2138
Environmental Toxicology and Chemistry, Vol. 18, No. 10, pp. 2138?2141, 1999
q 1999 SETAC
Printed in the USA
0730-7268/99 $9.00 1 .00
ESTIMATION OF MERCURY-SULFIDE SPECIATION IN SEDIMENT PORE WATERS
USING OCTANOL?WATER PARTITIONING AND IMPLICATIONS FOR AVAILABILITY
TO METHYLATING BACTERIA
JANINA M. BENOIT,*?? ROBERT P. MASON,? and CYNTHIA C. GILMOUR?
?The University of Maryland, Center for Environmental Studies, Chesapeake Biological Laboratory, P.O. Box 38, Solomons,
Maryland 20688, USA
?The Academy of Natural Sciences, Estuarine Research Center, 10545 Mackall Road, St. Leonard, Maryland 20685, USA
(Received 8 October 1998; Accepted 22 January 1999)
Abstract?The octanol?water partioning of inorganic mercury decreased with increasing sulfide, supporting a model that predicts
decreased fractions of neutral Hg-S species with increasing sulfide. These results help explain the decreased availability of Hg to
methylating bacteria under sulfidic conditions, and the inverse relationship between sulfide and methylmercury observed in sediments.
Keywords?Mercury Methylmercury Partitioning Bioavailability Methylation
INTRODUCTION
An inverse relationship between dissolved sulfide concen-
tration and methylmercury (MeHg) production and/or concen-
tration has been observed in sediments from a number of aquat-
ic ecosystems [1?6]. Sulfide inhibition of Hg methylation may
result from a decrease in the availability of substrate Hg to
bacterial cells. However, this inhibition is not simply caused
by decreased concentration of dissolved inorganic Hg (HgD),
due to precipitation of HgS(s), as is commonly speculated [2?
4,7]. Filterable Hg concentrations do not decrease across sul-
fide gradients in natural sediments, but may increase [6,8].
Further, no correlation is found between HgD and MeHg in
sediments [9]. An alternative explanation is that shifts in the
complexation of HgD in pore waters may affect Hg bioavail-
ability to bacteria. We have hypothesized that uptake of Hg
by methylating bacteria is diffusive and that the observed sul-
fide inhibition can be explained by a decreasing fraction of
neutral dissolved Hg complexes with increasing sulfide [6,9].
It has previously been shown that neutral chloride complexes
of inorganic Hg are lipid soluble and that Hg uptake by phy-
toplankton [10,11] and Hg permeability across artificial mem-
branes [12] both occur by passive diffusion.
The existence of a neutral Hg?monosulfide complex was
proposed by Dyrssen and Wedborg [13,14], who estimated the
concentration of that is in equilibrium with cinnabar0HgS(aq)
through the reaction HgS(s) 5 . The reaction constant0HgS(aq)
(termed the intrinsic solubility) was extrapolated from the ex-
perimentally determined intrinsic solubilities of ZnS(s) and
CdS(s) [14]. The existence of this complex and the magnitude
of its formation constant remain conjectural, and several pub-
lished models for cinnabar dissolution do not include 0HgS(aq)
[15?17]. Our own modeling efforts using formation constants
gleaned from the literature, and including , suggest that0HgS(aq)
at near neutral pH, the concentration of will decrease0HgS(aq)
with increasing sulfide as it is replaced by disulfide complexes,
primarily by [6]. This trend is consistent with observed2HgHS2
* To whom correspondence may be addressed
(benoit@acnatsci.org).
decreases in MeHg production in high-sulfide sediments if
neutral species limit HgD availability to methylating bacteria.
