Dimethylmercury Degradation by Dissolved Sulfide and Mackinawite.

Potential degradation pathways of dimethylmercury (DMHg) remain as one of the critical knowledge gaps in the marine biogeochemical cycle of mercury (Hg). Although Hg is known to be highly reactive with reduced sulfur, demethylation of DMHg in the presence of sulfide has until now remained experimentally untested. Here, we provide the first experimental support for demethylation of DMHg to monomethylmercury (MMHg) in the presence of both dissolved sulfide and mackinawite (FeS(s)m). The degradation of DMHg was shown to be pH dependent, with higher demethylation rates at pH 9 than pH 5. At room temperature and environmentally relevant DMHg to sulfide molar ratios, we observed demethylation rates up to 0.05 d-1. When comparing the number of active sites available, FeS(s)m was found to have a higher capacity to demethylate DMHg, in comparison with dissolved sulfide. Our study suggests that dissolved sulfide and FeS(s)m mediated demethylation of DMHg may act as a sink for DMHg, and a potential source of MMHg, in aquatic systems.

Introduction

Global efforts are currently undertaken to lower human risks of Hg exposure by mitigating anthropogenic Hg sources. Processes controlling the pool of monomethylmercury (CH3Hg-X, where X represents a counteranion, hereafter referred to as MMHg) available for bioaccumulation in marine food webs are critical when linking natural and anthropogenic emissions of Hg with human Hg exposure, but some processes remain poorly constrained. One of the major knowledge gaps urgently needing attention, in this respect, is the unknown role of dimethylmercury ((CH3)2Hg, henceforth referred to as DMHg) in the marine biogeochemical cycle of Hg. Dimethylmercury constitutes around 30–80% of the methylated Hg pool, and composes, together with MMHg, around 20% of the total Hg in open marine waters.[1,2] Despite its hydrophobic properties, it is not clear if DMHg bioaccumulates in aquatic food webs or not.[3] It may, however, still potentially play a role in the amount of Hg accumulated in food webs by acting as a source of MMHg when demethylated. DMHg has, for example, been suggested to be an important source of MMHg in snow and coastal areas after volatilization from marine waters.[4−6] Recent reports also suggests DMHg production in rice paddies, and a role for DMHg in the amounts of MMHg accumulated in the rice.[7,8] Pathways for DMHg demethylation are poorly known, however.

Early work suggested DMHg to be fairly stable in aqueous solutions in the pH range of 5–8, at low temperatures, and in the absence of light.[31,32] When exploring novel DMHg degradation pathways, a logical starting point is to look at elements known to be highly reactive with Hg as well as pathways already identified for the degradation of MMHg. Reduced sulfur, or sulfide, is known to have a strong affinity for Hg and to play a key role in many of the known transformation pathways of the Hg cycle. For example, Jonsson et al. recently demonstrated that MMHg may react with dissolved sulfide and sulfide minerals forming inorganic Hg and DMHg (2 MMHg + 2 ≡ −S– or S2–(aq) → 1 DMHg + ≡ 1 −S2–Hg or 1 HgS(s)).[9] Reduced sulfur also plays a key role in bacterial detoxification of MMHg via the MerB enzymatic pathway, in which the Hg–C bond is broken after binding to the thiol groups of cysteine residues at the active site of the protein.[10]

With the help of density function theory (DFT) and quantum theory of atoms in molecules (QTAIM), Ni et al.[11] proposed a pathway for DMHg demethylation by sulfide. This pathway begins by (i) binding of a HS– group to the Hg atom of DMHg (forming [Hg(CH3)2SH]− as an intermediate) and then (ii) proton transfer from H2S to one of the methyl groups (after formation of a [Hg(CH3)2SHSH2]− intermediate). The calculated energy barriers for the two steps (in gaseous phase) are low, 0 and 20.4 kcal mol–1 for (i) and (ii), respectively, and support demethylation of DMHg by reduced sulfur. So far, support for this degradation pathway is limited to this theoretical model. Here, we present the first experimental support for degradation of DMHg in the presence of dissolved sulfide. We have also tested the stability of DMHg in the presence of mackinawite (FeS(s)m), and cadmium sulfide (CdS(s)). Finally, degradation products were quantified and DMHg degradation was tested as a function of DMHg:dissolved sulfide (DMHg:S(aq)) and DMHg:FeS(s)m molar ratios and pH.

Material and Methods

Caution

Dimethylmercury is a volatile and extremely toxic compound! Adsorption through the skin has proven fatal even for small amounts. Work with concentrated solutions should be avoided and all work needs to be performed with extreme caution and suitable protective equipment.

