Tobacco Documents Online

Table 3. Intra-lab replicate analyses for chrysotile, (high cities, blinds external audit). Lab A Site Sample period (ng/m3) Average % absolute latation 1971 (quarter) 1st 2nd Mean deviation from mean A 1 1.7 1.2 1.5 33 4 2.1 1.8 2.0 8 B 1 4.0 6.7 5.2 25 4 7.4 7.2 7.3 3 C 1 4.0 3.7 3.9 4 4 5.3 1.5 3.4 56 D 1 9.4 4.4 6.9 39 4 11.0 3.1 7.0 57 E 1 8.4 8.0 8.2 3 4 3.0 4.6 3.8 23 1972 (quarter) A , 1 1.6 26.7 14.2 88 2 3.7 2.8 3.2 34 B 1 6.1 2.8 4.5 73 2 6.6 1.4 4.0 53 3 9.6 1.6 5.6 71 Z 2 0.4 27.7 14.1 97 C 1 4.2 2.5 3.4 25 2 0.7 1.2 1.0 25 D 1 6.8 2.0 4.4 50 2 0.8 2.8 1.8 56 E 1 18.8 11.8 15.3 23 2 3.1 1.6 2.3 31 Average 40 359 0 a w ~ 0 w ~ w ~

(Cs Table 5. Replicate analysis for airborne chrysotile between laboratories (in ng/m3). City Quarter Lab B Lab A Mean Average % absolute deviation from mean (Samples collected in 1969) A 2 0.4 1.8 1.1 63 3 95 3.9 49 202 4 0.7 15.6 8.2 182 B 2 3.9 5.3 4.6 15 3 4.2 6.7 5.5 22 4 8.0 3.5 5.8 39 Z 2 0.4 0.5 0.5 11 3 1.2 0.5 0.9 41 4 38 0.4 19.2 98 C 2 1.3 11.1 6.2 80 3 1.5 0.7 1.1 36 4 40.0 0.5 20.3 98 0 2 1.1 0.4 0.8 47 3 11.8 0.5 6.2 92 4 0.7 0.5 0.6 17 E 2 4.4 1.0 2.7 63 3 0.7 1.1 0.9 22 4 2.1 0.6 1.4 56 (Samples collected in 1970) A 1 1.0 1.4 1.2 17 B 1 6.5 1.3 3.9 67 Z 1 1.2 0.9 1.1 14 C 1 2.2 0.6 1.4 57 D 1 1.5 0.8 1.2 30 E 1 4.6 1.5 3.1 51 Average 59 Table 6. Mass methods comparisons, count vs. volume (in ng/m3). Mass by Mass computed from Ratio Sample fibril count (C) "fiber" volume-density (V) C/V 2 216 141 1.5 3 1,674 476 3.5 p a 361 ~ 0 ~ ~ w

Table 4. Intra-lab replicates for chrysotile, internal Q.C. Lab B Site Sample period (ng/m3) Average % absolute location quarter-year 1st 2nd Mean deviation from mean F 1-70 0.8 0.4 0.6 33 G 4-69 20.3 12.6 16.5 23 A 3-69 110 80 95 16 H 3-69 25.3 13.5 4.5 14.4 48 I 3-69 5.4 3.3 4.4 23 ,] 2-70 1.1 0.1 0.6 83 8 3-69 5.2 3.1 4.2 24 C 4-69 62.3 17.7 40 56 D 4-69 1.3 0.0 0.7 85 Z 3-69 50 27 38 31 K 2-70 5.2 1.0 3.1 . 68 L 2-69 1.7 1.1 1.4 21 L 1-70 6.3 2.4 4.4 44 M 2-69 5.3 2.1 3.7 43 N 2-69 25 5.3 1.3 10.2 95 0 3-69 19.3 16.6 18 7 Average 43 data given in Table 5 one may note that of the 5 value sets (of the 24 given) which are not within a factor of 10 of each other, findings above 10 nanograms per cubic meter (which is twice the average value for the set including the high values) are involved, and that three of the high values were reported by one lab, and two by the other. It is possible that the samples of high value (for which the agreement is the poorest) have large particles of asbestos and are thus more inhomogenous than are the samples with lower asbestos contents. It is also of interest that in a comparison of mass by the sample count versus standard count method with a mass computed from fiber volume from direct fiber counts of replicates, a bias of the mass method toward higher readings is noted in Table 6. 360

