Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The enhancing effect of pyrite on the kinetics of ferrous iron oxidation by dissolved oxygen Littlejohn, Patrick Oliver Leahy 2008

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2009_spring_littlejohn_patrick_oliver_leahy.pdf [ 5.2MB ]
Metadata
JSON: 24-1.0066862.json
JSON-LD: 24-1.0066862-ld.json
RDF/XML (Pretty): 24-1.0066862-rdf.xml
RDF/JSON: 24-1.0066862-rdf.json
Turtle: 24-1.0066862-turtle.txt
N-Triples: 24-1.0066862-rdf-ntriples.txt
Original Record: 24-1.0066862-source.json
Full Text
24-1.0066862-fulltext.txt
Citation
24-1.0066862.ris

Full Text

THE ENHANCING EFFECT OF PYRITE ON THE KINETICS OF FERROUS IRON OXIDATION BY DISSOLVED OXYGEN  by PATRICK OLIVER LEAHY LITTLEJOHN B.A.Sc., The University of British Columbia, 2006  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTERS OF APPLIED SCIENCE  in  THE FACULTY OF GRADUATE STUDIES (Materials Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  December 2008  (C) Patrick Oliver Leahy Littlejohn, 2008  Abstract The oxidation of ferrous in acidic sulfate media by dissolved oxygen is an important reaction in any sulfide mineral leach process that uses ferric as a surrogate oxidant. Ferric is reduced as it oxidizes metal sulfides, and the resulting ferrous is re-oxidized by dissolved oxygen. The oxidation of ferrous to ferric by dissolved oxygen is quite slow outside of elevated pressure-temperature autoclaves. However, pyrite in solution has been found to have a catalytic effect on the reaction, speeding it up significantly. This effect is particularly significant in the context of the Galvanox™ acidic sulphate atmospheric leach process. To quantify the kinetics of this reaction and the effect of pyrite, tests were run in an atmospheric batch reactor with constant tracking of pH and redox potential. The kinetics of this reaction were quantified with respect to primary variables such as acidity, pyrite pulp density, temperature, and total iron concentration. Secondary factors such as copper concentration, gas liquid mixing rate and the source of pyrite mineral were also considered. Redox potential is a logarithmic function of the ratio of the activity of free ferric to free ferrous and is complicated by speciation within the Fe(III)-Fe(II)-H2SO4-H2O system. Correlating redox potential data with extent of reaction was achieved by using permanganate redox titration and the isokinetic technique [1] to link redox potential data directly to the fraction of ferrous reacted. This technique is effective over the potential range of interest – 360 to 510 mV vs Ag/AgCl. Under these conditions the leaching rate of pyrite is appreciable, so the rate of pyrite dissolution was predicted with the shrinking sphere model developed by Bouffard et al. [1]. Ferrous oxidation in solution was simulated with an adjusted version of the model of Dreisinger and Peters [2], which also accounts for the catalytic effect of dissolved copper. Oxygen  ii  solubility was predicted using the model of Tromans [3]. Experimental data show a clear enhancing effect of pyrite on ferrous oxidation. A mathematical model of this effect applicable to the conditions of Galvanox™ leaching is presented.  iii  Table of Contents Abstract ............................................................................................................................... ii Table of Contents............................................................................................................... iv List of Tables ..................................................................................................................... vi List of Figures ................................................................................................................... vii Acknowledgements.......................................................................................................... viii 1 Background ...................................................................................................................... 1 1.1 Surrogate oxidants .................................................................................................... 2 1.2 Redox potential ......................................................................................................... 4 1.3 Industrial use of the ferric-ferrous couple................................................................. 6 1.4 Galvanox™ ............................................................................................................... 8 2 Literature Review........................................................................................................... 10 2.1 Ferrous oxidation in solution with dissolved oxygen ............................................. 10 2.2 Pyrite leaching ........................................................................................................ 14 2.2.1 Reaction stoichiometry .................................................................................... 20 2.2.2 Pyrite sample source ........................................................................................ 23 3 Introduction.................................................................................................................... 25 3.1 Kinetic tests............................................................................................................. 26 3.2 ORP titrations.......................................................................................................... 28 4 Experimental .................................................................................................................. 29 4.1 Equipment ............................................................................................................... 29 4.1.1 Kinetic tests...................................................................................................... 29 4.1.2 ORP titrations................................................................................................... 30 4.2 Experimental procedure .......................................................................................... 31 4.2.1 Kinetic tests...................................................................................................... 31 4.2.2 ORP titrations................................................................................................... 32 4.3 Pyrite characterization ............................................................................................ 33 4.3.1 Multi-acid digestion and ICPMS - elemental assay......................................... 34 4.3.2 XRD Rietveld analysis - mineralogy ............................................................... 35 4.3.3 Nitrogen sorption testing - surface area ........................................................... 36 4.3.4 Particle size distribution................................................................................... 37 5 Results............................................................................................................................ 41 5.1 Data smoothing ....................................................................................................... 41 5.2 ORP titrations with permanganate .......................................................................... 41 5.3 Ferrous oxidation without pyrite............................................................................. 45 5.4 Ferrous oxidation with pyrite.................................................................................. 50 6 Analysis and Discussion ................................................................................................ 57 6.1 Ferrous oxidation in solution .................................................................................. 57 6.2 Factors affecting oxygen solubility......................................................................... 62 6.3 Pyrite leaching ........................................................................................................ 66 6.4 Catalysis of ferrous oxidation by pyrite.................................................................. 70 6.5 Initial potential increase and pyrite weathering ...................................................... 79 6.6 Model formulation .................................................................................................. 81 6.7 Model algorithm...................................................................................................... 84 6.8 Model results........................................................................................................... 85 iv  7 Conclusions.................................................................................................................... 88 References......................................................................................................................... 91 Appendix A - Aqueous speciation modeling .................................................................... 94 A.1 Equilibrium balancing............................................................................................ 94 A.2 Gibbs free energy minimization............................................................................. 98  v  List of Tables Table 1 - Proposed models for ferrous oxidation kinetics ................................................ 11 Table 2 - Summary of solution chemistry effects on pyrite oxidation kinetics ................ 19 Table 3 - Pyrite elemental assay by ore type .................................................................... 34 Table 4 - XRD analysis of pyrite ore samples .................................................................. 35 Table 5 - Surface area per unit weight of pyrite ores........................................................ 36 Table 6 - Particle size distribution data for pyrite samples............................................... 37 Table 7 - Tromans’ oxygen solubility parameters - sulphate ions.................................... 63 Table 8 - Legendre root points and corresponding particle sizes ..................................... 70  vi  List of Figures Figure 1 - Schematic of oxygen concentration across phase boundaries............................ 2 Figure 2 - Oxidation rate of sulphide minerals as a function of ferric/ferrous ratio ........... 5 Figure 3 - Chalcopyrite passivation as a function of solution potential ............................. 6 Figure 4 - Sulphate yield versus solution potential........................................................... 22 Figure 5 - Experimental setup........................................................................................... 30 Figure 6 - Particle Size Distribution of Huanzala pyrite................................................... 38 Figure 7 - Particle Size Distribution of Zacatecas pyrite.................................................. 39 Figure 8 - Particle Size Distribution of El Potosi pyrite ................................................... 40 Figure 9 - Data analysis and curve fitting for conversion curves ..................................... 42 Figure 10 - ORP titrations with permanganate at 80°C .................................................... 43 Figure 11 - ORP titrations with permanganate at 60°C .................................................... 44 Figure 12 - Ferrous oxidation in solution, 60 g/L H2SO4 ................................................. 46 Figure 13 - Ferrous oxidation in solution, 15 g/L H2SO4 ................................................. 47 Figure 14 - Ferrous oxidation in solution, 60 g/L H2SO4, 8 g/L Cu ................................. 48 Figure 15 - Ferrous oxidation without pyrite as a function of temperature ...................... 49 Figure 16 - Ferrous oxidation with pyrite, varying pyrite pulp density............................ 52 Figure 17 - Ferrous oxidation with pyrite, varying ore type............................................. 53 Figure 18 - Ferrous oxidation with pyrite, varying acidity............................................... 54 Figure 19 - Ferrous oxidation with pyrite, effect of impeller speed ................................. 55 Figure 20 - Ferrous oxidation with pyrite as a function of temperature ........................... 56 Figure 21 - Cu-H2O Pourbaix diagram ............................................................................. 60 Figure 22 - Oxygen solubility as a function of temperature ............................................. 65 Figure 23 - Huanzala cumulative particle size distribution .............................................. 68 Figure 24 - Model of ferrous oxidation on pyrite, pyrite pulp density ............................. 75 Figure 25 - Model of ferrous oxidation on pyrite, pyrite ore source ................................ 76 Figure 26 - Model of ferrous oxidation on pyrite, copper ................................................ 77 Figure 27 - Model of ferrous oxidation on pyrite, temperature ........................................ 