Predicting Soil−Water Partitioning of Polycyclic ... - ACS Publications

May 4, 2001 - We evaluated a method to determine organic carbon- normalized soil-water partition coefficients (KOC) of 20 PAHs and 12 PCBs by desorpti...
0 downloads 0 Views 106KB Size
Environ. Sci. Technol. 2001, 35, 2319-2325

Predicting Soil-Water Partitioning of Polycyclic Aromatic Hydrocarbons and Polychlorinated Biphenyls by Desorption with Methanol-Water Mixtures at Different Temperatures MARTIN KRAUSS* AND WOLFGANG WILCKE Institute of Soil Science and Soil Geography, University of Bayreuth, D-95440 Bayreuth, Germany

We evaluated a method to determine organic carbonnormalized soil-water partition coefficients (KOC) of 20 PAHs and 12 PCBs by desorption in the presence of a cosolvent (methanol fractions of 0.1-0.9) and at different temperatures (20-80 °C). The KOC values, the deviation factor from ideal sorption R, and the desorption enthalpies ∆Hdes were estimated by nonlinear regression of log KOC on the methanol fractions and on T. The KOC values of individual compounds varied up to a factor of 100 among the studied 11 urban soils. The calculated R and ∆Hdes of individual compounds varied considerably among the soils (coefficients of variation 5-20% and 20-30%, respectively), R increased with increasing hydrophobicity of the compounds. A sequential extraction with four temperature/methanol fraction combinations followed by a nonlinear regression allowed for the direct determination of the KOC, R, and ∆Hdes. The use of less temperature/methanol fraction combinations requires a suitable estimation of R and ∆Hdes, as their choice may change the obtained KOC values by up to a factor of 10. The proposed method is suitable for a routine determination of KOC values of PAHs and PCBs for small soil samples (2-6 g) and low concentrations (down to 0.3 mg kg-1 of ∑20 PAHs and 1.2 µg kg-1 of ∑12 PCBs).

Introduction The distribution of hydrophobic organic contaminants (HOCs) between solution and solid phase determines their availability for degradation, uptake, or leaching and their ecotoxicity in soils (1, 2). The most common approach to describe soil-water distribution involves the use of the organic carbon-normalized equilibrium distribution coefficient KOC. Originally, the KOC was introduced as a compound-specific constant, based on the equilibrium partitioning model and the assumed analogy between hydrophobic sorption and partitioning in nonmiscible solvent-water systems (3, 4). In recent years, substantial evidence was presented that the equilibrium partitioning model is not sufficient to describe sorption of HOCs in soils, as organic matter is heterogeneous and sorption is governed by nonequilibrium processes (5* Corresponding author telephone: ++49 921 55 2318; fax: ++49 921 55 2246; e-mail: [email protected]. 10.1021/es001616r CCC: $20.00 Published on Web 05/04/2001