SS Discussion 0. MENIS: Are you familiar with the work of Spurny et al. on the Nuclepore and membrane filter retention and would you like to comment on it, because there appears to be some loss, 20 percent internal loss. Also, the question of prefiltering a lot of junk beforehand seems to be attractive. THOMPSON: Let me tell you why I don't like prefiltering. When you use a filter that is composed of a bunch of fibers matted together as are our glass fiber filters and the membrane filters we use (that's not so with Nuclepore, obviously; they drill holes in their's) then you are dealing with a brush pile of fibers, and as you put that particular matter on there I am convinced you go from a surface of the fiber to a particulate laden surface. After your first few minutes of sampling on a 24 hour basis, it's my belief that what you are dealing with is particulate matter filtration, and not filter filtration anymore. The particulate matter itself is now your filter, and to support that I will tell you of a two-week sampling shot I made to collect massive quantities of materials for detailed chemical analysis. I wanted the total elemental composition of particulate matter so I would know what I was up against analytically. Nobody had ever done that. You are filtering with particulate matter and here is why I think that. You start off with a high volume sampler at about 60 cfm, and if you run about 10 days you find that the flow is down to a constant of about 30 cfm. You put in 8xlO cellulose acetate filter on the same type of device, calibrate it to draw what it should be, 60 cfm with the glass fiber filter, and you sample about 35 cfm through that membrane filter. After about 10 days you will be filtering about the rate of about 30 cfm. If you throw another kind of filter substrate on there you see the same thing. That loading, I think, is your terminal loading of particulate matter that affects flow, but I am convinced you are filtering with particulate matter. I do not like prefiltration for that reason. You are going to get stuff hung on there; you are going to lose material, and that filter is not smart enough to open up and let whatever it is you want through quantitatively. It just won't do it. We tried it and I have had notable lack of success with that approach. It sounds nice, that you could screen out the lumps, but in practice it doesn't work that way. I don't think it feasible, and I have never been able to accumulate data that were very satisfying. N O 363 W ~ 0 ~ .~ w w

e The consideration of the health effects of asbestos fibers, fibrils, fiber size, etc., will be considered elsewhere in this symposium. For the problem of monitoring for the definition of air quality on a long-term basis which conceivably could be used for regulatory purposes for citing standards and for control measures and possibly for interpretations with rpspect to human health, in my judgment the mass method outlined in this paper is a superior method. It may avoid two very significant problems in the estimation of the chrysotile content of air as measured in collected particulate matter. One is the problem of homogeneity which is a problem with every sample that one obtains from the air. For example, asbestos fibers may be put into the air by construction/dem- olition. The other is the problem of what constitutes a fiber? This is in a sense another aspect of the same problem. If one had a uniform distribution of fibrils over a sample, the fibrilar estimation would probably be comparable by both methods. If however, one obtains a fiber or two, here and there, obviously then the sample is automatically inhomogenous since fibers could conceivably consist of 105 fibrils. In the method where the fibers and free fibrils are ground ultrasonically, the resulting particle size distribution should be a function of the energy put in. The procedure described should then yield a homogenous mixture. It is not suggested that this approach is the final answer for all monitoring problems, or that it addresses anything at all concerning fiber length in real air samples of any form of asbestos, or fiber size distribution. It is patently apparent that information of this nature cannot be obtained reliably using a method wherein the material has been subjected to diminution. References [1] Walter C. McCrone Associates, Inc. reports to E. C. Tabor, EPA, 1966_ [2] Henry, W. M. et al., Development of a Rapid Survey Method of Sampling and Analysis for Asbestos in Ambient Air, Battelle Columbus Report, Contract No. CPA 22-69-110 (February 1972). [3] Thompson, R. J. and Morgan, G. B. , Determination of Asbestos in Ambient Air. Pro- ceedings, International Symposium on Identification and Measurement of Environmental Pollutants, 154-157 (June 1971). [4] Rickards, Anthony L., Estimation of Submicrogram Quantities of Chrysotile Asbestos by Electron Microscopy. Anal. Chem., 45, 809-811 (1973). [5] Spiel, S. and Fenner, E. , personal communication. [6] Nicholson, W.J. et. al., Measurement of asbestos in Ambient Air, Mount Sinai School of Medicine, Contract CPA 70-92 (1971). 362

