EPA-600/2-77-059
February 1977
Environmental Protection Technology Series
             EVALUATION OF ELECTRON MICROSCOPY
                        FOR  PROCESS  CONTROL IN THE
                                    ASBESTOS INDUSTRY
                                  Industrial Environmental Research Laboratory
                                        Office of Research and Development
                                       U.S. Environmental Protection Agency
                                 Research Triangle Part, North Carolina 27711

-------
               RESEARCH REPORTING SERIES

Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
enviionmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The live series are:
    1.   Environmental Health Effects Research
    2.   Environmental Protection Technology
    3.   Ecological Research
    4.   Environmental Monitoring
    5.   Socioeconomic  Environmental Studies

This report has  been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate instrumentation, equipment, and methodology to repair or prevent
environmental degradation  from point and non-point sources of pollution. This
work provides the new or  improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
                    EPA RE VIEW NOTICE

This report has been reviewed by the U.S. Environmental
Protection Agency, and approved for publication.  Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, nor does mention of trade
namos or commercial products constitute endorsement  or
recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.

-------
                                       EPA-600/2-77-059

                                       February 1977
EVALUATION  OF  ELECTRON  MICROSCOPY

           FOR  PROCESS CONTROL

        IN  THE  ASBESTOS  INDUSTRY
                         by

            R. M. Gerber and R. C.  Rossi

             The Aerospace Corporation
                  P.O.  Box 92957
            Los Angeles , California 90009
                 Grant No. R802394
                ROAP No. 21AFA-011
             Program Element No.  1AB015
        EPA Project Officer:  D.  Bruce Harris

     Industrial Environmental Research Laboratory
       Office of Energy, Minerals, and Industry
          Research Triangle Park, NC  27711


                    Prepared for

    U.S. ENVIRONMENTAL PROTECTION AGENCY
          Office of Research and Development
               Washington, DC  20460

-------
                               ABSTRACT
      The overall objective of this study was to evaluate the transmission
electron microscope and scanning electron microscope as potential tools
for counting fine-particle asbestos fibers  for process control in the asbestos
industry. The capabilities  and  limitations of both instruments were defined
in applications where asbestos specificity is not necessarily required and
where cost of analysis must be  minimal.  It •was shown that both microscopes
are equally capable  of counting  all fibers in the full particle-size distribution.
However, because of fiber agglomeration  and difficulty in distinguishing
fibers from background,  each microscope is capable of observing only 75 per-
cent of the distribution.   In contrast,  present standard light microscopy
methods observe only the coarser 10 percent of the distribution; the fine
fibers are not resolved.

      Optimum asbestos fiber counting was  carried out at 15,000 X magnifi-
cation and at fiber concentrations on the filter of 40, 000 to 80, 000 fibers/
mm .  The minimum number of fibers counted to obtain high  statistical con-
fidence was 200 fibers per data point.  Standard preparation techniques for
the filter samples were found to have no effect for either instrument.  Ash-
ing of filters to remove nonasbestos  fibers was responsible for an asbestos
fiber loss of 85 percent.

      This  report was submitted in fulfillment of Grant No. R802394-02-0
by The Aerospace Corporation under the sponsorship of the U.S. Environ-
mental Protection Agency.  This  report covers the period January 1974 to
January  1977,  and work was completed as of 30 December 1976.
                                    111

-------
                                CONTENTS
Abstract	    iii
Figures	    vi
Tables	    vii
Acknowledgments	   viii

      1.     Conclusions  	     1
      2.     Introduction	     3
                  Health Effects	     3
                  Problem Definition	     4
      3.     Instrumentation	     7
                  Transmission Electron Microscope	     8
                  Scanning Electron Microscope	     8
      4.     Experimental Objectives	    10
                  Preparation Techniques  	    10
                  Detectability Limits	    11
                  Instrument Operating Conditions	    11
                  Effects of  Concentration Variations	    12
                  Summary of Objectives	    12
      5.     Experimental Procedure	    13
                  Sample Production	    13
                  Preparation for Observation	    14
                  Observation and Counting	    16
                  Data Analysis	    17
      6.     Experimental Results	    19
      7.     Effects of Test Parameters	    30
                  Fiber Deposition Uniformity	    30
                  Effect of Magnification	    32
                  Effect of Sample Preparation Procedures	    32
                  Effect of Fiber  Concentration	    34
                  Effect of Instrument  Performance	•.	    36

References	    40
Appendices

      A.    Size  Distributions of Asbestos Dusts	    43
      B.    Data Corrections	    45
                                     v

-------
FIGURES
Number
1


2

3

4

5
6

7
8

9
10
11
12
13
14
15

16

Schematics of scanning electron and transmission
electron microscopes and reflected light and
transmitted light microscopes 	
Schematic of general test plan and examination
plan for typical set 	
Percent standard deviation vs fibers counted
per sample 	
Comparison of measured fiber densities vs magnifi-
cation for each group 	 	 	
Distribution of fiber lengths for all fibers counted 	
Composite fiber diameters for all groups and
magnifications 	
Total fiber length and diameter distributions 	
Distribution of aspect ratios for each magnification
and preparation 	
Group I fiber length distributions 	
Group II fiber length distributions 	
Group III fiber length distributions 	
Group I fiber diameters 	
Group II fiber diameters 	
Group III fiber diameters 	
TEM and SEM photomicrographs from Nuclepore
filters 	
TEM and SEM photomicrographs from Millipore
filters 	
Pag


