Irreversible Adsorption of Chlorinated Benzenes to Natural Sediments

Environ. Sci. Technol. 2000, 34, 385-392

Irreversible Adsorption of Chlorinated Benzenes to Natural Sediments: Implications for Sediment Quality Criteria W . C H E N , † A . T . K A N , * ,‡ A N D M. B. TOMSON‡ Department of Environmental Science and Engineering, MS-317, Rice University, Houston, Texas 77005 and Brown and Caldwell, Inc., 1415 Louisiana, Suite 2500, Houston, Texas 77002

Desorption of many hydrophobic compounds from natural sediments has been observed to be biphasic, containing reversible and irreversible compartments. Therefore, it is necessary to reevaluate the currently adopted sediment quality criteria (SQCs) that are based on equilibrium partitioning models. In this study, the characteristics of irreversible adsorption were further studied with five chlorinated benzenes and four natural sediments. The five compounds differ significantly in Kow and solubility, and the four sediments cover a wide range of organic carbon content from 0.27 to 4.1%. With each chemical-sediment combination, a irr , fixed maximum irreversible adsorption capacity, qmax was observed. The apparent organic carbon-based partition coefficient associated with this irreversible fraction is essentially constant for different chemical-sediment 5.42(0.17. The desorption data combinations, with Kirr oc ) 10 were modeled with a previously proposed irreversible adsorption-desorption model, in which a Langmuiriantype expression (representing the irreversible fraction) is added to a linear model. For the five chlorinated benzenes and Dickinson sediment, there is little correlation between irr qmax and Kow. For 1,4-dichlorobenzene in different irr is proportional to soil organic carbon sediments, qmax content. The potential impact of this model on sediment quality criteria is also discussed.

Introduction Sediment contamination is a serious problem. The U.S. EPA estimates that approximately 10%s1.2 billion cubic yardss of the sediment underlying the nation’s surface waters is contaminated with toxic pollutants (1). These contaminated sediments serve as pollutant reservoirs and therefore pose ecological and human health risks for prolonged periods of time. Sediment-related problems also impose a significant economic impact on the country. For many years, both government and research institutes have tried to streamline a strategy to control and improve sediment quality. Unfortunately, these practices have been greatly hindered by our lack of knowledge on the uptake, release, toxicity, mobility, availability, and biodegradability of contaminants in aquatic systems. * Corresponding author phone: (713)285-5224; fax: (713)285-5203; e-mail: [email protected]. † Brown and Caldwell, Inc. ‡ Rice University. 10.1021/es981141s CCC: $19.00 Published on Web 12/31/1999

 2000 American Chemical Society

Adsorption and desorption have received considerable attention as the most important processes controlling the interaction between hydrophobic organic contaminants and sediments. The equilibrium partition approach is widely used to guide sediment quality management and decision-making (2). Numerous studies have observed that the release of organic contaminants is biphasic, including an equilibrium fraction and a highly resistant (irreversible) fraction (3-17). For sediment-associated hydrophobic organic contaminants, the irreversibly adsorbed fraction is the greatest concern and uncertainty, because it significantly affects chemical fate, toxicity, risk to human and aquatic life, and efficiency of most remediation technologies (10, 18, 19). Therefore, characteristics of desorption from the irreversible compartment need to be further understood to develop more useable adsorption-desorption models (10, 18-20). Previous research by the authors (15, 18-22) has shown that, for most chemical-sediment combinations, a significant fraction of the chemical is irreversibly bound to the sediment. The irreversible compartment has a finite maximum capacity: approximately 1/3-1/2 of the initially sorbed chemical is irreversibly boundsup to a specific maximum for each sediment-compound combination. At high solid-phase concentration ranges, the overall desorption is dominated by desorption from the reversible fraction, and this desorption exhibits an apparently linear isotherm that is predictable from conventional Koc-Kow relationships. At low-concentration loadings, still only 1/3-1/2 of the sorbed compound is irreversibly bound (15, 21). After the maximum irreversible capacity is reached, subsequent adsorption and desorptions become reversible. This finite maximum capacity is different for different compound-sediment combinations. The apparent Koc values associated with the irreversible fraction are nearly constant for most chemical-sediment combinations. Recent studies (19, 20) have suggested that the observed biphasic field and laboratory data can be modeled by incorporating the contribution of the irreversible compartment into a conventional linear adsorption-desorption model. The present authors (18) proposed an irreversible adsorption-desorption isotherm, based upon laboratory-observed parameters. The isotherm predicts that the corresponding aqueous concentration can be orders of magnitude different from that predicted from conventional Koc-Kow relationships. The generality and limitation of this isotherm have been examined with desorption data of aged chlorinated hydrocarbons from Lake Charles sediment (19). However, since this sediment was weathered for years, the observed results only cover a relatively small range of the proposed isotherm. In this paper, the irreversible adsorption compartment was further characterized using five chlorinated benzenes and four natural sediments. The compounds include 1,2dichlorobenzene, 1,4-dichlorobenzene, 1,2,4-trichlorobenzene, 1,2,3,4-tetrachlorobenzene, and hexachlorobenzene. All five compounds belong to the same chemical class but cover a wide range of solubility and Kow values. The four sediments used in this study varied significantly in physical and chemical properties such as particle size and organic carbon content. The sediments were first saturated with one of the five chlorinated benzenes and then stripped extensively with electrolyte solution and Tenax resin until the irreversible fraction was reached. The properties of the irreversible adsorption compartment of each chemical-sediment combination are compared. The dependency of the irreversible adsorption capacity on chemical Kow value and sediment organic carbon content is also studied. The observations of VOL. 34, NO. 3, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

