Document k9XpYXkXKbRRZpXzeem2Dp9ZB

T^J FIELD POINT OR DEPT. & BLDG. NO. DATE YOUR LETTER Woody Ban__________________________________Oaktree Blvd._______________________________________ FROM FIELD POINT OR DEPT & BLDG. NO James .Summers, Ashok Shah SUBJECT ALTC, Oaktree DATE THIS LETTER July 3, 1992 REVIEW OF BFG DATA ON VCM EXPOSURE. We have reviewed previous BFGoodrich work which validated ambient vinyl chloride monomer (AVCM) measurement procedures, reviewed work on mathematical models which set a requirement on residual vinyl chloride monomer, and reviewed work which measured worker exposure to AVCM in customer plants and BFG plants. T--< in <S vrr o MEASUREMENT PROCEDURES & A method which collects 250 ml air samples, absorbs the VCM (y in these samples onto activated charcoal, then uses a gas , <r\ chromatograph and flame ionization detection to measure vinyl 0 chloride is a reliable method, validated by O'Mara and Quisenberry (1>. A method which collects VCM from air over a period of time by pumping air through activated carbon is a good method, developed by Dow and quickly used by the rest of the industry. However it was not validated at BFG by Zakriski until the late 1970s or early 1980s (3] MODELS A mathematical model was developed by Ahmed & Haller for AVCM concentrations during thermal processing of PVC (3) ,~T- AVCM = fRVCMHweiaht of PVC in 8 hrs^ f fraction of VCM lost'! R T 28.8 (number of air turnovers/hr)<air volume in building) This model assumed but did not verify that VCM was instantly distributed uniformly into the whole room. There were also two unknown factors in the equation. RVCM was not measured at the time AVCM was measured, thus requiring a probability calculation of RVCM based on an average for the year and a distribution measured seven , years later. And the number of air turnovers per hour was not known at the time of the AVCM measurements, thus requiring a calculation "-O based on the above calculated probability of RVCM. The conclusion that resin must be at or below 8.5 ppm to assure that AVCM < 0.5 ppm is weak. Another mathematical model was developed by O'Mara, et al, for AVCM under various conditions of storage (4,5>. AVCM = (1.97X10-7) (Temp)2-96 (Ventilation) (Loading) (RVCM) (time)31 - 0.66 Ln time for 5 < time < 170 hours IX 00 CJ o o 0* o-.. 7/?5 LITHO IN U.S A. PRIVILEGED AMD CONFIDENTIAL ATTORNEY-CLIENT COMMUNICATION AVCM = (0.553) (Ventilation)"1'29 (Loading)1'02 (RVCM)1'2 (time)"-"* for time > 170 hours where porosity is 0.2. The model is good only for suspension PVC. Experimental data showed that fans were needed to assure the assumed uniform distribution in the building. Air turnover was assumed to be 0.5-0.6/hour, without verification. Verification showed actual AVCM 2-4 times the AVCM expected from the model. The conclusion that resin must be at or below 8.5 ppm to assure that AVCM < 0.5 ppm is weak. WORKER EXPOSURE Personal monitoring of AVCM, where RVCM is also known, indicates in several instances that, if the PVC contained 8.5 ppm, then the worker would have been exposed to greater than 0.5 ppm AVCM (fi). Data Johnson Plastics Mixer operator Mixer operator helper Extruder operator feeder Andersen Window forman opening rail car mixer operator Cooley Inc. Materials handler Hopper feeder Extruder forman Extruder operator RVCM, ppm AVCM From model? If RVCM = 8.5; Calculated AVCM <1 Cl <1 0.55 0.55 <1 Cl C.1 Cl 0.4 0.06 0.1 0.46 0.23 0.06 0.12 0.31 0.2 >3.4 >0.5 >0.9 7.1 3.6 >0.5 >1.0 >2.6 >1.7 The problem VCM levels are concentrated on powder mixing operations and calendering, where large surface areas of melt are exposed. INFORMATION NEEDED This review points to a need to better understand the range of ventilation rates in powder mixing and melt mixing (at BFG customers), and a need to better understand local AVCM levels in relation to RVCM in operations where ventilation rates are known (in BFG powder mixing and melt compounding operations). The range of warehouse ventilation rates needs to be better known. And the warehouse (storage) model needs to be expanded to other PVC resins such as dispersion resins. V* O Nl C&1 REFERENCES PRIVILEGED AND CONFIDENTIAL ATTORNEY-CLIENT COMMUNICATION 1. M. M. O'Mara, J. c. Quisenberry, "A Method for Analyzing VCM at the Part-Per-Billion Level", BFG Staff Technical Service Report, June 7, 1974. 