granular Ag coated with HCl. In other cases preoxidation is necessary. For ferric reduction to ferrous, the treatment with stannous chloride, and the addition of HgCl2 to destroy any excess of SnCl2 are often used. To transform a substance into an oxidized state, auxiliary oxidizing agents are used, for example, ammonium peroxydisulfate and hydrogen peroxide. These reagents are capable of oxidizing 2– Mn2+ to MnO4 , Cr3+ to Cr2O7 , and Ce3+ to Ce4+. Any excess of auxiliary oxidizing agents is easily destroyed by a short boiling of the solution. There are several methods to find the end point in redox titrimetry. A few titrants, such as permanganate, have oxidized and reduced forms whose colours in solution are significantly different, and it is not necessary to use any indicator. Specific indicators are substances that indicate the presence of oxidized or reduced species. For example, starch forms a dark blue complex with iodide; thiocyanate forms a red complex with ferric. The most important class of redox indicators are substances that may be reversibly oxidized and reduced, and change the colour upon oxidation or reduction. Redox indicators impart a colour that depends on the solution’s electrochemical potential. A list of general redox indicators16, 17 is shown in Table II. 458


457-465_Bazhko:text

2/14/09

12:07 PM

Page 459

Table I Common oxidizing and reducing agents Agent

Redox half-reaction

Standard reduction potential (v)

Ions that can be determined

+ 1.7

As (III), Cu (I), Fe (II), V (IV), Mo (V), Ti (III), U (IV)

+1.49

Fe (II), Sb (III), Mn (II), V (IV), W (V), U (IV), Tl (I), Cr (III), H2O2, (indirect determination Cr (II), V (II), Ti (III), Nb (III), Mo (III), Cu (I), Sn (II), Ca (II), Cd (II), Zn (II), Pb (Il), (II), Ni (II))

+ 1.42

As (III), Sb (III), Fe (II)

+ 1.33

Fe (II), As (III), Cu (I), U (IV), Sn (II), Te (IV), W (III), Mo (III), 2Ti (III), V (IV), SO3 , Ag (I), Ba (II), Pb(II)

+ 0.994

Fe (II), As (III), Hg (I), T1 (I), Sn (II), Sb (III), Ti (III), V (III), Mo (III, V), 2W (V), U (IV), SO3

Oxidizing agents 2-

Cerium IV Ce4+ IV

Ce(ClO4)6 + e- = Ce3+ + 6 ClO4 -

Permanganate MnO4

-

-

Bromate BrO3

-

+

+

-

-

MnO4 + 8 H + 5 e = 2+ Mn + 4 H2O

-

BrO3 + 6 H + 6 e = Br + 3 H2O 2-

Dichromate Cr2O7

-

2-

-

Cr2O7 + 14 H + 6 e = 3+ 2 Cr + 7 H2O

+

Vanadate VO3

+

+

-

2+

3+

-

2+

+ 0.771

Cu (I), U (IV), Sn (II), U (IV)

-

-

+ 0.521

As (III), Sn (II), SO3 , Sb (III)

-

2+

+ 0.771

VO3 , Cr2O7 , MnO4 , MnO2

-

+0.521

Fe (III), Cu (II) ), Cr2O7

VO2 + 2 H + e = VO + H2O

Ferric Fe3+

Fe + e = Fe

Iodine I2

I2 + 2 e = 2 I

2-

Reducing agents 3+

Ferrous Fe2+

Fe + e = Fe

-

-

Iodide I

I2 + 2 e = 2 I

Ascorbic acid C6H606

+

-

3 C6H606 + 6 H + 6e = 3 C6H8O8

4+

Stannous Sn2+ 2+

+

2-

2-

-

-

2-

-

2-

-

3-

+ 0.185

VO3 , CrO4 , MnO4 , AsO4 , Ce (IV), Sn (IV), Ag (I), Hg (II), PbO2, MnO2 -

2-

-

2+

+ 0.15

Cu (II), Fe (III), VO3 , Cr2O7 , MnO4

-

3+

+ 0.1

Cu (II), Fe (III), Sb (V), Cr2O7 , MoO4

+ 0.09

Cu (II), Fe (III), VO3 , Sb (V), 2Cr2O7 , MnO4 , MnO2

TiO + 2 H + e = Ti + H2O

Thiosulphate S2O3

2-

-

Sn + 2 e = Sn

Titanium III Ti3+

-

2-

S4O6 + 2 e = 2 S2O3

2-

-

Another method for locating the end point of a redox titration is the potentiometric method. It is based on the use of an appropriate electrode to monitor the change in electrochemical potential as titrant is added to a solution of analyte. The end point can then be found from a visual inspection of the titration curve.1 While selecting the method for analysis it is necessary to estimate the composition of analytes and possible chemical interferences that can cause erroneous analytical results. Unfortunately, analytical methods are rarely selective toward a single species. Different separation techniques such as precipitation, extraction, or chromatography can be used to remove either the analyte or the interferent from the sample matrix. A simpler variant that analytical chemists use to reduce matrix effects to either a negligible or a minimal magnitude APPLICATION OF REDOX TITRATION TECHNIQUES

459

457-465_Bazhko:text

2/14/09

12:07 PM

Page 460

Table II Redox indicators Indicator

Transition potential (v)

Colour of Ox form

Colour of red form

N-phenylantranilic acid (2-carboxy-diphenylamine)

+ 1.08

purple

colorless

Ferroin (tris-(1,10-phenanthroline) iron (II) sulphate)

+ 1.06

cyan

red

o-Dianizidine (5,3 '-dimethoxybenzidine)

+ 0.85

red

colourless

Sodium diphenylamine sulfonate

+ 0.84

red-violet

colourless

Diphenilamine

+ 0.76

violet

colourless

Diphenylbenzidine

+ 0.76

violet

colourless

3,3'-Ddimethylnaphtidine (3,3'-(dimethyl-4,4'-diamino-1, 1 '-dinaphthyl))

+ 0.7

purple red

colourless

Variamine blue (4-amino-4 '-methoxy-diphenylamine)

+ 0.575

violet blue

colourless

Nitroferroin (tris(5-nitro-1, l0-phenanthroline) iron (II) sulphate)

+ 1.25

cyan

red

5,6-Dimhethylferroin (tris-(5, 6-dimethyl-1, I 0-phenanthroline) iron(II) sulphate)

+ 0.97

yellow-green

red

Sodium o-Cresol indophenol

+ 0.62 (pH =0) + 0.19 (pH =7)

blue

colourless

Methylene blue (Bis-3,9-dimethylamino phenazothionium chloride)

+ 0.53 (pH =0) + 0.01 (pH =7)

blue

colourless

Thionine (diaminophenathiozine)

+ 0.56 (pH =0) + 0.06 (pH =7)

violet

colourless

Indigocarmine (indigodisulfonic acid)