385

TABLE 1. Typical Experimental Protocolsa experiment 1: 1,2-dichlorobenzene

experiment 2: 1,2,4-trichlorobenzene experiment 3: 1,4-dichlorobenzene experiment 4: 1,2,3,4-tetrachlorobenzene experiment 5: hexachlorobenzene

experiment 6: 1,4-dichlorobenzene



Dickinson Sediment 6 repetitive adsorptions (C0 ∼ 50 mg/L, 3-5 d per step); 15 repetitive desorptions (1-4 d, each); 1 adsorption (C0 ∼ 50 mg/L, 5 d); 18 repetitive desorptions (2-10 d, each); 1 Tenax desorption (3 d); 5 repetitive desorptions (6 h-7 d, each) 10 repetitive adsorptions (C0 ∼ 17 mg/L, 3-5 d); 12 repetitive desorptions (1-4 d); 1 adsorption (C0 ∼ 17 mg/L, 5 d); 17 repetitive desorptions (2-10 d); 1 Tenax desorption (3 d); 5 repetitive desorptions (6 h-7 d) 1 continuous adsorption (4 weeks); 5 repetitive desorptions (2-10 d); 3 Tenax desorptions (3-7 d); 21 repetitive desorptions (12 h-10 d) 1 continuous adsorption (4 months); 3 repetitive desorptions (3-6 d); 3 Tenax desorptions (2-4 d); 13 repetitive desorptions (1-23 d); 3 Tenax desorptions (1-5 d); 7 repetitive desorptions (1-13 d) 1 continuous adsorption (4 months); 3 repetitive desorptions (3-6 d); 3 Tenax desorptions (2-4 d); 7 repetitive desorptions (1-23 d) Lula Sediment 1 continuous adsorption (65 d); 8 repetitive desorptions (2-6 d); 3 Tenax desorptions (4-6 d); 6 repetitive desorptions (1-8 d) Lake Charles Sediment 1 continuous adsorption (65 d); 8 repetitive desorptions (2-6 d); 3 Tenax desorptions (4-6 d); 6 repetitive desorptions (1-8 d); 2 Tenax desorptions (2 d); 7 repetitive desorptions (1-13 d) Utica Sediment 1 continuous adsorption (65 d); 8 repetitive desorptions (2-6 d); 3 Tenax desorptions (4-6 d); 6 repetitive desorptions (1-8 d); 2 Tenax desorptions (2 d); 7 repetitive desorptions (1-13 d)

a Solution matrix: 0.01 M NaCl, CaCl , and NaN . Solid/solution ratio: 2 g/42 mL. Mixing: tumbling at 1 rpm at room temperature. Continuous 2 3 adsorption: 10 g of sediment and solid form of compound in a dialysis bag equilibrated with 35 mL of electrolyte solution. Repetitive adsorption: 90% supernatant replaced with freshly prepared sorbate solution. Tenax desorption: Tenax replaced with 0.25 g of clean Tenax. Repetitive desorption: 90% supernatant replaced with fresh electrolyte solution.

this study cover a wider range of experimental conditions and chemical and sediment types and, therefore, extend the range of applicability of the previously proposed irreversible adsorption-desorption model. Implications of the results to sediment quality criteria and management will also be discussed.

Experimental Section Sorbents. Four sediments, including Dickinson, Lula, Lake Charles, and Utica sediments, were used as sorbents in this study. The physical and chemical properties of these sediments have been reported previously (15, 18, 20, 23). Both Dickinson and Lula sediments do not contain a detectable quantity of hydrocarbon pollutants. Utica sediment was contaminated with a number of polycyclic aromatic hydrocarbons (PAHs). Lake Charles sediment was contaminated with various compounds including petroleum hydrocarbons, PAHs, as well as polychlorinated aromatic and aliphatic organic compounds. Dickinson, Lula, and Utica sediments were dried, ground, and sieved before being used in experiments. Upon receipt, Lake Charles sediment was centrifuged and stored in a refrigerator. Wet sediment was used in all Lake Charles experiments. The organic carbon (OC) contents of Dickinson, Lula, Lake Charles, and Utica sediments are 1.50, 0.27, 4.10, and 2.75% respectively. Sorbate and Chemicals. 1,4-Dichlorobenzene (Matheson Coleman & Bell, Los Angeles, CA), 1,2,3,4-tetrachlorobenzene, and hexachlorobenzene (Ultra Scientific, North Kingstown, RI) were used as received. Stock solutions of 1,2-dichlorobenzene and 1,2,4-trichlorobenzene (Ultra Scientific, North Kingstown, RI) were prepared in methanol. EPA Chlorinated Hydrocarbons Mix 8120 (Supelco, Inc., Bellefonte, PA) was diluted in isooctane and used as a GC/ECD calibration standard. Aqueous solutions of 1,2-dichlorobenzene and 1,2,4-trichlorobenzene were prepared in an electrolyte solution containing 0.01 M NaCl, CaCl2, and NaN3, respectively, 386

