Metallomics View Article Online

Published on 07 November 2013. Downloaded by Umea University Library on 20/08/2014 22:31:18.

PAPER

Cite this: Metallomics, 2014, 6, 1131

View Journal | View Issue

Stabilities of thiomolybdate complexes of iron; implications for retention of essential trace elements (Fe, Cu, Mo) in sulfidic waters† George R. Helz,*a Britt E. Ericksonab and Trent P. Vorlicekc In aquatic ecosystems, availabilities of Fe, Mo and Cu potentially limit rates of critical biological processes, including nitrogen fixation, nitrate assimilation and N2O decomposition. During long periods in Earth’s history when large parts of the ocean were sulfidic, what prevented these elements’ quantitative loss from marine habitats as insoluble sulfide phases? They must have been retained by formation of soluble complexes. Identities of the key ligands are poorly known but probably include thioanions. Here, the first determinations of stability constants for Fe2+–[MoS4]2 complexes in aqueous solution are reported based on measurements of pyrrhotite (hexagonal FeS) solubility under mildly alkaline conditions. Two linear complexes, [FeO(OH)MoS4]3

and [(Fe2S2)(MoS4)2]4 , best explain the observed solubility variations.

Complexes that would be consistent with cuboid cluster structures were less successful, implying that such clusters probably are minor or absent in aqueous solution under the conditions studied. The new data, together with prior data on stabilities of Cu+–[MoS4]2

complexes, are used to explore computationally

Received 12th August 2013, Accepted 7th November 2013

how competition of Fe2+ and Cu+ for [MoS4]2 , as well as competition of [MoS4]2 and HS for both metals would be resolved in solutions representative of sulfidic natural waters. Thiomolybdate complexes

DOI: 10.1039/c3mt00217a

will be most important at sulfide concentrations near the [MoO4]2 –[MoS4]2 equivalence point. At lower sulfide concentrations, thiomolybdates are insufficiently stable to be competitive ligands in natural waters

www.rsc.org/metallomics

and at higher sulfide concentrations HS ligands out-compete thiomolybdates.

Introduction Many trace elements exist in O2-bearing natural waters predominantly in their highest oxidation states, as environmentally mobile (hydr)oxyanions. Examples include [VO2(OH)2] , [CrO4]2 , [GeO(OH)3] , [AsO3(OH)]2 , [SeO4]2 , [MoO4]2 , [TcO4] , [Sb(OH)6] , [WO4]2 , and [ReO4] . As essential nutrients or as hazardous substances, many of these trace elements can play significant roles in aquatic ecosystems. In poorly ventilated, sulfide-bearing waters, the (hydr)oxyanions become unstable. Some are reduced rapidly by sulfide (e.g., [VVO2(OH)2] - [VIII(OH)3]0 or [CrVIO4]2 - [CrIII(OH)3]0),1–3 but others are transformed to thioanions without reduction (e.g., [MoVIO4]2 - [MoVIS4]2 or [ReVIIO4] - [ReVIIS4] ).4–6 a

Department of Chemistry and Biochemistry, University of Maryland, College Park, MD, USA. E-mail: [email protected]; Tel: +1 301-405-1797 b American Chemical Society, 1156 16th St. NW, Washington, DC 20036, USA. E-mail: [email protected]; Tel: +1 202-872-4545 c Department of Chemistry and Geology, Minnesota State University, Makato, MN 56001, USA. E-mail: [email protected]; Tel: +1 507-389-1598 † Electronic supplementary information (ESI) available: Includes details of experimental procedures, solubility data, comparison of Fig. 2 and literature spectra and analytical data on selected euxinic basins. See DOI: 10.1039/c3mt00217a

This journal is © The Royal Society of Chemistry 2014

In both instances, the biogeochemical properties of the trace element in aquatic ecosystems will be altered profoundly. Much remains to be learned about the environmental behavior and effects of the thioanions. Compared to concentrations of the oxyanions in surface waters, thioanions of the same element in deep sulfidic waters can be much less abundant (e.g., Mo) or much more abundant (e.g., As).7–9 (In exceptional cases in sulfidic waters, thioanions of As exceed 0.1 mM,9 which is 10 000 times the public health standard for As in drinking water.) Additionally, whereas oxyanions as a rule form weak complexes with trace metal cations, thioanions can form complexes of extraordinary stability.10,11 High stability arises in part from the chelate effect; thioanions can act as bidentate, or even as tridentate, S-donating ligands.10 Thus thioanion formation not only modifies the environmental behavior of the participating element, but has the potential, through complexing, to affect availability of large numbers of other biologically active trace elements in sulfidic natural waters. Here our first focus is on the complexes formed between [MoS4]2 and Fe2+ in aqueous solution. Chemists aspiring to replicate the catalytic activity of the MoFe7S9C cluster in nitrogenase have produced a rich literature on compositions and structures of Fe–Mo–S complexes.12–14 Clusters containing

Metallomics, 2014, 6, 1131--1140 | 1131

View Article Online

Published on 07 November 2013. Downloaded by Umea University Library on 20/08/2014 22:31:18.

Paper

two dozen or more atoms often are observed to self-assemble spontaneously, implying that they possess thermodynamic stability under the conditions of synthesis. For example, [Mo2Fe6S8(SR)9]3 , one of the most widely studied double cuboid clusters, is made simply by mixing FeCl3, [MoVIS4]2 and NaSR (a thiol salt) in a suitable solvent.14 Unfortunately almost all of these Fe–Mo–S complexes have been produced and studied under conditions not relevant to nature: e.g., high reactant concentrations, non-aqueous solvents and sometimes redox potentials that are unattainable in Nature.15 While these studies provide useful guidance regarding oxidation states, bonding and structure, they shed no light on whether Fe–Mo–S complexes could form spontaneously in Nature. Quantitative information about stabilities of thioanion complexes in aqueous solution remains sparse, so it is currently not possible to assess the influence of thioanion complexing on the trace element economies of natural waters. Extraordinarily large stability constants for Cu(I) complexes with thiomolybdates and thioarsenites have been measured.10,11 Copper thiomolybdate complexes famously explain molybdenosis, a Mo-induced copper deficiency disease among ruminants.11 Additionally, Cu thiomolybdate complexes probably control Cu availability in the anoxic zones of certain lakes.11 Theory predicts that large stability constants will be found for thioarsenite and thioantimonite complexes of other coinage metals, Ag+ and Au+.16 That these complexes can be important in nature is strongly hinted by the plethora of known sulfosalt minerals containing As and Sb thioanions as structural units.17 Owing to its much greater abundance in nature, Fe potentially might displace coinage metals like Cu from thioanion ligands. X-ray spectroscopic studies of sediments from Lake Cadagno,18 as well as from ancient black shales,6 demonstrate intimate association of Mo with S and Fe, implying that Fe–Mo–S compounds indeed are important in nature. The Mo–S and Mo–Fe interatomic distances measured by X-ray spectroscopy in these samples are similar to those in reduced cuboid Fe–Mo–S clusters (example in Fig. 1). Such clusters appear also to form spontaneously on pyrite (FeS2) surfaces exposed to [MoS4]2 at mildly alkaline pH.19 Here we employ solubility measurements to develop a thermodynamic model and use it to explore how Fe thiomolybdate complexes in sulfidic natural waters could affect ecosystem availability of Fe, Cu and Mo. These three trace elements play key roles in many fundamental biological processes in nature, including for example in cycling of nitrogen. Iron controls biological productivity in large parts of the ocean in part by limiting the rate of N2 fixation.20 Molybdenum plays a similar role in certain lakes.21 Copper-, molybdenum- and iron-containing enzymes are critical for some of the steps in NO3 assimilation and denitrification.22

