Salting out of chemicals in estuaries: implications for contaminant partitioning and modelling

The Science of the Total Environment 314–316(2003) 599–612

0048-9697/03/$ - see front matter� 2003 Elsevier Science B.V. All rights reserved.doi:10.1016/S0048-9697(03)00076-7

Salting out of chemicals in estuaries: implications for contaminantpartitioning and modelling

Andrew Turner*

Department of Environmental Sciences, University of Plymouth, Plymouth PL4 8AA, UK

Received 30 September 2002; received in revised form 4 November 2002; accepted 1 January 2003

Abstract

Many neutral chemicals are salted out of aqueous solution via electrostriction and exhibit increased sorption withincreasing concentration of dissolved salt. Salting out has significant implications for the reactivity, transport and fateof chemicals discharged to estuaries, but attempts to define or model the effect in such environments have beenlimited. This paper examines new and existing data on the sorption of neutral chemicals(specifically, a tetrachlorinatedbiphenyl, a phthalic acid ester, and neutral species of tributyltin) to estuarine suspended particles in order to evaluatethe potential application and limitations of salting theory to estuarine chemical modelling. It is shown that the salinitydependence of sorption may be empirically modelled using a salting equation, but salting constants derived fromdata-fitting are often significantly greater than those derived by calculation or from conventional aqueous solubilitystudies. This suggests that the hydrophobicity of sediment organic matter is modified by interactions with dissolvedseawater ions, and(or) chemical solubility is enhanced in river water via hydrophobic interactions with dissolvedorganic matter. In some estuaries, trace metals also appear to be salted out, suggesting that stable neutral complexesare formed between transition metals and a specific, but undefined pool of dissolved organic ligands. Despitesuccessful empirical modelling of the effect in estuaries, predictive modelling of salting out is currently hampered bya lack of understanding or definition of the precise interactions between(i) neutral solute(or trace metal) anddissolved and sediment organic matter, and(ii) sediment organic matter and dissolved seawater ions.� 2003 Elsevier Science B.V. All rights reserved.

Keywords: Salting out; Sorption; Estuaries; Neutral organic compounds; Tributyltin; Organic matter; Trace metals

1. Introduction

Salting out refers to the reduction in aqueoussolubility of neutral solute in the presence ofdissolved ions. Neutral solute is ‘squeezed out’ or‘salted out’ of solution because water is moreordered and compressible when bound up in hydra-tion spheres(Millero, 1996). This effect has sig-

*Tel.: q44-1752-233041; fax:q44-1752-233035.E-mail address: [email protected](A. Turner).

nificant implications regarding the reactivity,transport and fate of neutral chemicals in estuaries,where the concentration of dissolved salts mayincrease from-10 mol l in river water to iny3 y1

excess of 0.5 mol l in seawater. Specifically,y1

neutral chemicals at or near saturation in riverwater must volatilise, rise to the surface in theform of a slick, or, more important for largermolecules, sorb to suspended particles as salinityincreases. Accordingly, estuaries are predicted to

600 A. Turner / The Science of the Total Environment 314 –316 (2003) 599–612

reduce or retard the flux of many neutral chemicalsto the ocean.Despite the number and diversity of harmful

chemicals discharged to rivers and estuaries, andthe availability of salting data for most types oforganic chemical, attempts to model the sediment–water partitioning and transport of such chemicalsin estuaries by saltation theory have been limited(Readman et al., 1987; Hegeman et al., 1995;Means, 1995; Brunk et al., 1997; Turner andRawling, 2001). Moreover, treatment of dissolvedand sediment organic matter and consideration oflinear free-energy relationships(LFERs) have beeninconsistent among these studies. To this end, thecurrent paper provides a critical review of thepotential applications, limitations and implicationsof saltation theory in the context of chemicalpartitioning and modelling in estuaries. The dis-cussion focuses on the sorption of hydrophobic,neutral organic compounds(NOCs), including pol-ycyclic aromatic hydrocarbons, chlorinated hydro-carbons and phthalic acid esters, and neutralorganometallic compounds(NOMCs), since, forpractical and theoretical reasons, experimental sal-tation studies generally target such chemicals.However, it is argued that, under certain condi-tions, charged inorganic species, including tracemetals, may be neutralised via interactions withdissolved and particulate organic matter, and thatthe resulting neutral assemblages are subject tosalting out and may also be modelled by saltationtheory.

1.1. Derivation of solubility data and distributioncoefficients

While most of the data reviewed in this paperhave been taken from the literature, some newsolubility and sorption data generated by our lab-oratory are also presented. The basic approach ofthese studies involves monitoring the aqueousconcentration of C-labelled NOCs(or gamma-14

emitting trace metals) added to mixtures of filteredriver water and seawater, either in the presence orabsence of suspended estuarine sediment, at 208C. Full details of sampling, addition of radio-chemical, incubation, sediment–water separationand radiochemical analysis are given elsewhere

(Rawling et al., 1998; Le Roux et al., 2001; Turnerand Rawling, 2001).