78 Figure 28 - pH and Eh increase after pyrite addition........................................................ 80 Figure 29 - Relative effect of pyrite pulp density on ferrous oxidation kinetics .............. 86 Figure 30 - Relative effect of copper concentration on ferrous oxidation kinetics .......... 87  vii  Acknowledgements Scientific research is achieved by standing on the shoulders of giants. This thesis would not have been written without the support and help of many individuals. Foremost is my supervisor, Dr. David Dixon, who has been a constant guide both through my undergraduate as well as graduate studies. His strongly pragmatic sense of engineering and thorough knowledge of the subject matter has encouraged me to dig deep, while his vivacious, no holds barred sense of humour has made him a pleasure to work with.  The UBC Hydrometallurgy Group has been a world class group to study with. Dr. David Dreisinger has been an excellent teacher for the many courses I’ve taken from him. Berend Wassink’s encyclopediac knowledge of analytical techniques and the nuts and bolts of chemistry has been invaluable. Berny Rivera-Vasquez and Gonzalo Viramontes have been patient and thorough in answering my many questions on electrochemistry. Kodjo Afewu, Jaime Trejo, Maziar Eghbalnia, and Adam Fischmann have been great colleagues in the lab over the years.  The UBC Materials Engineering has been an enjoyable and educational home for the past 7 years and the Chemical Engineering and Mining Engineering departments have helped me see the larger picture of the resource industry.  Finally, thanks to my family for the constant love and support along the way.  viii  1 Background Many hydrometallurgical leach operations are electrochemical in nature, involving separate oxidation and reduction reactions on the mineral surface. The most common oxidant is oxygen. However, it is difficult to run an atmospheric reactor solely with oxygen for several reasons. First, dissolved oxygen tends to be intrinsically slow to react with mineral surfaces directly. This is exhibited by the slow reaction rate of chalcopyrite or pyrite or other sulphide minerals directly with oxygen. Second, as shown in Figure 1, two phase boundaries separate gaseous oxygen from the surface of the mineral – one between the gas bubble and the bulk solution and another between the bulk solution and the mineral surface.  1  Figure 1 - Schematic of oxygen concentration across phase boundaries These boundaries increase resistance to mass transfer and further depress the kinetics of dissolved oxygen reaction with mineral surfaces. Third, the solubility of oxygen in leach liquors tends to be quite low, due to both the effects of elevated temperature and salting out in solutions of high ionic strength. As a result, an oxygen-driven oxidation reactor has very little buffering capacity. Such a reactor is more likely to be limited by oxygen mass transfer than by the intrinsic chemical kinetics.  1.1 Surrogate oxidants The alternative to using dissolved oxygen to oxidize minerals directly is to use a surrogate oxidant or redox catalyst. A surrogate oxidant is an oxidizing reagent that can  2  be easily regenerated after it has been reduced. Iron is the most important and industrially relevant surrogate oxidant, as it is used in zinc pressure leaching, chloride chalcopyrite leaching, heap bioleaching, and sulfate atmospheric leaching, including the Galvanox™ process. Iron has two stable aqueous states, ferrous or Fe(II) and ferric or Fe(III). It is cheap, effective, and often already present in hydrometallurgical leach streams. Some sulphide minerals contain iron (ie. chalcopyrite, pyrite, or bornite). In such cases, even if the leach begins by using only oxygen, iron will progressively act as a surrogate oxidant. Ferric is reduced to ferrous on mineral surfaces thus:  Fe3+ + e– → Fe2+  (1)  Specifically in the case of chalcopyrite leaching, ferric reduction occurs as follows: CuFeS2 + 4 Fe3+ → Cu2+ + 5 Fe2+ + 2 S0  (2)  The ferrous is re-oxidized by dissolved oxygen as follows:  4 Fe2+ + O2 + 4 H+ → 4 Fe3+ + 2 H2O  (3)  This gives the following overall reaction for ferric leaching of chalcopyrite:  CuFeS2 + O2 + 4 H+ → Cu2+ + 4 Fe2+ + 2 S0 + 2 H2O  (4)  3  It is clear that the role of the surrogate oxidant is essentially to act as a redox catalyst, since the surrogate oxidant is not consumed in the overall reaction.  1.2 Redox potential In electrochemical leaching the redox potential (Eh) plays an important role. In a system with dissolved iron, Eh is determined by the ratio of the activities of free ferric to free ferrous, as given by the Nernst equation:  Eh = E O (T ) +  RT  aFe 3+ ln nF  aFe 2+       (5)  It is important to note that the Nernst equation uses the activity ratio rather than the concentration ratio to determine redox potential [4]. As a result, the speciation of the system is important in determining the redox potential, as it determines the relative amounts of free and total (free and complexed) ferric and ferrous in solution. The temperature-dependent portion of the Eh term is adequately described by Yeager and Salkind [4]:  E O (T ) = 0.771 VSHE at 25°C  (6)  dE O = 1.19 mV/K dT  (7)  4  In the context of the Galvanox™ system, the redox potential plays two important roles. The first is determining the leach rate of the minerals in the system. In general, the leach rate of a given mineral increases proportionally with the redox potential.  Figure 2 - Oxidation rate of sulphide minerals as a function of ferric/ferrous ratio However, if the mineral has a passive region on its polarization curve, the redox potential must be chosen carefully to avoid passivation. Minerals that do passivate will leach much more slowly if the redox potential is within the passive region of their polarization curves. Chalcopyrite is notorious for passivation, as can be seen in Figure 2 [6] and Figure 3. As a consequence, in an industrial chalcopyrite leach reactor the redox potential must be carefully controlled. If the redox potential is either too high or too low the leach rate will decrease. For Galvanox™ processing of chalcopyrite the operational redox potential range is roughly 460 to 480 mV vs Ag/AgCl [7] (N.B.: all potentials in this paper are given relative to the standard saturated Ag/AgCl electrode unless otherwise noted). Typically redox potential is controlled indirectly via the rate of oxygen or air  5  sparging and the sulfide mineral pulp density. When the redox potential is too high, decreasing the rate of sparging reduces the oxidant supply, and increasing pulp density increases the oxidant demand. Both techniques will reduce the solution potential.  Figure 3 - Chalcopyrite passivation as a function of solution potential  1.3 Industrial use of the ferric-ferrous couple The use of iron as a surrogate oxidant is common in industrial hydrometallurgical reactors. It is used for a variety of mineral concentrates, including primary and secondary copper sulfides, nickel and zinc. Iron is stable under a wide range of temperatures and pressures, making it usable in both autoclaves (Sepon Copper autoclave, pressure oxidation of zinc sulphides) as well as atmospheric leach tanks (Galvanox™, CESL, the  6  Mt. Gordon process) [8]. Iron is stable in both sulfate and chloride media below pH 4. Above pH 4 it is no longer appropriate due to the precipitation of ferric hydroxide. Jarosite formation can also be problematic if there is not enough acid to keep the iron in solution [9].  Unfortunately, the intrinsic kinetics of ferrous oxidation in acid solutions are quite slow. The limitation of reaction rate is reached before mass transfer considerations become relevant, so increasing the impeller speed or the rate of gas sparging is ineffective at increasing the ferrous oxidation rate. Industry has developed several strategies to work around this problem. The most obvious solution is the use of an autoclave. By increasing the temperature over the atmospheric boiling point of water and pressurizing the autoclave with high-purity oxygen gas, the kinetics of ferrous oxidation can be greatly increased. According to Henry’s law, the solubility of a sparingly soluble gas increases in proportion to its partial pressure. The downside is that autoclaves have a high capital cost and are therefore not economical for treating anything but a high grade concentrate. In addition, all sulfur in sulfide ores will be oxidized to sulfate rather than elemental sulfur, representing a waste of oxidizing and neutralizing agents unless the resulting sulfuric acid is required elsewhere.  Another strategy which has found success, primarily in heap leaching, is to use specific strains of bacteria to catalyze the rate of ferrous oxidation. There are many strains of Acidithiobacillus ferrooxidans and other species which have been adapted for this purpose on a variety of mineral ores [10]. However, with the use of bacteria comes the  7  necessity for tight control of solution chemistry, as fluctuating values of pH and redox potential can negatively impact the survival of the chosen bacteria. High ionic strength solutions are also inappropriate for bacteria. As a result, this process is more suited to heap leaching where solutions are typically more dilute. Some researchers have proposed using a small auxiliary oxidation circuit populated with A. ferrooxidans bacteria to regenerate the oxidant needed for an atmospheric leach reactor, but there has been no commercialization of this process to date [26].  Changes to plant configuration can be made to increase ferrous oxidation as well. Blowing higher purity oxygen can make a significant difference. Enriched vacuum pressure swing absorption (VPSA) oxygen, with purity in the 85-95% range is a significant improvement. Cryogenic oxygen (99% purity) is ideal but is less cost effective than VPSA oxygen unless the rate of oxygen demand is large. A less elegant solution is to increase the slurry residence time in the leach reactors, but this represents an inefficient use of equipment. Poor gas liquid mixing may be combated by increasing the power to the mixers.  1.4 Galvanox™ Galvanox™ is a chalcopyrite leach technology currently under development which uses the galvanic interaction between pyrite and chalcopyrite particles in physical contact to leach chalcopyrite concentrates [11]. Physical contact between the pyrite and chalcopyrite allows the two minerals to act as a single electrochemical couple. Chalcopyrite is the electron-donating anode which dissolves, leaving behind a porous  8  layer of elemental sulfur. Pyrite is the electron-accepting cathode which reduces ferric to ferrous. During this process the pyrite remains mostly unleached, acting as a redox catalyst for copper dissolution.  The role of pyrite in Galvanox™ is twofold. First, by taking the role of the cathode, pyrite prevents chalcopyrite passivation, thereby increasing the kinetics of atmospheric chalcopyrite leaching to industrially relevant levels. Second, pyrite catalyzes the oxidation of ferrous to ferric. Depending on iron concentration, acidity, and pyrite pulp density, the rate of ferrous oxidation in the presence of pyrite is increased by a significant factor. This rate increase renders it possible to maintain effective chalcopyrite leach kinetics with higher pulp densities and lower oxygen sparge rates.  9  2 Literature Review 2.1 Ferrous oxidation in solution with dissolved oxygen The kinetics of ferrous oxidation have been studied by many researchers under a variety of conditions. As the ferric-ferrous couple is ubiquitous in hydrometallurgical systems the kinetics of the reaction have been looked at in both atmospheric and autoclave conditions under a range of temperatures and in various aqueous media, including sulphate, chloride, phosphate, and perchlorate. Dreisinger did a literature survey in 1989 summarizing the various proposed models [2]. Several of the proposed models for sulphate media are presented in Table 1.  10  Table 1 – Proposed Models for Ferrous Oxidation Kinetics [2] Medium Sulphate  Temperature (K) 298–358  Perchlorate  298-313  Sulphate  343–363  SulphatePerchlorate  343-363  PhosphatePerchloratePyrophos.  