+ 0.29 (pH =0) - 0.13 (pH =7)

blue

colourless

Phenosafranin

+ 0.28 (pH =0) - 0.25 (pH =7)

red

colourless

Safranin T

+ 0.24 (pH =0) - 0.29 (pH =7)

red-violet

colourless

Neutral red

+ 0.24 (pH =0) - 0.33 (pH =7)

red

colourless

pH independent indicators

pH dependent indicators

is masking. Masking is technically not a separation technique because the analyte and interferent are never physically separated from each other. The masking agent binds the interferent as a soluble complex or changes the oxidation number of the interferent, preventing it from affecting determination of the analyte. A wide variety of ions and molecules are used as masking agents (Table III). Titration methods of Cu and SO2 in hydrometallurgical solutions One of the most important applications of redox titration in hydrometallurgy is a determination for copper. Various redox titration methods have been developed for copper analysis including iodometry, chromatometry, cerimetry, mercurometry, or permanganatometry.18 For analysis of copper electrowinning solutions an iodometric titration of copper is used. 460


457-465_Bazhko:text

2/14/09

12:07 PM

Page 461

Table III Masking agents Masking agent

Ions that can be masked

-

Cyanide CN

Fe (II), Fe (III), Co (II), Ni (II), Cd (II), Zn (II), Hg (II), Ag (I), Au (III), Pt (II), Pd (II)

-

Fluoride F

Al (III), Fe (III), Mg (II), Ca (II), Sr (II), Ba (II), Ti (IV), Cr (III) 3-

2+

Phosphate PO4

Fe (III), UO2 2-

Thiosulphate S2O3

Ag (I), Fe (III), Pb (II), Cu (II), Au (III), Pt (II), Pd (II), Hg (II), Cd (II)

Hydroxylamine NH2OH

Fe (III), Co (II), As (V), Sb (V)

Thiourea CS(NH2)2

Fe (III), Hg (II), Cd (II), Ag (I), Sb (III) 2-

EDTA (CH2N)2(COOH)2(COO)2 2-

Citrate (CH2)2C(OH)(COO)3 2-

Tartrate (CHOH)2(COO)2

Fe (III), Cu (II), Co (II), Ni (II), Cd (II), Zn (II), Mg (II), Mn (II) Ag (I), Fe (III), Sb (III), Cu (II), Hg (II), Cd (II) Fe (III), Al (III), Cu (II), Hg (II), Cd (II), Pb (II), Zn (II)

Ascorbic acid C6H606

Fe (III), Sn (IV)

The iodometric determination of copper is a type of displacement redox titration. Cu2+ reacts with an excess of potassium iodide in an acid medium: [1] The reaction is accompanied by the liberation of a stoichiometric amount of iodine. A large excess of potassium iodide is also necessary to minimize the losses of iodine by evaporation from the solution. Free iodine is not lost in the presence of excess KI, due to the formation of I3 ion: [2] Once the reaction is complete the iodine produced is titrated with sodium thiosulfate: [3] The end point of the iodine titration with thiosulfate is indicated by the colour change of the starch indicator. The amount of standard sodium thiosulphate solution required to titrate the liberated iodine is equivalent to the amount of copper. 2Ions that reduce iodide such as Fe3+, As (V), and Sb (V), or oxidize iodine such as Cr2O7 , As (III), and Sb (III) can exert an influence on the copper determination. Pb and Bi form coloured substances with iodide and create difficulty for titration; they must be masked or removed. Since iron forms stable fluoride complexes, Fe 3+ can be masked by sodium, ammonium or potassium fluoride. As (V) and Sb (V) react with iodide only in strong acid media. The samples that are usually analysed at Mintek originate from hydrometallurgical processes and contain not only Cu but Fe (II), Fe (III), Co (II), Ni (II), Zn, (II), Mn (II), and Cr (III). Thus, addition of fluoride only is sufficient for accurate copper determination. The procedure for analysis includes the following steps: • NaF is added for ferric masking • an excess of KI is added and the mixture is shaken. After about 10 minutes the solution becomes deep-brown and contains solids • the mixture is titrated with the standard thiosulfate solution until the slurry has a light yellow tint. At this step, 1 ml of starch indicator is added and titration must be continued until the mixture in the flask turns white. APPLICATION OF REDOX TITRATION TECHNIQUES

461

457-465_Bazhko:text

2/14/09

12:07 PM

Page 462

The use of this method for real solutions provided good results. Moreover, titrations were performed on plant samples, quickly, without additional capital inputs for equipment. Sulphur dioxide and sulphites are often used in hydrometallurgy as reducing agents for metal recoveries.19, 20 Their concentrations can be determined by oxidizing titrations, but if the analyte contains other reducing agents interferences occur. Our objective was to analyse samples that contained both sulphite and ferrous ions. In this case we determined sulphite by indirect titration. 2In the first step the amount of Fe 2+ and SO 2 (SO 3 ) can be determined by a dichromatometric method, which is based on reactions: [4] [5] The titration is performed in an acid medium (the addition of sulphuric acid). Phosphoric acid is added to mask ferric ions. Diphenylamine is used as a reversible oxidation-reduction indicator. In the second step the amount of iron (III) is determined by titration with EDTA. Iron (III) forms with EDTA stable complex at pH value of 2 or lower. Other base metals do not react with ADTA at low pH. Sulphosalicylic acid is used as indicator for titration of iron (III). And in the third step the amount of iron (II) and iron (III) is determined by titration with EDTA after oxidizing Fe2+ with hydrogen peroxide and decomposing of peroxide excess by heating. The concentration of sulphite ions is calculated by subtracting the results of the second and third steps from the result of the first titration. This method of redox titration was applied successfully for analysis of the solution after the SO2 leaching of an ore. References 1. HARVEY, D. Modern analytical chemistry, 2000. 798 pp. 2. International conference ANALYTICAL CHEMISTRY AND CHEMICAL ANALYSIS (AC&CA-05), devoted to 100 anniversary of Anatoly Babko / Book of abstracts: Kyiv Ukraine. September 12–18, 2005, 480 p. 3. MAKSIMOVA, I.M., KOROLEV, D.N., MOROSANOVA, E.I., and YU, A. Zolotov Photometric determination of metal ions in continuous-flow systems in the presence of a surfactant J. Analyt. Chem., vol. 50, no. 9, 1995. pp. 842–845. 4. SAMCHUK, A.I. and PILIPENKO, A.T. Analytical chemistry of minerals, Utrecht, The Netherlands. VNU Science Press, 1987. 5. BULATOV, M.I. and KALINKIN, P.I. Textbook for photocolorimetric and spectrophotometric methods of analysis. Leningrad, Ghemistry, 1976. 376 pp. (Russian). 6. PESHKOVA, V.M. and GROMOVA, M.I. Methods of absorption spectroscopy in analytical chemistry, 1976 (Russian). 7. BARKOVSKY, V.F. and GANOPOLSKY, V.I. Differential spectrophotometric analysis, 1969 (Russian). 8. MARCHENKO, Z. Photometrical determination of elements. M.:Mir, 1971 (Russian) 462