2. Saltation theory

Aqueous dissolution of a neutral chemicalinvolves overcoming water–water interactions andthe formation and occupation of a cavity(Tanford,1973; Schwarzenbach et al., 1993). Displacedwater molecules salvage their water–water inter-actions elsewhere, but water molecules lining cav-ities must interact with fewer molecules. Cavityformation therefore becomes increasingly energet-ically unfavourable with an increase in the molec-ular size or diffusion molar volume of the neutralsolute (Miller et al., 1985; Hawker and Connell,1988). In the presence of dissolved salt, water ismore ordered and compressible, since a proportionof water molecules is bound up in hydrationspheres, an effect known as electrostriction(Mil-lero, 1996). The increased compressibility of salinewater results in a reduction in the potential foraqueous cavity formation, and the solubility ofmost, but not all neutral chemicals decreases withincreasing salt content(Stoessell and Byrne, 1982;Bu and Warner, 1995; Inaba et al., 1995; Bennettand Larter, 1997; de Bruyn and Saltzman, 1997;Xie et al., 1997; Bullister and Wisegarver, 1998;Zhang et al., 1998). The extent of ‘salting out’ isdefined by an empirical salting(or Setschenow)equation(Eganhouse and Calder, 1976):

0 sw w xlog C yC ss salt (1)Ž .

where C and C are the relative or saturated0 sw

solubilities of the compound in pure and salinewater, respectively,wsaltx is the concentration ofdissolved salt(mol l ) in saline water, ands (ly1

mol ) is a salting(or Setschenow) constant, withy1

a positive sign for salting out. The effects ofdifferent salts are generally additive(Rossi andThomas, 1981). Thus, for a mixture of salts insolution, such as seawater(NaCl, KCl,MgSO ,«), s represents the weighted sum of4

salting constants for the individual salts:

ssN s qN s qN s « (2)1 1 2 2 3 3

601A. Turner / The Science of the Total Environment 314 –316 (2003) 599–612

Table 1Octanol–water partition coefficients, pure water–seawater salting constants, and values ofa andb wEq. (4)x for various neutralcompounds at 20–258C

Compound(abbreviation) log Kow a b s Referencea

(l mol )y1

Naphthalene 3.36 0.87 0.69 0.230 Hashimoto et al., 1984Anthracene 4.54 0.87 0.69 0.326 Hashimoto et al., 1984Phenanthrene 4.57 0.87 0.69 0.272 Hashimoto et al., 1984Pyrene 5.13 0.87 0.69 0.294 Hashimoto et al., 1984Benzo(a)pyrene(BaP) 6.50 0.87 0.69 0.157 Whitehouse, 1984Biphenyl 4.09 0.85 0.73 0.259 Hashimoto et al., 19842,29,5,59-tetrachlorobiphenyl(2,29,5,59-TCB) 6.18 0.85 0.73 0.116 Turner and Rawling, 2001Di-(2-ethylhexyl) phthalate(DEHP) 7.20 1.06 -0.22 1.250 Turner and Rawling, 2000Tributyltin hydroxide(TBTOH) 3.53 – – 0.610b Arnold et al., 1997Triphenyltin hydroxide(TPTOH) 4.09 – – 0.360b Arnold et al., 1997

Reference for salting constant only. Values forK are taken from Schwarzenbach et al.(1993) and references therein, with theaow

exception of organotin species(Arnold et al., 1997) and DEHP, for which a median value from Brooke et al.(1990) has been used.Values ofa were derived from the LFERs shown in Fig. 1.

Determined in NaCl solution, pH 10.b

whereN is the mole fraction of thenth componentsalt.Since, for neutral chemicals, solubility is

inversely related to sorption or partitioning, analternative form of Eq.(1) is as follows(Schwar-zenbach et al., 1993):

sw 0 w xlog K yK sas salt (3)Ž .ow ow

whereK is the octanol–water partition coeffi-ow

cient, a measure of the hydrophobicity of thechemical, anda is a data-fitted parameter thatdefines the gradient of the LFER betweenK andow

aqueous solubility for a congeneric group ofchemicals:

0 0logK syalogC qb (4)ow

whereb is a constant. Eq.(3) may be rewrittenin terms of salinity,S, as follows:

sw 0K sK exp 0.0352asS (5)Ž .ow ow

where 0.0352 accounts for the conversion to anexponential function, and from moles of dissolvedsalts to salinity(Turner and Rawling 2001). Eq.(5) represents a practical and theoretical basis formodelling the partitioning of neutral chemicals inestuaries, sinceK ands are readily determined0

ow

from conventional chemical methods and, for thelatter, by calculation(Barba et al., 1989; Gschwendand Schwarzenbach, 1992), and estimates ofaexist for a wide range of chemicals(Chiou et al.,1982; Schwarzenbach et al., 1993; Chu and Chan,2000).

3. Partitioning of NOCs in estuaries

3.1. Ideal partitioning

Table 1 lists values forK ands, derived fromow

experimental(solubility) studies, for a variety ofNOCs and NOMCs(with abbreviated names usedin subsequent discussion also defined). Values ofa derived from the gradients of LFERs for eachcompound type, and as illustrated in Fig. 1, arealso given. Fig. 2 shows the change in octanol–water partitioning for these compounds as a func-tion of salinity, calculated according to Eq.(5).Although the calculations for organotin compoundsare based on the partitioning and salting out of asingle neutral species, it must be borne in mindthat a combination of species, including chargedentities that are subject to additional electrostaticinteractions, are present in natural waters, andwhose relative concentrations vary with salinity.For the remaining compounds, which do not com-bine with inorganic ions, such diagrams afford a


Fig. 1. Relationships between octanol–water partitioning and saturated solubility(at 20 8C) for different types of NOC(PAH,polycyclic aromatic hydrocarbons; PCB, polychlorinated biphenyls). Data are taken from Schwarzenbach et al.(1993) and referencestherein. The gradients of the relationships define the magnitude of the parametera.

first-order predictive framework for the magnitudeand change in partitioning on estuarine mixing.Thus, salting out is not predicted to be significantfor chemicals such as BaP and naphthalene, forwhichK values are relatively invariant along theow

salinity gradient, but should be highly significantfor DEHP, which exhibits a five-fold increase inK from freshwater to seawater. The extent ofow

salting out is not solely related to the compoundhydrophobicity(BaP is one of the most hydropho-bic compounds shown), but is a complex functionof its diffusion molar volume, molecular diameter

and hydrophobic surface area, i.e. the factors thatdictate the volume and shape of the aqueous cavityrequired for dissolution(Miller et al., 1985;Gschwend and Schwarzenbach, 1992). Specifical-ly, the alkylated groups of phthalic acid esterseffect a more irregular shape and an increase inflexibility compared with the flatter and more rigidstructure of fused benzene molecules.Although a partitioning algorithm of the form

of Eq. (5) has been encoded into hydrodynamicframeworks such as the Estuarine ContaminantSimulator (ECoS; Harris et al., 1984; Natural


Fig. 2. Octanol–water partitioning as a function of salinity forNOCs calculated according to Eq.(5) and using parametervalues given in Table 1.