One way to test the existence of neutral sulfide complexes
is to measure partitioning from water into a hydrophobic sol-
vent. In this investigation we report results of determinations
of octanol?water partitioning (Dow) of HgD across a sulfide
gradient. Because octanol?water partitioning depends on the
hydrophobicity of Hg species, changes in Dow across the gra-
dient provide direct evidence for the existence of a neutral
complex whose concentration depends on that of sulfide. Fur-
thermore, because partitioning provides a surrogate for passive
uptake [10,11], this study addresses a potential mechanism
whereby sulfide may limit MeHg production and accumulation
in natural sediments.
MATERIALS AND METHODS
Partitioning experiments were carried out in 20-ml de-
gassed 40 mM phosphate buffer containing 1 mg/L resazurin
as a redox indicator. Buffer was adjusted to pH 6 for the first
experiment and pH 7 for the second experiment using HCl
or NaOH. Buffer aliquots were dispensed anaerobically into
prepurged glass serum bottles. All labware was rigorously
acid-leached and deionized-water rinsed, and trace-metal-
clean laboratory protocols were used during the experiments.
Teflont-faced septa were applied to the serum bottles, and
the head space was flushed with nitrogen. Titanium nitrilo-
triacetic acid reductant [18] was added via syringe to a con-
centration of 100 mM. The standard redox potential of Ti(III)
is 2480 mV, and resazurin becomes colorless at an Eh of
about 2100 mV [19]. Buffer solutions turned from pink to
clear upon addition of the titanium nitrilotriacetic acid, and
only solutions that remained clear were used.
In the first experiment (pH 6), degassed Hg(II) standard
in dilute HNO3 was added via syringe to each serum bottle
to a final concentration of 500 pg/ml. The Hg was added after
addition of sulfide. In the second experiment (pH 7), Hg(II)
standard was added to the entire batch of buffer before dis-
pensing in an effort to reduce the variability among replicates.
In this experiment, sulfide was added after Hg. In both ex-
Mercury-sulfide speciation Environ. Toxicol. Chem. 18, 1999 2139
Table 1. Results of the octanol?water partitioning experiments
Sulfide
concentration
(log M) pH Dowa
% HgD
present as
HgS0(aq)
% HgD
present as
Hg(HS)02
Experiment 1
25.8 6 0.04
25.2 6 0.08
24.3 6 0.04
23.2 6 0.01
22.1 6 0.01
6.2
6.2
6.2
6.2
7.0
25 6 6.9
14 6 2.5
5.8 6 2.6
1.5 6 0.65
0.46 6 0.12
92
61
14
2
0
0
5
11
12
2
Experiment 2
26.0 6 0.03
25.4 6 0.01
24.2 6 0.01
23.2 6 0.02
21.8 6 0.02
7.0
7.0
7.0
7.0
7.9
24 6 5.8
11 6 3.9
2.3 6 1.8
0.84 6 1.1
20.17 6 0.09
83
33
5
0
0
0
1
2
2
0
a Determined octanol?water partitioning.
Fig. 1. Mercury speciation in the experimental solutions. The percent
of total dissolved Hg (HgD) present as various sulfide complexes is
shown versus the sulfide gradient used in the experiments.
periments, solutions were shaken for 2 h before addition of
octanol. Therefore, the equilibration period for Hg with sul-
fide was the same for both experiments.
Sulfide stock solutions were prepared in sealed, degassed
bottles using degassed 40 mM phosphate buffer. All transfers
were via syringe. Saturated Na2S was diluted to produce a
series of solutions ranging from 2 M to 0.2 mM. These were
added to the buffer solutions to provide a sulfide gradient of
10 mM to 1 mM. Each concentration was produced in qua-
druplicate for experiment 1 and in triplicate for experiment 2.
Subsamples from two of each concentration were preserved
in sulfide antioxidant buffer [20] and sulfide was measured
using an Oriont (Beverly, MA, USA) silver?sulfide ion-spe-
cific electrode.
