Environ. Sci. Technol. 2000, 34, 973-979
Free and Bound Benzotriazoles in Marine and Freshwater Sediments C H R I S T O P H E R M . R E D D Y , * ,† JAMES G. QUINN, AND JOHN W. KING Graduate School of Oceanography, University of Rhode Island, Narragansett, Rhode Island 02882
To investigate how some anthropogenic compounds are sequestered in sediments, we examined the free and bound fractions of six different substituted benzotriazoles (BZTs) in sediment cores from the Pawtuxet River and Narragansett Bay. The free fraction was operationally defined as the fraction of BZTs that was removed with several organic solvent extractions, and the bound fraction was that portion of BZTs removed by solvent extraction after saponifying the sediment residue remaining from the initial solvent extractions. The total concentrations (free and bound) of BZTs were as large as 10 mg g-1 in the riverine core and 0.05 mg g-1 in the estuarine core. The percent bound of the BZTs ranged from 0 to 9% of the total and varied with each compound, sediment depth, and location. BZTs that had alkyl substitution at the 3′ position were less likely to be found in the bound fraction than compounds that did not have this substitution. On the basis of these results, it appears that these compounds may be chemically associated with the sediments, and this association is hindered by alkyl substituents on some of the BZTs. These results have important implications in understanding the bioavailability and geochemical fate of organic contaminants in sediments as well as the basic reactions of sedimentary organic matter.
Introduction Understanding the fate of hydrophobic organic contaminants (HOCs) in sediments is important when assessing their bioavailability, toxicity, and remediation, and over the past few decades, thousands of marine and freshwater sediments have been analyzed to address this issue. These studies, in almost all cases, have only measured the free or unbound fraction, i.e., the amount removed when the sample is extracted with a single organic solvent (e.g., methanol, acetonitrile, methylene chloride, chloroform, or hexane) or a mixture of solvents. However, another fraction may be released when the residual sediment, after solvent extraction, is hydrolyzed and then extracted by an organic solvent; this fraction is called the bound or nonextractable fraction. Although there have been many studies in marine and lacustrine sediments on the free and bound fractions of biogenic compounds such as fatty acids, fatty alcohols, fatty aldehydes, and sterols (1-7), surprisingly the “bound” fraction of organic contaminants in sediments or suspended matter has only been studied a few times (8-11). However, a wide range of compounds, which have different sources * Corresponding author e-mail:
[email protected]; telephone: (508)289-2316; fax: (508)457-2164. † Present address: Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, MA 02543. 10.1021/es990971i CCC: $19.00 Published on Web 02/09/2000
2000 American Chemical Society
and environmental chemistries, have been observed in the bound fraction in these limited studies. In addition, the percentage of the total contaminant that was bound has been shown to vary depending on the structure of the compound; for example, Bellar and Simoneit (8, 9) reported that >99% of the total hexachlorophene was bound while less than 1% of the total n-alkanes (n-C17-n-C31) were bound. These differences suggest that different mechanisms are responsible for binding the various contaminants. For hexachlorophene, the authors (8, 9) suggested that it was covalently bound through one of its hydroxy groups to humic material. Conversely, they suggested that the n-alkanes may have been “entrapped” or caged in humic material, which may act as a molecular sieve (12). In another study, Richnow et al. (11) detected bound chlorinated benzoic acids in riverine, seawater, and sediment humic substances and suggested that these residues are metabolites of polychlorinated biphenyls. These results show that the binding of toxic organic contaminants or their metabolites to sediments may play an important role in the biogeochemistry of these compounds. To expand on this previous work, we investigated six compounds from a class of synthetic organic chemicals called benzotriazoles (BZTs) (Figure 1). These compounds were produced from 1961 to 1985 by a major chemical plant located on the Pawtuxet River, RI. They were used as ultraviolet light absorbers to protect polymers from photochemical deterioration. Previous research has used these compounds as specific tracers of inputs from the Pawtuxet River into Narragansett Bay sediments (14-16). Little is known about their ultimate fate in sediments (i.e., microbial degradation). We chose to study the BZTs because (a) they are highly enriched in the sediments of the Pawtuxet River and Narragansett Bay, with µg g-1 to mg g-1 concentrations in the free fraction (14-16); (b) they have a hydroxy group, which has been shown to be a potential site for a chemical association to organic matter (8, 9, 11); and (c) they have varying degrees of alkyl substitution, which can be used to investigate if certain substituents hinder the extent of chemical association.
