Document k9e4BX3wj7Jj1ym5rE1eV1y0y

Electric Power RmmtcTi Institute Topics: PCB Chsmtesl analysis Tlanstormsrs Capacitor* PCOF-PCDD losulstlng oil EPftl EL'EA^ew IMums 2 PtOiKti 2026-5, -6, -9 Fins) Report December 1966 Polychlorinated Dibenzofurans (PCDF) and Polychlorinated Dibenzo-p-Dloxins (PCOD) In Utility Dansfbrmere and Capacitors Volume 2 Prepared by ITT Reeearch Irwtttute Chicago, ttmow HONS 2L54Q7 r REPORT SUMMARY SUBJECTS Hazardouartoxfc substance* / Danamiaaton: Substations / Distribution: Substations TOPICS PCB Chsmicai analysts Tiansformsrs Capacitors PCDF-PCDD Insulating oil AUDIENCE Environmental managers / Distribution engineers Polychlorinated Dlbonzofurans (PCDF) and Polychlorinated Dlbenzo-p-DloxIns (PCDD) In Utility Itaneformers end Capacitors Volumes 1-3 Many hearth affects originally attributed to PCBs appear to result from their pyrolysis and combustion by-products. PCDF and PCDD. This study developed Improved techniques for characterizing these products and shewed that PCDF and PCDD do not form during catastrophic arcing failure in transformers and capacitors. BACKGROUND Partial oxidation of aakarsIs--mixtures of potychlorinatsd biphenyls (PCB) and trf- and tetraehlorobenzenee--produces polychlorinated dibensofurens (PCDF) and polychlorinated dibens>-p-diOKins (PCDO). However, analytic technique# tor separating the active components in thtee mixture* are not yet mature. Information to atoo sparse on the quantity and typee of PCDF and PCDO in ae-manufactured PCB and on their formation during normal operation or arcing toNura In uWty equipment. Becauee of foeee uneertwntlet, utiilttoe and regulalors assume the wont-case PCDF-PCDO formation when there to a PC8-retoted accident and thereby overstate potential toxicity OBJECTIVE 1b Improve techniques tor measuring PCOF-PCDD in the presence of PCB end in utility transformers end capacitors. APPROACH Researchers developed a rationale tor eetecdng transformers and capaci tors to be sampled tor PCDF-PCDO content. They atoo simulated cata strophic transformer tenure through arcing in askerel and contaminated mineral on to investigate PCDF-PCDD formation. Several laboratories then performed round-robin gas ehromatography-mees spectrometry analyses of PCDF-PCDD samples using spedafly prepared carbon-13 CK) PCDF spiking compounds. After the Diet set of analyses, each laboratory modified its techniques end analyzed a larger group of samples from utility equip ment. Another organization statistically analyzed art laboratory results. RESULTS These three volumes are the flrat relates In a abMtolume report of PCOF-PCDD analytic findings Results from aN participating laboratories correlated remarkably well. Some of those results todow. EFRt EUEA4MSI VWe 1-3 RONS 21S40B  Making uet of the **C compounds. reaearchars successfully sepa rated tha components In several doeety eluting pain of phyelotogtcaity active PCOF and toss-active component Tha researcher* found no PCOF or PCOD In tha arctng of PCS, trt- and tatrachlorobenzena. or oontaminaiad mlnani oil. Anaiyaaa of samples from utility equipment auggaatad that tha PCOF-PCDO pftaanl did not result from highroad, Ngh-t*mperaturw oparatlon. - Wum* t presents raaaarehan' sampling recommendations tor Investigating PCDF-PCDD In utility equipment h afao daacribat arcing ahnuiationa and results. VMumaa 2 and 3 describe several analytic tachntquae, with wriume 3 Including a report on tha aynthaato of naw PCOF spiking compound*. Later volume* wW complete tha daacrlpiiona of laboratory tachniquaa. present oomprahanalv* atatlatical analyses, and ptmrid* an executive aummary. EPfli PERSPECTIVE Thie profact appaan to ba a landmark In PCOF-PCDO analysis. On* important result la tha aynthaafa of naw PCOfa tor ua* at spiking compounds In gas chromatography-mass apactromatry anaiyai* Tha ability to sapartta many of tha acta* PCOF compounds from moreInnocuou* matarlais to alto of graat value. Th* fact that laboratory raaufta conataiad so wad indicate* that researchers can have graat ttotobHIty In ealaettng techniques and devatoping facdUea. Utilities benefit horn this work In several way*. Tha study provides data showing thto HtHe or no PCOF-PCDD forms in slectrtcai aquipmant aNhar during normal oparatlon or during arcing Mure. Moreover, lm> prewad analytic tachniquaa make It dear that past predfctione of toxicity hare bean unnacatearity high. PROJECTS RP202S-5, RP202M, RP2028-P EPRl Prefect Managers: ON Addto; Jacques Quartin Etootrtoal Syatams DMtIon; Energy Anatyato and Erwtronmant Division Contractors: Radian Corporation; General Electric Company; IfT Raaaarch Inatituta For further information on EPRl research programs, cai EPRl Technical intormation Spadattats (41S) K5-24TT. RONS 215409 Rofychkxinated Dbenzofurans (PCDF) and Polychlorinated Dtoenzo-p-Oioxhs (PCDO) in Utfty transformers and Capacitors Volume 2 EUEA-4S58, Ntolum* 2 Research Project* 2026-5, -6, -9 Final Report, December 1966 Prepared by IIT RESEARCH INSTITUTE 10 WIM Thirty-fifth Street Chcaga Minors 60616 investigators S. M Cordon M. Miner EPRl Protect Managers G. Add** Transmission Substations Program Electrical Systems Divtson J. Guertin Environmental Physes and Chemistry Program Energy Analyen and Environment Oivison RONS 215410 ordering information Raqumts lor oopa* of ttut raport ihotAJ be tftractad to Reefaren Reports Cener (RRC). Bor 50490. Palo Alia CA 94303. (415) 965-4001 There no charge lor report* requested by EPRI member utttie* and affliatat. US uMty ataoMbona. US qowmment agenoee (lederal, tiaMt and local). meAa, and loraion orgeruMione wan when EPRI hat an information eechang* agreement On raouwL RRC wi aand a cauiog ct EPRI report*. CapaV* m Brant oa Miio maae* me Ai npa imumh notice Th* pwi wat awnfl by r afnmcrlil nmo Mew M an aoeeut a at etmt * rt EWeme aer*en mng en MM a ay a man t*| maw* tty mmn* acraa a neiee. nenawnMiM a nenmiMw. im--i nwmae. a lane acMea n tm toon a> m men " Mi ownee ripak a (N mwim try imeet #> mibm me um a, a tar awnaenn amang *an t* tee a tty amaL aw "wmoa-a m aMeaeam N meat Nama BT Hnatti baau* CMeaoa Hnaa MDNS 215*11 ABSTRACT The objective of this study was to enelyze dielectric fluids contaminated with polychlorinated biphenyls (PCBs), Including in-service transformer fluids, for specific polychlorinated dlbenzofurans (PCDFs) and dlbenzo-p-dloxlns (KDOs) using l>C-labeled Internal standards together with the best extraction and analytical gas chronatography/mass spectrometry (GC/NS) techniques available In-house. On completion of the first phase of the Investigation, the methodology was assessed and several Improvements were Incorporated Into the procedure before continuing with a round-robin analytical study. Samples were spiked with *>C-labe1ed PCDF and PCOD Internal standards before extraction and cleanup using basic and acidic alumina columns. The concentrated extracts were analyzed by high-resolution GC/low-resolutlon MS In the selected- 1on-monitoring mode. The methodology was applied to eight spiked baseline samples (Aroclors, trl-/ tetrachlorobenzenes, and aged mineral oils) and 11 In-service dielectric fluids. For the baseline samples, good agreement was obtained between measured and expected values for the mineral oil and chlorobenzene mixtures. Two of the three Aroclors gave substantially higher concentrations for hexa- through octa-CDFs than expected. In the case of the In-service samples. PCDF levels ranged from 'not detectable" to about 30 ppm (w/w). MOMS 215412 111 ACKNOWLEDGMENTS Tiie valuable assistance of Mr, l. L, Daniels, In preparing the samples for analysis, is gratefully acknowledged. The authors also extend their sincere appreciation to Dr. G. Addis and Or. J. Quart In of EPAI for their constant encouragement and constructive suggestions. HONS 215413 v CONTENTS Section 1 INTRODUCTION 2 ANALYTICAL METHODOLOGY Beckground Preperetlon of CellbretlonStenderds end Senples Phese I Phese II Senple Extrectlon end Cleenup SC/MS Anelysls 3 RESULTS AMO DISCUSSION Phese I Cellbretlon end QuelltyAssurence Dete Best line Senples In-Service Senples Conclusions Phese II Quellty Assurence endCellbretlon Oete Besellne Senples In-Service Senples '4 CONCLUSIONS 5 REFERENCES APPENDIX A SELECTED ION CURRENT PROFILES FOR CALIBRATION SOLUTION E7K-SS-OB APPENOIX B CALIBRATION CUIVES Page 1-1 Z-l 2-1 2-2 2*2 2-4 2-5 2-6 3-1 3-1 3*1 3-1 3-10 3-13 3-14 3-14 3-20 3-25 4-1 5-1 A-l B-l HONS 215*14 ILLUSTRATIONS LLa^rs 1-1 Pyrolytic Formation of PCOFt and PCOD* 1-2 The Mott Toxic PCOO and PCOF Itomert 2-1 Overview of Analytical Protocol 2-2 Selected Ion Current Profile for Matt 322 fro* GC/MS Analytlt of Performance Check Solution Containing TCOOt at Noted In the Plot, Utlng a 60 SP-2330 Futed Silica Capillary Coluan 2-3 Selected Ion Current Profile for Matt 304 froai GC/NS Analytlt of Performance Check Solution Containing TCDft at Noted on the Plot, Ustng a 60 a SP-2330 Futed-SIllca Capillary Coluan 3-1 Selected ton Current Profllet of Chromatography Performance Solution Using SP-2330 Capillary Coluein 3-2 Summary of Batalina Sample! and Splket (Phate I) 3-3 Selected ton Current Profllet of Chromatography Performance Solution Using SP'2300 Capillary Column pjj? 1-2 i,,j 2-2 2-8 2-9 3-2 3-3 3-15 MOMS 215*15 TABLES 1-1 Chlorinated Isomers of PCOFs, PCOOs, PCBs, CPs, and CBzs 2-1 Analytical Sequence In GC/NS Oetenelnatton of PCOFs and PCOOs 3-1 Average Response Factors for PC00/PC0F Calibration Compounds (Phase 1) 3-2 PCDF/PCDD Concentrations In Baseline Sample: Aroclor 1016 (Radian 1.0. E729-09-011 3-3 PCDF/PCOD Concentrations In Baseline Sample: Aroclor 1242 (Radian [.0. E729-10-01| 3-4 PCOF/PCDO Concentrations In Baseline Sample: Aroclor 1260 [Sample No. E729-11-Q1] 3-5 PCDF/PCOO Concentrations In Baseline Sample: Chlorobenzenes (Radian 1.0. E729-12-01| 3-6 PCOF/PCDO Concentrations In Baseline Sample: Mineral Oil (Radian 1.0. E729-08-011 3-7 PC0F/PC00 Concentrations In In-Service Sample: Oil-Filled Transformer (Battelle I.D. ISL-4-oj 3-8 PCOF/PCDO Concentrations In In-Service Sample: Askarel Capacitor (Battelie I.D. ISL-l-A) 3-9 PCDF/PCOO Concentrations In In-Service Semple: Askarel Load Center Network Transformer (Battelie 1.0. ISL-2-AI 3-10 PCDF/PCOO Concentrations in In-Service Sample: Askarel Load Center Network Transformer (Battelie 1.0. ISL-3-A) 3-11 Response Factors for PCDOfPCDF Calibration Compounds 3-12 Average Response Factors Over Entire Concentration Range for PC00/PC0F Calibration Compounds 3-13 Response Factors from Verification Runs 3-14 Analytical Oata Confidence Tiers PiSt 1-3 2-10 3-4 3-5 3-6 3-7 3-B 3-9 3-10 3-11 3-12 3-13 3-18 3-20 3-21 3-22 MONS 215416 x1 TABLES (continued) Table 3*15 PCDF/PCDD Concentrations in Baseline Simple: Aroclor 1016 {Radian 1.0. 729-53-021 3*16 PCOF/PCDD Concentrations In Baselfne Sample: Aroclor 1260 ~ {Radian 1.0. E729-S3-031 3-17 PCDF/PCDD Concentrations In Baseline Sample: Mineral Oil [Radian I.D. E729-53-011 3-IS PCDF/PCDO Concentrations In In-Service Sample: Load Center Network Transformer Askarel {Sample No. <1159-11-11; Battel la I.D. ISL-03-A) 3-19 PCOF/PCOD Concentrations In In-Service Sample: Load Center Network Transformer Askarel (Sample No. 41159-11-12; BattelIt 1.0. 