Document Z8ovo4eBBp9xD2B2YDD715bEd

-i R E P 0 R T . S U M M A R Y *, / * \ ---- "" ~~ ii - SUBJECTS Hazardous and toxic waste management / Transmission substation design s and operation / Distribution substations TOPICS PCB Chemical analysis Transformers / ' ' ' " " ' Capacitors ' PCDF-PCDD Insulating oil . AUDIENCE Environmental managers / Distribution engineers Analysis of Polychlorinated Dibenzofurans and Polychlorinated Dibenzo-p-Dioxins in Itansform ers and Capacitors V olu m es 1 -3 PCDF and PCDD-- by-products of PCB pyrolysis and combus tion-- apparently cause many of the health effects originally attributed to PCB. In parallel and round^robin studies, indepen dent laboratories found that norm al.equipm ent operation and arc ing failure cause little if any oxidation and do not significantly modify PCB fluids. BACKGROUND Partial oxidation of askarels--mixtures of polychlorinated biphenyls (PCB) and tri- and tetrachlorobenzenes--produces polychlorinated dibenzofurans (PCDF) and polychlorinated dibenzo-p-dioxiris (PCDD). However, analytic techniques for separating the active components in these mixtures had not matured prior to this study. Information had also been insufficient on the quantity and types of PCDF and PCDD in as-manufactured PCB and on the formation of these by-products during normal operation or arcing failure in utility equipment. Because of these uncertainties, utilities and regulators have assumed the worst-case PCDF-PCDD formation when there is a PCBrelated accident, thereby overstating the potential toxicity. EPRI report EUEAp4858 describes portions of the research performed in these projects. OBJECTIVE lb improve techniques for measuring PCDF/PCOD in the presence of PCB. APPROACH Five laboratories (including two whose studies are discussed in report EUEAr4S58) performed round-robin gas chromatography-mass spectrome try analyses of PCDF-PCDD samples using specially prepared carbon-13 (13C) PCDF spiking compounds. After the first set of analyses, each labora tory modified its techniques and analyzed a larger group of samples from utility equipment. Another organization statistically analyzed all the labora tory results. RESULTS * These three volumes report the second group, of PCDF-PCDD analyses. Results from all participating laboratories correlated remarkably well. Some of the results follow. EPRI EUE/V5443S Vola. 1-3 gBNP 011059 783870  Use of the 13C compounds allowed successful separation of the com ponents in several closely eluting pairs of physiologically active PCDF and less-active compounds. Analyses of samples from utility equipment suggested that the PCDF-PCDD present did not result from high-load, high-temperature operation. Volumes 1 and 2 describe several analytic techniques, with the appen dix of Volume 2 including a report on the Analysis by the University of Umea in Sweden, as well. Volume 3 summarizes all the laboratory techniques, presents comprehensive statistical analyses, and provides an executive summary. EPRI PERSPECTIVE This research is a landmark in PCDF-PCDD analysis. One important result is the synthesis of new PCDFs for use as spiking compounds in gas chromatography-mass spectrometry analysis. The ability to sepa rate many of the active PCDF compounds from more innocuous materi als is also of great value. The high correlation of results from several laboratories indicates that researchers can exercise great flexibility in selecting techniques and developing facilities. PROJECTS RP2028-7, RP2028-6, RP202B-10 EPRI Project Managers: Gil Addis; Jacques Guertin Electrical Systems Division; Environment Division Contractors: New York State Department of Health; Battelle Columbus Laboratories; Research Triangle Institute For further information on EPRI research programs, call EPRI Technical Information Specialists (415) 855-2411. 783871 G E N P 011060 A nalysis of P olychlorinated D ibenzofurans and Polychlorinated D ibenzo-p-D ioxins in Transform ers and C ap acito rs Volume 2: Formation of P C D F and PCDO in Askarei and Contam inated M ineral Oil Equipm ent EL/EA-5443, Volume 2 Research Project 2028-6 Final Report, March 1988 Prepared by BATTELLE MEMORIAL INSTITUTE Battelle Columbus Division505 King Avenue Columbus, Ohio 43201-2693 Principal Investigators W. M. Cooke F. L. DeRoos With contributions from UMEA UNIVERSITY Department of Organic Chemistry Umea, Sweden S-901 87 Principal Investigator C. Rappe Prepared for Electric Power Research Institute 3412 Hillview Avenue Palo Alto, California 94304 EPRI Project Managers G. Addis Transmission Substations Program Electrical Systems Division J. Guertin Land and Water Quality Studies Program Environment Division 783872 ORDERING INFORMATION Requests for copies of this report should be directed to Research Reports Center (RRC). Box 50490, Paio Alto, CA 94303, (415) 965-4081. There is no charge tor reports requested by EPR I mem ber utilities and affiliates, U.S. utility associations, U.S. government agencies (federal, stale; and local), media, and foreign organizations with which EPRI has an information exchange agreement. On request, RRC will send a catalog of EPRI reports. B ectnc Power Research Institute and EPRI are registered service m a rts at B e a n e Power Research In srtu ta Inc. CopyngW 1988 E lectric Power Research Institute, In c A ll rights reserved. NOTICE This report was prepared by BATTELLE as an account of w ork sponsored by the E lectric Power Research Institute. In c {EPRI). N either EPRI. m em bers o t EPRI. BATTELLE. nor any person a a in g on behalf o f any of them : (a) makes any warranty, express o r im plied, w ith respect to the use of any inform ation, apparatus m ethod, or process disclosed in this report o r th a such use m ay not infringe pm eiety owned rights; o r (b) assum es any liabilities w ith respect to the use of. o r fo r dam ages resulting (torn the use of. any inform ation, apparatus, m ethod, o r process disclosed in this re p o rt Furtherm ore if any item (fo r exam ple a m odel, prototype o r test-piece) is deihered pursuers to m is Agreem ent IT SHALL CARRY NO WARRANTY OR GUARANTEE WHATSOEVER INCLUDING WARRANTIES OF FITNESS FOR PURPOSE OR OF MERCHANTABILITY.' 783873 ABSTRACT A study was conducted by five Independent laboratories to evaluate methods for the measurement of trace levels of polychlorinated dlbenzofurans (PCDFs) and polychlorinated d1benzo-p-d1ox1ns (PCDDs) in utility dielectric liquids. Each laboratory evaluated a different analytical method. The method evaluated by Battelle involved spiking with labelled Internal standards, analyte enrichment by column chromatography, and analysis by high resolution gas chromatography/high resolution mass spectrometry (HRGC/HRMS). Baseline analyses of five dielectric fluids were conducted by each laboratory as an initial test of their methods. The variability of results between the five methods was within expectations, and the five methods were deemed suitable for round robin evaluation. In the round robin evaluation, the five methods were used to analyze 10 liquid samples, including seven utility dielectric fluids. The liquid samples were spiked with native PCDD/PCDF to estimate accuracy. The Battelle results were generally within 70 percent of the spike value except for HxCDF which exhibited a strong positive bias. This result was considered an outlier. The Battelle results were consistently lower than the average results obtained for the five laboratories. Overall, the methods proved to be both rugged and reliable when used with labelled quantification and recovery standards. The in-service liquids used in this study were found to have relatively low levels of PCDDs and trace amounts of PCDF. Liquids from transformers with more than 25 years of service had higher PCDF levels than liquids from newer transformers. Liquids from utility appliances exposed to thermal and electrical stresses did not have elevated PCDO or PCOF levels. PCDF levels increased as the Aroclor number (and relative amount of chlorination) increased. GENP011063 iii 783874 * -* ACKNOWLEDGMENTS The authors would like to thank the EPRI technical management team led by Dr. Gilbert Addis. Dr. Addis and h1s colleagues, Dr. Jacques Guertln, and Dr. Ralph Komal, guided this program and provided encouragement to all the participants. Dr. Thomas Rouse (General Electric Company) is acknowledged for his contributions concerning utility use of dielectric liquids and development of the engineering stress test protocols. The authors are grateful for the technical assistance provided by Battel le scientists who developed the methods, and performed the analyses reported in this document. Sample preparation was coordinated by Joe Hatchel, data computation was conducted by Susan Watson, and mass spectral analyses were performed by Dan Aichele, Charles Sueper, and Michael Zimmerman. We would also like to thank our colleagues who contributed the parallel analyses to this interlaboratory collaborative study. They are Dr. Sydney Gordon (Illinois Institute of Technology Research Institute), Drs. George Eadon and David Hilker (New York State Department of Health), Ms. Maureen Kilpatrick (Radian Corporation), Dr. Edo Pellizzari (Research Triangle Institute), and Dr. Christoffer Rappe (University of Umea). GENP 011064 V 783875 CONTENTS Section 1 INTRODUCTION 2 EXPERIMENTAL ACQUISITION OF IN-SERVICE SAMPLES ANALYTICAL TECHNIQUES SAMPLE PREPARATION AND SPIKING ANALYTE ENRICHMENT ANALYSIS QUALITY ASSURANCE RECOVERY OF INTERNAL STANDARDS QUANTIFICATION CALIBRATION CURVE DETECTION LIMITS BASELINE ANALYSES IN-SERVICE ANALYSES 3 BASELINE ANALYSES BACKGROUND RESEARCH STANOAROS USED FOR SPIKING RESULTS Chemical Analyses Baseline Recovery Data Method Reliability 4 SUPPLEMENTAL ANALYSES CROSS CHECK ANALYSES ANALYSES FROM THE UNIVERSITY OF UMEA 5 ROUND ROBIN ANALYSES OF IN-SERVICE LIQUIDS Accuracy Estimate 6 Comparison with Parallel Analyses DISCUSSION GENP 011065 V11 783876 Pass 1-1 2-1 2-1 2-2 2-2 2-4 2-5 2-5 2-5 2-6 2-7 2-3 2-8 2-8 3-1 3-1 3-2 3-4 3-4 3-4 3-10 4-1 4-1 4-1 5-1 5-1 6-15-13 Section METHODS IN-SERVICE LIQUID ANALYSES SUMMARY APPENDIX A APPENDIX B APPENDIX C IN-SERVICE LIQUID QUESTIONNAIRE ANALYTICAL CURVE DATA FOR THE CROSS CHECK ANALYSES LETTER REPORT FROM UMEA Page 6-1 6-2 6-3 A-l B-l C-l 783877 v111 GENP 011066 ILLUSTRATIONS Figure 4-1 Calibration curve for 2,3,7,8-tetrachlorodi benzofuran over the range 40 pg to 2.5 ng PS 4-2 GENP011067 1x 783878 TABLES Table 2.1 EQUIPMENT RECOMMENDED FOR SAMPLING FOR PCOF ANO PCOO ANALYSES 2.2 SURVEY OF LIQUID SAMPLE5 AVAILABLE FOR THE IN-SERVICE TRANSFORMER LIQUIDS STUDY 3.1 BASELINE TEST LIQUIDS 3.2 CONCENTRATION OF PCDF/PCDD STANDARDS SPIKED INTO TRANSFORMER LIQUIDS 3.3 BATTELLE FURAN AND DIOXIN DATA FOR THE BASELINE ANALYSES STUDY 3.4 IITRI FURAN AND OIOXIN OATA FOR THE BASELINE ANALYSES STUDY 3.5 NYSOOH FURAN AND DIOXIN OATA FOR THE BASELINE ANALYSES STUDY 3.6 RADIAN FURAN AND DIOXIN DATA FOR THE BASELINE ANALYSES STUDY 3.7 ABSOLUTE RECOVERIES OF LABELLED STANDARDS IN THE BASELINE STUDY 3.8 FIVE LABORATORY RESULTS OF ANALYSIS OF AGED MINERAL OIL 5.1 CHLORINATED FURAN AND DIOXIN RESULTS - SAMPLE ISL-02-A 5.2 CHLORINATED FURAN AND OIOXIN RESULTS - SAMPLE ISL-03-A 5.3 CHLORINATED FURAN AND DIOXIN RESULTS - SAMPLE ISL-04-0 5.4 CHLORINATED FURAN AND DIOXIN RESULTS - SAMPLE ISL-08-A 5.5 CHLORINATED FURAN ANO DIOXIN RESULTS - SAMPLE ISL-17-A 5.6 CHLORINATED FURAN