On the Relationship between Dow and Kow in Natural Waters

All Publications/Website .... fish exposed to pharmaceuticals and personal care products with the fish plasma model and whole-body tissue concentratio...
0 downloads 0 Views 161KB Size
Environ. Sci. Technol. 2005, 39, 8719-8727

On the Relationship between Dow and Kow in Natural Waters A N D R E W T U R N E R * ,† A N D IAN WILLIAMSON‡ School of Earth, Ocean and Environmental Sciences, University of Plymouth, Drake Circus, Plymouth PL4 8AA, U.K., and Environment Agency South West, Starcross Laboratory, Staplake Mount, Exeter EX6 8PE, U.K.

The relationship between the overall octanol-water partition coefficient of a mixture of related chemical species, Dow, and the octanol-water partition coefficients of its components, (Kow)i, is explored. One form of the relationship (model 1) is generally applicable but relies on definition of aqueous phase speciation at equilibrium with octanol. An alternative form of the relationship (model 2) circumvents this requirement but assumes that related species are conserved during the partitioning process and is explicitly dependent on the water to octanol volume ratio, Vw/Vo. The potential applications and limitations of each model for defining the hydrophobic characteristics of chemical species in natural waters are examined in the light of experimental partition results for dissolved Cu and Pb in river waters. Given the general difficulties in accurate speciation modeling of trace metals in natural samples, model 1 was only able to estimate a Kow (typically in the range 0.03-0.3) for a computed organically complexed fraction of metal (generally > 90%). However, by conducting partition “isotherms” as a function of Vw/Vo and, because of the buffering capacity of natural waters, by treating a sample as two distinct hydrophilic and hydrophobic “pools”, model 2 was able to estimate both the abundance and Kow of a more specific group of species. Parameter values derived from the latter approach indicated that river waters comprise a relatively small pool (about 4-20%) of metal whose octanol-water partitioning is in the region of 15-150. Given that the free ion activity of strongly binding metals in natural waters is extremely small, the hydrophobic fraction may, in many cases, represent the most biologically and environmentally significant component of metal. Accordingly, the experimental and modeling approaches described herein could be of great significance to an improved understanding of the fate and impacts of trace metals in the aquatic environment.

Introduction Octanol is an amphiphilic solvent whose dielectric properties are similar to those of a generalized lipid phase. The octanolwater partitioning of a chemical affords a measure of its hydrophobicity and, in many cases, its lipophilicity. Accordingly, it is recognized as a fundamental chemical characteristic whose applications in environmental studies * Corresponding author phone: +44 1752 233041; fax: +44 1752 233035; e-mail: [email protected]. † University of Plymouth. ‡ Environment Agency South West. 10.1021/es050135a CCC: $30.25 Published on Web 10/07/2005

 2005 American Chemical Society

range from predicting the extent of sorption onto natural solids to modeling food-web accumulation (1). For neutral, nondissociating organic compounds, octanol-water partitioning is reasonably straightforward to determine and is given in terms of an absolute coefficient, Kow

Kow )

Co Cw

(1)

where Co and Cw represent the equilibrium concentrations of the chemical in octanol and water, respectively. For multiple species of the same chemical constituent in simple model systems, including components of ionogenic compounds (organic acids and bases) (2, 3) and different complexes or organocompounds of a given metal (4-6), related species are not always analytically distinguishable. Thus, an overall octanol-water partition coefficient, Dow, is often employed that defines the ratio of the summed concentration of species in octanol to the summed concentration of species in water:

Dow )

∑(C ) ∑(C )

o i

(2)

w i

Recently, undefined (but presumably organic) complexes of various trace metals that are sufficiently hydrophobic to partition into octanol have been detected in natural waters (7). Because the free ion activity of many metals in the aquatic environment is extremely low (8, 9), the hydrophobic, and potentially lipophilic fraction of metal may well represent the most biologically significant form in many circumstances (10, 11). To gain a qualitative and quantitative understanding of these species, we require a relationship between the measured overall octanol-water partition coefficient of a given trace metal, Dow, and the partition coefficients of its component species, (Kow)i, that is readily solvable. To this end, the current paper examines two approaches to the problem based on separate models that have been presented in the environmental and toxicological literature. Applications of each approach are exemplified using results of octanolwater partitioning experiments and speciation modeling for the strongly binding trace metals, Cu and Pb, in a variety of river water samples. First, however, and as a general framework for the models and results, the behavior of related species of a given trace metal in the octanol-water system is addressed.

