Document YGw01JkpBz34vjqd7q2Y6NDDk

C fcem osphere, V o i. 1 5 , N o s.9 - 1 2 , p p 1 2 6 5 -1 2 7 1 , 1986 P rin te d in C re s t B r it a in 0 0 4 5 -6 5 3 5 /8 6 < 3 .0 0 + .0 0 Pergam on J o u rn a l L td . PYROLYSIS WO COMBUSTION OF AROCLOR 1254 CONTAMINATED DIELECTRIC FLUIDS Addis Electric Power Research Institute DIOXIN 85-5th International Symposium on Chlorinated Dioxins and Related Compounds ABSTRACT Pyrolysis and combustion products were determined for several levels of PC8 contamination In mineral oil and other dielectric fluids. Yields of PCOFs were roughly proportional to the quantity of PCB-In the feed. HAOtfSOIM) The electric utility Industry as a major purchaser of PCBs 1n the past has been left with a legacy of PCR problems of two different dimensions. Although the reality of PC8 as a prob lem Is still being debated and evaluated, the Industry is retrof1ll1ng or replacing Its PCB equipment at a steady pace, A short time ago, there were about 40,000 (_1_) PCB transformers 1n utility hands with perhaps an equal or larger number owned by nonutllTtyentitles. Several well publicized PCR fires (2_,J) 1 the United States have accelerated the effort toward removal of the askarel (generic term for PCR or PCB/tr1-/tetrachlorobenzene) equipment. Two options are open for elimination of PCBs In transformers. The first is retrofllling and the second 1s replacement of the transformer. Each method has Its adherents and Its detractors. Retrofllling^ 1n the absence of further Improvement of the technology, takes more than a year before a transformer can be reclassified as uncontamlnoted (In the United States, below SO ppm PCR). During most of this time, PCB concentration 1n the transformer fluid 1s still above 500 ppm because even 1f one uses the best available technology, between 2 and 5* of the old liquid remains behind after draining a transformer as thoroughly as possible. It Is estimated that this residual takes approximately 3 months to come to equilibrium with the replacement fluid. The dilution process mist therefore be repeated several times before an appropriately low PCB level 1s reached. We understand that there is cnnslderahle,research presently underway to speed up the removal of this trapped material; however, the work has not yet come to fruition. Where physically possible, complete replacement of the transformer appears to be the easy way out. However, disposal of the transformer requires draining of the PCB followed by flushing with a solvent. The liquids must then he destroyed in a licensed Incinerator while the transformer carcass must go to a certified landfill. Certified landfills In the USA are rapidly disappearing and 1t 1s also considered possible that these may become the next generation of problem cleanup sites. A second area of concern 1s the roughly 2,000,000 (11 mineral oil transformers, which over the years have Inadvertently become contaminated with PCBs at the SO ppm level or higher. About IDs of these are contaminated above 500 ppm and must be treated as PCB transformers. This pair of problems has fostered a need to learn more about the pyrolysis and combustion of PCB, both 1n Its concentrated state, and at various levels of contamination In retrofUl fluids, such as silicone and mineral oil. PCB 1s also of Interest as a contaminant 1n tetrachloroethylene, because 1t 1s a potential retroflll fluid, although 1t 1s more likely to be used as a replacement fluid. HORN PLAN A project has been sponsored and partially funded by the Electric Power Research Institute (EPRI) for the New York State Department of Health to study pyrolysis and combustion 1265 G EN P0107* 783604 fl; 1266 produets of PCBs, both as a concentrated fluid and as a contaminant 1n minerai oli and several replacement fluids. In-this project we define pyrolysis as the high-temperature decomposition of a fluid 1n an oxygen-depleted atmosphere. Combustion 1s defined as oxldatlqn In the presence of an open flame and an excess of