Environ. Sci. Technol. 1996, 30, 1866-1871
γ-Ray Destruction of PCBs in Isooctane and Transformer Oil ROD E. ARBON* AND BRUCE J. MINCHER Idaho National Engineering Laboratory, P.O. Box 1625-4107, Idaho Falls, Idaho 83415
WALTER B. KNIGHTON Montana State University, Bozeman, Montana 59717-0340
The radiolytic degradation of PCBs in air-equilibrated isooctane and transformer oil was investigated. Significant degradation was observed in both solvents studied at moderate absorbed dose. Scavenging studies implicate the electron as the major reactive species responsible for decomposition. This is consistent with what has been observed in neutral 2-propanol; however, surprising differences between the radiolytic behavior of PCBs in polar 2-propanol and nonpolar isooctane are observed. PCB degradation was found to be enhanced in isooctane relative to neutral 2-propanol. Chlorine number and substitution pattern have been found to influence degradation efficiency but to a lesser extent in isooctane than in 2-propanol. Deviations in PCB degradation behavior between isooctane and 2-propanol are interpreted in terms of differences in electron behavior between the two solvents. The products of PCB decomposition appear to be less chlorinated PCBs and PCB-solvent adducts.
Introduction The persistence of polychlorinated biphenyls (PCBs) in the environment, coupled with their apparent toxicity (1, 2), led to the passage of the Toxic Substances Control Act (TSCA) in 1976. This act regulates the use and disposal of PCB-containing materials. It is estimated that somewhere between 1.1 × 109 and 2.3 × 109 L of PCB-contaminated transformer oil fall under the disposal requirements of TSCA (3). The current TSCA-approved methods for destroying PCBs are high-temperature incineration or high-efficiency boilers (1). These methods, however, destroy both the solute and the solvent. Removal of the solute is often desirable. As pointed out by Exner (3), if the PCB fraction were destroyed in PCB-contaminated transformer oil, the resulting PCB-free transformer oil could theoretically be placed back into service. The radiolytic dechlorination of PCBs has been investigated thoroughly in alkaline 2-propanol (4-6) and neutral 2-propanol (7, 8). Both the mechanism (4-7) and the * Author to whom correspondence should be addressed; telephone: (208) 526-2806; fax: (208) 526-2304.
1866
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 6, 1996
factors that influence congener degradation (8) have been throughly characterized in polar 2-propanol. Although there is a demonstrated need, there have been very few radiolytic studies investigating the dechlorination of PCBs in nonpolar solvents. We are aware of only one study conducted in transformer oil (9). Webber (9) examined the radiolysis of air-equilibrated Aroclor 1260 in transformer oil and concluded that, even at an absorbed dose of 1540 kGy, there was no evidence of degradation having taken place. However, radiolytic degradation of PCBs has been observed to occur in cyclohexane and petroleum ether (10, 11), but no effort was made to determine the reactive agent responsible for the degradation or the factors that influence individual PCB congener degradation susceptibility. Given the fundamental differences between polar and nonpolar solvents, the radiolytic chemistry of PCBs in these solvents may be quite different. To our knowledge no study has ever compared PCB degradation in polar and nonpolar solvents. In addition, for a treatment process to be viable, one must know both the mechanism and the final degradation products. To investigate PCB radiolytic degradation in a nonpolar solvent, studies were performed on a variety of individual PCB congeners in isooctane. Isooctane was chosen because it is more amenable to mass spectrometric analysis than transformer oil, which requires extensive cleanup to remove high molecular weight hydrocarbons from the PCB elution window. This will greatly facilitate the analytical work necessary to characterize the radiolytic chemistry of PCBs in a nonpolar solvent. Scavenging experiments will demonstrate that the electron is the primary reactive agent responsible for the observed PCB degradation in isooctane. This is in agreement with what has been observed in the radiolytic degradation of PCBs in neutral 2-propanol (7, 8). However, surprising differences between the radiolytic degradation behavior of PCBs in nonpolar isooctane and the more familiar polar 2-propanol were discovered and are illustrated. Mass balance information is provided in isooctane. Finally, radiolytic degradation of Aroclor 1260 in transformer oil is demonstrated to be possible.