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 3, 2000

before each adsorption experiment. Previous studies have shown little effect on irreversible sorption from widely differing electrolyte and buffer solutions (15, 16, 18). Adsorption and Desorption Experiments. In Table 1, the experimental parameters for the eight adsorption and desorption experiments are listed. In experiments 1 and 2, the sediments were saturated with 1,2-dichlorobenzene and 1,2,4-trichlorobenzene, respectively, using the repetitive batch adsorption approach (21). First, the aqueous solution of each compound was prepared by mixing a certain amount of methanol stock solution in 100 mL of electrolyte solution, so that the concentration is approximately half of the solubility. Approximately 42 mL of the solution was added to a 45-mL glass vial containing 2 g of sediment to initiate adsorption. Then the vial was sealed with a Teflon-lined cap (Fisher Scientific, Fair Lawn, NJ) and tumbled at 1 rpm at room temperature for 3-5 days. Afterward, the vial was centrifuged at 3000g for 1 h. Ninety percent of the supernatant was separated from the sediment, and the aqueous-phase concentration was measured. Finally, the vial was filled with freshly prepared 1,2-dichlorobenzene or 1,2,4-trichlorobenzene solution to conduct another adsorption. Six repetitive adsorptions were completed in experiment 1, and 11 adsorptions were completed in experiment2. In experiments 3-8, the sediments were saturated with one of the chlorinated benzenes using a continuous adsorption approach. About 10 g of sediment was added to a 45-mL glass vial. A measured mass of compound was added to a dialysis bag, and the bag was added to the vial. The amount of compounds added to the dialysis bag was estimated to be in excess of the amount needed to saturate the solution and sediment. The vial was immediately filled with electrolyte solution and tumbled at 1 rpm for 28-122 days. The aqueousphase concentration was measured periodically during adsorption. At the end of the continuous adsorption, the vial containing sediment and solution was hand-shaken vigorously for 1 min, and 0.5 mL of the sediment slurry was

TABLE 2. Summary of the Results of Adsorption and Desorption Experiments adsorption exp

sediment (% OC)

compound

1 2 3 4 5 6 7 8

Dickinson (1.5) Dickinson (1.5) Dickinson (1.5) Dickinson (1.5) Dickinson (1.5) Lake Charles (4.1) Lula (0.27) Utica (2.8)

1,2-DCB 1,2,4-TCB 1,4-DCB 1,2,3,4-TeCB HCB 1,4-DCB 1,4-DCB 1,4-DCB

desorption

log Kow

q0b (µg/g)

log Koc (L/kg of OC)

d qirr max (µg/g)

e Cmax w (µg/L)

log Kirr oc (L/kg of OC)

3.38 4.00 3.38 4.55 5.50 3.38 3.38 3.38

960a 442a 562 688 6.12 2330 116 2660

3.05 ( 0.05c 3.42 ( 0.06c 2.78 3.87 4.93 2.97 2.86 3.18

6.37 13.6 14.5 15.3 0.450 271 0.970 202

0.81 ( 0.19 1.3 ( 0.29 1.4 ( 0.55 2.0 ( 0.53 0.15 ( 0.02 6.9 ( 1.1 0.92 ( 0.16 3.28 ( 0.16

5.44 ( 0.10 5.46 ( 0.11 5.38 ( 0.17 5.33 ( 0.10 5.31 ( 0.05 5.32 ( 0.08 5.26 ( 0.08 5.81 ( 0.02

a Concentrations at the end of repetitive adsorptions, calculated from mass balance. b Measured by solid extraction after continuous adsorption. Average values of individual steps in repetitive adsorptions. d Values fitted with the least-squares method using the irreversible isotherm proposed by Kan et al. (19) and experimental adsorption and desorption data. Kow values were obtained from Schwarzenbach et al. (20). e Cmax w , the average concentration observed in the last few desorption steps. c

immediately taken by a syringe and transferred to a 20-mL glass vial to measure the solid-phase concentration. Similarly, about 9 mL of slurry was withdrawn from the vial and transferred to a 45-mL glass vial to conduct desorption experiments. The amount withdrawn was weighed, and the mass of sediment transferred was calculated on the basis of the sediment/solution ratio used in the adsorption experiment. Desorption was conducted with both electrolyte solution and Tenax. In repetitive desorption experiments, the sediment was equilibrated with about 42 mL of electrolyte solution for specific periods of time (Table 1). Then, the vial was centrifuged at 3000g for 1 h, and about 90% of the supernate was removed. A portion of the supernate was extracted to measure the aqueous-phase concentration. Fresh electrolyte solution was then added to the vial to initiate another desorption. Tenax desorption was designed to accelerate desorption, and the procedures have been discussed in detail in a previous paper (22). In each Tenax desorption, about 0.3-0.5 g of Tenax was added to the vial. The vial was tumbled for 3 days, and Tenax was removed and extracted with acetone to determine the amount of that compound desorbed. Next, clean Tenax resin was added to initiate additional desorption. During successive desorptions, the sediment-phase concentration was measured for selected experiments to obtain mass balance. Also, for all the experiments, the sediment-phase concentration was measured upon the completion of the experiment to determine the amount of compound remaining. To determine solution-phase concentrations, aqueous solutions were extracted with isooctane. About 35 mL of aqueous solution was transferred to a 45-mL glass vial, and 1,2-dichlorobenzene or 1,2,4-trichlorobenzene solution was added as a surrogate standard. Then 5 mL of isooctane was added to the vial. The vial was sealed immediately and handshaken vigorously for 5 min. Next, it was left undisturbed until the solvent and aqueous phases separated. The isooctane volume was reduced to 0.5-1.0 mL and analyzed with GC/ECD. The detection limits on the GC for the compounds used in this study are from 0.5 µg/L for hexachlorobenzene to 3 µg/L for 1,2-dichlorobenzene. The concentration factor during extraction varied from 35 to 70, as needed, which yielded a detection limit in the original sample of 0.007-0.04 µg/L. Concentrations of chlorinated benzenes in sediments were determined by a method similar to that developed by Huang and Pignatello (24). About 0.25 g of wet sediment was transferred to a 20-mL glass vial, and a surrogate standard (in methanol solution) was added. Then methanol/water solution (85%:15%, v:v) was added to the vial. The vial was sealed, sonicated for 20 min, and horizontally shaken in a water batch at 70 °C for 16 h. Finally,