1SL-08-A) 3-20 PC0F/PC00 Concentrations In In-Service Sample: Capacftor Aroclor (Sample No. 41159-11-13; Battalia 1.0. ISL-27-AI 3-21 PCDF/PCDD Concentrations In In-Service Sample: Precipitator Transformer Aroclor [Sample No. 41159-11-14; Battalia 1.8. ISL-17-A] 3-22 PCDF/PCDD Concentrations In In-Service Sample: Arc Furnace Transformer Mineral Oil [Semple No. 41159-11-15; Battalia I.D. ISL-23-0] 3-23 PC0F/PCD0 Concentrations In In-Service Sample: Substation Distribution Transformer Oil [Sample No. 41159-11-16; Battalia I.D. ISl-4.0] 3-24 PCDF/PCDD Concentrations In In-Service Sample: Load Center Network Transformer Askarel [Sample No, 41159-11-17; BattelU I.D. ISL-02-A] Page 3-23 3-24 3-26 3-27 3-29 3-31 3-32 3-33 3-3S 3-36 HONS 215417 SIJMWtV There Is Increasing fnterest In the potential formation of polychlorinated dibenzofuran* (PCOFs) and polychlorinated dlbenzo-p-dloxlns (PCOOs) from uncontrolled fires involving polychlorinated biphenyls (PCBs) and chlorobenzenes (PCSzs) in dielectric fluids. At the temperatures that prevail In transformer fires, PCBs may react to form PCDF contaminants and other toxic oxidation products. Vary little published Information is available on the occurrence and distribution of PCDFs and PCOOs in the PCB-contalnlng dielectric fluids (generally known as askarels) that arc used by the electric utility Industry. Even less Is known about the effect of service life on the contaminant concentration In PC8-f 11 led electrical equipment. The possibility also arises that abnormal operation (arcing, overheating) eay create conditions that lead to the formation of PCDFs and PCOOs In the equipment. Tests using laboratory animals suggest that these compounds at an equivalent concentration are much more toxic than the more abundant PCBs. Moreover, the toxlclttes of PCOFs and PCOOs /ary over a range of five orders of magnitude depending on the speclffc compound (i.e., degree of chlorination and location of chlorine atoms). In order to measure Isomer-speclf 1c PCOFs/PCODs In PCS-contamlnoted dielectric fluids, the Electric Power Research Institute Initiated the present study, with the following major objectives: (1) Perform a round-robin analytical methods evaluation, using the best available In-house extraction and analytical techniques (2) Analyze In-service dielectric fluids for PCOFs ono BCDDs of interest. The round-robin method evaluation was designed to validate :ne analytical methodology and to determine the eccurecy with which specific Isomers could be quantified. It Included the synthesis of native and Isotop'eally labeled standards, and the analysis of spiked baseline samples. Five laboratories were S-I MONS 215418 Involved In this effort; Bettelle Columbus laboratories, I IT Research Institute, New York State Department of Health, Radian Corporation, and the Chemistry Department of the University of Umea. Sweden. The In-service dielectric-fluid study consisted of developing appropriate selection criteria, obtaining representative samples, and analyzing the selected samples. General Electric Company developed the selection criteria, Battelle Columbus Laboratories collected the samples, and the analyses were performed by the same group of five laboratories mentioned above. The study was conducted In two phases. In Phase I, five spiked baseline samples (Arodor 1016, A rod or 1242, Aroclor 1260, a trl- and tetrachlorobenzene mixture, and aged minaret oil) and four In-service dielectric fluids were analyzed using the best techniques available In-house. Response factors for the components of Interest were generated at a single concentration using a standard calibration solution. Phase II was undertaken after the results from Phase 1 were assessed and Improvements to the methodology were Implemented. In the second phase, calibration curves were generated over the linear response range of the analytical system at five concentrations using calibration solutions containing eight native compounds and five labeled compounds. The resulting response factors were used to characterize three spiked baseline samples (Aroclor 1016, Arodor 1260, and aged mineral oil) and seven In-service dielectric fluids. The results were evaluated to determine precision and accuracy. This report sumaerlzes I IT Research Institute's participation In the program, outlines the analytical methodology used, and discusses results obtained from the analysis of the baseline and In-service samples. ANALYTICAL APPROACH Combined capillary column gas chromatography/mess spectrometry (GC/MS) Is the method of choice for the Isomer-specific analysis of trace concentrations of PCDFs and PCDOs In PCS-contalnlng dielectric fluids. The selectivity of high resolution GC together with the sensitivity and specificity of NS yields detection limits In the low ng/g (parts-per-bl 11 Ion) range for these compounds. Sine* capillary GC columns cannot handle these complex sample matrices directly, the samples are extracted first using a basic alumina column to separata the analytes of Interest (l.e., PCBs, PCDFs, PCDOs) from them. This Is followed by an enrichment step (acidic alumina column) to remove co-extracted Interferences and S-2 HONS 215*19 to conctntrate the PCDFi and PCOOs In the sample extract. Finally, /> analyil* of the extracts. In the selected-1 on-monitoring mode, allows the PCDFs and PCOOs of Interest to be Identified and quantified. This Is based on the promise that etch chlorinated compound class can be separated by SC so that a unique set of mass spectral Indicator ions can be monitored for each of several appropriate SC retention time windows. The samples are spiked with J|C-labe1ed PCOF and PCOO internal standards before extraction and cleanup to determine the recovery efficiency and improve quantitation accuracy. - METHOD EVALUATION The round-robin method evaluation was designed to validate the analytical procedures and determine the reliability with which specific compounds could be identified and quantified. The methodology was applied to eight baseline samples spiked with known amounts of native and isotoplcally labeled PC0F/PC00 isomers. Very good agreement was obtained between measured and expected values for the mineral oil and chlorobenzene mixtures. The baseline Aroclor samples showed a strong Increase In PCOF content as the weight percent of chlorine was Increased, thus. Aroclor 1016 give no measurable levels of PCOF, other than those which had been spiked Into the samples, whereas Aroclor 1260 showed significantly higher concentrations, especially for the higher congener classes. ANALYSIS OF IN-SERVICE SAMPLES The analysis of the 11 In-service samples gave results that ranged from 'not detectable* levels of PCOFs to about 30 ug/g. It was difficult, however, to establish clear correlations between PCOF concentrations and dielectric fluid composition as the co^osltlon of several of the In-service fluids was not known. Nevertheless, In some cases, it appeared that the concentrations of PCDFs in the In-service fluids could be explained in terms of the amounts present In the original Aroclor 1260 fluid. CONCLUSIONS The procedures developed for the sensitive and compouno-speclflc characterization of PCDFs and PCOOs In PCB-contamlnated dielectric fluids proved to be reliable when applied to a variety of matrices and over a wide range of concentrations. The study succeeded in characterizing the PCDF/PCDD content of the samples, and permitted the assessment of the precision and accuracy of the procedure under carefully controlled experimental conditions. HONS 215*20 S-3 Since the occurrence and distribution of PCDFs In In-service fluids My be dua in large part to the aaiounts present in the original Aroclor 1260 fluid, the askarels that have been In use In electrical equipment may not contain higher levels of PCDFs than unused fluids. Although the data obtained In this study provide som Indications that this may be the case, there is Insufficient InfQneatlon available to draw firm conclusions, and further work is clearly needed. MONS 215421 $-4 Section 1 INTRODUCTION For many years, polychlorinated biphenyls (PCBs) and chlorobenzenes (PCBzs) have been widely used as dielectric fluids In transformers and capacitors. Nany of the oil-filled transformers In service also contain trace quantities of PCIs as a result of manufacturing and transformer-servIcing practices lit. When scientific evidence began to accumulate In the late 1960s about their adverse toxic proper* ties, the sale of PCBs was halted In 1971 for all uses except In closed electrical systems. The U.S. Congress promulgated the Toxic Substances Control Act In 1976 and Included special provisions for the regulation of PCBs. By 1979 the produc tion of PCBs was banned. However, roughly 130,000 transformers and 2.B million capacitors that contain PCBs or mixtures of PCBs and trl- or tetrachlorobenzenes, also known as askarels, remain In service. Another 2 million mineral oil trans formers contain fluid contaminated with PCBs In concentrations of SO parts per million (ppm w/w) or greater (21. Since 1979, the formation of the highly toxic polychlorinated dlbenzofurans (PCDFs) and polychlorinated dlbenzo-p-dloxlns (PCDDs) has been linked to uncon trolled fires Involving PCB-contalnlng dielectric fluids. At the temperatures that prevail In such fires, PCBs may react to form PCDFs and other toxic oxidation products. Laboratory studies [3| have shown that pyrolysis of PCBs at tempera tures of 200* to 600*C In air could result In the formation of significant amounts of PCDFs (Figure 1-1). Similarly, PCDFs and PCDDs may form when PCBzs are heated In the presence of air (4). In fact, several transformer fires have been reported in which the resultant soot and ash contained relatively large amounts of PCDFs and PCDDs [S,|. Although the ambient concentrations of PCDFs and PCDDs are normally very Tow, tests using laboratory animals suggest that these compounds at an equivalent concentration are much more toxic than the more abundant PCBs. Information is relatively sparse on the occurrence and distribution of PCDFs and PCDDs In the original PCB fluids [7,81. Even less Is known about how these levels are affected by normal operation of PCB-fllled electrical equipment or by abnormal MONS 215422 l-l PCB* Flgurt l-l. Pyrolytic Formation of PCOFs and PCDDs operation, such arcing or overheating. One study ha* suggested that PCDf levels In contaminated transformer askarels Increase with the time the transformer Is In service (9). This study, however, was performed using analytical techniques that are generally regarded now as outdated. A more recent study [jgl has shown that askarels that have Men In use In electrical equipment do not contain higher levels of PCDF! than the unused fluids. There are 13S PCDF and 7$ PCDO Isomers (Table 1-1). The toxic 1 ties of these compounds vary over about five orders of magnitude, depending on the degree of chlorination and the location of the chlorine atoms ||J|. Generally, the most toxic Isomers are the 2,3,7,8-substltuted congeners* shown In Figure 1-2. A highly sensitive and Isomer-specific analytical method is needed to measure specific compounds. To address these issues, the Electric Power Research Institute (EPRI) initiated a program with the following major elements: MONS 215423 1-2 Table 1-1 CHLORINATED ISOMERS Of PCOFt, PCOOs, PCBs, CPs, ANO CBzs Chlorine Substitution MonoD1TrlTetraPentaHexaHeptaOct aNonaDeca- PCPFi PCDPt PCBt CPs CBzs 4 2 3 3 _1 16 10 12 6 3 28 14 24 6 3 38 22 42 3 3 28 14 46 1 1 16 10 42 -- 1 4 2 24 -- -- 1 1 12 -- -- ---- 3-- 1 -- 13S 75 209 19 12 2,3,7,BTCDD 1,2,3,7,0PnCDO 1,2,3,4,7,8HxCDO 1,2,3,6,7,8- 1,2,3,7,B,9- HxCDD HxCDD 2,3,7,$TCDF jsg&C 1,2,3,7,8PnCDF A 2,3,4,7,8PnCDF A 1,2,3,4,7,0HxCDF "n$r aT oTa Cl Cl 2,3,4,4,7,8HiCDF Figure 1-2. The Host Toxic PCDD and PCDF I toners 1-3 MOWS 215424 * Round-robin method evaluation using the best evtlltble In-house extrectlon end analytical techniques Analysis of in-service dielectric fluids for PCDFs end PCDDs of Interest. The round-robin method evaluation was designed to valldete the intlytlcel metho dology and to determine the accuracy with which specific Isomers could be quantified. It Included the synthesis of native and Isotoplcally labeled standards, and the analysis of spiked temple matrices (baseline samples). The standards ware synthesized by Radian Corporation, who also prepared the spiked matrix samples. The samples were analyzed `blind* by Battalia Columbus Labora tories, I IT Research Institute, New York State Department of Health, Radian Corporation, and the University of Umea, Sweden. The In-service dielectric study consisted of developing appropriate selection 'criteria, obtaining representative samples, and analyzing the selected samples. General Electric Company prepared the selection criteria document, Battelle Columbus Laboratories collected the samples, and the same group of five labora tories listed above performed the analyses. The study was conducted In two phases. In Phase I, five spiked baseline samples (Aroclor 1016, Aroclor 124?, Aroclor 1260, a trl- and tetrachlorobcnzene mixture, and aged mineral oil) and four In-service dielectric fluids were analyzed using the best available in-house extraction and analysis techniques. Response factors for tha components of Interest were generated at a single concentration using a standard calibration solution. Phase II was Initiated after Phase I results were assessed and several Improvements to the methodology were Implemented. In this phase, calibration curves were generated over the linear response range or the GC/MS system at five concentrations using calibration solutions containing eight native compounds and five labeled compounds. The resulting response factors were used to characterize three spiked baseline samples (Aroclor 1016, Aroclor 1260, and aged mineral oil) and seven in-service dielectric fluids. The results were evalueted to determine precision and accuracy. This report documents IITRI's participation in the program, and Includes a summary of the analytical methodology used, the results obtained from the baseline fluid study, and the results obtained from the In-service fluid study. MOWS 215425 1-4 Section 2 ANALYTICAL METHOOOLOGY BACKGROUND - Th* Isomer-specific analysis of tract conctntrat loot of PCDFs and PCOOs In PCB- containing fluids Is a challenging problem. The analytical difficulty Is compounded by the prtsence of Interfering substances such as PCBs or polychlori nated diphenyl ethers (PCDEs). Combined capillary gas chromatography/mats spectrometry (GC/HS) is the method of choice for such measurements. The selec tivity of high resolution GC, together with the sensitivity and specificity of mass spectrometry, yields detection limits In the low ng/g range for these compounds. Since capillary GC columns cannot handle PCB-contamlnated fluids directly, the samples must first be extracted with a suitable solvent to separate the analytes of Interest (PCBs, PCDFs, PCOOs) from the