Sediment-Water Partitioning ofInorganic Mercury in EstuariesA N D R E W T U R N E R , *G E O F F R E Y E . M I L L W A R D , A N DS O P H I E M . L E R O U X

Department of Environmental Sciences, University ofPlymouth, Drake Circus, Plymouth PL4 8AA, U.K.

The sediment-water partitioning and speciation ofinorganic mercury have been studied under simulatedestuarine conditions by monitoring the hydrophobicity anduptake of dissolved 203Hg(II) in samples from a variety ofestuarine environments. A persistent increase in thedistribution coefficient with increasing salinity is inconsistentwith inorganic speciation calculations, which predict anincrease in the concentration of the soluble HgCl42- complex(or reduction in sediment-water distribution coefficient)with increasing salinity. Partition data are, however, definedby an empirical equation relating to the salting out ofnonelectrolytes via electrostriction and are characterizedby salting constants between about 1.4 and 2.0 L mol-1.Salting out of the neutral, covalent chloro-complex, HgCl20,is predicted but cannot account for the magnitude ofsalting out observed. Since Hg(II) strongly complexes withdissolved (and particulate) organic matter in naturalenvironments, of more significance appears to be thesalting out of Hg(II)-organic complexes. Operationalmeasurements of the speciation of dissolved Hg(II) using Sep-Pak C18 columns indicate a reduction in the proportionof hydrophobic (C18-retained) dissolved Hg(II) complexeswith increasing salinity, both in the presence and absenceof suspended particles. Ratios of hydrophobic Hg(II)before and after particle addition suggest a coupled saltingout-sorption mechanism, with the precise nature of Hg-(II) species salted out being determined by the characteristicsand concentrations of dissolved and sediment organicmatter.

IntroductionSediment-water partitioning affects the reactivity, transport,and fate of chemicals in the aquatic environment (1, 2). Inestuaries, partitioning is of particular importance becauseof the human perturbation to chemical concentrations andthe strong temporal and spatial gradients of reaction-controlling variables (salinity, pH, type and concentration ofdissolved organic matter, etc). For many trace metals(including Ag, Cd, Ni, and Zn), partitioning is reasonablywell understood in terms of dissolved inorganic speciation,including the tendency of metals to form stable, solublecomplexes in saline water (3), and empirical equations existwhich define an inverse relationship between sediment-water distribution coefficient (KD) and salinity (4).

For inorganic mercury(II), however, the relatively fewexperimental results that exist indicate an increase in KD

with increasing salinity (5, 6). This effect is not predictedfrom its calculated inorganic speciation since Hg(II) has agreater tendency to form chloro-complexes than the otherGroup IIb metals, Cd and Zn (7). Taking organic matter intoconsideration, speciation calculations indicate that dissolvedinorganic Hg is strongly complexed by organic ligands,especially those containing thiol functional groups (8), infreshwater (9). In seawater, chloro-complexes are predictedto dominate since humic and fulvic material is complexedby more abundant seawater cations (10), although fieldmeasurements indicate that organic forms of Hg(II) may stillcomprise a significant fraction (> 50%) of the dissolved pool(11, 12). Inorganic mercury also has a strong affinity forparticulate organic matter (13, 14) and biomagnifies at alllevels of the foodchain (15), properties which are more usuallyassociated with hydrophobic organic chemicals. It is possible,therefore, that the nature and extent of sediment-waterpartitioning of Hg(II) are governed by its hydrophobiccharacteristics, especially in the presence of organic matter,including a tendency to be salted out of solution (or becomeless soluble) by the action of dissolved seawater ions.

In this study, we review KDs and sorption data for Hg(II)which are relevant to estuaries and present partition andspeciation results derived under simulated estuarine condi-tions using natural samples from contrasting estuarineenvironments. We demonstrate that sediment-water par-titioning of Hg(II) in estuaries may be empirically modeledby a salting process which applies to the aqueous solubilityof nonelectrolytes, and propose that the neutral chloro-complex and, more significantly, organic complexes ofHg(II) are removed from the aqueous phase via a coupledsalting out-sorption mechanism.

Materials and MethodsSampling and Sample Characterization. Water and sedimentsamples were collected during 1998 from three U.K. estuaries,selected for their diversity in physical and geochemicalconditions and degrees of contamination. The catchment ofthe mesotidal Beaulieu Estuary is largely contained withinan afforested National Park, and river water is characterizedby high concentrations of natural dissolved organic carbon(DOC), mainly in the size range 103-105 Da, and lowturbidities (16). The Plym drains granitic upland moors of aNational Park and river water is characterized by low DOCconcentrations and low turbidities. Plym riverine andestuarine particles are not chemically contaminated to anysignificant extent but contain a high proportion of kaolinitederived from clay extraction in the catchment. The macrotidalMersey Estuary drains one of the most densely populatedand industrialized catchments in Europe. The estuary ischaracterized by high DOC concentrations, of mainly an-thropogenic origin, elevated suspended particle concentra-tions, resulting from tidal stirring of bed sediment and inputof turbid (and poorly oxygenated) river water, and significantcontamination (including Hg derived from chloralkali andpaper mill industries (17)).

