Environ. Sci. Techno/. 1995, 29, 2460-2463
Zero-Valent Iron-Promoted Dechlorination of Polychlorinated
PCB mixtures undergo hydrogenolytic dechlorination and other reactions in the presence of iron powder that lead to the virtually complete loss of chlorinated congeners.
Materials and Methods FEI-WEN CHUANG, RICHARD A. LARSON," A N D M A R G A R E T SCULLY W E S S M A N Institute for Environmental Studies, University of Illinois, 1102 West Peabody Drive, Urbana, Illinois 61801
Introduction Polychlorinated biphenyls (PCBs) were produced in large volumes in the United States beginning in 1929. Their excellent stability and thermal properties made PCBs suitable as a heat-transfer medium in electricaltransformers and capacitors (1). Because of widely publicized biological effects such as induction of skin lesions and tumors in animals, manufacture of PCBs ceased in 1977 under the Toxic Substances Control Act. However, because of their persistence, a large fraction of all the PCBs ever produced are still present in the environment. Remediation of PCBcontaminated soils, groundwaters, and sediments has become a major problem in environmental management. PCBs currently are principally being destroyed by incineration ( I ) , which has become the most widely used technique for their removal (2). Buser and Bosshardt reported that at temperatures higher than 800 "C, PCBs were thermodynamically unstable and the pyrolysis products were C(s), CO, COz, HC1, and Clz (3). Incineration, however, often produces more toxic compounds if it is not carefullycontrolled. Erickson etal., for example, indicated that polychlorinated dibenzofurans (PCDFs) and polychlorinated dibenzodioxins (PCDDs)were both observed in the combustion of PCBs ( 4 ) . About 1%of the PCBs in the system were converted into PCDFs. The optimum conditions for PCDF formation were 675 "C with 8% excess oxygen. Other methods for the destruction of PCBs that have been proposed include wet air oxidation (3,biodegradation (6),sodium metal-promoted dehalogenation (7),reaction with superoxide (81, photolysis in the presence ofhydrogen donors (9),and electrolytic reduction (10). However, none of these techniques has been widely adopted. Reynolds etal. indicated that, in the presence of metallic well casings, several aliphatic halogenated hydrocarbon solvents are unstable (11). Subsequently, several research groups have reported that iron metal suspended in water rapidly dehalogenates a variety of solvents (12-14). The observed reductive dechlorination reactions may occur by one or more of the following three mechanisms: direct electrolytic reduction at the metal surface (currentlythought to be most likely),reduction by hydrogen produced during the corrosion process, or reduction by dissolved (ferrous) iron that is also produced by corrosion (13). Less information is available on the reactions of halogenated aromatic compounds with zero-valent metals, although copper has been reported to dechlorinate PCDD and PCDF derivatives (15). We report that at temperatures greater than 300 "C, * Address correspondence to this author.
2460
1
ENVIRONMENTAL SCIENCE &TECHNOLOGY / VOL. 29, NO. 9.1995
PCB mixtures (Aroclor 1221 and 1254) were provided by Monsanto Company (Sauget,IL). Iron metal powder (FeO) was purchased from Fisher Scientific (Fair Lawn, NJ). Described as a certified, electrolytic powder, it contained 95.63%iron (analyzed as Fez+)accordingto the lot analysis. High-purity methylene chloride (CH2Cl2)was from Burdick & Jackson (Muskegon, MI). Benzyl bromide (98%) was purchased from Aldrich Chemical Company, Inc. (Milwaukee, WI). Deuterium oxide was from Sigma Chemical Company (St. Louis, MO). All reagents were used as received. Experimental Procedures. In a typical experiment, a mixture of PCBs (Aroclor 1221, 0.28 mg) was introduced into 10-mLKimax glass ampules by adding 0.5 mL of a 0.56 mg/mL PCB stock solution in CHZCIZ.The ampule therefore contained approximately 1.5pmol of PCBs and 94 pmol of O2 (assuming 10 mL of 21% 02 at STP). The solvent was allowed to evaporate in a ventilated fume hood overnight, and the ampules were flame-sealed after adding FeO powder (0.5 g, 8.9 mmol). In some experiments, 50 p L of distilled water (D20) or CH& was also added to the sample. The samples were heated in a muffle furnace at 200-600 "C. (We assume that the reaction occurred under anoxic conditions since the 1OO:l molar excess of FeO probably consumed the small amount of oxygen initially present inside the ampule.) The heating time of 400 "C samples varied from 10 min to 1 h; the others were heated for 1 h. After opening the ampules, the residue was Soxhletextracted with methylene chloride (approximately 80 mL) for 8 h. Three milliliters of an external standard (concentration = 1.46 x g of benzyl bromide/mL of CH2C12) was added after extraction. The solution was concentrated by evaporation under reduced pressure, brought to 3.0 mL, and analyzed by gas chromatography (GC: Varian 3300 Model with FID) and gas chromatographylmass spectrometry (GUMS: Finnigan Model 800 ion trap with a HewlettPackard Model 5890A gas chromatograph). Both instruments were fitted with DB-5 capillary columns, 30 m x 0.32 mm i.d. U.