2. Conversation with Paul Zakriski, 6/9/92. 3. M. J. Ahmed, H. S. Haller, "Modeling of VCM Concentrations During Thermal Processing of PVC", BFG Technical Document, 3/31/83. 4. M. M. O'Mara, L. B. Crider, R. L. Bowles, C. J. Tomanek, "A Physical Model for the Diffusion of Vinyl Chloride Monomer from PVC Under Various Conditions of storage", BFG Technical Document, August 25, 1975. 5. Linwood B. Crider, Michael M. O'Mara, Robert L. Bowles, "A Model for the Diffusion of vinyl Chloride Monomer from PVC Under Various Conditions of Storage", Society of Plastics Engineers, "Safety and Health with Plastics", National Technical Conference, Denver, page 195, Nov. 8-10 1977. 6. Richard J. Kruszynski, "OSHA - NIOSH - EPA, 2nd Quarter, 1975 Update", BFG Inter-Organization Correspondence, 7/8/75. S 0O C 82B 2 BFG52516 PRIVILEGED AND CONFIDENTIAL ATTORNEY-CLIENT CO.ViMUNlCAS^NPROCESSlNG VENTILATION SURVEY Company Name_____________________________________Contact:__________________________Phone: PVC Processing Powder mixing Separate room? Size of room, ft3 Local ventilation, ft3/min Area ventilation, ft3/min or turn over/hr Rate of mixing, Ibs/hr Flexible or rigid Extrusion Separate room? Size of room, ft3 Local ventilation, ft3/min Location Area Ventilation, ft3/min or turn over/hr Rate of PVC extrusion (total), lbs/Hr Flexible or rigid_____________________ Injection Molding__________________________ Separate room? Size of room, ft3 Local ventilation, ft3/min Location Area ventilation, ft3/min nr* fnrn / hr BfG5251 7 <V7 00 oO tQ PRIVILEGED AND C0NF1DEN 11AL BJ,GATTORNEY-CLIENT COMMUNICATION pLANT VENTILATION DATA AND MONITORED EFFECTIVENESS LTRasiftipikntj: l;: Resin Type Lot No. RVCM as manufactured Date ppm by weight RVCM as loaded Date ppm by weight Rail, Truck, Bags, or silo 1 Compounding riant Powder Mixing stage Compound No. Lot No. Rigid or Flexible Powder Rate, lbs/hr. (total lines) Personal Monitoring in Pwd. Mixing Date ppm (AVCM) by weight Worker & Location * Size of powder mixing room, ft3 Separate from melt mixing room? How many lines? Ventilation, list all local & area T.rv/-A+- ? BFG52518 JO &T JO 05 CO a references BFG52519 STAFF TECHNICAL SERVICE RETORT NO. 308 B. F. Goodrich Chemical Company A (DIVISION OP THE B. P. GOODRICH COMPANY DEVELOPMENT CENTER A METHOD FOR ANALYZING VCM AT THE PART-PER-BILLION LEVEL / O' ID ^ .v Date Completed: M. M. 0*Ma*<Oid J<& Qut senberry June 5, 1974 A/ Dote Issued*. June 7, 1974 Project No.: 3690 Deportment No.: 5026 Abstract; o A gas chromatographic method based on flame ionization decoction and coconut charcoal adsorption has been developed for detecting VCM in the atmosphere at the 1 part per billion (ppb) level. The basic principle involves the adsorption of a 250cc air sample on activated chorcoal contained in the sample loop of a gas chromatograph. The loop is heated to 200C. to desorb VCM and the VCM is then analyzed chromatographlcally. A calibration curve ranging from 5 ppb to 280 ppb was experimentally determined. The curve was linear and a linear regression analysis of the data revealed a * 3.07. deviation in thu slope of the calibration cui.c. At the 20 ppb level, n 9 deviation in reproducibility (5 runs) exists. BFG52520 A complete description of the method is provided. The general method is potentially applicable at the parts-per-trillion level and lower. Pi31r Ibution: Akrop E. K. Dean E. B. Kaczcnmeycr R. W. Strassburg M* N. Johnson B^ec^syiHc R. J. Fawcett D. E. Loy C. H. Luftcr Library *R. A. Krueger *A. R. Siobort *H. (Col.) Tucker P. Zakriski J. Pausch Production Plants Plant Min;i;.