9

15

23

25
25

26
26

27
27
28
28
29
29
29

37

37
    VI

-------
                                TABLES


Numbe r                                                             Page

   1    Fiber Count for Group I	   20

   2    Fiber Count for Group II	   21

   3    Fiber Count for Group III	   22

   4    SEM Comparison of Filters A and D	   31

   5    Comparison of Fiber Statistics for Filters Prepared
       for SEM and TEM Examination	   33

   6    Comparison of Fiber Statistics for Filters Prepared
       for SEM Examination and by Ashing for TEM
       Examination	   34

   7    Sample As Observed by SEM and TEM	   38
                                   Vll

-------
                    ACKNOWLEDGMENTS
The authors wish to acknowledge the efforts of Richard A. Brose,
who prepared all SEM and TEM samples and performed the SEM
examinations, Ethel J. Watts,  who performed all TEM exami-
nations,  Martha A. Perez, who counted and categorized the fibers
from all SEM and TEM photographs, and Ronald V. Peterson,
who set: up the aerosol generator system.
                              vin

-------
                               SECTION 1

                              CONCLUSIONS
      Evaluation of air-pollution abatement measures in the asbestos
industry has been hampered by the lack of a. cost-effective asbestos fiber
counting method that is capable of detecting fine asbestos fibrils.  Previous
studies have shown that the TEM and the SEM offer this capability in appli-
cations where all fibrous materials can be assumed to be asbestos.  The
objective of the present study was to evaluate the capability and limitations
of the TEM and SEM for counting asbestos fibers from a particle distribution
expected from effluents in process industries controlled by dust-abatement
equipment.

      The results of this  study indicate that the SEM and TEM are equally
capable of counting asbestos fibers in process  control applications where
fibrous materials can be  assumed to be asbestos.   The study was made on
filter samples  prepared specifically for each microscope; fiber length and
diameter were measured on more than 60,000  asbestos fibers.  A compari-
son of fiber counts on specifically selected filters showed that each instru-
ment failed to observe 26 percent of the fibers observed by the other instru-
ment.  The  SEM was not  capable of resolving individual fibers in dense
agglomerates;  the  TEM confused the asbestos fibers  with the texture of
support filters. If the same fiber distribution  were measured by phase
contrast light microscopy,  90 percent  of the fibers would not be  observed
because  of inadequate resolution.

      In  an evaluation of the statistical confidence in the data, it was deter-
mined that the  standard deviation of individual  measurements, defined as the
fiber count of a single field of view,  obeyed a chi-squared distribution with
respect to the number of  fibers counted.   From this  relationship, it was con-
cluded that the deposition of fibers on a. set of filters was uniform, and that
the individual fiber counts were distributed in a normal distribution about
the mean.  It was also determined from this  relationship that a count of
200 fibers represents the cost-optimum fiber count for maximum confidence
in the data.

      The particle  size of all fibers counted was found to be  log-normal in
both fiber length and fiber diameter, with a mean fiber length of 1. 5 |j.m and
a mean fiber diameter of 0. 17 |j.m.  Only  16 percent  of the fibers were longer
than 5  fim; 1 percent were shorter than 0. 1 [im.

-------
      A study of test parameters revealed that a magnification of 15, 000 X
is optimum for both SEM and TEM examination.  At this magnification, the
resolving power permits  observance of the finest asbestos fibril, and the
field of view is adequate for cost-effective  fiber counting in routine analyses.
A correlation between concentration of fibers on a filter and the accuracy of
the fiber count showed that 40, 000 to 80, 000 fibers/mm^ is the optimal
concentration.

      Results also indicate that neither SEM nor TEM standard preparation
techniques affect the accuracy of the resulting count statistics.  However,
when filters were ashed to  remove nonasbestos fibers,  the resulting  count
indicated an 85-percent loss of the asbestos fibers on the filter.

      For asbestos fiber counting in routine applications, it was shown that
both the SEM  and the TEM yield comparable results that are much better
than those obtained with light microscopy methods.  However, capital and
operating costs for TEM analyses are 5 to  10 times  higher than those for light
microscopy.  In contrast, SEM analyses can be performed for approximately
twice the cost of light microscopy methods. Moreover, although TEM anal-
yses are not amenable to new cost-saving techniques,  automation of the SEM
is possible; through use of newly developed instrumentation for image anal-
yses, total costs for the SEM may be reduced to nearly those of light
microscopy.

-------
                               SECTION 2

                             INTRODUCTION
HEALTH EFFECTS

      The health problem that results from inhalation of asbestos fibers by
workers in the asbestos industry was first recognized in 1900.  That this
hazard could affect the general population was recognized little more than
a decade ago (1, 2).  A large quantity of information has since been pub-
lished on the physiological effects of asbestos inhalation,  but the mecha-
nisms that cause these effects are still unknown (3).