air. The wort plan Includes the Investigation of Aroclor 1254, "neat," as well as at a 50, 500, and 5000 ppm Impurity level 1n mineral oil silicone and tetrachloroethylene. Since a mixture of tri- and tetrachlorobenzene Is frequently found as diluent 1h askarels, these two compounds, both separately and as a blend, also are being Investigated. EQUIPMENT PROCESS Pyrolysis Pyrolysis trials as described by Eadon (41 were conducted using a simple thermostatically controlled apparatus, capable of accorrrao3at1ng a 6-cm diameter metal block, within Its 9-cm-long heated region. In order to minimize hazards and disposal problems, yet permit sufficient product formation to facilitate detection of PCDQs and PCQFs, pyrolyses were performed on 100 pi samples. In an attempt to differentiate this work from other earlier and ongoing Investigations, as well as to simulate more accurately certain catastrophic Incidents, pyrolyses were conducted at atmospheric pressure. The necessity of containing the starting materials and nongaseous pyrolysis products 1n an open system led to the use of an 8 mu ID x 0.5 m glass tube, sealed at one end, and mounted vertically. The material to be pyrolyzed was deposited at the sealed end, then Inserted Into a tight-fitting hole In the preheated metal block 1n the heating apparatus. Typically, the liquid refluxed up the Inner walls of the tube; the length of the tube and Its comparatively large unheated volume kept the reflux level well below the open end 1n all experiments. To assure containment, the topmost 5 cm of the tube was chilled 1n dry 1ee and the end of the tube was connected to a charcoal trap. No visible material was trapped 1n the chilled region, and excellent mass balances were generally observed. Combustion The design of the combustion apparatus was also constrained by the necessity of assuring that all -discharged gases pass through a trapping system capable of efficient removal of PCBs and any potentially toxic products. The combustion chamber consists of a 1 m quartz tube with a 22-nn ID. One end of the chamber accommodates a modified blast burner and Inlet tube through which the end of a 1/16" x 24" (1.6 x 635 mm) syringe needle Is Introduced via airtight connections. The Inlet tube and needle are mounted to allow sample Introduction Into the flame of the blast burner. Sample addition 1s accom plished using a 10-ml syringe mounted 1n a worm syringe drive. The combustion chamber Figure 1; Pyrolysis Equipment 1s mounted in a furnace capable of main taining temperatures of l00-lG0fl*C 1n three Independently controlled zones. The effluent of the combustion chamber is passed throuan a water filled Implnger, then through an XAD-2 packed adsorbent tube which in turn Is attachea through an orifice to a vacuum line. XAD-2 has been shown here and elsewhere to be an efficient trapping agent for polychlorinated dihenzofurans (PCOes) and dihenzo-o-d1oxins (pcnos). 783605 Collection < --------- Syringe needle and makeup gas / Burner ------- Syringe needle Figure 3: Feed Syringe ANALYSIS Because of the hazardous nature of the compounds to be prepared and to check out equipment expeditiously using rapid analytical techniques, the trials were approached stepwise. Initial runs 1n each case were made with unchlorinated biphenyl to determine the suitability of the.equipment and to find a first approximation of proper operating conditions. These runs were followed by trials using Individual PC8 congeners known to form specific PCDF mixtures. It was anticipated that the relatively simple products formed could be analyzed by capillary GC/EC or GC/FID after chromatographic cleanup. This technique was successful, except 1n the pyrolysis of contaminated mineral oil. The mineral oil itself produced a complex mixture, and 1t was necessary, after appropriate up-front cleanup, to resort to the use of GC/HS frwi the start. Having thus optimized as far as possible operating conditions and analytical procedures, runs were finally made using Aroclor 1254 as the test fluid. 