Experimental methods A detailed description of the materials and procedures is given in the ESI.† Synthetic pyrrhotite was prepared by heating high purity Fe and S in evacuated silica tubes. Pyrrhotite is commonly

1132 | Metallomics, 2014, 6, 1131--1140

Metallomics

Fig. 1 Structures of two Fe–Mo–S complexes discussed in the text. (A) Linear complex consisting of a 2Fe–2S cluster coordinated by two [MoS4]2 ligands. (B) Cuboid cluster; L1 and L2 are external ligands; L1 ligands complete 4-fold coordination around Fe and L2 ligands complete 6-fold coordination around Mo; L1 and L2 can be the same, but in most known cases are different. Oxidation states of Mo and Fe in Fe–Mo–S complexes are often non-integers owing to electron delocalization. The linear complex contains nominally Fe2+ and Mo6+, but Mossbauer spectroscopy and Fe–S bond lengths suggest that actual oxidation states approximate Fe2.4+ and Mo5.6+. The cuboid contains a [Mo2Fe2S4]4+ core that nominally contains Fe2+ and Mo4+.

non-stoichiometric (Fe1 xS), but X-ray diffraction line positions demonstrated that x o 0.02 in our product; thus within uncertainty, we produced stoichiometric FeS (troilite), which is the most stable (least soluble) isomorph. Mo-containing stock solutions were prepared from (NH4)6Mo7O24
4H2O reagent. Aqueous sulfide stock solutions were prepared by saturating NaOH solutions with H2S gas; this method minimizes polysulfide contaminants that are sometimes present in Na2S
9.5H2O reagent. Individual test samples were prepared in a N2-filled glove box using ampoule methods described previously.23 Fusionsealed samples were allowed to react at room temperature (25  2 1C) with periodic shaking by hand to aid mixing. Reaction times in the ampoules ranged from 22 to 264 days. After the specified reaction time, each ampoule was opened in a N2 filled glove box. Solutions were filtered (0.02 mm pore diameter, Whatman Anotop 25), and the pH and total sulfide concentrations were immediately determined. In some cases, a UV-visible absorption spectrum was obtained. P Total sulfide ( S II) was determined by potentiometric titration with HgCl2 at pH 13. Total Fe and Mo were determined by flame atomic absorption spectroscopy. For the Mo analyses, sulfide was first oxidized to sulfate with BrCl; then Al was added to suppress Fe interference and analyses were done with a N2O– P acetylene flame. Analytical uncertainties in S II, Mo and Fe are on the order of 5%. Exploratory experiments were conducted to test reversibility of the dissolution reactions. In a glove box, a deaerated solution

This journal is © The Royal Society of Chemistry 2014

View Article Online

Published on 07 November 2013. Downloaded by Umea University Library on 20/08/2014 22:31:18.

Metallomics

of 70 mM Fe2+ (from Fe(NH4)2(SO4)2
6H2O) was buffered with 20 mM borate, adjusting pH to 8.5 with small amounts of HCl or NaOH. When 20 mM NaHS was added, nanoparticulate mackinawite (metastable, tetragonal FeS) immediately precipitated. To this black slurry was added 70 mM [MoS4]2 . Aliquots of this solution were taken periodically and analyzed by UVvisible spectroscopy and atomic absorption spectroscopy after filtering through 0.45 mm filters. In modeling the solubility data, ion activity coefficients were calculated from the Davies equation.24 The activity of OH was obtained from pH. Activities of NH30 and HS were obtained P P from NH3 and S II after making appropriate corrections for ionization equilibria. In our experiments, sulfide was in large excess to dissolved Mo and Fe, so it was not necessary to consider these latter components when calculating aHS from P II measured S . In cases where complex formation involved reduction (for example, the cuboid complex in Fig. 1), we assumed that the oxidized complement was S0, which in our highly sulfidic solutions would form polysulfide ions. We applied a constraint to the calculations that the electron equivalents of S0 in polysulfides must equal the electron equivalents of reduced Fe complex. This constraint permitted calculation of aS0 from polysulfide equilibria according to procedures described previously;25,26 we employed the most recent polysulfide stability constants.27 Reportedly, Fe species catalyze hydrolysis 2 sulfidation reactions among thiomolybdates (e.g., [MoS4]2 + OH 2 [MoOS3]2 + HS ),28 so we assumed that on the multipleweek time scale of our experiments, the various thiomolybdates were in equilibrium with one another. In support of this assumption, we found that the same stability constants and speciation model described solubilities in solutions prepared from a stock solution containing predominantly [MoS4]2 initially and a stock solution containing predominantly [MoO3S]2 and [MoO2S2]2 initially (see ESI†). At equilibrium, [MoO3S]2 , [MoO2S2]2 and [MoOS3]2 are always minor species relative to [MoO4]2 and [MoS4]2 .4 Therefore these intermediate thiomolybdates are unlikely to be significant as ligands in nature. Consequently we focused our experiments on characterizing MoS42 complexes. In the Discussion section, we present some equilibrium calculations pertaining to Fe, Mo and Cu speciation in natural waters. We explicitly account for effects of competing Cl and OH ligands, making use of published stability constants.29 Complexing of Fe2+ and Cu+ by HS is evaluated only using stability constants determined by approaching equilibrium from undersaturation.30,31 Equilibration from supersaturation (e.g. by spiking metal cations into sulfide solutions or sulfide into metal cation solutions) yields markedly different speciation and stability constants. These may be artifacts arising from misdescribing colloidal precipitation reactions as complexing reactions.32,33 We neglected possible complexing by competing organic ligands. In oxic waters, ubiquitous O- and N-donating macromolecular organic ligands often dominate FeIII and CuII speciation, but in sulfidic waters they compete ineffectively for FeII and CuI against S-donating ligands, including thioanions. On an equimolar basis, organic thiols potentially can compete

This journal is © The Royal Society of Chemistry 2014

Paper

with HS and [MoS4]2 ligands, but thiols ordinarily are present at much lower concentrations in natural waters.11

