Document vy60DL8Gg9QxQMYnG4zw6Oqdq

DISTRIBUTION Akron R. W. Strassburg Avon Lake General J. M. Whitney D. G. Desrosiers M. D. Rider Chemical Avon Lake Technical R. D. Hardesty J. A. TePas L. B. Crider Center Brecksville R & D Center C. H. Lufter J. B. Pausch R & D Files Calvert City C. J. Orsborn E. E. Atkins Cleveland K. J. Kaminski M. E. Schmiederer W. C. Becker W. C. Holbrook J. A. Klupar J. W. Lewis Henry K. Konter C. McCrosky Long Beach R. W. Birch E. D. Hancock W. D. Robb Louisville R. R. Taylor 0. W. Robben Pedricktown 0. P. O'Keefe J. A. Kiel Corporate Environmental Service Project 8504-78 October 18, 1978 c25-(*n{ THE B.F. GOODRICH CO. NOV201978 RESEARCH LIBRARY STATUS REPORT A Comparison of Collection Devices for Vinyl Chloride at Part-per-Billion Levels: Bendix Flasher Tubes Versus Tedlar Bags by John W. Born, Paul M. Zakriski Andrew M. Fairlie, Jr. 24819001 B F GOODRICH Corporate Environmental Service 8504-78, SR, 10-18-78 A Comparison of Collection Devices for Vinyl Chloride at Part-per-Billion Levels: Bendix Flasher Tubes Versus Tedlar Bags by John W. Born, Andrew M. Fairlie, Jr., and Paul M. Zakriski SUMMARY Our study involved sampling and measuring VCM concentrations in air at three ppb levels and four relative humidities. We achieved both main purposes of the study. In the process we learned what changes in the current BFG sampling method are necessary to obtain accurate measurements. We also devised the instru mentation and procedure to accomplish such changes. The current BFG method employs PMCC's (i.e., Bendix Flasher tubes) to sample ambient air for 24 hours at 10 cc per minute. The method does not correct the measurements of VCM concentrations for environmental variables such as relative humidity, air volume sampled, and VCM load. The first purpose of our study was to show whether the current BFG method gives accurate results without such corrections. The study showed that the current results are not accurate. It shows the quanti tative effect of the environmental variables on the collection and measurement of VCM concentrations. With corrections for such variables the current BFG method would be capable of measuring ppb concentrations of VCM in air accurately. Consideration of our findings led us to devise instrumentation and a pro cedure to correct for such variables. That instrumentation prepares three PMCC calibration samples plus two PMCC monitor samples per 24-hour period. The five PMCC's all sample the same ambient air. They collect the same amount of VCM from the air. However, the three PMCC calibration samples also contain VCM "spikes" from a standard VCM permeation vial. The "spike" weights of VCM permit accurate assignment of the GC peak and concurrent calibration. The second purpose of our study was to learn whether the California Air Resources Board's method of sampling and measuring such VCM test atmospheres is practicable. It requires the gas chromatographic detection and measurement of only about 0.04 percent as much VCM per sample as the BFG method measures. Our results show that the Board's Tedlar bag method is not practicable. It is not capable of measuring VCM concentrations in the 5.0 cc samples from the bag accurately below 100 ppb. The Tedlar bag method would require concentration of the total sample to compete with the PMCC method. B F GOODRICH ZUOGT8VZ Corporate Environmental Service -2- 8504-78', SR, 10-18-78 CONCLUSIONS Our conclusions can be briefly summarized. The current BFG method of sampling and measuring property line VCM concentrations gives inaccurate results. The measurements of VCM concentrations at the property line prior to this time are low. They can serve only as a lower limit of the true concentration. It is not possible to correct the data of record to the higher and more accurate VCM concen trations. Our study showed how to correct that PMCC (i.e., Bendix Flasher tube) method to give accurate, reliable results at VCM-air concentrations of 5-40 ppb. We also showed that the California Air Resources Board's Tedlar bag method (with direct injection) is not capable of even detecting ambient VCM concentrations below about 100 ppb reliably without concentration. Our experimental results showed that temperature, humidity, volume of air sampled, and VCM weight critically affect accurate measurements of VCM concentrations. Simultaneous loadings of four PMCC's and one Tedlar bag from the same VCMair stream permitted direct comparisons. Loadings for 6 hours at 40 cc per minutes versus 24 hours at 10 cc per minute gave equivalent results. The order of decreas ing effect of the environmental variables was humidity, volume of air sampled, and VCM weight. The ambient temperature effect ranks high with the relative humidity of from 44 to 95 percent. The effects of such variables on the PMCC collection and analysis of VCM can be controlled. In field testing BFG can achieve such control best by con currently measuring the effect of the variables. That approach is much more feasible than by