      Inhalation of asbestos has been shown to cause asbestosis  (4-6), pleu-
ral or peritoneal mesothelioma (7-9), and bronchial  carcinoma (10-12), each
of which can be fatal.  One obstacle to the detection and correlation of the
diseases is  the very long induction period (20 to 30 years) between  initial
exposure and evidence of biological effects.   In addition, the incidence  of
these diseases among nonindustrial workers has created further uncertainty
as to the causes  (13).  Because "asbestos" refers  to two mineral groups
that encompass several distinct fibrous  minerals,  the physiological effects
of specific mineral types are not understood and have not been included in
most medical reports (14).

      A further complication was  introduced by a recent study (15)  that
showed  a higher  incidence of lung cancer among cigarette-s'moking asbestos
workers than among nonsmoking asbestos workers.  One possible explana-
tion for many of the uncertainties about asbestos diseases is the belief  (3)
that asbestos serves  as a carrier of carcinogenic agents, specifically poly-
cyclic aromatic hydrocarbons.  The source of these  hydrocarbons could be
the natural oils of the asbestos fibers (16), adsorbed hydrocarbons from the
urban environment (17), or hydrocarbons adsorbed in cigarette smoking by
in vivo asbestos  fibers (3).
                                                    f,
      The presence of relatively high concentrations of asbestos fibers  in
urban environments and the uncertainty as to the cause of the associated
diseases have prevented the establishment of a dose-response relationship
between exposure levels and disease. Consequently, a safe level of asbestos
fiber exposure cannot be determined (18).  The problem is further  aggravated
by the presence of asbestos fiber  in foods (19), drinking water supplies (20),
and parenteral drugs (21), and the uncertainty with respect to the health
effects.

-------
      The  sources of asbestos fibers in the urban atmosphere are mines,
manufacturing processes,  and users (3).  Abatement measures within the
asbestos industry include elaborate ventilation systems that are designed to
protect both workers and the local community (22).  Because asbestos fibers
that can be inhaled are of a size that renders them easily airborne (23), the
effectiveness of abatement measures must be  nearly absolute if adequate pro-
tection is to be provided.

PROBLEM DEFINITION

      Although asbestos exposure was identified as a potentially fatal occu-
pational risk, as early as 1907,  serious disease among asbestos workers still
persists,  and the  disease is also found in the general population (24).  A
major reason for  the persistence  of the disease is the  long period of latency
between exposure and overt symptoms.  Although  industry has adopted mea-
sures to protect workers and the general public, insufficient knowledge  of the
physiological causes of the disease makes it difficult to ascertain whether or
not present abatement measures are capable of preventing the disease by the
year 2000.  One reason for this uncertainty is the  limited effectiveness  of
present techniques for measuring airborne asbestos.  Although methods for
determining concentrations of fibers have been developed for industrial
hygiene purposes  (25-26),  the methods do not  provide an understanding of
the relative biological effect of different  sizes of fibers, their concentra-
tions,  or the fiber mass (27).

      Standard methods for asbestos measurement use light microscopy with
phase-contrast illumination to count fibers collected on a membrane filter.
In current practice, only fibers longer than 5  |om with  diameters greater than
1 |j.m  are counted.  However, of the airborne asbestos particles, the long
asbestos fibers that can penetrate  to the  alveolar regions are those with
diameters  less than about  3 um. '   Several investigations (Appendix A) have
indicated that the  mean particle diameter and  mean length of airborne asbes-
tos distributions  lie at about 0. 3 and 1  fj,m, respectively, depending on the
source  of the sample.  Thus, more than  50 percent of  the total fiber number
in the airborne distribution may not be observed and is not counted by optical
microscopy.  These data indicate that standard light optic methods of asbestos
fiber  counting are capable  of detecting  only the coarser portion of the airborne
asbestos distribution and,  of the fibers counted, only those between 1  and 3 (Jim
*
 Several investigations (28,  29,  30) have determined 10  jam to be the maxi-
 mum aerodynamic diameter of dust particles that are inhaled and remain
 deposited within the lungs.  For asbestos  fibers, identified by a length/
 diameter ratio greater than 3, the equivalent maximum aerodynamic diam-
 eter is 3. 3 (Jim (31).  Observation of fibers with lengths up to 200  |jum deep
 within the lungs is not inconsistent with this size limitation, because the
 aerodynamic behavior of asbestos fibers is controlled by the fiber diameter
 rather than the length.

-------
diameter have biological  significance.  Thus, the standard methods for
determining fiber concentration may not provide control on a biologically
significant portion of the airborne asbestos fiber distribution.

      Most  of the histological record has been concerned with the effect of
long fibers,  probably because techniques for the preparation of histological
samples and their observation are made easier  by long fibers.  Because of
the record,  many investigators have evidently concluded that long fibers are
more dangerous  than short fibers.  However, Hold (32) reported that these
findings,  i.  e. , very fine dust not seen in light microscopes, coupled with
experience with rats,  resulted in the inevitable  conclusion that very  small
dust particles are at least as lethal as long fibers.  Because the relationship
between fiber size  and asbestos disease has not been clearly established,
and the mechanisms that contribute to  asbestos  diseases are not yet  under-
stood, it is  not known whether or not the uncounted fibers contribute  to the
epidemiology.

