Document 6R2Ym3KBM3Voan4YrzKD42G3R

DownloadSearchable PDFRandom document
By R. j. Lee Basic Research-Physics D. K. Scott Mathematics D.te May 15, 1978 Rf. No. 18-D-543(064) Subject CRITIQUE OF ELECTRON MICROSCOPE MEASUREMENT OF AIRBORNE ASBESTOS CONCENTRATIONS--A PROVISIONAL METHODOLOGY MANUAL by A.V. Samudra, C.F. Harwood and J.D. Stockham, IIT Research Institute, Chicago, Illinois. EPA 60C/2-77-178, August 1977. Project Officer, J. Wagman, Emissions Measurement and Charac terization Division, Environmental Sciences Research Laboratory, Research Triangle Park, North Carolina A review of the subject Provisional Methodology Manual for ambient air sampling indicates that the methods described are valid for monitoring commercial asbestos emissions from highly localized sources. However, we. have serious reservations about the validity of the sample collection methods', particulate identification methods, data analysis, and the estimates of minimum detectable limits when monitoring ambient air quality or emissions from extended sources. Specific criticisms of a number of the proposed procedures are de tailed and recommended changes presented. Aoa2ggs:^flSiS!es55Sggfisaseas CRITIQUE OF ''PROVISIONAL METHODOLOGY MANUAL ON ELECTRON MICROSCOPE MEASUREMENT OF AIRBORNE ASBESTOS CONCENTRATIONS" by A. V. Samudra, C. F. Harwood--and J. D. Stockham, IIT Research .Institute, Chicago, Illinois- ETA 600/2-77-178,' August 1977. : Project .. Officer, J. Wagman, Emissions Measurement .and Characterization Division, Environmental Sciences Research Laboratory, Research Triangle-Park, North Carolina________________________________ ___ R. J. Lee and D. M.. Scott . .. U. S- Steel Corporation, Research .Laboratory Monroeville, Pa. 15146 Summary A review of the subject Provisional.Methodology Manual for ambient air sampling indicates that the methods described are valid for monitoring commercial asbestos emis sions from highly localized sources. However, wc have serious reservations about the validity of the sample-collection methods, particulate-identification methods, data analysis, and estimates of minimum detectable limits when monitoring ambient-air quality or emissions from extended sources. Specific criticisms of a number of the proposed procedures are led and recommended changes presented. Atfr LL-IC7 HON MIC50-LMHNT O" AI RUORN'K :;cl`N7P.a" : ons ->_ O i scjss ion lor several years much attention has been focused on : '.::.: r.'.ulth hazards resulting from exposure to particulates : r- i :i low concentrations m ambient air, food, or water. tv thr absence of direct medical data, concern for the public altn has ben aroused by certain mineralogies! analyses and r.rupolation of established effects due to occupational o.'.ure to commercial asbestos. In response to wide variations'; t;.e results c: sample analyses reported in various labora- t>:i>s the bPA awarded a contract to the IIT Research Institute Ci.iCaoo to develop methods of identifying and counting ) r.-.. r. -a:; discussed in the Foreword to the report (reproduced Vr.fortur.ately, detailed examination of the Provisional .`'.t;.-'. lolcgy Manual has revealed that many unansworc . guest ions r. and that we have a number of specific objections to the pro:;-. ;r-c methods, detailed below. The format used is to reproduce the uui ` s recommendation, comm.ent on the recommendation, .j.-.d, where appropriate, make an alternative recommendation. best OS c\ asbust-t js-tr. ituif iat s include icvctat s - - upi ^ibneus c-lus Zztli < l' iub&tanci'i with Syccmi md cicct-~<CJ. pncpcntici fiiat have tuu, cncou\apcii r., . ^; <. n the menu&&cu-tc e * i a c /: 1* on , btnfee jUie P* : c * p cc'ucfi ai nee i inn f cu rtiu ns , etc. 7 ii c < 1 6 yc < L a fa.-: ': in tin- C "b i c i: * ad : and m; i a !.;/ ( f * . : ' J - c il.v-OItj 1 *i$ Cu 1 L d reni < d r "iC.i i. c r e i: con r. IM- d ;i ' * * .. : 'tcnzL cipciutih to a i b c i f . , i liavi1 L) 0 ? 1! t > i >; u U C I- C :.. i r/ ' k . pizu a a n d r- * *. / t c r. C urn, i. a:d iCr ^ j: c 1; , a f r : ui Kh'it r.! a n c 5` k i a f r i: t : c < o d t, * / . , ' ` ' it :j CS 5 . 1 : : ". : - '-Lit r t t.; t . ' < . t. \ ! *. ' r'. .* - K t c> * l' -c c d ( l * if a: ti C '. a *. a c t c u: : csbvs ti - i < ( ; 1 ` M* li.J ni.'. t i crr.f.cs. r. ccause e *' the pl'C. *i i* r. i / / < V < : -j \" ' v e ; c ii f . i: -f.4a < clh, u /. .#C au K tj cci ni; t (ll'Ci , ryC t i :i o c i/ i i a 4 i e b ; > no used &c `cutine mens a t n: i : t .`.a I . t o -. i:, aubt-i fr s c e i: c C >: t nations. T ki 4. t' v % * .i c * <-.c. : c . ' * l*' such z; a uscs .) ' >; tu l il a h ll v l- * J :!. :/ . . .. ' :i. I1 ', > ; f1 i :i />- . af,u ll i* t. . r * i" `."I -; , *' t : ' I.LCTPC,.'. XICPO- potions I. `-fif-rr-ti-j l. u < d\ i k " > I'k : a i`.