Theory Conservation versus Interaction of Chemical Species in the Octanol-Water System. Figure 1 conceptualizes the reactivity and octanol-water partitioning of a divalent trace metal, Me, in the presence of different divalent complexing ligands. Note that the partitioning of ligands and their protonated forms may also take place, but this is not shown for clarity, and it is implicitly assumed that no chemical interactions take place in the octanol phase. In the first case (Figure 1a), an excess of hydrophobic ligand or a ligand forming a hydrophobic complex, L, is shown along with ambient hydrophilic ligands, X, which may be organic and (or) inorganic, and whose complexes, MeX, along with Me2+, have little propensity to partition into the solvent. Such conditions ([L] > [Me]) are typically required in simple model solutions in order to effect maximum complexation of the metal by L (12-14) but may also arise in natural waters under certain conditions. For example, L could represent a single VOL. 39, NO. 22, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8719

FIGURE 1. Conceptual representation of the octanol-water partitioning of trace metal species in (a) an interactive (nonconservative) system and (b) a noninteractive (conservative) system. Legend: Me ) divalent trace metal; L ) hydrophobic ligand in excess; L* ) strongly complexing hydrophobic ligand of limited availability; X ) ambient hydrophilic ligands; Kow ) octanol-water partition coefficient of an individual species. Broken arrows signify that octanol-water partitioning is small, and the single arrow shown in (b) represents the formation of a very strong complex. Model applications for each system are also annotated. or number of anthropogenic hydrophobic ligand(s) in a highly contaminated water body, or a generalized and nonspecific charged complexant, such as a humic substance, which is partly or completely neutralized at multiple binding sites by a number of metals. Depending on coordination and stoichiometry with L, there potentially exist different metal species of contrasting hydrophobicity, and for the case shown in Figure 1a it is reasonable to assume that the neutral species, MeL, is more hydrophobic than the charged form, MeL22-, which in turn is considerably more hydrophobic than remaining ambient species; that is, (Kow)MeL > (Kow)MeL22- . (Kow)MeX, (Kow)Me2+. Partial loss of MeL22- and, more significantly, MeL to octanol requires that these species are gained in the aqueous phase in order to reestablish equilibrium, and this is achieved at the expense of ambient hydrophilic species, including the free ion. A process of “extraction” ensues which may be accentuated or partly offset, depending on the relative hydrophobicity of the protonated ligand, (Kow)H2L (not shown in Figure 1a), hence the availability of L, with the possible consequence that new and different equilibria are established in the aqueous phase. Since individual species are nonconservative in the octanol-water system, a relationship between Dow and Kow cannot be gained from the mass balance of each species, and an approach that caters for chemical interactions in the aqueous phase (model 1, see below) is required. In the second case (Figure 1b), the availability of a single (or group of) hydrophobic ligand(s), L*, is limited ([Me] > [L*]). This ligand is strongly complexing and a single arrow is, therefore, used to denote the formation of its complex, which is assumed to have 1:1 stoichiometry and partitions between octanol and water according to the magnitude of (Kow)MeL*. Remaining aqueous metal speciation is, however, unaffected because of an excess of additional hydrophilic ligands, X (where [X] . [Me]), in the sample. This situation is more likely to arise in complex solutions than in simple model solutions and, in particular, in environmental samples. Here, trace concentrations of poorly defined, and often metalspecific and strongly complexing organic ligands, including components of the humic phase (8, 15, 16), appear to exist among a rich milieu of buffering anions and counterions. In such cases, the hydrophobic metal complex (or complexes) is (are) conserved in the octanol-water system, and hydrophilic and hydrophobic “pools” of metal are effectively 8720

