Competition among thiols and inorganic sulfides and polysulfides for Hg and MeHg in wetland soils and sediments under suboxic conditions: Illumination of controversies and implications for MeHg net production



[1] Current research focus in mercury biogeochemistry is on the net production and accumulation of methyl mercury (MeHg) in organisms. The activity of iron- and sulfate-reducing bacteria (FeRB and SRB) has been identified as important for MeHg production. There are indications of a passive uptake of neutral Hg-sulfides by SRB, as well as of a facilitated bacterial uptake of Hg complexed by small organic molecules. In order to understand these processes, the chemical speciation of Hg and MeHg, and most important, the competition among organic thiols and inorganic sulfides and polysulfides, needs to be clarified under suboxic conditions (nM to low μM range of total sulfide concentrations) in wetland soils and sediments. In this paper the chemical speciation of Hg and MeHg is modeled at pH 4.0 and 7.0 in a conceptual wetland soil/sediment with typical concentrations of thiols, sulfides, Hg, and MeHg. Effects of precipitated HgS(s), the formation of Hg-polysulfides, and the size of the controversial stability constant for the formation of HOHgSH0 (aq) are emphasized. The outcome of the modeling is discussed in light of chosen stability constants for Hg complexes with thiols, sulfides, and polysulfides. It is concluded that organic thiols are competitive with inorganic sulfides in the approximate total sulfide concentration range 0–1 μm. It is also concluded that increases in absolute aqueous concentrations of MeHg, or the molar ratio of dissolved MeHg/Hg, are not appropriate as indirect measures of MeHg net production, unless changes and differences in solubility of MeHg and Hg are corrected for.

1. Introduction

[2] Mercury biogeochemistry is very complex, involving bacteria active in methylation, demetylation and reduction processes, as well as interactions with carbon, sulfur and iron geochemistry. Of major concern are processes resulting in a net formation of methyl mercury (MeHg). The concentration of MeHg in soils and sediments is considered to be the net effect of methylation, demethylation and input/output processes. In surface sediments methylation has been shown to dominate over the other two processes [e.g., Hintelmann et al., 2000; Drott et al., 2007, 2008]. Sulfate-reducing bacteria (SRB) have conclusively been identified as actively contributing to MeHg formation [e.g., Compeau and Bartha, 1985; King et al., 2001], and more recently also iron-reducing bacteria (FeRB) have been shown to methylate inorganic Hg [Fleming et al., 2006; Kerin et al., 2006]. Studies in different environments indicate that net MeHg production reach an optimum under suboxic and low anoxic conditions, with sulfides concentrations below ∼10–50 μM [e.g., Gilmour et al., 1998]. The methylating process is believed to be taken place within bacterial cells. Because of this, the methylating activity of bacteria is controlled by the availability of electron-donors, electron-acceptors and bioavailable forms of inorganic Hg. Laboratory experiments have indicated that forms of Hg-sulfides that partition into octanol from water is taken up by SRB. These forms have been interpreted as being dissolved, neutral Hg-sulfides [Benoit et al., 1999, 2001a]. Experiments with Hg in chloride and DOC solutions show that the neutral form HgCl20 is preferentially taken up by bacteria over negatively charged HgCl3 and Hg-DOC complexes [Barkay et al., 1997]. Similar experiments with other types of bacteria indicate that also Hg complexed by small organic molecules like amino acids may be taken up [Golding et al., 2002].

[3] Obviously the chemical composition of bioavailable forms of Hg, for methylating bacteria, and of MeHg, for demethylation reactions, is of major importance for the production and subsequent bioaccumulation of MeHg in food webs. The bioavailable forms of Hg and MeHg are determined by the competition among dissolved thiols, inorganic monosulfides, bisulfides and polysulfides, as well as their competition with organic and inorganic particle surfaces and solid phases. Crucial for the outcome of chemical speciation calculations is that properly determined stability constants exist and are correctly used. The problem today is that there are several controversial stability constants, and sometimes an inconsequent use of these. This may result in misconceptions and misinterpretations about the behavior of Hg and MeHg in soils, sediments and waters.

[4] In this paper I will briefly review the available data on concentrations of relevant organic and inorganic sulfur ligands in wetland soils and sediments and their stability constants with Hg and MeHg. By a modeling approach I will demonstrate the competitiveness of thiols, sulfides and polysulfides for Hg (used both as a general term for inorganic Hg and at times for the Hg2+ ion) and MeHg (used as a general term for MeHg and at times for the MeHg+ ion). The outcome of the modeling is used to discuss and clarify three important issues or controversies which are central when it comes to our understanding of Hg and MeHg biogeochemistry:

[5] 1. Stability constants used for Hg–natural organic matter complexes are often unrealistically small, which lead to incorrect conclusions about the competitiveness of thiols for Hg as compared to inorganic bisulfides.

[6] 2. The stability constant for the proposed species HOHgHS0 (aq) is highly controversial and depending on the size of the constant used, the concentration of dissolved, neutral Hg-sulfides will vary tremendously. Also the competitiveness of thiols is highly affected.

[7] 3. There is a widespread perception that absolute or relative increases in pore water concentrations of MeHg can be considered equal to a net MeHg production. This is, however, only true after proper correction for differences in dissolution of MeHg and Hg caused by complexation and adsorption to organic and inorganic sulfur compounds.

2. Three Controversies in Mercury Biogeochemistry

2.1. Stability Constants for Hg–Natural Organic Matter Thiol Complexes

[8] There is spectroscopic evidence for an association of Hg with two thiols [Skyllberg et al., 2006] and an association of MeHg with one thiol [Qian et al., 2002] in organic soils and humic streams. Some of these thiols may be free, low molecular weight (LMW) molecules like cysteine or glutathione. Concentrations of LMW thiols in estuarine, limnic and marine environments can reach 10–100 nM [Zhang et al., 2004; Han et al., 2005; Hu et al., 2006], in great excess of concentrations of Hg and MeHg. Spectroscopic studies of penicillamine reveal that Hg associates with two thiols at low pH (pH = 4), and with two, three or even four thiols at alkaline (pH = 11) conditions [Leung et al., 2007]. The majority of the thiols, however, is associated as functional groups to high molecular weight (HMW) dissolved or particulate natural organic matter (NOM). Even if the bonding of Hg and MeHg to NOM from estuarine and marine environments has not been explicitly studied by spectroscopic methods, there is no reason to believe that the bonding of Hg and MeHg to LMW and HMW thiols should be principally different than in organic soils and humic streams.