Materials and Methods Sample Locations and Collection. The chemical plant that manufactured the BZTs was located on the Pawtuxet River (Figure 2), which flows into the Providence River section of Narragansett Bay via Pawtuxet Cove. A dam at the end of the river prevents saltwater intrusion. Two sediment cores were examined in this study: one from the Pawtuxet River and the other in Narragansett Bay (Figure 2). The Pawtuxet River core was collected in 1989, sectioned at 2-3-cm intervals, and then stored at -20 °C (17). Eleven sections from 0-2 cm to as deep as 50-52 cm were analyzed in this study. The sedimentation rates in this section of the river are 2-3 cm yr-1 (18). These sediments are mostly sandy silts with some intervals at 0-2 and 45-47 cm (not analyzed) where they are silty sands. The redox discontinuity, determined visually, was in the top 2 cm of this core. The Narragansett Bay core was collected in 1997 and was processed as described above. Six sections from the top 13 cm of the core were analyzed (19). Salinity in the overlying water in this area is ∼27 ppt. Sediments in this area are clayey silts, become anoxic within a few millimeters of the surface, and have a sedimentation rate of ∼0.3 cm yr-1 (17). In this study, the deepest sections of both cores were the approximate depths of where the BZTs were no longer detected and should be roughly equivalent to the initial date VOL. 34, NO. 6, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
973
FIGURE 1. Structures of the six BZTs investigated in this study. Listed below each compound is an estimate of the logarithm of the octanol-water partition coefficient (log Kow) and the acid dissociation constant (pKa) of the hydroxy group (13).
FIGURE 2. Map of sample locations. The Pawtuxet River core was collected in 1989. The Narragansett Bay core was collected in 1997. of production of these compounds (1961-1970) (14). After acidification, organic carbon and nitrogen were measured on every sediment sample with an elemental analyzer. The precision on triplicate analysis, as expressed as the relative 974
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 6, 2000
standard deviation (RSD), was less than 5% for both carbon and nitrogen. Free Fraction. Wet sediment (∼0.5-5 g) was added to a 22-mL glass vial with a Teflon-lined cap and then spiked
with the internal standard, androstanol. Fifteen milliliters of a 60/40 mixture of acetonitrile/hexane was added to the vial. The vial was capped, shaken for 10 s, and then placed in a boiling water bath for 30 min (the solvents were one phase at 100 °C). After being cooled, the vial was centrifuged, and the liquid layers (organic and aqueous phases) were drawn off into a 125-mL separatory funnel leaving the sediment residue in the vial. The funnel contained 50 mL of deionized water adjusted to a pH of 4, which would ensure that the phenolic hydroxy groups on the BZTs would be protonated. The sediment residue was then extracted four more times in the same manner. (In our preliminary experiments, when we analyzed each extract after each extraction, we found that a total of four extractions completely removed the free contaminants.) The liquid layers were also added to the original separatory funnel. The hexane layer was removed from the separatory funnel, and the remaining aqueous/ acetonitrile solution was extracted with 25 mL of hexane two more times. All of the hexane extracts were combined, rotaryevaporated to dryness, and acetylated. This fraction is the operationally defined free fraction. Bound Fraction. The sediment residue from the above procedure was spiked with androstanol and saponified in 2 N KOH in 90/10 methanol/water in a boiling water bath for 4 h. After being cooled, the pH of the mixture was adjusted to ∼4 with 4 N HCl, and the mixture was extracted three times with hexane. All hexane extracts were combined, rotaryevaporated, and acetylated. This fraction is the operationally defined bound fraction. Acetylation and Chromatography. Samples were acetylated with acetic anhydride and pyridine (19), and the resulting product was solvent exchanged into hexane. The hexane extract was chromatographed, using nitrogen pressure, on a 0.5 cm (i.d.) × 15 cm column containing activated silica gel, and two fractions were collected. The first fraction (F1) was eluted with 15 mL of an 80/20 mixture of hexane/ methylene chloride and not analyzed in this study. The second fraction (F2), which contained the benzotriazole acetates, was eluted with 15 mL of a 50/50 mixture of methylene chloride/acetonitrile, rotary-evaporated, spiked with an external standard, o-terphenyl, and then injected onto the GC/MSD. Apparatus. A Hewlett-Packard 5890 series II gas chromatograph equipped with