Spontaneous Desorption of Organic Solvents from

Transcription

1 Industrial Health, 1987, 25, Spontaneous Desorption of Organic Solvents from Carbon Cloth Miyuki KASAHARA and Masayuki IKEDA Depertment of Environmental Health, Tohoku University School of Medicine, Seiryo-cho 2, Sendai 980, Japan (Received December 2, 1986 and in revised form March 17, 1987) Abstract : The velocity of spontaneous desorption of various solvents from carbon cloth K-filter TF-1500 (CC) housed in a diffusive sampler holder was examined. Among the 16 most popular organic solvents (acetone, benzene, butyl acetate, ethanol, ethyl acetate, n-hexane, isopropyl alcohol, methanol, methyl ethyl ketone, methyl isobutyl ketone, styrene, tetrachloroethylene, toluene, 1,1,1-trichloroethane, trichloroethylene and m-xylene), the desorption half-time (T1/2) was various depending on the solvent and was shortest with methanol (57 min) followed by ethanol (20 h) and acetone (38 h). As T1/2 values observed were mostly larger than 24 h, it was possible to conclude that diffusive sampling with CC is generally applicable to the exposure monitoring of a wide range of solvents. The application to methanol is, however, practically impossible due to its rapid spontaneous desorption. The solvent with short T1/2 of spontaneous desorption can be best predicted from log (n-octanol/water partition coefficient) [log (Po/w)] ; a solvent with log (Po/w) of less than zero will have T1/2 of less than 24 h, and diffusive sampling may not be applicable to monitoring of such solvents. Packing of CC in aluminum foil immediately after the termination of exposure and storage at 4 Ž jointly improved the preservation of the impregnated solvents so that no significant loss was detected for any solvent tested except for methanol in 4 days. Key words : Carbon cloth-desorption half-time-diffusive sampling-organic solvent-prediction of desorption-spontaneous desorption INTRODUCTION Light weight, simple structure and no need of power supply for air sampling are among the obvious advantages of diffusive sampling (also called passive sampling). Efforts have been made to identify the range of chemicals, especially organic solvents, to which the diffusive sampling is applicable to the determination of the time-weighted average (TWA) concentration either in the breath zone air of workers or at fixed sampling locations during the shift of a day. Although Offprint requests to : Prof. Masayuki Ikeda, Department of Environmental Health, Tohoku University School of Medicine, Seiryo-cho 2, Sendai 980, Japan.

2 74 M. KASAHARA and M. IKEDA studies in our laboratory have proved the validity of this method as a practical procedure to measure TWA exposure of workers to various solvents,1-7) it has also been disclosed in previous studies on carbon cloth (to be abbreviated as CC) employed as adsorbent that the reduced capacity of CC under very humid conditions8) and the spontaneous desorption of solvents, especially methanol," from CC were two major problems that disturbed the extensive application of diffusive sampling. While it is possible to solve the former problem by selecting CC with better performance,8) the latter cannot be readily solved as the problem is associated with the physical characteristics of the solvent to be monitored. In the present study, the latter problem was quantitatively studied by the measurement of the half-time (T1/2) of solvent desorption from CC. The effects of the storage conditions on the solvent preservation after the exposure of CC to solvent vapors untill analysis were also examined. MATERIALS AND METHODS Experimental exposure of carbon cloth to solvent vapors. The CC examined was K-filter TF-1500, a commercial product supplied from Toyobo Co., Osaka, Japan. For the experimental exposure of CC to various solvent vapors, a servomechanized exposure chamber system of the flow-through typel10,11) was employed. Square pieces (3 cm ~ 5 cm) of CC housed in holders8) individually were exposed to the solvent vapor at 1,000 ppm for 30 min for impregnation. The vapor diffused into the holder through a glass filter GA 100 (Toyo Roshi Co., Ltd., Tokyo, Japan) as a wind shield, and reached CC at 12 mm distance from the filter. Analysis of solvent present in CC. Each CC piece was extracted with 10 ml of either water (spiked with n-propanol as an internal standard) or CS2 (with sec.-butylbenzene or toluene) as described in Table 1. An aliquot, 2 Đl, of the former extract was injected to a Hitachi FID-GC (Model 163) to be analyzed on a stainless steel column (3 mm in inner diameter and 2 m in length) packed with 25% polyethyleneglycol-1000 on Celite 545 (60-80 mesh)9) and that of the latter extract on a column with 10% free fatty acid polyesters on Chromosorb WAW-DMCS ( mesh).8) The extraction recoveries of acetone, ethanol and methanol with water were 25%, 69% and 93%, respectively, and that of isopropyl alcohol with CS2 was 64%. Although the extraction recoveries for acetone, ethanol and isopropyl alcohol were rather low, the extraction procedures were considered applicable to the experiments because the recoveries were reproducible with coefficients of variation (SD/mean of the extraction rate) of 10%, 5% and 5%, respectively, under the conditions employed. Other solvents could be extracted into CS2 with an recovery of >90% in accordance with the reported findings.12,13)