457-465_Bazhko:text

2/14/09

12:07 PM

Page 463

9. BARD, A.J. and FAULKNER, L.R. Electrochemical methods such as voltamperometry, coulometry, potentiometry are also used for purpose of analysis in hydrometallurgy but often for single component samples. Electrochemical Methods. Wiley: New York, 1980. 10. GARCIA-ARMADA, P., LOSADA, J., and DE VICENTE-PEREZ, S. Cation Analysis Scheme by Differential Pulse Polarography, J.Chem. Educ., 1996, vol. 73, pp. 544–547. 11. EVANS, A. Potentiometry and Ion-Selective Electrodes. Wiley: New York, 1987. 12. KOROSTELEV, P., BERKA, A., VULTERIN, J., and ZYKA, J. New redox methods in analytical chemistry. 1968, p. 137–58. (Russian) 13. LAYTINEN, G.A. and HARRIS, V.E. Chemical analysis. 1979. 14. NOROOZIFAR, M. and KHORASANI-MOTLAGH, M. The Application of the SolidPhase Jones Reagent as a Reductant in the Speciation Flow Injection Analysis of Fe(III) and Fe(II) in Real Samples. Chem. Anal., Warsaw, vol. 49, 2004. 929 pp. 15. MORAIS, C.A. and CIMINELLI, V.S.T. Recovery of europium from a rare earth chloride solution, Hydrometallurgy, vol. 49, no. 1–2, June 1998, pp. 167–177. 16. HULANSKY, A. and GLAB, S. Redox indicators. Characteristics and applications. Pure & Appl. Chem., vol. 50, 1978. p. 463–498. 17. LURIE, U.U. Handbook for analytical chemistry. Chemistry. 1971. 456 p. (Russian) 18. KOROSTELEV, P.P. Titrimetric and gravimetric analysis in metallurgy. Handbook. Moscow. 1985. 320 p. (Russian) 19. DAS, G.K., ANAND, S., DAS, R.P., MUIR, D., and SINGH, P. Sulfur dioxide—a leachant for oxidic materials in aqueous and non-aqueous media. Mineral Processing and Extractive Metallurgy Review, vol. 20, no. 4–6, 2000. 20. NAIK, P. and SUKLA, L.B. Aqueous SO2 leaching studies on Nishikhal manganese ore through factorial experiment. Hydrometallurgy, vol. 54, no. 2–3, Feb. 2000.

Volha (Olga) Bazhko Post-doctoral fellow, Mintek, South Africa 2003–2007: Analytical Chemistry Department of Chemical Faculty, Belarusian State University, Republic of Belarus, teaching staff. Research in the field of sorption and ion-exchange, synthesis and investigation of immobilized sorbents, developing of quick analytical methods, investigation of environmental behaviour of metals and other toxicants. Presently: Hydrometallurgy division, Mintek, South Africa, Post-doctoral fellow. Developing of titrimetric, spectrophorometric and atomic absorption methods for the determination of base and noble metals, uranium and other components in the hydrometallurgical streams. APPLICATION OF REDOX TITRATION TECHNIQUES

463

457-465_Bazhko:text

464

2/14/09

12:07 PM

Page 464


465-474_Masiya:text

2/14/09

12:09 PM

Page 465

MASIYA, T.T. and GUDYANGA, F.P. Investigation of granular activated carbon from peach stones for gold adsorption in acidic thiourea. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Investigation of granular activated carbon from peach stones for gold adsorption in acidic thiourea T.T. MASIYA* and F.P. GUDYANGA† *Institute of Mining Research, University of Zimbabwe, Zimbabwe †Department of Metallurgy, University of Zimbabwe, Zimbabwe

This paper reports on an investigation into the effectiveness of using granular activated carbon produced from peach stones, an agricultural waste, for the adsorption of gold from acidified thiourea solutions. In this work the activated carbon was produced by chemical activation of the peach stones with either phosphoric acid or zinc chloride. The effects of several parameters on the adsorption kinetics and equilibra were examined. Gold adsorption was found to depend significantly on: solution pH, rate of stirring, activated carbon dosage, carbon particle size, and the initial concentration of gold and thiourea. The adsorption equilibrium experimental data fitted well both the Langmuir and Freundlich isotherm models with high correlation coefficient. The adsorption capacity calculated from the Langmuir isotherm was 31.2 mg Au/g for phosphoric activated peach stones and 69.0 mg Au/g for zinc chloride activated peach stones.

Introduction The ability of activated carbon to adsorb dissolved metal species has been known for a considerable time. It has found increasing application, especially in the mining industry, as an adsorbent for the extraction of gold from leached solutions (Petersen and van Deventer, 1994; Hurter, 1986). The adsorption of gold cyanide onto activated carbon, especially of coconut shell origin, has been studied extensively and is well known (Yapu et al., 1994; Van Deventer and Van der Merwe, 1993; Marsden and House, 1992; Adams and Fleming, 1989). Activated carbon has also been used for gold recovery from non-cyanide solutions (Haque, 1989). Lignite and activated bagasse were successfully assessed for adsorption of gold from acidic thiourea solutions (Syna and Valix, 2003; Zouboulis et al., 1994). It is of both technological and economic interests to examine the potential of activated peach stones for the adsorption of gold species from acidic thiourea solutions. Peach stones are agricultural by-products that are currently of no economic value, and have a hard lignocellulosic material shell that gives them the potential to be used as raw materials for production of granular activated carbon. In Zimbabwe more than 90% of the activated carbon used by the mineral industry is imported at a very high cost (Thixton, 1998). INVESTIGATION OF GRANULAR ACTIVATED CARBON FROM PEACH STONES

465

465-474_Masiya:text

2/14/09

12:09 PM

Page 466

The peach stones used for this study were activated chemically by phosphoric acid and zinc chloride. The effect of different process conditions such as pH, particle size, carbon dosage, stirring speed, initial gold and thiourea concentration on per cent gold recovery was investigated to ascertain the mechanism for adsorption. Experimental The activated carbon particles used in this study were manufactured from peach stones by chemical activation with either phosphoric acid or zinc chloride. Except where stated otherwise, the activated carbons in the size range -1.0 +0.5 mm were used for the investigations. ZnCl2 activation 3.3 g of ZnCl2 was dissolved in 100 ml distilled water. 100 g of the raw peach stones were then soaked in this solution for 24 hours. At the end of this time, the solution was filtered and the impregnated peach stones were dried in an oven overnight at 105°C after which they were activated at 700°C for 2 hours in a muffle furnace. The sample was then removed from the furnace, allowed to cool and then washed in hot HCl solution and rinsed in distilled water until the wash water was almost neutral. H3PO4 activation The raw peach stones were impregnated with 43 wt% H3PO4 for 24 hours in the weight ratio 1:1 (peach stones: H3PO4), followed by drying in an oven overnight at 105°C. The dried sample was then activated in a muffle furnace at 300°C for 1 hour. The sample was allowed to cool and subsequently washed first in hot NaOH solution followed by a series of soaking and decanting in hot distilled water until there were no traces of phosphates (identified by adding a few drops of Pb(NO3)2 into the wash water, which turns to a white precipitate in the presence of phosphates). Equilibrium adsorption studies Equilibrium data was collected by taking 200 ml solution of known gold concentration into a series of 500 ml containers. To each container, different activated carbon dosages (0.05 to 2.0 g) were added. Prior to their use activated carbon samples were washed thoroughly with water until most of the fines were removed, and then dried at 105°C in an oven for 24 hours. The containers were sealed and bottle rolled for 72 hours until equilibrium was assumed to have been reached. The solutions were filtered and the filtrates were analysed for residual gold concentration using the Spectra Atomic Absorption Spectrometer (AAS). Batch kinetic adsorption studies The kinetic adsorption studies were carried out by stirring between 0.25 g and 1.0 g of activated carbon with 300 ml acidified aqueous gold-thiourea solution of desired initial concentration in a series of 800 ml Duran beakers on a magnetic stirrer (HeIdolph type) with an adjustable stirring speed. Temperature was maintained at room temperature during the investigations. Aliquots of 10 ml were withdrawn regularly (after 15, 30, 45, 60, 90, 120, 180 and 240 minutes) and filtered through a filter paper. Filtrates were analysed for residual gold concentration using an atomic absorption spectrometer (AAS). 466