AND OIOXIN RESULTS SAMPLE ISL-23-0 5.7 CHLORINATED FURAN AND DIOXIN RESULTS SAMPLE ISL-27 5.8 CHLORINATED FURAN AND OIOXIN RESULTS - SAMPLE EZ29-53-01 ' 5.9 CHLORINATED FURAN AND DIOXIN RESULTS - SAMPLE E729-53-02 5.10 CHLORINATED FURAN AND DIOXIN RESULTS - SAMPLE EZ29-53-03 5.11 ACCURACY DETERMINATION USING SPIKED 8ASELINE LIQUID ANALYSES 5.12 MEAN FURAN RESULTS FOR SEVEN UTILITY DIELECTRIC FLUIDS EaSi 2-2 2-3 3-1 3-3 3-5 3-6 3-7 3-8 3-9 3-11 5-2 5-3 5-4 5-5 5-6 5-7 5-8 5-9 5-10 5-11 5-12 5-14 783879 X1 011068 GENP SUMMARY SYNOPSIS This report describes a study to develop reliable methods for the chemical analysis of polychlorinated dibenzo-p-dioxins (PCDD) and polychlorinated dibenzofurans (PCOF) in utility dielectric liquids. Prior to this study a great deal of data was available on levels of PCDD and PCDF in utility fluids; but since the methods used to obtain this data had never been rigorously challenged in a statistically valid study, the reliability of the existing data bases was suspect. Current analytical methods are affected by high levels of polychlorinated biphenyl (PCS). Since PCB is an interfrent in most methods proposed for the analysis of PCDD/PCDF, and many existing utility test appliances contain PCB as the major dielectric fluid or as a contaminant of other fluids (due to residual PCB), investigation of reliable analytical methods was warranted. Five independent laboratories participated in this program- Each laboratory employed a different analytical procedure to measure trace levels of PCDF and PCDD in dielectric liquids. The method used by Battel1e involved spiking with labelled internal standards, analyte enrichment by multiple column chromatography, and analysis by high-resolution gas chromatography/high-rsolutIon mass spectrometry (HRGC/HRMS). BASELINE ANALYSES The first phase of this study, baseline analyses, involved comparing methods for analyzing difficult matrices; e.o.. Aroclor. Baseline analyses of five dielectric fluids were conducted by each laboratory as an Initial test of their methods. The variability of results among the five methods was within expected levels. The five analytical procedures proved to be rugged and reliable when used with isotoplcally labelled, quantification and recovery standards. Use of labelled standards 1s especially important in the Askarel and mineral oil matrices since the high organic burden found in these materials requires secondary partitioning to remove interfering PCB and background hydrocarbons. GENP 011069 s-i 783880 The relative average deviation is the absolute difference between individual results and the mean (or true value), divided by the number of determinations. For samples such as Aroclor 1242, Aroclor 1016, and mineral oil, the average deviation for baseline samples was often less than 50 percent. This result was found in individual laboratory replicate analyses and in grouped data from different laboratories. All five methods were deemed suitable for use in the second phase of the program. ROUND ROBIN STATISTICAL ANALYSIS The second phase of the study involved a more severe test, a round robin statistical evaluation that employed field samples collected from in-service utility applications. In the round robin evaluation, the five independent laboratories used different methods to analyze 10 liquid samples, including seven utility dielectric fluids. The in-service dielectric liquid samples were collected from four Askarel-filled network transformers, one arc furnace transformer, one electrostatic precipitator (ESP) transformer, and one capacitor. The Battelle results were generally within 70 percent of the spike value except for hexachlorodibenzofurans, which exhibited a strong positive bias. This result was considered an outlier. The Battelle results were consistently lower than the average of results obtained for data from all five laboratories. All samples of the in-service liquids tested had relatively low amounts of PCDD. This result supports the hypothesis that dioxins are not a major constituent of Askarel and PCB-contain1ng fluids. Except for the mineral oils, trace amounts of polychlorinated dlbenzofurans were found in all of the Aroclor samples tested. The mineral oil samples had extremely low amounts, If any, of PCDF. Even the inservice liquid from Transformer 4, a network distribution transformer that was involved in a high-energy excursion, had minimal amounts of PCDD and PCDF. Transformer 4 had visible carbon, and major signs of scorching on the case and in the area of the electrodes. The Askarel transformers with more than 25 years of service tended to have higher levels of PCDF than the appliances put into service for a shorter time. It is possible that manufacturing controls improved as the production of Askarel matured in the 1960s such that Askarel was produced with lower levels of PCDF in the asdelivered fluids. GENP 011070 S-2 783881 Two utility appliances were selected by the EPRI Technical Advisory Committee for special attention: an Askarel-filled ESP transformer and an oil-filled arc furnace transformer. These two samples were the "worst case" for testing service stress on the dielectric fluids. The arc furnace had essentially no PCDF or PCDD, and the PCDF levels in the ESP transformer appeared to be dictated by the age of the unit (27 years of service) and the fluid composition (Aroclor 1260). This extended period of stress did not appear to elevate the PCDF levels above typical Aroclor 1260 levels. One capacitor was tested in the in-service liquid screening program. It was found to have relatively low levels of PCDF compared to the other Askarel-filled units. CONCLUSIONS The in-service liquids used in this study were found to have minimal levels of PCDO and only trace amounts of PCDF. Liquids from transformers with more than 25 years of service had higher PCDF levels than liquids from newer transformers. Liquids from utility appliances exposed to thermal and electrical stresses did not show elevated PCDD or PCDF levels, indicating that high-energy events (electrical and thermal excursions) do not appear to generate PCDD or PCDF in utility electrical appliances that employ PCB. PCDF levels increased as the Aroclor number (and relative amount of chlorination) increased. This study was successful in developing a set of methodologies that reliably determine trace levels of PCDF and PCDD in a very difficult matrix -- chlorinated organic oil. The development of commercially available reference standards, and the statistical validation of reference methods that can be used to analyze utility fluids was a major contribution .to the quality and defensibility of data. The method verification program showed that a range of instrumentation can be successfully used to measure trace levels of PCDF and PCDD in utility fluids. In general, the methods employing low-resolution mass spectrometry (LRMS) require additional column chromatography to isolate PCDF/PCDD-enriched fractions for final HRGC/LRMS analysis. The precision of blind, spiked samples was superior to that expected by experienced analysts who had previously performed these analyses. The availability of labelled quantification standards helped improve method performance and lowered data scatter. genpou 7 1 783882 Section 1 INTRODUCTION The Electric Power Research Institute Initiated this program to Investigate the chemistry and analysis of polychlorinated dlbenzofurans (PCDF) and polychlorinated dibenzo-p-dioxins (PCOO) as they relate to the utility use of Askarel, a dielectric liquid which contains polychlorinated biphenyl (PCB), frequently diluted with tri/tetra chlorinated benzene. Askarel is sometimes contaminated with traces of PCDF/PCDD. The program had three goals; (1) to develop reliable, well characterized analytical methods to measure PCDF/PCDD, (2) to measure PCDF/PCDD levels in typical dielectric liquids used in utility operations, and (3) to investigate possible PCDF/PCDD production in unusual transformer events such as high energy arcing. This report describes analytical methodology and results of chemical analyses conducted by Battel!e Columbus Division in support of EPRI Research Project 20286, "Formation of PCOD and PCDF in Askarel and Contaminated Mineral Oil Equipment". Chemical laboratories of four additional organizations conducted parallel measurements during this program: the Illinois Institute of Technology Research' Institute, the New York State Department of Health, Radian Corporation, and the University of UMEA (Sweden). The data presented 1n this report were replicated by these four laboratories and separate reports are available from each of the study participants. Combined data from the replicate analyses conducted by the five collaborating laboratories (Including Battelle), are presented in a summary report prepared by the Research Triangle Institute. The summary report examines pooled data and gives a rigorous statistical analysis, including variability and reliability of analytical results reported by laboratories routinely conducting PCDF/PCDD analyses in Askarel, mineral oil, and related sample matrices. 783883 Reliable analytical methods are especially difficult to develop for Askarel since PCB constitutes the major interference in most PCDF/PCOD analytical methods. In Askarel, the normal PCB concentration is 60-70 percent by weight. Thus, Askarel presents perhaps the most difficult matrix for these analyses. Parts-per-billion (ppb) levels of PCDF, and even lower concentrations of PCDD must be measured. When this program was initiated in early 1984, the accuracy of PCDF analyses was hampered by a lack of accurate quantification standards. A major product of this 1-1 GENP 011072 study was the synthesis of a series of ^-labelled pcDF standards that allowed use of sensitive isotope dilution techniques to measure PCDF. Quantification and recovery measurements based on labelled standards permit the chemist to measure accurate mass spectral response ratios for naturally occurring PCDF (and PCDD) compared to the response factors of labelled standards spiked Into the sample matrix. The method development phase involved testi ng methods proposed by the five participating laboratories; Battelle, the Illinois Institute of Technology, Research Institute (IITRI), the New York State Department of Health (NYS), Radian Corporation, and the University of Umea. These five methods were challenged by analyzing PCDF/PCDD in baseline liquids, a series of five typical sample matrices: Aroclor 1260, Aroclor 1242, Aroclor 1016, chlorobenzene, and mineral oil containing approximately 50 ppm of Aroclor 1260. The baseline Askarel liquids contained native PCDF/PCDD, and the mineral oil was spiked with labelled PCDF. The presence of PCDF/PCDD in typical utility liquids was investigated by measuring characteristic compounds and congener groups 1n samples provided by EPRI member utilities. A report was generated by the General Electric Company which outlined the main types and distribution of PCB containing liquids'(1). A repository was established at Battelle where samples of transformer oils provided by several EPRI members were stored. A questionnaire was developed, and a data base was accumulated on the repository samples. Seven of these liquids were selected by the EPRI Technical Advisory Committee. The seven samples, along with three control liquids, were analyzed by the participating laboratories. The seven "in-service" liquids selected for analysis represented either high energy operation, or utility applications that typify PCB usage. The data collected In the in-service liquid study were analyzed by chemists and statisticians at the Research Triangle Institute (RTI). . A separate report will be generated by RTI that summarizes the results of all five laboratories. In summary, the Independent methods worked well, indi cati ng that good results can be expected when approprlate quali ty control procedures are applied. Arcing experiments were conducted by General Electric. Transformer test units were fabricated out of the normal materials of transformer construction. These test units were exposed to high electric energy such as might be expected in catastrophic transformer failure. In the engineering tests, final PC0F/PC00 levels were either the same or lowered from initial levels, indicating that short term GENP011073 1-2 783884 electrical discharges are not a primary source of production for PCDF/PC0 in transformer liquids. 