Environmental Research Council, 1991), it mustbe appreciated that such a saltation model issubject to important assumptions and limitations.For example, potential effects arising from thepresence of dissolved organic matter in naturalwaters have been neglected, and it is implicitlyassumed that a simple or all-encompassing rela-tionship exists between octanol–water partitioningand sediment–water partitioning for neutral sol-utes. Potential sources of deviation from idealsalinity-dependent partitioning are examined indetail below.

3.2. Effects of dissolved organic matter

For the estuarine modeller, a more accurateparameterisation ofs is required that accounts forpossible additive or competing effects arising fromthe presence of additional dissolved chemicals innatural waters. The electrolytic properties of riverwater are generally similar to pure water, but riverwater may contain significant quantities of com-plex organic molecules and colloids. NOCs mayassociate with organic matter of relatively highmolecular weight or high aromaticity by complex-ation, sorption or entrapment in structural voids(Engebretson and von Wandruszka, 1998). Suchinteractions enhance the apparent solubility ofNOCs in river water(Tanaka et al., 1997; Uhle etal., 1999), but appear to have little effect inseawater(Turner and Rawling, 2001), possiblybecause hydrophobic riverine organic matter issalted out on estuarine mixing(Alberts et al.,1989).Table 2 shows the solubility of 2,29,5,59-TCB, a

planar polychlorinated biphenyl, in various naturaland artificial river waters relative to its solubilityin pure (Milli-Q ) water, along with salting con-stants calculated with respect to its solubility innatural seawater. Compared with pure water, sol-ubility in river water is enhanced by up to approx-imately 35%(equivalent to a three-fold increasein the value ofs). For experiments employing acommercial humic acid, solubility enhancementincreases with concentration of dissolved organiccarbon (DOC), but for experiments employingnatural samples, the extent of solubility enhance-ment bears no relationship to DOC. For example,


Table 2Solubility of 2,29,5,59-TCB at 208C in various natural and arti-ficial river waters(S-0.3), relative to its solubility in Milli-Qwater(MQ) and natural seawater(ass)

Water sample DOC Solubility s

(mg l )y1 relative to MQ (l mol )y1

MQ 0.27 1.00 0.116Beaulieu 8.97 1.15 0.241River Carnon 0.92 1.30 0.354Dart 2.08 0.98 0.102Plym 0.98 1.09 0.192MQq1 mg l HAy1 0.60 1.01 0.125MQq10 mg l HAy1 4.20 1.21 0.287MQq35 mg l HAy1 11.99 1.35 0.385

DOC, dissolved organic carbon; HA, Aldrich humic acid(catalogue number HI675-2). Solubility values are relative(orunsaturated) values derived from glass–water partitioning stud-ies; mean values of four experiments are shown(relative stan-dard deviations were typically-5%).

solubility enhancement is greatest in a river dom-inated by acid mine drainage(Carnon), possiblybecause of the protonation and partial neutralisa-tion of dissolved organic matter, but is negligiblein a (relatively) organic-rich river draining uplandmoor and bog(Dart). This is a clear indicationthat the effect is highly dependent on the natureof the dissolved organic matter, including itsmolecular weight, structure, aromaticity and func-tionality, and its ability to interact with inorganicsolutes(Chin et al., 1997; de Paolis and Kukkonen,1997).

3.3. Presence of suspended sediment

Estuaries are characterised by the presence ofsuspended particles of variable concentration andcomposition, and of more practical value to con-taminant modellers is a salting expression involv-ing the sediment–water partitioning of thechemical of interest rather than its aqueous solu-bility. The octanol–water partition coefficient inEq. (5) may be replaced with the sediment-waterdistribution coefficient,K , or the organic carbon-D

normalised distribution coefficient,K (sK y f ,oc D oc

where f is the fraction of sediment organicoc

carbon) as follows(Chin and Gschwend, 1992):

sw 0K sK exp 0.0352a9sS (6a)Ž .D D

sw 0K sK exp 0.0352a0sS (6b)Ž .oc oc

wherea9 anda0 define the respective relationshipsbetweenK andK and aqueous solubility. SuchD oc

parameters are difficult to quantify, since the natureand availability of heterogeneous sediment organicmatter are highly variable and poorly defined.Moreover, sediment–water partitioning is affectedby factors not directly related to the concentrationand nature of sediment organic matter, includingorganic matter–ion interactions(Murphy et al.,1994; Means, 1995) and sorption to inorganicminerals(Huang et al., 1996; Mader et al., 1997).Perhaps a more appropriate term to include in Eq.(6a) and Eq.(6b), therefore, is a sorption saltingconstant,s , that empirically defines the effectads

of salting out of the aqueous compound, togetherwith the aforementioned additive or competingeffects:

sw 0K sK exp 0.0352s S (7)Ž .D D ads

To evaluate the nature and extent of deviationof an NOC or NOMC from ideal salting behaviourin an estuary(or the significance of additionalprocesses such as complexation by dissolvedorganic matter and interaction of sediment organicmatter with seawater ions) requires determinationof s via empirical fitting of environmental data,ads

and comparison of this value with the aqueoussalting constant,s, as described below.