[9] In Table 1 stability constants for the formation of two-coordinated Hg–LMW thiol complexes are listed. It should be noted that there are many obviously incorrect constants reported for Hg–thiol (as well as for several other Hg–complexes) in the literature, owing to methodological problems [cf. Casas and Jones, 1980]. Both cysteine and penicillamine have one carboxyl, one amino and one thiol group. The pKa of the carboxyl is 1.9–2.0. The thiol group normally is assigned an intermediate acid strength (pKa2 = 8.5 in cysteine and 7.9 in penicillamine) and the amino group is supposed to be the weakest acid (pKa3 = 10.5 in cysteine and 10.4 in penicillamine). In reality, however, the latter two pKa values are macroscopic constants representing a combined acidity of the thiol and the amino group. This means that if the fully protonated cysteine or penicillamine ligand (L) is designated H3L, the two-coordinated Hg–thiol complex Hg(SR)2 could be represented by either Hg(HL)2 or HgL22−. The formation constant of each of these complexes differs substantially (theoretically by 2 times pKa3 − pKa2), but depending on the assumptions made, both may be relevant to describe the binding of Hg to two thiols. In mercaptoacetic acid, which does not have any amino group, the thiol group has a pKa value of 10.0. If the thiol groups in cysteine and penicillamine are assigned a pKa of 10.0 (subtracting the log βHgL2 value by 2 × (pKa3 − 10.0); corresponding to 0.8 log-units for penicillamine and 1.0 log-units for cysteine), an “average” stability constant for the formation of Hg(SR)2 complexes in LMW thiols with a pKa of 10.0 may be calculated. Such a calculation yields an average of 42.5 for the log βHgL2 constant of the three LMW thiols in Table 1. Given the variability in literature data, an average constant of log K = 42 ± 2 for the formation of two-coordinate Hg(SR)2 complexes with LMW thiols may be used as a reference for evaluating constants reported for the association of Hg to NOM (reaction (1a) and Table 2). This average constant is in good agreement with log KHg(SR)2 = 41.6 (pKa = 9.34), calculated from Hg titration data of a Black Sea water in which organic thiols and inorganic sulfides were separated [Dyrssén and Wedborg, 1986].

Table 1. Selected Stability Constants Reported for the Complexation of Hg to LMW Thiolsa
 log KHgLlog βHg(HL)2log βHgL2I (M)Reference
  • a

    I, ionic strength; n.a., not available.

pKa = 2.0, 8.5,10.5ý and Kratzer [1988]
 37.838.344.0n.a.Basinger et al. [1981]
pKa = 1.9, 7.9, 10.4 38.2 0.5Koszegi-Szalai and Paál [1999]
 38.3 44.40.1Casas and Jones [1980]
 37.837.844.5n.a.Basinger et al. [1981]
Mercaptoacetic acid34.5 40.5n.a.Basinger et al. [1981]
pKa = 3.4, 10.0  43.81.0Martell et al. [1998]
Table 2. Stability and Conditional Constants for a Two-Coordinated Hg–Thiol Complex Expressed Following Different Conventionsa
 ReactionpHlog K
  • a

    The pKa for RSH is set to 10.0, and I = 0.

  • b

    Conditional constant at pH 7 if Hg complexed by two thiols is (incorrectly) expressed as bound to only one thiol. [RS] is set to 10−9 M ([RSH] = 10−6 M). KHgSR+ = KHg(SR)2 × [RS].

  • c

    The concentration of ligands [L] = [RSH] + [RS] is set to 10−6 M.

  • d

    The concentration of DOC (mol L−1) is back-calculated from [RSH] = 10−6 M, assuming that thiol-S corresponds to 0.15% of DOC on a mass basis [Qian et al., 2002; Skyllberg et al., 2003, 2005].

equation image
0–14log KHg(SR)2 = 42 ± 2 (M−1)
equation image
0–14log KHg(SR)2 = 22 ± 2 (M−1)
equation image
7.0log KHgSR+ = 33 ± 2b unitless
equation image
7.0log KHgL = 30 ± 2c unitless
equation image
7.0log KHg-DOC ∼ 27 ± 2d unitless
equation image
4.0log KHg-DOC ∼ 21 ± 2d unitless

[10] Stability constants reported for Hg associations to NOM are often conditional and are defined by different conventions, which complicate comparisons. Only in a few studies have organic S groups been quantified, enabling calculations of stability constants for Hg(SR)2 complexes. Skyllberg et al. [2000] reported a log K of 31.6–32.2 for a mixed complex involving one carboxyl and one thiol group at pH 3.0–3.4. Using data from Skyllberg et al. [2000], a log K (M−1) for the formation of a Hg(SR)2 complex in agreement with reaction (1a) can be calculated by the expression log KHg(SR)2 = [Hg(SR)2]/([RS]2 × {Hg2+}). The calculated log KHg(SR)2 varied between 43.2 and 43.6 for the six organic soils studied, of which the two soils with the smallest and greatest values are reported in Table 3. Khwaja et al. [2006] reported an average log KHgL of 38.3 for a bidentate Hg–thiol complex (i.e., the two thiol groups were assumed to be attached to the same molecule) for IHSS Pahokee Peat Humic Acid at pH 3.0. Using data reported by Khwaja et al. [2006], a recalculation to a log KHg(SR)2 for reaction (1a), in which two independent thiols are involved, resulted in an average of 45.3 for the IHSS Pahokee Peat Humic Acid and a range of 43.4–47.7 for extracted humic acids (HA) and intact organic soils (three of these samples are reported in Table 3). It should be noted that in these calculations thiol groups are assumed to comprise 30% of the reduced organic S groups (which in turn were determined by sulfur X-ray absorption near edge spectroscopy; XANES), in agreement with results from Qian et al. [2002] and Skyllberg et al. [2006]. The ratio of reduced organic S groups to Hg was in the range of 27–800 in the study of Khwaja et al. and 13000–77000 in the study of Skyllberg et al. [2000]. We can conclude that conditional constants reported for Hg association to organic soils and humic substances extracted from organic soils are in fair agreement with similarly defined constants (reaction (1a)) determined for Hg(SR)2 complexes in LMW thiols.

Table 3. Data Input for Calculation of the Formation Constant logKHg(SR)2a
 Org Carbon (g (C) L−1)pHRSH (mmol kg−1 C)log {Hg2+}Hg(SR)2 (μmol kg−1 C)log KHg(SR)2 (M−1)
  • a

    The pKa value for RSH is 10.0, and 30% of reduced organic S groups are assumed to be RSH. The concentration of organic carbon (g C L−1) was used to recalculate the units of Hg(SR)2 (μmol kg−1 C) and RS (mmol kg−1 C) to mol L−1.

  • b

    Data from Skyllberg et al. [2000, Table 4]; org-SRED = RSH, {Hg2+} is calculated by Davies equation at I = 0.47 M.

  • c

    Data from Khwaja et al. [2006, Table 2]; SRED = RSH, I = 0.5 M, activities calculated by Davies equation.

  • d

    Data from Khwaja et al. [2006, Table 4]; SRED = RSH, I = 0.5 M, activities calculated by Davies equation.

Bog uplandb0.83.1514.44.0 × 10−293.643.3
Fen upland-dischargeb2.33.2014.45.0 × 10−302.443.6
IHSS Pahokeec1.93.0029.44.6 × 10−3014645.3
S3 – 1HAd1.84.08683.0 × 10−329944.4
S2 – HAd1.84.301036.0 × 10−3323144.7
S3 – 3 SOMd0.54.826.55.2 × 10−3426047.7

[11] In order to compare and evaluate constants on Hg–NOM associations reported in the literature, the log KHg(SR)2 of 42 for reaction (1a) needs to be transformed to follow conventions used in other studies. A first step would be to express the complexation of Hg to two thiols as a one-coordinate complex in agreement with reaction (2). This conversion can only be done at a fixed concentration of dissociated thiols [RS], at which KHgSR+ = KHg(SR)2 × [RS]. Several studies have focused on the association of Hg to strong ligands with a typical concentration in the nM range. If we set the total concentration of thiols ([RSH] + [RS]) to 1 μM, the calculated log KRSHg+ will be 33 at pH 7.0 (Table 2), i.e., with [RS] = 1 nM. Important to note is that this log K value should not be mistaken for the log KRSHg+ value reflecting the true one-coordinated Hg–SR+ complex, which should be on the order of 20 [Dyrssén and Wedborg, 1991]. The log KHgSR+ of 33 for reaction (2) is in turn easily converted to a log KHgL of 30 for reaction (3), if [L] = [RSH] + [RS] = 1 μM. Finally, given the assumption that RSH corresponds to approximately 0.15% of DOC on a mass basis [Qian et al., 2002; Skyllberg et al., 2003, 2005], the log KHgL of 30 can be converted to conditional constants for the formation of Hg–DOC complexes, as defined by reactions (4a) and (4b), at pH 7.0 and 4.0, respectively (Table 2).

