Environ. Sci. Technol. 2001, 35, 1111-1117
Partitioning, Extractability, and Formation of Nonextractable PAH Residues in Soil. 2. Effects on Compound Dissolution Behavior GRANT L. NORTHCOTT* AND KEVIN C. JONES Department of Environmental Science, Institute of Environmental and Natural Sciences, Lancaster University, Lancaster, LA1 4YQ, United Kingdom
This study was carried out to assess the influence of PAH sequestration in sterile sewage sludge-amended arable soil on their dissolution behavior. Radiolabeled phenanthrene (14C-9-Phe), pyrene (14C-4,5,9,10-Pyr), and benzo[a]pyrene (14C-7-B[a]P) were spiked and aged for up to 525 days in sterile soil microcosms. The dissolution behavior of the compounds was determined at various times using an in-situ XAD-2 sorbent method. The XAD procedure allowed the detection of hydrophobic PAHs released from the soil and demonstrated that a fraction of PAH was resistant to dissolution. The experimental data generated was analyzed using an empirical two-site model, and fitted values were obtained for the magnitude of the rapidly released compound fraction and the rate of release for the rapid and slowly released compound fractions. Increasing sequestration did not reduce the size of the rapidly released compound fraction in this soil. We did, however, observe a decrease in the rate of the fast released fraction of 14C-7-B[a]P which formed a significantly larger sequestered fraction. The duration of the dissolution experiments proved insufficient to obtain precise estimations for the rate of slow compound release from the soil, and we were unable to determine whether sequestration affected the rate of release for the slowly released compound fraction. These results demonstrate the necessity of conducting desorption experiments for a period of months if complete data on the long-term release of desorption resistant compound fractions is to be obtained.
Introduction The partitioning of hydrophobic organic chemicals (HOCs) between water and soils or sediments continues to be the subject of extensive research. Much of this research is currently focused on kinetic aspects of compound partitioning, or the rate of compound sorption and desorption from soil and sediments, under nonequilibrium conditions. Compound desorption processes result from a continuum of diffusive interactions operating over a wide range of time scales. At one extreme there is rapid and reversible desorption of a labile compound fraction, at the other a slow, or “resistant”, fraction desorbing much more slowly. This desorption resistant fraction is believed to be due to molecular retention or retarded diffusion of contaminants within soil * Corresponding author phone: (+44) 1524 593300; fax: (+44) 1524 593985; e-mail:
[email protected]. 10.1021/es000072q CCC: $20.00 Published on Web 02/10/2001
2001 American Chemical Society
micropores (intraparticulate diffusion (IPD)) (1, 2), the macromolecular structure of organic matter (intraorganic matter diffusion (IOMD)) (3, 4), and the enhanced affinity of some HOCs to highly aromatic and high surface area, carbonaceous soot-like organic carbon phases (5, 6). Biphasic desorption kinetics is commonly observed for HOCs sorbed to soil and sediment, including PAHs, PCBs, chlorobenzenes, pesticides, and herbicides (7-11). The sorption of nonionic compounds in soil and sediment with organic carbon (OC) >0.1% is dominated by partitioning and diffusive interactions with organic carbon (12). Therefore, in many soils and sediments, slow phases of solute sorption and desorption may be attributed to IOMD and soot-carbon interactions. As a consequence of experimental observation and the development of conceptual models, the desorption resistant compound fraction in soil and sediment is often referred to as being sequestered within the sorbent matrix, particularly within, or by, organic carbon phases. The size of this resistant compound fraction increases with contact time, or aging, in both laboratory-spiked and field samples. Organic chemicals aged in soil exhibit a reduction in the amount of chemical available for extraction (13), microbial degradation (14), toxicity (15), and bioaccumulation (16). These observations are explained by a decreasing amount of chemical in the fast desorbing fraction as it becomes progressively degraded by biotic and abiotic processes, reduced by leaching and volatilization processes, and sequestered within soil. Rapidly released HOCs in soil and sediment can be readily degraded and depleted by resident microorganisms. The biodegradation of organic compounds in soil is often limited by the mass transfer, or rate of release, of the slowly desorbing fraction into interstitial water where they are available for utilization by microorganisms (17). The rate and extent of solute release from soil and sediment are, therefore, important parameters for determining the long-term fate of organic contaminants in these environments and in the wider ecosystem. Various methods have been adopted to determine the rate of release of HOCs from soil and sediment. These include aqueous batch equilibration desorption methods (7), gaspurging (18, 19), miscible displacement (20, 21), in-situ adsorption using polymeric resins (9, 11, 22), and, recently, supercritical fluid extraction (23). The highly hydrophobic nature of soil-bound PAHs results in very low aqueous concentrations that compromise their detection. To facilitate the detection of PAHs released from soil in our experiment, we adopted the use of an in-situ solid phase adsorbent, XAD-2, for compound concentration. Amberlite XAD-2 resin is a nonpolar, hydrophobic styrenedivinylbenzene copolymer that has been widely used for the extraction and concentration of PAHs from water samples (24, 25). The macroreticular nature of the macrospherical resin beads, together with their large surface area and hydrophobicity relative to their mass, promotes fast adsorption of HOCs from aqueous media and high extraction efficiencies. The XAD in-situ adsorption technique has been adopted in a number of studies examining the kinetics of release of HOCs, including PCBs, PAHs, and linear alkylbenzenes, from soil and sediment (9, 22, 26). Adding an appropriate quantity of XAD-2 to a soil or sediment water suspension provides an “infinite” sink for the adsorption of released HOCs. This promotes compound release to the aqueous phase in order to attain solute equilibrium within the batch equilibrium VOL. 35, NO. 6, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1111
reactor. HOCs are then concentrated, improving the detection of sparingly water-soluble compounds released from soil to the aqueous phase. This is the second paper examining the sequestration of PAHs in sewage sludge-amended arable soil. In the first, we reported the sequestration of PAHs in sterilized sewage sludge-amended arable soil by abiotic diffusive processes in the absence of microbially ameliorated processes (27) and observed an increase in the sequestration of PAHs by abiotic processes in sterile soil with increasing contact time and compound MW and Kow. The objective of the experiments reported in this paper was to determine whether the sequestration of PAHs in soil affected their dissolution behavior into aqueous solution and, therefore, the biodegradation potential and bioavailability of aged PAHs.
Materials and Methods Chemicals. Nonradiolabeled phenanthrene (Phe), pyrene (Pyr), and benzo[a]pyrene (B[a]P) and the 14C-radiolabeled analogues 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P were obtained from Sigma Aldrich Co. Ltd., U.K. Toluene and methanol were obtained from Rathburn Chemicals Ltd. U.K. and sodium azide (NaN3) from MerckBDH, U.K. Sample oxidizer reagents and consumables, Ultima Gold XR (UGXR) scintillation cocktail and 23 mL plastic scintillation vials were all obtained from Canberra Packard, U.K. A-C-S reagent potassium carbonate (K2CO3) and analytical grade calcium chloride (CaCl2) were obtained from Aldrich, U.K. Equipment. Supelpak-2, a precleaned high purity Amberlite XAD-2 macroreticular styrenedivinylbenzene copolymer resin, was obtained from Sigma Aldrich Co. Ltd., U.K. Polycarbonate centrifuge tubes (50 mL Oak Ridge) with polypropylene screw cap lids were obtained from MerckBDH, U.K.