the vial was centrifuged at 3000g for 30 min, and the supernate was analyzed with GC/ECD.

Results Adsorption and Resistant Desorption. Adsorption and desorption experiments were conducted with five chlorinated benzenes. These chlorinated benzenes differed widely in solubility and Kow. In the continuous adsorption experiments, the adsorption time for 1,4-dichlorobenzene was from 1 (experiment 3) to 2 months (experiments 6-8), while in experiments 4 and 5, sediments were equilibrated with 1,2,3,4-tetrachlorobenzene and hexachlorobenzene solutions, respectively, for as long as 4 months. For the repetitive adsorption of 1,2-dichlorobenzene (experiment 1) and 1,2,4trichlorobenzene (experiment 2), the time for each adsorption step was from 3 to 5 days. The overall contact time for adsorption was 22-35 days. Previously, we estimated that the equilibrium time for adsorption was from 1 day to 1 week for compounds with Kow values from 102.7 to 106.3 (18). Similarly, according Huang and Weber (25), equilibrium would have been reached within these times because all of the samples we used would be considered “geologically young” in their classification. Therefore, under the protocols of this research, it is reasonable to assume that adsorptions have reached equilibrium in all of the experiments. Previous experiments with naphthalene have shown that, if the equilibrium solution-phase concentration is greater than about 1/3-1/2 of the aqueous solubility, the irreversible compartment can be filled in one step (18, 19, 21). In experiments 1 and 2, the equilibrium concentrations of 1,2dichlorobenzene and 1,2,4-trichlorobenzene at the end of repetitive adsorptions were both higher than half of their respective solubilities. In all of the continuous adsorption experiments (experiments 3-8), the aqueous concentrations were observed to be significantly higher than half of the solubility during the adsorption. The results of the adsorption experiments are summarized in Table 2. In experiments 1 and 2, adsorption was initiated with solutions containing high concentrations of 1,2-dichlorobenzene and 1,2,4-trichlorobenzene, respectively. Repetitive adsorption was continued until there was no significant uptake of sorbate from the solution applied to the sediment. For both compounds, the organic carbon-normalized partition coefficient, Koc, was similar in each individual adsorption step. The average log Koc value of 1,2-dichlorobenzene was 3.05 in six successive adsorptions, with a small standard deviation of 0.05. The average log Koc value for 1,2,4trichlorobenzene during 11 successive adsorptions was 3.42 ( 0.06. These values are comparable to their Kow valuess 103.38 and 104.00, respectively (26). They are also similar to the values observed previously using Lake Charles sediment and VOL. 34, NO. 3, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

387

solutions containing much lower aqueous concentrations of these two compounds (19, 22). In experiments 3-8, the contaminants were continuously loaded onto the sediments until the maximum adsorption capacity was reached. Both the aqueous- and solid-phase concentrations were measured at the end of the continuous adsorption. The Koc values of the adsorption experiments were calculated as the ratio of the organic carbon-normalized solid-phase concentration to the equilibrium aqueous-phase concentration. The observed Koc values of 1,2,3,4-tetrachlorobenzene (experiment 4) and hexachlorobenzene (experiment 5) were 103.87 and 104.93, respectively, and were also comparable with their Kow valuess104.55 and 105.50, respectively (26). The Koc values of 1,4-dichlorobenzene observed among four different sediments were similar. They were 102.78 in Dickinson sediment, 102.97 in Lake Charles sediment, 102.86 in Lula sediment, and 103.18 in Utica sediment. The desorption of these compounds deviated significantly from adsorption after a few desorption steps. For example, in experiment 3, the observed Koc value of 1,4-dichlorobenzene was 102.78 at the end of adsorption. The value increased to 103.94 after only five repetitive desorption steps. Similarly, in both experiments 6 and 8, the Koc values of 1,4-dichlorobenzene increased by a factor of 10 in eight desorption steps. The solid-phase concentrations in these experiments eventually dropped to a relatively constant value after extensive desorption with both electrolyte solution and Tenax. The five compounds used in this study cover a wide range of solubility: 1,2-dichlorobenzene (130 mg/L), 1,4-dichlorobenzene (60 mg/L), 1,2,4-trichlorobenzene (41 mg/L), 1,2,3,4tetrachlorobenzene (8.2 mg/L), and hexachlorobenzene (0.0058 mg/L). Thus, the equilibrium concentrations of these compounds differed considerably at the end of adsorption. With repetitive desorption, the aqueous-phase concentrations dropped exponentially in the first few desorption steps. Afterward, very little desorption was observed in each desorption step. Tenax was used to accelerate the desorption. The aqueous concentrations of these five compounds eventually leveled off at relatively constant values after extensive desorption. The final solution-phase concentrations of these five compounds were similar, close to one or a few micrograms per liter (column 8, Table 2). These solutionphase concentrations are comparable to that of a previous study (22) in which concentrations of 1,2-dichlorobenzene and 1,2,4-trichlorobenzene in Lake Charles sediment leveled off to between 1.2 and 2.6 µg/L, respectively, after 18 repetitive desorptions with electrolyte solution. Similar to previous observations (15, 18, 19, 21), the nearconstant concentrations observed in this study appear to be independent of desorption time between 2 and 24 days. For the last several desorption steps in experiments 1-5, the desorption time was varied from 3 h to 24 days; desorption time of each step was chosen randomly and was not necessarily increased from one step to the next step. In Figure 1 are plotted the aqueous-phase concentrations versus the time applied in each desorption step. Among the five compounds, 1,2-dichlorobenzene and 1,2,4-trichlorobenzene showed some time dependency between 1 and 2 days of desorption. For instance, the aqueous-phase concentration of 1,2,4-trichlorobenzene at the end of a 1-day desorption was 0.8 µg/L, while a concentration of 1.3 µg/L was observed in a 2-day desorption step. However, the concentration did not change appreciably when desorption time was increased to 7 days in another desorption step. The aqueous concentration of 1,2,3,4-tetrachlorobenzene was nearly identical in six successive desorption steps in which the desorption time varied from 18 h to 4 days. The concentration increased slightly when the desorption time was increased to 13 days. The aqueous-phase concentrations of 1,4-dichlorobenzene and hexachlorobenzene did not show statistically significant 388