783606 1268 *r-s. Figure 4: Combustion Product Trap RESULTS Pyrolysis of Aroclor 1254 in Mineral 011, Silicone, and C2CI4 Under the conditions chosen for this project, conversion of PCBs to PCDFs in the trial runs reached an approximate maximum around 550C. Most of the subsequent runs with Aroclor 1254 were then made at this temperature setting. Where other temperatures were used, they have been noted 1n the text or tables. All runs were of 15 minutes duration. To minimize run-to-run experimental variations found 1n the preliminary work, and to produce larger samples for analysis, a series of six runs was made at each set of conditions. The six runs of each set were always made during a single day and combined randomly (prior to cleanup) into two composites for analysis. Results are tabulated in Tables l and 2. Note T A BU l: P O F Foraed (ng^Sran of mature* pyrelyzed: Aroclor 12S4 In Insulating Fluids Trele Oescrlotlon 2582 <2583 soasa 50855 50856 50857 asi 50859 45298 295 <2571 100t Aroclor 1254 100S Aroclor 1254 5,000 ppa* In nlnaral oil 5,000 ppn In nlneral oil 500 ppn In nlneral ell 500 ppn In nlneral oil 50 ppn in nlntraVoll SO ppn In nlneral oil 5,000 ppn3 *56* In silicone 5,000 ppn* 1n tttraehloroethylene Native M i * In nlneral oil Added Found 2378* TCOf 1,300 965 15.0 11.8 1.6 1.1 . 9.9 1.2 9.3 9.1 Total TCOF 8,200 5,500 45.0 .33.0 3.1 107.0 2.9 1237812 Base Peak PrCOF PrCOF Total PrCOF 17.0 17.2 1.8 1.9 0.4 0.9 4.5 4.6 - 62.0 55.0 6.7 5.8 l.l 1.7 . 16,000 10.000 165 171 17.2 18.2 . 3.1 3.4 70.0 28.0 Base Peak HiCOf Total HiCOF 8,000 5,100 87.0 60.0 6.3 6.8 1.5 1.3 2.2 5.5 8.000 5,100 162 205 16.8 18.1 3.5 . 6.9 12.5 17 .-2 (23478) 17.5 32.4 (234678) 23.0 Total teCDF 60 . 24.0 14.0 < 10 2.2 . QCOF - <2 0.23 C.1S 0.9 56 (OCOO) 47 (1) Caen nlneral Oil reit represent! 1 coreosIte or three seprate pyrelysts -Men re coaofned prior to analysis to ntnlnlzv run-to-run trillion. M l pyrolyses tt 5S0*C ror 15 ntnutes. (?) Includes coeluters on OB-S colum. (3) 650`C. (*) Chlorinated fluorenest?) at 10* higher level. (5) 60Q'C. (6) Mineral ell spiked H n PCDFs to test netted. 783607 GENP 01079( 1269 fairTMLC 2: Toned (ng) per Grm of 'Areclor 1254 f jr o ljm d In Insulating flu id Saeplf O n e r lotIon 42562 42SB3 soas4 S06S6 50556 50857 50858 50859 45298 44295 IPOS Areclor 1254 100X Areclor 1254 5,000 ppfll In elntnl ell 5,000 pc In atntral oil 500 pea tn aimral oil SCO pea In Blneral oil 50 ppa In otnaral oil SO ppa In atntral oil 5,000 ppa* In *111cent 5,000 pea* In tetracftloroatftylant 2376* TCOf 1,300 965 3,000 2.360 3.200 2.200 - - 1,980 240 Total tcot 1.200 5,500 9,000 6,600 6.200 - - 21.400 560 12378* l u e Teak WDT PeCOf - - 3,400 3,440 3.600 3,800 1,000 16,000 900 960 - - 12,400 11.000 13,400 11.600 22,000 34,000 - - Total FeCOF 16,000 10,000 33,000 34,209 34,400 36,400 62,009 60,000 14,000 5.600 last Pttk Total HiCOf 1,000 a.ooo 6.100 5,100 17,400 32,400 12,000 41,000 12,600 *33,600 13.600 36,200 30,000 70,000 26.000 0 - 1.380 1,100 2,500 Total HtCOf 60 4,800 2.800 4,400 - 30 180 OCOf - 64 46 - - * - (1) Caeft afnaral oil tuplo repreitnta coueoalte of three uparat* p y ro iy m. T h e were conined prior to analyst* to nlnlolie run-to-run variation. All pyrolysM at SSO'C for 15 tnutas. (2) Include* coelutert on DB-5 caluai. (3) 650*C. (4) 600*C. that Tables 1 and 2 report the sane data but expressed In different form. Table 1 shows ng PCOF formed per gram of total mixture pyrolyzed. while Table 2 shows ng PCDF per gram of Aroclor 1254. It may be seen that conversions of PCBs to PCOFs are substantially identical for all congeners and chlorination groups measured for 5000 and 500 ppm. and are less than an order of magnitude different for 1001 PCB. Conversion appears to Increase slightly at 50 ppm, but this may be the result of measuring error due to analytical difficulties at this level. Pyrolyses In silicone and tetrachloroethylene show conversion efficiencies qualitatively