Tobacco Documents Online

Cs 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) TRANSMISSION ELECTRON MICROSCOPICAL METHODS FOR THE DETERMINATION OF ASBESTOS Ian M. Stewart Walter C. McCrone Associates, Inc. Chicago, IL 60616 Abstract Three 'criteria are given for the identification of a mineral fragment as asbestos: morphology, crystallography, and chemistry. The derivation of this information in the transmission electron microscope is discussed. Quantification of asbestos fiber content in an environmental sample is considered and currently practiced techniques for quantification both by mass and by number are reviewed. Key Words: Analysis; amphibole; asbestos; electron diffraction; electron microscopy; fibers; transmission electron microscopy; x-ray energy analysis. The first meeting on methodology for determination of asbestos by electron microscopy was held almost exactly seven years ago. Sponsored by, as it then was, the National Air Pollution Control Administration, it was attended by about a dozen people. The explosion of interest in asbestos has led to a series of methodology meetings, particularly over the last two or three years, culminating in the massive attendance at the present meeting. It is clear, therefore, that there is considerable interest in asbestos and in particular, asbestos methodologies. There is thus no need to reiterate the reasons for this interest here. What may be less obvious however, is why there should be such a necessity for the development of electron microscopical methods. Figure I shows an electron micrograph of a standard suspension of an ultrasonerated chrysotile sample which has been prepared to simulate material shed from asbestos filters used for parenteral drugs. The size range represented is quite wide and very closely approximates that which has been found in liquids filtered through an asbestos filter. If such a sample were to be characterized entirely by light microscopical methods, much of the material which can be seen in the electron microscope, for example fibers A and B in Figure 1, would be completely omitted. Figure 2 is an environmental sample, taken approximately three miles down stream from an asbestos plant and here again we have material below the detection range of the light microscope. The level of asbestos fibers determined by electron microscopy in this case was of the order of 108 - 10s fibers/liter, several orders of magnitude higher than would have been determined if the light microscope was used. Again, in water samples from the Duluth and Silver Bay areas, the number of asbestos fibers that were identified by light microscopy was virtually zero, fewer than one dozen fibers being detected in over fifty samples by this method. Nevertheless, transmission electron microscopy, as shown in Figure 3, established that there were indeed high levels of fibrous amphiboles in these samples. Clearly then, in order to satisfactorily characterize the asbestos content of such samples, electron microscopy is a necessity. Preceding page blank .~ 0 ~ 0 ~

Figure 1. Ultrasonerated chrysotile suspension simulating size distribution of fibers shed from asbestos filters used for parenteral drugs - 3200 X. 272

Figure 2. Filtered river water 3 miles downstream from an asbestos processing plant - 20.000 X. N 0 273 w r. 0 ~ 0 4 1

Figure 3. Water from western arm of Lake Superior - 12,600 X. Before discussing methods of preparing samples for examination in the electron micro- scope or for counting them, it is necessary to be sure what information we need to derive from the electron microscope in order that we can characterize a particular particle as an asbestos fiber. If one accepts the Federal Register definitions of asbestos and, from a legal standpoint, that is all that one can use at the present time, then to determine an asbestos fiber, one must show first that the material is fibrous, that is, that it has an aspect ratio of greater than 3:1 and, second, that it is a mineral of the type which is classed as asbestos by the Federal Register. The determination of the aspect ratio is quite straightforward. One measures the length and the width of the particle. The determination that the particle is indeed asbestos, however, is not so straightforward. There are basically two criteria which must be satisfied for a positive identification, certainly on the amphiboles, although for chrysotile perhaps only one of these criteria will suffice. These criteria are, firstly, that the particle in question belongs to the correct crystallographic system and has the correct crystallographic parameters for one of the asbestos minerals. Because of the unique structure of chrysotile, which will not be discussed here, the diffraction pattern of chrysotile can be regarded as sufficiently definitive without the addition of chemical information (Figure 4). In the case of the amphiboles, the diffraction patterns are less characteristic and careful diffraction work must be performed to establish that the particle is indeed an amphibole. Having established that it is an amphibole, one must then differentiate which of the several 274