matrix. An enrichment step then removes co-extracted interferences and concentrates the PCDFs and PCDOs In the sm>1e extract. Finally, GC/MS analysis Identifies the sample extract and provides a quantitative measure of the PCDFs and PCOOs, based on the premise that GC separa tion of each chlorinated compound class can be achieved so that a unique set of mass spectral indicator Ions can'be monitored for eech of the appropriate GC retention time windows. To Improve Identification and quantification, appropriate l,C-lab*1ed Isotopic analogs of the polychlorinated dlbeuo-p-dloxlns and dlbenzofurans are added to the samples prior to extraction and analysis. `C-cont*1n1ng compounds are not present In significant quantities In naturally occurring materials and are dearly distinguishable from native compounds (1 *C) by mess spactroarftry. Spiking the samples with labeled coapounds thus provides unique Internal standards for accurst* analysis and determination of the recovery efficiency of the method. An overview of the general analytical protocol is shown In Figure 2-1. MONS 215426 2-1 Figure 2-1 Overview of Analytical Protocol PREPARATION OF CALIBRATION STANDARDS AND SAMPLES PMW I Calibration Standards. A set of cell brat Ion standards was prepared and distri buted by Radian Corporation, The uniformly labeled 1'C-standards, In Individual ampoules, were mixed with the "Radian PCDF/TCOO Cocktail" to give a calibration solution containing the following compounds at a concentration of 100 ng/ml etch In n-nonane: 2.3.7.8-TCDO 2.3.7.8-TCDF 1.2.3.7.8-PnCDF 1.2.3.4.7.8-HxCOF 1.2.3.4.6.7.8-HpCDF l)C-2,3,7-,8-TC00 >C-2,3,7,8-TC0F >C-l,2,3,7,8-PnCDF ' 'C-OCDD. Baseline 5--les. The five baseline saftples were prepared at Radian Corporation as solutions In n-hexane at a concentration of 0.25 g/c. Each sample was spiked with four i>C-1abe1ed Internal standards---C-2,3,7,8-TCOF, >C-2.3,7,8-TCOO, l|C-l,2,3,7,8-PnCDFt and 'C-0CDD--at concentrations of 100 ng/g each. In addi tion, the three Aroclors and the tr1-/tetrach1orobcnzene mixture were fortified with the native isomers 1,2,3,4,7,8-HxCDF and 1,2,3,4,6,7,8-HpCDF at concentra tions of 100 ng/g each. MOMS 21542? 2-2 The aged Mineral oil sample contained a number of netive PCDF/RCOO isomers, t level* ranging from 0 to 150 ng/g, as well as 510 wg/g Aroclor 1260 and 530 ug/g tri-/tetrechlorobenzene mixture, to simulate the composition of a "real-world* transformer filled with mineral oil (see Figure 3-2 for details). The tr1-/tetrachlorobenzene mixture was prepared by Radian Corporation to approximate a "typical* coamerclal dielectric mixture. It had the following composition: ~ 1.2.4-trlchlorobenzene 1,2,3-trlchlorobenzene 1,2,4,5-tetrachlorobenzene 1.2.3.4-tetrachlorobenzene pentachlorobenzene 51.26 13.46 2.66 28.46 4.46 The baseline samples were identified by Radian as follows: Compound Aroclor 1016 Aroclor 1242 Aroclor 1260 Tr1-/tetrachlorobenzene mixture Aged mineral oil Radianl.D,Code E729-09-01 E729-1D-01 729-11-01 E729-12-01 E729-08-01 In-Service Samples. The preliminary round-robin study was conducted by three of the participating laboratories (Including IITRI), using four dielectric fluids taken from In-service transformers and capacitors. The samples were selected by Sattelle Columbus Laboratories from a repository of 28 In-service samples obtained by following selection criteria laid down by General Electric Company. The fluids consisted of a mineral oil and three askarels. One of the askarels was taken from a transformer after 20 years of service, while a second was taken from a transformer (of a different manufacturer) after 31 years of service. The ira askarel was removed from a capacitor that had bulged, but not ruptured, while in service. The mineral oil sample contained PCBs and was taken from a transformer that had felled In service by arcing. Upon receipt at IITRI, 0.1-g samples of the four In-service fluids were spiked with 20 ng of each of the Internal standards, l)C-2,3,7,8-TC00, >C-2,3,7,8-TC0F, and >>C-l,2,3,7,8-PnC0F, and 28 ng of the Internal standard '*C-OCOO. HONS 215428 2-3 The in-service temples were identified by BatteUe as follows: Compound Transformer mineral oil Capacitor askertl Transformer eskarel (20 years' service) Transformer eskarel (31 yeses' service) BatteUe 1.0, Code ISL-4-0 ISL-l-A ISL-2-A I5L-3-A Phase II CallPrat Ion Standards. Radian Corporation prepared a set of calibration standards containing eight unlabeled components: 2.3.7.0-TCOO 2.3.7.8-TCOF 1.2.3.7.0-PnCOF 2.3.4.7.8-PnCOF 1.2.3.4.7.8- HxCOF 2.3.4.6.7.8- HxCOF 1.2.3.4.6.7.8- HpCOF OCOF and five uniformly labeled >>C standards: >>C-2,3,7,8-TCD0 1 *C-2,3,7,B-TCDF >>C-l,2,3,7,8-PnCDF 1>C-l,2,3,4,7,8-HxCQF >>C-0CD0. The standards were prepared at seven concentrations of each of the native s=Ccomponents: 2.5 ng/ml, 10 ng/ml, 40 ng/ml, 150 ng/ml, 500 ng/ml. 2,500 ng/nl. and 7,500 ng/ml. In each standard, the concentration of the *C-components was held constant at 500 ng/ml each. Betel ine Sawlat. Three baseline samples were prepared at Radian Corporation as solutions In n-hexene at a concentration of about 0.2 g/ml. Each sample was spiked with 100 ng/g of each of the five *C-compounds listed above. Each sample was also spiked with 'blind* amounts of the following unlabeled compounds: 2.3.7,B-TC00 2.3.7,B-TCOF 2,3,4,B-TCDF 1.2.3.7,B.PnC0F, 2,3,4,7,B-PnCDF ' 1,2,3,4,7,8-HxCOF 1,2,3.6.7,8-HxCOF 2,3,4,6,7,8-HxCOF 1,2,3,4,6,7,8-HpCOF OCOF In addition, the mineral oil sample was spiked with 625 jg/g Aroclor 1260 and 1250 wg/g tr1-/tetrachlorobenaenes. MONS 215429 The baseline samples were identified by Radian as follows Compound Mineral oil Aroclor 1016 Aroelor 1260 Radian 1,0. Code E729-53-01 E729-S3-02 E729-53-03 In-Service Samples. Seven In-service samples were selected by Battelle Columbus Laboratories for characterization by the participating laboratories. The fluids consisted of five eskarels and two mineral oils, all of which were prepared at a concentration of approximately ZOO g/mL In n-hexanc. Each sample was spiked at Battelle with the five isotopicaliy labeled internal standards at concentrations of 100 ng per gram of oil. The in-service samples were Identified by Battelle as follows: Compound Sample Mo. Battelle 1.0, ho. Transformer askarel Transformer askarel Capacitor Aroclor Precipitator transformer Aroclor Transformer mineral oil Transformer mineral oil Transformer askarel <1159-11-11 41159-11-12 41159-11-13 41159-11-14 41159-11-15 41159-11-16 41159-11-17 1SL-03-A ISL-OB-A ISL-27-A ISL-17-A ISL-23-0 1SL-04-0 ISL-02-A SAMPLE EXTRACTION AND CLEAMOP The extraction and cleanup procedure was based on the general method described by Albro and coworkers 112*131 for the fractionation and class determination of complex mixtures of chlorinated aromatic compounds. Briefly, the procedure consists of the following major steps: - 1. To the sample (-0,1 g), spiked with a mixture of >Clebeled PCDf and/or ROOD Internal standards, about 30 ng of 2,3,7-trl-COD Is added to serve as a "carrier" compound. 2. The saaple Is chromatographed on a gel permeation column (20 g Sephadfx LH-20), eluting with 50X methylene chlorideIn-methanol and collecting the 60-150 mL fraction, to separate halogenated aromatic compounds from any allphatlcs present. 3. About 10 uL of propylene glycol are added as a "keeper" compound, to prevent sample loss during evaporation. HONS 215430 2-5 4. After reducing the volume of the helogeneted erometlc frectlon by roto-evaporatlon, the friction 1* chrometo- orephed on en A-540 basic alumina column (20 g topped with a layer of anhydrous sodium sulfate), eluting the PCS fraction with MO ml of 2* methylene chlorlde-ln-hexane. 5. The PCOFS, PCDOs, polychlorinated quadphenyls, and less chlorinated PCBs are recovered by eluting with 20* methylene chlorlde-ln-hexane (200-220 ml). _ 6. About 10 ul of propylene glycol are added, then the fraction is concentrated to near dryness and loaded onto a 10 g EM-107B addle alumina column, eluting with 80 ml of 1* methylene chlorlde-ln-hexane, to purify the PCOF/PCOO fraction. The eluate is discarded. 7. The purified PCOF/PCOO fraction is eluted with 20* methylene chlorlde-ln-hexane (120 ml), 8. The extract Is concentrated to 25 wl linedletely before analysis after carrying out a solvent exchange into toluene. 9. If necessary, the extract Is purified further by loading It In hexene onto a small (1 cm) column of Certoopeck C mixed with Cellte 545, eluting with 2 ml hexane, 1 ml of SO* methylene chlorlde-1n-cyclohexane, and 1 ml of methylene chloride {methanol*, benzene (75:20:5). The column flow Is reversed and the PCOF/PCOO fraction Is eluted with 8 ml of toluene* GC/NS ANALYSIS Samples and standards were analyzed using combined capillary column gas chromatography/low-resolution mass spectrometry (HNGC/IW6 or GC/MS). The major elements of our analysis protocol were as follows: HAGC/LRMS --Unit mass resolution -1,000 --SP-2330, OV-17 columns Selected Ion Monitoring --Two Ion masses monitored per PCOF/PCOO congener class Identification Criteria --Coincident GC retention times --Correct Intensity ratios for monitored ions Quantification --Direct comparison with added internal standards --MS response factors from standards. MONS 215*31 2-5 The gas chromatograph. a Varlan 3700, was equipped with a 60 m * o.2S an ID fusedSilica SP-2330 (Supelco) column. Samples were injected in the splltless mode, with the column programmed from 8$*C to 250*C at a rate of 15'C/nln, and held at the upper limit for 60 min. The SP-2330 column has been shown to separate the 2,3,7,8-TCDD from all other tetra-isomers, and to partially separate the 2,3,7,8TCDF from the 2,3.4,8-TCDF |14,15|. However, It Is unable to resolve 1,2.3,7,8PnCDF from 1.2.3,4,8-PnCOF and 1,2,3,4.7,8-HxCDF from 1,2,3,4,7,9-HxCOF. These Isomers can be separated to some extent on less polar columns, suCh as OV-17 or OB-5 [14,15!. An example of the chromatographic resolution obtained using the SP-2330 column is shown In Figure 2-2 for 2,3,7,8-TCDO and several closely eluting tetradloxln isomers. The plot clearly shows that this column is Isomer-specific for 2,3,7,8TCOO. Figure 2-3 shows the separation obtained between the 2,3,7,8- and 2,3,4,8TCDF isomers. Here the resolution Is not at good as In the previous case, but it is nonetheless sufficient to permit quantification of the two Isomers in most cases. All data were acquired by software-controlled multiple Ion detection using two ion masses from the molecular ion cluster for each of the PCOF and PCDO levels of chlorination. Two Ions were also monitored for each Internal standard. This permitted the calculation and comparison of the isotope ratios with their theoretical values to verify their identity. Several additional ton masses were monitored to indicate possible Interferences from chlorinated diphenyl ethers El,17|. The presence of these compounds could give rise to fragment ions with the same masses as those monitored for the PCOFs. Therefore, the absence of a co response for the diphenyl ether Ion mass when a PCDF compound snows a positive response was taken to Indicate that the signal for the PCDF was "real." The massas that were monitored, along with the theoretical isotope ratios, are listed In Table 2-1. In order to obtain maximum sensitivity and selectivity during an experiment, the run was divided into four adjoining*time windows. Each window contained up to 16 Ion masses (for specific native and labeled PCOFs and PCDDs, and chlorinated diphenyl ethers); the sample dwell times and interchannel delay times were chosen to give the most sensitive and rapid cycle time. Since the GC column used in this work did not yield complete separation of each chlorination level of PCOFs/PCDOs from the other groups In a given time window, HONS 215*32 2-7 t%x> fjHtvsis w c,33aDPtMeeee.M:st un uindon< i TITLE TCOO PC MIX #2> 1UL $PLITLSS |KjiSP273KtM2>l KU OFEPftTCtt' TCCO.rrH/7, SCAN SPEED 690.9 MSEC &C 3 SAfFLE IO CCTtllSSIOHi 000736 DATE 3-MOM-84 10 19 Figure 2-2. Selected Ion current profile for ness 322 fron SC/MS analysis of performance check solution containing TCOOs as noted In the plot, using a 60 SP-2330 fused silica capillary colunn. HONS 215433 2-8 ^MLVSIS NPME 0W*CW3,W3XTmiWl .rtI*M UB9.S UINDOM' 1 TITLE 0*0*1 TEST BIX l-2ePC^JL> 1UL .>VZ33K tl227S>>S9-E9. I30<41 OPERATOR1 CyTH<Ul32W,M2-3lC7,U>179S.WM379S SPC> 5 SAfPlE 10' DATE' 2-JPN-M e<34<4S COW ISSI ON1 r**SBCNU"CF*tPE.r**>* Figure 2-3. Selected ion current profile for sets 304 fro* GC/MS analysis of performance check solution containing TCOFs as noted on the plot, using a 60 m SP-2330 fused-slllca capillary column. HONS 215434 2.9 Tifel* t-l ANALYTICAL SEQUENCE IN 6C/MS DETERMINATION OF FCDFl AND PCDOs Tim. Tin* Start tlM SC Col can Prog *:*0 Hn) lam) 7*a*. CO f"C/afa> I '*00 It'OO 43 U Crcn 0ai i T>a* Tlaa (at) <*> 3200 192 on* Monitor** Caaiauwaa ta/11 Monitor** Haii i Nil 1 r. flgu P*COO HaOPt* H*0P| 303.9 >13.9 319.9 331.9 332.9 349.9 353.9 323.1 *02.4 305.9 317*9 321.9 333.9 339.9 >51.9 352.9 -- tetoa* Patio (Mat* :/ M*ia 21 0.22 0.27 0.72 0.72 0.41 0.41 1.54 -- " 2 21:00 33.30 250 -- 3142 193 PnCOT PnCOT** 332.9 224.9 33--9.9 0.41 PaCOO FftCOD'* 333.9 352.9 290.9 .. 