291-303  Sulphate  413  Chloride  303-338  Chloride  291-308  PhosphatePerchlorate  303  Sulphate  373–403  Sulphate  293–353  Sulphate  293–323  Sulphate  313–408  Sulphate  373–403  Sulphate  333–403  Sulphate  393–428  Rate Expression  Reference  d [Fe( II )] k [Fe( II )] PO2 = dt [ H + ]0.36 d [Fe( II )] 2 = k [Fe( II )] PO2 dt 2 d[Fe( II )] 2 2 = k1[Fe(II )] PO2 + k2 [Fe( II )] SO42− PO2 dt 2 [ d Fe( II )] 2 2 = k1[Fe(II )] PO2 + k2 [Fe( II )] SO42− PO2 dt 2 d [Fe( II )] = k1 [Fe( II )]PO2 H 2 PO 4 dt + k 2 [Fe( II )]PO2 H 2 P2 O 72 − 2  [ [  [  ]  Iwai et al. 1982  [  ]  Awakura et al. 1986  ]  [  2  King and Davidson 1957  ]  d [Fe( II )] 2 = k [Fe( II )] PO2 dt d [Fe( II )] = ka HC 1,t [Fe( II )]PO2 dt d [Fe( II )] = k [Fe( II )]PO2 dt 2 d [Fe( II )] = k [Fe( II )]PO2 H 2 PO42− dt d [Fe( II )] 2 0.25 = k [Fe( II )] PO2 [Cu( II )] dt 1.84 1.01 d [Fe( II )] k [Fe( II )] PO2 = dt [ H + ]0.25  d [Fe( II )] k [Fe( II )] PO2 = dt [ H + ]0.35 d [Fe( II )] 2 = k [Fe( II )] PO2 dt  Verbaan and Crundwell 1986 George 1954  ]  Plasket and Dunn 1986 Awakura et al. 1986 Posner 1953 Cher and Davidson 1955 McKay and Halpern 1958 Mathews and Robins 1972 Keenan 1969  1.04  Chmielewski and Charewicz 1984 Cornelius and Woodcock 1958 Ronnholm 1999  d [Fe( II )] 2 = k [Fe( II )] PO2 dt d [Fe( II )] k1 [Fe( II )] [O2 ] = dt 1 + k 2 [Fe( II )] 2  Dreisinger and 2 d [Fe( II )] = ( k1 Fe 2+ PO2 + k 2 Fe 2+ [FeSO4 ]PO2 Peters 1989 dt 2 0.5 + k 3 [FeSO4 ] PO2 ) 1 + k 4 [CuSO4 ]  [  ]  (  [  ]  )  11  The majority of the suggested models have several commonalities: the ferrous term is second order, the oxygen term is first order, acid has a negative effect, and copper has a positive effect.  Ferrous concentration. In sulphate systems the literature agrees that the effect of ferrous concentration is second order. The only dissent on this point is Mathews, reporting a 1.84-order effect. In chloride, phosphate, and perchlorate systems the effect of ferrous concentration is reported by Awakura et al., Posner, Cher and Davidson, and King and Davidson as first order. George reports a second-order effect in perchlorate media.  Oxygen pressure. In most of the models outlined above, oxygen partial pressure is the relevant variable relating to O2. All of the literature from the last 25 years agrees that the effect of oxygen is first order. Keenan and Mathews suggested a slightly higher order effect, at 1.04 and 1.01 order respectively. This is a workable approximation, but using oxygen pressure as the variable excludes consideration of gas liquid mixing effects or changing oxygen solubility as a function of solution ionic strength. It has been found that the intrinsic kinetics of the ferrous oxidation reaction are such that gas liquid mixing is not rate determining, but this may not be the case in industrial reactors.  Acidity. In several models the negative effect of sulphuric acid is reported. Since protons are a reactant in the ferrous oxidation reaction, one might assume that their presence would speed the kinetics. However, in sulphate media protons have a strong affinity for forming bisulphate (HSO4–) with sulphate (SO42–) ions. The proton’s affinity for forming  12  bisulphate is stronger than free ferrous’ affinity for complexation with sulphate. When sulphuric acid (H2SO4) is added to solution, one proton stays with the sulphate to form bisulphate, and the other acts to rob sulphate from any other weakly complexed cations in solution such as Fe(II), Fe(III), and Cu(II) [11–12]. Complexed ferrous has been found to oxidize faster than free ferrous due to the lower electro-repulsive force acting on the ions as they come together to react. As a result, the addition of acid lowers the complexation of ferrous ions even though more sulphate is present. The net effect of these factors is to decrease the reaction speed. This effect is accounted for in the model of Dreisinger and Peters [2].  Copper. It has been recognized in several studies that copper has a positive effect on the kinetics of ferrous oxidation. This effect has been widely reported in literature by Dreisinger and Peters [2], Ronnholm et al. [12], and others. The mechanism for this effect is not entirely clear, but Ronnholm et al. and Chmielewski and Charewicz [13] suggest that the presence of copper in solution changes the reaction path and decreases the activation energy. The reaction path as proposed by Huffman and Davidson [14] is as follows: Cu2+ + Fe2+ ↔ Cu+ + Fe3+  (8)  Cu+ + O2 + H+ ↔ Cu2+ + HO2*  (9)  Huffman and Davidson suggest that the oxidation of cuprous to cupric results in the formation of the perhydroxyl radical. Quantification of this effect ranges from copper concentration to the 0.25 power in the case of McKay and Halpern to the 0.5 power  13  reported in Dreisinger and Peter’s model. Depending on how the copper was added to solution, this enhancing effect could also be influenced by the additional sulphate coming in as CuSO4.  Gas liquid mixing. In any kinetic test involving oxygen there is the question of whether the kinetics as measured reflect the intrinsic rate of the reaction or the rate of gas liquid mixing. In previous work it has been found that, with pure O2 gas and sufficient solution agitation, the rate is limited by the intrinsic kinetics of ferrous oxidation rather than gas liquid mixing [12, 15].  2.2 Pyrite leaching Pyrite is the most common sulphide mineral. It is present in many mineral deposits, from primary copper sulphides to gold bearing ore bodies. It is sought after in gold and cobalt processes, as those elements are often associated with pyrite and related iron sulphides. Pyrite is also used for sulphuric acid generation in autoclave operations. However, in most base metal operations pyrite is treated as a gangue mineral to be rejected during the concentration process.  As pyrite is a rather ubiquitous mineral and is present in so many mineral bodies and concentrates, many studies of pyrite leaching have been undertaken. Pyrite oxidation can occur in several contexts such as acid mine drainage, where exposed sulphide minerals are oxidized through bacterial contact in an acidic environment, and bacteria catalyzed heap bioleaching, where sulphur oxidation by the bacteria provides the thermal energy  14  needed to keep the heap warm in addition to any acid required. Pyrite leaching can also occur abiotically through direct oxidation by oxygen or oxidants such as ferric or permanganate. In the case of bacterial oxidation the rate of reaction is 10 to 50 times that of purely chemical oxidation under similar circumstances [16].  Pyrite oxidation has great importance in industrial autoclaves for the purpose of heat and acid generation. Long and Dixon surveyed a number of autoclave oxidation models in sulphuric acid systems in their 2004 paper [17]. On the basis of several studies they found that the oxygen dependence ranged from first-order at lower oxygen partial pressures (<20 atmospheres) to half-order at higher oxygen partial pressures. This is corroborated by the many atmospheric models which show a first-order relation to oxygen partial pressure.  Many hydrometallurgical plants treat sulphide mineral concentrates in an acidic sulphate environment where conditions are such that pyrite oxidation is thermodynamically favourable. Whether the reactor is a relatively low-temperature atmospheric tank or a high-temperature high-pressure autoclave, pyrite is thermodynamically favoured to oxidize. Moses et al. [18] found that in both aerobic and anaerobic systems the oxidation reaction proceeds with dissolved ferric acting as the oxidant.  Pyrite is similar to other minerals in that its direct oxidation by oxygen is quite slow. However, one of the products of this reaction is ferrous, which can be oxidized to ferric, at which point it proceeds to act as a surrogate oxidant for the remaining pyrite. In this  15  way, direct oxidation of pyrite by oxygen becomes oxidation via iron acting as a surrogate.  The effect of other elements and solutes on the rate of pyrite leaching has been examined by other researchers. Small amounts of copper have minimal effect on pyrite pressure leaching, while amounts of up to 1 M nickel sulphate, calcium sulphate and zinc sulphate decrease the rate by up to 30% [19]. The atmospheric leaching of coal pyrite showed that of a wide range of cations and anions, the only notable effect was depression of the rate by phosphate at concentrations greater than 200 mg/L. In the system studied in this work the only ions with a significant effect on pyrite leach rates are ferrous and ferric.  Table 2 [1] summarizes the work of various researchers to model the effects of solution chemistry on pyrite leaching kinetics. These models describe leaching in both chloride and sulphate media and encompass a temperature range from 293–373 K. This review will cover all models, but the models most relevant to this research relate to work done in higher temperature sulphate.  Ferrous and ferric concentrations. McKibben and Barnes, Wiersma and Rimstidt and Boogerd et al. report that the leach rate of pyrite is a function of ferric alone. In these models the experiments were conducted beginning with iron purely in the ferric form. However, as pyrite is oxidized, ferrous is released into solution. Even if the solution begins with pure ferric, the ratio of ferric to ferrous (and by extension, the solution potential), drops rapidly as the reaction proceeds. The models that explain pyrite kinetics  16  purely through ferric concentration are in the minority. All other surveyed researchers report the rates to be dependent on both ferrous and ferric concentrations. These dependencies vary from functions involving the ratio of ferric to ferrous, as proposed by Williamson and Rimstidt and Bouffard et al., to the more complex functions of Kawakami et al. and Nayak et al. It is important to note that the important variable relating to ferrous and ferric concentrations is the ratio of one to the other. The total concentration of iron does not significantly affect pyrite oxidation kinetics. Increasing the absolute value of ferric concentration does not improve kinetics unless the ferric-toferrous ratio is increased as well.  Acidity. Fowler et al. reported that the pyrite oxidation rate is −0.5 order with respect to H+ concentration near room temperature, in sulphate media and in the absence of bacteria. McKibben and Barnes report the same effect in chloride media. Williamson and Rimstidt report a −0.32 order effect in sulphate media. Mathews and Robinson present a model with a −0.44 order at higher temperature in both sulphate and chloride media. Other researchers present the effect of solution chemistry on the rate of pyrite oxidation as strictly a function of ferrous and ferric concentrations.  Copper. Long and Dixon [17] reported that under autoclave conditions at high temperature copper has a catalytic effect on pyrite oxidation. At low pyrite pulp density 2 g/L copper increased the oxidation rate by 14% and at high pyrite pulp density the enhancement was 50%. This effect was theorized to be the result of the near instantaneous oxidation of Cu(I) by dissolved oxygen [20], unrelated to galvanic  17  interactions of copper on the mineral surface. As a result of the fast oxidation of Cu(I), a larger proportion of ferrous iron can be oxidized in the homogeneous bulk solution rather than at the mineral surface, reducing the number of electron-transfer steps at the anode and thus increasing the solution potential at the pyrite surface.  Mineral topography. Using scanning electron microscopy, McKibben and Barnes [21] found that pyrite oxidation does not occur uniformly on the mineral surface. Instead the reaction is much faster at grain boundaries, defects in the crystal structure, around inclusions, and near fracture or cleavage sites. As a result, the reactive surface area differs from the total surface area. A related effect is that of ultra fine particles agglomerating on the mineral surface. This is a particular problem when using freshly ground material. Kawakami et al. [22] built on this idea and proposed a model that accounts for high reactive sites and low reactive sites through adsorption and desorption expressions. Rinsing the mineral sample and drying with acetone prior to testing will reduce agglomeration and surface oxide, but there will still be high and low reactive sites [23].  Activation energy. The activation energy for pyrite oxidation has been reported by a number of studies to be in the range of 50 to 90 kJ/mol. This is well above the range that would be expected for a diffusion-controlled reaction (~10 kJ/mol), indicating that neither the mass transfer of oxidant to the mineral surface nor the mass transfer of products away from the surface is rate limiting [19].  