4. Application of saltation model to estuarinesorption data

4.1. NOCs

Figs. 3 and 4 show the salinity-dependent dis-tribution coefficients for DEHP and 2,29,5,59-TCB,respectively, the NOCs that have been most exten-sively studied under simulated estuarine conditions(Zhou and Rowland, 1997; Rawling et al., 1998;Turner and Rawling, 2000, 2001; Zhou and Liu,2000), together with empirical fits to the dataaccording to Eq.(7). Also shown is the predictedpartitioning based on empirically derived distribu-tion coefficients in freshwater and aqueous saltingconstants reported in Table 1. The fact that all data


Fig. 3. Sediment–water partitioning of DEHP under simulatedestuarine conditions. Solid lines are best fits to experimentaldata(shown as squares) according to Eq.(7), for which thecoefficients are defined in Table 3. Broken lines are distribu-tions calculated from freshwaterK values(derived from fit-D

ting of experimental data) and the aqueous salting constantreported in Table 1. Data are derived from Zhou and Rowland(1997), Turner and Rawling(2000) and Zhou and Liu(2000).

Fig. 4. Sediment–water partitioning of 2,29,5,59-TCB undersimulated estuarine conditions. Legend as Fig. 3. Data arederived from Turner and Rawling(2001) and Turner and Tyler(1997).

are defined by an exponential equation, the coef-ficients of which are reported in Table 3, suggeststhat salting out, either directly or indirectly, exertsthe principal control on variations in partitioningof such chemicals in estuaries, at least for auniform, conservative particle population.For DEHP in the Conwy and Humber estuaries

(Fig. 3), the sorption salting constants derivedfrom data-fitting are in close agreement with theaqueous salting constant given in Table 1, sug-gesting that variations inK are controlled, prin-D

cipally, by the effects of dissolved salts on aqueoussolubility. In the Beaulieu Estuary(Fig. 3) dataonly exist for the river and marine end-members,


Table 3Details of exponential regressions of sediment–waterK vs. salinity for DEHP and 2,29,5,59-TCB in various estuaries(Figs. 3 andD

4)

Estuary K0D K0

oc sads r2 s ysads Reference(ml g )y1 (ml g )y1 (l mol )y1

DEHPBeaulieua 2200 94800 2.100 – 1.68 Turner and Rawling, 2000Conwy 2060 111000 1.100 0.99 0.88 Zhou and Liu, 2000Humber 49900 875000 1.080 0.75 0.86 Zhou and Rowland, 1997

2,29,5,59-TCBDartb 45000 900000 0.516 0.97 4.48 Turner and Rawling, 2001Dartc 159000 3180000 0.437 0.97 3.79 Turner and Rawling, 2001Humber 60200 1200000 0.480 0.52 4.14 Turner and Tyler, 1997

The ratios ys defines the deviation of the compound from ideal salting behaviour.ads

Parameter values are derived from river water and seawaterK values only.aD

Particle concentration 500 mg l .b y1

Particle concentration 50 mg l .c y1

but s 4s. Experiments undertaken in Beaulieuads

end-members and Milli-Q water in the absence ofparticles indicated little solubility enhancement byriverine dissolved organic matter, suggesting thatthe hydrophobicity of sediment organic matter isenhanced by interactions with seawater ions in thisestuary. Potential effects of particle–ion interac-tions include a reduction in the negative surfacecharge of estuarine particles with increasing salin-ity through compression of the electrical doublelayer and specific adsorption of divalent cations(Loder and Liss, 1985; Beckett and Le, 1990),and salt-induced reorientation of adsorbed organicmatter and consequent entrapment of looselysorbed chemicals(Means, 1995).Data for 2,29,5,59-TCB in the Dart Estuary(Fig.

4) afford the clearest evidence for modification ofsediment organic matter by dissolved seawaterions. Thus, solubility studies employing filteredestuarine end-members and mixtures thereof indi-cate a salting constant of approximately 0.1 lmol , which increased to between;0.4 and 0.6y1

l mol (dependent on particle concentration, buty1

in excess of all aqueous values reported in Table2) in the presence of Dart Estuary particles(Turnerand Rawling, 2001). In the Humber Estuary(Fig.4), distribution coefficients are more scatteredsince they were determined on unfiltered samplesof different particle characteristics and concentra-tions collected along the estuarine axis. Neverthe-

less, as in the Dart,K values are significantlyD

greater than predicted from salting out of theaqueous compound(Table 1).The implication of these observations is that the

effects of salinity on the sediment–water partition-ing of NOCs may be significant, and, in theabsence of other processes, such as those respon-sible for a particle concentration effect(see laterdiscussion), can be empirically modelled withreasonable accuracy using saltation theory. How-ever, until the structure and composition of naturalorganic matter and its ability to interact with NOCsare better characterised, accurate predictive mod-elling of the particle–water interactions of NOCswill rely on site-specific salting and sorption con-stants derived by experiment.

4.2. Tributyltin

Tributyltin (TBT) appears to be the only organ-ometallic chemical for which the sediment–waterpartitioning has been studied under simulated estu-arine conditions. However, data presented do notgenerally discriminate between different neutraland charged species, and distribution coefficientstherefore reflect the combined sorptive behaviourof each form. The speciation of TBT is sensitiveto many environmental variables, including salin-ity, pH and concentration of dissolved and partic-ulate organic matter, and this may partly explain


Fig. 5. Sediment–water partitioning of TBT under simulatedestuarine conditions. Solid lines are best fits to experimentaldata(shown as squares) according to Eq.(7), for which thecoefficients are defined in Table 4. Broken lines are distribu-tions calculated from freshwaterK values(derived from fit-D

ting of experimental data) and the aqueous salting constant forTBTOH (Table 1). Data are derived from Harris et al.(1996).