1.54 ,, HlOOF 323.4 ''C-MaCOF 313.9 MaCOf " 310.9 323.4 342.9 -- 1.23 1.23 -- MaCOO HaCOD** H*0P|* OOPf* 399.4 324.9 *02.4 *43.4 391.4 -- -- 1.23 -- -- " 1 35:50 54:00 250 -- 3250 322 MaCOf 323.4 325.4 (.23 HaCOO 319.9 393.4 1.23 MpCOP 02,4 *09.4 1.03 HaCOO OOPf* *23.4 473.4 **3.4 1.03 .. NOPC* *22.2 -- -- 4 54:00 44:00 250 ~ 3250 53* ocor *41.2 443.2 0.44 tRa. *52.2 459.2 0.44 OOPC* *49.4 421.4 511.2 0.44 -- * MaOPf, HaQPC, OOPf, KW, OOPf Oaalgnata nan-, naeta-, x'i>, Kyia-, jnt aacac-ioro* )**<:* I trttara, raaaat'iaiy. Matt aontrora* *1 l*aat<T fragaaat ion (M-C0C>I. 2-10 HONS 215435 some of the ion mw were included In more thin one window to ensure thet ell Isomers froa I pirtlculir congener dess were monitored. Three major criteria were used to conflra the presence of specific PCOFs/PCDOs In the samples analyzed: correct retention times for each PCOF/PCDO of Interest relative to the appropriate Isotoplcally labeled internal standerd(s) - Intensity ratio for M+/(H+2)+ within 10* of the theoreti cally expected ratio (see Tapia 2-1) slgnal-to-nolse response for each PCOF/PCDO of Interest greater than 2.5:1 for both ion masses monitored. Quantification. Ideally, each of the 210 separate isomers of PCDF and PCOO should be quantified using tha Instrument response of the corresponding labeled Internal standard as a reference. Since only a limited number of these standards ara available In practice, the approach generally followed Is to use an appropriate set of internal standards that Includes representative isomers from each chlorinated class of PCDFs and PCDOs. and further assume that the data obtained for these are representative of all Isomers In each group. Relative response factors were determined from the analysis of standard solutions containing, the Internal standards and Isomers representative of each chlorination class, Response factors were calculated from the following equation: RF - Ax(Q1s/A1s)Qx (2-1) where Ax - sum of the Integrated Ion abundances of the masses for the unlabeled compound Als sum of the integrated Ion abundances of the masses for the appropriate labeled compound Qls - amount of the appropriate labeled compound, ng Qx - amount of the unlabeled compound, ng. * The PCOFs and PCOOs In the sample extract are quantified by calculating the ratios of the mass spectral responses obtained for the Ions characteristic of the labeled 2-11 HOMS 15436 PCOFs/PCDDs to those of the appropriate Internal standards, corrected for differences In response fictor. The equation used for quantification was: where w weight of sample, g RF - response factor. (2-2) - Because the labeled internal standard was added before the sample was extracted and analyzed, and the internal standard was quantified at the same time as the native components, any losses of PCDO/PCDF Incurred during the analysis were accounted for by the above approach. Detection Halts, l.e., the minima detectable concentrations required to produce a signal 2.5 times the averege background signal, were calculated for each PCDF/PCDD congener class. For each class, the average width of the baseline noise-band for the native compounds and the peek height for a known concentration of the associated >]C-labe1ed analog were measured manually. Measurements of the noise band were made In a region of the selected Ion current profile that was fret of Interferences and as close as practicable in the plot to the peak for the corresponding `>C-labtled compound. The detection limit, OL, was calculated using the relationship: (2-3) where Ax height of the noise band of the selected mass for tha unlabeled compound Ais - height of the peak corresponding to the labeled compound of the same congener class. 2-12 HONS 215437 Section 3 RESULTS AHO OISCUSSION PHASE 1 Calibration and Quality Assurance Oata Before using the SP-2330 GC coluem to generate response factors end analyze the besellne end In-service sanples, the coluem performance was evelueted In terns of Its ability to resolve the Isoeiers of interest end to cover the entire tetrathrough octe- renge of compounds. The resolution of the coluem for the highly toxic 2,3,7,8-TCDD end TCDF with respect to closely eluting Isomers Is shown In Figures 2-2 end 2-3. Figure 3-1 shows the seperetlon of the tetra- through octa- congeners In a total analysis time of about 1 hour, using the teaiperature program given earlier. The Ion masses used to monitor the PCOFs and PCDOs are indicated on the plots, along with the Identity of the specific components detected. The response factors generated using the standard "Radian PCOF/TCDO Cocktail" solution described earlier are listed in Table 3-1. The solution, which was analyzed In triplicate, did not contain either the OCDF or OCDD components. OCDF was. however. Included In the "Rappe Toxic Cocktail," which was also mixed with equal amounts of the '^-labeled standards. Analysis of this solution gave an average response factor of 0.92 tO.13 for OC0F with respect to >>C-OCOO. The response factors used to calculate the concentrations of the other PCOFs end PCDOs of Interest In the Phase I baseline and in-service samples ere narked with asterisks In Table 3-1. Baseline Salles To test the various procedures, the five baseline samples were extracted and analyzed In duplicate for the native spikes listed in Figure 3-2, and for total PCOFs and PCDOs. The results obtained for the analysis of the baseline fluids art susnarlzed In Tables 3-2 through 3-6. It should be noted that the values reported for the Isomer pairs l,2,3,7,8-/l,2,3,4,8-PnC0F and 1.2.3,4,7.S-/l,2,3.4,7,9-HxCDF are shown as composite amounts, since these Isomers cannot be resolved using the HONS 215436 3-1 <d) l3C-0CW> l Figure 3-1. Selected Ion current profiles of ehroeetography perfones nee lolutlon using SP-2330 capillary coluen. (a) TCDF/TCDO/FnCOF coepontnt* In 1st tlee window; tb) HxCDF component In 2nd tlee window; (c) HpCOF component In 3rd tlee window; (d) OCOO In 4th tlee window. MONS 215439 3-2 13C-Labeled Internal . Standards (100 ng/g each] Samples 12C-Natfve SpNces (100 ng/g each] Flsurt 3-2. Swury of Baseline Singles and Spikes (Phase I) HONS 215440 3-3 Table 3-1 AVERAGE RESPONSE FACTORS FOR PCOO/PCOF CALIBRATION COMPOUNDS (PHASE I) Compound '"C-2".JT.8- TCOO 2.3.7.8- TCOD 2.3.7.8- TCOF 1.2.3.7.8- PnCOF 1,2,3,4,7,8,-HxCOF 1.2.3.4.6.7.8- HpCOF 0.98 t0.03* 1.39 t0.12 1.04 0.08 0.95 0.04 1.24 0.11 TCPF 0.86 0.07 1.21 0.01* 0.91 0.02 0.84 10.06 1.09 10.14 PnCOf 1.24 0.03* 1.13 10.03* 1.48 10.14* `K-OCPO 1.40 10.06 1.79 10.68* *Response factors used to calculate concentrations in tablet. SP-Z330 capillary column. In the case of the partially overlapping 2,3,7,8/2,3,4,8-TCOF peaks, the amounts of each present in the sables were estimated from the relative areas of the peaks. These areas were determined by dropping a perpendicular line from the valley between the two peaks to the baseline. Arocler 1011 (Radian 1.0. 729-09-01). The Aroclor 1016 sa^le was extracted and analyzed In triplicate (Table 3-2) instead of duplicate, largely because of the marked differences observed for several of the component concentrations in the first two determinations. In fact, the results show that far better agreement was observed between the second and third determinations; we are unable to explain the values obtained In the first determination. In terms of the second and third determinations, the data In Table 3-2 show that the Aroclor 1016 fluid does not contain any native PCDfs or PCOOs other than those that were spiked Into the sample before analysis. The values obtained for the spiked components, l,2,3,4,7,*8-/l,2,3,4,7,9-HxC0F and 1,2,3.4,6,7,8-HpCOF, are roughly 509 of the expected concentrations. The reason for these low recoveries is not clear, but may be due to an adverse matrix effect. HONS 215441 3-4 Teble 3-2 PCDF/PCOD CONCENTRATIONS IN BASELINE SAMPLE: AAOCLON 1016 [Redlen 1.0. 729-09-011 Component! PCDF/PCDD Totel Tetre-COF Totel Pente-COF Totel Nexe-COF Totel Nepte-COF Octe-COF Totel Tetre-COD Octe-COO Compound 2.3.7.8-Tetre-CDO 2.3.7.8-Tetre-COF 2.3.4.8-Tetre-COF 1.2.3.7.8-Pente-COF* 1.2.3.4.8-Pent*-COF* 2.3.4.7.8-Pente-COF 1.2.3.4.7.8-Hexe-CDF* 1*2,3,4,7,9-Htxe-COF* 1.2.3.7.8.9-Htxe-COF 1.2.3.6.7.8-Hexe-COF 2.3.4.6.7.8-Hexe-CDF 1.2.3.4.6.7.8-Heptt-COF Spited 100 100 Concentretlon Ino/al tiet. 1 .. Neesured let. 2 Qet^ 3 90 125 198 63 152 . 45 59 56 45 15 8 34 109 63 56 83 45 45 Ave. 92 81 76 58 * Isomers coelute on SP-2330 column. HONS 215492 3-5 Aroclor 1242 (Radian 1.0. E729-10-01). Two extractions and analysts wera performed on the Aroclor 1242 sample (Table 3-3). In general, a larger number of PCDFs nd PCDDs was observed it significantly higher levels then In the Aroclor 1016 sample. The first determination geve very good agreement between the spiked end observed levels of HxCDF end HpCOF components. The meesured velues, however, were much lower for the second detenu 1 net Ion. 8y contrest, the velues obtelned for the other tetre- end pente- COFs of Interest showed good agreement with one enother. Teblt 3-3 PCDF/PCOO CONCENTRATIONS IN BASELINE SANPLE: AROCLOR 1242 (Radian 1.0. E729-10-01] Components PCDF/PCDO Total Tetra-COF Total Pente-CDF Total Hexa-COF Total Hepta-COF Octa-CDF Total Tetra-COO Octa-COO Compound 2.3,7,8-Tetra-CDO 2,3,7,8-Tetra-COF 2,3,4,8-Tetra-COF 1.2.3.7.8-Ptnta-COF* 1.2.3.4.8-Penta-COF* 2,3,4,7,8-Penta-COF 1.2.3.4.7.8-Mexa-CDF* 1.2.3.4.7.9-Hexa-CDF* 1.2.3.7.8.9-Hexa-COf 1.2.3.6.7.8-Hexa-COF 2.3.4.6.7.8-Hexa-COF 1,2,3,4,6,7,8-Hepta-CDF Concentration (no/el_________ Spiked Neesured Pet. 1 Pet. 2 Tm 8S4 878 866 382 323 353 152 58 105 144 74 109 120 134 127 70 69 70 153 161 157 47 36 42 67 34 51 100 100 25 63 100 110 44 77 * Isomers coelute on SP-2330 column. 3-6 HONS 215443 Aroclor 1260 (Redlan 1.0. E729-11-01). The Aroclor 1260 sample (Tibia 3.4) showed a rather complex mixture of PCDFs at significantly increased levels in comparison with either Aroclor 1016 or Aroclor 1242. Here, the agreement between the two determinations was generally very good. The values obtained for the two spiked HxCOF and HpCDF components are six to seven times higher than the expected values, probably as a result of the levels of these compounds which, together with the other PCOFs listed in the table, are present as contaminants in the Aroclor 1260. These data emphasise the point that the Aroclors, especially Aroclor 1260, Table 3-4 PCOF/PCDD CONCENTRATIONS IN BASELINE SAMPLE: AROCLOR 1260 (Sample No. E729-U-OU Components PCOF/PCOD Total Tetra-COF Total Penta-COF Total Hexa-CDF Total Hepta-CDF Octa-COF Total Tetra-CDD Octa-CDO Compound 2,3,7,6-Tetra-CDO 2.3.7.8- Tetra-COF 2.3.4.8- Tetra-COF 1.2.3.7.8- Penta-COF* 1.2.3.4.8- Penta-COF* 2.3.4.7.8- Penta-COF 1.2.3.4.7.8- Mexe-COF* 1.2.3.4.7.9- Hexa-CDF* 1.2.3.7.8.9- Hexa-CDF 1.2.3.6.7.8- Htxa-COF 2.3.4.6.7.8- Htxa-COf 1.2.3.4.6.7.8- Hepta-COF Concentration (ng/g) : Spiked set. i MewasEur/e3d--jvr 629 1039 1372 1810 3833 59S 1034 1493 2159 4474 2 612 1037 1433 1985 4154 145 144 145 48 47 48 236 225 231 89 86 88 568 634 601 626 742 684 * Isomers coelute on SP-2330 column. MOMS 215444 3-7 contain considerable eeognts of PCDFs, despite the fact thit they were "fresh* fluid* that htd not teen eny service In electric utility equipment. Tr1-/tetrach1orobenzenes (Radian 1.0. E729-12-01). Both measurements of the PCOFs tnd PCOOf in the trU/tatrachlorobenzene staple gave values that were in excellent agreement with etch other and with the amounts of the HxCDF tnd NpCOF coeponents spiked into the fluid (Table 3-5). The results suggest that the polychlorinated benzene matrix is relatively "clean* and dots not have a negative influence on the extraction and analysis of PCOFs and PCOOs using our procedure. Table 3-5 PCDF/PC80 CONCENTRATIONS IN BASELINE SAMPLE: CHLOROBENZENES tRadian l.D. E729-12-011 Components PCDF/PCDD Total Tetra-CDF Total Penta-CDF Total Hexa-CDF Total Hepta-COF Octa-COF Total Tetra-COO Octa-COO Compound 2.3.7.8- Tetra-COO 2.3.7.8- Tetra-COF 2.3.4.8- Tetra-COF 1.2.3.7.8- Penta-COF* 1.2.3.4.8- Penta-COF* 2.3.4.7.8- Penta-COF 1.2.3.4.7.8- Hexa-COF* 1.2.3.4.7.9- Haxa-COF* 1.2.3.7.8.9- Hexa-COF 1.2.3.6.7.8- Hexa-COF 2.3.4.6.7.8- Hexa-COF 1.2,3,4,6,7,8-Hepta-COF 100 loo 105 100 103 101 95 98 105 100 103 101 95 98 * Isomers coelute on SP-2330 column. 