Studies on the Dissolution of Polycyclic Aromatic Hydrocarbons from

The Castle Business Park, Stirling FK9 4TR, U.K.. Assessment of risk and remediation strategies at contaminated sites requires that both the amounts o...
0 downloads 0 Views 134KB Size
Environ. Sci. Technol. 1999, 33, 2118-2126

Studies on the Dissolution of Polycyclic Aromatic Hydrocarbons from Contaminated Materials Using a Novel Dialysis Tubing Experimental Method P A U L A J . W O O L G A R * ,†,‡ A N D KEVIN C. JONES† Environmental Science Department, Institute of Environmental and Natural Sciences, Lancaster University, Lancaster LA1 4YQ, U.K., and Scottish Environment Protection Agency, Head Office, Erskine Court, The Castle Business Park, Stirling FK9 4TR, U.K.

Assessment of risk and remediation strategies at contaminated sites requires that both the amounts of contaminants present and their potential for release from materials and soils be evaluated. The release, or dissolution, of polycyclic aromatic hydrocarbons (PAHs) from contaminated materials to water was therefore investigated. To facilitate investigations of PAH dissolution from physically disparate materials such as solid coal tars, creosote, oil, and spent oxide, an experimental method for measuring dissolved PAHs was developed employing dialysis tubing in a batch-type system. This was validated and compared to aqueous-phase PAH concentrations measured using more traditional techniques and also predicted using Raoult’s law. The experimental procedure was successfully used to determine ‘near equilibrium’ aqueous concentrations of PAHs, but it could only be used to determine relative rates of approach to equilibrium as the dialysis tubing effected the rate constants. It was found that the contaminant materials influenced dissolution, in particular the close to equilibrium concentrations. For materials chemically similar to PAHs, such as nonaqueousphase liquids (NAPLs), the concentrations could be predicted using Raoult’s law. For materials that were chemically dissimilar to PAHs, such as spent oxide, release was more thermodynamically favorable than for NAPLs.

Introduction Polycyclic aromatic hydrocarbons (PAHs) are constituents of a number of anthropogenic materials that enter and persist in the environment and are of concern because individual compounds exhibit carcinogenic, mutagenic, and toxic effects (1). Residual PAH containing material, either mixed in soil or sediments or occurring as a discrete phase, can lead to the long-term desorption and dissolution of PAHs into groundwater (2). Dissolution is governed primarily by solubility, and as PAHs are sparingly soluble, depletion from the source material is slow. Furthermore, because PAHs are nonpolar and hydrophobic, they reside primarily in residual oil phases, * Corresponding author telephone: +44 1786 457700; fax: +44 1786 446885; e-mail: [email protected]. † Lancaster University. ‡ Scottish Environment Protection Agency. 2118

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 12, 1999

although partitioning may occur to other organic phases such as soil organic matter (3). Risk assessment and evaluation of remediation strategies at contaminated sites require that both the total amounts of contaminants present and their potential for release from materials, sediments, and soils be evaluated. There are two aspects to be addressed when considering release: the concentrations of the compounds released and the rates of release (4). Equilibrium conditions indicate the maximum concentration of compound that could potentially be present in solution, but as such a state is rarely achieved in the field, it tends to be the rate of dissolution that determines the impact on groundwater quality and the effectiveness of remediation (5). Robust and reproducible experimental methods are required to enable the dissolution of PAHs to be measured. Particular care needs to be taken as once in solution hydrophobic compounds such as PAHs are liable to become sorbed onto any exposed surface, in particular glass, Teflon, and other plastic surfaces. The sorption and desorption of compounds between soil and water has been widely investigated using static equilibration (batch) techniques (6). Batch procedures typically consist of mixing soil and water in a defined ratio for a set time period during which equilibrium is reached, followed by separation and analysis of the water and soil (7). Separation of water and soil is typically achieved using centrifugation, although nonsettling particles (NSPs) such as microparticles or organic macromolecules may remain in the solution phase even after ultracentrifugation (8-10). If sorption of PAHs onto these occurs, then the concentrations detected in the aqueous phase do not represent dissolved PAHs. The use of filtration as an alternative separation procedure is not without problems either, as hydrophobic compounds are liable to become sorbed onto filters (7) and undissolved droplets of material may pass through glass fiber filters giving erroneous results (11). The sequestration and dissolution of compounds between oily materials and water has been investigated, although not so widely as the sorption of compounds in soil- and sediment-water systems. Batch experiments have been successfully applied to coal tars where water is in direct contact and can be easily removed (12) and also to light nonaqueous-phase liquids (LNAPLs), where water is removed from underneath the nonaqueous layer by means of a stopcock (13, 14). One problem with batch experiments is that NAPLs tend to form emulsions when shaken (15), making separation of the aqueous phase difficult. Column experiments have also been used, either packing solid materials into a column (16) or coating liquid materials onto glass beads or firebrick support through which water can flow (17, 18). Given the fact that no one method could be applied to solid materials, aqueous materials, and nonaqueous materials that were both heavier and lighter than water, a method employing dialysis tubing was developed. The method had potential merit in that it addressed some of the problems associated with previously published methods. In particular, it offered an easy method of separation as by encasing a material in dialysis tubing, the transfer of microparticles and organic macromolecules from solid materials into the aqueous phase (19) and the introduction of NAPL emulsions into the aqueous phase was prevented. The dialysis method was based on similar techniques used by other researchers for measuring aqueous solubilities of 10.1021/es980638z CCC: $18.00