why results generated from partitioning studies inestuaries are very different. Randall and Weber(1986) and Harris et al.(1996) showed that thedistribution coefficient for TBT increased withincreasing salinity, consistent with hydrophobicinteractions and salting out of neutral species(Inaba et al., 1995). However, Unger et al.(1988)showed that the sorption of TBT on estuarinesediment declined with increasing salinity, consis-tent with competitive sorption between chargedforms of TBT and seawater ions. It is possiblethat the latter, electrostatic effect is at least partlypromoted in experiments employing unrealisticallyhigh concentrations(G10 g l ) of suspendedy1

particles, which act to reduce the buffering capac-ity of the medium and favour the formation of thecationic form (TBT ). Hydrophobic interactionsq

are more likely at higher pH when TBT speciationis dominated by the neutral compound, TBTOH(Arnold et al., 1997).The data of Harris et al.(1996) are replotted in

Fig. 5, and the best exponential fits have beenused to calculate sorption salting constants, asgiven in Table 4. The magnitude of these valuesexceeds the salting constant for TBTOH(Table1), suggesting the presence of additional speciesthat are subject to salting out, or, more likely,alteration of the hydrophobicity of the particlesurface by seawater cations, as discussed above.Also shown in Table 4 are salting constants derivedfrom a study of the sorption of TBT to hydrousiron oxide in artificial brackish water(Ss5) andseawater(Randall and Weber, 1986). These resultsindicate a particle concentration dependence ofsalting out, an effect that was evident from studiesof 2,29,5,59-TCB sorption to Dart estuarine particles(Table 3), and that has significant implicationsregarding chemical modelling in estuaries(seelater discussion).

4.3. Trace metal complexes

The discussion thus far has focussed on theeffects of dissolved salts on the solubility andsorption of discrete neutral molecules or stableneutral compounds(e.g. uncharged species ofTBT). For many charged chemicals, includingtrace metals, inorganic speciation calculations pre-

dict a reduction in sorption, or increase in apparentsolubility, with increasing salinity because of com-petition for particle sorption sites by seawatercations, and the formation of stable and relativelysoluble complexes with seawater anions(Cl ,y

SO «) (Hegeman et al., 1992; Paalman et al.,2y4

1994). However, recent evidence suggests thatsaltation theory may apply to neutral assemblagesformed by electrostatic or hydrophilic interactionsbetween trace metals and specific dissolved organ-ic ligands (Turner et al., 2001). That is, trace


Table 4Details of exponential regressions of sediment–waterK vs. salinity for TBT in various estuaries(Fig. 5)D

Estuary K0D K0

oc sads r2 Reference(ml g )y1 (ml g )y1 (l mol )y1

Crouch 384 19200 0.78 0.93 Harris et al., 1996Tamar 380 7600a 1.32 0.97 Harris et al., 1996Fe oxide in artificial 113000c – 0.929c – Randall and Weber, 1986brackish wateryseawaterb 1420d – 0.330d – Randall and Weber, 1986

Crouch and Tamar data were derived from radiotracer experiments using C-labelled TBT; Fe oxide data were derived from14

experiments employing various butyltin chlorides and determined by hydride-generation atomic absorption spectrometry.Organic carbon concentrations were not reported with original TBT data, but a value of 5%, based on previous studies in thea

estuary(Millward et al., 1990), was used to calculateK .ocParameter values are derived from brackish water(Ss5) and seawaterK values only.b

D

Particle concentration 10 mg l .c y1

Particle concentration 1000 mg l .d y1

metals may partially or completely neutralise cer-tain organic ligands, and the resulting, more hydro-phobic complexes are subject to salting out onincreasing salinity.Fig. 6 shows the salinity distributions ofKD

values for trace metals in a variety of estuaries.Data are also defined in terms ofK to beoc

consistent with previous presentation(Figs. 3–5),although it is important to appreciate that, unlikeNOCs, uptake of trace metals and organic com-plexes thereof is not necessarily dominated bysorption into sediment organic matter. Zinc andHg data were derived in our laboratory by estab-lished radiotracer techniques(Le Roux et al.,2001), while Cr and Pb data are based on analysesof dissolved and suspended particulate metal insamples collected in the Mersey Estuary(Comberet al., 1995). Despite these different approaches,all data sets are reasonably well defined by asalting equation, for which the coefficients aregiven in Table 5. Deviations from the best-fit linesmay be attributed to the effects of additionalchemical processes(for example, complexation byand competition with inorganic seawater ions, andsolubility enhancement by complexation withhydrophilic organic ligands) and, for Cr and Pb,the presence of additional uncontrolled variables,such as the concentration and composition ofsuspended particles.Although not a general observation for trace

metals, an examination of the literature revealsthat an exponential increase inK with increasingD

salinity is most significant for certain metals, suchas Hg(Turner et al., 2001), or limited to estuariescontaining relatively high concentrations of dis-solved organic matter or that are highly contami-nated(Barbeau and Wollast, 1994; Baskaran et al.,1997). This suggests that the effect may be con-trolled by the abundance or availability of specificligands. Salting constants for trace metals aresignificantly greater than those for most of theNOCs reported in Table 1, but are close to thoseof n-paraffins of carbon number 16–26(Suttonand Calder, 1974). Although such compounds orderivatives thereof are unlikely to be responsiblefor the effect, they afford an indication of the sizeand molecular mass(or diffusion molar volume)of more representative ligands. Little informationexists about the salting tendency of metal com-plexes(Zhang et al., 1998), but the most hydro-phobic appear to be formed between transitionmetals and a variety of anthropogenic ligands,including alkylated xanthogenates and phenanthro-lines (Florence et al., 1992; Campbell, 1995).

4.4. The particle concentration effect

The model presented thus farwEqs.(5) and(7)xhas not accounted for potential effects arising fromvariations in particle character and concentration.Such effects may be significant, as demonstratedin laboratory sorption studies employing differentparticle concentrations or populations(see data for2,29,5,59-TCB sorption to different concentrations


Fig. 6. Sediment–water partitioning of trace metals in various estuaries. Chromium and Pb data are derived from analysis of samplescollected in the Mersey Estuary(Comber et al., 1995), and Hg and Zn data are derived from radiotracer sorption studies conductedby our laboratory. Solid lines are best fits to experimental data according to Eq.(7), for which the coefficients are defined in Table5; note that the lowest salinity data point for Pb has been omitted from the empirical fit.