Experimental Section Irradiations. Samples were γ-ray irradiated using spent nuclear fuel at the Advanced Test Reactor located at the Idaho National Engineering Laboratory. Spent nuclear fuel is an excellent source of γ-rays with an average kinetic energy of 700 keV. The dose rate is dependent upon the age of the fuel with an achievable maximum of up to 25 kGy h-1. Few neutrons are present, so there is no neutron activation. PCB samples were contained in 1.5-mL glass septum vials that were then placed in stainless steel vessels along with dosimetry and lowered into a dry tube in the fuel storage canal for a predetermined exposure time. The details involved in calibrating the dosimetry and accessing the spent fuel have been described elsewhere (7, 8). Analytical Methods and Reagents. Identification and quantification of the radiolytically induced degradation products of the irradiated individual PCB congeners in isooctane were performed using a Hewlett Packard 5995 GC/MS operated in electron impact mode as previously described (8). The analytical column was a DB5-625, 30-m
S0013-936X(95)00467-6 CCC: $12.00
1996 American Chemical Society
TABLE 1
Ballschmiter Numbers, Structure, and Dose Constants (kGy-1) in Isooctane and 2-Propanol Bz no.
structure
isooctane (kGy-1)
2-propanol (kGy-1)
Bz no.
Monochlorobiphenyl 0.034 ( 0.004 0.036 ( 0.008
0.006 ( 0.001 0.008 ( 0.002
101 118 126a
2,2′,4,5,5′ 2,3′,4,4′,5 3,3′,4,4′,5
Pentachlorobiphenyl 0.060 ( 0.016 0.066 ( 0.003 0.073 ( 0.003
0.014 ( 0.002 0.015 ( 0.002 0.028 ( 0.003
structure
isooctane (kGy-1)
2-propanol (kGy-1)
1 3
2 4
4 11
2,2′ 3,3′
Dichlorobiphenyl 0.042 ( 0.001 0.007 ( 0.001
0.009 ( 0.001 0.005 ( 0.002
155 167a 169a
2,2′,4,4′,6,6′ 2,3′,4,4′,5,5′ 3,3′,4,4′,5,5′
Hexachlorobiphenyl 0.046 ( 0.018 0.043 ( 0.004 0.085 ( 0.003
0.014 ( 0.002 0.025 ( 0.002 0.036 ( 0.004
33 35
2,3,4 3,3′,4
Trichlorobiphenyl 0.056 ( 0.003 0.066 ( 0.015
0.012 ( 0.003 0.017 ( 0.001
183
Heptachlorobiphenyl 2,2′,3,4,4′,5′,6 0.11 ( 0.010
0.019 ( 0.002
47 54 77a
2,2′,4,4′ 2,2′,6,6′ 3,3′,4,4′
Tetrachlorobiphenyl 0.065 ( 0.004 0.041 ( 0.004 0.068 ( 0.005
0.009 ( 0.002 0.008 ( 0.001 0.019 ( 0.002
194 200
Octachlorobiphenyl 2,2′,3,3′,4,4′,5,5′ 0.12 ( 0.014 2,2′,3,3′,4,5′,6,6′ 0.10 ( 0.014
0.025 ( 0.002 0.024 ( 0.002
a
Planar or almost planar. The uncertainties shown are at the 90% confidence level and were calculated according to Noggle (28).