[12] Several authors have reported constants for Hg complexed by dissolved and extracted NOM as a log KRSHg+ for reaction (2). Haitzer et al. [2002] reported a log KRSHg+ of 28.5 (M−1) at pH 7.0 (I = 0.1 M), Drexel et al. [2002] reported a log KRSHg+ of 25.8–27.2 (M−1) at pH 6.0 (I = 0.01 M) and Benoit et al. [2001b] reported a log KRSHg+ of 22.4–23.8 at pH 6.0 (I = 0) for isolated DOM and peat soils from the Florida Everglades. All these constants are substantially smaller than the log KRSHg+ value of 33 for reaction (2). Even if the reported values are adjusted to I = 0 and pH = 7.0 (the values at pH 6.0 should be increased by 1.0 log-units to be compared to the value 33 at pH 7.0) these constants are too small to reflect a coordination of Hg with two thiol groups. In the study of Drexel et al. the concentration of RSH was approximated by reduced S, as determined by XANES, whereas Haitzer et al. [2002] determined the density of strong sites by Hg titration of DOM. If instead S XANES data reported by Haitzer et al. were used, the log KRSHg+ would be ∼27. As described by Gasper et al. [2007], there are many methodological uncertainties involved in the determination of Hg stability constants. These could easily explain deviations in binding affinity data from the expected two-coordination of Hg with thiols in NOM. Gasper et al. [2007] used (and recommended) a competitive ligand exchange with solid-phase extraction (CLE-SPE) method and reported log KHg-DOC values in the range 25–31 at pH 7.0 (I = 0.1 M) for hydrophobic and hydrophilic organic fractions isolated from the Florida Everglades (similar material as studied by Haitzer et al. [2002] and Drexel et al. [2002]). The concentrations of ligands [DOC] were not reported, but some of these constants seem to be in the range for twofold complexation of Hg to thiols, as reflected by reaction (4a) and Table 2.

[13] Hsu and Sedlak [2003] used the CLE-SPE method to determine the complexation of Hg to ligands in wastewater. Black et al. [2007] used the same method in natural freshwaters. Hsu and Sedlak reported a log KHgL > 30, with [L] in the range 90–500 pM (pH 6.6–7.4), and Black et al. reported log KHgL in the range 29.9–33.5, with [L] = 0.02–11 nM (pH = 7.8). Given the concentrations of ligands and pH, these constants are best compared with the log KRSHg+ of 33 for reaction (2). We can conclude that a two-coordination of Hg with thiols is in fair agreement with values at the higher end reported by Black et al. [2007], whereas values at the lower end and the log KHgL reported by Hsu and Sedlak [2003] are several orders of magnitude too small to reflect a pure twofold coordination of Hg with thiols. The latter also holds for log KHgL in the range 26.5–29.0 ([L] = 0.01–9 nM; pH 9.5–9.8), reported for marine environments [Han and Gill, 2005], log KHgL in the range 27.9–28.3 ([L] = 0.07–0.16 nM; pH 9.6), reported from an estuary [Han et al., 2007], and log KHgL in the range 21–24 ([L] = 0.3–60 nM; pH 7.5) reported by Lamborg et al. [2003] for various types of freshwaters and marine waters. In all these studies of natural waters involvement of reduced organic (and inorganic) S ligands in the complexation of Hg has been proposed (given the large stability constants). In addition to methodological constraints, a mixture of complexes with organic and inorganic S ligands may explain the variability as well as the deviation from the theoretical value for a twofold coordination with thiols. Because of the large variability in the values of the reported constants, it is recommended that only constants in agreement with the spectroscopic evidence for a two-coordinate bonding to thiols are used to calculate concentrations of Hg–NOM complexes. This is not the case today.

[14] Partly because of a utilization of too small constants for Hg–NOM interactions (and too big constants for Hg–inorganic sulfide complexes; see below), it is a widespread perception that thiols are much weaker complexing ligands than bisulfide (HS) and sulfide (S2−) in the complexation of Hg [e.g., Hudson et al., 1994; Zhang et al., 2004; Miller et al., 2007; Han et al., 2007]. Goulet et al. [2007] applied the generic Hg–fulvic acid and Hg–humic acid constants included in WHAM 6, for Hg–NOM complexes. These constants are based on data where the much weaker carboxyl groups are involved in the complexation [Tipping, 2007]. In two recent studies, both concluding that thiols are outcompeted by inorganic sulfides (even at sub μM level of S(-II)), the complexation of Hg to LMW thiols have been included. When modeling the chemical speciation of Hg in an estuary, Han et al. [2007] included glutathione (47–61 nM), but only considered the one-coordinated Hg–glutathione complex with a log KRSHg+ of 27.3, i.e., approximately 6 orders of magnitude less than the log KRSHg+ of 33 that we could expect for the two-coordinate Hg–glutathione complex, if expressed as one-coordinated. For DOM, a log KHgL of 27.9–28.3 ([L] = 0.3–60 nM) was used, a value approximately 5 orders of magnitude lower than expected for two-coordinated Hg–thiol complexes. Zhang et al. [2004] included two-coordinate complexes with glutathione (maximum concentration 40 nM) and mercaptoacetic acid (maximum concentration 220 nM) when the chemical speciation of Hg (and MeHg) was modeled in wetland pore waters. A log KHg(SR)2 of 34.0 was used for the former thiol, which likely is too low (see reaction (1a)), and a log KHg(SR)2 of 43.8 was used for the latter thiol (pKa = 10.6), in fair agreement with the average log KHg(SR)2 for reaction (1a). Unfortunately, the calculated concentration of the Hg(SR)2 was not reported. Given the above background, it is obvious that further clarification of the competitive effect between organic thiols and inorganic sulfides is needed.

2.2. Controversy About the Stability Constant for the Species HOHgSH0(aq)

[15] Dyrssén and Wedborg [1991] summarized the current knowledge on Hg and MeHg speciation in sulfidic environments. In focus was the competition between thiols and bisulfides. For the proposed complex HOHgSH0 (aq) (this complex is also designated HgS0(aq), the only difference being a water molecule) one theoretically derived constant was reported for the equilibrium with solid HgS(s):

equation image

[16] When corrected for the solubility product (log Ks1) of HgS(s) = −38.9 and the ionic product of water (pKw = 13.7), the constants for formation of a mixed hydroxide/sulfide complex fits very well into the sequence of two-coordinated Hg complexes with hydroxide and bisulfide (Table 4).