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 22, 2005

independent, as indicated in Figure 1b by Me2+ + MeX and MeL*, respectively. Consequently, a noninteractive relationship between Dow and Kow may be gained from mass balance, as described in model 2 below. General Relationship between Dow and Kow - Model 1. A generally applicable relationship between Dow and (Kow)i that is valid for species of the same chemical that are either interactive (Figure 1a) or noninteractive (Figure 1b) may be gained by rewriting eq 2 in terms of Kow as follows:

Dow )

∑(C ) (K ∑(C )

ow)i

w i

(3)

w i

Thus, since the denominator in eq 3 represents the total concentration of the chemical (i.e. all species) in the aqueous phase at equilibrium with octanol, a molar fractional term, (Rw)i, may be introduced

(Rw)i )

(Cw)i

∑(C )

(4)

w i

and a weighted average expression derived:

Dow )

∑(R ) (K w i

ow)i

(5)

The practical difficulty with this relationship, hereafter termed model 1 and which has had widespread use in the environmental literature (4-6, 17, 18), is that (Rw)i represents the fractional concentration of a chemical species in the aqueous phase at equilibrium with octanol. Unlike the fractional concentration of such in the original, prepartitioned sample, this term is not straightforward to calculate or measure. For interacting species of the same chemical (Figure 1a), for instance, it relies on a knowledge of the partitioning of all related species, including the different complexants. In practice, prediction of either Dow, as an indicator of the hydrophobicity or lipophilicity of a mixture of related species, or Kow, defining the hydrophobicity or lipophilicity of an individual species, may rely on a number of assumptions or simplifications about aqueous phase speciation. Clearly, therefore, an alternative expression involving the fractional

concentrations of chemical species in the original sample is more desirable. Relationship between Dow and Kow for Noninteracting Species - Model 2. For a mixture of noninteracting, conservative species or pools (as defined in Figure 1b), the octanol-water partitioning of each species is independent, and the combined partitioning of all species is additive. In this case, a relationship between Dow and (Kow)i that is equivalent to the model presented by Verbruggen et al. (19) for mixtures of organic chemicals may be gained from the mass balance of each species

Qi ) (Cw)iVw + (Co)iVo

(6)

where Qi is the total quantity of a particular species, and Vw and Vo are the volumes of water and octanol, respectively. By rearranging this equation in terms of both (Co)i and (Cw)i, and thence introducing (Kow)i, the following expression for the overall octanol-water partition coefficient is obtained:



Vw Dow ) Vo

1+

(



Qi -



Qi Vw

)

Vo(Kow)i Qi Vw 1+ Vo(Kow)i

(7)

FIGURE 2. The overall octanol-water partition coefficient, Dow, arising from the mixture of two conservative chemical species ((Kow)1 ) 1; (Kow)2 ) 100), calculated as a function of the relative abundance of each species (r2 or 1-r1), and for different water to octanol volume ratios (Vw/Vo), using model 2 (eq 8).

Given that the (molar) fractional concentration of each species in the original, prepartitioned aqueous sample, Ri, is related to the corresponding total quantity of species, Qi, and that ΣRi ) 1, we may rewrite eq 7 as follows:



Vw Dow )

(

Vo 1 -

1+

Ri Vw

)

Vo(Kow)i Ri Vw 1+ Vo(Kow)i



(8)

This equation, hereafter referred to as model 2 and which is derived more explicitly in the Supporting Information (eqs S1-S8), highlights the dependence of Dow on the water to octanol volume ratio, Vw/Vo. Note that, although model 1 circumvents Vw/Vo, the magnitude of (Rw)i is implicitly dependent on this parameter. The effects of varying Vw/Vo on the Dow of hypothetical binary mixtures of noninteracting chemical species, calculated according to eq 8, are exemplified in Figures 2 and 3. Thus, in the former case, Dow is calculated as a function of the fractional composition of a mixture of two species of contrasting hydrophobicity ((Kow)1 ) 1; (Kow)2 ) 100) for ratios of Vw/Vo between 0.1 and 100. In the latter case, Dow is shown as a function of Vw/Vo for equimolar concentrations of two noninteracting species, one whose Kow is 100 and the other whose Kow is varied between 10-7 and 10. Clearly, the dependence of Dow on Vw/Vo in eq 8 precludes its use as a general predictor of the combined hydrophobicity or lipophilicty of noninteracting mixtures (20). With respect to mixtures of organic chemicals, alternative approaches based on experimental sum parameters have, therefore, been pursued (21). However, it is demonstrated in the current paper that, by measuring Dow as a function of Vw/Vo and fitting the data with model 2, the hydrophobic properties of trace metals in natural samples may be quantitatively defined.

