The mechanisms of pyrite oxidation and leaching: A ...amulsar.com › pdfs › Science-Direct-The-mechanisms-of-pyrite-oxidat… · A.P.Chandra,A.R.Gerson/SurfaceScienceReports65(2010)293–315

Surface Science Reports 65 (2010) 293–315

Contents lists available at ScienceDirect

Surface Science Reports

journal homepage: www.elsevier.com/locate/surfrep

Review

The mechanisms of pyrite oxidation and leaching: A fundamental perspectiveA.P. Chandra, A.R. Gerson ∗

Applied Centre for Structural and Synchrotron Studies, University of South Australia, Mawson Lakes, 5095 Adelaide, Australia

a r t i c l e i n f o a b s t r a c t

Article history:Accepted 18 August 2010editor: W.H. Weinberg

Keywords:PyriteLeachingSurfaceOxidationElectrochemicalPhotoemission electron microscopy (PEEM)Scanning photoelectron microscopy (SPEM)

Pyrite is the earth’s most abundant sulfide mineral. Its frequent undesirable association with minerals ofeconomic value such as sphalerite, chalcopyrite and galena, and preciousmetals such as gold necessitatescostly separation processes such as leaching and flotation. Additionally pyrite oxidation is a majorcontributor to the environmental problem of acid rock drainage. The surface oxidation reactions of pyriteare therefore important both economically and environmentally.

Significant variations in electrical properties resulting from lattice substitution of minor andtrace elements into the lattice structure exist between pyrite from different geographical locations.Furthermore the presence of low coordination surface sites as a result of conchoidal fracture causesa reduction in the band gap at the surface compared to the bulk thus adding further electrochemicalvariability. Given the now general acceptance after decades of research that electrochemistry dominatesthe oxidation process, the geographical location, elemental composition and semi-conductor type (n orp) of pyrite are important considerations.

Aqueous pyrite oxidation results in the production of sulfate and ferrous iron. However otherproducts such as elemental sulfur, polysulfides, hydrogen sulfide, ferric hydroxide, iron oxide and iron(III)oxyhydroxide may also form. Intermediate species such as thiosulfate, sulfite and polythionates are alsoproposed to occur. Oxidation and leach rates are generally influenced by solution Eh, pH, oxidant type andconcentration, hydrodynamics, grain size and surface area in relation to solution volume, temperature andpressure. Of these, solution Eh is most critical as expected for an electrochemically controlled process,and directly correlates with surface area normalised rates. Studies using mixed mineral systems furtherindicate the importance of electrochemical processes during the oxidation process.

Spatially resolved surface characterisation of fresh and reacted pyrite surfaces is needed to identifysite specific chemical processes. Scanning photoelectron microscopy (SPEM) and photoemission electronmicroscopy (PEEM) are two synchrotron based surface spectromicroscopic and microspectroscopictechniques that use XPS- and XANES-imaging to correlate chemistry with topography at a submicronscale. Recent data collected with these two techniques suggests that species are heterogeneouslydistributed on the surface and oxidation to be highly site specific.

© 2010 Elsevier B.V. All rights reserved.

Contents

1. Introduction........................................................................................................................................................................................................................2942. The structure of pyrite .......................................................................................................................................................................................................294

2.1. Chemical structure.................................................................................................................................................................................................2942.2. Crystal structure ....................................................................................................................................................................................................2942.3. Electronic structure ...............................................................................................................................................................................................2952.4. Semiconductor properties of pyrite......................................................................................................................................................................296

3. Pyrite oxidation ..................................................................................................................................................................................................................2973.1. Atmospheric oxidation ..........................................................................................................................................................................................2973.2. Aqueous oxidation .................................................................................................................................................................................................2983.3. Mechanism and kinetics of aqueous oxidation....................................................................................................................................................301

3.3.1. Summary .................................................................................................................................................................................................306

∗ Corresponding author. Tel.: +61 8 8302 3044; fax: +61 8 8302 5545.E-mail address: [email protected] (A.R. Gerson).

0167-5729/$ – see front matter© 2010 Elsevier B.V. All rights reserved.doi:10.1016/j.surfrep.2010.08.003

http://dx.doi.org/10.1016/j.surfrep.2010.08.003

http://www.elsevier.com/locate/surfrep

http://www.elsevier.com/locate/surfrep

mailto:[email protected]

http://dx.doi.org/10.1016/j.surfrep.2010.08.003

294 A.P. Chandra, A.R. Gerson / Surface Science Reports 65 (2010) 293–315

4. Spatially resolved surface characterisation ......................................................................................................................................................................3084.1. Scanning photoelectron microscopy (SPEM) .......................................................................................................................................................3084.2. Photoemission electron microscopy (PEEM) .......................................................................................................................................................3094.3. Topographic artifacts in SPEM and PEEM.............................................................................................................................................................3124.4. Sample preparation ...............................................................................................................................................................................................312

5. Conclusion ..........................................................................................................................................................................................................................312Acknowledgements............................................................................................................................................................................................................314References...........................................................................................................................................................................................................................314

1. Introduction

Pyrite (FeS2, iron disulfide) is the most abundant and wide-spread of the earth’s sulfide minerals and is frequently found inmassive hydrothermal deposits, veins and replacements, igneousrocks and sedimentary beds [1,2]. Pyrite has little economicvalue but is frequently associated with valuable minerals suchas sphalerite, chalcopyrite and galena. Occasionally pyrite mayalso carry dispersions of valuable metals such as gold and issometimes mined for this [3], in addition pyrite is commonly usedfor production of sulfuric acid [4].

Pyrite oxidation is economically important in mineral flotationand leaching, two industrial methodologies used to separate pyritefrom otherminerals of value. Oxidation can impart hydrophobicityor hydrophilicity to pyrite surfaces and hence influences interac-tions with collectors during flotation [5]. Inadvertent flotation ofpyrite leads to reduced concentrate grades, increased superfluousS and Fe contamination resulting in increased smelting costs [1,6].Moreover, pyrite can also contain ‘‘penalty’’ elements such as ar-senic that require costly removal prior to smelting [7].

Oxidation is a major process in the dissolution of pyritesduring leaching [8]. Aqueous oxidation processes also play asignificant role in the acidification of natural water systemsthrough the production of sulfuric acid as a result of naturalweathering of pyritic rocks and shale, and of waste dumps inmining operations [9]. Mining industry treatment costs for acidrock drainage, in the US alone, are over $1 million per day [10].It is also important in biogeochemical cycling of Fe and S, and inthe ecology of Fe- and S-oxidising bacteria [11].

Formulation of a consistent oxidation reaction mechanismshould involve identification of the rate controlling species andrate law formulations through solution based studies, surface stud-ies of fresh and reacted pyrite with detailed species evolution in-formation, and solution speciation and surface structuremodeling.While numerous studies have been conducted, there has been todate little agreement as to the different surface species formed af-ter oxidation, on the rate controlling species, proposed rate laws oreven whether the rate controlling process is chemical or electro-chemical. It is common knowledge that pyrite surfaces (like mostother mineral surfaces) are very heterogeneous however we haveidentified a lack of surface chemical information on a spatially rel-evant scale and as such site specific surface reactions are not suffi-ciently characterised.

We review the structure of pyrite, highlighting the structuraland electronic differences between the surface and bulk. Adetailed discussion is also presented on pyrite oxidation underatmospheric and aqueous conditions with emphasis on leaching,aqueous oxidation kinetics and the nature of the rate controlmechanisms. Finally two synchrotron based spectromicroscopysurface techniques are discussed with reference to some recentlycollected original data. We aim to identify the nature of pyriteoxidation processes from the fundamental studies reviewedand the factors that need to be considered during aqueousoxidation of pyrite. We also aim to highlight the chemical andtopographical heterogeneity of fresh and reacted pyrite surfacesand possible methods of obtaining spatially resolved surfacechemical information.

2. The structure of pyrite

2.1. Chemical structure

FeS2 is composed of a ferrous (Fe2+) cation and an S2−2anion with an ideal Fe:S ratio of 1:2 [4]. Deviations (<1%) fromthis stoichiometric relationship with either a higher cation oranion concentration are frequently reported [7]. Paszkowicz andLeiro [12] using the Rietveld refinement method found the S:Feratio of two different pyrite samples to be 1.978(6) and 2.027(6).Long and Dixon [13] reported the S:Fe ratio of 1.92 for a pyritesample from Mexico. A greater S:Fe ratio is reported by Huet al. [14] and Peng et al. [15]. These deviations are present dueto lattice substitutions of the Fe2+ or the S− ion with atoms ofsimilar radius and charge or net polarity [4]. These minor andtrace elements, listed in Abraitis et al. [7], can introduce significantvariations in the semi-conducting bulk properties of pyrite whichcan directly affect the reactivity of the pyrite surfaces [7,9]. Thedistribution of these impurity elements is not homogeneous [16]therefore site specific electronic variations are likely to exist. This,in turn, can have significant implications for the leaching andflotation behaviour of pyrite samples [17].

2.2. Crystal structure

FeS2 exits in two polymorphic forms, pyrite which is cubic,and marcasite which is orthorhombic [6] of which marcasiteis the less stable and also the less widespread [4]. Pyrite hasa face-centred crystallographic arrangement similar to that ofrock salt, halite (NaCl), with Na+ replaced by Fe2+ and thecentre of the S–S bond being located at the Cl− position [18–21]. Each Fe2+ is coordinated to six S in a distorted octahedralarrangement while each S is coordinated to one S and threeFe2+ in a distorted tetrahedral arrangement. The S2−2 dumbbellsare diagonally oriented and alternate in orientation in eachcrystal layer. This arrangements results in a reduction in thehigh symmetry close-packed NaCl structure [3]. Despite this thepyrite structure is still very dense, approximately 5.02 g/cm3

[4,22] and has a space group of Pa3 [19,23].Apart from the usual cubic {100} morphology, pyrite also

forms as a dodecahedron {210} with pentagonal shaped facesknown as pyritohedron, and octahedral {111} crystals withtriangular faces [4]. The crystal faces are usually striated as aresult of microscopic alternation of {100} and {210} growth [3].Natural pyrite is found as a single morphology or in a variedcombination of these structures [4]. Unlike other minerals ofsimilar structure pyrite shows poor {100} fracture which isfrequently conchoidal [4,21]. Paszkowicz and Leiro [12] reportedcrystallographic data of two different pyrite samples, a unitcell parameter of 5.41784(2) Å and x fractional coordinate of S(free positional parameter) of 0.3848(1) (from Spain) and a of5.41819(2) Å and x of 0.3840(1) (from Russia). These variationswere attributed to deviations from ideal stoichiometry. Whilestandard deviations (provided in parentheses) in values of a andxwere reported by Paszkowicz and Leiro [12], no discussions weremade regarding the contribution of estimated standard deviations

A.P. Chandra, A.R. Gerson / Surface Science Reports 65 (2010) 293–315 295

Table 1Rest potential of some common sulfide minerals determined under standardconditions.Source: Adopted from [25].

Sulfide mineral Rest potential (SHE) V

Pyrite 0.66Chalcopyrite 0.56Sphalerite 0.46Covellite 0.45Bornite 0.42Galena 0.40

to the observed variability in crystallographic parameters ofpyrite from different origins. Further crystallographic data andinteratomic distances are given elsewhere [17,22–24].

2.3. Electronic structure

The electronegativity difference between Fe and S suggestspredominantly covalent bondingwhich is defined by the overlap ofthe 3d valence electrons of Fe and the 3p valence electrons of S [7,22]. Pyrite has a relatively high rest potential of approximately 0.66V [25] making it the most electrochemically inert of the commonsulfide minerals (Table 1).

X-ray photoelectron spectroscopy (XPS), Mössbauer spec-troscopy, ultra-violet photoelectron spectroscopy (UPS), low en-ergy electron diffraction (LEED) and scanning tunneling mi-croscopy (STM) together with first-principles quantum chemicalcalculations (ab initio) have been used to study the electronic struc-ture of pyrites [7,17,22,24,26–33]. The shortest interatomic dis-tance occurs between the two anion pairs with lengths calculatedas dS–S of 2.20 Å [24] and dS–S of 2.14 Å [28]. The lengths betweenthe Fe and S is reported as dFe–S of 2.26 Å [24] and dFe–S of 2.27Å [28]. Unlike other sulfides such as chalcopyrite and sphalerite,where there is no S to S bonding, the short S-S distance in pyritesplits the S 3s orbital into bonding σ and antibonding σ ∗ orbitalswith a gap of approximately 3 eV [24,28]. UsingXPS andMössbauerstudies, van der Heide et al. [26] and Folmer et al. [27] constructeda molecular orbital (MO) diagram of the S2−2 anion showing the 3sand 3p overlap of the S atoms and the subsequent s–p hybridisa-tion. The 3s overlap results in a bonding 3s (σg ) and an antibonding3s (σ ∗

u ) molecular orbital with two electrons in each orbital. The3pz orbitals form an occupied bonding 3pz (σg ) and an unoccupied3pz (σ ∗

u ) molecular orbital while the 3pxy orbitals form occupiedbonding (πu) and antibonding (π∗

g )molecular orbitals. There is alsoextensive splitting of the Fe 3d orbitals by the strong field S ligandsresulting in the formation of triply degenerate t2g and doubly de-generate e∗

g molecular orbitals. Only the t2g molecular orbital is oc-cupied as Fe retains the spin-paired d6 configuration [27]. This t2gmolecular orbital is almost nonbonding in pyrite [26]. Moreover,as both the anion (S2−2 ) and the cation (Fe2+) in the pyrite struc-ture have paired electrons, pyrite is a low-spin diamagnetic com-pound with no magnetic moment [7,27,28,34]. The diamagneticelectronic configurations of the cation and the anion are shown inFig. 1a and b respectively.

Edelbro et al. [28] presented a complete band structure ofFeS2 calculated by a full potential density functional approachand state that the band structure determines the strength ofchemisorption bonds. This was similar to the calculations madeby Cai and Philpott [24]. The structure showed that the valenceband (valence band, spread across 1.25 eV) just below the Fermilevel is mainly composed of Fe t2g states with some anion states.The main bonding band is located just 0.91 eV below the valenceband and consists of S 3p orbitals and some Fe 3d orbitals,spread over 5 eV. Moreover, the t2g orbitals also overlap withempty S 3d orbitals resulting in electron delocalisation [7]. The

conduction band located approximately 0.9–0.95 eV above thevalence band is a mixture of antibonding σ ∗ and the Fe e∗

gstates [17,35]. Folmer et al. [27] showed that the π∗ branch ofthe valence band lies across the Fermi level and possible Fe e∗

gand S π∗

g mixing exists as the π∗ orbital is oriented towards themetal eg orbitals. Furthermore, Oertzen et al. [36] conducted abinitio quantum-chemical calculations of the optical and electronicproperties of pyrite, using density-functional, Hartree–Fock andhybrid functional methods, and compared the resulting bandstructure to the shape of S 1s-edge X-ray absorption spectraof pyrite, measured using synchrotron radiation. Based on theanalysis it was shown that the conduction band consisted of Sp and Fe p and d states. Further structural details can be foundelsewhere [37]. Ferrer et al. [38] cautioned on the use of certaintheoretical approximations for determining the electrical bandgap as they cite this reason for the widely reported band gapvalues. According to themmost approximations used are valid onlyfor parabolic band edges while for pyrite the band edges appearflat. While this is a valid argument, electrical variations betweendifferent samples and even within the same samples can exist, asa result of heterogeneous impurity distributions. As discussed insubsequent paragraphs the apparent disparity between bulk andsurface chemical states, and possibly within surface states can alsobe a contributing factor.

The photoelectron spectra of the pyrite valence band isgenerally characterised by two distinct regions, the outer valenceband and the inner valence band which are separated by aminimum at about 2 eV [30,31]. The outer valence band ischaracterised by a strong peak near 0.6–1 eV while the innervalence band stretches from the minimum to about 20 eV [30].The inner valence band has an almost broad contribution fromthe minimum to about 8–10 eV and a doublet like (of the regionbetween 0 to approximately 10 eV) feature from 10 to 20 eV [30,31].

Using both conventional and synchrotron XPS Nesbitt et al. [30]obtained a series of valence band spectra of the vacuum fracturedpyrite surface that ranged from surface sensitive to bulk sensitive.In order to test previous theoretical calculations of the pyritevalence band, the incident photon energy was carefully variedduring successive measurements and qualitative information onthe relative contributions by the S 3s and 3p and Fe orbitalswas obtained. From this range of valence band spectra, sevenpeaks were identified from 0.8 eV to 16 eV. Two peaks wereidentified in the doublet-like region, at 16 and 13 eV, which werefound to arise entirely from S 3s molecular orbitals as bondingσ and antibonding σ ∗ respectively. There was no evidence tosuggest any sp3 hybridisation of S molecular orbitals, hence itwas concluded that the hybridisation concept cannot be used toexplain tetrahedral S geometry in pyrite. The broad contributionof inner valence band was found to contain four distinguishablepeaks, at 7, 5, 4 and 2.5 eV. The peak at 7 eV was attributed to themixing of S 3p atomic orbitals to form bonding σ molecular orbitalresponsible for the S-S dimer, while the peak at 5 eVwas attributedto contributions from S 3p derived orbitals possibly from the S-S π bond. Negligible Fe 3d contributions to the 7 and 5 eV peakswere also found. The peak at 4 eV results from mixing of Fe 3d–S3p molecular orbitals and has a more S 3p character than Fe 3d.This represents the bonding (σ ) molecular orbitals of Fe-S bonds.The 2.5 eV peak also results from mixing of Fe 3d-S 3p molecularorbitals however it has amore Fe 3d character than S 3p and resultsmostly from Fe-S π bonding. The single peak of the outer valenceband, which is also the most prominent valence band peak, atapproximately 0.8 eV, is attributed to result largely from Fe 3d.The peak position and shape was found to vary slightly in spectrataken at different incident photon energies, and it was concludedthat there were additional contributions from S 3p and Fe 3d


a b

Fig. 1. Molecular orbital diagram showing diamagnetic configuration of (a) Fe in pyrite, bulk and surface (b) S2−2 anion.Source: Fig. 1(a) redrawn from Ref. [23].

surface states. The Fe-S π∗ bonding was suggested to contributeto the high binding energy side of this peak at about 1 eV whilethe Fe 3d of surface Fe atoms was suggested to contribute to thelower binding energy side of this peak at about 0.5 eV. The surfaceFe contributions may arise from Fe 3deg (non-bonding orbitals)dangling bonds as a result of Fe-S bond fracture.