&W. Scientific, Placerville, CAI, and their separations of the PCB mixtures were directly comparable. The column conditions for GC and GUMS were as follows: initial hold at 40 "C for 5 min, increase from 40 to 280 "C at the rate of 5 "C/min, and hold at 280 "C for 10 min. Areas of the individual GC-FID peaks were determined electronically using a Spectra-Physics integrator, and those for isomers with the same molecular weight (as determined by GC/MS) were summed then each sum was normalized to the peak area of the external standard. The extent of conversion was calculated by assuming that all PCB congeners had comparable FID ionization efficiencies,Le., that the relative peak area would remain constant after conversion of a highly chlorinated PCB congener to a less highly chlorinated one. [Thisassumption appears justified since it has been reported that PCB congeners with avariety of substitution patterns have relative molar response factors that vary from 0.94 to 1.09 (16).]
0013-936X/95/0929-2460$09.00/0
0 1995 American Chemical Society
1.5
1.2
I
rn c10 mrm c11 c12
I
M
I
C13 -c- c10 ec11 -t- c12 *C13
Y (d
d Q)
>
o r (
.! 0.5 2 c1
0
0.0
10
20
SO
40
60
60
70
Time (Min)
RT
200
300
400
600
600
Temperature (C) FIGURE 1. Distribution of biphenyl congeners after 1 h of heating Aroclor 1221 at the indicated temperatures. Bars marked RT indicate the composition of the unheated starting material. Clo = biphenyl; Cl, = monochlorobiphenylisomers; Clz = dichlorobiphenyl isomers; CIS = trichlorobiphenyl isomers.
Chloride analyses were performed by ion chromatography (Dionex, Sunnyvale, CAI. Reaction residues and blanks were extracted with deionized-distilled water, and concentrations of C1- were determined by interpolation using a standard curve containing 5 x 10-4-1 x M
FIGURE 2. Distribution of biphenyl congeners after heating Aroclor 1221 at 400 "C for the indicated times. TABLE 1
Aroclor 1254 Experiments Conducted at 400 "I: congener Clo Cll Cl2 CIS cl4
Omin
10rnin
20min
40min
60rnin
ND ND ND ND
0.404 0.067 0.012
0.564 0.026
0.528
0.552
ND ND ND ND ND ND
ND ND ND ND ND ND
ch
0.136 0.278
cis
0.109
ND ND ND ND
ND ND ND ND ND
c1-.
Results and Discussion Figure 1 summarizes the trend of reactions between a PCB mixture (Aroclor1221 and a mixture ofbiphenyl and mono-, di-, and trichlorobiphenyl congeners) and FeO heated for 1 hat temperatures in the range of200-600 "C. (The relative peak area, or RPA, recorded on the Y-axis is the ratio of the GC-FID peak area of PCB isomers having a particular molecular weight as established by GUMS, to that of the external standard, benzyl bromide.) At 200 "C or below, little loss of PCBs occurred. Significant dechlorination started to occur at 300 "C (extent of conversion, cu. 78%). Further dechlorination, with almost the same overall conversion, was observed at 400 "C; the fully dechlorinated product (biphenyl) made up about 95% of the total congener mixture. At 500 and 600 "C, the yield of biphenyl in the samples began to diminish and the total yield of all chlorobiphenyl congeners also decreased (to 59% and 2% conversion, respectively). Therefore, not only dechlorination but also other reactions must have taken place, especially at higher temperatures, giving rise to compounds undetected by GUMS under our conditions. These could have been oxidation or condensation reactions that produced polar or polymeric products. However, no peaks for possible reaction products were observed in the GC or GCI MS traces; only compounds such as phthalate esters and