S2 amphibole types it may be. This can best be performed by chemical analysis in the electron microscope. At the present time the most popular method of determining this analysis is by use of an energy dispersive x-ray analyzer, fitted to the transmission electron microscope. Figures 5 and 6 show, respectively, the electron diffraction pattern and the energy dispersive spectrum of an amphibole fiber which can be tentatively identified as the commercial asbestos "amosite" - actually a fibrous grunerite. The word 'tentatively' is used deliberately since there are -many problems associated with the interpretation of both the diffraction pattern and the energy dispersive spectrum. Thus, in general, it is prudent only to classify an amphibole as being within a certain series, such as the tremolite-actinolite series, or the cummingtonite-grunerite series. Figure 4. Chrysotile diffraction pattern. 275

Figure 5. "Amosite" diffraction pattern. © F. 4 o f Yf. . jY ~I ~~~ e.+ na.aa Co navo1e Figure 6. "Amosite" energy dispersive x-ray spectrum. 276

SS In order to determine these parameters simultaneously in the electron microscope, some idea should be given of how this information is derived. The morphology is obvious; this follows from the normal operation of the microscope as an image producing instrument. However, like all optical systems, the laws of diffraction apply in the transmission electron microscope. Thus, given an object with a periodic structure, the image of this object in the back focal plane of the objective lens will be a diffraction image related to the periodicity of the structure. In the transmission electron microscope, this image may be observed at higher magnification by adjusting the strength of one of the projection lenses, such that the back focal plane of the objective lens is in focus at the final viewing screen. As was stated above, the chemical nature of the particle under investigation can also be determined in the microscope by an energy dispersive x-ray system. This is because striking a target with a high energy electron beam will result in the emission of x-rays whose wavelengths or energies are characteristic of the chemical species at the point of impact. By suitably focusing the incident beam it is possible to isolate individual particles in the microscope and to either analyze their energies by an energy dispersive spectrometer or their wavelengths by a wavelength spectrometer. In practice the energy dispersive spectrometers are more common. They have the advantage that they detect all elements simultaneously from about sodium upwards in atomic number and they are also considerably cheaper to install on an instrument than the wavelength dispersive system which, although having a better signal to noise ratio, suffers a major disadvantage for rapid analysis in that it is sequential, analyzing only one element at a time. There are many factors which may interfere with or disturb the energy dispersive signal; factors such as particle size, shape, geometry, scatter from the instrument, and so forth, confuse the already complex chemistry of the amphiboles. These have been discussed in many other sources and will not be discussed in detail here. One should, however, be aware that such complications do occur and should interpret the spectia with appropriate caution. Having settled on criteria by which one would identify the fibers, the next problem is, "What does one wish to count or measure?" There are two philosophies which are current. One is that the important factor is to determine the number and size distribution of the fibers present as they exist in the sample. The other philosophy is that the mass concentration is important. We should discuss a little why these two schools of thought have arisen. It would seem to the lay observer, that, as yet, there is no sound medical reason in favor of determining one or the other. There are sound analytical reasons for suggesting either. The most attractive feature of the fiber number, size distribution and shape philosophy is that as well as giving information on levels that exist in the material, it also gives the size range, which may or may not be important, and much recent work has suggested that it is. It is also possible, by factoring in a geometric factor together with a density factor, to determine the mass of fiber present. One of the major drawbacks of such a method, however, is the tendency of fibers to overlap each other and also to overlap other material in the sample. It is particularly common in quarry samples, for example, to find intergrowths of chrysotile with the related serpentine mineral antigorite. Unless a good separation between the antigorite and the chrysotile is obtained, it may not be possible to positively identify the asbestos fibers and hence they will not be included in the count. Repeated over many fibers, and bearing in mind the multiplication factors which exist by virtue of the difference in area examined in the microscope relative to that represented by a membrane filter area, this can lead to quite dramatic differences in fiber counts or mass levels detected. In addition, the presence of one or two massive fibers can drastically skew the mass number, again because of the multiplication factors involved. Mass concentrations have been determined by several workers, and several methods exist for preparation of samples to determine mass reasonably accurately. These methods, developed principally by Battelle, Mt. Sinai, and Johns-Manville, and ideally, applicable only to chrysotile asbestos, all involve the reduction of more massive fibers to the so- called unit fibril of chrysotile. In some methods these fibrils are then individually measured for length, and by geometric calculations the mass is deteimined. In the Battelle method, the intercepts of fibrils along a line are counted and compared to similar counts performed on a standard mass concentration sample. The advantage claimed for such methods is that they will separate the fibrils from interfering material. One disadvantage is the several preparation steps which may be involved in preparing the sample and which may lead to either cross contamination of the sample or loss of material 277 2063105071