determining the absolute values of the variables. We have designed a method of simultaneous PMCC sampling and preparation of PMCC calibration samples. The calibration samples would collect the same test atmosphere as the field samples but would also receive a VCM "spike" of known weight. This procedure would provide accurate and reliable measurements of ambient VCM concentrations at property lines. The presence of relatively large concentrations of other organic pollutants in the ambient air could seriously interfere with reliable GC analysis. RECOMMENDATIONS 1. Display a copy of this report permanently in your file of VCM monitoring reports. It will serve as your first notice that ppb VCM measurements must include corrections for relative humidity, air volume sampled, and apparent weight of VCM collected. 2. Achieve all these corrections plus preparation of VCM calibration samples at the same time you collect the ppb VCM perimeter samples. Use the instrumen tation which the BFG Environmental Laboratory will assemble. A vacuum pump will draw ambient air through five cleaned PMCC's at 10.0 cc per minute for 24 hours. An Aalborg five-flowtube flowmeter with NRS needle valves will control the five equal flowrates. A portable calibrator (Analytical Instru ment Development, Inc. Model 340) will "spike" the ambient air for three of the five PMCC's at a standard VCM level of 40 ppb. The two remaining PMCC's will collect the "unspiked" ambient air. Thus, all the required corrections will be "built in" during the collection of the five PMCC samples. B F GOODRICH L 24819003 Corporate Environmental Service -3- 8504-78, SR, 10-18-78 3. As soon as it becomes available, use the latter instrumentation for all future monitoring at ppb levels of VCM. 4. Do not use the CARB direct injection method of measuring VCM-air concentrations below 100 ppb. Its required injection of a 5.0 cc aliquot of ambient air sample from a Tedlar bag does not give adequate GC response for reliable measurement. 5. Avoid confusing a VCM peak with another peak either at or very near the same retention time. 6. Prove out the new instrumentation and sampling procedure at a BFG plant before using it at all BFG plants. 7. Use a Hewlett-Packard 18850A Terminal (or its equivalent) to plot and integrate the GC signal and simultaneously record the results of the measurement. Use the Terminal's "slope sensitivity minus" feature to isolate the VCM peak and insure its quantitative detection. 8. After proving out the new instrumentation and VCM sampling procedure at a PVC plant, use this BFG sampling method to monitor ambient VCM levels at all PVC plant perimeters. INTRODUCTION The Los Angeles BFG plant is subject to the State of California Air Resources Board (CARB) ambient air quality standard for vinyl chloride.1 That regulation details how to collect and analyze ambient VCM-air samples. It requires collection in Tedlar bags. It does not consider the feasibility of using the Bendix Personnel Monitoring Collection Column (PMCC), also termed the "Flasher tube", instead of the Tedlar bag. The regulation sets "a maximum allowable level for ambient concen trations of vinyl chloride within the range of 0.01 to 0.05 parts per million, twenty-four hour average." The BFGoodrich Company has used and wishes to continue to use activated charcoal monitor tubes (namely, PMCC's) rather than the Tedlar bag. We have con ducted this study to learn whether the PMCC method is as effective as the CARB Tedlar bag method. We also wanted to define the reliability of the data already generated by the PMCC method. The study directly compares the reliabilities of the PMCC and the Tedlar bag methods at VCM concentrations of 0.005 to 0.040 parts per million in air. The measurements were made over a range of relative humidities at constant temperature. B F GOODRICH D. ,U n,-- * v u m is v z Corporate Environmental Service -4- 8504-78, SR, 10-18-78 EXPERIMENTATION The experiment involved simultaneous collection of vinyl chloride monomer (VCM) from the flowing test atmosphere by Bendix Personnel Monitoring Collection Columns (PMCC's) and Tedlar gas bags. The procedure included VCM concentrations of 5.0 to 40 parts per billion (ppb) and relative humidities of 44, 65, 75, and 95%. The temperature of the VCM-"zero air" test atmosphere was 241C throughout the collections. The constant temperature permitted comparisons on the basis of relative humidity instead of mg of water per m3. Preparation of the PMCC's and Tedlar Bags for VCM Collection The PMCC is a stainless steel tube containing 450 mg of a proprietary form of activated charcoal. A Pyrex glass wool filter followed by a stainless steel retainer screen keeps the charcoal in place. The PMCC collects VCM by adsorbing it on charcoal surfaces. The charcoal also adsorbs any water vapor and other organic pollutants in the test atmosphere. Thermal desorption at 250C removes . the adsorbed substances from the charcoal for either cleaning or analysis. A flow of helium gas through