Results Solubility measurements Several practical considerations limited the range of solution compositions that were investigated. The pH was restricted to the alkaline range to minimize H2S(g) loss during sample preparation and to ensure that [MoS4]2 remained soluble. At the same time, pH could not be so high in relation to aHS that major [MoS4]2 decomposition would occur by hydrolysis ([MoS4]2 + OH - [MoOS3]2 + HS . . . etc.). The lower limit of [MoS4]2 concentration was established by the need to have enough of this ligand to produce >1 mM Fe solubilities. Solubilities below this level are often erratic owing to the difficulty of completely removing small particles from samples by filtration. At the same time, [MoS4]2 could not be so high that uncertainty P associated with its interference in the S II assay compromised P calculation of the activity of free HS . Therefore, S II had to P be >10-fold in excess of Mo, but not so high that Fe(HS)20, which increases as aHS 2, overwhelmed Fe–[MoS4]2 complexes. Despite being hemmed in by these practical constraints, we were able to explore solubility over more than an order of magnitude range in each of the independent variables, aH+, aHS and aMoS42 . Table S1 in the ESI† presents data for 49 pyrrhotite solubility experiments. Powder X-ray diffractometry revealed no crystallographic changes in the pyrrhotite after equilibration with these solutions and no evidence of new phases, such as mackinawite or pyrite. This test would reveal major (>5%) new crystalline phases, but would not detect amorphous or minor new phases. Final dissolved Mo concentrations generally agreed within uncertainty with initial concentrations, demonstrating that the pyrrhotite charges took up immeasurably small amounts of Mo, if any. In all cases, dissolved Fe concentrations far exceed what could be explained by known Fe2+ species involving NH30, Cl , OH or HS ligands in these solutions. Based on solubility product measurements for pyrrhotite,34,35 and on published stability constants for Fe(II) complexes,29,30 the most abundant known complex would be Fe(HS)20. By calculation, its concentration would in all cases explain less than 0.01% of observed Fe solubility. Thus previously unidentified complexes formed from thiomolybdates are required to explain the observed solubilities. The chemical literature on Fe–Mo–S complexes offers limited but helpful constraints on the nature of these unidentified complexes. Where structures of complexes have been determined by X-ray crystallographic methods, FeII and MoVI are found to be 4-coordinated to S. Additionally, UV-visible spectra have been reported for a large number of complexes, albeit usually dissolved in organic solvents. Fig. 2 shows spectra from two of the solutions that had been equilibrated with pyrrhotite. The two panels depict spectra from solutions of similar composition except that the one on the left contains six times more [MoS4]2 . Because [MoS4]2 is in substantial excess to dissolved Fe in both cases, unprocessed spectra, shown in insets, are dominated by [MoS4]2 peaks at

Metallomics, 2014, 6, 1131--1140 | 1133

View Article Online

Published on 07 November 2013. Downloaded by Umea University Library on 20/08/2014 22:31:18.

Paper

Fig. 2 UV-visible absorption spectra from samples 13.3 (A) and 3.13 (B) (solution compositions in Table S1, ESI†). The insets show observed, [MoS4]2 -dominated spectra in each sample and in a corresponding blank sample that contains the same solution but no pyrrhotite. In the main panels are difference spectra in which the [MoS4]2 peaks have been eliminated by subtracting weighted blank spectra from FeS sample spectra. Numbers associated with peaks are wavelengths at peak maxima.

318 and 468 nm. The main panels were obtained by subtracting free thiomolybdate ([MoS4]2 , [MoOS3]2 etc.) absorption, thus uncovering spectra of Fe–Mo–S complexes. The concentration P difference between AA-measured total Mo ( Mo) and opticallymeasured free thiomolybdates is designated Mo* in the keys in Fig. 2; Mo* is taken to be a measure of the Mo contained in dissolved Fe–Mo–S complexes. The spectrum in Fig. 2B contains some of the same peaks as Fig. 2A, but includes an additional distinct peak at 390 nm. Although [MoO2S2]2 and [MoOS3]2 both absorb near 390 nm, their concentrations would be too small to explain the height of the 390 nm peak in Fig. 2B. Two important points emerge from Fig. 2. First, Mo* approxiP mately equals Fe in both cases, revealing that over a range of dissolved Mo concentrations average Fe/Mo E 1 in the dissolved Fe–Mo–S complexes. Second, the spectra in the two panels differ appreciably, implying that among the various solutions we studied more than one Fe–Mo–S complex must be present. In Table S2 in the ESI,† we compare wavelengths of the peaks in Fig. 2A with reported spectra for two known Fe–Mo–S complexes, [(Fe2S2)(MoO2S2)2]4 and [Fe(MoS4)2]3 . Our peaks agree reasonably well with published spectra for both complexes, yet only the first of these satisfies the Fe/Mo E 1 constraint. This comparison demonstrates a limitation of UV-visible spectroscopy as a guidepost for interpreting solubility data. Compositions and electronic structures of many Fe–Mo–S complexes are quite similar. UV-vis spectra robustly confirm the presence of Fe–Mo–S complexes, but they provide only tenuous information about the complexes’ actual identities. As described later, we will rely on the solubility data themselves to infer the identities of the complexes. Precipitation It is desirable in solubility studies to test whether the same final state can be reached from both undersaturation and supersaturation.

1134 | Metallomics, 2014, 6, 1131--1140

Metallomics

If this test succeeds, reversible equilibrium is demonstrated unequivocally. Unfortunately, this test often fails because the undersaturation pathway reaches equilibrium with the least soluble solid phase, whereas the supersaturation pathway, in accordance with Ostwald’s step rule,24 first reaches equilibrium with the most soluble (metastable) solid phase – often a colloid. In experiments with low solubility phases, such as transition metal sulfides, recrystallization of the metastable to the stable phase has been found in some cases to occur at negligible rates on laboratory time scales at 25 1C.33 Both in nature and in the laboratory near room temperature, mackinawite rather than stable pyrrhotite is the FeS phase that precipitates from supersaturation. Therefore, from Ostwald’s step rule we expected higher Fe solubilities in precipitation experiments because mackinawite is more soluble than pyrrhotite, but we decided to check. The consequences of adding 0.070 mM [MoS4]2 to a mixture of 0.070 mM Fe2+ and 10 mM HS at pH 8.5 are shown in Fig. 3. The absorption spectrum of a filtered sample taken at 0.007 days (10 min) has a high baseline owing to the colloidal character of the early-formed mackinawite. As a fresh precipitate, mackinawite has particle sizes o10 nm.36,37 Atomic absorption analysis showed that virtually all initial dissolved Fe and Mo was recovered at 0.007 days, demonstrating that the early precipitate passed through 0.45 mm filters. Superimposed on the high optical baseline at 0.007 days in Fig. 3 are two prominent peaks arising from dissolved [MoS4]2 . The peak heights indicate that only about half (0.033  0.001 mM) of the initial [MoS4]2 remains in solution, the rest having already been captured by the precipitate. By 0.08 days (1.8 h) all dissolved [MoS4]2 has been captured from solution according to the optical data. The colloidal precipitate also has coarsened so that it is now removed by filtration, eliminating the high optical baseline. Absence of dissolved Mo and Fe was confirmed by atomic absorption spectroscopy, proving that [MoS4]2 has indeed

Fig. 3 Time series of UV-vis absorption spectra following precipitation. The two large peaks in the 0.007 days (i.e. 10 min) spectrum belong to [MoS4]2 ; an unresolved peak at 396 nm belongs to [MoOS3]2 , which is an expected minor (o5%) component at equilibrium in this solution. An unresolved shoulder near 340 nm may belong to a small amount of Fe–Mo–S complex. The inset expands the vertical scale to reveal spectral characteristics of the solution phase during subsequent aging of the precipitate. All spectra were taken after filtration of the solutions (0.45 mm).