`wiU'i:u( ap-f < mtur. r f ' ? t < i; ; i i'" ''v - - .'j.'i- atbeit.-i , 1 v . i w u ^ i J * i ** * i ii (/ ( C J '1 * i i 4/ (,'U < UU u v * '' ** i - s" J - < < 5 / , fa c i-u i uf c f >': i> I'sucui nc (iH'riu <: j 4 ; <! f /; c Laoc tc ft- t< ,'i . S a < 4 r < c.'i ' 1**' ** 't-. .4 .* 4 ' i i.' i'il 4 J *' 4 \ ii V i` ,*1 |^ (' i *.4 C* ' t li k na II1/ ^ )* ^ i' 'l u f i / II ii ....i.\,.' an.. u t a <-.-v a i an i. C'.-tin c . npr i. t e picc ec . ~ i. u' r-.an.ial rfn'i nr* p'.cvid c fiit i'J4 f amem:' o (i c'a^: i; v- p * .'V t 4 ('l) liS 4' r.v.ilu.'dd i V lUJ . 1 > i 0 4 (id < d v 1 4' . v ii \ 111" *, *c *tii4 \epv *. t c r. CPA Contract i v, 6 S - v C " c C .*? i . .::."it -it ii-.f foreword clearly indicates the methods described i r(jv;:..onai Mo thocioiouy Manual were developed for the c;f airborne asbestos. However, the deflnit -.'s te icor: 11:,,ca 11 or methods mtsor j !,.ed jr. the Muni:-:!! \ oroade:. the definition of asbestos from that given :c:wrru to include non-asbestos amph.mole minerals and ; r.uii.be: of other minerals. * 1 . I. i . . I 1 *1 " I l Li \ t 4' * l S i t -C %.* Ii '..:v'd C it v ono*l: a ad b--\.adti. r > a C t ' 1 i bet 4 that . dipe:' t c oj j c 11 c. ft.c.: ;: 1 and have bubi t.an t ia [ Ln . Obscwc the ir.v'.phve cgtj e * c ct c *: ^iLu'-t ff;riir; >'i Lana Cuba' and note ichcthe' a XuLuba t a tauctu-ie :/ia tcc ft- * -!/4 i 4^ 4il C4 fa 14 4 >'4 04 p 11' 4 C. K 4 , Sli`4 4 C ii 4. i\ t C i A t` P . Cie've tiif db-, ; .iib.oi: patte-'.n. .' u;he. th e-i tan '/' a i 4- ci.'.iiO'fj.i,' c: anpi,' t->/., r t iehvthi". 4'-'. r.C'.riu", eh t v-ic f 4 i cc-t cr. hi .>. \ . - : '.no c r. i y svctio:. i:. whicii the ( err. ":..h*..-r" t'.-i i:i:t:or. :s :r.r,i rer. tl y taker, to be u uimdi :t 'ill and neiriy r.ara et::>s. The- .ichor:.' - r '.':nos.:.:: : : aspect ratio as the lower i it Cor title- tin asbestos fiber is not clt'.ir; it ccjid :,,4-.-:. ill, -: 1 , 10:1, /0:1, or ev'.M: -50:1. This las; : :: ::.o: v.- r epr.-s a t: of tr.-. : co: 1 aspect ratios . ' o ji: *. i air: tvr.r,trt;-.! bentos na ri c j * - * . . . *"* * ' *.., ' t i j i'. *.: I - - <: 1 o.. : i.' Vi * : .i!. , J l, \ v TTr: CRITJQUE OF ELECTRON MICRO SCOPE MEASUREMENT OF AIRBORNE ASBESTOS CONCENTRATIONS -4- of waterborne chrysotile, found the average aspect ratio'to be 1*00:1.In a recent round-robin study conducted by an' ASTM committee, the average aspect ratio of chrysotile fibrils was on the order of 30:1.3) Similar data have been developed in' our laboratory for airborne amphibole particles. In one study, the aspect ratios of airborne amphibole particlcn fell into two well separated categories.4^ One group of amphibole particles generally had very high aspect ratios on the order of 40 or more to 1. These particles are often observed to have defect properties such as those described for amphibole asbestos by Hutchinson, and others.5The second group of particles having aspect ratios on the order of 5 to 7:1 were generally well crystallized and bear little resemblance to the type of faulted amphibole asbestos particle observed in commer cial asbestos samples. Finally, the authors appear' to be extending the definition of asbestos from that given in the Foreword to include not only'chrysotile and amphibole asbestos, but all amphibole particles in all size ranges. Given the significant number'of studies which show no (or at least reduced) hazard from small' particles, 8) this ox tension does not appear warranted. Hence, the value in adopting a method for the sole purpose of resolving submicrometor particles is debatable. USE Recommendation If ambient asbestos levels are to be analyzed, the definition of fiber should be such that the particles counted are similar in aspect ratios, composition, and structure to ccr-v.orcial asbestos. Since it is pointed out in the Foreword -hat only occupational exposures to commercial asbestos minerals re known to induce biological response, it would seem prudent or any monitoring program to separate those amphibole particles v.-irh physical properties similar to commercial asbestos from those amphibole particles which would not be considered commer cial asbestos.