The most commonly found cubic pyrite morphology, termi-nates with the {100} surface while pyritohedral and octahedralmorphologies terminate with {210} and {111} surfaces respec-tively [4]. Rare {110} surface terminations are also found [23]. Allof these surfaces are of lower coordination as compared to the bulkstructure as bonds are fractured during cleavage. The {100} surfaceis frequently found to have five-fold coordinated surface Fe due tothe loss of S2−2 [17,23,39]. This leads to a square pyramidal field(point group C4v) around the Fe atom resulting in the loss of de-generacy of the Fe e∗

g orbitals [24]. Thus, the band gap reduces fromabout 0.9 eV in the bulk to only approximately 0.16 eV at the {100}surface (Fig. 1a) inducing a more metallic character in the surfaceas compared to the semi-conducting character of the bulk [17,23].

The {110} surface contains Fe of even lower coordination (<5)with four-fold coordination being the most frequent [23]. As seenin Fig. 1a, this leads to a further reduction in band gap and resultsin the {110} surface (<5 coordination) becoming spin polarised(paramagnetic) unlike the 5-fold coordinated and bulk Fe. Eventhough the {110} surface is rare, the 4-fold coordination typical ofthis surface also occurs on parts of the more common {100} pyritesurface. The {100} surface is the most stable of the terminatingsurfaces however, cleavage results in a high density of defectsand imperfections (steps and kinks) which have low coordination(<5) Fe sites similar to the {110} surface [23,24,39]. The {111}and {210} surface have also been shown to terminate with lowcoordinated (≤5) Fe atoms with similar structures to those shownin Fig. 1a [40].

The loss of coordination at the surfaces results in higherdangling bond density, making such sites highly reactive. Surfacestructures undergo considerable relaxation to stabilise these lowcoordinated sites [17,40]. This tends to shorten the S-S and Fe-S bond lengths as compared to the bulk [24]. Hung et al. [23],through their density functional calculations, suggested that lowcoordination sites (<5) tend to be spin polarised (while sites ofnormal coordination are spin neutral) and that species such astriplet molecular oxygen (O2) which is paramagnetic will be moreinclined to react with these low coordination defect sites. Qiuet al. [17] propose that during oxidation reactions the transfer ofelectrons will be much faster as a result of the reduced band gap

and enhancedmetallic character andwill occur preferentially fromthe Fe rather than surface S species, which may result from bondcleavage during fracture (refer Section 3.1). This is in agreementwith previous studies by Rosso et al. [33] who revealed for thefirst time the surface electronic heterogeneity of UHV fracturedsurfaces using STM microscopy and spectroscopy together withLEED, UPS and ab initio calculations. It was concluded that surfaceredox processes are initiated by first quenching of high energydangling bonds (at Fe sites) and leading to the formation of newsurface species.

2.4. Semiconductor properties of pyrite

Pyrite is a potential photovoltaic absorber material for solarcells due to its high electron mobility (230 cm2 V−1 S−1) and highoptical absorption coefficient (α > 6.0×105 cm−1 for hv> 1.3 eV;[38,39,41]. However, considerable variations exist in the semi-conductor properties of natural pyrites which affect the physic-ochemical processes of pyrite dissolution [42]. Abraitis et al. [7]comprehensively reviewed the semi-conducting properties ofpyrite and found that reported conductivities vary by four ordersof magnitude. Depending on geological conditions, natural pyritecan exist as either a n-type semiconductor or as a p-type semicon-ductor [43,44]. Pyrite formed at relatively high temperatures nor-mally exhibits n-type character while pyrite formed at relativelylow temperatures are p-type [7,22].

Abraitis et al. [7] calculated the mean conductivities of n- andp-type pyrites from the data they reviewed. Their calculationsclearly illustrate that n-type pyrites have higher conductivitieswith a mean of 56.8 (�cm)−1 while p-type pyrites are of lowerconductivity with a mean of 0.53 (�cm)−1. The variations insemi-conductor properties arise from deviations in stoichiometryand the presence of trace elements [9,22]. Trace elements caneither have electron donating (n-type) or electron accepting (p-type) properties and hence impart the same to pyrites. Pyriteshigh in arsenic are found to be p-type semi-conductors whilethose low in As are n-type [7,44]. Pyrites high in cobalt have alsobeen found to behave as a n-type semiconductor [44]. Similarly,pyrites with S:Fe stoichiometric ratios less than 2 are usually n-type while those above 2 are p-type. Buckely andWoods [5] foundevidence of the coexistence of S-deficient and Fe-deficient regionson abraded pyrite surfaces. Moreover, according to Lowson [22],Rimstidt andVaughan [9] andAbraitis et al. [7] natural pyrites havebeen reported to contain alternating n and p type properties. Infact, overall n or p behaviour of pyrites may result from a net of nor p properties.


Using n-type pyrites as microelectrodes, Wei and Osseo-Asare [42] investigated anodic and cathodic dissolution in 1 MHNO3 solution at a given potential. The dissolution rate was seento be much faster under anodic current compared to cathodiccurrent. Illumination of the pyrite microelectrode was found toincrease anodic dissolution with no effect on cathodic dissolution.Hence, it was concluded that n-type pyrites dissolve anodicallythrough ahole transfer (valence band) pathway.Moreover, Abraitiset al. [7] add that in mixed sulfide mineral pulps (without kineticconsiderations), pyrite with a comparatively higher rest potentialwill be cathodic while more reactive sulfides will be anodic.However, the resulting rates of preferential dissolution of theanodic sulfides will be dependent on variations in the pyrite’s restpotential resulting from impurities or semiconductor type.

3. Pyrite oxidation

3.1. Atmospheric oxidation

Oxidation of pyrite surfaces may occur upon exposure to atmo-spheric O2 and water [5]. The oxidation layer formed can passivateagainst further oxidation, and hence determine subsequent aque-ous phase oxidation processes [45]. The atmospheric reactivity ofa pyrite surface is dependant on the different surface species andthe abundance of such species on a freshly fractured pyrite surface.

XPS studies of vacuum fractured pyrite surfaces have shownthat three different S species are present, disulfide (S2−2 ), monosul-fide (S2−) and polysulfide (S2−n , n > 2) [11,46–49]. The main disul-fide peak for pyrite (fully coordinated S2−2 ) occurs at the S 2p3/2binding energy range of 162.3–162.7 eV [5,46–51]. During pyritefracture, which is conchoidal, the rupture of the Fe-S bonds cre-ates surface S2−2 where the outermost (surface layer) S in S2−2 is ina decreased 3-fold coordination of two bonds to Fe and one bondto S [46,49]. The reduction in coordination to Fe and the result-ing decrease in attractive electrostatic potential cause a negativebinding energy shift of about 0.7 eV compared to fully coordinateddisulfide [46,49]. Surface S2−2 has an S 2p3/2 binding energy of ap-proximately 162.0 eV [46,49].

On the basis of bonding energy considerations Nesbitt et al. [46]suggested that S-S bonding in pyrite is weaker than Fe-S bondingand that therefore during pyrite fracture, appreciable amountsof S-S bonds are broken. The breaking of S-S bonds gives riseto one S− species on each fracture face. According to Nesbittet al. [46] this highly reactive mononuclear species undergoesrelaxation through transfer of an electron from the Fe2+ to formone S2− and one ferric iron (Fe3+). The existence of Fe3+ onthe surface has been confirmed through synchrotron XPS andquantum chemical calculations [52,53]. It is also possible that theS− species can be stabilised through transfer of an electron fromanother nearby S− species, resulting in the formation of one S2−one S0. Nesbitt et al. [46] suggest that the elemental sulfur (S0)produced can undergo further reactions to form the polysulfidespecies S2−3 . The monosulfide, S2−, S 2p3/2 binding energy occursaround 1.4 eV below the bulk S2−2 binding energy in the rangeof 161.3–161.8 eV [5,46–49]. The intensity ratio (XPS) of surfaceFe3+ to Fe2+ was found to be 1.7:1 in a vacuum fractured pyritesurface [49], indicating the presence of surface S2−. Moreover,any surface loss of S atoms due to cleavage reduces the idealsurface S:Fe stoichiometry of 2:1 and this can add to the electricaldiscrepancies between the surface and bulk structures. Therefore,in addition to Fe2+ and fully coordinated (4-fold S)S2−2 , freshlyfractured (vacuum) pyrite surfaces also contain surface (3-foldS) S2−2 , Fe3+, S2−, and S0 or S2−n type species. Leiro et al. [54]however report the existence of an additional surface S state (inaddition to surface S2−2 and S2−) shifted by 2.0 eV to a lower

binding energy of 160.8 eV. They conducted synchrotron S 2p XPSmeasurements using 3 different incident photon energies, 210, 350and 780 eV, while cooling the sample with liquid N2 to reducephonon broadening. It was found that the extra component at160.8 eV reduced in intensity with increasing incident photonenergy. Estimated effective attenuation lengths (λ) were 4 ± 1Å for 210 eV, 8 ± 2 Å for 350 eV and 13 ± 1 Å for 780 eV photonenergies at the take-off angle of 45°. It was concluded that thisextra component was due to S2− species that occurred only nearsteps and kinks between terraces on a fractured pyrite surface. Thepyrite surface is therefore highly heterogeneouswith each of thesespecies having different reactivity which may influence the natureand direction of initial oxidation.

Numerous studies have been conducted to identify oxida-tion products formed after exposure to atmospheric gases. Syn-chrotron and conventional XPS have been widely used to examineshifts in the binding energy of S 2p, Fe 2p and O 1s electrons[5,45,48,49,55,56]. In addition synchrotron-based X-ray absorp-tion spectroscopy examination of the S 2p3/2 and 2p1/2, Fe 2p3/2and 2p1/2 Fe and O 1s absorption edges [45] along with DRIFTS(diffuse reflectance Fourier transform IR), UV spectroscopy andhigh performance liquid chromatography [55] have all been usedto study pyrite surface oxidation. Sulfate has been shown to bea major surface oxidation product with the possible presence ofFe oxy-hydroxide species [22,49] however controversy still existsas to the presence of elemental sulfur (S0) and polysulfides (S2−n )[5,49,55].

Pyrite oxidation begins within minutes of exposure to theatmosphere, commencing with the oxidation of S2− species.Schaufuß et al. [48] studied the reactivity of S2−, S2−2(surface) andS2−2(bulk) species by exposing pristine fractured pyrite surfaces to theatmosphere for various times. Nearly 80% of the S2− was oxidisedto sulfate within 1 min exposure to the atmosphere. Buckleyand Woods [5] also found evidence of Fe2+ sulfate productionwithin a fewminutes of atmospheric exposure. Schaufuß et al. [48]concluded that the S2− species is the most reactive species withan initial oxidation rate of 0.77 min−1, with the S2−2(surface) speciesbeing the secondmost reactive species while the bulk coordinatedS2−2(bulk) is the least reactive. Buckley and Woods [5] also exposedpyrite surfaces to the atmosphere for just a few seconds but foundno evidence of sulfate formation, although the O 1s spectrum (XPS)revealed appreciable amounts of oxygen containing species on thesurfacewhichwas attributed to the presence of chemisorbedwaterand/or hydroxide. Oxidation to sulfate also occurs within minuteseven if the pyrite is exposed to limited amounts of atmosphericgases [48] however Schaufuß et al. [48] add that under suchconditions intermediate sulfur and oxy-sulfur species dominatesulfate formation.

As stated previously sulfate is the main oxidation product ofprolonged atmospheric exposure. de Donato et al. [55] foundsulfate to represent 36%–39% of all surface oxidation products.Todd et al. [45] and de Donato et al. [55] identified that this to belargely present as Fe3+ sulfate, Fe2(SO4)3. Schaufuß et al. [48] andTodd et al. [45] found Fe3+ oxy-hydroxide (FeOOH) to be the mainoxidation product after sulfate and to extend below the uppermostFe layer. This view is however not shared by de Donato et al. [55]who found Fe3+ hydroxide, Fe(OH)3, and Fe2+ oxide, FeO, to bemuchmore prolific than FeOOH. Buckley andWoods [5] also foundFe oxide on air-exposed pyrite surfaces which they expected tobe hydrated. Furthermore with the appearance of possible Fe-oxyspecies on the surface they theorised that a metal-deficient sulfidemust be forming in addition to the sulfate, given that the ratio ofS to Fe is nearly 2. This conclusion was made on the basis that noelemental S signal was evident from S 2p spectra and no gaseousSO2 formed. Although not identified by Buckley and Woods [5]this Fe-deficient compound could have been polysulfide. Using


Fig. 2. Atmospheric pyrite oxidation mechanism.Source: Redrawn from [59].

synchrotron XPS Schaufuß et al. [48] observed significant increasein spectral intensity between 163 and 168 eV binding energyafter prolonged atmospheric exposure, which was attributed tothe occurrence of polysulfides and intermediate oxysulfur speciessuch as sulfite (SO2−

3 ), thiosulfate (S2O2−3 ) and a new stable S-OH

species (164.3 eV). In addition, de Donato et al. [55] not only foundevidence of polysulfides but also deduced this species to exist asa metastable species with S5S2− structure. Moreover, they alsofound evidence of elemental sulfur (S0) which they suggested wasin equilibrium with the polysulfide.

Jerz and Rimstidt [57] investigated the rate of pyrite oxidationin moist air, and found that the reaction rate decreased with time.This is consistent with the observation by Schaufuß et al. [49],where the reaction rate of S2−2(surface) was also seen to decreasewith time. Jerz and Rimstidt [57] linked the reduced reaction rateto the development of a thin layer of Fe2+ sulfate plus sulfuricacid solution on the pyrite surface with the latter being dueprimarily to the hygroscopic nature of Fe2+ sulfate. Itwas proposedthat this layer retards oxygen transport and hence induces apossible change from chemical rate control to transport controldetermined by diffusion (solubility) of oxygen through this film.Such a solution film has not been reported by other authors;however possible passivation by build up of oxidation productswas acknowledged [45,58]. Jerz andRimstidt [57] suggest that Fe2+sulfate under certain conditions may precipitate in cracks in pyriteand wedge apart the sample leading to physical disaggregation.

Using data from XPS, UPS (ultraviolet photoelectron spec-troscopy), STM (scanning tunnelling microscopy) and Monte Carlosimulation studies of atmospherically oxidised pyrite, Egglestonet al. [59] proposed an air oxidation mechanism involving surfacecycling of Fe2+ and Fe3+ along borders of oxidised and unoxidisedareas. Patches of Fe3+ hydroxide or oxide products form upon ini-tial oxidation and these Fe3+ products serve as a conduit for elec-tron transfer from the pyrite surface to molecular oxygen. This in-volves the transfer of electrons from the pyrite valence band to theoxide conduction band, which effectively is transfer of an electronfrom pyrite Fe2+ to oxide Fe3+. The process takes place preferen-tially from pyrite Fe2+ adjacent to the oxide. Electron transfer thentakes place between oxide Fe2+ to O2 resulting in the formationof O−

2 species. Electron transfer rate calculations estimate that thetransfer of electrons from oxide Fe2+ is nine orders of magnitudefaster than the transfer from pyrite Fe2+ [59]. The O−

2 species thenformsH2Ovia the formation of highly oxidising intermediates,•OHand H2O2. Fig. 2 shows the diagrammatic representation of themechanism proposed by Eggleston et al. [59] which includes anouter-sphere sulfur oxidation by an H2O film adsorbed from theatmosphere. The mechanism shows that polysulfide formation oc-curs separately from sulfate formation and that therefore polysul-fide is not an intermediate to sulfate formation.

Sulfur (surface) oxidation to sulfate proceeds through theformation of an intermediate S-OH species followed by the

formation of thiosulfate, sulfite and finally sulfate [49]. The bulkdisulfide is not directly oxidised, however there may be transferof electrons from the bulk disulfide to surface Fe3+ [49]. Thisleads to the formation of electron-deficient disulfide which mayrearrange to formpolysulfides. Continued oxidationwill also resultin rupture of further Fe-S and S-S bonds and formation of resultingsurface species. S oxidation also leads to the formation of patches(‘‘islands’’) of productswith in surrounding unreacted regions [48].