This journal is © The Royal Society of Chemistry 2014

View Article Online

Published on 07 November 2013. Downloaded by Umea University Library on 20/08/2014 22:31:18.

Metallomics

Paper

been removed from solution, not simply transformed to a species that does not absorb in the 275–500 nm wavelength range. Seemingly, the only spectral feature remaining at 0.08 days is a slight increase in absorbance at l o 300 nm; this is the edge of an immense 230 nm HS peak that swamps optical absorption at shorter wavelengths. Nonetheless, a scale expansion, shown in the inset, reveals a hint of absorption due to polysulfides (HxSnx n), which are the usual first products of HS oxidation. As the inset shows, by 1.0 days this has become a distinct feature and with further aging becomes even larger. By optical absorption methods,38,39 P total zero-valent sulfur ( S0) in polysulfides after 213 days of aging is found to be 0.069  0.001 mM. Although O2 contamination in the glovebox might explain oxidation of sulfide to polysulfide, we think that MoVI in the precipitate is more likely the oxidant. The amount S(0) in polysulfide after 213 days is P essentially equal to initial MoVI, suggesting that a slow, 2e reduction of the MoVI to MoIV has occurred. After 213 days, a small [MoS4]2 peak, indicative of 0.84 mM Mo, also has appeared. Thus aging appears to involve principally reduction of the precipitate but also involves a small release of initially precipitated MoVI. This reaction may explain the reduced Mo observed by X-ray spectroscopy in Fe–Mo–S phases in sediments.18 The activity of zero valent sulfur (aS0) after 213 days is 0.22 on a scale where aS0 = 1 at saturation with rhombic sulfur.38,39 Since rhombic sulfur is the most stable allotrope of this element, this solution is undersatured with all condensed forms of elemental sulfur. The pH and HS concentration in this precipitation experiment are similar to those in the pyrrhotite solubility experiments depicted in Fig. 2, but the spectra are entirely different. Whereas pyrrhotite took up negligible thiomolybdate, mackinawite has taken it up quantitatively. Contrary to our expectations from Ostwald’s step rule, the supersaturation pathway has produced a lower concentration of dissolved Fe than the undersaturation pathway. The reason is that the supersaturation pathway removed the Mo ligand from solution, preventing formation of Fe–Mo–S complexes at measurable levels. The amount of Mo lost from solution is too large to be explained by adsorption on mackinawite. This surprising result requires precipitation of an Fe–Mo–S solid of some kind. To our knowledge, such a solid has not been characterized in the laboratory. However, a proposal emerged from analyses of Rogoznica Lake waters that the following reaction controlled dissolved Mo in sulfidic natural waters.8 0.6H2S(aq) + FeMo0.6S2.8(s) 2 FeS(mack) + 0.6[MoS4]2 + 1.2H+, log K =

11.7

(1)

In this reaction, FeMo0.6S2.8 is an empirical formula deduced from Rogoznica Lake data. At a given pH and aH2S, the equilibrium [MoS4]2 concentration in reaction 1 would be higher if mackinawite were replaced by a more stable (more negative free energy of formation) FeS phase like pyrrhotite. Using the above log K for reaction 1, [MoS4]2 solubility in equilibrium with mackinawite would be o1 mM under the pH

This journal is © The Royal Society of Chemistry 2014

and aH2S conditions in Fig. 3, but would be about 500 mM under the same conditions in equilibrium with pyrrhotite. Much work remains to be done to fully characterize this precipitate. Nonetheless, this solubility estimate very plausibly explains why experiments with pyrrhotite produced measurable Fe solubilities, whereas Fe was immeasurable in contact with mackinawite, seemingly contrary to Ostwald’s step rule. In principle, Fe–Mo–S complexes also form in contact with mackinawite, which is the dominant FeS phase in low temperature natural environments, but their concentrations must lie below detection limits of the methods used in this study. We address this issue later, after first developing a thermodynamic model that permits calculation of solubilities under conditions existing in nature. With regard to the solubility measurements, the key result here is that reversibility cannot be demonstrated by precipitation. We will rely instead on the time-independence of solubility as evidence that equilibrium has been achieved. This will be shown below by comparing experiments covering time spans of 22 to 264 days.

Discussion Model derivation By investigating how Fe solubility varies with solution composition, it is possible to infer empirical formulas for the dominant Fe–Mo–S complexes in these solutions. This well-established approach involves predicting Fe solubilities by positing equilibrium solubility reactions. Optimum equilibrium constant values for the posited reactions are obtained by the non-linear least squares method. As in all fitting procedures, the goal is to find the smallest set of parameters (posited constants) that minimize deviations between predicted and observed solubilities while also eliminating trends in deviations with respect to the independent variables. A recognized shortcoming of the solubility method is that it yields information only about the species that are most abundant over the composition range investigated. Thus models derived from solubility data possibly oversimplify the actual chemistry. Table 1 presents fitting parameters for selected complexes that for the most part have been previously observed in nonaqueous solvents. Each of these complexes was first fit one-at-atime to the data. Standard deviations between observed and calculated solubilities are given in the right hand column. Species 1 to 11 as well as 15 and 16 are consistent with Fe/Mo = 1. Species 14 to 16 are based on cuboid cluster formulas. Although not shown, we tested a number of other plausible cuboid formulas containing [Fe3MoS4]3+ and [Fe2Mo2S4]4+ cores but differing in external Cl , OH and HS ligands. The external ligands depicted in species 14 and 15 provided the best fits for each core-type, but neither core-type achieved an optimum fit. The standard deviations between model-calculated vs. observed concentrations in Table 1 suggest that species 2, 3, 6, 8 and 9 are the most promising for explaining Fe solubility variations with respect to pH, aHS and aMoS4.

Metallomics, 2014, 6, 1131--1140 | 1135

View Article Online

Paper

Metallomics

Table 1 Fits of hypothesized Fe–Mo–S complexes one at a time to the solubility data. Constants in columns 3 to 7 are exponents in generalized mass action laws describing pyrrhotite solubility: K = (aFespecies)(aH+)a(aHS )b(aMoS42 )c(aCl )d(aS0)e. Negative exponents denote reactants and positive exponents denote products in reactions that form the indicated Fe species from FeS

Published on 07 November 2013. Downloaded by Umea University Library on 20/08/2014 22:31:18.