^ i p-'-. iUtcenuatndaZicn- Sample Collection ()) lake an aiK sample on a polycanbonatz membrane. -r.L-tec, C.4 urn, ui-Lng a nig h-volume on. peK&onal 6amptc.K. L'."S Comment The use of a polycarbonate membrane filter to collect : is; regarded as undesirable. Al the recent MBS ' ' --- 'orkshop held July IB through 20, 1977, a n umber r- o fa CRITIQUE OF ELECTROS MICRO SCOPE MEASUREMENT OF AIRBORNE ASBESTOS CONCENTRATIONS -5- objections to the use of polycarbonate filters, for .field. tl. ,, sampling were raised. These objections were reinforced at a ' meeting held by the EPA at Research Triangle Park on November 1, 2, and 3, 1977, which was attended by Dr. Samudra, the author . of the EPA manual. One of the problems with the polycarbonate filter for use in field sampling is collection efficiency; for example, Merchio, et al found that 39 percent of: the chrysotile fiters lest than or equal to 0.1 micron in diameter by 4. microns ip length passed through an 0.8 pm polycarbonate filter, , and 15 percent of the chrysotile fibers less than .0.1 micron in diameter by 15 microns in length were passed. Their data-did not show the extent of the variability of the. collection efficiency for different filters of nominally.the.same diameter. Other factors which would seem to preclude, the use of poly carbonate filters for field sampling are. bhe difficulties in handling during transport and the potential loss of material from the filters during transport and sample-preparation. UBS Recommendation - ; ', , ; t. Until problems with particle retention and filter handling are solved, field collection of .ambient air-samples rueold be carried out using Millipore filters, with the material transferred to polycarbonate filters for laboratory analysis. C PA C cmtr.ent . . ; .................. - Neithen the monphotogy, non the ele.ctn.on .di^inaction pattcnr., non both, can give inneutabZe pnoo that a given v-lct ib abbebtoi. Pobitive nebuttb &nom the tcbtb indicate .-u y that a&bebtii.cnm (ibem aae pm&ent. UBS Comment ... This again implies "asbestiform fibers" based on their aspect ratio and the general appearance of the selected circa diffraction pattern. These criteria encompass a much .larger group of minerals than those described as asbestos in the Foreword. As stated by the authors, the methods described ir. the Manual will not permit positive identification of a given particle. As a result, only in the case of highlylocalized sources emitting asbestos can the methods described be considered an acceptable measure of the concentration of fibers. They are not acceptable for ambient air samples where many unknown minerals may be present. CRITIQUE OF ELECTRON MICRO SCOPE MEASUREMENT OF AIRBORNE ASBESTOS CONCENTRATIONS -6- -3zo:r. ;-o-TOEJi ;. / T* CPA ReComnencfa-t-ton-F-ibe* CtcAi-itUcc-tion a {ib CfLt (c) ti bU i The {allowing nules should be {ollowed wkeh-.'classi{tying v:;-.'' . J j'.v'm *. 1>6ac*.vc a {iben at a TEM 6CAfc! magnet{icaiion o{ about 20.000K through a binoculaK- with a magni{icaticit o{ about 10K. At a AcVecn- mggr.i {ication' o { 20, 000K, the tubulaK' stnuctuKc- o{ chaysotilc `asbestos is usually apparent (compare with- st'andand sp'ccinc r.s) . FibcKS showing the tubulaa sltKuctuAe':'ma]f 'be 'classi{icd as chaysotilc asbestos with confidence.- -' TheAe 'o.\c only Ka>ic exceptions;amphibole asbestos usually have a lath shape; but sometimes appear similar in {oam to chaysotilc ^ebcA.6 without lumina. - .. .. - v.-i. .. . Election di{{aaction `patteKns--{aom :pa'aticlcs-with {ibaous moaphology {all into distinct groups. Chaysotilc asbestos has a chaaacteaistic 'sta'caked ' layea line thaough the central Spot and also a triple set o{ double spots' bn-tke :'6etv'nd' la'tjea line. Amphibole asbestos gives 'a-la'tyeK-patieari',-geneta11y with little ok no stKeaking'. y ',:'; i c, T/tan4m^.44<UH electKon micaogaaphs and selected anea electaon di{{aaction patterns obtained with standaad samples should be used as guides to {ibeK identi{ica- tion 1^,5]. ' \ 1. USS Comment In (a), the authors indicate that a tubular morphology is characteristic of chrysotile and no other minerals. .However, as pointed out in Grimms' book on Clay Mineralogy, halosite is tubular and the possibility exists that some kaolinite may exist in a tubular structure. Evidence for elongate forms of illite, atapolgjte, alophane, and montmorillonite with diameters 'vlO^-lO^A was also discussed by Grimm. As Beaman has pointed out, further complications arise when the chrysotile has been subject to thermal or chemical treatments.He finds the tubular structure is often not apparent when the chrysotile has undergone extensive physical' processing; Beaman also reported that only about 10 percent of- the singlefibni-chrysotile particles give good diffraction patterns, whale about 40 percent can be recognized as'crystalline. This presents a serious problem in identification as the very small particles which do not give well defined electron-diffraction patterns may be chrysotile or they may be some other mineral par f iciz-. Bfii t < .'a! CRITIQUE OF ELECTRON MICRO SCOPE MEASUREMENT OF AIRBORNE ASBESTOS CONCENTRATIONS . -7 c , In (b) , the authors indicate that electron-... diffraction patterns from particles with fibrous-morphology fall into distinct groups. If their definition of fibrous is simply that of 3:1 particles, this statement.is incorrect. As shown in Figures 1 through 3, taken from a paper by, Lee,. bally and