Experimental Setup

Information about preparations of chemicals for experiments is available in the Supporting Information (SI). Stock solutions of sulfide, sulfide mineral slurries (containing mackinawite, FeS(s)m, or cadmium sulfide, CdS(s)) and buffers were prepared and added to N2-degassed deionized water (>18 MΩ-cm) in precombusted 40 mL amber glass vials under oxygen free conditions using a glovebox with N2(containing up to 5% H2) atmosphere. Citrate, potassium phosphate, or carbonate buffers were used to achieve desired pH of the experimental solutions, ranging from pH of 5 to 9. The vials were sealed before taken out of the glovebox. In a fume hood, the incubations were then initiated by addition of DMHg through the septa using a Hamilton syringe. Different DMHg:reactant molar ratios were achieved in experiments with dissolved sulfide and FeS(s)m by altering the concentrations of dissolved sulfide or FeS(s)m while keeping the DMHg concentration constant between treatments. To minimize headspace, sample volumes were set to 39 mL. Detailed information on the experiments, including, for example, DMHg and sulfide amounts used is listed in SI Table S1.

Samples were incubated at room temperature (ca. 23 °C), as well as in temperature-controlled water baths (40 and 60 °C). Upon subsampling, samples were shaken, and 50 μL aliquots were taken with a Hamilton syringe and injected into precombusted 40 mL amber glass vials containing milli-Q water buffered to a pH of 4.7 (225 μL 2 M acetate buffer), giving a total volume of 30 mL. To avoid loss of DMHg during subsampling, syringe injections were always made beneath the water surface. The reproducibility of subsampling with a Hamilton syringe was tested to be within a few percent (SI Figure S1). After addition of the sample aliquot, 30 μL of a derivatization solution was added (1% sodium tetraethylborate; NaTEB) giving a final volume of 30 mL. No measurable breakdown of DMHg to MMHg with subsequent MMHg ethylation was observed in tests with DMHg in milli-Q, DMHg:S(aq) ratios of 3.6 × 10–6 or higher, or for DMHg: FeS(s)m molar ratio of 0.01. The subsample vial was then capped and shaken and analyzed as described below. Although the possibility of reusing septas has been previously suggested,[12] only unused septa were used in our experiments since lower recovery was observed in separate tests when using already pierced septa.

Experiments were mainly performed with treatments reproduced in triplicates with discrete sampling occasions. Where treatments were not replicated, sampling frequency was increased. While most experiments were sampled over a day, some incubations were stored at room temperature for months. SI Table S1 gives a complete overview of experiments performed and the parameters investigated. Potential degradation products in the experiments include elemental Hg (Hg0), MMHg and inorganic divalent Hg (HgII). These were directly or indirectly quantified as described below. To reduce the handling of DMHg, most of the experiments were conducted as short-term incubation experiment at temperatures up to 60 °C.

Chemical Analysis

Concentrations of DMHg were quantified using a Tekran 2700 methylmercury analyzer (cold vapor–atomic fluorescence spectroscopy (CV-AFS)) and external calibration (calibration range: 0.0–20 pM MMHg as Hg, stock solution: 1000 ppm MMHg standard (Alfa Aesar)). R2 for all calibrations were >0.97. In the methylmercury analyzer, all Hg forms are pyrolyzed and analyzed as Hg0. This allows us to use the calibration curve of MMHg also for quantification of the other Hg-species, assuming comparable recovery of species in the purging step of calibration standards. Comparable recovery can be assumed for Hg0 and DMHg from the experimental solutions, and MMHg from the calibration standards, as both Hg0 and DMHg are gaseous Hg species and not influenced by the efficiency of the derivatization step. Tests conducted by the manufacturer (pers. comm. Diana Babi, Tekran) have also confirmed close to identical calibration curves for MMHg and DMHg when applying a similar setup as used here (sensitivity differed less than 2%). Comparison of the DMHg concentrations detected at the start of the incubations (t0) shows good recoveries of DMHg within the concentration ranges of S(aq) and FeS(s)m tested (SI Figure S2). The limit of detection (LOD), when applying external calibration, was calculated for each species from analytical reagent blanks from all experiments (n = 93 (Hg0), 80 (DMHg), 88 (MMHg), 86 (HgII)) as average blank concentration +3·standard deviation (SD) (SI Table S2).