 2001 American Chemical Society

8). Therefore, KOC values were found to vary up to a factor of 100 between different soils or sediments. This variation depends on the contact time of HOCs with the sorbent (5) and the composition of soil organic matter (6-8). Furthermore, the KOC values differ between different HOC classes at a given KOW (9, 10). In particular, polycyclic aromatic hydrocarbons (PAHs) sorb to a larger extent to highly aromatic compounds (“soot” or “black carbon”) than polychlorinated biphenyls PCBs (10-12). If dissolved organic matter (DOM) or nonsettling colloids are present, the observed KOC values are lower than those expected from a system with sorbed and truly dissolved HOCs alone (13, 14). In the initial partitioning model, colloids or DOM were not considered. The apparent decrease in KOC when DOM is present is more pronounced for highly hydrophobic compounds because of their high affinity for DOM (15, 16). Thus, KOC values should be determined for each given soil and compound individually. However, the routine determination of KOC values by measuring concentrations in solution and solid phase in batch experiments has some shortcomings. For highly hydrophobic compounds, the concentrations in solution are often below detection limits. This may only be overcome by using a larger amount of soil and solution. If particle size or density fractions of soils or sediments are investigated, for which time-consuming and expensive preparation methods are necessary (e.g., refs 17-20), the use of large amounts of sample is impossible. Furthermore, the complete separation of solid and solution phase and a removal of colloids is difficult (13). Additionally, sorption to the vessels may create analytical artifacts. In soil-water systems, the time to reach near-equilibrium conditions may be in the scale of several months, particularly in desorption experiments (21). An easy routine method for the determination of KOC values should therefore (i) reduce the influence of sorption on colloids and vessels and (ii) allow for a fast achievement of equilibrium conditions. When only small amounts of samples are available, it should (iii) also produce detectable dissolved HOC concentrations. One possible way to meet these requirements is the use of a water-miscible cosolvent to accelerate desorption and increase equilibrium concentrations in the solution phase (22, 23). The theoretical framework to quantify the influence of cosolvents on the partitoning of organic contaminants was proposed by Rao et al. (24). Their “solvophobic approach” was successfully applied by several authors to describe HOC sorption equilibria and dynamics in soils and sediments (e.g., refs 22-29) and to predict KOC values of highly hydrophobic compounds (10, 30). A brief description of this theory is given in the next section. Furthermore, the use of elevated temperatures in desorption experiments also increases the HOC release rate from soil, shifts in most cases the equilibrium toward the solution phase and additionally allows for the calculation of desorption enthalpies as a measure for the binding strength of HOCs to soil (31). The objective of this study is to evaluate a method to determine KOC values in small field-contaminated soil samples by desorption in the presence of cosolvents and at different temperatures.

Theoretical Considerations The solvophobic theory was initially developed to describe the solubility of organic compounds in cosolvent-water mixtures (32). The mole fraction solubility Xmix of a compound VOL. 35, NO. 11, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2319

without polar groups in a cosolvent-water mixture is given by

log Xmix ) log Xwater +

fc(γw - γs) × TSA 2.303kT

(1)

where Xwater denotes the mole fraction solubility in pure water, fc is the volume fraction of cosolvent in the mixture, k is the Boltzmann constant (1.38 × 10-23 J K-1), and T is the temperature (K). γw and γs are the interfacial free energies between the solutes hydrophobic surface and water or cosolvent, respectively (J nm-2). TSA is the total molecular surface area of the compound (nm2). The partition coefficient of HOCs between organic carbon and pure water (KOC,water) can be estimated by regression from Xwater, the entropy of fusion ∆Sf (J mol-1 K-1), and the melting point Tm (K) of the compound (33):

log KOC,water ) -R log Xwater -

∆Sf(Tm - T) +β 2.303RT

(2)

where R and β are empirical constants and R is the gas constant (8.314 J mol-1 K-1). The partition coefficient between soil organic carbon and a water-solvent mixture (KOC,mix) can be expressed analogously as

log KOC,mix ) -R log Xmix -

∆Sf(Tm - T) +β 2.303RT

(3)

Combining eqs 1-3 gives an equation to assess KOC values as a function of the cosolvent fraction in solution, substituting (γw - γs) by ∆γ (24):

fc∆γ × TSA log KOC,mix ) log XOC,water - R 2.303kT

(4)

Assuming that partitioning of HOCs between soil and water is similar to a solvent-water partitioning process, the temperature dependency of KOC,water or KOC,mix in a desorption experiment can be described by (31)

d ln KOC ∆Hdes )d 1/T R

(5)

where ∆Hdes is the enthalpy of desorption (kJ mol-1). Assuming that ∆Hdes is independent of temperature in the investigated range (293-353 K), integration and conversion to decadic logarithms yields

log KOC(T) ) log KOC(T0) -

(

∆Hdes 1 1 2.303R T T0

)

(6)

To describe log KOC,mix as a function of T and fc, we combined eqs 4 and 6:

log KOC,mix(T) ) log KOC,water(T0) -

(

)