783885 1-3 GENP 011074 Section 2 EXPERIMENTAL ACQUISITION OF IN-SERVICE SAMPLES The selection of in-service dielectric fluids for this study was coordinated by the EPRI Technical Advisory Committee with the assistance of Dr. Thomas Rouse of the General Electric Company. It was important to select a series of in-service liquids that represented a range of utility applications. In the early planning for this program, Or. Rouse prepared a criteria document that discussed the distribution of Askarel among different types of electrical devices, the relative chemical stresses expected from these devices in normal usage, the dispersion of Askarels, heating processes in transformers, and situations that represent the worst case for Askarel usage. He estimated that the United States has 140,000 Askarel-filled transformers in service, 30 million oil-filled transformers, and 510 million power capacitors. The oil-filled transformers constitute a large number; however, only about 50,000 are considered PCB-containing (>500 ppm (w/w)), and another approximately 1 million are PCB-contaminated (50-500 ppm). Since PCOF and PCDD are minor constituents of PCB, PCB-containing and PCB-contaminated transformers are expected to contain very low concentrations of PCDF/PCDO (perhaps not measurable). Dr. Rouse published an EPRI report (1) in which he recommended establishing a repository of 30-40 In-service samples that would be reviewed by the technical advisory group to select a final set of In-service liquids for exhaustive testing by the methods developed in this program. The samples recommended by Or. Rouse for inclusion in the final test group are given in Table 2-1. In order to acquire suitable samples for this study, several participating utility representatives were asked to survey their records on available samples, searching for dielectric materials that met the criteria established by Or. Rouse. To prescreen the available samples, a questionnaire was developed and a data base of questionnaire information maintained. A copy of the questionnaire is given 1n Appendix A. 783886 TABLE 2.1 EQUIPMENT RECOMMENDED FOR SAMPLING FOR PCDF ANO PCDO ANALYSES ASKAREL-FILLED Load Center/Network Transformers Rectifier Transformers Precipitator Transformers Capacitors 8-10 1-2 2 4 OIL-FILLED WITH PCB Load Center/Network Transformers 4 After gathering completed questionnaires on a large number of candidate fluids, a repository was established, that contained the samples listed in Table 2.1. As shown in Table 2.2, suitable dielectric liquids were collected from each of the categories of interest. A final selection was made using information in the repository data base, and the*materials listed in the last column of Table 2.2 were selected for inclusion in the in-service testing program. ANALYTICAL TECHNIQUES The following method description gives the procedures that were used by Battel le to analyze all Askarel and mineral oil samples in this program. The samples preparation procedures, the spiking protocols, the instrumental conditions, and the equations used to calculate final results are given here. The oil samples (five "baseline'' samples) were spiked with labelled interna! standards prior to distribution to the participating laboratories. The sample preparation and spiking procedures for the baseline liquids are given in a separate report in this series prepared by Radian Corporation (Austin, TX). SAMPLE PREPARATION AND SPIKING The seven in-service liquids were diluted to a concentration of approximately 2 pg/mL with hexane prior to spiking with internal standards. Aliquots of each liquid were weighed by difference into 100 mL volumetric flasks which were then filled to the mark with D1st111ed-1n-Glass hexane. The flasks were placed on a 2-2 GENP 011076 783887 GENP 011077 N u> -- __ __ . ,- -vl 00 CjO 00 00 00 TADIf 2.? SURVEY OF LIQUID SAMPLES AVAILABIE FOR THE IN-SERVICE TRANSFORMER LIQUIDS STUDY AUptpii ll iintcye Service Life IYears) CAPACITORS Capacitor 1 Capacitor 2 Capacitor 3 T 7 T TRANSFORMERS Arc Furnace 1 Arc Furnace 2 a B CIRCUIT BREAKER Circuit Breaker 30(Est.) LOAD CENTER NETWORK TRANSFORMERS Network Load Center 1 Network load Center 2 Network Load Center 3 Network Load Center 4 Network Load Center S Network Load Center 6 Network Load Center 7 Network Load Center B Network Load Center 9 Network Load Center 10 Network Load Center 11 Network Load Center 12 Network Load Center 13 Network Load Center 14 - - -- Network Load Center 15" Network Load Center 16 Network Load Center 17 Network Load Center IB Network load Center 19 Network Load Center 20 " 7 30(Est.) 21 28 20 19 20 19 31 30(Est.) 30(Est.j 20(E&t.) 15(Est.) lOEst.J - 20<Elt.j 20(Est.) 20 Est.) 10(Est.J 20(Est.) 7 PRECIPITATORS Precipitator 1 Precipitator 2 Precipitator 3 20(Est.) 20(2E1st.) Rating Dielectric 7200 KVAR 7200 KVAR ? ? Aroclor 1242 12470/280 KVA Mineral Oil 12470/280 KVA Mineral Oil 34.5 KV Mineral Oil 4S0 KVA 1500 KVA 1000 KVA 500 KVA 750 KVA 500 KVA 750 KVA 500 KVA 500 KVA 243 KVA 559 K W 313 KVA 339 KVA 126 KVA ~ 11 KVA 11 KVA 11.5 KVA 11.5 KVA 11.5 KVA ? ' Minera) 011 Aroclor 1242 Inerteen Aroclor 1260 Aroclor 1260 Aroclor 1260 Aroclor 1260 Aroclor 1260 Aroclor 1260 Pyranol Pyranol 7 Pyranol Pyranol Inerteen Chlorestol Inerteen Inerteen Chlorexdol Mineral 011 440/53.5 KVA 440/53.5 KVA 440/53.5 KVA 7 7 7 PCB (p p m ) Vo)une Field Code Laboratory Code __ i ISL-28 1SL-27 - - 7 ISI-01-A 161 4.500 ISL-24-0 ISO 4.500 1SL-21-0 4)159-11-15 403 360 ISL-29-0 248 ? ISL-05-0 __ -- 500 ISL-06-A -- - - 3780 (Ib) 1SL-07-A -- -- 3175 (Ib) ISL-Oa-A 41159-11-12 - - 2650 (Ib) 1SL-02-A 41159-11-17 -- ? ISL-09-A -- -- 240 1SL-10-A -- - - 3200 (Ib) ISL-1I-A -- - - ? ISL-03-A 41159-11-11 -- 195 1SL-12-A -- 370 1SL-13-A -- - - 195 1SL-14 -- -- 360 ISL-15-A -- -- 490__ _ 1SL-16-A - _-- -- - 235 ISL-18-A -- -- 272 ISL-19-A -- 282 1SL-20-A 184 1SL-21-A -- 277 ISL-22-A -- 10 200 ISL-04-0 41159-11-16 94 ISL-26 - - 94 ISL-2S - - 143 ISL-17-A 41159-11-14 mechanical shaker and mixed overnight. After approximately 12 hours of mixing, the flasks were removed from the shaker and spiked with the five isotopically labelled internal standards. The internal standards were: 2,3,7,8-TCDD-l3Ci2* 2,3,7,8- 2TCDF-13Ci , l,2,3,7,8-Penta-CDF-13Ci2, l,2,3,4,7,8-Hexa-CDF-13C12, and Octa-CDD- 13012. ANALYTE ENRICHMENT One-milliliter aliquots (containing *250 mg of the original fluid) of each sample were transferred to separatory funnels and diluted with 49 mL of hexane. The hexane solutions were washed with 10-mL portions of concentrated sulfuric acid until the washes were only slightly colored. A maximum of five acid washes were used. Following the acid washes, each oil was washed with 20 mL of distilled water to remove any residual acid. The hexane solutions were transferred to glass collection tubes and concentrated to approximately 2 mL using gentle streams of nitrogen gas. The hexane solutions were transferred to multilayered silica gel columns containing activated silica gel, 44 percent concentrated sulfuric acid on silica gel, and 33 percent 1H sodium hydroxide on silica gel. The purpose of these columns was to remove acidic and basic compounds from the solutions as well as materials that were easily oxidized. The silica gel support provided a large surface area for contact with the sample solutions thus improving the cleanup efficiency. The PCDD/PCDF isomers were eluted from the columns using 70 mL of hexane and the entire eluents, including the original sample volume, were collected. The eluates from the multilayered silica gel columns were concentrated to approximately 1 mL using gentle streams of nitrogen and transferred to the top of columns containing 5 g of activated basic alumina. These columns were eluted using hexane, hexane/methylene chloride (97:3, v/v) and hexane/methylene chloride (1:1, v/v) as elution solvents. The 1:1 hexane/methylene chloride eluates were collected, solvent exchanged into hexane, and transferred to the top of alumina columns containing approximately 2 g of activated basic alumina. These columns were eluted with hexane, hexane/methylene chiorlde (97:3, v/v), and hexane/ methylene chloride (1:1, v/v). The 1:1 hexane/methylene chloride eluates were collected, concentrated to near dryness with nitrogen gas, and spiked with 10 ng of l,2,3,4-TCDD-l3Ci2' These solutions were stored at 0C until they were analyzed. Several, of the sample solutions were still too contaminated with interferences to be analyzed. These solutions (41159-11-12,13,14,17 and E729-53-02.03) were / 2-4 GENP 011078 783889 a*' chromatographed through florisil columns to provide additional analyte enrichment. Each of the florisil columns was prepared using 5 g of activated florisil (60-100 mesh) which was slurry packed with hexane. The sample solutions were diluted with approximately 5 mL of hexane and* transferred to the florisil columns. The columns were eluted with hexane, hexane/ethyl ether (94:6, v/v), and hexane/methylene chloride (25:75, v/v). The 25:75 hexane/methylene chloride fraction was collected and concentrated to approximately 20 pL. These solutions were stored at 0C until they were analyzed. ANALYSIS The sample solutions were analyzed and quantified for PCOD/PCDF using combined capillary column gas chromatography/high resolution mass spectrometry (HRGC/HRMS). The HRGC/HRMS consisted of a Carlo Erba Model 4160 Gas Chromatograph interfaced directly into the ion source of a VG Model 7070 high resolution mass spectrometer. The chromatographic column was a 60 M DB-5 fused silica column using helium carrier gas at a flow velocity of 30 cm/sec. The mass spectrometer was operated in the electron impact (El) ionization mode at a mass resolution of 9000-12000 (M/ M, 10 percent valley definition). All HRGC/HRMS data were acquired by multiple-iondetection (MID) using a VG Model 2035 Data System. QUALITY ASSURANCE The operation of the HRGC/HRMS was evaluated each day by analyzing standard mixtures of selected PCDD/PCDF compounds. These were used to define the chromatographic elution windows, the compound resolution of the chromatographic column, and to verify the quantitative accuracy of the analyses. The mass focus accuracy of the MID unit was evaluated at least every four hours by observing selected ion masses from perfluorokerosene (PFK) which was introduced