Table 5Details of exponential regressions of sediment–waterK vs. salinity for Cr, Pb, Hg and Zn in various estuariesD

Metal Estuary K0D K0

oc sads r2 Reference(ml g )y1 (ml g )y1 (l mol )y1

Cr Mersey 36400 728000b 2.040 0.79 Comber et al., 1995Hg Plym 59000 2810000 1.390 0.96 Turner et al., 2001Pba Mersey 53200 1060000b 1.670 0.95 Comber et al., 1995Zn Beaulieu 6960 101000 1.360 0.93 This study

Hg and Zn data were derived from radiotracer sorption experiments, and Cr and Pb data were derived from analysis of samplescollected along the axis of the estuary.

Regression analysis excludes data point for lowest salinity where competitive adsorption and(or) an external source of Pb isa

evident.Organic carbon concentrations were not reported with original metal data, but a value of 5%, based on previous studies in theb

estuary(Millward et al., 1990), was used to calculateK .oc


of Dart estuarine particles, Table 3; or data forTBTCl sorption to different concentrations of Feoxide precipitate, Table 4). Strictly, therefore, thesalting model is applicable to a single particlepopulation traversing the estuarine salinity gradi-ent, a condition approximated in microtidal strati-fied flows. In more dynamic, macrotidal estuaries,local and axial particle concentrations vary consid-erably over a tidal cycle and a particle concentra-tion effect(PCE) must be accounted for.The PCE refers to the inverse relationship

between sediment–water partitioning(K ) andD

suspended particle concentration that is displayedby many chemicals, and especially NOCs, in theaquatic environment(Schrap and Opperhuizen,1992; Bergen et al., 1993; Santschi et al., 1997).The effect is highly significant, since an order ofmagnitude increase in suspended particle concen-tration may result in an equivalent reduction in themagnitude ofK . The precise cause of the PCED

has not been unequivocally resolved, but it isprobably the net result of many different processes.These include the operational determination ofcolloidal particles(sorbents) in the aqueous phase,changes in particle surface area or release ofcomplexing ligands by particle–particle interac-tions, and, in estuaries, dilution of fine, organic-rich suspended particles by tidally resuspendedbed sediment of lower sorptive capacity. Empiricalalgorithms defining the PCE have been reported(Turner et al., 1999, and references therein). How-ever, the form of an equation that combines theeffects of salinitywEq. (7)x and particle concentra-tion is unclear, because the two effects appear tointeract; that is, the extent of salting out is depend-ent on particle concentration, or the magnitude ofthe PCE is dependent on salinity(Turner andRawling, 2000, 2001).

5. Conclusions and directions for further study

The salinity-dependent sediment–water parti-tioning of many neutral chemicals in estuaries canbe fitted using an empirical equation based on thesalting out of neutral solute via electrostriction.Predictive modelling of partitioning is not, how-ever, possible at present since the magnitudes ofobserved salting out and absolute distribution coef-

ficients are dependent on a number of effects thatare neither understood nor well defined. Theseinclude the nature and availability of heterogene-ous dissolved and sediment organic matter, andtheir interactions with seawater ions, and a poten-tial complex dependence of these effects on sus-pended particle concentration. Current mainstreamresearch is focussed on the refinement of LFERs(Goss and Schwarzenbach, 2001), and character-isation of the structure and composition of sedi-ment organic matter and its influence on thesorption of neutral organic chemicals(Huang andWeber, 1997). From an estuarine perspective, how-ever, improved modelling will also rely on anunderstanding of the effects of dissolved salts onthese relationships and characteristics. Othereffects that require definition for more accuratecontaminant modelling are the potential depend-ence of s or s on water temperature(i.e.ads

season) and the kinetics of particle–water inter-actions and their dependence on salinity. Theprecise mechanism of salting out of trace metalcomplexes and the nature and hydrophobicity ofmetal-complexing ligands also deserve more atten-tion, since the effect has not previously beenrecognised for trace metals, yet has significantimplications regarding their modelling and behav-iour in estuaries and coastal areas.

Acknowledgments

Drs Sophie Le Roux and Carl Rawling(UoP)are thanked for conducting some of the radiochem-ical experiments. Discussions with Dr John Harris(Plymouth Marine Laboratory) about the sorptionof TBT in estuaries, and the perceptive commentsof Dr Fauzi Mantoura(Plymouth Marine Labora-tory) were greatly appreciated.

References

Alberts JJ, Filip Z, Leversee GJ. Interaction of estuarineorganic matter with copper and benzo(a)pyrene. Mar Chem1989;28:77–87.

Arnold CG, Weidenhaupt A, David MM, Muller SR, Haderlein¨SB, Schwarzenbach RP. Aqueous speciation and 1-octanol–water partitioning of tributyl- and triphenyltin: effect of pHand ion composition. Environ Sci Technol 1997;31:2596–2602.


Barba D, Evangelista F, Marrelli L. A general thermodynamictool for predicting salting coefficients of non-electrolytes inbrackish or seawater. Desalination 1989;71:289–300.

Barbeau K, Wollast R. Microautoradiography(with controlledliquid scintillation) applied to the study of trace metaluptake by suspended particles: initial results using Ni as63

a tracer. Limnol Oceanogr 1994;39:1211–1222.Baskaran M, Ravichandran M, Bianchi TS. Cycling of Be7

and Pb in a high-DOC, shallow, turbid estuary of south-210

east Texas. Estuarine Coastal Shelf Sci 1997;45:165–176.Beckett R, Le NP. The role of organic matter and ioniccomposition in determining the surface charge of suspendedparticles in natural waters. Colloids Surf 1990;44:35–49.