National Bureau of Standards Special Publication 506. Proceedings of the Workshop on Asbestos: Definitions and Measurement Methods held at NBS, Gaithersburg, MD, July 18-20, 1977. (Issued November 1978) AMBIENT AIR MONITORING FOR CHRYSOTILE IN THE UNITED STATES Richard J. Thompson Analytical Chemistry Branch Environmental Monitoring Support Laboratory Environmental Protection Agency Research Triangle Park, North Carolina 27711 Abstract The only continuing national air monitoring has been conducted by the National Air Surveillance Network. The objective is long term trend assessment of air quality. The information has proven of value in setting standards, in consideration of health effects, in estimation of economic effects, and in showing patterns of pollutant distribution in both urban and non-urban areas. In order to provide samples which could be analyzed for constituents not determinable in particulate matter samples collected with glass- fiber filters, a membrane sampling network was instituted. The only analyses of the samples conducted thus far has been for airborne asbestos using in part a method developed under contract which provides for the determination of the mass of chrysotile in the particulate samples. A viewpoint will be presented on the method needed for air monitoring and an assessment of the mass method as the most suitable for this purpose. Data obtained will be examined which will include information on inter- and intra-laboratory replication. Key Words: Air monitoring; airborne particulate; asbestos; chrysotile; filters. National air monitoring had its inception in a Public Health Service survey of protein in airborne particulate matter conducted at seventeen sites in 1953-54. Sufficient amounts of material were collected using glass fiber filters to permit chemical analyses as well as the determination of total suspended particulate matter. In 1955 the Federal Air Pollution Research and Technical Assistance Act, Public Law 159, 84th Congress, was passed. The Network was expanded to 66 stations nationwide for every year sampling and 110 urban and 51 non-urban stations for an intermittent sampling and all the then 48 states and Alaska, Hawaii, and the Commonwealth of Puerto Rico. Currently some 270 stations collect particular matter in the National Air Surveillance Network (NASN). Certain constituents of the particulate matter collected could not be determined when glass fiber filters were used. As a result, a.embrane sampling network was instituted within the NASN in 1969. Untit recently 51 stations were maintained, but currently only some two dozen are operated. Samples collected on the cellulose acetate membrane filters can be analyzed for constituents of glass such as boron and silica. The only analyses of the samples conducted thus far, however, have been for the mass of chrysotile in the particulate samples using the method developed by EPA under contract, or a variation of this method. Ambient air samples collected by EPA had been analyzed by contract [1]i in 1966 by both ordinary light field techniques and with dispersion staining with the optical Figureckets indicate the literature references at the end of this paper. Preceding page blank 355 2063105147