a Hewlett-Packard 5971 mass selective detector (GC/MSD), operating in the selective ion monitoring (SIM) mode, was used to analyze extracts (19). Compounds, as acetate derivatives, were separated with a 30-m J&W Scientific DB-XLB fused-silica capillary column (0.25 mm i.d. and 0.25 µm film thickness). The quantitation ions for the acetate derivatives of androstanol, C1-BZT, C4-BZT, C4-Cl-BZT, C5-Cl-BZT, C8-Cl-BZT, and C10-BZT were m/z 243, 225, 267, 286, 300, 322, and 342, respectively. The GC/MSD was calibrated with acetylated standards of C1-BZT, C10-BZT, C8-Cl-BZT, and androstanol. The BZT standards were obtained under their trade names Tinuvin P, Tinuvin 327, and Tinuvin 328, respectively. The other BZTs were tentatively identified by their mass spectra and relative retention times (14), and the response factor of acetylated C8-Cl-BZT was used for quantifying these BZTs. All concentrations are reported on a dry weight basis. Quality Control. Laboratory blanks for both the free and bound fractions were at least three times below the lowest sample values. The relative percent difference (RPD) for duplicate samples was 3).
Results and Discussion Narragansett Bay Core. Organic carbon (OC), nitrogen, and atomic carbon to nitrogen ratios (C/N) are shown in Table 1. OC and nitrogen decreased down-core from 5.3 and 0.38% at 0-2 cm to 2.9 and 0.17% at 10-13 cm, respectively. C/N ratios were from 16 to 20 and increased down-core as expected. C1-BZT, C10-BZT, and C8-Cl-BZT were substantially more concentrated than C4-BZT, C4-Cl-BZT, and C5-Cl-BZT in these samples, and only their results are shown in Figure 3. This is consistent with other studies on these compounds in Narragansett Bay that have shown that these BZTs are the most adundant (14-16). All three compounds were detected at trace levels in the 10-13-cm section, and their concentrations generally increased up-core. C10-BZT and C8-Cl-BZT were more abundant, with concentrations as high ∼25 µg g-1. C1-BZT was less concentrated with a maximum value of 10.5 µg g-1 at 4-6 cm. VOL. 34, NO. 6, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
975
FIGURE 4. Narragansett Bay core (a) concentration of free C1-BZT, (b) concentration of bound C1-BZT, and (c) bound C1-BZT (%). The only BZT detected in the bound fraction was C1-BZT. This is the first time that any bound synthetic organic chemical has been detected in the sediments of Narragansett Bay. Relative to the total (free + bound), the percent bound was ∼9% at the surface and decreased to ∼2% at 8-10 cm (Figure 4). The concentration of bound C1-BZT was 614 ng g-1 at the surface, decreased down-core to 26.2 ng g-1 at 8-10 cm, and was not detected at 10-13 cm (Figure 4). There was no change in the sediment profile of bound C1-BZT when normalized to organic carbon. We also have examined several other sediment surface and core samples from Narragansett Bay (20) and found the same trend in that C1-BZT was the only bound BZT even though C10-BZT and C8-Cl-BZT were more abundant on a total basis. For most biogenic compounds, such as sterols (5), it has usually been found that the percent bound increases with depth relative to the percent free. This trend is attributed to a simple conversion of the free to the bound and/or the removal of the less stable free compound relative to the more stable bound compound. However, the main difference between most biogenic compounds and the BZTs is that the input of the biogenic compounds is relatively constant on a yearly basis, whereas the production of BZTs was much less constant. Hence, this makes it is very difficult to elucidate when and how fast the BZTs were being bound. When examining these results, two important questions were considered: (a) Is the bound C1-BZT an inadvertent product of the method? That is, could the C1-BZT not be bound but only an artifact of the sample extraction, i.e., insufficient removal or desorption of the free C1-BZT from the sediment. There are two compelling arguments against this possibility. First, we checked the extraction method and are confident that five extraction steps exhaustively removed the BZTs from the sediment (see Materials and Methods section). Consider that in the top 4 cm C10-BZT and C8-Cl-BZT were twice as abundant as the C1-BZT, hence an inefficient extraction would have left more of the C10-BZT and C8-Cl-BZT behind. Second, if the compounds were not actually bound, then the C10-BZT and C8-Cl-BZT should have been the slowest to desorb because of their larger molecular weights and higher octanol-water partition coefficients, ∼107 vs ∼104 (Figure 1). (b) Why is C1-BZT the only BZT detected in the bound fraction? One possible reason could be that it is the smallest of the three BZTs, and therefore, the only one that might be physically sequestered inside the structure of the sedimentary organic matter (e.g., humic molecules) or the micropores of the sediment (21). Or it could be chemically associated with the sedimentary organic matter, and the C10-BZT and C8-Cl-BZT are not as tightly bound because of their bulky alkyl substituents that hinder 976