18  Fe Type Fe(III) only STR  Reactor STR 74–150  do (µm) 125–250 306  T (K) 298–303  Chemical rate function  C  C  CFe( III)  +C  CFe( III)  Fe ( III )  C Fe ( III ) Fe ( II )  (C  Fe ( III )  Fe ( III )  C Fe ( III )  C Fe ( III )  2  )  + kl  Fe( III)  + KC Fe ( II ) )C H0.44 +  + KC Fe ( II ) )C H0.44 +  CFe( II) CFe(III) + KRCFe(II)  (C  (C  0.5  + KLCFe( II) )  CFe(III)    C Fe ( III )   − 0.5  K 1C H + + K 2 C Fe ( II )   0.93 Fe ( III ) 0.40 Fe ( II )  0.47 0.32 C Fe ( II ) C H +  0.30 C Fe ( III )  K + C Fe ( III )  C Fe ( III )  f (C ) = C Fe ( III )  0.58 −0.5 f (C ) = C Fe ( III ) C H +  Table 2 – Summary of Solution Chemistry Effects on Pyrite Oxidation Kinetics [1]  Fe(III) only 303–343  f (C ) = f (C ) =  Medium 0.0005–0.0032 M Fe(III) in Cl− 0.0004–0.001 M FeT 0.02–0.09 M Fe(III) in SO42− and Cl−  293–303  <0.003M FeT in SO42−  308  293–303  10  150–250  150–250  45–53  f (C) = kR  (KC  1 f (C ) = C H0.5+  333–363  0.05–1.0 M Fe(III) in Cl−  f (C ) =  313–373  303–343  0.02 M Fe(III) in SO42− and Cl−  f (C ) =  303  f (C ) =  0.04–0.4 M Fe(III) in SO42−  0.001–0.270 M Fe(III) 0.0001– 0.1 Fe(II) in SO42− 0.1–1.0 M Fe(III) in Cl−  f (C ) =  Shake Flask  STR & MFR  Fe(III) & Fe(II) Fe(III) & Fe(II)  44–105  53–500  150–600  53–76  <0.003M FeT in SO42−  Fe(III) only  Fe(III) & Fe(II) STR & MFR  STR  Fe(III) & Fe(II)  STR  STR  Fe(III) & Fe(II)  Fe(III) & Fe(II)  Airlift  STR  Fe(III) & Fe(II)  Reference McKibben and Barnes (1986) Wiersma and Rimstidt (1984) Boogerd et al. (1991)  Williamson and Rimstidt (1994)  Williamson and Rimstidt (1994)  Nayak et al. (1995)  King and Perlmutter (1977)  Kawakami et al. (1988)  Mathews and Robins (1972)  Boon and Heijnen (1998); Boon et al. (1999)  19  2.2.1 Reaction stoichiometry As a result of the complex nature of sulphur behavior in solution, and particularly its fate in sulphide leach systems, the stoichiometry of the pyrite oxidation reaction is not fixed. Moses [18] and others have reported in elevated temperature acidic media that the sulphur in pyrite reports as elemental sulphur or sulphate. Below pH 3.9 a negligible amount of sulphur reports as thiosulphate or polythionate [1]. Assuming that the pyrite sample is pure FeS2 then the stoichiometry of the reaction is determined by the amount of sulphur that is oxidized to elemental sulphur S(0) rather than to sulphate S(IV). The following two reactions apply:  FeS2 + 14 Fe3+ + 8 H2O → 15 Fe2+ + 2 SO42– + 16 H+  (10)  FeS2 + 2 Fe3+ → 3 Fe2+ + 2 S0  (11)  If the fractional yield of sulphur as sulphate (the fraction of the reaction that proceeds as Eq. 10 instead of Eq. 11) is given as β, then the following stoichiometry applies to pyrite oxidation: FeS2 + (2+12β) Fe3+ + 8β H2O → (3+12β) Fe2+ + 2β SO42- + (2-2β) S0 + 16β H+  (12)  The value of β varies from 0 to 1 depending on temperature, O2 pressure, and solution chemistry. A value of 0 corresponds to 2 electron moles per mole of pyrite oxidized, while a value of 1 corresponds to 14 electron moles per mole of pyrite oxidized. Understanding the sulphate yield in a given industrial operation is important as it  20  determines a significant portion of the oxygen demand of the reactor and also contributes to the amount of heat and acid generated. With higher temperature and pressure the value of β is closer to 1. This is why pyrite leaching in an autoclave can be used to generate acid and heat for a hydrometallurgical plant. The value of β is also affected by the solution chemistry, in particular the solution potential. Holmes and Crundwell [24] found that as the solution potential increases so too does β, as shown in Figure 4. In general, the more intense and strongly oxidizing the system is (increase in temperature, O2 pressure, solution potential, and/or total iron), the higher the value of β.  21  Figure 4 - Sulphate yield versus solution potential  22  Some researchers have also found that a minor portion of the sulphur reports as thiosulfate, polythionate, and sulfite. However, thiosulfate is very unstable in acid and quickly disproportionates to elemental sulphur and sulfite. This sulfite is this oxidized to sulphate by ferric iron [1].  Depending on the acidity of the leach solution, the reaction can also be a net acid consumer. This finding was reported by Bailey and Peters [25], who found that at a level of 0.17 M H2SO4 the system was acid-neutral. With increasing acid strength the reaction was acid-consuming, leading the system to buffer at pH ~1.  2.2.2 Pyrite sample source The majority of the present work was done using pyrite from the Huanzala mine in Peru. Two other samples were also used – one from El Potosi mine in the Chihuahua state of Mexico and the other from the Zacatecas pyrite behavior under Galvanox™ conditions. Different ore bodies of pyrite can behave differently in terms of flotation, leach rate, weathering, and reactivity. The reasons for the differing behavior of various pyrites are not fully understood. One contributing factor is the crystal structure of the pyrite, which ranges from strongly defined euhedral to amorphous anhedral structures, or a blend of the two (subhedral) [19]. The crystal structure of pyrite not only influences the rate of ferrous catalysis, but also impacts its suitability in Galvanox™ leaching. Pyrite crystal structure is also important in the larger view of an industrial plant as a whole, as the flotation behavior of pyrite can change. In addition to the influence of crystal structure, impurity elements that affect the properties of pyrite as a semiconductor are suspected to alter the  23  behavior of pyrite in Galvanox™. Pyrite aging and weathering, which can lead to the build up of surface layers, will also alter pyrite behavior. The mechanisms and details of the aforementioned factors are not fully understood. This topic is the subject of a current PhD study within the UBC Hydrometallurgy Group.  24  3 Introduction In most kinetic test work the kinetics of a given reaction are measured by analysis of solution samples using atomic absorption spectroscopy, titration, or some other chemical analysis technique. This necessarily limits the number of data points that can be taken from a test since taking many solution samples can significantly alter the overall mass balance and solution chemistry. In this work the primary data to be tracked are the pH and the redox potential. As a result many data points can be taken since taking them does not affect the solution chemistry. Taking a large number of data points also allows greater confidence when removing the noise from data curves. When data points are separated by hours it can be difficult to determine which, if any, are outliers. When data points are separated by seconds, this distinction is much clearer.  Of course, redox potential and pH are both functions of activity and solution speciation rather than simply the molal concentration of species. As a result they are difficult to correlate directly to hard values such as the proportion of ferrous to ferric or the amount of acid in solution. In order to work around this problem two regimes of testing took place. One regime was a set of kinetic tests where the effect of pyrite catalysis on ferrous oxidation was investigated within the context of several variables. The other regime was a set of permanganate ORP titrations where the precise number of moles of ferrous converted to ferric was tracked by measuring the amount of permanganate that had entered the system. By correlating the redox potential at a given point of the kinetics test to the same potential on an ORP titration curve, the ratio of ferrous to ferric in the kinetic test can be given precisely without resorting to speciation calculations.  25  3.1 Kinetic tests The goal of the kinetics tests was to accurately measure the effects of several variables on pyrite catalysis of ferrous oxidation. To this end, tests were conducted to investigate the following factors: •  Acidity. The level of acidity was varied from 15 to 90 g/L H2SO4, with 60 g/L as the baseline. The acidity of a solution has a noticeable effect on both the redox potential (higher acidity increases the redox potential for the same ratio of ferric to ferrous), and on the rate of ferrous oxidation in solution (increased acid has a negative effect on the rate of ferrous oxidation).  •  Pyrite pulp density. The rate of ferrous oxidation increases with increasing pyrite pulp density. This is consistent with the idea that pyrite catalysis is a function of available pyrite surface area. Pyrite pulp density was investigated from 0 to 40 g/L with 20 g/L as the baseline.  •  Pyrite ore type. The bulk of this work was done with Huanzala pyrite from Peru, which had been tested extensively in previous Galvanox™ test work. Pyrite from the El Potosi and Zacatecas mines in Mexico were also tested. It is known that pyrite types vary in reactivity and impurities, so their performance as ferrous oxidation catalysts was investigated here.  •  Copper concentration. Previous ferrous oxidation kinetic studies have shown that copper in solution has a catalytic effect on ferrous oxidation. Since this work is geared primarily towards use in Galvanox™ copper leach plants it seemed prudent to investigate it here as well. The majority of work was done with no copper, but levels as high as 8 g/L Cu were used.  26  •  Temperature. The majority of this work took place at 80°C, the optimal temperature for Galvanox™, but two tests were conducted at 60°C in order to estimate activation energies for ferrous oxidation in solution and ferrous oxidation on pyrite.  •  Total iron. The effect of the starting level of iron was investigated, with the baseline at 11 g/L, ranging down to 3.3 g/L. It is important to note that the iron level increased as pyrite leached during these tests, so the aforementioned values represent the initial iron concentration.  In these tests there were three reactions of note. First is the oxidation of ferrous by dissolved oxygen in solution (Eq. 3). Second is the oxidation of ferrous catalyzed by pyrite:  4 Fe2+ + O2 + 4 H+ FeS  2 → 4 Fe3+ + 2 H2O  (13)  Third is the oxidation of the pyrite by ferric, where β is the proportion of sulfur oxidized to sulfate:  FeS2 + (2+12β) Fe3+ + 8β H2O → (3+12β) Fe2+ + (2–2β) S0 + 16β H+ (14)  By modeling the known rate of ferrous oxidation in solution in the absence of pyrite as well as the rate of pyrite leaching, the rates of the three reactions could be delineated and analyzed separately.  27  3.2 ORP titrations The goal of the ORP titration regime was to accurately correlate the redox potential with the concentration ratio of ferric to ferrous (as opposed to the activity ratio). In order to do this, ferrous solutions with various acid levels were made and a standard oxidizing solution was dripped in slowly, oxidizing the ferrous. Permanganate was chosen as the oxidizing agent because of its strong oxidizing potential, rapid action, and previous use in similar tests. Because the permanganate standard was of known concentration and there were no other contaminants in the solution, all of the permanganate can be assumed to be oxidizing ferrous iron to ferric, thus:  MnO4– + 8 H+ + 5 e– → Mn2+ + 4 H2O  EO = 1.491 V (vs SHE)  5 Fe2+ + MnO4– + 8 H+ → 5 Fe3+ + Mn2+ + 4 H2O  (15) (16)  As a result, a solution containing concentrated H2SO4 was dripped into the reactor at a rate such that the overall acid level remained constant. Depending on the acidity, the starting potential for these tests ranged from 280 mV to 360 mV. The final potential was 500 to 530 mV, at which point manganese oxide began to precipitate. The information from these ORP titration curves of conversion versus redox potential was then used to determine the conversion as a function of redox potential in the kinetic tests.  28  4 Experimental 4.1 Equipment 4.1.1 Kinetic tests All tests were conducted in an airtight 2.1 L water-jacketed glass batch reactor (labeled A in Figure 5). The insulating jacket was fed by a Haake water heater/circulator with digital temperature control (B). The upper fixture of the batch reactor was modular, containing ports for the impeller, pH probe, redox electrode, temperature probe, baffles, oxygen sparger, and reagent addition (C). The impeller (D) was controlled via an Applikon stirrer controller (E). For control of the oxygen sparge rate, tracking of temperature, pH and redox potential data as well as potential control, an Applikon ADI 1030 Bio Controller (F) was used. The pH probe was a glass model made by AppliSens. The epoxy Ag/AgCl redox electrode was made by Analytical Sensors Inc and used 4.0 molar KCl with AgCl for fill solution. An Aalborg mass flow controller (G) regulated the flowrate of oxygen. A PC laptop (H) running a specially designed Excel macro recorded data from the Bio Controller.  29  Figure 5 - Experimental setup 4.1.2 ORP titrations In the case of ORP titrations the experimental setup remained much the same, except that permanganate acted as the oxidant instead of sparged oxygen. A permanganate standard and a strong acid solution were dripped into the reactor via peristaltic pumps. The amount of solution added was monitored by tracking the weight removed from reservoirs that were placed on digital scales. These scales were hooked up to the data tracking laptop, which recorded the weight reading of both scales as well as temperature, pH, and redox potential.  30  4.2 Experimental procedure The procedure for both types of tests began the same. The first step was to heat the acid and water to temperature. 1.5 kg of water was mixed with the appropriate amount of concentrated sulfuric acid (15 to 90 g per kg water) and added to the reactor. The heated water bath was given a set point of 83.6 to 84.0°C (depending on ambient temperature) in order to achieve a reactor temperature of 82.0°C. The impeller was started at a modest speed of 400 rpm in order to speed the heating process. During the heating phase the pH and redox probes were calibrated and secured in the reactor and the laptop began tracking data. The appropriate amount of ferrous salts, copper salts, and pyrite were weighed out and set aside. At this point, the procedures of the two testing regimes diverged.  