Arsenic mobilization and iron transformations during sulfidization of As(V)-bearing jarosite

Chemical Geology 334 (2012) 9–24

Contents lists available at SciVerse ScienceDirect

Chemical Geology

j ourna l homepage: www.e lsev ie r .com/ locate /chemgeo

Arsenic mobilization and iron transformations during sulfidization ofAs(V)-bearing jarosite

Scott G. Johnston a,⁎, Edward D. Burton a, Annabelle F. Keene a, Britta Planer-Friedrich b, Andreas Voegelin c,Mark G. Blackford d, Greg R. Lumpkin d

a Southern Cross GeoScience, Southern Cross University, Lismore, NSW 2480, Australiab Environmental Geochemistry, Bayreuth University, Universitatsstrasse 30, Bayreuth 95440, Germanyc Eawag, Swiss Federal Institute of Aquatic Science and Technology, Überlandstrasse 133, CH-8600, Dübendorf, Switzerlandd Institute of Materials Engineering, Australian Nuclear Science and Technology Organisation, Locked Bag 2001, Kirrawee DC, NSW 2232, Australia

⁎ Corresponding author.E-mail address: [email protected] (S.G. John

0009-2541/$ – see front matter © 2012 Elsevier B.V. Allhttp://dx.doi.org/10.1016/j.chemgeo.2012.09.045

a b s t r a c t

Article history:Received 14 August 2012Received in revised form 24 September 2012Accepted 26 September 2012Available online 5 October 2012

Editor: J. Fein

Keywords:ArsenicJarositeMackinawiteSulfideEXAFSOrpiment

Jarosite (KFe3(SO4)2(OH)6) is an important host-phase for As in acid mine drainage (AMD) environments andcoastal acid sulfate soils (CASS). In AMD and CASS wetlands, jarosite may encounter S(− II) produced by sul-fate reducing bacteria. Here, we examine abiotic sulfidization of As(V)-bearing K-jarosite at pH 4.0, 5.0, 6.5and 8.0. We quantify the mobilization and speciation of As and identify corresponding Fe mineral transforma-tions. Sulfide-promoted dissolution of jarosite caused release of co-precipitated As and the majority of mobi-lized As was re-partitioned to a readily exchangeable surface complex (AsEx). In general, maximum Asmobilization occurred in the highly sulfidized end-members of all treatments and was greatest at low pH, fol-lowing the order pH 5.0≈4.0>8.5>6.5. X-ray absorption spectroscopy revealed that most solid-phase Asremained as oxygen-coordinated As(V) when pH values were >5.0 — even during latter stages ofsulfidization and the presence of ≥100 μM dissolved S(− II). In contrast at pH 4.0, As transitioned fromoxygen-coordinated As(V) to a sulfur-coordinated orpiment-like phase. This transition coincided with amarked decrease in AsEx, attenuation of As(aq) and TEM-EDX spectra indicate concurrent formation ofnano-scale zones variably enriched in As (~1–15%). Although discordant with geochemical modeling, the for-mation of an orpiment-like precipitate appears to be a primary control on As mobility during the late stagesof complete jarosite sulfidization under acidic conditions (pH 4.0).Mackinawite was the main Fe-mineral end product in all pH treatments. However, at pH 8.0, jarosite rapidly(b1 h) transformed to a lepidocrocite intermediary. Although lepidocrocite efficiently adsorbed As(aq), thetransformation process itself was incongruent with electron transfer to Fe(III). Further investigation is re-quired to determine whether the electron donor triggering this transformation was direct via S(− II), or in-direct via surface complexed Fe(II) and hence akin to the widely-known Fe(II)-catalyzed transformation ofFe(III) minerals. The results demonstrate that abiotic sulfidization of As(V)-co-precipitated jarosite can mo-bilize substantial As and that pH exerts a major control on the subsequent As solid-phase speciation, electrontransfer kinetics and Fe mineralization pathways and products. The findings are particularly relevant to het-erogeneous sediments in which As-bearing jarosite encounters dissolved sulfide under a range of pHconditions.

© 2012 Elsevier B.V. All rights reserved.

1. Introduction

Jarosite is a common Fe(III)-mineral in coastal acid sulfate soils(CASS) and acid mine drainage (AMD) settings (Acero et al., 2006; Astaet al., 2009; Johnston et al., 2011a). Jarosite can effectively remove Asfrom solution by sorption or co-precipitation mechanisms (e.g. Savageet al., 2005; Egal et al., 2009; Asta et al., 2010; Johnston et al., 2011b),and is therefore an important mineralogical control on aqueous concen-trations of As in both AMD and CASS environments.

ston).

rights reserved.

The formula of jarosite can be represented as MFe3(SO4)2(OH)6,where M is usually K+, Na+, NH4

+, or H3O+. Jarosite typically formsduring oxic conditions at low pH (~1–3) and can transform toschwertmannite and goethite as pH increases (Bighamet al., 1996).With-in the sediments of constructed AMD wetlands or re-flooded CASS wet-lands, jarosite may be subjected to reducing and pH circum-neutralconditions that are well outside its stability field (Jones et al., 2006; Zhuet al., 2008; Johnston et al., 2011a). In such wetland environments thereis potential for jarosite to encounter sulfide [S(−II)] producedby themet-abolic activity of sulfate reducing bacteria (SRB) (Burton et al., 2011a).

Localization of organic matter and physical heterogeneity withinwetland sediments can lead to extreme spatial variability in pH, redox

10 S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

potential and the development of distinct micro-niches (Zhu et al.,2006; Robertson et al., 2009). In AMD wetlands and re-flooded CASSwetlands there is potential for organic-rich zones with active SRB tooverlap and occur in close proximity to jarosite (Johnston et al., 2009;Burton et al., 2011a). Hence, in such heterogeneous field environments,As-bearing jarositemay encounter dissolved sulfide under a range of pHconditions.

Sulfide is a powerful reductant. A variety of investigations haveexplored abiotic reduction of structural Fe(III) in iron oxides via elec-tron transfer from dissolved S(− II) (e.g. Dos Santos Afonso andStumm, 1992; Yao and Millero, 1996; Poulton et al., 2004; Hellige etal., 2012). This surface-controlled reaction is rapid and highly pH de-pendent and can be represented by Eq. (1) (Poulton et al., 2004):

2Fe OHð Þ3 þH2Sþ 4 Hþ↔2Fe2þ þ S 0ð Þ þ 6H2O: ð1Þ

Electron transfer from surface complexed S(− II) leads initially tothe generation of S(0) and Fe(II), driving the reductive dissolutionof the iron oxide and eventual precipitation of Fe(II) species. PotentialFe(II) precipitates can include mackinawite (FeSm), if there is sufficientHS− to reactwith Fe(II) (Dos Santos Afonso and Stumm, 1992; Peiffer etal., 1992; Peiffer and Gade, 2007), or other intermediate Fe(II)/Fe(III)minerals such as magnetite (Hellige et al., 2012). The subsequent for-mation of FeSm can be represented by Eq. (2) (Poulton et al., 2004):

Fe2þ þ HS−↔FeS sð Þ þHþ: ð2Þ

Most previous work has focused on interactions between dissolvedS(−II) and various iron oxides (i.e. Dos Santos Afonso and Stumm,1992; Peiffer et al., 1992; Yao and Millero, 1996; Poulton et al., 2004;Peiffer and Gade, 2007; Hellige et al., 2012). In contrast, the reaction be-tween dissolved S(−II) and jarosite has received relatively little re-search attention. Likewise, the consequences of this specific reactionfor the subsequent mobilization or attenuation of any As associatedwith jarosite has not been investigated. Reaction between jarosite anddissolved sulfide should cause reduction of structural Fe(III) to Fe(II)and drive jarosite dissolution. As a result, sulfidization of jarosite hasthe potential to cause considerable mobilization of jarosite-associatedAs. Sulfidization may also change the proportional abundance ofco-precipitated versus surface-complexed As during the subsequentprecipitation of new Fe(II) mineral phases (Lee et al., 2005).

If there is sufficient sulfide to react with jarosite, Fe-sulfide min-erals such as FeSm are likely to form (Ivarson and Hallberg, 1976)and this newly-formed mackinawite may sequester dissolved As viapH-dependent sorption processes (Farquhar et al., 2002; Woltherset al., 2005). Although mackinawite has a considerably lower sorptionaffinity for As when compared to Fe(III)-mineral phases such asferrihydrite (Charlet et al., 2011), there are also widely contrastingsorption affinities for As between different iron-sulfide mineral phases(i.e. pyrite, mackinawite) (Wolthers et al., 2005; Kirk et al., 2010).

In addition, free sulfide may facilitate the precipitation of discreteAs-sulfide phases — a process which is strongly influenced by pH andreaction kinetics (e.g. Rochette et al., 2000; Gallegos et al., 2007;Renock et al., 2009). Indeed, sulfidogenesis can be an important mech-anism for attenuating As in sulfur-rich sediments (e.g. Bostick et al.,2004; O'Day et al., 2004a; Root et al., 2009; Johnston et al., 2010).Nano-particulate FeSm is typically one of the first iron-sulfides to formin natural sulfidogenic environments and is regarded as an importantmineralogical control on As mobility (Wolthers et al., 2005; Gallegoset al., 2007).

However, it is overly simplistic to regard sulfidogenic environ-ments as generally unfavorable for As mobility (Kirk et al., 2004).For example, several recent studies demonstrate that there is consid-erable potential for mobilization of As when sulfide produced by SRBreacts with As-bearing iron oxides (e.g. Kocar et al., 2010; Burton et

al., 2011b). As mobilization and speciation in sulfidic systems is com-plex, with interactions between S(− II) and As leading to formation ofa variety of thiolated As(III) and As(V) anion species (Planer-Friedrichet al., 2007; Wallschläger and Stadey, 2007; Helz and Tossell, 2008).Contemporary understanding of As–S systems has evolved substan-tially in the last decade and analytical differentiation betweenthiolated As(III) and As(V) species is a relatively recent development(Suess et al., 2009; Planer-Friedrich et al., 2010).

In this study, we subjected synthetic K-jarosite containing co-precipitated As(V) to abiotic reactions with dissolved S(− II) at pH4.0, 5.0, 6.5 and 8.0. Our aim was to examine the effects of pH onthe subsequent mobilization and aqueous/solid-phase partitioningand speciation of As. In addition, we investigated the compositionand Fe mineralogy of the precipitates that formed under differentpH regimes and explored how their formation influenced the ob-served aqueous behavior of As.

2. Methods

2.1. General methods

All laboratory glass-ware was soaked in 5% (v/v) HNO3 for at least24 h, followed by repeated rinsing with deionized water. All chemicalswere analytical reagent grade. All reagent solutionswere preparedwithdeionizedwater (MilliQ). Solid-phase sampleswere prepared and driedunder oxygen-free conditions in an anaerobic chamber (1–5%H2 in N2),containing an O2 consuming Pd catalyst.

2.2. Experimental design

Approximately 500 g of K-jarosite was synthesized according toBaron and Palmer (1996). Sufficient Na2HAsO4.7H2O was dissolvedin the initial solution (prior to the addition of reagents) to generatea synthetic jarosite containing between 500 and 1000 ppm of co-precipitated As(V). Although this is well below the demonstrated ca-pacity of jarosite to incorporate As(V) (Paktunc and Dutrizac, 2003),this concentration range was selected to be consistent with priorobservations of As concentrations in naturally occurring jarosite(e.g. Dudas, 1984). The resulting suspension was allowed to settle,the supernatant decanted and replaced with deionised water andthe suspension thoroughly mixed. This was repeated 4 times to re-move soluble ions, prior to drying the final concentrated slurry at40 °C. The resulting dry material was finely ground using a mortarand pestle.