C Table 1. Replication filter section determination (TEM). Site Location . ng Chrysotile/m3 air Average A Near use site (1970) 280, 260 270 A Near use site (1971) 110, 86 98 A Near use site (1971) 7900, 7200, 9700 8200 B Near use site (1970) 28, 40 34 B Near use site (1971) 130, 117 124 A la Remote site 0.112, 0.102 , 0.147 0.12 A 1a Remote site 0.094, 0.119 , 0.106 0.10 A 2 Remote site 0.028, 0.024 0.026 0.03 a Two samples taken concurrently. Table 2. Method comparison check. Sample Radio assay TEM Difference % 1 8.3 11.0 29 2 34.0 40.0 15 3 17.6 20.8 15 4 4.7 4.0 18 Quarterly composites constructed fros the 51 network sites were analyzed for chrysotile. The average of the analyses of some 521 of these composites samples is 2.6 nanograms of chrysotile per cubic meter of air sampled. The samples analyzed were quarterly composites of those samples collected through the second quarter of 1973. In Table 3 the data replicate slices of quarterly composites are given; these samples were provided to the contract laboratory on a blind basis for the purpose of an external audit. The percent of absolute deviation from the average of 22 individual sample sets was 40 percent. In Table 4 the internal QC replication of a different laboratory is given; note that in the analyses of 16 sets of replicate sections from samples, the percent absolute deviation from the average is 43. In Table 5 the data are shown obtained from a sample split program between the two laboratories conducted on a blind basis to the participating laboratories. Note that some of these data are cammon to Tables 3 and 4 also. It is of interest that in 24 sample sets of samples analyzed by each laboratory in some cases the data shown are averages of replicates within one of the laboratories. The percent absolute deviation from the average is 59 percent. From an examination of these data one gets the impression that the average percent deviations are roughly the same between laboratories and within laboratories. It is also of interest that in the inter-laboratory 358

layer of carbon, sectioned, the filter substrate removed with acetone, and the carbon film placed on a 200 mesh copper electron microscope support grid. The sample is examined at 20,000 X with a transmitting electron microscope (TEM), and the chrysotile fibers (fibrils) determined by counting grid openings to obtain a count of 100, or a minimum of 5 grid openings if a count of 100 is not obtained. The mass equivalent of the count is obtained from a working curve constructed from data derived by counting samples containing known'amounts of added chrysotile. Synthetic standards were made by addition of known quantities of pulverized sea sand, particulate matter collected from air, (ignited to 800 °C to destroy chrysotile) fly ash, and water. The samples so prepared were subjected to the procedure described. It was established that the curve obtained by plotting count versus mass of known chrysotile added may be derived satisfactorily by adding weighed known chrysotile to water alone. Asbestos obtained from commercial supplies of mineralogical samples included chrysotile, amosite (which constitutes almost all of the non-chrysotile asbestos used in the U.S. commercially), and crocidolite. The chrysotile used in the developmental work was a "respirable pure" white chrysotile obtained from Johns Manville. Advantages to be noted by this procedure are that if the ultrasonification is complete, there should be a uniform distribution of fibrils of a spectrum of lengths which is related to the energy of ultrasonification applied and that the uniformity of the sample and the (comparatively) tremendous number of fibriTs of chrysotile to be determined enhances the statistical possibility of replication, at a reasonable level, using simplified counting techniques. There is of course the possibility of a recognition problem of any material in air which would behave as does chrysotile and appear as does chrysotile in the carbon replica which contains fibrils. The diameters of chrysotile fibrils from diverse geographical origins are said to be within a very narrow size distribution approximating 30 nm [4]. Confirmational data such as SAED and atom ratios by probe can be obtained. The method was applied to samples taken from ambient air in urban areas where asbestos might be expected to be found on the basis of industrial activity and to samples taken from a remote location. Representative data are given in Table 1. The remote location was chosen as being as far from road traffic as could be found in one day's commuting from the laboratory and at a site where power existed. The replication at low levels was surprisingly good; triplicate samples gave mean values of approximately 0.1 nanograms per cubic meter with a replication within 10 percent of the value measured. Even at the level of 0.03 nanograms per cubic meter the spread of measured values was within 10 percent of the value measured. For samples which contain tens of nanograms per cubic meter, replication was achieved within 50 percent of the average value noted. The method was checked by a phenomenologically different approach wherein samples of asbestos, activated by neutron irradiation, were blown into a chamber and recollected. As shown in Table 2, radioassays and the TEM estimates agreed within 30 percent of the average value of the two readings recorded by the different methods. The replication found within the method in the limited comparison between methods lead one to believe that the two phenomenologically independent methods gave comparable results and that the electron microscopy method gave replication in the vicinity of 50 percent. It may also be noted that the Stokes diameter was checked at sites downwind from a point source at distances of 1 and 2 miles, the predominant diameter distribution in terms of the mass seemed to be in the fraction of asbestos particles which were in the 8 to 16 micrometers diameter stage of the collection device. 357