      Light  microscopy  with an oil immersion lens cannot resolve fibers with
diameters less than 0.2 |j.m.  Electron microscopy methods are capable  of
resolving to less than 0. 001 jim with a transmission electron microscope
(TEM) or to less than 0.01 jxm with a scanning electron microscope (SEM).
Both instruments can provide a more complete description of the asbestos
fiber distribution.  The  U.S. Environmental  Protection Agency recognized
the potential role of electron microscopy, and in 1971 a procedure for TEM
examinations of airborne asbestos was proposed (33). The major short-
comings of the TEM method are the long, tedious  sample preparation and
the high cost of analysis (typically ten times  the cost of light optical micro-
scopy).  Additionally,  where there is an unequivocal need for asbestos  identi-
fication, the method involving electron diffraction is too expensive for routine
use.

      The SEM is a more recently developed instrument and has not been
extensively  evaluated for asbestos counting (34).  The cost of analysis may
be near that for the light microscopy, but this also has not been evaluated.
The major shortcoming  of the SEM is the high cost of distinguishing asbestos
fibers from other fibers by x-ray analyses.   (A scanning transmission elec-
tron microscope  (STEM) combines the features  of the SEM and TEM  and per-
mits the identification of asbestos  by electron diffraction; however., this  capa-
bility also increases analysis cost, which makes the method impractical for
routine use.)  Although this shortcoming may limit use of the SEM for ambi-
ent air monitoring, the instrument should be completely adequate for mea-
suring the effectiveness  of industrial control equipment and for certain
epidemiological studies.  In both of these applications,  the electron micro-
scope may be capable of measuring the total  fiber distribution.  This will
assure adequate industrial control of asbestos fibers and provide a tool for
gaining an understanding and control of asbestos diseases.

-------
      This study focused on evaluation of the TEM and SEM as tools for
counting fine asbestos fiber in routine application for the subsequent evalua-
tion of control equipment in the asbestos process industry. Alternative
sampling and preparation techniques were investigated,  and the two instru-
ments were compared in order to determine the limitations and consistency
of each.  An expected result of this study,  although not a study objective, is
the establishment of a measuring technique  for counting fine asbestos fiber
that can  provide a data base for the determination of the biologic effect of
fine fibers.

-------
                               SECTION 3

                           INSTRUMENTATION
      Electron microscopy has been available as a research technique for
20 years, and the TEM has been recognized as a tool for asbestos counting
for at least 10 years.  Although the TEM provides the capability for observ-
ing and counting the smallest asbestos fibril,  and development studies have
been completed (33), the instrument has not been used for asbestos measure-
ment,  except for selected research, primarily because of the high costs
involved.

      During the last five years, however, the SEM has  been developed suf-
ficiently to be considered an alternative tool to the TEM for  counting small
fibers  (34).  Although the SEM does not offer the ultimate resolution of the
TEM (7 nm versus 0. 3 nm), the resolution of the'modern SEM is sufficient
to detect the smallest asbestos fibrils  (20-nm dia).  In addition,  the SEM
offers  the advantage of much  simpler sample preparation, which increases
throughput and decreases the cost of routine asbestos examination and count-
ing. The SEM,  therefore,  could fill the gap between the light microscope
and the TEM, with all three instruments being used complementarily. Light
microscopy, with the lowest initial  equipment cost, can  provide estimates of
fiber concentrations.  The SEM, with higher  initial cost, provides the abil-
ity to count small particles.  The TEM, with highest overall cost, permits
research studies of individual fibrils and basic asbestos  structures.

      For a better understanding of each technique, it is important to know
the basic differences between light and electron microscopes.  For example,
because of differences in their modes of operation, the two types of micro-
scopes have different sample requirements.  The  electron microscope dis-
plays sample response to electrons  rather than light; therefore,  the electron
image  of a particular sample may be quite different than the  light image of
the same area.  The samples observed in an  electron microscope must be
electrically conducting in order to receive the electron beam and avoid charge
buildup that occurs when an electrically conductive path to ground does not
exist.  In addition, electron microscopy requires that the sample be examined
in a vacuum.  Thus, samples prepared for electron microscopy must be cap-
able of withstanding both'the vacuum environment during examination and the
turbulence that may occur during chamber evacuation prior to examination.

-------
TRANSMISSION ELECTRON MICROSCOPE

      The TEM can be thought of as an electron analog of the transmitted light
microscope,   The electron-light analog is basic to the optics of the two instru-
ments,  as shown in Figure  l(a).  Both instruments use a focused beam to
illuminate the sample, which,  after passing through  the sample, is further
focused and enlarged.  With the transmitted light microscope,  the enlarged
image is focused through an eye piece; with the TEM, the image is focused
on a phosphor screen.

      The S£.mple for the transmitted  light microscope must be translucent
to the illuminating light; in  the same manner,  the sample for the TEM must
be translucent to the electron beam.   Thus, for TEM observation, a sample
must be thin enough for electrons to pass through.  With asbestos fibers,
the normal method of sample preparation requires the transfer of particles
collected on a filter to a sample holder that is translucent to the electron
beam.  The asbestos  particles are then observed in  silhouette  against the
translucent background.   When the fibers are  small,  some of the electron
beam can pass through the fiber and reveal high-resolution  details of the
morphological features of each fiber particle.  The primary high costs of
this  technique are those  of preparing the sample for  observation.  Previous
work on developing TEM for asbestos  counting has been directed toward
sample  preparation methods.