465-474_Masiya:text

2/14/09

12:09 PM

Page 467

Results and discussion Effect of pH on gold adsorption Gold adsorption on peach granular activated carbon and commercial granular activated carbon as a function of initial pH was studied for the pH range 1 to 12, and the results are shown in Figure 1. In the pH range 1.4 to 4 the per cent gold recovery initially decreases for both activated carbons as the initial pH is increased. As solution pH is increased above pH 4.1 a sharp increase in gold recovery is observed, especially with activated peach stones. This is attributed to the fact that beyond pH 5, thiourea is known to decompose to sulphur and cyanamide through an intermediate product, formamidine disulphide. [1] Zouboulis et al. (1994) attributed this observed increase in gold recovery in the alkaline range to increased adsorption capacity by the finely divided elemental sulphur formed. Figure 2 presents the effect of initial pH value on the kinetics of gold adsorption from thiourea solutions. The gold adsorption kinetic is faster at lower initial pH than at higher initial pH values. The adsorption of metal ions depends on solution pH, which influences electrostatic binding of ions on the activated carbon to corresponding metal groups in solutions (Ahalya et al., 2005). Effect of activated carbon particle size on gold adsorption Figure 3 presents gold adsorption behavior of -0.150 mm peach activated carbon particles as a function of contact time in comparison with that of the fraction -1.0 +0.5 mm of the same activated carbon. As is expected, the smaller the size of granules, the faster the rate of adsorption of gold by the activated carbon. Smaller particle size, for the same mass of activated carbon, enhances gold adsorption (Yannopoulos, 1990). This is attributed to increase in surface area for adsorption and reduced mean pore length through which gold species travel within the activated carbon particles as particle size decreases. The ultimate carbon loading

Figure 1. Effect of initial solution pH on equilibrium gold adsorption on peach granular activated carbon (PGAC) and commercial granular activated carbon (CGAC) from thiourea solutions (initial solution concentration = 7.5 mg Au/L; solution volume = 200 ml; carbon dosage = 0.2 g)

INVESTIGATION OF GRANULAR ACTIVATED CARBON FROM PEACH STONES

467

465-474_Masiya:text

2/14/09

12:09 PM

Page 468

Figure 2. Effect of initial pH on the rate of gold adsorption from acidic thiourea solutions (initial solution volume = 300 ml, carbon dosage = 0.5g, initial gold content = 10.5 mg/L)

Figure 3. The effect of activated carbon size on rate of gold adsorption from thiourea solutions

Figure 4. Variation of gold recovery with time for different stirring speeds (initial solution volume = 300 ml; initial gold content = 5.0 mg/L; pH = 1.7; carbon dosage = 0.5 g)

468


465-474_Masiya:text

2/14/09

12:09 PM

Page 469

capacity, however, is virtually independent of particle size (Marsden and House, 1992). However, the use of fine activated carbon is not recommended, especially in carbon-in-pulp plants as it will result in higher gold losses via the carbon fines. Effect of stirring speed on kinetic adsorption of gold from thiourea solutions The effect of stirring (Figure 4) of the adsorbent/adsorbate system in acidic gold thiourea solutions was investigated for cases where agitation resulted in activated carbon being fully suspended in the gold solutions (i.e. between 200 and 400 rpm). Below 200 rpm it was observed that some of the carbon remained on the base of the beaker and at 500 rpm the magnetic follower became unstable. The observed increase in rate of adsorption of gold with stirring speed in the range 200 to 400 rpm is attributed to the improvement in contact between the gold species in solution and the active sites on the carbons, thereby promoting effective transfer of adsorbate ions to the adsorbent site. Effect of initial gold concentration on equilibrium adsorption The equilibrium removal of gold from solution decreases as the initial gold concentration of the solution is increased for both carbon dosages investigated (see Figure 5). This is because at lower concentration, the ratio of the initial moles of gold species to the available surface area is low and subsequently the fractional adsorption becomes independent of initial concentration. However, at higher concentration the available sites for adsorption become fewer compared to the moles of gold species present and hence the percentage removal of gold is dependent upon the initial gold concentration, i.e. there is increased competition among gold species for the available active sites on the carbon surface. For the same initial gold concentration there is an increase in per cent gold removal as the carbon dosage is increased from 0.25 g to 0.5 g. This is because there was an increase in number of available active adsorption sites for the same number of moles of gold species as carbon dosage was increased Figure 6 shows the variation of the actual amount of gold adsorbed per gramme of activated carbon with time. An increase in the initial gold concentration leads to an increase in the adsorption capacity of gold on activated carbon. This indicates that the initial gold concentration plays an important role in the adsorption capacity of gold on activated carbon.

Figure 5. Effect of initial gold concentration on equilibrium gold recovery for two different initial activated peach carbons concentrations (bottle rolled; solution volume = 200 ml; contact time = 5 hours; pH = 1.7)


469

465-474_Masiya:text

2/14/09

12:09 PM

Page 470

Figure 6. Variation of amount of gold adsorbed as a function of time for two different initial gold concentrations (initial solution volume = 300 ml; carbon dosage = 0.5 g; pH = 1.7)

Figure 7. Effect of initial thiourea concentration on gold recovery (bottle rolled: initial gold concentration = 4 mg /L; solution volume = 200 ml; carbon dosage = 0.25 g and 0.5 g; contact time = 5 hours; pH = 1.8)

Effect of initial thiourea concentration on gold recovery The effect of the amount of free excess thiourea concentration on equilibrium gold-thiourea complex adsorption on activated peach stones was examined and the results are presented in Figure 7. The loading capacity for gold on activated carbon was found to decrease as the amount of free excess thiourea concentration was increased. Thiourea concentration is an important parameter when considering the adsorption of gold from thiourea solutions on activated carbon. Thiourea is an organic molecule that has a high tendency to adsorb on activated carbon. When there is not enough carbon, or the number of surface active sites is not sufficient for adsorbing both gold-thiourea complexes and free thiourea, then thiourea will be loaded preferentially over gold, and gold adsorption will be suppressed (Petersen and van Deventer, 1994; Zouboulis et. al., 1994). 470