Higher concentrations of dissolved sulfide is known to lower the recovery of MMHg when measured through direct ethylation.[13,14] These results were corroborated in a test experiment where we observed a 75% loss in MMHg recovery at 2.5 uM Na2S concentrations (SI Figure S3). Concentrations of MMHg were successfully quantified in a subset of experiments with dissolved sulfide (SI Table S1) by applying a masking technique published by Yang et al. (2009)[14] where CuSO4 and sodium oxalate solutions was added to the subsamples (Cu2+ is used to complex the dissolved sulfide and thus minimize sulfide interference, oxalate is added to complex excess Cu2+). In preliminary tests of this method, highest MMHg recoveries were obtained when CuSO4 concentrations just exceeded the sulfide concentrations, with sodium oxalate in excess to CuSO4, and with the addition of 75 μL 1% NaTEB. However, consistently satisfactory recovery was not achieved in all instances. An isotopically enriched MM200Hg isotope spike tracer was thus added to all replicates for a subset of experiments, and MMHg concentrations were quantified by coupling the methylmercury analyzer to an inductively coupled plasma-mass spectrometer (ICP-MS, Thermo–Fisher X-series 2). The LOD was calculated for individual samples as described elsewhere[15] (SI Table S2). Concentrations of HgII were estimated in the same samples by comparing the peak intensity of 202HgII with added MM200Hg internal standard. We did not test the recovery of HgII in our experiments and thus presented values are to be viewed as estimates. Since the affinity for sulfide is higher for HgII than for MMHg, potential differences in the efficiency of the masking and derivatization step likely favors MMHg. We thus assume our estimates of HgII to be close to, or underestimate, the true concentrations. HgII was only considered for triplicate samples when detected in at least two of the triplicates. For experiments containing FeS(s)m, MMHg and HgII was analyzed directly through CV-AFS. Good recoveries of MMHg were also confirmed in separate experiment (SI Figure S4). Significant production of Hg0 was excluded as observed Hg0 peaks detected by CV-AFS did not exceed background values (with the exception from a few samples where concentrations were just above the calculated LOD).

Concentrations of S(aq) were monitored throughout the incubation in select experiments (SI Table S1). This was done by taking out a subset of a sample through the septa with a Hamilton syringe and transferring it to a test tube. Syringe injections were always made beneath the water surface. For high sulfide concentrations, sample subsets were diluted using carbonate buffer (pH 9) before transferring a fraction to the test tubes. An amount of 50 μL (4.75 mM) of a zinc acetate preservative solution was added to test tubes with 3 mL sample within 15 min of subsampling. Test tubes were left open for one or two nights to air out DMHg. Sulfide concentrations were analyzed as described in SI.

Data Analysis

To compare the rate of DMHg loss between the experiments, a set of demethylation rates were calculated based on changes in concentrations observed between time of 0 h (t0) and 3–6 or 140–480 h in short- and long-term experiments, respectively. First, a DMHg loss rate constant, kctr(d–1), was calculated for controls (kept at corresponding sample temperature and pH), using eq .[16]kctr were calculated for each treatment group separately. The kctr gives a total rate of the DMHg loss through processes other than the reaction of DMHg with sulfide, for example, evasion or adsorption of DMHg to the walls of the vials, and is assumed to be the same in controls and in the samples at a given temperature and pH (SI Table S3). A second DMHg loss rate constant, kobs(d–1), was then calculated for samples containing reduced sulfur as dissolved sulfide, FeS(s)m or CdS(s), using eq . The kobs is assumed to represent the combined DMHg loss rate for DMHg demethylation via the reaction of DMHg and sulfide, kDMHg↓, and the total rate for DMHg loss through other processes, kctr. Based on this assumption, kDMHg↓(d–1) was calculated as the difference between kobs and kctr (d–1, eq ). In addition, a rate of MMHg formation, kMMHg↑ (d–1) was calculated based on eq . It should be noted that both kDMHg↓ and kMMHg↑ are demethylation rate constants of DMHg, the difference being that one is calculated from the loss of reactant (DMHg) and the other an increase in product (MMHg), respectively. For all the calculations above, remethylation of degraded DMHg was assumed not to alter the concentrations of DMHg or MMHg during the first hours of the incubation.[9,17] Also, for kMMHg↑(d–1), formation of MMHg by other processes than the reaction of DMHg with sulfide was assumed neglectable as no MMHg formation was observed in control samples.Only kDMHg↓’s meeting the two quantification limit criteria (kobs > average kctr+2·SD(kctr) and kDMHg↓> 0.4·kobs) were considered as significant. Assuming normal distribution of kctr, these criteria are meant to ensure that kobs ≠ kctr(p < 0.05), and kDMHg↓ constitutes what is considered to be a significant fraction of kobs. For experiments with samples prepared without duplication, varying time points of measurement were used to calculate the average and SD of kctr. The kMMHg↑ was calculated in experiments where the concentrations of MMHg at t = 1–6 h were statistically higher than the concentrations of MMHg at the t = 0 (one-sided t tests, α = 0.05).