the heated PMCC purges the VCM from the tube. Appendix A describes the PMCC further. All desorptions in this study occurred in the Bendix Flasher oven at 250C. Helium flowed through the PMCC at 30 cc per minute for each 30 minute precollection cleaning cycle. We repeated the cleaning cycles for each PMCC until gas chromatographic (GC) analysis showed no detectable peak at the retention time for VCM. The GC analysis times and temperatures employed during each cleaning cycle were the same as used for the measurement of the VCM from loaded tubes. Cleaning the Tedlar bags involved filling each one with "zero" air and com pletely emptying it. We prepared "zero" air for cleaning by flowing compressed air through activated charcoal. Filling and emptying each bag three times cleaned the bag well. GC analysis of 5.0 cc samples from the bag after cleaning verified the absence of detectable VCM. Thus, each PMCC and Tedlar bag were free of detectable VCM before they were used to collect VCM from the test atmosphere. Preparation of the VCM Test Atmospheres Preparation of the required VCM-air test atmospheres began with cleaning the air from the compressed air line. (See Appendix B.) That air stream passed through an oil trap, a silica gel water vapor trap, and an activated charcoal trap for organic vapors in that order at room temperature. The resulting "zero" air was free of water vapor and organic impurities. The dynamic vapor generator (Appendix B) humidified the "zero" air. Humidification involved bubbling the "zero" air through double-distilled water inside glass reservoirs. Our study included relative humidities of 44, 65, 75, and 95%. We used a Taylor wet-and-dry bulb hygrometer to measure the relative humidity. The purified and humidified air became the diluent in preparing the ppb VCM test atmospheres. B F GOODRICH D .u r__ 2 4 8 1 .9 6 0 3 Corporate Environmental Service -5- 8504-78, SR, 10-18-78 The VCM for this study came from a Precision Gas Products, Inc. cylinder. The supplier certified that the VCM-air mixture in the cylinder consisted of 48.6 ppm VCM in pure ("zero") air. The National Bureau of Standards essentially verified that certification (Appendix C). The dilution involved mixing the 48.6 ppm VCM-air mixture from the cylinder with the purified and humidified air in the dynamic vapor generator (DVG). The flow rate of the diluent air was 48.6 1pm as controlled and measured with rotameters. We calibrated the two rotameters with a Singer Model 802 dry gas meter. Concentrations of 5.0, 10.0, and 40 ppb resulted from controlling the flow rate of the 48.6 ppm VCM-air mixture at 5.0, 10.0 and 40 cc per minute. An NRS needle control valve. Model No. 8514 (Brooks Instrument Division, Emerson Electric Co.), reproducibly controlled the flow of the 48.6 ppm VCM-air mixtures. We measured the flow rate periodically during the collection with a soap-film bubble meter. The above procedure produced three different VCM concentrations in air at four different relative humidities. The result was test atmospheres having seven different combinations of VCM concentration and relative humidity. Collection of VCM Test Atmospheres in PMCC's and Tedlar Bags After dynamic mixing, the ppb-level VCM-air stream flowed through the fiveport glass manifold (see Appendix B). The Tedlar bag collected a 14.4 liter sample of the test atmosphere at the middle port. That sample flowed from inside the supply manifold through a glass L-tube and Teflon tubing into the Tedlar bag. At the same time 14.4 liters of the same test atmosphere flowed through each of four PMCC's in the other four ports of the supply manifold. The flow rate for the PMCC's and the Tedlar bag was 40 cc per minute in all but one set. The correspond ing collection period was 6.0 hours. In one set we collected concurrent PMCC and Tedlar bag samples at 10 cc per minute for 24 hours. Past experience assured us that sampling for 6 hours at 40 cc per minute is equivalent to sampling for 24 hours at 10 cc per minute under the same conditions. We adjusted and monitored the flow rate through each PMCC during the VCM collection with an Aalborg four-flowtube flowmeter. We calibrated the flowrate through each flowtube-plus-PMCC both before and after each collection period. A vacuum pump drew the test atmosphere directly from the supply manifold into each PMCC. The air minus the VCM continued through the flow tube and the vacuum pump. It exited into the exhaust hood. A specially-fitted 55 gallon steel drum served as the expansion chamber to fill the Tedlar bag.2 Appendix D shows a sketch of the drum and associated equip ment. We installed a stainless steel quick-connect fitting through the drum lid. A snap-lock ring and a tube gasket made an airtight seal between the lid and the drum. A vacuum pump evacuated the air from the drum through Tygon tubing. We installed a stainless steel tube through the side of the drum with an airtight seal. We connected that tube to the Halkey-Roberts screw-valve of the Tedlar bag