Fi3her, which was presented at the NBS Asbestos.Workshop,^3) a veiricty of diffraction patterns can be obtained for individual particles, and thus the existence or absence of a."layer, pattern" is not sufficient to identify particles as amphibole or nonarr.phibole. Here we have taken the authors' meaning of "layer pattern" as any diffraction pattern having,a .row of spots, with a projected separation of s5A.. .This is no* the conventional use of the term "layer" pattern,- a term normally ,k1 '.i reserved for the diffraction pattern from textured materials, having an axis of azimuthal symmetry. . . . ...... .. -<( U3S Recommendation - The identification of particles in ambient .air samples is most efficiently accomplished by grouping the '1 particles into classes on the basis of their X-ray -spectra using either transmission electron microscopy (TEM) ". or' scanning electron microscopy (SEM) equipped with an energy-dispersive spectrometer. The homogeneity of each class can be established by using quantitative selected area diffraction., (SAD). techniques on randomly selected particles from each- class- ,This -willalso permit the analyst to establish confidence levels on the certainty of identification for particles within each class. LPa Zecommendation-Fiber Clarification From the examination of the electron-dlffraction pattern*6, *llbcr& are clarified cu belonging to one othe j {.cllcwlng categories -* chrysotlle, amphibole, ambiguous, non- H asbestos, unknown (nc pattern) . 'J It should be noted that other particles with fibrous n.crphc Logy alsc give layer pattern; ion example, hornblende. T!:l complete quantitative Indexing and deriving Interplanar i d-spac-engs from diffraction patterns Is a time consuming and complex undertaking and li net feasible for routine, analytic. USS Comment ............. i-iti Identification by inspection is not reliable--for -component samples. While the authors apparently do not intend to include hornblende in the amphibole-asbesfcos category, i- is an amphibole mineral. Separation of this amphibole from bj&afe] CRITIQUE OF ELECTRON MICRO SCOPE MEASUREMENT OF AIRBORNE ASBESTOS CONCENTRATIONS 8- - TM Da other non-asbestos amphibole particles is very difficult with seme types of elemental analysis. It should be noted that USS Research was hot' consulted with respect to the quantitative interpretation'ofelectron diffraction patterns or as to the suitability of our methods for routine analysis, even though these techniques have been presented at several meetings attended by IITRI personnel. Finally, an abundance of data suggests that significant physical differences exist which differentiate the asbestos' and non-asbestos amphibole varieties. ^* ^-2 ^3) 'a potential means for distinguishing these-minerals was'offered at the-NBS Workshop. This was based on differing'crystal'habits reflected in the orientation of particles in- the TEM.-'--'This difference and its effect on SAD patterns are illustrated in Figures 4 and S. Further study in this area is required to determine whether these differences have general applicability in the identifica tion of asbestos. USS Recommendation -' v- If electron optical'methods are chosen,'-our methods or their equivalent should be used. 'Amphibo'le 'particles with very large aspect ratios' -(>40:1) should be separated from"* those with moderate aspect ratios' (<10:1)".:: * L PA Comment- F-tbeA I de.ntj.i-ica.tj.on ........................' ' ' It -it not possible to inspect cleetKon- di^>iaeti.on pattern* t>on tone. fiibe-r* even when the.-in identity a* a*be*to* i'bca* i* known. Then-C axe ievexat xea&on* ox thz 'absence c& o. pattexn. Thzbz -include contamination ofi the fiibex', intex- ` {e-ience <xom neaxby paxticle*, too *mall a ibex, too thick a i-LLex, and non-*uitable.oxicntaticn o the fiibex. Some chxy*otile r.ibzx* axe dei.tx.oycd in the election beam, xe*ulting in pattern* that &adc away within second* of being tfoxmcd. Comment Here the authors describe numerous reasons for the absence of well defined diffraction patterns. While their statements here are in essence correct, they fail to give estimates of the percentage of particles of a given mineral type for which they are unable to obtain diffraction patterns. Ir. a similar vein, there is no estimate of the percentage of s'.: sidentif ied particles; this becomes very important when analysing ambient air samples or "any. sample .in which a variet-v mineral types can occur. amass o?cp.itic-ue electron micro scope MEASUREMENT OF AIRBORNE ASBESTOS CONCENTRATIONS _____ -9- ---.............. ^rnmiiT~TiTr The absence of any examples of electron-diffraction patterns from chrysotile, amphibole, and other: mineral '.particu lates is particularly troublesome. If, .indeed,- the * diffraction patterns from fibrous materials were