No.

a (H+)

Fe species

1 [FeCl2MoS4]2 1 [Fe(OH)2MoS4]2 3 2 [FeO(OH)MoS4] 2 1 [Fe(HS)2MoS4] 0 [Fe(OH)(HS)MoS4]2 0 [FeS(HS)MoS4]3 0 1 [(H2O)2FeMoS4] 4 1 [FeS2MoS4] 0 [(Fe2S2)(MoS4)2]4 4 [(Fe2S2)(MoO2S2)2]4 4 2 [(Fe2S2)(MoOS3)2] 4 1 [Fe(MoS4)2] 0.5 [Fe(MoS4)2]3 0.5 [(Fe3MoS4)(HS)3(OH)2Cl]3 4 2 [(Fe2Mo2S4)(HS)2(OH)6] 4 0 [(Fe2Mo2S4)(HS)4(OH)4] 1 P 2 Standard deviation = ( (pred obs) /(N 1))2,

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 a

b (HS ) 1 1 1 1 0 1 1 1 0 4 2 1 0.5 0.5 2 0 where N

c (MoS42 ) 1 1 1 1 1 1 1 1 2 2 2 2 2 1 2 2

d (Cl ) 2 0 0 0 0 0 0 0 0 0 0 0 0 1 0 0

e (S0) 0 0 0 0 0 0 0 0 0 0 0 0 0.5 0.5 2 2

Standard deviationa (mM)

log K 5.62 12.21 22.12 9.3 1.07 0.38 5.41 9.94 0.77 44.60 21.22 8.7 4.6 2.74 23.67 2.52

 0.14  0.05  0.04       

0.06 0.04 0.17 0.05 0.05 0.19 0.06

 0.07  0.08  0.26

0.029 0.020 0.018 0.029 0.021 0.019 0.030 0.018 0.020 0.030 0.023 0.030 0.027 0.024 0.026 0.031

1 = 48.

In a second stage of data analysis, we explored fitting pairs of the species in Table 1 in order further to minimize trends in deviations with respect to the independent variables. Fig. 4 presents the deviations for the best model, which improved the standard deviation to 0.015 mM. This model requires only two reactions: 2H2O + [MoS4]2 + FeS(pyrr) 2 [FeO(OH)MoS4]3 + HS + 2H+, log K =

22.12  0.04

(2)

2[MoS4]2 + FeS(pyrr) 2 [(Fe2S2)(MoS4)2]4 , log K = +0.77  0.05

(3)

The improvement in standard deviation with this twoparameter model vs. the best one-parameter model is significant at the 90% level of confidence according to an F test. In both complexes, Fe/Mo = 1, but their concentration dependence on aMoS4 differs. It is likely that [(Fe2S2)(MoS4)2]4 , which in equilibrium with pyrrhotite increases as the square of aMoS4, is the species mainly responsible for the spectrum of the more Mo-rich solution depicted in Fig. 2A. This species is closely related to the known complex, [(Fe2S2)(MoO2S2)2]4 , whose spectrum is compared in Table S2 (ESI†) to the one in Fig. 2A. Note in the lower left panel of Fig. 4 that deviations between predicted and observed solubilities show no consistent trend with reaction time over 22 to 264 days. The structure of [(Fe2S2)(MoS4)2]4 is known to be linear and is shown in Fig. 1A.40 To our knowledge, [FeO(OH)MoS4]3 is not a known complex, but [Fe(Cl)2MoS4]2 and [Fe(OAc)2MoS4]2 (OAc = acetate) are known in organic solvents.41,42 It seems plausible that these latter species would hydrolyze to [FeO(OH)MoS4]3 in mildly alkaline aqueous solution. In general, posited species containing Cl ligands produced poor predictions of observed solubilities in aqueous solution. Quantitative modeling suggested that Cl variations affected solubility primarily through their influence on the activity coefficients of Fe complexes. This finding is consistent with evidence that Cl forms

1136 | Metallomics, 2014, 6, 1131--1140

Fig. 4 Deviations between observed and model-predicted Fe concentrations resulting from fitting equilibrium constants for [FeO(OH)MoS4]3 and P [(FeS)2(MoS4)2]4 to the solubility data. Mofree = [MoOnS4 n]2 , n = 0–4 (i.e. all Mo not bound to Fe).

only weak complexes with Fe2+. For example, log b2 is 0.4 for [Fe(Cl)2]0 compared to 6.46 for [Fe(HS)2]0 and 7.39 for [Fe(OH)2]0. Solubility evidence places no constraints on the relative stabilities of isomers. Although we have written the formula, [FeO(OH)MoS4]3 , so as to imply that the O atoms terminally coordinate Fe (as they do in [Fe(OAc)2MoS4]2 ),41 the data are equally consistent with O in bridging positions between Fe and Mo or even terminal to Mo. The last two options suggest that this species alternatively could be described as a dithiomolybdate complex. This description might rationalize why the 390 nm peak in Fig. 2B, which we would now attribute to this species, falls where a major MoO2S22 peak falls. Employing literature data for reaction 4,34,35 we can describe the formation of [FeO(OH)MoS4]3 and [(Fe2S2)(MoS4)2]4 more conventionally in terms of formations from their component ions: FeS(pyrr) + H+ 2 Fe2+ + HS , Ksp = 10

5.20.1

(4)

This journal is © The Royal Society of Chemistry 2014

View Article Online

Metallomics

Paper

2H2O + [MoS4]2 + Fe2+ 2 [FeO(OH)MoS4]3 + 3H+, K = 10 2

2HS + 2[MoS4]

2+

+ 2Fe

16.9

(5)

2 [(Fe2S2)(MoS4)2]

K = 10

4

+

+ 2H ,

+11.2

(6)

Published on 07 November 2013. Downloaded by Umea University Library on 20/08/2014 22:31:18.

Cuboid clusters Discovery that the most efficient nitrogenase enzyme employs a MoFe7S9C cluster to catalyze reduction of N2 to NH3 sparked immense interest in Fe–Mo–S clusters. EXAFS evidence,6,18 that Fe–Mo–S cuboidal clusters might exist in sediments and shales, raises interesting questions about whether they could have formed in Hadean or Archean sediments and possibly served as prebiotic catalysts. Cuboid clusters (example, Fig. 1B) have a number of features that make them interesting as catalysts. They can undergo labile ligand exchange without loss of core structure, they are able to participate in both 1e and 2e redox reactions over a range of redox potentials, they possess several kinds of binding sites with differing affinities for external ligands and they have multiple binding sites, enabling them to interact with complex substrates at more than one site simultaneously. Demonstrably, synthetic [MoFe3S4]3+ cuboid clusters catalyze hydrogenation of acetylene to ethane, a reaction also catalyzed by nitrogenase.43,44 On the other hand, to date no synthetic cluster has performed nitrognease’s main job of catalyzing hydrogenation of N2 to NH3 under mild conditions. Geochemists have lately become curious about the possibility that Fe–Mo–S cuboid clusters had important catalytic roles leading up to the origin of life.45 Stoichiometrically, the examples of a cuboid cluster and a linear complex depicted in Fig. 1 are not very different. A hypothetical reaction relating one to the other would be:

equilibrium with pyrrhotite under the conditions of our experiments, but their presence as minor species cannot be dismissed. On the other hand, polysulfide absorbance did begin to appear in our mackinawite precipitation experiment, but only after precipitation and scavenging of thiomolybdates were complete. Tentatively, the implication is that reduction of MoVI and formation of cuboid-like chemical structures in sulfidic water bodies, like Lake Cadagno,18 occur not in the water column but in the sediments during early diagenesis. Whether these materials have roles as catalysts of organic reactions in sediments is unknown. Modeling natural waters The new thermodynamic data for Fe–Mo–S complexing, as well as previous data for Cu–Mo–S complexing,11 enable a generalized exploration of how availability of Mo, Fe and Cu in sulfidic waters functionally depends on pH and total dissolved sulfide. Fig. 5 presents a model for conditions that are common in deep waters of modern marine euxinic water columns. These waters characteristically have pH between 7 and 8 and Cu concentrations in the 0.1 to 1.0 nM range. Their Mo concentrations range from 0.1 mM, which is characteristic of oxic seawater, down to values near 0.001 mM in permanently sulfidic waters. P II The calculations indicate that with increasing S , the P Fe concentration steadily declines owing to the decreasing solubility of mackinawite. Experience indicates that euxinic waters are usually near saturation with respect to this metastable P phase.46–50 The downward trend in Fe is arrested above 1 mM P II S where [Fe(HS)2]0 replaces Fe2+ as the predominant dissolved