3-8 HONS 215445 Aged Mineral Dll (Radian 1.0. E729-08-01). The aged mineral oil sample was chosen to simulate a complex real-world situation, as Illustrated In Figure 3-2. The data In Table 3-6 show reasonably good agreement between the two determinations. In some cases, such as for 2,3,7,8-TCOO and 2,3,7,8-TCDf, both the repettablllty and accuracy wert almost 100X, while In others, notably 2,3,4,8-TCDF and i,2,3.4,7,8-/l,2,3,4,7,9-HxCOF, the agreement between the two determinations was poor. Furthermore, for 2,3.4,7,3-PnCOF and 1,2,3,4,6,7,8-HpCOF, the. measured Table 3-6 PCOF/PCOO CONCENTRATIONS IN BASELINE SAMPLE: MINERAL OIL (Radian I.D. E729-0B-01I Components PCDF/PCDO Total Tetra-CDF Total Penta-CDF Total Hexa-COF Total Kepti-CDF Octa-CDF Total Tetra-CDO Octa-COD Compound 2,3,7,8-Tetra-CDO 2,3,7,8-Tetra-COF 2,3,4,6-Tetra-COF 1,2,3,7,8-Penta-COF* 1,2,3,4,8-Penta-COF* 2,3,4,7,8-Penta-COF 1.2.3,4,7,8-Hexa-COF* 1,2,3.4,7,9-Hexa-COF* 1,2,3,7,8,9-Hexa-COF 1,2,3,6,7,8-Hexa-COF 2.3.4.6.7.8-Nexa-COF 1.2.3.4.6.7.8-Hepta-COF Concentration Spiked** Pet, 1 fisLJ 4a- 88 $2 70 172 158 165 130 284 207 24 56 40 160 430 355 393 41 43 42 40 41 41 41 20 22 20 21 100 66 32 49 140 113 111 112 100 59 45 52 120 130 281 206 100 100 24 52 38 * Isomers coelute on SP-2330 column. ** In addition, sample spiked with 510 vg/g Aroclor 1260 and 530 vg/g tr1-/tetrachlorobemenes. MOMS 215446 values were markedly lower than tha expected concantratlont. No cloarcut reasons can ba offarad for thata large discrepancies. ln-ServIcn Thraa of tha participating laboratory* undertook a preliminary round-robin study using four dielectric fluids taken from In-service transformers and capacitors. Because this represented our first attempt at characterizing the PCDF and PCOO constituents of In-service equipment, the only sped at Ion performed was for the 2,3,7,8-TCDF and TCDO. For the rest, only total congener class concentrations were measured. All of the samples were extracted and analyzed In duplicate. The results obtained are discussed In the following sections. 011-Fllled Substation Distribution Transformer (Battalia 1.0. I$1-4-0). Semple ISL-4-0, a mineral oil fluid containing 100 g/g PC8s, was taken from a substation distribution transformer. The transformer had failed In service as a result of major arcing. Analysis of this sample did not reveal the presence of any PCOFs or PCDDs (Table 3-7). Table 1-7 PC8F/PCD0 CONCENTRATIONS IN IN-SERVICE SMTUl OIL-FILLED TRANSFORMER I Battelle 1.0. ISL-4-01 Components PCDF/PCOO Total Tetra-COF Total Penta-COF Total Hexa-CDF Total Hepta-COF Octa-COF Total Tetra-CDO Octa-COO Compound 2.3.7.8- Tetra-CQO 2.3.7.8- Tetra-CDF nd nd nd nd nd nd nd nd nd Oetection Limit nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd 10 18 n 7 33 10 33 nd not detected 3-10 HONS 215447 Askarel Capacitor (Battelle t.D. ISL-l-A). Staple ISt-l-A wet taken from capacitor containing Aroclor 1242 which had failed In service. It had bulged but did not rupture, and showed evidence of burnt paper on the inside. Again, analysis of the fluid did not reveal the presence of detectable levels of PCOFs or PCDDs (Table 3-S). Table 3-8 PCDF/PCDD CONCENTRATIONS IN IN-SERVICE SAMPLE: ASKAREL CAPACITOR IBattelle I.D. ISL-l-A] Coaponents PCDF/PCDO Total Tetra-CDF Total Penta-CDF Total Hexa-COF Total Hepta-COF Octa-CDF Total Tetra-CDD Octa-COD Compound 2,3,7,8-Tetra-CDO 2,3,7,8-Tetra-COF Concentration (a/a) Pet. 1 6etTT Ave. nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd nd- nd- nd nd nd nd nd nd Detection Halt 1 10 9 6 4 1 4 nd not detected Askarel toed Center Network Transformer (Battalia l.D. ISL-2-A). Staple ISL-2-A was taken froa a load center network transformer that hed been In service for 20 years, and contained typical proportions of Aroclor 1260 and tr1-/tetrach1orobenienes. The results of the analysis (Table 3-9) revealed substantial concentre tlons (ug/g-levels) of all of the congener classes, especially HxCOFs, Further more, the agreement observed between the two sets of data was gulte satisfactory 3-11 MOMS 215448 tnd did not Indicate any analytical difficulties thit could be ascribed to the Mtrlx, The values shown here are * 111 Ur to those obtained earlier for the *fresh* tuple of Aroclor 1260 (Ttble 3*4), which had not seen any electric utility service. Asktrai Load Center network Trantfoner (Nettalle 1.6. ISL-3-A), Sample ISL-3-A contained 70S Aroclor 1260 , 29X trlchlorobenienes, and IX tetrachlorobeiuenes. The fluid was taken froa a transformer that had been In service for 31 years. Here, the concentrations for the PCOFs (Table 3-10) were significantly higher than for the previous Aroclor 1260-contalnlng transformer (ISI-2-A), especially with respect to the hexa-, hepta-, and octa-CDFs. This transformer bed been In service for 11 years longer than ISL-2-A. Table 3-9 PCDF/PCOD CONCENTRATIONS IN IN-SERVICE SAMPLE: ASKAftEL LOAO CENTER NETWORK TRANSFORMER Ifattelle 1.0. 1SL-2-A) Components ' PC0F/PC00 Total Tetra-CDF Total Penta-COF Total Hexa-COF Total Hepta-COF Octa-COF Total Tetra-COO Octa-COO Compound 2,3.7.8-Tetra-COO 2,3,7,B-Tetra-COF ^ Concfntratlon {ng/flj Det. l Pet. 2 Ave 962 2163 3247 1341 162 nd nd 549 921 1436 1844 962 nd 48 756 1542 2342 1593 562 nd nd nd nd 281 141 211 Detection Limit (iw/el 3 2 3 1 7 3 7 nd not detected 3-12 HONS 215*49 Table 1-10 PCOF/PCDD CONCENTRATIONS IN IN-SERVICE SAMPLE: ASKAREL LOW CENTER NCTORX TRANSFORMER {Battel1e 1.0. 1SL-3-A) Components PCDF/PCOD Total Tetra-COF Total Penta-CDF Total Hexa-COF Total Hepta-COF Octa-CDF Total Tetra-COD Octa-COO Concentration fno/ol Pet7 1 Pet, 2 Uva. 210 2,940 10,96$ 12,931 5,723 nd nd 68 716 3,967 11,484 6,893 nd 10 139 1.833 7.466 12,208 6,308 nd Detection Limit fno/o) ~ 1 1 2 1 6 1 6 Compound 2,3,7,8-Tetra-CDD 2,3,7,8-Tetra-COF nd nd 13 8 nd 11 no not detected Conclusio- n--s * The results obtained In the Phase I pilot study for the analysis of the five baseline and four in-service samples suggested that the extraction and analysis methods developed for PCDFs end PCODs In dielectric fluids are compound-specific and sufficiently sensitive. However, Indications were that the precision attainable with these methods could possibly be improved by attention to some aspects of the extraction and sample concentration procedure. These were addressed at the start of the Phase II study. 'Analysis of the baseline samples gave reasonable agreement with expected values In those cases where comparisons were possible, l.e., In those samples In which the native levels of the PCDF-contamlnants did not mask the cowonents that were spiked Into the samples. In the case of the In-service samples, significant quantities of PCOFs were found In 20* and 31-year-old transformers containing Aroclor 1260. 3-13 HONS 215450 PHASE II Upon completion of the Phese t study, sever* 1 changes were Hde to the simple extnctlon end concentration procedures In *n effort to Improve the overall precision and efficiency of the methodology. These changes, which are Incor porated In the method summary, were based on the work of Albro et al. |JJ| and Included the following*. _ 1. Investigate the effect of reducing the length of the A-MO basic alumina column, because elution through this column was found to be the most time-consuming step. Investigate the effect of Increasing the size of the EH-1078 acid alumina column--the least time-consuming step-in an effort to Improve overall cleanup efficiency. 2. Include propylene glycol as a "keeper" and 2,3,7-trl-COO as a "carrier" to minimize losses during evaporation of solvents and on glass surfaces. 3. Replace isooctane with toluene as the final solvent to reduce losses during sample concentration due to the strong adherence of some congeners to the glass walls. 4. Investigate the efficacy of the roto-eveporator for all sample concentration steps. 5. Include ions In the SC/MS anelysls mass-menu to Indicate the presence of polychlorinated dlphenylether Interferences. Host of these steps proved to be beneficial and were Incorporated Into the sample extraction protocol that was used In Phase It of the study. Quality Assurance and Calibration Data Column Performance. Before undertaking the Phase It stage of the Investigation, the performance of the SP-2330 GC column was checked to evaluate Its ability to resolve the isomers of Interest and to cover the entire congener range of interest. Figures 2-3 and 2-4 show the resolution obtained with this column for 2,3,7,8-TCOO and TCDf from their closely elutfng Isomers. Figure 3-3 shows the plots obtained for the chromatography performance standard, which was prepared at Battalia, to further evaluate the system and establish the order of elution on the SP-2330 fused-slllca column. To this mixture we added several more 2,3,7,8-substltuted PCOF/PCQO Isomers (denoted below with an asterisk) In order to establish their elution times as well. The performance 3-14 HONS 215451 TCDO TCDF _FnCDD PflCDF M HxCDD HxCDF FnCDF Figure 3-3. Selected Ion current profiles of chroaetogrepfty perforaence solution using S P-2300 ctpllltry coluan. (e) TCDO/TCDF/PnCOD/PnCDF coaponents In 1st tlae window; (b) HxCDO/HxCDF/PnCOF coaponents In 2nd tlae window; (c) HpCDO/HpCDF/HxCOF coaponents In 3rd tleo window. 3-15 HONS 215452 nlxturt was prepared with the following compounds at a conctntratIon of about 200 pg/uL tach In n-dectne, and tha ordar of tlutlon 1i shown bel*. Itoner 1,3,6,8-TCOO 1,2,3,4-TCOD 2.3.7.8-TCOO 1.2.8.9-TCDO 1,3.7,8-TCOF 2.3,4.8-TCOF 1.2.3.8-TCOF 2,3,7,8-TCOF 3.4.8.7-TCOF flutlon Order First tetra-COO elutar Co-elutar Last elutar Second tatra-COF elutar Co-alutar Co-eluter? " Second-last tatra-COF elutar 1.2.3.7.8-PnCOO* 1.2.4.8.8-PnCOF 1,2,3,4,8-PnCDF 1,2,3.7,8-PnCOF 2,3,4,7,8-PnCDF 2,3,4,8,7-PnCOF Second penta-COF elutar Co-eluter Co-eluter Last penta-COF elutar 1.2.3.4.7.8-HxCOO* 1.2.3.6.7.8-HxCDD* 1.2.3.7.8.9-NxCOO* 1,2,4,8,7,8-HxCOF 1.2.3.4.7.8-MxCOF 1.2.3.8.7.8-NxCOF 1.2.3.7.8.9-HxCDF* 2,3,4,8,7,8-HxCOF Second last elutar Second hexa-COF elutar Last hexa-COF elutar 1,2,3,4,8,7,8-HpCDO* 1,2,3,4,6.7,8-HpCDf* 1,2,3,4,7,8,9-HpCOF* Last elutar First elutar Last elutar. Calibration Data. Tha calibration standards prapartd by Radian wart usad to ganarate flva sats of calibration curvas and rasponsa factors for tha eo^ounds of Intarast. For exaeple, tha salactad Ion currant profllas obtalnad for tha Injectlon of a 2 uL aliquot of tha ISO ng/aL standard solution art shown In Appendix A (Flgura A-l), along with a listing of the raw data for each aass, tlae centroid, concentration, and corresponding peak area. The aasses used to aonltor for tha coa^onants of Intarast u wall as the aaasurad peak areas are shown on tha plots. Tha calibration curvas are shown In Appandlx 8, Figures 8-1 through 8-9. No data ware obtained for the 2.5 ng/aL and 7,500 ng/aL standards, since they gave responses that fell outside the working range of our SC/MS systea. Tha results In Figures 8-1 through 8-9 Indicate that all of the experlaentel data are wall approxlaated by straight lines, tha correlation coefficients for linear regressions ranging between 0.9920 and 0.9997. 3-16 MON5 215A53 Response factors for the compounds of Interest Mart calculated according to Eq. 2-1. The response factors generated In this way are listed In Table 3-U, along with the mean of replicate determinations (Ave.), the standard deviation (SD), the percent relative standard deviation (RSO, *), and the nuaiber of repli cate measurements performed (n). In general, the precision obtained for the indi vidual compounds at each concentration Is highly satisfactory, and Is less than 10X RSD for all of the co^wunds except OCDF. Because of chromatographic limita tions, OCDF did not give measurable peak areas for the 10 and 40 ngM standards, and the RSD for the higher native concentrations ranged between 7.IS and 16.OS. It Is Interesting to note that the response factors listed In Table 3-11 for 1.2,3,7,8-PnCDF end 2,3,4,7,8-PnCDF agree very closely with one another throughout the concentration range. This agreement Is expected on theoretical grounds. On the other hand, the response factors for 1,2,3,4,7,8-HxCOF and 2.3,4,6,7,8-HxCDF differ front one another by almost a factor of two. The reason for this Is that they were measured In adjacent time windows, with significantly different dwelltlme characteristics. The 1,2,3,4,7,8-HxCOF and Its Isotopic analog occur In the second time window; although the 2,3,4,6,7,8-HxCDF Isomer appears In the third window, its response factor is still calculated with respect to the labeled HxCDF In the second window. Thus, the response factor for the former Is closer to the theoretically expected value. According to the analytical scheme shown In Table 2-1, the dwell time for window 3 Is 1.91 times greater than that for window 2. This ratio is sufficient to account fully for the apparent difference In response factors for the two HxCDF Isomers shown In Table 3-11. In practice, this difference between the two values is of no real consequence since the baseline and In-service samples were analyzed under the same experimental conditions that were used to generate the response factors. Table 3-12 gives the average response factors for the compounds of interest over the entire range of concentrations measured for each analyte. The overall precision varies between 3.7X and 6.5X for all compounds, except for OCDF for which the relative standard deviation Is 16.5X. Again, this reflects the fact that this compound exhibited severe tailing on the SP-2330 column, thus decreasing the signal levels attainable and the reproducibility. Response Factor Verification bate. In order to verify the constancy of the response factors with time, the standard solutions were analyzed at regular Intervals throughout the study, and the response factors compared with the values 3-17 MONS 215454 S5rSTe SNOW 3-18 `1 10*0 1*0 ( rt wo U'i t * 0 0 00' 91*0 tro <ro tt`0 * 1 *l 9`l lt'l 9'l fO` f0' M'l * t'l 0t ' 600 *n 4 *' 10 44* 1 ! tO* ll'l *t in 60" 1 trt W' 1 *6'l 0f l*f 90"0 't t r* fl'O 0*1 4js *`l K'l til il'l ' 1 .