Table 4. Stability Constants for Two-Coordinated Inorganic Hg Complexesa
 ReactionLog K
equation image
equation image
equation image

[17] In a second step, Dyrssén and Wedborg compared and corrected the theoretically derived constant of reaction (5) by an equation (logKs1(calc) = 2.26 logKs1(exp)) derived by a comparison of theoretical (calc) and experimental data (exp) for the solubility of ZnS(s) and CdS(s) as determined by Gübeli and Ste-Marie [1967] and Ste-Marie et al. [1964]. This correction resulted in a log K value of −10.0 for reaction (5), thus a constant 12 orders of magnitude larger than the theoretical one. The much higher solubility obtained from experimental data was commented by Dyrssén and Wedborg as probably caused by an inclusion of colloids in the aqueous phase. This fact was further corroborated by experimental studies of Daskalakis and Helz [1993] showing much lower solubilities of ZnS(s) than reported by Gübeli and Ste-Marie [1967] and Ste-Marie et al. [1964]. If the larger log K value of reaction (5) is combined with solubility product of HgS(s) of −38.9 and the ionic product of water (pKw = 13.7), the calculated log K value for reaction (7) is 40.5. This value is even larger than the log K for the Hg(SH)2 complex (Table 4), which seems unreasonable and inconsistent with the preference of Hg for reduced S over O groups, as conclusively shown in several studies of S and O containing amino acids [e.g., Basinger et al., 1981; Leung et al., 2007]. Tossell [2001] made ab initio calculations and suggested that the complex likely is having the structure HOHgSH0 with a pKa above 7, but theoretical calculations were not in fully agreement with experimental data from Benoit et al. [1999] (see below), making comparisons about the relative stability versus Hg(SH)20 inconclusive.

[18] Neither direct spectroscopic evidence [e.g., Lennie et al., 2003] nor indirect macroscopic evidence has been presented for a significant contribution from HOHgSH0 to the dissolution of HgS(s). The latter point is shown in Figure 1a, in which solubility data from Jay et al. [2000] and Paquette and Helz [1997] are illustrated. Note that the sulfide concentrations are too high (>10−4.5 M) for a direct evaluation of a quantitative contribution from HOHgSH0 to the overall solubility. Note that a log K constant of −9.3 for reaction (5), based on calculations by Dyrssén and Wedborg, was used in the modeling by Jay et al. [2000]. There is one study in which experimental solubility data have been used to evaluate the existence of the HOHgSH0 molecule [Benoit et al., 1999]. They measured the octanol–water partitioning of Hg in a sulfide solution. The decrease in the partitioning of Hg to octanol with increasing sulfide concentration was explained as a decrease in the concentration of neutral Hg-sulfide complexes. Benoit et al. [1999] were able to explain their data using the larger constant of −10.0 for reaction (5), as reported by Dyrssén and Wedborg [1991]. A complication in the experiment was an initial oversaturation in relation to HgS(s), and as a result 96% of Hg added was sorbed onto the glassware prior to the octanol addition and shaking. This was corrected for by only considering the partitioning of Hg passing a 0.22 μm filter. If, however, some colloids of HgS(s) passed the filter and subsequently partitioned into the octanol phase, this Hg would be mistaken as dissolved, mainly neutral Hg–complexes. At this point we cannot with certainty tell the true value for the stability constant of HOHgSH0 (aq).

Figure 1.

Solubility of HgS(s) as a function of total inorganic sulfides (S-II)Tot, (a) without and (b) in presence of elemental S and polysulfides. Note that no experimental data (data from Jay et al. [2000], solid symbols; data from Paquette and Helz [1997], open symbols) are available below 10−4.5 M of S(-II)Tot. Reprinted with permission from American Chemical Society [Jay et al., 2000]. Copyright 2000 American Chemical Society.

[19] The choice of a stability constant of −22.3 or −10.0 for reaction (5) has extraordinary consequences for the interpretation of data and our understanding of Hg biogeochemistry, especially on the effect of passive uptake of neutral Hg-sulfides by SRB. If the larger constant is used, the neutral HOHgSH0 complex will dominate the overall chemical speciation of Hg at neutral pH in solution at sulfide concentrations below 10−5 M (Figure 1a). But the complex will be negligible as compared to Hg(SH)20 and its two dissociation products if the smaller constant is used.

2.3. Is an Increase in Absolute or Relative Pore Water Concentrations of MeHg a Reasonable Estimate of Net MeHg Production?

[20] In the study of Gilmour et al. [1992] it was conclusively shown that sulfate amendment increased the total amount or concentration of MeHg in sediment, owing to the activity of SRB. This conclusion seems sound since the total amount of MeHg in the sediment was determined. More problematic is when conclusions about net methylation, as a consequence of, e.g., sulfate additions to soils and sediments, are drawn exclusively on the basis of increased concentrations of MeHg in aqueous solution [e.g., Branfireun et al., 1999, 2001; Jeremiason et al., 2006; Mitchell et al., 2008b].

[21] The concentration of MeHg in pore water is a consequence of the total concentration of MeHg in the soil/sediment, the concentration of strong ligands that form soluble MeHg complexes and possible kinetic constraints regulating desorption and adsorption processes. In experiments with amendment of sulfate to potentially suboxic and anoxic environments an increase in the concentration of sulfide is expected. It is also possible that organic thiols and polysulfides may form. These circumstances will inevitably result in an increased concentration of dissolved MeHg. In none of the above mentioned studies were changes in the solubility of MeHg owing to changes in the concentrations of MeHg complexing ligands corrected for. One possible way to indirectly correct for an increased solubility would be to normalize MeHg to the concentration of Hg, assuming that MeHg and Hg are equally affected by changes in concentrations of organic and inorganic ligands. Kelly et al. [1997] showed that MeHg in % of total Hg increased seven times (from 4.5 to 32%) in a lake water as a consequence of flooding. It was supposed that the increase in % MeHg was mainly caused by a net MeHg production. Similarly, Mitchell et al. [2008a] used % MeHg as a proxy for net methylation when identifying “hot spots” of methylation in wetlands. In none of these studies, however, were relative differences in MeHg and Hg solubility as a function of sulfide and thiol concentrations estimated or corrected for. There is obviously a need for clarification of the effect of sulfide and thiol concentrations on the absolute solubility of MeHg, Hg and their molar ratio.