River and marine end-member water samples werecollected from the shore above the limit of saline intrusionand beyond the estuary mouth, respectively, in 5 L, acid-cleaned, polyethylene Nalgene bottles. Samples were filteredthrough acid-cleaned 0.45 µm Sartorius filters housed in aplastic Nalgene filtration unit and stored at 4 °C. DOCconcentrations were determined on samples that had beencollected in 250 mL borosilicate bottles, filtered through 0.7µm Whatman GF/C filters, and acidified with 250 µL of H3-

* Corresponding author phone: +44 1752 233041; fax: +44 1752233035; e-mail: aturner@plymouth.ac.uk.

Environ. Sci. Technol. 2001, 35, 4648-4654

4648 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 23, 2001 10.1021/es010933a CCC: $20.00 2001 American Chemical SocietyPublished on Web 11/01/2001

PO4, using a TOC-5000 Schimadzu organic carbon analyzer(Table 1).

A surface scrape (<5 mm) of oxic bed sediment wascollected concurrently with the river end-member and sievedthrough a 63 µm nylon mesh using river water. Aliquots ofsediment were dried at 60 °C and ground using a pestle andmortar, and specific surface areas (SSA) and particulateorganic carbon (POC) concentrations were determined,respectively, by multipoint BET nitrogen adsorption using aMicrometrics Gemini 2360 analyzer, and combustiometricallyusing a Shimadzu TOC-5000 CHN analyzer (Table 1). Analiquot of dried Mersey sediment was also digested in 30%H2O2/0.02 M HNO3 at 85 °C for 2 h in order to allow someexperiments to be undertaken in the absence of particulateorganic matter.

Sample Mixing and Incubation with 203Hg(II). Speciationand sorption experiments involved monitoring the hydro-phobicity and particle-water interactions of the gamma-emitting radioisotope, 203Hg(II), according to establishedmethods (4, 18). Experiments were undertaken in quintrip-licate, and samples were incubated with Hg(II) for no longerthan 36 h in order to minimize methylation of inorganic Hg,although conversion rates in aerated and moderately tur-bidized waters at near-neutral pH are generally less than0.05% d-1 (19).

Filtered end-members were mixed in varying proportionsto a final volume of 100 mL in acid-cleaned Nalgene bottlesand were spiked with a dilute solution of carrier-free 203Hg(as HgCl2 in HCl; Amersham International) to activities ofapproximately 100 Bq mL-1. The addition of isotope resultedin a 203Hg(II) concentration of about 20 nmol L-1 and a totalproton pH change of less than (0.2 units. Following 12 hincubation in the dark at 20 °C, 20 mL aliquots were pipettedinto scintillation vials. Further 20 mL aliquots were passedthrough reversed-phase Sep-Pak C18 columns (200 mg resinC18 Bond Elut LRC, Prolabo) by gravity at a flow rate of about20 mL h-1, to isolate hydrophobic complexes of the addedisotope, before being pipetted into scintillation vials. Thehydrophobic fraction of metal is often operationally definedas the organically complexed fraction. However, it is im-portant to appreciate that metal retained by the columnsmay include inorganic species sorbed to the C18 resin andthat some hydrophilic organic complexes may pass throughthe column (20).

An aliquot of sediment slurry was added to the remainingoriginal water mixtures (containing 203Hg(II) that had beenincubated for 12 h) to attain a suspended particle concen-tration of about 100 mg L-1. Experiments conducted over arange of turbidities (60-300 mg L-1) showed that particleconcentration did not affect the sediment-water partitioning(v/w) of Hg(II). However, a particle concentration of 100 mgL-1 was selected for the experiments because it resulted ina suitable quantity of isotope in both particulate and aqueousphases for radiochemical analysis. Following 24 h incubationin the dark under constant agitation, particulate and dissolvedphases were separated through preweighed 0.45 µm Sartoriusfilters. Twenty milliter aliquots of the dissolved phase werepipetted into scintillation vials, while further 20 mL aliquotswere passed through Sep-Pak C18 columns to isolate hydro-phobic complexes before being added to scintillation vials.

Sample Counting. Gamma-energies of the air-dried andweighed filters (in Petri dishes), 20 mL filtrates, and 20 mLsolutions passing the C18 columns were counted at 279 keVusing a high performance Wallac (Perkin-Elmer) 1480 “Wizard3” γ-counter with NaI detection. The hydrophobic fractionof dissolved isotope, fh, before and after addition of particles,was calculated from the difference between the total activityin the 20 mL filtrate and the activity passing through the C18

column. Distribution coefficients, KDs (mL g-1), were de-termined from the activities in the filters (corrected for filteradsorption) and filtrates, after appropriate correction forcontainer geometry and isotope decay.