An airtight 2 L glass reaction vessel (Asynt) with PTFE access portsand a PTFE paddle stirrer was used for all reaction series. A 2 L solu-tion of 0.1 M NaCl, buffered using 0.025 M MES 2(N-morpholino)ethanesulfonic acid and 0.025 M DEPP (N,N′-diethylpiperazine),was placed in the reaction vessel and purged with high purity N2

for 16 h. These buffers were chosen for their non-complexing proper-ties and the wide pH range spanned by their respective pKa values(Kandegedara and Rorabacher, 1999).

A Metrohm 836 Titrando pH stat was integrated with the reactionvessel enabling pH to be maintained (±0.05 pH units) via the addi-tion of N2 purged 0.5 M HCl or 0.5 M NaOH. The atmosphere equili-bration ports on the Metrohm 836 Titrando pH stat were connectedto a N2 filled gas-bag during each experiment, thereby ensuring thatthe headspace of the HCl and NaOH supply vessels remained com-posed of N2 throughout the experiments.

The desired experimental pH of the 2 L solution (4.0, 5.0, 6.5 or8.0) was established using the pH stat, immediately prior to adding10.0 g (~26.2 mM L−1 Fe equivalent) of synthetic As(V)-bearingjarosite to the reaction vessel while simultaneously purging the solu-tion and headspace with high purity N2. The suspension was stirredcontinuously at 300 rpm and purged with high purity N2 for an addi-tional 1 h. After 1 h, N2 purging was ceased and the reaction vessel

11S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

closed to maintain an O2-free environment for the duration of the ex-periments. The head-space in the reaction vessel was then equalizedto atmospheric pressure by connection to a partially-inflated N2

gas-bag.A concentrated S(− II) solution (~0.85 M Na2S) was then added at

a constant rate (100 μM of S(− II) per minute) via a syringe pumpinto the O2-free jarosite suspension while stirring continuously at300 rpm. The concentration of the S(− II) solution was determinedprior to addition via iodometric titration. The flow rate of the syringepump was adjusted to ensure 100 μmol of S(− II) per minute wasadded to the reaction vessel. This addition rate was chosen in orderto constrain dissolved S(− II) concentrations to within an environ-mentally relevant concentration range. The period of sulfide additionwas 10 h (i.e. 60 mmol of S(− II) added in total).

One hundred milliliters of the jarosite suspension was collectedfrom a PTFE sampling port via an enclosed syringe at intervals of 0,20, 60, 120, 240, 360, 480 and 600 min from the start of sulfide addi-tion. The sampling port line was back-purged with N2 after samplecollection to return any residual solution to the reaction vessel cham-ber. At the conclusion of the 10 h sulfide addition period, remainingsolution was transferred via syringe and needle to glass serum vialspre-filled with N2 and fitted with crimp-sealed, butyl-rubber stop-pers. Serum vial suspensions were aged for a further 1 week at~21 °C. The volume of all solutions added (i.e. S(− II), HCl, NaOH)and sample removed during the 10 h sulfide addition period wasrecorded, thereby enabling accounting for all dilution and mass losseffects. Each pH series was conducted in duplicate.

2.3. Aqueous and solid-phase methods

Syringes containing sample suspension were capped and immedi-ately transferred to a glove box. Aliquots were then filtered using0.45 μm enclosed syringe-driven filter units. The filtrate Eh was mea-sured using a freshly calibrated platinum electrode and a TPS WD90meter. Aliquots of filtrate were added to 1,10-phenanthroline solu-tions for determination of Fe2+ (APHA, 2005). Total aqueous Fe wasalso determined after pre-reduction of Fe3+ by hydroxylamine hy-drochloride. An additional aliquot of unfiltered sample was subjectto digestion by N2-purged 0.5 M HCl (Kostka and Luther, 1994),prior to analysis for Fe as described above. This enabled determina-tion of solid-phase reactive Fe(II) and reactive Fe(III) species (aftersubtraction of aqueous Fe(II) and Fe(III) respectively).

Aqueous S(−II) (which includes H2S, HS−,S2− and may include col-loidal aqueous FeS complexes/clusters) was measured on 0.45 μm fil-trate and total sulfide (i.e. dissolved plus FeS(s)) was measured onunfiltered sample by the methylene blue method of Cline (1969).AVS-Swas determined by subtracting dissolved sulfide from total sulfide.AVS-Fe was calculated assuming Fe:S stoichiometry of 1:1. Non-sulfidicFe(II)(s) was calculated by subtracting AVS-Fe from solid-phase reactiveFe(II). Elemental S (S(0)) was extracted from an unfiltered sample,after reaction with HCl, by shaking with 10 mL of N2-purged toluenefor 16 h. An aliquot of the toluene extract was analyzed for S(0) byhigh-performance liquid chromatography (HPLC) with a Dionex Ulti-Mate 3000 system (mobile phase=95% methanol; column=reverse-phase C18; flow rate=2 mL min−1; column temp=40 °C; UV detectionat 254 nm). However, although S(0) is an important reaction product,the data are not considered quantitative due to visible accumulation ofS(0) on thewalls of the glass reaction vesselwhich lead to incomplete re-covery. Aqueous K was determined on a 0.45 μm filtered aliquot via in-ductively coupled plasma–atomic emission spectrometry (ICP-AES;Perkin-Elmer DV4300).

Preservation of sulfidic solutions by acidification can precipitate Assulfides (Smieja andWilkin, 2003). In order to eliminate this possibility,the pH of an aliquot of filtered sample was raised to pH ~10 by the ad-dition of 2.0 M NaOH, followed by the addition of 30% H2O2. This meth-od avoids precipitation that can occur upon acidification (Beak et al.,

2008). Total As was then analyzed via Hydride Generation-Atomic Ab-sorption Spectroscopy (HG-AAS; McCleskey et al., 2004). The limit ofdetection for As was 0.2 μg L−1 and the analytical precision was betterthan 12% based on duplicate analysis of 30% of samples. An additional~10 mL aliquot of filtered sample (0.45 μm) was flash frozen in liquidnitrogen with minimal head space to preserve thiolated arsenic species(Suess et al., 2011). This was analyzed for arsenic speciation viaIC-ICPMS as described by Planer-Friedrich et al. (2010). However, dur-ing flash freezing in N2 to preserve samples for speciation, a black pre-cipitate formed (possibly FeSm) at pH 4.0 and 5.0. This precipitate hadto be filtered prior to IC-ICPMS analysis and appears to have led to var-iable and incomplete recovery of As(aq) at these pH values.

Remaining sample was transferred to a 50 mL centrifuge tube inthe anaerobic chamber and the lid sealed. The centrifuge tube wasthen removed and centrifuged at 3500 rpm for 5 min, before beingtransferred back to the anaerobic chamber where it was decantedand allowed to dry (under desiccant) prior to further solid-phase anal-ysis. Exchangeable As (AsEx) was extracted on a dried sub-sample witha deoxygenated 0.5 MNaH2PO4 solution in an anaerobic chamber usinga modified version of the method of Keon et al. (2001) as described byEiche et al. (2008). The extract was centrifuged at 3500 rpm for 5 minand aliquots of filtrate (0.45 μm) were analyzed for HG-AAS and totalFe as described above. Solid phase remaining after NaH2PO4 extractwas subject to aqua regia digest and analyzed for total Fe as describedpreviously (APHA, 2005).

2.4. Geochemical modeling

The Geochemist's Workbench software package (Bethke, 2007)was employed to model the equilibrium speciation and solubility ofAs. Redox coupling reactions were disabled for all calculations, whiletemperature and pressure were fixed at 25 °C and 1.013 bars, respec-tively. The default thermodynamic database was modified to includethermodynamic data for the solubility of orpiment, amorphous As2S3and realgar from Nordstrom and Archer (2003), Webster (1990) andEary (1992), with inclusion of additional constants for thioarsenic spe-cies fromHelz and Tossell (2008). Solubility calculations for As-bearingsulfide minerals were confined to phases that form via precipitation atlow temperature e.g. orpiment, realgar (O'Day et al., 2004a).

2.5. X-ray diffraction

Randomly orientated powders were mounted in the anaerobicchamber and analyzed by X-ray diffraction (XRD) with a BrukerD4-Endeavor diffractometer fitted with a 2.2 kW CoK X-ray sourceand a Lynx Eye detector. Samples were scanned stepwise from 5° to80° 2θ, using 0.06° steps and a 1.8 s count time. Data were analyzedwith the EVA software (DIFFRAC-plus evaluation package, BrukerAXS, Karlsruhe, Germany).

2.6. Electron microscopy

The morphology of selected specimens was determined by scan-ning electron microscopy (SEM; Leica 440) on carbon-coated samplesmounted on aluminum stubs. Elemental composition was deter-mined with an ISIS energy dispersive X-ray (EDX) microanalysissystem, utilizing a quantitative peak-to-background EDX method op-timized for rough surfaced specimens (Sullivan and Bush, 1997).

Samples for transmission electron microscopy (TEM) were pre-pared by suspending dried powder in ethanol and adding drops ofthe suspension onto holey carbon support films. TEM examinationwas carried out using a JEOL JEM 2010F (JEOL, Japan) equipped witha field emission gun (FEG) electron source operated at 200 kV. TheTEM was equipped with an energy dispersive X-ray (EDX) spectrom-eter and NORAN System SIX microanalysis system (Thermo ElectronCorporation, USA), and GIF2001 electron energy filter (Gatan, USA).

Table 1Summary properties of synthetic As(V)-bearing jarosite. Number in brackets is stan-dard deviation.

Total Fea

(mmol g−1)Ka

(mmol g−1)Sa

(mmol g−1)Asa

(μmol g−1)AsExb

(μmol g−1)Surfaceareac

(m2 g−1)

5.25 (0.19) 2.12 (0.06) 4.55 (0.02) 8.06 (0.22) 0.029 (0.002) 1.49 (0.04)

a Aqua-regia digest.b 0.5 M NaH2PO4 at pH 5.0 (Eiche et al., 2008).c 5 point BET surface area using Argon with free space determination using He

(T=83 K).

12 S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

High resolution TEM imageswere recorded using the 1 k×1 k CCD cam-era in the GIF 2001. Lower magnification images and selected area elec-tron diffraction (SAED) patterns were collected using a side mountedGatan model 782 CCD camera. Magnification and diffraction cameralength calibrations were conducted using a certified MAG-I-CAL sample(NORROX Scientific, Canada). SAED patterns were recorded at2000 mm camera length. A suite of well characterized syntheticand natural mineral samples were used for calibration of the EDXspectrometer.

2.7. X-ray absorption spectroscopy

Selected samples were examined by X-ray absorption spectrosco-py (XAS) to determine As and Fe speciation. Arsenic K-edge X-ray ab-sorption near-edge structure (XANES) and extended X-ray absorptionfine structure (EXAFS) spectra were collected on bending magnetbeamline 20B at the Photon Factory, Tsukuba Japan. Air-sensitive ref-erence standards and samples were placed within glass serum vialswhile in an anaerobic chamber and sealed in order to be kept freeof oxygen. Samples were mounted in an aluminum sample holder,sealed under Kapton tape and held at 15–26 K with a He-purgedcryostat to avoid changes in oxidation state. The X-ray energy resolu-tion was maintained by a Si(111) channel-cut monochromator andthe energy calibrated against the first inflection point of the absorp-tion edge of Na2HAsVO4 (11,874 eV). Four replicate spectra were col-lected in fluorescence mode using a 32 element array Ge solid-statedetector. Energy calibration and merging of replicates were performedusing Average software (ASRP, 2007). The Athena program was usedfor standard background subtraction and edge-height normalization, aswell as EXAFS spectra extraction using the AUTOBK algorithm (Rbkg=0.8; k-weight=3) (Ravel and Newville, 2005). Arsenic speciation insamples was quantified by least squares linear combination fitting(LCF) of both the As K-edge XANES spectra, and k3-weighted EXAFS os-cillations in the 2–10 Å−1 range, to reference standards using theATHENA program. No energy shift was included in the LCF. Referencestandards included As(V)-bearing synthetic jarosite, As(III)-sorbed toschwertmannite (prepared as described in Burton et al., 2009) and orpi-ment. Fourier-transformed EXAFS spectrawere calculated over a k-rangefrom 2 to 10 Å−1, using a Kaiser–Bessel apodization window (windowparameter=2.5). Selected end-member As K-edge EXAFS data were an-alyzed by first shell fits on the Fourier-transformed EXAFS signal(k-range 2 to 10 Å−1, r-range 0.8–2.4 Å) using the software code Arte-mis 0.8.012 (Ravel and Newville, 2005). Theoretical EXAFS scatteringpaths for As–S or As–O were calculated from the crystallographic struc-ture of orpiment and scorodite respectively using FEFF8 (Ankudinov etal., 1998). The amplitude reduction factor was constrained for all fits(S02=1 or 0.867) and fitting parameters included the radial distance(R), coordination number (CN) and Debye–Waller factor (σ2) and ener-gy shift (ΔE0). Although this procedure determines inter-atomic dis-tances accurately (±0.02 Å), the coordination number is less reliable(~±25%) due to high correlation between CN and σ2 (Kelly et al., 2008).