similar to mineral oil* for tetra- and penta-COF, but are apparently dropping off for higher chlorination levels. Added wort must be done to confirm these levels. Analysis of the pyrolysis products of PCBs In both silicone and tetrachloroethylene yielded qualitative Indications of several products related to PCOF/PCDO. In silicone, compounds tentatively identified as chlori nated fluorenes were found 1n the MS scans. Also, 1n pyrolyzlng 1.2.3,4,5 pentachlorobiphenyl 1n silicone, a series of compounds CH. (which may be methylated chlorinated fluorene) was found. Chlorine atoms plus methyl-groups total four 1n each case (Figure 5). 1-cWiy au Tetrachloroethylene adduct ot PCS -age Cl Methylated - chlorinated fluorene Pyrolysis of 2-ehlorobiphenyl 1n silicone has fluorene as a major product. This has been compared with an authentic standard. Other * compounds listed above as products In the si1leone and discussed 1n the tetrachloroethylene wort below will be compared with authentic standards before their presence Is considered confirmed. Polychlorinated fluorene Polychlorinated Biphenylene Figure 5: Compounds Related to PCOF/PCDO CrJBNP 01079 783608 V . . , ' . t , 1270 In the combustion and pyrolysis of 1254 1n tetrachloroethylene, preliminary work shows that some different tetri* end penta-CDFs ire being formed, compared to "neat" 1254 or 1254 1n mineral oils. Chlorinated fluorene may also be forming. In addition, a compound that may be a reaction product of PCS and tetrachloroethylene Is possible. Combustion of Aroclor 1254 In Mineral Oil, Silicone and C2C14 Conversion for the Isomer groups studied (Table 3) was near optimum for a combustion wall temperature of 550*C and a 3-second residence time. With feed solutions of mineral oil containing Aroclor 1254 at 50, 500, and 5,000 ppm, tr1-, tetra-, and penta-CDF conversion In ng/g PCB fed fell within a narrow range for each of the Isomer groups. Under all conditions, Next- and hepta-CDF formation was not detectable. Combustion trials with PCBs 1n silicone were not completed successfully. Large quant ities of S102 formed. The finely divided S102 tended to plug the equipment, particu larly the feed needle and the sample collection train. In C g C ^ solution, combustion yields were similar to those found with mineral oil for the -Cl3 isomers. There were substantially greater yields for the'more highly chlori nated Isomers, reflecting chlorination of lower chlor inated CDFs. Combustion in C2C1. under oxygen depletion provides a substantial yield of b1phenylenes. TABLE 3: Caftustto* of Aroclor U M tn Insulating fluid at S50*C - ] Seconds 1234 In to). PCB't CottustM destroyed SoWent_________ m/ol________t________ CL, Mineral Oil Mineral Oil Mineral Otl CjC14 c2 4 Silicone^ 00 soo M MOO SOO M _ 82-S8 ST-9? 68-90 68 92 72 . 0.72 0.4 0.3 0.44 0.24 0.44 /n 1 PCDfs foriffgd*1) Cl, d s c y CiT 0.S8 0.33 0.17 .(*) 0.034 M i JZ) .(1 0.3 0.0S6 M i Mi I.SO 1.35 0.33 0.50 0.7 0.66 0.23 0.70 1.32 1.0 0.38 0.16 (1) lascd on PCS In feed. (2) Son* dttcctod. (21 bins unsuccessful. Cqulpncnt plugged with SIOj. OTHER WORK IN PROGRESS Pyrolysis and combustion of tr1- and tetrachlorobenzene are underway 1n the laboratory. These results will be of value 1n assessing the potential fire products frotp askarels where the PCB is diluted with these solvents. Preparation of certain synthetic chemicals such as substituted fluorenes is being undertaken. These will be used to confirm the structure of several of the unknowns that appeared during the GC/HS analysis. DISCUSSION Part of the original Impetus 1n this project was to determine whether formation of partial oxidation products of PCB was linear with Increasing dilution In a solvent. This question appears to be answered. Within the constraints of variations 1n analytical conditions and recoveries during cleanup and analyses of extremely small quantities of material, the linearity should be considered good; tn most cases, there was no more than >a factor