continually into the ion source as a reference compound. Mass focus stability was assured by using a reference PFK "lock mass" to correct for any mass focus drift. Native spike and laboratory method blank samples were also processed during the extraction and cleanup of the samples. The native spike samples were used to evaluate the accuracy of the quantification while the laboratory method blank sample was used to demonstrate freedom from laboratory contamination. RECOVERY OF INTERNAL STANDARDS The recoveries of the internal standards were calculated by comparison to an external standard, 1,2,3,4-TCDD-^C|2 which was added following the column cleanup steps. Relative response factors were determined from triplicate analyses of a OETSTP011079 2-5 783890 standard mixture containing the labelled internal standards and the 1,2,3,4-TCDD- 2^ C i * The equation used to calculate the recoveries was: Recovery {%) Ais x Ors x 100 Ars x Qis x Rf Where: Ais - Sum of integrated areas for internal standard; Qrs - Quantity of recovery standard in ng; Qis - Quantity of internal standard in ng; Ars - Sum of integrated areas for recovery standard; and Rf - Response factor. QUANTIFICATION The PCDO/PCOF isomers and congener class concentrations were determined by comparing the sum of the two ion masses monitored for each class to the sum of the two ion masses monitored for the corresponding internal standard. It was assumed that the response factors for each of the individual isomers in a class were the same as the response factors calculated for the specific isomers present in the standard solutions. The response factors were calculated from six replicate analyses of standard solutions at four concentrations. The sum of the two most intense ion masses in the molecular ion region of eachisomer were used to generate the response factors. The formula used for quantifying the PCOD/PCDF isomers was: Quantlty/Sample (ng/g) Ai^ S'x*Rf1x m Where: Quantity - Total quantity in ng of target isomer or congener class Ac - Sum of Integrated areas for the target isomer or congener class Q1s - Quantity of internal standard 1n ng Ais * Total integrated areas for the internal standard Rf - Response factor. 783891 Each pair of resolved peaks in the selected-ion-current chromatograms was evaluated manually to determine if It met the criteria for a PCDD or PCDF isomer. By examining each pair of peaks separately, quantitative accuracy was improved over what 1$ obtained when all of the peaks in a selected chromatographic window are averaged. When averaged data are used, It 1s possible for pairs of peaks with high and low chlorine Isotope ratios to produce averaged data that meet the chlorine isotope ratio criterion. For example, two pairs of peaks having chlorine Isotope ratios of Q.56 and 0.96, both outside, of the acceptable range, would have an 2-6 GENP 011080 average ratio of 0.76. By checking the isotope ratio of these peaks separately, they would be eliminated from the class total. The criteria that were used to identify PCDD and PCDF isomers were: 1. Simultaneous responses at both ion masses. 2. Chlorine isotope ratio within +/-1S5S of the theoretical value. 3. Chromatographic retention times within windows determined from analyses of standard mixtures. 4. Signal-to-noise ratio equal to or greater than 2.5 to 1.0. The individual isomers for which an isotopically labelled analog was available included the additional criterion that they eluted within 2 seconds of the internal standard. The limit of detection (LOD) was calculated for samples in which isomers of a particular chlorine congener class were not detected. The formula used for calculating the LOD was: LOD/g (ngj Hc_x Qis x 2.5 His x Rf x W Where: LOD Single isomer limits of detection for a congener class in ng; He * Height of congener class isomer; Qis * Quantity of internal standard in ng; His - Peak height of internal standard; Rf * Response factor; and W - Weight in grams of sample. CALIBRATION CURVE Calibration curves were generated for each of the native/isotopically labelled isomer pairs. The calibration standards were analyzed at least five times at each of four concentration levels between 0.04 and 2.5 parts-per-million (ppm). The response factors were calculated by dividing the sum of the areas for the two most abundant ion masses 1n the molecular 1on cluster of each native PCDD/PCDF Isomer by the sum of the areas for the corresponding ion masses from the isotopically labelled standard. The Isomer pairs for which, response factors were calculated included: GENP 011081 2-7 783892 1 2.3.7.8- TCOD vs 2,3,7,8-TC0D-}3Ci2 2.3.7.8- TCDF vs 2,3,7,8-TCDF-13Ci2 OCOO vs 0C0D-}3Ci, OCOF vs 0C00-13ci2 1.2.3.7.8- Penta-COF vs l,2,3,7,8-Penta-CDF-}3Ci2 2.3.4.7.8- Penta-CDF vs l,2,3,7,8-Penta-CDF-13Ci2 1.2.3.4.7.8- Hexa-COF vs 1,2,3,4,7,8-Hexa-CDF-{31C2 2.3.4.6.7.8- Hexa-CDF vs l,2,3,4,7.8-Hexa-CDF-131C2 1.2.3.4.6.7.8- Hepta-COF vs OCOO-13C i2 It was assumed that each of the Isomers of a particular chlorine congener class had the same response factor. DETECTION LIMITS Detection limits were calculated for samples that did not contain PCDD/PCDF isomers in a particular class. These detection limits were calculated using the quantification equation with the addition of a factor of 2.5 to account for the signal-to-noise criterion. The noise level was measured as the average of the peak to peak noise that occurred in the baseline of the particular MID channel. If at all possible, the noise was measured at the retention time of the isotopically labelled internal standard. The height of the internal standard peak was measured to the average of the noise that occurred on the peak top. The reported limits of detection are for single PCDD/PCDF isomers and have been corrected for recovery losses during sample workup. BASELINE ANALYSES The three baseline samples, Aroclor 1260, Aroclor 1254, and Aroclor 1016, were analyzed using the methodology described above. The Aroclor 1016 presented the greatest analytical problem due to the relatively high amount of the sample that passed through the column cleanup steps. Even after the four column cleanup steps, a noticeable PCB background was evident 1n the single ion current chromatograms. IN-SERVICE ANALYSES The seven in-service samples were also analyzed using the methodology described above. 783893 2-8 GENP 011082 Section 3 BASELINE ANALYSES This section describes the baseline analyses performed by all five laboratories participating in this study. The baseline analyses were designed as a rigorous blind analysis of authentic Aroclorand mineral oil samples. If the proposed analytical methods performed well inthe baseline tests, the additional steps of method validation* and screening of in-service utility fluids could be performed with confidence in the reported results. Consequently, the baseline analyses were critical to the success of this program. BACKGROUND RESEARCH Before the baseline analyses could be performed, five test liquids were obtained by Radian Corporation and a combination of native and labelled (*3C) PCDF/PCDD were spiked into the test liquids to ensure a rugged test of the proposed analytical techniques. The test liquids selected for the baseline study are listed in Table 3.1. 01 electric Fluid Aroclor 1016 Aroclor 1242 Aroclor 1260 Chlorobenzenes Aged Mineral Oil TABLE 3.1 BASELINE TEST LIQUIDS Composition Narrow boiling point fraction of Aroclor 1242. 42% Cl (w/w) 60% Cl (w/w) -Synthetic Mixture 1.2.4- Trichlorobenzene (51.2% (w/w)) 1,2,3-Trichlorobenzene (13.4% (w/w)) 1,2,4,5-Tetrachlorobenzene ( 2.6% (w/w)) 1.2.3.4- Tetrachlorobenzene (28.4% (w/w)) Pentachlorobenzene ( 4.4% (w/w)) Spiked Aroclor 1260 (500 ppm (w/w)} Tri/Tetrachlorobenzene Mix Listed Above (500 ppm (w/w)) G E N P 011083 3-1 783894 RESULTS Chemical Analyses The results of chemical analyses performed by Battelle on the baseline samples are given in Table 3.3. All results are given as parts-per-billion by weight in the original oil (ng/g). These data are interesting because they show the expected trends in the Aroclor PCOF/PCDO concentrations. PCDD are not found at the detection limit of the high resolution mass spectrometer used for this analysis, approximately 10 ppb (w/w) in the mineral oil samples. Trace levels of PCDF are observed with the concen trations increasing as the level of chlorination increases in the supporting Aroclor. The PCDF levels increase as the Aroclor number (and relative amount of chlorination increase), Aroclor 1260 >Aroclor 1242 >Aroclor 1016. Even though Aroclor 1016 is similar in composition to Aroclor 1242, the wider distribution of PCB congener groups in Aroclor 1242 would produce higher quantities of the high chlorination number PCB, 6-10 chlorine atoms per molecule, which are believed to be the precursors of PCDF when HC1 elimination is the route for PCOF formation. An increase in PCDF concentrations is shown 1n Table 3.3 as the total high chlorine PCB concentration increases. Total PCDF levels increase from 280 ppb for Aroclor 1016, to 1,710 ppb for Aroclor 1242, to 10.4 ppm for Aroclor 1260. Data for three of the participating laboratories are given in Tables 3.4 through 3.6. These data show the same trends observed in the Battelle data. The measured levels of spiked PCDF/PCDF were also similar among the reporting laboratories. 8ase1ine Recovery Data The recovery of labelled standards 1s used to determine how much of the native PCOF/PCDD present in the sample is lost during the sample preparation steps of candidate methods. Since the Internal quantification standard corrects for method losses, the reported value should be accurate within the measured precision of the method; however, excessive loss of analytes during sample preparation will have a profound affect on both method sensitivity, and precision. Recoveries for six labelled PCOF/PCDO standards spiked into the baseline samples are given in Table 3.7. These data are interesting because they show the outliers that can affect individual data points. 