Octanol was deoxygenated by bubbling with N2 for several
hours at room temperature. After a 2-h equilibration of the
Hg- and sulfide-containing buffer solutions, 10- to 20-ml al-
iquots of degassed octanol were delivered into the serum
bottles. Solutions were shaken for 2 h, then subsamples were
taken from the water-only controls and the aqueous portion
of the octanol?water mixtures and filtered through 0.2-mm
Acrodisc filters (Gelman Sciences, Ann Arbor, MI, USA) for
Hg analysis. These subsamples were diluted, preserved with
1% HCl, and digested overnight with 0.5% BrCl before anal-
ysis for HgT using the cold-vapor atomic fluorescence spec-
trometry method of Gill and Fitzgerald [21] and Bloom and
Fitzgerald [22]. The Hg concentration in the octanol was
calculated by difference, taking into account the volumes of
the two liquid phases. Aqueous pH was measured on separate
aliquots.
Equilibrium speciation calculations were carried out using
the MINEQL1 program (Environmental Research Software,
Hallowell, ME, USA) to estimate the fraction of HgD present
as a given complex. Because the experiments were performed
at room temperature, the MINEQL1 simulations were run at
258C. The formation constants chosen for Hg-S complexes that
were used are given in the Appendix. These values represent
average literature values rounded to the nearest 0.5 log units
(see [6] for details). A value for the formation constant of
can be derived from the intrinsic solubility (Ks1 5 210)0HgS(aq)
of cinnabar reported by Dyrssen and Wedborg [14] and the
solubility product (Ksp 5 36.7) for cinnabar originally deter-
mined by Schwarzenbach and Widmer [17], to yield a rounded
estimate for log Ks0 of 26.5 for the reaction Hg21 1 HS2 5
1 H1. This value of Ks0 provided good fit of a Hg0HgS(aq)
speciation model to data from two disparate aquatic ecosys-
tems [6]. All other equilibrium constants were from the MI-
NEQL1 database.
RESULTS AND DISCUSSION
Partitioning coefficients (Dow 5 [HgD-octanol]/[HgD-water]) for
the two experiments are given in Table 1, along with the chem-
ical equilibrium model-estimated percent of HgD present in
neutral complexes. Increasing sulfide concentration decreased
the hydrophobicity and partitioning of Hg into octanol. A de-
crease in octanol solubility is consistent with decreased passive
uptake of Hg across hydrophobic cell membranes with in-
creasing sulfide concentration. This decline in bioavailability
provides a mechanistic explanation for the frequently observed
inhibition of Hg methylation in sulfidic sediment pore waters.
Water-only controls from the experiments had an average
HgD concentration lower than the calculated solubility of HgS(s)
(i.e., ,20 ng/L). These controls indicated that 96% of the
added Hg was sorbed to glassware, and that adsorption rather
than precipitation of cinnabar controlled HgD. Adsorption was
rapid, and it was complete within the 2-h equilibration period,
before addition of octanol. Therefore, the concentration in the
controls at the end of the experiment was assumed to represent
the steady-state pool of dissolved Hg available for partitioning
into the two phases.
Figure 1 shows the calculated speciation of mercury in the
experimental solutions as a function of sulfide concentration.
Together, sulfide complexes account for 100% of the HgD
across the sulfide gradient. Two neutral dissolved Hg-S com-
plexes are present in our model. Notice that the effect of in-
creasing pH was to decrease the importance of relative0Hg(HS)2
to . At the neutral and higher pH encountered in many0HgS(aq)
2140 Environ. Toxicol. Chem. 18, 1999 J.M. Benoit et al.
Fig. 2. Partitioning curves calculated with the chemical speciation
model compared to measured octanol?water partitionings (Dows) from
the two experiments. Model curves are shown for a value of Kow 5
25 for neutral sulfide complexes ( and ).0 0HgS Hg(HS)(aq) 2
aquatic sediments, dominates as the most important0HgS(aq)
neutral Hg complex in the presence of excess sulfide.