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 6, 2000
this binding process (Figure 1). To investigate further, we examined a core in the Pawtuxet River (Figure 2), which was closer to the chemical plant and was more enriched in the other three less abundant BZTs. These compounds have a degree of alkyl substitution that lies between the range of C1-BZT to C8-Cl-BZT and C10-BZT. Pawtuxet River Core. Both OC and nitrogen have subsurface maxima at 5-7 cm and then followed similar down-core trends, oscillating from 2 to 9% and from 0.1 to 0.5%, respectively (Table 1). The C/N ratios were highly variable, from 11 to 29, and were lowest in the sections from 7 to 15 cm. All six BZTs (Figure 1) were detected in the free fraction, with concentrations ranging from trace levels to 5 mg g-1 (Figure 5). C1-BZT and C8-Cl-BZT were the most abundant, and each were observed down to 50-52 cm. The other four BZTs were only present in the top 20 cm of the core. Because the concentrations of the BZTs were so elevated, they actually were a significant amount of the total organic carbon and nitrogen in this core. For example, in the 10-12-cm section, the nitrogen from the BZTs accounts for 22% of the total nitrogen, and the carbon was 10% of the total carbon (19). Five of the six BZTs were detected in the bound fraction in at least the top 15 cm of the core. On an absolute basis, the C1-BZT and C8-Cl-BZT were the most abundant bound compounds. For example, the concentration bound was 29 µg g-1 for C1-BZT and 1.0 µg g-1 for C8-Cl-BZT in the 1012-cm section (Table 2). There, however, was a clear trend in percent bound with the C1-BZT, C4-BZT, and C4-Cl-BZT being at least 0.10% bound, but the C5-Cl-BZT and C8-ClBZT were always 0.04% bound or less (Table 3). It appears that the BZTs that do not have an alkyl group at the 3′ position have a greater chance to be bound than the BZTs that have an alkyl group at that position (Figure 1). Consider the binding of C4-Cl-BZT and C5-Cl-BZT (Table 3). They are very similar compounds, except that the C5-Cl-BZT has one additional alkyl carbon and its tert-butyl group is adjacent to the hydroxy group. Hence, they have approximately the same size, molecular weight, and molecular diffusivity in water. Yet, the amount of C4-Cl-BZT that is bound in the Pawtuxet River core is ∼20 times more than the C5-Cl-BZT, even though the total concentration of C5-Cl-BZT is more than that of C4-Cl-BZT. This suggests that size and diffusion are not the most important factors controlling the binding of these compounds. The possibility that the contaminants are mainly bound due to some type of strong hydrophobic interaction is unlikely since the more hydrophobic BZTs (like C8-BZT) do not have the highest percentage of bound species. This is seen in all of the Narragansett Bay and Pawtuxet River core samples as well as several other sediment samples from
FIGURE 5. Pawtuxet River core concentrations of free BZTs (a) C1-BZT, (b) C4-BZT, (c) C4-Cl-BZT, (d) C5-Cl-BZT, (e) C8-Cl-BZT, and (f) C10-BZT. Narragansett Bay. These results are consistent with the hypothesis that the binding reactivity of these compounds is controlled or hindered by the presence or absence of alkyl groups at the 3′ location. This hindrance may be sterically induced and/or due to the differences in the acid dissociation of the adjacent hydroxy group and, hence, its reactivity. The unhindered BZTs have a smaller pKa than the others (Figure 1), which makes their hydroxy groups more easily dissociated. For example, the hydroxy on C1-BZT can be 11-18 times more dissociated than the hydroxy group of C8-Cl-BZT for pH values from 5 to 8, the normal range encountered in the environment (22). While it is not clear, based on these initial results, what type of chemical association is binding BZTs to sediments (11), it is probably a covalent bond between the BZTs and the sedimentary organic matter. This is because the extraction for the free fraction method would have removed any BZTs that were hydrogen bonded or had some type of charge-transfer complex with the sediment. Two other possibilities are the formation of an ether bond through the hydroxyl group or a Michael addition reaction, where the