783895 3-4 o o o00 L/t to un 0o0o 0CO0D TABLE 3.3 BATTELLE FURAN AND DIOXIN DATA FOR THE BASELINE ANALYSES STUDY (ppb (w/w)) Compound Aroclor Aroclor Aroclor Tri and Tetra 1016 1242 1260 Chlorobenzene Mineral Oil Run No. 1 Mineral Oil Run No. 2 Polychlorinated Dibenzofurans Total TCDF Total PnCOF Total HxCDF Total HpCDF OCDF Total PCDF 0 5 190 85 NR* 280 990 400 84 184 52 1710 640 2100 3100 2100 2500 10440 0 2 82 100 . NR 184 119 115 270 228 278 246 128 113 152 157 947 859 Polychlorinated Dibenzodioxins Total TCDD OCOD 0 00 NR NR NR 0 NR 32 34 NR NR Specific Molecules and Unresolved Molecular Clusters 2,3,7,8-TCDD 2t3,7,8-/2,3,4,8-TC0F 1.2,3,7,8-/l,2,3,4,8-PnCDF 2,3,4,7,8-PnCDF 1,2,3,4,7,8-HxOF 1,2,3,7,8,9-HxCDF 2,3,4,6,7,8-HxCDF 1,2,3,4,6,7,8-HpCDF 0 00 0 84 150 0 54 480 0 00 190 84 1700 NR NR NR NR NR NR 85 147 680 0 0 0 0 82 NR NR 100 32 34 119 115 170 143 98 95 246 262 NR NR NR NR 113 120 * NR=Not Reported. / TABLE 3.4 11TRI FURAN AND DIOXIN DATA FOR THE BASELINE ANALYSES STUDY (ppb (w/w)) Compound Aroclor 1016 Aroclor 1242 Aroclor 1260 Tri and Tetra Chlorobenzene Mineral Oil Aqed Total TCDF 30 866 612 0 70 Total PnCOF 42 353 1037 0 165 Total HxCDF 92 105 1433 103 207 Total HpCDF 81 109 1985 98 40 OCOF 20 127 4154 0 393 Total TCDO 0 00 0 42 OCOO 0 01 1 0 2,3,7,8-TCDD 2,3,7,8-/2,3,4,8 TCDF 0 8 2 2 07 ' 0 193 0 0 42 70 1,2,3,7,8 /1,2,3,4,8-PnCDF 11 42 231 0 112 2,3,4,7,8-PnCDF 0 51 88 0 52 OCITJi 1 ,2,3,4,7,8-HxCDF 1,2,3,7,8,9-MxCDF 2,3,4,6,7,8 HxCDF 76 63 601 0 00 0 00 103 0 0 206 0 0 1,2,3,4,6,7,8 HpCDF 58 77 684 98 68 4* GENP 011086 ~0C0CvO00Di TABLE 3.5 NYSDOH FURAN AND DIOXIN DATA FOR THE BASELINE ANALYSES STUDY (ppb (w/w)) ,80UoaKao Comoound Total TCDF Total PnCDF Total HxCDF Total HpCDF OCDF Total TCDD CDD 2,3,7,8-TCDD 2,3,7.B-/2,3,4,8-TCOF 1,2,3,7,8/1,2,3,4,8-PnCDF 2,3,4,7,8-PnCDF 1,2,3,4,7,8-llxCDF 1,2,3,7,8,9-HxCOF 2,3,4,6,7,8-llxCDF 1,2,3,4,6,7,8-HpCDF Aroclor 1016 I 7 350 92 0 0 0 0 0 0 0 0 0 0 0 Aroclor 1242 1900 584 140 130 2300 0 0 0 150 1000 0 0 0 0 0 Aroclor 1260 940 996 5500 2430 NA 0 0 7 190 220 0 0 0 0 0 Tri and Tetra Chlorobenzene 3 13 1000 91 0 0 0 0 0 1 0 0 0 0 0 Mineral Aaed 95 0 1100 0 0 42 0 43 34 140 0 0 0 0 0 GENP 011088 TABLE 3.6 RADIAN FURAN AND DIOXIN DATA FOR THE BASELINE ANALYSES STUDY (ppb (w/w)) Com DO und Total TCDF Total PnCDF Total HxCDF Total HpCDF OCDF Total TCDD COD 2,3,7,8-TCDD 2,3,7,8-/2,3,4,8-TCDF l,2,3,7,8-/l,2,3,4,B-PnCDF 2,3,4,7,8-PnCOF 1,2,3,4,7,8-HxCDF 1,2,3,7,8,9-HxCDF 2,3,4,6,7,8-HxCDF 1,2,3,4,6,7,8-llpCDF Aroclor 1016 0 0 97 80 0 0 0 0 0 0 0 97 0 0 60 Aroclor 1242 0 0 60 46 24 0 0 0 0 0 0 60 0 0 46 Aroclor 1260 0 477 872 807 1786 0 0 0 477 0 872 0 0 807 Tri and Tetra Chlorobenzene 0 0 161 170 0 0 386 0 0 0 0 161 0 0 170 Mineral Oil Aaed 125 226 178 36 126 0 0 0 125 159 66 177 0 0 36 o00o00 CO CO j GENP 011089 2,3,7,8-TCDF-13Ci2 I,2,3,7,8-PCDF-13C i2 2 >3,7(B-TCDD-^Cj2 0CDD-13Ci2 1,2,3,4,7,8-HxC0F-,3C )2 1,2,3,4,6,7,8-HpCDF *O u t)ie rs TABLE 3.7 ABSOLUTE RECOVER1ES OF LABELLED STANDARDS IN THE BASELINE STUOY (UNIVERSITY OF UMEA) Spike Level iBRbl Aroclor Aroclor Aroclor Chloro Minerai 1016 1242 1260 benzenes on 100 93 95 95 97 116 100 111 134 140 97 114 100 67 86 86 101 118 100 102 89 157 86 121 100 83 90 317 110 NA 100 57 45 1224* 68 NA o0o0 ocoo <3 One effect is evident in these data, the recoveries from chlorobenzene and mineral oil appear to be superior to recoveries from Aroclors. The Aroclors contain the major interferents in the analysis of PCDF and PCDD. It appears that there may be matrix dependent difficulties in the analytical preparation of complex samples. Method Reliability The baseline data are grouped to show the relative performance of the four laboratories reported here. A compilation of data for aged mineral oil, baseline sample results is given in Table 3.8. m Method reliability is compared by computing the Student's-t variability for four independent determinations around the mean value, the known spiked amount of standard PCOF/PCDD. In the four determinations shown the Student's-t variability of the mean varies from 20 to 167 percent. This is the degree of expected scatter in a group of four PCDD/PCDF determinations conducted by independent laboratories using the Student'st distribution to approximate variance at the 95 percent confidence level. 783901 GETTP 011090 GENP 011091 TABLE 3.8FOUR LABORATORY RESULTS OF ANALYSIS OF AGED MINERAL OIL (ng/g (w/w)) Compound Tota) 1CDF Total PnCDF Total HxCDF Total HpCDF OCOF Total TCOO 2,3,7,-TC0D 2,3.7,8/2,3,4,8-TCDF 1,2,3,7,8/1,2,3,4,8-PnCOF 2,3,4,7,8-PnCDF 1,2,3,4,7,8-HxCDF 1,2,3,4,6,7,8-HpCDF Battelle 115 228 246 113 157 34 34 115 143 95 262 120 1TTRI 70 165 207 40 393 42 42 70 112 52 206 68 NYSDOH 95 0 1100 0 0 42 43 34 140 0 0 0 Radian 125 226 178 36 126 0 0 125 159 66 177 36 Group Mean 101 155 433 47 169 30 30 86 139 53 161 56 Native Spike Level Confidence Interval i95%) 200 109% 160 167% 100 20% 100 41% 00 oCO CO INJ Section 4 SUPPLEMENTAL ANALYSES CROSS CHECK ANALYSES To validate the analytical curve for each method proposed by the participating laboratories, a series of calibration runs were performed to establish the variability, linear dynamic range, and the limit of detection for each instrument. The calibration range is different from the method range. The validation of the calibration curve range only deals with standard solutions eluted through a tuned gas chromatograph or mass spectrometer (GC/MS) system. The quantities used to develop the calibration curve were 40 pg, 150 pg, 500 pg, and 2.5 ng. Each concentration level was analyzed with five replicates. The analytes used for the calibration curve validation were 2,3,7,8-TCOD, OCDO, 2,3,7,8-TCDF, 1,2,3,7,8-PCDF, 2,3,4,7,8-PCDF, 1,2,3,4,7,8-HxCOF, 2,3,4,6,7,8-HxCOF, and 1,2,3,4,5,7,8-HpCOF. An illustration of the type of calibration plot generated in this study is given in Figure 1, which shows the standard curve for 2,3,7,8-TCDF. The complete set of calibration curve plots, along with statistical parameters describing calibration precision are given in Appendix B. The standard deviation and the relative standard deviation at each point show the variability of response which increases at higher concentrations. The calibration curves were linear over the range of concentrations measured. ANALYSES FROM THE UNIVERSITY OF UMEA Under a subcontract to Battelle, the University of Umea participated in the baseline and round robin analyses. A letter report of the data from Umea is included as Appendix C. 783903 4-1 GEMP 011092 GENP'011093 -vj C0O0 Figure 4-1. Calibration curve for 2,3,7,6-tetrachlorodibenzofuran over the range 40 pg to 2.5 ng oCO Section 5 ROUND ROBIN ANALYSES OF IN-SERVICE LIQUIDS The main thrust of this study was to evaluate a series of congeners and Individual molecular species in typical in-service utility transformer and capacitor liquids. Ten (10) liquid samples, including the seven utility dielectric fluids, were analyzed for PCDD/PCDF using the methods developed in this program. All five participating laboratories contributed independent data sets, which will be statistically analyzed by an independent contractor, the Research Triangle Institute. The results of the Battelle analyses are given in Table 5.1 to 5.10. Accuracy Estimate The in-service analyses were designed with a quality control check on accuracy of reported PCDD/PCDF concentrations. The "Baseline" liquids included with actual utility operating fluids were spiked with known amounts of several PCDD/PCDF standards. These known spike concentrations allowed an estimate of accuracy using the mean of three determinations performed by each participating laboratory compared with the "true" or spiked concentration. The "Baseline" analyses revealed that the aged mineral oil sample and the Aroclor 1016 had minimal concentrations of the PCDD/PCDF standards present as contaminants. Thus, native PCDD/PCDF were suitable spike materials. The Battelle results for in-service liquids are presented in Table 5.11 for the mineral oil and Aroclor 1016 samples. Table 5.11 gives the spike levels and compares the percent difference in spiked levels with measured levels of PCD0/PC0F. The Battelle results are generally within 70 percent of the spike value except for one notable exception. HxCDF exhibited a strong positive bias. Most of the elevated HxCDF levels were 2,3,4,6,7,8-HxCDF. The HxCDF measured levels were 438 percent higher than the HxCDF spike levels 1n mineral oil, and 1325 percent higher in Aroclor 1016, even after subtraction of the native HxCDF found in the earlier baseline investigations. These HxCDF data are deemed to be outliers. 783905 5-1 GENP 011094 TABLE 5.1 CHLORINATED FURAN AND OIOXIN RESULTS (ppb (w/w), ng/g) SAMPLE ISL-02-A Conoaner Determination Det 1 Det 2 Det 3 Mean Relative Standard Standard Deviation Deviation PCDF Conaeners Total TCDF Total PnCDF Total HXCDF Total HpCDF OCDF Total PCDF Furan Comoounds 360 69 8,100 1,600 4,300 230 72 19,000 1,400 4,800 290 71 20,000 2,000 5,100 293 71 15,700 1,667 4,733 14,429 25,502 27,461 22,464 53 1 5,389 249 330 5,738 18% 2% 345 15* 7* 26* 2,3,7,8-/2,3,4,8-TC0F 2,3,4,7,8-PnCDF 1,2,3,7 ,8-/1,2,3,4,8-PnCDF 1 2,3,4,7,8-/1,2,3,4,7,9-H^CDF 1,2,3,6,7,8-HXCDF 2,3,4,6,7,8-Hx CDF 1,2,3,4,6,7,8-HpCOF 1,2,3,4,7,8,9-HpCDF - 190 300 69 2,700 760 1,100 180 280 Dioxin Conaeners 140 260 72 6,400 2,300 3,400 230 250 180 290 71 6,400 2,700 3,800 320 320 170 233 71 5,167 1,920 2,767 243 283% 22 17 1 1,744 836 1,190 58 29 13* 6* 2* 34* 44* 43* 24% 10* Total TCDD OCDD PCDD Compounds NA NA NA NA NA NA 18 18 10 15 4 25* 2,3,7,8-TCDQ NA NA NA NA NA NA O E U P O '1095 5-2 783906 TABLE 5.2 CHLORINATED FURAN AND DIOXIN RESULTS (ppb (w/w), ng/g) SAMPLE ISL-03-A Conaener PCDF Congeners Determination Det 1 Oet 2 Det 3 Mean Relative Standard Standard Deviation Deviation Total TCDF Total PnCDF Total HxCOF Total H0CDF OCDF Total PCDF Furan Comoounds 170 2,300 52,000 3,800 22,000 110 2,100 11,000 13,000 28,000 40 720 9,700 11,000 22,000 170 1,707 24,233 9,267 24,000 80,270 54,210 43,460 59,313 53 702 19,641 3,951 2,828 15.455 50% 41% 81% 43% 12% 26% 2,3,7,3-/2,3,4,8-TCDF 31 2.3,4,7,8-PnCDF 1,2 3,7,8-/1,2,3,4,8-PnCDF 130 35 1,2,3,4,7,8-/1,2,3,4,7,9-H*C0F 29,000 1 , 3 ,6,7,8-HxCDF 6,000 2 *3,4,6,7,8-HxCDF 1,800 1.2,3,4,6,7,8-HoCDF 1,100 1 2,3,4,7,8,9-HpCDF 1,200 Dioxin Conoeners 30 130 49 6,100 1,100 240 4,500 2,600 40 110 26 5,100 1,200 210 3,100 2,600 34 123 37 13,400 2,767 750 2,900 2,133 % 4 9 9 11,038 2,287 743 1,395 660 13% 8% 26% 82% 83% 99% 48% 31% Total TCDD OCDD PCDD Comoounds NA NA NA NA NA 18 12 NA 10 . 7 NA 75% 2,3,7,8-TCDD NA NA NA NA NA NA 783907 OBtiP 0U096 TABLE 5.3 CHLORINATED FURAN AND DIOXIN RESULTS (ppb (w/w), ng/g) SAMPLE ISL-04-0 Congener PCOF Congeners Determination Pet 1 Pet 2 Pet 3 Total TCDF Total PnCDF Total HXCDF Total HpCDF OCDF Total PCDF Furan ComDOunds NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 2,3,7,8-/2,3,4,8-TCDF NA NA NA 2 13 ,4,7,8-PnCDF NA NA NA l,2,3t7,8-/l,2.3.4,8-PnCDF NA NA NA 1 ,2,3,4,7,8-/! ,2,3,4,7,9-HxC0F NA NA NA 1,2,3,6,7,8-HxCDF NA NA NA 2,3,4,6,7,8-HxCDF NA NA NA 1,2,3,4,6,7,8-H0CDF NA NA NA 1,2,3,4,7,8,9-HpCDF NA NA NA Dioxin Congeners Total TCDD OCPO PCDD Comoounds NA NA NA NA 13 11 2,3,7,3-TCDD NA NA NA Mean NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 8 NA Relative Standard Standard Deviation Deviation NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS 6 71S NA NAS GENP 011097 5-4 783908 I TABLE 5.4 CHLORINATED FURAN AND DIOXIN RESULTS (ppb (w/w), ng/g) SAMPLE ISL-08-A Congener Determination Pet 1 Pet 2 Pet 3 Mean Relative Standard Standard Deviation Deviation PCOF Conqeners Total TCDF Total PnCDF Total HXCDF Total HnCDF OCDF Total PCOF Furan Comoounds 62 79 1,500 1,500 4,000 7,141 64 230 1,300 2,100 4,800 8,494 99 310 990 2,700 4,900 8,999 75 206 1,263 2,100 4,567 8,211 17 96 210 490 403 784 23% 46% 17% 23% 9% 10% 2,3,7,8-/2,3,4,8-TCDF 20 20 30 23 2,3,4,7,8-PnCDF 39 41 37 39 1,2,3,7,8-/1,2,3,4,8-PnCDF 22 25 34 27 1,2,3,4,7,8-/1,2,3,4,7,9-H.COF 880 740 640 753 1 ,2,3,6,7 ,8-HxCDF 54 50 36 47 2,3,4,6,7,8-HxCDF 47 25 43 38 1,2,3,4,6,7,8-HoCDF 350 530 750 543 1,2,3,4,7,8,9-HpCOF 340 450 580 457 Dioxin Conqeners 5 2 5 98 8 10 164 98 20% 4% 19% 13% 17% 25% 30% 21% Total TCDD OCDD PCDD Comoounds NA NA NA NA NA NA NA NA NA NA% NA NA% 2,3,7,8-TCOO NA NA 1NA NA NA NA% 783909 G ENPoj 1098 TABLE 5.5 CHLORINATED FURAN AND DIOXIN RESULTS (ppb (w/w), ng/g) SAMPLE ISL-17-A Congener Determination Det I Det 2 Det 3 Mean Relative Standard Standard Deviation Deviati' PCDF Congeners Total TCDF Total PnCDF Total HXCDF Total HpCDF OCDF Total PCDF Furan Compounds 970 11,000 14,000 3,100 150 580 6,700 36,000 2,800 170 610 6,500 16,000 4,200 220 720 8,067 22,000 3,367 180 29,220 46,250 27,530 34,333 177 2,076 9,933 602 29 8,455 25% 26% 45% 18% 16% 25% 2.3.7.8- /2 *3,4,8-TCDF 2 3,4,7 ,8-P-CDF 1 .2.3.7.8- /1,2,3,4,8-PnCDF 1 213 4,78-/12 3 47 9-HvCDF 12 ,3,6,7 ,8-HxCDF 2.3.4.6.7.8- HxCDF 1.2.3.4.6.7.8- HpCDF 1.2.3.4.7.8.9- HpCDF \ Dioxin Congeners 400 2,300 870 4,700 1,800 620 920 740 380 2,400 900 7,200 3,200 6,800 560 860 350 1,800 780 4,100 2,300 930 1,300 990 377 2,167 850 5,333 2,433 2,783 927 863 21 262 51 1,342 579 2,832 302 102 5% 12% 6% 25% 24% 102% 33% 12% Total TCDD OCDD PCDD Compounds NA NA NA NA NA 18 NA NA NA NA% 8 141% 2,3,7,8-TCDD NA NA NA NA NA NA% G E N P 0 I 1099 5-6 783910 TABLE 5.6 CHLORINATED FURAN AND DIOXIN RESULTS (ppb (w/w), ng/g) SAMPLE ISL-23-0 Conaener Determination Det 1 Det 2 Det 3 PCDF Congeners Total TCDF Total PnCDF Total HXCDF Total HpCDF OCDF NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA Total PCDF NA NA NA Furari Comoounds 2,3,7,8-/2 3,4,8-TCDF 2,3,4,7 f8-PnCDF 1,2.3,7.8-/1.2,3,4,8-P,,CDF 1,2,3>4,7I8-/1I2 13,4,7,9-HXCDF 1,2,3,6,7,8-HxCDF 2,3,4,6,7,8-HxCDF 1,2,3,4,6,7,8-H0CDF 1,2,3,4,7,8,9-HpCDF Dioxin Congeners Total TCDD OCDD PCDD Comoounds NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 12 NA NA NA NA NA NA NA NA NA NA 2,3,7,8-TCDD NA NA NA Mean NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA 4 NA Relative Standard Standard Deviation Deviation NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS NA NAS 6 141% NA NAS 783911 5-7 GENP 011100 TABLE 5.7 CHLORINATED FURAN AND DIOXIN RESULTS (ppb (w/w), ng/g) SAMPLE ISL-27 Conoener Determination Det 1 Det 2 Det 3 PCDF Conoeners Total TCDF Total PnC0F Total HXCDF Total Hq CDF OCDF H 580 570 350 250 270 220 140 130 160 26 28 6 46 44 10 Total PCDF 1,042 1,042 746 Furan Compounds 2,3,718-/2,3,4,8-TCDF 300 280 270 2 ,3 ,4 7 ,8-PnCDF 1,2,3,7,8-/1,2 *3,4,8-PnCDF 71 74 53 580 570 350 1.2,3l4 17 i8-/l,2,3,4>7 I9-H,eCDF 37 16 31 1.2,3,6,7,8-HxCDF 15 15 27 2 3,4,6,7 ,8-HxCDF 25 17 39 1.2,3,4,6,7,8-HoCOF 1,2,3,4,7,8,9-HpCDF 792 551 Dioxin Conqeners Total TCDO OCDO NA NA NA NA NA NA PCDD Compounds 2,3,7,8-TCDD NA NA NA Mean Relative Standard Standard Deviation Deviati 500 106 21% 247 21 . 8% 143 21 20 10 33 17 9% 51% 50% 943 140 ' 15% 283 12 66 9 500 106 28 9 19 6 27 9 63 42 4% 14% 21% 32% 30% 34% 50% 45% NA NA NA NA NA% NA% NA NA NA% GENPOlUOl 783912 TABLE 5.8 CHLORINATED FURAN AND DIOXIN RESULTS (ppb (w/w), ng/g) SAMPLE EZ29-53-01 (Spiked Mineral Oil) Conaener Determination Det 1 Det 2 Det 3 Mean Relative Standard Standard Deviation Deviation PCDF Conaeners Total TCOF Total PnCDF Total HXCDF Total HnCDF OCDF 220 160 310 550 680 6,600 530 33 NA NA 190 190 260 373 390 2,557 220 261 31 24 127 2,862 205 1 13% 34% 112% 79% 141% Total PCDF 1,740 7,343 1,063 3,382 2,815 83% Furan Compounds 2,3,7,8-/2,3,4,8-TCDF 2,3,4,7 ,8-PnCDF I ,2,3,7,8-/1,2,3,4,8-PnCDF 1,2,3,4,7,8-/1,2,3,4,7,9-HxCDF 1,2,3,6,7,8-HxCDF 2,3,4,6,7,8-HxCDF 1.2,3,4,6,7,8-HnCDF 1,2,3,4,7,8,9-HpCOF 220 160 170 370 140 170 310 550 140 120 200 5,900 530 33 NA NA 190 190 120 220 140 150 260 373 76 112 140 2,080 220 261 NA NA 24 108 14 127 27 2,701 205 NA 13% 49% 9% 34% 24% 130% 79% NA% Dioxin Congeners Total TCDD OCDD 19 19 23 20 NA NA 1 NA 2 9% NA 141% PCDD Compounds 2,3,7,8-TCDD 19 19 23 20 2 9% 783913 GETSn? 01U02 TABLE 5.9 CHLORINATED FURAN AND DIOXIN RESULTS (ppb (w/w), ng/g) SAMPLE E729-53-02 (Spiked Aroclor 1016) Conaener Determination Det 1 Det 2 Get 3 PCDF Conoeners Total TCDF Total PnCDF Total HxC0F Total HnCDF OCQF K 57 600 5,400 65 110 67 500 2,300 150 120 66 460 780 340 120 Total PCDF 6,232 3,137 1,766 Furan Comoounds 2,3,7 ,8-/2 3 4,8-TCDF 57 2,3 ,4,7 ,8-PnCDF 410 l,2,3,7,8-/l,2,3,4,8-PnCDF 180 1 ,2,3,4,7,8-/1,2,3,4,7,9-HxCDF NA 1 ,2,3,6,7,8-HxCDF 100 2,3,4,6,7,8-HxCDF 5,200 1,2,3,4,6,7,8-HDC0F 65 1,2,3,4,7,8,9-HpCOF NA 67 310 180 NA 84 2,100 150 NA 66 260 190 NA 81 680 180 NA Dioxin Conoeners Total TCDO OCDD 46 46 44 NA NA NA PCDD Compounds 2,3,7,3-TCDO 46 46 45 Mean Relative Standard Standard Deviation Deviation 63 520 2,827 185 117 3,712 4 59 1,923 115 5 1,868 oLO n III 68% 622 4 63 327 183 NA 88 2,660 132 NA 4 62 5 NA 8 1,887 49 NA 7% 192 32 NA2 7792% 37% NA2 45 1 22 NA NA NA2 46 NA 12 GENP 011103 5-10 783914 TABLE 5.10 CHLORINATED FURAN AND DIOXIN RESULTS (ppb (w/w), ng/g) SAMPLE EZ29-53-03 (Spiked Aroclor 1260) Conaener PCDF Conaeners Determination Det 1 Det 2 Det 3 Mean Relative Standard Standard Deviation Deviation Total TCDF Total PnCDF Total HXCDF Total HnCDF OCDF Total PCDF Furan Comoounds 1,400 1,100 20,000 1,700 5,000 1,700 1,300 4,700 3,000 4,900 970 1,200 11,000 660 2,600 1,357 1,200 11,900 1,787 4,167 29,200 15,600 16,430 20,410 300 82 6,279 957 1,109 6,225 22% 72 532 542 272 302 2,3.7,8-/2,3,4,8-TCDF 1,200 2 3,4,7,8-PnCDF 1,2,3,7,8-/1,2,3,4,8-PnCDF 590 1.400 1,2,3,4,7,8-/1,2,3,4,7,9-HYCDF 1,900 1,2,3,6,7,8-HxCDF 1,200 2.3,4,6,7,8-HxC0F 10,000 1,2.3,4,6,7,8-HoC0F 1,2,3,4,7,8,9-HpCDF 55 300 Dioxin Conoeners 1,300 410 1,170 1,800 93 770 850 360 880 800 970 1,600 780 6,500 49 180 1,127 627 1,357 1,767 691 5,757 318 280 179 194 300 125 456 3,805 376 75 162 312 222 72 662 662 1182 272 Total TCDD OCDD PCDD Compounds 110 100 80 97 NA 12 NA 4 12 132 6 1412 2,3,7,8-TCDD 110 100 80 97 12 132 783915 5-11 GENP 011104 TABLE 5.11 ACCURACY DETERMINATION USING SPIKED BASELINE LIQUID ANALYSES GENP 011105 5_12 Total TCDf Total PnCDF Total HxCDF Total HpCDF OCDF Total PCDF Mineral Oil E729-53-01 Spike Amount Percenti I) Level Found Difference 200 300 475 .400 0 1375 100 373 2557 261 1 3382 . -5 24 438 -35 146 Aroclor 106 E729-53-02 Spike Level Amount Found Percent Di f ferente 53 63 316 520 185 2827 526 185 132 117 19 65 1428 1212 3712 206 Corrected^) Aroclor 1016 E729-53-02 63 5IS 2637 100 117 3432 Total TCOD OC DO 25 2 00 -92 53 45 00 -15 45 0 2.3,7,8-TCDD 25 2 -92 53 46 -13 46 2.3.7,8-/2.3,4,B-TCDF 200 190 -5 53 63 19 63 1,2,3,7,B-/1,2.3,4 .B-PnCDf 150 150 0 158 183 16 183 2.3.4,7,8-PnCDF 150 220 47 150 327 107 327 1.2.3,4.7.8-/ 1.2.1.4.7.9-HxCDF 250 373 49 00 -- -190 1,2,3,6,7,8-HxCDF 50 0 -100 53 88 66 88 2,3,4,6.7.0-HxCDF 175 2080 1089 132 2660 1915 2660 1,2,3,4,6,7,8-HpCDF 400 261 -35 526 132 -75 132 1.2.3,4,7,8.9-HpCDF __ 0 -- 0- - - - -85 (1) Percent Difference 100 (Measured Value * True Spike)/True Spike (2) Corrected by subtracting the Aroclor 1016 "Baseline" PCOD/PCDF concentrations from the measured values. Corrected Percent Difference 19 63 1325 -81 -11 183 -15 -13 19 16 107 -- 66 1915 -75 -- 783916 f -pciilts are compared with the results reported by the other four Battel I ' '** 'es participating in this program in Table 5.12. The data presented in i;jrator1 ^ other laboratories are taken from a draft statistical report (2) .mS tab^6 ^ d0 peiiizari (Research Triangle Institute, Research Triangle Park, ;^sen ^ ^ meeting of the five participating laboratories h.eld in Washington, D.C. November 19. 1986. ttelle8 results averaged over three replicate determinations are presented in 6 r n uhpre the average reported values by the five collaborating laboratories 'atjle 5-u w,,c ..,,,,,,cling parallel analyses are also given. he two sets of data compare favorably; except for hexachlorinated furans, the =ittelle results are consistently lower overall than the data contributed by other *oratories. The general trends in the data such as very low values for PCDO/PCDF *.n mineral oil samples was apparent in all the reported data. The 30-year-old -ietwork transformer containing Askarel (Sample Number 3 in Table 5.12) had the highest reported PCOF level of all the in-service samples tested, approximately 32 :pffl (w/w) using the average of all reported determinations. 783917 5-13 G ENP011106  ^ >-- l V-- * O cin i 00 CCOO 00 (AUI F 5.12 W A N FURAN RCSUl IS fOR SI ViN UI II IIV DinrCIKlC HUIOS Network Transformer No. 1 Sample Identifier fluid Years Service Field Code laboratory Code Askarel 20 ISL-02 A 17 Battelle Furan Results Total TCOF Total PnCDF Total HxCOF Total HpCDF OCDF 293 71 1S700 249 330 Total PCDF 16643 Combined Furan Results (II Total TCOF Total PnCDF Total HxCOF Total HpCDF OCDF 415 991 4886 1277 3346 Total PCDF 10914 Network Transformer __ Ho- 2__ Askarel 28 1SL-0B-A 12 17 96 1263 490 403 2269 128 279 1012 1822 6244 9486 Network Transformer No. 3 Askarel 30 ISL 03 A 11 . 53 702 24233 3951 2828 31767 158 1672 10824 6587 21830 4107) Network Transformer No. 4 Mineral Oil .UNK ISL-04 0 16 0 0 0 0 0 0 1 45 80 15 14 IS4 Arc furnace Transformer Mineral Oil UNK ISL 23 0 IS 0 0 0 0 0 0 1 4 2 8 18 34 Precipi tator Transformer Askarel 27 1SL-17 A 14 177 2076 22000 602 29 24884 782 7746 12931 1767 188 23413 Capacitor Askarel UNK ISL-27 13 106 21 143 10 17 297 566 221 67 12 21 906 Percent Difference in Furan Congener Results (21 521 -761 -231 (3) (3) (1) Combined Mean = Average reported values from the five participaiing laboratories. (2) Percent Difference (Combined Mean - Battelle ValueJ/Combined Mean X 1001 (3) 7ero value division not possible. 61 -671 Section 6 DISCUSSION This study produced a great amount of Insight Into the variability of analytical methods applied to trace analysis 1n a complex dielectric fluid matrix. PC6 constitute the principal class of interferents for modern mass spectrometric methods when used to analyze polychlorinated aromatic species such PCOF and PCDO. Several conclusions can be drawn concerning performance of the methodology and the accurate screening of in-service dielectric fluids. METHODS The analytical procedures proved to be both rugged and reliable when used with labelled quantification and recovery standards. The quantification standards are used to calculate errors from extraction and sorbent partitioning losses. This is especially important in the Askarel and mineral oil matrix since the high organic burden found with this matrix requires secondary partitioning to remove interfering PCB and the background hydrocarbons from mineral oil. Random outliers were observed in this study. One example was the elevated levels of hexachlorodibenzofuran found in the in-service analyses. This kind of acute error can arise from a mistake in the gravimetric preparation and dilution of the labelled recovery standard. Another possible cause for this random error is selective partitioning losses during sample clean up. Observing this type of spurious result indicates that the methods should be used with enough quality control checks to "flag" this type of error during the analysis. Using a second labeled Internal standard to quantify the mass recovery of the quantification standards would identify this problem if it 1s related to quantification standard losses.. The observed data variability in this program was smaller than the participants had anticipated before the program began. In this very difficult matrix, a chlorinated oil, most analysts would predict a variability of one to two orders of magnitude (lOx to lOOx scatter around the true value). In most cases, the relative standard devia-tions were found to be less than 100 percent. The relative average deviation is the absolute difference between' Individual results and the mean (or true value), divided by the number of determinations. For samples such as the Aroclor 1242, G E T S tP 0 1 1 1 0 8 783919 Aroclor 1016, and mineral oil, the average deviation was most often less than 50 percent. This result was seen in individual laboratory replicate analyses and in grouped data from different laboratories. IN-SERVICE LIQUID ANALYSES The in-service liquids tested in this program were found to have relatively low amounts of PCDD in all samples tested. This result supports