Fe K-edge XANES and EXAFS spectra were collected on the highfield (1.9 T) wiggler XAS beamline at the Australian Synchrotron,Melbourne. Samples and reference standards were mounted as thinpowders between kapton tape in an anaerobic chamber atmosphereand sealed in O2-free glass serum vials prior to transport to thebeamline. Mounted samples were placed in a He-purged sample cham-ber at room temperature and spectra collected in fluorescence modeusing a 100 element solid state Ge detector with energy calibratedagainst an in-line Fe(0) foil (7110.75 eV). X-ray energy resolution wasmaintained by a crystal Si(111) monochromator. The Athena programwas used for standard background subtraction and edge-height nor-malization, as well as EXAFS spectra extraction using the AUTOBK algo-rithm (Rbkg=1.0; k-weight=2) (Ravel and Newville, 2005). Ironspeciation in samples was quantified by least squares LCF of both FeK-edge XANES spectra and k3-weighted EXAFS oscillations in the

2–10 Å−1 range, to selected reference standards using the ATHENAprogram. No energy shift was included in the LCF and the sum of thefitted fractionswas not constrained (LCF data are presented normalizedto 100%). Reference standards used in LCF include As(V)-bearing syn-thetic jarosite, lepidocrocite, goethite, magnetite, mackinawite and sul-fate green-rust, which were prepared as described in Burton et al.(2008a) and Claff et al. (2010). Fourier-transformed EXAFS spectrawere calculated over a k-range from 2 to 12 Å−1, using a Kaiser–Besselapodization window (window parameter=3).

3. Results

3.1. Initial jarosite properties

The composition of the synthetic As(V)-bearing K-jarosite used inthe experiments is summarized in Table 1. The ratio of K:Fe:S was1.2:3:2.6, which is close to the ideal stoichiometry of 1:3:2, but some-what deficient in Fe. Such deficiencies in Fe are not uncommon injarosite group minerals (e.g. Paktunc and Dutrizac, 2003). The totalAs(V) content was 8.06 μmol g−1 (~600 ppm) and the majority ofthis As was co-precipitated, with only 0.029 μmol g−1 (0.36%) ex-tractable by 0.5 M NaH2PO4. The synthetic jarosite comprised aggre-gated spheroids of approximately 1–3 μm in diameter (Fig. 1) andhad a surface area of 1.49 m2 g−1 (Table 1).

3.2. Aqueous and solid-phase speciation and partitioning

At the conclusion of all experiments, the final molar ratio ofdissolved sulfide added to total Fe was between 1.41 and 1.46 (Fig. 2).This is close to the stoichiometric quantity of sulfide required(1.5 S(− II)/Fe(III)) to drive reduction of all Fe(III) to Fe(II) and enablethe subsequent precipitation of all Fe(II) as FeSm according to Eqs. (1)and (2). Although the rate of S(−II) addition was constant in each pHtreatment and concentrations of total Fe and S(−II) were very similarthroughout (see Supporting Information Fig. SI1), there were large dif-ferences in dissolved S(−II) accumulation (Fig. 2).

At pH 6.5 and 8.0 dissolved sulfide reacted rapidly, which limited itsaccumulation in the reaction vessel to ≤100 μM (Fig. 2). In contrast,dissolved sulfide accumulated at lower pH values, with a maximum of1.6 mM and 6.8 mM at pH 5.0 and 4.0 respectively — although thetiming of the respective maxima differed considerably (Fig. 2). S(−II)oxidation rates were calculated for each time step (difference betweendissolved S(−II) added and total S(−II) remaining — accounting formass removal by sampling). Mean S(− II) oxidation rates for the 10 hreaction period spanned 17–30 μmol min−1, displayed a high degreeof pH dependence andwere greatest at pH6.5 (see Supporting Informa-tion Fig. SI2).

The behavior of Fe2+(aq) was consistent with the strongly pH de-pendent activity of Fe2+(aq). For example, Fe2+(aq) accumulated to amaximum of ~16 mM at pH 4.0 by the conclusion of S(− II) addition,whereas at pH 5.0 the peak in Fe2+(aq) concentrations was 8.6 mMand occurred at a S(− II):Fe ratio of about 0.5. At pH 6.0 the maximumFe2+(aq) concentration was ~3.2 mM and at pH 8.0 it remained below0.01 mM after a brief initial peak of 0.3 mM (Fig. 2).

Fig. 1. SEM image of initial As(V)-bearing jarosite spheroids.

13S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

Themobilization of K+(aq) can be regarded as a proxy for jarosite dis-

solution. The increases in K+(aq) during the progression of sulfidization

displayed a clear pH dependencewhich indicates that the rate of jarositedissolution was fastest at pH 8.0>6.5>5.0>4.0 (Fig. 2). At pH 8.0 thedissolution of jarosite as inferred by K+

(aq) was both rapid and completeby a S(−II):Fe ratio of ~0.2. For all other pH treatments the concentra-tions of K+

(aq) indicate near complete dissolution of jarosite as theS(−II):Fe ratio approached 1.5.

Concentrations of As(aq) increased during sulfidization and spannedover 3 orders of magnitude among pH treatments, following the order5.0≈4.0>8.5>6.5. Despite slower dissolution of jarosite, As(aq) mobili-zation was more rapid and of greater magnitude at pH 5.0 and 4.0 com-pared to pH 8.0 and 6.5 (Fig. 2). As(aq) concentrations generally increasedat each successive sampling interval, with one notable exception beingthe final sulfidization step at pH 4.0, where As(aq) was attenuated by50% compared to the preceding value. However, regardless of initial dif-ferences As(aq) concentrations converged toward a similar maximumnear the final sulfidization step (~5–12 μM). At pH 6.5, concentrationsof As(aq) remained b0.1 μM until the final sulfidization step. In controlexperiments without S(−II) addition, As(aq) concentrations remained

Fig. 2. Aqueous concentrations of Fe2+, S(−II), K+ and As during reaction of As(V)-bearingtotal dissolved sulfide added to total Fe. Shaded region represents maximum theoretical K+

K+(aq). Error bars (±10%) are smaller than symbol size on log scale.

both very low (b0.01 μM) and stable across all pH treatments (seeSupporting Information, Fig. SI3).

As previously described, the recovery of As(aq) species duringIC-ICPMS analysis was incomplete and highly variable, particularly atlow pH, and hence the speciation results are of limited quantitativeuse (Supporting Information, Table SI1). Nonetheless several useful ob-servations can be made. In most of the samples, arsenate, arsenite andmono- and di-thioarsenate species were evident to some extent. Initialsamples from pH 4.0 and 5.0 at low S(−II):Fe ratios (0.24 and 0.12 re-spectively) had higher As recoveries (69% and 82% respectively) andwere composed of arsenite, mono- and di-thioarsenate species, whilearsenate was below detection limits. Arsenate was detectable atpH 4.0 and 5.0 at higher S(− II):Fe ratios. Samples from the pH 8.0treatment at S(− II):Fe ratios >0.8 also had reasonable As recover-ies (68–103%) and were dominated by arsenate species (SupportingInformation, Table SI1).

The transfer of electrons to structural Fe(III) in jarosite duringsulfidization (as measured by the ΣFe(II) aqueous and solid-phasespecies) was most efficient at pH 8.0 and least efficient at pH 4.0(Fig. 3a). Near-complete Fe(III) reduction was observed at the finalsulfidization step across all pH treatments. Although rates of S(− II)oxidation and Fe(III) reduction vary throughout the sulfidization pro-cess, the relationship between the rates approximates the expected2:1 slope based on Eq. (1) (Fig. 3b). At pH 4.0, 5.0 and 6.5, the mobi-lization of K+ into solution was broadly congruent with electrontransfer to Fe(III), with a normalized plot displaying a slope close to1:1 (Fig. 3b). However, at pH 8.0 the mobilization of K+ into solutionwas incongruent with electron transfer to Fe(III) and was extremelyrapid. This suggests that at pH 8.0 the process of jarosite dissolutionwas via a process that was disconnected from the reduction of struc-tural Fe(III). There was a reasonable agreement between the sum ofelectron equivalents measured analytically relative to the knownsum electrons added via addition of dissolved S(− II), with the slopeof linear regressions ranging from y=0.86 to 1.06 (see Supporting In-formation Fig. SI4).

The pH treatments displayed major differences in solid-phasepartitioning of Fe during sulfidization of jarosite (Fig. 4). AVS-Fedominated the solid-phase Fe pool at the final sulfidization stepand comprised >85% of total Fe at all pH values except pH 4.0,

jarosite with dissolved sulfide at pH 4.0, 5.0, 6.5 and 8.0. S(− II):Fe is the molar ratio of(aq) assuming complete dissolution of the initial jarosite and conservative behavior of

Fig. 3. (a) Electron flow to Fe(III) (ΣFe(II) species) during reaction of As(V)-bearingjarosite with dissolved sulfide at pH 4.0, 5.0, 6.5 and 8.0, where S(− II):Fe is themolar ratio of total dissolved sulfide added to total Fe; (b) Fe(III) reduction rate as afunction of sulfide oxidation rate. Dashed line is 2:1 slope; (c) relationship betweenjarosite dissolution [aqueous K+ normalized to total solid-phase K+ (KT)] and electronflow to Fe(III) [ΣFe(II) species normalized to total solid-phase Fe (FeT)]. Dashed line is1:1 slope.

14 S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

where Fe2+(aq) remained the dominant Fe phase and AVS-Fe wasb30%. Assuming that AVS was stoichiometric mackinawite, thefinal mass of mackinawite precipitate in the reaction vessel suspen-sion was ~0.5 g L−1 at pH 4.0, compared to 1.7–2.0 g L−1 at higherpH values (see Supporting Information Fig. SI5). At pH 5.0 an initialrapid increase in Fe2+(aq) was followed by a decline correspondingto AVS-Fe precipitation. At pH 6.5 and 8.0 there was a substantial ac-cumulation of solid-phase non-sulfidic Fe(II). This accounted for amaximum of 50% of the total Fe pool at pH 8.0 at an S(− II):Fe ratioof ~0.5 (Fig. 4). There was also a temporary formation of reactiveFe(III)(s) during the early stages of sulfidization at both pH 6.5 and8.0.