465-474_Masiya:text

2/14/09

12:09 PM

Page 471

However, when there is excess activated carbon, or the number of surface active sites exceeds the total number of molecules to be adsorbed, then free thiourea does not interfere with adsorption of gold-thiourea complexes. Deschenes and Ghali (1988) explained this decline in gold adsorbed differently. They attributed it to the fact that thiourea is less stable at high concentrations and decomposes to cyanamide and elemental sulphur. Similar observations were also reported by Amer (2002). They believed that the elemental sulphur formed blocks on some of the macropores on the carbon, resulting in gold species being blocked from reaching the micropores. Adsorption isotherm equilibria To quantify the adsorption capacity of activated peach stones for removal of gold from acidic thiourea solutions, the adsorption isotherm data was evaluated using the Langmuir and Freundlich adsorption isotherms. The basic assumption of the Langmuir adsorption process is the formation of a monolayer of adsorbate on the outer surface of the adsorbent and after that no further adsorption takes place. A linear form of the Langmuir equation is given by: [2] where Qe is the equilibrium quantity of gold adsorbed on activated carbon (mg/g), Qm the maximum monolayer adsorption, KL the langmuir equilibrium constant for the adsorption reaction (L/mg) and Ce the equilibrium gold concentration in the solution (mg/L). A linear plot of 1/Qe versus 1/Ce was employed to give the values of Qm and KL from the intercept and slope of the plot (Figure 8). The Freundlich adsorption isotherm, on the other hand, is an indicator of the extent of heterogeneity of the adsorbent surface. A linear form of the isotherm is given by:

Figure 8. Comparison of the Langmuir isotherms for ZnCl2 and H3PO4 granular activated peach carbons (GAC) and commercial activated carbons (CGAC) (bottle rolled for 72 hours; pH = 1.8; initial gold solution = 11.5 mg/L)


471

465-474_Masiya:text

2/14/09

12:09 PM

Page 472

Figure 9. Comparison of the Freundlich isotherms for ZnCl2 and H3PO4 granular activated peach carbons (GAC) and commercial activated carbons (CGAC) (bottle rolled for 72hours; pH = 1.8; initial gold solution = 11.5 mg/L)

Table I Adsorption isotherm constants for chemically activated peach and commercially activated carbons Langmuir constants KL (L/mg) H3PO4 GAC ZnCl2 GAC C.GAC

15.3 48.3 63.0

Qm (mg/g) 31.2 69.0 158.7

Isotherm

Freundlich isotherm constants

R2

KF (mg/g)

n

R2

0.9802 0.9033 0.9758

26.5 110.6 203.4

3.8 3.0 3.9

0.9067 0.9938 0.9696

[3] where KF and n are the Freundlich constants and represent the significance of adsorption capacity and intensity of adsorption, respectively. Values of KF and n are calculated from the intercept and slope of the plot log Qe versus log Ce (Figure 9), which is a straight line. The constants for the two isotherms, as calculated from the plots in Figure 8 and Figure 9, are shown in Table I, together with their correlations coefficients. Conclusions It can be concluded that the adsorption of gold (I) ions from acidic thiourea solutions by activated peach stones is dependent on several parameters, which include: solution pH, rate of stirring, carbon dosage, gold and thiourea concentration in solution and carbon particle size. Zinc chloride activated peach stones fitted well the Freundlich isotherm, with a very high correlation coefficient (R2 = 0.9938), while the phosphoric acid and commercial activated carbons produced good fits with the Langmuir isotherm, with high correlation coefficients (R2 = 0.9802 and 0.9758 respectively). 472


465-474_Masiya:text

2/14/09

12:09 PM

Page 473

References ADAMS, M.D. and FLEMING, C.A. (1989), The mechanism of adsorption of aurocyanide onto activated carbon, Metallurgical and Materials Transactions B, vol. 20B, 1989. pp. 315–325 AHALYA, N., KANAMADI, R.D., and RAMACHANDRA, T.V. Biosorption of Chromium (VI) from aqueous solutions by the husk of Bengal gram (Cicer arientinum), Environmetal Biotechnology: Electronic Journal of Biotechnology, vol. 8, no. 3, 2005. AMER, A.M. Processing of copper anode-slimes for extraction of metal values, Physicochemical Problems of Mineral Processing, vol. 36, 2002. pp. 123–134. DESCHENES, G. and GHALI, E. Leaching of gold from chalcopyrite concentrate by thiourea, Hydrometallurgy, vol. 20, no. 2, 1988. pp. 179–202. HAQUE, K.E. Gold Leaching from Refractory Ores-Literature survey, Mineral Processing and Extractive Metallurgy Review, vol. 2, no. 3, 1987. pp. 235–253. HURTER, M.F. A review of advances and established procedures in the carbon in pulp (CIP) gold recovery plant at Western areas Gold Mining Company Limited, Proceeding of Internatioanl Conference on Gold, vol. 2, Extractive Metallurgy of Gold, SAIMM, 1986. pp. 335–351. KADIRVELU, K. and NAMASIVAYAM, C. Agricultural by-products as metal adsorbents: sorption of lead (II) from aqueous solutions onto coir-pith carbon, Environmental Technology, vol. 21, no. 10, 2000. pp. 1091–1097. MARSDEN, J. and HOUSE, C.I. The chemistry of gold extraction, Ellis Horwood, London. 1992. MCKAY, G., BLAIR, H.S., and GARDENER, J.R. Adsorption of dyes on Chitin I: Equilibrium studies, Journal of Applied Polymer Science, vol. 27, no. 2, 1982. pp. 151–155. PETERSEN, F.W. and VAN DEVENTER T.S.J. Comparative performance of porous adsorbents in presence of gold cyanide, organic foulants and solid fines, Hydrometallurgy ’94, IMM and SCI, Cambridge, 1994. pp. 501–515. SYNA, N., and VALIX, M. Assessing the potential of activated bagasse as gold adsorbent for gold-thiourea, Minerals Engineering, vol. 16, no. 6, 2003. THIXTON, D.H. Carbon technology for the recovery of gold: A comprehensive guide to current carbon technology for the recovery of gold with special reference to practice in Zimbabwe, Ministry of Mines, Environment and Tourism, Publication No. 21, Dept. of Metallurgy. 1998. VAN DEVENTER, J.S.J. and VAN DER MERWE P.F. The reversibility of adsorption of gold cyanide on activated carbon, Metallurgical and Materials Transactions B, 1993. pp. 433–439. YANNOPOULUS, J.C. The extractive metallurgy of gold, Van Nostrand Reinhold, New York. 1990. YAPU, W., SEGARRA, M., FERNANDEZ M., and ESPIELL, F. Adsorption kinetics of dicyanoaurate and dicyanoargentate ions in activated carbon, Metallurgical and Materials Transactions B, 1994. pp. 185–191. ZOUBOULIS, A.I., KYDOS, K.A., and MATIS, K.A. Adsorption of gold-thiourea complex on Greek lignite, Hydrometallurgy ’94, July, 1994. pp. 546–559. INVESTIGATION OF GRANULAR ACTIVATED CARBON FROM PEACH STONES