the hypothesis held by many chemists who have investigated Askarel and PCB-containing fluids that dioxins are not a major constituent of these mixtures. Except for the mineral oils, polychlorinated dibenzofurans, however, were found in trace amounts in all of the samples tested. The mineral oil samples were found to have extremely low amounts of PCDD/PCDF if any were found at all. Even the in-service liquid from Transformer 4, a network distribution transformer that was involved in a high energy excursion, was found to have minimal amounts of PCDD and PCDF. Transformer 4 had visible carbon, and major signs of scorching on the case and the area of the electrodes. The Askarel transformers with more than 25 years of service tended to have higher levels of PCDF than the appliances put into service for a shorter time. It is possible that manufacturing controls improved as the production of Askarel matured in the 1960s such that Askarel was produced with lower levels of PCDF in the asdelivered fluids. It would be interesting to obtain replacement Askarel with a known lot number which matched a sample with extended service life. Analyzing samples of an unused Askarel and comparing the PCDF levels in a sample subjected to heavy load and environmental stress for an extended period would challenge the theory that utility operation contributed to elevated PCDF levels in Askarel fluids used in transformer and capacitor applications. Two utility appliances were selected by the EPRI Technical Advisory Committee for special attention. An Askarel-filled electrostatic precipitator (ESP) transformer, and an oil-filled arc furnace transformer. Both of these applications involve extended thermal and electrical excursions during daily operation. These two samples were the "worst case" for testing service stress on the dielectric fluids. The arc furnace had essentially no PCDF or PCDD, and the PCDF levels in the ESP transformer appeared to be dictated by the age of the unit (27 years of service) and the fluid composition (Aroclor 1260). This extended period of stress did not appear to elevate the PCDF levels above typical Aroclor 1260 contamination levels. One capacitor was tested in the in-service liquid screening program. It was found to have relatively low levels of PCDF compared to the other Askarel-filled units. 6-2 783920 I1 I The observed distribution of PCOF seemed to follow a trend of higher concentrations; of PCDF arising with the highest degree of chlorination Aroclor. The Aroclor 1260 with 60 percent chlorine by weight tended to have higher levels of PCDF in every comparable sample than Aroclor 1242, or Aroclor 1016, both of which have approximately 42 percent chlorine 1n the base Aroclor. A second effect was observed 1n these data, the relative concentrations of PCOFj congeners in any sample tends to favor the analogs with higher chlorination as the chlorination number of the underlying Aroclor increases. For instance the proportion of hexa, hepta, and octachlorodibenzofurans are higher in Aroclor 1260 than Aroclor 1242. If PCDF are formed by direct internal condensation with the addition of oxygen, HC1 elimination can occur. This reaction would preferentially! produce higher levels of chlorination in the product PCDF when high degrees of j chlorination occur in the reactant PCB. j iii SUMMARY This study was successful in developing a set of methodologies that reliably determine trace levels of PCDF and PCDQ in a very difficult matrix -- chlorinated organic oil. The development of commercially available reference standards and the statistical validation of reference methtfds that can be used to analyze utility i fluid was a major contribution to the quality and defensibility of data in this ' field. The method verification program showed that a range of instrumentation can be successfully used to measure trace levels of PCDF and PCDD in utility fluids. In general, the methods employing low resolution mass spectrometry require additional column chromatography to isolate PCDF/PCDD-enr1ched fraction for final gas chromatography. The precision of blind spiked samples was superior to the variance expected by experienced analysts who had previously performed this analysis. The availability of labelled quantification standards helped improve method performance and lower data scatter. | Although the in-service liquids tested in this program were a very limited set, there were interesting findings about the type and distribution of polychlorinated dioxin and furan species in the systems tested. Dioxin levels were very low, or not measurable 1n the liquids analyzed. The mineral-oil-filled equipment had levels that were found only at the quantification limits of the methods employed. Aroclor 1260 appears to have higher levels of PCOF with enrichment of the higher chlorine-number analogs, when compared to Aroclor 1242, 1016, or Aroclors with fewer chlorine atoms per molecule than 1260. This data base can not support the \ 1 783921 6-3 GENP011110 premise that the age of Askarel transformers tends to favor higher PCDF levels in older transformers since only four units were investigated. GENP011111 6-4 783922 Appendix A IN-SERVICE LIQUID QUESTIONNAIRE 783923 GENP 011112 Respondant: GENERAL INFORMATION QUESTIONNAIRE EPRI IN SERVICE LIQUIOS STUDY (RP 2028) Samples provided Askarel Load Center/Network Transformers Rectifier Transformer Precipitator Transformer Capacitor Oil Filled Load Center/Network Transformer Number Shipped ______________ ______________ ______________ ______________ ______________ GENP 011113 A-l 783924 DATA SHEET FOR EACH SAMPLE EPRI IN SERVICE LIQUIDS STUDY (RP 2023) General Information Voltage Rating; Liquid Volume: Manufacturer and Year: PCB/Askarel Manufacturer: (if available) PCB/Askarel Type: PCB Levels: (for oil filled, if known) Additional Equipment Identification: (name plate data) Equipment Maintenance History Length of Actual Service: _ _ _ _ _ _ _ _ _ _ Major Maintenance: Recorded Failure in Service: 783925 GENP 011114 A-2 DATA SHEET FOR EACH SAMPLE Was Fluid Changed at Failure: Any Record of Temperature Excursions: Are Maintenance Records Available: Any Other Pertinent Information About This Unit or the Sample Liquid: GENPoiiijj A-3 783926 Appendix B ANALYTICAL CURVE DATA FOR THE CROSS CHECK ANALYSES 783927 G E N P 011116 0 EPRI. COMPOSITE STO CURVE. 2.J.7.8*retnCBF A8eA R A T 5 coNComnncN EPRI. COMPOSITE STO CURVE. 2.J.7.8-T,tr*CDF X-WALUCS 34 AAA* I*A AAAO *AA AAAO 344A. 4400 Y - V M . U 19 0091....0031743440474A9* 0 0003 0.334A 91..41***003 X-VAUJK Y-VALUC STO DEV 00 037364*4 91..71043037 A 30* 00..03441300 41.41*33700 MES. C. 00 0384441* 41.4133333* A. A. DATA WITH ACSFCCT TO T M Y-VACUCS 31403000...000000000100 3300.0000 0031....04310334*33413*7 00..00009433 00..0*101197 1*31t....7400*100 0000....000410134081**09 DATA WITH ACSFtCT TO T X AEORCSSION L I X 3331003O000....0000000000001000 0031....03418334*3431*73 0.0371 0.0139 00.0M*l3*9 313331....3440*89* 0000....4000947313111743 A A T S S X L k O G D P F A ! C TA or OF HA * A30.. 00303310*0 A-CORMCLA O0ivVcCa8 TION C A a O M m E F INTCMCC TNTCACC FICIENT PT CT O F o O f F0 444-04A3. 093908**09 0000....000300143004034* 0.0343 000..,009144134933 GENP 011117 B-l - 783928 EPflI, COMPOSITE STO CUflvE, 1.2.J.7.8-PnuC0F CQCEKTMTIOM PflI. COMPOSITE STO CURVE, 1.2.J.7,S-PnuCOF I-VALUES Y-VALUES 40 0000 1SO.0000 44* 4**0 34** 4*00 0 09* 0 3347 1 31 S3 6. SOSO o.loai 0. 3333 1 3171 *. 3310 X-VALUS Y-VACm a m dev 0. 0**1 0.3349 1 3301 9. 9900 ASOX 0 0*3* 0 3931 1 1449 9 9030 ACS. C. 0.0*3* 0 3*03 1 1469 9. 3400 r .R. e. DATA WITH H P I C T TO THE Y-V4LUSO 40. 0000 190. 0000 900. QOOO 3900. 0000 0. OSO 9 0.3441 1. 1S33 9.1333 0. 03** 0. 0119 0 03*fl 0 41*9 4*. 1* 3. 33 3. 39 7 19 0. 0*2* 0. 0103 0. 0*3S 0. *9*9 DATA WITH RESPECT TO TM REGRESSION LINE 40. 0000 190. 0000 900. 0000 3900. OOOO 0 0*09 0 3441 1 1*33 9 *333 0 0419 0 0133 0 044* 0.41*9 91 91 3. 9* 3. 7* 7 19 0. 0*9* 0. 01*4 0. 0709 0. *90* A SLOPE OF 0.00333 0IV5 AM INTERCEPT OP -0.00137 A SLOPE OP 430 03700 01 VES AM 1NTCRCECT OP 0 *7317 THE d a t a MAS A COMPILATtON eOCPPICIENT OP O. 9***1 0. 0337 0. 00* 0. 033* 0. 394* 0 0393 0 0109 0 03*0 0. 394* 783929 G B N P0111,8 B-2 EMl, COMPOStTE STO CURVE. Z.J.4.7.0-PenuCOF AR EA R OI 9 . 624.99 cacEM iso1n2o4H9.99 1974,99 2499.99 S34. COMPOSTE STO CURVE. 2 , 3 ,,1.7,S-?e<ttlC3F -VM.US3 Y-VALUSS 40. 0000 130. 0000 4*0 ***0 34*0 **00 0. 0703 0. 434* 1. 3337 7. 4333 0.0743 0 4403 1. 4380 7 4430 -VALUO Y -VALUO STO OCV 0. 0714 0. 4473 1. 4433 7. 0*40 ASO 0 00*3 0 3304 1 1333 9 1430 M I . C. 0. 0010 0. 33*4 1. 1114 9. 3**9 4 4. S. OATA W1TH 4C04CCT TO THC v-vALuca 40. 0000 130.0000 900.0000 3300. OOOO 0. 0773 0. 4009 1. 3403 4. 3040 0 OOOO 0. 0344 0 SOIS 1 13*4 10. 33 14 04 19. 03 17. 34 OATA U1TH A I W t C T TO TH 4CO4C9SI0N L I M 40. OOOO 130. OOOO 300. OOOO 3300. OOOO 0 0773 0 4003 1 3403 4. 3040 0. 034* 0 0344 0 SOSO 1 13*4 47 43 14. 13 i3. a* 17. 34~ * 9L0P* 00 Q. 00340 OIVSS AN INTtACtPT 00 O 00934 * SLQAC 00 304 30*00 Ovea *M INTEMCCCT 00- -a 04303 TMC DAT* HA * COftaeLATtOH COCFOTCISNT 00 O **0*9 0. 0113 0. 07*7 0. 3B90 1. 9*74 0. 0931 0 OOOl 0 3*00 1. 3*79 0. 0099 0. 0333 0. 1370 0. 7*73 0. 0391 0 0399 0; 13*4 0. 7*7* GETSn? 011119 B'3 783930 &0 COCEHT5AT10N ?[, COMPOSITE s:: c u r v e . 1 .2 ,,3.4.7.8-HtxiCOf X-VALUH Y-VAUJE3 40. OOOO 190. 0000 *** ***o 3440 7*00 0. 0 7 03 0 313* 1 173* 9. 6339 0 0777 0. 3101 1. 0739 9. 7743 X-VtUK r-VAtm STO OCV 0 0703 0 3073 1 3033 9 4700 RSDt 0 03*4 0 41*3 t 09*7 4 *040 ' 41. E. 0 0733 0, 370* l. 03*1 4. 3130 R. R D A TA W I T H ftCCFCCT TQ TV V~ V* tU C3 0 0000 190.0000 900. 0000 3900. 0000 0. 07*4 0. 3439 1 1149 9 *373 0.00*9 0 0447 0 0*7* 0 7**1 S. 3* 13 04 * 09 13. 19 0. 0103 0. 0707 9 10*3 1. 9790 OATA UITM ACSriCT T TW* RIBRESSICM tir 40. 0000 190. 0000 900. 0000 3900. 0000 0 07*4 a 3439 1 1149 9. 4373 0 01**0 0499 0 06S30 ***t 39 *4 13. 33 *. 17 1. 19 * stara * stari thc o a t a ourrma4oa.990c.0c3aa1im*iooliaetvtievcsmsacnaomiirnintncetreicreecnptetcotorro0r o 009*7 -a. 9s*9 ****8 0. 0314 0. 0730 0. 1033 l. 9790 0. 0093 0 0403 0 0*0* 0 3*74 0 017* 0. 0410 0 0*30 0 8*74 783931 S N P 011120 AE a A T 5 cac&fltanw E?R!. wHPQSITS ST3 C'JfWE. 2,J,4,6.7.3*m C3F -VACUI r-vM.ua 40. 0000 130. OOOO 499. 9990 3499. 9900 0. 039 0. 2134 l. 0319 3. 4339 0. 0437 0. 379 1 0143 3. 3473 1-VM.UC Y-UACUK STO OCV 0. 0373 0. 3134 0, 9733 3. 3113 ASSI 0. 0710 0. 3733 1. 0334 4 3340 Cl. 1. 0. 0713 0. 2904 1. 039 7. 3400 1. A. DATA WITH R U 7 I C T t o r m y- v a c u o 40. 0000 130. OOOO 300. OOOO 2300. OOOO 0. 0447 0. 3044 1. 0313 3. 3937 0. 0043 0 0402 0. 0419 1. 093 DATA U1TH A IS T K T TO T W H0AK8SICM L IM 9 99 13. 19 4 03 19. 44. 0. 0090 0. 0949 0. 0393 1. 3403 40. OOOO 130. OOOO 300. OOOO 2300. OOOO 0. 0447 0. 3044 1. 0303 3. 3937 0 0240 0. 0429 0. 0429 1. 0192 3: 29 14 04 4. 09 19. 44 0 0330 0. 0. 0G4S0149 1. 3404 A SLOT o r * scare or 0. 4*3. 030030203 O IV otvci am am INTCTCCFT in t ia c ic t or or -O20. .0749M33*0 TMl d a ta MA a C Q W C a T I ON COCFFICICNT or O. 99993 0. 0032 0. 0323 0.0339 0. 1791 0. 0200 O 0347 0. 030 0 179 GENP OII121 B-5 783932 i cawENTmrim E P R I . LflMPOSITE ST3 CURVE. 1 , 2 . 1 ,4,6,2,8'rtaociCSF K-VAUUCS Y-VALUCS 40. o o o o 190. O O O O 4TT TTTO 24TT TTOO 0 3*73 1 347* 9. 3 0 7 * 33. 3 3 7 0 0. 3 0 9 7 t. 1 3 9 9 4 3144 33.97T4 0. 3 4 3 9 1 413* 3. T 3 0 9 3*. 0 1 * 9 1-WACU WAUUC STO OCV *803 0 390* 0 3T39 4 0030 33.09QO A M . C. 0. 