Hdes 1 fc∆γ × TSA 1 (7) -R 2.303R T T0 2.303kT Using eq 7, it will be possible to calculate KOC,water values of HOCs at room temperature T0 from the KOC,mix value obtained from desorption with methanol-water at an elevated temperature T. For the application of eq 7, we had to make several assumptions regarding the parameters TSA, ∆γ, R, and ∆Hdes. Total surface areas (TSA) are calculated from van der Waals radii and bonding lengths of the molecules (32), which are considered as constant at the used temperature range. The difference of the interfacial free energies ∆γ is specific for a 2320

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 11, 2001

particular organic cosolvent-water system and independent of fc (32). Although surface tensions decrease with increasing temperature, these differences can be neglected at the investigated range. Values reported for ∆γ of the methanolwater system are 1.99 × 10-20 (25), 2.16-2.30 × 10-20 (29), and 2.37 × 10-20 J nm-2 (32). We selected the mean value of 2.16 × 10-20 J nm-2 for ∆γ in our study. The choice of the remaining parameters R and ∆Hdes is more difficult. If these are known and constant for different soils, a determination of KOC,mix at one particular temperature and methanol fraction is sufficient to calculate log KOC,water at room temperature using eq 7. If R and ∆Hdes vary for different soils, we can estimate them from eq 7 if we determine KOC,mix at least at three different methanol fraction-temperature combinations. The empirical parameter R represents the deviation of the sorption - fc relation from an “ideal” solubility - fc relation according to eq 1 (R ) 1). This deviation is caused by cosolvent-sorbent interactions. Conceptually, R values smaller than 1 are attributed to stronger resistance of compounds in the soil matrix against desorption by the cosolvent-water mixture. Values reported for R vary over a wide range. While Bouchard (23) and Brusseau et al. (22) found that R is close to 1 for sorption of PAHs to soils from methanol-water mixtures, Fu and Luthy (27) and Kimble and Chin (26) calculated considerably smaller values down to 0.35 in sorption experiments with various HOCs. The latter stated that R depends on sorbate properties and type and amount of cosolvent used. Walters and Guiseppi-Elie (28) found that R increases with increasing log KOW of the compounds. Spurlock and Biggar (34) questioned the use of constant R values, as they seem to be influenced by a variety of effects. They showed that the cosolvent type used had a large influence on the size and variation of R. Sorption or desorption enthalpies (∆Hs or ∆Hdes) of HOCs are generally considered to be small, because hydrophobic sorption is mainly an entropy-driven process (31, 35). Sorption enthalpies should increase with increasing molecular size, as London-van der Waals interactions are the predominant enthalpy-related force (31). The rarely reported values of HOC sorption enthalpies vary considerably (overview given by ten Hulscher and Cornelissen, ref 31) and cover the whole range from -25 to +25 kJ mol-1.

Experimental Design Due to the empirical nature of R and the reported variation of both R and ∆Hdes, it is necessary to determine these parameters in desorption experiments under different conditions and for different soils. For our study, we selected methanol as the organic cosolvent, because the methanolwater system shows a linear log KOC,mix - fc relation (25) and sorbent-solvent interactions seem to have less influence as compared to other solvents (34). Additionally, many solubility and sorption data are available from other studies for comparison (22-30). We conducted experiment 1 with soil A2 (0-5 cm, alluvial grassland, 39.9 g kg-1 SOC, 7.2 mg kg-1 ∑20 PAHs, 58.1 µg kg-1 ∑12 PCBs) using solutions with methanol fractions of 0.1, 0.25, 0.5, 0.75, and 0.9 at 20, 40, 60, and 80 °C. This approach was chosen to investigate the applicability of the suggested model (eq 7) and to determine R and ∆Hdes for different compounds at various temperatures and methanol fractions. Additionally, KOC,water values were directly determined for soil A2 by desorption with demineralized water. In experiment 2, we extended the database to a set of 11 soils (including also soil A2), while we reduced the number of methanol fractions and temperature steps. We used a sequential desorption procedure to determine KOC,mix values at four methanol fraction-temperature combinations (0.35,

40 °C; 0.65, 40 °C; 0.35, 60 °C; 0.65, 60 °C). On the basis of this data set, we examined the variation of R and ∆Hdes in different soils and calculated KOC,water values at 20 °C by nonlinear regression. In experiment 3, we determined KOC,mix at a fc ) 0.5 and 60 °C assuming constant values for R and ∆Hdes and calculated KOC,water values at 20 °C for these 11 soils.