In order to test the hypothesis that sulfide complexation
decreased the partitioning of Hg by causing a shift in the
speciation away from neutral toward charged com-0HgS(aq)
plexes, we modeled Dow for the experimental solutions using
the relationship Dow 5 S ai?(Kow)i, where Kow is the parti-
tioning coefficient of individual chemical species and a is
the fraction of Hg present as species i [after11,12]. In ex-
periment 1, all of the neutral dissolved Hg is present as
at the lowest sulfide concentration and as at0 0HgS Hg(HS)(aq) 2
the highest sulfide concentration (see Table 1), so Kow for the
two neutral complexes can be calculated using the endpoints
of this experiment. Assuming that only neutral species par-
tition significantly into octanol, at the high endpoint Dow 5
25 5 0.92(Kow)HgS0 and at the low endpoint Dow 5 0.46 5
0.02(Kow)Hg(HS)2; therefore Kow 5 27 for and Kow 5 230HgS(aq)
for . For simplicity, we used Kow 5 25 for both com-0Hg(HS)2
plexes when calculating the expected Dow for Hg across the
sulfide gradients.
The model curves are compared to the experimentally de-
termined Dow distributions in Figure 2. The decline in Dow
across the sulfide gradient is consistent with the calculated
decrease in the concentration of neutral sulfide species, which
suggests that the observed change in partitioning across the
sulfide gradient is driven by shifts in Hg?sulfide speciation.
At low sulfide dominates, but disulfide complexes be-0HgS(aq)
come more important as sulfide concentration increases. Near
neutral pH, the major disulfide complex ( ) is charged2HgHS2
and hydrophilic, so partitioning of HgD is inhibited.
CONCLUSIONS
The results demonstrate the existence of neutral dissolved
Hg complexes in sulfidic solution. A chemical equilibrium
model including two neutral complexes successfully repro-
duced experimental Dows for Hg. The model indicated that
is the dominant dissolved neutral Hg complex deter-0HgS(aq)
mining lipid-solubility in sulfidic solutions at near neutral pH.
The concentration of neutral dissolved Hg complexes decreas-
es with increasing sulfide concentration, which is consistent
with observed patterns of MeHg production and accumulation
in aquatic ecosystems [5,6]. These results support our hy-
pothesis that passive uptake of neutral dissolved Hg-S com-
plexes may control the bioavailability of Hg to methylating
bacteria. On the other hand, pore-water Hg complexation may
depend on the presence of ligands other than sulfide, including
dissolved organic carbon and polysulfides, in many natural
sediments. Chemical equilibrium models of dissolved Hg com-
plexation in pore waters may be useful in identifying ecosys-
tems that are vulnerable to MeHg production and bioaccu-
mulation.
Acknowledgement?This work was supported by the South Florida
Water Management District (C-7690), the Florida Department of En-
vironmental Protection (SP-434), and U.S. Geological Survey Co-
operative Agreement Z929801. J. Benoit was supported by a Ches-
apeake Biological Laboratory Fellowship and a U.S. Environmental
Protection Agency Star Fellowship.
REFERENCES
1. Craig PJ, Moreton PA. 1983. Total mercury, methyl mercury and
sulphide in River Carron sediments. Mar Pollut Bull 14:408?411.
2. Compeau G, Bartha R. 1983. Effects of sea-salt anions on the
formation and stability of methyl mercury. Bull Environ Contam
Toxicol 31:486?493.
3. Compeau G, Bartha R. 1987. Effect of salinity on mercury-meth-
ylation activity of sulfate-reducing bacteria in estuarine sedi-
ments. Appl Environ Microbiol 53:261?265.
4. Winfrey MR, Rudd JMW. 1990. Environmental factors affecting
the formation of methylmercury in low pH lakes. Environ Toxicol
Chem 9:853?869.
5. Gilmour CC, Riedel GS, Ederington MC, Bell JT, Benoit JM,
Gill GA, Stordal MC. 1998. Methylmercury concentrations and
production rates across a trophic gradient in the northern Ever-
glades. Biogeochemistry 80:327?345.