Schaufuß et al. [49,48] also provide an explanation for theinitial formation of Fe3+ oxides on fresh pyrite surfaces, similarto that suggested in the Eggleston et al. [59] model. The surfacemonosulfide attached to a Fe3+ site (resulting from e− transfer toS−) readily oxidises into sulfate. The sulfate is then displaced bycompetitive adsorption of H2O, O2 and OH−. An electron is thentransferred from an adjacent Fe2+ (through attached S-bridge) tothe attached O2 resulting in the formation of O2− species. A secondFe3+ adjacent to the first results and Fe3+ oxide propagationoccurs. Further electron transfer may rupture Fe-S bonds at theinitial Fe3+ site resulting in the formation of FeOOH.

Becker et al. [60] provides a mechanism of adsorption andreaction of surfaces species through what they call the ‘‘proximityeffect’’. Since pyrite is a semi-conductor and its surface electronicproperties are affected by defect sites (steps and kinks) andimpurities, surface reactions at different sites are coupled throughelectron transfer. The proximity effect suggests that the oxidationof one site near terraces renders the nearest neighbor moresusceptible to oxidative attack in comparison to some site furtheraway. This leads to the enlarging of the existing oxidation patchesrather than the creation of new ones.

Information as to the correlation between topography and sur-face chemistry would benefit the development and substantiationof the proposed mechanisms however there is currently a lack ofsuch a data set.

3.2. Aqueous oxidation

Studies have proposed that aqueous oxidation may involvechemical, electrochemical or bacterially catalysed pathways[1,9,22,61]. The latter normally involves catalysis of an oxidationreaction step by Thiobacillus bacteria such as T. ferrooxidans, how-ever we do not review this here as it is not within the scope of thisarticle.

The aqueous oxidation of pyrite is generally described by thefollowing overall stoichiometric chemical reactions which wereoriginally characterised by Garrels and Thompson [62] and Singerand Stumm [63].

FeS2 +72O2 (aq) + H2O → Fe2+ + 2SO2−

4 + 2H+ (1)

Fe2+ +14O2 (aq) + H+

→ Fe3+ +12H2O (2)

FeS2 + 14Fe3+ + 8H2O → 15Fe2+ + 2SO2−4 + 16H+. (3)

According toMoses andHerman [64] the pyrite oxidation reactionsare far from equilibrium with essentially no reverse rate.

O2 and Fe3+ have been recognised as the two most importantoxidants for pyrite oxidation [1,11,61,64–66]. Moses et al. [11]studied rates of pyrite oxidation at 22–25 °C, in oxygen saturated(DO — dissolved O2) and Fe3+ saturated solutions with initialpH values from 2 to 9, by monitoring the increase in eithersulfate or total S concentrations over the oxidation period. InFe3+ saturated experiments, an initial rapid increase in sulfateconcentration was observed at all pH values (2–9) which slowedconsiderably after a relatively short period of time. A similarobservation was made by Moses and Herman [64] in experimentsconducted at circumneutral pH with only Fe3+ addition, however,


linear sulfate production was seen to resume after O2 purging. Theexperiments conducted with DO saturated solutions, [11] on theother hand showed linearity throughout the oxidation period andin addition to sulfate, a number of different sulfoxy intermediates(SO2−

3 , S2O2−3 , SnO2−

6 ) were observed to be formed during the DOexperimentswhichwere absent in the Fe3+ saturated experimentsespecially at low pH. McKibben and Barnes [65], in a similarexperiment at 30 °C and low pH found sulfate to be the dominantproduct along with tetra-, penta- and hexa-thionates. It wasfurther shown that no Fe3+ production occurs (by Eq. (2)) duringshort term (1–4 days) DO saturated oxidation of pyrite at lowpH [65] while at pH 7 no Fe2+ is evident throughout the oxidationprocess [64].

An investigation byMoses et al. [11] of overall rates showed thatrates due to the presence of Fe3+ were two orders of magnitudehigher than those due to DO at low pH. Even at higher pH,where Fe3+ solubility diminishes, rates were almost an order ofmagnitude higher than the rates observed in DO saturated media.Therefore, with the lack of observed intermediates, it was apparentthat Fe3+ is a more aggressive and effective than O2 for pyriteoxidation. This is also true even at circumneutral and higherpH where the solubility of Fe3+ diminishes [64]. However, thepresence of DO is critical to sustain the oxidation process. Thereduced oxidation rate observed in Fe3+ saturated experimentsoccurred as a result of Fe3+ depletion [11]. It was seen that aqueousFe3+ made available (through Fe(III)-oxyhydroxides) to pyrite wasless than what could be consumed. Hence, O2 or DO is needed toreplenish the diminishing Fe3+ through Fe3+/Fe2+ cycling.

Moses et al. [11] postulated that since O2 in the ground stateis paramagnetic (two unpaired electrons, triplet), its reaction withdiamagnetic pyrite (no unpaired electrons) would not be possibledue to spin restrictions. However, asmentionedpreviously, densityfunctional calculations byHung et al. [23] have shown the presenceof low coordination (<5) sites on pyrite surfaces which can haveunpaired electrons and may become sites of preferential attackby O2. Fe3+ is also a paramagnetic species however it has beenproposed that it is able to react readily with pyrite surfaces asit is usually complexed to a maximum of six diamagnetic H2Omolecules in solution. It is suggested that the resulting aquo-Fe3+ complex can react with a diamagnetic surface by transferringhydroxyl radical from its complexed water molecules to thesurface [11]. This proposal by Moses et al. [11] does not explainhow a paramagnetic Fe3+ is able to react with diamagnetic waterother than by dipole interaction. An alternate explanation isthat the Fe3+ also reacts with paramagnetic sites on the pyritesurface but that the activation energy required for this reaction,in comparison to that required to break O-O bonds is significantlylower and hence the reaction proceeds much more readily.

In the experiments conducted by Moses et al. [11], solutionpH was seen to decrease in DO saturated experiments whilesurprisingly it increased slightly in experiments saturated withFe3+. Moreover, only a slight pH rate dependency was observedfor DO saturated experiments while McKibben and Barnes [65]found no pH rate dependency in the pH range 2–4. The presenceof intermediates seen during the DO saturated experiments washowever pH-dependent with most appearing above pH 3.9 at highstirring speeds. The presence of intermediates was described bythe Wackenroder reaction (Eq. (4)) with reaction balance shiftingleft above pH 7 and right below 7 [11].

SnO2−6 + S2O2−

3 ↔ Sn+1O2−6 + SO2−

3 . (4)

In addition to Fe3+ and O2, McKibben and Barnes [65] alsoinvestigated pyrite oxidation by hydrogen peroxide at 30 °C andlow pH. Pyrite oxidation by hydrogen peroxide can be representedby Eq. (5).

FeS2 +152

H2O2 → Fe3+ + 2SO2−4 + H+

+ 7H2O. (5)

Fe3+ and sulfate were the only two detectable species with nometastable sulfur oxyanions detected. Fe2+ was only detectedinitially while some Fe hydroxide precipitates were found towardsthe end of the experiments. No pH dependency was observedfor the range of pH 2–4. However the initial rate of peroxideconsumption was nearly double the rate of initial total Feproduction, implying significant catalytic peroxide decompositionby Fe3+.

SEM studies have indicated oxidant attack on pyrite surfaces atonly specific sites of high surface energy [65]. Thus, it appears that‘‘effective’’ or ‘‘reactive’’ surface area is more significant than totalsurface area in determining rates of reaction. Such reactive sitesare related to the grain edges and corners, defects, solid and fluidinclusion pits, cleavages and fractures. The preponderance of thesefeatures and their geometry may significantly vary from grain tograin, introducing variations in the observed rates of reaction onthis spatial scale.

Atomic O reacted with pyritic S and subsequent prod-ucts is believed to originate mostly from water molecules[64,67]. The identification of the source of O can provide valu-able insights into the actual mechanism of aqueous pyrite oxida-tion [66]. Reedy et al. [67] showed that the majority of the sul-fate formed (90%) derived all four O atoms from water ratherthan O2 (under conditions of atmospheric O2 with and withoutadded Fe3+, at pH 1 and 20 °C or 70 °C) using 18O2 and H16

2 O andthat the major product formed under all conditions was S16O2−

4 .Significant amounts of two other isotopomers, S16O2−

318O2− and

S16O2−2

18O2−2 , were also formed when Fe3+ was not added while

only S16O2−3

18O2− was formed as a minor product when Fe3+ wasadded. This was felt to strengthen the notion that Fe3+ is thedominant oxidant. According to Holmes and Crundwell [66], thefact that O atoms are mostly derived from water molecules dur-ing the oxidation process, points directly to an electrochemicalmechanism of reaction, where electrochemical reaction steps atthemineral-solution interface control the dissolution rate of pyrite.The detection of significant levels of two sulfate isotopomers mayalso indicate mechanisms involving different intermediate sulfoxyspecies [67].

Moreover, the Fe(III)-O and Fe hydroxides found on thesurface are the result of the oxidation processes and not fromprecipitation of dissolved Fe from solution. Fe hydroxides andoxides have previously been found not only on pyrite surfaces fromaqueous oxidation [5,45,50,68–70] but also on surfaces exposed toatmospheric gases [45,47,49,51,55,59,71]. Such Fe species on thesurface are suggested to occur as an accumulation of oxidationproducts or as part of the oxidation mechanism. From Fe 2p XPSanalysis of air-exposed pyrite, Eggleston et al. [59] concludedthat hematite-like oxide or hydroxide type oxidation productsform in patches on the surface. Such products promote electrontransfer process from the valence band of pyrite to the conductionband on the oxide/hydroxide products and therefore aid inelectrochemical oxidation of pyrite. Nesbitt and Muir [69] alsomade a similar observation on naturally weathered pyrite samplesfrom mine dumps. Fe 2p and O 1s XPS were used to confirm theexistence of goethite or hematite like products on tarnished pyritesurfaces. It was concluded that such Fe(III)-oxide or oxyhydroxideproducts represented oxidant reservoirs that helped in oxidation.Fe-hydroxy products have however been found to effectivelypassivate chalcopyrite surfaces during bioleaching [72]. Proposedelectrochemical oxidation mechanisms of pyrite also suggestthe occurrence of Fe3+ oxide/hydroxide as one possible species(possibly cathodic) involved in electron transfer steps [9,59,70,73].The Fe(III)-O and hydroxide/oxide type products seen on thesamples leachedunder different conditions are therefore as a resultof the direct oxidation process and not as a result of precipitation of


dissolved Fe from solution. Precipitation of dissolved Fe on pyritesurfaces has been shown to occur only under neutral to alkalineconditions, where a thick coating of goethite retards oxidation byblocking oxidant diffusion [1,22,45,74].

Pyrite leaching has been studied under conditions of varyingpH, Eh, oxidant (Fe3+ and O2), stirring speeds, grain size andtemperatures [11,13,14,61,64–66,75,76]. These experiments havegenerally shown that the addition of oxidants, Fe3+ and O2,significantly increases the rate of pyrite dissolution. King andLewis [75] found that O2 partial pressures up to 120 psig and initialFe3+ concentrations up to 0.5 M had a significant influence, whilerates at ambient PO2 were similar to rates under N2. Moreover,it was also established that the addition of both oxidants yieldedbetter results than just one oxidant alone [11,65,66,75].

King and Lewis [75] further found that increased temperature(80–100 °C) also positively affected pyrite dissolution whilestirring speeds above 680 rpm had no effect. Long and Dixon [13]also studied pyrite leaching at stirring speeds of 650, 800 and950 rpm. They found that maintaining or increasing speeds above800 rpm had no significant effect on the initial rate of dissolution.Apart from the effect on rates, speeds below 800 rpm were idealto minimise Fe-species precipitation through solution splashing.Long and Dixon [13] also established that increasing temperaturehad a significant influence on the rate of dissolution. They foundthat at 230 °C nearly all the pyrite dissolved within 20 min. Kingand Lewis [75] achieved a 90% conversion in 4 h at 100 °C, with120 psig O2, 0.5 M Fe3+ and 20 g/L pyrite loading.

To a lesser extent, the acidic anions in solutions can also beexerting an influence on leach rates. Previous studies of aqueousoxidation of pyrite by H2O2 with different leach media found oxi-dation rates to vary according to HClO4 < HCl < H2SO4 [77–80]. Itwas found that increasing HCl or H2SO4 concentrations in theirrespective leaches had a negative effect on oxidation rates [77,80]. Additions of Cl− or SO2−

4 had a similar effect. It wassuggested that Cl− and SO2−

4 adsorb onto the pyrite surfaceand inhibit the access of oxidants. Increasing HClO4 concen-tration or addition of ClO−

4 had no effect on the oxidationrates, while addition of SO2−

4 in HClO4 leach solutions de-creased oxidation rates [79]. Cl 2p XPS measurements of pyritesamples leached in Cl− solutions with Fe3+ oxidant showedevidence of Cl− adsorption on to the leached surfaces [81]. How-ever, Cl− intensity was found to be independent of solution Cl−concentrations. In addition limited SO2−

4 was found (through XPS)on pyrite surface in SO2−

4 rich leach solutions which was also in-dependent of SO2−

4 solution concentrations. Studies by Lehmannet al. [82] have suggested a role of Cl− ions in inhibiting the depo-sition of S0 and S2−n on reacting pyrite surfaces and also prevent-ing the buildup of passive Fe hydroxy/oxide coatings. Using ring-disk (made of pyrite) voltammetry and in-situ Raman spectroelec-trochemistry they propose that Cl− promotes oxidation of thio-sulfate intermediate into tetrathionate which readily dissolves. Inabsence of Cl− (experiment conducted in H2SO4 solution) thiosul-fate undergoes acid decomposition to form S0 which accumulateson the surface. Furthermore, Cl− ions are aggressive anions andreadily adsorbs on to the surface [82]. Being a strong Lewis base,Cl− is able to replace hydroxyl ions or water molecules in the hy-drated Fe hydroxy/oxide surface layers. This results in the forma-tion of Fe-chloride complexes which readily dissolve in solution.Lehmann et al. [82] further suggest that the adsorption of Cl− mayalso inhibit pyrite oxidation by blocking reactive sites for Fe3+ ad-sorption. It may however be argued that the removal of S0, S2−n andFe oxyhydroxides/oxides from the surface by Cl− may prevent theinhibition of oxidation itself, by allowing continued diffusion of theoxidants to the surface and removal of products into the solutionand may as such be beneficial to oxidation rate. Beneficial effects

of Cl− has been noted in chalcopyrite leaching. It has been shownthat leaching in Cl− media has a faster rate with moremetallic dis-solution compared to leaching conducted in SO2−

4 media [83–85].Inhibition of pyrite oxidation by Cl− has not been widely noted inthe literature. In fact experiments conducted by Williamson andRimstidt [8] did not show any appreciable difference in the reac-tion rates of pyrite oxidation conducted in Cl− and SO2−

4 solutions.Investigations of electrochemical properties (such as rate of estab-lishment and value of corrosion potential, Tafel’s slope, corrosioncurrent and electrodic order) of pyrite oxidation also found negli-gible difference upon anodic oxidation in HCl, H2SO4 or HClO4 so-lutions [78,86].

Furthermore, the formation of different complexes in differentleach media and their ability to act as an oxidant could alsobe a factor causing the observed variation in the leach rates.The acidic anions in solutions tend to complex with free Fe3+

and limit its availability for pyrite oxidation. Sasaki et al. [81]studied the effect of anionic ligands on the oxidation of pyriteby Fe3+ ions in acidic solutions where dissolution experimentswere conducted in N2 purged and uncontrolled (pH and Eh)solutions, with varied concentrations of different ligands at roomtemperature. It was found that ligands suppressed pyrite oxidationin the order Cl− < SO2−

4 ≪ PO3−4 ≈ C2O2−

4 . This ligand behaviourwas explained by comparing the standard redox potential (E°) ofdifferent Fe3+ complexes formed during the dissolution reactionsand the potential of dominant complexes to act as oxidants.Generally, a more positive E° signifies a greater tendency to acceptelectrons and act as oxidants. Oxidationwas found to be extremelylow in oxalate and phosphate anion solutions. In oxalate solutionsthis was due to a sharp decrease in complexes with higher (morepositive) E° and emergence of complexes with lower E°. In PO3−

4solutions it was due to the formation of colloidal or polynuclearFe3+ species with further suppression through deposition ofsuch colloidal species on the pyrite surface. The dissolution wasrelatively faster in SO2−

4 and Cl− solutions due to the formationof complexes with higher E° than those formed in either PO3−

4or C2O2−

4 solutions. Cl− solutions caused less suppression thanSO2−

4 solutions due to formation of Cl− complexes with higher E°compared to SO2−

4 complexes.The exact effect of acidic anions on the aqueous oxidation

rate remains poorly understood, in part due to other morepredominant factors such as solution Eh not being controlledduring experiments, which may mask effects due to anions. Thereare contrasting views on the effect of Cl−, whether it has a positiveor negative effect on rates and the actual mechanism involved isunknown. Further studies in this area are needed to resolve theseissues.

Moses and Herman [64] pointed out that the reaction progressvariable (RPV) is a critical factor for judging the pyrite dissolutionprocess. Most studies have used either different forms of S or Fe asthe RPV or a combination of these. Sulfate has been widely usedas the RPV in many studies [11,64–66,87] and so has the sum ofaqueous S species [11,65]. The emergence of sulfoxy intermediates(thiosulfate and polythionates) may reduce the accuracy of sulfateas a the RPV, as accumulation of sulfate may not reflect the rateof pyrite oxidation [64]. Under such conditions the sum of S mayprove to be a better RPV [65]. However, under certain conditions(such as low pH or anoxic Fe3+) where little or no intermediatesare evident, ΣS = [SO2−

4 ] [11]. Total Fe, Fe3+ concentrations andFe3+/Fe2+ ratios have also been used as the RPV [13,65,75]. Thereare however concerns that some Fe (Fe3+) may be lost from thesolution due to precipitation as oxyhydroxides [64].