3 0 * 9 1. 1 * 3 0 3. 7 3 1 0 31. 7 0 0 0 IT ft. C. s a t * y It h n u l t e r m t h c y -v a u u u 40. OOOO 190. OOOO 900. OOOO 3900. OOOO 0. 291* l. IT37 3. T*37 31. T099 0. 0340 0. 1*1* 0. 847* 3. 7443 13. 4* 19. 31 31. 4T 17. 0* 0. 0937 0.3971 1. 3404 9. T300 OAT* WITH R O F C C T TO TMC ftCOMCSHON L I M 40. OOOO 190. OOOO 900. OOOO 3900. OOOO 0. 391T 1. 1T37 3. T437 21. T099 0. 1399 0 20*1 0 8TI9 3. 7449 93. *0 17 37 33. *0 17. OT- 0. 3143 0. 3390 1. 40T* 9. T309 A 1U0F* OF O 0Q**3 O I V AN INTERCEPT OF -O. 3313*' A LOFC OF 113.30*00 01VCS AN INT1RCECT OF 39.04*10 t h i o a t * h a * A CG*ACUATION COCFFICICNT OF O TATOS 0. 007T 0. 0431 0. 1T49 0. 8*7* 0. 0314 0.0479 0. 304* 0. 8*77 783933 B-6 GENP011122 EP*l, COMPOSITE STO CURVE, 2.3.7.0-TetrCOO EPRI, COMPOSITE STO CURVE. 2,2.7,S-TtrCD0 -VALU Y-VALU59 40 0000 190. 0000 444 4440 3444. 4400 0 0*03 0. 3*00 0. 4703 * 0330 0. 0*09 0. 3033 0. 4934 4. 339 X-VAUJC V-VALUt STO DCV 0 0*1* 0. 3703 0. *117 4. 7193 450 0 07*9 0 3790 0 9476 4 4340 * u . c. 0. 0*77 0. 3517 0 4313 *. 1340 n . . t. <<11 OATA WITH F U FfCT TO T X 40. 0000 190. 0000 900. 0000 3900.0000 0. 0*43 0.37*0 0. 4303 4. 4*40 0. 00** 0. 00*1 0. 037 . 0. *0** 4 7 3. 14 3. 4 13. 30 0. 010 0. 004* 0. 0*34 0. 4943 ATA WITH tSPCCT TO T X 4 C M M S 1 0 N LINK 40. 0000 190. OOOO 900. OOOO 3900. OOOO 0.0**3 0.37*0 0. 4353 4. 4*40 0. 0333 0. 013 0: 0443 0. *0*7 39. 10 4 *3 9. 30 13. 30 0. 03*0 0. 0303 0. 0 7 7 4 0. 4943 * .OF OF 0. 001TP QlVtl AM t N T m V T OF >0. 0340' * SLOPI OF 900. 33400 9 1 V U AM tMTCRCKCT OF 17 09*30 T X 0ATA HA a COMCLATION C03FFCI3NT OF O 444*3 0.00*7 0. 00*3 0. 03*3 0. *144 0. 033* 0. 0130 0. 0444 O. *190 GENP0J1123 B-7 783934 EPICI, COPOSITE STO CtfiUE. OCOO CtMZHTWrtW ERRI. COMPOSITE STO CURVE. OCOO VALUES T-VALUSS 40. 0000 190. 0000 4 N . ***0 34**. 7900 0. 0*70- 0. 93*4 3. 3410 13. 3**9 0. 1037 0. 9040 a. oat i 13. 3449 (v a l u e T-VALUE STD DEV 0. 1093 0. 941 3. 1*39 13. 94*9 RESE 0. 1933 0 4*43 3. 4300 * *900 RES. E. 0. 1919 0. 99*1 a. 1330 * 1030 a r .e. OAT* WITH REIPECT TO THE v- valuee 40.0000 190. 0600 900.0000 3900.0000 0. 1347 0. 9944 3. 1309 10. *141 0. 03*0. 0. 0713 0. 1. 31 30*43 OATA WITH RBIPECT TO THE REGRESSION LINE at. 4* 13. *9 10. *0 17. 30 0. 0433 0. 1137 0. a. 3* *0 39 30 40. 0000 190. 0000 900. 0000 3900. 0000 0. 1347 0. 3944 3. 1309 10. *141 0.03*4 0. 077* 01. 3*30 0*43 31 9* 13. 10. S*O* 17. 30 0. 0*33 0. 1334 0. 3. 3* *0 39 30 A SLOPE o r 0. 0 0 4 3 V gives AN INTERCEPT OP -0 07793 A SLOPE OP 337. 47100 GIVES AN INTENCECT OP 17 *43*0 T> DATA M A E A CORRELATION COEPPtCIENT OP 0. ***** 0.0134 0. 033* 0. 1097 0. 070* 0. 01*3 0 0397 0. 1097 0. *70* GENP 011124 B-8 783935 Appendix C LETTER REPORT FROM UMEA GENP 011125 783936 Urnei Universitet 901 87 Umea 86-06-23 Analys results BPRI-pro jekt 2028 Formation of PCDD and PCDF in askarel and contaminated mineral oil equipment. Contents. Normalized raw counts for regression lines. Levels in baseline liqids Levels in inservice liqids: TCDF/TCDDs PnCDF/PnCDDs and HxCDF/HxCDDs HpCDF/HpCDDs and OCDF/OCDDS GC/MS Conditions. Clean-up By Lars-Owe Kjeller UmeA Universitet 901 87 umeA Sweden. 783937 G E N P 011126 LARS-OWE KJELLER Ume Universitet 901 87 UME Sweden Urne 85-06-18 EPRI-prodjekt 2028 Calibration curve, raw counts normalized for ^C-internalstandard. Concentration injected pg 2.37 8-TCDF 2.3.7.8- TCDD 1.2.3.7.8- PnCDP 2.3.4.7.8- PnCDP 1.2.3.4.7.8- HxCDF 2.3.4.6.7.8- HxCDF 1.2.3 . 4 . 6 . 7.8-HpCDP OCDP 10 9.5 3.4 6.0 4.9 3.9 1.7 1.4 ND 10 12.4 7.9 9.5 8.5 7.8 4.6 3.9 ND 10 10.8 7 .0 8.1 7.5 7.3 4.5 3.8 ND Concentration injected pg 2.3.7.8-TCDF 2.3.7.8-TCDD 1.2.3.7.8-PnCDP 2.3.4.7.8-PnCDF 1.2.3.4.7.8-HxCDP 2 . 3 . 4 . 6 . 7 . 8-HxCDP' 1.2.3.4.6.7.8-HpCDP OCDP 40 38.S 18.0 24.3 17.8 19.1 8.8 5.8 0.9 40 42.3 27.6 35.0 29.0 27.3 18.2 13.0 2.0 C-2 40 42.6 27.8 37.8 29.0 27.3 18.2 13.0 1.5 ^83938 G E N P 0 1 JJ2 7 Concentration injected pg 2.3.7.8- TCDF 2.3.7.8- TCDD 1.2.3.7.8- PnCDF 2.3.4.7.8- PnCDF 1.2.3.4.7.8- HxCDF 2.3.4.6.7.8- HxCDF 1.2.3.4.6.7.8- HpCDF OCDF 150 160.9 100.1 133.4 100.7 105.3 51.7 39.3 5.6 150 155.1 94.2 124.5 98.2 82.2 43.1 33.7 2.1 150 160.1 105.7 137.9 118.1 103.3 68.9 53.4 6.7 Concentration injected pg 2.3.7.8- TCDF 2.3.7.8- TCDD 1.2.3.7.8- PnCDF 2.3.4.7.8- PnCDF 1.2.3.4.7.8- HxCDF 2.3.4.6.7.8- HxCDF 1.2.3.4.6.7.8- HpCDF OCDF 500 545.3 370.9 470.0 403.5 339.8 214.3 182.4 23.5 500 570.0 371.0 471.0 401.0 338.0 203.0 177.0 17.0 500 550.9 359.9 467.7 402.1 .341.9 216.2 181.9 17.4 X Concentration injected pg 2.3.7.8- TCDF 2.3.7.8- TCDD 1.2.3.7.8- PnCDF 2.3.4.7.8- PnCDF 1.2.3.4.7.8- HxCDF 2.3.4.6.7.8- HxCDF 1.2.3.4.6.7.8- BpCDF OCDF 2500 2791.7 1925.0 2412.8 2012.5 1780.1 1118.9 973.4 166.4 2500 2769.4 1806.1 2384.7 1979.0 1760.3 1145.3 889.8 125.8 2500 2767.9 1924.3 2413.1 2061.7 1793.3 1100.5 952.1 169.1 783939 GENP 011128 Concentration injected pg 2.3.7.8-TCDF 2.3.7.8-TCDD 1.2.3.7.8-PnCDF 2.3.4.7.8-PnCDF 1.2.3.4.7.8-HxCDF 2.3.4.6.7.8-HxCDF 1.2.3.4.6.7.8-HpCDF OCDF 7500 8731.9 5966.9 7401.8 6219.0 5565.6 3532.6 3067.2 513.6 7500 8593.7 5480.0 7264.2 5897.8 5502.2 3577.6 2790.8 419.3 7500 9037.3 5735.6 6989.2 5980.2 5494.7 3638.1 2768.4 433.1 Characterization for the regression line y = a*x + b y * amont of Dioxin/furans. x = normalized counts. 2.3.7.8-TCDF 2.3.7.8-TCDD 1.2.3.7.8-PnCDF 2.3.4.7.8-PnCDF 1.2.3.4.7.8-HxCDF 2.3.4.6.7.8-HxCDF 1.2.3.4.6.7.8-HpCDF OCDF1 a 0.853 1.306 1.037 1.240 1.357 0.487 2.595 0.061 b 29.89 18.28 9..24 6.76 24.50 -21.26 31.83 -4.44 Residual standard deviation. 135.7 117.5 77.5 76.1 38.0 34.1 159.6 22.6 1 10 pg cone, not included. 783940 GENP 011129 C-4 783941 Octn LP o l9Li0An2eRASa-OsuHnEUiHvKEeAJrEsLSiLwtEeeRtd en timed 0 6 -0 6 -1 7 E P K l-prodject 2028 L e v e ls o f diQMihH and d ib e n s o fu ra n s in in sert* ic e 1ig u id 1 12 A tt n r. 41159-11TCDP 11111111411112126464744487889766////1111123264717807 /1 2 6 8 /1 4 7 8 /1 1 6 9 T22231111222111111O342311233442222228664747466646T633399778899867987498////122T1233C4744D6979F7 /2 4 6 8 /1 2 3 6 ABC l i C -2378-T C D T 11 8<<1<<54S982285549441111 ......345104 ..........4591351100 2 42<<<<2104081111 .1 .1 ..11 451 12 13 14 15 16 17 <11068222421110............417442025419 6 4 1<<<<4047930000.........86051191114444 8<<600 ..11 44 501 n1II nn129<<13o10592952t7t00......05600111 71I 4n84<<<<<360070Bt00000 .0 3 ..80 1 ....B000 131 47% 1 .2 a6 ..19 17221704 .8 116<11106950..73 2 91162677<<<<4<<3B070110000000 ..31 22 .*12 .1 2 ..31 22 54% N<<<<<<<<<<<<<<<<<<<<<<<<<O0000000000000000000000000........................1.00000000000000000000000014444444444444444444444444 65% N<<<<<<<<<<<<<<<<<<<<<<<<<O0000000000000000000000000 .........................0000000000000000000000000 S555555555555555555555555 67% 1 5<7<14<<<21311<615340638553032000000............044014131113 444444 4 76<<<04000 .3 ..33 4 44 01% TCDO BBC 11C -2378-TCD O No TCDO isomers d e t e c t e d . Detection limit 0 . 2 n g /g 41% 48% 49% 50% 56% 44% 26% ieiTIOdNHO Zfr6C8L A tt n r. 4 U S 9 -U - HxCDO Ho IlkClit) iso m e rs d e te c t e d . D e te c tio n l i m i t 0 .2 n q /g *we tvm*>m*ujs*w)eykt>*ne m t m et u e *j ^ tet kin et* hn B t f wN e. * v e v v tnx*. i e sNB * * smn pn S10 HHQ y*MB MawP-yeM*aM***Mrwun*tMvuJwv*ui**-ou**>K wei-*wM-*Mww0Mww' VWy*wi *Kre*J>w*wMi*t-io*^*ww*wwV*Mu*'MD*w--i s * M *s0 0n t_CI -- C-J ^-u Vi bu>/ * * *1C Bo tB o0 yBoO^ uO oO^Bw Bo 'O i" oe ^O iO' Uet oOotH 9V *> MN BH By O< w> MI ICB >w>*y '> BB w* Nw U--t Uyt B t ' O AD t ` > ' K I KWI * 0 * O S uUta m ou1a a a Wut B Bet Wut ae t oKOoJkuO*efB"uiBai yBi *'B>VO'uPO6"a h*OD iD*P toBaavBo"iBay*DO> UCD B AO AB AB AB AB AB BA OA AB BA OA AO AO ho oV oh aB aO aB aBaOab aBaB aB aB IePt Oz aB aO aD aO aO aO aB aB aB aB aO aB aB ^ BBOBOOBBO vUe) MeOO>Mw *t We tUMt wUt wMB e Bi BwWB WO" **Utt B* MJ Ut B A y W f t A A A A O A U O ' l l l ' J A C l I O A A l O A c 3 n 0 o *<J I9--9 i OA Wi J * OA y DA SAOAMKJ* * B OO A* Ky wV > J I*A OM AH V i M aa Ut te KiBaa BIt * j aa ait ut ot Bet t i * * ^p yto Oa AO AO AB tvDU O W A *" 3 A A w B tn m m ft o o a U ut e> e se ut B B n ft m *a Bais L* OOO0 e0^s^O0^O0AW0-f19f ^i 0A ut ut *O w u t UM* * M AB BB AO O^ Oe yt 'B Bo Oi OC UA te OB uu ut ut ut n 9o ZB AB AO AO AB AB AB AB AB AB AB AB AB AB AB AB BA OA AB OA AO AB AO AB *B- Oa aO aB uO tOu Bt Ba Ba Ba Ba Oa aB aB aB aO a 0a 3a Ba Ba aB aS aB aB B m ewt b ab ab ab ab ao ab ao ao ao ab ab ab ao ab ab ab ab ab ao ao ao ab ab 3 tC ebt 0b ao ab ba ba sa sa betba oa ba oa ba betbetbetbetoetbetbetbetbet tfl be t oOetuB< aO eBta* Ba Bi-- M Ba Oa wB mB e>weM ABeBt Bu te twmOB*i a6 AOASW AO w D w O Ut w BW w WJ At* wui we t Mw n A tt n r. 41159-11H pcor3 T1111O2222333T344447666S8877p9989C D r BpCDD OCDF 12346789 OCDO 12346789 RKC 13C-OCOO 11 12 13 14 15 16 17 1I4S5081O0006 11205040006400000 < 01130 .....27022 7 7 2 184618071700000 N<<<<D0000 ....00006666 <<N<<D0000 ....00007777 NO HpCDD is o m e r s d e t e c t e d . D e t e c t i o n l i m i t 0 .5 n q /q 21425313 .8 .8 20000* 10000 <10 7.7 ND 45% 5.4 100 < 0 .7 3 41% < 4 .7 69% <3.3 <3.8 5400 <1.2 8.5 <5.2 43%' 41% 100% * A ll q u a n t i f i c a t i o n s a r e m ade w ith t h e l 2 C-c o tn p u n d s i n RADIAN c a l i b r a t i o n s t a n d a r d s s e t . 2 C orection fo r recovery is made. 3 C o r e c t i o n f o r r e c o v e r y m ade w i th 3'3C -123 478-H xC O P s p i k e . 4 The valu es a r not co rected fo r reco very . 783943 C-7 GENP 011132 High Resolution Gas Chromatography and Mass Spectrometry Conditions ,Column SPELC0 SP-2330 Fused Silica Length 60 m Diameter 0.25 mm Film Thickness 0.20 urn GC-Conditions HP 5890 with Helium as carrier gas Column pressure 40 psi Temperatur program: 100C for 2 min splittless, to 18QC with 2QC/min, to 250C with 3C/min, hold at 250C for 30 min. Injections volym 1-3 ul MS-Conditions Instrument: VG 12-250 Low Resolution Quadrupole in HI mode Electrone Energi 37 eV Source Potential 9v Repeller Voltage 9V Amplifier Gain Low sens SIR Multiplier Voltage 2000 V Tuning Calibrant Source Temp. PFK 200 Transfer Line Temp. 250 783944 C-8 GENP 011133 Clean-up technique for the determination of PCDF in askarel and miniral oil 783945 GENP 01II34 C-9 LARS-OWE KJELLER UmeA U n i v e r s i t e t 90L 87 UMEA Sweden fJmeA 86 -06-18 E P R I - o r o d j e c t 2028 Levels o f d io x in s and d ib ensofurans in b a s e lin e l i q u i d s Att nr. E729-53- 2.3.7.8-TCDF 2,3,4, 8-TCDP Tot. TCDP's REC I3C-2378-TCDP 2.3.7.8-TCDD Tot. T C D D 's REC 13C-2378-TCDD 1.2.3.4.8-/ 1.2.3.7.8-PnCDP 2.3.4.7.8-PnCDP Tot. P n C D P 's REC I3C-12378-PnCDP 1.2.3.7.8-PnCDD Tot. PnCDD's 1.2.3.4.7.9-/ 1.2.3.4.7.8-HxCDP 1.2.3.6.7.8-HxCDP 1.2.3.7.8.9-HxCDP 2.3.4.6.7.8-HxCDP Tot. HxCDP's REC 13C-123478-HxCDP 1.2.3.4.7.8-HxCDD 1.2.3.6.7.8-HxCDD 1.2.3.7.8.9-HxCDD Tot. HxCDD's 1,2,3,4,6,7,8-HpCDF Tot. HpCDP's Tot. HpCDD's OCDP OCDD REC 13C-OCDD 01 89 100 190 60% 14 14 67% 160 160 320 59% <1.5 ND 290 31 <2.1 170 490 59% <2.7 <2.7 <2.7 ND 450 450 ND ND2 ND2 C-10 ND2 02 03 62 <0.8 62 60% 38 38 58% 1300, ND`L 2900 67% 92 92 62% 190 180 370 63% <1.3 ND 400 370 1800 58% <1.6 ND <1.1 40 <1.1 110 150 1400 370 <1.8 350 2800 66% 69% <2.2 <2.2 <2.2 ND <2.4 <2.4 < 2 .4 ND 630 1000 630 2300 ND ND 56 3800 <10 <8.1 76% 110% GENP011135 783946 The v a lu e s a re g i v e n i n n g /g ND = Not D e t e c t e d . ND^ = Not d e t e c t e d due t o i n t e r f e r e n s from th e b i g 2,3,7,8-TCDF peak. ND2 = Not d e t e c t e d , no s i n g n a l f o r 13C-0CDDn. GENP011136 c-n 783947