Formation of Mercury Sulfide from Hg(II)âThiolate Complexes in

Article pubs.acs.org/est

Formation of Mercury Sulfide from Hg(II)−Thiolate Complexes in Natural Organic Matter Alain Manceau,*,† Cyprien Lemouchi,†,‡ Mironel Enescu,§ Anne-Claire Gaillot,∥ Martine Lanson,† Valérie Magnin,† Pieter Glatzel,⊥ Brett A. Poulin,#,∇ Joseph N. Ryan,# George R. Aiken,∇ Isabelle Gautier-Luneau,‡ and Kathryn L. Nagy*,@ †

ISTerre, Université Grenoble Alpes, CNRS, 38000 Grenoble, France Institut Néel, Université Grenoble Alpes, CNRS, 38000 Grenoble, France § Laboratoire Chrono Environnement, Université de Franche-Comté, CNRS, 25030 Besançon, France ∥ Institut des Matériaux Jean Rouxel (IMN), Université de Nantes, CNRS, 2 Rue de la Houssinière, 44322 Nantes, France ⊥ European Synchrotron Radiation Facility (ESRF), 71 Rue des Martyrs, 38000 Grenoble, France # Department of Civil, Environmental, and Architectural Engineering, University of Colorado Boulder, 428 UCB, Boulder, Colorado 80309-0428, United States ∇ U.S. Geological Survey, 3215 Marine Street, Boulder, Colorado 80303, United States @ Department of Earth and Environmental Sciences, University of Illinois at Chicago, MC-186, 845 West Taylor Street, Chicago, Illinois 60607, United States

Downloaded by UNIV OF CAMBRIDGE on August 22, 2015 | http://pubs.acs.org Publication Date (Web): July 28, 2015 | doi: 10.1021/acs.est.5b02522

‡

S Supporting Information *

ABSTRACT: Methylmercury is the environmental form of neurotoxic mercury that is biomagnified in the food chain. Methylation rates are reduced when the metal is sequestered in crystalline mercury sulfides or bound to thiol groups in macromolecular natural organic matter. Mercury sulfide minerals are known to nucleate in anoxic zones, by reaction of the thiol-bound mercury with biogenic sulfide, but not in oxic environments. We present experimental evidence that mercury sulfide forms from thiolbound mercury alone in aqueous dark systems in contact with air. The maximum amount of nanoparticulate mercury sulfide relative to thiolbound mercury obtained by reacting dissolved mercury and soil organic matter matches that detected in the organic horizon of a contaminated soil situated downstream from Oak Ridge, TN, in the United States. The nearly identical ratios of the two forms of mercury in field and experimental systems suggest a common reaction mechanism for nucleating the mineral. We identified a chemical reaction mechanism that is thermodynamically favorable in which thiolbound mercury polymerizes to mercury−sulfur clusters. The clusters form by elimination of sulfur from the thiol complexes via breaking of mercury−sulfur bonds as in an alkylation reaction. Addition of sulfide is not required. This nucleation mechanism provides one explanation for how mercury may be immobilized, and eventually sequestered, in oxygenated surface environments.

■

INTRODUCTION Soils are a main depository of mercury (Hg) from geologic and anthropogenic sources and a persistent long-term source of this toxic metal to aquatic wildlife and ultimately humans.1 Mercury(II) deposited from the atmosphere appears to become largely immobilized within days to weeks1,2 likely by reaction with thiol sulfur ligands of macromolecular soil organic matter (SOM) to form Hg−SOM complexes.3−5 The mercury in these complexes eventually may transform to bioavailable methylmercury, or relatively inert mercury sulfides,2 but reaction mechanisms are difficult to study in situ because of the predominance of Hg(II) in forms with long residence times and low reactivity. Understanding the conversion of newly deposited Hg to less reactive forms is necessary to mitigate the © 2015 American Chemical Society

impact of this potent toxin on the biosphere and to predict mercury mobility and bioavailability as global emissions and environmental conditions change.6 The current paradigm is that Hg sulfide minerals form only where sulfide [S(-II)] is produced from porewater sulfate by microbial activity,7 which requires suboxic to anoxic conditions. This and other biogenic processes can also generate toxic methylmercury.8 Mercury would precipitate mostly as metacinnabar (β-HgS),9,10 which thermodynamically should recrysReceived: Revised: Accepted: Published: 9787