SCANNING ELECTRON MICROSCOPE

      The SEM may be compared and  contrasted with the  reflected light
microscope £.s shown in  Figure l(b).  Both instruments can examine opaque
specimens,  ?.llowing observation of a  considerably broader  range  of speci-
mens than can be  observed  with transmission  systems.   Sample preparation
is usually simpler for reflected light microscopy than for transmitted light
microscopy;  in the same way,  SEM sample preparation is simpler than TEM
preparation.   The electron-light analog can be carried a step further in that
an opaque feature viewed by transmission microscopy will appear dark,
whereas the ssame feature viewed by reflected microscopy will frequently
appear  light.

      The SEM and reflected light microscope differ in that the reflected
light microscope focuses light to form an image after the light has left the
sample, whereas  the  SEM focuses  the  electron beam before it  strikes the
sample. The electrons that leave the  sample  as a result  of the electron
beam are collected and displayed on a  cathode ray tube (CRT).   The finely
focused electron  beam is scanned synchronously with the  beam in the CRT,
and each point on  the  sample is represented by a  point on the CRT screen.
This forms an image  much  like a television image.

-------
                    TRANSMISSION
                    LIGHT MICROSCOPE
                      TRANSMISSION
                      ELECTRON MICROSCOPE
               LIGHT SOURCE
             CONDENSER LENS
             OBJECTIVE LENS
                           I
             PROJECTOR LENS
             (eyepiece)
         OBSERVATION SCREEN'
         (photographic plate)
                  ELECTRON SOURCElJ
                     MAGNETIC  ,
                     CONDENSER

                -SAMPLE
                     MAGNETIC
                     OBJECTIVE^
               INTERMEDIATE   ||f
               IMAGE PROJECTOR1*1*1
                                     (a)
                                 'SECOND STAGE
                                  MAGNIFIED IMAGE
                                  (viewing screen or
                                  photographic plate)
                 REFLECTED LIGHT
                 MICROSCOPE
                        SPECIMEN
          OBJECTIVE LENS,
        CONDENSER
        LENS
       LIGHT
       SOURCE
PROJECTOR
LENS (eyepiece)
   OBSERVATION SCREEN"
   (photographic plate)
                           SCANNING
                           ELECTRON MICROSCOPE
                        ELECTRON
                        SOURCE
   MAGNETIC
   CONDENSER


   MAGNETIC
   OBJECTIVE II
SCANNING COILS

       SPECIMEN
                                                                CAMERA
                                                                      DISPLAY
                                                                      TUBE
                                 SIGNAL
                                 DETECTOR
                                     (b)
Figure 1.   Schematics  of scanning electron and transmission
              electron microscopes and  reflected light and
              transmitted light microscopes.

-------
                                SECTION 4

                      EXPERIMENTAL OBJECTIVES
      The overall objective of this study was to evaluate both the TEM and
SEM as potential tools for counting asbestos fibers in process control appli-
cations.  Additional objectives were to make direct comparisons of the two
instruments in order to determine their relative usefulness in process control
applications, and to identify and develop suitable sample preparation methods
in order to reduce operating costs.

PREPARATION TECHNIQUES

      An important objective of this study was to determine the degree to
•which asbestos count statistics  are affected by the specific preparation tech-
niques required for  each  instrument.  This  investigation included evaluation
of the individual stages in sample collection and handling procedures.  It was
generally believed that the probability of fiber loss increased as the number
of steps  (or complication of the step)  increased during sampling and handling
procedures.

      SEM preparation is  basically very simple.  In this procedure,  fiber
loss is most likely to occur during handling in sample transportation and
during evacuation of the instrument.

      For TEM examination, removal of the asbestos fibers from the collec-
tion filter is basic to preparation because filters are too thick for electron
transmission.  In this procedure,  loss of fibers is most likely to occur dur-
ing dissolution of the filter.  In addition, handling of the sample at each stage
may contribute to fiber loss.

      Ashing is not required for either SEM or TEM examination; however,
ashing provides a means  of removing most  nonasbestos fibers from samples
collected from the ambient air and is therefore part of the 1971 EPA-
proposed TEM technique.   Because the  technique involves burning the  sup-
porting filter, there is a  strong potential for loss or breakage of asbestos
fibers.  Further loss may occur during the  refiltering operation following
ashing.  After ashing, the sample is filtered and ready for subsequent SEM
or TEM preparation techniques.

      Selection of a  suitable filtering  medium is important, because the
proper filter will simplify counting and  facilitate examination.  An unsuitable
                                    10

-------
filter can contribute to excessive fiber loss during preparation and, therefore,
to inaccurate results.  Two types of filters are available commercially for
airborne asbestos collection:  cellulose fibrous filters and perforated plastic
membrane filters.

      Handling can result in sample loss or alteration at any stage  of prep-
aration.  Thus, one objective of this study was to determine losses caused
by handling and to identify procedures particularly susceptible to handling
losses.

DETECTABILITY LIMITS

      Another important  objective of this study was to investigate limits of
asbestos fiber detectability for each instrument and to determine whether or
not the difference in ultimate detectability  between the two instruments is
significant in asbestos counting statistics.   If a difference in detectability
limits were found, it was expected  to be caused by the difference in ultimate
resolution  between the two  instruments.  Whether or not the difference was
significant would depend  on the minimum size of asbestos fibers.  If all
fibers were large enough to be easily observed with both instruments, the
difference  in resolution was not expected to be significant for statistics on
asbestos counting.