in an airtight way. With that connection we could open and close the Tedlar bag valve inside the closed drum. We connected the outer end of the stainless steel tube with Teflon tubing to the glass L-tube inside the supply manifold. The Tedlar bag filled with the test atmosphere as the vacuum pump evacuated the drum. A soap-film bubble meter between drum and vacuum pump measured the evacuation rate. B F GOODRICH Research Center SOUGrstJj Corporate Environmental Service -6- 8504-78, SR. 10-18-78 A Whitey needle valve adjusted that rate. We measured the rate periodically during the collection. At the end of the collection period we closed the quick-connect fitting in the drum lid. Finally we closed the Tedlar bag valve without opening the drum. Measurement of the VCM Content of PMCC's and Tedlar Bags We measured the weight of VCM in each PMCC by thermal desorption followed by GC analysis. The thermal desorption involved heating the PMCC for 2.0 minutes at 250C in the Bendix Flasher oven with no flow through the tube. Then helium flowed through the PMCC and carried the desorbed VCM through the GC column. The 20 ft x 1/8 inch stainless steel column contained 10% DC-200 on 80/100 mesh Chromosorb PAW (Supelco, Inc.). The Hewlett-Packard 5830A GC detected the VCM by flame ionization. The Hewlett-Packard 18850A Terminal graphed the electronic output from the GC detector, integrated the areas under the peaks, and printed the results. The tem perature-time cycle included 2 minutes' desorption, 10 minutes' GC analysis at 70C, 3 minutes' temperature rise to 150C, and 10 minutes' thermal cleaning of the column at 150C. Thus, the measurement equalled all the VCM collected by and re covered from the PMCC. We measured the VCM content of the air in the Tedlar bag in the manner required by CARB. That involved withdrawing a 5.0 cc sample of the VCM-air mixture from the bag and injecting it from a syringe into the GC column. The 5.0 cc sample represents only (5/14, 400) or 0.00035 of the total VCM-air sample in the bag. The VCM measurements were made with a Hewlett-Packard 5711A GC coupled with a HewlettPackard 3380A Integrator. Measurement of Relative Humidity The relative humidity measurements came before and after each set of VCM-air loadings. We closed the ports of the supply manifold and attached a glass T-tube to its bottom outlet tube (Appendix B). We stoppered one branch of the T-tube. A Taylor Comfortguide Hygrometer measured the wet bulb and dry bulb temperatures of the emergent air stream. The two temperatures enabled us to calculate the relative humidity. The relative humidity and the temperature of the test atmo sphere remained constant throughout each set of measurements. Calibration of the Gas Chromatographs The calibration of the Hewlett-Packard 5830A GC involved loading thermallycleaned PMCC's directly from the 48.6 ppm VCM cylinder. We flowed the VCM-air mixture through the PMCC's at 40 cc per minute. The loading time was 30 minutes for four PMCC's and 3.0 minutes for one PMCC. The loadings thus represented 990 and 99 ppb if the same weights of VCM were collected during a 24-hour period. We analyzed the five PMCC samples in the manner described above. The calibration of the Hewlett-Packard 5711A GC for the analysis of Tedlar bag samples followed a different procedure. We first injected a measured weight of pure liquid VCM into a measured weight of carbon disulfide in a closed vial. The resulting solution became our primary standard. A series of successive dilu tions with carbon disulfide resulted in standard solutions having VCM concentra tions of 31.54 ng/pl (i.e., nanograms per microliter) and 0.0282 ng/pl. Syringe injections of from 2.0 to 50 pi of the two solutions resulted in analyses of from 0.0705 ng to 1580 ng. The VCM content of 5ccof a lOppb VCM-air mixture is 0.128 ng. B F GOODRICH Research Center 24819007 Corporate Environmental Service -7- 8504-78, SR, 10-18-78 RESULTS AND DISCUSSION From the start our purpose became threefold. First was to learn whether the Tedlar bag method of collection and analysis is practicable. Second was to com pare the results of the PMCC method with the results of the Tedlar bag method. Concurrent collection of identical VCM-air samples in the PMCC's and Tedlar bag achieved both of these purposes. Our third purpose was to determine if the historical VCM ambient data, which were collected without regard to relative humidity and temperature, are reliable. This also has been accomplished by the sampling and analysis of test atmospheres of constant composition. Calibrations of the Gas Chromatographs Calibrations of the two gas chromatographs were the necessary first step of analysis. Table I and Figure 1 show that the calibration curve for the PMCC analyses was a straight'1ine. The VCM calibration factor within the 95% confidence limits was 1104.0 au/ng. Tables II and III give the data for two separate GC cali brations for the Tedlar bag analyses. The first calibration (Table II) gave the wider spread of