There were substantial increases in exchangeable As associated withsolid-phase precipitates that formed during sulfidization (Fig. 5). This in-crease was evident at all pH treatments and indicates a large shift frommainly co-precipitated to surface-complexed As during the sulfidizationprocess. Initial concentrations of AsEx were 0.0055 μmol per mmol Fe(s).This increased by 3 orders of magnitude, reaching a maximum of

2.6 μmol of AsEx per mmol Fe(s) at pH 4.0, S(−II):Fe ratio of 0.81. Afterattaining thismaximumAsEx at pH 4.0, therewas a subsequent steep de-cline to 0.3 μmol of AsEx per mmol Fe(s). At both pH 5.0 and 6.5 AsExpeaked around 1.7 μmol of As per mmol Fe(s) at S(−II):Fe ratio of 0.5,and thereafter steadily declined. At pH 8.0 AsEx remained reasonablystable after reaching a maximum of ~1.2 μmol of As per mmol Fe(s) at aS(−II):Fe ratio of 0.25. Further decreases in AsEx were evident followingaging for an additional 7 days.

The calculated re-sorption or (co)-precipitation of As (AsRe-sorbed)exhibited a strong pH dependence during sulfidization and wasgreatest at pH 6.5 (Fig. 6a). Throughout most of the sulfidization pro-cess the pH 4.0 treatment displayed the least effective re-sorption or(co)-precipitation of As, followed by pH 5.0 and pH 8.0. The propor-tion of mobilized As that was re-sorbed or (co)-precipitated generallydecreased with increasing S(− II):Fe ratios. An exception was thefinal sulfidization step at pH 4.0, where AsRe-sorbed increased substan-tially. Additional aging for a further 7 days also increased AsRe-sorbedacross all pH treatments. There was reasonably good agreement be-tween the measured AsEx and the calculated AsRe-sorbed with theslope of the regression close to 1:1. A notable exception was thefinal sulfidization step at pH 4.0 (Fig. 6b). This latter observation isconsistent with a process removing As from the aqueous-phase tothe solid-phase in a form that is not readily exchangeable withPO4

3−, such as an arsenic sulfide precipitate.

3.3. Mineralogy and solid-phase speciation

3.3.1. X-ray diffractionExamination of precipitates by XRD demonstrates that mackinawite

(FeSm)was a principalmineralogical end product of jarosite sulfidizationat all pH treatments (Fig. 7). However, the preceding mineral transfor-mation pathways varied substantially among treatments. In addition,rates of mineral transformation displayed a degree of pH dependenceand were generally faster at higher pH values. At pH 4.0, jarositepersisted until the S(−II):Fe ratio exceeded 1.1, after which point bothS(0) and mackinawite were dominant mineral end products. At pH 5.0,diffraction peaks attributable to jarosite were evident in the final stagesof sulfidization, but mackinawite formed much earlier than at pH 4.0[S(−II):Fe ratio of ~0.51], which is broadly consistent with solid-phasedata in Fig. 4. At pH 6.5, diffraction peaks associated with jarosite disap-pear at lower S(−II):Fe ratios and are replaced mainly by mackinawitewith minor lepidocrocite at the final sulfidization step.

At pH 8, there is clear evidence of very rapid formation oflepidocrocite (S(− II):Fe ratio of 0.04; 20 min after initiating the exper-iment). In particular, by the time the S(−II):Fe ratio had progressed to0.12, jarosite had been almost completely replaced by lepidocrocite asthe principal mineral phase. This is broadly consistent with the solid-phase data in Fig. 4 which shows an increase in reactive Fe(III) duringthe early stages of the sulfidization process. However, diffractionpeaks at pH 8.0 indicate that lepidocrocite eventually transforms tomackinawite as sulfidization progressed beyond a S(−II):Fe ratio of1.1. In contrast, jarosite remained the only detectable mineral phase inall no-sulfide control treatments (Supporting Information Fig. SI6).

3.3.2. Scanning and transmission electron microscopySEM and TEM were performed on selected samples during

sulfidization to identify solid-phase reaction products. During theearly stages of sulfidization at pH 5.0 (S(− II):Fe ratio of 0.25), jarositespheroids display clearly visible etching and pitting, consistent withS(−II) promoted dissolution (Fig. 8a). At pH 8.0 and a S(−II):Fe ratioof 0.12, an aggregated overgrowth of lepidocrocite laths is evident(Fig. 8b), displaying a spheroidal architecture with a size distributionsimilar to the initial jarosite. Investigation of this precipitate by TEMconfirmed the occurrence of distinct laths aggregated together as sphe-roidal overgrowths (Fig. 9a) displaying SAED diffraction patterns con-sistent with lepidocrocite (Fig. 9b). High-resolution TEM confirms

Fig. 4. Proportional speciation of iron during reaction of As(V)-bearing jarosite with dissolved sulfide at pH 4.0, 5.0, 6.5 and 8.0. Fe[x] is iron species; FeT is total iron; S(− II):Fe ismolar ratio of total dissolved sulfide added to total iron.

15S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

lepidocrocite (200) lattice fringes with a spacing of ~6.27 Å (Fig. 9c)and EDX spectra indicate that the laths were composed primarily of Feand O (Fig. 9d).

At the final sulfidization step at pH 8.0 (S(− II):Fe ratio of 1.45),mackinawite displayed plate-like morphology aggregated as over-growths with an underlying spheroid architecture, in a similar mannerto lepidocrocite (Figs. 8c, 10a). SAED d-spacings are consistent withmackinawite (Fig. 10b) while high-resolution TEM reveals (001) latticefringing with a spacing of ~5.03 Å (Fig. 10c). Corresponding EDX spec-tra are dominated by ~equimolar S and Fe (Fig. 10d). Mackinawiteoccurred as fused nanocrystals composed of ~5–10 atomic layers per-pendicular to the (001) plane and displayed poor outline definition.

Fig. 5. Evolution of exchangeable arsenic (AsEx) in precipitates (μmol As per mmol oftotal Fe(s)) as a function of solution pH and increasing S(− II):Fe ratio. Additionaldata points after dashed line represent solution aging for 7 d. No data available forpH 4 at 7 d.

Precipitate from the final sulfidization step of the pH 4.0 treatment(S(− II):Fe ratio of 1.41) also displays diffuse SAED d-spacings consis-tent with mackinawite and EDX spectra indicate approximately equi-molar S and Fe (Fig. 11). Additional EDX analysis revealed highlylocalized zones (nm scale) with large concentrations of As (up to15% by weight; Table 2; Fig. 11d, f) and substantial spatial variabilityin those concentrations. Despite such high concentrations of As beingsomewhat suggestive of the formation of a discrete As precipitate,SAED d-spacings at corresponding locations were inconclusive andgenerally diffuse, indicating the absence of long-range structuralorder.

3.3.3. X-ray absorption spectroscopyFe K-edge XANES and EXAFS spectra are presented in Fig. 12 and

the results of least-squares best fits summarized in Table 3. XAS analysisis broadly consistentwith XRD data and demonstrates that mackinawitewas the dominant Femineral phase at the conclusion of sulfidization ex-periments at all pH treatments (69–88% based on EXAFS LCF).

At both pH 4.0 and 5.0 the transformation from jarosite tomackinawite proceeded without detection of any other intermediateFe mineral phases. In contrast, during the pH 8.0 treatment XAS con-firms the extremely rapid transformation of jarosite to lepidocrocite.Linear combination fits indicate that ~81–97% of solid-phase Fe atpH 8.0 was transformed to lepidocrocite by an S(− II):Fe ratio of0.12. Fourier transforms (FT) indicate the development of backscat-tering at an absorber-scatterer distance consistent with Fe–Fe cornersharing, as expected for a lepidocrocite-dominated precipitate. Linearcombination fitting at pH 8.0 also suggests the occurrence of green-rustsulfate (8–12%) at an intermediate S(−II):Fe ratio of 0.51.

At pH 6.5 and a S(−II):Fe ratio of 0.12, FT shows a more intense Fe–Fe corner sharing peak compared to the initial jarosite and LCF indicateslepidocrocite accounted for 25–32% of solid-phase Fe at this point. Assulfidization progressed to a S(−II):Fe ratio of 0.51, there is an increasein the intensity of FT peaks consistent with Fe–Fe edge sharing and LCF

Fig. 6. (a) The proportion of arsenic released during jarosite dissolution that is eitherre-sorbed to the solid-phase or (co)-precipitated (AsRe-sorbed) as a function of pH andincreasing S(− II):Fe ratio. AsRe-sorbed=AsMax−As(aq); where AsMax is the maximumtheoretical aqueous As concentration possible due to dissolution of As(V)-bearingjarosite, assuming both congruency and complete partitioning of As to the aqueousphase. Jarosite dissolution is calculated from K+

(aq) concentrations, assuming conser-vative K+

(aq) behavior. In (a) AsRe-sorbed is presented normalized to AsMax. (b) Variationin exchangeable arsenic (AsEx) relative to AsRe-sorbed as a function of pH, where thesolid line represents 1:1 slope. The arrow in (b) corresponds to the final sulfidizationstep at pH 4.0, S(−II):Fe=1.41.

Fig. 7. X-ray powder diffractograms of precipitate formed during reaction ofAs(V)-bearing jarosite with dissolved sulfide at pH 4.0, 5.0, 6.5 and 8.0. The notationfor each sample indicates both the time period following initiation of reaction andthe molar ratio of dissolved sulfide added to total Fe. J, jarosite; S0, elemental sulfur;L, lepidocrocite; m, mackinawite; h, halite.

16 S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

suggests the occurrence of both goethite (33–50%) and green-rust sul-fate (16–36%).

Arsenic K-edge XANES and EXAFS spectra are shown in Fig. 13 andthe results of least-squares best fits summarized in Table 4. At theconclusion of the experiments at pH 5.0, 6.5 and 8.0 treatments, themajority of solid-phase As remains as oxygen-coordinated As(V),with As(III) species comprising 11–30% (S(− II):Fe ratio of >1.45).The proportion of As(III) apparent at pH 5.0 and 8.0 increased afteran additional 7 days of aging. At pH 4.0, there was a distinct transition(beginning at an S(− II):Fe ratio of 0.79) from oxygen-coordinatedAs(V) to sulfur-coordinated As species with a valency and EXAFS oscil-lations consistentwith that of orpiment. At the final sulfidization step atpH 4.0 (S(−II):Fe ratio 1.41), LCF indicates that this orpiment-likephase accounted for >86% of solid-phase As, which subsequently in-creased to 95% with an additional 7 days aging.

Modeled first shell fits for key As-bearing solid-phase end mem-bers (i.e. initial jarosite and pH 4.0, S(− II):Fe ratio 1.41) are shownin Table 5. For As(V)-bearing jarosite, the modeled first shell As–Ointer-atomic distance of 1.68 Å and coordination number (4.9) close-ly conforms to that reported in previous studies of As(V)-jarosite

(Paktunc and Dutrizac, 2003). For the pH 4.0 sulfidization end mem-ber, the modeled As–S first shell inter-atomic distance (2.26 Å) isconsistent with theoretical expectations for orpiment (Gibbs et al.,2010). The rationale for choosing orpiment crystallographic structurefor modeling first shell EXAFS scattering for As–S is based on a com-parison of published EXAFS spectra for realgar and orpiment (O'Dayet al., 2004a; Root et al., 2009). This comparison reveals that orpimentoscillations in k-space better represent the sulfidization end-membersample for pH 4.0, as realgar has large peaks in the FT between 2 and3.5 which are absent in the pH 4.0 end member.

3.3.4. Geochemical modelingThermodynamic solubility predictions during sulfidization indicate,

for most pH treatments, undersaturation of low temperature As–Smin-eral phases such as orpiment, amorphous As2S3 and realgar (Fig. 14).While this ismostly consistentwith observed data, at pH 5.0 themodel-ing shows that the aqueous phase is above saturation with regard toamorphous As2S3 and crystalline orpiment. However, there is no evi-dence for the actual formation of these phases at this pH — whichmay simply reflect kinetic constraints and a system far from equilibri-um. In addition, the thermodynamic modeling does not predict the

Fig. 8. SEM images of (a) etched and pitted jarosite spheroids after 2 h at pH 5.0, S(−II):Fe ratio=0.25, (b) lepidocrocite overgrowths with lath-like morphology after 1 h at pH8.0, S(−II):Fe ratio=0.12, (c) mackinawite with aggregated plate-like morphology ar-ranged in spheroids consistent with the original jarosite, after 7 d at pH 8.0, S(−II):Feratio=1.45.