Results and Discussion

Degradation of DMHg by Dissolved Sulfide

The stability of DMHg in the presence of dissolved sulfide was tested in short (<24–48 h) and long-term (<62 days) incubation experiments (SI Table S1). Decreasing concentrations of DMHg during the incubation were observed in both samples containing sulfide as well as in controls not containing sulfide. The observed loss in control samples (about 1% h–1) is likely due to diffusion of DMHg through the septa, degradation of DMHg by other pathways than the reaction of sulfide, and/or adsorption of DMHg to vial surfaces. The similar loss of DMHg observed in controls at pH 5 and 9 suggests the pH in our experiments to be high enough to avoid significant DMHg loss due to acidolysis[18] (SI Figures S5 and S6). Under a wide range of experimental conditions (discussed further below), a greater loss of DMHg was observed in samples containing dissolved sulfide than in controls. In experiments where a masking agent was used to recover ionic forms of Hg in the presence of sulfide, increasing concentrations of MMHg with time further supported demethylation of DMHg to MMHg in the presence of dissolved sulfide. Demethylation of DMHg by dissolved sulfide was first hypothesized in theoretical work by Ni et al.[11] Here, we provide the first experimental support for this reaction.

Experimental conditions of the incubations with dissolved sulfide are listed in SI Table S1 and cover pH of 5 and 9, temperatures up to 60 °C and DMHg:S(aq) molar ratios between 6.28·10–8 to 1.16·10–4. To compare data between experiments, we calculated the demethylation rate of DMHg when reacting with sulfide (kDMHg↓) by subtracting the average rate of DMHg loss in control samples (kctr) from the measured rate of DMHg loss in samples containing sulfide (kobs) as outlined in eqs , 2, and 3.

We found the demethylation rates to be higher at pH 9 than at pH 5 (kDMHg↓ = 2.1 d–1 at pH 9 vs 0.70 d–1 at pH 5, both at 60 °C, SI Figure S6). As this means a shift from dominantly H2S at pH 5 to HS– at pH 9, these results imply that the speciation of sulfide plays a role in the demethylation rates of DMHg. In the different experiments, DMHg:S(aq) molar ratios ranged from ∼10–3 to ∼10–9 (at pH 9). The highest demethylation rates were observed at DMHg:S(aq) molar ratios around ∼10–5. This is 50 000 times higher concentrations of sulfide than that corresponding to the stoichiometric ratio for the reaction mechanism proposed by Ni et al.[11] In one subset of experiments, concentrations of sulfide were monitored throughout the incubation. For the highest DMHg:S(aq) ratio tested ([S(aq)] = 22.4 μM), we observed loss of sulfide over the incubation, likely due to oxidation of the sulfide by trace concentrations of oxygen and/or evaporation of H2S(g). Despite this, we argue this to not be the main reason behind the lower kDMHg↓ at this DMHg:S(aq) ratio (Figure ). This argument is based on lower kDMHg↓ values calculated from individual samples where we still have detectable concentrations of sulfide lower than the theoretical stochiometric DMHg:S(aq) ratio of 0.5 (SI Figure S7). We propose that the increase in demethylation rate as DMHg:S(aq) ratio approaches 10–5 is attributable to an increase in the concentration of a sulfide-coordinated DMHg species that undergoes decomposition. As the concentrations of sulfide increased beyond DMHg:S(aq) molar ratios of ∼10–5, lower rates of DMHg demethylation were observed (Figure ). The reason behind this decline is unclear. As previously mentioned, MMHg is known to react with dissolved sulfide to form (CH3Hg)2S, which decomposes to form DMHg and HgS.[19] We therefore investigated the possibility that the apparent increase in DMHg stability is the effect of a fast transformation reaction converting (CH3Hg)2S back to DMHg. Using the activation energy for DMHg formation through this pathway published by Kanzler et al. (2018),[17] we calculated a first-order rate constant of 9.25·10–3 d–1 at 60 °C for the production of DMHg from (CH3Hg)2S. Coupled with the fact that only a small fraction of product MMHg will exist as (CH3Hg)2S at pH 9 at mM sulfide concentrations, DMHg regeneration is an unlikely explanation for the lower demethylation rates observed at low DMHg:S(aq) molar ratios. Thus, the cause for lower kDMHg↓ observed at low DMHg:S(aq) molar ratios remains unclear.

Figure 1

Demethylation rate constants kDMHg↓ (calculated from the loss of DMHg) as a function of DMHg:S(aq) molar ratio (experiments conducted with DMHg and dissolved sulfide) at pH 9, 60 °C. Error bars represent standard deviation of triplicate measurements.