Experimental Section Spiked and Aged Soil. Field-wet arable crop soil, amended with concentrated sewage sludge (equivalent to 40 tonnes/ hectare), was spiked with 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P using a previously evaluated and optimized procedure (28). Dry weight of the prepared soil was 82.0 ( 0.2 % as determined by drying to a constant weight at 105 °C (n ) 20). Soil TC, TIC, and TOC values were 2.25 ( 0.09%, 0.15 ( 0.02%, and 2.10 ( 0.09%, respectively (n ) 20) (Carlo Erba CHNSO Elemental Analyzer EA 1108-R). Sewage sludgeamended soil microcosm replicates were spiked to a concentration of 10 mg kg-1 for nonradiolabeled Phe, Pyr, and B[a]P and 79.6 (4776 DPM.g-1), 96.1 (5766 DPM.g-1), and 89.9 Bq.g-1 (5394 DPM.g-1) for 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P, respectively. Full details regarding soil sampling location, organic carbon analysis and soil spiking, sterilization by gamma irradiation, and storage methods are provided elsewhere (27, 28). Individual microcosms were sampled at predetermined times over the following 18 months for total activity, extractability, nonextractable residue, and dissolution behavior of the aged PAHs. Blank samples for all analyses were sampled from blank solvent spiked soil microcosms. Activity of Spiked Soil, Extracted Soil Residues, Solvent Extracts, and Aqueous Solutions. Full details of the procedures used to extract and determine the activity of radiolabeled PAHs in soil and solvent extracts are detailed in the first paper in this series and elsewhere (27, 29). Briefly, the activity of radiolabeled compound in soil and extracted soil residues was determined by total sample oxidation using a Packard 307 Sample Oxidizer. The activity of radiolabeled compound in organic solvent extracts and aqueous solutions was determined by mixing aliquots with UGXR scintillation cocktail in 23 mL plastic scintillation vials and counting with 1112
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 6, 2001
FIGURE 1. Centrifuge tube and contents used in the dissolution experiments. a Canberra Packard Tri-Carb 250CA liquid scintillation analyzer (LSC). Prepared vials were capped, shaken, and stored overnight in the dark to stabilize and to counter the effects of induced chemiluminescence and photoluminescence. The samples were counted for a period of 10 min or sufficient time to provide 50× background count values. Quenching of emitted photons during liquid scintillation counting was corrected by preparing a quench calibration curve from methanol:toluene extracts obtained from blank soil:XAD-2 equilibrations. This overcame the problem of chemical quenching by components of the solvent mixture and color/light quenching by natural organic matter components released from the soil. Samples were corrected by blank subtraction.
Desorption/Dissolution Experiment XAD-2 Resin Pretreatment. Desorption of aged PAH residues was determined using a modified version of the in-situ XAD resin method of Carroll et al. (9). Supelpak-2, a high purity Amberlite XAD-2 resin, was prepared by soaking in methanol (MeOH) to remove entrapped air and traces of residual monomeric compound and preservative agents (30). Over the following 24 h the MeOH was periodically decanted and replaced with fresh solvent. The MeOH was exchanged with a freshly prepared solution of 0.01 M CaCl2 and 0.01 M NaN3 in MilliQ treated water. The aqueous solution was changed 10 times to ensure complete removal of MeOH. The equilibrated XAD-2 resin was refrigerated as a slurry in a glass Schott bottle with excess 0.01 M CaCl2 solution, to prevent dehydration. NaN3 was added to inhibit microbial activity in the aqueous solution and microbial growth on the XAD-2 resin beads. Excess CaCl2 solution was decanted from the resin immediately before beginning a dissolution experiment, leaving the resin as a damp solid. XAD-2 Dissolution Experiment. Approximately 3.5 g of soil and 1 g of wet XAD-2 resin were weighed directly into a 50 mL polycarbonate centrifuge tube. Thirty milliliters of 0.01 M CaCl2/0.01 M NaN3 solution was then added (Figure 1). An aqueous solution of 0.01 M CaCl2 was used to prevent the dispersion of soil clay particles (22), while 0.01 M NaN3 was included to prevent microbial growth. The tubes were sealed, placed in an end-over-end rotating shaker (22 rpm), and maintained at a temperature of 22 ( 1 °C. Duplicate tubes were removed at predetermined times for analysis. Sampled tubes were stood vertically for five minutes to allow settling of the soil-XAD-2-water slurry, after which 2.7 g of K2CO3 was added to promote phase separation. The tubes were resealed, vigorously shaken by hand to mix the contents, and centrifuged (MSE-Centaur 2) at 5000 rpm for 30 min to separate the soil and XAD-2. This treatment increases the density of the ionic solution and causes the XAD-2 to float to the surface of the centrifuge tubes. The resin was recovered by vacuum filtration into an inverted glass Pasteur pipet, where it was retained by a plug of glass wool (Figure 2). The