4H2O + [(Fe2S2)(MoS4)2]4 - [(Fe2Mo2S4)(HS)4(OH)4]4 + 2Spolysulfide0

(7)

In this case, where we have arbitrarily chosen equal amounts of HS and OH as ligands external to the 4+ cuboid core, the equilibrium ratio of the two complexes is independent of aHS and pH. For other ratios of these external ligands, some aHS and pH dependence would exist. Regardless of the external ligands, the key feature of the above reaction is that MoVI is reduced to MoIV in the cuboid with concomitant oxidation of external S( II) to S(0). In order that the linear complex can draw together into the compact cuboid cluster, internal coulombic repulsion must diminish. This is accomplished both by chemical reduction and by charge delocalization over the core. In fitting models to our solubility data, we explored both linear and cuboid complexes. Comparison of lines 9 and 16 in Table 1 shows that the linear complex in the above reaction fitted the data considerably better than the cuboid complex. The smaller standard deviation obtained with the linear complex is a significant improvement on that obtained with the cuboid complex at the 99.9% level of confidence (F test). This evidence, combined with absence of any sign of polysulfide absorbance in UV-visible spectra disfavors cuboid clusters as major species in

This journal is © The Royal Society of Chemistry 2014

Fig. 5 Calculated effect of thiomolybdate complexing on the speciation of Fe, Mo and Cu under conditions common in modern euxinic marine P waters. Consistent with widespread field observations, Fe is assumed to be controlled by mackinawite (FeSmack) saturation. Total dissolved Mo is P P taken to be the smaller of either Mo in seawater (10 7 M) or Mo at saturation with respect to FeMo0.6S2.8 and FeSmack. Total Cu is fixed at 0.5 nM. In the middle graph, minor concentrations of intermediate thiomolybdates ([MoO3S]2 , [MoO2S2]2 and [MoOS3]2 ) are omitted for clarity. In this model, ligand competition from OH , HS and Cl were accounted P P for.27,28 Definitions: S II = H2S + HS ; Mo = all thiomolybdates + Mo P in Fe–thiomolybdate complexes; Fe = Fe2+ + [Fe(HS)2]0 + Fe in Fe–thiomolybdate complexes.

Metallomics, 2014, 6, 1131--1140 | 1137

View Article Online

Published on 07 November 2013. Downloaded by Umea University Library on 20/08/2014 22:31:18.

Paper

Fe species. Thiomolybdate complexing plays an insignificant role in Fe speciation at this pH. P On the other hand, Mo at its seawater value of 0.1 mM is independent of sulfide concentration until saturation with P respect to FeMo0.6S2.8 is reached near 0.01 mM S II. Beyond P that point, higher sulfide at first causes Mo to decline, but this trend subsequently reverses near the [MoO4]2 –[MoS4]2 equivalence point, and still higher sulfide causes FeMo0.6S2.8 to redissolve in accordance with reaction 1. As shown in Table S3 (ESI†), these patterns predicted by the model are expressed qualitatively in modern euxinic marine basins. Thiomolybdate complexing accounts for a maximum of only about 10% of P Mo (middle panel). In contrast, the thiomolybdate complex of Cu contributes up P to about 30% to Cu. By stabilizing a dissolved form of Cu, thiomolybdate complexing is acting modestly to curtail loss of Cu from sulfidic waters to sediments. During extensive periods in the Proterozoic and Mesozoic when large parts of the ocean are believed to have been sulfidic,51–53 thiomolybdate complexing of Cu may have helped to prevent Cu impoverishment of the ocean and thus helped to forestall a ‘‘laughing gas greenhouse.’’54 Thiomolybdate complexing is restricted at pH 7.5 (Fig. 5) because the [MoS4]2 concentration in equilibrium with P FeMo0.6S2.8 is very low in the mid-range of S II. At its miniP mum, the calculated Mo concentration is only a few nanomolar, as indeed is observed in the modern Black Sea. Minimum P Mo becomes even lower under more acidic conditions because H+ suppresses [MoS4]2 solubility according to reaction 1. Thus [MoS4]2 will be insignificant as a ligand for Fe and Cu in acidic natural waters. On the other hand, under more alkaline conditions higher [MoS4]2 concentrations are possible. For example, P at pH 8.5 the minimum Mo in equilibrium with FeMo0.6S2.8 is 36 nM. Under such conditions, [FeOOH(MoS4)]2 accounts for P 96% of dissolved Fe and 56% of dissolved Mo at 1.0 mM S II, 2 while [Cu2(HS)2MoS4] accounts for 80% of dissolved Cu. This might explain the Fe-, Mo- and Cu-rich character of sulfidic waters in some alkaline lakes, such as Walker Lake.11,55 Thus [MoS4]2 complexing has the potential to affect trace metal speciation in nature profoundly, but at the pH of most marine and freshwater basins, this potential cannot be realized because of limited Mo solubility. Thioanions of As and other elements that are enriched, rather than depleted in sulfidic waters are likely to have a greater impact on trace element biogeochemistry. However the solubility limitation relaxes at higher pH. The elevated solubility of Mo in alkaline waters is consistent with recent ideas that Mo could have played an important catalytic role at the advent of life.45 An important point illustrated by Fig. 5 is that Fe and Cu thiomolybdate complexes are quantitatively most important at sulfide concentrations in the vicinity of the crossover from [MoO4]2 to [MoS4]2 predominance (the switch point; refer to the middle panel). At sulfide concentrations well below this point, [MoS4]2 (and all other thiomolybdates) are insufficiently stable to be meaningful ligands for metals; at sulfide concentrations well above this point, higher sulfide complexes ([Fe(HS)2]0, [Cu(HS)2] ) out-compete them.

1138 | Metallomics, 2014, 6, 1131--1140

Metallomics

Conclusions Measurements of Fe concentrations in [MoS4]2 solutions in contact with pyrrhotite can be explained by positing two complexes, [FeOOHMoS4]2 and [(Fe2S2)(MoS4)2]4 . Stability constants derived by fitting the measured Fe concentrations are the first obtained for Fe–Mo–S complexes in aqueous solution. No evidence of aqueous-phase cuboidal clusters was found, although their presence as minor species is not excluded. Even though thiomolybdate ligands occur at very low concentrations in natural waters, thermodynamic modeling indicates that the complexes that they form with metals could significantly affect metal biogeochemistry, especially in mildly alkaline waters.