column with helium carrier gas at a flow of 2 mL min-1. Free chloride ion analysis was performed by vigorously shaking a known aliquot of sample with water to extract the chloride. The resulting aqueous solution was then analyzed for chloride ions by ion chromatography as previously described (8). PCB standards were purchased from Ultra Scientific (North Kingston, RI) and were greater than 99% pure. Some experiments included a 14C-labeled tetrachlorobiphenyl, labeled in all 12 positions, which was purchased from Sigma Chemicals (St. Louis, MO). As discussed in a previous reference (8) carbon-14 measurements were made using a Model 2250CA Packard Tricarb liquid scintillation counter and Ecolume scintillation cocktail. Some samples were derivitized with diazomethane. Diazomethane was generated by treatment of diazald with potassium hydroxide. Aroclor 1260 samples in transformer and hydraulic oil were analyzed using a Hewlett Packard 5890 Series II gas chromatograph fitted with an electron capture detector. A DB-1701, 30 m × 0.32 mm column was used with a helium carrier flow of 7.5 mL min-1 and a 95% argon/5% methane makeup gas at 45 mL min-1. To prevent excessive column contamination, a 1:10 dilution in isooctane was performed on the oil samples prior to analysis. Injections were made using a 7673 Hewlett Packard automatic sampler onto a column at an initial temperature of 60 °C, ramped at a rate of 10.0 °C min-1 to 180 °C, held 6.00 min, and then ramped at 5 °C min-1 to a final temperature of 250 °C. When discussing specific PCB congeners used in this study, we adopted the numbering system suggested by Ballschmiter (12). Ballschmiter, Bz, numbers and the corresponding congener structure are given in Table 1. Solvents were reagent grade and were used without further purification. Determination of Degradation Efficiency. The figure of merit typically used in radiolysis studies is the G value. G values relate the radiation chemical yield to the absorbed dose in molecules/100 eV. G values are reported in this study as Ginitial values. Ginitial values were determined by evaluating the amount of starting material degraded between the unirradiated point and the first irradiated data point. Since the radiolytically induced change in PCB concentration was observed to decrease exponentially with absorbed dose, the resulting Ginitial values are dependent on both the starting PCB concentration and the amount of
dose received. While these G values are useful, some caution must be exercised due to their concentration and dose dependence. To allow for the direct comparison of individual PCB congener degradation susceptibility in isooctane to a previous study performed in 2-propanol (8), dose constants are also reported. In 2-propanol (8) when the logarithm of PCB concentration is plotted against absorbed dose, a linear relationship was obtained indicating a first-order or pseudo-first-order kinetic process. The slope of the line when the natural logarithm of PCB concentration versus absorbed dose is plotted is the dose constant. The dose constant is analogous to a first-order rate constant but with units of reciprocal dose. When individual congener degradation was measured in isooctane, apparent first-order kinetics was also observed. In contrast to 2-propanol, the dose constant in isooctane was found to vary inversely with initial PCB concentration (13). Thus, to allow for meaningful congener degradation comparison, the beginning PCB concentrations were all approximately 250 mg L-1.
Results and Discussion Figure 1a shows a plot of Bz 200 concentration in airequilibrated isooctane as a function of absorbed dose. The octachlorobiphenyl concentration steadily decreases with increasing dose. As the degradation of octachlorobiphenyl proceeds, the simultaneous growth and subsequent destruction of the less chlorinated hepta-, hexa-, penta-, tetra-, tri-, and dichlorobiphenyl congeners are observed and shown in Figure 1b. The dechlorination process produces a variety of different isomers allowed by the starting material but not necessarily in equal proportions. At the maximum absorbed dose obtained in Figure 1b, negligible amounts of monochlorobiphenyl congeners and biphenyl were found. At the highest absorbed dose, only about 43% of the original PCB material remains in the form of less chlorinated PCBs. A more complete discussion of the degradation products observed is presented in the Mass Balance section. Degradation Efficiency. To quantitate the destruction efficiency, dose constants and Ginitial values were evaluated for individual congeners in isooctane. The larger the dose constant or Ginitial value, the greater the susceptibility toward radiolytic degradation. A summary of the degradation
VOL. 30, NO. 6, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1867
TABLE 2
Relative Destruction for a 250 mg L-1 Solution of Bz 1 and Bz 200 in Isooctane with and without the Addition of Radical Scavengers
FIGURE 1. (a) Observed destruction from the irradiation of Bz 200. Note the half-life dependency of Bz 200 in relation to absorbed dose. (b) Degradation products produced from Bz 200. Curves a-e correspond to hepta-, hexa-, penta-, tetra-, tri-, and dichlorobiphenyl congeners, respectfully.
efficiencies for 18 PCB congeners obtained in isooctane as well as those previously measured in 2-propanol (8) are given in Table 1. The Ginitial values for all the congeners studied lie in the range of 0.1-2.4, indicating that the degradation reaction is only moderately efficient. That is, given the G values obtained, a chain reaction does not appear to occur. Interestingly, all of the congeners studied were more efficiently destroyed in isooctane, 1.4-9 times, than in 2-propanol. An explanation of this intriguing observation is given in the comparison of PCB Degradation in Isooctane vs 2-Propanol section. Scavenging Experiments. In an attempt to understand the radiolytic dechlorination mechanism in isooctane, scavenging experiments were conducted to identify the reactive species responsible. The yield and identity of the radicals produced during the radiolysis of hydrocarbons have been studied extensively (14). The major species produced are hydrogen atoms, various alkane radicals, positive ions, and electrons (14). It is conceivable that any of the reactive agents mentioned could be responsible for PCB degradation. Table 2 shows the effect various scavenging agents have on the dose constants obtained for Bz 1 and Bz 200 in
1868
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 6, 1996
PCB Bz no.