the fraction of each PAH released from the soil at each sampling time and to construct compound desorption curves. The fraction of released compound was calculated by dividing the amount of compound activity extracted by the XAD-2 resin by the amount of compound activity present in the mass of soil taken from each spiked soil microcosm. An empirical two-compartment model incorporating a fast and slowly desorbing PAH fraction in soil (31) was modified and used to describe the observed desorption kinetics (eq 1) (32). This equation consists of two first-order terms with the first describing the fast released compound fraction and the second the slowly released fraction
Sx ) FSoe-k1t + So(1-F)e-k2t
(1)
where t ) time (hours); Sx ) mass of chemical on XAD-2 after time t (mg kg-1 dry soil); So ) mass of chemical in the soil (mg kg-1 dry soil); F ) fraction of chemical released rapidly from the soil (labile phase); (1-F) ) fraction of chemical released slowly from the soil (nonlabile phase); k1 and k2 are first-order rate constants for the rapid and slowly released fractions respectively (hours-1). This equation was modified to represent the fraction of chemical released as a function of time (eq 2) (22, 32) FIGURE 2. Vacuum extraction method for the recovery of XAD-2 after compound dissolution. XAD-2 within the Pasteur pipet was rinsed with 20 mL of 0.01 M CaCl2/0.01 M NaN3 solution to remove K2CO3 and then dried under vacuum for approximately 2 min. XAD-2 Extraction. After vacuum-drying, the tip of the Pasteur pipet was removed using a glass cutter, and the glass wool plug removed with tweezers. The XAD-2 resin was rinsed into a glass chromatography column (150 mm × 5 mm diameter, 10 mL solvent reservoir) fitted with a Teflon tap, using 2 × 2.5 mL volumes of MeOH:toluene (1:1). PAHs adsorbed to the XAD-2 resin were eluted with a total of 15 mL of MeOH:toluene (1:1). The eluted solvent containing extracted PAHs was collected as 3 individual 5 mL portions in plastic scintillation vials containing 15 mL of UGXR scintillation fluid. The activity of compound in each vial was determined by LSC. Compound Recovery and XAD-2 Resin Extraction Reproducibility. Individual spiking solutions of radiolabeled and nonradiolabeled Phe and B[a]P were prepared in MeOH. Adding 40 µL of each PAH spike solution to 45 mL of water gave an aqueous compound concentration no greater than 10% of the corresponding compound water solubility. Forty microliters of each solution was spiked into Oak Ridge centrifuge tubes containing 45 mL of 0.01 M CaCl2/0.01 M NaN3 solution and 1 g of wet XAD-2. The tubes were sealed and equilibrated for 6 h, the resin was separated, and the XAD-2 adsorbed compound was recovered and measured as described above. The presence of soil released PAHs in the aqueous phase was checked at various times during the XAD-2/soil equilibration experiments. After equilibration, the tubes were centrifuged at 5000 rpm for 30 min to separate the soil and XAD-2 resin from the aqueous phase. PAHs were also measured in the aqueous phase of replicate soil-water equilibrations prepared in the absence of XAD-2. Soil-water mixtures equilibrated for 48 h were centrifuged to separate the soil and aqueous phase. Ten milliliters of these aqueous solutions was combined with 10 mL of UGXR and compound activity determined by LSC. Desorption Data Evaluation. The data collected at the conclusion of the experiment included total soil PAH activity and the activity of PAHs desorbed from soil during duplicate XAD-2/soil equilibrations. These data were used to determine
R ) (1-(Sx/So)) ) 1 - F * exp(-k1*t) (1-F) * exp(-k2*t) (2) where R ) (1-(Sx/So)) ) fraction of chemical released after time t. Values for So and Sx, obtained by total oxidation of the sampled soil and the activity of PAH extracted from XAD-2 at each sampling time, were used to calculate R. Values for the three unknown parameters in eq 2 (F, k1 and k2) were determined using the nonlinear regression function of SPSS for Windows (Release 7.5.1). Parameter R was defined as the dependent variable, t the independent variable, and F, k1 and k2 the fitting parameters. The Levenberg-Marquardt estimation method was used to calculate the parameter values that provided the best nonlinear least-squares fit to the data. Starting values for F, k1, and k2 were taken from previously published results for PAH release from contaminated soils (22). Using these starting values, the solution required 10 to 20 iterations for 14C-9-Phe and Pyr, and 16-38 for B[a]P, to reach constant values for the specified parameters.