9


variation with desorption time, even with desorption times as long as 24 days (experiment 5). Therefore, desorption from the irreversible compartment may have reached equilibrium. Irreversible Partition Coefficient, Kirr oc , and Maximum irr Irreversible Capacity, qirr max. As will be shown below, Koc and qirr are the most important parameters characterizing the max irreversible adsorption compartment. In Table 2, the observed irr Kirr oc and qmax values of the five chlorinated benzenes are summarized with respect to the four sediments. The irreversible partition coefficient, Kirr oc, similarly to Koc, is defined to be the ratio of the organic carbon-normalized solid-phase concentration to the solution-phase concentration after extensive desorption, i.e., when the aqueous-phase concentration reaches the significantly low and constant value. As in previous studies (18, 27), when log Koc values are plotted versus successive desorption steps (data not shown), irr the curves increase to a fixed value (log Kirr oc). The Koc values irr listed in Table 2 were the average Koc values in the last 5-9 desorption steps, wherein the desorption was accomplished by successive replacement of the solution with clean electrolyte solution. Since both the solid and aqueous concentrations in each experiment were relatively constant for the last few desorption steps, the Kirr oc values in these steps were very close. This is indicated by the small standard deviations of each average Kirr oc value in Table 2. Interestingly, the Kirr oc values of these eight experiments were very close. The Kirr oc values of 1,2-dichlorobenzene, 1,2,4-trichlorobenzene, 1,2,3,4-tetrachlorobenzene, and hexachlorobenzene were 105.44, 105.46, 105.33, and 105.31, respectively, even though their Koc values in adsorption differ by more than 2 orders of magnitudes102.78-104.93. The Kirr oc values of 1,4-dichlorobenzene with respect to the four different sediments were also similar (experiments 3 and 6-8). The observed values are 105.38, 105.32, 105.26, and 105.81 for Dickinson, Lake Charles, Lula, and Utica sediments, respectively. The eight experi5.42(0.17. This is in ments gave an average Kirr oc value of 10 agreement with previous studies in which an average Kirr oc value of 105.53(0.48 was reported (18) for compounds with Kow values from 102.3 to 106.3 and sediments with organic carbon contents from 0.27% to 4.1%. These observations support the hypothesis that the Kirr oc value is independent of either chemical or sediment type. The qirr max values listed in Table 2 were obtained by curvefitting the experimental adsorption and desorption data using the irreversible adsorption-desorption isotherm proposed previously (18). The least-squares method was used to fit the irr experimental data, with qirr max as the only variable. The qmax values in the eight experiments varied from 0.450 to 271 µg/ g, depending on the specific chemical-sediment combination. Previous research (18, 21) has shown that the value of qirr max is independent of the number of adsorption steps used and the adsorption time, beyond a few days. Both Lake Charles and Utica sediments are field-contaminated sediments containing chlorinated benzenes and PAHs that are resistant to desorption. However, in these two fieldcontaminated sediments, the contaminant concentrations are considerably below the qirr max values in Table 2 (18). For example, the Lake Charles sediment, as received, contains 16.7 µg/g of 1,4-dichlorobenzene vs qirr max ) 271 µg/g (Table 2) and Utica sediment contains 1.0 µg/g of naphthalene, as received, vs qirr max ) 202 µg/g for 1,4-dichlorobenzene (Table 2). The low sediment concentrations in the field samples would indicate that the contaminants in the irreversible compartment may have been depleted during weathering.