relative destruction rate
condition
200 200 200 200 200 1 1 1 1 1 1
1 1 0.18 0.052 0.72 1 1 0.19 0.081 1 1.3
no additive 400 mg L-1 benzene 400 mg L-1 nitrobenzene SF6 sparged cyclopropane sparged no additive 400 mg L-1 benzene 400 mg L-1 nitrobenzene SF6 sparged cyclopropane sparged N2 sparged
isooctane. The data have been normalized to the scavengerfree congener. Normalizing the dose constants allows for easy comparison of scavenging agent effects. The addition of nitrobenzene significantly reduced the degradation of Bz 200 and Bz 1, while benzene had no effect. Nitrobenzene has a diffusion-controlled rate of reaction with aqueous electrons while benzene has a reaction rate with aqueous electrons that is over 3 orders of magnitude smaller (15). Both benzene and nitrobenzene have similar reaction rates with hydrogen atoms (15). Given this, a reasonable conclusion is that hydrogen atoms do not contribute to PCB degradation in isooctane. The above experiment is not definitive toward solvated electrons because nitrobenzene is also known to scavenge alkane radicals (14). However, sulfur hexafluoride, known to have a high rate of reaction with electrons, is believed to have a reaction rate significantly less for alkane radicals. This conclusion is based on work performed by Johnson and Salmon (16) where the rate of nitrobenzene reacting with a methanol radical was found to be 5 orders of magnitude greater than was the rate for sulfur hexafluoride. Assuming a similar trend, the reaction of alkane radicals with sulfur hexafluoride should be negligible. When samples were sparged with sulfur hexafluoride, the dose constants were greatly reduced, implicating the solvated electron as a major reactive species. The effect dissolved oxygen has on degradation also implicates solvated electrons. Oxygen is a known free radical scavenger capable of efficiently scavenging solvated electrons (14). In the absence of oxygen, degradation should be enhanced. As shown in Table 2, sparging a solution of Bz 1 with nitrogen increased degradation while sparging with oxygen reduced Bz 1 degradation. Radiolysis literature identifies cyclopropane as an efficient positive ion scavenger (17). A modest reduction in degradation was observed for Bz 200 when sparged with cyclopropane while no reduction in degradation for Bz 1 was observed. This interesting result required further investigation. PCB congeners 169, 155, and 77 were evaluated with and without cyclopropane. No observable reduction in degradation due to the presence of cyclopropane occurred. Positive-ion PCB reactions, if they occur, do not appear to be significant. Another potential degradation mechanism is energy transfer from excited isooctane to the PCBs followed by dissociation of the PCB molecule. Holroyd (18) suggested
that electronic excitation energy transfer from nonpolar cyclohexane to N2O readily occurs and results in large nitrogen yields when N2O is irradiated in cyclohexane. As reported by Topchiev (19), electronic excitation energy transfer, if it is occurring, can be experimentally determined by the addition of aromatic inhibitors, such as benzene, which reduce the yield of radiolysis products. As discussed, benzene had no effect on PCB degradation. Therefore, this process does not appear to occur to a significant extent in our system. Consequently, a reasonable conclusion is that dissociative electron capture is responsible for the PCB degradation observed. This conclusion is supported by work performed by Warman et al. (17). In their study, chlorinated and brominated hydrocarbons were found to efficiently capture electrons in nonpolar cyclohexane. In fact, dissociative electron capture was found to be the major pathway for dechlorination. Interestingly, at the concentration of nitrobenzene and sulfur hexafluoride employed, degradation was not completely eliminated. Apparently, the PCBs are capable of competing with these scavengers for the available electrons. Comparison of PCB Degradation in Isooctane vs 2-Propanol. Electrons are the primary reactive species responsible for PCB degradation in neutral 2-propanol (7, 8) and, as just demonstrated, in isooctane. In nonpolar solvents such as isooctane, the majority of electrons formed are expected to undergo geminate recombination with their parent ion; thus, the free electron yield available for reaction is quite small, approximately 0.1 electrons 100 eV-1 (20). This value is over an order of magnitude less than the free electron yield in 2-propanol of 1.4 (14). How can the larger dose constants obtained in isooctane than 2-propanol be reconciled if the free electron yield is an order of magnitude greater in 2-propanol than in isooctane? Free ion yield does not necessarily reflect the total electron yield available for reaction with the PCBs. Because 2-propanol has a measurable electron capture rate, the amount of electrons available to react within it is reduced. The electron attachment rate constant for 2-propanol is 1.5 × 106 M-1 s-1 (21) which taken in conjunction with its high concentration of 13.1 M leads to a pseudo-first-order rate constant for electron attachment to the solvent of 2 × 107 s-1. By comparison, the