Document VJ2mqXK094Kxrp8745a9eOz8j

^ u JIn tiffs EXHIBIT WCD-371 v, To: J. J. Travers From: W, C. Forbes Subject: Evaluation of proposed FDA teat for "Determination of Asbestos in Talc by Optical Microscopy". In compliance with your letter of September 14th, I have evaluated the test for "Determination of Asbestos in Talc by Optical Microscopy" which has been proposed by the FDA. In my opinion the'test is completely unsuitable for the evaluation of asbestos abundance in talc. j The FDA procedure states that mineral abundances can be determined by counting the number of grains of that mineral which occur within a grain mount. This approach is both naive and totally incorrect. The basic flow of the FDA procedure is that it completely ignores the size of the grains being counted. A chrysotile grain 1000 x 100 microns in size is treated exactly the way as a grain measuring 10 x 1 microns. In this case the actual volume abundance varies by 4 orders of magnitude, ' The only valid method for estimating mineral abundances using the polarizing microscope is the point count method.' To briefly describe the method, a thin section or grain mount is examined at regular intervals, the interval spacing being selected to achieve a uniform coverage of the entire sample'. At each sampling site, the mineral variety under the cross-hair is identified and individually tabulated. After covering the entire slide, the number of "hits" for each individual mineral species are summed and recal culated to percentage values. The number of times a given mineral is encount ered is, of course, proportional to the area occupied by that mineral in the entire sample. The areal percentage obtained is meaningful only when compared to the areas of all other mineral components. Thus, the PDA procedure has absolutely no value if isolated grains are counted with no regard to the size of these grains or the abundance of other mineral species. Objections to the FDA test on technical grounds are summarized below: 1, The FDA test is predicated on the assumptions of: (a) the "optical uniqueness" of asbestos-, (b) uniformity of the sample material and^ (c) uniform distribution of the sample over the slide. These assumptions were found to be invalid. - a. Chrysotile is quoted as having indices of refraction substan tially less than 1.574 and therefore should be easily detectable in an oil of that index. It should be noted that chrysotile invariably consists as an aggregate of fibers variously disoriented so that a great deal of optical Goodhue WCD 000016 ' ' ' -2- corapensation exists. The true maximum and minimum indices are probably never measured. Also talo flakes and fibers oriented with their c_ crystallographic axes parallel to the. stage will exhibit an index of refraction between 1.536 and 1.590 and on cursory examination may be identified as ohrysotile. Tho probability for operator error is very large and will likely produce over estimates of ohrysotile abundances. ', An additional complication is that iron and aluminum isubstition in chrysot|le will raise the indices of ohrysotile to levels d o s to those of talc. However, the chanoe of this happening is slight. b. The mineral contents of the sample must b e .completely uniform if bias is to be avoided. Asbestiform minerals usually occur as fibrous aggregates even after grinding.Xf the fibers of one such clot were oounted, - and this "count" extrapolated over the entire slide area, a highly exaggerat ed asbestiform content would result. Optical examination of several talc samples supplied by Avon showed substantial inhomogenity in mineral abundances. c. A very important assumption is that the sample can be distribut ' ed evenly over the slide area (l8ram square). This could not be achieved in several attempts. Fine grained materials such as ground talo tend to aggregate into large multi-grain particles which resist mechanical attempts to achieve a uniform particle distribution. Thorough mixing of the talo sample with the immersion oil also failed to achieve a uniform grain distribution. A talc --- water slurry was evaporated onto the slide,- however, this did not produce a marked improvement in sample distribution. The homogenity of the sample distribution was tested in the following way. One milligram samples of 815-157 and "Western Treated" talc were prepared following the FDA procedure. For 815-157 the number of diopside grains observed in a 0.64mm2 field of view were tabulated; the slide was then shifted 0.5mm and the procedure repeated. Fifteen such counts were made per soan. Four other such scans were made at 4 intervals. The . average number of diopside grains observed at each "sample site" and the standard deviations are listed in Table 1. The large spread in the averages (0.13-5.8grains/site) is immediately evident, indicating an almost complete lack of homogeneity. This is reflected in the standard deviations which are almost as large as the averages themselves. The-results for the "Western Treated" talo are listed in Table 2 and show a similar lack of homogenity. Goodhue WCD 000017 ' -3- Por this elide, carbonate, quartz and. dropside grains were oounted. There is a six-fold difference between the maximum and minimum averages with corres pondingly high standard deviations. The lack of homogeneity precludes any accurate estimate of mineral abundances based on a ''grain count". It should bo noted that the scanning and counting procedure is much more rigorous than that called for by the PDA. Only clearly recognizable grains were counted and the scan was made at regular intervals. This avoids the bias introduced by judgement decisions |(chrysotile vs. fibrous talc?) and irregular scanning methods. . 2. A grain mount is not a two-dimensional surfaoe. The immersion oil + grains may have a thickness of up to 50**75 microns. This can result in mineral segregations based on density differences. As microscopes have a ' 5-7 micron depth of focus at 400 magnification, apparent mineral abundances will ba a function of the deptly of the focal plane within the grain mount. 3. The PDA test calls for a 1 milligram sample. As fine grained materials may he hydroscopic, a large error in sample weight oan he introduced at an'early stage of the test. 4 . The test calls for counting of asbestos fibers using an un specified scanning procedure, When I50 fibers are counted, total asbestos is "determined" by the ratio of area scanned to total slide area. As the area scanned is an estimate and the scan steps may be irregular, large errors will be introduced, 5. Amphibole grains in the 5 1 2 micron size bracket will not be clearly apparent in 1.590 immersion oil as the test procedure implies. the small grain size reduces relief and many small amphibole grains will be overlooked. 6. The test does not specify the grain size of the sample to be examined. The obvious extreme is to have a 1 mg. sample which consists of a single talc flake and thus has 0$ impurities. 7 . This procedure is extremely tedious and there is a Btrong ' possibility of operator error after 2-3 hours work. 8 . At one sample site (768-206), 26 grains were found in a * ,65mm area which initially met the PDA criteria as chrysotile, i.e. index of refraction less than 1 .574, and a length/width ratio greater than 3 . Subsequent investigation showed all 26 grains to be talc, either as fibrous talc or the curled edges of talc flakes. There is a very strong possibility Goodhue WCD 000018 -Ar- Ot misidentifiction,. resulting In overestimates of chrysotile abundances. Respectfully submitted, September 5/ 1973 Goodhue WCD 000019 Table 1 Average number of dropside grains per sample site (0,64 ffl2) Sample #815-157. Percentage of slide covered equals 15$. Soan # 1 * 2 3' 4 i 1: Mean number 3-2 ; 5*8 1*4 0.13 dfopside grains Standard 3.7 4.9 12 -- -- Deviation - ,% , * Interval between scans is 4mm. . Area of individual scan site equals 0.64mm2 Distance between scan sites equals 0.5mm.' 5! 5.5 3.3 Goodhue WCD 000020 Table 2 Average number of carbonate, quartz, and dropside grains per sample site. Sample #"Y/estern Treated Talc". ' Percentage of slide covered equals 30$. . Scan # ' V) 12 5 6 7 8 9 10 r 4 Mean number couPfcIinS 7-3" 11.8 io.o 8.0 5.4 7.0 5.8 4.0 3.8 2.0 Standard Deviation 2.7 4.7 3.5 3.1 3.0 5.0 2.6 2.3 2.0 1.5 * n + v vai befwJ ***' &ca.n& 1 l m<n . ' ? Goodhue WCD 000021