4.2.1 Kinetic tests As the solution approached temperature, oxygen sparging at a rate of 250 mL/min was started in order to allow time for a pure oxygen atmosphere to build up over the solution, and to ensure that the solution was saturated with dissolved oxygen. The impeller speed was increased to 1200 rpm to increase gas liquid mixing. The ferrous salts were added to the solution after 15 minutes of oxygen sparging. The endothermic dissolution reaction of the ferrous salts served to bring the temperature of the solution down from 82°C to 80°C, and the water heater control was adjusted downward to maintain an 80°C set point. After allowing 30 seconds for the ferrous to dissolve, the charge of ground pyrite was introduced, marking time zero.  31  Values for pH and redox potential were recorded every 15 seconds for the first 2 hours, and every minute thereafter. Tests continued for 20 to 22 hours. At the end of the test the solution was filtered through 2-µm filter paper to collect the pyrite residue, which was then dried and weighed.  4.2.2 ORP titrations For the ORP titrations standard solutions of permanganate and strong acid were made. The permanganate standard was made up to 0.3 molal as this was found to be significantly below the solubility limit of potassium permanganate (approximately 0.4 molal). The permanganate salt was added to the appropriate amount of water and allowed to mix with a magnetic stir bar for 2 to 3 hours prior to use. During this time the beaker was wrapped in foil to prevent premature reduction of the manganese by UV radiation. The permanganate standard was made fresh for each test, as permanganate solutions gradually precipitate if left in storage for any length of time. The acid standard was 6.0 molal H2SO4. The acid standard could not be mixed with the permanganate standard as the former would quickly cause precipitation and premature reduction of the latter. Instead they were kept in separate beakers. A large amount of the acid standard was made initially and used for all the tests, as sulphuric acid does not experience aging effects over a one-week time span.  Once standards were made, the reactor was heated up just as in the kinetic test procedure. Once the reactor reached temperature, nitrogen was sparged in order to purge the solution of any dissolved oxygen. After the solution had been purged for 15 minutes the ferrous  32  salts were added and the impeller speed was reduced to 400 rpm. The lower impeller speed was chosen to reduce gas liquid mixing through the surface and to prevent any residual oxygen from entering the solution. After allowing 30 seconds for the ferrous to dissolve, the peristaltic pumps controlling the permanganate and acid standards were engaged. The acid solution was pumped in at a rate proportional to the rate of permanganate addition, calculated by the stoichiometry of acid consumption in the permanganate reaction.  pH, redox potential, temperature, and the readings on the acid and permanganate digital scales were recorded every 15 seconds. The tests were concluded when MnO2 began precipitating, typically after 2 to 3 hours. The presence of a red, cloudy precipitate was a clear indicator of solid MnO2 in the reactor.  4.3 Pyrite characterization The three samples of pyrite used in this test were characterized by 1) multi-acid digestion and ICP to determine the amount of each element in the sample, 2) XRD Rietveld Analysis in order to determine the minerals reporting in each sample, 3) nitrogen gas sorption testing to determine the surface area per unit weight of each ground sample, and 4) particle size analysis using laser diffraction measurement to determine the particle size distribution.  33  4.3.1 Multi-acid digestion and ICPMS - elemental assay Elemental assays were conducted on the three pyrites to determine impurity elements. The samples were finely ground and completely digested in a multi-acid solution. The resultant solution was then analyzed using inductively coupled plasma mass spectroscopy (ICPMS). The results are summarized here in Table 3.  Table 3 – Pyrite elemental assay by ore type in ppm Al Ca Cr Co Cu Fe Pb Mg Ni Zn  El Potosi 200 300 246 818 73 499100 <2 <100 551 52  Zacatecas 107 270 159 346 49 489030 <2 111 394 19  Huanzala <2 17010 <2 <2 5000 467000 2000 2000 1000 15400  34  4.3.2 XRD Rietveld analysis - mineralogy X-ray diffraction (XRD) Rietveld analysis was conducted on all 3 samples in order to determine the mineralogy of the samples; in particular looking for the presence of gangue rock. XRD Rietveld analysis alone is not accurate enough to predict the elemental breakdown of the samples, but the information obtained from such analysis is invaluable in determining the mineralogy. The results of the XRD analysis are summarized in Table 4.  Table 4 - XRD Analysis of Pyrite Ore Samples Mineral Pyrite  Ideal Formula FeS2  Huanzala 97.4  Melanterite  Fe2+SO4 7H2O  2.6  1+  3+  El Potosi 84.1  Zacatecas 92.4  Rhomboclase  (H5O2) Fe (SO4)2·2H2O  6.4  Rozenite  Fe2+SO4·4H2O  6.6  Szomolnokite ?  Fe2+SO4 H2O  1.8  Quartz  SiO2  1.1  Enargite  Cu3AsS4  5.8  Chalcopyrite  CuFeS2  1  Total  100  100  0.8  100  35  4.3.3 Nitrogen sorption testing - surface area Gas sorption is a common test to determine the surface area and pore size of solid powders. The sample is placed under vacuum and nitrogen is pumped into the chamber. The amount of gas sorbed is determined by pressure variations. Knowing the area occupied by one molecule of gas, the surface area of a given weight of sample can be measured. This analysis was performed with a Quantachrome Autosorb Automated Gas Sorption system. A summary of results is presented here in Table 5.  Table 5 - Surface area per unit weight of pyrite ores Pyrite Type Huanzala Zacatecas El Potosi  Surface Area (m2/g) 2.06 0.46 1.17  36  4.3.4 Particle size distribution Particle size distribution analysis was done with an Elzone II 5390. This equipment works on the electrical sensing zone method, which detects changes in the electrical resistance as particles suspended in an electrolyte move through an aperture. The results can be seen in Figure 6 for Huanzala, Figure 7 for Zacatecas, and Figure 8 for El Potosi. This information is summarized in tabular form in Table 6.  Table 6 - Particle size distribution data for pyrite samples Huanzala Do (µm) F 1.195 0 1.998 0.01176 2.499 0.02374 4.996 0.1338 7.5 0.1988 9.999 0.2452 12.49 0.2861 15 0.326 20 0.3938 25 0.455 29.99 0.5176 34.98 0.5845 37.98 0.6288 39.99 0.6592 43.97 0.7205 49.99 0.8094 53.96 0.8609 59.97 0.9142 69.95 0.9666 74.98 0.9817 79.96 0.9903 89.95 0.9963 98.89 1 -  El Potosi Do (µm) F 0 0 1.998 0.00841 2.499 0.01749 4.996 0.1857 7.5 0.2975 9.999 0.365 12.49 0.417 15 0.4638 20 0.541 25 0.6025 29.99 0.6566 34.98 0.7056 37.98 0.7338 39.99 0.7521 43.97 0.7878 49.99 0.8364 53.96 0.8691 59.97 0.9125 69.95 0.9623 74.98 0.9753 79.96 0.9822 89.95 0.9916 96.13 0.9933 114.2 0.9956 128.1 0.9973 147.8 1  Zacatecas Do (µm) F 1.195 0 1.998 0.00639 2.499 0.01419 4.996 0.0879 7.5 0.1566 9.999 0.2205 12.49 0.2784 15 0.3347 20 0.4334 25 0.5133 29.99 0.5798 34.98 0.6356 37.98 0.6655 39.99 0.6862 43.97 0.7268 49.99 0.7781 53.96 0.8107 59.97 0.8533 69.95 0.9069 74.98 0.9273 79.96 0.9447 89.95 0.9732 99.97 0.9897 119.9 0.9965 128.1 1 -  37  Figure 6 - Particle size distribution of Huanzala pyrite  38  Figure 7 - Particle size distribution of Zacatecas pyrite  39  Figure 8 - Particle size distribution of El Potosi pyrite  40  5 Results 5.1 Data smoothing During these tests data points were taken very frequently, leading to complete curves consisting of 900–1500 points. As a result of the nature of this data it was often quite noisy, with many outliers. Analysis of data of this sort is difficult without first smoothing it and removing outliers to determine the true trends of the tests. This is doubly true when the raw solution potential versus time data is being correlated to the actual ferrous conversion via ORP titrations. In each of these tests, this problem was solved using the following approach, visually represented in Figure 9. First, raw solution potential versus time data (the upper solid line in Figure 9) was correlated directly to conversion values. This yields a rather noisy curve of conversion versus time (the lower solid line). Then, this noisy curve was fit with a numerical trend line (dashed line). It was this dashed line that was compared to model fits.  5.2 ORP titrations with permanganate Figure 10 and Figure 11 show the results of permanganate ORP titrations.  The  conversion refers to the amount of ferrous oxidized as a function of the amount of permanganate added. Five titrations were conducted at acid levels ranging from 15 to 90 g/L. Points represent the data while the smoothed lines represent a numerically fit trend line.  41  Figure 9 - Data analysis and curve fitting for conversion curves  42  Figure 10 - ORP titrations with permanganate at 80°C  43  Figure 11 - ORP titrations with permanganate at 60°C  44  5.3 Ferrous oxidation without pyrite The conversion curves shown in Figure 12, Figure 13, and Figure 14 were achieved by smoothing the noise and outliers out of the solution potential curves before correlating them to the ORP titration data. Solution potential data is inherently noisy when points are taken as frequently as in this work; smoothing helps reveal the true trends in the data. These tests show the curves of conversion versus time as a function of acidity and of copper concentration. Figure 15 shows the significant effect of temperature on the kinetics of ferrous oxidation without pyrite.  45  Figure 12 - Ferrous oxidation in solution, 60 g/L H2SO4  46  Figure 13 - Ferrous oxidation in solution, 15 g/L H2SO4  47  Figure 14 - Ferrous oxidation in solution, 60 g/L H2SO4, 8 g/L Cu  48  Figure 15 - Ferrous oxidation without pyrite as a function of temperature  49  5.4 Ferrous oxidation with pyrite Figure 16 shows the difference in conversion as a function of pyrite pulp density from 0 to 40 g/L. In these tests, because the concentration of iron is changing as pyrite is leached, the conversion is expressed as the fraction of the total iron in the system that is converted to ferric, not the fraction of the initial iron in the system converted.  Figure 17 shows the difference in performance between pyrite samples from three different sources. The solution conditions are identical except for differing sources of pyrite. The amount of solids added was not varied to control for actual weight of pyrite added, as all 3 samples consisted of over 90% pyrite by weight. The El Potosi sample was more reactive than the Huanzala and Zacatecas samples. In these tests it can also be seen that the conversion curves for tests with pyrite do not begin at 0. This is because immediately after introducing the pyrite to the environment the potential would increase by 35 to 80 mV, depending on the pyrite pulp density. This corresponds to between 2 and 6% conversion. This jump was 9.5% in the low iron test (0.06 molal Fe initially rather than 0.2 molal), indicating that the effect is not related to the rest potential of the pyrite. Instead, it is caused by the rapid dissolution of ferric iron salts present on the pyrite due to pyrite weathering and aging.  Figure 18 shows the effect of acidity on the rate of ferrous oxidation with pyrite, ranging from 15 to 90 g/L H2SO4. The differences are minor, indicating that the acidity does not have a significant effect on the rate of ferrous oxidation on pyrite. However, the acidity  50  will have an effect on ferrous oxidation in solution, which is happening in parallel with ferrous oxidation on pyrite, albeit at a considerably lower rate.  Figure 19 shows the effect of changing the impeller speed from 1200 rpm to 1800 rpm. As can be seen, the curves are virtually identical, indicating that the reaction rate is limited by chemical kinetics and not mass transfer and gas-liquid mixing.  Figure 20 shows the effect of temperature on the kinetics of ferrous oxidation on pyrite. This effect is not nearly as dramatic as that shown in Figure 15. This indicates that the reaction path on the pyrite surface is less sensitive to temperature than the reaction path in solution without pyrite. The effect of pyrite leach rate and the addition of soluble iron from the dissolution of pyrite could also play a role in these curves.  