Mass balance calculations showed that a maximum of15% activity was lost to container walls in the absence ofparticles and that almost complete recovery (> 95%) wasachieved in the presence of particles, indicating little loss ofisotope by volatilization from dried sediment. The reproduc-ibility of the experiments was generally better than 10%,except when the activity in the aqueous phase approachedthe detection limit of about 100 counts per minute (for somepartitioning experiments containing > 50% seawater in theadmixture) or the hydrophobic fraction (calculated bydifference) was less than about 15% (in some speciationexperiments using Beaulieu and Plym samples).

Results and DiscussionSorption Constants and KDs. Table 2 is a compilation ofsorption constants and field measurements for Hg(II) whichare pertinent to estuaries, including results derived byradiotracer experiments in this study. Although this reviewencompasses experiments undertaken using a variety ofsynthetic and natural sorbents under different conditions(duration, pH, particle-water ratio, particle size, degree ofagitation), some general observations can be made.

First, in the absence of organic matter, KDs (mL g-1)defining Hg(II) sorption to synthetic inorganic phases (clays,oxides, silica) are on the order of 102-104 mL g-1 and exhibita decrease with increasing salinity. This behavior is consistentwith that of most other trace metals, including the otherGroup IIb metals, Cd and Zn (4), and is attributed to thetendency of Hg(II) to form soluble and stable complexeswith Cl- (principally as HgCl4

2-) in the presence of seawaterions (7).

Second, the presence of organic matter in the aqueousphase enhances the sorption of Hg(II) to mineral surfaces.For example, sorption of Hg(II) to calcite is reduced by factorsof about 50 and 4 when organic matter is removed from riverwater and seawater, respectively (23). Sorption enhancementin the presence of dissolved organic matter is due to thegreater affinity of organic-complexes of Hg(II) (comparedwith inorganic species) for mineral surfaces (30) and en-hancement of the general sorptive properties of the particlesurface by adsorbed organic matter (31).

Third, sorption of Hg(II) is significantly enhanced by thepresence of preexisting particulate organic matter. Forexample, Gagnon and Fisher (22) demonstrated that thesorption of Hg(II) to montmorillonite in seawater wasenhanced 30-fold when the clay was coated with fulvic acid,while experiments conducted as part of this study showed

TABLE 1. Characteristics of the Sieved Sediment and End-Member Water Samples

river water/seawater sediment

estuary salinity pH DOC, mg L-1 turbidity, mg L-1 POC, % SSA, m2 g-1

Beaulieu <0.3/33.9 6.9/8.1 17.7/1.6 100 6.9 6.9Mersey <0.3/32.5 7.0/8.1 9.7/1.8 100 7.5 (<0.1)a 9.3Plym <0.3/32.8 6.7/8.1 2.0/1.5 100 2.1 6.1

a Following digestion in H2O2.

VOL. 35, NO. 23, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4649

that sorption of Hg(II) to Mersey estuarine sediment wasreduced by 1-2 orders of magnitude once particulate organicmatter had been digested in H2O2. The association of Hg(II)with sediment organic matter is exemplified by the relation-ship between the distribution coefficient defining Hg(II)sorption to natural particles in seawater (for which moredata exist than river water; Table 2) and particulate organiccarbon (POC) concentration (Figure 1). The scatter of thedata suggests that the nature (e.g., polarity, age, aromaticity)as well as quantity of organic matter is critical to the partitionprocess.

Fourth, the sorption of Hg(II) to natural sediment exhibitsa significant increase with increasing salinity or the oppositeeffect to that predicted from simple inorganic speciationconsiderations and observed for other Group IIb trace metals(4). Distribution coefficients (as KD and organic carbon-normalized values, Koc) defining 203Hg(II) sorption to estuarinesediments measured in this study are plotted against salinityin Figure 2. To our knowledge, and with the exception ofthree KDs reported by Li et al. (5) and river and marine end-member values given by Stordal et al. (6), these are the onlydata defining the salinity-dependence of Hg(II) sorption tonatural sediment that exist. Despite contrasting concentra-tions and types of organic matter (and different Koc values)in each estuary, a common, exponential increase is evident.This apparently anomalous salinity-dependent partitioning,which is reflected in many field measurements of Hg(II) (Table2), has important environmental and biological implications

for Hg in estuaries and coastal seas and is the focus of theremaining discussion.