6. Benoit JM, Gilmour CC, Mason RP. 1999. Sulfide controls on
mercury speciation and bioavailability to methylating bacteria in
sediments pore waters. Environ Sci Technol 33:951?957.
7. Blum JE, Bartha R. 1980. Effect of salinity on methylation of
mercury. Bull Environ Contam Toxicol 25:404?408.
8. Bloom NS, Gill GA, Cappellino S, Dobbs McShea L, Driscoll
R, Mason R, Rudd J. 1999. Speciation and cycling of mercury
in Lavaca Bay, Texas sediments. Environ Sci Technol 33:7?13.
9. Benoit JM, Gilmour CC, Mason RP, Riedel GF, Riedel GS. 1998.
Behavior of mercury in the Patuxent River estuary. Biogeochem-
istry 80:249?265.
10. Mason RP, Reinfelder JR, Morel FMM. 1996. Uptake, toxicity,
and trophic transfer of mercury in a coastal diatom. Environ Sci
Technol 30:1835?1845.
11. Mason RP, Reinfelder JR, Morel FMM. 1995. Bioaccumulation
of mercury and methylmercury. Water Air Soil Pollut 80:915?
921.
12. Gutknekt JJ. 1981. Inorganic mercury (Hg21) transport through
lipid bilayer membranes. J Membr Biol 61:61?66.
13. Dyrssen D, Wedborg M. 1989. The state of dissolved trace sul-
phide in seawater. Mar Chem 26:289?293.
14. Dyrssen D, Wedborg M. 1991. The sulphur?mercury (II) system
in natural waters. Water Air Soil Pollut 56:507?519.
15. Paquette K, Helz G. 1995. Solubility of cinnabar (red HgS) and
implications for mercury speciation in sulfidic waters. Water Air
Soil Pollut 80:1053?1056.
16. Paquette K, Helz G. 1997. Inorganic speciation of mercury in
sulfidic waters: The importance of zero-valent sulfur. Environ Sci
Technol 31:2148?2153.
17. Schwarzenbach G, Widmer M. 1963. Die Loslichkeit von Me-
tallsulfiden. I. Schwarzes Quecksilbersulfid. Helv Chim Acta 46:
2613?2628.
Mercury-sulfide speciation Environ. Toxicol. Chem. 18, 1999 2141
APPENDIX
Mercury?sulfide complexes and formation constants (Kf) used in
the chemical equilibrium model for dissolved Hg speciation
Reaction Log Kf
21 2 0 1Hg 1 HS 5 HgS 1 H 26.5(aq)
21 2 1Hg 1 HS 5 HgS 1 H 36.5(s)
21 2 1Hg 1 HS 5 HgSH 30.5
21 2 0Hg 1 2HS 5 Hg(HS) 37.52
21 2 2 1Hg 1 2HS 5 HgS H 1 H 32.02
21 2 22 1Hg 1 2HS 5 HgS 1 2H 23.52
18. Moench TT, Zeikus JG. 1983. An improved method for a tita-
nium(III) media reductant. J Microbiol Methods 1:199?202.
19. Gerhardt P. 1981. Manual of Methods for General Bacteriology.
American Society for Microbiology, Washington, DC.
20. Brouwer H, Murphy TP. 1994. Diffusion method for the deter-
mination of acid-volatile sulfides (AVS) in sediments. Environ
Sci Technol 13:1273?1275.
21. Gill GA, Fitzgerald WF. 1987. Picomolar mercury measurements
in seawater and other materials using stannous chloride reduction
and two-stage gold amalgamation with gas phase detection. Mar
Chem 20:227?243.
22. Bloom N, Fitzgerald WF. 1988. Determination of volatile mercury
species at the picogram level by low temperature gas chroma-
tography with cold vapour atomic fluorescence detection. Anal
Chim Acta 208:151?161.