pseudo-first-order rate constant for a 200 mg L-1 solution of Bz 200 is estimated to be 8 × 106 s-1, based on a PCB molar concentration of 4 × 10-4 M and assuming a very fast electron attachment rate of 2 × 1010 M-1 s-1. This shows that, at best, about 40% of the electrons will be scavenged by the PCB molecules. This figure will be lower still if the PCB concentration or the electron attachment rate constants are smaller. In isooctane, the solvent and the PCBs do not directly compete for electrons because the solvent does not capture electrons. Indeed, PCB molecules are able to scavenge electrons that normally would undergo geminate recombination in isooctane. This is confirmed by evaluating Ginitial values obtained for Bz 155, which varied from 0.37 to 2.4 molecules 100 eV-1 depending upon beginning Bz 155 concentration. These Ginitial values significantly exceed the free ion yield of 0.1. In fact, a G value of 2.4 is approaching the theoretical yield of both free electrons and geminate electrons (17). Thus, the conclusion that the recombination reaction is less efficient than PCB electron capture appears to be justified. The net effect is that more electrons are available for reaction with the PCB molecules in isooctane
FIGURE 2. Plot of observed dose constant in isooctane as a function of chlorine content on the biphenyl ring. Specific congeners can be identified by their Bz numbers.
than in 2-propanol, which is reflected by the higher dose constants observed. The practical significance of these high dose constants is that less dose is needed to achieve significant levels of destruction in isooctane than in neutral 2-propanol! Congener Substitution and Degradation. Figure 2 shows a plot of dose constant as a function of chlorine number in isooctane. In general, the dose constant increases with increasing chlorine content. Careful review of Table 1 shows that chlorine position within a homolog influences the dose constants obtained in isooctane but to a lesser extent than 2-propanol. This is demonstrated by observing the variability that exists within a homolog in Figure 2 and Table 1. If the high and low dose constant obtained for each homolog is evaluated, greater variability exists in 2-propanol than isooctane. Table 1 also shows that the overall difference in dose constants is smaller in isooctane relative to that in 2-propanol. The notable exception is Bz 11 where the dose constant is essentially identical in both isooctane and 2-propanol. Congener degradation appears to be less dependent on chlorine number and substitution in isooctane than in 2-propanol. Given that the dechlorination mechanism proposed for both solvents is the same, greater similarity between the two data sets might be anticipated. However, the available reaction rate data for electron scavenging solutes in nonpolar solvents indicate a complex chemistry for the electron (22-24). In contrast to polar solvents, reactions of electrons in nonpolar liquids can bear a similarity to the reactions of electrons in the gas phase, where the rates of reaction may depend upon the kinetic energy of the electron (22). In nonpolar liquids, the kinetic energy of the electron is dependent on the choice of the solvent. In a study performed by Allen et al. (22), they observed that the measured electron capture rate constant for electron scavengers such as sulfur hexafluoride, carbon tetrachloride, ethyl bromide, and trichloroethylene were dependent on the solvent system chosen. Holroyd et al. (22) also determined that low electron affinity solutes exhibit a greater dependence upon the energy of the electron than compounds with high electron affinity. This work could help in interpreting the results obtained in isooctane. Considerable differences in the lowest unoccupied molecular orbital (LUMO) energy (25) and thus the electron affinity (26) exist between PCB congeners within a homolog series, particularly between the planar and nonplanar
VOL. 30, NO. 6, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1869
congeners. Even greater differences in LUMO energy are found between homologs (25). Conceivably, lower electron affinity PCBs such as the mono- and dichlorobiphenyls may exhibit maxima in their electron attachment cross section at a different electron energy than the high electron affinity PCBs. Thus, if the electron energy in isooctane occurs at or near the maxima for the electron capture rate constant for low electron affinity PCBs (such as mono- and dichlorobiphenyl), the dose constant obtained would be increased relative to high electron affinity PCBs. As observed, the result would be a relative compression of the dose constants. To determine definitively, experiments designed to probe the dependence of reaction rate on electron energetics should be performed. Mass Balance. When a mass balance analysis is performed on the data shown in Figure 1b at the largest absorbed dose, about 43% of the original carbon and chlorine are recovered as residual Bz 200 and less chlorinated PCB congeners. This leaves a significant fraction of the beginning mass missing. In an attempt to identify the location of the missing mass, a 14C-labeled Bz 47 was