Application of Salting Theory to Salinity-Dependent KDs.The effects of increasing pH on estuarine mixing have beenshown to slightly reduce the sorption of Hg(II) to estuarine

TABLE 2. Distribution Coefficients Defining Hg(II) Sorption to Natural and Synthetic Particles (as log KD) or Solvent (as log Kow)in Fresh Water (Salinity < 1) and Seawater (Salinity > 30)

log KD, mL g-1 (or log Kow)

sorbent water POCa, % freshwater seawater ref

Synthetic Particlesillite artificial river water and seawater 3.72 3.41 21kaolinite artificial river water and seawater 4.28 3.95 21silica natural seawater 2.36 22montmorillonite natural seawater 3.30 22fulvic acid-coated montmorillonite natural seawater 4.83 22calcite natural estuarine water and seawater 4.38b 3.83 23

natural estuarine water and seawater, 2.69b 3.19 23UV-photolyzed

Solventsoctanol (Hg as HgCl20) pure water 0.52 15octanol (Hg as Hg(OH)2

0) pure water -1.30 15

Natural ParticlesHudson estuary suspended sediment natural river water and seawater 10c 4.76 5.08 5Narragansett Bay sediment natural seawater 2.0 5.23 24Carbonate ooze natural seawater 0.5 4.40 25silicious clay natural seawater 1.3 3.47 25deep-sea sediment natural seawater 2.5 4.85 25Long Island sediment natural seawater ∼5 5.23 22Trinity River particles natural river water and seawater 5.40 5.50d 6Beaulieu estuarine sediment natural river water and seawater 6.9 4.56 5.42 this studyPlym estuarine sediment natural river water and seawater 2.1 4.73 5.42 this studyMersey estuarine sediment natural river water and seawater 7.5 4.74 5.74 this studyMersey estuarine sediment, natural river water and seawater <0.1 3.70 3.40 this studyorganic matter removed

Field StudiesGironde estuarine end-member particles natural river water and seawater 5.35 5.08 26Lena estuarine end-member particles natural river water and seawater 5.20 5.40 27Loire estuarine end-member particles natural river water and seawater 5.30 5.60 28Ob estuarine end-member particles natural river water and seawater 4.90 5.20 27Rhone estuarine end-member particles natural river water and seawater 5.75 5.65 29Scheldt estuarine end-member particles natural river water and seawater 5.70 5.70 10Seine estuarine end-member particles natural river water and seawater 6.00 5.90 28Yenisei estuarine end-member particles natural river water and seawater 5.30 5.20 27

a POC values are usually reported but cannot always be linked to corresponding sorption data. b Salinity ) 3. c As loss on ignition. d Salinity) 24.

FIGURE 1. Distribution coefficients defining the sorption of inorganicHg(II) to natural particles in seawater versus particulate organiccarbon (POC) concentration. The equation defining the regressionline is y ) 53.7 × 103x + 1940; r2 ) 0.69.

4650 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 23, 2001

particles as a function of salinity (5), presumably becausesorption of organic complexes of Hg(II) is favored at lowerpH. It appears, therefore, that aqueous Hg(II) species (orligands binding Hg) are subject to some form of salting outon estuarine mixing; that is, their solubility is reduced (orsorption enhanced) with increasing salinity through eitherelectrostriction or destabilization-coagulation.

Electrostriction refers to the more ordered and com-pressible nature of water in the presence of dissolved salt,resulting from the incorporation of water molecules intohydration spheres. Electrostriction acts to decrease thesolubility of neutral chemicals by reducing the aqueous cavityvolume able to accommodate neutral solute (i.e. neutralsolute is “squeezed out” or “salted out” of solution). Therelationship between the solubilities of a neutral chemicalin pure and saline water (C0 and Csw, respectively) and molarsalt concentration, [salt], is defined by an empirical saltingequation (32, 33):

where σ (L mol-1) is a salting constant (of positive sign forsalting out) whose magnitude is a function of the molarvolume of the solute and the chemical properties of thedissolved salt. The effects of different salts on the solubilityof a neutral chemical are generally additive (34). Thus, fora mixture of salts in solution, like seawater, σ represents theweighted sum of salting constants for the individual salts:

where N is the mole fraction of each component salt. In thepresence of suspended particles, the effect may be definedin terms of the sediment-water distribution coefficient ororganic carbon-normalized value thereof (35, 36):

Here, σads is a sorption salting constant which, in theory,should equal, or be close to, the salting constant defined ineq 1 but, in practice, may be different because interactionsbetween sediment organic matter and seawater ions tend tomodify the sorptive characteristics of the particle surface(36). From an oceanographic standpoint, since the distribu-tion coefficient is measured in river water (an electrolyte)rather than pure water, and data are considered with respectto salinity, S (g L-1), a more practical equation is:

where the factor 0.0352 accounts for the conversion toan exponential function, and from moles of salt perliter to salinity, given that the total molar salt content ofaverage seawater of salinity 34.66 g L-1 is 0.531 mol L-1

(36).Applying exponential regressions to the Hg(II) data in

Figure 2 resulted in estimates of KD0, Koc, σads, and σads* (the

salting constant on a mass basis, L g-1, ) 0.531σads/34.66) asshown in Table 3. The goodness of fits to the data suggeststhat the partitioning of Hg(II) is at least partly controlled bysome salting mechanism involving electrostriction. Also