data. Its calibration factor within 95% confidence limits.was 1,523 1,157 au/ng. The second calibration gave a factor of 1,930 260 au/ng (95% limits). Loss of small amounts of VCM is difficult to avoid during the repeated dilutions in preparing standard ppb solutions of VCM in carbon disulfide. Even a small loss is critical in the 0.0135 - 0.108 ng/pl range. Therefore, we used greater skill from practice and a higher range of VCM concentrations in preparing the second series of standard solutions. Figure 2 shows that the calibration curve for the data in Table III is a straight line through the origin. We have used the calibration factor of 1,930 au/ng throughout our subsequent calculations. The response factors for the Tedlar bag samples involved measurements at the 0.06 ng VCM level. That level is virtually the GC detection limit. In sharp contrast the PMCC response factor involved measurements at the 60,000 ng VCM level. The range of 106 ng between the two VCM levels is very wide. It partially explains the difference in calibration factors. Figure 3 illustrates how that difference is consistent with the nature of the calibration curve. The chief reason for the difference is the usual negative inflection in the calibration curve in the vicinity of the limit of GC detection (i.e., "zero"). The inflection is probably due to a combination of instrument "noise" and physical interactions of the VCM with the trapping charcoal, the column, and the detector. In short, the GC becomes proportionately less sensitive as the VCM weight approaches zero. Thus the rate of decrease of the ratio (au/ng) speeds up in the vicinity of zero VCM. Collection of VCM from Air Samples in PMCC's The relative humidity will certainly vary during 24-hour collections of VCM-air samples. In fact, both the relative humidity and the ambient temperature will probably vary during a single 24-hour collection. The concentration of water vapor varies with the ambient temperature at any fixed relative humidity. In an earlier study of VCM analyses at 1.0 ppm concentration3, we found that relative humidity had a profound effect on VCM detected by the PMCC tube. The effect becomes more pronounced as sampling volume increases. We evaluated the data from the present study in terms of both relative humidity and the weight of water charged to the tube. B F GOODRICH Research Center 24819008 Corporate Environmental Service -8- 8504-78, SR, 10-18-78 Table IV summarizes the experimental data from the collection of VCM-air samples in PMCC's. The values on each line of the table are for several simultane ous PMCC collections. Table VIII states the number of PMCC's per corresponding line in Table IV. Table IV shows the following in order: (a) the VCM-air concentration in parts per billion, (b) the percent relative humidity, (c) the GC response in area units, (d) the calculated weight of VCM recovered from the PMCC, (e) the GC response factor in au/ng, and (f) the percent recovery of VCM. We calculated the weight of VCM in each PMCC by multiplying the measured VCM-air volume by the VCM-air concentration. The GC response factor was the GC response divided by the charged weight of VCM. We calculated the percent recovery by dividing the sample response (in au/ng) by the calibration factor (110 au/ng). We needed to know how the experimental variables affected the GC peak area. Interpreting the actual GC area is difficult. It must change with variations in either the VCM concentration or the volume of "contaminated" air passing through the PMCC. Therefore, we calculated an "adjusted GC response". The "adjusted GC response" is defined as the response per 15 liters of sampled test atmosphere x 10"3. The 15-liter standard volume was selected because it approximates many of the actual volumes in the sampling program. That is. Equation (1) is adjusted PMCC peak area = ~ctuaacftuasal mGpCledpevaokluamreea (15 1 x 10"3) Mathematical modeling was used to show the effect of the experimental vari ables on the GC response. We employed multiple regression analysis using adjusted GC area versus ppb VCM, total sampled volume (liters), and either relative humidity or total mg of water vapor. Looking at the GC area per unit volume then enabled us to look at and graph the varying efficiencies of detecting VCM. We thus evaluated the effects of different levels of humidity and total volume for the range of VCM levels. The mathematical levels based on mg of water vapor were consistently inferior to those using relative humidity. We therefore dropped the mg H20 models from further consideration. Models based on VCM percent recovery did not show the effect of varying VCM levels as did models based on adjusted GC response. We required a model which passes through the origin of the graph. The model based on VCM recovery did not do so. Table V shows the adjusted GC peak areas, volumes of test atmosphere, ppb VCM, and % R.H. data. The fifth set of data in Table V had an abnormally low adjusted GC area relative to the corresponding