Bennett B, Larter SR. Partition behavior of alkylphenols incrude oil brine systems under subsurface conditions. Geo-chim Cosmochim Acta 1997;61:4393–4402.

Bergen BJ, Nelson WG, Pruell RJ. Partitioning of polychlori-nated biphenyl congeners in the seawater of New BedfordHarbor, Massachusetts. Environ Sci Technol 1993;27:938–942.

Brooke D, Nielsen I, de Bruijn J, Hermens J. An interlabora-tory evaluation of the stir-flask method for the determinationof octanol–water partition coefficients(log P ). Chemo-ow

sphere 1990;21:119–133.Brunk BK, Jirka GH, Lion LW. Effects of salinity changesand the formation of dissolved organic matter coatings onthe sorption of phenanthrene: implications for pollutanttrapping in estuaries. Environ Sci Technol 1997;31:119–125.

Bu X, Warner MJ. Solubility of chlorofluorocarbon 113 inwater and seawater. Deep-Sea Res 1995;42:1151–1161.

Bullister JL, Wisegarver DP. The solubility of carbon tetra-chloride in water and seawater. Deep-Sea Res1998;45:1285–1302.

Campbell PG. Interactions between trace metals and aquaticorganisms: a critique of the free-ion activity model. In:Tessier A, Turner DR, editors. Metal speciation and bioa-vailability in aquatic systems. Chichester: John Wiley, 1995.p. 45–102.

Chin YP, Gschwend PM. Partitioning of polycyclic aromatichydrocarbons to marine porewater organic colloids. EnvironSci Technol 1992;26:1621–1626.

Chin YP, Aiken GR, Danielsen KM. Binding of pyrene toaquatic and commercial humic substances: the role ofmolecular weight and aromaticity. Environ Sci Technol1997;31:1630–1635.

Chiou CT, Schmedding DW, Manes M. Partitioning of organiccompounds in octanol–water systems. Environ Sci Technol1982;16:4–10.

Chu W, Chan KH. The prediction of partitioning coefficientsfor chemicals causing environmental concern. Sci TotalEnviron 2000;248:1–10.

Comber SDW, Gunn AM, Whalley C. Comparison of thepartitioning of trace metals in the Humber and Merseyestuaries. Mar Pollut Bull 1995;30:851–860.

de Bruyn WJ, Saltzman ES. The solubility of methyl bromidein pure water, 35‰ sodium chloride and seawater. MarChem 1997;56:51–57.

de Paolis F, Kukkonen J. Binding of organic pollutants tohumic and fulvic acids: influence of pH and the structureof humic material. Chemosphere 1997;34:1693–1704.

Eganhouse RP, Calder JA. The solubility of medium molecularweight aromatic hydrocarbons and the effects of hydrocar-bon co-solutes and salinity. Geochim Cosmochim Acta1976;40:555–561.

Engebretson RR, von Wandruszka R. Kinetic aspects of cation-enhanced aggregation in aqueous humic acids. Environ SciTechnol 1998;32:488–493.

Florence TM, Powell HKJ, Stauber JL, Town RM. Toxicity oflipid-soluble copper(II) complexes to the marine diatomNitzschia closterium: amelioration by humic substances.Water Res 1992;26:1187–1193.

Goss KU, Schwarzenbach RP. Linear free energy relationshipsused to evaluate partitioning of organic compounds. EnvironSci Technol 2001;35:1–9.

Gschwend PM, Schwarzenbach RP. Physical chemistry oforganic compounds in the marine environment. Mar Chem1992;39:187–207.

Harris JRW, Bale AJ, Bayne BL, Mantoura RFC, Morris AW,Nelson LA, Radford PJ, Uncles RJ, Weston SA, WiddowsJ. A preliminary model of the dispersal and biologicaleffects of toxins in the Tamar Estuary, England. Ecol Model1984;22:253–284.

Harris JRW, Cleary JJ, Valkirs AO. Particle–water partitioningand the role of sediments as a sink and secondary source ofTBT. In: Champ MA, Seligman PF, editors. Organotin.London: Chapman & Hall, 1996. p. 459–473.

Hashimoto Y, Tokura K, Kishi H, Strachan WMJ. Predictionof seawater solubility of aromatic compounds. Chemosphere1984;13:881–888.

Hawker DW, Connell DW. Octanol–water partition coefficientsof polychlorinated biphenyl congeners. Environ Sci Technol1988;22:382–387.

Hegeman WJM, van der Weijden CH, Zwolsman JJG. Sorptionof zinc on suspended particles along a salinity gradient: alaboratory study using illite and suspended matter from theRhine River. Neth J Sea Res 1992;28:285–292.

Hegeman WJM, van der Weijden CH, Loch JPG. Sorption ofbenzowaxpyrene and phenanthrene on suspended harbor sed-iment as a function of suspended sediment concentrationand salinity: a laboratory approach using the cosolventpartition coefficient. Environ Sci Technol 1995;29:363–371.

Huang W, Weber WJ Jr. A distributed reactivity model forsorption by soils and sediments. 10. Desorption, hysteresis,and the chemical characteristics of organic domains. EnvironSci Technol 1997;31:2562–2569.

Huang WL, Schlautman MA, Weber WJ Jr. A distributedreactivity model for sorption by soils and sediments. 5. Theinfluence of near-surface characteristics in mineral domains.Environ Sci Technol 1996;30:2993–3000.


Inaba K, Shiraishi H, Soma Y. Effects of salinity, pH andtemperature on aqueous solubility of four organotin com-pounds. Water Res 1995;29:1415–1417.

Le Roux SM, Turner A, Millward GE, Ebdon L, Appriou P.Partitioning of mercury onto suspended sediments in estu-aries. J Environ Monitor 2001;3:37–42.

Loder TC, Liss PS. Control by organic coatings of the surface-charge of estuarine suspended particles. Limnol Oceanogr1985;30:418–421.