INSTRUMENT OPERATING CONDITIONS

      An objective of this study was to establish optimum operating conditions
for each instrument so that the use of electron microscopy for asbestos count-
ing could be  evaluated.  Among the conditions to be investigated for each
machine were instrument response to alternative specimen preparation tech-
niques and to filter medium, most suitable magnifications for examination,
and contrast enhancement methods.  In all cases, an evaluation was made of
standard operating conditions as well as newly identified conditions revealed
in the  course of the study.

      Proper magnification selection is necessary  to provide accurate count-
ing statistics in minimal time.  If magnification is too high, large fibers  are
not easily observed,  and excessive time is needed  to count the small fibers,
as few fibers are visible in each frame.  On the other hand, if magnification
is too  low, small fibers may not be resolved; this will skew the measured
size distribution as well  as the total concentration  measurements.

      Contrast enhancement is particularly important for the SEM, where
fibers are  viewed against'a background that tends to obscure the  fibers if
insufficient contrast is available.  Lack of contrast is generally less  of a
problem for the TEM than for the SEM, but it can be a problem if specimens
are not prepared properly.
                                     11

-------
EFFECTS OF CONCENTRATION VARIATIONS

      Variations in the amount of asbestos collected from the air onto the
filter can influence the accuracy of fiber counting.  If too many fibers are
present, agglomeration or simple overlapping of fibers can occur, which
complicates the counting procedures.  If too few fibers are present,  the pre-
cision of the counting statistics will be doubtful,  and the counting procedure
will  require excessive time and cost.  An important experimental objective,
therefore, was to determine the effect of variations in fiber-counting statis-
tics  for each instrument.

SUMMARY OF OBJECTIVES

      The  effect of each of these four variables  on asbestos fiber-count sta-
tistics was determined by evaluating and comparing total particle count,
particle density, and particle size distributions. Comparison of the count
statistics produced by each technique resulted in conclusions  regarding the
variables inherent in the operation of both types of electron microscopes.
Through these comparisons and resulting conclusions,  both the TEM and
SEM were evaluated as potential tools for counting, asbestos fibers in pro-
cess control applications.
                                    12

-------
                                SECTION 5

                      EXPERIMENTAL PROCEDURE
      The  experimental plan of this  investigation was divided into four
discrete activities:

      1.     Samples that could be used for examination with both micro-
            scopes were produced.  The intent was to provide both instru-
            ments with identical inputs that represented samples that might
            be collected from field sites.

      2.     Samples were  prepared for  observation.  Because the prep-
            aration techniques are different for each microscope, the
            effects of these techniques were among the major variables
            to be evaluated.

      3.     The asbestos fibers were observed and counted by the method
            most suitable for each instrument.

      4.     The data were analyzed.  Total fiber counts for each data set
            were corrected for geometric and magnification effects and
            normalized for direct comparisons of fiber concentrations and
            particle-size distributions.

SAMPLE PRODUCTION

      All samples in this experiment were produced by aspirating samples
of asbestos with a Royco model 256 aspirator  and collecting the airborne
fibers on filters held in a Nuclepore double filter holder.   In all cases, the
asbestos used was Canadian chrysotile from a standardized sample (Duke
Standards, Inc.,  Palo Alto,  California).  In the  standard  sample, 70 per-
cent of the fibers were 8 |xm in length or less  (determined by optical micro-
scopy),  and the particle-size range was similar to that expected in air sam-
ples from process industries.   The immediate objective during  sample gen-
eration was to deposit asbestos as uniformly and consistently as possible to
provide equivalent samples for each technique.  The samples prepared were
free from nonasbestos fibrous contamination,  as most process plant environ-
ments are  expected to be,  because consideration of specificity was not within
the scope of the investigation.

      Seventy-five percent of the filters  used in the experiment were Nucle-
pore membrane filters (Nuclepore Corporation,  Pleasanton,  California),
                                    13

-------
characterized by  a smooth  surface with uniform,  round holes.   All
Nuclepore falters were precoated with carbon to provide conductivity dur-
ing examination.  The remainder were Millipore filters (Millipore Corpora-
tion,  Bedford,  Massachusetts), characterized by a fibrous surface.

      Asbestos  from the aerosol generator  was deposited on  36 filters.
The filters were  divided into three groups  (I through  III) of  12 filters
each, with e-ach group representing different fiber concentrations on the
filter.  Each group was composed of three replicate sets (1 through 3) of
four filters per set; the four filters in each replicate set were labeled A
through D.  A schematic diagram of the test  plan is presented in Figure 2,
which shows the organization of the test matrix and the course of
examination.

      Group I was produced  such that each filter had 150,000  fibers/mm .
Each Group II filter had 80,000 fibers/mm2,  and each Group III filter had
40,000 fibers/mm .  Within each set,  filters A,  B, and D were carbon-
coated Nuclepore filters with 0.2-um pore diameter.  Filter C  was a Milli-
pore filter with 0.45-um equivalent pore size.  (The different pore size
between Nuclepore and Millipore filters gave experimentally equivalent
retention.)  All four filters in a set  received  asbestos loading as part of the
same aerosol generator run.  Filters A and  D of each set were the first and
last,  respectively, to receive asbestos during each  run, and direct compari-
son was made between  their  fiber concentrations in order to ascertain that
asbestos loading did not vary through the aerosol generator run.