*,er Technleal Socvie. ClcYolnt^d A. Vittonc B. A. DiLlddo E. J. Schm E. G. Schwaogorlc G. E. Thompson B. M. G. Zwicker-M. E. Roha F. J. Donat R. D. Scott J. L. Nelson E. W. Harrington H. Woltomato C. R, Flynn W. C. Holbrook *W. F. Blxby *W, E. Brodino *K. Greene ITC *R. J. Fulton *D. E. Wright ALTC *Mnnagors R. M. Krcagcr A. L. Schultz C.T.F. (3) ITL - R. J. Meyer L. B. Crider F. E. Krause J. G. Quisenborry M. M. O'Hara E. A. Collins E. G. DeCapita J. L, Dorsch (2) JO CvT jo 03 CO CO Xt Introduction The need to analyze vinyl chloride monomer (VCM) in the low parts per billion (ppb) range is obvious as monitoring proceeds away from the point source. A number of approaches can be used to monitor at this level but the moot obvious and best developed method is based on the use of gas chromatography. Since neither flame ionization nor electron capture are sensitive enough to measure at this level (recent udvanccs in microcouloroetrie detectors might make this a viable approach) n pro-chromatographic concentrating step is called for. The literature contains a number of publications in this area but the articles by West^) and Scheei^) provide an adequate background to this type of approach. About 3 years ago, Scheel personally described to me his work in this area. His studies on the thermal desorption rather than the solvent desorption of coconut charcoal were of interest to our pyrolysis work at that time. In fact, wo have Dreviously studied the thermal desorption of materials from molecular sieves.(3) West, in his work, described the use of gas chrom atography to analyze air pollutants which were thermally desorbed from activated alumina, silica gel and charcoal, d) The work described in our report is an extension of West's method, optimized to the analysis of VCM In air and modified for part per billion analysis. II. Analytical Approach w In essence, the sample loop of an 8-port gas chromatographic sampling valve is replaced with a 6" x stainless steel tube which contains ~-0.2g of coconut charcoal. Two of the sample ports are used to flush a 250 ml gas samplo container with helium through the charcoal bed. Once this flushing operation is complete, the charcoal bed is heated to 20Q*C. At that point, the gas sampling valve is actuated thereby diverting the chromatographic carrier gas through the carbon bed and onto the chromatographic column. In this way, the VCM is removed from the carbon bed and deposited on the chromatographic column. A typical analysis La then'carried out. The limiting factor in terms of sensitivity in this analysis is the 250 ml sample container. With this constraint, analysis is limited to >-! port- per-bill ion. III, problems Although the approach was rather straightforward, two problems arose which should be recognized by those attempting to duplicate the procedure. Tho carbon bed contains glass wool to prevent carbon movement and loss inside tho sampling leap. Wo found that unless the glass wool is thermally conditioned, ghost peaks occur in the chromatographic uuulyslz (see Appendix). Tho second problem occurred in making up standards at the low part-per-billion level. Our previous work in preparing standards involved tho paarrtt--ppoorr-million level and at that time no problems developed in those aannaalylyssese.s.However, in the part-por-billion range we noticed that Initially it was very difficult to reproduce our standards especially in the 5-20 ppb region. The problem was finally traced to residual VCM in O 2 our make-up air. An analysis of the dir near Che air compressors inlet (roof of boiler house) revealed 100 ppb of residual VCM. We solved this problem , by putting an activated carbon "bed on our air stream. However, to assure ourselves of true zero VCM grade air, all standards were prepared from cylinder air which was tested for VCM before use. Once those two problems were recognized, the analytical development of the method was relatively 6 itnplc . IV. VCM Calibration Curve All standards wore prepared by mixing o 19 ppm VCM standard with air and flowing this mixture into the 250 ml sample container. With our flow meter ranges, we arc limited (lowest) to a 2 ppb calibration standard* A 1 ppm VCM standard will bu ordered. This will allow us to prepare standards r.' ?!'