Results Compound Recovery and Data Variability. The recovery of 14 C-9-Phe and 14C-7-B[a]P extracted from spiked aqueous solutions with XAD-2 was excellent with values of 98.7 ( 2.5% (n ) 6) and 97.3 ( 2.2% (n ) 6), respectively, for 14C9-Phe and 14C-7-B[a]P. This is consistent with previous studies that have demonstrated the high extraction efficiency of PAHs from water samples (84-110%), both at trace levels and higher concentrations representative of those in wastewater streams (24, 33). The low RSDs obtained by the XAD-2 adsorption replicates demonstrate the excellent reproducibility of the adopted method. None of the aqueous solutions sampled from the soil/ water/XAD-2 equilibration experiments contained any measurable activity above natural background levels. We therefore assume that the XAD-2 resin efficiently adsorbed the radiolabeled PAHs released from soil to the aqueous phase. By contrast, aqueous solutions from soil/water equilibrations carried out in the absence of XAD-2 contained measurable amounts of desorbed 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C7-B[a]P. The fractions of compound in the aqueous solution were approximately 1/40, 1/74, and 1/80 the value of those obtained for 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P using the XAD-2 method. This further illustrates both the efficiency and advantages of increased solute concentration obtained VOL. 35, NO. 6, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1113
FIGURE 3. Representative experimental and model-fitted dissolution curves obtained for the PAHs (fraction released ) activity of compound measured on XAD-2 resin divided by the total activity of compound in the mass of added soil). by the XAD-2 in-situ adsorption method when determining compound release from soil. Variability in this method arises from the heterogeneous nature of compound in the bulk soil within each microcosm and the variability of the employed analytical methodology. In particular, separation and vacuum recovery of XAD-2 were identified as critical sources of data variability. With proper care and attention paid to the separation technique, which must be conducted in a reproducible manner, variability among replicate measurements is very low. Regardless of the inherent variability of the spiked soils and adopted methodology, the results of the duplicate measurements were usually in excellent agreement. RSDs for the fraction of released compound from duplicate equilibrations ranged from 0.1 to 5.8%, 0.02 to 6.5%, and 0.6 to 10.1%, respectively, for 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P. The higher RSDs obtained for B[a]P reflect small differences between minute fractions of released compound. They do not indicate an increased inherent variability in the release of B[a]P from soil, or extraction from solution, by XAD-2. Mass balances were determined at various times during dissolution experiments by total sample oxidation of the soil residues in centrifuge tubes after equilibration with, and the removal of, XAD-2. Comparison between the initial amount of compound measured in sampled soils with the sum of compound extracted by XAD and remaining on soil after equilibration showed that quantitative compound recovery was maintained during the dissolution experiments. Quantitative recovery of PAHs during dissolution experiments and spiked aqueous solution equilibrations demonstrates negligible loss of desorbed PAHs occurred to the polycarbonate container materials in the presence of XAD-2. Compound Dissolution. The fraction of released PAH was plotted against equilibration time to provide compound dissolution kinetic plots, or plots of compound release. An example of the dissolution kinetic plots obtained by the XAD-2 adsorption method for the three studied PAHs is shown in Figure 3. The dissolution curves are typical of those obtained for the PAHs at each microcosm sampling time. The plots exhibit characteristic biphasic patterns of compound release, with a relatively fast phase of compound release followed by significantly slower release of the remaining compound. The dissolution plots show that the amount of PAH released from the soil is approaching a plateau, or constant fraction, for 14C-9-Phe and 14C-4,5,9,10-Pyr after approximate equilibration times of 72 and 120 h, respectively. In contrast, 14C-7B[a]P does not appear to have reached a plateau after 335 h equilibration. This illustrates the different dissolution behavior exhibited by more hydrophobic and polycondensed 1114
