Document XKX4NgEn82xyDRLJaqox60Zx

;> r.i; .- ^i.i li'V. .tjttlA 123 es ). irp oJ. 1969), 377 19. hem. (1969). . 827, 1. 1 7. ; ,72). 20 ntam. ss, 0). ~*r T*'' jih I, .J '.igtl .ii:<Vi* ' j.77>i .L\is.A.i.. i.e ~ ViJ>J.-^.'ii >--'.`.V.rT-/J, | f Polychlorinated Biphenyl Emission from Fluorescent Lamp Ballasts C.by Donald Staief Field Studies Section UJi. Environmental Protection Agency ff'enatchee, H ash. 98801 Griffith E. Quinbv Consulting Medical Toxicologist Wenatchee, Wash. 98801 and G.David L. Spencer1 and Herbert Starr, Jr. Colorado Community Study on Pesticides Greeley, Colo. 80631 Polychlorinated biphenyls (PCBs) have been used in a vide variety of industrial and consumer applications over the past 40 years, but only recently did evidence appear that those materials were becoming widely distributed in the environment (GUSTAFSON 1970; DUSTMAN et al. 1971; RISEBROUGH et a_l. 1968). PCBs are used extensively by the electrical industry as components of certain types of transformers and condensers. L-An important use is as insulating liquid in ballasts for [ fluorescent light, fixtures. Although light fixtures with I bullasts are fourid in homes, they find greater use in offices, laboratories, or industrial plants. Certain types of ballasts, when they reach lie end of their life, may be accompanied by spontaneous leaking, smoking, and a markedly objectionable odor which tends to penetrate throughout the adjacent area. In ballasts manufactured more recently, thermal protective cut-out switches are usually incorporated^ into the units to prevent "overheating, thus preventing ^smoking, leaking of fluid, and the accompanying objectionable odor. Ballast life is ~~ normally estimated at around twelve years (IES LIGHTING HANDBOOK 1966) ; however, 'this is dependent primarily upon temperature o2 operation. TTris, shorter life or very much longer service can be obtained depending on the circumstances of operation such as the use of heat shields for deflecting heat from ballast. Although worn-out ballasts are usually replaced with newer types having thermal protective switches, a great many of the earlier units witibiut such protective devices are still in use throughout the country. Because ballast fluids are known to contain PCBs, the occasional burnout of ballasts at this laboratory has stimu lated our Interest in the problem of PCB emission, not only. from Che standpoint of potential exposure of people working nearby but also from the potential of the compounds emitted being undesirable..contaminant8_in_au.analytical chemistry laboratory where'analyses for trace amounts of relateH~coinpouhds ate~belngTfcrrled out. Prior to 1952 the liquid used as'the insulating mediunTTiTTHe types of ballasts under study was Aroclor 1254;^ however, after that It was replaced by Aroclor 1242 which has better electrical properties and a more favorable dielectric constant (GENERAL ELECTRIC COMPANY 1972). Present address: Analytical Development Corporation, 1772 Lake Woodmoor Drive, Monument, Colorado 80132. <55 Bulletin of nr4rfncnttl CoaUtnlootlon A Toxicology, Vl. 11, No. 4 1) 19J4 by Springer*Verlag Nov Yotk Inc. DSW 026346 r1^ST* '!. ,i" [wiyw!li ' ' . ' ' u ' * . ` STLCOPCB4010308 ----- :------ :---- ... 'V ^( i'.; HA >.f. AI-. v: i.