May 22, 2015 July 10, 2015 July 13, 2015 July 13, 2015 DOI: 10.1021/acs.est.5b02522 Environ. Sci. Technol. 2015, 49, 9787−9796

Article


Environmental Science & Technology

assumption that the Elliott Soil humic acid is similar to the macromolecular SOM at the field site. Five experiments were conducted at equilibration times of 15 h, 3 days, 10 days, 1 month, and 6 months. In experiments longer than 15 h, the reaction vessels were opened periodically, the suspensions were stirred in contact with air, and the pH was readjusted to 6 with 0.1 M HNO3 as needed (Table S1 of the Supporting Information). The dissolved SOM formed a stable colloidal suspension at all times with no visual evidence of biofilms. After equilibration, the SOM was separated by filtration on a stirred ultrafiltration cell (Millipore 8200) and the wet powder freeze-dried. A maximum of 3% of the mercury remained in solution, presumably bound to small organic molecules that passed through the filter. The sample equilibrated for 15 h was used as a spectroscopic reference, termed Hg−SOM, for a linearly coordinated mercury−thiol sulfur complex [Hg(SR)2].5 Two aqueous solutions of Hg(cysteine)2 complexes ([Hg(SCH2CH(NH3)CO2CH2CH3)2]2+ linear complex as confirmed by 1H and 13C nuclear magnetic resonance) were prepared at pH 4.5 and aged for up to 80 days in the dark. Headspace gas, either Ar or air (∼3:1 gas:solution ratio), was replaced every 4 days (Supporting Information Part 1). Formation of poorly crystallized β-HgS in each experiment was confirmed by powder XRD (Figure S1B of the Supporting Information), and the nanoparticulate β-HgS synthesized in the Ar-equilibrated experiment was used to fit the soil spectra. Twenty well-characterized Hg complexes and inorganic compounds, including well-crystallized β-HgS prepared by solvothermal synthesis, were also prepared or purchased. Detailed information about the synthesis procedures and characterization of the samples is given in the Part 1 of the Supporting Information. HR-XANES Spectroscopy. All spectra were measured in fluorescence yield detection mode with analyzer crystals on ID26 at the European Synchrotron Radiation Facility (ESRF). The flux on the sample was approximately 1 × 1013 photons/s in a beam footprint on the sample of ∼500 μm (H) × 80 μm (V) full width at half-maximum (FWHM). The Hg Lα1 (3d5/2 → 2p3/2) fluorescence line was selected using the 555 reflection of five spherically bent (radius of 1 m) Si analyzer crystals (diameter of 100 mm) aligned at an 81.8° Bragg angle in a vertical Rowland geometry. The diffracted intensity was measured with a Si detector in single-photon counting mode. Measuring the fluorescence intensity with an X-ray spectrometer had two main advantages. First, it increased the signal-to-noise ratio by eliminating most of the unwanted background radiation as only the Lα1 line was detected. For this reason, this method is also termed high energy-resolution fluorescence detection X-ray absorption spectroscopy (HERFD-XAS), and partial fluorescence yield XAS (PFY-XAS) to differentiate it from standard XAS in which both the Lα1 and Lα2 (or Kα1 and Kα2) lines are detected with a solid-state detector.21−24 Second, the natural line width is approximately halved, yielding a high spectral resolution. The effective energy resolution, obtained by convoluting the total instrumental energy bandwidth (spreads of the incident and emitted rays) and the 3d5/2 hole width from the Lα1 line, was ∼3.0 eV, compared to ∼6.1 eV in a standard fluorescence experiment. The samples were prepared as pressed pellets and cooled to 10 K (He-T) using a liquid He cryostat to minimize radiation damage, such as the photoreduction of Hg2+ to Hg0. In addition, spectra were collected in quick-scan mode to reduce the exposure time, and the sample was moved horizontally and vertically to