17S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

formation of an As–S solid-phase at pH 4.0; yet there is strong evidence(i.e. TEM-EDX, XAS, wet-chemistry) supporting the formation of anorpiment-like species during the later stages of sulfidization at pH 4.0.

4. Discussion

The aqueous speciation, rates of electron transfer and mineraliza-tion products during sulfidization of As(V)-bearing jarosite all displaya high degree of pH dependence. The magnitude of arsenic mobiliza-tion generally increased steadily during sulfidization, attaining maxi-ma at the most highly sulfidized end-member in all except the pH 4.0treatment. For all pH treatments, jarosite was largely replaced bymackinawite as the principal mineral end-product.

4.1. Arsenic mobilization

Increased mobilization of arsenic during sulfidization of jarositecan be attributed to a variety of factors. A primary mechanism is thesulfide-promoted dissolution of the jarosite host-phase causing re-lease of co-precipitated As, combined with a subsequent decrease inthe jarosite surface area available for re-sorption of As. In addition,the steady replacement of jarosite by mackinawite as sulfidizationprogressed has important consequences for As mobilization, due tothe widely contrasting sorption affinity of these mineral phases forAs (discussed below).

Our findings are consistent with several recent studies (i.e. Kirk etal., 2010; Kocar et al., 2010; Burton et al., 2011a, 2011b), which foundincreased mobilization of As during sulfidization of the ferrihydritehost-phase and relatively weak partitioning of As to newly formedmackinawite. Wolthers et al. (2005) reported fast, but relativelyweak As sorption to disordered mackinawite, with an observed max-imum sorption of 4.9 and 1.3 μmol of As(V) and As(III) respectivelyper gram of FeS. Another more crystalline FeS-phase (i.e. troilite)was reported by Bostick and Fendorf (2003) to have a sorption max-imum of 14 μmol of As(III) per gram of FeS. In contrast, a study byAsta et al. (2009) at low pH (1.5–2.5) found a comparatively highlevel of As sorption to jarosite, with reported maxima of 280 μmolof As(V) per gram of jarosite; i.e. about 2–3 orders of magnitudehigher than mackinawite. However, to the authors knowledge, therehave been no systematic studies to date of As(V) and As(III) sorptionto jarosite across the broad pH range examined here.

Changes in aqueous As speciation during sulfidization of jarositeare also likely to have enhanced As mobility and contributed to theobserved pH-dependent variability. Rochette et al. (2000) demon-strated that reduction of arsenate by hydrogen sulfide can be rapidand is ~300 times faster at pH 4 than pH 7. Despite limited quantita-tive recovery, some As(III) species were found in all aqueous solu-tions subject to speciation analysis, reflecting reduction of As(V).Sorption of As(III) to mackinawite is weaker than As(V) (Woltherset al., 2005). The fact that aqueous samples from pH 4.0 and 5.0 atlow S(− II):Fe ratios (0.24 and 0.12 respectively) comprised arsenite,mono- and di-thioarsenate species with no detectable arsenate, sug-gests that changes in As speciation may have been an important fac-tor driving mobilization of As during the early stages of jarositesulfidization.

Recent investigations demonstrate that both As(V) (Stauder et al.,2005; Wallschläger and Stadey, 2007) and As(III) thioanion species canbe dominant in sulfidic solutions (Planer-Friedrich et al., 2010).According to Planer-Friedrich et al. (2010), the proportion of thioarsenitespecies increases with increasing aqueous S:As ratios, yet these speciesare highly reactive and readily decompose to more stable thioarsenatespecies. Mono- and di-thioarsenatewas detected in almost every samplefrom this study thatwas subject to As speciation analysis. Few systematicstudies exist to date on the sorption behavior of thioarsenates to ironmineral phases. Suess and Planer-Friedrich (2012) have investigatedthioarsenate sorption to α-goethite at pH 9–10 and shown slowersorption kinetics and a lower amount of total sorbed arsenic formonothioarsenate compared to arsenite and arsenate. Wilkin (2001)also suggested thiolated aqueous As species as having weaker sorptionaffinity for mackinawite than their oxygen coordinated counterparts.However, there is still a considerable gap in our understanding of Asthioanion sorption behavior in iron-sulfide systems which is an impor-tant topic for future research.

Except for the final sulfidization step at pH 4.0 (where As(aq) was at-tenuated by 50% compared to the preceding sampling point), the mobi-lization of As was greatest at lower pH values and followed the order5.0≈4.0>8.0>6.5. This is broadly consistent with Wolthers et al.(2005) who found As(V) and As(III) sorption maxima for mackinawitenear circumneutral pH. It is also consistent with Bostick and Fendorf(2003) study of As(III) sorption to troilite, whereby sorption affinity

Fig. 10. Precipitates formed at pH 8.0, S(−II):Fe ratio=1.45 (a) representative TEM image showing mackinawite, arranged in spheroids consistent with the original jarosite (b) SAEDd-spacings consistent with mackinawite, (c) high-resolution TEM showing mackinawite lattice fringes (001, ~5.03 Å) and (d) corresponding EDX spectrum.

Fig. 9. Precipitates formed at pH 8.0, S(− II):Fe ratio=0.12 (a) representative TEM image showing lepidocrocite with lath-like morphology occurring as overgrowths on the originaljarosite spheroids (b) SAED d-spacings consistent with lepidocrocite, (c) high-resolution TEM showing lepidocrocite lattice fringes (200, ~6.27 Å) and (d) corresponding EDXspectrum.

18 S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

decreased with decreasing pH and little sorption occurred below pH 6.Importantly, Bostick and Fendorf (2003) also demonstrated that S(−II)can inhibit As(III) sorption to troilite (FeS) via competitive adsorption.Hence, the higher concentrations of dissolved S(−II) at pH 4.0 and 5.0(typically 1–2 orders of magnitude higher than at pH 6.5 and 8.0;Fig. 2) are likely to have enhanced any competitive anion adsorption ef-fects between S(−II), As and the mackinawite surface.

Fig. 11. Precipitates formed at pH 4.0, S(− II):Fe ratio=1.41 (a) representative TEM imenergy-dispersive X-ray (EDX) spectrum. (d) TEM image showing the location of EDX spot

4.2. Solid-phase arsenic speciation

During sulfide-induced jarosite dissolution and associated mineraltransformations, much of the mobilized As was re-partitioned to areadily exchangeable surface complex on remaining precipitate(Fig. 5). The efficiency of this As re-sorption is consistent with previ-ously described sorption behavior and was strongly pH dependent,

age with (b) SAED d-spacings consistent with mackinawite and (c) correspondings that correspond to Table 2, (e) EDX spectrum at spot ‘Y’.

Table 2Spatial variation in As content of precipitates formed at pH 4, S(−II):Femolar ratio=1.41. Based on energy dispersive X-ray (EDX) spot analysis(*See Fig. 11d for spot locations).

EDX spot* As(% w/w)

V 1.4W ndX 15.6Y 11.1Z 5.0

nd = below detection limit.

19S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

following the order 6.5>8.0>5.0>4.0 throughout most of thesulfidization reaction (Fig. 6a). At pH 6.5 and 8.0, over 90% of the Asreleased by jarosite dissolution was effectively either re-sorbed or

Fig. 12. Fe K-edge XANES spectra, k3-weighted EXAFS spectra and Fourier transform (uAs(V)-bearing jarosite reacted with dissolved sulfide. The molar ratio of S(−II):Fe for eacand colored dashed lines are least-squares best fits. XANES and EXAFS fits are provided in Tsitions of peaks due to the backscattering contributions of coordinating atoms.

(co)-precipitated with remaining solid phase until the latter stagesof sulfidization were reached i.e. S(− II):Fe ratio >1.4. In general, Asre-sorption was least effective at the highest S(− II):Fe ratio. Howev-er, a marked exception occurred at the final sulfidization step of thepH 4.0 treatment, where several independent lines of evidence sug-gest the formation of an orpiment-like species (discussed below).Additional decreases in AsEx following aging for 7 days suggestincreasing resistance to desorption over time and are consistentwith time dependent sorption behavior previously reported for As(Zhang and Selim, 2005; Yang et al., 2012).

Solution pH greatly influenced the speciation and coordination ofsolid-phase As during jarosite sulfidization. Despite high S(− II):Fe ra-tios and the co-occurrence of≥100 μM S(−II) in solution, at pH values>5.0 the majority of solid-phase As remained as oxygen-coordinatedAs(V). Some As(III) was evident (~11–34%), yet this also remainedoxygen coordinated. This is broadly consistent with the findings of

ncorrected for phase shift) for reference compounds and experimental samples ofh pH treatment is shown after the semi-colon. Solid lines are the experimental dataable 4. The vertical dashed lines in the Fourier transform indicate the approximate po-

Table 3Summary of Fe mineralogy of solid-phase precipitates based on XANES and EXAFS linear combination fits. EXAFS fit over k-range 2–10.

XANES/EXAFS (%)

pH/S(−II):Fe

Jarosite Lepidocrocite Goethite Green rust Mackinawite R-factor

4.0/0.12 100/100 – – – – 0.0002/0.00374.0/0.50 99/98 – – – 1/2 0.0001/0.00344.0/1.41 4/3 –/9 – – 96/88 0.0003/0.04615.0/0.12 100/100 – – – – 0.0001/0.00455.0/0.51 72/70 – – – 28/30 0.0001/0.00695.0/1.45 35/31 – – – 65/69 0.0002/0.04596.5/0.12 75/64 25/32 –/4 – – 0.0002/0.00436.5/0.51 8/8 – 33/50 36/16 23/26 0.0001/0.02936.5/1.45 –/1 – 8/13 – 92/86 0.0003/0.04618.0/0.12 3/6 97/81 –/9 –/5 – 0.0002/0.00688.0/0.51 – 51/51 –/12 12/8 37/29 0.0001/0.01398.0/1.45 – 17/7 –/17 – 83/76 0.0001/0.0270

20 S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

Farquhar et al. (2002), who reported As(V) and As(III) formedouter-sphere complexes during adsorption to the mackinawite surface(pH 5.5–6.5), whereby As(V) and As(III) retained their original oxida-tion states and first-shell oxygen coordination environment. It is alsocomparable with Wolthers et al. (2005) who invoked outer-spherecomplexation to explain As(V) and As(III) adsorption behavior tomackinawite.

Fig. 13. As K-edge XANES spectra, k3-weighted EXAFS spectra and Fourier transform (uAs(V)-bearing jarosite reacted with dissolved sulfide. Solid lines are experimental data andpH treatment is shown after the semi-colon. XANES and EXAFS fits are provided in Table 5of peaks due to the backscattering contributions from first shell As–O and first shell As–S r

Solid-phase As speciation and coordination at pH 4.0 evolved in amanner distinctly different to all other pH treatments. Both XANES andEXAFS clearly indicate a transition from oxygen-coordinated As(V) toan As–S coordination with an orpiment-like valency at S(−II):Fe ratios>1. This is supported by first-shell modeling fits and also coincideswith a marked decrease in AsEx — a feature which is consistent withthe formation of an orpiment-like species. The fact that geochemical

ncorrected for phase shift) for reference compounds and experimental samples ofcolored dashed lines are least-squares best fits. The molar ratio of S(− II):Fe for each

. The vertical dashed lines in the Fourier transform indicate the approximate positionsespectively.