. 0.1 [' hi" 1 if lhl:i 'i'Civ'-'v n.*c.*,,'rnrv. T.'*'lc l c '. Im a sur.-iviry 'C th . i ;........ ' gas chror .. to^raph i c peak aieu-s for tho.i: : i.............. c. 1 . ti\n -1 \o arc alc.> in Figure i. A typical chr- ^ to le I '' ft The j ;r'h l.vr.L : in Figure 2. It l; rt'-t t*' l'y L Lho chr.-, t.'-;ra:'h was not at the moat sensitive setting during this 5 ppb analysis. If it were, the peak would bo 4 times the size shown Ln Figure 2. Thus with a steady baseline nt maximum sensitivity, an analysis of 1 part-pcr-billion of VCM is quite reasonable. The gas chromatographic data ln Table l was hand calculated because of a ma1function in PACE (off-line computer). PACE-obtained data (see Appendix) w.\-; reliable above 20 ppb but for VCM concentrations below this, the data was not i>e 1 C-conaia tent. A linear regression analysis of the expanded calibration data in Tabic 2 (see Appendix) yielded the following equation: [VCM] ppb - 1.99 (C.C, Area) - 4.81 ppb The slope waq l . ,0ft ppb/nte/i with a correlation coefficient of 0.^90. The excellent Linearity of the cu 1 ibrut curve l; i;.J i C i vo of a number of Important phenomena. If there is any loss of VCM during the purging operation, It La independent of the concent rat I on of VCM. Also if any VCM remains on the eluarco.il, this too is concentration independent. Wo wore concerned about this latter possibility In that vo might be building a "memory" on the charcoal bed. However, repeated blank runs have boon carried out and to date this does not appear to be a problem. If very hLgh levels of VCM are deposited on the charcoal subsequent analyses could shown a "memory" to this initial deposition. *0 BFG52523 o O'? 1 3 L Table 1 Calibration Data for VCM Adsorbed On Carbon and Chromatographically Analyzed After Thermal Desorption VCM Concentration (ppb) 5 10 15 20 AO 60 120 190 280 Peak Areq^a^ 2.5 5.0 6.2 1A.9 22.A 35.1 69.7 90.0 1AG.4 Defined as "the peak height times the. peak width at half height times the range times the attenuation". PACE data was not reliable in the 5 to 20 ppb region; the PACE problem is correctable. These values are average values of 2 or more determinations. X* F ig u re l : VCH Response A f i T h e n a a l D e s o r p tio n frora C h a rc o a l -- CT------l I t Figure 2: 5 ppb Vinyl Chloride i I 5 BFG52526 C4 o Q) 6 V, Discuss log Two approaches were passible wLth this technique. One (that chosen.) involved making one carbon adsorption bed for the gas chromatograph and sampling with the 250 ml glass containers. The other approach involved making n series of carbon beds and sampling directly with these. We chose the former method for a number of reasons. Ultimately we had control over the adsorption experiment. In the field, sampling with the eampling container, neither the sampling rate or the sampling time are especially critical. For example, sampling at n rate of l or 2 llters/min. for any period of time over a few minutes is adequate. However, much tighter control over both the sampling rata and the time would be absolutely critical If sampling was done with the carbon bed directly. Wc also checked the flowmeter on one of the portable pumps with a precision flowmeter and found it to be out of calibration by 117.. In terms of sampling with the gloss vessels, this deviation is unimportant. However, It is not unimportant in sampling with the carbon bed. before any annLysis is