-: % V l t -i.t..r f, . * r S' V "^sirr^iUru^i-jj., The purpose of the present study was to determine If to determlniF"wiTlch compounds^ ITTny,-Are Involved; to explore the analytical problems encountered In PCB analyses; and to attempt to quantitate any emission at different distances from burned out, defective, or properly operating ballasts. EXPERIMENTAL Testa for PCB emission were carried out under several experimental conditions at the Wenatchee Research Station laboratory, U.S. Environmental Protection Agency, Wenatchee, Washington. In one series, samples of air were taken at different distances from a spent ballast that had burned out under actual operating conditions. Sampling was started 5 minutes after objectionable odor and fluid leakage Indicated a burnout.. This was followed three days later by the collec tion of additional air samples at the 6ame respective locations. Several weeks later air samples were also taken at different distances from similar ballasts that appeared to be operating satisfactorily in the fluorescent light units. In a second series of tests, both burned-out and nonburned-out ballasts were heated at various temperatures in 61 x 122 x 122 cm test chamber boxes. Air samples were drawn from the teat chambers during 30-mlnute periods when the temperatures of the lower sides of the ballast casings were approximately 150, 300, and 400C to determine emission of PCBs and leakage of fluid as related to temperature under artificitl conditions. In She above tcst6 two sizes of ballasts were Involved. One was 33 cm in length and the ocher type was 26.7 cm in length. Each was designed for operation of two 96T12 or two 72T12 fluorescent lamps at 425 or 430 milliamps. All air sampling noted above was accomplished by use of portable battery-powered vacuum pumps connected in series to two midget implngers- Ethylene glycol was used as the trapping solvent. A flow ra.tc of 2.83 liters per minute was employed. Samplers were operated for periods of 30 minutes for the series in test chambers amd from 2 to 6 hours following conventional burnout or during normal operating conditions. A third series of tests to detect low-level PCR emissions from fluorescent lights was carried out at the Colorado Community Study on Pesticides, Greeley, Colorado, using the nylon chiffon screen method of TESSAR1 and SPENCER (1971). This method involves exposure of an ethylene glycol-treated cloth screen in an area to be tested for the presence of chem icals in air. The screen was left in place for a sampling period of 144 hours during fluorescent light operation, thus allowing contact with a larger volume of air than was utilized ^Use of trade names is for identification purposes only and does not constitute endorsement by the U.S. Environmental Protection Agency. 456 OSW 026347 , ' , ; t. " ;i Vu'*r x**"v' - . . . ' . * ,,.' , .----^ .a.- - . STLCOPCB4010309  .V ( 1 -|/ `rt" ^ J^1 r --'l ifTl*1 ^ ;a jVi idintb ^--J it 4 V i ; I \ t I i i 4 i < 1 * I Jr * t 1 II I t n with the impinger. This made It possible to detect smaller amounts of PCBs than was possible with the impinger technique. Following a sampling period, the cloth screen is extracted with a suitable solvent to be used in chemical analysis. The present study involved exposure of one-half square meter of ethylene glycol-treated nylou chiffon cloth having mesh openings of 0.0477 sq cm, thread sire 0.0076 cm, 90 mesh, and 477. open area. This cloth was suspended from a laboratory light fixture so that its maximum distance from the fixture was 35.6 cm. The fluorescent light fixture held two i0-watt bulbs 122 cm long and was equipped with a high power factor lead log ballast 20.3 cm long, 118 volts, 60 cycles, and line current 0.85 amperes. Sample Work-up! Ethylene glycol from the implngers was transferred to a separatory funnel and the implngers were washed with water. Dilute sodium sulfate solution was added to the combined ethylene glycol-water mixture. The solution was mixed well and extracted 3 times with nanograde hexane. The combined hexane extracts were filtered over anhydrous sodium sulfate and collected In 100 ml volumetric flasks. The solutions of hexane then could either be diluted or concentrated to give an appropriate working concentration for gas chromatographic or thin layer ch roam tog raphlc analyses. Details for extraction and work-up of the extract from the nylon chiffon mesh screen