It is clear from our experiments that while dissolved sulfide plays a role in the observed degradation of DMHg, the reaction mechanism may be more complicated than originally proposed. Ni et al. suggested the demethylation reaction to be initiated by the binding of HS– to DMHg, followed by proton transfer to a methyl group to release CH4 and MMHg.[11] Their modeling, for reactions in the gas phase, based their models on H2S as a H+ donor. In aqueous media, H2O and other H+ donors are also available. Inconsistent with a first- or second-order (if H2S is H+ donor) dependence of DMHg degradation on sulfide concentration, we observed increasing rates of DMHg demethylation at sulfide concentrations up to about 0.2 mM, and then decreasing rates at sulfide concentrations higher than this level. Increasingly, a role for Hg–Hg interactions has been recognized in Hg transformations,[17] including methyl transfer reactions. The involvement of a dimeric DMHg species[20] with significant Hg–Hg interaction, favored at low HS– concentrations, in DMHg protonolysis facilitated by sulfide, could explain the observed maximum demethylation at intermediate DMHg:S(aq) molar ratios. The mechanism requires further experimental and computational investigation, however.

Formation of MMHg was consistently observed in sulfide-containing samples where kDMHg↓ exceeded 0.67 d–1, but not at lower sulfide concentrations or in controls. Where MMHg was detected during the incubation, a second DMHg demethylation rate constant (kMMHg↑), based on the formation of MMHg, was calculated. A linear relationship was observed between kDMHg↓ and kMMHg↑ (R2 = 0.95), with kMMHg↓ = 0.62 kDMHg↑, suggesting MMHg to be the first and primary breakdown product from the decomposition of DMHg in the presence of dissolved sulfide (SI Figure S8). At the lowest sulfide concentrations (0.01–0.23 mM), formation of bismethyl–Hg[17] may partly account for the fraction of Hg not accounted for at these S(aq) concentrations (Figure b,c). Estimated concentrations of Hg0 (semiquantitative analysis of the Hg0 peak from the methylmercury analyzer) rarely rose above the detection limit, and Hg0 as a significant product was thus excluded. In samples for which kDMHg↓ and kMMHg↑ were high, MMHg concentrations eventually decreased at the same time as the estimated concentrations of HgII increased (Figure ). While our experimental data is limited, this supports a stepwise demethylation of Hg from DMHg to MMHg which then demethylates to HgII. As MMHg was not observed in controls, we cannot evaluate if sulfide played a role in the demethylation of MMHg or not.

Figure 2

Concentrations of DMHg (circles), MMHg (squares), and HgII (triangles) as a function of time in incubations without sulfide (a) and with dissolved sulfide at increasing DMHg:S(aq) molar ratios of (b) 1.18·10–4, (c) 1.18·10–5, (d) 1.19·10–6, (e) 2.38·10–7, and (f) 5.94·10–8. All incubations were performed at 60 °C at pH 9, but for d–f, temperature reverted to ∼23 °C after ∼500 min. Shaded areas represent measured sum of detected DMHg, MMHg, and HgII. Sulfide recovery (yellow triangles) is reported as % of expected concentrations. Error bars represent standard deviation of triplicate measurements.

Degradation of DMHg on Sulfide Minerals

Demethylation of DMHg on sulfide mineral surfaces was tested using FeS(s)m, a metastable iron sulfide phase, and CdS(s). For FeS(s)m, experiments were conducted under various experimental conditions covering pH of 5, 7.5, and 9, temperatures up to 60 °C and DMHg:FeS(s)m molar ratios of 9.65·10–7–9.65·10–3 (SI Table S1).

