Document QXp4oBEb4k2KEYyxxX6BdLkNo

waiter c. me crone associates, inc. CONSULTING ULTRAMICROANALYSIS MICROSCOPY SMALL PARTICLE PROBLEMS SOLID-STATE CHEMISTRY 3 December 1973 Hearing Clerk Food and Drug Adm inistration Room 6-86 5600 Fisheries Lane Rockville, M aryland 20852 Gentlemen: We a r e w ritin g to com m ent on th e C o m m issio n e r's p ro p o sa l co v erin g "A sbestos P a rtic le s in Food and D rugs" as published in the F ed eral R eg ister, Vol. 38 No. 188, pages 27076-27081, 28 Septem ber 1973. We a re in terested in helping the FDA choose a p ractical analytical procedure for the detection and identification of asbestos. Our laboratory has analyzed many hundreds of asbestos and asbestos-containing sam ples by m ost of the various analytical procedures: polarized light m icroscopy, x -ray diffraction, scanning electron m icroscopy with energy-dispersive x -ra y analysis, transm ission electron m icroscopy with selected a rea electron diffraction, electron m icroprobe analysis and electron m icroscope m icroprobe analysis. We would like to begin by giving you our evalu ation of each of these methods. The methods to consider are: 1. Morphology by light and electron m icroscopy 2. Optical properties by light m icroscopy 3. Diffraction by x -ray o r electron diffraction 4. Elem ental analysis by electron o r ion m icroprobe The various instrum ents available a re capable of a variety of combinations of these methods (See Table 1 attached to this le tte r). The advantages and disadvantages of each a re a a re discussed individually. The polarizing light m icroscope (LM) req u ires g re a t skill and background obtained only with long p ractice. A m icroscopist trained in optical crystallography and fam iliar with the serpentines and amphiboles still req u ires hours of tim e p er sam ple to find and identify each of the possible asbestos types. The problem is com plicated by the presence of interfering fibers (3:1 aspect ra tio !), e. g . , paper, talc on edge, diatom s, q u artz, m ineral wool, plant h airs and fibers etc. Finally, all of the asbestos types show considerable variation in all of th eir properties and these variations m ust be fam iliar to the m icroscopist. Page Two FDA D ispersion staining (DS) g reatly speeds up the detection and identification of asb esto s fib e rs. It is the only practical way to apply optical crystallography to sam ples containing sm all p e r centages of asbestos. The technique can be learned much m ore rapidly than optical cry stal lography yet it is securely based on optical crystallographic properties. X -ray diffraction (XRD), also based on c ry stal p ro p e rtie s, is som ewhat le ss sensitive than dispersion staining but requires no m ore background and skill. The sensitivity can be in c re a s e d fro m 0. 5 -1 . 0% to 0 .1 - 0 .5 % by ste p -sc a n n in g . One advantage o v e r LM and DS is s m a lle r size lim it -- about 50 nm com pared with about 5, 000 nm (5 /am). Scanning electron m icroscopy with an energy dispersive x -ra y analyzer is an overrated tool. Many substances besides asbestos a re fibrous and it is nearly im possible to obtain a good ch em ical a n aly sis of s u b m ic ro m e te r p a rtic le s in a p a rtic u la te m a trix w ith EDXRA in th e SEM; th ere is too much interference. Fitting the SEM with a wavelength dispersive x -ra y analyzer greatly im proves it usefulness. The tran sm issio n electron m icroscope (TEM) with selected a re a electron diffraction is an ideal to o l when th e a sb e sto s fib e r s iz e is < 0 .2 /m. The shape of fib e rs down to 0 .0 0 5 fxm is im m ediately apparent on scanning a sam ple and the diffraction pattern is im m ediately discernible. The electron m icroprobe analyzer (EMA) with its wavelength d isp ersiv e x -ra y analyzer is b e tte r than the SEM fitted with EDXRA but is su rp assed by an SEM fitted with wavelength d ispersive x -ra y analyzer. The b e tte r m orphological im age of the SEM tips the balance. EMMA, the combination tran sm issio n electron m icroscope-electron m icroprobe analyzer is, by far, the m ost useful of the analytical tools for asbestos. It may not be needed if the fib e r s iz e is la rg e enough (about 0 .5 am in s m a lle s t d im en sio n ), but s m a lle r p a rtic le s can be characterized by morphology, diffraction and chem ical analysis in this one instrum ent. It is ideal for w ater sam ples containing suspended asbestos fibers. The ion m icroprobe, although a highly sensitive tra c e analytical tool, is not usually used for asbestos fiber analysis. It does, however, detect fine fibers and identifies them by qualita tive chem ical analysis (quantitative when com pared with standards). Sum m arizing then -- if the size of the asbestos fibers is very sm all (as it nearly always is for w ater sam ples) LM is u sele ss and a diffraction method m ust be used. If the size is larg e enough, say 1 x 3 /m, and the p ercen tag e high enough, say < 0 . 