Table 4Summary of As speciation of solid-phase precipitates based on XANES and EXAFS linearcombination fits. EXAFS fit over k-range 2–10.

pH/S(−II):Fe XANES/EXAFS (%)

As(V)a

(%)As(III)b

(%)Orpiment(%)

R-factor

4.0/0.12 100/94 –/5 –/1 0.00003/0.00994.0/0.50 92/92 8/5 –/3 0.00013/0.01134.0/0.79 72/87 24/2 4/11 0.00069/0.01274.0/1.09 41/31 –/18 59/51 0.00037/0.05164.0/1.41 11/11 –/3 89/86 0.00082/0.04714.0/1.41 (7 d) 5/3 –/2 95/95 0.00098/0.06005.0/1.45 88/67 12/33 –/– 0.00697/0.03445.0/1.45 (7 d) 68/39 32/53 –/8 0.00448/0.06606.5/1.46 85/62 15/34 –/4 0.00172/0.08976.5/1.46 (7 d) 85/62 15/36 –/3 0.00105/0.07888.0/1.45 88/66 11/32 –/2 0.00082/0.06468.0/1.45 (7 d) 83/43 17/54 –/3 0.00582/0.0880

a As(V)-synthetic jarosite.b As(III)-sorbed schwertmannite.

21S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

modeling did not predict the formation of an As–S solid-phase at pH 4.0may reflect the possible inclusion of colloidal FeS clusters in the‘dissolved’ S(−II) fraction, which would lead to an over-estimate of‘dissolved’ S(−II) species. Importantly, TEM-EDX data from the finalsampling point [S(−II):Fe ratio 1.41] indicate the formation of highly lo-calized zones of nanometer scale displaying considerable As enrichment(up to 15%). Combined, these data lead us to conclude that the formationof an orpiment-like specieswas a primary control on aqueous As concen-trations during the later stages of jarosite sulfidization at pH 4.0.

Hence, lower pH (~4–5) appears to have a contrasting dual effectwhereby As mobilization is favored over neutral pH conditions at midS(− II):Fe ratios, yet as S(− II) ratios increase As(aq) may be attenuat-ed by formation of an orpiment-like species. These observations areconsistent with the findings of Renock et al. (2009) who reportedformation of discrete, nanometer-scale As-sulfide precipitates whenAs(III) interacted with mackinawite at low pH values (pH 5.0).Gallegos et al. (2007) also reported As(III) uptake from solution bymackinawite via the formation of a ‘realgar-like’ precipitate and, im-portantly, that the formation of this precipitate was favored atlower pH and higher As concentrations.

4.3. Electron transfer, Fe(III) reduction and mineral transformations

The proposed reaction mechanism between dissolved sulfide andFe(III) oxyhydroxides is regarded as reasonably well understood(Dos Santos Afonso and Stumm, 1992; Peiffer et al., 1992; Yao andMillero, 1996; Poulton et al., 2004; Peiffer and Gade, 2007; Helligeet al., 2012). The reaction is thought to involve the formation ofinner-sphere surface complexes prior to electron transfer. As a conse-quence it is highly pH dependent, with maximum S(− II) oxidationrates near circumneutral pH (Dos Santos Afonso and Stumm, 1992;Peiffer et al., 1992).

Table 5Results of first shell fits of the As K-edge EXAFS spectrum for the key end member sam-ples (a) initial starting phase As(V)-synthetic jarosite (amplitude reduction factor of 1) and(b) final sulfidization precipitate from pH 4 treatment, S(−II):Fe=1.41; 7 d (amplitudereduction factor of 0.867).

Sample Path CN R (Å) σ2 (Å2) ΔEo R-factor

(a) As(V)-syntheticjarosite

As–O 4.9[4.6–5.4]A

1.676±0.0123[1.68]A

0.0017 13.0 0.00112

(b) pH 4, S(−II):Fe=1.45; 7 d

As–S 2.8 2.259±0.0069[2.24–2.31]B

0.0023 11.0 0.00021

A CN and As–O inter-atomic distance for arsenate substituted jarosite (Paktunc andDutrizac, 2003).

B As–S first shell inter-atomic distance for orpiment (Gibbs et al., 2010).

Solution pH clearly had a large effect on rates of electron transferduring sulfidization of jarosite. Observed behavior was broadly consis-tent with prior studies of electron flow between dissolved S(− II) andFe(III) oxyhydroxides. For example, mean sulfide oxidation ratesattained amaximumat pH 6.5 andwere lowest under acidic conditions,which accords with the findings of Dos Santos Afonso and Stumm(1992) and Peiffer et al. (1992). Correspondingly, Fe(III) reductionwas also least efficient under acidic conditions. However, despitepH-dependent differences inmean ratemagnitude, the relationship be-tween rates of Fe(III) reduction and corresponding S(−II) oxidation ap-proximates the 2:1 slope expected based on Eq. (1).

Mackinawitewas the primary Femineral endproduct in all pH treat-ments. The fundamental reaction stoichiometry outlined in Eqs. (1) and(2) indicates that a total of 1.5 M S(− II) is sufficient to reduce 1 MFe(III) to Fe2+, and then react with the Fe2+ to form FeSm. The domi-nance of mackinawite at S(− II):Fe ratios >1.4 is consistent with thisstoichiometry. Mackinawite morphology varied considerably betweenpH treatments. At pH 8.0, mackinawite was morphologically well de-fined, with distinct platelets aggregated in spheroids of similar size dis-tribution to the original synthetic jarosite (Fig. 8c). In contrast,mackinawite had indistinct morphology at pH 4.0. This difference mayreflect the higher solubility ofmackinawite at lower pH and the relativeinstability of its surface under acidic conditions.

There was a considerable variation among pH treatments in theFe(III) reduction rates and the formation of intermediate Fe and S min-eral phases. In particular, the reduction of Fe(III) was slowest at pH 4.0.Additionally, at pH 4.0most Fe(II) remained in solution as Fe2+(aq), pre-sumably due to faster dissolution from protonated surface sites underacidic conditions (Poulton, 2003; Poulton et al., 2004). Considerable el-emental S accumulated at pH 4.0, and the final quantity of mackinawite(0.5 g L−1) was only ~25% of the other pH treatments, which may beattributed to the higher solubility of FeSm under acidic conditions.Although faster than at pH 4.0, the transformation of jarosite tomackinawite at pH 5.0 proceededwithout detecting any Fe or Smineralintermediaries via the techniques applied here. In contrast, at pH 6.5some intermediate Fe(III) phases were detected via XAS, includinglepidocrocite, goethite and green rust sulfate (GRS). However, neithergoethite nor GRS was evident by XRD nor independently verified byTEM-SAED. This suggests they may have been poorly-crystalline andlacking long-range order or possibly an artifact of XAS LCF fitting(O'Day et al., 2004b). At pH 8.0, jarosite rapidly transformed tolepidocrocite (see Section 4.3.1 for more detailed discussion), followedby a gradual transformation to mackinawite.

An important feature at pH 6.5 and 8.0 was the occurrence ofnon-sulfidic solid-phase Fe(II). At pH 8.0 there was little Fe2+ in solu-tion (b10 μmol), yet during the intermediate stages of sulfidizationthis non-sulfidic solid-phase Fe(II) accounted for up to 50% of totalFe. This may partially comprise a sorbed Fe(II) phase and the lack ofFe2+ in solution yet continued Fe(III) reduction suggests that detach-ment of Fe(II) from the surface is not required for Fe(III) reduction tocontinue. This is consistent with the findings of Poulton (2003) andHellige et al. (2012), who also observed a considerable fraction ofFe(II) in a non-sulfidic surface-associated solid-phase.

4.3.1. Rapid transformation of jarosite to lepidocrocite at pH 8At pH 4.0, 5.0 and 6.5, the dissolution of jarosite was broadly con-

gruent with the overall electron transfer to Fe(III). This is consistentwith the notion that reduction of structural Fe(III) was the primarycause of jarosite dissolution at these pH values. However, at pH 8.0jarosite underwent a rapid dissolution that was clearly incongruentwith the overall electron transfer to Fe(III) (Fig. 3c). Instead, jarositerapidly transformed (b1 h) to lepidocrocite at pH 8.0. The aggrega-tion of lepidocrocite laths as overgrowths with an underlying spheri-cal morphology matching that of the original jarosite suggests thatthe dissolution/re-precipitation reaction was both fast and occurred at

Fig. 14. Variation in aqueous arsenic and sulfide concentrations at (a) pH 4.0 and 5.0and (b) 6.5 and 8.0 in relation to aqueous arsenic speciation and the solubility of arse-nic sulfide phases. Boundaries were calculated using The Geochemist's Workbench(Bethke, 2007) with thermodynamic constants as provided by Kirk et al. (2010) [derivedfrom Helz and Tossell (2008) and Nordstrom and Archer (2003)]. Activities are;SO4

2−(aq)=10−3 M; Cl−(aq) 10−1 M; Na+(aq) 10−1 M.

22 S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

or in very close proximity to the initial jarosite surface. Transformationdid not occur in pH 8.0 control experiments without added S(−II).

The extremely rapid transformation of jarosite to lepidocrocite atpH 8.0 following interaction with dissolved S(− II) is an unexpectedand somewhat novel finding that, to the authors knowledge, has notbeen reported before. This transformation of jarosite to lepidocrocitecould be triggered by direct electron transfer from S(−II), or alternative-ly, by indirect and catalytic electron transfer from surface complexedFe(II) that was formed via S(−II) oxidation. Our suggestion that thetransformation process may be catalytic has considerable precedent.For example, a range of studies have demonstrated that Fe(II) can cata-lyze the structural transformation of meta-stable Fe(III) minerals tomore crystalline and thermodynamically stable phases (e.g. Hansel etal., 2003; Pedersen et al., 2005; Burton et al., 2007, 2008b; Jones et al.,2009). Of particular relevance to this study are the findings by Jones etal. (2009), who reported Fe(II)-catalyzed transformation of jarosite tolepidocrocite over 7 days at pH 6.5. The catalytic transformation processrequires electron transfer from Fe(II) that is sorbed to the Fe(III) mineralsurface and involves a rapid dissolution/re-precipitation mechanism

(Williams and Scherer, 2004). Higher pH values appear to speed upthis catalytic transformation process; presumably by creating surfacecharge conditions that are more favorable for Fe(II) sorption at theFe(III) mineral surface (Burton et al., 2008b).

Hence, it remains uncertain whether the transformation of jarositeto lepidocrocite described here is due to direct electron transfer fromS(−II), or by indirect catalytic electron transfer from surface complexedFe(II). Further research is required to resolve the precise mechanism.However, an important question arising from this finding is “do similarrapid Fe mineral transformations occur when S(− II) interacts with otherFe(III) oxides at slightly alkaline pH?”. The answer to this question mayhave considerable implications for the interpretation of prior studiesof S(− II) oxidation by different Fe(III) oxides at slightly alkaline pHwhich did not consider possible Fe(III) re-crystallization during the re-action process (e.g. Peiffer et al., 1992; Poulton et al., 2004). This is atopic worthy of further investigation.

4.4. Environmental implications

Although this study demonstrates that sulfide-promoted reduc-tive dissolution of As(V)-bearing jarosite can enhance As mobility, itis important to note that this is an abiotic study which lacks the com-plexity of natural systems. In natural sediments there is a wide rangeof other factors thatwill influence As behavior during As-bearing jarositesulfidization, including; hydraulic gradients, which may transport As(aq)away from the initial point of mobilization; microbial consortia; humicsubstances; competing ions and the presence of a wide range of sorbentsurfaces (i.e. clays, other minerals). In addition, sulfidization duringthis experiment was forced and relatively rapid, thereby generatingnon-equilibrium conditions in the reaction vessels. In natural sedi-ments the reaction kinetics are likely to be much slower, whichmay favor different mineralization pathways and alternative solid/aqueous phase partitioning.