sample volume, ppb VCM, and % R.H. The value shown is 41.1 compared to a predicted 84.2 au. We deleted that row of data and chose Equation (2) as our final mathematical model: GUOGTSI'2 B F GOODRICH Research Center Corporate Environmental Service -9- 8504-78, SR, 10-18-78 adjusted GC peak area/15 liters = (1.3683) [(volume) (ppb VCM)] + (0.001044) [(ppb VCM) (% R.H.)2] - (0.02100) [(volume) (ppb VCM) (% R.H.)] This model explains 97.2% of the observed variation in adjusted GC areas (i.e., R2 = 0.972). The 95% confidence limits are 40 au x 10~3. Using Equation (2) we generated Table VI. It predicts the adjusted GC area/ 15 liters covering the range of ppb VCM and % R.H. Predictions were made only at the total volumes of 12 and 15 liters of test atmosphere. Only two volumes were lower, and the risk of extrapolation is too great. Figures 4 and 5 illustrate the results via a plot of the data in Table V. To clarify the reader's examination of Figures 4 and 5, we remind that a change in adjusted GC peak area with a change in total air flow reflects a change in the efficiency of detecting VCM. The following Equation (3) permits calculating the predicted actual GC peak area: predicted actual GC area - 1 dsted,GC_area) (actual volume,, 1L1L.M1 The plots in Figures 4 and 5 illustrate how well Equation (2) fits the data. The curves show that % R.H. has a dominant influence on the adjusted GC area (and therefore on the actual GC area). For example (Figure 5), at 15 liters and 40 ppb the adjusted peak area is 340,000 au at 45% R.H. but almost 0 at 95% R.H. Above an average 75% R.H. taking a PMCC ppb VCM sample is hardly worth while. Some compensa tion for % R.H. is needed at all humidity levels; Increasing the ppb VCM concentration increases the adjusted peak area as expected. Finally, comparing Figures 4 and 5 shows the effect of changing the volume of test atmosphere through the PMCC. The plots indicate that a lower total flow reduces the effect of high humidity. This reduction is expected on a theoretical basis since less air volume means less total water to interfere with VCM adsorption and desorption during collection and analysis. Collection of VCM-Air Samples in Tedlar Bags Table VII summarizes the experimental data from the collection of VCM-air samples in Tedlar bags and the subsequent analyses. The column headings were described before in this report for Table III. Multiple regression analysis was attempted again. As before it was based on actual GC response versus ppb VCM and % R.H. (or ng of H20). Constant sample volume (5.0 cc GC injections) made adjustment of GC response unnecessary. We were not able to find a significant relationship among the variables for the Tedlar bag samples. The apparent sampling and testing error was essentially the same as the total observed variation in GC response. Figure 6 helps to show this almost total lack of significant relationships. Note that the scatter of results within a par ticular VCM-humidity combination is about equal to the total scatter of data points. The regression analysis does not seem to respond at these extremely low VCM levels. Number of "Hits" in PMCC and Tedlar Bag Analyses It is time now to compare "hits and misses" for individual ppb VCM samples from PMCC's versus Tedlar bags. Tables IV and VII show our estimates of GC peak areas for PMCC and Tedlar bag analyses. However, they don't tell the whole story. As the VCM concentration drops to extremely low values, reliable results become B F GOODRICH Research Center OIOGTSl'Z -10- Corporate Environmental Service 8504-78, SR, 10-18-78 more and more difficult to obtain. Frequently VCM cannot be detected at all in a VCM-air sample at such levels. The following results from Table VIII illustrate the number of "hits" for PMCC and Tedlar bag analyses versus ppb VCM. VCM Concn. (ppb) PMCC Samples Analyses Ho. of Successes Attempted "Hits" (*) Tedlar Bag Samples Analyses No. of Successes Attempted "Hits" (%) 5 12 11 92 15 2 13 10 39 29 74 120 47 39 40 19 19 100 43 28 65 Total: 70 59 84 At 5 10 ppb: 51 40 78 178 77 135 49 43 36 The results show how questionable the Tedlar bag method of analysis really is. They suggest that this type of analysis should not be attempted at VCM levels much below 100 parts per billion with 5.0 cc samples at the present state of tech nology. In contrast the PMCC method of collection and analysis is practicable under the conditions of this experiment. Relationship of Tedlar Bag Samples to Adjusted GC Peak Areas We used multiple regression analysis to learn whether adjusted GC peak areas of PMCC VCM samples could be predicted from measurements of Tedlar bag samples and % R.H. Table IX lists the data used and shows the ppb VCM in each test atmosphere. We obtained a significant relationship by using the following model (Equation 4): Adjusted PMCC GC peak area = = 94.19 + (0.3085) (Tedlar GC peak area) - (1.2421) (% R.H.) This model explains 74.7% of the observed variation in adjusted GC peak area (i.e., R2 = 0.747). The 95% confidence limits are 44 au x 10 . We dropped the ninth set of data in Table IX from the analysis as an outlier. Its predicted value was 45.5 au compared with an observed value of 170 au.