At room temperature, we observed significant DMHg demethylation in samples with DMHg:FeS(s)m molar ratios of 9.65·10–7 (SI Figure S9). Observed demethylation of DMHg in the presence of FeS(s)m can either be explained by reactions of DMHg on the surface of FeS(s)m or with dissolved ligands present in the FeS(s)m slurries. Based on the solubility constant of FeS(s)m (Ksp = 10–3.5 for FeS(s)m + H+ → Fe2+ + HS–)[21] we would not expect the DMHg:S(aq) ratio in FeS(s)m slurries to be lower than 5·10–3. These dissolved sulfide concentrations are too low to explain observed degradation of DMHg (DMHg:S(aq) > 5·10–3, Figure ). In addition, no demethylation was observed at a DMHg:FeS(s)m molar ratio of 9.7·10–5 where similar concentrations of dissolved sulfide would be expected. The potential role of dissolved ligands that may be present in the FeS(s)m slurries was further examined by incubating DMHg with Mohr’s salt ((NH4)2Fe(SO4)2, used in the synthesis of FeS(s)m). These tests confirmed DMHg to be stable at DMHg:((NH4)2Fe(SO4)2) molar ratios far exceeding what would be expected in any of the experiments conducted with FeS(s)m (tested molar ratio was 10–6). Potential dissolved residuals from the Mohr’s salt, or dissolved sulfide present, thus seems unlikely to explain observed loss of DMHg in FeS(s)m experiments. In experiments conducted with pH of 9 and where kDMHg↓ exceeded 0.70 d–1, we also observed production of MMHg, and a linear relationship between kDMHg↓ and kMMHg↑ (R2 = 0.98, kMMHg↑ = 0.52 kDMHg↓, SI Figure S8), suggesting MMHg to be the main breakdown product. Together, these experiments demonstrate that DMHg can be demethylated to MMHg on FeS(s)m surfaces.

To further understand why no demethylation was observed at DMHg:FeS(s)m molar ratios of 9.7·10–5 and 9.7·10–3, solubility of FeS(s)m and distribution of DMHg between the dissolved and solid phase was considered. Assuming a solubility product (Ksp) of 10−3.5 for FeS(s)m, all FeS(s)m is predicted to dissolve as the slurry was diluted to achieve at the highest DMHg:FeS(s)m ratio. For the DMHg:FeSm molar ratios of 9.7·10–5, dissolution of the FeS(s)m was however less than 2% and unlikely to explain the lack of DMHg demethylation. Preliminary adsorption experiments conducted suggests solid water partition coefficient (KD) of DMHg in the FeS(s)m slurries to be somewhere in the range of 102.5 −104 L kg–1. Although additional work is required to, with certainty, determine the KD of DMHg in FeS(s)m slurries, these preliminary data allows us to estimate distribution ranges of DMHg between the dissolved and solid phase. At the DMHg:FeS(s)m molar ratios of 7·10–3 and 7·10–5, less than 2% and between 6 to 66% of the DMHg, respectively, was estimated to be adsorbed on FeS(s)m. As the demethylation was concluded to be a surface mediated reaction, the nonsignificant kDMHg↓ at DMHg:FeS(s)m ratio of 7·10–3 is thus likely explained by the small fraction of adsorbed DMHg.

We hypothesize that demethylation of DMHg is mainly mediated by monocoordinated sulfide atoms on the FeS(s)m surface[22] (≡Fe1S1). At pH of 5 and 9, where these sites are fully protonated (≡Fe1S1H2+) and deprotonated (≡Fe1S1–), respectively, we observe kDMHg↓ of 0.65 and 2.6 d–1, respectively (at incubation temperature of 60 °C, Figure ). At the point of zero charge (pH 7.5) suggested by Wolters et al.,[22] no demethylation of DMHg was observed. In contrast, Jonsson. et al. demonstrated the decomposition of MMHg on FeS(s)m surface (to DMHg and HgS(s)) to be independent of protonation state of the ≡Fe1S1 sites.[9] Methylation of the Hg atom is however known to lower the affinity of Hg to particulate surfaces, as illustrated by the lower KD typically observed for MMHg (102.5 −106L kg1–) than HgII (103–106.5L kg1–).[23] It is reasonable to assume that the affinity of Hg toward mineral surfaces (including ≡Fe1S1 sites) is further weakened with the second methyl group of DMHg, as the electrophilicity of the Hg atom should be lowered with each methyl group added. Our preliminary data also suggested lower KD of DMHg (102.5 −104L kg1–), than typically observed for MMHg. Protonation of the active sites may thus have lowered the adsorption capacity of the FeS(s)m. For both dissolved sulfide, and FeS(s)m, negatively charged complexing or binding sites appear to promote demethylation of DMHg, in comparison to neutrally charged sulfide sites (H2S0, ≡FeSH0). This is in line with theoretical work suggesting the reaction to be initiated by a nucleophilic attack of a deprotonated reduced sulfur atom.[11] The mechanism for degradation of DMHg on ≡FeSH2+ surfaces is less obvious. As discussed above, one alternative pathway for the demethylation of DMHg after the binding of the reduced sulfur to the Hg atom, is the formation of Hg–Hg bonds and finally protonation of one of the methyl groups. A plausible explanation can thus be that the acidic binding sites at low pH facilitates this last step. Formation of MMHg was observed in experiments with deprotonated ≡Fe1S1 sites (pH > 7.5). In contrast, no MMHg was observed at pH of 5 (≡FeSH2+), although significant loss of DMHg was observed at this pH at both 40 and 60 degrees (Figure , SI Figure S10). Based on the available data, we are not able to tell whether the lack of MMHg is because of it not being produced, or that it is forming an insoluble species that cannot be quantified through direct ethylation.[17]