from the sample, leading to high and low readings, respectively. Additionally, such methods may liberate fibers which would not normally be considered free fibers and therefore presumably not hazardous. Details of these procedures have been published previously and will not be reiterated here. As regards methods for sample preparation for fiber counting without destroying the identity of the fibers, one might say there are as many variations of sample preparation methods as there are electron microscopists working in this field. The state of the art does, however, seem to have boiled down to two basic direct transfer methods, one using condensation washing and one using a wicking technique. These methods will be discussed by Dr. Anderson, who has prepared an excellent document entitled "A Preliminary Interim Procedure for Determining Fibrous Asbestos", which spells out the basic steps in preparing samples and the criteria for asbestos identification. I believe this document represents the most acceptable state of the art on asbestos determination by transmission electron microscopy at the present time. Although there have been other methods proposed, these have not received as wide favor as the direct transfer methods. These other methods include placing a drop of the fluid suspected to contain asbestos on an electron microscope grid with a calibrated micro pipette. Assuming that all the material from the drop is deposited on the grid uniformly, and knowing the volume of the micro pipette, it is possible to derive the number of fibers per unit volume of fluid. In a similar method a calibrated micro pipette is not used, but a small drop of the liquid is placed on a grid and the diameter of the area occupied by the deposited solids after the droplet has dried is measured. It is assumed that the diameter of the evaporated circle represents the diameter of the original drop and hence the volume of the drop may be calculated and again the number of fibers per unit volume determined. One of the major drawbacks of many of these direct drop emplacement methods is the difficulty in holding the liquid in such a manner that none of the drop is transferred off the grid to its surroundings, for example by wicking up between the arms of a pair of tweezers or by contact with the substrate on which the grid may be supported. An additional disadvantage is the tendency for size separation to occur within the drying drop, resulting in an uneven distribution of fibers on the grid. - In any event, in any method involving direct transfer either from a liquid or from a filter it should be borne in mind that due to the overlapping nature of the particulate species present, the possible ambiguities of interpretation of diffraction patterns and/or chemistry due to such overlaps and the inability to see many of the fibers, the number of fibers counted will, in all cases (with the exception of bad housekeeping resulting in contamination), result in a minimal number for the total fiber loading per unit volume. A truer estimate of the loading per unit volume may be made by applying corrections for such overlaps or by additionally counting those ambiguous fibers which cannot be directly identified. There is, however, no hard and fast rule as to the magnitude of such corrections. In the case of methods reducing fibers to unit fibrils and estimating mass, these will again be minimal numbers if the criterion used is that the fiber must be positively identified, as, it is more difficult in general to obtain a positive identification of a small fiber than a large one either by electron diffraction or by chemical characterization. Although this may paint a rather pessimistic picture in terms of establishing a standard using electron microscopy, some positive suggestions may be put forward. For example, if it is decided that the standard should be a certain number of asbestos fibers per unit volume, then it should be possible to set up the microscope parameters such that the microscopist can determine all fibers in a unit area quite rapidly. This can, in turn, be calibrated in terms of fibers per unit volume of the sample source. If none of these fibers are asbestos and the number is still below the statutory limit, then clearly it is not necessary to perform any identification on the fibers to determine if they are asbestos or not. Such a procedure could well be used for screening purposes. Again, a subjective opinion could be made by the microscopist as to what percentage of those fibers are asbestos. If the total fiber content was 2 or 3 times that which is permitted by the regulation but the asbestos content is clearly, say 10 percent, of the total fiber content, then again there should be no major problem. This would dramatically reduce the number of marginal cases in which the total asbestos content may be close to or exceed the statutory limit. Only in such cases would it be necessary to perform a complete and detailed analysis. It would be necessary, of course, to ensure valid documentation uf the 278