tallize to cinnabar (α-HgS), although the rate of this reaction can be altered by impurities or reduced sulfur species11 under ambient anoxic conditions. In addition, β-HgS formed in experimental systems with dissolved organic matter (DOM), added Hg(II), and added sulfide is nanoparticulate, suggesting that interactions with organic matter slow aggregation and crystal growth.12−14 Here we report results for Hg speciation in the top few centimeters of a contaminated floodplain soil and in the products of experiments designed to characterize the evolution of Hg complexation in dissolved SOM. We applied high energyresolution X-ray absorption near-edge structure (HR-XANES) spectroscopy15−17 at the mercury L3 edge to identify and quantify the dominant forms of mercury. In both systems, the dominant Hg species were nanoparticulate β-HgS and linear Hg−SOM complexes. The ratio of Hg in Hg−SOM complexes to nanoparticulate β-HgS in the fine fraction of the topsoil was the same as the ratio obtained after reacting Hg(II) with dissolved SOM in contact with air for 6 months at pH 6. Reaction of model Hg(cysteine)2 thiolate complexes in solution also produced β-HgS, supporting the experimental results seen with the Hg−SOM complex. A sequence of plausible chemical reactions and structural transformations obtained from ab initio calculations is proposed to explain the formation of β-HgS from Hg−thiol complexes.

■

MATERIALS AND METHODS Materials. The soil was collected in a dry riparian floodplain of a shallow creek initially contaminated in the 1950s and 1960s by the discharge of predominantly dissolved Hg(II) from the U.S. Department of Energy Oak Ridge Y-12 Plant into the headwaters approximately 17 km upstream from the sampling site (Supporting Information Part 1).18 We examined the speciation of Hg in the clay-sized (clay) fraction from the topsoil below the litter layer (O horizon) and the clay- and silt-sized (silt) fractions from the well-drained underlying layer (A horizon) situated above a deeper redoximorphic horizon. The three fractions contain 48, 191, and 81 mg of Hg/kg dry weight, respectively. Organic material was not removed prior to size fractionation. The higher Hg concentrations in the clay and silt fractions from the A horizon correlate with higher historical Hg releases and the subsequent burial of older deposits with relatively less contaminated river sediments.10 Mercury is also more concentrated in the separated clay fraction relative to the silt fraction of the A horizon because the mercury is present mainly as submicrometer β-HgS particles.9,10 Some mercury in the soil previously was inferred to be associated with organic material, including plants and coal coke,9 but the nature of this mercury has not been determined. Mercury was complexed with Elliott Soil humic acid [International Humic Substances Society (IHSS), catalog no. 1S102H] at 20.0 ± 0.1 °C and pH 6 in closed glass vessels in the absence of light, as in previous work.5,19 The vessels contained ambient air and solution in a ratio of approximately 1:1.4 (v/v). A mass of 66 mg of freeze-dried SOM was suspended in 120 mL of 0.01 M NaNO3 at pH 6 for 5 h. Then, 658 μL mercury from a 0.1 mM Hg(NO3 )2 stock solution was added to obtain a concentration of 200 mg of Hg/kg of SOM (200 ppm of Hg), and the pH was readjusted with 0.1 M NaOH. The experimental molar ratio of mercury to thiol sulfur20 is 1:25, a reasonable approximation for the ratio in the fine fraction of the O horizon soil based on the measured Hg concentration and the 9788

DOI: 10.1021/acs.est.5b02522 Environ. Sci. Technol. 2015, 49, 9787−9796

Article


Environmental Science & Technology access unexposed material for every scan. The scan time varied from 20 to 120 s depending on the sensitivity of the sample to radiation damage. The total number of scans per sample was adjusted to yield optimal signal-to-noise ratios. The incident energy was scanned from 12260 to 12360 eV in 0.2 eV steps. The HR-XANES spectra were normalized to unity at an E of 12360 eV. Computational Methodology. Thermochemical calculations and geometry optimizations were performed with Gaussian 09,25 using a computational scheme tested previously on the modeling of the structure and stability of monomeric Hg− thiolate complexes.26 All calculations were performed in water that was represented by the polarizable continuum model (PCM).27 Single-point energy calculations were performed at the CCSD(T)28 level of theory on molecular geometries preoptimized using the second-order Møller−Plesset perturbation theory (MP2).29 Thermal corrections to the Gibbs free energy at 298 K and 1 atm were obtained by performing frequency calculations at the MP2 level of theory. The standard Gibbs free energy (Gaq) of a species S in aqueous solution was calculated according to the formula Gaq (S) = Eaq (S) + ΔGc,aq (S)

where Eaq(S) is the potential energy (the sum of electronic energy and nuclear repulsion energy) and ΔGc,aq(S) is the thermal Gibbs free energy correction. The R group in the model systems was represented by the methyl group for simplification, as the dissociation energy of the R−SH bond is almost independent of R.30 Detailed information is given in Part 1 of the Supporting Information, and other accounts of MP2 calculations on Hg are described in refs 31 and 32.