diagnostic,any methodology making use of electron diffraction should .include:specific- diffraction patterns to be used in identifying each mineral species. '` i . i. *A Riicomniemialion- Vaxticie Counting i Counting at Low Leading Level -ken than 50 fibexA pex gxid opening ate cr.cuatitenid, the pxefexxed toanting method i'A to Acan the ' ent+xe gxid opening and define the full gxid opening a& one fxeld. With the mi.cxoAC.ope magnification. at 20,000X, a AexicA oft paxallel AcanA acxoAA the. gxid Aqaaxe axe made-Ataxting w-ith the top coxnex of the .Aquaxe .and ending .at the-bo t-tom U<:t F<Qu.te 2(c)). IWith the tilting section of the flucieA cent Arxcen uAcd ai a Aingle field c$ view; appxoximately -500-400 fieldA uiiil be obAcxved if, the entixc gxid opening it Acanned. ) FiLexA noted in each fall gxid opening -(ox Aingle field) - axe clcLAiified in accoxdanct with the pxoceduxe dcAcxibed above. Additional gxid opening A axe A elected,- .scanned, and r. ounied until the total numbex of fibexA counted exceedA 100, ox a minimum of 10 gxid openlngA have been -Acanned, whlchevex ocu\a flxAt. Counting at Medium Loading Level When the loading on ike filtex it in the xange of 50 O SCO fibexi pex gxid opening, counting iA done on xandomly selected fieldA of view. At a Acxeen magnification of 20,000X, j-'.i.ida axe xandomly a elected within a gxid opening until a total of 20 fibeXA have been counted, Aized, and claAA'ified. Genexatly 20-40 fieldA of view axe cbAexved pex gxid opening.) Aftcx about 20 fibexA have been counted, anothex gxid opening -ic-ttcted and an additional 20 fibexA (appxox.! axe counted. IhiA pxoceduxe iA xepeated fox 5 gxid openingA until a minimum of 100 fibexA axe counted. (When eAtimating fibexA of a paxticulax type of aAbeAtoA, counting iA continued until 501C0 fibexA of that type axe counted.) USE Comment The authors describe three levels of counting. The r u p-s are from'-O to 50 fibers, from' 50 to 300, and greater mm CRITIQUE OF ELECTRON MICRO SCOPE MEASUREMENT OF AIRBORNE ASBESTOS CONCENTRATIONS -10- * * / ........ than 300. In our experience, .acquired while 'analyzing ambientair samples for the purpose of developing satisfactory methods, we have found that the occurrence of 300 fibers per ;grid - opening in an ambient-air sample which was not taken-very'near a point source is unlikely. In fact, the average number-of - particles per grid opening generally ranges from 0;1-10.: These loadings are most suitable for analysis. When counting particles in the range from 0.1 particles per grid opening to 1 particle per grid opening, the use of 10 grid openings as a limiting value necessarily implies extremely poor accuracy in the resulting statistics.- In addition, in this range, the background levels of 3:1 particles on the filters takes on an important role. In a study by Cook, et al, :of '.Duluth tapwater, they reported from one to five-chrysotile'and from 1-5 ambiguous particles per grid opening on nominally .blank grids. Computation of a concentration, therefore;- based on the analysis of a very small number of particles, could suggest the existence of higher than ambient levels of chrysotile amphibolein the sample, when, in fact, the particles observed were on the original filter. Another problem occurs when analyzing larger numbers of particles. The "stop after counting.100 fibers" rule will often dictate that the final field not be completely inspected, and unless the portion of the field inspected is used in the estimate of the area, the computed fiber density will'be biased. Since-the variance of the estimate is inversely proportional to the number of the fields sampled, the effect of chis fixed-count sampling rule is to lower" the precision of the estimates of large concentrations. USS Recommendation Counting particles using electron optical methods should aim at establishing the concentration at a -predetermined minimum level of precision. The unit field to be counted should be adjusted so the expected number of fibers per field j.s between 1 and 10. a minimum should be set on the number of particles to be identified before an estimate of the concentration is made. Whatever the definition of field used, complete fields should be analyzed. If the data obtained does not satisfy the Poisson distribution, the analysis should be rejected.' As an alternative to the manual's stopping rule we v.ciid propose a technique called sequential testing. This * technique was developed for use in quality-control situations v.'!:oru the number of samples to be taken is not predetermined. l!.e approach i-to try to classify -the value of the parameter --r-tiir CRITIQUE OF ELECTRON MICRO SCOPE MEASUREMENT OF AIRBORNE ASBESTOS CONCENTRATIONS -II- 1/ .