Figure 3

Concentrations of DMHg (circles) and MMHg (squares) as a function of time for pH 5 (red), 7.5 (green) and 9 (blue) in experiments with FeS(m) at 60 °C at DMHg:active site ratio of 7.9·10−5. Rates of DMHg loss (kDMHg↓) and MMHg formation (kMMHg↑) are marked out where significant. Projected loss of DMHg in controls based on calculated control loss rates (kctr) is represented by dashed lines (pH 9 = blue, pH 5 = red), with faded areas representing uncertainty range based on kctr standard deviation. Error bars represent standard deviation of triplicate measurements.

DMHg degradation experiments were also conducted with CdS(s). Assuming comparable number of binding sites on the CdS(s) per m2,[9] demethylation of DMHg was mediated by FeS(s)m, but not CdS(s), at comparable molar ratios of DMHg and active sulfide sites on the mineral surfaces (SI Figure S11). One possible explanation for the lack of demethylation on the CdS(s) surface could be other properties of the active sites in comparison to the active sites on FeS(s)m surface. For example, the basicity of surface sulfhydryl groups is lower for CdS(s) than for FeS(s)m, due to the stronger metal-sulfide bond of CdS(s) (log Ksp = −14.36)[35] compared to FeS(s)m ((log Ksp = −3.5).[21] Interestingly, we also observe (as discussed above) pH to control DMHg demethylation when reacted with dissolved sulfide or FeS(s)m, which further supports the chemical property of the reactant to be of importance. In contrast, similar reactivity of MMHg on FeS(s)m and CdS(s) surfaces, respectively, and under different pH regimes has been previously shown,[9] presumably due to the higher binding affinity of MMHg to such sites.

To compare kDMHg↓ when the demethylation of DMHg was mediated by dissolved sulfide and FeS(s)m, respectively, DMHg:active sites (monocoordinated sulfides, ≡Fe1S1–) molar ratios were calculated from the determined surface area of synthesized FeS(s)m, and assuming 2 ≡Fe1S1– sites per nm2.[22] At the DMHg: ≡ Fe1S1– molar ratio of 7.9·10–5, kDMHg↓, ranged from 2.1 ± 0.1 to 2.6 ± 0.2 d–1 at 60 °C and from 0.3 ± 0.1 to 0.7 ± 0.1 d–1 at 40 °C (Table ). At comparable DMHg:active sites ratio for experiments with dissolved sulfide (DMHg:S(aq) of 1.2·10–4), kDMHg↓ were <0.34 and 0.42 ± 0.5 d–1 at 60 °C and <0.21 at 40 °C. It should be noted that as our experiments were not conducted at FeS(s)m concentrations where the majority of the DMHg can be assumed to bind to the surface, the values reported here likely do not reflect the maximum kDMHg↓ by FeS(s)m at optimal DMHg:FeS(s)m ratio. In long-term experiments at room temperature, comparable kDMHg↓ was observed at DMHg:S(aq) molar ratios 5000 times higher than the DMHg:≡Fe1S1– tested. Both short- and long-term experiments thus support higher kDMHg↓ when the reaction is mediated on the surface of FeS(s)m in comparison to when mediated by dissolved sulfide. This is aligned with the reactivity of MMHg with the same two reactants and could be related to the electron rich properties of the ≡Fe1S1–. While DMHg formation has previously been reported for incubations with DMHg and FeS(s)m, direct comparison of reaction rates is not possible due to difference in tested Hg:FeS(s)m ratios.[9] It is, however, worth noting that optimum ratios (between the reactants and sulfide) for the degradation of DMHg can be assumed to be more commonly occurring in natural systems than the corresponding ratio for sulfide mediated DMHg formation.

Table 1

DMHg Demethylation Rate Constant, Determined from the Loss of DMHg between t0 (Time 0 h) and t(h), kDMHg↓ (d–1), at Temperatures of 20, 40, and 60 °C in the Presence of Dissolved Sulfide (S(aq)) or FeS(s)ma

			kDMHg↓(d–1)
sulfide	t(h)	DMHg:active sites	20 °C	40 °C	60 °C
Short-Term Incubations
FeS(s)m	3–6	7.9·10–5	<0.36	0.61 ± 0.2	2.5 ± 0.9
			<0.36	0.70 ± 0.1	2.1 ± 0.1
			<0.36	0.30 ± 0.15	2.6 ± 0.2
S(aq)	3–6	1.2·10–4	0.40 ± 0.1	<0.21	<0.34
		1.2·10–5–1.4·10–5			4.3 ± 0.02
					4.1 ± 0.4
		1.0·10–6–5.8·10–8	<0.36	<0.21	<0.34 to 2.2 ± 0.1
Long-Term
FeS(s)m	140	7.1·10–5	0.052 ± 0.021
	480		0.042 ± 0.013
S(aq)	170	1.1·10–6	0.023 ± 0.024
	360		<0.0058
	170	2.2·10–8	0.046 ± 0.013
	360		0.022 ± 0.0041