■

Figure 1. High-resolution (HR, blue) and normal-resolution (gray) Hg L3-edge XANES spectra at 10 K (liquid helium temperature, He-T) of Hg(II) complexed to (A) soil organic matter (Hg−SOM; Elliott Soil humic acid aged for 15 h in the dark with pH 6 air-equilibrated water), (B) cinnabar (α-HgS), and (C) metacinnabar (β-HgS), with ball-andstick representations of the Hg−S bonds. Magenta, yellow, gray, and light gray spheres represent mercury, sulfur, carbon, and hydrogen atoms, respectively. R in panel A represents the remainder of the natural organic molecule(s) bonded to Hg. Preparation and characterization of well-crystallized β-HgS (solvothermal synthesis) and Hg−SOM are described in Part 1 of the Supporting Information.

RESULTS AND INTERPRETATIONS Coexistence of Thiol-Bound Hg(II) and Nanoparticulate β-HgS in Soil. Our initial interest was to identify the Hg− SOM complexes expected in the soil, which are overwhelmed by the abundance of β-HgS. To this end, we measured precisely the bulk ratios of Hg−SOM complexes to β-HgS at 10 K by HRXANES spectroscopy. This method has enhanced sensitivity to Hg speciation, as seen by the prominent near-edge peak for the Hg−SOM complexes at 12279.2 eV (2p3/2 → 6s/5d electronic transition), which is weak in standard fluorescence measurement (Figure 1 and Part 2 of the Supporting Information).33−35 The near-edge peak occurs when Hg(II) is coordinated linearly, for instance, to two thiol sulfur ligands as in SOM, Hg(cysteine)2 complexes [Hg(SR)2] and cinnabar (α-HgS); to one thiol sulfur and one methyl group (CH3HgSR complex); or to oxygen (HgO, Hg acetate), chloride (HgCl2), or iodide (HgI2) ligands (Figure 1 and Part 3 of the Supporting Information). The β-HgS spectrum at He-T (Figure 1C) lacks the near-edge peak because Hg is tetrahedrally coordinated. The β-HgS spectrum can be distinguished from spectra of Hg(SR) 4 compounds by modulations beyond the absorption edge. In this region of energy above the electronic transitions to bound 6s/5d states, photoelectrons emitted during X-ray absorption have a large mean free path (i.e., low kinetic energy) and can be scattered by distant atoms.36,37 For β-HgS, scattering from distant Hg atoms is observed at He-T, but not at room temperature (RT; arrows in Figure 2), because of the large thermal motion of mercury.5 Therefore, HR-XANES spectroscopy at He-T can be used to resolve coexisting species and, in particular, to quantify key Hg−

SOM complexes in multicomponent soils containing β-HgS (Part 4 of the Supporting Information). The soils of the O and A horizons of the floodplain (Figure 2A) have different mercury speciation (Figure 2B). In the A horizon, the average speciation is the same in the silt and clay fractions, as shown by their statistically identical HR-XANES spectra at He-T (Part 5 of the Supporting Information). These spectra resemble most closely those of β-HgS at both He-T and RT (Figure 2C), confirming that Hg is β-HgS in this deeper soil horizon. In addition, the β-HgS is dominantly nanoparticulate because the A horizon spectra match better the RT β-HgS spectrum in the above-edge region (Figure 2B) and are also similar to our reference spectra for nanocrystallized β-HgS (Figure S7 of the Supporting Information), indicating that the Hg atoms have on average fewer Hg neighbors (i.e., lower coordination) than in micrometer-size crystals. Mercury may also be sorbed on Al- or Fe-oxyhydroxides in well-drained soils, in addition to being complexed to organic matter.38 The mineral surface complexation is with singly or doubly coordinated 9789