<D*H 1'iivtri 11* CO'I *i* n't M`l *t>*h -'f*9,*,t*f t M 10*0 tn f 9*f 0*0 91*1 f 0*4 90*0 1*1 1*1 *1*1 1*1 fl'l *l"t tri fl*l 0*1 11*1 11*1 *1*1 0'l * 1 ** 1 f'l t f' too O'l ( `t *0*0 fl'l f *t WO i*i n 1*1 to*i *0*1 iO'l 01*1 u*i on in rt 1*1 i*i 0*1 i * i ll'l f ' 0*0 *0'l t rt too i*i f ** 600 1*1 00*1 M'O 44*0 U`t 01*1 jay* -9'i**'t'`l 91*1 tl*l 91*1 ' 1 * 1 D*W -V**t* in O'l l* 1 trt 0C*t in -0*`t`*l *jotomj noA% f *t *0'0 *1*1 Ol'l 1*1 1*1 trt irt t t*> 0*0 1*1 90*1 *1*1 *1*1 f 1 * on t **t WO tn tn 0*1 r i 91*1 60*1 -`*t* f 'f too iro oro *0 *'0 '0 06*0 4 ** 0*0 M'O '0 M'O *0* M'O 10*1 f ' 90** 6*0 l*0 `0 00`t fO't o'l 0BJ1 -*'t' (f) 0 a j**i W** 041 II if) aw Of W* 09 if) am os *V* 04 fO w||*Jin9 (; jo I a*4) SmnOdHO) NOIiVWnvD M3d/OOM HQJ SWUDVJ 3$MM$38 II* *l<m Tafcle 3-11 RESPONSE FACTORS FOR PCOB/PCOF CM.UMTION COMPOUNDS (pope 2 of 2) CqpNtrttlM 04 too a|M 2.V.*TCDO 0.07 *.N a.M t.H *. J.3.7.*war t.OO 1.10 .IT 1,11 1.1} A*ae SO 0)0 tl> 2,300 i*M. I.M t.os 3.2 t . 0.00 0.M *.*2 0.04 I.M 0.0* }.t 9 i.to l.7 1.17 Ml l.lt P0CDT 1.24 1.21 l.lt 1.20 1.1* 1.21 0,02 1.7 9 .0* I.M .< 1.21 1.17 A*ae SO OSO (J) 0.00 I.U J.T l.l* 0.03 1.) 9 . 0.0* 3.} 9 OiHiM factors Pad* 1.24 1.22 1.20 t.2l 1.13 V.J.4.7,0HoCOF 1.0} I.M 1.01 1.00 1.14 1.20 o.ot 2.0 3 I.M 1.27 1.13 Ml 1.17 1.10 0.04 3.4 3 1.10 l.l* 1.14 1.17 1.10 .* 0.09 4.* 3 >.)} 0.0} 2.0 3 2,).4,*.7.0MaCOF 1.01 7.07 7.09 2.0* 2.14 I.7.J.4.V.*- *CDF* Otff* 1.22 1.2* 1.27 1,2* 1.27 0.1* 0.20 0,10 0.14 7.0} 0.00 4.1 3 1.07 2.10 2.07 2.00 2.11 1.2* 0,02 1.7 3 1.23 1.23 1.32 1.2* 1.23 0.17 0.0) t.o 4 0.1* 0,1* 0.2) 0.24 2.07 0.0* 2.7 3 1.27 0.0} 2.9 3 0.21 0.0] 12,4 4 * Rmiow Iscror calcalataO altk rmpact to < * >1111111 ttor ulMltM alt* tiMict to 1*C-OC0O (! --I '^C-OCDO MIm Cvo., prlvota CMMlutloi). ot*or tilnliM otto raioatt to ,3C-1atalaO c Ava iinrIf valao; SO ttwOrt aavlatloa; HO nlIttvo ttmlTl tovI .t Ion; a anOor at Ovlltlont. All 3-19 2l9%56 kons Table 3-12 AVERAGE RESPONSE FACTORS OVER ENTIRE CONCENTRATION RANGE FOR PCDO/PCDF CALIBRATION COMPOUNDS Components 2,3,7,8-TCOO 2,3,7,8-TCOF 1,2,3,7,8-PnCOF 2,3,4,7,8-PnCOF 1,2,3,4,6,7,8-HpCOF* OCOF Ave. 0.94 l.M 1.17 1.17 1.10 2.02 1.24 0.18 JL 0.06 0.04 0.06 0.05 0.05 0.12 0.D7 0.03 BSP L%) 6.5 3.7 4.g 4.1 4.8 5.8 5.3 16.5 n 25 25 25 25 25 25 25 12 * Response factor calculated with respect to >C-1.2,3,4,7,8-H*C0F. ** Response factor celculated with respect to <C-0CD0 (assumed >C-OCDO concentration 300 ng/nL; Alan Nichols, Radian Corp., private coaunlcatlon). All other response factors celculated with respect to their C-labelad analogs. obtained during the calibration phase. The results are siMearlzed In Table 3-13 and nay be compared with the average values aentloned above. At no time did the values obtained In the verification experiments differ from the average response factors by tore than a few percentage points. This high degree of consistency indicates that the GC/MS system reealned very stable end reproducible in terns of .signal detectability and response throughout the study. ftTln . The three spiked baseline samples (Aroclor 1016, Aroclor 1260, end aged mlnerel oil) were extracted and analyzed In triplicate, using the average response factors given In Table 3-12. The results obtained are summarized In Tables 3-15 through 3-17. Each table also shows the amounts of the various PCOFs and PCDOs that were 3-20 MONS 215457 Table 3-13 RESPONSE FACTORS FROM VERIFICATION RUNS Cotroonents 2.3.7.8-TCDO 2.3.7.8-TCDF 1.2.3.7.8-PnCDF 2.3.4.7.8-PnCDF 1.2.3.4.7.8-HxCOF 2.3.4.6.7.8-HxCDF 1.2.3.4.6.7.8-HpCOF* 0CDF*+ Response Factors from Calibration Oata 0.94 1.14 1.17 1.17 1.10 2.02 1.24 0.18 0.06 0.04 0.06 0.05 0.05 0.12 0.07 0.03 Dally Response Factors per Concentration of 1JC w 0.88 1.09 1.18 1.15 1.10 2.04 1.22 0.17 0.92 1.12 1.19 1.19 1.05 1.98 1.18 0.15 RC 1.18 1.21 1.22 1.15 2.15 1.28 0.19 NC 1.11 1.22 1.21 1.08 1.97 1.27 0.23 m 0.85 1.12 1.21 1.18 1.14 2.07 1.25 0.14 * Response factor calculated with respect to >C-l,2,3,4,7,8-HxCOF. M Response factor calculated with respect to 1}C-QCOO (assuaed >JC-OCOO concentration > 300 ng/L; Alan Nichols, Radian Corp., private cornunlcation). All other response factors calculated with respect to their `'C-lebeled anal09s. NC Not calculated due to Interferences In 2,3,7,8-TCOO. spiked into the samples before analysis, as well as the average Measured values and the detection Halts estlasted using Egs. 2-2 and 2-3. Analytical datt tiers, which Identify data groups, have been used to evaluate the data and coapare the Measured results with the spiked values. Specific analytes that have been quantified using a spiked, labeled calibration standard of the same isoaerlc compound are expected to have a higher Inherent reliability and accuracy than, say, the sumed congener results. The defined tiers are listed In Table 3-14. 3-21 MONS 215458 Tier 2 3 4 Tabic 3-14 analytical qata confidence tiers Description Unambiguous; all Isomers fully resolved, and iden- tic*1 labeled standard used Compounds for which labeled Isomeric analogs used for quantification; not fully resolved 2,3,7,8-SUbStltuted compounds for which no labeled analogs are available Summed congener classes Example 2,3,7,8-TCOD OCDO 2.3.7.8-TCDF 1.2.3.7.8-PnCDF 2,3,4,6,7,8-HxCOF All HxCDF Aroclor 1016 (Radian 1.0. E729-53-02). Triplicate extractions and analyses of the Aroclor 1016 baseline sample gave high levels of precision for almost all of the compounds of Interest, except 1,2,3,4,6,7,8-HpCDF (Table 3-15). The agreement between the spiked and measured concentrations for the Tier-1 compound 2,3,7,8TCOD was excellent, and the agreement for the Tier 2 compounds was almost as good. Overall, It Is clear that the unspiked Aroclor 1016 fluid contained virtually no chlorinated furans, since the concentrations measured in this sample are adequately explained In terms of the levels spiked In before analysis, furthermore, the recoveries obtained for the labeled analogs also support the view that the Aroclor 1016 matrix Is satisfactorily handled by our extraction and analysis procedures. Aroclor 1260 (Radian I.D. E72D-53-03). The Aroclor 1260 sample (Table 3-16) shmed significantly higher levels of PCDFs than were found for the Aroclor 1016. The same trend--increasing levels of PCDF contaminants with degree of chlorination--was observed In the earlier Phase I analysis of these Aroclors. In general, the reproducibility of the values Is almost as good as was the case with the less complex Aroclor 1016. One of the measurements of the Tier 1 compound 2,3,7,8-TCOD was based on the >*C-2,3,7,8-TCDF as the internal standard. This was necessary because the >>C-2,3,7,8-TCDD peaks were unusable due to interferences. Nevertheless, the average measured concentration for 2,3,7,8-TCOO was in good agreement with the expected value. The average values obtained for the two Tier 2 3-22 HONS 215459 Tabic 3-lb PCDF/PCOO COHCCtmtATIONS IN BASELINE SAMPLE: AROCLOR 1016 I Kadi an 1.0. E729-53-WJ Coopotitut* pcor/nxo T*tl Tttrs-COF Totsl Totsi PH|nit|<-Cv0f T*tsi xjiti-eor Oet-CDf Totsl Tstrt-COO Octs-COO Cbusouiis 2,1.T,4-Ttrs-C00 2,J,7,4-Ttri-CDF 2.3,4,4-Tttrs-eW l.2.1,T,4-4ft*s-eDF* I.2.3.4,4-*mTs-CDM 2.1.4,?,4-4*nti<0F l,I.3.4,T.^os-CDf 1 ,2,l,.7,4-***>i*-C0f' l,2,l,T,4.04Wttr 1.2,1.4, T,|-mo<0F 2,1,4,4,7.4-ttoM-CDr 1.2.1.4,l.7,4-H#t*-C l,C-isboisd Coosovtd llC*2,1.7,4-TC0r IJC-1.2,l.T.l-TC0r *C-l.2,3.,T,xtaCDF 1Jc-ocoo lotus ' Oof. 1 4COT/4COO <M/ Dst. 1 1 i in 42 41 41 110 114 112 224 232 227 224 132 141 It? 144 404 124 134 T1 in 11 13 11 14 (>< 1041 w_ "SO 111 lilit 44/4) 11 2t too1 202 42t 44 2 24 1.2 13 13 111 11 42 11 41 14 41 1 1 2t 0.1 O.T 114 111 144 144 144 1 2 0.7 111 144 144 144 14T 32 2 1 2 2 t 11 1.0 t1i3l 42 ISO n4o1 44 42 11T 142 22 T4 124 143 m 404 124 104 20 2.4 Ptrcsat ftscovsry** <4 43 100 4? 4 4 101 41 41 T2 ICO 21 44 11 44 i4 44 T! 42 44 21 14 > iair coslot* Oft If*1330 cdw. " Psrtsut rocovory SstftCftiuftS troa tko r#so*s of fix 1 tc-utsrn#! itiMtrH vs. 1 JC-2.3,7,1-TCCO. 3-23 MONS 215460 T*l 3-18 KOF/KOO CONCEITTIUTIONS IN 8ASEUNC SAMPLE: MOCLOft 1280 IPtdlu I.D. E729-53-03I Coaoo*a*tt xar/Kdo Total Tatra-CD? Total Paflta-COf Total Haaa-COr Total n*ta-COf 0eta-CD7 Total T*tro-CDO Octa-COO CoaoouiU *COF/*COO l*o/ol RSO 0>K Oat. 2 0*7, 8 1 P<8 --IT -ill Ott*enort lata* 1330 1114 1447 1217 1343 13*3 10*3 1141 4V ii 3 # 2334 2243 2242 2244 49 2 120* 1072 440 1047 17# 17 222 3414 2330 I07 0* >*4 2712 30 31 *7 *4 9 f 32 *.* 2,3.7,4-Tatra-COO 2,3,7,4-Tatra-eW 2,3,4,4-Tatra-CDF III 107** 4* *7 *4 * v 33 33 04 41 47 34 4 1*4 244 334 234 247 43 14 0*4 1.1 l.2,3,7,4-*onta-CDF' 1,2.3.4.|-**nt*-CDF 2,3,4,7,4<*aftta-CDF 111 323 344 322 330 14 4 111 244 273 232 244 12 4 U1 1,2,3,4.7,4-Ho*.-CDF* i,2.3.4,7.-Hooa-CDr 1.2,3.7.4.4-Noiia-CDr l,2.3.4.7>*oo*-C0r 2,3.4 .*.7 ,*-Hoa-CCf MO *47 *21 *42 *30 34 4 271 341 3*3 371 342 12 3 34 314 2*0 2M 302 14 3 3,4 l,2.3,4.*,7,-Hata-C0r HI 347 311 444 307 42 12 2.0 1 ,CHaoai4 Ceaoooni *o/eant Nocovary*** `*C-2,3.7.4-Tt0r }iC-l,2.3.T,4-OT ! }C-l,2,3,*.7,l-ttC0l <>c-oa NC *3 101 47 NC 3 1} #3 NC 4* 32 41 NC 37 14 24 * Itoaart a* **-2330 (Olaw, * Coe#fltr*tlo* catCulMlr*t raaoact to >*C-I.3,T,*-rcOf. *trcat rOQ.*f7 aataraloa* trm nm roaaaAt* *1 tha 1 >C-Itaral itMUrll va. 1 *C>2.3.T,4-TC0D. NC Hat caicolatao M to latartaraacoo (a l*C-2.3.7,4-TCOO. 3-24 MOMS 215461 coapounds, 2,3,7,8-TCDF end 1,2,3,7,8-PnCDF, however, were considerably higher then the expected values. Although neither of these peeks we$ fully resolved froe neighboring Isoaers, It Is more likely thet this dlscrepency wes due to the coapounds present In Aroclor 1260 es contaminants. Examination of the velues obtelned for the remaining compounds of interest, elong with the totel tetrethrough octe-CDFs In Teble 3-16, Indlcetes thet the seme conclusion epplles here es well. Oesplte the reletlve complexity of the Aroclor 12(0 Mtrlx, the recoveries obtelned for ell of the lebeled enelogs, except T'C-OCDO, were quite setlsfectory. Mineral Oil (kadian I.D. E729-53-01). The dete In Teble 3-17 for the eged mineral oil sample show e high degree of reproducibility throughout, end excellent egreenent between measured end expected velues for ell compounds of Interest. This epplles not only to the Tier 1 end 2 coapounds but elso to those In Tiers 3 end 4. The reeson for this lies In the feet thet the coapounds spiked Into the alnerel oil were not subject to eny Interferences during the enelysls beceuse of e totel ebsence of PCOF contealnents from the seaple. Although the metsured precision end eccurecy were very good for the alnerel oil, the recoveries besed on the labeled PCDFs were much higher then expected, except for the '*C-2,3,7,8-TCDF, which wes 64X. Me ere unable to account for this discrepancy at present. In-Servlcw The seven samples of dielectric fluids taken from In-service transformers and capacitors were extracted and analyzed In triplicate In the saae way as the three baseline saaples. The results obtained are discussed In the following sections. Askarel load Center Network Transformer* 41111-11-11 (Battalia I.D. ISL- 03--A). Seaple 41159-11-11 was the seme transformer fluid thet was analyzed in Phase I as Saapla ISL-3-A. It consisted of 70S Aroclor 1260, 29X trlchlorobenzenes, and IX tetrachlorobenzenes. The transformer had been in service for 31 years when the fluid was removed. The analytical results are shown in Table 3-18. A cursory examination of the data shows that the precision obtained was somewhat lower than for the three baseline saaples. This presumably reflects the increase in complexity of tha seaple matrix. Oesplte this, the recoveries obtained for the labeled coapounds were generally satisfactory for the congener classes measured. 3-25 HONS 215462 Table 3-17 PCOf/PCDO CONCENTRATIONS IN BASELINE SAMPLE; MINERAL OIL |Radltn I.O. E729-S3-01) CogpOOOIttt 4COF/4COO Total T*tr*<V T*t#l Pant*-CDF Tot#! N*M-CDF Total M**t#<or Octa-CDF Total T*tra^00 Octa-COO Cotount 2,3,7,6-T#tr-CDO 2,3,7,6-T#tri-COF 2,3,4,f-T#tra-eDF T,-r*ot#-cor* l,2,l,,-P*nt#-CDF* 2,1,4,7,|-P*nt*-CDF l,2,3,i,7,*^aii#-C0f l,2.3.4,7,t-+i*-C0F4 l,2.3,7,f,6-M#>a-eDF 1,2,1.6,7 ,|-+t*.#-COF 2.3,4.6,7,t-t4*M-C0F 1,2,1,4,,7 ,-tf#ot#-COf 1*0*1*4 Cooooan* ;jc-2.J,7,-Tcor >c-i .2,i,.7.a^*cof UC-OCtt OttacttoA W/PCOO (114/61 WO Halt ~0*1. 1" Pol. 1 *v. ~s: 111 (*4/41 174 150 147 137 13 253 253 310 244 t3 471 462 467 467 31 m 400 377 166 44 If 24 17 21 21 4 17 34 10 n 24 100 100 100 74 130 131 130 142 no 2*1 10 62 175 166 400 322 17 21 21 4 17 2 64 44 10 ! 