phenolic hydroxyl group forms a keto and reacts at the 4′ position (β to the hydroxyl). A Michael addition would form a carbonscarbon bond and be also hindered by alkyl substitution at the 3′ position. However, the saponification step presumably would not cleave an ether linkage or a direct carbonscarbon bond and then also return the original adduct. Hence, we hypothesize that the BZTs are reacting with a carboxylic acid group in the organic matter and forming an ester linkage. Future work will be directed at this issue. For the three compounds (C1-BZT, C4-BZT, and C4-ClBZT) without an alkyl group adjacent to the 3′ position, there was some differences in the percent bound in the top 15 cm of the core (where all three compounds were found in the bound fraction) (Table 3). Generally, the percent bound of C4-Cl-BZT was the highest or equivalent to the percent bound of C1-BZT and was 2-5 times greater than the percent bound of C4-BZT. The average of the total (free + bound) concentrations in these sections of the core are 2.6, 0.15, and 0.045 mg g-1 for C1-BZT, C4-BZT, and C4-Cl-BZT, respectively, suggesting that other factors besides total concentration and VOL. 34, NO. 6, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
977
TABLE 2. Concentrations of Bound and Free BZTs in the 10-12-cm Section of the Pawtuxet River Core
TABLE 3. Percentage (%) of BZTs Bound in the Pawtuxet River Corea
p
a
NA, not applicable because both free and bound were not detected.
alkyl substitution adjacent to the hydroxy group may also control the percentage of bound BZTs. When considering C4-BZT and C4-Cl-BZT, the only apparent reason that C4-Cl-BZT has a higher percentage of bound compounds is that the addition of the Cl in the benzotriazole moiety also enhances the binding of this compound. Another factor we considered was whether the amount of sedimentary organic carbon affected the percent bound of all of the BZTs in this core; however, we found no obvious trends or strong correlations. Comparison of C1-BZT in the Two Cores. The dynamics of bound C1-BZT were quite different in the Pawtuxet River core than the Narragansett Bay core. While the percent bound of the C1-BZT in the top 15 cm of the riverine core remained relatively constant at 0.5 ( 0.0.2%, it increased to ∼3% below that depth (Table 3). This increase at 15 cm coincides with the general decrease in the concentration of free C1-BZT and the other BZTs (Figure 5) and suggests that the high 978
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 6, 2000
concentrations of BZTs may have affected the binding of these compounds. It may be that the high concentrations of BZTs saturated or overwhelmed the specific binding site in these sediments. This may not only affected the percent bound of each compound but also the relative proportions of each bound compound. In the Narragansett Bay core, with its much lower concentrations of BZTs, the percent bound C1-BZT decreased from 9% at 0-2 cm to ∼2% at 8-10 cm (Figure 4). Since this is a limited study and initial investigation of only two cores, it would be difficult to draw any major conclusions about this difference. One major factor may be the overall differences in riverine and estuarine sedimentary organic matter, including their degree of aromaticity, amount and distribution of functional groups, and composition. In addition, other environmental factors may be responsible for these differences. For example, the Narragansett Bay sediments are finer grained and have a much slower sedimentation rate. Another factor such as
porewater pH, which was not measured, may be important based on the pKa values of these compounds (Figure 1). Last, our treatment for removing the bound BZTs is operationally defined and may only remove a portion of the total bound BZTs in each sediment sample. In summary, BZTs were found in our operationally defined free and bound fractions of marine and freshwater sediments. For every sample, BZTs that did not have alkyl groups in the 3′ position were more likely to be bound than compounds that did. From these results, one cannot conclude that the BZTs are actually covalently bound to the sediment; hence, we chose to refer to the BZTs being chemically associated with the sedimentary organic matter. Nevertheless, these initial results in “real-world samples” show how alkyl substituents can effect the fate of organic contaminants, provide some insight in how organic contaminants interact with sedimentary organic matter, suggest that high concentrations of organic contaminants in sediments may affect binding reactions, and reveal their importance in understanding the long-term fate and bioavailability of organic contaminants in sediments. Additional work will focus on elucidating the mechanisms of these reactions with abiotic and biotic laboratory incubations of BZTs with sediments and using other chemical reagents to cleave the bound BZTs.