attempted, a 11*4 error is introduced Immediately. Another demand wu placed on the analysis was rapid turnover. A recent analysis of automobile exhaust demonstrated the problem if too much material is adsorbed on the charcoal. The chromatograph must be baked out after every run. With the automobile exhaust samples, it took an hour to clean the chromatograph of the high boiling components that were trapped out of the exhaust. This, of course, would be no problem if we had a number of chromatographs available for the aualyata. In light of the number of samples that need to bo run, rapid turnover was a necessary constraint. This seemed to eliminate direct use of the carbon bed because of the other components that would bo deposited in relatively high yields. Finally, the sampling rate was a potential problem from another standpoint besides control. Very little carbon (--0.2g) is involved in the adsorption. We initially started out with 3q of carbon but this had to bo abandoned because It led to a rather dLffuso VCM chromatographic peak. To keep the adsurptLon efficiency high, the sample containers are purged at n flow rate of 100 cc/tnln. or 1/I0ih the sampling rate that would ho used in the field. Efllcicncy might hr quLto low at this Held rate especially in light of the small amount of carbon that must he used to obtain acceptable chromatographic results. Needless to say all of these problems can no overcome if analysLs below a part-per-blllIon is important. Efficiency studies can be carried out, the portable pumps cun be fitted wLth hLgh precision flowmeters, and the chromatographic ana Lysis time cun be Lengthened, for example. In terms of developing a rapid, reliable method for analyzing VCM in the L ppb region, wo feeL the method chosen was the best compromise. -g-pG525^ JO Appendix I Procedural Information Relating To The Ann 1 vain of VCH at the I ppb Level BFG52528 ?V5 /! 05 Cj. o ow 8 I., Chroma toaraph Model: Hewlett-Packard Model 571CA He : 3 7.5 cc/min. H: l Air* j factory specifications Column Temperature: inttl.nl - 80C. for 2 minutes program - 16*C./minute final - 200*C. for 2 minutes Injection Port Temperature: 150*0. Detector Temperature: 250"C. II. Column 8* X 1/8*' fornpak QS, 50/80 mesh 111, Carbon Adsorption I'-ed A 6" X 14" stainless steel column containing 0.-2306g of Fisher Scientific activated coconut charcoal (Catalog Number 5-685-B), mesh size 6-14, was prepared. Glass wool was placed at boLh ends of the bod to prevent movement. It is important that the class wool be heated to 250WC. overnight to avoid o.host peaks during analysis. The L/4" column Is then placed in a vertical position, interfaced at bath end;; with 1/8" stainless steel Cubing and connected to a gas sampling valve i.n the position occupied by the gas sampling Loop. The photograph below (Part IV) shows the gas sampling valve and carbon bed. The carbon bed was wrapped with a heating oler:vnt and a thermocouple was inserted to monitor temperature. Adsorption la carried out at 25*C. and desorption at 200* C. w*- IV. Gas Snmnlliy. Valve Any 8-port -.-.ns sampling valve can be used to Interface the carbon bed to the chromatograph. Two of the porta arc used tor the carbon bed, two of the ports are vised to purge the .sample vessel through the carbon bed and two of the port3 are vised to flush the carbon bed onto the chromatographic column. The two remaining ports are "looped" together to provide u by-paos for purging. The location of the helium l Live, gas sampling container, effluent line, etc. aro shown below and labeled accordingly. Two metering valves arc also shown in the photograph. These are necessary to close off the carbon bod during the heat up procedure. BFG52529 JO 7 JO 05 CO o JO 9 i. A - helium purge line to bottom# of sampling vessel B - sampling vessel C - metering valves for isolating carbon bed D - 8-port gas sampling valve E - carbon bed F - thermocouple