51  Figure 16 - Ferrous oxidation with pyrite, varying pyrite pulp density  52  Figure 17 - Ferrous oxidation with pyrite, varying ore type  53  Figure 18 - Ferrous oxidation with pyrite, varying acidity  54  Figure 19 - Ferrous oxidation with pyrite, effect of impeller speed  55  Figure 20 - Ferrous oxidation with pyrite as a function of temperature  56  6 Analysis and Discussion In this work three separate reaction models must be used, as there are three distinct processes occurring. The first is the oxidation of ferrous in solution without interaction with pyrite. In the case where no pyrite is added, this is the only reaction causing a change in ferrous concentration. When pyrite is added, this reaction occurs in parallel to the pyrite catalysis. The second process is the leaching of pyrite. This occurs at an increasing rate as the redox potential increases and adds both acid and ferrous to solution, while consuming ferric. The third process is the catalysis of ferrous oxidation on the surface of pyrite. This process contributes to the ferrous oxidation kinetics but must be considered separately from solution oxidation. Models of ferrous oxidation in solution and pyrite leaching have been presented in literature previously. Some of the techniques used by previous researchers have been adapted for this work. In addition to these three reaction models, Tromans’ oxygen solubility model was also used to link oxygen pressure to oxygen concentration in solution [3].  6.1 Ferrous oxidation in solution Regardless of the pyrite concentration, ferrous oxidation will occur in solution according to Eq. 3. Analysis of the kinetics of this reaction has been undertaken a number of times in previous literature. Dreisinger and Peters present a summary of the different proposed differential forms in their 1989 paper on the topic [2]. The majority of the proposed models are of the following form:  57  dCFe(II) dt  = −k  a CFe(II) POb2  (17)  CHc +  where 1.84 ≤ a ≤ 2, 1 ≤ b ≤ 1.04, and 0 ≤ c ≤ 0.35. Dreisinger and Peters proposed a more complex model involving solution speciation and separate rate terms for each ferrous species, as follows:  dCFe(II) dt  2 = − KT (k1CFe2 2+ + k2CFe 2+ CFeSO 0 + k3CFeSO CCuSO 0 ) 0 ) PO (1 + k 4 2 4  4  (18)  4  where KT is an Arrhenius activation energy term and k1, k2, k3 and k4 are constants. This model accounts for the higher reactivity of ions with lower positive charge due to lesser electrostatic repulsion, changes in speciation due to varying levels of acid and sulfate, and the enhancing effect of dissolved copper on ferrous oxidation.  However, considering kinetics as a function of the partial pressure of oxygen is not strictly accurate, since this neglects the effect of ionic strength on oxygen solubility and also the effect of temperature on oxygen partial pressure. In this work the effect of ionic strength on oxygen solubility was modeled using Tromans’ model, outlined below in section 6.2. The model shows that as the ionic strength increases the solubility of oxygen decreases. In addition to its direct effect on the Henry’s Law coefficient, temperature also affects oxygen solubility indirectly by altering the vapor pressure of the system and thus decreasing the oxygen partial pressure at a constant total pressure. In this work the reactor was an atmospheric sealed tank fed with pure oxygen gas. However, the partial  58  pressure of oxygen was not 1 atmosphere, but rather 1 atmosphere minus the water vapor pressure. When a kinetic expression is derived for a single temperature this correction is embedded in the pre-exponential constant, but for variable temperatures this must be corrected for separately.  In this system the oxygen sparge rate was high relative to the oxygen demand by ferrous oxidation. One test was conducted at a higher impeller speed (1800 rpm versus the baseline 1200 rpm) and no difference in kinetics was observed. As a result, it can be assumed that the rate of oxygen consumption by ferrous oxidation is equal to the rate of oxygen dissolution from gas liquid mixing. Therefore, it was assumed that for this system the concentration of dissolved oxygen was equal to the oxygen solubility limit given by Tromans’ model. This assumption may not be valid for some systems and must be investigated on a case by case basis.  In this work the following expression was used to model ferrous oxidation in solution:  dCFe(II) dt  2   1  CFe(II)CO 2 1 = −49700 exp − 9660 −  0.25  T 423.15  CH +    1 + 5.75 CCu(II)CFe(II)  CFe(II) + CFe(III)        (19)  It was confirmed in this study that increased acidity has a negative effect on ferrous oxidation kinetics. This is because protons have a strong tendency to form bisulfate (HSO4–) with any sulfate in the system [11-12]. This strips the sulfate from FeSO40 ion pairs, leaving ferrous ions more positively charged. The stronger positive charge results  59  in stronger electrostatic repulsion between ferrous ions, and thus, slower kinetics. This is the reason for the acid term in the denominator of Eq. 19.  This study also confirmed that copper has a catalytic effect on ferrous oxidation kinetics. This is owed both to the presence of copper in solution as well as to the increased sulfate when copper is added as sulfate salts. It was found that the catalytic effect of copper decreased somewhat as the redox potential increased, leading to the multiplication of the ferrous to total iron ratio to the copper concentration. The positive effect of copper on ferrous oxidation kinetics is more noticeable at lower solution potentials and lower levels of ferric iron.  Figure 21 - Cu-H2O Pourbaix diagram  60  As can be seen on the Cu-H2O-H2SO4 Eh-pH diagram Figure 21, at a redox potential of  ~0.4 V there is a dividing line between solid Cu0 and aqueous cupric Cu2+. However, EhpH diagrams only show the dominant ion under any given set of conditions. Aqueous copper is also sparingly stable as Cu+, or cuprous, existing simultaneously with cupric. As the solution potential increases the ratio of cuprous to cupric decreases. Since the catalytic effect of copper on ferrous oxidation is related to the rapid reversible cuprouscupric reaction: Cu2+ + e– → Cu+  (20)  it stands to reason that as solution potential increases and the cupric ion becomes less stable, the catalytic effect is diminished. The subsequent re-oxidization of cuprous by oxygen is virtually instantaneous [20], as cuprous ions have a very strong affinity for dissolved oxygen:  4 Cu+ + O2 + 4 H+ → 4 Cu2+ + 2 H2O  (21)  The kinetics of the reaction of cuprous ions and dissolved oxygen are so fast that dissolved oxygen and cuprous can not co-exist in solution [20]. For this reason the reaction can be used to measure gas-liquid mixing kinetics in reactors.  The use of a speciation procedure in this model was considered, but was rejected in the interest of keeping the model simple. It proved unnecessary to introduce another level of complexity when the experimental data could be simulated effectively without directly accounting for speciation.  61  6.2 Factors affecting oxygen solubility According to Henry’s Law the solubility of a gas in a given fluid is the partial pressure of the gas multiplied by the solubility limit, which is a function of temperature and ionic strength of the solution. In this work Tromans’ model of oxygen solubility is used to determine the effect of dissolved salts on oxygen solubility. In this model, the effects of temperature and ionic strength are decoupled and calculated separately:  COsat2 = K O (T ) f (I )PO 2  (22)  In the case of pure oxygen gas, the partial pressure is calculated by subtracting the water vapour pressure from the total pressure. The vapour pressure of water as a function of temperature is easily calculable by a number of expressions and Excel add-ins.  The effect of temperature on oxygen solubility in aqueous media is modeled as follows also shown in Figure 22:   68623 − 1430.4 T − 0.046 T 2 + 203.35 T ln T   K O (T ) = exp  8.3144 T    (23)  Modeling the effect of ionic strength is slightly more complicated but no more difficult. For a single electrolyte in aqueous solution the form is as follows:  62  f (I ) = Φ = (1 + κC y ) − η  (24)  where κ, y, and η are electrolyte-specific fitting parameters. In Tromans’ paper the following table of values for κ, y, and η is laid out (Table 7).  Table 7 - Tromans’ oxygen solubility parameters – sulphate ions Compound H2SO4 Na2SO4 K2SO4 MgSO4 ZnSO4 CoSO4 NiSO4 CuSO4 Al2(SO4)3  Κ 2.01628 0.629498 0.55 0.119674 0.232671 2.23207 2.23207 2.23207 0.641163  y 1.253475 0.911841 0.911841 1.107738 1.010428 1.115617 1.115617 1.115617 0.954719  η 0.168954 1.440175 1.440175 5.455537 2.655655 0.222794 0.222794 0.222794 3.033594  For a solution with multiple electrolytes the value of Φ must be calculated for each individual ion as though it were alone in the system. Then the multiple Φ values are combined as follows, where Φ1 is the lowest Φ value (and thus has the most severe effect on oxygen solubility) of all the solutes in the system:  z  f ( I ) = Φeff = Φ1 ∏ Φi0.8  (25)  2  It can be seen using this model that as the ionic strength of a solution increases, the oxygen solubility decreases.  In a typical test for this project the oxygen solubility  63  decreased by about 15% over the course of the test, due to pyrite leaching and the higher average valence of iron in solution.  64  Figure 22 - Oxygen solubility as a function of temperature  65  6.3 Pyrite leaching The rate of pyrite leaching is limited by reaction at the surface of the unreacted core underneath the layer of elemental sulfur which forms. Modeling of pyrite leaching is best achieved by using the shrinking sphere (linear leaching) model which expresses the rate of conversion over time as follows:  dX ∂d 3(1 − X ) = ⋅ dt ∂t d0  2/3  (26)  where d0 is the initial particle size, and ∂d/∂t is the shrinkage rate, which is a function of temperature and solution chemistry. The shrinkage rate was taken from the work of Bouffard et al. [1], albeit with a slightly different rate constant (Eq. 27):   ∂d 1  CFe( III)  1 = 0.0129 exp − 9937  −  ∂t  T 333  CFe( II )    0.572  (27)  Using this relation to calculate the particle shrinkage rate, the overall rate of conversion on monosize particles can be found. However, in this work the pyrite was not monosized and so the overall rate of conversion can only be determined by integrating over a range of particle sizes.  To do this, Gauss-Legendre quadrature (GLQ) was used, again following Bouffard et al. [1]. GLQ is a method of numerically solving integrals over a finite range by evaluating the function at several Legendre polynomial root points and then weighting and summing 66  the resulting values. In this case, the variable was particle size. By expressing the particle size distribution as a cumulative distribution function F(d0) the data can be expressed as shown in Figure 23, where F is the volume fraction of the sample passing a given particle size d0.  67  Figure 23 - Huanzala cumulative particle size distribution  68  Legendre root points used in GLQ are normally taken over the range of –1 to +1. In this case they had to be normalized to the range of 0 to 1 in order to fit the bounds of the cumulative particle size distribution function (Table 8). Once the 15 Legendre root points were calculated they were correlated to particle diameters corresponding to those passing mass fractions by using the Gaudin-Meloy particle size distribution function. These values are given in Table 8.  By evaluating Eq. 27 at these specific Legendre root point values, weighting them according to GLQ weights, and summing the values, the leach rate for a given solution chemistry and temperature can be determined. The effective formula is as follows:  ∂X ∂d 15 3(1 − X ) = wj ∑ ∂t ∂t j =1 d0 ( z j )  2/3  (28)  where wj is the jth GLQ weight, and d0(zj) is the specific particle size corresponding to the jth GLQ root. In this system the chemistry is changing constantly as ferrous is oxidized to ferric, so this formula must also be integrated over the variable chemistry. In this work this was accomplished by evaluating Eq. 28 at each time step and summing the total for the entire 20-h run.  69  Table 8 - Legendre Root Points and Corresponding Particle Sizes 15-Point Legendre Root – 0.988 – 0.937 – 0.848 – 0.724 – 0.571 – 0.394 – 0.201 0.000 0.201 0.394 0.571 0.724 0.848 0.937 0.988  Normalized Legendre Root 0.006 0.031 0.076 0.138 0.215 0.303 0.399 0.500 0.601 0.697 0.785 0.862 0.924 0.969 0.994  Particle Size d0 (um) 0.27 1.44 3.52 6.49 10.33 14.98 20.40 26.52 33.27 40.57 48.35 56.50 64.91 73.47 81.98  6.4 Catalysis of ferrous oxidation by pyrite It is important to note that as there are three processes which contribute to the net change in ferrous iron over time, the conversion curves are not direct reflections of the rate of ferrous oxidation on pyrite but show the cumulative effect on ferrous concentration. As can be seen in Figure 16 and Figure 17, the presence of pyrite in the system has a clear enhancing effect on the kinetics of ferrous oxidation at the beginning of the curve. As the pyrite pulp density is increased, the slope during the first 2 to 3 hours becomes steeper. However, as the test continues, each test with pyrite goes through an inflection point between 3 and 5 hours, after which the rate of conversion becomes roughly linear. Between 7 and 10 hours the equivalent test in the absence of pyrite has overtaken the pyrite tests in the fraction of ferrous converted.  