are given in the paper by TESSARI and SPENCER (1971). . Gas Chromatography; Air samples obtained from the burned- out ballast and subsequent experimentally heated ballasts, as well as liquid residues from these ballasts, were analyzed by gas chromatography. A Microtek MT-220 gas chromatograph equipped with * (tritium) electron capture detector was used. Columns employed were 47. SE-30/67. QF-1 on Chromosorb W, H.P. 80/100 mcalH and 1.57. OV-17/1.957. QF-l on Chromosorb W, H.P. 100/120 rafish. Additional confirmation of PCBs from the cloth screen vas obtained by gas chromatography on a 37. 0V-1 column using a chlorine specific Coulson electroconductivity detector system. Thin Layer Chromatography: The gas chromatographic analyses were confirmed by reverse phase thin layer chroma tography according to the method of DE VOS and PEET (1971) . Uniplate TLC plates were used. Liquid residues were analyzed with appropriate dilutions, directly on the TLC plates, con currently with Aroclor standards.^ Air samples, in some cases, required concentration of the hexane extract before visuali sation could be accomplished. More specific Identification f PCBa from the cloth screen was by the thin layer chroma tography procedure according to MULH0RN et: a_l. (1971) . 2 Provided by Monsanto Chemical Co., St. Louis, Mo. 4S7 '.-ismvT-f DSW 026348 I. [. STLCOPCB4010310 ^ : /-'- ' U - / *. '" - ^ ' ` \` . * ' " ' -^ " .. . - ` t ' f'- -I ' ' / " ' ' - . ^f jjjftMn*jw rjifiirfli`VJ-l-^'/-`1'-J,~I------- -'-* * --.-:-t.X.V,-,:,.'. --- - ' 1 ~ - >' , ' - ".,.' > '-1-'* `i.-iinV.'r tut V" 1--' h.' Vk ` RESULTS AND DISCUSSION Results of analyses of the volatile emissions from a burned-out fluorescent light ballast indicated a preponderance of Aroc lor 1242 formulation of polychlorinated biphenyls. Certain other peaks characteristic of polychlorinated biphenyls were noted, but were not Identified. Analysis of samples of the dark, oily fluid which dri ppod--from--tht s -same hal last during its" time of malfunctioning yielded an average assay value of 6.2 x 10^ ppm (627.) Aroclor 1242 with no evidence of other polychlorinated biphenyls. The approximate distribution of PCBs emitted from a burned-out ballast in a laboratory room is shown in Table 1. Calculations based on results of air samples collected during a 4-1/2-hour period starting 5 minutes after detection of ballast burnout indicated a level of 0.166 mg/M^, the highest level found in this series of tests. This was at a point 1 M below the fluorescent light fixture Bnd ballast, and approxi mately nose height level for workers in the area. Air samples taken at the sane height but 2 M from the ballast showed a Maximum value of 0.046 mg/M^. At 3 to 6 M distance at the (same height the values were from 0.012 to 0.018 ng/M^. All values are below the 1.0 mg/M^ threshold limit established fffr Aroclor 1242 by the Ameriran Cnafarencj nf P.nv^rpjienta Industrial liygi.enists (1972) . Thus, unless several ball at burn out at approximately the.!same-time, -health hazajrd^to colsons exposetl^in -the-arp.rt should be mir-Lmn 1 .^_L~ ` In order t determine if there was any persistence of PCBs in the rocm after a period of 3 days,' another series of air samples were collected at the same locations as given above (Table 2). In general. Aroclor 1242 was still detectable but at very low levels'." ' One meter below the ballast the highest concehTtfaTTorT found at that time was 0.004 mg/M^, and at the 2 M distance it was 0.002 tng/M-*. At 4.5 to 6 M distance the values were 0.001 mg/M . Several weeks after the original burnout Recurred, no ECBs could~Be detected flower limit of sensitivity of test Tub "approximately 0.0005 mg/Tl^ in air samples taken during a five-hour perlodl (1019 liters per sample) at different distances from properly operating fluorescent light ballast