01%, then LM can do the job. If high sensitivity is not needed, say about 0.2% o r g re a te r, then XRD is the b est method. Very fine p articles require the TEM with selected area diffraction. It should not be n ecessary to em phasize that no method is b e tte r than the analyst who makes the analysis. Few laboratories have all of the above instrum ents and highly skilled analysts fo r each. Instead, each tends to use the tools available and the methods with which th eir personnel a re m ore o r le ss fam ilia r. In a situation as im portant as th is, the choice should not be based on availability of tools and skills. Fortunately, the tools for detection and identification of asbestos are neither the m ost expensive nor the m ost sophisticated. Page Three FDA _ O ur p ro ced u re (in b rie f), based on the analysis of many hundreds of sam p les, is: 1. S creen the sam p le by DS using C arg ille high d is p e rsio n liquid 1 .5 5 0 . 2 . If positive: a) identify the a sb e sto s type by DS o r XRD, b) d e te rm in e am ount by counting using c rite ria suggested in the FDA method of 28 Septem ber 1973, o r by XRD. (We m ight add that a 3:1 length to width ratio is not exactly a fib er. The P article Atlas specifies 10:1; quartz flakes a re often > 3:1). 3. If negative: a) conclude that less than 0.01% of chrysotile o r amphiboles are p re se n t, b) u se TEM o r EMMA w ith EDXRA and SAED to be su re subm icroscopic asbestos particles are not present. This is a com plete procedure which can usually be shortened considerably in practice. F or example, a m ineral sam ple o r a sam ple containing a m ineral fraction will generally not have been subjected to a procedure that elim inates la rg e r fib ers, e. g. , u ltras one ration o r sed im en tatio n . Hence TEM is not n e c e s s a ry and DS can b e u sed . B efore describing in m ore detail p recisely how we proceed, let me comment on the proposed FDA procedure. 1. To do optical crystallography on talc the p a rtic le s m ust be well d ispersed. One mg under an 18 m m cover slip is fa r too crowded if one hopes to find individual fine asbestos fibers; one-tenth that concentration is much better. 2. It is tru e that all refractiv e indices of chrysotile a re le ss than 1. 5T4, but so are the indices of num erous other elongated substances including ubiquitous paper and glass fibers. In o th er w ords, 3:1 aspect ratio p a rtic le s having all indices < 1 .5 7 4 a re very probably not asbestos. To be certain they are chrysotile requires: a) rem ounting another aliquot in, say, 1. 560 to see if n || is about th at value and if the other two n 's a re about 1 .5 5 0 . W orse s till, one has to find those now n early invisible (w ithout DS) fib e rs am ong m illio n s of o th e r p a rtic le s; o r b) scan the sam ple at room tem perature in a hot stage so that a suspected fiber in 1.574 could be shown to d isappear with sodium light on heating at about 25 + [(1 .5 7 4 -1 . 560)/ 0 .0 0 0 4 5 ] o r 56 C fo r n |j and about 2 5 + [ ( 1 . 5 7 4 -1 . 5 5 0 )/ 0 . 00 04 5] o r 78 C fo r n ^ . One would at le a st have the sam e p article in view throughout the te st but it would still take hours, o r even days, to check 100 p articles by this procedure. 3. T alc, often on edge, has one index le ss than 1. 574 and it takes tim e to check the other index to be su re they a re not chrysotile. 4. It is likewise tru e that all amphiboles have all indices higher than 1. 590 but the same difficulties hold tru e here as a re given above fo r chrysotile. The amphiboles a re even w orse, however, because th ere a re m ore of them , each one varies considerably in optical properties and there are many possible interfering substances. 5. Counting the num ber of fibers is not a good m easure of % asbestos. We find the size of individual fib e rs v a rie s 10-fo ld , at le a s t. Page Four FDA 6. It req u ires a highly skilled m icroscopist to identify the amphiboles; especially in sm all am ounts, in confusing m ixtures and as sm all fib ers. Of all the m ieroscopists in the w orld few er than 1% would be so qualified. N early all of these qualified m ieroscopists are m ineralogists and a few chem ical m ieroscopists. V ery, very few industrial lab o rato ries, especially in food and drugs, have such personnel nor are they available on the job m arket in the num ber that would be n ecessary if th is procedure is adopted as w ritten. Now, what would we sug g est a s m odifications th a t m ight im p ro v e the p ro p o sed m ethod ? We recognize th at the proposed method is basically good in that it is based on crystallographic data, the c la ssic a l approach to the problem . We would keep th is base but shift to diffraction ra th e r than cry stal optics. The diffraction method is e a sie r to teach, e asie r to learn , just as dependable, works for all fibers (large o r sm all) and is m ore rapid. It is b etter than crystal optics in assessing the type of amphibole especially in complex m ineral series like trem olite -- actinolite -- ferroactinolite o r the cummingtonite -- grunerite series. It is, how ever, somewhat less sensitive than light m icroscopy. With step scanning, a tim e consum ing but poten tially autom atic p ro c e d u re , 0 . 