3.3. Mechanism and kinetics of aqueous oxidation

Formulating an accurate mechanism of reaction generally in-volves successive collection of chemical data such as species con-trolling the reaction rate, rate law formulation (as a function ofpotential, reagent concentration, isotopic constitution of reagentsand temperature), reaction products and intermediate identifica-tion, tracing of product/intermediate speciation, activation energy,bond cleavage and the influence of free energy on the rate [8,88]. While numerous mechanisms of abiotic aqueous oxidation ofpyrite have been proposed, it is very difficult to prove any specificmechanism with a high degree of certainty. Table 2 provides ki-netic data and proposedmechanisms from some leach studies con-ducted with pyrite samples over the last two decades.

Most of the studies reported in Table 2 have used Fe3+,O2 or H2O2 as oxidants while a range of pH, particle size,temperature and leach medium has been used. High activationenergies (33–92 kJ/mol) are reported which suggests chemical orelectrochemical control and not physical or diffusion control ofthe oxidation process. Most of these studies have either identifiedpyrite oxidation as being controlled by a chemical process withmolecular adsorption reactions of chemical nature occurring onthe surface or by an electrochemical process with electron transferreactions from distinct anodic and cathodic sites on the pyritesurface. The rate dependencies do not however agreewell betweenthe studies reported in Table 2 especially where Fe3+ is used as theoxidant.

According to Bockris and Reddy [89] and Bockris et al. [90]there are fundamental differences between chemical and elec-trochemical reactions. For a chemical reaction to occur the reac-tants collide directly to form a product while in electrochemicalreactions reactants do not directly meet but collide with an elec-tronically conducting substrate (electrode), such as a mineral sur-face. The direct collision and product formation that occurs for thechemical reaction produces heat while the free energy change inelectrochemical reactions produces electricity without heating thesurroundings. However according to Bockris and Reddy [89] it isdifficult to separate electrochemical reactions from heterogeneouschemical reactions occurring at a solid-solution interface, such asaqueous pyrite oxidation. Heterogeneous processes on surfaces(both chemical and electrochemical) may in general involve a cat-alyst surface (electrocatalyst for electrochemical reactions), trans-port and adsorption of reactants to the surface, reaction betweenadsorbed species (charge transfer for electrochemical reactions)and finally release of products. FurthermoreBockris andReddy [89]state that since both processes involve ‘‘potential-energy-distancerelations and barriers and activated states’’, both may follow thesame mathematical expression for the rate (Eq. (6)):

r =kTh

∏i

cie−∆G0=/RT (6)

where Πci refers to product of reactant concentrations and ∆G0=

is the standard free energy of activation. However, in contrast tochemically controlled reactions, electrochemical reactions involvea net charge transfer and have potential dependant rates, normallyexpressed in terms of current density, i, given by Eq. (7).

rateelectrodic =iF

(7)

where the rate of the electrodic reaction is expressed asmoles m−2 s−1 and F is Faraday’s constant.

Eq. (7) can further be expanded and expressed in terms ofanodic current density and surface area, given by Eq. (8).

r =iaAZaF

(8)

where ia is the anodic current density, A is the surface area and Zais the moles of electrons transferred during mineral oxidation.

According to Bockris and Reddy [89] the rate of the electrodicreaction is normally expressed in A cm−2 however the currentdensity, i, is divided by the amount of charge, F , transferred permole of reactant. Pyrite oxidation is however complex as it isa multi-step process with the possibility of both homogenousand heterogeneous processes. ‘‘Purely chemical’’ reactions do notform a major part of most natural processes, including corrosion,photosynthesis and biological respiration [90]. According to Korytaet al. [91] chemical reactions can be part of electrochemicalreactions in that they may produce an electroactive speciesprior to an electrochemical step, consume a product afteran electrochemical step or occur simultaneously to regenerateoriginal species fromproducts. If this is the case for pyrite oxidationthan the rate determining step in this multi-step oxidation processcan either be a chemical reaction or an electrochemical transferprocess. Depending on the leaching conditions such as, pH, Eh,oxidants and their concentrations, solution to mass or volumeratios, and even the type of pyrite sample, the rate control maychange from electrochemical to chemical and vice-versa.

Apart from those listed in Table 2, earlier studies also posedconflicting views on pyrite oxidation rate control processes. Usingpyrite samples from three different localities (Utah, Vermontand Spain) to study pyrite oxidation Garrels and Thompson [62]proposed that the instantaneous rate was controlled by adsorptionof Fe3+ and Fe2+ on to the pyrite surface while Singer andStumm [63] showed that rate is controlled by Fe2+ oxidation togenerate Fe3+ in a cyclic oxidation process. These two articles havesince been amongst the most cited in pyrite oxidation research.

Garrels and Thompson [62] conducted their experiment instirred acid (H2SO4) Fe3+ sulfate solution at 33°C in a nitrogen(N2(g)) purged environment. The Fe3+ was generated from Fe2+sulfate through quantitative potentiometric titration with cerateor permanganate prior to the experiment. The solution potentialfor each experiment was recorded as a function of time. AccordingtoGarrels and Thompson [62] the solution potentialmay be relatedto Fe3+ and Fe2+ molalities according to Eq. (9):

Eh = E◦′

+ 0.059 logmFe3+

mFe2+. (9)

This relationship and results from various potential versus timeplots of each experiment, with different pyrite and different initialFe3+ concentrations, was used to trace the reduction of Fe3+. Thereduction of Fe3+ was used as an indicator of pyrite oxidation.It was found that pyrite of different origin oxidised at distinctlydifferent rates albeit following a common mechanism. It was alsofound that the average rate of Fe3+ reduction was independent oftotal Fe in solution while the instantaneous rate decreased withtime and Fe3+ molality. It thus appeared that the overall oxidationrate is dependent on the oxidation potential of the solution or theFe3+ to Fe2+ ratio. Garrels and Thompson [62] proposed that therate is proportional to the portion of the surface occupied by Fe3+which would be a function of the relative Fe3+ to Fe2+ adsorptionand the surface area available. The fraction of Fe adsorbed asFe3+ would in turn be proportional to the fraction of Fe3+ insolution. Fe3+ would normally (that is at condition of high Fe3+concentration or solution potential) oxidise pyrite irreversibly toFe2+ and sulfate according to Eq. (1). However, under equilibriumconditions (low Eh approx. 250–300 mV), S0 may be producedaccording to Eq. (10). If an excess of Fe3+ is present the S0 will befurther oxidised to sulfate via a number of intermediate steps. Inaddition Garrels and Thompson [62] also found that pH does notexert a significant influence on pyrite oxidation rate in the range ofpH 0–2, however it may be significant at higher pHwhere Fe(OH)3is an additional oxidation product.

FeS2 → Fe2+ + 2S° + 2e. (10)


Table2

Pyrite

oxidationmec

hanism

spr

opos

edfrom

leachstud

iesco

nduc

tedwith

naturalp

yrite

samples

over

thelast

twode

cade

sun

derv

arious

diffe

rent

cond

ition

s.Py

rite

type

(sizeµm)

T°C

/pH

Leachmed

ium

Oxida

ntus

edRa

tede

pend

ence

Activ

ation

energy

,kJ

/mol

(T°C

)

Rate

cons

tant

(k)

Mec

hanism

Ref.

Variou

sna

tural

(75–

150)

25/2.0

(HCl

adjusted

)Fe

Cl3so

l.Fe

3+10

−3M

[Fe3

+]1

92(25–

50)

1×

10−4 –

2.7

×10

−4s−

1Su

rfacech

emical

controlle

dan

dmass

tran

sport

[ 6]

Natur

al(125

–250

)30

/1.98(H

Clad

justed

)

FeCl

3so

l.Fe

3+mM

[Fe3

+]0

.5,[H

+]−

0.5

–1

×10

−9.74

moles

cm−2 m

in−1

Non

epr

opos

edsu

ggestedsu

rfacech

em.

andelec

troc

hem.

possibilitie

s

[ 65]

H2O

2so

l.H

2O2mM

[H2O

2]1

–1

×10

−1.43

cmmin

−1

O2pu

rged

sol.

PO21a

tm[O

2]0.5

56.9

±7.5

(20–

40)

1×

10−6.77

moles

0.5cm

−0.5min

−1

Variou

sna

tural

(108

)25

/6.7–8

.5Ca

rbon

ate-bu

ffered

sol.(5

mM

NaH

CO3)

O20.06

%–04

.9%

Non

-linea

rwith

-respe

ct-to

[O2]

88(3–2

5)0.64

×10

−12–1

2.9

×

10−12

moles

S−1g−

1Su

rfaceco

ntrolle

dby

prod

uctd

ecom

positio

n[61]

Natur

al(38–

45)

23–2

3.5/6–

7Dea

erated

0.01

MNaC

lCo

mbina

tionof

O2,

Fe3+

andFe

2+1s

tord

erw.r.t.su

rfacearea

toso

lutio

nvo

l.ratio

––

Elec

troc

hemical

appr

oach

.Lim

itedby

Fe2+

oxidation

[64]

Natur

al(150

–250

)25

/≈2(H

Clad

justed

)Fe

Cl3so

l.Fe

3+[Fe3

+]0

.62

–3.0

×10

−5moles

m−2s−

1Non

prop

osed

.Sug

gests

inhibitora

mon

gstthe

prod

ucts

[92]

Natur

al(150

–250

)25

/<3(H

Clan

dH

2SO

4ad

justed

)Fe

Cl3an

dFe

2(SO

4)3

sol.(O

2or

N2pu

rged

)

Fe3+

ForN

2pu

rge:

[Fe3

+]0

.3,

[Fe2

+]−

0.47,[H+

]−0.32

–1.0

×10

−8.58

moles

m−2s−

1Su

rfaceElec

troc

hemical

controlle

d[8]

Fe3+

andO

2Fo

rO2pu

rge:

[Fe3

+]0

.93 ,

[Fe2

+]−

0.40

–1.0

×10

−6.07

moles

m−2s−

1

O2

ForO

2pu

rge:

[O2]

0.5 ,

[H+]0

.11

–1.0

×10

−8.58

moles

m−2s−

1

Natur

al(50–

63)

20–5

0acidic

Acidified

(HClO

4)pe

roxide

sol.

H2O

2[H

2O2]

157

–Ch

emically

controlle

d[79]

Natur

al20

–50acidic

Acidified

(H2S

O4)

peroxide

sol.

H2O

2[H

2O2]

168

–Su

rfacech

emical

control

bysh

rink

ingco

remod

el[77]

Natur

alsp

ain

35/≈1.6–

3.5

(H2S

O4ad

justed

)

Fe2(SO

4)3(O

2or

N2

purged

)and

O2

purged

sol.

Fe3+

and/or

O2

Anod

icdissolution

[H+]−

0.5

79.9

–Elec

troc

hemical

control

with

anod

icox

idation

andcathod

icredu

ction

[66]

Ferric

redu

ction

[Fe3

+]0

.97–

1.07

45–

O2redu

ction

[O2]

1.05–1

.09 ,

[H+]0

.14

––

Ove

rall:

[Fe3

+]0

.5,[Fe2

+]−

0.5 ,

[O2]

0.5 ,

[H+]−

0.5 ,

[H+]−

0.18

(Fe

absent)

––

Natur

al(various

fractio

nsfrom

149–

44)

170–

230

O2pu

rged

0.5M

H2S

O4so

l.O

2[O

2]0.5(at2

10°C

)33

.2DO

(170

–230

)–

Elec

troc

hemical

controlle

d[13]

Natur

al≈84

%Py

(<38

)45

–75acidic

0.1M

Fe(Fe3

+/Fe

2+)

in0.5M

H2S

O4so

l.KM

nO4

[Fe3

+/Fe

2+]0

.57

83–

Elec

troc

hemical

with

fast

ferric

redu

ction

[87]


Singer and Stumm[63] studied the relative rates of pyrite oxidationusing both Fe3+ and O2 as oxidants, in a bid to identify the ratedetermining step in order to control acid mine drainage. Theyconsidered pyrite oxidation as a cyclic process involving first aninitiation step (Eq. (1)) which produces Fe2+. After the initiationstep a cycle is established where Fe2+ is oxidised (Eq. (2)) toform Fe3+ which then is responsible for continued pyrite oxidation(Eq. (3)). Investigations of the rate of reduction of Fe3+ by pyrite,with and without O2 (DO) at pH 1, demonstrated Fe2+ to bethe dominant and more rapid oxidiser of pyrite. Experimentsexamining Fe2+ oxidation rates as a function of pH demonstratedthe following rate expression (Eq. (11)) at pH values above 4.5:

−dFe2+

dt

= kFe2+

[O2]

OH−

2 (11)

where k = 8.0 × 1013 l2 mole−2 atm−1 min−1 at 25 °C. Howeverat pH values below 3.5 Fe3+ oxidation is independent of pH; asimilar observation to that of Garrels and Thompson [62] for theoverall pyrite oxidation rate between pH 0 and 2. Below pH 3.5Fe2+ oxidation follows the following rate law (Eq. (12)):

−dFe2+

dt

= k′Fe2+

[O2] (12)

where k′ equals 1.0 × 10−7 atm−1 min−1 at 25 °C. The Fe2+oxidation by O2 was identified as the rate limiting step [63].Further investigation by Singer and Stumm [63] with natural minewaters showed increased pyrite oxidation rates due to microbialacceleration of Fe2+ oxidation by a factor of more than 106.The authors suggest use of kinetic control measures to suppressbacterial action and retard Fe2+ oxidation to control acid minedrainage.

Three to four decades of exhaustive research has sincebeen conducted into identifying the control mechanisms ofthis economically and environmentally important process. Theremainder of this section in most part examines the importantstudies conducted in this period. As there are countless variablesinvolved within each experiment, it is best to consider each studyindividually first before collating relevant important comparableinformation.

McKibben and Barnes [65] carried out pyrite oxidation in lowtemperature acidic chloride solutions to determine the rate lawsfor oxidation of pyrite by solution Fe3+, O2 and peroxide. Theyfound that the dissolution rates generally obey the following ratelaw (Eq. (13)):

Rvol = −kAV

∏i

(Mnii ) (13)

where Rvol is the volumetric dissolution rate, k is the rate constant,A is total mineral surface area, V is solution volume, Mi aremolarities of species i affecting rate and ni the order of dependence.Using this approachMcKibben and Barnes [65] used the initial ratemethod to determine the rate laws as they reasoned that this wasa more sound approach than the integration method as a rate lawis not assumed and there are no large variations in A and V duringthe initial periods of the reaction in contrast to the longer periodsrequired for the integration method. In addition to the data givenin Table 2, McKibben and Barnes [65] obtained the following ratelaws (Eqs. (14)–(16)) for the three oxidants examined:

Rsp,Fe3+ = −10−0.974M0.5Fe3+M

−0.5H+ (14)

Rsp,O2 = −10−6.77M0.5O2

(15)

Rsp,H2O2 = −10−1.43MH2O2 . (16)

McKibben and Barnes [65] suggest that a rate law containing afactor of M0.5 involves the surface dissociation of a species. Whilethis is not the case for Fe3+, O2 may undergo such a dissociationprocess. The authors have proposed a partial mechanism whichinvolves the dissociation of O2 adsorbed on the pyrite surface intoindividual O species or peroxide (Eqs. (17) and (18)).

O2 → 2O (17)O2 + 2H2O → 2H2O2. (18)

They further propose that this form of molecular mechanismshould be ‘‘reconciled’’ with electrochemical mechanisms involv-ing cathodic reduction of O2 (Eqs. (19) and (20)) and anodic oxida-tion of pyrite.

O2 + 4H++ 4e−

→ 2H2O (19)

O2 + 2H++ 2e−

→ H2O2. (20)

Wiersma and Rimstidt [6] identified chemical control as ratelimiting from their leach studies on three different categoriesof pyrite samples (lower-temperature/early diagenetic, marcasite,and higher-temperature hydrothermal/metamorphic) in acidicFeCl3 solutions to determine the dependence of pyrite oxidationrate on Fe3+. It was shown that the Fe3+ data was best describedby the following rate law (assumed, Eq. (21)) that is first orderwithrespect to Fe3+:

− dmFe3+/dt = k(A/M)mFe3_ (21)

where k is the rate constant, A/M the ratio of surface area ofthe reacting solid to mass of solution and mFe3+ is the molarconcentration of uncomplexed Fe3+. A high activation energy of92 kJmol−1 was found fromanArrhenius plot and itwas concludedthat the rate determining step was the breaking of the relativelystrong covalent bonds at the surface, and hence a chemicallycontrolled reaction. It was also noted that the activation energyfell to 25 kJ mol−1, at low stirring (400 rpm) speeds and hightemperatures (>35 °C), indicating a change in rate control fromsurface chemical control to solution mass transport control. Otherstudies under similar conditions did not witness any dependenceof rate of reaction on stirring speeds and thus ruled out solutionmass transport as being rate limiting [65,79]. Because a surfacechemical control was identified, Wiersma and Rimstidt [6] alsoemphasised the need to incorporate the surface area to mass ofsolution ratio when calculating rate constants.