Materials and Methods Soils. We used 11 urban or peri-urban topsoils (0-5 cm) sampled in the city of Bayreuth (Northern Bavaria, Germany, 76 000 inhabitants) or in the surrounding area up to 7 km away from the city center. These include one forest stand (F2), one roadside (R2), three house gardens (G2, G3, and G7), two alluvial grassland sites near the Rotmain river (A2 and A4), one park area (P2), one agricultural site (AG1), one former landfill (I3), and one former gaswork site (I4). Soil samples were air-dried, sieved to 6) were lower than expected from all calculation methods. As equilibrium may not have been reached after 8 d of desorption, the KOC values may have been generally overestimated. However, this effect compensated for more hydrophobic compounds by their association with DOM or colloids. With our approach, it is not possible to distinguish between methanol- and DOMinduced desorption. However, we consider the DOM-induced desorption small as compared with the methanol-induced desorption, because the solubility enhancement of methanol should outweigh the effect of DOM largely (30).

To detect systematic differences between the different calculation methods, we plotted the KOC values of all compounds obtained from experiment 2 (sequential extraction and regression analysis) for the 11 soils against those from experiment 3 (single extraction at 60 °C and fc ) 0.5; Figure 3). In Figure 3a, compound-specific R and a constant ∆Hdes of -10 kJ mol-1 were assumed. This led, on average, to a slight overestimation of KOC values as compared with those obtained in the sequential extraction (experiment 2), as indicated by the intercept of the regression line as compared to the 1:1 line. If both constant R ) 0.85 and ∆Hdes ) -10 kJ mol-1 were used, the differences between KOC values from experiment 3 and KOC values from the sequential extraction (experiment 2) increased for compounds with low KOC values (Figure 3b), indicated by a smaller slope of the regression line. In most cases, the difference between the respective KOC values calculated with both methods was smaller than 0.5 log unit, but in some cases it was even larger than 1 log unit. These differences must be attributed to the use of constant R and ∆Hdes in experiment 3, besides random variations of the analytical method. In experiment 2, ∆Hdes (average of all compounds) ranged from -5 to -15 kJ mol-1 between different soils. This would result in variabilities of approximately (0.5 log unit as compared to a constant ∆Hdes of -10 kJ mol-1. The variability of R between different soils of 0.4 unit would also result in differences of (0.5 log unit. The use of a too large R for all low molecular weight PAHs is responsible for the systematic differences observed in Figure 3b. Applicability of the Method. Our study showed that the desorption with methanol-water mixtures at elevated temperatures is suitable to determine KOC,water values of PAHs VOL. 35, NO. 11, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2323

schaft (DFG Ze 154/38-1, -3), which we gratefully acknowledge. W. Wilcke works under a Heisenberg grant of the Deutsche Forschungsgemeinschaft.