I microscope and by scanning electron microscopy using magnifications up to 18,500 X. Fibers were noted which were believed to be chrysotile. Because chrysotile comprises approximately 95 percent of the total asbestos used in the U.S., chrysotile was the form of asbestos for which a monitoring method was desired. It was also decided that a mass method would be more appropriate for a survey tool than the fiber count method traditionally used in he`alth effects studies. Although adverse health effects of asbestos on man is the reason for the interest in asbestos in air (airborne asbestos having no known adverse economic effect of significance), the optical method did not seem to be appropriate for analysis of ambient air monitoring samples, albeit effective in the work room where asbestos fibers of optically detectable size were known to prevail. An electron microscopic method would obviously be desired since submicroscopic particles were demonstrated. Chemical analyses obviously would not be an appropriate survey tool, unless sufficiently high concentrations of chrysotile would be found to permit x-ray diffraction as a possible tool for application. Using brake lining consumption figures and assuming that all of the asbestos remains airborne, it was estimated that the chrysotile content of air could under these conditions be in the nanograms per cubic meter range. The objective of monitoring using the mass method would be to determine the quantity of chrysotile asbestos in ambient air. Thus the objective did not include a need for a knowledge of fiber length, fiber size distribution, and other factors that could be obtainable if the asbestos fibers per se as collected were to be examined. Since a known standard of asbestos in air was perceived a philosophical nightmare, the problem of a quantitative recovery of fibers from air without diminution or destruction, and a quantitative estimation of fiber size and length was considered to be unobtainable for routine application for monitoring chrysotile levels in ambient air. One problem in particular was the problem of what is to be counted if one is to count that material which is taken from air directly without alteration and if particle recognition, using the morphology of chrysotile was to be employed as a working tool (chrysotile fibrils are cylindrical tubules). One would be posed with the problem of fibrils of chrysotile in ambient air samples of some 30 nm in diameter being present and fibers of chrysotile composed of hundreds of thousands of fibrils being also possibly present in the same sample (if taken at an urban sight where construction/demolition might be ensuing). How would one then handle the problem of counting a fiber and counting an artifact fibril from that fiber? Would they both be counted as fibers (assuming they met aspect ratio criteria)? Fibrils might be produced by one handling technique which might not be produced by what was thought to be the identical technique in the hands of another operator. Furthermore, the non-homogeneity of such a sample would make the counting statistics very unfavorable toward application at a reasonable level. Reasonable in this case is defined as being of the accustomed precision expected by analytical techniques that are more objective and less subjective than particle recognition based on morphology. The method developed for monitoring of chrysotile in ambient air for EPA under contract [2] has been described in detail elsewhere [2,3]. The method was discussed in detail in a conference held at the Research Triangle Park, North Carolina, by EPA in July 1970 attended by representatives of all U.S. laboratories then known to be working on asbestos estimation from air samples and included representation of the United Kingdom. A method similar to the EPA method in some details was developed independently [4]. The method developed for EPA differed slightly from other mass methods employed at that time. The method in use at Johns-Manville used a gold labeling technique to achieve quantitation of chrysotile; a watch glass was used to grind a sample suspended in aqyl acetate on a microscope slide. Gold 198 was added prior to ashing and the efficiencies of recovery was estimated by radioactivity measurements of the gold, the chrysotile was assumed to behave as did the gold standard [5]. The method then employed by Mount Sinai Laboratories involved grinding with a glass and spiking the samples with known weights of chrysotile from which a recovery factor was derived and used in calculation [6]. The method developed for EPA for the determination of the mass of chrysotile In ambient samples involves starting with a portion of a particulate matter sample taken on a cellulose acetate filter, ashing at low temperature, suspending in water with the aid of a surfactant, and grinding to fibrils by ultrasonicating at high energy. The now homogenous samples containing shattered fibrils is filtered on a membrane filter coated with a 20 n. 356