      Fiber deposition  was visually uniform  over the surface  of each filter.
Concentration,  uniformity of deposition,  and operating conditions for asbes-
tos deposition  (in particular,  flow  rates and deposition times)  were estab-
lished,  with the SEM and carbon-coated Nuclepore  filters used  for calibra-
tion.  Intentional fiber  concentration differences were accomplished by vary-
ing deposition times; all other parameters were held constant.  During
generation of Group I filter samples, the glass nebulizer in the aerosol
generator broke and was replaced.   Only after all data were collected was
it realized that the fiber length distributions  were different as obtained from
the two nebuli.zers.

PREPARATION FOR OBSERVATION

      The preparation techniques for observation with the two microscopes
are considerably different.   For the SEM,  an electrically conductive filter
is  used to collect asbestos from the ambient  air.  The filter is  then removed
from its sample collection holder, mounted on a  sample mount  for the micro-
scope, and inserted into the  SEM.  The lack  of processing after sample col-
lection is made possible by the use  of precoated filters.  Where the filters
are coated wif.h carbon before sample collection, it is possible  to achieve the
electrical conductivity  necessary for SEM  observation without disturbing the
sample.   The procedure often provides better contrast,  because the poorly
conducting asbestos fibers produce  a strongly charged image  against the
carbon-coated filter.
                                     14

-------
                                        OVERALL TEST PLAN ORGANIZATION
GROUP 1
150,000
FIBERS/mm?
X
X
                                              TYPICAL SET
                                                                 FIBER DEPOSITION
                                           TEM       TEM       TEM  TEM EXAMINATION
                                                                 SEM REEXAMINATION
Figure  2.   Schematic of general test plan and  examination plan for typical set.
                                                15

-------
      The preparation needed for TEM observation is considerably more
involved. As a starting point, the TEM preparation technique recommended
in 1971 by the EPA (33) was used to prepare specimens for TEM observa-
tion.  Briefj.y outlined, the procedure consists of (1) ashing the filter,
(2) redepositing on a second filter, (3) carbon coating the asbestos-bearing
filter, (4) dissolving the filter away, (5) floating the remaining asbestos-
bearing carbon film on water,  and (6) depositing the film on a TEM grid for
observation,,  Ashing is not an essential step in the TEM technique and is
primarily intended to eliminate nonasbestos material.   Ashing, therefore,
was used for only two sets of Group I filters processed for TEM examination;
this permitted observation and evaluation of the effects of ashing on the
asbestos samples.  Because ashing of TEM samples is not required,  all
other preparation involved carbon coating of the asbestos-bearing filter,
followed by  dissolution of the filter.  The carbon film bearing the asbestos
was then water-floated and placed on a TEM grid for subsequent
examination.

OBSERVATION AND COUNTING

      Observation and counting with the SEM was accomplished by inserting
the prepared sample into the evacuated SEM chamber through an airlock and
observing a number of areas on the filter surface.  The filter was held per-
pendicular to the electron beam, and 45 predetermined, arbitrary locations
were  photographed on each filter.  Asbestos fibers that appeared  in the
photographs were then manually counted and categorized by length, width,
and aspect ratio.

      Most fibers were counted and categorized while they were being ob-
served in the TEM;  the rest were measured from photographs.  All fibers
in five arbitrarily selected 80- by 85-fim TEM grid squares were counted
and categorized as with the SEM.  In addition, 20 TEM photographs were
taken for subsequent direct comparison with the SEM.

Initial SEM  Examination

      Filters A  and D from each set were examined with the SEM for a check
of deposition distribution and uniformity.  In addition,  examination of these
two filters provided a baseline for comparison of changes caused  by subse-
quent handling.  Both A and  D filters were  photographed 20 times at 6000 X
magnification and 25 times at 15,000 X magnification.   The 6000 X photographs
were  taken in a  4 by 5 matrix.  Ten of the photographs  taken at 15, 000 X were
of areas already photographed at 6000 X.   This provided more detailed infor-
mation on fiber  size, detectability limits, and possible tradeoffs  between
these limits and the advantages offered by the larger field of view in low-
magnification views.  The 15 remaining photographs taken at 15,000 X were
of sites other than those used for the 6000 X photographs.
                                    16

-------
TEM Examination

      Subsequent to SEM examination of filters A and D of each set, all four
filters in each set were prepared for TEM examination in accordance with
the described technique  (except for sets 1 and 2 of Group I,  which were first
ashed).  The filters  prepared by this technique were examined in the TEM
at 30, 000 X.   All fibers present in five grid squares were counted.  In addi-
tion,  for several filters, all particles in each of the five squares were also
counted and characterized from photographs as a check on the technique.
For all filters, photographs were taken at 10 preselected locations within
each  square at 30, 000 X magnification, and 10 photographs were taken at
15,000 X at these same locations.