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 6, 2001
PAHs, such as B[a]P, from soil. These results also demonstrate that desorption experiments performed for short time intervals (hours) may not reach solution phase equilibria and identify slowly desorbed compound fractions. This is particularly important for more hydrophobic HOCs with very low water solubility and high Koc. The dissolution data for 14 C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P displayed in Figure 3 was modeled using eq 2 to obtain fitted values for F, k1, and k2 which were used to generate model fitted-curves. Comparison of the experimental and model-fitted dissolution curves for each PAH shows that the model provides a good fit to the experimental data (Figure 3). Plots of the compound dissolution profiles obtained for the individual PAHs at each soil sampling time are shown in Figure 4A-C. Initial dissolution experiments were carried out for a total of 90 h. Previous experiments had shown this to be an appropriate length of time for obtaining dissolution profiles for PAHs from a range of contaminated source materials including sewage sludge and industrial soil samples (34). Our initial results showed that while this equilibration time was sufficient to reach a constant fraction of released 14C-9-Phe and 14C-4,5,9,10-Pyr, 14C-7-B[a]P would require longer equilibration times (see Figure 4A-C). The equilibration time was increased to 220 h for samples taken at 53 days aging (in line with Carroll et al. (9)) and still further to 335 h for the last three sampling occasions at 170, 259, and 525 days. The fraction of 14C-9-Phe released after 10 and 21 days aging was significantly greater than the fraction released after 53 days aging and longer (Figure 4A). The dissolution curves for 14C-9-Phe aged in soil for 170, 259, and 525 days were similar and overlap at most equilibration times. Similar results were obtained for 14C-4,5,9,10-Pyr (Figure 4B). In comparison, the dissolution curves obtained for B[a]P fell into two distinct groups (Figure 4C). Unlike those for 14C-9-Phe, 14C-4,5,9,10Pyr, the dissolution curves for 14C-7-B[a]P after 10, 21, and 53 days aging were similar and could be approximated by a single dissolution curve. The dissolution curves for 14C-7B[a]P at 170 days aging and longer were also similar and could be approximated by a single dissolution curve. The fraction of compound released from the aged soil decreased in the order Phe > Pyr > B[a]P with approximate mean values of 0.8, 0.7, and < 0.3, respectively. Estimates of the parameters F, k1, and k2 were calculated from the dissolution data obtained for the three PAHs at each sampling time during the duration of the experiment (525 days). Values for F, k1, k2, and R2 (regression fit), obtained by the best fit of the experimental data to the empirical twocompartment model (eq 2), are shown in Table 1. Before discussing these results a number of limitations of the XAD-2 method, and assumptions regarding the application of the two-compartment model (eq 1) to the experimental data, should be mentioned. Namely (1) the results represent the upper estimate for compound release due to optimized soilwater-XAD mixing and contact, (2) XAD-2 has an infinite sorption capacity for released study compound(s), and (3) the results are for the specific soil sample and compound(s) that were analyzed. And, in regard to the model (1) k1 and k2 are pseudofirst-order rate constants representing all types of compound release which may be better described by more complicated expressions, (2) k1 and k2 describe the rate of both compound rapid release from soil to water and compound sorption to XAD, (3) eqs 1 and 2 assume that all the sorbed compound will be released after an infinite time, (4) the model assumes there is no re-release of XAD sorbed compound back to the soil, and (5) the empirical curve fitting equation does not provide data on solute binding and release mechanisms. These assumptions are not unique to this procedure and are equally applicable to other methods employed to study