Mader BT, Goss KU, Eisenreich SJ. Sorption of non-ionic,hydrophobic organic chemicals to mineral surfaces. EnvironSci Technol 1997;31:1079–1086.

Means JC. Influence of salinity upon sediment–water parti-tioning of aromatic hydrocarbons. Mar Chem 1995;51:3–16.

Miller MM, Waslik SP, Huang GL, Shiu WY, Mackay D.Relationships between octanol–water partition coefficientand aqueous solubility. Environ Sci Technol 1985;19:522–529.

Millero FJ. Chemical oceanography. 2nd ed.Boca Raton, Flor-ida, 1996. (469 pp)

Millward GE, Turner A, Glasson DR, Glegg GA. Intra- andinter-estuarine variability of particle microstructure. Sci TotalEnviron 1990;97y98:289–300.

Murphy EM, Zachara JM, Smith SC, Phillips JL, WietsmaTW. Interaction of hydrophobic organic compounds withmineral-bound humic substances. Environ Sci Technol1994;28:1291–1299.

Natural Environmental Research Council. Plymsolve: an estu-arine contaminant simulator—user manual. Swindon:NERC, 1991. (91 pp).

Paalman MAA, van der Weijden CH, Loch JPG. Sorption ofcadmium on suspended matter under estuarine conditions:competition and complexation with major seawater ions.Water Air Soil Pollut 1994;73:49–60.

Randall L, Weber JH. Adsorptive behavior of butyltin com-pounds under simulated estuarine conditions. Sci TotalEnviron 1986;57:191–203.

Rawling MC, Turner A, Tyler AO. Particle–water interactionsof 2,29,5,59-tetrachlorobiphenyl under simulated estuarineconditions. Mar Chem 1998;61:115–126.

Readman JW, Mantoura RFC, Rhead MM. A record ofpolycyclic aromatic hydrocarbon(PAH) pollution obtainedfrom accreting sediments of the Tamar Estuary, UK: evi-dence for non-equilibrium behaviour of PAH. Sci TotalEnviron 1987;66:73–94.

Rossi SS, Thomas WH. Solubility behavior of three aromatichydrocarbons in distilled water and natural seawater. Envi-ron Sci Technol 1981;15:715–716.

Santschi PH, Lenhart JL, Honeyman BD. Heterogeneous pro-cesses affecting trace contamination distribution in estuaries:the role of natural organic matter. Mar Chem 1997;58:99–125.

Schrap SM, Opperhuizen A. On the contradiction betweenexperimental sorption data and the sorption partitioningmodel. Chemosphere 1992;24:1259–1282.

Schwarzenbach RP, Gschwend PM, Imboden D. Environmentalorganic chemistry. New York: John Wiley, 1993. (681 pp).

Stoessell RK, Byrne PA. Salting-out of methane in single-saltsolutions at 258C and below 800 psi. Geochim CosmochimActa 1982;46:1327–1332.

Sutton C, Calder JA. Solubility of higher-molecular-weightn-paraffins in distilled water and seawater. Environ Sci Tech-nol 1974;8:654–657.

Tanaka S, Oba K, Fukushima M, Nakayasu K, Hasebe K.Water solubility enhancement of pyrene in the presence ofhumic substances. Anal Chim Acta 1997;337:351–357.

Tanford C. The hydrophobic effect. New York: John Wiley,1973. (200 pp).

Turner A, Rawling MC. The behaviour of di-(2-ethylhexyl)phthalate in estuaries. Mar Chem 2000;68:203–217.

Turner A, Rawling MC. The influence of salting out on thesorption of neutral organic compounds in estuaries. WaterRes 2001;35:4379–4389.

Turner A, Tyler AO. Modelling adsorption and desorptionprocesses in estuaries. In: Jickells TD, Rae JE, editors.Biogeochemistry of intertidal sedimentsCambridge Univer-sity Press, 1997. p. 42–58.

Turner A, Hyde TL, Rawling MC. Transport and retention ofhydrophobic organic micropollutants in estuaries: implica-tions of the particle concentration effect. Estuarine CoastalShelf Sci 1999;49:733–746.

Turner A, Millward GE, Le Roux SM. Sediment–water parti-tioning of inorganic mercury in estuaries. Environ SciTechnol 2001;35:4648–4654.

Uhle ME, Chin YP, Aiken GR, McKnight DM. Binding ofpolychlorinated biphenyls to aquatic humic substances: therole of substrate and sorbate properties on partitioning.Environ Sci Technol 1999;33:2715–2718.

Unger MA, MacIntyre WG, Huggett RJ. Sorption behavior oftributyltin on estuarine and freshwater sediments. EnvironToxicol Chem 1988;7:907–915.

Whitehouse BG. The effects of temperature and salinity onthe aqueous solubility of polynuclear aromatic hydrocarbons.Mar Chem 1984;14:319–332.

Xie W-H, Shiu W-Y, Mackay D. A review of the effect ofsalts on the solubility of organic compounds in seawater.Mar Environ Res 1997;44:429–444.

Zhang YH, Danielsson LG, Becze Z. Partition of copper andcadmium betweenn-octanol and aqueous phases usingsegmented flow extraction. Water Res 1998;32:2073–2080.

Zhou JL, Rowland SJ. Evaluation of the interactions betweenhydrophobic organic pollutants and suspended particles inestuarine waters. Water Res 1997;31:1708–1718.

Zhou JL, Liu YP. Kinetics and equilibria of the interactionsbetween diethylhexyl phthalate and sediment particles insimulated estuarine systems. Mar Chem 2000;71:165–176.

Date post:	15-Sep-2016
Category:	Documents
Upload:	andrew-turner
View:	215 times
Download:	0 times

Salting out of chemicals in estuaries: implications for contaminant partitioning and modelling

Documents