3. Modeling of the Speciation of Hg and MeHg in a Pore Water–Wetland Soil System

[22] The chemical speciation of Hg and MeHg in a conceptual wetland soil/sediment was modeled by considering reactions in aqueous and solid phases. The following conditions were applied in the modeling: pH = 4.0 and 7.0 and temperature 25°C. The temperature was chosen for simplicity and a change to lower, more realistic temperatures will have little effect on the modeling outcome. Davies equation was used to calculate activities at an ionic strength of 0.5 mM in the soil/sediment pore water, relevant for freshwater sediments and non-marine wetlands. The wetland soil/sediment was considered to be composed of 15 mass-% organic C (on a dry mass basis). If the NOM to organic C mass ratio is assumed to be 1.7, approximately 25% will be organic matter and the remaining 75% is considered mineral matter (silicates) with negligible affinity for Hg and MeHg (as compared to organic and inorganic reduced S ligands). The water content was set to 90%. Thus, the total organic C concentration in the soil\sediment, expressed in relation to the aqueous phase, was 16.7 g C L−1. In the aqueous phase the concentration of dissolved organic carbon (DOC), defined as passing a 0.45 μm filter, was set to 50 mg C L−1 which is a quite typical value for soil solutions and pore waters of organic rich soils and sediments [e.g., Skyllberg et al., 2003]. Data on the concentration of reduced organic S in boreal wetland soils [Skyllberg et al., 2000], organic soils and humic streams [Qian et al., 2002; Skyllberg et al., 2006], as well as for organic matter in 25 Scandinavian humic streams originating from different types of wetlands and lake ecosystems (U. Skyllberg et al., Concentrations of reduced organic sulfur in Scandinavian humic streams in relation to catchment properties, submitted manuscript, 2008a) together, give a range of 0.2–1.1% (of dry mass) and an average concentration of reduced organic S groups of 0.5 mass-% of organic C. Given that Qian et al. [2002] calculated the concentration of thiols in organic substances to be in average 30% of reduced organic S (by combining S XANES and Hg extended x-ray absorption fine structure (EXAFS) spectroscopy, using MeHg as a probe for RSH), [RSH] was calculated as 0.15 mass-% of organic C. This results in a concentration of [RSH (aq)] = 2.3 μM and a concentration of thiols pertaining to particulate organic C (in the solid phase) [RSH (total)] of 780 μM, if expressed in relation to the pore water volume. The pKa value for the reaction H2S (aq) = HS + H+ was set to 7.0 and the total concentration of inorganic sulfides was designated S(-II)Tot = [H2S(aq)] + [HS]. The total Hg in the organic soil/sediment was set to 100 ng g−1 (5.5 × 10−8 M if expressed in relation to the pore water volume) and the concentration of MeHg was set to 0.1 ng g−1 (5.5 × 10−11 M if expressed in relation to the pore water volume). Concentrations of Hg and MeHg are in a typical range for wetlands soils, estuarine and freshwater sediments [e.g., Skyllberg et al., 2000, 2003; Goulet et al., 2007; Fitzgerald et al., 2007], not affected by point sources of Hg.

[23] Model calculations were undertaken using the stability constants in Table 5. Stability constants for the formation of halide complexes (Cl, Br, I) and hydroxyls were included but data for these complexes are not reported because they will be of minor importance (as shown in previous studies [e.g., Skyllberg et al., 2003]) as compared to thiol and inorganic sulfide and polysulfide complexes. This holds for both Hg and MeHg. Not even in marine environments, with 0.6 M of NaCl, will chloride complexes contribute significantly to the solubility of Hg and MeHg in presence of thiols and/or inorganic sulfides and polysulfides. Neither concentrations of the monovalent HgSH+ complex (Hg2+ + HS = HgSH+, log K = 20.0 [Dyrssén and Wedborg, 1991]) nor mixed halide–bisulfide complexes are included in the models because of minor influence on the solubility of Hg.

Table 5. Stability Constants for Hg and MeHg Complexes Used in the Chemical Modeling of Aqueous and Solid/Surface Phasesa
 ReactionLog KReferences
  • a

    The pKa value of RSH is 10.0, the pKa value of H2S is 7.0, and the pKw of H2O is 13.7. For reaction (10) both the theoretical and experimentally corrected (in parentheses) constants are given.

equation image
22.0average for LMW thiols (this study)
equation image
−5.2Dyrssén and Wedborg [1991]
equation image
7.1 (19.4)Dyrssén and Wedborg [1991]
equation image
23.7Schwarzenbach and Widmer [1963]
equation image
17.5Schwarzenbach and Widmer [1963]
equation image
9.2Schwarzenbach and Widmer [1963]
equation image
29.4Schwarzenbach and Widmer [1963]
equation image
−3.9Paquette and Helz [1997]
equation image
−11.7Jay et al. [2000]
equation image
−15.7Jay et al. [2000]
equation image
6.5Carty and Malone [1979], Karlsson and Skyllberg [2003]
equation image
−4.4Schwarzenbach and Schellenberg [1965]
equation image
7.5Dyrssén and Wedborg [1991]

[24] Three models were used: Model A excludes the solid phase HgS(s). It has been argued that precipitation of cinnabar (α-HgS, trigonal) and metacinnabar (β-HgS, cubic) is inhibited in organic rich environments, owing to the strong complexation of Hg to NOM [Ravichandran et al., 1999; Waples et al., 2005]. Therefore Model A is restricted to an organic solid/surface phase with RSH functional groups, and sulfides do only exist in solution. Formation of polysulfides is not included. Thus Model A is restricted to include reactions (1b), (9)(13) and (18)(20). Reactions (1b) and (18) apply to Hg and MeHg associated with thiols both in solution and with thiol functionalities in particulate organic matter (i.e., the solid phase). Model B includes the formation of metacinnabar through reaction (14). The constant corresponds to a log Ks1 (HgS(s) + H+ = HS + Hg2+) of −36.4, determined by Schwarzenbach and Widmer [1963] and adopted by Martell et al. [1998] as a critically selected constant for metacinnabar dissolution. Model C includes the formation of both HgS(s) and polysulfides, as described by reactions (15)(17). For simplicity the activity of elemental S0(s), which is needed for the formation of polysulfides, is set to 1.0. Thus, elemental sulfur is available as a solid phase in the soil/sediment (if elemental S only exists as a soluble species, the activity is well below 1). The complexation of MeHg by thiols, bisulfides and hydroxyls (reactions (18)(20)) is considered in all three models, in order to compare the solubilities of Hg and MeHg. At this point there are no conditional constants for the adsorption of Hg or MeHg to surfaces of HgS, FeS(s) or mixed Hg/FeS(s) surfaces. Neither are data available for the solubility of a possible mixed Hg/FeS(s) phase. These limitations of the models need to be taken into consideration.

4. Results

4.1. Competition Between Thiols and Inorganic Sulfides With and Without HgS(s) Present

[25] The competition between dissolved thiols and inorganic sulfides for Hg can be illustrated in different ways. In Figure 2a the absolute concentrations are illustrated as a function of (S-II)Tot in the 0–2 μM, suboxic region. Hg is associated to the solid phase as surface complexes with two thiol groups, and when the concentration of inorganic sulfides increase the solubility of Hg increase due to the formation of [HgS2H] > [Hg(SH)20] > [HgS22−]. The relative concentration of these three species is determined by the pH value (in this case 7.0) in relation to the pKa1 = 6.3 and pKa2 = 8.2 of Hg(SH)20 and HgS2H, respectively. Interesting to note is that the absolute concentration of Hg(SR)2(aq) does not change very much in solution.

Figure 2.

Absolute and relative concentrations of organic thiol [Hg(SR)2] and inorganic monosulfide, bisulfide, and polysulfide complexes as a function of total sulfide concentrations, S(-II)Tot. Stability constants and conditions used in Models A, B, and C are given in section 3 and Table 5.