Article



Figure 2. Mercury speciation in the soil O and A horizons (shown below plant litter in panel A) by Hg L3-edge HR-XANES spectroscopy. (B) Spectra of the A horizon (light brown) and O horizon (magenta) measured at 10 K (He-T). (C) Comparison of the A horizon spectrum (light brown) to those of well-crystallized β-HgS (black) at He-T and 293 K (room temperature, RT). Modulations in the above-edge region of the He-T β-HgS spectrum (blue arrows) indicate Hg−Hg pairs; structural disorder resulting from thermal agitation dampens the signal at RT. The He-T A horizon spectrum is similar to the RT well-crystallized β-HgS spectrum, and also to the He-T nanoparticulate β-HgS spectrum (Figure S7 of the Supporting Information), which is evidence of β-HgS nanoparticles in the soil. Polyhedral representation is of the β-HgS structure. (D, top) Linear least-squares fit (black) to the O horizon spectrum (magenta) with 74 ± 6 mol % nanoparticulate β-HgS [black precipitate from Hg-(L-Cys-OEt)2 aged for 80 days under an argon atmosphere] and 26 ± 6 mol % organically bound Hg (Hg−SOM from Figure 1A). Ball-and-stick representations at the bottom right are of the two Hg species. (D, bottom) Spectra of the O horizon (magenta) and the Hg−SOM complex aged for 6 months at pH 6 in contact with air-equilibrated water (black). Mercury is present in the same forms and proportions in both materials. The normalized sum-squared residual (NSS) is the normalized difference between two spectra expressed as ∑[(yexp − yfit)2]/∑yexp2.

α-HgS and β-HgS was also verified on the derivative spectra (Figure S2 of the Supporting Information). Likewise, the amount of Hg−SOM was within the 6% precision when added to the A horizon model (Part 5 of the Supporting Information). Adding a methylmercury−SOM complex also did not improve the fit. In summary, Hg occurs as β-HgS in the subsurface A horizon and as both β-HgS and Hg−SOM complexes in the surface O horizon. Thiol Groups in Soil Organic Matter as the Source of Reduced Sulfur in β-HgS. We expected the Hg−SOM complex to dominate in the oxygenated surface soil based on previous results for the top 1 cm of bulk soil at the field site obtained using sequential extraction analysis, which showed that approximately 90% of the total mercury was complexed to the organic matter.43 Therefore, we questioned why there should be such a large proportion of nanoparticulate β-HgS in the O horizon clay fraction. We considered that multiple processes might explain the presence of this metacinnabar, given that the field site is 17 km downstream from the mercury point source. For example, the mineral could be derived from contaminated sediments eroded upstream and deposited in the floodplain or from floodwaters carrying sulfide nanoparticles complexed to DOM, as reported for copper and zinc in rivers.44 Metacinnabar

oxygens,39,40 which provide a distinctive spectral signature from the sulfur ligands observed here (Figure S7 of the Supporting Information). If present, oxygen-bound Hg species are minor. The spectroscopy results provide information that cannot be obtained with electron diffraction, the technique used previously to identify micrometer to submicrometer β-HgS crystallites in similar soils.10 Diffraction preferentially detects the larger particles in a powder and underestimates the proportion of nanoparticles, whereas every atom in every particle is detected equally via spectroscopy.41 The clay fraction of the O horizon has two main spectral features that distinguish it from the A horizon: a well-resolved peak for organically bound Hg at ∼12279 eV and a rightward shift of the trailing edge (arrow in Figure 2B). An unconstrained linear fit to all the reference spectra (“Combo fit”20,42) identified two components, Hg complexed to SOM and nanosized β-HgS, with fractional amounts of 26 ± 6 and 74 ± 6%, respectively (Figure 2D). The β-HgS component is represented equally well by our nanoparticulate β-HgS reference or the A horizon spectrum (Figure S7 of the Supporting Information). When αHgS was added, its amount was estimated to be within the 6% precision of the fitting approach. The good distinction between 9790