3 DESORPTION OF SOLVENTS FROM CARBON CLOTH 75 Solvents and other reagents. The 16 solvents tested, i.e., acetone, benzene, butyl acetate, ethanol, ethyl acetate, n-hexane, isopropyl alcohol, methanol, methyl ethyl ketone, methyl isobutyl ketone, styrene, tetrachloroethylene, toluene, 1,1,1- trichloroethane, trichloroethylene and m-xylene were selected as those most frequently found in commercially available solvent-containing industrial products. 14,15) All reagents including the solvents for vapor generation were of extra pure grade obtained from Wako Pure Chemicals, Osaka, Japan. Physical parameters. Three physical parameters of the solvents, i.e., boiling point, vapor pressure and n-octanol/water partition coefficient (Po/w) were cited from a literature.16) Statistical analysis. First order kinetics was assumed when the half-time (T1/2) of spontaneous desorption of the solvent was determined. The T1/2 was calculated by the least squares method. Statistical significance of the difference betwepn means was examined by Student's t-test and that of correlation by regression analysis. RESULTS T1/2 of spontaneous solvent desorption from CC To determine T1/2, five pieces of the CC were analyzed immediately after the termination of exposure to measure the amount of solvent impregnated (0 time analysis). Other pieces were kept in individual holders even after the termination of the exposure, and left in fresh air; four CC pieces each were analyzed either every 2 h up to 8 h in case T1/2 was expected to be a few hours, or every day up to 4 days when it was expected to be a few days. The T1/2 of the desorption of various solvents are summarized in Table 1. It is evident that the T1/2 varies over a wide range depending on the solvent examined; methanol has the smallest T1/2 of 57 min ( day) followed by ethanol (20 h or day). In contrast, butyl acetate, styrene, methyl isobutyl ketone and m-xylene have a T1/2 of over a month. Spontaneous desroption rates under various storage conditions In order to examine the effects of the storage conditions of CC on spontaneous solvent desorption, CC pieces were taken out of the holders and packed in aluminum foil immediately after the termination of solvent impregnation, and then kept either at 4 Ž or at room temperature for up to 4 days. At each time period presented in Table 2, four pieces each were analyzed for the amount of solvent in CC. For comparison, additional solvent-impregnated CC pieces were kept in holders and left in fresh air at room temperature as in the case of T1/2 determination; four pieces each were analyzed on the 2nd and 4th days. Five

4 76 M. KASAHARA and M. IKEDA other pieces of CC were subjected to 0 time analysis. The results are summarized in Table 2; minute discrepancies between the half-time values in Table 1 and the recoveries given in Table 2 are due to the fact that each of the former values was calculated from a set of the data obtained at five time periods by the least-squares method, while the latter describe the findings at 2 individual time periods. In addition to ethanol and methanol with T1/2, of less than 24 h, significant amounts (p<0.01) of solvents were lost when CC pieces housed in holders were kept in fresh air for 2 days after impregnation with either acetone, benzene, ethyl acetate, isopropyl alcohol, methyl ethyl ketone, 1,1,1-trichloroethane or trichloroethylene. Loss was also significant (p <0.05) with n-hexane, tetrachloroethylene and toluene when CC in holders was left in fresh air for 4 days. When CC was taken out of the holder immediately after the exposure and packed in aluminum foil, however, the loss was significant (p <0.05) in 2 days only in the case of the most labile methanol and in 4 days with ethanol, whereas there was no significant loss (p>0.05) with other solvents. The preservation of the solvent was further improved when the CC was packed in aluminum foil and kept in a refrigerator at 4 Ž; under this condition, the loss was significant Table 1. Analytical conditions and desorption half-time of the 16 enlventc examined