473

465-474_Masiya:text

2/14/09

12:09 PM

Page 474

Francis Gudyanga Secretary for Science and Technology Development, Zimbabwe Professor Francis Gudyanga is currently the Permanent Secretary of the Ministry of Science and Technology Development in Zimbabwe while at the same being on the academic staff on a part-time basis in the Department of Metallurgical Engineering at the University of Zimbabwe (UZ). He obtained (1988) a PhD in Minerals Technology and a DIC in Electrochemical Engineering from the Royal School of Mines, Imperial College, after working on the electrohydrometallurgical reduction of cassiterite associated with sulphide minerals. In 1989 he joined he teaching staff in the Department of Metallurgy, UZ, carrying out research in hydrometallurgy principally in the reductive decomposition of sulphidic mineral ores. He was Deputy Dean (1991–1994) and Dean (1994–1997) of Engineering at UZ. He worked at Bindura Nickel Corporation’s refinery on the production of Ni, Cu and Co by the Outokumptu process. He spent a year’s sabbatical (1996) at Mintek, Randburg, South Africa, working on the bacterial leaching of sulphide ores. In 2000 he was appointed Deputy Director General (Technical) of the Scientific and Industrial Research and Development Centre (SIRDC) in Harare. He was Chairman of the Research Council of Zimbabwe (2000–2007). He has been on the board of the Zimbabwe Mining and Development Corporation and/or its subsidiaries since 1991. He was a Member of the Executive Board of the International Council for Science (ICSU) (2002–2008) and is a member of the ICSU Regional Committee for Africa since 2004. He serves on several other board and committees.

474


475-484_Jeffrey:text

2/14/09

12:10 PM

Page 475

JEFFREY, M., PLEYSIER, R., and BUNNEY, K. Elution behaviour of metals from carbon. Hydrometallurgy Conference 2009, The Southern African Institute of Mining and Metallurgy, 2009.

Elution behaviour of metals from carbon M. JEFFREY, R. PLEYSIER, and K. BUNNEY Parker Centre (CSIRO Minerals), Australian Minerals Research Centre, Australia

This study investigated the elution of precious and toxic metals from activated carbon. Gold and mercury are of primary interest, due to their value and environmental impact respectively. A laboratory elution rig was commissioned to simulate industrial Zadra and AARL carbon elution conditions. Mercury was found to elute most effectively in strong caustic cyanide solutions; gold was found to elute most effectively in hot low ionic strength solutions. As cyanide decomposes on contact with activated carbon at high temperatures, optimum elution conditions for gold conflicts with those for mercury —no practical compromise in conditions was found. However, sequential elution allowed reasonable recovery of both metals. Ongoing work is looking at the deportment of other metals under elution conditions.

Introduction Gold can be extracted industrially from ores by a variety of processes, and one of the most successful methods is the carbon-in-pulp (CIP) process. In CIP processing, ore is mined, ground, and leached with alkaline cyanidic solution to solubilize the gold as a gold cyanide complex. The gold cyanide is adsorbed by activated carbon, which is subsequently eluted to produce a concentrated gold solution. The eluted solution is electrowon to recover the precious metals (alternatively, zinc precipitation may be used), and the carbon is treated in a kiln under a flowing steam atmosphere to reactivate its surface for reuse. Many gold plants process ores containing traces of mercury, but the deportment of mercury through the entire plant under different operating conditions is poorly understood. It is known that generally mercury follows gold through the cyanide circuit (Zárate, 1985) and loads onto the activated carbon, but only partially elutes with the gold. This partial elution compounds the treatment difficulties, since a portion of the mercury reports to the electrowinning cells, and the remainder reports to the carbon regeneration kilns. Mercury is a heavily regulated effluent from industrial processes and, in Western Australia, is listed as a controlled waste under Schedule 1 of the WA Environmental Protection (Controlled Waste) Regulations 2004. Compounds or solutions of mercury may not be discharged into the environment under Schedule 1 of the WA Environmental Protection (Unauthorized Discharges) Regulations 2004, as it is classified as ‘waste that can potentially accumulate in the environment or living tissue’ under Part 3 of the WA Environmental Protection Regulations (1987). ELUTION BEHAVIOUR OF METALS FROM CARBON

475


2/14/09

12:10 PM

Page 476

This paper presents results from fundamental studies of the elution behaviour of mercury and gold under simulated industrial conditions. Conditions that favour mercury and gold desorption have been established, and a process for the treatment of mercury-containing CIP carbon is proposed. Method Elution rig A laboratory-scale elution rig was commissioned to simulate Zadra and AARL carbon elution conditions, see Figure 1. The majority of the system was constructed from stainless steel tubing and Swagelok fittings. The elution column was ~600 mm long with an internal diameter of 25 mm, and a capacity of ~100 g of granulated activated carbon per experiment. The void volume of the system, when loaded with carbon, was ~150 mL. Solutions were pumped through the column with a high pressure positive displacement pump with a flow rate of 9.9 mL/min unless noted otherwise. A pressure relief valve (Figure 1) was set to 1 400 kPag as a safety precaution, but was not triggered in the experiments as routine operating pressures were well below this threshold. Heat was maintained with electrical heating tape wound around the main column, controlled by a thermocouple inserted through the top of the column, connected to a PID controller. Temperature control was accurate to within ±2ºC over the normal course of an experiment. Solutions were preheated prior to entering the column by five turns of tubing, wound with the same heating tape. Metal foil loosely bound around the column and heating tape provided convective insulation for the system. Samples Samples of granulated activated carbon were received from an industrial CIP plant. The samples were either barren, loaded, or acid washed (barren was taken after elution but before regeneration, loaded was from the adsorption train, and acid washed was taken immediately before elution). The metal loadings on each sample are given in Table I. Gold loadings were measured by FAAS on aqua regia digestions of ashed carbon samples, while mercury loadings were measured by CVAAS on microwave-assisted acid digestions of as-received samples.