1 66 63 66 6 6 130 136 132 1 2 143 134 147 44 236 240 240 2t 62 62 62 0 0 163 163 163 32 400 377 366 40 11 Pareant 6*covry 1,3 3,0 0.6 1.3 1.7 66 7 36 6* 6 4 114 202 213 177 33 31 103 272 103 227 144 46 77 262 262 214 III 35 * liiiiri toaivt* on SP-23M eolta, Vexit racovarr Oataralt* froo to# raaaona# of th# 1 >C-mt*r*t tlwtifft *%, l lC-2,S,7,6-TCOO. 3-26 MOMS 215463 Ttbl* 3-18 PCDF/PCOO CONCENTRATIONS IN IN-SERVICE SAMPLE: LOAD CENTER NETWORK TRANSFORMER ASKAREL (S*>1e 41159-11-11; Battelle I.O. ISL-03-AJ Components PCOF/PCDO Total Tetra-COF Total Penta-CDF Total Haxa-COF Total Hepta-COF Octa-CDF Total Tetra-COO Octa-CDD Pet. 1 103 1601 5449 7035 29722 17 Compound 2,3,7,8-Tetre-CDD 2,3,7,8-Tetre-CDF 2,3,4,8-Tetra-CDF 26 29 1,2,3,7,8-Panta.CDF* 1 *2,3,4,8-Penta-CDF* 2,3.4,7.8-Penta-COF 867 118 1.2.3.4.7.8-Hexa-CDF* 1.2.3.4.7.9-Hexa-CDF* 1.2.3.7.8.9-Hexa-CDF 1,2,3,6,7,8-Hexa-COF 2,3,4,6,7,8-Hexa-CDF 3130 401 87 1,2,3,4,6,7,8-Hepta-COF 3212 `C-Labalad Compound `C-2,3,7,8-TCOF '*C-l,2,3,7,8-PCOF iC-1.2.3,4.7,8-H*CDF "C-OCOO 85 124 136 91 PCOF/PCDO (na/a) Det. 2 Pet. 3 Ave. Detection RSD Limit JL 121 fne/g) 99 1911 6789 7798 33265 20 87 1649 7194 10095 46148 15 96 1720 6477 8309 36378 8 167 913 1593 8644 17 3 9 10 14 19 24 15 60 5 10 16 984 105 3953 9 18 926 112 4086 15 10 64 21 7 33 926 59 6 112 7 6 3723 518 14 545 595 514 101 67 83 79 11 3925 4609 3915 699 Percent Recovery** 20 13 18 89 100 91 8 9 129 106 120 12 10 127 90 118 24 21 76 69 79 11 14 0.5 0.5 1 1 1 * Isomers coelute on SP-2330 column. ** Percent recovery determined fro* the response of the 'JC-1ntarnel standards vs. t>C-2,3,7,8-TC00. 3-27 HONS 215464 In general, the levels of PCOFs measured here were relatively high end, except for OCOF, were In quite good Agreement with the Mounts found In the Aroclor 1260 betelIne sample end In the ISI-3-A tuple analyzed In Phase I. This lends added support to en earlier comment that most of the PCOFs observed In such In-service samples may be accounted for In terms of the PCOFs present as contaminants In the original Aroclor, The differences between the two OCOF concentrations In Phase I and-Phase II can be explained largely In terms of the response factors used to calculate the concentrations. In Phase It the OCDF/>C-OCDO response factor was calculated from the "Reppe Cocktail.' However, some of the response factors determined from this mixture were suspect due to the manner In which the mixture had been prepared for transport and distribution to the participating laboratories. The Phase H calibration standards, which contained well-defined amounts of OCOF, yielded an OCDF/i>c-OCD0 response factor roughly one-fifth that obtained from the "Rappe Cocktail.' The high detection limit listed for this compound and OCDO, hmever, Is In large measure due to the poor chromatographic behavior of these compounds on the capillary column used In these analyses, and may account to some extent for the differences noted. Askarel Load Center network Transformer; Sample 41159-11-12 (Battel It 1.0. ISL06-A). Sample 41159-11-12 was taken from a load center network transformer that had been In service for 26 years, and contained 676 Aroclor 1260 , 306 trichlorobenzenes, and 26 tetrechlorobenzenes. Inspection of the fluid at the time that the sample was taken Indicated the presence of carbon throughout the askarel. Generally, the precision attained in the triplicate analysis of this sample was relatively high, and was significantly better than that obtained for the previous sample (Table 3-19). Although this sample appeared to share most of the major characteristics of Sample 41159-11-11, the levels of most of the PCOFs of interest were somewhat lower. This may be due to the presence of the carbon particles mentioned above, which may have acted as an absorbent and thus reduced the meesureble amount of the PCOFs In the extracteble portion of the fluid. As In the previous case, the recoveries measured In terms of the labeled compounds were very good and do not Indicate unusual matrix effects. 3-26 HONS 215465 Table 3-19 PCDF/PCOO CONCENTRATIONS IN IN-SERVICE SAMPLE: LOAD CENTER NETWORK T8ANSFORKR ASKAREL (Staple No. 41159-11-12; flatten* I.O. ISL-OO-AI Componentt PCDF/PCDD Total Tetra-COF Total Pentt-CDF Total Hexa-CDF Total Kepta-CDF Octa-COF Total Tetra-CDO Octa-COD Pt. 1 PCDF/PCDO (na/a) Pet. 2 Pet. 3 Ave. Detection RSO JL 1*1 Llalt 138 329 939 1511 7617 125 339 1002 1504 7702 22 125 309 1002 2106 9475 17 129 326 981 1707 8265 6 15 36 346 1049 20 3 5 5 4 20 13 16 66 7 Compound 2,3,7,8-Tetra-COO 2,3,7,8-Tetra-COF 2,3,4,8-Tetra-COF 31 25 27 28 3 11 10 13 14 12 2 17 1,2,3,7,8-Penta-CDF* 1.2.3.4.8-Ptnta-CDF* 2.3.4.7.8-Penta-COF 110 115 121 115 6 5 43 43 51 46 5 10 1.2.3.4.7.8-Htxa-COF* 501 549 571 540 36 7 1.2.3.4.7.9-Hexa-COF* 1.2.3.7.8.9-Nexa-COF 1,2;3,,7,8-Hexa-C0F 2,3,4,6,7,8-Nexi-COF 80 87 79 82 4 5 13 15 19 16 3 20 1,2,3,4,,7,8Hepta-COF 569 563 772 635 119 19 1 0.5 1 1 2 Percent Recovery** uC-L*b*l*d Coapound `C-2,3,7,8-TCOF 80 91 106 92 13 14 >*C-1,2,37,8PCDF 127 110 113 117 9 8 >C-l,2.3,4,7.8-MxCOF 109 91 116 105 13 12 l >c-ocoo 57 46 101 68 29 43 * Isoaert coelute on SP-2330 coluan. M Percent recovery determined froa the r*$pon*e of the *'C-1nternal itenderdi v*. `*C-2,3,7,fl-TCD0. 3-29 MOMS 215466 CxpiHtnr* Wit 41159-11-13 (Battalia t.O. ISL-27-A). Sample 41159-11-13 was taken fro* a capacitor, the composition and age of which were unknown. The results of the analysis of this sample are summarized in Table 3-20. The results Indicate nich lower levels for the PCDFs of Interest than found In the previous two askarel transformer fluids. As noted earlier, poor chromatographic behavior In the time window corresponding to 0CD0 and OCDF resulted In much higher detec* tlon limits for these compounds than for the remaining compounds of Interest. In this context, it 1* interesting to note that, although a peek was observed for OCDD, the value reported In the table is Just below the estimated detection limit, and thus should probably be Ignored. Once again, the recoveries estimated from the labeled compounds were quite satisfactory, and suggest that the sample matrix was compatible with our extraction method. Precipitator Transformer: Sa1e 41159-11-14 (battella I.D. ISL-17-A). Sample 41159-11-14 was taken from a 27-year-old 50 KVA precipitator transformer. The analytical results obtained are listed In Table 3-21. for this sample, the precision attained was generally of the sane order as was seen for the three baseline samples. Concentrations of the PCDFs of Interest In this sample were highest for the penta- and hexa-CDFs In contrast to the previous Aroclor 1260contalnlng In-service samples (41159-11-11 and 41159-11-12), In which the maxima for the total-COFs were "skewed* towards the hepta- and octa-CDFi. Sample 41159-11-14 also showed a value for OCDD that was approxImately equal to the estimated detection limit. As In the previous case, this Indicates that the measured value for this compound should be disregarded. Once again, the recoveries calculated from the <>C-1abeled analogs were not exceptionally high in this sample. Mineral Dll Arc furnace Transformer; Sample 41159-11-15 (Battalia t.O. ISL23-0). Sa^le 41159-11-15 was a mineral oil drawn from an arc furnace transformer that had undergone heavy electrical service for eight years. According to the data In Table 3-22, the sample was essentially free of all PCOF and PCOD components, except for smell amounts of HpCOf (Including 1,2,3,4,6,7,8-HpCOF) and OCOO. Because of the very low levels encountered, the reproducibility of the data was quite poor. The average recoveries obtained for the >C-labe1ed analogs varied between 69X and 209X, Indicating an unusual matrix effect for this sample. 3-30 HONS 215467 Ttble 3-20 PC9F/PC00 CONCENTRATIONS IN IN-SERVICE SAMPLE: CAPACITOR AROCLOR {Staple No. 41159-11-13; Settelle 1.0. ISL-27-AI Component! PC0F/PC00 Tottl Tetre-COF Tottl Pente-COF Tottl Hext-CDF Tottl Heptt-COF Octt-CDF Tottl Tetrt-COO Octt-coo Get. 1 S72 276 54 17 9 Compound 2,3,7,8-Tetrt-COO 2.3,7,8-Tetrt-COF 2,3,4,8-Tetrt-COF 27 78 1,2,3,7,8-Pentt-COF* 1,2,3,4,8-Pentt-COF* 2,3,4,7,8-Pentt-COF 23 63 1,2,3,4,7,8-Htxt-COF* 1,2,3,4,7,9-Hext-COF* 1,2,3,7,8,9-ttext-COF 1,2,3,6,7,8-Htxt-COF 2,3,4,6,7,8-Hext-COF 1,2,3,4,6,7,8-Heptt-CDf 19 6 3 9 l)C-lttoe1ed Compound l)C-2,3,7,8-TC0F l)C-l,2,3,7,8-PCOF 1 *C-1,2,3,4,7 t8-HxC0F i iC-OCOO 85 117 138 132 PCOF/PCOO fno/ol Get. 2 Pet. 3 Ave. 473 461 502 253 222 250 52 83 63 15 30 21 8 88 27 28 27 72 60 77 22 20 22 59 67 63 18 27 21 7 10 8 3 54 7 13 10 Percent Recovery** 85 88 86 132 121 123 141 106 128 104 88 108 Detection RSO Unit 1*1 (ng/9) - 61 12 27 11 17 28 8 39 42 17 10 12 45 27 46 S 23 2 27 l 31 3 32 O.S 1 0.5 1 1 22 86 19 15 22 21 * lsomert coelute on SP-2330 column. ** Percent recovery determined from the reiponee of the *>C-1nternt1 ittndtrdi v*. >>C-2,3,7,8-TCOO. 3-31 MONS 215468 Tifelt 3-21 PCDF/PCDO CONCENTRATIONS IN IN-SERVICE SAMPLE: PRECIPITATOR TRANSFORMER AROCLOR (Sample No. 41159-11-14; 8*ttelle I.D. ISL-17-A) Components PCOF/PCOO Total Tetra-COf Total Penta-CDF Total Hexa-CDF Total Hepta-CDF Oct*-CDF Total Tetra-CDO Octa-CDD Pet, l 970 9236 10598 1725 293 8 Compound 2.3.7.8- Tetra-CDD 2.3.7.8- Tetra-CDF 2.3.4.8- Tetra-COF 151 146 1.2.3.7.8- Ptnta-COF* 1.2.3.4.8- Penta-CDF* 2.3.4.7.8- Penta-CDF 1115 1772 1.2.3.4.7.8- Hexa-CDF* 1.2.3.4.7.9- Hexa-COF* 1.2.3.7.8.9- Hexa-COF 1.2.3.6.7.8- Hexa-CDF 2.3.4.6.7.8- Hcxa-COF 3802 174 1195 387 1.2.3.4.6.7.8- Hepta-CDF 728 `C-laheled Compound * )C-2,3,7,8-TCDF 1 lC-l,2,3,7,8-PC0F :>C-1.2,3,4,7t8-HxC0F nC-OCOD 101 110 115 77 Detection PCDF/PCOO fno/ol RSD Unit Pet. 2 Pet. 3 Avt. "3E: m fnq/gl 617 6589 9918 2196 626 13 576 6002 10666 2254 374 5 721 7276 10394 2058 431 217 1723 414 290 174 94 30 24 4 14 40 47 37 8 132 141 141 116 141 134 1039 1023 1059 1584 1726 1694 3711 3775 3763 187 1155 472 233 1137 $09 198 1162 456 905 884 839 Percent Recovery** 10 7 16 12 49 5 98 6 47 1 31 16 30 3 63 14 97 12 79 103 94 13 14 183 133 142 37 26 230 102 149 70 47 238 91 135 89 66 0.5 1 1 2 1 * Isomers coelute on SP-2330 column. ** Percent recovery determined fro* the response of the `*C-1nterne1 standards vs. 'JC-2,3,7,8-TCD0. 3-32 HONS 215469 Tablt 3-22 pcof/reoo concentrations in in-service win MC FURNACE TRANSFORMER MINERAL OIL ($Mplt No. 41159-11-15; Sattelle 1.0. ISL-23-0I Components PCDF/PCDD Total Tatra-CDF Total Penta-CDF Total hexa-CDF Total Hepta-CDF Octa-CDF Total Tttra-CDD Octa-CDD Cwxpound 2.3.7.8-Tetra-CDO 2.3.7.8-Tetre-CDF 2.3.4.8-Tatra-COF 1.2.3.7.8-Penta-CDF* 1.2.3.4.8-Penta-CDF* 2.3.4.7.8-Penta-CDF 1,2,3,4,7,0-Hexa-CDF* 1.2.3.4.7.9-Hexa-COF* 1.2.3.7.8.9-Hexa-CDF 1.2.3.6.7.8-Hexa-CDF 2.3.4.6.7.8-Mexa-CDF 1,2,1,4,6,7,8-Hepte-CDF 13 13 s `K-labeled Coa^ound 1JC-2,3,7,8-TCOF `C-l,2,3,7,6-PC0f 'JC-l,2,3,4,7,O*MxC0F >'C-0C00 72 137 152 109 Detection ffiHPCDf/PCOO RSD UeU fisTTSE! ISX fng/e) 16 5 9 5 $4 10 8 10 3 24 31 3 V J 0.5 7 1 24 Percent Recovery** 2 52 68 67 69 3 4 217 274 209 69 33 199 265 20$ $7 28 127 176 137 3$ 2$ 0.5 * isoawrs coelute on SP-2330 colmn. *# Percent rtcovery determined fro* the response of the vs. >>C-2,3,7,8-TCDD. >C-Internal standards 3-33 HONS 215470 Substation Distribution Transformer: Sample 41159-11-16 (Battalia I.D. ISL4*0). Sample 41159-11-16 wtt taken front a substation distribution transformer that contained 100 ppit of either Aroclor 1254 or Aroclor 1260, and had undergone major arcing and in-service failure. The results obtained from the triplicate extraction and analysis of this sample are summarized In Table 3-23. The data In the table show relatively large coefficients of variance for almost all compounds of Interest. The reasons for this lack of reproducibility are not clear, and suggest that the concentrations obtained for this sample should be treated with some caution. Recoveries here, measured In terms of the labeled analogs, also vary widely, suggesting that the analysis of this matrix ws not as successful as some of the other samples discussed earlier. Askarel Load center network Transformer: Saxcle 41159-11-17 (Battelle I.D. ISl2-A). This was the same sample that was analyzed In Phase I as ISL-2-A, and was taken from a transformer that had bean In service for 20 years and contained typical proportions of Aroclor 1260 and tr1./tetrach1oroben2enes. The results obtained are shown In Table 3-24 and may be compared with the results obtained earlier (Table 3-9). In general, agreement Is very good except In the case of OCOF. The discrepancy between the Phase 1 and Phase II OCOF concentrations can again be explained by the response factors used to calculate the concentrations (see earlier discussion). Overall, the precision attained In the most recent measurement was quite satisfactory, as were the measured recoveries across the congener range. HONS 215471 3-34 Table 3-23 PCOF/PCOO CONCENTRATIONS IN IN-SERVICE SAMPLE: SUBSTATION DISTRIBUTION TRANSFORMER OIL (Swplt No. 41159-11-16; Bettelle I.D. ISL-4-0I exponents PCDF/PCOD Totel Tetre-CDF Total Penta-COF Total Hext-COF Total Hepta-CDF Octt-COF Total Tetra-COD Qcta-CDO Oat. I PCOF/PCOO fno/ol Get. 2 oet. 3 Ave. Detection ASO Unit JL 121 Ing/ol 358 87 215 220 136 62 619 191 301 397 214 54 114 33 62 70 41 59 2S 23 28 25 3 10 37 5 Compound 2,3,7,8-Tetra-CDO 2,3,7,8-Tttra-CDF 2,3,4,B-Tetrt-C0F 3 5 1.2.3.7.8-Penta-CDF* 1.2.3.4.8-Penta-COF * 2,3.4,7,8-Penta-CDF 72 19 111 29 40 44 27 61 0.5 65 68 41 60 1.2.3.4.7.8-Hexa-COF* 1.2.3.4.7.9-Hexa-CDF* 1.2,3,7,8,9-Htxa-CDF 1,2,3,5,7,8-Hexa-COF 2,3,4,6,7,8-Hexa-COF 243 IS 87 32 65 144 151 89 59 4 8 9 6 62 21 46 51 33 65 11 19 21 11 51 1 1,2,3,4,6,7,8-Hepti-CDF 43 15 23 27 14 53 0.5 Ptrcent Recovery** ,JC-LiPeled Compound i)C-2,3,7,B-TCDF 48 68 73 63 13 21 l)C-l,2,3,7,8-PC0F 228 212 196 212 16 8 ' lC-l,2,3,4,7,8-HxCOF 260 125 165 183 69 38 uc-ocoo 182 87 90 120 54 45 * ItoMrt coelute on SP-2330 colunn. ** Percent recovery determined froe the response of the uC-Internal sttndtrds vs. iC-2,3,7,B-TCDD. HONS 215B72 3-35 Table 3-24 PCOF/PCOO CONCENTRATIONS IN IN-SERVICE SAMPLE: LOAO CENTER NETWORK TRANSFORMER ASKAREL (Sample No. 41159.11.17; Smell* 1.0. (S1.