5 DESORPTION OF SOLVENTS FROM CARBON CLOTH 77 Table 2. Loss of the solvent from CC during the storage under various conditions (p <0.05) only with methanol after 4 days of storage. DISCUSSION The present study clearly demonstrated that spontaneous desorption takes place when solvent-impregnated CC in a holder is left in fresh air, that the desorption half-time, T1/2, varies depending on the solvent, and that the T1/2 is shortest for methanol followed by ethanol among the 16 solvents studied. It was further disclosed that desorption is suppressed when CC is packed in aluminum foil immediately after exposure to solvent vapors and kept at 4 Ž thereafter. For the application of diffusive sampling for solvent exposure monitoring, one of the worst cases from the viewpoint of spontaneous desorption will be short-term peak exposure at the very beginning of the shift of the day followed by essentially no exposure during other periods of the shift. In this case, CC will be impregnated at time 0 with the solvent and spontaneous desorption will follow. The assumption of first-order kinetics for spontaneous desorption can

6 78 M. KASAHARA and M. IKEDA Table 3. Relationship between T1/2 and three physical parameters Figure. Relationship of T1/2 with 3 physical parameters of [A] vapor pressure, [B] boiling point and [C] Po/w. Each dot represents one organic solvent. Those identified with numbers are : 1,methanol ; 2, ethanol ; 3, acetone ; 4, 1, 1, 1-trichloroethane ; and 5, n-hexane. The oblique line in each figure is a calculated regression line, the equation for which is given in Table 3. The vapor pressure and the boiling point are expressed in mmhg and Ž, respectively.

7 DESORPTION OF SOLVENTS FROM CARBON CLOTH 79 be expressed as log At= at + ƒà, where At is the amount of the solvent in CC at time t, and a and ƒà are positive (i.e., 0) constants in which T1/2 = (log 2)/a and log A0 = ƒà by definition. Thus, log Ao --log At= ƒ t. Assuming that <5% loss of the solvent in CC is acceptable, i.e., At>0.95 Ao, then log A0 log 0.95 Ao>ƒ t, or a<(2 log 95)/t. Substituting a with (log 2)/T1/2, T1/2>(log 2) x t/ 1_2 (log 95)] or T112> ~ t. On the assumption that <10% desorption is allowable, a similar calculation resulted in T1/2>6.578 ~ t. In practice, 95% recovery is assured for a solvent with T1/2 of longer than 122 h (5 days), even in the worst case when the worker is equipped with a diffusive sampler for 9 h (i.e., 8-h worktime plus 1-h lunch break), or for a solvent with T1/2 longer than 54 h (2.25 days) when the sampler is exchanged during lunch break (i.e., 4 h each for the morning and afternoon shifts). It is evident that the sampler is not applicable to the measurement of TWA intensity of exposure to methanol because the T1/2 is too short (less than 1 h). In the case of ethanol with the T1/2 of day (20 h), 90% retention will be barely expectable when the sampler is exchanged every 3 h (i.e., 20/6.578 is nearly equal to 3). Thus, among the 16 most popular solvents14,15) examined in the present study, methanol (considering that ethanol will be seldomly used as a solvent in factories) was found to be the only solvent to which diffusive sampling is not applicable. It would be convenient if the inapplicability of diffusive sampling to a given solvent can be predicted from readily available physical parameter(s) of the solvent. In fact, Pozzoli and Cottica17) recently compared the desorption of tetrafluoroethylene (boiling point; 76.3 Ž), hexafluoropropylene (-29.4 Ž) and cyclooctafluorobutane (-4.0 Ž) from CC, and concluded that the spontaneous desorption of the substances adsorbed on CC will decrease with an increase in the boiling point of the substance concerned. Accordingly, trials were made to correlate T1/2 with 3 well-examined parameters of vapor pressure, boiling point and Po/w in terms of either antilogarithm or logarithm (Table 3). Both the antilogarithm and logarithm of the three physical parameters correlated significantly (p<0.01) with the antilogarithm and logarithm of T1/2, of which three typical cases are depicted in the Figure. Despite the fact that the absolute value of the correlation coefficient between log (T1/2) and log (Po/w) is not the smallest (Table 3) among the three correlations depicted in the Figure, it should be noted that the two problematic solvents (i.e., ethanol and methanol) can be best separated from other solvents to which diffusive sampling is applicable in this particular case (Figure C). Such separation can not be achieved when vapor pressure (Figure A) or boiling point (Figure B) is employed as a predicator of T1/2 even though the correlation with T1/2 is better than that of Po/w (Table 3). Thus, it is possible to deduce from Figure C that difficulties due to rapid desorption may be encountered when diffusive sampling is applied to a solvent with a negative (ie., 0) log (Po/w). The theoretical explanation for this empirical conclusion, however, remains yet to be developed further.