Acknowledgments We would like to thank Dr. R. Cairns, P. Hartmann, S. Sherwood, S. Sylva, and M. Traber of URI for their help in this project. Dr. P. Gschwend of MIT also was very helpful. C.M.R. was supported by a Narragansett Electric Coastal fellowship and a GSO alumni award. The GC/MSD was donated through the Hewlett-Packard University Grants Program, and an ALCOA Foundation grant was used for partial support of this work.
Literature Cited (1) Farrington, J. W.; Quinn, J. G. Geochim. Cosmochim. Acta 1973, 37, 259. (2) Van Vleet, E. S.; Quinn, J. G. Geochim. Cosmochim. Acta 1979, 43, 289. (3) Van Vleet, E. S.; Quinn, J. G. Deep-Sea Res. 1979, 26A, 1225. (4) Prahl, F. G.; Pinto, L. A. Geochim. Cosmochim. Acta 1987, 51, 1573.
(5) Lee, C.; Gagosian, R. B.; Farrington, J. W. Geochim. Cosmochim. Acta 1977, 41, 985. (6) Lajat, M.; Saliot, A.; Schimmelmann, A. Org. Geochem. 1990, 16, 793. (7) Cranwell, P. A. Org. Geochem. 1981, 3, 79. (8) Bellar, H. R.; Simoneit, B. R. T. In Organic Marine Geochemistry; Sohn, M. L., Ed.; ACS Symposium Series 305; American Chemical Society: Washington, DC, 1986; pp 198-214. (9) Bellar, H. R.; Simoneit. B. R. T. Bull. Environ. Contam. Toxicol. 1988, 41, 645. (10) Jeng, W. L.; Han, B. C. Estuarine, Coastal, Shelf Sci. 1988, 38, 727. (11) Richnow, H. H.; Seifert, R.; Hefter, J.; Kastner, M.; Mahro, B.; Michaelis, W. Org. Geochem. 1994, 22, 671. (12) Khan, S. U.; Schnitzer, M. Geochim. Cosmochim. Acta 1972, 36, 745. (13) Estimated by Advanced Chemical Developments, Inc., Toronto, Canada. (14) Lopez-Avila, V. Ph.D. Dissertation, Massachusetts Institute of Technology, 1979. (15) Lopez-Avila, V.; R. A. Hites Environ. Sci. Technol. 1980, 14, 13821390. (16) Pruell, R. J.; Quinn, J. G. Estuarine, Coastal Shelf Sci. 1985, 21, 295. (17) Corbin, J. M.S. Thesis, University of Rhode Island, 1989. (18) King, J. W.; Corbin, J.; McMaster, R.; Quinn, J. G.; Gangemi, P.; Cullen, J. D.; Latimer, J. S.; Peck, J.; Gibson, C.; Boucher, J.; Pratt, S.; LeBlanc, L.; Ellis, J. T.; Pilson, M. E. Q. A Study of the sediments of Narragansett Bay. Vol. 1: Final report to the Narragansett Bay Project; Graduate School of Oceanography, University of Rhode Island: Narragansett, RI, 1995. (19) Reddy, C. M. Ph.D. Dissertation, University of Rhode Island, 1998. (20) Reddy, C. M. The University of Rhode Island, unpublished results. (21) Steinberg, S. M.; Pignatello, J. J.; Sawhney, B. L. Environ. Sci. Technol. 1987, 21, 1201. (22) Harris, J. C.; Hayes, M. J. In Handbook of Chemical Property Estimation Methods; Lyman, W. J., Reehl, W. F., Rosenblatt, D. H., Eds.; Americal Chemical Society: Washington, DC, 1990; Chapter 6, pp 1-28.
Received for review August 18, 1999. Revised manuscript received December 14, 1999. Accepted December 15, 1999. ES990971I
VOL. 34, NO. 6, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
979