Experimental Section Sampling. Water samples for this study were collected from the rivers Beaulieu (southern England) and Aire and Calder

FIGURE 3. The overall octanol-water partition coefficient, Dow, arising from an equal mixture of two conservative chemical species, calculated as a function of the water to octanol volume ratio (Vw/Vo), and for different hydrophobicities of one species, using model 2 (eq 8). (northern England). The Beaulieu is enriched in natural fulvic and humic substances because of its low-lying, forested catchment (22), while the Aire and Calder drain highly populated and industrialized catchments and are contaminated by a variety of organic and inorganic chemicals (23). One liter samples were collected in acid-cleaned, highdensity polyethylene bottles and stored below 10 °C before being filtered (within 24 h) through 0.45 µm pore size Millipore cellulose acetate filters using a Nalgene polysulfone filtration unit. Additional 100 mL sample aliquots for the analysis of dissolved organic carbon (DOC) were collected in ashed Pyrex bottles and were filtered through 0.7 µm pore size Whatman GF/F filters using a Sartorius Pyrex filtration kit. Experimental. For each sample, the octanol-water partitioning of Cu and Pb, the metals considered in this study, was determined according to the difference method outlined by Turner and Mawji (7). For concentrations of dissolved trace metal encountered in the environment, such an approach gives more reproducible results and is less prone to contamination than one based on the analysis of acid VOL. 39, NO. 22, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

8721

TABLE 1: Bulk Inorganic Composition of the Filtered River Water Samples Used in This Study, Both Before (C) and After (Cw) Sample Partition into Octanola C (Cw) constituent pH conductivity @ 20 °C, µS cm-1 chlorinity, mM alkalinity, mM SO42-, mM Na+, mM K+, mM Ca2+, mM Mg2+, mM

River Aire

River Beaulieu

River Calder

7.14 (7.14) 352 (351)

6.99 (7.00) 161 (162)

7.02 (7.03) 334 (339)

2.50 1.16 0.57 2.81 ( 0.02 (2.84 ( 0.03) 0.207 ( 0.002 (0.205 ( 0.003) 1.59 ( 0.01 (1.63 ( 0.03) 0.371 ( 0.004 (0.380 ( 0.005)

1.80 0.21 0.35 0.643 ( 0.008 (0.636 ( 0.006) 0.065 ( 0.001 (0.062 ( 0.002) 0.411 ( 0.003 (0.407 ( 0.010) 0.134 ( 0.003 (0.133 ( 0.003)

1.70 0.34 0.43 1.67 ( 0.15 (1.66 ( 0.02) 0.090 ( 0.008 (0.092 ( 0.001) 0.565 ( 0.009 (0.564 ( 0.003) 0.280 ( 0.002 (0.283 ( 0.002)

a Concentrations of the major cations are given as the mean and standard deviation of three separate experimental determinations. In all cases, there was no significant difference (p > 0.1) between C and Cw according to a two-tailed, paired t-test.