8 80 M. KASAHARA and M. IKEDA The method of preservation of the exposed CC is also an important factor for good practice, as it is usually inevitable that there is a lapse of time between the termination of exposure and the analysis.5) The present results recommend the procedure of taking CC out of the holder immediately after the exposure, packing it in aluminum foil, and keeping the package in a refrigerator until analysis. In case this procedure is followed, it is possible to expect that no significant (p <0.05) loss will take place in the cases of popular solvents other than methanol. ACKNOWLEDGEMENTS The authors are grateful to Dr. H. Kishi, Faculty of Science and Technology, Keio University, Yokohama, Japan for kind supply of Po/w values of the following chemicals; butyl acetate, n-hexane, isopropyl alcohol and methyl isobutyl ketone. Thanks are also due to Mr. M. Harper, Department of Occupational Health, London School of Hygiene and Tropical Medicine, London, United Kingdom for his valuable discussion on the correlation of T1/2 with physical parameters. REFERENCES 1) Ikeda M, Koizumi A, Miyasaka M, Watanabe T. Styrene exposure and biologic monitoring in FRP boat production plants. Int Arch Occup Environ Health 1982; 49: ) Miyasaka M, Kumai M, Koizumi A, Watanabe T, Kurasako K, Sato K, Ikeda M. Biological monitoring of occupational exposure to methyl ethyl ketone by means of urinalysis for methyl ethyl ketone itself. Int Arch Occup Environ Health 1982; 50: ) Hasegawa K, Shiojima S, Koizumi A, Ikeda M. Hippuric acid and o-cresol in the urine of workers exposed to toluene. Int Arch Occup Environ Health 1983; 52: ) Ohtsuki T, Sato K, Koizumi A, Kumai M, Ikeda M. Limited capacity of humans to metabolize tetrachloroethylene. Int Arch Occup Environ Health 1983; 51: ) Ikeda M, Kumai M, Aksoy M. Application of carbon felt dosimetry in field studies distant from analytical laboratory. Ind Health 1984; 22: ) Ikeda M, Watanabe T, Kasahara M, Kamiyama S, Suzuki H, Tsunoda H, Nakaya S. Organic solvent exposure in small scale industries in north-east Japan. Ind Health 1985; 23: ) Inoue O, Seiji K, Kasahara M, Nakatsuka H, Watanabe T, Yin S-G, Li G-L, Jin C, Cai S-X, Wang X-Z, Ikeda M. Quantitative relation of urinary phenol levels to breath zone benzene concentrations; a factory survey. Br J Ind Med 1986; 43: ) Hirayama T, Ikeda M. Applicability of activated carbon felt to the dosimetry of solvent vapor mixture. Am Ind Hyg Ass J 1979; 40: ) Koizumi A, Ikeda M. Dynamics of methanol adsorption on carbon felt with special reference to interaction with coexisting toluene. Ind Health 1982; 20: ) Koizumi A, Ikeda M. A servomechanism for vapor concentration control in experimental exposure chambers. Am Ind Hyg Ass J 1981; 42: ) Kumai M, Koizumi A, Morita K, Ikeda M. An exposure system for organic solvent vapor. Bull Environ Contam Toxicol 1984; 32:

9 DESORPTION OF SOLVENTS FROM CARBON CLOTH 81 12) Sidzu, KS. An assessment of collection efficiency of some environmental contaminants on activated carbon. Arch Environ Contam Toxicol 1983; 12: ) Pannwitz K-H. Diffusion coefficients. Drager Review 1984; 52: ) Inoue T, Takeuchi Y, Hisanaga N, Ono Y, Iwata M, Ogata M, Saito K, Sakurai H, Hara I, Matsushita T, Ikeda M. A nationwide survey on organic solvent components in various solvent products: Part I. Homogeneous products such as thinners, degreasers and reagents. Ind Health 1983; 21: ) Kumai M, Koizumi A, Saito K, Sakurai H, Inoue T, Takeuchi Y, Hara I, Ogata M, Matsushita T, Ikeda M. A nationwide survey on organic solvent components in various solvent products: Part 11. Heterogeneous products such as paints, inks and adhessives. Ind Health 1983; 21: ) Kagakubushitsu Yoran (Environmental Chemicals Handbook), Reports of a working group to the Environment Agency, the Government of Japan, To be published from Maruzen Publishers, Tokyo. (In Japanese) 17) Pozzoli L, Cottica D. An overview of the effects of temperature, pressure, humidity, storage and face velocity. Presented at International Symposium on Workplace Air Monitoring by Diffusive Sampling An Alternative Approach, held in Luxembourg on 22nd-26th September, 1986.