data in- those cases where it is said that the level does not exceed the statutory limit. A similar approach could also be applied to the mass method and indeed may be more readily applied if one is already estimating mass on the basis of number intercepts per unit area. In the foreseeable future it is quite conceivable that automated methods for deter- mining asbestos in the electron microscope may come to be a reality. The application of computer solution to the electron diffraction pattern as described by Fisher and Lee in these proceedings could be combined with the capability for electronically recording such diffraction patterns which is offered by the technique of scanning electron diffraction. This could then be integrated in one instrument with an x-ray energy dispersive x-ray system, and electron energy loss analysis system operating in the scanning transmission mode to provide a valuable and powerful tool for automating the asbestos identification process. It is unlikely, however, that such a tool would be applied on a routine basis, in view of the capital cost which would be involved. Thus, there remains the major problem of characterizing asbestos particles in the submicroscopic size range and doing this economically. Work is currently in hand to effect separation of asbestos from other mineral species; separation from organic material may already be achieved by such techniques as low temperature ashing. Assuming that such separation can be both successful and complete the analytical procedures may well be simplified. Until such time, however, transmission electron microscopy must remain primarily a technique applicable to the research situation and is not presently an economically viable tool for monitoring and control programs on an extensive scale. References In trying to put together specific references to techniques mentioned in this pa~er, I realized how much of my information had been absorbed through discussions both privately and at meetings such as this -- a sort of mental osmosis. The following list is therefore not complete and should be more properly regarded as suggestions for further reading. I apologize in advance to those who may feel slighted by the omission of references to their work. Descriptions of the mass method can be found in the following: Leineweber, J. P., "A Method for Determination of the Fiber Content of Water", Johns- Manvilie Research and Engineering Center, Report No. E 404-37, August 1968. Thompson, R. J. and Morgan, G. B., "Determination of Asbestos in Ambient Air," Proc. International Symposium on Identification and Measurement of Environment Pollutants, p. 154, June 1971. Descriptions of direct transfer methods from filters are given in: Anderson, C. M., "Preliminary Interim Procedure for Determining Fibrous Asbestos," July 1976. Available from Dr. C. M. Anderson. See also Dr. Anderson's paper in these proceedings. Overlap, loss, and similar problems: Knight, G., "Overlap Problems in Counting Fibers." AIHA Journal, p. 113-114, February 1975. Beaman, D. R. and File, D. M., "Quantitative Determination of Asbestos Fiber Concentrations," Anal. Chem., 48, No. 1, p. 101-110, 1976. Energy dispersive x-ray analysis is discussed in: Beaman's paper cited above and Ruud, C. 0., Barrett, C. S., Russel l, P. A., and Clark, R. L. ,"Selected Area Electron Diffraction and Energy Dispersive X-ray Analysis for the Identification of Asbestos Fibers, A Comparison," Micron, 7, p. 115-132, 1976. 279 2063105073

Several general papers on the characterization, identification, and quantification of asbestos also appear in the Proceedings of the First FDA Office of Science Summer Symposium on Electron Microscopy of Microfibers, August 1976, currently in press. Discussion C. ANDERSON: Ian, it strikes me that to determine mass and to determine the number of fibers at a certain period of time are entirely incompatible for the reason that you state-that 10 percent or even less of the fibers contribute to 90 percent of the mass. I. STEWART: That's exactly right. ANDERSON: Therefore, for any kind of precision of mass you must count possibly 100 large fibers. STEWART: Or 1000 fibrils or you just look at your intercept. But I wasn't putting it forward as being a way that we should go. You see the big problem is that you're like me, you're an analyst too, and the medical people haven't decided what they want from us -- mass data or fiber counts and sizes. If that problem is resolved, so too will many of the analytical problems. ANDERSON: I wonder if you agree that determining mass and the number of fibers in the same amount of time is almost incompatible within a certain precision? STEWART: Yes and no. You can get a mass number out. If you're too lazy to look at the statistics of the size distribution, the mass will give you an idea of whether you've got a lot of big fibers there; not always, but sometimes. Written comments by Prof. J. Zussman to Dr. Stewart's paper. J. ZUSSMAN: Dr. Stewart mentioned that fiber counts by electron microscopy would be expected to be in error on the low side, especially through overlapping particles. This effect can be lessened, of course, if specimen preparation is such as to produce not too dense a fiber population on the elm grid. I would also like to mention that there are two factors leading to erroneously high fiber counts - the use of the rub out technique, and the process of ultrasounding if too vigorous. STEWART: I agree in part. However, in the case of overlaps due to other suspended particulates, dilution may produce too low a fiber population for the data to be statisti- cally valid. I also agree on the comments an erroneously high fiber counts. The rub out technique is only valid for mass data although I know that fiber counts produced by this technique have been quoted by some people. 280