back-extracts of octanol. Briefly, six 5 mL aliquots of filtrate were pipetted into acid-cleaned 50 mL polyethylene centrifuge tubes at room temperature (20 ( 2 °C). To three aliquots were added 5 mL of 1-octanol (99%+, HPLC-grade; Aldrich), and the contents were shaken laterally at 100 rpm for 12-16 h, during which time we assume equilibrium is attained. Octanol and water were separated by centrifugation (3600 rpm for 30 min) before 4 mL aliquots of the aqueous phase were carefully pipetted into clean centrifuge tubes. The six aqueous aliquots were then analyzed within 24 h, and the fraction of metal in octanol, hence its overall octanolwater partitioning, was determined from the difference between the mean concentrations of the original aliquots and the aliquots having undergone octanol extraction, provided that the difference was significant (p < 0.05) according to a two-tailed t-test. Procedural blanks involving Millipore Milli-Q water (resistivity ) 18 MΩ cm) revealed that a small quantity of Cu was introduced to the aqueous phase from the solvent, and this was corrected for in the calculations. Results of control experiments employing Milli-Q water spiked with ionic metal standards indicated that adsorption and entrainment of metal ions and inorganic complexes were not significant. The effects of the sample to octanol volume ratio on metal partitioning were investigated by repeating experiments with different volumes of filtrate (up to 25 mL). Additional 20 mL sample aliquots were also processed with 20 mL aliquots of octanol as above and analyzed for pH, conductivity, major solutes (all < 0.45 µm filtrates), and DOC ( 0.1 according to a two-tailed, paired t-test). Also of significance, the results for Ca in the high alkalinity River Aire suggest that inorganic colloids (e.g. CaCO3 and Ca3(PO4)2) are not subject to entrainment or interfacial accumulation. In contrast, aqueous concentrations of trace metals (Cu and Pb) and dissolved organic matter (defined analytically as DOC), shown in Table 2 both before and after the partition experiments (C and Cw, respectively), indicate that variable proportions of these constituents (between about 3 and 20%) were sufficiently hydrophobic to partition into the solvent. Accordingly, the overall octanol-water partition coefficient, Dow, and the fractional loss to octanol, fo, of each constituent are also presented. With respect to Cu and Pb, values of Dow are at the lower end of those reported previously for a number of surface water environments (7, 24). Regarding DOC, Dow values are consistent with those reported for a variety of natural and commercial humic substances at circumneutral pH (25). Metal Speciation in River Water and Application of Model 1. According to the general Dow-Kow relationship defined by eq 5 (model 1), in order to determine the (Kow)i value(s) of trace metal species in river water that partition into octanol we require a knowledge of metal speciation in the aqueous phase at equilibrium with the solvent. Specifically, since a proportion of the organic rather than the inorganic component of the samples partitioned into octanol, definition of the organic complexation of Cu and Pb is required.

TABLE 2: Concentrations and Fractional Speciation of Cu and Pb and Dissolved Organic Matter (as DOC, and Comprising FS and HS) in Filtered River Watersa chemical constituent

C, M

DOC

3.53 × 10-4

Cu

4.76 ( 0.01 × 10-8

Pb

2.75 ( 0.06 × 10-9

DOC

7.88 ( 0.27 × 10-4

ri

Dow

fo

Cw, M

(rw)i

River Aire 0.029 0.028

3.43 × 10-4

CuFS ) 0.738 CuHS ) 0.238 CuX ) 0.023 Cu2+ ) 0.001

0.051

0.048

4.53 ( 0.05 × 10-8

CuFS ) 0.738 (0.761) CuHS ) 0.237 (0.213) CuX ) 0.024 (0.025) Cu2+ ) 0.001 (0.001)

PbFS ) 0.814 PbHS ) 0.094 PbX ) 0.083 Pb2+ ) 0.009

0.257

0.204

2.19 ( 0.02 × 10-9

PbFS ) 0.811 (0.821) PbHS ) 0.094 (0.082) PbX ) 0.085 (0.087) Pb2+ ) 0.010 (0.010)

River Beaulieu 0.097 0.089

7.18 ( 0.21 × 10-4

Cu

2.28 ( 0.02 ×

10-8

CuFS ) 0.833 CuHS ) 0.166 CuX ) 0.001 Cu2+ < 0.001

ns

ns

2.26 ( 0.07 × 10-8

CuFS ) 0.833 (0.899) CuHS ) 0.166 (0.100) CuX ) 0.001 (0.001) Cu2+ < 0.001 (