Figure 1. Photograph of the elution rig

476



2/14/09

12:10 PM

Page 477

Table I Metals on carbon samples Sample C1 loaded C1 acid washed C2 acid washed

Gold loading (g/ton)

Mercury loading (g/ton)

909 1 140 355

187 229 92

Due to the greater ease and precision with which higher element concentrations can be measured, the majority of experiments were performed with representative portions of C1 acid washed carbon. Conditions The standard operating procedure was to load the column with a known weight of acid washed carbon, fill the rig with deionized water, pump 150 mL of 2% w/v NaCN + 2% w/v NaOH solution into the rig, seal and soak the carbon at 130ºC for ~30 minutes, then pump 9.9 mL/min of deionized water through at 130ºC (AARL type elution). Output from the rig was collected in ~100 mL portions, weighed, and assayed for gold and mercury concentrations. Pressure inside the elution rig was maintained by restricting the flow rate from the outlet manifold with the PD piston pump able to maintain pressures in excess of 1 MPa in the system. When pumping was stopped (such as during the soak stage), the rig was sealed to maintain the high pressures required to prevent the solution boiling. Variations on this base method are noted for each run. The effects of modifying temperature, pressure, cyanide concentration and eluant composition were investigated. Results and discussion The concentration profiles shown are calculated from the metal assays of each sample and the masses of each solution. The recovery profiles shown are calculated as the normalized integrated concentrations of each metal, assuming that the density of each solution sample is equal to that of pure water. Simple density measurements on selected solution samples indicated this to be a reasonable approximation. Commissioning data During the commissioning of the column, a number of experiments were conducted at various temperatures and pressures to ascertain that the elution circuit was operating with similar characteristics to that of industry. Initially the C2 acid washed carbon was eluted at a typical elution pressure and temperature of 200 kPag and 130°C. The gold profile data is shown in Figure 2. The elution profiles (closed symbols) represent the metal assays from each sample with the recovery profiles provided by the cumulative metal values (open symbols). From the concentration profile, it was clear that during the initial stages of elution, which corresponds to the flushing of the NaCN + NaOH solution from the column, the elution of gold was poor. However, as the soak solution was displaced with distilled water, there was a large peak in the concentration profile, followed by a tail which was typical of elution processes. The recovery profile showed that within 2 200 g of eluant (~15 BV), the recovery of gold was >90%, and the profile was similar to that reported in the literature (Marsden and House, 2006). An elution experiment was also carried out at 700 kPag, and the data (Figure 2) showed that both ELUTION BEHAVIOUR OF METALS FROM CARBON

477


2/14/09

12:10 PM

Page 478

Eluant mass (g)

Figure 2. Effect of pressure on the concentration of gold in the eluant and the cumulative recovery for an AARL style elution at 130°C

the recovery and concentration profiles were similar at the two elution pressures. Such a result was not surprising, as it would be expected that pressure would not impact on the elution process, providing that it was maintained above the vapour pressure of the eluant solution. An important operational observation from the elution experiments at the two different pressures, however, was that pressure control was less reproducible at 200 kPag. The outlet valve needed to be continually manipulated to prevent large changes in pressure, and hence flow rate, which affected the stability of the column temperature. In contrast, at 700 kPag, the elution operated more smoothly, with little change required in the valve position during the course of the experiment. This variability is evident in the gold concentration profile, where the main peak is less uniform at the lower pressure. Therefore, the remainder of the experiments were conducted at the higher column pressure to take advantage of the improved pressure, flow and temperature control. The gold elution profiles at 130°C and 700 kPag for the other two carbon samples, C1 acid washed, and C1 loaded were compared to the profiles for the C2 acid washed carbon, as shown in Figure 3. From the gold recovery curves, it was evident that both the C1 and C2 carbon samples eluted in a very similar manner, despite the loading for the C2 carbon being significantly less than the C1 carbon. It was also clear that the acid washing stage had little impact on the elution of gold. For each of the carbon samples eluted at 130°C and 700 kPag, the concentration and recovery profiles for mercury were also measured, as shown in Figure 4. These profiles are significantly different from those for gold, with the concentration of mercury peaking in the second eluant batch collected, and then rapidly dropping to zero. The recovery profiles indicated that only < 60% of the mercury had been eluted from the carbon, compared to > 90% for gold. These results indicate that distilled water is not effective at eluting mercury from carbon. The recovery of mercury from both the acid washed carbon samples was similar at ~57%, but recovery from the loaded (not acid washed) carbon was significantly lower at 41%. The mechanism by which acid washing improves mercury desorption has not yet been determined. During the commissioning trials, elutions were conducted at a range of temperatures, with data obtained at 130 and 160°C, as shown in Figure 5. As expected, the elution of gold was more effective at higher temperatures, producing a sharper gold concentration profile, and 478



2/14/09

12:10 PM

Page 479

Eluant mass (g)

Figure 3. The concentration of gold in the eluant and the cumulative recovery for an AARL style elution at 130°C of three difference carbon samples

Eluant mass (g)

Figure 4. The concentration of mercury in the eluant and the cumulative recovery for an AARL style elution at 130°C of three difference carbon samples

Eluant mass (g)

Eluant mass (g)

Figure 5. Effect of temperature on the concentration of gold and mercury in the eluant and the cumulative recovery for an AARL style elution

ELUTION BEHAVIOUR OF METALS FROM CARBON

479


2/14/09

12:10 PM

Page 480

hence a more concentrated pregnant eluant at the higher temperature. The overall recovery was also superior at the higher temperature, 97%, compared to 93% at 130°C. These results are consistent with the data reported in the literature (Marsden and House, 2006). An interesting observation was that the elution of mercury was less effective at the higher temperature, with the peak concentration and recovery of mercury, lower at 160°C (40% recovery) than at 130°C (57% recovery). The data shown in Figure 4 suggest that cyanide is important for the elution of mercury, since the peak mercury concentration corresponded to the period in which the caustic cyanide soak solution was flushed from the column. The data obtained at 160°C are consistent with this hypothesis, since cyanide is known to decompose on activated carbon at elevated temperatures, resulting in a lower free cyanide concentration at higher elution temperatures. Effect of cyanide concentration in the AARL soak stage The effect of varying the cyanide concentration in the soak stage of the AARL elution on the metal elution was investigated. This work was undertaken as the commissioning data suggested that cyanide played an important role in the elution of mercury. The C1 acid washed carbon was soaked at 130ºC in 2% w/v NaOH solutions containing 0, 2, and 5% w/v NaCN prior to the standard deionized water elution. The results are shown in Figure 6. The data obtained for the gold profiles indicate that although a soak solution containing no cyanide could be utilized prior to the elution, the recovery of gold under these conditions was slow and that the recovery of gold was incomplete. Gold elution rates increased with increasing NaCN in the soak solution, with profiles approaching a similar overall recovery of gold. This data is also consistent with the literature, which shows the rate of gold elution from carbon utilizing a Zadra system at 95°C increases with increasing cyanide concentration. (Adams and Nicol, 1986). Mercury desorption was even more markedly affected by the presence of cyanide, with the peak mercury concentration very low from a caustic-only soak, resulting in an overall mercury recovery of only 6%. This data clearly indicated the importance of cyanide in the elution of mercury, and were consistent with the data in Figures 4 and 5. When 5% NaCN was adopted in the presoak, the concentration peak and overall recovery of mercury were higher, but still nowhere near complete (<65%). The reason for this is that for the AARL type elution, cyanide is used only in the presoak solution, and hence the elution of mercury ceases once this is flushed from the column with the distilled water.