FIGURE 2. Distribution coefficients (as KD and Koc) defining thesorption of inorganic Hg(II) to estuarine particles versus salinity inthe Beaulieu, Mersey, and Plym estuaries. The data are defined byan exponential salting equation (eqs 3 and 4) whose coefficientsare given in Table 3. Error bars represent the standard deviationof five replicate experiments.

log(C0/Csw) ) σ[salt] (1)

TABLE 3. Estimates of Salting Constants Defining Hg(II)Sorption to Estuarine Particlesa

estuarylog KD

0,mL g-1

log Koc0,

mL g-1σads*,L g-1

σads,L mol-1 r2

Equation 4Beaulieu 4.30 5.46 0.0313 2.04 0.88Mersey 4.74 5.86 0.0303 1.99 1.00Plym 4.77 6.45 0.0213 1.39 0.96

Equation 3

(i) LaboratoryHudson 4.76 5.76 0.0097 0.63Trinity 5.40 0.0047 0.31

(ii) FieldGironde 5.35 -0.0086 -0.56Lena 5.20 0.0060 0.39Loire 5.30 0.0092 0.60Ob 4.90 0.0093 0.61Rhone 5.75 -0.0029 -0.19Scheldt 5.70 ns nsSeine 6.00 -0.0029 -0.19Yenesei 5.30 -0.0029 -0.19

a Beaulieu, Mersey and Plym values were derived from exponentialregressions of KD versus salinity (eq 4); remaining values were derivedfrom end-member KDs, reported in Table 2, according to eq 3; ns ) nosignificant difference between river and marine end-member KDs.

σ ) N1σ1 + N2σ2 + N3σ3 + ... (2)

log(KDsw/KD

0) ) σads[salt] (3a)

log(Kocsw/Koc

0) ) σads[salt] (3b)

KDsw ) KD

0exp(0.0352σadsS) (4)

VOL. 35, NO. 23, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4651

shown in Table 3 are estimates of σads and σads* derived fromend-member KDs reported from additional laboratory ex-periments and field studies (shown in Table 2) using eq 3,assuming that riverine (S < 1) and pure water KDs areequivalent. Values derived from the additional laboratorystudies are generally similar to those derived from theexponential regressions, while values derived from fieldstudies are of variable magnitude and sign, presumablybecause the salting mechanism is obscured by other pro-cesses (e.g. pore water inputs, methylation-demethylation)and variables (e.g. particle concentration and type), and theanalysis of particulate Hg(II) is not specific to the sorbedfraction.

Salting out of HgCl20. Because of the inefficient shielding

of the nucleus by 4f and 5d shells, Hg(II) tends to formuncharged complexes with anions of moderate electrone-gativity, like Cl-, which are covalent, nonpolar, and lipophilic(37). For instance, the octanol-water partition coefficient,Kow, for HgCl2

0 is about 3.3 (Table 2). Thus, it is possible thatthe increase in KD with increasing salinity is at least partlythe result of an increase in the proportion of the relativelyhydrophobic and lipophilic HgCl2

0 complex, which is subjectto salting out, compared with the more hydrophilic Hg(OH)2

0

complex. Neutral chloro-complexes of the other Group IIbmetals, Cd and Zn, are ionic, more soluble (and tend todissociate), and less lipophobic (Kow for CdCl2

0 ) 0.21 (15))and are, rather than salting out, subject to an increase intheir solubility (or reduction in sorption) with increasingsalinity (4).

Qualitatively, salting out of the neutral chloro-complex isconsistent with an increase in KD with increasing salinity(Figure 2) and the observation that chloro-complexes ofHg(II) are lower in seawater than predicted (12, 38), possiblyreflecting the sorption of HgCl2

0 into particulate organicmatter. However, the magnitudes of the salting constantsderived from the exponential regressions (about 1.4-2.0 Lmol-1) far exceed aqueous or sorption salting constants fororganic compounds of similar molecular mass or molarvolume to HgCl2

0 (generally less than 0.5 L mol-1 (36, 39)).Moreover, speciation calculations indicate that HgCl2

0 com-prises only about 20% of total chloro-complexes in seawater(10), and the greater abundance of the more soluble HgCl4

2-

complex above salinities of about 10 is predicted to concealor even offset any salting effects of the neutral complex.

Salting out of Organic Complexes of Hg(II). The absenceof a salting effect observed for the sorption of Hg(II) to puremineral phases (Table 2), coupled with the detection of asignificant fraction of organically complexed Hg(II) inseawater (11, 12), suggests that dissolved organic matter and/or particulate organic matter is a prerequisite for efficientsalting out of Hg(II). Should the interaction of Hg(II) withorganic matter result in a neutral or unstable assemblage,then salting out could occur through either electrostrictionor coagulation (40). [Note that in the present context,electrostriction and coagulation are neither easy to dis-criminate nor entirely independent]. However, the mech-anism is evidently effected or catalyzed by the addition ofinorganic Hg (e.g. acting as a bridging agent, or neutralizingdissolved organic ligands) or particles (e.g. acting as nuclei),rather than destabilization of organic matter by the additionof salt (41), since sorption of Hg(II) is greatest in the marineend-member where laboratory mixing (or addition of salt topreexisting organic matter) did not occur.