irradiated. The irradiation was run in a series of doses, as done previously for dose constant measurements, to a maximum dose of 89 kGy. Following irradiation, liquid scintillation counting of the sample aliquots revealed that the activity level was identical for all of the samples and also identical to the activity level for the unirradiated solution. Thus, the beginning mass is still in solution but is unidentified. To determine in which fraction the 14Clabeled products reside, aliquots were nitrogen-sparged at room temperature with no reduction in activity. Consequently, the unidentified mass is not in a volatile form. Aliquots were then distilled at 100 and 150 °C. No activity was removed at either temperature. This is significant evidence that the unidentified degradation products are present in either semivolatile or nonvolatile fractions. Additional mass spectrometric work was undertaken to identify the remaining semivolatile degradation products. Careful review of irradiated congener mass spectra yielded the existence of a variety of solvent-altered PCB adducts and relatively high molecular weight hydrocarbons in addition to less chlorinated PCB congeners. These reaction products are produced by various radical-radical recombinations. PCB solvent adducts were observed for congeners containing three to seven chlorines. The largest class of PCB-solvent adducts observed was the addition of an isooctane radical to a PCB radical. Methyl-radical PCB-radical additions were also observed. The isooctanePCB adducts were confirmed by the presence of a molecular isotopic cluster that was 112 m/z larger than the lowest mass in the molecular cluster for a given homolog. For example, a tetrachlorobiphenyl-isooctane adduct contained a molecular isotopic cluster at m/z 402 with two major isotopic fragment ions at m/z 303 and 331 with the base peak being m/z 333. The mass spectra obtained is consistent with the isooctane being attached to the biphenyl ring via the terminal carbon on isooctane. Accurate quantification of the observed adducts is not currently possible because of the lack of standards. The best that can be done is an estimation. If typical response factors are used, it is reasonable to assume that as much as 1015% of the original carbon and chlorine can be identified in the form of these measurable adducts. This brings the measurable total of beginning carbon and chloride up to approximately 60% of the beginning material. Derivati-
1870
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 6, 1996
FIGURE 3. Aroclor 1260 in transformer oil as a function of absorbed dose. There is a steady decrease in Aroclor 1260 concentration with increasing dose. Aroclor 1260 concentration has been reduced from 5000 to 1800 mg L-1.
zation work was also undertaken using diazomethane in an attempt to identify any polar chlorinated fragments that may have been formed. No compounds were identified. Detection and quantification of the remaining material is undoubtedly hindered by the fact that the products are many steps down a chain of reactions in very small quantities. For example, starting with Bz 200, five unique heptachlorobiphenyl congeners and five heptachlorobiphenyl-isooctane solvent adducts could theoretically be produced in just the first dechlorination step. The number of unique hexachlorobiphenyl and hexachlorbiphenylsolvent radical congeners that could conceivably be produced from the various heptachlorobiphenyl congeners is 50! This number does not reflect the methyl-radical combinations or other potential solvent-radical combinations but only the isooctane-PCB adduct and less chlorinated PCBs. This trend greatly complicates obtaining an accurate mass balance. Based on the radiolabeled work and the degradation products identified, the location of the remaining material is most likely in a wide variety of
less chlorinated homologs and solvent-altered PCB homologs all at concentrations too low to measure. Transformer Oil. Given the results obtained in isooctane, radiolytic degradation of PCBs in transformer oil was believed to be possible with much lower doses than predicted by Webber (9). Figure 3 demonstrates this to be the case. Figure 3 consists of a series of γ-irradiated Aroclor 1260 spiked transformer oil chromatograms obtained with an electron capture detector. The dose each sample received is shown as an insert on the chromatogram. A significant change in the Aroclor 1260 fingerprint occurs with increasing dose. The original peaks have been reduced, with the simultaneous appearance of smaller peaks at shorter retention times. The Aroclor 1260 concentration has been reduced from approximately 5000 to 1800 mg L-1 at an absorbed dose of 229 kGy. The sample was not sparged or modified in any way. To confirm the reactive agent in transformer oil, individual congeners were sparged with sulfur hexafluoride. Reductions in degradation similar to that observed in isooctane occurred as a result. Thus, a reasonable conclusion is that electrons are responsible for the radiolytically induced degradation of PCBs in transformer oil. Given these results, the results obtained by Webber are currently not understood.