Literature Cited

FIGURE 3. Plot of the KOC,water values calculated from experiment 2 (sequential extraction) against KOC,water values calculated from experiment 3 (desorption at 60 °C with methanol-water 1:1) using (a) constant ∆Hdes ) -10 kJ mol-1 and compound-specific r from Table 2 or (b) constant ∆Hdes ) -10 kJ mol-1 and constant r ) 0.85. and PCBs in soils with naturally aged PAHs and PCBs showing concentrations down to 0.3 mg kg-1 of ∑20 PAHs and 1.2 µg kg-1 of ∑12 PCBs and with samples of 2-6 g. The proposed sequential extraction with four temperature/methanol fraction combinations followed by a nonlinear regression seems a reliable tool within a moderate time and cost scale. This approach additionally allows for the calculation of R and ∆Hdes. The latter probably indicates sorption strength for different compounds or soils, but this assumption remains to be tested by further studies. If less data points are used for the determination of KOC,water, a proper choice of R and ∆Hdes is necessary. Both parameters vary between different soils, and additionally, R depends on the hydrophobicity of the compounds. In this case, additional experiments should be conducted to check the suitability of the chosen R and ∆Hdes. All methods for the determination of KOC,water values are operational and depend on the particular method chosen. In reality, the solid-water disribution of contaminants varies in time and space and the in situ conditions may deviate considerably from equilibrium. Nevertheless, the KOC value is still the most widely used predictor of HOC mobility and bioavailability, and simple methods to determine soil- and compound-specific KOC values as proposed here are needed.

Acknowledgments We thank W. Zech for his important support and A. Bergmann for her help with the analyses. We are indebted to the Bayreuther Institut fu ¨ r Terrestrische O ¨ kosystemforschung (BITO ¨ K) for providing the accelerated solvent extractor (ASE). This study was funded by the Deutsche Forschungsgemein2324