[26] The situation is different when HgS(s) is allowed to precipitate (Figure 2b). The competition between organic thiols in the soild phase and inorganic sulfides results in a precipitation of metacinnabar above approximately 1.2 μM S(-II) at pH 7.0 (shown by the kink in the curves in Figures 2b, 2c and 2e) and above 0.6 μM S(-II) at pH 4.0. With increasing S(-II) concentration the formation of HgS(s) will gradually outcompete Hg(SR)2 complexes in the solid phase of the soil (Table 6). This process is dependent on pH because the thiols are more competitive at higher pH. Note also the importance of organic C content, where HgS(s) is not thermodynamically stable below 9.8 μM of (S-II)Tot at pH 7.0 in a soil/sediment with 50% organic C. The incorporation of Hg into HgS(s) will decrease the concentration of Hg in solution, having an absolute effect on the concentration of Hg(SR)2(aq). In Figure 2c it can be seen that the decrease in [Hg(SR)2 (aq)] continues log linearly with increasing (S-II)Tot. The precipitation of HgS(s) will, however, also limit the concentrations of HgS2H, Hg(SH)20 and HgS22−. In relative terms, the percentage contribution of Hg(SR)2(aq) to total dissolved Hg remains the same in Models A and B. The effect of pH on the relative contribution of Hg(SR)2 (aq) to total dissolved Hg is not large, as illustrated in Figure 2d. This is because all considered S-ligands have pKa values above 7 (except Hg(SH)20, which has a pKa of 6.2), and therefore concentrations of deprotonated ligands decrease in a similar way when pH is decreased from 7 to 4. It can be seen that 50% of total dissolved Hg is complexed by organic thiols, and 50% by inorganic sulfides, at approximately 0.5 μM of S(-II). Given that the total concentration of dissolved thiols is 2.3 μM, it can be concluded that the competitiveness of thiols and inorganic bisulfides are quite equal. Note that the constant for two-coordinate Hg–thiols is somewhat uncertain, and if log KHg(SR)2 is increased one log-unit, the concentration of Hg–thiols increases from 58% to 93% of total dissolved Hg at pH 4.0 and (S-II)Tot = 0.4 μM.

Table 6. Distribution of Hg Associated to Particulate Organic Matter [Hg(SR)2] and as HgS(s) in the Solid Phase of a Soil/Sediment With 15 and 50 Mass-% Organic C as a Function of Total Concentration of Inorganic Sulfides, S(-II)Tot, in Solutiona
org CSolid Phase Hg SpeciesConcentration of S(-II)Tot, μM
  • a

    Output for Model B at pH 7.0.

15%Hg(SR)2 - tot, % of total Hg1001006633115.4
15%HgS(s), % of total Hg0034678995
50%Hg(SR)2 - tot, % of total Hg1001001001009948
50%HgS(s), % of total Hg00001.152

4.2. Competition Between Thiols and Polysulfides for Hg

[27] The studies of Paquette and Helz [1997] and Jay et al. [2000] have convincingly shown that polysulfides contribute to the solubility of Hg in presence of HgS(s) and elemental S in solid form. As can see by a comparison of Figures 2c and 2e, introduction of polysulfides (Model C) has a tremendous effect on the chemical composition of Hg in solution. The polysulfide species HgSnSH, as suggested by Paquette and Helz [1997], as well as its dissociated form Hg(Sn)22−, as suggested by Jay et al. [2000] both predominate over Hg–sulfide/bisulfide complexes at pH 7.0. The third polysulfide species, HgSnOH as suggested by Jay et al. [2000], take over completely at S(-II) concentrations below 10−5 M. This species may be seen as an analog to the proposed HOHgSH0 species, combining one hydroxyl with one reduced S ligand (and 3–5 elemental S). Similar to HOHgSH0, the experimental evidence for the stability constant of this species is not as clear as for the other two polysulfides species. As can be seen in Figure 1b, there are no experimental data in the low-sulfide region, where HgSnOH is the dominant species. The scientific basis for the reported log K value is derived from pH-dependent studies, but the constant was less well-constrained in comparison with the constant for Hg(Sn)22− [Jay et al., 2000]. As a consequence of the very strong binding of Hg to polysulfides, the competitiveness of thiols decreases dramatically at pH 7.0 (Figure 2f). At pH 4.0, however, the polysulfides are less competitive and in such acidic environments thiols are of quantitative importance (>5%) at total S(-II) concentrations below approximately 2 μM.

4.3. Effect of Choice of Stability Constant for the Formation of HOHgSH0 (aq)

[28] The concentration of HOHgSH0 is independent on both the concentration of inorganic sulfides and pH, as reflected by reaction (5). In Figures 2c and 3a the effect of using a constant of −22.3 or −10.0 of reaction (5) (corresponding to a log K of 7.1 or 19.4, respectively, for reaction (10)), is illustrated. If the smaller constant is used, HOHgSH0 will have no quantitative influence on the competitiveness of organic thiols, but if the greater constant is used the HOHgSH0 species will dominate completely (Figure 3b). The point at which Hg in solution will be 50% complexed by thiols will be change from approximately 0.5 μM to less than 0.01 μM. In presence of HgS(s), these numbers are independent of pH.

Figure 3.

The effect on HgS(s) solubility using a log K of 19.4 for reaction (10) (Figure 3a) can be compared with Figure 2c, where a log K of 7.1 is used. The effect of using a constant of 7.1 or 19.4 for reaction (10) on the relative concentration of Hg(SR)2 (aq) is illustrated for the 0–5 μM range of S(-II)Tot (Figure 3b). Calculated log Kd values for Hg and MeHg at pH 7.0 and 4.0 for Models A, B, and C are given in Figures 3c and 3d.

4.4. Modeled Partitioning Between Solid Phase and Pore Water (log Kd)

[29] In Figures 3c and 3d the outcome of the modeling is summarized and compared as plots of log Kd (L kg−1) versus the total concentration of S(-II)Tot. It can first be noted that a decrease in pH results in an increased log Kd for Hg. This is in opposition to the commonly adopted view that most metals are solubilized under acidic conditions. The reason for this pH dependency is that even if the free concentration of Hg2+ increase with decreasing pH, this species is negligible (practically non-existent), as compared to complexes with organic and inorganic S ligands. In the case of Model A, a decrease in pH below the pKa values of Hg(SH)20 (6.2) and HgS2H (8.3) will make the sum of Hg–sulfide/bisulfide complexes to decrease in solution, and log Kd to increase, when pH decrease from 7.0 to 4.0. In the case of Model B, the absolute aqueous concentrations of Hg–sulfides are smaller than in Model A, resulting in greater relative change in log Kd when pH is decreased from 7.0 to 4.0. The most pronounced effect of pH is seen in the presence of polysulfides (Model C), as described above. At pH 7.0 polysulfides keep the log Kd constant between 2.5 and 4.5 throughout the suboxic region, as compared to a log Kd of 5–6 at pH 4.0. The effect of using the greater log K for the formation of HOHgSH0 is illustrated by an incorporation of the constant in Model C. In presence of polysulfides, HOHgSH0 does not add very much to the solubility at pH 7.0 (Figure 3c). When pH is decreased, however, the pH-independent solubility of HOHgSH0 will have a very strong influence on the Hg solubility, and thus on log Kd, if the log K = 19.4 of reaction (10) is used (Figure 3d). For MeHg, the log Kd decreases slightly with a decrease in pH from 7.0 to 4.0 (Figures 3c and 3d). An increase in organic C content of the soil from 15 to 50% (e.g., an organic peat soil) will increase the log Kd by one log-unit throughout the suboxic region (0–20 μM S(-II)Tot) for Model A, whereas the log Kd for Models B and C would be less affected because of opposing effects of Hg binding to solid phase thiols and inhibition of HgS(s) formation. Once HgS(s) is formed, the organic C content will have a very small influence on log Kd. The total Hg concentration will have no effect on log Kd for Model A, whereas in presence of HgS(s) log Kd for Models B and C will increase one log-unit for a 10 times increase in total Hg.