* -t' *% ; . of the population being sampled as belonging, to one-of; .two. groups of values. In effect, two hypotheses are-proposed .and sampling takes place until one hypothesis, is .accepted.. : For exar.pla, let <*. be the total number of fibers of .a particular type on a filter. The two hypotheses proposed are . V A < xo Hi: . * > Sequential sampling directs that one field at a .time .is sampled and a cumulative count of the fibers of, interest is. kept. -. After eath field is sampled, and counted -tables.;are; consulted ; and if the total count is above a critical .value,.;: By is-, accepted if it is below another critical value,. Bq is accepted; .-and if . .. the current total is between the -two table values,sampling continues. For filter data 1q and could-be -.chosen'o be symmetrically placed about the maximum allowable count, of specific fibers on a filter. In this -way.;the acceptance, of H0 cr would signify whether or not certain airrguality standards are met. . , .. .................. LPA Rc.commfcttdaff.on-PA.tcxA.ton of TEM Eitimatu> W/ien mote than one TEM gnid i&- u*ed, . it ii po&tiblc ic obtain -the mean value-6 and 95 pencent confidence level* on Che mean*. Thi* ii done fen. each important panametet. The rr-rthed comiiti of obtaining the mean, x, the itandand ennoi o* the mea.n, SLm, and t-value [6] .(0.025,-n - ?) fon n - ! dca-'.ec* of fneedom, whete n * numbei of TEM gnidi examined and hence r. nepl-icatei available. The 9 5 peneent confidence limiCi ane given by x z t (SEm). USE Comment The Manual's authors use the *'T" statistic to estimate the precision of their mean TEM concentration levels. An underlying assumption of the "T" statistic is that the.distribu tion of the sample mean concentration is normal, and. a.-powerful statistical theorem, the Central Limit Theorem, proves that regardless of the underlying distribution of the data, as long ct the data had a mean u and a variance o^, the distribution o: the sample means calculated from that data is assymptotically normally distributed with parameters u and o^/n, where n is the number of samples used to calculate each sample mean. Qj.\u_._ r nrwnm'iv. .... CEITILTE 0-r ELECTRON MICRCECLr-E MEASUREMENT OF AIRBORNE AL.-.ES70S CONCENTRATIONS -12- ,.::i"'.".p*_oticalIy normally it is meant that the distribution becomes more closely normal as n gets larger.) Of course, if the underlying distribution of the samples is normal the distribution of sample means is exactly normal. If it is not, for any fixed n, the distribution of sample means is only jj.pr'jxinately normal, and hence the precision derived from the use of the "T" statistic is affected. As shown by Merchio in 1972 the distribution of r:t data is often a Poisson with parameter A2. This assump tion can be confirmed by using contingency table tests which are given in many statistical methods books (Snedcor and Cochran's Statistical Methods, pp. 236-7). The test statistic f-ir` these tests is a calculated y2 value which can be compared to a table value. If the table value is not exceeded by the calculated one, the data can be assumed to have an underlying F-oisson distribution, with this information the precision of the mean concentration estimates can be increased. An example nr the application of the Poisson distribution is given in Tables 1 and II using the data presented in the manual (Tables D2-D5, p. 19). Table I shows the data for chrysotile fiber count. The null hypothesis was that data from both grids was distributed as a Poisson with parameter >.2 = 1.1625. The calculated x2 value of 4.662 was less than the critical value of 9.490 for the a = 0.05 level of significance; therefore the data could be assumed to be distributed as a Poisson. The estimate of concentration was then calculated along with the bounds for a 95 percent confidence interval. Note that this interval falls considerably inside the corresponding limits in the manual obtained using the "T" statistic. This procedure r-.seated in Table II for all fibers with comparable results. ' data are not distributed as a Poisson, then the appeal to the Central Limit Theorem and estimates of the mean and confidence intervals calculated using "T" statistics can be .tsCv . It should be noted that the half-fiber counting rule tr.e oasic unit of count in terms of a Poisson distribution a fiber, and these data were analyzed on that basis. Th.e nu: i-i io'-r rule presents a problem since, when counting whole of.emr.gs, presumably a long fiber could be counted as ar.al; fiber several times. In addition, each time it i3 counted '. must be movc-d to the center of the field to perform the : .-J'-c'.ud area-diffraction analysis, which results in noncert'inty repos.i tionir.y of the sample. nr ':':t:cel or electron micro.CO:-E MLAS L' REMEN T OF AIRBORNE ASBESTOS CONCENTRATIONS -13- rc.::jr.