Acknowledgments Work performed under contract to the U.S. Department of Energy, Field Office, Idaho, under Contract DE-AC0776ID01570.
Literature Cited (1) Erickson, M. D. In Analytical Chemistry of PCBs; Butterworth Publishers: Boston, 1986; Chapter 1. (2) De Voogt, P.; Wells, D. E.; Reutergardh,L.; Brinkman, U. T. Int. J. Environ. Anal. Chem. 1990, 40, 1-46. (3) Exner, J. H. In Detoxification of Hazardous Waste; AnnArbor Science: Ann Arbor, 1992; Chapter 1. (4) Singh, A.; Kremers, W.; Smalley, P.; Bennett, G. S. Radiat. Phys. Chem. 1985, 25, 11-19. (5) Sawai, T.; Shinozaki, Y. Chem. Lett. 1972, 865-869.
(6) Sawai, T.; Shimokawa, T.; Shinozaki, Y. Bull. Chem. Soc. Jpn. 1974, 47, 1889-1893. (7) Mincher, B. J.; Arbon, R. E.; Knighton, W. B.; Meikrantz, D. H. Appl. Radiat. Isot. 1994, 45 (8), 879-887. (8) Arbon, R. E.; Mincher, B. J.; Knighton, W. B. Environ. Sci. Technol. 1994, 28, 2191-2196. (9) Webber, I. Canadian Electrical Association, Phase 11; CEA: Montreal, Quebec, 1983; pp 70-77. (10) Lepine, F.; Masse, R. Environ. Health Perspect. 1990, 89, 183187. (11) Lepine, F.; Milot, S.; Gagne, N. J. Agric. Food Chem. 1990, 38, 1873-1876. (12) Ballschmiter, K.; Zell, M. Fresenius’ Z. Anal. Chem. 1980, 302, 20-21. (13) Arbon, R. E. Ph.D. Dissertation, Montana State University, 1985. (14) Spinks, J. W. T.; Woods, R. J. In An Introduction to Radiation Chemistry, 3rd ed.; Wiley-Interscience: New York, 1990; Chapter 9. (15) Buxton, G. B.; Greenstock, C. L.; Helman, W. P.; Ross, A. B.; J. Phys. Chem. Ref. Data 1988, 17, 513. (16) Johnson, D. W.; Salmon, G. A. J. Chem. Soc., Faraday Trans. 1 1979, 75, 446. (17) Warman, J. M.; Asmus, K. D.; Schuler, R. H. J. Phys. Chem. 1969, 73, 4. (18) Holroyd, R. A. J. Phys. Chem. 1968, 72, 759. (19) Topchiev, A. V. Radiolysis of Hydrocarbons; Elsevier Publishing Company: Dordrecht, 1964; p 15. (20) Freeman, G. R.; Fayadh, J. M. J. Chem. Phys. 1965, 43 (1) 86-92. (21) Holroyd, R. A. Fundamental Process in Radiation Chemistry; Ausloos, P., Ed.; Wiley-Interscience: New York, 1968; p 413. (22) Allen, A. O.; Gangwer, T. E.; Holroyd, R. A. J. Phys. Chem. 1975, 79 (1), 25-31. (23) Holroyd; Gangwer, T. E. Radiat. Phys. Chem. 1980, 15, 283-286. (24) Beck, G.; Thomas, J. K. J. Chem. Phys. 1972, 57, 3649. (25) Holroyd, R. A. Presented at the 4th Tihany Symposium, Keszthely, Hungary, 1976. (26) Greaves, J.; Harvey, E.; MacIntyre, W. G. J. Am. Soc. Mass Spec. 1994, 5, 44-52. (27) Kebarle, P.; Chowdury, S. Chem. Rev. 1987, 87, 513-534. (28) Noggle, J. H. In Physical Chemistry, 2nd ed.; Scott, Foresman, and Co.: Glenview, IL, 1989; Appendix 1.
Received for review June 30, 1995. Revised manuscript received January 30, 1996. Accepted February 6, 1996.X ES950467N X
Abstract published in Advance ACS Abstracts, April 1, 1996.
VOL. 30, NO. 6, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1871