Concurrent experiments undertaken in our laboratoryinvolved the use of C18 columns to operationally discriminatehydrophobic (organic) and hydrophilic (both organic andinorganic) Hg(II) species in solution (Table 4). Before theaddition of sediment particles, the hydrophobic fraction (fh)exhibited a general decline with increasing salinity, inaccordance with a predicted reduction in the organically

bound proportion of Hg (10). However, after equilibrationwith particles, the hydrophobic fraction (fh

p) increasedsignificantly at all salinities in the Beaulieu and Plym samples.The hydrophobic fraction in Mersey river water was greaterafter addition of particles but declined rapidly with increasingsalinity such that the hydrophobic fraction was significantlygreater in the absence of particles in seawater.

The ratio of the hydrophobic fraction of dissolved Hg(II)after and before addition of particles (fh

p/fh) is plotted as afunction of salinity in Figure 3. Thus, in the Mersey, thereduction in fh

p/fh with increasing salinity suggests that themore hydrophobic complexes of Hg(II) preferentially sorbto sediment particles, while in the Beaulieu and Plym, anincrease in fh

p/fh with increasing salinity suggests that morehydrophilic complexes are preferentially removed from theaqueous phase. Significantly, that the data are defined by arelationship of the form of eq 4:

suggests that sorption of Hg(II) to estuarine particles iseffected, principally, by salting out of aqueous organiccomplexes of varying degrees of hydrophobicity. The saltingconstant in eq 5 (σh, L mol-1) defines the change in relativesolubility or sorption of hydrophobic (or hydrophilic) Hg(II)complexes on estuarine mixing, and magnitudes of thisparameter (Table 5) are of the same order but not identicalto salting constants derived from the corresponding sorptionstudies shown in Figure 2. This suggests that additionalprocesses, such as salinity-induced desorption via theformation of the soluble HgCl4

2- complex, or particulate

TABLE 4. Fraction of Hydrophobic (C18-Retained) Hg(II) inEstuarine End-Member Mixtures before (fh) and after (fh

p) theAddition of Particlesa

salinity fh, % fhp, %

Beaulieu<0.3 53.0 ( 4.3 80.8 ( 4.3

1.9 63.6 ( 2.1 85.4 ( 1.14.8 61.2 ( 2.7 79.8 ( 5.29.7 57.4 ( 3.9 88.4 ( 2.9

19.4 47.6 ( 6.1 88.4 ( 2.533.9 15.4 ( 5.4 82.6 ( 0.9

Mersey<0.3 50.6 ( 3.0 78.0 ( 3.7

1.9 54.6 ( 2.7 70.2 ( 6.44.6 63.6 ( 4.0 58.2 ( 5.59.3 60.6 ( 2.1 61.2 ( 8.8

18.6 58.0 ( 3.1 26.4 ( 7.232.5 39.8 ( 4.3 1.0 ( 1.0

Plym<0.3 52.6 ( 2.6 53.1 ( 4.0

1.9 41.4 ( 5.9 54.4 ( 4.84.7 40.2 ( 6.0 56.1 ( 4.59.4 38.9 ( 5.1 57.3 ( 1.8

18.7 9.8 ( 4.8 38.9 ( 5.032.8 0.9 ( 0.3 6.8 ( 3.1

a The mean and standard deviation of five replicate experiments areshown.

TABLE 5. Estimates of Salting Constants for HydrophobicHg(II) Species in Mixtures of Beaulieu, Mersey, and PlymEstuarine End-Members (Eq 5)

estuary (fhp/fh)0 σh*, L g-1 σh, L mol-1 r2

Beaulieu 1.19 0.0170 1.11 0.85Mersey 1.98 -0.0516 -3.37 0.90Plym 1.04 0.0276 1.80 0.97

(fhp/fh)sw ) (fh

p/fh)0exp(0.0352σhS) (5)

4652 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 23, 2001

organic matter-seawater ion interactions may be significantand/or that the fraction of Hg(II) which undergoes sorptionto estuarine particles is not identical to the hydrophobic orhydrophilic fraction as defined by retention on Sep-Pakcolumns.