 1999 American Chemical Society Published on Web 06/15/1999

organic compounds (11), for sorption/desorption studies (20, 21), and for investigations of dissolved organic matterhydrophobic compound interactions (21-23). Sorption onto dialysis materials was a potential problem (20) but had not posed difficulties for some researchers where cellulose membranes were utilized and the systems were allowed to re-equilibrate to compensate for sorption (22, 23). More significant problems were retarded diffusion through the dialysis membrane (24) affecting the rate of dissolution and degradation of cellulose membranes. The validity of the dialysis method therefore needed to be established. The dialysis method was validated by using 14C tracer studies and by comparing results obtained to those from more commonly used batch-type methods utilizing centrifugation and direct contact. The method was then used to investigate the dissolution (and potentially desorption) of PAHs from 11 different materials, which provided further validation as the equilibrium concentrations measured were compared with those predicted using Raoult’s law (excluding those materials for which Raoult’s law was not applicable). The experimental results obtained have been used to comment on the ideality exhibited by the study materials and the affect that different contaminant materials have on the release of PAHs to water, considering both the approach to equilibrium and equilibrium aqueous phase concentrations.

Theory Dissolution (and desorption) arises out of equilibrium partitioning and mass transfer phenomena between two phases, one of which is an aqueous phase and the other is a material phase. Aqueous solubility is a fundamental parameter in assessing the rate and extent of dissolution (25). When dealing with liquid mixtures of organic chemicals dissolving in water, there is a need to account for dilution of the chemicals by other substances in the mixture. There is also the potential for the dissolution of other components of the mixture to enhance solubility (3). Raoult’s law can be applied to predict equilibrium aqueous-phase concentrations associated with complex liquid organic mixtures (e.g., refs 12, 13, and 26-29), and it assumes that the properties of the mixture are determined by the properties of its pure components and their concentrations, i.e., the chemicals behave ideally. Under these conditions, the concentration of a chemical in the aqueous phase in contact with a complex mixture can be predicted based on the following equation (4):

Cw ) XoSl

(1)

where Cw is the chemicals concentration in the aqueous phase (mol/L) in equilibrium with the organic phase, Xo is the mole fraction of the chemical in the organic phase, and Sl is the aqueous solubility of the pure liquid chemical (mol/L) that can be applied to solid solutes (all PAHs at standard state) by employing hypothetical subcooled liquid solubilities. At equilibrium, there is no net transfer between two phases, and partition coefficients can be used to represent the ratio of the concentration of the same chemical species in two phases. In some systems, it can take many years for equilibrium to be reached, and it is recognized that the partition coefficients derived in this paper may represent apparent coefficients. Partition coefficients are widely available for systems such as octanol-water and organic carbonwater, representing “solubilities” in one phase relative to that in water. Indeed, relative concentrations of PAHs between two distinct phases provide an explicit measure of the propensity of PAHs to exist in each phase (3). For material-water partitioning, coefficients (Kmw, L/kg) were defined as the ratio of the chemical concentration in the

material (Cm) and water phase (Ce) at equilibrium:

Kmw ) Cm (mg/kg)/Ce (mg/L)

(2)

In the experiments, the approach to equilibrium was measured, and mass transfer theory was used to derive firstorder mass transfer coefficients and equilibrium aqueousphase concentrations by fitting the data to

Cw ) Ce[1 - exp(-kt)]

(3)

where Cw is the aqueous-phase concentration at any point in time (ng/mL), k is the lumped mass transfer coefficient (h-1), Ce is the equilibrium aqueous-phase concentration (ng/ mL), and t is the contact time with water (h). Fitting of the data to the equation was achieved using a nonlinear curvefitting program (SigmaPlot), and 95% confidence limits for the two parameters were derived from the asymptotic errors generated. Such an approach has been used widely by researchers (e.g., refs 30 and 31). In the experiments described, fitting data to eq 3 enabled all the data points to be used in the calculations, thereby reducing the requirement for replication.