<_-nc. a ion We would suggest that the analysis should include investigation of the applicability of using the Poisson distribu te:.. Estimates made from this distribution will yield improved '-1: util . ty and evaluation of the concentration over those o.-. u "T" statistic. ' nreridii-t.ion-Limits o{, Detection The minimum detection limit the electron microscope /j, i the enumeration o$ airborne asbestos fibers u .:. to and depend* upon the amount ot total extraneous ue.i tc natter in the sample and the contamination level the laboratory environment. This txm.it also depends on the i i-. pa.-.amc-fc'ii, Loading level, and the electron pe parameter* used. In the provisional method proposed, 100 fields, each ::itn an area O.H x 10~'s cm` arc scanned. Assuming -j.ber count has an accuracy o=. / ^gbc-t then the detection Velection Limit * ICO Area cj Titter lcm 0.IS x 10 16 {cmL i t Air !c." ) The authors suggest that the minimum detectable i: .one fiber per sample. However, the use of this ...... i toi-s not account for the probability of contamination, ; t allow any estimate of the certainty of the limit, reasoi.ar.ie rule proposed by Rickards^) is that the - -- o: detection be determined on the basis of a giver, a. c-. r f: bars per unit field. However, we know of no .j '.stir.ate fer the limit of detection. R - c z: r, :! ,i 11 o r. L c : r : iuctu r--d rc by `-h<- u ihorr the min; mu:: j.y ti:- ur.it change it concentra t.: o;. i * usee ns an upper hour. J M ::E O'-t'-c: '-r' ; r. t.- a ' e-' j . rr-b'-r o: fiber:: are de :'_-iLjpiLj; >x l- 'MVr*` * *. ./"*, > i>iXV *-."* - v* : '- oa&* %3Sjgk /*-**r >*/. $' - V . Y-tfs*. v <*.?; VA-..'ASV&'.-'- ..* ' / * *o> . . ^^:.rSfcv!V/ 5KSrepE*> s-'v.v.'to-~>S u :5:s S&irx - *'.. ' . j'crt.. ; w,.- ysj;; ' * -*'-r}6'. ,, *. "'" `.i^/v".; *. *-----~--J *~- -- i' * **vr*>.>?>-->T%.V/ .: .'4^y 'T'/ n-^f ,-.`..'0.- ^3^;' . * . ,*;> - c?.:r:0L'L of electron kicpo:-C',rL MEASL'REMrJNT .OF AIRBOP-NE A EELS 70S CONCENTRATIONS -1-5- only a comparison of the number with the results of analysis of blank saxplfs will q:ve the analyst an indication of whether .I. - concentration measured on the sample is significantly uO.'-ve background. "one 1 usior.s The methods and definitions proposed in the ERA minnal are not suitable for measuring the concentration of :s in erne asbestos without extensive modif ication. As presented, the techniques cio not offer a clear advantage over the current, optical methods except in the superior resolution of the l i.-.ctron microscope. Since adoption of the electron microscope ~i:-. a routine monitoring device will increase the time and expense associated with such analyses, the quality of the data i :*i r a ted should reflect this expenditure and make full use of hv information available about crystal structure ..:\d comucsition. Detailed comments concerning many aspects of the provisional methodology manual have been made. In areas where it seemed appropriate specific suggestions for improved methods are described. In summary, we would suggest the following portions of the Manual or Method should be revised. A. Sample collection procedure. E. Counting rules. C. Particle identification. D. Estimate of precision. E. r.stinste of the minimum detectable limit. v.'e would recommend that particle counting using vic-ccro.'i optical methods incorporate the following requirements: 1/ The os tab! is recent of concentrations at a predef--r" -: - precision and level of confidence; or the use of sequent u:i .'ting procedures w.nich permit upper or lower bounds to be t: I'-eo on tne concentretion. 2i Cr.e estimate of uncertainty in the identificatie;: : ; . ; '..cits . 1/ A .mini .mum number of particles to be identi fj r-1 QVi.r. sp>.-cie. oelore u.r. estimate of the concentre > t->j; '\ or electron micro- 5'.Cw-;\t-. >;eaeurexent or airborne a.. :: .0: or conc entpati ons -15- 4j A minimum size ar,ci aspect ratio for particles - :-------t-_.it. wi tii pbysjcai data on asbestos, and reliable identifiivt.'j!. ui.ir.y electron optical methods. 5) Selection of field size based on the requirement n: -1C? particles per field and analysis of data on the basis R-. i Statistics. 6j Rejection of samples not satisfying Poisson -. : but ion. RJL/DMS/d p: d R. M C. F %j H, L. G J. S, H. W D. E. P. A. Bleifuss Bullock Fisher Glick Gross/W. T. Hendrickson Lally Paxton Splitstone Stoll Lankford, Jr. : arr-...\*r CLITIQUE OF ELECTROS MICROSCOP?: MEASUREMENT OF AIRBORNE ALBESTOS CONCENTRATIONS -16- References 1. J. C. Merchio, W. c. Cooper, and A. DeLeon, "Asbestos Fibers m Ambient Air of California," School of Public Health, University of California, Berkeley: For Air Resources Board, Resources Agency, State of California, Sacramento, PB-226-202, 1973. 2. D. R. Beaman and F. M. File, "Quantitative Determination of Asbestos Fiber Concentrations," Anal. Chem., 4j8, 101, 1976. 