Tobacco Documents Online

CO 3 Figure 6. Fiber clump found in water sample. Transmission electron micrograph. 91 N O W H 0 ~ 0 ~ tA

"' Sample Preparation Errors After a sample has been collected on a filter surface, additional processing is necessary prior to examination in the electron microscope. A variety of methods can be used and each can be the source of significant errors. Perhaps the most serious of all is the loss of a significant number of fibers during the clearing or dissolution of the filter. The "cold finger" apparatus is commonly used to clear cellulose ester (Millipore) membranes, and the Jaffe wick method is used for clearing Nuclepore membranes. Both depend on dissolving the polymer in solvent vapors with the subsequent deposition of the entrapped particles on the carbon substrate. Some particles will always be washed away as the polymer is removed. How many and how consistently are very difficult to quantify. Beaman et al. [8], estimate that the losses can be as high as 50 percent for amphibole fibers. Extreme care must be exercised to avoid flooding when using the Jaffe wick method and to control the rate of boiling when clearing by the "cold finger" method. In many cases, a sample might be contaminated with excessive organic material which interferes with the examination of the sample. Removal of the organic material can be accomplished by low temperature ashing followed by redispersion and deposition on a second membrane filter. Although this may be a necessary step, it can lead to serious clumping of fibers. Furthermore, the redispersion can alter the size distribution of the fibers. Chrysotile asbestos, for example, is extremely sensitive to dispersing agents such as Aerosol OT. Another technique that is sometimes used in conjunction with low temperature ashing is the so-called rub-out method. This is useful for reducing the size of large extraneous particles, but does result in a radical change in the fiber dimensions. This method should not be used if the analyst is required to report fiber counts and fiber dimensions. It can only be used to estimate the total mass of fiber present. In general, sample preparation errors lead to an understatement of the number of fibers present in a sample and can distort the size distribution. Some analysts multiply the counts by a factor which was established on the basis of a few controlled experiments. This•practice could only be considered valid if the factor was determined for conditions identical to the reported analysis. This would require the analysis of a standard sample along with each group of unknown samples. Fiber Identification Errors The identification, or mis-identification, of the fiber spgcies present can lead to either positive or negative errors in total fiber counts. With extremely fine fibers positive identification using electron beam techniques is very difficult. Diffraction patterns may have only a few discernible spots and can also be quite fugative. Elemental analyses by x-ray emission can also be erroneous due to the influence of nearby particles. Fiber identification errors can be minimized by adequate operator training. Cer- tainly, critical samples should be analyzed only by experienced operators. Laboratory Contamination Because of the extremely low levels of fibers encountered in environmental samples and the very small sample size, contamination of the specimens can be a serious source of error. Most laboratories concerned with fiber analysis have handled bulk fibers for many reasons. Fibers can also be present in the other media used to process the samples. Good housekeeping practices can keep laboratory contamination to a minimum. It is advisable to handle all samples in an isolated area. A clean air hood equipped with HEPA Filters is most desirable. Obviously, no bulk fibers should be handled in this area. Finally, all solvents should be filtered immediately prior to use. Never rely on the fact that distilled water or other solvents, regardless of their purity, will be fiber free. Finally, it is advisable to run a blank sample through all of the steps of the procedure, along with each group of samples being analyzed. 292

Work to be Done It is obvious from the foregoing discussion that the analysis of environmental samples for asbestos fiber is far from precise. Large errors can be the result of normal random variations and also the manipulations required for sample preparation. It is further obvious that additional work should be done to establish techniques which will minimize the controllable errors. First, and foremost, among the tasks to be accomplished is to establish an acceptable standard procedure for fiber analysis. Work of this type is currently underway in several laboratories. This should be pursued with vigor so that methodology can be specified as soon as possible. Second, and concurrent with the methodology development, should be a systematic study of filter clearing techniques. The objectives of this task would be to better describe the losses which can occur, and to seek imporvements which might give smaller and more consistent losses. Finally, serious consideration should be given to the preparation of a standard dispersion which could be used for comparative studies between laboratories. Such a standard dispersion would also be useful to assist in the quantification of the errors introduced by the various analytical steps. Reporting Results Because of the variety of procedures currently employed and the magnitude of the errors, it is important that as much information as possible be included with fiber analysis reports. This information should include: Sampling conditions Volume filtered Sample preparation method Number of fibers and fields counted Blank counts Identification problems Fiber dimensions This information is absolutely essential. Too many reports are published which show only the number of fibers found in an environmental sample without any background information. Without this information, it is impossible to evaluate the true significance of any and all fiber analyses. References [i] Beckett, S. T. and Attfield, M. D., Inter-Laboratory Comparison of Asbestos Fibers Samples on Membrane Filters, Ann. Occup. HIg. 17, (1974). [2] Curtis, P. A. and Bierbaum, P. J., Technological Feasibility of the 2 Fibers/cc Asbestos Standard in Asbestos Textile Facilities, Amer. Ind. Hyg. Assn. J., 115-125 (February 1975). [3] Rajhans, G. S. and Bragg, G. M., A Statistical Analysis of Asbestos Fiber Counting in the Laboratory and Industrial Environment, Amer. Ind. 1!yg. Assn. J., 909-915 (December 1975). [4] Walton, W. H., Attfield, M. D., and Beckett, S. T., An International Comparison of Counts of Airborne Asbestos Fibers Sampled on Membrane Filters, Ann. Occup. ~Yg. 19, 215-224 (1976). 293 2063105087

[5] Cunningham, H. M. -and Pontrefract, R., Asbestos Fibers in Beverages and Drinking Water, Nature London , 232, 332-333 (1971). [6] Durham, R. W., and Pang, T., Asbestos Fibers in Lake Superior, Water ualit Param- meters, ASTM STP573, American Society for Testing and Materials, pp 5-13 (19 [7] Rohl, A. N., Langer, A. M., and Selikoff, I. J. , Environmental Asbestos Pollution Related to Use of Quarried Serpentine Rock, Science 196, 1319 (June 1977). [8] Beaman, 0. R. and File, 0. M., Quantitative Determination of Asbestos Fiber Concen- trations, Anal. Chem. 48, 101-110 (1976). [9] Anderson, C. H. and MacArthur Long, J. Preliminary Interim Procedure for Fibrous Asbestos, Analytical Chemistry Branch, USEPA, Athens, Georgia (July 31, 1976). [10] Miller, I. and Freund, J. E., Probability and Statistics for Engineers, Second Edition, pp. 77-82 (Prentiss-Hall, New Jersey, 1977 . [11] Brown, A. L. Jr., Taylor, W. F., and Carter, R. E. , The Reliability of Measures of Amphibole Fiber Concentration in Water, Environmental Research 12, 150-160 (1976). Discussion D. SARVADI: Are you familiar with the NIOSH proficiency analytical testing program, and do you have any feel for the inter- and intra-laboratory work they are doing on asbestos counts? J. LEINEWEBER: They have done a fairly credible job on making inter- and intra- laboratory comparisons on standard samples, and even within one laboratory in attempting to compare the results of a group of operators. They have come a lot farther with optical counting than we have with EM counting. There are still problems, but I think they have their situation under a little better control than we do. 294