Figure 4 shows a relationship between the salting constantfor Hg(II) calculated from the sorption studies (as σ*ads, Lg-1) and riverine dissolved organic carbon concentration forthe three estuaries studied. Such a relationship suggests thatthe extent of salting out is also dependent on the quantityof organic matter or, more likely, the availability or abundanceof a specific pool of complexing ligands. Organic compoundshaving aqueous salting constants closest to sorption saltingconstants for Hg(II) are n-paraffins containing 16-26 carbonatoms (σ about 0.7-2.2 L mol-1 (39)), suggesting that suchligands may be large, alkylated molecules containing Hg-binding functional groups (especially thiolic groups (42)).Interestuarine differences in the concentration, and precisestructure and hydrophobicity of these ligands, together withthe nature and extent of their interaction with sedimentorganic matter, are therefore predicted to determine the

overall magnitude of salting out of Hg(II) on estuarine mixing.Specifically, that σh in the Mersey is negative reflects thegreater hydrophobicity of dissolved organic matter in con-taminated environments, due, in part, to the higher sulfurcontent and aromaticity of anthropogenic ligands comparedwith natural organic matter (43). Differences between σads inthe Beaulieu and Plym may be attributed to interestuarinedifferences in the availability of naturally occurring Hg(II)-complexing ligands.

General Implications. Without further study it is notpossible to define the precise salting-sorption mechanism(e.g. electrostriction versus coagulation) and composition ofdissolved and sediment organic matter affecting the parti-tioning of Hg(II) in estuaries. Nevertheless, the experimentalresults presented herein and the proposed mechanism arein qualitative agreement with results from field studies,including an increase in KD from river to sea (Table 2) andremoval, rather than addition, of dissolved Hg on estuarinemixing (11, 28, 44-46). Although this study has onlyconsidered species of inorganic Hg, it is predicted that neutralmethylated forms behave in a similar fashion since they havea strong affinity for particulate organic matter (22) and areknown to be salted out from aqueous solution via electro-striction (47).

The results presented for Hg(II) in this study add to theexisting body of evidence (for instance, its affinity forparticulate organic matter and tendency to biomagnify (15))that the biogeochemical behavior of Hg in the aquaticenvironment bears more semblance to a neutral organicchemical rather than an inorganic ion. From a modelingperspective, definition of Hg(II) partitioning as a function ofsalinity by an empirical salting equation is extremelysignificant since most estuarine geochemical models im-plicitly assume a power law dependence of KD on salinity fortrace metals due to competitive adsorption and complexationby seawater anions (4, 48). Increased sorption at high salinitiesfavors the retention of Hg(II) in estuaries since net particletransport in the coastal zone is generally directed landwards.Sedimentation of such particles acts to enhance the supplyand, through methylation, availability and toxicity of Hg tobenthic organisms.

AcknowledgmentsThis study was a component of the Natural EnvironmentResearch Council’s Environmental Diagnostics program,Grant Number GST/03/1567. The authors thank Dr. P. Jonesof the Environment Agency (North West region) for logistical

FIGURE 3. Ratio of the fraction of hydrophobic dissolved Hg(II)after and before addition of suspended particles, fh

p/fh, in mixturesof estuarine end-member waters as a function of salinity. The dataare defined by an exponential salting equation (eq 5) whosecoefficients are given in Table 5. Error bars represent the standarddeviation of five replicate experiments.

FIGURE 4. Salting constants defining the sorption of Hg(II) toestuarine particles versus river water end-member dissolved organiccarbon (DOC) concentration. The equation defining the regressionline is y ) 4.80 × 10-3 ln x + 0.018; r2 ) 0.97.

VOL. 35, NO. 23, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 4653

support during the Mersey sampling. The critical commentsof three anonymous reviewers were greatly appreciated.

Literature Cited(1) Sigleo, A. C.; Means, J. C. Rev. Environ. Contam. Toxicol. 1990,

112, 123.(2) Luthy, R. G.; Aiken, G. R.; Brusseau, M. L.; Cunningham, S. D.;

Gschwend, P. M.; Pignatello, J. J.; Reinhard, M.; Traina, S. J.;Weber, W. J., Jr.; Westall, J. C. Environ. Sci. Technol. 1997, 31,3341.

(3) Turner, D. R. In Metal Speciation and Bioavailability; Tessier,A., Turner, D. R., Eds.; John Wiley: Chichester, 1995; pp 149-203.

(4) Turner, A. Mar. Chem. 1996, 54, 27.(5) Li, Y.-H.; Burkhardt, L.; Teraoka, H. Geochim. Cosmochim. Acta

1984, 48, 1879.(6) Stordal, M. C.; Santschi, P. H.; Gill, G. A. Environ. Sci. Technol.

1996, 30, 3335.(7) Turner, D. R.; Whitfield, M.; Dickson, A. G. Geochim. Cosmochim.

Acta 1981, 45, 855.(8) Dyrssen, D.; Wedborg, M. Water, Air, Soil Pollut. 1991, 56, 507.(9) Mason, R. P.; Sullivan, K. A. Water Res. 1998, 32, 321.

(10) Leermakers, M.; Meuleman, C.; Baeyens, W. Water, Air, SoilPollut. 1995, 80, 641.

(11) Guentzel, J. L.; Powell, R. T.; Landing, W. M.; Mason, R. P. Mar.Chem. 1996, 55, 177.