Acknowledgements This work partially supported by U.S. National Science Foundation Grant EAR 9405432 and acknowledgement is made to the Donors of the American Chemical Society Petroleum Research Fund for additional support (PRF # 52201-UR2).

Notes and references 1 R. B. Wanty and M. B. Goldhaber, Thermodynamics and kinetics of reactions involving vanadium in natural systems: Accumulation of vanadium in sedimentary rocks, Geochim. Cosmochim. Acta, 1992, 56, 1472–1483. 2 M. Pettine, F. J. Millero and R. Passino, Reduction of chromium(VI) with hydrogen sulfide in NaCl media, Mar. Chem., 1994, 46, 335–344. 3 C. Kim, Q. Zhou, B. Deng, E. C. Thornton and H. Xu, Chromium(VI) reduction by hydrogen sulfide in aqueous media: stoichiometry and kinetics, Environ. Sci. Technol., 2001, 35, 2219–2225. 4 B. E. Erickson and G. R. Helz, Molybdenum(VI) speciation in sulfidic waters: Stability and lability of thiomolybdates, Geochim. Cosmochim. Acta, 2000, 64, 1149–1158. 5 G. R. Helz and M. K. Dolor, What regulates rhenium deposition in euxinic basins?, Chem. Geol., 2012, 304, 131–141. 6 G. R. Helz, C. V. Miller, J. M. Charnock, J. F. W. Mosselmans, R. A. D. Pattrick, C. D. Garner and D. J. Vaughan, Mechanism of molybdenum removal from the sea and its concentration in black shales: EXAFS evidence, Geochim. Cosmochim. Acta, 1996, 60, 3631–3642. 7 C. H. Van der Weijden, J. J. Middelburg, G. J. De Lange, H. A. Van der Sloot, D. Hoede and J. R. W. Woittiez, Profiles of the redox-sensitive trace elements As, Sb, V, Mo, and U in the Tyro and Bannock Basins (eastern Mediterranean), Mar. Chem., 1990, 31, 171–186. ´, N. Mikac and I. Ciglenecˇki, 8 G. R. Helz, E. Bura-Nakic New model for molybdenum behavior in euxinic waters, Chem. Geol., 2011, 284, 323–332. ¨rleyu ¨k, R. Jellison, S. G. 9 J. T. Hollibaugh, S. Carini, H. Gu ¨ger, Joye, G. LeCleir, C. Meile, L. Vasquez and D. Wallschla Arsenic speciation in Mono Lake, California: Response to

This journal is © The Royal Society of Chemistry 2014

View Article Online

Metallomics

10

Published on 07 November 2013. Downloaded by Umea University Library on 20/08/2014 22:31:18.

11

12

13

14

15

16

17

18

19

20

21

22

23

seasonal stratification and anoxia, Geochim. Cosmochim. Acta, 2005, 69, 1925–1937. M. B. Clarke and G. R. Helz, Metal-thiometalate transport of biologically active trace elements in sulfidic environments. 1. Experimental evidence for copper thioarsenite complexing, Environ. Sci. Technol., 2000, 34, 1477–1482. G. R. Helz and B. E. Erickson, Extraordinary stability of copper(I)-tetrathiomolybdate complexes: possible implications for aquatic ecosystems, Environ. Toxicol. Chem., 2011, 30, 97–102. D. Coucouvanis, Synthesis, structures, and reactions of binary and tertiary thiomolybdate complexes containing the (O)Mo(Sx) and (S)Mo(Sx) functional groups (x = 1, 2, 3), Adv. Inorg. Chem., 1998, 45, 1–73. S. Groysman and R. H. Holm, Biomimetic chemistry of iron, nickel, molybdenum and tungsten in sulfur ligated protein sites, Biochem., 2009, 48, 2310–2320. C. D. Garner, S. R. Acott, G. Christou, D. Collison, F. E. Mabbs and V. Petrouleas, Iron-molybdenum-sulphur clusters, Philos. Trans. R. Soc. London, Ser. A, 1982, 308, 159–166. R. P. Pesavento, C. P. Berlinguette and R. H. Holm, Stabilization of reduced molybdenum-iron-sulfur single- and double-cubane clusters by cyanide ligand, Inorg. Chem., 2007, 46, 510–516. J. A. Tossell, Theoretical studies on metal thioarsenites and thioantimonides: Synergistic interaction between transition metals and heavy metalloids, Geochem. Trans., 2000, 1, 16–22. Y. Moelo, E. Makovicky, N. N. Mozgova, J. L. Jambor, N. Cook, A. Pring, W. Paar, E. H. Nickel, S. Graeser, S. Karup-Moller, T. Balic-Zunic, W. G. Mumme, F. Vurro, D. Topa, L. Bindi, K. Bente and M. Shimizu, Sulfosalt systematics: a review. Report of the sulfosalt subcommittee of the IMA Commission on Ore Mineralogy, Eur. J. Mineral, 2008, 20, 7–46. T. W. Dahl, A. Chappaz, J. P. Fitts and T. W. Lyons, Molybdenum reduction in a sulfidic lake: Evidence from X-ray absorption fine-structure spectroscopy and implications for the Mo paleoproxy, Geochim. Cosmochim. Acta, 2013, 103, 213–231. B. C. Bostick, S. Fendorf and G. R. Helz, Differential adsorption of molydate and tetrathiomolybdate on pyrite (FeS2), Environ. Sci. Technol., 2003, 37, 285–291. P. G. Falkowski, R. T. Barber and V. Smetacek, Biogeochemical controls and feedbacks on ocean primary productivity, Science, 1998, 281, 200–206. J. B. Glass, R. P. Axler, S. Chandra and C. R. Goldman, Molybdenum limitation of microbial nitrogen assimilation in aquatic ecosystems and pure cultures, Front. Microbiol., 2012, 3, DOI: 10.3389/fmicb.2012.00331. J. B. Glass and V. J. Orphan, Trace metal requirements for microbial enzymes involved in the production and consumption of methane and nitrous oxide, Front. Microbiol., 2012, 3, DOI: 10.3389/fmicb.2012.00061. K. D. Daskalakis and G. R. Helz, Solubility of CdS (greenockite) in sulfidic waters at 25 1C, Environ. Sci. Technol., 1992, 26, 2462–2468.