Materials and Methods Chemicals. Acenaphthene-d10 and chrysene-d12 were used as recovery standards, and phenanthrene-d10, benz[a]anthracene-d12, and perylene-d12 were used as internal standards. These deuterated PAHs were obtained as individual solutions prepared by Ultra Scientific, North Kingstown, RI. A PAH mixture prepared by the same company and containing the U.S. EPA priority PAHs was used for instrument calibration and check purposes as well as the spike studies. Cold and hot [14C]-9-phenanthrene supplied by Sigma, U.K., was utilized in the tracer studies. Milli-Q water was supplied by a Millipore water purification system. Silica gel 60 (70-230 mesh) was used as supplied by Merck, Lutterworth, U.K. All solvents used were supplied by Rathburns, U.K. and were glass distilled grade or better. Study Materials. Eleven materials of widely differing physicochemical characteristics and representing a range of industrial materials were chosen for study. The materials ranged from NAPLs of varying viscosity, some less dense and some more dense than water, to solid organic (coal tars) and inorganic (spent oxide) materials as well as an aqueousphase material (sewage sludge). Homogeneous samples were prepared by grinding and sieving the solids to a defined particle size (1-0.5 mm) and shaking the liquid materials. Further details on the study materials can be found in Supporting Information Table 1. Dialysis Tubing Experimental Method. The dialysis tubing method involved encasing the study material in a sealed length of dialysis membrane and placing this in contact with water, enabling dissolved PAHs to permeate through the membrane into the surrounding water. The procedure adopted consisted of introducing 2.5 mL of liquid materials or 2.5 g of solid materials as a water slurry into a 10 cm length of 3500 molecular weight cutoff regenerated cellulose tubing (Specta/Por 3, Medicell International Ltd, London, U.K.), knotted at one end and subsequently sealed with a Spectrum polyethylene closure of grip width 23 mm at the other (Medicell International Ltd, London, U.K.). The material contained in the tubing was then placed in a borosilicate glass reagent bottle coated with UV opaque film (100 mL nominal volume, Schott Duran), which was filled with 130 mL of Milli-Q water, fitted with a Teflon ring to assist pouring and sealed using a Teflon-lined screw cap. One bottle for each contact time period was placed horizontally on a flat bed shaker (IKA-labortechnik Janke & Kunkel KS250) rotating at 300 rpm in a room with the temperature controlled at 8 VOL. 33, NO. 12, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2119

2120

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 12, 1999

0.08 ( 0.034 0.96 194 ( 84 14 300 1450 ( 83 89 920 3.8 ( 1.8 5 209 158 ( 19 2 650 nd 2 968 2750 ( 54 112 980 28 ( 1.9 508 215 ( 34 12 937 nd 1 865

582 ( 52 25 646

0.01 ( 0.001 0.02 ( 0.004 0.04 ( 0.005 0.12 ( 0.023 0.02 ( 0.002 0.10 ( 0.054 0.12 ( 0.041 0.06 ( 0.025 0.08 ( 0.037 0.15 ( 0.087 0.09 ( 0.040 0.07 ( 0.032 994 ( 151 1400 ( 161 1220 ( 116 3250 ( 294 970 ( 162 2090 ( 232 1580 ( 78 632 ( 185 579 ( 181 724 ( 238 470 ( 144 197 ( 73 13 700 ( 438 1470 ( 130 8020 ( 566 18 700 ( 1290 8180 ( 1031 11 400 ( 1002 8500 ( 664 449 ( 0241 4290 ( 177 4730 ( 565 3620 ( 157 1370 ( 158 115 ( 2.8 1080 ( 63 1650 ( 36 1210 ( 153 959 ( 59 69 ( 3.3 81 ( 4.6 18 ( 1.8 11 ( 0.7 6.3 ( 2.6 5.4 ( 2.0 nd 7.0 ( 1.4 49 ( 6.1 128 ( 22 544 ( 27 162 ( 6.9 93 ( 1.3 346 ( 14 299 ( 28 335 ( 27 224 ( 17 248 ( 4.3 57 ( 7.9 351 ( 31 1060 ( 206 548 ( 56 744 ( 37 135 ( 11 11 ( 0.3 97 ( 12 9.0 ( 1.0 13 ( 1.6 nda nd nd 28 400 ( 2143 4450 ( 203 10 700 ( 576 22 600 ( 1994 4880 ( 464 11 100 ( 495 16 400 ( 1136 2290 ( 540 2160 ( 203 2920 ( 260 2840 ( 309 1490 ( 143 1.1 ( 0.3 1.5 ( 0.1 1.2 ( 0.8 121 ( 4.4 5.2 105 ( 8.3 65 ( 5.3 31 ( 1.3 50 ( 6.5 69 ( 7.9 5.8 ( 0.8 23 ( 1.9 39 ( 7.7 245 ( 53 428 ( 58 3540 ( 355 768 ( 80 595 ( 53 2780 ( 193 1200 ( 163 1970 ( 166 362 ( 28 709 ( 44 86 ( 27

acenaphthylene acenaphthene fluorene phenanthrene anthracene fluoranthene pyrene benz[a]anthracene chrysene benzofluoranthenes benzo[a]pyrene indeno[123-cd]pyrene/dibenz[ah]anthracene benzo[ghi] perylene total PAH

60 ( 4.4 148 ( 32 565 ( 4.6 888 ( 29 17 ( 1.4 8 ( 2.0 140 ( 12.3 6 ( 3.1 30 ( 4.8 1.9 ( 0.5 0.8 ( 0.4 nd

2370 ( 219 452 ( 45 1780 ( 245 5020 ( 295 1920 ( 190 4310 ( 248 3120 ( 234 1390 ( 79 1270 ( 100 1840 ( 122 1040 ( 144 552 ( 61

sewage sludge coal carb gas oil liquid coal tar creosote shale oil light cycle oil Irish solid coal tar gas oil

English solid coal tar spent oxide heavy cycle oil study material PAH (µg/g)