Article



Figure 3. Formation of β-HgS from aged Hg−SOM and Hg−cysteine complexes. (A) He-T HR-XANES spectra of Hg(II) reacted with the humic acid fraction from Elliott Soil at pH 6 in air-equilibrated water in the dark for 15 h (Hg−SOM in Figure 1A), 3 days, 10 days, 1 month, and 6 months. Approximately 74% of the Hg atoms, initially coordinated to two thiol S atoms, progressively formed nanoparticulate β-HgS with each Hg atom bonded to four S atoms (Figure 1C). (B) Transformation of Hg−SOM complexes to β-HgS nanoparticles as a function of log time. Amounts of β-HgS were obtained by fitting each aged Hg−SOM sample with nanoparticulate β-HgS and 15 h Hg−SOM spectra (Figure 1A). The sigmoidal curve indicates a nucleation−growth mechanism, here described by the Avrami−Erofeev equation. (C) HRTEM image of β-HgS nanocrystals ∼3−5 nm in diameter from the Hg−SOM complex aged for 6 months. Fourier-filtered lattice fringes (right) and Fourier transform (middle) are of a single nanocrystal closely oriented along the ⟨110⟩ zone axis. (D) Photographs of experiments conducted in the dark at 293 K in the presence of air or argon showing the transformation of Hg(II) complexed to L-Cys-OEt into β-HgS.

relatively large amount of β-HgS from Hg−SOM complexes that does not involve microbial production of sulfide. Could thiol sulfur in natural organic matter in the floodplain be a source of sulfide in the nanoparticulate β-HgS? We tested the hypothesis that β-HgS could have formed in situ from mercury complexed to the soil organic matter by aging the Hg−SOM complex used to fit the spectrum of the O horizon

also might be transported upward from the A horizon, for instance, through bioturbation or formed in reducing microenvironments during periods of heavy rainfall by microbial production of sulfide from porewater sulfate. However, because the O horizon must be an overall oxygenating environment to support the mixed hardwood forest that exists in the floodplain, we asked if there could be a mechanism for generating the 9791


Article

Environmental Science & Technology field sample for periods of up to 6 months and examining the products using HR-XANES spectroscopy. With time, the peak at 12279.2 eV from linear Hg(SR)2 vanished and weak modulations from Hg−Hg pairs appeared in the above-edge region (Figure 3A). After 6 months, the spectrum at He-T was nearly identical to that of the O horizon, indicating spontaneous formation of nanoparticulate β-HgS (Figure 2D). Direct transformation of Hg(SR)2 to β-HgS is demonstrated by isosbestic points where all HR-XANES curves meet (Figure 3A), which are characteristic of binary (i.e., two-component) systems. The transformation versus log time curve is sigmoidal and approaches a plateau at a mole fraction, f, of Hg as β-HgS equal to 0.75−0.80 (Figure 3B). Fitting to the Avrami−Erofeev equation

particles formed in a two-step reaction sequence that was initiated with the Hg(SR)2 complex in SOM. Our model steps are similar to the reaction mechanisms observed in the formation of early HgxSy clusters in sulfidic aqueous solutions.50 In these solutions, aqueous Hg(II) and S(-II) initially bonded as linear [SHg-S]2− units that evolved after a few seconds into mercury sulfide chains. The chains then rapidly transformed to a disordered (pseudocubic) form, followed after 3−4 h by an ordered (cubic) β-HgS precipitate in which Hg is fourcoordinate with S [d(Hg−S) = 2.53 Å].51 The first reaction step of our model is RS−Hg−SR + RS−Hg−SR → RS−Hg−S−Hg−SR + R−S−R


f = a[1 − exp( −kt n)] = a[1 − exp( −0.693(t /t1/2)n )]

(1)

The chain can be extended by the addition of Hg(SR)2 complexes:

for nucleation−growth reactions in the solid state,45 where a is a constant, t is time, the Avrami rate coefficient k is described in terms of the crystallization half-time t1/2, and n is the reaction order, yielded a = 0.73 ± 0.02, t1/2 = 7.4 ± 0.5 days, and n = 2.0 ± 0.3. The a value of

Formation of Mercury Sulfide from Hg(II)âThiolate Complexes in

Recommend Documents

Formation of Mercury Sulfide from Hg(II)âThiolate Complexes in