70  The nature of the inflection point is not fully understood as it is a complex system. However, several factors must be in play. One is the increase in the rate of ferric consumption by pyrite leaching and the leaching to completion of small pyrite particles and the subsequent decrease in pyrite surface area. A second is the formation of residual sulfur layers from pyrite leaching that could contribute to the reduced catalysis of the pyrite surface. In the pyrite leach study of Bouffard et al. [1] it was shown that under atmospheric conditions only about 60% of the sulfur in pyrite is oxidized to sulfate, the rest remaining as elemental sulfur.  Another possibility is that the reaction rate of ferrous on pyrite is limited by adsorption/desorption kinetics on the pyrite surface. In this case the rate would not only be a function of ferrous and dissolved oxygen concentrations, but also on the proportion of the pyrite surface area available for ferrous adsorption. The adsorption/desorption kinetics may also be a function of redox potential.  As a result of these complexities, the formulation of an analytical model explaining this phenomenon is difficult at best. However, in the context of a Galvanox™ circuit these considerations may not be relevant. In a Galvanox™ leach the pyrite does not leach until after all more reactive copper sulphides have already electrochemically dissolved. In a continuous circuit additional charge material is constantly being added. In this situation the pyrite will not leach. If the inflection point and subsequent ferrous oxidation slowdown is related to pyrite leaching, then this problem will not present itself in a real Galvanox™ circuit. In this light, the understanding of this inflection point and slowdown  71  takes a back seat to the more industrially relevant question of how pyrite affects ferrous oxidation kinetics prior to significant pyrite leaching.  To that end, the data presented in Figure 16 through Figure 20 was analyzed with an emphasis on the initial part of the curve; typically the first 2 to 3 hours. This generally corresponded to ferrous conversions of 0.18 to 0.22 at a solution potential of 470 to 490 mV. Here again is another reason why understanding the latter part of the curve is less relevant in the context of Galvanox™. In previous work it has been found that the optimal solution potential for operating a Galvanox™ leach is 470 mV. Thus, the crucial part of any ferrous oxidation model is how it predicts the kinetics of the reaction below 470 mV.  After these constraints were placed on the model, the following relationship was formulated to predict the effect of pyrite on ferrous oxidation kinetics:  dCFe(II)  − Ea  1 1  2/3 = k exp  −   CPy APy ⋅ 3(1 − X Py ) CFe (II ) CO 2 f (CCu (II ) ) dt  R  T 353    (   CFe (II )CCu (II ) f (CCu (II ) ) = 1 + kCu, Py  CFe (II ) + CFe (III )   )       (29)  where k is the rate constant at 353 K, Ea is the activation energy, CPy is the concentration of pyrite in solution in g/L, APy is the surface area per unit mass of pyrite as given by nitrogen sorption testing, XPy is the fraction of pyrite oxidized, and CFe(II) and CO2 are the concentrations of ferrous and oxygen respectively. The presence of copper in the system  72  has a further enhancing effect on the pyrite, given by the f(CCu(II)) term, where kCu,Py is the copper-pyrite enhancing factor, and CFe(II), CFe(III) and CCu(II) are the molal concentrations of ferrous, ferric, and copper respectively. This relationship does not describe the kinetics of the reaction that well over the whole range of tests, but at the conditions most relevant to an operating Galvanox circuit it does significantly better.  Several points of the above model are worth elaborating on. First, the activation energy for this reaction was found to be very low, on the order of 10 kJ/mol. This can be seen in Figure 20, where the conversion curve of the 60°C test is quite similar to the 80°C test. There is not nearly the marked difference there as there is in Figure 15, the comparison of the temperature effect on kinetics of ferrous oxidation in solution without pyrite. The pyrite enhancement is weakly affected by temperature compared to pyrite leaching and ferrous oxidation in solution. Second, there was little or no effect of acid on the kinetics of ferrous oxidation on pyrite. The three tests differed, but not markedly so. Third is the importance of tailoring the kinetic constants to reflect the specific pyrite ore body. As outlined in the pyrite characterization section, different bodies of pyrite behave differently in the context of Galvanox leaching. As a result, it is important to account not only for the differing reactivity of the pyrite via the pyrite oxidation kinetic constants, but also in the surface area and enhancing effects of Eq. 29.  Next, the kinetics of this reaction appear to be first order with ferrous rather than second, as in the case of solution oxidation. This indicates that the reaction mechanism on the pyrite surface differs significantly from that in solution. Finally, it appears that the effect  73  of copper on this reaction can be modeled in roughly the same way as in the case without pyrite. The fit resulting from this modeling is presented in Figures 24 through 27.  74  Figure 24 - Model of ferrous oxidation on pyrite, pyrite pulp density  75  Figure 25 - Model of ferrous oxidation on pyrite, pyrite ore source  76  Figure 26 - Model of ferrous oxidation on pyrite, copper  77  Figure 27 - Model of ferrous oxidation on pyrite, temperature  78  6.5 Initial potential increase and pyrite weathering In each of the tests where pyrite is added to heated ferrous sulphate solution it can be seen in Figure 28 that the solution potential instantaneously increases by 30 to 80 millivolts, accompanied by an increase in pH of 0.05 to 0.1 units. The potential and pH spikes are proportional to the amount of pyrite added, indicating that the spike is not related to pyrite’s rest potential. As the spike occurs nearly instantaneously it is believed that the cause is rapid dissolution of iron salts from the surface of the exposed pyrite. It is known that pyrite samples age over time due to the oxidative nature of the atmosphere. As pyrite ages the exposed surfaces are covered by ferric salt. While the amount of ferric sulphate on the surface is small, it is significant enough to change the ratio of ferric to ferrous iron and increase the potential. The increase in pH is due to the added sulphate ion, which preferentially forms bisulphate, thereby lowering the activity of the proton in the system.  79  Figure 28 - pH and Eh increase after pyrite addition  80  6.6 Model formulation The modeling of this system was undertaken in several stages. To begin with, the simplest system was looked at: the oxidation of ferrous iron by molecular oxygen in the absence of pyrite or copper. Analyzing this reaction involved two models; Tromans’ oxygen solubility model, and an original model calculating the rate of ferrous oxidation including factors for oxygen concentration, acidity, and iron concentration. An equation expressing these kinetics was assumed to have the following form, where l, m and n are constants, k(T) is a function of temperature, and f(CCu(II)) is a function of copper concentration:  dCFe(II) dt  l = k (T )CFe(II) COm2 CHn 2SO 4 f (CCu(II) )  (30)  The k(T) function was assumed to be an Arrhenius activation energy function, and so for multiple tests at the same temperature this value could be reduced to a single constant. The other parameters l, m, and n were fit based on a simple square error minimization. In these tests the effect of copper was not present, so the last term of Eq. 30 could be set to 1 and ignored.  Once this simple set of kinetic data was successfully modeled, the next step was to do the same test at 60°C instead of 80°C to determine the value of k(T) as temperature varies. This fixes both the activation energy of the reaction as well as the reference rate constant. After these factors were successfully modeled tests involving copper were analyzed in  81  order to determine the form of the copper term f(CCu(II)) in Eq. 29. It was found that the effect could be quantified with one additional parameter, thus:   f (CCu (II ) ) = 1 + k     CFe(II )CCu (II )     C  + C Fe ( III )   Fe(II )   (31)  This aspect of the model is very similar to the one presented by Dreisinger & Peters (1989), as copper concentration shares a square root dependency in both. The key difference is the inclusion of the iron terms. As a whole these terms account for the depression of copper’s catalytic effect as solution potential increases. This is consistent with the idea that copper’s catalytic effect comes from the Cu(II)-Cu(I) couple. As the solution potential increases, the fraction of copper stable as cuprous (Cu(I)) diminishes. This is also reflected in Cu-H2O-H2SO4 Eh-pH diagrams (Pourbaix diagrams). Once the effect of copper was expressed, the next step was the analysis of kinetic data involving pyrite. This involved two models in addition to those mentioned above – one for pyrite oxidation, and the other for ferrous oxidation on the surface of pyrite. For the former, the starting point for the model was the work presented by Bouffard et al. on atmospheric pyrite oxidation. In this work the pyrite oxidation rate was taken as follows:   ∂d 1  CFe( III)  1 = 0.0129 exp − 9937  −  ∂t  T 333  CFe( II)    0.572  (32)  82  The three parameters in the above expression are the reference rate constant, the activation energy, and the iron ratio exponent. Applying their model to this system, a set of residue weights was predicted for the experiments of this work. Comparing the predicted results to the actual weight of the residue it was found that the Bouffard et al. model predicted lower residue weights than was obtained. By altering the reference rate constant a tighter fit was obtained. This reference rate constant was taken to be a weak function of the mineralogy of the sample. As such, the model used a unique value for the reference rate constant for each of the three pyrite ore samples used in this work. The different particle size distributions for the three pyrite ore samples was accounted for explicitly in this model.  The fourth and final model to be determined was that of ferrous oxidation catalyzed by the pyrite surface. This reaction was modeled separately from ferrous oxidized in solution. It was assumed that the rate of this reaction was proportional to the surface area of pyrite.  83  6.7 Model algorithm Once suitable models were established for ferrous oxidation in solution, ferrous oxidation on pyrite, pyrite oxidation, and oxygen solubility they were stitched together to model the entire system. 1. Global inputs: Temperature, XPy, CPy, Pyrite Type, CFe(II), CFe(III), CCu(II), CH2SO4 2. O2 solubility model: a. Inputs – Temperature, CFe(II), CFe(III), CCu(II), CH2SO4 b. Outputs – CO2,sat 3. Ferrous oxidation in solution: a. Inputs – Temperature, CFe(II), CFe(III), CCu(II), CH2SO4 b. Outputs – dCFe(II)/dt, dCFe(III)/dt, dCH2SO4/dt 4. Ferrous oxidation on pyrite: a. Inputs – Temperature, XPy, CPy, Pyrite Type, CFe(II), CFe(III) b. Outputs – dCFe(II)/dt, dCFe(III)/dt, dCH2SO4/dt 5. Pyrite oxidation a. Inputs – Temperature, XPy, CPy, Pyrite Type, CFe(II), CFe(III) b. Outputs – dCFe(II)/dt, dCFe(III)/dt, dCH2SO4/dt, XPy 6. Summation of changes and time advancement a. Inputs – dCFe(II)/dt, dCFe(III)/dt, dCH2SO4/dt (from #3 to 5) 7. Global Outputs – Updated values of XPy, CPy, CFe(II), CFe(III), CH2SO4  84  6.8 Model results Figures 29 and 30 show the relative effect of pyrite enhancement of ferrous oxidation kinetics as a function of pyrite pulp density and copper concentration, respectively. The conditions chosen are typical for Galvanox operation. The effect of pyrite pulp density is linear, with 20 g/L pyrite providing a reaction path for ferrous oxidation with close to twice the kinetic rate of ferrous oxidation in solution. This effect can also be thought of as the available pyrite surface area. For example, if the pyrite surface area was doubled the effect would be roughly the same as doubling the pyrite pulp density, although this would change the leaching behavior of the pyrite.  Figure 30 shows that the enhancing effect of pyrite on ferrous oxidation is further increased with increasing copper concentration. This effect is significant when moving from a solution with no dissolved copper to 15-20 g/L, but further increases are minimal.  