aliphatic hydrocarbons, probably derived from the extraction thimbles, were observed. Figure 2 shows the reactions occurringwithAroclor1221 at 400 "C at different reaction times. Most of the dechlorination reactions took place within 10 min and the lower chlorine-containing PCBs were somewhat more efficiently dechlorinated. (Biphenyl was stable at 400 "C under these conditions.) When Aroclor 1254was used, similar results were noted (Table 1). The Aroclor 1254 sample we used contained approximately 28% tetrachlorobiphenyls along with 55% pentachloro and 17% hexachloro congeners. After 10 min ofheating at 400 "C, the Aroclor 1254mixturewas converted in 93% yield to a mixture consisting of 83% biphenyl, 15% monochlorobiphenyls,and 2% dichlorobiphenyls. If heating was continued beyond 20 min, biphenyl was the only observable product. The chlorine atoms removed during dechlorinationwere shown, using ion chromatography, to have been largely or entirely converted to chloride ion. The residue from a reaction between Aroclor 1221 and iron (400"C, 20 min) was extracted with deionized-distilled water. (Underthese conditions, approximately 75-80% ofthe organicallybound chlorine in the starting PCB congeners was lost during the conversion to biphenyl.) After subtraction of blanks due VOL. 29, NO. 9,1995 I ENVIRONMENTAL SCIENCE &TECHNOLOGY 2461
TABLE 2
Aroclor 1221 Experiments with Modified Iron MetaP subtractiva conganar
darting material
CHzCI,
"Cl
CHzCIl 1500 "Cl
additive CHzCIzl300 "C)
water (300"Cl
wnter (500 %I
010IUIO "C)
'Numbers in the body of the tabla are relative peak areas of biphenyl isomers containing the indicated number of chlorine atoms. Each is the mean of four replicates except for the subtractive CH2CI, experiments isix samples). the StsRing material 111 sampled, and the DIO experiment ftwo ramnlesl. ND = not detected.
to water, reagents, and glassware, a [CI-I of 73% of the theoretical maximum was calculated. Control experiments in which Aroclor 1221was heated without iron (data not shown) indicated, at less than 300 "C, a similar component distribution to that of unheated controls. At 400 "C, the proportion of biphenyl congeners was about the same as the initial reactants (no obvious dechlorination had taken place), hut the total quantity of biphenyls decreased to about half the starting concentration, and no additional product peaks were observed by GC or GUMS. At 500 and 600 "C, PCBs were totally converted to compounds undetected by GC. The formation of the hydrogenated PCB analogs is of mechanistic interest since the reactions were performed without solvent or other additives that could have been donors of protons or H atoms. At high temperature, PCBs presumably accept one or two electrons released by FeQ, lose chlorine as Cl-, and undergo further reactions to form biphenyl (13). Inthis process, ahydrogenorprotondonor must participate. One possibility, that the hydrogen atoms of unreacted PCBs are transferred to the site where CI was removed, was ruled out by the high yields of biphenyl observed in the reactions. Further mechanistic experiments with Aroclor 1221 are summarized in Table 2. In "subtractive" experiments, we attempted to remove potential hydrogen donors from the iron by extracting it with methylene chloride before use. The material extracted by methylene chloride was examined by GC/MS to reveal a complexmixtureofaliphatichydrocarbonshavingaboiling point distribution reminiscent of a fuel oil. Reactions with extracted Fea were performed under the same conditions as experiments with unextracted Feo. The sole product observed in the 500 "C samples was biphenyl (RPA = 0.74; conversion, 76%). Therefore, all the PCBs were completely dechlorinated,and the biphenyl yieldwas comparable, even slightlyhigherthan that in the experimentswithunextracted FeO. These data appear to rule out the organic impurities as a source of hydrogen atoms. We next tested the potential involvement of two hypothetical hydrogen donors in "additive" experiments in which we deliberately added methylene chloride (CH2Cb), thesolvent forthe PCB stocksolution, orwater, which could be present in significant amounts in the air or in the solid phases of the ampule. We sealed a known amount of PCBs in the presence of 500 mg of FeQ(extracted or unemacted) and 50 p L of methylene chloride or water in 10-mL ampules and heated them to 300 or 400 "C for 1h. A drop of CH2Clzinhibited the dechlorination; after the reaction, the quantity and distribution of biphenyl congeners was similar to that of the control with a possible small increase in the relative amount of the more highly chlorinatedmaterials. The yields ofthe sampleswithadded 2482 m ENVIRONMENTAL SCIENCE 8 TECHNOLOGY i
VOL. 29. NO. 9.1995
iio 100 80
--s ..-s
--
80
70
eo
3
60
.z -
40
I
% .
2
30 20 10
0
MI2
FIGURE 3. Mass spectroscopic fragmentation panern of biphenyl molecular ions derived from Aroclor 1721 in the absence (a) and in the presence (b) of 020.