4.5. Molar Ratio of MeHg to Hg in Aqueous Phase as a Function of S(-II)Tot

[30] The molar ratio of MeHg to Hg in pore water is illustrated in Figures 4a and 4b. At zero concentration of inorganic sulfides, the [MeHgTot (aq)]/[HgTot (aq)] ratio is 0.33 in solution. Thus, the effect of 1000 times higher concentration of Hg in the soil, as compared to MeHg, is almost canceled out by the mathematical effect of the square on the [RS] for the formation of the Hg(SR)2 complex (in solution and at surface). When only dissolved sulfide complexes are considered (Model A), the ratio first increases with increasing HS (Figure 4a) owing to the fact that two HS is required for each Hg2+ ion, whereas only one HS is needed to form the MeHgSH0(aq) complex. A maximum ratio of 0.4 (pH 7.0) and 0.5 (pH 4.0) is reached at 0.1–0.4 μM of S(-II)Tot. Then the ratio decreases, because the relative affinity for HS, as compared to organic thiols, is greater for Hg than MeHg, given the stability constants used in Model A. At a S(-II)Tot of 10 μM less than 3% (of HgTot) is MeHg in solution at pH 4.0, and 2% at pH 7.0. When a solid HgS(s) phase is allowed to precipitate (Model B), the situation is different. The incorporation of Hg into HgS(s) makes the [MeHgTot (aq)]/[HgTot (aq)] ratio to deviate from Model A above approximately 0.6 μM S(-II)Tot at pH 4.0 and above 1.2 μM S(-II)Tot at pH 7.0 (Figure 4a) and then the ratio only slowly decreases with increasing concentration of S(-II)Tot. The increasing (or less decreasing) [MeHgTot (aq)]/[HgTot (aq)] ratio is explained by a decreased competitiveness of dissolved bisulfides for Hg2+ (due to the requirement of two HS ligands), as compared to formation of HgS(s) and the MeHgSH0 complex, which only requires one HS. If polysulfides are formed (Model C), the ratio is much lower at pH 7.0 than at pH 4.0 (Figure 4a). This is of course owing to the strong complexation of Hg to polysulfides at neutral pH. In presence of polysulfides (Model C) at pH 7.0, the choice of the constant for reaction (10) for the formation of HOHgSH0 will not change the molar MeHg/Hg ratio significantly. At pH 4.0, however, the choice of the size of the constant has tremendous effects on the molar ratio, making it similar to the output at pH 7.0 in Figure 4a. Methyl mercury polysulfide complex formation is currently unknown, and therefore not included in the model. In Figure 4b the relative concentration of total Hg and MeHg in the sediment still is 1000:1, but the absolute concentrations have been increased 10 times. The increase in the molar MeHg/Hg ratio is explained by precipitation of HgS(s) already at 0.1 μM S(-II) and an increased concentration of MeHg in solution as a consequence of an increased total concentration.

Figure 4.

The effect of conditions expressed by Models A, B, and C on the molar ratio of dissolved MeHg/Hg in pore water. Total concentrations in the sediment are (a) 100 ng g−1 Hg and 0.1 ng g−1 MeHg and (b) 1000 ng g−1 Hg and 1.0 ng g−1 MeHg.

5. Discussion

5.1. Stability Constants for Hg–Natural Organic Matter Thiol Complexes

[31] It is not surprising that thiols can compete with bisulfides for Hg, given that the stability constant for bidentate Hg–thiol association is 22 ± 2 (reaction (1b)) and the constant for the formation of Hg(SH)20 is 23.7 (reaction (11)). These two constants can be directly compared since they involve the same number of ligands and the pKa values of the ligands are accounted for. It should be noted that the selected pKa value of 10.0 for the thiol group can be discussed, and if a pKa value of 8.5 is chosen (the lower end for thiols), the stability constant for reaction (1b) would have been 25.0. Thus, a proper selection of pKa values and stability constants for LMW thiols and thiols associated to NOM cannot be enough emphasized. It is not surprising that many macroscopically determined constants for Hg–NOM interactions are inconsistent with a bidentate bonding with thiols. In some cases the thiols are simply oversaturated by excessive additions of Hg, but many other experimental difficulties may also arise, such as problems with contamination. Because of the extremely low concentrations of Hg2+ in equilibrium with thiols, CLE is currently the only reliable method for determination of stability constants for Hg–NOM, but also this method may yield very different values, as summarized by Gasper et al. [2007]. Instead of selecting an experimentally derived value for a particular organic substance of interest (or a particular environment), that easily could be biased owing to experimental deficiencies, it may be more appropriate to use a well-established value for known two-coordinate Hg–thiol complexes. The concentration of thiol groups does not seem to vary tremendously among different environments (in average 0.15% of DOC in organic soils and wetlands of boreal regions), and can be determined using S XANES spectroscopic analysis. A relatively small error in the estimate of the concentration of thiol groups is to prefer over huge errors owing to selection of unreasonable stability constants.

5.2. Controversy About the Stability Constant for the Formation of HOHgSH0(aq)

[32] It is obvious from Figures 3a3d that the stability constant chosen for HOHgSH0(aq) have dramatic effects on the model output. Given the fact that Daskalakis and Helz [1993] conclusively have shown that the experimental basis used by Dyrssén and Wedborg [1991] for correcting their theoretically derived value was incorrect, the constant should be considered substantially lower than the commonly used log K value of −10 for reaction (5). Until macroscopic and spectroscopic evidence is reported where HOHgSH0 conclusively is shown to contribute significantly to the solubility of HgS(s) below a total S(-II) concentration of 10−4.5 M, caution need to be taken about results relying on a log K for reaction (5). At neutral pH and in the presence of polysulfides, however, the effect of HOHgSH0 on the solubility of HgS(s) is negligible (Figure 3c). Therefore, studies focusing on the overall solubility of Hg under such circumstances should not be substantially biased by the choice of the size of the stability constant for HOHgSH0.

5.3. Solubility of Hg and MeHg in Wetland Soils and Sediments

[33] The relevance of the Models A, B and C can be evaluated by a comparison of modeled log Kd with actual observations. Log Kd for MeHg in acidic forested, organic wetland soils (pH 3.6–5.0) with presumably low (<0.5 μM) concentrations of inorganic sulfides, showed a range between 2.6 and 4.0 [Skyllberg et al., 2003]. Goulet et al. [2007] reported an average log Kd of 3.0 in a riverine wetland with a pH between 7 and 8 and sulfide concentrations in the 0–4 μM range. Similarly, log Kd varied between 3 and 4 in a sediment of a northern freshwater lake (pH 6.5–7.5) with less than 0.5 μM of S(-II) (U. Skyllberg et al., Mercury biogeochemistry in a seasonally suboxic lake sediment, submitted manuscript, 2008b). All these data show a reasonable agreement with the model output under oxic or suboxic conditions (Figures 3c and 3d). In a recent review, Fitzgerald et al. [2007] reported log Kd for MeHg in the range 1.5–3.5 for coastal marine sediments. The lower value is reasonable in highly sulfidic environments. In some estuarine and marine environments log Kd values can be greater than 4.0 and values between 4.14 and 4.79 have been reported for sediment in San Francisco Bay, Galveston Bay and the Florida Everglades, as summarized by Choe and Gill [2003b]. These environments show generally high concentrations of sulfides and FeS(s) phases, and the high log Kd suggests that adsorption of MeHg to metal sulfides like FeS(s) needs to be included in an improved model for sulfidic environments. Unfortunately, experimental data (and stability constants) for these adsorption processes are currently lacking.