Discussion Previous studies, using many classes of compounds and sediments, have shown that the unique characteristics of

FIGURE 1. Plot of the aqueous-phase concentrations, with respect to the irreversibly adsorbed fraction, of 1,2-dichlorobenzene, 1,4dichlorobenzene, 1,2,4-trichlorobenzene, 1,2,3,4-tetrachlorobenzene, and hexachlorobenzene with varied desorption times. The data points are observed aqueous-phase concentrations of the last 5-7 desorption steps in experiments 1-5. Desorption time in the last few desorption steps in each experiment was randomly chosen and was not necessarily increased from one step to the next step. the irreversible compartment may be related to the physical and chemical properties of both chemicals and sediments. In this study, experiments were designed so that these dependencies could be further understood. The eight experiments in this study fall into two divisionssfirst, the resistant desorption of closely related compounds in the same sediment (experiments 1, 2, 4, and 5); second, the resistant desorption of the same compound in different sediments (experiments 3 and 6-8). Also, most experiments conducted previously covered a relatively low range of both solid and aqueous concentrations. In this study, the sediments were saturated with chemicals, so that both the maximum adsorption capacity and the irreversible capacity were filled after adsorption. The observed adsorption and desorption behaviors in this study are consistent with the previous observations, even though the range of conditions has been

greatly expanded. First, a maximum irreversible adsorption capacity exists for each compound-sediment combination. Second, after about 1 or 2 days, the aqueous-phase concentration in equilibrium with the irreversible compartment is not time-dependent, indicating that the observed phenomenon is probably not a simple kinetic or diffusion-limited process. Third, Kirr oc is constant among different compounds and sediments. Irreversible Adsorption-Desorption Isotherm. An irreversible adsorption-desorption isotherm has been proposed by the authors (18) to describe the biphasic adsorption and desorption of both the laboratory and field observations. It is proposed that the labile fraction follows a linear isotherm, while the nonlinear behavior associated with the irreversible fraction can be described by a Langmuirian type of isotherm. VOL. 34, NO. 3, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

389

FIGURE 2. (a) Plot of the model predicted vs experimentally observed isotherms of 1,2-dichlorobenzene, 1,2,4-trichlorobenzene, 1,2,3,4tetrachlorobenzene, and hexachlorobenzene in Dickinson sediment. Solid lines are predicted isotherms using the irreversible adsorption model proposed by the authors (19). Dotted lines are the linear isotherms plotted using the Koc values obtained in adsorption experiments. The diamond symbols are experimental observations. (b) Plot of the model predicted vs experimentally observed isotherms of 1,4dichlorobenzene in Dickinson, Lake Charles, Lula, and Utica sediments. Solid lines are predicted isotherms using the irreversible adsorption model proposed by the authors (19). Dotted lines are the linear isotherms plotted using the Koc values obtained in adsorption experiments. The diamond symbols are experimental observations. The irreversible compartment has a well-delineated maximum adsorption capacity, qirr max. Thus, the overall isotherm is the combination of the isotherms contributed by both the labile and resistant fractions, as

q ) Koc focC +

irr Kirr oc focqmax fC irr qmax f + Kirr oc focC

(1)

where f is the fraction of qirr max filled during adsorption and is assumed to be 1 for all the experiments in this study because the irreversible compartment should have been filled based on the protocol of this study. As can be seen with eq 1 and in Figure 2, the specific value of f is only important in the plateau transition region of the isotherm; this was discussed previously (18, 19). In Figure 2, the experimentally observed adsorption and desorption data in experiments 1-8 are fitted with both the linear reversible isotherm and the irreversible adsorptiondesorption isotherm. The solid lines in Figure 2 are fitted isotherms using eq 1 and the observed Koc and Kirr oc values from the adsorption and desorption experiments (summarized in Table 2). The dotted lines are the linear isotherms plotted with the Koc values observed in adsorption experi390

9


ments (in Table 2). The data points are experimental irr observations. Similar values for Koc, qirr max, and Koc can be obtained simultaneously using eq 1 in a rigorous nonlinear least-squares program. For example, using the PSI-Plot program, the 1,4-dichlorobenzene data with Lula sediment in Figure 2 yielded log Koc ) 2.80, qirr max ) 0.96 µg/g, and log Kirr oc ) 5.52, very similar to the values reported in Table 2. As shown in this figure, the irreversible isotherm and the linear isotherm overlap at high aqueous- and solid-phase concentrations. As the aqueous- and solid-phase concentrations decrease, the irreversible isotherm deviates increasingly from the linear model and on a log-log plot eventually becomes parallel to the linear isotherm. This can be illustrated using eq 1. At high aqueous-phase concentration, C, the Langmuirian term reduces to qirr max, which is usually considerably smaller than the solid-phase concentration in the reversible fraction. Thus, the total sediment-phase concentration is dominated by the contribution of the linear fraction. At very low aqueous concentration, the irreversible isotherm reduces to Kirr oc focC. This equation is similar to the linear model, except that Koc is substituted with Kirr oc. At intermediate concentrations, neither the contribution of the linear fraction nor that of the irreversible fraction is negligible.