TABLE 1. PAH Concentrations in the Study Materials

( 2 °C for the set time period. All the water was removed from the bottle at the end of this period by pouring into a glass separating funnel for solvent extraction. Any degradation of the tubing and bio/photodegradation of PAHs was minimized by conducting the experiments at a low temperature. This also helped to minimize volatilization, already anticipated to be negligible due to the limited headspace. The time intervals for which the materials were placed in contact with water were 1.5, 3, 6, 12, 24, 48, 96, and 192 h. In addition to the eight bottles, one of which was sacrificed for each contact time, one blank bottle comprising of 2.5 mL Milli-Q water contained in dialysis tubing and one replicate bottle comprising of the study material in dialysis tubing, both left in contact with water for 96 h, were included. Validation Procedures. Various trials to evaluate sorption associated with the dialysis method were conducted using Milli-Q water spiked with the PAH standard mixture. The same borosilicate bottles and shaker were used as described above. The glass reagent bottles, some containing dialysis tubing and closure, were filled with Milli-Q water, spiked, and placed horizontally on the flat bed shaker rotating 300 times a minute in the temperature-controlled room. Spiking was carried out using 10-1000 ng of PAHs, and the bottles were left for time periods varying between 1.5 and 24 h. 14C tracer studies were used to investigate sorption and retardation effects of the dialysis tubing as low concentrations of PAHs in water could be easily detected; therefore, water inside the tubing could be spiked and movement into the water outside the tubing monitored. A spike solution was prepared 10 min before each experiment by adding 50 µL of 26 µg/mL phenanthrene in methanol to a glass scintillation vial and then spiking 5 µL of 56 990 dpm/µL of 14C-labeled phenanthrene in toluene into this. A total of 8 mL of Milli-Q water was added, and the vial was shaken and allowed to stand for 2 min. A total of 2 mL of this spike solution was added to the inside of the dialysis tubing, which was then sealed and placed in the bottle containing 130 mL of water. Two milliliters was also placed in a scintillation vial for counting and in another bottle just containing 130 mL of water. Three experiments were set off at different times to enable replication and full sampling over a 48-h time period. For each experiment there were two bottles; one with water spiked inside the dialysis tubing and another just containing spiked water. The 0.5-mL samples were taken from each bottle every half hour for the first 3 h, every hour for 6 h, every 2 h for 12 h, and every 4 h for 24 h and placed into scintillation vials. The samples were counted (Canberra Packard TriOCarb 2205CA liquid scintillation analyzer) after 6 mL of Milli-Q water and 9 mL of Ultima Gold XR scintillant (Canberra Packard) had been added to each vial. At the end of the experiment (48 h), the water inside the tubing, the combusted tubing, and the detergent and acetone rinses of the bottles and closure were counted. Direct Contact Experimental Method. A direct contact method, designed so that experimental artifacts were minimized, was used to compare dissolution from the four LNAPL study materials with results obtained using the dialysis method. A total of 100 mL of Milli-Q water was poured into a borosilicate glass reagent bottle (100 mL nominal volume, Schott Duran) to which a 10 cm length of glass tubing (o.d. 5 mm, i.d. 3 mm) was added. A total of 2.5 mL of material was then floated on the surface of the water, and the bottle was sealed with a screw cap top and placed upright on the flat bed shaker rotating at 150 rpm at 8 °C for the set contact time period. The bottles were swirled gently to enhance diffusion while preventing emulsion and droplet formation, which would have made isolation of the aqueous phase difficult. The water was removed by inserting a syringe through the glass tube. As with the dialysis method, a bottle

FIGURE 1. Aqueous phase concentrations of 14C-labeled phenanthrene measured over time. was sacrificed for each of the 1.5, 3, 6, 12, 24, 48, 96, and 192 h contact time intervals, and a replicate and a blank were included for the 96 h contact time. Centrifugation Experimental Method. A centrifugation method was used to compare dissolution from the three solid study materials with dissolution using the dialysis method. In comparing the experimental methods, one 0.5-g portion of solid material was placed directly in a Teflon centrifuge tube (50 mL nominal volume, Nalgene), and the other 0.5-g portion was slurry packed in a 15 cm length of dialysis tubing knotted at both ends and placed in an identical tube. Each tube was filled with 40 mL of Milli-Q and placed on the flat bed shaker rotating at 300 rpm in an 8 °C room. A tube was prepared and sacrificed for each of the 1.5, 3, 6, 12, 24, 48, 96, and 192 h contact time intervals, and a replicate and a blank were included for the 96 h contact time. After the required contact time with water had passed, the tube without the dialysis tubing was centrifuged at 27200g for 30 min in a Beckmann J2-21 centrifuge. A total of 30 mL of water was removed from this into a separating funnel for extraction using a glass syringe as well as from the centrifuge tube containing the dialysis tubing. Glass centrifuge tubes would have been preferential to minimize sorption, but these were not available. However, given that the samples contained high PAH concentrations, any sorptive losses were likely to be compensated for by re-equilibration. Method for Analysis of the Contaminant Materials and Aqueous Samples. The solid materials were mixed with powdered anhydrous sodium sulfate in pre-extracted cellulose thimbles and extracted with 50 mL of dichloromethane in a Tecator Soxtec HT 1043 extraction unit for 2 h boiling and 4 h rinsing (the boiling cycle provides the opportunity for faster extraction times than a conventional Soxhlet unit). The extract was then quantitatively transferred and made up to volume. Water samples were spiked with 500 pg of each recovery standard and extracted by partitioning three times with 20 mL of dichloromethane in a separating funnel. The water extracts were dried by passing through granular anhydrous sodium sulfate, and the extract was concentrated by rotary evaporation. All extracts and the LNAPL materials dissolved in dichloromethane were cleaned up using silica activated at 120 °C for 17 h or longer and cooled in a desiccator for 10 min prior to use. The silica was slurry packed in hexane to a height of