[34] For Hg, reported data on log Kd are numerous and highly variable. In estuarine environments with pH above 7.0 and high concentrations of sulfides, FeS(s) and possibility for HgS(s) formation, values cover the range 4–7 for sediments and bottom waters, with values between 5–6 most common [e.g., Choe and Gill, 2003a; Conaway et al., 2003]. These data are well covered by model B or C (Figure 3c). Depending on the activity of elemental S (the activity may range between 0–1) and sulfides in solution, Model C can take any log Kd value between 2.5 (activity of elemental S = 1 and S(-II)Tot > 1 μM) and 5.0–6.0 (activity of elemental S = 0 and S(-II)Tot < 0.5 μM) in suboxic environments at pH 7.0. In a neutral, suboxic freshwater sediment with elemental S (determined by S XANES) and S(-II)Tot < 0.5 μM, log Kd for Hg varied between 4.0 and 5.0 and could be fairly well modeled by Model C [Skyllberg et al., 2008b]. Goulet et al. [2007] reported an average log Kd of 3.8 for Hg in riverine wetland sediments with the likely formation of Hg polysulfides. The concentration of dissolved elemental S and concentrations of dissolved sulfides in pore waters varied between 0 and 4 μM. The log Kd of 3.8 is on the same order of magnitude as predicted by Model C, and suggests a significantly lower log Kd under suboxic conditions in presence of elemental S. Formation of elemental S and polysulfides in the suboxic region may also explain the minimum in log Kd observed in samples with low concentrations of acid volatile sulfide by Hammerschmidt and Fitzgerald [2004] in marine nearshore sediments.

5.4. Molar Ratio of MeHg to Hg in Aqueous Phase: Implications for Studies of MeHg Production

[35] As shown in Figures 4a and 4b, the molar ratio of MeHg/Hg in pore water may vary between 0.01 and values above 1 or more depending on the environmental circumstances. When sulfate is experimentally added to an anoxic environment, or when oxic soils are flooded, the production of inorganic sulfides (and possibly thiols) will inevitably change absolute concentrations of MeHg and Hg and their molar ratio in the pore water. The question is how much? The reported 7-fold increase of the MeHg/Hg molar ratio in solution from 0.045 to 0.32 after flooding a wetland [Kelly et al., 1997], and increases in MeHg/Hg molar ratios from approximately 0.05–0.10 before to 0.20–0.40 after addition of sulfate (and organic electron donor molecules) to peat mesocosms [Mitchell et al., 2008b] are changes on a similar scale as illustrated in the suboxic region in Figures 4a and 4b. The same is true for the study by Harmon et al. [2004], in which absolute concentrations of MeHg increased three times and the MeHg/Hg molar ratio in pore water doubled after addition of sulfate to mesocosms of a wetland soil. The inorganic sulfide concentration was in the 0.1–10 μM range, with only small changes between control and treatments. Only small treatment effects were observed when the concentration of total MeHg in the bulk soil was analyzed, with non-significant difference between control and the largest addition of sulfate.

[36] Given the results at 100 ng g−1 Hg and 0.1 ng g−1 MeHg (Figure 4a), it seems that the overall trend is that the MeHg/Hg molar ratio will initially increase and then decrease above 1 μM S(-II)Tot. This may suggest that increases in the MeHg/Hg molar ratio, as observed in the above cited studies, in fact may be attributed to a net methylation. It should, however, be noted that in presence of HgS(s), the outcome of the modeling is highly dependent on the total concentrations of Hg and MeHg. If the total concentration of both Hg and MeHg is increased 10 times (from 100 to 1000 ng g−1 Hg and from 0.1 to 1.0 ng g−1 for MeHg) the maximum MeHg/Hg molar ratio in pore water will increase more than one order of magnitude (reaching values of 3–5, Figure 4b) within the critical suboxic region when S(-II)Tot is below 1 μM.

[37] The message here is that without a proper correction for absolute and relative changes in the pore water speciation of MeHg and Hg, conclusions about net MeHg production by measurements of pore water concentrations alone are not valid. Also the possibility of kinetic effects of adsorption/desorption should be considered in field experiments (in the modeling exercise reported in this paper only equilibrium reactions are considered). At this point, notably with incomplete understanding of MeHg and Hg adsorption to FeS(s) phases, it is highly recommended that total MeHg in soil/sediment is measured in experiments with the emphasis on quantifying methylation reactions.

6. Conclusions and Implications

[38] 1. Considering the variability in stability constants reported (±2 log-units), organic thiols (RSH) have an approximately similar affinity for Hg (and MeHg) as inorganic bisulfides. This means that, in the absence of polysulfides, pH and the concentrations of RSH (aq) and H2S (aq) determine the chemical composition of Hg (and MeHg) in solution. In the pore water of a typical wetland soil or sediment, having a concentration of 50 mg DOC L−1 (and RSH constituting 0.15% of DOC), approximately 50% of Hg will be complexed by organic thiols and 50% by inorganic sulfides in the pH range 4–7 and a total sulfide concentration of 0.5 μM. Precipitation of HgS(s) will decrease the concentration of all species in solution, but has no effect on their relative concentrations. This means that Hg complexed by both organic and inorganic ligands are available for uptake by iron- and sulfate-reducing bacteria in the suboxic (0.01–10 μM (S-II)Tot) region.

[39] 2. The effect of polysulfides on the solubility of Hg is highly pH dependent. At pH 7.0 thiols will be outcompeted by Hg–polysulfide complexes at a S(-II)Tot concentration of 0.01 μM or above. If pH is decreased to pH 4.0 the competitiveness of polysulfides is much less, resulting in 50% complexation of Hg to thiols and 50% to bisulfides and polysulfides at approximately 0.3 μM S(-II).

[40] 3. The choice of one of the two proposed values of the stability constant for the formation of the mixed hydroxide–bisulfide complex HOHgSH0 (aq) of log K = −10 or −22.3 for the reaction HgS(s) + H2O = HOHgSH0 (aq), will have dramatic effects on the predicted chemical composition of Hg in solution. Given the high uncertainty in this constant, a conservative choice of the theoretically derived constant of −22.3 is recommended until solubility experiments conducted at a concentration of S(-II)Tot below 10−4.5 M and independent spectroscopic evidence give a scientific basis for an experimentally well-established constant.

[41] 4. Chemical modeling shows that the MeHg/Hg molar ratio in pore waters may vary substantially depending on the concentrations of Hg, MeHg, thiols, bisulfides, polysulfides and precipitation of HgS(s). This means that pore water concentration alone cannot be used to estimate net production of MeHg (by methylation) unless differences in solubility of Hg and MeHg under different conditions are corrected for. Therefore determination of the total concentration of MeHg in soil/sediment is highly recommended in studies of MeHg production/consumption. It should be noted that observed log Kd values for MeHg in high-sulfidic environments commonly are higher than calculated by the model used in this study. The reason for this is that MeHg likely adsorbs to solid phases like FeS(s) and possibly other metal sulfides, including HgS(s). Conditional constants for adsorption of MeHg, as well as Hg, to FeS(s) are currently not available. This gap of knowledge needs to be filled if the accuracy of the chemical speciation of MeHg and Hg is to improve.