(12) Mason, R. P.; Sullivan, K. A. Deep-Sea Res. 1999, 46, 937.(13) Benoit, J. M.; Gilmour, C. C.; Mason, R. P.; Riedel, G. S.; Riedel,

G. F. Biogeochemistry 1998, 40, 249.(14) Wallschlager, D.; Desai, M. V. M.; Spengler, M.; Windmoller, C.

C.; Wilken, R. D. J. Environ. Qual. 1998, 27, 1044.(15) Mason, R. P.; Reinfelder, J. R.; Morel, F. M. M. Environ. Sci.

Technol. 1996, 30, 1835.(16) Moore, R. M.; Burton, J. D.; Williams, P. J. leB.; Young, M. L.

Geochim. Cosmochim. Acta 1979, 43, 919.(17) Jones, P. D. Water Environ. Manage. 2000, 14, 124.(18) Le Roux, S. M.; Turner, A.; Millward, G. E.; Ebdon, L.; Appriou,

P. J. Environ. Monit. 2001, 3, 37.(19) Matilainen, T.; Verta, M. Can. J. Fish. Aquat. Sci. 1995, 52, 1597.(20) Gerringa, L. J. A.; van der Meer, J.; Cauwet, G. Mar. Chem. 1991,

36, 51.(21) Frenet-Robin, M.; Ottmann, F. Estuar. Coastal Mar. Sci. 1978,

7, 425.(22) Gagnon, C.; Fisher, N. S. Environ. Sci. Technol. 1997, 31, 993.(23) Bilinski, H.; Kozar, S.; Plasvic, M.; Kwokal, Z.; Branica, M. Mar.

Chem. 1991, 32, 225.(24) Nyffeler, U. P.; Li, Y.-H.; Santschi, P. H. Geochim. Cosmochim.

Acta 1984, 48, 1513.

(25) Buchholtz, M. R.; Santschi, P. H.; Broecker, W. S. In Applicationof Distribution Coefficients to Radiological Assessment Models;Sibley, T. H., Myttenaere, C., Eds; Elsevier: London, 1986; pp192-206.

(26) Cossa, D.; Noel, J. Mar. Chem. 1987, 20, 389.(27) Coquery, M.; Cossa, D.; Martin, J.-M. Water, Air, Soil Pollut.

1995, 80, 653.(28) Coquery, M.; Cossa, D.; Sanjuan, J. Mar. Chem. 1997, 58, 213.(29) Cossa, D.; Martin, J.-M. Mar. Chem. 1991, 36, 291.(30) Xu, H.; Allard, B. Water, Air, Soil Pollut. 1995, 56, 709.(31) Zhou, J. L.; Rowland, S.; Mantoura, R. F. C.; Braven, J. Water Res.

1994, 28, 571.(32) Poulson, S. R.; Harrington, R. R.; Drever, J. I. Talanta 1999, 48,

633.(33) Millero, F. Mar. Chem. 2000, 70, 5.(34) Rossi, S. S.; Thomas, W. H. Environ. Sci. Technol. 1981, 15, 715.(35) Brunk, B. K.; Jirka, G. H.; Lion, L. W. Environ. Sci. Technol. 1997,

31, 119.(36) Turner, A.; Rawling, M. C. Water Res. 2001, 35, 4379.(37) Jackson, T. A. In Metal Metabolism in Aquatic Environments;

Langston, W. J., Bebianno, M. J., Eds.; Chapman and Hall:London, 1999; pp 77-158.

(38) Stordal, M. C.; Gill, G. A.; Wen, L. S.; Santschi, P. H. Limnol.Oceanogr. 1996, 41, 52.

(39) Xie, W.-H.; Shiu, W.-Y.; Mackay, D. Mar. Environ. Res. 1997, 44,429.

(40) Alberts, J. J.; Filip, Z.; Leversee, G. J. Mar. Chem. 1989, 28, 77.(41) Sholkovitz, E. R. Earth Planet. Sci. Lett. 1978, 41, 77.(42) Loux, N. T. Chem. Spec. Bioavailab. 1998, 10, 127.(43) Santos, E. B. H.; Duarte, A. C. Water Res. 1998, 32, 597.(44) Campbell, J. A.; Chan, E. Y. L.; Riley, J. P.; Head, P. C.; Jones, P.

D. Mar. Pollut. Bull. 1986, 17, 36.(45) Cossa, D.; Gobeil, C.; Courau, P. Estuar. Coastal Shelf Sci. 1988,

26, 227.(46) Bilinski, H.; Kwokal, Z.; Plavsic, M.; Wrischer, M.; Branica, M.

Water Res. 2000, 34, 2001.(47) Meuleman, C.; Laino, C. C.; Lansens, P.; Baeyens, W. Water Res.

1993, 27, 1431.(48) Ng, B.; Turner, A.; Tyler, A. O.; Falconer, R. A.; Millward, G. E.

Water Res. 1996, 30, 63.

Received for review May 2, 2001. Revised manuscript re-ceived August 24, 2001. Accepted August 29, 2001.

ES010933A

4654 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 23, 2001