This journal is © The Royal Society of Chemistry 2014

Paper

24 W. Stumm and J. J. Morgan, Aquatic Chemistry, John Wiley & Sons Inc., 1996. 25 D. Shea and G. R. Helz, The solubility of copper in sulfidic waters: sulfide and polysulfide complexes in equilibrium with covellite, Geochim. Cosmochim. Acta, 1988, 52, 1815–1825. ´ osovic ´, 26 E. Bura-Nakic´, G. R. Helz, I. Ciglenecˇki and B. C Reduced sulfur species in a stratified seawater lake (Rogoznica Lake, Croatia); Seasonal variations and argument for organic carriers of reactive sulfur, Geochim. Cosmochim. Acta, 2009, 73, 3738–3751. 27 A. Kamyshny, J. Gun, D. Rizkov, T. Voitsekovski and O. Lev, Equilibrium distribution of polysulfide ions in aqueous solutions at different temperatures by rapid single phase derivatization, Environ. Sci. Technol., 2007, 41, 2395–2400. 28 F. Bonomi, S. Iametti and D. M. Kurtz, Acceleration by Fe(II) of thiomolybdate formation from aqueous molybdate and sulfide. A simplified synthesis of [Fe(MoS4)2]3 , Inorg. Chim. Acta, 1992, 193, 125–128. 29 E. W. King and R. Farlow, Role of carbonate speciation on the oxidation of Fe(II) by H2O2, Mar. Chem., 2000, 70, 201–209. 30 W. Davison, N. Phillips and B. J. Tabner, Soluble iron sulfide species in natural waters: Reappraisal of their stoichiometry and stability constants, Aquat. Sci., 1999, 61, 23–43. 31 B. W. Mountain and T. M. Seward, The hydrosulphide/ sulphide complexes of copper(I): experimental determination of stoichiometry and stability at 22 1C and reassessment of high temperature data, Geochim. Cosmochim. Acta, 1999, 63, 11–29. ´ and G. R. Helz, Voltammetry of 32 I. Ciglenecˇki, D. Krzaric copper sulfide particles and nanoparticles: Investigation of the cluster hypothesis, Environ. Sci. Technol., 2005, 39, 7492–7498. 33 K. Daskalakis and G. R. Helz, The solubility of sphalerite (ZnS) in sulfidic solutions at 25 1C and 1 atm pressure, Geochim. Cosmochim. Acta, 1993, 57, 4923–4931. ¨ger, Solubility constants and free 34 R. Heindl and H. Gamsja enthalpies of metal sulphides, Part 6: A new solubility cell, Monatsh. Chem., 1977, 108, 1365–1369. ¨ger, H. F. Reiterer and R. Heindl, Solubility con35 H. Gamsja stants and free enthalpies of metal sulphides and carbonates, Ber. Bunsen-Ges. Phys. Chem., 1982, 86, 1046–1049. 36 D. Rickard, A. Griffith, A. Oldroyd, I. B. Butler, E. LopezCapel, D. A. C. Manning and D. C. Apperley, The composition of nanoparticulate mackinawite, tetragonal iron(II) monosulfide, Chem. Geol., 2006, 235, 286–298. 37 H. Y. Jeong, J. H. Lee and K. F. Hayes, Characterization of synthetic nanocrystalline mackinawite: Crystal structure, particle size and specific surface area, Geochim. Cosmochim. Acta, 2008, 72, 493–505. 38 T. P. Vorlicek, M. D. Kahn, Y. Kasuya and G. R. Helz, Capture of molybdenum in pyrite-forming sediments: Role of ligand-induced reduction by polysulfides, Geochim. Cosmochim. Acta, 2004, 68, 547–556. 39 G. R. Helz, T. P. Vorlicek and M. D. Kahn, Molybdenum scavenging by iron monosulfide, Environ. Sci. Technol., 2004, 38, 4263–4268.

Metallomics, 2014, 6, 1131--1140 | 1139

View Article Online

Published on 07 November 2013. Downloaded by Umea University Library on 20/08/2014 22:31:18.

Paper

¨gge, E. Bill and ¨ller, A. Hildebrand, M. Penk, H. Bo 40 A. Mu A. Trautwein, [Fe2S2(MoS4)2]4 : the first complex with two MoS42 ions coordinated to a ferredoxin type central unit, Inorg. Chim. Acta, 1988, 148, 11–13. 41 B. Zhuang, J. W. McDonald and W. E. Newton, Synthesis of the [(AcO)2FeS2MS2]2 (M = Mo, W) ion and observation of a new bonding mode for complexes of [MoO2S2]2 , Inorg. Chim. Acta, 1983, 77, L221–L223. 42 H. Tieckelmann, H. C. Silvis, T. A. Kent, B. H. Huynh, J. V. Waszczak, B.-K. Teo and B. A. Averill, Synthetic molybdenum-iron-sulfur clusters. Preparation, structures and properties of the [S2MoS2Fe(SC6H5)2]2 and [S2MoS2FeCl2]2 ions, J. Am. Chem. Soc., 1980, 102, 5550–5559. 43 L. J. Laughlin and D. Coucouvanis, Use of [MoFe3S4]3+ single cubanes in the catalytic reduction of acetylene to ethylene and ethane. Identification of molybdenum and iron atoms as catalytic sites during substrate reduction and implication for nitrogenase action, J. Am. Chem. Soc., 1995, 117, 3118–3125. 44 R. A. Henderson, Mechanistic studies on synthetic Fe-S-based clusters and their relevance to the action of nitrogenases, Chem. Rev., 2005, 105, 2365–2437. 45 B. Schoepp-Cothenet, R. van Lis, P. Philippot, A. Magalon, M. J. Russell and W. Nitschke, The ineluctable requirement for the trans-iron elements molybdenum and/or tungsten in the origin of life, Sci. Rep., 2012, 2, DOI: 10.1038/srep00263. 46 L. Jacobs and S. Emerson, Partitioning and transport of metals across the O2/H2S interface in a permanently anoxic basin: Framvaren Fjord, Norway, Geochim. Cosmochim. Acta, 1985, 49, 1433–1444.

1140 | Metallomics, 2014, 6, 1131--1140

Metallomics

47 W. M. Landing and S. Westerlund, The solution chemistry of iron(II) in Framvaren Fjord, Mar. Chem., 1988, 23, 329–343. 48 B. L. Lewis and W. M. Landing, The biogeochemistry of manganese and iron in the Black Sea, Deep Sea Res., 1991, 38, S773–S803. 49 B. L. Lewis and W. M. Landing, The investigation of dissolved and suspended-particulate trace metal fractionation in the Black Sea, Mar. Chem., 1992, 40, 105–141. 50 X. Li, G. A. Cutter, R. C. Thunell, E. Tappa, W. P. Gilhooly, T. W. Lyons, Y. Astor and M. I. Scranton, Particulate sulfur species in the water column of the Cariaco Basin, Geochim. Cosmochim. Acta, 2011, 75, 148–163. 51 H. C. Jenkyns, Geochemistry of oceanic anoxic events, Geochem., Geophys., Geosys., 2010, 11, DOI: 10.1029/ 2009GC002788. 52 S. W. Poulton, P. W. Fralick and D. E. Canfield, The transition to a sulphidic ocean B1.84 billion years ago, Nature, 2004, 431, 173–177. 53 B. Kendall, G. W. Gordon, S. W. Poulton and A. D. Anbar, Molydenum isotope constraints on the extent of late Paleoproterozoic ocean anoxia, Earth Planet. Sci. Lett., 2011, 307, 450–460. 54 R. Buick, Did the Proterozoic ‘Canfield Ocean’ cause a laughing gas greenhouse?, Geobiology, 2007, 5, 97–100. 55 J. L. Domagalski, H. P. Eugster and B. F. Jones, Trace metal geochemistry of Walker, Mono and Great Salt lakes, in Fluid-Mineral Interactions: A tribute to H. P. Eugster, ed. R. J. Spencer and I.-M. Chou, 1990, pp. 315–353.

This journal is © The Royal Society of Chemistry 2014