30 mm from the sinter in a 300 mm × 10 mm i.d. glass column fitted with a porosity 1 sinter and Teflon tap. Sample extracts were added onto the columns in less than 1 mL of dichloromethane. Hydrocarbons were eluted with 30 and 15 mL of 60% hexane:40% dichloromethane. The PAH fraction was eluted by adding a further 25 mL of 60% hexane:40% dichloromethane and collected into a 25-mL volumetric flask. The solvent from this fraction was evaporated as required to enable detection. Analysis for PAHs was carried out on a Hewlett-Packard (HP) 5972 mass selective detector (MSD) configured to a 7673 autosampler and coupled to a 5890 gas chromatograph (GC) equipped with an on-column injection system incorporating electronic pressure control and temperature programming. The column used was a HP-5MS (cross-linked 5% phenyl methyl silicone) 30 m × 0.25 mm i.d. with 0.25 µm film thickness. This was connected by means of a glass push fit connector to a 1.2 m length of 0.53 mm i.d. deactivated fused silica, which acted as a retention gap and facilitated on-column injection. The on-column injector temperature was set 3 °C above the oven temperature, and the transfer line temperature was 300 °C. The oven temperature program was 50 °C for 1 min, 15 °C/min to 300 °C, and 300 °C for 11.5 min. Helium was used as a carrier gas, with a pressure program of 6 psi for 1 min, 0.5 psi/min to 14.5 psi, 10 psi/min to 35, and 35 psi for 10 min. The autosampler was set up to inject 1 µL on-column. The MSD was tuned at least once a week using the maximum sensitivity autotune and was operated in selected ion monitoring (SIM) mode with the electron multiplier 300 V above that set in the autotune. The GC/MSD system was calibrated over the ranges 1000-100, 100-10, and 10-1 pg on-column using the deuterated internal standards to compensate for any instrumental variations between and during injections. A total of 50 pg of the standard PAH mix was injected on-column after every six samples to check instrument performance, and blanks indicated no cross contamination.

Results and Discussion PAH Concentrations and Mixtures in the Study Materials. The concentrations of the individual PAHs determined in each material are provided in Table 1 as a mean of triplicate VOL. 33, NO. 12, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2121

FIGURE 2. Phenanthrene dissolution measured using the dialysis tubing method compared to results for the centrifugation (A, spent oxide) and direct contact (B, petroleum gas oil) experimental methods. analyses, with a 95% confidence interval quoted for the PAHs (µg/g) in each material. From the table it can be seen that the creosote, gas oil, and the light cycle oil contain predominantly low molecular weight PAHs, while the other materials contain the complete range. For all materials, phenanthrene was either the most or second-most abundant PAH. Validation of the Dialysis Procedure. Spike Studies. The spiked water studies indicated that losses due to sorption increased with molecular weight of the compound, contact time, and in the presence of dialysis tubing and a closure. Sorption ranged from 10-20% for the lower molecular weight PAHs to 50-60% for the higher molecular weight PAHs. This increased by 10% and 20%, respectively, when dialysis tubing and closure were present. Solvent rinsing recovered the 2122

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 12, 1999

sorbed PAHs, but this was not feasible in the actual experiments as PAHs could be solubilized from the materials themselves. The results of the trial using 14C-labeled phenanthrene are presented in Figure 1. From this, it was concluded that phenanthrene spiked in water inside dialysis tubing takes up to 2 h to diffuse into the water outside the tubing and that the concentration decreases over time due to sorption. It was found that, for the bottles containing spiked water, 98.7% of activity was associated with the water and the remainder was recovered in the rinses. This compared well to a 2% loss to glass vials for [14C]phenanthrene reported in the literature (32). Where water inside the tubing had been spiked, 82.3% of activity was associated with the water outside the tubing, and the remainder was recovered from rinsing the closure

TABLE 2. Comparison of Measured and Predicted (Underlined) Aqueous Phase Concentrations (ng/mL) Associated with the Study Materialsa PAH acenaphthylene

heavy cycle oil co

creosote

shale oil

liquid coal tar

gas oil

co

co

co

nq

6.29 2.69 ( 0.23 fluorene 8.75 2.95 ( 0.12 phenanthrene 23.8 9.64 ( 0.45 anthracene 5.1 1.44 ( 0.10 fluoranthene 1.53 0.31 ( 0.04 pyrene 4.47 1.20 ( 0.07 benz[a]anthracene na

15.5 7.60 ( 0.33 18.9 9.57 ( 0.72 4.56 3.20 ( 0.72 3.58 1.56 ( 0.26 0.10 0.08 ( 0.03 0.07 0.05 ( 0.03 na

1.23 1.57 ( 0.23 2.55 2.60 ( 0.46 3.57 3.98 ( 0.64 1.05 1.02 ( 0.25 0.23 0.12 ( 0.03 0.54 0.34 ( 0.06 na