(see Table X). We used Equation 4 to develop Table X. That table predicts the adjusted PMCC GC peak area for a variety of Tedlar bag GC peak area - % R.H. combinations within our general experimental range. Figure 7 displays curves based on Equation (4) and plots the data used in developing that equation. The preceding section of this report shows how the difficulty of detecting VCM in test atmospheres by the Tedlar bag method increases as the concentration goes from 40 to 10 to 5 ppb. Figure 6 suggests the same trend. That is, for 75% R.H. or less the PMCC gives measurable VCM. However, comparable Tedlar bag samples give an estimate of zero PMCC VCM by Equation 4. Figure 6 shows 3 to 5 ppb VCM by the PMCC method when the Tedlar bag shows 0 ppb VCM. Table XI gives the data which were calculated from Equation 4 and used to plot the curves in Figure 7. TTOGTSt-Z B F GOODRICH Research Center -11- Corporate Environmental Service 8504-78, SR, 10-18-78 REFERENCES 1. State of California Administrative Code, Title 17, Health and Safety Code, Section 39606(b). 2. Code of Federal Regulations, Title 40, Chapter I, Part 61, Subpart I "National Emission Standard for Benzene Emissions from Maleic Anhydride Plants", Figure 111-1: Integrated-bag sampling train, page 48. 3. Born, J. W.; Zakriski, P. M.; and Hardesty, R. 0.; Evaluations of Bendix Flasher Tubes and MSA Tubes as VCM Monitors, Status Report, 8504-76, August 24, 1978. 8 B F GOODRICH Research Center -12- Corporate Environmental Service 8504-78, SR, 10-18-78 TABLE I Calibration of the Gas Chromatograph for PMCC Sample Analyses PMCC No. J1 J2 J3 J4 VCM Concentration in Air (ppb) 486. 486. 486. 48.6 Weight of VCM in PMCC (pq) 89.58 92.57 89.58 8.958 GC Peak Area (au) 9,870,000 10,040,000 9,866,000 993,600 Peak Area/VCM Wt (au/nq) 110. 108. no. 111. Average (Peak Area/VCM Weight) = 109.8 au/ng Standard deviation = 1.26 au/ng 95% confidence limits = 4.0 au/ng Coefficient of variation = 1.15% TABLE II Calibration of the Gas Chromatograph for Tedlar Bag Sample Analyses Std. Solution C Volume (pi) 2.5 2.5 2.5 2.5 5.0 5.0 5.0 20.0 20.0 20.0 20.0 Weight of VCM (nq) 0.0675 0.0675 0.0675 0.0675 0.135 0.135 0.135 0.540 0.540 0.540 0.540 GC Response (area units) 150 131 81 112 142 297 253 804 697 700 284 Factor (au/nq) 2,222. 1,941. 1,200. 1.659. [1,756.] 1,052. 2,200. 1,874. [1,709.] 1,489. 1,291. 1,296. 526. [1,150.] Standard Sol'n. C (ul) 2.5 5.0 20. VCM Weight (ng) 0.0675 0.135 0.540 Overall Average: Calibration Factor (au/nq) 1,756 1,709 1,150 1,523 Standard 1Deviation (au/nq) 436 592 426 519 Coefficient of Variation (%) 24.9 34.6 37.0 34.1 95% Confidence Limits (au/nq) 1,386 2,545 1,357 1,157 & CO u* to h* & B F GOODRICH Research Center -13- Corporate Environmental Service 8504-78, SR, 10-18-78 TABLE III Calibration of the Gas Chromatograph for Tedlar Bag Sample Analyses Standard Solution B Volume (yl) Weight of VCM (nq) GC Response (area units) Calibration Factor ______ (au/ng) 10.0 15.0 25.0 30.0 50.0 315. 473. 788. 946. 1 ,577. 642,679 944,632 1 ,440,660 1,738,106 3,055,728 2,040. 2,000. 1 ,830. 1,840. 1,940. Average Calibration Factor = 1,930 au/ng Standard Deviation = 94 au/ng Coefficient of Variation = 4.9% 95% Confidence Limits = 260 au/ng VCM (PPb) 5.0 5.0 5.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 40.0 40.0 40.0 40.0 40.0 TABLE IV Summary of Data from the Collections of VCM in PMCC's R.H. (%) GC Response (area units) VCM Wt. (mg) Factor (au/ng) 44 8,096 73.6 110 16 44 54,540 202. 270 75 44 25,564 166. 154 21 44 4,648 43.2 1 08. 44 39,483 369. 107 32 44 63,491 367. 173 30 65 29,117 371. 68. 818 65 51,078 382. 154 29 75 57,222 374. 153 10 74 35,980 378. 84. 627.4 74 19,810 308. 64. 32 3.1 95 11,279 398. 37. 013.0 95 38,237 368. 104 12 44 35,535 1,545. 230 12 67 146,135 1,457. 100 7.8 67 143,573 1 ,267. 100 29 oCVl CVJ +1 to 95 33,697 1,398. 24.,5 5.3 95 11,180 1,480. 4. VCM Recovery (%) 100. 245. 140. 98. 97. 157. 62.5 140. 139. 86.7 58.4 33.6 94.5 209. 90.9 90.9 22.2 4.22 B F GOODRICH Research Center VCM (ppb) 5.0 5.0 5.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 40.0 40.0 40.0 40.0 40.0 -14- Corporate Technical Support 8504-78, SR, 10-18-78 TABLE V PMCC Volumes Sampled, Water Charged, and Adjusted GC Peak Areas R.H. (*) 44 44 44 Water (mq) 57.3 160. 251 . GC Response (au) 8,096 54,540 25,564 VCM-Air Volume Sampled (1) 5.74 15.8 13.0 Adjusted GC Response (au x 10-3/15 1) 21.16 51.78 29.59 44 17.7 44 146. 44 133. 4,648 39,483 63,491 1.74 14.4 13.1 40.07 . 41.13 72.76 65 200. 65 221. 29,117 58,623 14.4 15.5 30.29 56.66 75 251. 74 239. 74 206. 57,222 35,980 19,810 14.5 14.5 13.4 59.36 37.30 22.24 95 330. 95 310. 11,279 38,237 15.0 14.2 11.24 40.51 44 153. 355,350 15.1 349.2 67 221. 67 200. 146,135 143,573 14.3 12.6 153.5 170.4 95 297. 95 309. 