85  Figure 29 - Relative effect of pyrite pulp density on ferrous oxidation kinetics  86  Figure 30 - Relative effect of copper concentration on ferrous oxidation kinetics  87  7 Conclusions In this work experimental evidence is presented for the enhancing effect of pyrite on ferrous oxidation in sulfate media by dissolved oxygen. Kinetics tests were conducted to investigate the effects of acidity, pyrite pulp density, pyrite ore type, and copper concentration on the rate of ferrous oxidation. The primary data was constant tracking of redox potential, which was correlated to conversion of ferrous through the novel use of a series of ORP titrations.  In this system there are three major chemical processes occurring simultaneously: ferrous oxidation in solution, ferrous oxidation on pyrite, and pyrite leaching. Ferrous oxidation in solution was simulated using an adjusted version of the model of Dreisinger and Peters [2]:  dC Fe(II) dt  2 1  C Fe(II)C O2  1 = −49700 exp − 9660 −  0.25  T 423.15  C H +    C Cu(II)C Fe(II) 1 + 5.75  C Fe(II) + C Fe(III)     (33)    Pyrite leaching was modeled in the same fashion as Bouffard et al. [1] using GaussLegendre quadrature to account for the particle size distribution of the pyrite.  ∂X 1  C Fe ( III )   1 = 0.0129 exp − 9937  −  ∂t  T 333  C Fe ( II )    0.572  15  ∑wj j =1  3(1 − X ) d0 (z j )  2/3  (34)  88  Tromans’s model [3] was used to determine the effect of temperature and ionic strength on oxygen solubility, as outlined in section 6.2.  A kinetic expression for ferrous oxidation on pyrite was derived for the range of conditions most relevant to Galvanox™ operations. This relationship does not apply if the pyrite begins to significantly leach, or at solution potentials above ~480 mV. However, across a range of acidity, at modest solution potential, and at or around 80C this relationship did describes the data well as a function of temperature, pyrite pulp density, pyrite ore type, and copper concentration.  dCFe(II)  − Ea  1 1  2/3 = 4000 exp  −   CPy APy ⋅ 3(1 − X Py ) CFe (II ) CO 2 f (CCu (II ) ) dt  R  T 353    (   CFe (II )CCu ( II ) f (CCu (II ) ) = 1 + 20  CFe (II ) + CFe (III )        )  (35)  The conclusions of this study most relevant to Galvanox™ are as follows. First, the quality of the pyrite in any candidate material should be investigated thoroughly, as there are significant differences in reactivity and ferrous oxidation enhancement from one ore body to another. Second, acidity should be limited to the stoichiometric requirement plus a modest excess in order to avoid slowing down the ferrous oxidation reaction, and to increase oxygen solubility in water. Third, the leaching of pyrite should be avoided as much as possible in order to maintain a high and steady level of ferrous oxidation enhancement.  89  More broadly, this study has shown the value of ORP titrations in correlating real ferrous and ferric molal concentrations with solution potential by avoiding the problem of speciation. In addition, the use of Eh and pH as indicators of how far a reaction has proceeded is effective. Having many data points makes kinetic analysis much more effective, particularly for relatively fast reactions such as this.  90  References [1]  S.C. Bouffard, B.F. Rivera-Vasquez, D.G. Dixon.  Leaching kinetics and  stoichiometry of pyrite oxidation from a pyrite-marcasite concentrate in acid ferric sulfate media. Hydrometallurgy 84 (2006) 225–238. [2]  D.B. Dreisinger, E. Peters. The oxidation of ferrous sulfate by molecular oxygen under zinc pressure-leach conditions. Hydrometallurgy 22 (1989) 101–119.  [3]  D. Tromans. Oxygen solubility modeling in inorganic solutions: concentration, temperature and pressure effects. Hydrometallurgy 50 (1998) 279–296.  [4]  M.J. Dry, A.W. Bryson. Prediction of solution potential in concentrated iron sulfate solutions. Hydrometallurgy 21 (1989) 59–72.  [5]  E. Yeager, A.J. Salkind. Techniques of Electrochemistry, Vol. 1. Wiley and Sons, New York, 1972.  [6]  Petersen, J. and Dixon, D.G. Competitive bioleaching of pyrite and chalcopyrite. Hydrometallurgy 83 (2006) 40–49.  [7]  D.G. Dixon, D.D. Mayne, K.G. Baxter. Galvanox™ – a novel process for recovery of copper from primary copper concentrates. Can. Metall. Q. 47 (2008) 327–336.  [8]  Jansen, M and Taylor, Alan. Key elements in the selection of sulphide leach processes for copper concentrate leaching. International Project Development Services Proprietary Limited, Alta Copper 2000, August 2000.  [9]  J.M. Casas, C. Paipa, I. Godoy, T. Vargas. Solubility of sodium-jarosite and solution speciation in the system Fe(III)-Na-H2SO4-H2O at 70°C. Journal of Geochemical Exploration 92 (2007) 111–119.  91  [10] B. Verbaan, F.K Crundwell. An electrochemical model for the leaching of sphalerite concentrate. Hydrometallurgy 16 (1986) 345–359. [11] Casas, J.M., Crisostomo, G. and Cifuentes, L. Speciation of the Fe(II)-Fe(III)H2SO4-H2O system at 25°C and 50°C. Hydrometallurgy 80 (2005) 254–264. [12] M.R. Rönnholm, J. Warna, T. Salmi, I. Turunen, M. Luoma. Kinetics of oxidation of ferrous sulfate with molecular oxygen. Chem. Engng. Sci. 54 (1999) 4223–4232. [13] T. Chmielewski, W.A. Charewicz. The oxidation of iron(II) in aqueous sulfuric acid under oxygen pressure. Hydrometallurgy 12 (1984) 21–30. [14] R.E. Huffman, N. Davidson. Kinetics of the ferrous iron-oxygen reaction in sulphuric acid solution. J. Am. Chem. Soc. 78 (1956) 4836-4842. [15] R.Ž. Vraĉar, K.P. Cerović. Kinetics of oxidation of Fe(II) ions by gaseous oxygen at high temperatures in an autoclave. Hydrometallurgy 44 (1998) 113–124. [16] M. Boon, J.J. Heijnen. Chemical oxidation kinetics of pyrite in bioleaching processes. Hydrometallurgy 48 (1998) 27–41. [17] H. Long, D.G. Dixon. Pressure oxidation of pyrite in sulfuric acid media: a kinetic study. Hydrometallurgy 73 (2004) 335–349. [18] C.O. Moses, D.K. Nordstrom, J.S. Herman, A.L. Mills. Aqueous pyrite oxidation by dissolved oxygen and by ferric iron. Geochimica et Cosmochimica Acta 51 (1987) 1561–71. [19] R.T. Lowson. Aqueous Oxidation of Pyrite by Molecular Oxygen. Chemical Reviews 82 (1986) 461–96. [20] H.T.H. Kimweri. Enhancement of Gas-liquid Mass Transfer in Hydrometallurgical Leaching Systems. PhD dissertation (2001).  92  [21] M.A. McKibben, H.L. Barnes. Oxidation of pyrite in low temperature acidic solutions: Rate laws and surface textures. Geochimica et Cosmochimica Acta 50 (1986), 1509–20. [22] K. Kawakami, J. Sato, K. Kusunoki, K. Kusakabe, S. Morooka. Kinetic Study of Oxidation of Pyrite Slurry by Ferric Chloride. Ind. Eng. Chem. Res. (27) (1988) 571–576. [23] N. May, D.E. Ralph, G.S. Hansford. Dynamic Redox Potential Measurement for Determining the Ferric Leach Kinetics of Pyrite. Minerals Engineering (10:11) (1997) 1279–1290. [24] P.R. Holmes, F.K. Crundwell. The kinetics of the oxidation of pyrite by ferric ions and dissolved oxygen: An electrochemical study. Geochimica et Cosmochimica Acta 64 (2000) 263–274. [25] L.K. Bailey, E. Peters. Decomposition of pyrite in acids by pressure leaching and anodization: the case for an electrochemical mechanism. Can. Metall. Q. 15 (1979) 333–344. [26] P.J. van Staden, M. Gericke, P.M. Craven. Minerals Biotechnology: Trends, Opportunities, and Challenges. Proceedings of Hydrometallurgy 2008, Phoenix, Arizona (2008) 6-13.  93  Appendix A - Aqueous speciation modeling The problem of aqueous speciation can be approached in a variety of ways. Two methods will be outlined here; the method of equilibrium balancing, and that of the Gibbs free energy minimization using Lagrange unspecified multipliers.  A.1 Equilibrium balancing The equilibrium balancing method is the most obvious method of calculating aqueous speciation. To use a generalized example of a metal ion M, and a ligand L as an example, where i is the complexation number :  M + i L → MLi  (A1)  The equilibrium equations all take the following form:  Ki =  [MLi ] [M ][L]i  (A2)  Similarly, in the event a precipitate forms:  MLj ↔ M + j L  (A3)  The equilibrium equation for the formation of the precipitate is as follows, assuming the activity of the solid is unity:  94  K  sp , j  = [M  ][L ] j  (A4)  Equations A3 and A4 can be rearranged as follows:  ln (K i ) = ln[MLi ] − ln[M ] − i ln[L]  (A5)  ln (K sp , j ) = ln[M ] + j ln[L]  (A6)  In the absence of precipitation the left hand side of equation A6 will exceed the right. If precipitation does occur, both sides will be equal. In addition, the system is constrained by mass balances on the M and L components.  [M ]T = [M ] + ∑ [MLi ] + [ML j ]  (A7)  [L]T = [L] + ∑ i[MLi ] + j[ML j ]  (A8)  n  i =1  n  i =1  Thus in total there are n + 2 (no precipitate) or n + 3 (precipitate) non-linear equations in the system. This system can be solved using the Newton-Raphson technique. For a system of equations:  f i (x j ) = 0 ; 1 ≤ i , j ≤ N  (A9)  95  The system of differentials can be written:  N  ∂f  df i = ∑  i dx j   j =1  ∂x j   (A10)  Assuming the increments are small these can be written as differences:  N  ∂f  ∆f i = ∑  i ∆x j   j =1  ∂x j   (A11)  If an initial guess for the values of xj is provided that results in non-zero values for fi, the next guess for xj,new can be determined by adding the resultant ∆ xj vector to xj,old. When this system is written in matrical form it is as follows, where J is the Jacobian matrix:  − f = J ⋅ ∆x  (A12)  For a 3x3 system:  ∂f 1 ∂x1  ∂f 1 ∂x2  ∂f 1 ∂x3  J = ∂f 2 ∂x1 ∂f 2 ∂x 2  ∂f 2 ∂x3  ∂f 3 ∂x1  ∂f 3 ∂x2  (A13)  ∂f 3 ∂x3  ∆x1 ∆x = ∆x 2  (A14)  ∆x3  96  − f1 f = − f2  (A15)  − f3  For this method of speciation calculation, the f vector consists of the equilibrium equations and mass balances for each species, as per equations A6-A8. The ∆x vector corresponds to the concentrations of each species in the system. These concentrations are written in log form, such that x1 = ln[ML ]. The problem is first solved assuming no precipitation, where the Jacobian matrix has the following form (i=3):  J =  1  0  0  −1  −1  0  0  1  0  −1  −2  0  0  0  1  −1  −3  0  exp(ln[ML])  exp(ln[ML2 ])  exp(ln[ML3 ])  exp(ln[M ])  0  0  exp(ln[ML]) 2 exp(ln[ML3 ]) 3 exp(ln[ML3 ]) 0  0  0  0 0  (A16)  exp(ln[L]) 0 0  1  By doing a check against the Ksp it can be determined if a precipitate is forming. If it is, the matrix is as follows:  J=  1  0  0  −1  −1  0  0  1  0  −1  −2  0  0  0  1  −1  −3  0  exp(ln[ML])  exp(ln[ML2 ])  exp(ln[ML3 ]) exp(ln[M ])  exp(ln[ML]) 2 exp(ln[ML3 ]) 3exp(ln[ML3 ]) 0  0  0  0 1  0  exp(ln[MLj ])  j  1  (A17)  exp(ln[L]) j exp(ln[MLj ])  97  Once the system is set up, iteration of equation K and updating of x values continues until the f and ∆x vectors are approximately equal to zero, at which point a solution has been reached.  A.2 Gibbs free energy minimization Another method of calculating aqueous speciation is through Gibbs Free Energy Minimization (GFEM) using Lagrange unspecified multipliers. Considering the same system of one metal ion M, a ligand L, n complexes MLi and a solid precipitate MLj, for each independent equilibrium k the following can be written:  ∆Gk = ∆Gko + RT ln (K k ) = 0  (A18)  ∆Gko = ∑υ lk ∆Glo  (A19)  Where:  l  RT ln (K k ) = RT ∑υ lk ln (a l )  (A20)  l  Where υlk is the stoichiometric factor for species l in equilibrium k, positive for products, negative for reactants.  Defining a new variable, the Lagrange unspecified multiplier, λ`m , such that for each species l:  98  ∆Glo + RT ln (a l ) + ∑ α ml λ`m = 0 m  (A21)  Where α ml is the stoichiometric number of element m in species l. The equilibrium condition is satisfied when:  ∑υ ∑ α lk  k  ml  λm = 0  (A22)  m  For example, for the following complexation reaction:  M + i L → MLi  (A23)  The relevant Gibbs free energy balance is as follows:  ∆Gk = ∆Gko + RT ln (K k )  [  (A24)  ]  ∆Gk = ∑υ lk ∆Glo + RT ln (a l )  (A25)  l o ∆Gk = (∆G ML − ∆G Mo − i∆G Lo ) + RT (ln (a MLi ) − ln (a M ) − i ln (a L )) i  (A26)  Replacing the free energy terms for each species with equation Q yields the following, where the Lagrange multiplier terms eliminate each other:  ∆Gko + RT ln (K k ) + λM` − λ`M + iλ`L − iλ`L = 0  (A27)  99  Hence for this example chemistry the system of equations is made of n + 3 free energy equations, one for each species, and m mass balance equations, one for each component:  ∆Glo + RT ln (a l ) + ∑α ml λ`m = 0  (A28)  m  mm = ∑ α ml ml  (A29)  l  If the unknown concentrations are written in log form, such that x n = ln[MLn ] , while the Lagrange unspecified multipliers are left in standard form x n = λ`n then the Jacobian matrix is set up in a fashion identical to the matrix given above for the equilibrium balancing method. All solute free energy functions should be along the top, component balances in the middle, and precipitate free energy functions at the bottom. This will insure that the Jacobian is never singular, and thus never requires pivoting. Once the Jacobian is formulated and the system of equations is laid out, the problem can be solved using the Newton-Raphson method, using iterative matrix multiplication as outlined above.  The advantage of using the GFEM method is that there is no need to assemble the equilibrium equations of every species and precipitate. Instead only the free energy and components of the species need be known.  100  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0066862/manifest

Comment

Related Items