3.03 3.66 ( 3.76 9.22 6.19 ( 0.94 4.77 4.47 ( 0.66 0.09 1.13 ( 0.20 0.02 0.05 ( 0.01 0.18 0.09 ( 0.01 na

nq

chrysene

na

na

na

acenaphthene

a

na

114 69.9 ( 6.04 87 41.1 ( 4.27 38 13.1 ( 1.33 20 5.06 ( 1.00 9.5 3.69 ( 0.88 7.3 0.80 ( 0.50 14.3 0.94 ( 0.50

light cycle oil co

Irish solid coal tar nq

English solid coal carb gas coal tar oil

301 225 ( 16.4 12.3 nq 10.7 8.20 ( 1.27 26.7 ( 3.88 5.09 204 33.7 3.38 ( 0.31 247 ( 18.4 42.3 ( 2.90 2.27 142 31.3 1.86 ( 0.26 135 ( 7.47 37.6 ( 3.14 0.41 30.3 11.8 0.44 ( 0.88 23.7 ( 2.37 8.63 ( 0.39 0.01 26.6 10.3 0.06 ( 0.03 11.1 ( 1.10 5.79 ( 1.08 0.07 24.6 4.63 0.07 ( 0.05 14.2 ( 1.55 3.16 ( 0.69 na 5.02 na 0.41 ( 0.12 9.67 na 0.36 ( 0.12

120 48.8 ( 12.6 31.5 40.8 ( 27 21.9 12.6 ( 1.31 19.1 12.0 ( 1.37 5.66 4.08 ( 0.65 4.71 1.89 ( 0.51 2.22 1.08 ( 0.33 na na

na, not available. nq, not quantified, concentrations outside calibration range. co, not quantified due to closely eluting sample components.

(12.4%), rinsing the bottle (1.2%), inside the tubing (1.5%), and the actual tubing (2.6%). Losses due to sorption were higher in the systems containing the dialysis closure and tubing, mainly due to sorption onto the closure. However, losses to the bottle were identical for the two setups. Comparison of Experimental Methods. Discussion of the results of the comparison of the dialysis tubing method has been limited as it is presented elsewhere (33), and despite refining experimental procedures, the results for the other methods may be influenced by the problems highlighted in the Introduction. Results are presented in Figure 2 for phenanthrene; panel A, centrifugation-dialysis; panel B, direct contact-dialysis comparison. The results indicate a slower rate of dissolution for the dialysis method but comparable equilibrium concentrations. The mass transfer coefficients for all PAHs and methods are provided in Supporting Information Table 2. The slower rate of dissolution, which was greater than that seen in the 14C studies, may be attributable to the dialysis tubing introducing diffusion constraints on the removal of solutes from the surface of the material into bulk solution, perhaps due to the formation of a stagnant film of water around the material surface. If the mass transfer coefficients for each method are compared separately, the PAHs in the different materials are seen to have similar coefficients, and the majority of the materials give comparable coefficients for each PAH. Mukherji et al. (34) also found that transfer coefficients did not vary between different NAPLs or PAHs. These results therefore suggest that while the method cannot be used to determine absolute mass transfer rates, it may have utility in comparing rates. Although the validation studies indicated that use of dialysis tubing, and in particular a closure, increases sorption of PAHs in the experimental system, this was not seen to influence the results for the study materials as equilibrium concentrations were not significantly different when dialysis tubing was used. This suggests that further release occurred to restore equilibrium. This suggestion is reinforced by comments in the literature that sorption does not affect the final equilibrium (22, 23) and that sorption to equilibration vessels is negligible when the compounds of interest are present in large quantities (12). This aspect of the method may cause problems where low concentrations limit further release. However, this is true of all experimental methods. Comparison of Predicted and Measured Concentrations. Close to equilibrium, aqueous-phase concentrations were

derived by fitting eq 3 to the data, with 95% confidence limits based on the asymptotic standard error. Interpretation was restricted to the lower molecular weight (LMW) PAHs, which showed an inverse exponential decay with time. The higher molecular weight PAHs were excluded as the concentrations were close to detectable limits. Equation 1 was used to predict aqueous-phase concentrations, using experimentally determined average molecular weights (Supporting Information Table 1) and aqueous solubilities from the literature (33). The measured and predicted aqueous phase concentrations were compared and are summarized in Table 2 for all materials excluding the sludge and spent oxide, to which Raoult’s law is not applicable. From Table 2, it can be seen that the measured aqueousphase concentrations are similar to those predicted for the majority of PAHs in the shale oil, gas oils, light cycle oil, and solid coal tars with the exception of lower measured concentrations of the higher molecular weight PAHs for the Irish solid coal tar. The measured concentrations were almost twice as low as those predicted for the heavy cycle oil, creosote, and liquid coal tar. However, other researchers have not considered differences to be significant if within a factor of 2 (4) or even an order of magnitude (2). The measured close to equilibrium concentrations were used to calculate material-water partition coefficients (Kmw) for the PAHs in each material using eq 2, and the data are presented in Supporting Information Table 3. In calculating the partition coefficients, the concentrations (mg/kg) of PAHs in the study materials presented in Table 1 were used, and the aqueous-phase concentrations (mg/L) were as experimentally determined (Table 2). While the normal convention is to determine concentrations in both phases at the end of the experiment (28), the inaccuracy introduced in the chosen approach was negligible given the concentrations involved; indeed the same approach has been used by other researchers (34). Furthermore, when account was taken of the mass of component and material released to water, a method used by Picel et al. (26), no difference to the calculated coefficients was observed. All measured material-water partition coefficients were greater than published octanol-water partition coefficients, Kow (33). This was in accord with the work of Rostad et al. (35), who observed that coal-tar water partition coefficients (Kctw) were greater than Kow values. High correlation between Kctw and Kow has already been demonstrated by Rostad et al. (35) and Picel et al. (26). The measured Kmw values compared VOL. 33, NO. 12, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2123