Dimitrijevic et al. [79] and Antonijevic et al. [77] conductedsimilar experiments and also found the pyrite oxidation process tobe chemically controlled. Using H2O2 as the oxidant, Dimitrijevicet al. [79] studied the kinetics of pyrite oxidation in perchloric(HClO4) acid solutions while Antonijevic et al. [77] used sulfuric(H2SO4) acid solutions. Both studies found that pyrite dissolutionwas best described by a surface reaction controlled shrinking coremodel with a reaction order of one with respect to peroxide,which was the same as the reaction order found by McKibbenand Barnes [65]. In addition a high activation energy (Table 2) andinverse relationship between rate constantwas found. Both studiesconcluded that the reaction was surface chemically controlled.Dimitrijevic et al. [79] proposed a mechanism involving pyriteoxidation by peroxide through a ‘thio’ intermediate (S2O2−

2 andS2O2−

3 ) formation followed by either oxidation to sulfate ordecomposition to elemental sulfur and bisulfite in perchloric acidsolutions. Antonijevic et al. [77] also proposed a similarmechanismfor oxidation in sulfuric acid solution, however in addition theysuggested that the initial peroxide adsorption onto the pyritesurface or the subsequent formation and decomposition of apyrite-peroxide activated complex could be the rate determiningstep.


Fig. 3. Schematics of the electrochemical mechanism proposed by Moses andHerman [64].Source: Redrawn from Ref. [64].

Numerous studies have suggested that electrochemical pro-cesses control the rate of pyrite oxidation and dissolution [8,13,22,64,66,87]. Moses and Herman [64] conducted numerous pyriteleach experiments at circumneutral pH using either Fe3+ or O2,with orwithout added initial Fe2+, as oxidants. They developed thefollowing rate law (Eq. (22)) which is first order with respect to theratio of surface area to solution volume:

rate = kAV

n

. (22)

To derive this rate law they assumed zero-order kinetics withrespect to [SO2−

4 ] and [Fe(II)(aq)], which are products of theoxidation process. They also assumed zero-order with respectto [Fe(III)(aq)], [O2] and [H+

], although the rate of change ofthese species was not determined during their experiment. Lineardependency of O2 pyrite oxidation rate on surface area hasalso been reported under buffered carbonated conditions at pH6.7–8.5 [61].

McKibben and Barnes [65] have cautioned against assuminglinear dependency of pyrite oxidation rates on surface area. Asstated previously, SEM studies by McKibben and Barnes [65]showed that pyrite oxidation is mostly confined to regions withhigh excess surface energies, which results from the mineralgrowth history and surface preparation techniques. Their resultsdemonstrated that ‘‘effective’’ or ‘‘reactive’’ surface area was moreimportant and significantly different from total surface area. Inexperiments purged with O2 and with added initial Fe2+, Mosesand Herman [64] found that there was a very rapid loss of Fe2+during the initial stages of the reaction which oxidation (to Fe3+)alone could not account for. Moreover, in experiments with Fe2+alone, they found that the reaction ceased after a rapid initial rate,despite the reaction solution having enough Fe2+ to sustain thereaction, with far less Fe2+ in solution then expected from Fe3+reduction on the pyrite surface. From these results they concludedthat Fe2+ was preferentially (over Fe3+) adsorbing onto reactivesurface sites and blocking Fe2+ attachment. They also suggestedthat Fe2+ adsorption also affects oxidation by O2 (alone) but lessseverely as the amount of Fe2+ produced by O2 oxidation is far lessthan that produced by Fe3+ (Eqs. (1) and (3)).

Rimstidt and Newcomb [92] also suggest the existence of aninhibitor amongst the oxidation products but rule out Fe2+, sulfateor chloride. From their results Moses and Herman [64] proposedan electrochemical mechanism of pyrite oxidation, which theyconsidered as an extension of the model proposed by Singer andStumm [63]. Since Fe2+ will always be present in natural systems,their mechanism involves Fe2+ adsorption on to the reactive sitesas the first step followed by oxidation to Fe3+ by O2. A schematic ofthe mechanism is shown in Fig. 3. The hydrated Fe2+ first adsorbsonto an electron-rich anodic site on the pyrite surface and thenforms a termolecular complex with O2 through hydrogen bonding.Electrons from the Fe2+ then flow to the O2 to produce an Oreduction product and Fe3+ (Fig. 3b). Electrons then rapidly flow

from the pyrite surface to the Fe3+ to reduce it back to Fe2+ andcause a transfer of a hydroxyl to the surface sulfur group (Fig. 3c).After rehydration the Fe2+ (Fig. 3c) can repeatedly undergo thesteps shown in Fig. 3a, b and c until a stable sulfoxy speciesdissociates into the solution. According to Moses and Herman [64]the rate limiting step of this mechanism at circumneutral pH isthe oxidation of Fe2+ to produce Fe3+. However if Fe2+ is notpresent then the first S to O bond formation becomes rate limiting.Williamson and Rimstidt [8] argued that at pH 2 the oxidation ofFe2+ to produce Fe3+ for pyrite oxidation is not significant. Themechanism shown in Fig. 3 indicates that O2 does not directlyattack the pyrite surface (although this may be possible) andthat the oxygen added to the sulfur group originates from water.Isotopic tracer studies by Reedy et al. [67] have shown this to becorrect. Fe3+ is the dominant oxidant and if adsorbed prior to Fe2+begins the oxidation process from Fig. 3b.

Williamson and Rimstidt [8] compiled the available literaturerate data and combined it with their own experimental data toproduce rate laws using multiple linear regression analysis. Theirrate laws also showed dependence on Fe2+ concentration whenFe3+ was present, thus highlighting the importance of Eh in thecontrol of the rate of reactions. When DO was present duringpyrite oxidation by Fe3+, rates were only affected by Fe2+ and Fe3+concentrations (Eq. (23)):

r = 10−6.07(±0.57)m0.93(±0.07)

Fe3+

m0.40(±0.06)Fe2+

(23)

where r is the rate of pyrite dissolution inmolm−2 s−1. However, inthe absence of DO the pH also plays a significant role in rate control(Eq. (24)):

r = 10−8.58(± 0.15)m0.30(±0.02)

Fe3+

m0.47(±0.03)Fe2+

m0.32 (±0.04)H+

. (24)

When only DO is present as the oxidant, rates are predominantlydependant on DO concentration andwithminor contribution frompH (Eq. (25)):

r = 10−8.19(± 0.10) m0.5(±0.04)DO

m0.11(±0.01)H+

. (25)

Moreover, Williamson and Rimstidt [8] also found that the rateof pyrite oxidation by Fe3+ increases when DO is present athigh Fe3+ to Fe2+ ratios, however the rate is faster when DOis not present at low Fe3+ to Fe2+ ratios. The rates were alsoseen to positively correlate with Fe3+ and negatively with Fe2+concentrations, which indicated a dependence of rate on the Ehof the solution. Williamson and Rimstidt [8] developed rate lawsdependant on solution Eh for Fe3+ oxidation. When DO was alsopresent the rate is given by Eq. (26):

r = 10−19.71 (±0.86)Eh12.93 (±1.04)pH1.0 (±0.29). (26)

And for N2 purged solutions the rate is given by Eq. (27):

r = 10−12.7 (±0.11)Eh6.10 (±0.19)pH0.37 (±0.04). (27)

These rate laws show that solution Eh plays a significant rolein controlling the rate of pyrite oxidation by Fe3+, which mayprovide evidence for an electrochemical rate determined oxidationprocess. According to Williamson and Rimstidt [8] the leachmechanism may not follow a site specific adsorption processas they found that the reaction kinetics (fractional orders) andsaturation of the mineral surface by oxidants cannot be explainedby a simple Langmuir isotherm model and therefore adsorptionof oxidants or desorption of products may not control the


rate of reaction as suggested by McKibben and Barnes [65].Williamson and Rimstidt however found that the data couldbetter be represented by a Freundlich isotherm with a multilayernonsite-specific process. Williamson and Rimstidt [8] proposedan electrochemical mechanism for aqueous oxidation of pyritewith non-site specific, multilayer adsorption of oxidants withelectron transfer from the pyrite surface as the rate limiting step.This electron transfer was proposed to occur at a cathodic siteon the pyrite surface with the oxidant not necessarily directlybinding to the surface. The electron transfer may occur withina ‘‘discrete zone’’ of solution in close proximity to the cathodicsite. Williamson and Rimstidt also suggest an analogous anodicreaction, at a separate site, involving electron-deficient sulfur andwater resulting in the release of a stable sulfur product. Thiselectrochemical mechanism is different from the one suggested byMoses and Herman [64] in that the oxidants approach/adsorb onthe cathodic site and not on the reduced (anodic) site. Also there isno termolecular complex (Fig. 3) formation and the sulfoxy speciesare produced at a different site from the site approached by theoxidants.

Holmes and Crundwell [66] conducted electrochemical exper-iments to study the kinetics of pyrite dissolution by Fe3+ and/orDO through steady-state voltametric measurements from a stan-dard three electrode (pyrite working electrode, platinum counterelectrode and saturated calomel electrode, SCE) cell. The pyriteused in the electrodewas from Logrono, Spain andwas determinedby thermoelectric and Hall Effect measurements to be an n-typesemiconductor. The oxidation and reduction processes were stud-ied separately by applying anodic and cathodic potentials with re-spect to the rest potential while measuring the current produced.The mixed potential of the pyrite electrode leached for 14 daysin acidic (pH 1.6) Fe2(SO4)3 (9 g/L total Fe) solutions at 35 °Cwith solution redox potential of 0.65 V (vs SCE) was determinedwith and without DO. The mixed potential is defined as a sponta-neous process where ‘‘two or more coupled redox reactions occurat equal but opposite rates at the same interface, one componentbeing oxidised and the other reduced’’ [93]. According to Holmesand Crundwell [66] pyrite dissolution involves anodic oxidation ofpyrite and cathodic reduction of oxidants (Fe3+ and/or O2) whichoccur simultaneously on the pyrite surface. Thus pyrite dissolutionmay be represented by the following half reactionswhich show theanodic oxidation of pyrite by water (Eq. (28)) and cathodic reduc-tion of the oxidants (Eqs. (19) and (29)).

FeS2 + 8H2O → Fe2+ + 2SO2−4 + 16H+

+ 14e− (28)

Fe3+ + e−→ Fe2+. (29)

It was proposed that the anodic and cathodic processes aredependent on the potential across the mineral-solution interfaceand the overall process occurs at the mixed potential or corrosionpotential (Em) where the current due to the anodic and cathodicprocesses are equal with no net electron production. The rates ofthe individual half reactions determine the mixed potential forpyrite dissolution and may be given by Eq. (30) in the presenceof aqueous Fe3+ and Fe2+, with out O2. This equation was derivedfrom the individual half reactions and their relationship to currentdensity, i,

Em =RTF

ln

kFe3+

Fe3+

kFeS2 [H+]−

12 + kFe2+

Fe2+

(30)

where kFeS2 is the rate constant for anodic oxidation, kFe2+ is thecathodic oxidation rate constant and kFe3+ the cathodic reductionrate constant. According to Holmes and Crundwell [66] under con-ditions where kFeS2 [H

+]−

12 ≫ kFe2+ [Fe2+], the mixed potential

and rate become independent of Fe2+ concentration and depen-dant on pH. This occurs at low Fe2+ concentrations typical ofacid mine drainage conditions. However, under conditions wherekFeS2 [H

+]−

12 ≪ kFe2+ [Fe2+], the mixed potential depends on both

Fe2+ and Fe3+ concentrations, and approaches the value of the re-dox potential of the solutionwhich is given by the followingNernstequation (Eq. (31)),

Eredox = E0redox

RTF

ln

a(Fe3+)

aFe2+

, (31)

where a(Fe3+) and a(Fe2+) represents Fe3+ and Fe2+ solutionactivities. The authors have cautioned against the assumptionthat the mixed potential, which is a kinetic quantity, and redoxpotential, which is a thermodynamic quantity, are the same. Atconstant Fe3+ to Fe2+ ratios the mixed potential is independent oftotal solution Fe concentration.When both Fe3+ andO2 are presentthe expression for the mixed potential of pyrite modifies to (Eq.(32)),

Em =RTF

ln

kFe3+

Fe3+

+ kO2 [O2]

H+0.14

kFeS2 [H+]−12 + kFe2+

Fe2+

(32)

where kO2 is the cathodic reduction rate constant. Examinationof the pyrite anodic electrode (used to study pyrite oxidation)using Raman spectroscopy and XPS showed increasing amounts ofpolysulfides (S2−n ) on the surface with increasing applied potential.No apparent reduction in oxidation rates was observed due tothe accumulation of these surface products as current was seento increase exponentially with applied potential. Moreover themixed potential of samples leached for different time periodsand an unleached sample were the same. The study of thecathodic pyrite electrode for Fe3+ or O2 reduction showed thatboth processes are controlled by reaction kinetics and not bymass transfer, which is also evident from the high activationenergies for these processes (Table 2). The cathodic studyand mixed potential measurements further showed that pyritedissolution occurs much more slowly as a result of oxidationby O2 as compared with Fe3+. Holmes and Crundwell [66]developed the rate laws for pyrite dissolution by expanding theEq. (7). The electrochemical rate law for pyrite dissolution withonly Fe3+ is given by Eq. (33),

rFeS2 =kFeS2

H+− 1

2

14F

kFe3+

Fe3+

kFeS2 [H+]−

12 + kFe2+

Fe2+

1

2

, (33)

and for dissolution by only DO:

rFeS2 =kFeS2

H+−0.18

14F

kO2 [O2]kFeS2

12

, (34)

and for dissolution by both Fe3+ and DO the rate law is(Eq. (35)),

rFeS2 =kFeS2

H+− 1

2

14F

kFe3+

Fe3+

+ kO2 [O2]

H+0.14

kFeS2 [H+]−12 + kFe2+

Fe2+

1

2

. (35)

The study finally concluded that pyrite dissolution was entirelygoverned by the proposed electrochemical/mixed potential pro-cess and the rate determining step is the charge transfer at thesurface-solution interface.

Recent studies by Long and Dixon [13] and Bouffard et al. [87]have also pointed to an electrochemical control mechanism ofpyrite oxidation. Long and Dixon [13] conducted a pressureoxidation study of pyrite from Zacatecas, Mexico (obtained from


Wards Natural Science Establishment, Ontario, Canada) in sulfuricacid solutions and at O2 partial pressures from 345 to 1035 kPa inthe temperature range of 17–230 °C. Their results showed that atlow pulp densities (1 g dm−3 pyrite) oxidation initially conformsto shrinking sphere model given by Eq. (36):

dXdt

=3(1 − X)

23

τor

dσdτ

= −1τ

(36)

where σ = (1 − X)1/3 = d/d0 (d = particle diameter, d0 = initialparticle diameter), X = pyrite conversion, τ = time (mins) forcomplete oxidation.

However at later stages (at low pulp densities) the oxidationfollows a ‘‘passivating shrinking sphere’’ model, proposed by Longand Dixon [13], as a result of possible passivation of the pyritesurface by elemental sulfur (S0). Earlier studies have also proposedpossible rate control by product decomposition on the surface [61].According to Long and Dixon [13] this surface passivation may berelated to a decrease in surface electrochemical potential evidentfrom the low Fe3+ to total Fe ratio. Moreover, they suggest that theelectrochemical component of pyrite oxidation can be representedby Eq. (37). The thiosulfate (S2O2−

3 ) produced is unstable in acidsolutions and may readily convert to elemental sulfur, and sulfite(SO2−

3 , Eq. (38)). The sulfite may further be oxidised to sulfate byFe3+. They further postulate that the potential near the surfacemaydecrease as a result of the possible increase in local concentrationof Fe2+, causing a build-up of the passivating elemental sulfur filmon the surface. It was further shown that increasing pulp densityto 20 g dm−3 also prevented the onset of surface passivation inaddition to increasing the rate of oxidation. According to Long andDixon [13] the increased surface area relative to Fe concentrationmost likely maintained the electrochemical potential at a levelthat prevented passivation. This together with the evidence of thedissolution rate being proportional to oxygen partial pressures tothe power of 0.5 led Long and Dixon [13] to suggest that the pyriteoxidation mechanism may be consistent with anodic dissolutionand cathodic reduction with the rate controlled by the initialcharge transfer step of both half-cell reactions represented byEqs. (37) and (38).