on carbon bed V. Prue~ v! ui'c The fol l l np, dose r i p L i on covers the procedural aspects of analysing VCM with the carbon bod. The 250 ml sampling vessel containing vinyl chloride numwwr is Inter faced to the sampling vulvc (see photograph) the helium purge line, teed from the main line through a metering valve, is connected to the bottom of the sampling vessel. The two motoring valves on the gas sampling valve arc opened, the stopcock at the top of the campling vessel Is opened, followed by the stopcock at the bottom. The helium flow through the vessel onto the carbon bed (at 25*C.) is maintained between 100 - 110 cc/min. for 7.0 minutes. At the end of this purge time, the bottom stopcock, top Stop cock and the two metering valves on the sampling valve are closed (in the order described). At this point, the hunter on the carbon bed is turned on. In our system, it takes nppruxlnci te1y 3 minutes to reach 200C. Onco tho thermocouple on the carbon bed Indicates 2009C., the valve is "pulled" or npu*:lH'dM to brLng th' carbon bet! on stream with the chroma togrnph Ic column. BFG52530 CeO C4 o :3 *%) 10 Xho temperature on the bed is allowed to rise.to 225'C. (about 1 minute after the valve is actuated). Wo beliova this may be necessary to avoid memory on tho carbon bed, Tho hooter on the bed is then turned off. Once the valve is actuatod, the chromatographic process is initiated ((1) 2 minute delay at 80*C, (2) program to 200"C. at 16C./minute, (3) hold at 200C, for 2 minutes]. VI, Complete Calibration Data In this section, the average data contained in Table 1 will be expanded to Include all data. As mentioned above, the PACE data was not self-consistent. A summary of all data including all standards with PACE and hand calculated analyses are contained In Table 2. We discovered tho error in PACE while running the 10 ppb standards. These were repetitive runs and the two peaks were visually superioiposabie on an area basis. The two areas calculated by hand allowed a differ ence of about 37. which correlated veil with the visual observation. However, PACE data indicated n difference of '407.. From 20 ppb to 280 ppb, PACE data correlates well with tho hand calculated data. In these eases, the VCM is 90% of tho chromatograph and this may relate to the current PACE problem. The 97. deviation in rcproduc lb i l Ity previously mentioned was determined from thu 20 ppb data. While tho 120 ppb data shows a greater spread, this data may bo in error because of VCM contamination in the make-up air of the calibration gas. VII. Linear Kc * Ion Analysis Tho linear regression analysis was carried out on. all the data in Table 2. The following data relate, to the regression analysis: (1) slope: 1.99 (2) y-lntercept: -4.81 r\t Other Amilicationa The general technique described in this paper can be used in a number of other applications. For example, it can be used for analyzing pollutants on a thormal conductivity gas chromatograph where increased amounts of sample arc required, For the analysis of < ompounds that aro not sensitive to flame ionization, this technique couplet to thermal conductivity gas chromatography may be very adequate. Tho true potential of tho technique has not been evuluntod. It offers the potential to analyze quantitatively at almost any level. /T oO Q) CO 10 n Table 2 PACE and Hand Calculated Data for VCM Standardn VCM Standard, nob 5 5 10 10 20 20 20 20 20 40 15 15 60 60 12 0 120 120 190 L90 280 2 80 PACE G.C. Area fia nr^-Ca 1 dilated ^ ^ .000993 .000393 2.60 ' 2.34 .00904 .01616 5.10 4.95 .00645 .00655 .00731 .00796 .00767 14.0 13.5 13.8 17.8 15.2 .0111 22.4 .0198 .0082 6.50 6.00 .0187 .0174 34.9 35.3 . O'* 10 .034? .02 92 31.9 69.8 57.7 .O'. 4 8 . 04 3 3 93.3 86.6 . 064 1 .07H 134,2 L4 6.6 (L) Peak height x peak width x a ttcimat Ion x ranp.c. a 1.198 MV Cull scale G.C. recorder. Those values arc based on BpG52532. j\) Pr jp) Cj p