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 11, 2001

(1) Alexander, M. Environ. Sci. Technol. 1995, 29, 2713-2717. (2) Reid, B. J.; Jones, K. C.; Semple, K. T. Environ. Pollut. 2000, 108, 103-112. (3) Karickhoff, S. W.; Brown, D. S.; Scott, T. A. Water Res. 1979, 13, 241-248. (4) Chiou, C. T.; Porter, P. E.; Schmedding, D. W. Environ. Sci. Technol. 1983, 17, 227-231. (5) Pignatello, J. J.; Xing, B. Environ. Sci. Technol. 1996, 30, 1-11. (6) Rutherford, D. W.; Chiou, C. T.; Kile, D. E. Environ. Sci. Technol. 1992, 26, 336-340. (7) Kile, D. E.; Wershaw, R. L.; Chiou, C. T. Environ. Sci. Technol. 1999, 33, 2053-2056. (8) Lueking, A. D.; Huang, W.; Soderstrom-Schwarz, S.; Kim, M.; Weber, W. J., Jr. J. Environ. Qual. 2000, 29, 317-323. (9) Chiou, C. T.; McGroddy S. E.; Kile, D. E. Environ. Sci. Technol. 1998, 32, 264-269. (10) Jonker, M. T. O.; Smedes, F. Environ. Sci. Technol. 2000, 34, 1620-1626. (11) Gustafsson, O ¨ .; Haghseta, F.; Chan, C.; MacFarlane, J.; Gschwend, P. M. Environ. Sci. Technol. 1997, 31, 203-209. (12) Bucheli, T. D.; Gustafsson, O ¨ . Environ. Sci. Technol. 2000, 34, 5144-5151. (13) Gschwend, P. M.; Wu, S. Environ. Sci. Technol. 1985, 19, 90-96. (14) McCarthy, J. F.; Zachara, J. M. Environ. Sci. Technol. 1989, 23, 496-502. (15) McCarthy, J. F.; Jimenez, B. D. Environ. Sci. Technol. 1985, 19, 1072-1076. (16) Maxin, C. R.; Ko¨gel-Knabner, I. Eur. J. Soil Sci. 1995, 46, 193204. (17) Guggenberger, G.; Pichler, M.; Hartmann, R.; Zech, W. Z. Pflanzenernaehr. Bodenk. 1996, 159, 565-573. (18) Kukkonen, J.; Landrum, P. F. Chemosphere 1996, 32, 10631076. (19) Wilcke, W.; Zech, W.; Kobza, J. Environ. Pollut. 1996, 92, 307313. (20) Cornelissen, G.; van Zuilen, H.; van Noort, P. C. M. Chemosphere 1999, 38, 2369-2380. (21) Kan, A. T.; Fu, G.; Tomson, M. B. Environ. Sci. Technol. 1994, 28, 859-867. (22) Brusseau, M. L.; Wood, A. L.; Rao P. S. C. Environ. Sci. Technol. 1991, 25, 903-910. (23) Bouchard, D. C. J. Contam. Hydrol. 1998, 34, 107-120. (24) Rao, P. S. C.; Hornsby, A. G.; Kilcrease, D. P.; Nkedi-Kizza, P. J. Environ. Qual. 1985, 14, 376-383. (25) Nkedi-Kizza, P.; Rao, P. S. C.; Hornsby, A. G. Environ. Sci. Technol. 1985, 19, 975-979. (26) Kimble, K. T.; Chin, Y.-P. J. Contam. Hydrol. 1994, 17, 129-143. (27) Fu, J.-K.; Luthy, R. G. J. Environ. Eng. 1986, 112, 346-366. (28) Walters, R. W.; Guiseppi-Elie, A. Environ. Sci. Technol. 1988, 22, 819-825. (29) Woodburn, K. B.; Rao, P. S. C.; Fukui, M.; Nkedi-Kizza, P. J. Contam. Hydrol. 1986, 1, 227-241. (30) Hegeman, W. J. M.; van der Weijden, C. H.; Loch, J. P. G. Environ. Sci. Technol. 1995, 29, 363-371. (31) ten Hulscher, T. E. M.; Cornelissen, G. Chemosphere 1996, 32, 609-626. (32) Yalkowsky, S. H.; Valvani, S. C.; Amidon, G. L. J. Pharm. Sci. 1976, 76, 1488-1494. (33) Karickhoff, S. W. J. Hydraul. Eng. 1984, 110, 707-734. (34) Spurlock, F. C.; Biggar J. W. Environ. Sci. Technol. 1994, 28, 1003-1009. (35) Hassett, J. J.; Banwart W. L. In Reactions and Movement of Organic Chemicals in Soils; Saxhney, B. L., Brown, K., Eds.; Soil Science Society of America Special Publications 22; Soil Science Society of America: Madison, WI, 1989; pp 31-44. (36) Krauss, M.; Wilcke, W.; Zech, W. Environ. Sci. Technol. 2000, 34, 4335-4340. (37) Ballschmiter, K.; Zell, M. Fresenius Z. Anal. Chem. 1980, 302, 20-31. (38) Mackay, D.; Shiu, W. Y.; Ma, K. C. Illustrated Handbook of Physical-Chemical Properties and Environmental Fate for Organic Chemicals. Vol. II: Polynuclear Aromatic Hydrocarbons, Polychlorinated Dibenzodioxins and Dibenzofurans; Lewis Publishers: Boca Raton, FL, 1992.

(39) Hawker, D. W.; Connell, D. W. Environ. Sci. Technol. 1988, 22, 382-387. (40) Pearlman, R. S.; Yalkowsky, S. H.; Banerjee, S. J. Phys. Chem. Ref. Data 1984, 13, 555-562. (41) McGroddy, S. E.; Farrington, J. W.; Gschwend, P. M. Environ. Sci. Technol. 1996, 30, 172-177. (42) Mackay, D.; Shiu, W. Y.; Ma, K. C. Illustrated Handbook of Physical-Chemical Properties and Environmental Fate for Organic Chemicals. Vol. I: Monoaromatic Hydrocarbons, Chlorobenzenes, and PCBs; Lewis Publishers: Boca Raton, FL, 1992.

(43) Means, J. C.; Wood, S. G.; Hassett, J. J.; Banwart W. L. Environ. Sci. Technol. 1980, 14, 1524-1528. (44) Girvin, D. C.; Scott, A. J. Chemosphere 1997, 35, 2007-2025.

Received for review August 23, 2000. Revised manuscript received January 29, 2001. Accepted February 20, 2001. ES001616R

VOL. 35, NO. 11, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2325