The interaction between S(− II) and jarosite in re-flooded CASSwetlands relies on sediment heterogeneity and the development ofhighly contrasting geochemical micro-niches. A conceptual representa-tion of interaction between S(−II) and As-bearing jarosite is providedin Fig. 15, alongside an example of such sediment. In this example ofnatural sediment, zones of distinct black coloration are evident (charac-terized by sulfate reduction) and occur adjacent to residual jarosite pre-cipitates (yellow color). This study shows that the magnitude of Asmobilization during jarosite sulfidization is strongly influenced byboth pH and S(−II):Fe ratios. Both of these factors are likely to varysharply over short distances (mm scale) in the example sedimentshown in Fig. 15. In a recent study, Masue-Slowey et al. (2010) demon-strated that considerable heterogeneity can be established in Fe and Asspeciation at the mm-scale across redox zones in soil aggregates.Detailed spatial investigations (perhaps utilizing micro-XRF andmicro-XANES) may be required to unravel the potentially complexpatterns of As speciation and distribution that are likely to occuracross mm-scale redox zones of the wetland sediments describedhere.

5. Conclusions

Sulfide-promoted reductive dissolution of As(V)-bearing jarositeand its replacement by mackinawite greatly enhanced As mobility.The magnitude of mobilization was a function of both S(− II):Fe ratiosand pH, with generally greater mobility observed under acidic condi-tions and with increasing S(− II):Fe ratios. An exception occurred athigh S(− II):Fe ratios at pH 4.0, where As(aq) was attenuated viaformation of a solid-phase orpiment-like species. Large increases inthe pool of phosphate-exchangeable surface-complexed As reflectsre-partitioning of mobilized As to remaining precipitate and has im-portant implications for As-bearing jarosite in sulfidogenic environ-ments. The findings presented here are of particular relevance to As

Fig. 15. Photograph of a re-flooded acid sulfate soil (left) and conceptual representa-tion of redox-pH micro-niche (right). Circle inset ‘a’ highlights a location where sulfatereduction by sulfate reducing bacteria (SRB) and subsequent S(− II) generation (blackcolor) co-occurs adjacent to As-bearing jarosite (yellow color). Far left scale is in cm.Soil profile photograph is from the site described in Johnston et al. (2011a) [Transect1, 80 m].

23S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

cycling in CASS and AMD wetlands whereby minerals such as jarositemay encounter biogenic S(− II) under a range of pH conditions.

Acknowledgments

Research expenses and salary support for Scott Johnston was pro-vided by the Australian Research Council Future Fellowship (grant no.FT110100130). Salary support for Ed Burton was provided by theAustralian Research Council (grant no. DP110100519). The As XAScomponents were undertaken at the ANBF at the Photon Factory inJapan, supported by the Australian Synchrotron Research Program(ASRP) (grant no. AS111/ANBF/3271). We acknowledge the LIEF Pro-gram of the ARC for financial support (LE0989759) and the High En-ergy Accelerator Research Organisation (KEK) in Tsukuba, Japan, foroperations support. The Fe XAS components were supported by theASRP (grant no. AS112/XAS3855). We thank Micheal Cheah and ChrisGlover for their technical assistance and expert advicewith XAS data col-lection. The TEMcomponents of this studywere funded by theAustralianInstitute of Nuclear Science and Engineering (grant no. ALNGRA11079).Arsenic speciation analysis at Britta Planer-Friedrich's laboratory wasfunded by the German Research Foundation within the Emmy Noetherprogram (grant # PL 302/3-1). We thank Maxine Dawes for assistancewith SEM analysis and Michelle Bush for assistance with XRD analysis.Many thanks to William Burgos for BET surface area measurements ofsynthetic jarosite.

Appendix A. Supplementary data

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.chemgeo.2012.09.045.

References

Acero, P., Ayora, C., Torrentó, C., Nieto, J.M., 2006. The behavior of trace elements duringschwertmannite precipitation and subsequent transformation into goethite andjarosite. Geochimica et Cosmochimica Acta 70 (16), 4130–4139.

Ankudinov, A.L., Ravel, B., Rehr, J.J., Conradson, S.D., 1998. Real-space multiple-scatteringcalculation and interpretation of X-ray-absorption near-edge structure. Physical Re-view B 58 (12), 7565–7576.

APHA, 2005. Standard Methods for the Examination of Water andWastewater. AmericanPublic Health Association — American Water Works Association, Baltimore, USA.

ASRP, 2007. Average v 2.0 for Windows. Australian Synchrotron Research Program.Asta, M.P., Cama, J., Martínez, M., Giménez, J., 2009. Arsenic removal by goethite and

jarosite in acidic conditions and its environmental implications. Journal of HazardousMaterials 171 (1–3), 965–972.

Asta, M.P., Ayora, C., Román-Ross, G., Cama, J., Acero, P., Gault, A.G., Charnock, J.M.,Bardelli, F., 2010. Natural attenuation of arsenic in the Tinto Santa Rosa acid stream

(Iberian Pyritic Belt, SW Spain): the role of iron precipitates. Chemical Geology 271(1–2), 1–12.

Baron, D., Palmer, C.D., 1996. Solubility of jarosite at 4–35 °C. Geochimica et CosmochimicaActa 60 (2), 185–195.

Beak, D.G., Wilkin, R.T., Ford, R.G., Kelly, S.D., 2008. Examination of arsenic speciation insulfidic solutions using X-ray absorption spectroscopy. Environmental Science &Technology 42 (5), 1643–1650.

Bethke, C.M., 2007. The Geochemist’s Workbench. Hydrogeology Program, Universityof Illinois.

Bigham, J.M., Schwertmann, U., Traina, S.J., Winland, R.L., Wolf, M., 1996. Schwertmanniteand the chemical modeling of iron in acid sulfate waters. Geochimica et CosmochimicaActa 60 (12), 2111–2121.

Bostick, B.C., Fendorf, S., 2003. Arsenite sorption on troilite (FeS) and pyrite (FeS2).Geochimica et Cosmochimica Acta 67 (5), 909–921.

Bostick, B.C., Chen, C., Fendorf, S., 2004. Arsenite retentionmechanismswithin estuarine sed-iments of Pescadero, CA. Environmental Science & Technology 38 (12), 3299–3304.

Burton, E.D., Bush, R.T., Sullivan, L.A., Mitchell, D.R.G., 2007. Reductive transformationof iron and sulfur in schwertmannite-rich accumulations associated with acidifiedcoastal lowlands. Geochimica et Cosmochimica Acta 71 (18), 4456–4473.

Burton, E.D., Bush, R.T., Sullivan, L.A., Johnston, S.G., Hocking, R.K., 2008a. Mobility ofarsenic and selected metals during re-flooding of iron- and organic-rich acid-sulfate soil. Chemical Geology 253 (1–2), 64–73.

Burton, E.D., Bush, R.T., Sullivan, L.A., Mitchell, D.R.G., 2008b. Schwertmannite transfor-mation to goethite via the Fe(II) pathway: reaction rates and implications for iron-sulfide formation. Geochimica et Cosmochimica Acta 72 (18), 4551–4564.

Burton, E.D., Bush, R.T., Johnston, S.G., Watling, K.M., Hocking, R.K., Sullivan, L.A.,Parker, G.K., 2009. Sorption of arsenic(V) and arsenic(III) to schwertmannite. Envi-ronmental Science & Technology 43 (24), 9202–9207.

Burton, E.D., Bush, R.T., Johnston, S.G., Sullivan, L.A., Keene, A.F., 2011a. Sulfur biogeo-chemical cycling and novel Fe-S mineralization pathways in a tidally re-floodedwetland. Geochimica et Cosmochimica Acta 75 (12), 3434–3451.

Burton, E.D., Johnston, S.G., Bush, R.T., 2011b. Microbial sulfidogenesis in ferrihydrite-rich environments: effects on iron mineralogy and arsenic mobility. Geochimicaet Cosmochimica Acta 75 (11), 3072–3087.

Charlet, L., Morin, G., Rose, J., Wang, Y., Auffan, M., Burnol, A., Fernandez-Martinez, A.,2011. Reactivity at (nano)particle–water interfaces, redox processes, and arsenictransport in the environment. Comptes Rendus Geoscience 343 (2–3), 123–139.

Claff, S.R., Sullivan, L.A., Burton, E.D., Bush, R.T., 2010. A sequential extraction procedurefor acid sulfate soils: partitioning of iron. Geoderma 155 (3–4), 224–230.

Cline, J.D., 1969. Spectrophotometric determination of hydrogen sulfide in natural wa-ters. Limnology and Oceanography 14, 454–458.

Dos Santos Afonso, M., Stumm, W., 1992. Reductive dissolution of iron(III) (hydr)ox-ides by hydrogen sulfide. Langmuir 8 (6), 1671–1675.

Dudas, M.J., 1984. Enriched levels of arsenic in post-active acid sulfate soils in Alberta.Soil Science Society of America Journal 48, 1451–1452.

Eary, L.E., 1992. The solubility of amorphous As2S3 from 25 to 90 °C. Geochimica etCosmochimica Acta 56 (6), 2267–2280.

Egal, M., Casiot, C., Morin, G., Parmentier, M., Bruneel, O., Lebrun, S., Elbaz-Poulichet, F.,2009. Kinetic control on the formation of tooeleite, schwertmannite and jarosite byAcidithiobacillus ferrooxidans strains in an As(III)-rich acid mine water. ChemicalGeology 265 (3–4), 432–441.

Eiche, E., Neumann, T., Berg, M., Weinman, B., van Geen, A., Norra, S., Berner, Z., Trang,P.T.K., Viet, P.H., Stüben, D., 2008. Geochemical processes underlying a sharp con-trast in groundwater arsenic concentrations in a village on the Red River delta,Vietnam. Applied Geochemistry 23 (11), 3143–3154.

Farquhar, M.L., Charnock, J.M., Livens, F.R., Vaughan, D.J., 2002. Mechanisms of arsenic up-take from aqueous solution by interaction with goethite, lepidocrocite, mackinawiteand pyrite: an X-ray absorption spectroscopy study. Environmental Science & Tech-nology 36, 1757–1762.

Gallegos, T.J., Hyun, S.P., Hayes, K.F., 2007. Spectroscopic investigation of the uptake ofarsenite from solution by synthetic mackinawite. Environmental Science & Tech-nology 41 (22), 7781–7786.

Gibbs, G.V., Wallace, A.F., Zallen, R., Downs, R.T., Ross, N.L., Cox, D.F., Rosso, K.M., 2010.Bond paths and van der Waals interactions in orpiment, As2S3. The Journal ofPhysical Chemistry. A 114 (23), 6550–6557.

Hansel, C.M., Benner, S.G., Neiss, J., Dohnalkova, A., Kukkadapu, R.K., Fendorf, S., 2003. Sec-ondarymineralization pathways induced by dissimilatory iron reduction of ferrihydriteunder advective flow. Geochimica et Cosmochimica Acta 67 (16), 2977–2992.

Hellige, K., Pollok, K., Larese-Casanova, P., Behrends, T., Peiffer, S., 2012. Pathways of fer-rous ironmineral formation upon sulfidation of lepidocrocite surfaces. Geochimica etCosmochimica Acta 81, 69–81.

Helz, G.R., Tossell, J.A., 2008. Thermodynamic model for arsenic speciation in sulfidicwaters: a novel use of ab initio computations. Geochimica et Cosmochimica Acta72, 4457–4468.

Ivarson, K., Hallberg, R., 1976. Formation of mackinawite by the microbial reduction ofjarosite and its application to tidal sediments. Geoderma 16, 1–7.

Johnston, S.G., Keene, A.F., Bush, R.T., Burton, E.D., Sullivan, L.A., Smith, D., McElnea,A.E., Martens, M.A., Wilbraham, S., 2009. Contemporary pedogenesis of severelydegraded tropical acid sulfate soils after introduction of regular tidal inundation.Geoderma 149 (3–4), 335–346.