3. "Inter-Laboratory Measurements of Amphibole and Chrysotile Fiber Concentration in Water," ASTM E-4 Subcommittee' XI Task'' Group Report Submitted for Publication in the ASTM Journal of,Testing and Evaluation, July, 1978, in press. R. J. Lee, "Basic Concepts of Electron Diffraction and Asbestos Identification Using Selected Area Diffraction. Fart I: Current Methods of Asbestos Identification Using SAD. Part II: Single Crystal and SAD," SEM '78, ed. O. Johari. To be published. 5. J. L. Hutchinson, M. C. Irusteta, and E. J. W. Whittaker, "High Resolution Electron Microscopy and Diffraction Studies of Fibrous Amphiboles," Acta Cryst. A31, p. 794, 1975. . C. J . E. Chisholm, "Planar Defects in Fibrous Amphiboles," Jour. Mat. Sci., 8, p. 475, 1973. 7. D. R. Veblen, P. R. Buseck, and C. W. Burnham, "Asbestiform Chain Silicates: New Minerals and Structural Groups," Science, Vol. 198, No. 4315, p. 359, 1977. J. M. G. Davis, Current Concepts in Asbestos Fiber Pathogenisity ?roc. Conf. on "Occupational Exposure to Fibrous and Particulate Dust and Their Extension into the Environment," Sponsored by SEOA, December 4-7, 1977. 5. Vv. J. Campbell, R. L. Blake, L. L. Brown, E. E. Cather, and J. J. Sjoberg, "Selected Silicate Minerals and Their Asbestiforir. Varieties; Mineralogical Definition and Identification- Characterization," U. S. Bureau of Mines, Int. Cir. 8759, 1977. 0. A. Samudra, C. F. Harwood, O. Johari, and J. Wagman, "Vari ability in Quantitative Measurements of Microfibers Using i....:ctrc,n Microscopy, " Proceedings of the First FDA Office of Science Summer Symposium on Electron Microscopy of Microfibers h'.-io at the Pennsylvania State University, 1976. m CRITIQUE OF ELECTRO.'.' MICRO SCOPE MEASUREMENT OF AIRBORNE ASBESTOS CONCENTRATIONS 17- 11. R. E. Grimm, "Clay Mineralogy," Second Edition, McGraw-Hill, Hew York, 1968. 12. D. Beaman and H. J. Walker, "Difficulties Encountered in the Identification of Asbestos Fibers by Analytical Transmis sion Electron Microscopy," Proceedings of the First FDA Office of Science Summer Syposium on Electron Microscopy of Microfibers held at the Pennsylvania State University, 1967. '3. P.. J. Lee, J. S. Lally, and R. M. Fisher, '"Identification and Counting of Mineral Fragments," Proceedings, Workshop on Asbestos; Definitions and Measurement Methods, National Bureau of Standards, Gaithersburg, MD, July 18 to 20, 1977. 14. P. M. Cook, I. B. Rubin, C. J. Moggiore, and W. J. Nicholson, "X-Ray Diffraction and Electron Beam Analysis of Asbestiform Minerals in Lake Superior Waters," Proc. Int. Conf. on Environmental Sensing and Assessment, 34^ 1, 1976. 15. A. L. Rickards, "Estimation of Trace Amounts of Chrysotile Asbestos by X-Ray Diffraction," Anal. Chem., 44, 1872; 1972. .. -7-- ,, C::Rz;-:o`: :lk nurp-s S -i ~.p I o 1 Field : -> 3 *4 5 6 7 5 10 11 12 13 14 15 16 17 18 19 20 Tots 1 Fiber Count Grid 2 1 4 3 1 .5 2 1 0 .5 1.5 2 1 1 1 1 0 1.5 2.5 1 i 26.5 Fiber Count Grid 2 1 2 1 1 1 0 1 .5 1 1.5 .5 1 1.5 .5 1 .5 1 2.5 20.0 1 : r .r.c: err Av*-7 ;* 1.221 A UU H 2.1625 2.325 r*.. rv;v. v* 3 : 5 . 5C2 0.604 2.607 '';ria!u:r 0.9 53 0.320 0 . f? `t 0 2.534 Met. Cor-. 292 . 242. 28 . '-'''.'ra.u.':tci iz distribu Poisson (? = 1 .102 5) conti oo.-.cy table "ooocness-of -fit" tost- .. . Cr 2 cur.:..._ . o iff.' .'it. ! 0.05 ( .i j interval f:,,-r ' cc~b. :r.c"i concurs trot: ion " : 2a = 0.0r- "PP---r 2 ir:t : 257 . , ,, r , . "3 x j*;' particj.os/rr.-' .ctirx.to : 268. K X 1 Q 0 li::: i : 1 / C . S2 .. .i ftb j > * Vn : o /.o : trtk-.-:. ai 1' -: i Lc- 1 W );..< TO! > Mill' 'UrUM., J.1 I.WJJIU 41'J^^ Gv'j'i*-. , >. /UiJLL i 1 % * T 4. Far.: Fiber Count :`l Grid.I .^tfahiggg^- oi$rr5 Fiber Count Grid 2 JVV'l-MB*-*.U jrqr r** !rr III* rlfTfTn ..-.-VMlAi -fit TTfcM u{,mV.iPt* llnritf aHMS*iM Selected area electron diffraction patterns of mineral particles which all show a c-spacing of about S X. Ficure 1 Indexed electron diffraction patterns of representative amphiboles ar.d nor: a<nphibo.l es showing essentially similar appearance. Figure 2 V- Ih '`-i-wsw .4. .r .. j;~-- Indexed electron diffraction patterns of different orientations of grunerite achieved by tilting specimens in the electron micro scope. I-'icrure k- * *- * 4 - - "V v. ,* . - - - . . ; ' * /; -V "2~ . _t. '- Elc.-c tron micrographs and diffraction patterns of anosite and <-!ritc- particles, which indicate that the asbestos and non- {i c L' c - '----patt :-rni. varieties can hbe distinguish.Ve--dJ are interpreted pprrooDpeerrllyv i- f- t-h* e-----d--i--f4 frwa13 cCtl li, ion Fi gure -} ic c, f o crite and arrosite looking down the c-axis. r~ 1 traces and (ICO) parting arc- till .; . f t:.in j i procucod by fir.o-sceio : ; rt .. : r shown. This lea dr; to the : ..." . t .t ;. .-rti wi th largo (100) facer.. i i i i