CHzC12showed a total RPA equal to 0.81, close to that of the initial starting PCB mixture (RPA = 0.97). It is possible that the CHzClz could have been pyrolyzed at high temperature to give CI', CH2CI',and other radicals. By this mechanism, CI' radicals might become attached to the aromatic rings of PCBs at high temperature to give more highly chlorinated biphenyls. Samples containing added water were heated at 300500 "C. The PCBs were largely dechlorinated, and the amount of product, biphenyl, in these samples was slightly larger than that in the samples without water addition. In a 500 "C experiment, the amount of biphenyl (RPA= 0.76; conversion, 79%) was very close to that observed in the subtractive experiment with CH2Ch-emacted iron. The result may indicate that adventitious water may act as a proton donor to a formal biphenyl anion, which could conceivably be associated with or bonded to cationic iron produced during the redox process. Further evidence for
this mechanism was obtained in an experiment withhoclor 1221 and D20at 400 "C; the isolated biphenyl contained deuterium,as shown by the appearance of greatly enhanced ions at mlz > 155 in its mass spectrum (Figure 3). From a practical standpoint, this basic technique, being simple and inexpensive, may merit consideration at some future time as a method for cleaning up PCB-contaminated environments. No attempt has been made to optimize even the laboratory process, however, in regard to oxygen and moisture content or the effects of co-occurringimpurities, such as hydrocarbons or soil constituents, to give only a few obvious variables. Therefore,much additional research is necessary.
Conclusions Zero-valent Fe promoted the dechlorination of PCBs in the absence of solvent in a high-temperature environment. Most of the PCBs were dechlorinated to biphenyl within 10 min at 400 "C. The dechlorination product, biphenyl, was stable at 400 "C. At temperatures higher than 500 "C, dechlorination and other reactions of PCBs occurred. Water appeared to be the hydrogen donor for the formation of biphenyl, as shown by an experiment with added D20.
Acknowledgments We thank the IllinoisWater Resources Center and the Illinois Office of Solid Waste Research for financial support and Francisco Bosca, Karen Marley, Paul Tratnyek, and Eric Weber for helpful discussions.
Literature Cited (1) Wentz, C. A. Hazardous WusteMunugement; Clark, B. J., Morris, J. M., Eds.; McGraw-Hill Chemical Engineering Series; McGraw-Hill: New York, 1989. (2) Hawarl, J.; Demeter, A.; Samson, R. Environ. Sci. Technol. 1992, 26, 2022. (3) Buser, H. R.; Bosshardt, H. P. Chemosphere 1978, 1, 109. (4) Erickson, M. D.; Swanson, S. E.; Flora, J. D., Jr.; Hinshaw, G. D. Environ. Sci. Technol. 1989, 23, 462. (5) Randall, T. L.; Knopp, P. V. J. Water Pollut. Control Fed. 1980, 52, 2117. (6) Thomas, D. R.; Carswell, K. S.; Georgiou, G. Biotechnol. Bioeng. 1992, 40, 1395. (7) Davies, W. A.; Prince, R. G. H. Process Suf Enuiron. Prot. 1994, 72, 113. (8) Sugimoto, H.; Shigenobu, M.; Sawyer,D. T. Enuiron. Sci. Technol. 1988, 22, 1182. (9) Epling, G. A.; Florio, E. M.; Bourque, A. J. Enuiron. Sci. Technol. 1988, 22, 952. (10) Zhang, S.; Rusling, J. F. Enuiron. Sci. Technol. 1993, 27, 1375. (11) Reynolds, G. W.; Hoff, J. T.; Gillham, R. W. Enuiron. Sci. Technol. 1990, 24, 135. (12) Gillham, R. W.; O'Hannesin, S. F.; Orth, W. S. Metal enhanced abiotic degradation of halogenated aliphatics: laboratory tests and field trials. Presented at HazMat Central Conference, Chicago, IL, 1993. (13) Matheson, L. J.; Tratnyek, P. G. Enuiron. Sci. Technol. 1994,28, 2045. (14) Schreier, C. G.; Reinhard, M. Chemosphere 1994, 29, 1743. (15) Hagenmaier, H.; Brunner, H.; Haag, R.; Kraft, M. Enuiron. Sci. Technol. 1987, 21, 1085. (16) Krupcik, J.; Kocan, A.; Petrik, J. Chromutogruphiu 1992,33, 514.
Received for review December 7, 1994. Revised manuscript received May 19, 1995. Accepted June 8, 1995. ES940741Y
VOL. 29. NO. 9.1995 I ENVIRONMENTAL SCIENCE &TECHNOLOGY a 2463