Eluant mass (g)

Eluant mass (g)

Figure 6. Effect of cyanide concentration in the soak stage on the concentration of gold and mercury in the eluant and the cumulative recovery for an AARL style elution

480



2/14/09

12:10 PM

Page 481

Pressure Zadra elution with high cyanide The effectiveness of eluting metals with cyanidic eluant in place of deionized water was tested by utilizing a pressurized Zadra type elution at 130ºC with 2% w/v NaCN + 2% w/v NaOH. This was undertaken in an attempt to establish whether the Zadra type systems are more effective at eluting mercury than the AARL process. The gold and mercury profiles are shown in Figure 7. It is interesting to note that the concentration profile for gold was poorer for the Zadra system, which was expected due to the high ionic strength of the eluant. However, the elution had not reached a plateau when the experiment was ceased, and the recovery profile indicates that given more time, the ultimate recovery of gold would be comparable to the AARL elution. The concentration profiles for mercury do not tail off as rapidly for the Zadra elution as they do for the AARL elution, providing further evidence that the presence of free cyanide is necessary to elute mercury from carbon. Even though the recovery of mercury with the pressure Zadra system was higher than the AARL system (76% and 56% respectively), the barren carbon still retained significant concentrations of mercury. It should be noted that there is a small hump in the gold profile after 1 200 g of eluant; at this point, the flow was ceased, the column sealed and soaked at 130ºC for 30 minutes prior to recommencing the flow. This was primarily done to assess whether additional soaking would improve the mercury elution. It is clear from Figure 7 that the gold concentration peaks and the mercury concentration dips temporarily, most likely as a result of the increased degradation of cyanide during the 30 minute soak period. A further Zadra style elution was performed at the lower temperature of 110ºC with 2% w/v NaCN + 2% w/v NaOH solution and no soak stage. The resulting profiles for gold and mercury are shown in Figure 8, along with the standard AARL elution profiles for comparison. At 110ºC, not surprisingly, the gold elution was significantly curtailed due to the combined effect of the higher ionic strength and lower temperature. The maximum gold concentration was only 19% of the peak value obtained with the AARL elution, and the gold recovery was only 34% after 1 900 g of eluant. Mercury elution, however, was much more effective at the lower temperature with cyanide, with an overall recovery of 87% being obtained. These results indicate that at 110°C, the degradation of cyanide in the column is significantly reduced, resulting a higher cyanide concentration, and hence better mercury recovery at the lower temperature.

0

500

1000 1500 Eluant mass (g)

2000

2500

0

500

1000 1500 Eluant mass (g)

2000

2500

Figure 7. Comparison of Zadra and AARL style elution systems showing the concentration of gold and mercury in the eluant and the cumulative recovery of each metal at 130°C

ELUTION BEHAVIOUR OF METALS FROM CARBON

481


2/14/09

0

12:10 PM

500

Page 482

1000

1500

2000

2500

0

500

1000

1500

2000

2500

Eluant mass (g)

Eluant mass (g)

Figure 8. Comparison of Zadra elution at 110°C and AARL elution at 130°C, showing the concentration of gold and mercury in the eluant and the cumulative recovery of each metal

0

500

1000

1500

2000

Eluant mass (g)

2500

3000

0

500

1000

1500

2000

2500

3000

Eluant mass (g)

Figure 9. Split (two-stage) elution, with Zadra type eluant for the first 1 500 g, then a soak, followed by AARL elution at 130°C

Split elution: recovery of mercury and gold in separate eluant streams An ideal elution is rapid, complete, and produces a concentrated solution of the metal of interest. The results presented so far indicate that optimum elution of mercury occurs under Zadra type conditions with a high concentration of cyanide and low temperature. In contrast, the optimum conditions for gold elution requires hot solutions with low ionic strength solution (i.e. AARL conditions). These conditions are mutually exclusive, and no viable balance between them was found where a high recovery of gold and mercury was obtained with a single eluant in a reasonable period of time. However, a split elution could be adopted where the mercury is initially eluted with a cyanide solution, followed by gold elution with distilled water. Figure 9 shows the profiles for such an arrangement, where ~1.5 L of NaCN + NaOH solution was passed initially at 90ºC, followed by a soak stage at 130°C, and then ~1 L of deionized water. It can be seen that in this arrangement, almost complete mercury recovery was obtained within the first stage of elution. In the second stage of elution, the gold concentration profile was similar to the standard AARL profiles, and the recovery incomplete only as the experiment was truncated. The split elution produces a gold-rich/mercury-cyanide-poor stream, and a mercurycyanide-rich/gold-poor stream, and hence there is the possibility of treating the two streams separately. For example, the mercury-rich stream could be treated with sulphide to precipitate 482



2/14/09

12:10 PM

Page 483

the mercury as cinnabar (HgS, see Equation [1]) before recycle to avoid loss of gold, while the gold-rich stream can be directly electrowon. [1] Conclusions Under the conditions employed, mercury elutes only in the presence of free cyanide. Gold elutes best in low ionic strength solutions at high temperatures (after being soaked in concentrated cyanidic solution). Cyanide decomposes on hot activated carbon, so mercury and gold elution require opposing conditions for efficient desorption. A split (or two-stage) elution system may allow for the practical elution of both metals and the possibility of separate treatment for each stream. Future work Optimization of the split elution conditions to achieve economic recovery and better separation of the two metals of interest are underway. Investigation of the deportment of other trace metals under elution conditions is also in progress. References Environmental Protection Act. Environmental Protection Regulations 1987. Western Australia, www.slp.wa.gov.au. 1987. ADAMS, M.D. and NICOL, M.J. The kinetics of the elution of gold from activated carbon. Fivaz, C.E. and King, R.P. (eds.) Gold 100. 2nd ed. Johannesburg, South Africa, The South African Institute of Mining and Metallurgy. 1986. MARSDEN, J.O. and HOUSE, C.I. The Chemistry of Gold Extraction, Littleton, Colorado, Society for Mining, metallurgy, and Exploration, Inc. 2006. ZÁRATE, G. Copper and Mercury Behaviour In Cyanidation of Gold and Silver Minerals. Oliver, A.J. (ed.) 15th Annual Hydrometallurgical Meeting. Vancouver, Canada, CIM Metallurgical Society. 1985.

Matthew Jeffrey Research Program Leader, CSIRO, Australia Matthew Jeffrey was awarded a Bachelor of Engineering (Chemical) with first class honours from the University of New South Wales in 1994, and a Doctor of Philosophy, from Curtin University (AJ Parker CRC for Hydrometallurgy) in 1998. Matthew has extensive research experience in hydrometallurgy, being based at Monash University for 8 years as a senior lecturer prior to joining CSIRO Minerals. He has published over 70 papers in the field, and holds a patent. Matthew is currently the CSIRO Research Program Leader for Gold Hydrometallurgy, and the Parker Centre gold market leader. Matthew’s expertise and research interests include: • Processing of gold using cyanide or thiosulfate • Leaching and electrochemistry of gold and base metals • Electrowinning • Resin and carbon adsorption and elution • Modelling and solution speciation • Ion chromatography (for determining sulphur and cyanide speciation). ELUTION BEHAVIOUR OF METALS FROM CARBON

483


484

2/14/09

12:10 PM

Page 484


Hydrometallurgy_2009.pdf

Recommend Documents