4FeS2 + 7O2 + 4H+→ 4Fe3+ + 4S2O2−

3 + 2H2O (37)

S2O2−3 → S0 + SO2−

3 . (38)

Bouffard et al. [87] studied the leaching kinetics and stoichiometryof pyrite oxidation in acidic Fe3+ sulfate (Fe2(SO4)3) solutionsunder controlled conditions (isokinetic) in the temperature rangeof 45–75 °C. KMnO4 was used as the oxidising agent which alsoserved as a medium for controlling solution potential (Eh) andfor tracing leaching progress. Fe3+ to Fe2+ ratios from 10 to300 were used in a range of experiments. The sample used inthis study contained 84.0 wt% pyrite, 11.2 wt% marcasite, 2.7wt% arsenopyrite and 0.6 wt% chalcopyrite, determined throughquantitative XRD, Rietveld analysis. It was found that the leachdata, under all conditions studied, conformedwell to the shrinkingsphere model. The stoichiometry of dissolution was only slightlydependant on solution potential and independent of temperature,with nearly 64% sulfate formed from each unit of sulfide sulfurthat was oxidised. The remaining proportion was found to beelemental S0 particles of 2 µm diameter. On the basis of theleach data a rate expression for pyrite oxidation based on mixedpotential theory, an Arrhenius temperature function and shrinkingsphere model was developed. The authors considered that Fe3+reduction and mineral oxidation, occur simultaneously on thepyrite surface, and are governed by surface mixed potential withno net electron production. The Eq. (8) was further expanded byincluding surface galvanic contributions from Fe(III) and Fe(II),

and including effects of changing grain morphology (shrinkingsphere model) and temperature (Arrhenius function) to developthe following overall rate law (Eq. (39)):

∂d∂t

= 0.0176 exp[−9937

1T

−1

333

] CFe(III)

CFe(II)

0.572µmh

(39)

where ∂d/∂t is the particle shrinkage rate, CFe(III) and CFe(II) aremolal concentrations of Fe(III) and Fe(II) respectively and T isabsolute temperature. In addition to the rate lawBouffard et al. [87]also created a mathematical expression for pyrite leaching:

1 − X (t, d0) =

1 −

td0

0.0176 exp

[−9937

1T

−1

333

]CFe(III)

CFe(II)

0.5723

(40)

where X is pyrite dissolution and d0 is initial particle diameter. Thepyrite oxidation rate was found to be dependent on temperatureand proportional to the square root of the potential determiningcouple, Fe3+/Fe2+ ratio. According to the authors fast Fe3+reduction occurs at the pyrite surface where mixed potentialsapproach the reversible potential of the Fe3+/Fe2+ couple.

Rimstidt and Vaughan [9] further strengthened the notion ofelectrochemical control by combining results from their previousstudies and other published data to provide a detailed mechanismshowing each elementary step involved in the cathodic reductionof oxidants (Fe3+ and O2) and anodic oxidation of pyrite. Theirproposed mechanism is shown in Fig. 4, where the first step isthe transfer of charge from a cathodic mineral surface to oxidantspecies followed by electron transport from an anodic site to thecathodic site and finally attachment of an oxygen atom of water toa positively charged S of disulfide sulfur (S2−2 ) at an anodic site. Theoxidants (Fe3+ and O2) react with pyrite according to Eqs. (1) and(3).

Rimstidt and Vaughan [9] propose that the electron transfer atthe cathodic site occurs from a Fe2+ site on the pyrite surface toeither a hydrated Fe3+ or O2 through formation of an activatedcomplex between Fe2+ and the oxidant, and that this step is viewedas the rate determining step for the overall oxidation process. Theadsorbed Fe3+ finally leaves the cathodic site as a Fe2+ after a singleelectron transfers while O2 is converted to 2 hydroxyl groups afterthe successive transfer of four electrons from the same surfaceFe2+. After each successive electron transfer from the cathodic Fe2+site on the mineral surface (which renders the cathodic Fe2+ to bea Fe3+), an electron is transferred from an anodic site to reduce theFe3+ to reformas a Fe2+, to allow further cathodic electron transfer.This then leads to the final step to complete the oxidation processwhere the O of water attaches (through nucleophilic attack) to aterminal S of positive polarity (due to electron transfer to cathode)to form sulfate as the final product after 7 or 8 electron successivetransfer steps. This S may pass through several oxidation states toproduce S(VI)O2−

4 from S(−I)2−2 . Rimstidt and Vaughan [9] suggestthat the last step of themultistep sulfur oxidation process dependson pH which can result in nearly 100% sulfate formation at low pHto substantial amounts of thiosulfate and other sulfur products aswell at high pH. Along with the sulfate, Fe2+ from the anodic sitewill simultaneously be released into the solution where it will becoordinated by 6 water molecules which will change it to a highspin configuration thus easing its oxidation to Fe3+.

3.3.1. SummaryRecent studies seem to generally conclude that the control

mechanism for pyrite oxidation follows an electrochemical processrather than a purely chemical reaction. In summary the followingconclusions can be drawn.


Fig. 4. Diagrammatic representation of the electrochemical mechanism of pyrite oxidation suggested by Rimstidt and Vaughan [9]. The first step is cathodic reduction,followed by electron transfer from anode and finally anodic oxidation. However the three steps above represent the attachment of just a single oxidant (either O2 or Fe3+)and oxidation of single sulfur and as such during pyrite oxidation there will be numerous such processes that will occur simultaneously. The cathodic and anodic sites canbe present either on the same pyrite surface or different surfaces. In the above diagram the notation ‘‘× number’’ represents the successive addition of four water moleculesor the successive release of eight hydrogen ions after each step. It also represents the number of e− added successively to the oxidant or removed from the anodic surfaceto the cathode. In the above processes the first step in cathodic reaction is electron transfer to the oxidant while in the anodic reaction an electron removal from the S atomgenerally occurs prior to the addition of a water molecule. For the production and release of a thiosulfate to the solution at least 7 electrons are successively removed fromone disulfide sulfur while for the production of sulfate at least 8 electrons are removed from disulfide sulfur.

(1) In addition to Fe3+, Fe2+ also adsorbs onto the surfaceof reacting pyrite. One would expect that with continuedleaching and with a limited supply of complexing ions insolution the concentration of Fe2+ in the leach solution willincrease (even with continued oxidation to Fe3+) and thismay increase the competition between Fe2+ and Fe3+ surfaceadsorption. What effect may this have on the rate of theoxidation process? Moses and Herman [64] and Garrels andThompson [62] suggest that such Fe2+ adsorptionwill reducethe rate and the effects may be more severe if Fe3+ is presenttogether with O2. This is consistent with reported fractionalrate law orders which are negative for Fe2+ (Table 2). It ispossible that oxidation may also proceed from adsorbed Fe2+sites by a similar mechanism to that proposed by Moses andHerman [64], or the adsorbed Fe2+ may behave like surfaceFe2+ and follow the mechanism suggested by Rimstidt andVaughan [9].

(2) Rate control by decomposition of surface products has beensuggested [61] and with the suggested presence of elementalsulfur [13,62,87] this seems highly likely. However veryfew studies have suggested the presence of surface leachinginhibitors [92] and such passivation has only been shownto occur under high pressure conditions [13]. The impact ofthe accumulation of surface products on rate may be greaterunder unstirred conditions which are typical of acid rockdrainage however this seems to require further investigationand verification.

(3) The rate is generally reported to be slightly inverselydependent on pH under acidic conditions but pH may playa greater role at high pH where Fe-oxy/hydroxide formationalso occurs.Williamson and Rimstidt [8] showed that the ratedependence on pH was greater in the absence of O2 when

Fe3+ was present and only slightly dependent on pHwhen O2was the only oxidant. While Mckibben and Barnes [65] foundsimilar results for the former case (− 1

2 order dependence),no dependence on pH was found for the latter. Holmes andCrundwell [66], with their electrochemical study under acidicconditions, showed that pH can play a significant role indetermining rates under certain conditions.

(4) The surface area of the reacting pyrite in relation to solutionvolume or mass is another very important factor thatdetermines the rate [6,64,65]. Studies [61] have reported alinear rate dependency on surface area (for O2 oxidation)while others have found beneficial effects of increasingpulp densities, which increases surface area in relation toFe concentrations [13]. Garrels and Thompson [62] havesuggested a proportionality of rate to the surface area coveredby the Fe3+ which is consistent with the SEM results ofMcKibben and Barnes [65] who highlighted the importanceof ‘‘reactive’’ surface area over total surface area. Studiessuggesting chemical control [6,77,79] and electrochemicalcontrol [13,87] have both found a dependence of rate on theinitial particle size where results conform to the shrinkingsphere model. Certainly the surface plays a determining rolefor attachment of reactants and electron transfer processes.The ratio of surface area to solution volume needs to besufficient to minimise solution transport effects and is animportant consideration in rate law formulations.

(5) Studies such as that reported by Wiersma and Rimstidt [6]and Holmes and Crundwell [66] suggest that under certainconditions, especially at low stirring speeds, solution trans-port can exert an influence on the rate. Note that nearly allstudies reported here have conducted leach studies undersufficiently stirred conditions to overcome solution transport.


While industrial leaching is generally carried out with suffi-cient stirring to overcome bulk solution diffusion this will bequite different for acid rock drainage where transport controlcan exert a significant influence.

(6) Fe3+ is the dominant and the more rapid oxidiser comparedto O2 however O2 is needed to sustain the oxidation reaction.Peroxide has also been shown to be also an important oxidiserwith a reported first order rate dependency [65,77,79].

(7) The Fe3+ to Fe2+ couple is rate determining even when O2is also present. Reported rate laws involving Fe3+ generallyfollow half order kinetics with a few studies reporting up tofirst order. Order with respect to Fe2+ is also generally halforder but is negatively correlated with rate. The Fe3+ to Fe2+

ratio determines the solution’s redox potential [62,87] andtherefore leach rate depends on the solution Eh [8].

(8) The situation is however less clear when only molecularoxygen is present as an oxidant. The rate dependence ofoxidation on O2 has not been shown to be dependent on Eh.Oxidation by DO produces Fe2+ in solution and this has thepotential to oxidise to Fe3+ to become a potential oxidiseritself. However, the rate of pyrite oxidation when only O2is present has been shown to be dependent only on O2concentration and pH (Table 2). It has also been shown thatunder such conditions Fe2+ oxidation to produce Fe3+ is notsignificant [8,65]. According to Williamson and Rimstidt [8]the O2 concentration apparently determines the solution Ehin absence of aqueous Fe.

(9) Therefore when both Fe3+ and DO are present as oxidants thepyrite oxidation rate is directly correlated with Eh and thistends to support an electrochemical control mechanism. Therate of pyrite oxidation may be defined by mixed potentialtheory and expressed in terms of current density [66,87] mayfollow a multistep electrochemical mechanism as suggestedby Rimstidt and Vaughan [9]. However, with the strongevidence that surface area also plays a critical role this factorit needs to be incorporated into any proposed electrochemicalrate laws, similar to Bouffard et al. [87].

(10) The evidence for electrochemical control does not howeverrule out the presence of chemical processes amongst theelectrochemical processes. The isotopic study conductedby Reedy et al. [67] also showed that minor amounts ofsulfate also derived at least one oxygen from molecularoxygen in addition to oxygen fromwater. The electrochemicalmechanisms proposed do not however explain the existenceof such sulfate-species.

Recent reviews by Chandra and Gerson [94] have characterised thesurface processes occurring during copper activation and flotationof pyrite and sphalerite. An electrochemical control of the processwas also suggested especially under mixed mineral conditionswhere galvanic interactions may be rife. Mineral processingplants dealing with pyrite (for both leaching and flotation)can benefit positively by controlling electrochemically drivenoxidation processes within the circuit. Further studies, resulting inthe provision of information on product/intermediate speciationand spatial distribution, are required to further validate theexistence of electrochemical processes as a dominant mechanismand to verify the co-existence of chemical processes and theirimpact under various different leach conditions. Spatially resolvedsurface studies are also needed to study site-specific variations inreactivity and speciation.

4. Spatially resolved surface characterisation

Pyrite (as well as many other minerals) has a very heteroge-neous surface, both topographically and chemically. Numerous dif-ferent S and Fe species have been suggested to occur on fresh andreacted pyrite surfaces. However, to date very few studies have fo-cused on characterising the lateral chemical heterogeneity to geta better understanding of how distinct topographic sites react andhow and what local speciation occur during reactions. Recent re-views have shown that pyrite surface studies have mostly beenconfined to macroscopic techniques [95]. Macroscopic techniquesprovide area averaged information and may miss critical site spe-cific chemical variations. For example, it has been shown that afresh pyrite surface contains four different main surface species,S2−2 (surface) (3-fold coordinated), S2−2 (bulk) (4-fold coordination),S2− and S2−n species [46–49]. However no information is so faravailable to see how each of these species is distributed on thesurface and if there are particular regions of the surface whereone species may be concentrated. If a particular species does havesite-specific localisation then that area of the surface will reactdifferently to other areas. Such information will be vital to theunderstanding of pyrite oxidation during leaching and acid rockdrainage, and suggestions of electrochemical mechanisms may beable to be verified. Spatially resolved surface speciation informa-tion may also help in the understanding of copper activation andxanthate adsorption processes during mineral flotation.

Synchrotron based imaging and spectroscopic techniquesare ideally suited for surface studies of heterogeneous mate-rials. Synchrotrons offers a highly collimated and wide radi-ation energy range of high brightness (typically greater than1018 photons/s/mm2/mrad2/0.1% bandwidth). The high bright-ness together with the high degree of collimation enables a con-siderable degree of demagnification and therefore high spatialresolution. The incident beam can be focused down to the sub-micron and nanometre scale while maintaining sufficient in-tensity to enable good quality analyses. Scanning photoelectronmicroscopy (SPEM) and X-ray photoemission electron microscopy(PEEM) provide spectromicroscopy and microspectroscopy withsub-micron scale spatial resolutionwith fast acquisition times thatenables surface elemental and speciation imaging correlated tosurface morphology.

4.1. Scanning photoelectron microscopy (SPEM)

Summary of capabilities:

• Highly surface sensitive imaging with high spatial resolution(≈100–150 nm);

• Nano spot (150 nm) spectroscopy (XPS and XANES) with highspectral resolution;

• Chemical and elemental imaging;• Ability to correlate topography with chemistry at a submicron

scale;• Ability to extract spectroscopic information from any region of

the recorded image;• In situ sample preparation with UHV sample fracture and

exposure to atmospheric gases;• The sample does not need to be atomically flat;• However the focused beam can induce surface carbon deposi-

tion which may obstruct measurements.

SPEM measurements involve the use of X-ray optics to demagnifythe incident photon beam to sub-micron size. Measurements aremade by scanning the sample on an x–y stage with respect to theincident beam.

In a SPEM setup the sample is always normal to the incidentbeamwhile the hemispherical electron analyser (energy filter) has


Fig. 5. Schematic of a SPEM setup showing the incident beam and the glancingangle geometry of the hemispherical analyser.

a grazing acceptance angle for high surface sensitivity (Fig. 5).The grazing angle geometry makes this technique very surfacesensitive and depending on the core level binding energy ofthe element examined the information depth can be in theorder of Angstroms. The incident beam is focused to sub-micronsize using a Fresnel zone plate lens. The two typical types ofchromatic optical elements used to reduce photon beam size arereflective (such as Schwarzschild objectives) and diffractive (zoneplates). Zone plates are however preferred for SPEM as they offerthe optimal spatial resolution i.e. the smallest focal spot, whileSchwarzschild objectives are mostly preferred for photon energiesbelow100 eV [96]. An order-sorting aperture (OSA) is used to blockunwanted diffraction orders from reaching the sample’s surface, sothat a high signal-to-noise ratio is maintained. The pressure in theSPEM chamber is maintained at a pressure in the order of 10−10

mbar.SPEM can be used to either conduct imaging spectromicroscopy

or selectedmicrospot spectroscopy. In the spectroscopic mode it ispossible to collect XPS spectra of selected elements in addition toconducting XANES measurements (as total electron yield). How-ever, XANES measurements require a synchronised movement ofthe zone plate opticswith changing the incident energy to keep thesample in focus [96]. The SPEM instrument incorporates a multi-channel electron detector where photoelectrons with specifickinetic energy are measured by each channel, within a specifiedenergy window or pass energy [97]. During imaging mode anenergywindow is specified andmultiple images (equal to the num-ber of detector channels) of photoelectronswithin this energywin-dow are simultaneously recorded. This also provides the option ofextracting XPS spectra, of energy range determined by the energywindow, from any selected region of the total image.

Since intensities of photoelectrons with a particular kineticenergy within the energy window are stored in different channels,chemical maps showing chemical state information can beobtained from elemental maps by plotting information from onlythose channels corresponding to the energies of a particularcomponent. Lateral chemical variations at different depths canalso be imaged for a single element by using different core leveltransition energies, for example Fe 2p and Fe 3s images willprovide information fromdifferent depths due the different kineticenergies of the resulting photoelectron.

Moreover, the final image contains topographical information(in addition to any possible artifacts as discussed below) ofthe sample’s surface, similar to an SEM image. Fig. 6 shows animage of a freshly fractured pyrite surface taken at the ESCA

Fig. 6. A S 2p image showing planes and ridges on the UHV fractured pyrite. Theimage was taken at the ESCA Microscopy beamline, Elettra, with a 150 nm beam,0.25 µm steps and 200 ms dwell. The energies (approx. 8 eV window) were storedin 48 detector channels at 0.164 eV per channel.

Microscopy beamline, Elettra, Italy. The image, recorded at theS 2p photoelectron energy shows the topography of the surfacein a similar manner as to what would be observed with a SEM.This image also contains the elemental or chemical contrastinformation across the sample area, which is convoluted withthe topographical effects. Processing of the image, such asdivision by a background image (taken at some energy belowthe target photoelectron peak) of the sample area, is necessaryto remove the topographic effects and to reveal any chemicalcontrast present on the surface. To obtain the image in Fig. 7,the detector channels corresponding to S2−2 from Fig. 6 werechosen and normalised against the sum of the intensities of thedetector channels of other energies. This removes the topographiccontribution to the image and only intensities corresponding todisulfide energy are consequently displayed. Fig. 8 shows a micro-XPS spectrum with approximate energy/channel distribution andchannels corresponding to different S species. Some topographiceffects will remain if there is a significant difference betweenthe kinetic energies of the background image and the elementalimage as information will be from different depths. Divisionby a background image also helps to remove any variation inbeam intensity which is essential for quantitative evaluations.Further details of artifact removal are available elsewhere [98,99].The SPEM setup also enables in situ sample treatments, such asexposure to different gases, and pyrite surfaces before and aftertreatment can be studied.