TABLE 1. Fraction of Chemical Rapidly Released and Calculated Chemical Release Rate Constants for PAHs Aged in Sterile Arable Soil for 525 Days days aging
Fd
10a 21a 53b 170c 259c 525c
0.69 0.71 0.76 0.75 0.69 0.71
10a 21a 53b 170c 259c 525c
0.54 0.56 0.59 0.62 0.57 0.60
10a 21a 53b 170c 259c 525c
0.04 0.03 0.10 0.05 0.03 0.06
k1 (h-1)
k2 (h-1)
R2 e
0.014 0.008 0.003 0.001 0.002 0.002
0.991 0.999 0.994 0.988 0.980 0.909
0.006 0.004 0.002 0.001 0.001 0.001
0.999 0.999 0.993 0.997 0.992 0.994
0.0016 0.0016 0.0008 0.0007 0.0008 0.0007
0.992 0.993 0.990 0.992 0.985 0.989
14C-9-Phe
0.39 0.38 0.18 0.26 0.45 0.41 14C-4,5,9,10-Pyr
0.16 0.13 0.11 0.07 0.08 0.08 14C-7-B[a]P
0.47 0.39 0.04 0.05 0.05 0.03
a Total equilibration time of 90 h. b Total equilibration time of 216 h. Total equilibration time of 335 h. d Fraction of chemical released rapidly from the soil. e R squared for nonlinear regression fit ) 1 - (residual sum of squares/corrected sum of squares). c
FIGURE 4. Experimental dissolution curves for PAHs aged in sterile arable soil: A. 14C-9-Phe; B. 14C-4,5,9,10-Pyr; C. 14C-7-B[a]P. compound desorption and rate of release from soil and sediment. By standardizing our experimental methods and using a single soil and specific compounds in this study, we can obtain comparative parameters for the fast and slowly released compound fractions desorbing from the soil. The match between the fitted model and the experimental data to which it was fitted was excellent, with values for R2 ranging from 0.909 to 0.999 and an average value of 0.99 ( 0.02 (n ) 18) (Table 1). The fitted values of F for each compound show a reduction in the amount of fast released compound with increasing compound molecular weight and kow. Average values of F for 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P decreased in the order 0.72 > 0.58 > 0.05 and were significant at p ) 0.05. Values of F obtained for the individual compounds did not display any distinct trend with increased aging in soil. The fitted F values ranged from 0.69 to 0.76 for for 14C-9-Phe, 0.54 to 0.62 for 14C-4,5,9,10-Pyr, and 0.03 to 0.01 for 14C-7-B[a]P with respective RSDs of 4.2%, 5.2%, and 50%. The RSD for the fitted F values obtained for 14C-7-B[a]P decreased to 31% when the significantly higher value obtained for day 53 is excluded from the calculation. The wider range and higher RSD of fitted F values for 14C7-B[a]P reflect the inherent difficulty in obtaining dissolution
data for more hydrophobic compounds which are desorbed from soil in sparing quantities. Fitted values for k1, the rate constant for the rapidly released compound fraction, show a similar trend to the F values, generally decreasing with increasing compound MW. This is not surprising, since the rate of rapid compound release and fractions of rapidly released compound are codependent. The values of k1 for 14C-9-Phe show no definite trend over the 525-day compound aging period. In contrast, k1 values for 14C-4,5,9,10-Pyr decreased by a factor of 2 and 14C-7-B[a]P by an order of magnitude during 525 days aging. This decrease became significant for 14C-4,5,9,10-Pyr after 53 days aging and after 21 days aging for 14C-7-B[a]P. Values of k2, the rate constant for the slowly released compound fraction, exhibit a slowly decreasing trend for all three PAHs with increased aging. The k2 value for 14C-9-Phe decreases by a factor of 2 between 10 and 21 days aging (see Table 1) and remains approximately the same after 53 days aging (0.0027 h-1). The k2 values for 14C-4,5,9,10-Pyr decrease by a factor of 5 between 10 and 525 days aging. Values of k2 for 14C-7-B[a]P decrease by a factor of 2 after 21 days aging and thereafter remain essentially constant.