FIGURE 3. Graphic illustration of the impact of the proposed irreversible adsorption isotherm on sediment quality criteria (SQC). The solid line and dotted lines are the predicted isotherms of 1,4-dichlorobenzene in Lake Charles sediment with the irreversible model and the equilibrium model, respectively. As in Figure 2, the experimentally observed desorption results were well-described by the irreversible isotherm. The isotherms of 1,2-dichlorobenzene, 1,2,4-trichlorobenzene, 1,2,3,4-tetrachlorobenzene, and hexachlorobenzene in Dickinson sediment are plotted in Figure 2a. These compounds differ significantly in both Kow value and aqueous solubility; however, the shapes of these isotherms are similar. The resistance of different compounds to desorption is also clearly illustrated with these isothermssthe deviation of desorption from sorption decreases as the hydrophobicity increases. For compounds with small Kow values, such as dichlorobenzene, the difference between the reversible and irreversible model is more significantsas indicated by the fact that the irreversible isotherm deviates significantly from the linear reversible isotherm. For compounds with large Kow values, such as hexachlorobenzene, the irreversible isotherm comes closer to the linear isotherm, which is consistent with the previous paper (18). In Figure 2b, the isotherms of 1,4dichlorobenzene in the four different sediments are plotted. Again, the irreversible isotherms have the same shape, while the plateaus shift among different experiments. This is because the organic carbon contents of these four sediments are very differentsfrom 0.27% for Lula sediment to as high as 4.1% for Lake Charles sediment. This difference in organic carbon content results in a large difference among the irreversible adsorption capacities, qirr max. It is important to notice that although the irreversible model was developed with experimental data covering relatively low concentration range, it is equally effective for the data obtained in this study, which cover a wide range of both solid- and solution-phase concentrations. For example, in experiment 6, the solution-phase concentrations of 1,4dichlorobenzene at the end of adsorption steps was 59 mg/ L, near the solubility of this compound. After extensive desorption, the concentration dropped to less than 1.0 µg/L, about 5 orders of magnitude lower. The solid-phase concentration in this experiment also covered a range of more than 2 orders of magnitude. The experimental results, however, were well-represented by the proposed irreversible isotherm. Dependency of qirr max on Kow and foc. Among the parameters describing the properties of the irreversible adsorption,

irr qirr max is probably the most difficult one to determine. The qmax values in all the experiments were obtained by curve-fitting the experimental data using eq 1. In previous studies, it was proposed that the irreversible adsorption was a consequence of an interaction between hydrocarbon molecules and organic matters in sediments. Therefore, for the adsorption and desorption in a specific sediment, the qirr max value should be dependent on the properties of the organic adsorbate. Nevertheless, the qirr max values seem to be poorly related to the Kow values in this irr study. In experiments 1-5, the observed qmax values with respect to Dickinson sediment were 6.37, 14.5, 13.6, 15.3, and 0.453 µg/g for 1,2-dichlorobenzene, 1,4-dichlorobenzene, 1,2,4-trichlorobenzene, 1,2,3,4-tetrachlorobenzene, and hexachlorobenzene, respectively. These values are approximately constant except for hexachlorobenzene. In a previous study (19), it was found that the organic carbon-normalized irr qirr max valuesqmax/focsvaried less than 1 order of magnitude when Kow value increased from 102.4 to 106.4. The qirr max of hexachlorobenzene observed in this study was much lower than those for the other four compounds. This may be due to the small maximum adsorption capacity of hexachlorobenzene in this sedimentsonly about 6.12 µg of hexachlorobenzene per g of Dickinson sediment (Table 2, column 4). In other words, even though hexachlorobenzene has a high affinity for organic matter, as indicated by its high Kow value, its qirr max value is low. One measure of desorption resistance might be the percentage of originally sorbed compound (q0), which is eventually identified as irreversibly sorbed (qirr max, see Table 2). This varies from 0.66% for 1,2dichlorobenzene on Dickinson sediment to 12% for 1,4dichlorobenzene on Lake Charles sediment. For sorption to Dickinson sediment, the percentage irreversibly bound appears to increase with log Kow (r 2 ) 0.74), but clearly more work is needed to establish such a correlation. For the adsorption of a certain compound to different sediments, the irreversible capacities of this compound in different sediments are proportional to the organic carbon 3.80 f content of these sediments. A correlation of qirr oc max ) 10 was obtained with the qirr values of 1,4-dichlorobenzene in max the four sediments (experiments 3 and 6-8). This equation

VOL. 34, NO. 3, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

391

is in agreement with a recent study (19) in which the authors summarized the experimental data in many field and laboratory studies. Implications for Sediment Quality Criteria. Thus far, regulations such as the EPA’s sediment quality criteria (SQCs), sediment quality advisory levels (SQALs), and chemicalspecific fate scores are directly or indirectly derived on the basis of the linear equilibrium partition model (2). Since most chemical-sediment combinations exhibit an irreversible adsorption compartment, it is necessary to reevaluate the validity of these regulations. The purpose of SQCs is to ensure that the pore water concentration of a certain compound does not exceed the final chronic water quality criteria (FCVs). Thus, the maximum sediment-phase concentration of a chemical is correlated to its pore water concentration using the linear equilibrium partition model as shown in eq 2 (2):

SQC (µg/g) ) Kp (L/kg) × FCV (µg/L) × l03 (kg/g)

(2)

This equilibrium partition approach often significantly overestimated the desorption and the associated risk to the aquatic environment. Since a chemical may exhibit great resistance to desorption, the real partition coefficient could be orders of magnitude higher than the equilibrium partition coefficient, Kp, as in eq 2. This effect is even more significant for compounds with lower Kow values. As in Table 2, the Kirr oc and Koc values of hexachlorobenzene only differed by a factor of 2.4, but the Kirr oc value of 1,4-dichlorobenzene is more than 2 orders of magnitude higher than its Koc value. Therefore, it is necessary to incorporate a more realistic model into the above equation to better reflect the potentially lower toxicity of sediment-associated organic chemicals. The following equation is proposed for the SQC:

SQC ) Kp × FCV × 103 +

irr 3 Kirr oc focqmax × FCV × 10 irr irr qmax + Koc foc × FCV × 103

(3)