4.2. Photoemission electron microscopy (PEEM)

PEEM generally offers similar capabilities to SPEM. However,some differences are:

• Higher spatial resolution is possible (<100nmdown to<10nmon new generation PEEM);

• Conducting XANES imaging is easier;• Problem of surface carbon contamination offset with XANES

imagingwhichhas greater informationdepth thanXPS imaging.

PEEM is a spectromicroscopic technique which is capable ofproviding surface images with high spatial resolution togetherwith detailed spectroscopic information via the use of an electronimaging system for magnification and projection of emittedelectrons. The microscopic and spectroscopic information isgathered through parallel measurements where detailed imagesare taken at each energy point across the absorption edge of


Fig. 7. Disulfide distribution on a freshly fractured pyrite surface. The S2−2distribution is not uniform on the surface and its abundance is relatively lower onthe fracture ridges as identified from Fig. 6.

Fig. 8. A micro-XPS spectra showing the energy channels, from 48 detectorchannels, corresponding to different sulfur species.

a selected element. Spectroscopic information (XANES) is thenderived from this series of images (or stacks) by first selecting anarea (or areas) of interest and plotting the intensity of the regionthrough the stack. A diagrammatic explanation is shown in Fig. 9.Prior to obtaining the spectroscopic information the images in thestack need to be normalised against the incident beam intensityto account for natural decay or any instabilities. The images alsoneed to be aligned well as the sample position may shift duringmeasurements as a result of vibrations or sample stage instability.In simple terms PEEM is usually capable of XAS (XANES)-imagingby measurement of a fixed electron kinetic energy as a functionof incident X-ray energy and/or XPS-imaging by measurement ofa range of kinetic energies as a function of fixed incident X-rayenergy. Fig. 10 shows the incident beam and sample geometryinside a PEEM UHV chamber.

PEEM offers rapid imaging due to parallel detection (simultane-ously acquisition of photoelectrons emitted from different regionsof the microscopic field of view) with acquisition times of a fewtenths of a millisecond. This capability makes it possible to under-take in situ studies of chemical and physical processes [100].

PEEM measurements can be performed using a variety ofphoton sources such as a UV-lamp, laboratory X-ray sourcessuch as Al Kα1 and synchrotron X-rays [101]. Imaging can alsobe done using other light sources such Hg arc lamps whichmay provide better (than X-ray) image contrast due to the

Fig. 9. Schematic of a PEEMmeasurement. A series of images (stack) is taken acrossthe absorption edge (or photoelectron peak, in the case of XPS) of an element ofinterest. The final image is usually an image of total average intensities of all imagesin the stack. Spectroscopic area averaged data is obtained from the intensities of thestack.

Fig. 10. Schematics of PEEM setup showing incident beam and emittedphotoelectrons.

narrow energy distribution [102]. Contrast in PEEM images usuallydepends on the illumination source, sample surface compositionand sample topography [103]. Themost powerful attributes of thistechnique has been made possible through the development of3rd generation synchrotron X-ray sources which produce a wideselectable energy X-ray range and a highly brilliant X-ray beamwhich can also be polarised (depending on the insertion deviceused) The chemical and magnetic sensitivity offered by PEEMmakes this technique one of the most important in the areas ofsurface and material science, thin films, geology, medicine andbiology [104].

In its simplest form a PEEM instrument contains an objectivelens and a projective lens between the sample and the detector(Fig. 11). Further lenses, apertures, beam separator, astigmatismcorrectors and mirrors may be added to this basic set-up forbetter spatial and energy resolution and more spectroscopicoptions [98,102,104]. These lenses in PEEM instruments can eitherbe electrostatic or magnetic with the latter having better spatialresolution due to smaller objective lens aberrations and highertransmission [102,104]. The sample in the PEEM UHV chamberis normal to the objective lens with homogenous illumination ofa relatively large area of the sample by photons incident at aglancing angle. A strong electric field exists between the sampleand the objective lens (known as cathode lens) which causes theemitted electrons to accelerate from the sample surface to theobjective lens. The image of the sample within the field of view(FOV) is first magnified by the objective lens, after which the


Fig. 11. A basic electron optics layout of PEEM without energy filtering.Source: Redrawn from Ref. [98].

image undergoes further magnification by a series of additionallenses and finally the projective lens magnifies it onto a 2Dimage detector. The detector is usually an image intensifier (multi-channel plate electron multiplier) coupled with a phosphor screenwhich converts the electron image into visible light detectable bya CCD camera with high dynamic range and high sensitivity [98].

Three categories of PEEM designs exist; PEEM without anenergy filter, PEEM with an energy filter and aberration-correctedPEEM with and without energy filters [98,102]. The currentgeneration of PEEM instruments can achieve a spatial resolutionof a few tens of nanometres and an energy resolution of a fewtenths of an eV [102,104]. PEEM also offers picosecond scale timeresolution [104].

The spatial resolution in PEEM is limited by chromatic andspherical aberrations of the objective lens. PEEM receives all pho-toelectrons emitted from the surface and PEEM images are domi-nated by low energy secondary electrons, the source of chromaticaberrations [103]. Aberration-corrected PEEM instruments such asSMART at BESSY II and PEEM-3 at ALS are now being commis-sioned. These incorporate an electron optical mirror with a highlysymmetric beam separator for correcting the chromatic and spher-ical aberrations of the objective lens [100]. Spatial resolution below2 nm is theoretically possible [98,100]. However, even with theseaberration-corrected setups PEEM will not be able to achieve the0.1 nm range resolution currently possible with TEM [102]. Spa-tial resolution also depends on the degree of contrast offered bythe sample which is a function of the sample material and surfacesmoothness. The maximum resolution cannot be smaller than twopixels. The ALS PEEM currently does not have the correction mir-rors installed, however spatial resolution below150 nm is possible.

Fig. 12 shows a feature on a pyrite surface leached in HClO4solution under controlled solution Eh of 700 mV (SHE), 75 °C andpH 1 with added Fe3+. This image was taken with the PEEM 3instrument at the beamline 11.0.1 at the Advanced Light Source(ALS). Fe XANES spectra can be obtained by first selecting the areaof interest and then integrating the intensities of the selected areathrough the stack. The stack images needs to be aligned well andintensities normalised prior to extracting the spectroscopic data.Fe 2p XANES spectra obtained from the area selected inside thefeature (area 1) and outside the feature (area 2) (Fig. 13) showcontrasting profiles. It is also be possible to process the image inFig. 12 to remove topographic contributions and to show the Fe

Fig. 12. A feature on a pyrite surface leached in HClO4 solution at pH 1, 75 °C andcontrolled solution Eh of 700 mV (SHE), taken with the ALS PEEM 3 instrument,beamline 11.0.1. The image shown is part of the PEEM stack or series of imagestaken at the Fe 2p edge. The image shown is just above Fe edge at 710 eV.

Fig. 13. Fe 2p edgeXANES spectra obtained from the selected areas (1 and 2) shownin Fig. 12.

distribution by normalising this image with an image (part of thestack) taken below the Fe 2p absorption edge.

PEEM instruments can be used to perform X-ray spectromi-croscopy which may require high pass or band pass energy fil-ters. Incident X-ray photons result in emitted electrons with awide range of energies such as primary unscattered or elasticallyscattered photoelectrons, inelastically scattered electrons, Augerelectrons and secondary electrons [104]. Spectroscopic imagingthat is conducted by selecting specific kinetic energies of primaryor Auger electrons, such as XPS and AES (Auger electron spec-troscopy) requires the use of an energy filter. Energy filtering alsohelps improve the spatial resolution of the images as it allowsa narrow energy window to be defined without substantial re-duction in transmission, especially for secondary electron imag-ing [102]. The depth from which these electrons are emitted isusually dependant on the energy of the incident photons and theminimum detectable kinetic energy of photoelectrons from spe-cific core level transition. For XPS-imaging using PEEM this depthis approximately 1 nm [105].

The transmission of electrons through PEEM optics is propor-tional to 1/E, where transmission decreases quickly with increas-ing kinetic energy. The signal levels are higherwhen kinetic energy


Fig. 14. A 10 µm sized feature on HCl leached pyrite surface at pH 1, 75 °C andcontrolled solution Eh of 900 mV (SHE), taken with the CLS PEEM instrument, SMbeamline. The image shown is part of the PEEM stack or series of images taken atthe S 2p edge.

Fig. 15. S 2p XPS spectra of the 10 µm feature and of the area outside the featureas shown in Fig. 14. The spectra were obtained through S 2p spectromicroscopicmeasurements of the area shown in Fig. 14.

is lower and optimum signals are obtained when electron kineticenergy is between 50 and 100 eV making the XPS-imaging modePEEMhighly surface sensitive. XPS-imaging is best done at energiesabove any secondary electron peak to avoid the resulting strongbackground [102]. Fig. 14 shows a S 2p-edge image taken at 181.0eV photoelectron kinetic energy at the PEEM endstation, SMbeam-line, Canadian Light Source from of an HCl leached pyrite sample.The CLS PEEM, known as CaPeRS PEEM (Canadian PhotoemissionElectron Research Spectromicroscope) is a commercial instrumentpurchased from Elmitec GmbH (www.elmitec.de). The setup in-cludes an energy filterwhich allowsXPS imaging in addition toXASspectromicroscopy. The best resolution possiblewith an ideal sam-ple (high contrast features and low topography) is 25 nm [106]. Us-ing XPS spectromicroscopy, S 2p XPS spectra were obtained fromwithin the 10µm feature (Fig. 15) and outside the feature with theformer being dominated by high binding energy S components.

X-ray absorptionmeasurements are conducted by themeasure-ment of the flux of the secondary electrons (resulting from filling ofcore holes created by the photoionisation process) which does notrequire an energy filter. Secondary electrons have a relatively largemean free path due to their low energies (<10 eV) and therefore X-ray absorptionmeasurements using PEEMhave an approximate in-

formation depth of 5–10 nm [104,105]. Diffraction measurements(X-ray photoelectron diffraction, XPD) may also be possible withPEEM setups that use magnetic lenses [102,104].

4.3. Topographic artifacts in SPEM and PEEM

The grazing angle geometry can introduce topographic artifactsdue to the angular dependence of the emitted photoelectronprobing depth [98]. This causes curved or protruding surfacefeatures to appear brighter on the side facing the analyser or theincident beam, due to enhanced flux while the side away fromthe analyser or incident beam appears darker (shadowing effect)as either the emitted photoelectrons or the incident beam areobstructed. Similar effects also occur around pits or depressions onthe surface. Other artifacts due to diffraction, back-scattering anddeflection of emitted photoelectrons may also occur [98]. A simpleapproach to removing these artifacts from the image is by divisionof the image with a background image (taken at some energybelow the absorption edge or photoelectron peak), provided thatthe kinetic energies of the background and the image are similar.An example of a topographic artefact on a pyrite surface is shownin Fig. 16.

4.4. Sample preparation

Being photoelectron-based techniques, SPEMand PEEM requirethe sample to be conductive or of low resistance to avoid samplecharging. Pyrite being a semi-conductor (band gap 0.95 eV) meetsthis requirement. One of the advantages SPEM has over PEEM isthat it does not require the sample to be flat, hence is ideal forstudying freshly (in situ) fractured surfaces.

Samples for PEEMmeasurements need, ideally, to be atomicallyflat with highly parallel front and back surfaces so that sampletilt can be optimised (only few degrees of angle adjustment isavailable) and so that the surface normal is parallel with theelectron optic axis. This ensures a uniform sample illumination fora good focus of the field of view and sufficient signal. However,the most important reason for a highly flat surface is due to thepossible field emissions that may be created by rough surfacetopography. The sample in the PEEM UHV chamber is part ofthe cathode lens and is usually maintained under a negativevoltage (approx. −15 to −20 kV) while the objective lens is atground, approximately 2 mm from the sample. This lens fieldcan be distorted by sample charging or surface roughness. Sharpedges on the sample surface tend to field-emit very strongly andcause the field emission intensity to exceed safe levels for theimage intensifier. In the most extreme case this can possibly leadto multi-channel plate damage or in more moderate cases fieldemissions may cause strange image patterns. Discharges from thesample may also cause contamination of the sample surface withobjective lens material such as Fe, Ni and Cr. Imaging may still bepossiblewithmoderate sample topography however the collectionefficiency will vary between depressions and hills causing falseintensity variations and loss of spatial resolution. Mineral samplescan be polished to a mirror finish using an automatic polisherwhich ensures parallel surfaces (Fig. 17).

5. Conclusion

Pyrite is often viewed as a ‘‘penalty’’ mineral that not onlyincreases smelting costs but is also a significant contributor toacid rock drainage. Pyrite from different geographical regions canhave different electrical properties with samples also exhibitingelectrical structure variations between the surface and the bulk.Pyrite is composed of Fe2+ cations and S2−2 anions (with covalentbonding) in an ideal stoichiometric ratio of 1:2, however this

http://www.elmitec.de


Fig. 16. S 2p SPEM image of pyrite surface collected at ESCA Microscopy beamline, Elettra, Italy. Image A is a raw image which shows a protruding surface feature whichappears brighter towards the analyser while the side away from the analyser is dark. After normalising the imagewith background channel, image B is obtainedwhich showsthat there was no real chemical contrast but just an artifact of the glancing angle geometry.

Fig. 17. Pyrite crystal chunk and polished pyrite slab that can be used for SPEM andPEEMmeasurements.

rarely occurs in nature. Lattice substitution of these ions by otherminor and trace elements is directly related to ore genesis andgeographical location. This tends to introduce significant electronicvariations due to band gap variability and may cause the pyriteto be either an n-type semiconductor or a p-type semiconductor.Furthermore with the possibility of electrochemical control ofpyrite oxidation such variationsmay be the reason for the differingreaction rates observed for pyrites from different locations.Generally pyrite formed in an high temperature environment, withlow arsenic content and with stoichiometric S to Fe ratios of lessthan 2 are n-type semiconductor while where the opposite is thecase p-type semiconducting pyrite tends to form. The distributionof elemental inclusions within the pyrite structure does notseem to be uniform with evidence of alternating n and p semi-conductivities within a single pyrite sample. On comminution thismay result in the liberation of n and p-type particles, thus adding tothe variability in reaction rates. The geographical origin, elementalcomposition and semi-conductor type of pyrites are thereforeimportant considerations for the study of pyrite or in industrialplant processing.

Low coordination sites are present on the pyrite surface,compared to the bulk, as a result of conchoidal fracture. Ferrousiron, with low coordination, tends to be present at sites ofdefects and imperfections (steps and kinks) which have smallerband gaps and higher dangling bond densities. This makes thesesites prone to oxidant (ferric iron and molecular oxygen) attackwhere the rate of charge transfer will be much faster comparedto sites of normal (6-fold) coordination. Pyrite exists in a low-spin diamagnetic configuration with unoccupied Fe 3d states. Ifoxidants chemisorb to the pyrite surface then these empty 3d

states will accommodate electrons from the adsorbates followingstrong hybridisation. Consequently ferrous iron is an ideal speciesfor electron transfer reactions with oxidants, as is consistent withthe electrochemical oxidation mechanism proposed by Rimstidtand Vaughan [9].

Atmospheric oxidation of pyrite begins within minutes ofexposure, resulting in the production of sulfate, iron oxy-hydroxide species and possibly elemental sulfur and polysulfide.The oxidation reaction is more rapid when both molecular oxygenand water are present. The formation of ‘islands’ of surfaceproducts provides further evidence for the concept of reactive andless reactive surface areas.

Aqueous oxidation of pyrite generally results in the productionof sulfate as themain product and the release of unoxidised ferrousiron into solution. It may also produce species such as elementalsulfur, polysulfide, hydrogen sulfide gas, ferric hydroxide precipi-tate, iron oxide, iron(III) oxyhydroxide and intermediates such asthiosulfate, sulfite and polythionates. However, the presence andabundance of these species depends on the oxidising conditions.Solution pH, Eh, oxidant type and concentration, stirring speeds,grain size and surface area in relation to solution volume, temper-ature and pressure are all critical and generally influence the ratesof reaction. Of these, solution Eh and surface area or grain size exerta more significant influence on rates as compared to any one otherfactor. The direct correlation of rates with Eh strongly indicates thedominance of electrochemical reactions during pyrite oxidation.This may involve the anodic oxidation of pyrite with simultane-ous cathodic reduction of oxidants with the overall process occur-ring at amixedpotentialwhere the individual reactions are definedby current density. The proposed mechanism involves numerouscharge transfers with a multi-step disulfide oxidation process toproduce sulfate as the final product through several intermediatesand release of ferrous iron. The proposed rate determining step isthe transfer of charge from the pyrite surface at the cathodic site.Given the reactivity of other surface species (on freshly fracturedsurface) such as monosulfide, these species will be oxidised dur-ing atmospheric exposure or during the initial period of aqueousreaction.