Discussion The pattern of biphasic release observed in this experiment is consistent with previous results published for PAH dissolution from soil and sediments (11, 26, 35). The large differences in magnitude between the rate constants for rapid and slow compound release support the hypothesis that a fraction of PAH sequestered within soil is less available to partition to the aqueous phase. This is supported by concurrent experiments which demonstrated that a fraction of each 14C-labeled PAH could only be extracted from the aged soil samples using a base saponification procedure (27). The Rapidly Released Fraction, F. Average values of F for 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P decreased in the order 0.72 > 0.58 > 0.05. This corresponds to the fraction of compound released between zero and 12, 48, and 24 h respectively for 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P VOL. 35, NO. 6, 2001 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1115
(Figure 4A-C). As outlined in the result section, the fitted values of F for 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P did not display any distinct trend with increased soil aging. The small but distinguishable differences in the compound dissolution plots (previously discussed) are not sufficient to make a discernible change in the fitted F value for the three PAHs. This suggests that the fraction of compound available for rapid dissolution into the aqueous phase remains unaffected by increasing compound sequestration with aging. If we assume that the fraction of rapidly released compound approximates, or is equal to the bioavailable compound fraction, our results suggest the observed increase in compound sequestration may not have affected the bioavailability of the aged compounds. However, in the absence of a direct measure of compound bioavailability in our experiment we must treat this assumption with a degree of caution. In the previous paper we reported increased sequestration of 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P in this soil with increased aging (27). The percentage of nonextractable compound increased over the 525 day experimental period from 1.8 to 5.2%, 3.0 to 9.3%, and 6.3 to 22.7%, respectively, for 14C-9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P. Since the amount of 14C-9-Phe sequestered over the experimental period is very small, we are not surprised by the absence of a corresponding change in the amount of rapidly released compound, F. However, since 14C-4,5,9,10-Pyr and 14C-7B[a]P displayed a significant increase in the amount of sequestered compound with increased aging, we expected to observe a corresponding decrease in F values. This is illustrated by the results obtained for 14C-7-B[a]P, which displayed a significant increase in the amount of sequestered compound during 525 days aging in soil (27) without a significant change to the fitted F values. Values for F and k1 can be estimated with high precision if a sufficient number of samples are taken near the beginning of the dissolution study between the times 1/k1 and 1/k2 (32). The order of precision for the fitted parameters is generally F > k1 > k2 (36). Since the bend in the dissolution curve (or change in the rate of compound dissolution) defines F, a large ratio of k1/k2 provides a well-defined change in the shape of the dissolution curve. The ratios of k1/k2 for all our study compounds are large, indicating that a relatively precise estimate for F has been obtained (36). Therefore, we can be confident that the absence of a trend in the fitted F values with increased aging is not the product of obtaining insufficient data points for the experimental fitting and artifact of the experimentally generated data. There are a number of reasons that may explain why we did not observe any significant change in F values for 14C9-Phe, 14C-4,5,9,10-Pyr, and 14C-7-B[a]P or in k1 for 14C-9Phe and 14C-4,5,9,10-Pyr with increased aging. First, microbial degradation and/or abiotic loss mechanisms, including volatilization, leaching, and photodegradation, significantly decrease the amount of labile PAHs in soil, particularly PAHs with four or fewer aromatic rings (37, 38). Since these processes were absent in this experiment (27, 28), the fraction of labile compound in the soil was not markedly reduced during the 525 day aging period. Second, the fraction of each compound released from soil into the aqueous phase at each sampling time was much less than the amount extracted from the soil by solvent extraction and sequential MSE (27). While an increasing fraction of each PAH was sequestered during the experiment, it may not have been enough to significantly decrease the size of the labile compound fraction. A sizeable portion of compound appears to have remained available for partitioning to the aqueous phase at all sampling times. Last, the fitted values for F may merely reflect changes to the physical structure of the soil aggregates. Water-stable soil particles that survive mild agitation in aqueous suspen1116
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 35, NO. 6, 2001
sion can be aggregated (39). The action of mixing and abrasion of the soil particles will result in a degree of disaggregation of larger and less water-stable soil aggregates to much smaller particle size ranges. Wu and Gschwend noted a substantial shift from larger (840 to 177 µm) to smaller (