The table includes short-term experiments conducted at the pH of 9 and long-term incubation experiments conducted at a pH of 5 (S(aq)) and unbuffered MQ water (FeS(s)m). Error represents standard deviation of triplicate measurements.

Environmental Implications

Although the formation of DMHg in natural systems was shown more than 50 years ago,[24] the role of DMHg in the biogeochemical cycle of Hg remains largely unknown. This is partly due to that the research field traditionally has been focused on freshwater systems, where concentrations of DMHg typically are assumed to be low. It should, however, be noted that DMHg has been observed in, for example, rice paddies,[7] and that studies looking for DMHg in freshwater systems still remain limited. In addition, the research community has been hesitant to work with DMHg due to its high toxicity. However, DMHg composes a large fraction of the methylated pool in marine waters and marine harvested food is the major exposure pathway of Hg for many populations worldwide. Closing the current knowledge gap of the role of DMHg by identifying its potential formation and degradation pathways in natural environments is thus needed.

Here, we provide the first experimental support for DMHg demethylation by reduced sulfur. Our findings highlight the need to further explore both biotic and abiotic sulfide assisted demethylation pathways of DMHg. For example, thiols are already known to have a key function in the degradation of MMHg by the MerB pathway,[25] but bacterial demethylation of DMHg remains to be tested. Intracellular demethylation of DMHg could also be a potential accumulation pathway of Hg into aquatic food webs as DMHg readily diffuses through the cell membrane[3] and, if degraded, could biomagnify as MMHg. Our results further emphasize the potential importance of abiotic transformation pathways. We show demethylation of DMHg on the surface of FeS(s)m, a metastable iron sulfide mineral ubiquitous in sediments. In such systems, the amounts of FeS(s)m present is unlikely to be limiting for reaction, as significantly lower DMHg:FeS(s)m molar ratios would be expected than those evaluated in our experiments. While we demonstrate that the surface reactivity of FeS(s)m controls the reaction rates, further work is needed to determine DMHg degradation rates in the presence of dissolved organic matter and other competing ligands. Demethylation of DMHg by dissolved sulfide may also be of importance in marine waters, as we observed demethylation of DMHg under environmentally relevant DMHg:S(aq) molar ratios. For example, assuming an average marine DMHg concentration of 0.1 pM,[26] and dissolved sulfide concentrations up to 25 μM in anoxic zones of marine snow[27] and up to 6000 μM in anoxic bottom waters[28] (i.e., DMHg:S(aq) molar ratio in the 10–9–10–11 range), such environments could be important sites for sulfide mediated DMHg degradation. Interestingly, surprisingly low concentrations of DMHg have been observed in parts of the Black Sea, waters also rich in dissolved sulfide.[29,30] Very few studies have, however, reported degradation rates of DMHg in natural environments. Early work based on existing concentration gradients suggested DMHg demethylation rates of 0.0002 d–1 for deep Pacific waters.[31] Incubation of marine surface waters further suggested demethylation rates of 0.1–0.2 d–1 under dark conditions.[32] It should, however, be noted that bottle effects were suggested to explain some of the lost DMHg in these incubations. Although these early rates should be considered as rough estimates, it is worth to note that observed demethylation rates in our laboratory experiments falls within the range of rates reported for marine systems. Reported MMHg to DMHg methylation rates for Arctic and Mediterranean water range up to 0.0016 and 0.015 d–1, respectively.[33,34] This is lower than the upper range of rates we report here from our experimental system. Our reported range of demethylation rates (up to 0.05 d–1 at room temperature in MQ water matrix), together with reported methylation rates of HgII to MMHg (∼0.01 d–1)[33] in marine waters, and known distribution of Hg forms in marine waters (∼80% HgII, ∼10% MMHg and ∼10%DMHg), suggests that sulfide mediated degradation of DMHg could act as an important source of MMHg. Finally, our work stresses the need to further elucidate the role or sulfide mediated degradation of DMHg in the biogeochemical cycle of Hg, in marine as well as freshwater systems.