-02-A] Components PC0F/PC00 Totel Tetre-COF Total Pente-COF Total Hexa-COF Total Hepta-COF Octa-COF Total Tetra-COO Octi-COO bet. 1 431 1221 2311 1912 6046 22 Compound 2,3.7.8-Tetra-COO 2,3,7,8-Tetra-COF 2,3,4,B-Tetra-CDF 102 $0 1.2.3.7.8-Penta-COF* 1.2.3.4.8-Penta-COF* 2,3,4,7,8-Penta.CDF 320 180 l,2,3,4,7,8.Mexa-CDF* 1,2,3,4,7,9-Htxa-CDf* 1,2,3,7,8,9-Hexi-COF 1,2,3,6,7,8-Hexa-COF 2,3,4,6,7,8-Hexa-COF 1218 119 74 1,2.3,4,6,7,8-Hepta-CDF 1095 1 ^-Labeled Compound '>C-2.3,7,B.TCOF * >C-1,2,3,7,8-PCOF iC-l,2t3,at7,8-MxC0F iiC-OCOO 99 US no 77 Detection PCOF/PCOO (nafol R50 Limit Pet. 2 Oet. 3--S ve. ~J3E x*r fngfel 344 27$ 350 78 22 387 810 973 218 22 2006 1745 2021 283 14 2273 1430 1872 423 23 6846 5801 6231 547 9 23 13 19 6 28 32 8 103 71 92 18 20 39 27 39 12 30 292 237 283 42 IS 176 139 165 23 14 1131 920 1090 153 14 115 119 118 2 2 81 55 70 13 19 1270 776 1047 250 24 Percent Recovery** 79 77 85 12 14 138 120 124 12 10 139 98 116 21 18 124 59 87 34 39 0.5 2 0.5 0.5 1 * Isomers coelute on SP-2330 column. ** Percent recovery determined from the response of the ^C-Internal sttndirds vs. i>C-2.3,7,8-TCD0. 3.36 HONS 215473 Section 4 CONCLUSIONS The method thtt we have developed for the sensitive end compound-spec If1c measure ment of PCDFs end PCDDs in PCB-contamlnated dielectric fluids het proven to be reliable when tpplled to e variety of matrices and over a wide range of concentra tions. The study was designed to characterize the PC0F/PC00 content of the samples, and to assess the precision and accuracy of the procedure under carefully controlled experimental conditions. As the sample matrix increases in complexity, there Is a concomitant decrease In the overall precision attainable. Furthermore, the precision obtained for the individual Isomers is generally significantly tetter than for the total congener classes. The results obtained for the baseline Aroclor samples show a strong increase In PCDF content as the weight percent of chlorine In the formulation Increases. Thus, Aroclor 1016 gives no measurable levels of PCDF, whereas Aroclor 1260 shows significant concentrations, especially for the higher congener classes. Some of the Aroclor 1260-contalnlng in-service samples also have significant levels of PCDF contaminants, but It Is difficult to establish a strong correlation because the composition of several of the In-service fluids is not known with any degree of certainty. Nevertheless, In some Instances, it appears that the concentrations of PCDFs In the In-servlca fluids may be explained in terms of the levels present In the original Aroclor 1260 fluid. If this observation Is generally valid, It suggests that askarels that have been In use In electrical equipment do not contain higher levels of PCDFs than unused fluids. Rather, the PCOFt present may be due to the original synthesis techniques used to manufacture the askarels. Although the data generated In the current study provide some Indications that this may be the case, we do not have enough Information to draw definitive conclusions. Further work is clearly needed. This should focus on characterizing the PCOF composition of several more PC8contaminated electric fluids taken from In-service equipment, as well as unused HONS 215474 4-1 higher-chlorinated Aroclors, such as Aroclor 1254 and 1260, as a function of a9e. One* this Information Is available. It may be possible to estimate the level of PCOF contamination In electrical utility equipment on the basis of known composi tion and usafe without the need to perform analyses on Individual units each time. HONS 215475 4-2 Section $ REFERENCES 1. J. Vuceta, J. R. Marsh, S. Kennedy, l. Hlldemann, end S. Wiley. State-ofthe-Art Review; PCOOs end PCOFs In Utility PCB Fluid. Palo Alto, Calif.: Electric ^owcr Research Institute, November 1983. (5-3308. 2. A. Lee. Assessment of PCOOs end PCOFs froei PCB Transformer end Capacitor Fires. U.5. Environmental Protection Agency, May IMS. (PA/6to/s2-8$/036. 3. H, R. Buser, H. P. Bossherdt, end C. Rappe. "Formation of Polychlorinated Dlbenzofurens (PCOFs) from the Pyrolysis of PC8s." Chemosphere. vol. 7, 1978, pp. 109-119. 4. H. R. Buser. "Formation of Polychlorinated Olbenzofurans end Olbenzo-pOloxlnt from the Pyrolysis of Chlorobenzenes." Chemosphere. vol. 8, 1979, pp. 415-424. 5. P. W. O'Keefe et al. "Chemical and Biological Investigations of a Transformer Accident at Binghamton, N.Y.* Environ. Health PertPCCt.. vol. 60, 19B5, pp. 201-209. 6. C. Rappe, S. Marklund, L. 0. KJcller, P.-A. Bergqvlst, and M. Hansson. "Strategics and Techniques for Sample Collection and Analysis: Experience from the Swedish PCB Accidents." Environ. Health Persoect.. vol. 60, 1985, pp.. 279-292. 7. j. Q. vos, J. H. Koeman, H. 1. van der Maas, M. C. ten Noewer de Brauw, and R. K. de Vos. "Identification and Toxicological Evaluation of Chlorinated Dlbenzofuran and Chlorinated Naphthalene In Two Cooeerdal Polychlorinated Biphenyls." Food Cosmet. Toxicol., vol. 8, 1970, pp. 625-633. 8. 0. Hutzlnger, 6. 6. Choudhry, 6. 6. Chlttln, and l. E. Johnston. "Formation of Polychlorinated Olbenzofurans and Oloxlns Ourlng Combustion, Electrical Equipment Fires end PCB Incineration." Environ. Health Persoect.. vol. 60, 1985, pp. 3-9. 9. B. Chlttle, B. S. Clegg, S. Safe, and 0. Hutzlnger. "PCOFs and PCOOs: Detection and Quantitation In Electrical Equipment and Their Formation During the Incineration of PCBs:- Report to Environment Canada, 1979. (Cited In Ref. 8) 10. D. K. Shipp. (National Electrical Manufacturers Association, Washington, O.C.). Correspondence, June 24. 1982. (Cited in Ref. 8). 11. National Research Council of Canada. Polychlorinated Olbenzo-o-dloxlns: Criteria for Their Effects on Man and n(s Environment. Ottawa, oecember i981. Fubl TcatTon lio. TBs7^. 25 r MONS 215476 5-1 12. P. H. Albro end C. E. Parker. "General Approach to the Fractionation and Class Determination of Complex Mixtures of Chlorinated Aromatic Compounds". J. Chroaiatoor.. vol. 192. I960, pp. 155*169. 13. P. W. Albro, J. S. Schroeder. o. J. Harvan, and 6. J. Corbett. "Characteristics of an Extraction and Purification Procedure for Chlorinated 01benio-p-d1ox1ns and Olbenzofurans In Soil and liver." J. Chromatoor.. vol. 312, 1984. pp. 165-182. 14. C. Aappe. "Analysis of Polychlorinated Oloxlns and Furans." Environ. Scl. Techno1.. vol. 18, 1984, pp, 78A-90A. 15. C. Aappe, S. Marklund, M. Nygren. and A. Sara. "Parameters for Identifica tion and Confirmation in Trace Analyses of Polychlorinated Dlbenio-p-dloxlns and Dlbenzofurans." In Chlorinated Dioxins and Olbenzofurans In the Total Environment. Boston: Buttervorths, 1983. 16. L, M. Smith and J, L, Johnson. "Evaluation of Interferences from Seven Series of Polychlorinated Aromatic Compounds In an Analytical Method for Polychlorinated Dlbenzofurans and Ofbenzo-p-dloxlns In Environmental Samples." In Chlorinated Oloxlns and Olbenzofurans in the Total Environment. Boston: Buttenvorths, 19B3, " t 17. T. 0. Tiernan et al. "Factors Affecting the Reliability of Quantitative Analyses for Chlorinated Olbenzodloxlns and Dlbenzofurans In PCB-Contalnlng Transformer fluids and Their Products of Combustion." In Workshop Proceedlnos: PCB By-Product Formation. Palo Alto, Calif.: Electric Power Research Institute, July 1995. CS/El-4104. MQNS 215*77 5-2 Appendix A _ SELECTEO ION CURRENT PROFILES FOR CALIBRATION SOLUTION E7Z9-S5-05 HONS 215478 2-V ssx> ANALYSIS NTWC< itcee3,e33XPST<4aie.nisj i uas.e mihdou< TITLE. EPRI STD 4-15PCAJLlE3D-5i2UL SL>SP2338l83-2300. lKV'Mf OPERATOR' CVTI1' l,H2-3167,M3-3730.W4373e SPC* SAMPLE ID* DATE* 28-JAM-06 COmiSSION* mNi2 .!WV 4H02 13.C-2378-TCD0 fl 4306 13.C-2378-TCOO II 3419 2378-TC00 LI 1186 2378-TCW) A 942 13,C-2378-TCDF 13,C-2378-TCDF 2378-TCOF 1 I 88 16*1 10-4 2378-TCOF Figure Al. Selected Ion current profiles for nesses corresponding to PtOF end PCDp labeled end native coapounds present In calibration solution E729-55-05. Concentration of labeled coapounds * 500 ng/aL each; concentration of unlabeled coapounds a 150 ng/nt each. Peaks of Interest and aeasured peak areas (in arbitrary units) are indicated on the plots. HONS 215479 $sx> ANALYSIS NANE* OR00>OB030333EPST4010,NI8j I UB9.0 UIKXM<1 TITLE* EPRI SID 04~t90PC/tLJE95-05j2UL SLj$P2330j85-2S0D 130/Hi 1.9fCU OPERATOR* CYIR*Hl>SlaiCrll479i.H4^379t 8PC* S SMPLE ID* DATE* 20-JAN-06 I3*44* CQNN1S8I0N* IKWPIPgllffE.WUa NASS* 397.980 _ __ "5L 399.900 13C-12378-PnCDF^4055 351.900 ^49.900 13C-12378-PnCDF /\2728 ms 3C9.900 12378-PnCOF fll482 "5W5ST ^ 12378-PnCOF ~ J\*36 . NABS* 407.000 ----------------------------------------------------------------- 1------------ m*------- 573.64* - - ------------------------- --A-- ^I^l 1 T" -- St 1C`OS II'N ____A-A.. M a. 1 - I' > r t T" ! 24*1 ' Figure Al. (continued) HONS 215480 ssx> ANALYSIS NAME* OR00*E003#0333EPST4010.MISj 1 U03.0 HIKWH' 2 TITLE* EFRI SID 4-t3OPG/U.je39-09j2UL SLiSP2330J63-2360. ISD^TIj 1 9KU OPERATOR* CTO1*Ml-3200.M23l67,M33730.M4373e SPC* 3 SPWJE ID* OATE* 20-JAH-06 13-44 02 COmiSSION* IKOTCOEFRIPE.Wtlj 2 A-4 l*i T>Trn 27*M 20*00 29*0 r^rrr^T nui'Min *i*ii*im'r >*00 31*00 32*00 33*00 34*00 33-60 Figure Al. (continued) HONS 215441 ssx> ANALYSIS MAfC* 0R99*CB93eg3aBPSHei9.mSJ I U93.0 MINOCW' 2 TITLE* EPRI STD *-1S0PG'U.jE3S-9S#2U. SLjSP2330f 03-2300, lSO^tli 19KV OPERATOR' CYTN>Ml-320e,M2*3167l&i3750U*"3750 SPC 3 SAMPLE ID' DATE' 29-JAN-66 IS'44-02 COmiSSION' tM98HDUPRIFE.miJ2 Figure Al. (continued) HONS 215482 ssx> ANALYSIS NAME' OR00*GO03*8333EP$T4010.MISJ 1 1*03.0 MIKX* 3 TITLE' EFRI STD #4-130PC/U_jE33-3i2UL SLiSP2330i83-2300* 13D*T1> 1 9KU OPERATOR* CYm>Ml-3200.H2-3167,U3-3730.M4*3730 SPC* 3 SAMPLE ID* DATE* 28-JAN-06 13*44*02 COTtllSSION* HA65tOtPftlPe.HMNj2 Figure Al. (continued) HONS 219483 ssx> ANALYSIS MA* CB03,0339EPSr4610.NISjl U09.0 UIMXM* 4 TITLE* EFW STD *4-190PG/U.jE95Hj2UL SLjSP2330;O3-250O, lSVfl; 19KV OPERATOR* CYTH* l,U2-3107,U3-37T,U4-373 SPC* 3 SAMPLE 10* DATE* 20-JAH-86 13*44 02 COMMISSION* 0*Ij2 'rw'r"w ' "01V 90S l3C-OCOO /\l312 MASS* 471.770 f\13C-OCOO 1224 MASS* 409.780 -*#39.790 A-7 ocof ^ jyg- . 443.740 _ JWOS* 441^740. S7*00 99*00 99*00 00*00 01*00 02*00 03*00 04*00 09*00 06*00 Figure Al. (continued) MO NS 215484 m m sm m w m SSX- PLIST OF DR00-C003,033XPST4ei0.PECj 1 MAR 31 86 IS138*30 UERSION U09.0 PAGE1 COHC. * 1.7383E-01 * AREA 2.3S24E+82 R-2 MAUDE FOR AREA - 3.3841E-R1 COHC. - 7.76B6E-R1 * ICIOfT 3.3983E+82 R^2 MAUDE FOR WIOfT - 4.WE3C 81 nass caTnIMnoEn) area wight area conc. SOURCE^ ht. conc. type 212 1132 3 3 1 1 3 134 IRS 174 SSI 103 1SS 361 281 3 3 1 1 3 3 1 1 3 3 3 3 1 1 3 3 3 3 1 1 K/ttl ICstfl K>rri Ksi-ri tc/m IC/ttl torn ICstfl Ksttl K/ttl torn ICstfl Kst1 K/HH K-'MM ks*n icstti Kstti icstti ICstfl figure A2. Computer-generated suenary report of peeks selected end ueesured In sun>1e file EPST401Q.NIS;1 (Figure Al) fro* calibration solution 729-55-05. Pedes ere listed In order of Increasing ness, end are narked either "K/WH* or "U/HM-, where A Indicates pedes of known concentration (standards), U Indicates peaks of unknown concentration, and AM shows that the peaks were warually selected for processing. HONS 215485 ssx- plist of ome< NOR 31 96 IS* 98*38 HASS TIfC CEKTROIO 487, 441.7488 443.7488 469.7888 471.7788 36*94 68*57 68*57 61*41 61*48 9 8. 1. 1. 1. #833JEP8T4919.FECj 1 UERSION U89.8 S$X> MIT PW0CC891MC 0V9lTE K/M tom vsm K/m K/Jfi K^m Figure A2. (continued) hohs Appendix B CALIBRATION CURVES HONS Zli**? 1 C_> CO c(_j) l~ -< cz MI q: *< X -1 C o o X z' X LOG COflC Figure Bl. Calibration Curve for 2.3.7,8-TCOO with Respect to '*C-2.3.7.8-TCD0 MONS 215*88 LOG core Figure 82. Calibration Curve for 2.3,7,8-TCDF with Respect to ,JC-2.3.7,8-TC0f HONS 215489 LOG CONC Figure B3. Calibration Curve for 1.2,3,7.8-PnCOF with Respect to ,JC-1.2,3.7,8-PnCOF HONS 215490 1 CJ CD N CJ (\i < CL Ixl cl 41* < -i * < UJ Q. oa -2 LOG CONC FIqure B4. Calibration Curve for 2.3,4.7.8-PnCDF with Respect to ,sC-l,2,3,7.8-RnC0F ............... ----- wspeci to ''C-l,2.3.4,7.8-HxCDF MONS 215491 LOG PEAK AREA RATIO (12C/13C) LOG CQNC Figure 85. Calibration Curve for 1.2.3,4,7,8-HxCOF with Respect to '*-1.2.3,4.7.8-11x0* HOHS z LOG CONC Figure 86. Calibration Curve.for 2,3,4,7.8-HxCOf with Respect to ,JC-l,2.3.%7%8-H>ia>F HONS 215493 LOG CONC Figure 87. Calibration Curve for 1.2.3.4.6,7,8-MpCOF with Respect to ' ^-I^.I.^.B-HkCDF HOHS 21"'* LOG CCNC Figure B8. Calibration Curve for 1,2,3,4,6,7,6-HpCOF with Respect to ,JC-OC0O HONS 215495 1 & -c O' O' < c oo LOG CONC Figure B9, Cellbretlon Curve for 0C8F with Respect to ,5C-OCDO hohs kt Utility Transformers and Capacitors HONS 215497 Dec 1986