The proposed mechanisms have not yet defined the nature ofadsorption (if any) of the oxidants on to the surface. Furthermore,the proposed mechanisms do not account for the presence ofsulfoxy species with at least one oxygen atom abstracted frommolecular oxygen. Therefore the simultaneous presence of minorchemical processes, with the possibility of chemisorption ofoxidants, should also be considered.

While numerous detailed studies of the pyrite surface havebeen carried out, there is currently lack of any chemical informa-tion on a spatially resolved scale. Recent data suggests that species


are heterogeneously distributed on the surface and oxidation isseen to be site specific. A detailed characterisation of freshly frac-tured surfaces by imaging the spatial distribution of different sur-face species is required. Such a study should also be coupled witha spatially resolved examination of species evolution on reactedpyrite surfaces. It would then be possible to trace the reactivity ofdifferent surface regions and verify if the proposed electrochemicalmechanism conforms to the spatially observed surface speciation.Scanning photoelectron microscopy (SPEM) and photoemissionelectron microscopy (PEEM) are two synchrotron-based surfacespectromicroscopic and microspectroscopic techniques that canprovide such information. The two techniques use XPS-imagingandXANES-imaging to provide surface chemicalmaps enabling thecorrelation of chemistry and topography at a sub-micron scale.

Acknowledgements

The support from the University Presidents Scholarshipawarded by the University of South Australia to Mr. Anand Chan-dra is gratefully acknowledged. This gratitude is further extendedto the Premiers Science and Research Fund (PSRF) of South Aus-tralia, Rio Tinto and BHP-Billiton for financial support of theresearch project New Information forMinerals Processing. Supple-mentary funding was also provided by Access to Major ResearchFacilities (AMRF) fund for the measurement of synchrotron datareported here. Contributions by Dr. Joe Cavallaro to themanuscriptgraphics are also acknowledged. We are also thankful to beamlinescientists and staff, Dr. Luca Gregoratti, Dr. Mateo Amatti and Dr.Majid Kazemian of ESCA Microscopy beamline, Elettra, Dr. UdayLanke of SMbeamline, PEEMendstation, Canadian Light Source andDr. Andreas Scholl and Dr. Tony Young of beamline 11.0.1 (PEEM 3)of Advanced Light Source.

References

[1] J.B. Hiskey,W.J. Schlitt, Aqueous Oxidation of Pyrite, SME, Littleton, CO, 1982,pp. 55–74.

[2] V.L. Tauson, R.G. Kravtsova, V.I. Grebenshchikova, E.E. Lustenberg, S.V. Lipko,Geochemistry International 47 (2009) 231–243.

[3] H.-R. Wenk, A.G. Bulakh, Minerals: Their Constitution and Origin, CambridgeUniversity Press, Cambridge, New York, 2004.

[4] J.D. Dana, C.S. Hurlbut, C. Klein, Manual of Mineralogy (after James D. Dana),Wiley, New York, 1977.

[5] A.N. Buckley, R. Woods, Applied Surface Science 27 (1987) 437–452.[6] C.L. Wiersma, J.D. Rimstidt, Geochimica et Cosmochimica Acta 48 (1984)

85–92.[7] P.K. Abraitis, R.A.D. Pattrick, D.J. Vaughan, International Journal of Mineral

Processing 74 (2004) 41–59.[8] M.A. Williamson, J.D. Rimstidt, Geochimica et Cosmochimica Acta 58 (1994)

5443–5454.[9] J.D. Rimstidt, D.J. Vaughan, Geochimica et Cosmochimica Acta 67 (2003)

873–880.[10] V.P. Evangelou, Pyrite Oxidation and its Control, CRC Press, New York, 1995.[11] C.O. Moses, D. Kirk Nordstrom, J.S. Herman, A.L. Mills, Geochimica et

Cosmochimica Acta 51 (1987) 1561–1571.[12] W. Paszkowicz, J.A. Leiro, Journal of Alloys and Compounds 401 (2005)

289–295.[13] H. Long, D.G. Dixon, Hydrometallurgy 73 (2004) 335–349.[14] H. Hu, Q. Chen, Z. Yin, P. Zhang, G. Wang, Hydrometallurgy 72 (2004) 79–86.[15] Y. Peng, S. Grano, D. Fornasiero, J. Ralston, International Journal of Mineral

Processing 69 (2003) 87–100.[16] J.R. Craig, F.M. Vokes, T.N. Solberg, Mineralium Deposita 34 (1998) 82–101.[17] G. Qiu, Q. Xiao, Y. Hu,W. Qin, D.Wang, Journal of Colloid and Interface Science

270 (2004) 127–132.[18] W.L. Bragg, The Analysis of Crystals by the X-ray Spectrometer, The Royal

Society, 1914, pp. 476–478.[19] P. Bayliss, American Mineralogist 74 (1989) 1168–1176.[20] S.L. Finklea III, L. Cathey, E.L. Amma, Acta Crystallographica Section A 32

(1976) 529–537.[21] I. Uhlig, R. Szargan, H.W. Nesbitt, K. Laajalehto, Applied Surface Science 179

(2001) 222–229.[22] R.T. Lowson, Chemical Reviews 82 (1982) 461–497.[23] A. Hung, J. Muscat, I. Yarovsky, S.P. Russo, Surface Science 513 (2002)

511–524.[24] J. Cai, M.R. Philpott, Computational Materials Science 30 (2004) 358–363.[25] H. Majima, Canadian Metallurgical Quarterly (1969) 269–282.

[26] H. van der Heide, R. Hemmel, C.F. van Bruggen, C. Haas, Journal of Solid StateChemistry 33 (1980) 17–25.

[27] J.C.W. Folmer, F. Jellinek, G.H.M. Calis, Journal of Solid State Chemistry 72(1988) 137–144.

[28] R. Edelbro, A. Sandstrom, J. Paul, Applied Surface Science 206 (2003) 300–313.[29] M. Bronold, Y. Tomm,W. Jaegermann, Surface Science 314 (1994) L931–L936.[30] H.W. Nesbitt, A.G. Berlich, S.L. Harmer, I. Uhlig, G.M. Bancroft, R. Szargan,

American Mineralogist 89 (2004) 382–389.[31] H.W. Nesbitt, I. Uhlig, G.M. Bancroft, R. Szargan, American Mineralogist 88

(2003) 1279–1286.[32] C. Pettenkofer, W. Jaegermann, M. Bronold, Berichte der Bunsenge-

sellschaft/Physical Chemistry Chemical Physics 95 (1991) 560–565.[33] K.M. Rosso, U. Becker, M.F. Hochella Jr., American Mineralogist 84 (1999)

1535–1548.[34] F. Hulliger, E. Mooser, Journal of Physics and Chemistry of Solids 26 (1965)

429–433.[35] I. Opahle, K. Koepernik, H. Eschrig, Computational Materials Science 17

(2000) 206–210.[36] G.U. vonOertzen, R.T. Jones, A.R. Gerson, Journal of Electron Spectroscopy and

Related Phenomena 144–147 (2005) 1245–1247.[37] R.T. Jones, Electronic structures of the sulfide minerals sphalerite, wurtzite,

pyrite, marcasite, and chalcopyrite, University of South Australia, Ian warkResearch Institute, 2006.

[38] I.J. Ferrer, D.M. Nevskaia, C. de las Heras, C. Sanchez, Solid State Communica-tions 74 (1990) 913–916.

[39] G. Qiu, X. Qi, H. Yuehua, Computational Materials Science 29 (2004) 89–94.[40] A. Hung, J. Muscat, I. Yarovsky, S.P. Russo, Surface Science 520 (2002)

111–119.[41] H.D. Banerjee, N. Godhaunkar, U. Pal, Materials Letters 10 (1990) 99–104.[42] D. Wei, K. Osseo-Asare, Journal of Electrochemical Society 143 (1996)

3192–3198.[43] V.P. Gupta, N.M. Ravindra, V.K. Srivastava, Journal of Physics and Chemistry

of Solids 41 (1980) 145–148.[44] K.S. Savage, D. Stefan, S.W. Lehner, AppliedGeochemistry 23 (2008) 103–120.[45] E.C. Todd, D.M. Sherman, J.A. Purton, Geochimica et Cosmochimica Acta 67

(2003) 881–893.[46] H.W. Nesbitt, G.M. Bancroft, A.R. Pratt, M.J. Scaini, American Mineralogist 83

(1998) 1067–1076.[47] H.W. Nesbitt, I.J. Muir, Geochimica et Cosmochimica Acta 58 (1994)

4667–4679.[48] A.G. Schaufuß, H.W. Nesbitt, I. Kartio, K. Laajalehto, G.M. Bancroft, R. Szargan,

Surface Science 411 (1998) 321–328.[49] A.G. Schaufuß, H.W. Nesbitt, I. Kartio, K. Laajalehto, G.M. Bancroft, R. Szargan,

Journal of Electron Spectroscopy and Related Phenomena 96 (1998) 69–82.[50] F. Demoisson, M. Mullet, B. Humbert, Journal of Colloid and Interface Science

316 (2007) 531–540.[51] M. Descostes, F. Mercier, C. Beaucaire, P. Zuddas, P. Trocellier, Nuclear

Instruments and Methods in Physics Research Section B: Beam Interactionswith Materials and Atoms 181 (2001) 603–609.

[52] H.W. Nesbitt, M. Scaini, H. HoÌˆchst, G.M. Bancroft, A.G. Schaufuss, R. Szargan,American Mineralogist 85 (2000) 850–857.

[53] G.U. von Oertzen, W.M. Skinner, H.W. Nesbitt, Radiation Physics andChemistry 75 (2006) 1855–1860.

[54] J.A. Leiro, S.S. Mattila, K. Laajalehto, Surface Science 547 (2003) 157–161.[55] P. de Donato, C. Mustin, R. Benoit, R. Erre, Applied Surface Science 68 (1993)

81–93.[56] S. Mattila, J.A. Leiro, M. Heinonen, Surface Science 566–568 (2004)

1097–1101.[57] J.K. Jerz, J.D. Rimstidt, Geochimica et Cosmochimica Acta 68 (2004) 701–714.[58] G. Hu, K. Dam-Johansen, S. Wedel, J.P. Hansen, Progress in Energy and

Combustion Science 32 (2006) 295–314.[59] C.M. Eggleston, J.J. Ehrhardt, W. Stumm, American Mineralogist 81 (1996)

1036–1056.[60] U. Becker, K.M. Rosso, M.F. Hochella, Geochimica et Cosmochimica Acta 65

(2001) 2641–2649.[61] R.V. Nicholson, R.W. Gillham, E.J. Reardon, Geochimica et Cosmochimica Acta

52 (1988) 1077–1085.[62] R.M. Garrels, M.E. Thompson, American Journal of Science 258-A (1960)

57–67.[63] P.C. Singer, W. Stumm, Science 167 (1970) 1121–1123.[64] C.O. Moses, J.S. Herman, Geochimica et Cosmochimica Acta 55 (1991)

471–482.[65] M.A. McKibben, H.L. Barnes, Geochimica et Cosmochimica Acta 50 (1986)

1509–1520.[66] P.R. Holmes, F.K. Crundwell, Geochimica et Cosmochimica Acta 64 (2000)

263–274.[67] B.J. Reedy, J.K. Beattie, R.T. Lowson, Geochimica et Cosmochimica Acta 55

(1991) 1609–1614.[68] B.F. Giannetti, S.H. Bonilla, C.F. Zinola, T. Raboczkay, Hydrometallurgy 60

(2001) 41–53.[69] H.W. Nesbitt, I.J. Muir, Mineralogy and Petrology 62 (1998) 123–144.[70] M.B. Goldhaber, American Journal of Science 283 (1983) 193–217.[71] S.W. Knipe, J.R. Mycroft, A.R. Pratt, H.W. Nesbitt, G.M. Bancroff, Geochimica

et Cosmochimica Acta 59 (1995) 1079–1090.[72] M.B. Stott, H.R. Watling, P.D. Franzmann, D. Sutton, Minerals Engineering 13

(2000) 1117–1127.[73] G.W. Luther III, Geochimica et Cosmochimica Acta 51 (1987) 3193–3199.


[74] D.M.C. Huminicki, J.D. Rimstidt, Applied Geochemistry 24 (2009) 1626–1634.[75] W.E. King Jr., J.A. Lewis, Industrial & Engineering Chemistry Process Design

and Development 19 (1980) 719–722.[76] M.J. Nicol, I. Lazaro, Hydrometallurgy 63 (2002) 15–22.[77] M.M. Antonijevic, M. Dimitrijevic, Z. Jankovic, Hydrometallurgy 46 (1997)

71–83.[78] M.M. Antonijevic, M.D. Dimitrijevic, S.M. Serbula, V.L.J. Dimitrijevic,

G.D. Bogdanovic, S.M. Milic, Electrochimica Acta 50 (2005) 4160–4167.[79] M. Dimitrijevic, M.M. Antonijevic, Z. Jankovic, Hydrometallurgy 42 (1996)

377–386.[80] M. DimitrijevicÌ, M.M. AntonijevicÌ, V. DimitrijevicÌ, Minerals Engineering 12

(1999) 165–174.[81] K. Sasaki, M. Tsunekawa, K. Hasebe, H. Konno, Colloids and Surfaces A:

Physicochemical and Engineering Aspects 101 (1995) 39–49.[82] M.N. Lehmann, M. Stichnoth, D. Walton, S.I. Bailey, Journal of the

Electrochemical Society 147 (2000) 3263–3271.[83] J.E. Dutrizac, Metallurgical Transactions B 12 (1981) 371–378.[84] J. Liddicoat, D. Dreisinger, Hydrometallurgy 89 (2007) 323–331.[85] R.G. McDonald, D.M. Muir, Hydrometallurgy 86 (2007) 206–220.[86] T. Biegler, D.A. Swift, Electrochimica Acta 24 (1979) 415–420.[87] S.C. Bouffard, B.F. Rivera-Vasquez, D.G. Dixon, Hydrometallurgy 84 (2006)

225–238.[88] P.A. Christensen, A. Hamnett, Techniques and Mechanisms in Electrochem-

istry, Blackie Academic and Professional, Glasgow, 1994.[89] J.O. Bockris, A.K.N. reddy,Modern Electrochemistry, PlenumPress, NewYork,

1970.[90] J.O. Bockris, A.K.N. Reddy, M. Gamboa-aldeco, Modern Electrochemistry 2A:

Fundamentals of Electrodics, Kluwer Academic, New York, 2000.[91] J. Koryta, J. Dvorak, V. Bohackova, Electrochemistry, Methuen and Co Ltd.,

London, 1970.[92] J.D. Rimstidt, W.D. Newcomb, Geochimica et Cosmochimica Acta 57 (1993)

1919–1934.[93] G.H. Kelsall, in: D.J. Vaughan, R.A.D. Pattrick (Eds.), Mineral Surfaces: The

Mineralogical Society Series, vol. 5, London, 1995.

[94] A.P. Chandra, A.R. Gerson, Advances in Colloid and Interface Science 145(2009) 97–110.

[95] R. Murphy, D.R. Strongin, Surface Science Reports 64 (2009) 1–45.[96] A. Barinov, P. Dudin, L. Gregoratti, A. Locatelli, T. Onur Mentes, M. Ángel

Niño, M. Kiskinova, Nuclear Instruments and Methods in Physics ResearchSection A: Accelerators, Spectrometers, Detectors and Associated Equipment601 (2009) 195–202.

[97] L. Gregoratti, M. Marsi, G. Cautero, M. Kiskinova, G.R. Morrison,A.W. Potts, Nuclear Instruments and Methods in Physics Research Sec-tion A: Accelerators, Spectrometers, Detectors and Associated Equipment467–468 (2001) 884–888.

[98] S. Günther, B. Kaulich, L. Gregoratti, M. Kiskinova, Progress in Surface Science70 (2002) 187–260.

[99] M. Marsi, L. Casalis, L. Gregoratti, S. Günther, A. Kolmakov, J. Kovac, D. Lonza,M. Kiskinova, Journal of Electron Spectroscopy and Related Phenomena 84(1997) 73–83.

[100] T. Schmidt, et al., Surface Review and Letters 9 (2002) 223–232.[101] M. Merkel, M. Escher, J. Settemeyer, D. Funnemann, A. Oelsner, C. Ziethen,

O. Schmidt, M. Klais, G. Schönhense, Surface Science 480 (2001) 196–202.[102] E. Bauer, Journal of Electron Spectroscopy and Related Phenomena 114–116

(2001) 975–987.[103] S. Christensen, U.D. Lanke, B. Haines, S.E. Qaqish, M.F. Paige, S.G. Urquhart,

Journal of Electron Spectroscopy and Related Phenomena 162 (2008)107–114.

[104] A. Locatelli, E. Bauer, Journal of Physics Condensed Matter 20 (2008).[105] K.V. Kaznatcheev, C. Karunakaran, U.D. Lanke, S.G. Urquhart, M. Obst,

A.P. Hitchcock, Nuclear Instruments andMethods in Physics Research SectionA: Accelerators, Spectrometers, Detectors and Associated Equipment 582(2007) 96–99.

[106] U.D. Lanke, S.G. Urquhart, Canadian Photoemission Electron ResearchSpectromicroscope, CaPeRS, Saskatoon Canadian Light Source, 2009.

The mechanisms of pyrite oxidation and leaching: A ...amulsar.com › pdfs › Science-Direct-The-mechanisms-of-pyrite-oxidat… · A.P.Chandra,A.R.Gerson/SurfaceScienceReports65(2010)293–315

Documents