Johnston, S.G., Keene, A.F., Burton, E.D., Bush, R.T., Sullivan, L.A., McElnea, A.E., Ahern, C.R.,Smith, C.D., Powell, B., Hocking, R.K., 2010. Arsenic mobilization in a seawater inun-dated acid sulfate soil. Environmental Science & Technology 44 (6), 1968–1973.

Johnston, S.G., Keene, A.F., Bush, R.T., Burton, E.D., Sullivan, L.A., Isaacson, L., McElnea,A.E., Ahern, C.R., Smith, C.D., Powell, B., 2011a. Iron geochemical zonation in a tid-ally inundated acid sulfate soil wetland. Chemical Geology 280 (3–4), 257–270.

24 S.G. Johnston et al. / Chemical Geology 334 (2012) 9–24

Johnston, S.G., Keene, A.F., Burton, E.D., Bush, R.T., Sullivan, L.A., 2011b. Iron and arseniccycling in intertidal surface sediments during wetland remediation. EnvironmentalScience & Technology 45 (6), 2179–2185.

Jones, E.P.J., Nadeau, T.L., Voytek, M.A., Landa, E.R., 2006. Role of microbial iron reduc-tion in the dissolution of iron hydroxysulfate minerals. Journal of Geophysical Re-search 111 (G1), G01012.

Jones, A.M., Collins, R.N., Rose, J., Waite, T.D., 2009. The effect of silica and natural or-ganic matter on the Fe(II)-catalysed transformation and reactivity of Fe(III) min-erals. Geochimica et Cosmochimica Acta 73 (15), 4409–4422.

Kandegedara, A., Rorabacher, D.B., 1999. Noncomplexing tertiary amines as “better”buffers covering the range of pH 3–11. Temperature dependence of their acid dis-sociation constants. Analytical Chemistry 71 (15), 3140–3144.

Kelly, S.D., Hesterberg, D., Ravel, B. (Eds.), 2008. Analysis of Soils and Minerals Using X-rayAbsorption Spectroscopy. Methods of Soil Analysis, Part 5 — Mineralogical Methods.Soil Science Society of America, Madison, WI, USA.

Keon, N.E., Swartz, C.H., Brabander, D.J., Harvey, C., Hemond, H.F., 2001. Validation of anarsenic sequential extraction method for evaluating mobility in sediments. Envi-ronmental Science & Technology 35 (13), 2778–2784.

Kirk, M.F., Holm, T.R., Park, J., Jin, Q., Sanford, R.A., Fouke, B.W., Bethke, C.M., 2004. Bac-terial sulfate reduction limits natural arsenic contamination in groundwater. Geol-ogy 32, 953–956.

Kirk, M.F., Roden, E.E., Crossey, L.J., Brealey, A.J., Spilde, M.N., 2010. Experimental anal-ysis of arsenic precipitation during microbial sulfate and iron reduction in modelaquifer sediment reactors. Geochimica et Cosmochimica Acta 74 (9), 2538–2555.

Kocar, B.D., Borch, T., Fendorf, S., 2010. Arsenic repartitioning during biogenic sulfidizationand transformation of ferrihydrite. Geochimica et Cosmochimica Acta 74 (3), 980–994.

Kostka, J.E., Luther, G.W., 1994. Partitioning and speciation of solid phase iron insaltmarsh sediments. Geochimica et Cosmochimica Acta 58 (7), 1701–1710.

Lee, M.-K., Saunders James, A., Wilkin Richard, T., Mohammad, S., 2005. Geochemicalmodeling of arsenic speciation and mobilization: implications for bioremediation,advances in arsenic research. ACS Symposium Series. American Chemical Society,pp. 398–413.

Masue-Slowey, Y., Kocar, B.D., Bea Jofré, S.A., Mayer, K.U., Fendorf, S., 2010. Transportimplications resulting from internal redistribution of arsenic and iron withinconstructed soil aggregates. Environmental Science & Technology 45 (2), 582–588.

McCleskey, R.B., Nordstrom, D.K., Maest, A.S., 2004. Preservation of water samples forarsenic(III/V) determinations: an evaluation of the literature and new analyticalresults. Applied Geochemistry 19, 995–1009.

Nordstrom, D., Archer, D., 2003. Arsenic thermodynamic data and environmental geo-chemistry: an evaluation of thermodynamic data for modeling the aqueous envi-ronmental geochemistry of arsenic. In: Welch, A.H., Stollenwerk, K.G. (Eds.),Arsenic in Ground Water. Springer, US, pp. 1–25.

O'Day, P.A., Vlassopoulos, D., Root, R., Rivera, N., 2004a. The influence of sulfur and ironon dissolved arsenic concentrations in the shallow subsurface under changingredox conditions. Proceedings of the National Academy of Sciences of the UnitedStates of America 101 (38), 13703–13708.

O'Day, P.A., Rivera, N., Root, R., Carroll, S.A., 2004b. X-ray absorption spectroscopicstudy of Fe reference compounds for the analysis of natural sediments. AmericanMineralogist 89, 572–585.

Paktunc, D., Dutrizac, J.E., 2003. Characterization of arsenate-for-sulfate substitution insynthetic jarosite using X-ray diffraction and X-ray absorption spectroscopy. TheCanadian Mineralogist 41 (4), 905–919.

Pedersen, H.D., Postma, D., Jakobsen, R., Larsen, O., 2005. Fast transformation of ironoxyhydroxides by the catalytic action of aqueous Fe(II). Geochimica et CosmochimicaActa 69 (16), 3967–3977.

Peiffer, S., Gade, W., 2007. Reactivity of ferric oxides toward H2S at low pH. Environ-mental Science & Technology 41 (9), 3159–3164.

Peiffer, S., Dos Santos Afonso, M., Wehrli, B., Gaechter, R., 1992. Kinetics and mecha-nism of the reaction of hydrogen sulfide with lepidocrocite. Environmental Science& Technology 26 (12), 2408–2413.

Planer-Friedrich, B., London, J., McCleskey, R.B., Nordstrom, D.K., Wallschläger, D., 2007.Thioarsenates in geothermal waters of Yellowstone National Park: determination,preservation, and geochemical importance. Environmental Science & Technology41 (15), 5245–5251.

Planer-Friedrich, B., Suess, E., Scheinost, A.C., Wallschläger, D., 2010. Arsenic speciationin sulfidic waters: reconciling contradictory spectroscopic and chromatographicevidence. Analytical Chemistry 82 (24), 10228–10235.

Poulton, S.W., 2003. Sulfide oxidation and iron dissolution kinetics during the reactionof dissolved sulfide with ferrihydrite. Chemical Geology 202 (1–2), 79–94.

Poulton, S.W., Krom, M.D., Raiswell, R., 2004. A revised scheme for the reactivity of iron(oxyhydr)oxide minerals towards dissolved sulfide. Geochimica et CosmochimicaActa 68 (18), 3703–3715.

Ravel, B., Newville, M., 2005. ATHENA, ARTEMIS, HEPHAESTUS: data analysis for X-ray ab-sorption spectroscopy using IFEFFIT. Journal of Synchrotron Radiation 12, 537–541.

Renock, D., Gallegos, T., Utsunomiya, S., Hayes, K., Ewing, R.C., Becker, U., 2009. Chem-ical and structural characterization of As immobilization by nanoparticles ofmackinawite (FeSm). Chemical Geology 268 (1–2), 116–125.

Robertson, D., Welsh, D.T., Teasdale, P.R., 2009. Investigating biogenic heterogeneity incoastal sediments with two-dimensional measurements of iron(II) and sulfide. En-vironmental Chemistry 6 (1), 60–69.

Rochette, E.A., Bostick, B.C., Li, G., Fendorf, S., 2000. Kinetics of arsenate reduction bydissolved sulfide. Environmental Science & Technology 34 (22), 4714–4720.

Root, R.A., Vlassopoulos, D., Rivera, N.A., Rafferty, M.T., Andrews, C., O'Day, P.A., 2009.Speciation and natural attenuation of arsenic and iron in a tidally influenced shal-low aquifer. Geochimica et Cosmochimica Acta 73 (19), 5528–5553.

Savage, K.S., Bird, D.K., O'Day, P.A., 2005. Arsenic speciation in synthetic jarosite. Chem-ical Geology 215 (1–4), 473–498.

Smieja, J.A., Wilkin, R.T., 2003. Preservation of sulfidic waters containing dissolvedAs(III). Journal of Environmental Monitoring 5 (6), 913–916.

Stauder, S., Raue, B., Sacher, F., 2005. Thioarsenates in sulfidic waters. EnvironmentalScience & Technology 39 (16), 5933–5939.

Suess, E., Planer-Friedrich, B., 2012. Thioarsenate formation upon dissolution of orpi-ment and arsenopyrite. Chemosphere 89, 1390–1398.

Suess, E., Scheinost, A.C., Bostick, B.C., Merkel, B.J., Wallschlaeger, D., Planer-Friedrich,B., 2009. Discrimination of thioarsenites and thioarsenates by X-ray absorptionspectroscopy. Analytical Chemistry 81 (20), 8318–8326.

Suess, E., Wallschläger, D., Planer-Friedrich, B., 2011. Stabilization of thioarsenates iniron-rich waters. Chemosphere 83 (11), 1524–1531.

Sullivan, L.A., Bush, R.T., 1997. Quantitative microanalysis of rough soil surfaces in thescanning electron microscope using a peak-to-background method. Soil Sci. 162,749–757.

Wallschläger, D., Stadey, C.J., 2007. Determination of (Oxy)thioarsenates in sulfidic wa-ters. Analytical Chemistry 79, 3873–3880.

Webster, J.G., 1990. The solubility of As2S3 and speciation of As in dilute and sulphide-bearing fluids at 25 and 90 °C. Geochimica et Cosmochimica Acta 54 (4), 1009–1017.

Wilkin, R.T., 2001. Iron sulfide–arsenite interactions: adsorption behaviour onto ironmonosulfides and controls on arsenic accumulation in pyrite. USGS Workshop onArsenic in the Environment, Denver.

Williams, A.G.B., Scherer, M.M., 2004. Spectroscopic evidence for Fe(II)–Fe(III) electrontransfer at the iron oxide–water interface. Environmental Science & Technology 38(18), 4782–4790.

Wolthers, M., Charlet, L., van Der Weijden, C.H., van der Linde, P.R., Rickard, D., 2005.Arsenic mobility in the ambient sulfidic environment: sorption of arsenic(V) andarsenic(III) onto disordered mackinawite. Geochimica et Cosmochimica Acta 69(14), 3483–3492.

Yang, W., Zhao, N., Zhang, N., Chen, W., Kan, A.T., Tomson, M.B., 2012. Time-dependentadsorption and resistant desorption of arsenic on magnetite nanoparticles: kineticsand modeling. Desalination and Water Treatment 44 (1–3), 100–109.

Yao, W., Millero, F.J., 1996. Oxidation of hydrogen sulfide by hydrous Fe(III) oxides inseawater. Marine Chemistry 52 (1), 1–16.

Zhang, H., Selim, H.M., 2005. Kinetics of arsenate adsorption−desorption in soils. Envi-ronmental Science & Technology 39 (16), 6101–6108.

Zhu, Q., Aller, R.C., Fan, Y., 2006. Two-dimensional pH distributions and dynamics in bio-turbated marine sediments. Geochimica et Cosmochimica Acta 70 (19), 4933–4949.

Zhu, L., Lin, C., Wu, Y., Lu, W., Liu, Y., Ma, Y., Chen, A., 2008. Jarosite-related chemicalprocesses and water ecotoxicity in simplified anaerobic microcosmwetlands. Envi-ronmental Geology 53 (7), 1491–1502.