of 10 variation 1n a range of concentrations from 50 ppm to 1001 of the askarel tested. Improvements In the pyrolysis and combustion schemes during the course of the project provided this degree of resolution. No doubt, this can be Improved upon with further sophistication 1n the work. The results of these laboratory trials are found to be significantly different from those of other workers In the field (5-9). This fact brings with It a word of caution 1n applying any of these results to real-world PCB fires except for use as guiding principles because each real PCB fire, under completely random conditions, 1s different from all others. Even GENP 010798 783609 1271 within i given fire, there will be in infinite n n g e of competing reactions. It c m be assumed that only a small portion of the combustion process has optimum conditions, and that the combustion process varies with time, leading to the partial destruction of the various combustion by-products. Thus, all of the research being done must be considered as setting a boundary or worst-case condition that gives a general direction to the Investigation of the real world. A number of new avenues for exploration have been seen here and 1n the work of others. Among the more significant ones are the finding of relatively large quantities of other chlorinated polycyclic aromatic compounds (PCAs). A second 1s the need for a more rapid method of analysis. A bioassay designated the Flat-Cell Assay (10), based on a change 1n In vitro growth and morphology of a line of skin cells has been sKown tp be very sensitive and relatively specific for the more toxic of the PCOF and PC00 compounds. The biological mechanism of this effect 1s considered to be related to the development of chloraene In humans after 2,3,7,3-TCOF exposure and Is thus, a relevant end point. This method Is already applicable to certain products of combustion. It Is being modified to detect and assay these products 1n the presence of solvents such as mineral oil, which currently Interfere with the test. Calibration against a range of chlorinated PCAs would follow. AOCROMLEDGKEffTS The help of K. Aldous, R. Briggs, G. Eadon, 0. Hllker, K. Kidd, A. Narang, R. Narang, P. O'Keefe, and R. Smith are gratefully acknowledged. REFEREKCES 1. RPC. Volume III - Report of the Study of PCBs In Equipment Owned by the Electric Utility Industry. Prepared for the Edison Electric Institute, Washington, D.C., February 1982. 2. N. C. K1m and G. Eadon. The Binghamton State Office Building. Workshop Proceedings: PCB By-product Formation, EPRl S/EL-4l04, July 1985, p. 5-1. 3. R. L. Wade. Utilization of Quantitative Risk Assessment Techniques In the Development of Decontamination Standards (A asc Study - San Francisco, alifornia). Ibid, p. 5-16. 4. G. Eadon. Pyrolysis and Combustion of Mineral Oil. Tetrachloroethylene and Silicone 011 Mixtures Containing Aroclor 125TI Ibid., p. 3-24. - 5. S. E. Swanson, H. D. Erickson, and L. Moody. Products of Thermal Degradation of Dielectric Fluids. Prepared for the U.S. Environmenta1 Protection Agency, Washington, QC. Interim Report No. 2, May 1985. 6. H. R. Buser, H-P. Bosshardt, and C. Rappe. Formation of Polychlorinated Dlbenzofurans (PCDFs) from the Pyrolysis of PCBs. Chemosphere /(l), 19>8, pp. 1Q9-119. 7. H. R. Buser and C. Rappe. Formation of Polychlorinated Dlbenzofurans (PCDFs) from the Pyrolysis of Individual Isomers, cnemosphere b u ) , 19/9, pp. 157-1/4. 8. B. Dellinger, V. Rubey, 0. L. Hall, and S. L. Mazer. Laborary Investigation of the Hlgh-Temperaturc Formation and Destruction of PCDFs. workshop Proceedings: P C B T y product Formation, ephi C5/EL-41U4, Ju ly 1985,'p. 3-17. 9. C. Rappe, S. Marklund, and L-0. Kjeller. Formation of PCDFs from PCBs. Ibid., p. 3-28. 10. J. F, Glerthy and D. Crane. In Vitro Bioassy for Dtoi1n-L1ke Activity Based on inAlterations In Epithelial Cell Proliferation and Morphology. Fundamentals in Aool1ed Toxicology, Vol. V, 1975', press. 783610 GENP 01079.