The Kp value in this equation can be estimated readily with a number of equations [e.g., Kp ≈ 0.63Kowfoc (28)], Kirr oc is a irr 5.5 constant (Kirr oc ) 10 ), and the range of qmax can also be estimated (19). It should be noted that the specific values of qirr max and f used in eq 1 only impact the isotherm in the narrow plateau values of q (see Figures 2 and 3); therefore, f equal to 1 in eq 3 is a conservative assumption and has little impact on the overall implications of proposed SQC in eq 3. The impact of the proposed irreversible model on SQC can be illustrated with Figure 3. It is shown that the SQC of 1,4dichlorobenzene would be nearly 2 orders of magnitude less strict when the resistant fraction in sediment is taken into account. This implies that many contaminated sediment sites need only moderate treatment or can be safely left alone without any significant environmental concern.

Acknowledgments This research has been conducted with the support of Hazardous Substance Research Center South and Southwest, the Gulf Coast Hazardous Substance Research Center, Office of Exploratory Research of the U.S. Environmental Protection

392

9


Agency, and the Defense Special Weapon Agency. We also thank Dr. C. R. Demas of the U.S. Geological Survey Louisiana District for assistance in collecting Lake Charles sediments.

Literature Cited (1) EPA’s Contaminated Sediment Management Strategy; EPA 823R-97-001; U.S. Environmental Protection Agency: Washington, DC, 1998. (2) The incidence and severity of sediment contamination in surface waters of the United States, Vol. 1; EPA 823-R-97-006; U.S. Environmental Protection Agency: Washington, DC, 1997. (3) Hutchins, S. R.; Tomson, M. B.; Ward, C. H. Environ. Toxicol. Chem. 1983, 2, 195-216. (4) McCall, P. J.; Agin, G. L. Environ. Toxicol. Chem. 1984, 4, 37-44. (5) Coates, J. T.; Elzerman, A. W. J. Contam. Hydrol. 1986, 1, 191210. (6) Steinberg, S. M.; Pignatello, J. J.; Sawhney, B. L. Environ. Sci. Technol. 1987, 21, 1201-1208. (7) Vaccari, D. A.; Kaouris, M. J. Environ. Sci. Health 1988, A23, 797-822. (8) Pereira, W. E.; Rostad, C. E.; Chiou, C. T.; Brinton, T. I.; Barbar, L. B., II. Environ. Sci. Technol. 1988, 22, 772-778. (9) Weber, W. J., Jr.; Miller, C. T. Water Res. 1988, 22, 457-464. (10) Brusseau, M. L.; Rao, P. S. C. Crit. Rev. Environ. Control 1989, 19, 33-99. (11) Pignatello, J. J.; Frink, C. R.; Marin, P. A.; Droste, E. X. J. Contam. Hydrol. 1990, 5, 195-214. (12) Pignatello, J. J. Environ. Toxicol. Chem. 1990, 9, 1117-1126. (13) Ball, W. P.; Roberts, P. V. Environ. Sci. Technol. 1991, 25, 12371249. (14) Pavlostathis, S. G.; Jaglal, K. Environ. Sci. Technol. 1991, 25, 274-279. (15) Kan, A. T.; Fu, G.; Tomson, M. B. Environ. Sci. Technol. 1994, 28, 859-867. (16) Fu, G.; Kan, A. T.; Tomson, M. B. Environ. Chem. Toxicol. 1994, 13, 1559-1567. (17) McGroddy, S. E.; Farrington, J. W. Environ. Sci. Technol. 1995, 29, 1542-1550. (18) Kan, A. T.; Fu, G.; Hunter, M.; Chen, W.; Ward, C. H.; Tomson, M. B. Environ. Sci. Technol. 1998, 32, 892-902. (19) Chen, W.; Kan, A. T.; Fu, G.; Vignona, L. C.; Tomson, M. B. Environ. Toxicol. Chem. 1999, 18, 1610-1616. (20) Gustafsson, O.; Haghseta, F.; Chan, C.; MacFarlane, J.; Gschwend, P. M. Environ. Sci. Technol. 1997, 31, 203-209. (21) Kan, A. T.; Fu, G.; Hunter, M. A.; Tomson, M. B. Environ. Sci. Technol. 1997, 31, 2176-2185. (22) Chen, W. Sorption and desorption of hydrocarbons to and from historically-contaminated Lake Charles sediments. M.S. Thesis, Rice University, 1997. (23) Chandra, S. D.; Ward, C. H.; Hughes, J. B. Hazard. Waste Hazard. Mater. 1996, 13, 375-385. (24) Huang, L. Q.; Pignatello, J. J. J. Assoc. Off. Anal. Chem. 1990, 73, 443-446. (25) Huang, W.; Weber, W. J. Environ. Sci. Technol. 1998, 31, 225622569. (26) Schwarzenbach, R. P.; Gschwend, P. M.; Imboden, D. M. Environmental Organic Chemistry; John Wiley & Sons: New York, 1993. (27) Pignatello, J. J. Environ. Toxicol. Chem. 1989, 9, 1107-1115. (28) Karickhoff, S. M.; Brown, D. S.; Scott, T. A. Water Res. 1979, 13, 241-248.

Received for review November 4, 1998. Revised manuscript received September 20, 1999. Accepted November 11, 1999. ES981141S

Irreversible Adsorption of Chlorinated Benzenes to Natural Sediments

Recommend Documents