FIGURE 3. Inverse correlations between measured and predicted PAH partition coefficients and subcooled aqueous solubility. favorably to those reported in the literature. Considering phenanthrene as an example, log Kmw values measured for the liquid and English and Irish solid coal tars were 5.66, 5.31, and 5.13, respectively, which compared well with values of 5.2 (28) and 5.44 (26) reported for coal tar. Material-water partition coefficients were calculated using the predicted aqueous-phase concentrations in Table 2 applied to eq 2 and compared to the measured coefficients. As well as confirming that there was good agreement between measured and predicted partition coefficients (see Supporting Information Table 3), the coefficients were used to examine whether the materials exhibited ideal behavior. Application of Raoult’s law and the assumption of ideal behavior yields the following relationship (13):

log Kmw ) -log Sl - log (AMWt/F)

(4)

where F is the density (g/L) and AMwtm is the average molecular weight of the material. Lines were fitted to the logarithmic plot of predicted partition coefficients against subcooled aqueous solubility (Sl), as shown in Figure 3. The measured partition coefficients have been plotted on the graphs to enable deviations away from ideality to be observed. In interpreting the graphs in Figure 3, agreement within a factor of (0.3 of experimental 2124

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 12, 1999

data and predicted data was used to verify ideality, in keeping the approach commonly used by other researchers (12). The majority of materials were seen to exhibit ideal behavior with some deviations, particularly for the heavier molecular weight PAHs. Both the heavy cycle oil and liquid coal tar (viscous DNAPLs) appeared to exhibit nonideal behavior. This was difficult to attribute to an obvious experimental artifact, particularly as both materials contained high PAH concentrations. Dissolution Results. According to mass transfer theory (18), the rate of release to water slows down with a reduction in the driving force as equilibrium is approached. The release of all PAHs to water from the different study materials was seen to be in accord with mass transfer considerations, exhibiting an inverse exponential decay with time. This was demonstrated by graphical presentation of the LMW PAH aqueous-phase concentrations measured at the different contact time intervals, shown in Figure 4. Curves have been fitted to the data points using eq 3. The approach to equilibrium for PAH dissolution was typical of work reported in the literature (e.g., refs 17 and 31), and it was estimated that a close to equilibrium situation was approached after 50 h. From Figure 4, it can be seen that the more soluble and abundant PAHs equate to the highest aqueous-phase con-

FIGURE 4. PAH dissolution from the study materials using the dialysis tubing method. centrations. Interestingly, the results do show that PAH dissolution varies between materials; for the organic materials, approximately 0.01% weight of phenanthrene, for example, is released at equilibrium. This is approximately 1 order of magnitude less than that released from the spent oxide and 2 orders of magnitude less than that released from the sewage sludge. This can be directly correlated to the low percentage weight of the material extracted with dichloromethane (see Supporting Information Table 1), an indicator of organic matter content. The dissolution experiments therefore indicate that the materials in which PAHs are entrained do influence equilibrium aqueous phase concentrations. Dissolution of PAHs was found to be less favorable for those materials that were chemically similar to PAHs. The results are consistent with

observations made in the literature that equilibrium is a function of liquid-liquid partitioning (36) and that dissolution of a solute in a solvent of a “similar” nature is greater than in a “dissimilar” solvent (37). Given that the release of PAHs to water varies according to the material in which they are entrained, the importance of the contaminant material should not be overlooked when assessing risk and formulating remediation strategies. Indeed, in many cases, the focus of remediation must be on residual material phases if the continual release of PAHs into the water environment is to be prevented. Comments on Application and Limitations of the Dialysis Method. From the method validation and comparison work, it was apparent that the dialysis-based experimental method had a number of advantages and VOL. 33, NO. 12, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

2125

disadvantages. The main benefits of the experimental method were that it offered the only feasible way to measure equilibrium aqueous-phase concentrations from the physically and chemically disparate study materials and that it was an easy means of ensuring separation of the aqueous phase without creating experimental artifacts arising from the presence of emulsions, droplets, or particles. This is an important consideration when dealing with highly contaminated materials. The method may also be used for comparing rates of dissolution. However, there are also some disadvantages. The introduction of additional sorptive surfaces may result in losses for materials in which sorption cannot be countered by reequilibration. The integrity of the tubing was observed to deteriorate with time, limiting studies to less than 2 weeks duration. Furthermore, the dialysis membrane retards the diffusion of solutes into bulk solution, and so the method cannot be used for absolute determination of rates.

Acknowledgments We thank Lancaster University for the studentship to P.J.W., BP Oil for the average molecular weight determinations, and the various companies who donated study materials. We also acknowledge financial support from the Ministry of Agriculture, Food and Fisheries for funding at Lancaster on sewage sludge.