units in the labsratory building. One of the samples was taken within 5 cm of a ballast casing having a bottom surface temperature of 72C. Bottom surface temperatures of 10 nor mally operating ballasts in the laboratory ranged from 67C to 72C. Results of analyses of artificially Induced PCB emissions flora ballasts heated at various temperatures in a test chamber are shown in Table 3. Samples 14 and 15 involved additional heAtlng of the same ballast that emitted PCBs in the laboratory as reported in Tables 1 and 2. PCB samples 16-18 and PCB 459 . u." i i / hj~i. in. ^ i i ""`''w'.t'- osu 026350 l I j ! ' ;i i ! 1} ./ 1 1 */: STLCOPCB4010311 STLCOPCB4010312 -n . > v:&m : .. ... * * '; > - -*' .. * ' /V'-f :'. - L:-r$ *.. V./ ; './ . .: ' ^ i: ' ' " *. - ' > . ' '. i : V, ' ,p -- *'; . . ' ` . i'l $ J , / " o <A s: o fsj a U) U1 ro TABLE 3 Artificially Iiwlueed ?C3 Smiaslna frca Hellaets Resulting from Heating* Air Sample sampling no. tine Description of ballast** Temp. heated ___________ Qua)__________________ __________________________________________ (C) PCBc recovered (*jr) Air concentration in test chamber (me/MJ) 14 0.5 33 cm burned-out ballast. Some leakage of fluid (during heating). Previously burned-out (ruptured) in laboratory* 15 0.5 Same as 14 300 144 400 2.400 1.696 28.268 16 0.5 Same ballast type as 14 (nonruptured) 17 0.5 Same as 16 18 0.5 Same a3 16 19 0.5 26.7 cm burned-out Instant start ballast (nonruptured) . Some leakage of fluid during heating. 20 0.5 Same as 19 21 0.5 Same as 19 150 none 300 54 400 230 150 none 300 65 400 580 eld 0.636 2.709 elc 0.765 6.833 22 0.5 Same type as 19 (nonruptured) but had not 150 none burned out. Leaked black fluid above 300C. 23 0.5 Same as 22 300 47 24 0.5 Same as 22 400 214 eld 0.553 2.521 0n hot plate In test chambers. ^Designed for use with two 96T12 or 72T12 fluorescent lamps at 425 MA without protective thermal trip devices. cIdentified as Aroclor 1242. dBelow lower limit (0.0005 m3) of sensitivity of test. V / t- ! t 7 l k\ STLCOPCB4010313 J tViin 1.'a. -^Tir'-tli' ir r < 1 >- A.;. samples 19-21 represent two additional ballasts which had previously burned out. PCB samples 22-24 represent a ballast that had been removed from a properly operating light unit that had not burned out. As can be seen in Table 3, no identifiable PCBs could be detected at 150C. At 300C the values ranged from 0.553 to 1.696 mg/H^ for Aroclor 1242 in the test chamber and at 400C the values ranged from 2.521 to 28.268 mg/M . It can also be seen in comparing these data that the heated ballast that had not burned out yielded results similar to the burned out ballasts. The above temperatures were chosen in an effort to bracket the possible temperature during which an actual burnout may occur. These rpsnlfg indicate that emission of PCBs from hnllasts that hflvp become defective and burned out occur at temperatures in the range qfITorZgreater than,-il>a-hig}iej:_teiiiperatures . us ed.in the_ experiment.. This can be considered a safety factor when compared to the average surface operating temperature of approximately 67C to 72C we found for fluorescent light ballasts that appeared to be operating normally. At the higher temperatures in the test chamber much smoke and fumes were emitted and black, thick, tar-like fluid seeped from the ballast casings. This material on cooling became hardened like tan as contrasted to the dark, oily liquid which dripped from the unit during actual burnout. Analysis of the cooled 7* tar showed no PCB content. -------- ------------------------------- In, the present study each of the samples were run con comitantly with all suspected Aroclor formulations (1242, 1248, 1254, and 1260). Although in some air samples certaLn other PCB peaks were noted, only Aroclor 1242 was reported because the few additional peaks were very small. In the study using the cloth screen technique, compounds identified as PCBs were detected in the range of 4.5 to 4.8 ji?