33,697 11,180 13.3 14.4 38.12 11.67 CO CO C/1 -15- Corporate Environmental Service 8504-78, SR, 10-18-78 TABLE VI Predictions of Adjusted PMCC GC Peak Areas for 12 and 15 Liters R.H. (%) VCM-Air Volume Sampled (1) Adjusted GC Peak Area (au)* Minus the 5th Set Including All Sets 10 30 12 98 83 10 45 12 72 65 10 65 12 45 41 10 75 12 34 29 10 95 12 19 5 20 30 12 196 165 20 45 ' 12 144 130 20 65 12 89 82 20 75 12 68 58 20 95 12 38 10 30 30 12 294 248 30 45 12 215 195 30 65 12 134 123 30 75 12 102 87 30 95 12 57 16 40 30 12 392 331 40 45 12 287 259 40 65 12 178 164 40 75 12 135 116 40 95 12 76 21 10 30 15 120 103 10 45 15 85 81 10 65 15 45 51 10 75 15 28 36 10 95 15 0 7 20 30 15 240 206 20 45 15 169 162 20 65 15 89 103 20 75 15 56 73 20 95 15 1 13 30 30 30 45 30 65 30 75 30 95 15 15 15 15 15 360 253 134 83 1 40 30 40 45 40 65 40 75 40 95 15 15 15 15 15 480 338 178 111 1 These values were calculated using Equation 2. 310 243 154 109 20 05 413 b* 324 to 205 a 146 26 Ci B F GOODRICH VCM (PPb.) 5 10 10 10 10 10 40 40 10 10 40 40 -16- Corporate Environmental Service 8504-78, SR, 10-18-78 TABLE VII Summary of Data from the Collections of VCM in Tedlar Bags R.H. _i*L 44 GC Response (area units) 0 VCM Wt. _ (ng) 0.129 Factor (au/ng) 0 65 109. 65 189. 0.130 0.128 838 277 1,480 109 74 0 74 56.4 75 254. 0.128 0.128 0.129 0 441 134 1 ,968 1, 271 67 340. 67 112. 0.510 0.510 667 231 220 86 95 131. 95 252. 0.132 0.130 992 167 1,938 148 95 183. 95 119. 0.517 0.517 354 217 230 85 VCM Recovery m 0 43.5 76.6 0 22.9 102. 34.6 11.4 51.4 101. 18.3 11.9 CD CoO B F GOODRICH -17- Corporate Environmental Service 8504-78, SR, 10-18-78 TABLE VIII Comparison of GC Responses vs. Injections for Tedlar Bags and PMCC's VCM Concentration (PPb) 5.0 5.0 5.0 10.0 10.0 10.0 Relative Humidity (%) 44 44 44 44 44 44 Tedlar Baq Number of Number of Injections Responses -- 15 2 __ -- -- PMCC Number of Number of Injections Responses 44 44 43 41 43 43 10.0 10.0 65 27 12 65 13 7 44 43 10.0 10.0 10.0 75 22 3 74 15 0 74 11 5 33 43 42 10.0 10.0 95 12 5 95 20 15 44 43 40.0 44 - 44 40.0 40.0 67 10 7 67 15 11 44 44 40.0 40.0 Totals: 95 12 6 95 6 _4 178 77 33 _4 _4 70 59 Responses/Injections: (77/178) = 432 ; (59/70) = 842 Analysis Time per Set Tedlar bag =1.0 hour; PMCC = 2.0 hours, Comparative Analysis Times: Tedlar bag = 178 hours; PMCC = 100 hours &T06T8P2 B F GOODRICH -18- Corporate Environmental Service 8504-78, SR, 10-18-78 TABLE IX Data Used to Develop Equation 4 to Relate Tedlar and PMCC GC Peak Areas VCM Concentration (ppb) R.H. (%) Observed Tedlar Peak Area (au) Predicted Adjusted PMCC Peak Area (au) 5.0 44 0 29.59 10.0 10.0 10.0 10.0 10.0 10.0 10.0 65 109.0 65 189.0 74 0 75 56.40 75 254.0 95 131.0 95 252.0 30.29 56.66 37.30 22.24 59.36 11.24 40.51 40.0 40.0 40.0 40.0 67 112.0 67 340.0 95 119.0 95 183.0 170.4 153.5 11.67 38.12 TABLE X Comparison of Adjusted PMCC Peak Areas: Observed vs. Predicted3 Adjusted. PMCC VCM GC Peak Areas (au) Observed0 Predicted0 59.3589 29.5880 37.2979 11.2415 40.5053 38.1188 11.6701 22.2418 30.2881 153.5032 170.3794 56.6588 79.3980 39.5395 2.2769 16.6088 53.9393 32.6517 12.9066 18.4352 47.0840 115.8671 45.5254 71.7653 Difference (au) (*) -20.0391 -9.9515 35.0209 -5.3673 -13.4340 5.4671 -1.2364 3.8066 -16.7959 37.6360 124.8540 -15.1065 -33.7592 -33.6337 93.8953 -47.7450 -33.1660 14.3423 -10.5950 17.1146 -55.4536 24.5181 73.2800 -26.6622 a The predicted values are based on the observed Tedlar bag VCM GC peak areas. They were calculated to show the relationship between the observed Tedlar peak areas and the corresponding PMCC peak areas. k The observed adjusted PMCC GC peak areas come from Table V as calculated by Equation 1. c The predicted adjusted PMCC GC peak areas were calculated from the observed Tedlar bag GC peak areas by Equation 4. e im x s iz B F GOODRICH -19- Corporate Environmental Service 8504-78, SR, 10-18-78 TABLE XI Data Derived from Equation 4 and Used to Plot Figure 7 Assigned Tedlar Peak Area (au) R.H. (X) Predicted Adjusted PMCC Peak Area (au) 0 45 150 45 300 45 38 85 131 0 65 150 65 300 65 13 60 106 0 75 150 75 300 75 1 47 94 0 95 150 95 300 95 -24 23 69 o z a s T s t-z B F GOODRICH Corporate Environmental Service 20 8504-78, SR, 10-18-78 B F GOODRICH -21- Corporate Environmental Service 8504-78, SR, 10-18-78 Calibration of H P 5 7 IIA -I6 A ' Column*48 Figure 2. 4J> u B F GOODRICH -22- Corporate Environmental Service 8504-78, SR, 10-18-78 B F GOODRICH -23- Corporate Environmental Service 8504-78, SR, 10-18-78 ce i B F GOODRICH Corporate Environmental Service 24 8504-78, SR, 10-18-78 CS o B F GOODRICH Figure 7. PMCC Adjusted G.C. Area vs. Tedlar Bay Observed G.C.Area and % Relative Humidity CXJ-Q-X ------^ o^aod ---- , tno iIo'lijAft- Corporate Environmental Service 8504-78, SR, 10-18-78 T T I X r B F GOODRICH -26APPENDIX A Corporate Environmental Service 8504-78, SR, 10-18-78 Descriptions of the Bendix Flasher & Flasher Tube Bendix PMCC Bendix Flasher Tube f~ IPB~PMCC-ModiPfed Charcoal Length : it mm O.D. : & mm ; I.D. : 4 mm Charcoal : ~ 4-50 Bendix Flasher 24819027 B F GOODRICH