2 - 0 . 5% of a given a sb e sto s m ight be detected; by o rd in ary slow continuous scanning 0. 5 -1 . 0%. C rite ria for the certain identification of each asbestos type can be found in the lite ra tu re . We use the following c rite ria for chrysotile and trem olite. C h ry so tile -- An x -ra y pow der p a tte rn showing both the 002 ( 7 . 3 0 .1 A; J /lQ = 1. 00) and 004 (3 .6 5 0. 05 A; l / l 0 = 90 2 0 ) re fle c tio n s and showing no lin e in th e 1 4 .6 A chlorite region is necessary evidence for the presence of chrysotile. A line at 1 4 .6 A in addition to, and of the p ro p er intensity relativ e to, the 7 .3 A and 3 .6 5 A lines identifies chlorite rath er than chrysotile. Trem olite -- N ecessary evidence for the presence of trem olite is an x -ray powder d iffractio n p a tte rn w ith lin es at 8 . 4 0 .1 A, 3 . 2 7 0. 05 A, 3. 0 3 0. 05 A and 2 .9 2 0. 05 A. (Note: zinc and m agnesium s te a ra te m ay in te rfe re a t 8 .4 A and calcite at 3. 03 A .) S im ilar c rite ria have been se t fo r the other am pluboles. If g re a te r sensitivity is required we recom m end step-scanning. This m ust be done through the regions of the three strongest lines fo r each of the asbestos types. If still g re a te r sensitivity is d esired o r if, fo r any reaso n , light m icroscopy is the p re fe rre d m ethod, we then su g g est d isp e rsio n stain in g . The o v e rrid in g re a so n fo r DS r a th e r than the c lassica l m ethods of optical crystallography is the following. One cannot by the la tte r methods mount a sam ple in the diagnostic refractiv e index mounting liquid for, say, chrysotile because chrysotile thereby becom es nearly invisible and, especially in trace am ounts, sm all needles would not be detected. F o r tins reason your proposed method req u ires mounting in 1.574 so that chrysotile fibers can be seen and th eir index relative to 1.574 determ ined. However, many other substances also have indices low er than 1.5 74 and, although birefringence, extinction, sign of elongation and contrast may elim inate som e, this takes tim e to check. DS, on the o th e r hand, allow s the an aly st to mount the sam ple in 1 .5 5 0 , the liquid m ost likely to malm chrysotile disappear when c la ssic a l methods a re used. Now, how ever, with DS c h ry so tile fib e rs show two b rillia n t and d istin c tiv e c o lo rs , one fo r n,, and one fo r n^ . Page Five FDA This uniquely identifies chrysotile fibers alm ost at a glance; glass fib ers, paper fib ers, talc on edge, quartz, splinters etc all show, at least, different colors and an experienced m icroscopist can quickly identify them . Sim ilarly fo r the am phiboles, we mount the sam ple in 1 .6 0 5 ra th e r than 1 .5 9 0 and for a corresponding reaso n . In 1 .6 0 5 , tre m o lite is easily identified by e ith e r yellow and blue colors on cry stals with oblique extinction o r yellow and golden m agenta on c ry stals with p a ra lle l extinction (n || and n. , resp ectiv ely ). F u rth e rm o re , actinolite and anthophyllite have distinctive and different colors in this sam e liquid. One is left only w ith am osite (positive elongation) and crocidolite (blue fib ers with negative elongation). All five of th ese amphiboles can thus be detected and identified in this one liquid. In conclusion, if forced to analyze a sam ple by the proposed method, we would req u ire sev eral (2-8 ) hours of a skilled cry stallo g rap h er's time; this would average $250 p er sam ple. The tim e can be cut to 0. 5-1 hour and the cost to about $50 when dispersion staining is used. It is also p ractical to use less highly trained m icroscopists for the dispersion staining procedure. Although w e're not su re, reading the proposed method, w hether we are excluded from using dispersion staining, it is obvious the w rite r of the procedure did not visualize doing so. We have ju st com pleted, for exam ple, an intensive 2-day course in these methods for EPA personnel and have scheduled a sim ila r course fo r food and drug industry personnel 25-26 F ebruary 1974. It is feasible to tra in a m icroscopist with little previous knowledge of crystallography how to detect and identify asbestos by dispersion staining in two days. Not so by the c lassical method; h ere, we would need 10 days and even then less than half of the students could dependably analyze any and all sam ples for serpentine and amphiboles. We have also enclosed copies of two pap ers from o u r labo rato ry on asb esto s. One is, u n fo rtunately, s till in p r e s s and w ithout the cited F ig u re s . We hope, in any c a s e , that som e of this m a te ria l based on our own experience m ay be helpful to you. If we can help in any fu rth er way, please be su re of our d esire to cooperate. Sincerely WCM:bb offese. W alter C. McCrone Scientific Adviser