> depending on tke method of calculation. Using the method of TESSARI and SPENCER (1971) to estimate the level of the coitaminant in air utilizing values obtained in the cloth screen tests, approximately 3 to 15 ng/M^ could be calculated as present at the sampling zone. Determination of lower levels than vene found with the impinger was possible because the cloth .screen was subjected to a larger volume of air during the test! period. ThH results of the present paper, even though based on limited testing, indicates that PCBs are emitted during burnout of certain types of ballasts contained in fluorescent light ~ xmi-trs. Even thpngh the level c Hpl-erJ-nd i n-lr -worn wal l - hel nu the level whic_h_ might-present--a -heabth -hazard-to-persons-- working in a^contamiuated area, the PCBs -aiay--be -undesirable ' contaminants In certain lftboiatarica-Hhere detection of trace amourrta~Ol related compounds is necessary. Dripping of oily materials, tars, and PCBs from ballasts onto a workbench and glassware where samples are being prepared could present even more serious problems in The-analyses of"such'samples by gas 462 "i "'*J"T'mm DSW 026353 i .. ? ft. t STLCOPCB4010314 , rr,y'A'.,;v:r. jv'. `ST. .1 . J- Results also indicate that emission from "normally operating" tluorescent light units Is very low. The levels found are Insignificant in comparison to the established threshold limit of 1 mg/M^ for Aroclor 1242 (AMERICAN CONFERENCE OF GOVERNMENTAL INDUSTRIAL HYGIENISTS 1972). It is possible that new fluorescent light ballasts may have no PCB emission and that certain older ballasts In "normally operating" Tight units may emit" even more than was indicated -in our t RTFs'; "however," this could not be clarified with the limited experimentation conducted. In recent years manufacturers have Incorporated thermal protective switches In ballasts to prevent overheating and burnout. Thu, units now used as replacements should not overheat, rupture. And emit PCBs. However, because of the long life of ballasts (12 yrs average) many older thermally unprotected units are still In operation. Where PCB contami nation may ba a serious problem, resulting from ballast burnouts, consideration should be given to replacement of such nonprotected units with thermally protected units before burnouts occur. If a burnout should occur In a laboratory where trace amounts of PCBs may cause interference in ana lytical rcsslts, all glassware and surfaces exposed to either contaminated air or to liquid emission should be thoroughly decontaminated before use. REFERENCES AMERICAN CONFERENCE OF GOVERNMENTAL INDUSTRIAL HYGIENISTS: Threshold Limit Values of Airborne Contaminants and Physical Agents with Intended Changes Adopted by ACCIH for 1972. DE VOS, R.B., and E.W. PEET: Bull. Environ. Contam. Toxicol. 6, 164 (1971). DUSTMAN, E.H., L.F. ST1CKEL, L.J. BLUS, W.L. REICHEL, and S.N. W1EMEYER: The Occurrence and Significance of Polychlori nated Biphenyls in the Environment, Transactions of the Thirty-Slxtft North American Wildlife and Natural Resources Conference, March 7-10, 1971. Washington, D.C.: Wildlife Management Institute 1971. GENERAL ELECTRIC COMPANY: The Role of Polychlorinated Biphenyls ia Electrical Equipment, 29 pp., Feb. 4, 1972. GUSTAFSON, D.G.: Environ. Sci. Technol. 4, 814 (1970). 1ES LIGHTING HANDBOOK, J.E. Kaufman, Ed., 4 ed. New York: Illuminating Engineering Society 1966. MULHERN, B.M., E. CHROMARTIE, W.L. REICHEL, and A.A. BELISLE: JAOAC 54, 548 (1971). .ow RISEBROUGH, R.W., P.R. RIECHE, and D.B. FEAKALL: Nature 220. 1098 (1968). TESSARI, J.D.; and D.L. SPENCER: JAOAC 54, 1376 (1971). :e f I in 463 f DSW 026354 \ | J I I, ) v I ,I :( l | fl I I vv . JJ \ `^y\ , STLCOPCB4010315