Supporting Information Available Three tables decribing the study materials details; the mass transfer coefficients measured using the dialysis tubing, centrifugation, and direct contact methods; and the logarithms of measured and predicted material-water partition coefficients. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Toxicological profile for polycyclic aromatic hydrocarbons (update); ATSDR, Department of Health and Human Services: 1995. (2) Lane, W. F.; Loehr, R. C. Water Environ. Res. 1995, 67, 169. (3) Zemanek, M. G.; Pollard, S. J. T.; Kenefick, S. L.; Hrudey, S. E. Environ. Pollut. 1997, 98, 239. (4) Estimating release of polycyclic aromatic hydrocarbons from coal tar at maufactured-gas plant site; Electric Power Research Institute: Palo Alto, CA, 1992; TR-101060. (5) Luthy, R. G.; Dzombak, D. A.; Peters, C. A.; Roy, S. B.; Ramaswami, A.; Nakles, D. V.; Nott, B. R. Environ. Sci. Technol. 1994, 28, 266A. (6) Brusseau, M. L.; Jessup, R. E.; Rao, P. S. C. Environ. Sci. Technol. 1991, 25, 134. (7) Roy W. R. In Migration and fate of pollutants in soils and subsoils; Petruzzelli, D., Helfferich, F. G., Eds; Springer-Verlag: Berlin, Germany, 1993; pp 169-188. (8) Karickhoff, S. W.; Brown, D. S. J. Environ. Qual. 1978, 7, 246.

2126

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 12, 1999

(9) Voice, T. C.; Rice, C. P.; Weber, W. J. Environ. Sci. Technol, 1983, 17, 513. (10) Gschwend, P. M.; Wu, S. Environ. Sci. Technol, 1985, 19, 90. (11) Etzweiler, F.; Senn, E.; Schmid, H. W. H. Anal. Chem. 1995, 67, 655. (12) Lee, L. S.; Rao, P. S. C.; Okuda, I. Environ. Sci. Technol. 1992, 26, 2110. (13) Lee, L. S.; Hagwall, M.; Delfino, J. J.; Rao, P. S. C. Environ. Sci. Technol. 1992, 26, 2104. (14) Groves, F. R., Jr. Environ. Sci. Technol. 1988, 22, 282. (15) Yeom, I. T.; Ghosh, M. M.; Cox, C. D.; Robinson, K. G. Environ. Sci. Technol 1995, 29, 3015. (16) Grathwohl, P.; Gewald, T.; Pyka, W.; Schu ¨ th, C. In Contaminated Soil ‘93; Arendt, F., Annokkee, G. J., Bosman, R., van den Brink, W. J., Eds.; Kluwer Academic Publishers: Dordrecht, The Netherlands, 1993; pp 175-184. (17) Priddle, M. W.; MacQuarrie, K. T. B. J. Contam. Hydrol. 1994, 15, 27. (18) Billington, J. W.; Huang, G. L.; Szeto, F.; Shiu, W. Y.; Mackay, D. Environ. Toxicol. Chem. 1988, 7, 117. (19) Allen-King, R. M.; Groenevelt, H.; Mackay, D. M. Environ. Sci. Technol. 1995, 29, 148. (20) Harkey, G. A.; Landrum, P. F.; Klaine, S. J. Chemosphere 1994, 28, 583. (21) Kukkonen, J.; Pellinen, J. Sci. Total Environ. 1994, 152, 19. (22) McCarthy, J. F.; Jimenez, B. D. Environ. Sci. Technol. 1985, 19, 1072. (23) Carter, C. W.; Suffet, I. H. Environ. Sci. Technol. 1982, 16, 735. (24) Saltonstall C. W. Am. Biotech. Lab. 1992, August. (25) May, W. E.; Wasik, S. P.; Freeman, D. H. Anal. Chem. 1978, 50, 997. (26) Picel, K. C.; Stamoudis, V. C.; Simmons, M. S. Water Res. 1988, 22, 1189. (27) Cline, P. V.; Delfino, J. J.; Rao, P. S. C. Environ. Sci. Technol. 1991, 25, 914. (28) Lane, W. F.; Loehr, R. C. Environ. Sci. Technol 1992, 26, 983. (29) Luthy, R. G.; Ramaswami, A.; Ghoshal, S.; Merkel, W. Environ. Sci. Technol. 1993, 27, 2914. (30) Efroymson, R. A.; Alexander, M. Environ. Sci. Technol. 1994, 28, 1172. (31) Grimberg, S. J.; Aitken, M. D.; Stringfellow, W. T. Water Sci. Technol. 1994, 30, 23. (32) Carmichael, L. M.; Harkness, M. R.; Bracco, A. A.; Balcarcel, R. R. Environ. Sci. Technol 1997, 28, 253. (33) Woolgar, P. J. The influence of contaminant source materials on the aqueous dissolution of polycyclic aromatic hydrocarbons. Ph.D. Thesis, Lancaster University, U.K., 1998. (34) Mukherji, S.; Peters, C. A.; Weber, W. J., Jr. Environ. Sci. Technol. 1997, 31, 416. (35) Rostad, C. E.; Pereira, W. E.; Hult, M. F. Chemosphere 1985, 14, 1023. (36) Brusseau, M. L. Water Resour. Res. 1992, 28, 33. (37) Mingelgrin, U.; Gerstl, Z. J. Environ. Qual. 1983, 12, 1.

Received for review June 22, 1998. Revised manuscript received March 29, 1999. Accepted March 31, 1999. ES980638Z