SEM  Reexamination

      After TEM examination, each TEM sample was reexamined in the SEM
at both  15, 000 X and 30, 000 X and in the same locations as in the TEM exami-
nations.  Count statistics were compiled and  photographs taken that corre-
sponded to those compiled and taken during TEM  examination.  This proce-
dure  provided a direct one-to-one comparison between the SEM  and TEM
when the same sample was examined.

DATA ANALYSIS

      For most TEM analyses, data were generated and collected by means
of in  situ fiber counting methods in accordance with recommended procedures
from a previous study (33).  Manual counting from photographs was used for
the remaining TEM analyses and for  all of the SEM analyses.  Each fiber was
categorized by length and diameter, and approximately 25 percent of the fibers
on filters A and D were also categorized with respect to aspect ratio.  Fiber
data from each photograph were corrected for magnification differences to
convert all data to a common base (Appendix  B).  These data were then nor-
malized with respect to total fiber count for development of length, diameter,
and aspect ratio distributions.  The distributions  were then used in compari-
sons  and for correlations of the various experimental parameters evaluated
in this study.  For each filter, a value of fiber concentration was determined
from a total of all areas examined.  Standard deviations of fiber concentra-
tion were then calculated from each sample,  defined as either individual
photographs or groups of photographs.   Standard  deviation of fiber concen-
tration was an additional test parameter used in comparing  and evaluating
data.

Effects  of Fiber Deposition Uniformity

      The uniformity of fiber deposition was determined by  comparing the
fiber  concentration measured between filters  A and D of each filter set.
The difference in fiber concentration between the A and D filters in each
set was evaluated with respect to the distribution of concentration differ-
ences among all sets.
                                    17

-------
Effects of Magnification

      The comparison of fiber-count statistics between filters observed at
different magnifications was used in determining the effect of viewing mag-
nification oa counting accuracy.  This comparison was also used in deter-
mining the relationship between instrument detectability limits and observa-
tion of the smallest fibrils.

Effect of Preparation Procedures

      The fiber concentration and size distribution as determined from SEM
analyses on filters A and D were compared with those values determined
from  TEM analyses on filters A,  B, C, and D of each set.  This compari-
son provided a measure of the effect of preparation procedures and included
data with and without ashing in the TEM procedure.  Count statistics from
filter B (Nuclepore) were compared with those from filter C (Millipore) to
determine the effect of filtering media.

Effect of Asbestos Fiber Concentration

      The concentration of the asbestos fibers on the collection filter may
affect the accuracy of the fiber count.   For high concentrations, the individual
fibers may pile up and overlap so that an accurate  count cannot be made.
For concentrations that are too low, the statistical significance of the fiber
count may be inadequate.  For a determination of the effect of fiber concen-
tration on count accuracy, the standard deviation in fiber concentration and
particle-size distribution of each set was  evaluated as a function of the
concentration of fibers on each filter.

Effects of Instrument Selection

      The relative count accuracy of the two electron microscopes was
evaluated by comparing the count statistics from filters A, B,  C,  and D
as measured on the TEM with the count statistics from these filters as
reexamined from  grids on the SEM. Additionally,  the count statistics from
filters A and D in the initial SEM examination were compared with the count
statistics from these same filters in the SEM re examination.
                                     18

-------
                                SECTION 6

                        EXPERIMENTAL RESULTS
      The fiber-count statistics gathered in this study are presented in
Tables  1 through 3 for Groups I through  III, respectively.  Fiber concentra-
tion and length and diameter distributions are expressed in units of thousands
of fibers per square millimeter and calculated directly from fiber counts
and viewing areas in accordance with the method described.  Each data
point presented in these tables from 6000 X magnification is an average value
that represents the summation of counts  from 20 photographs.  Each data
point from 15, 000 X magnification is an average value that represents the
summation of counts from 25 photographs.  The 30, 000  X data are average
values of all fibers counted in five TEM  grid squares.  Also included in these
tables are the mean values of each statistic calculated from the three repli-
cate sets.  Because set 3 in Group I was not a replicate of sets 1 and 2,  a
separate summation is given.

      For evaluation of the experimental significance of these data, the
variation in fiber concentration, normalized to percentage, was examined as
a function of the number of fibers  counted.

      The percent standard deviation of fiber concentration for  each sample,
calculated from the mean of all samples  from a particular filter, is plotted
in Figure 3 as a function of the total fibers counted on each filter.   (A  sample
is either an individual photograph  or a group of photographs from a single
filter.)  These data show that the statistical variation in  fiber  density obeys  a
chi-squared distribution function with one degree of freedom.  The distribu-
tion may be represented by


                                        /2      B     \1/2
                 % standard deviation = la +	-—___ I
                                        ^    n exp (n) j


where a - 10, 6=1x10"  ,  and n is the  number of fiberjs counted,  normalized
to the median count (n = N/x, x = 260, N = fiber count).   The  a term repre-
sents the effect of variations due to uncertainty in magnification and is
reflected in the minimum value of the function that asymptotically approaches
10-percent standard deviation rather  than zero (Appendix B).  The constant B
in the preceding equation was empirically calculated from the data in
Figure  3.   The agreement between the measured data and the chi-squared
distribution indicates  that the fiber concentration variation from the mean is
                                     19

-------
                    TABLE 1.   FIBER COUNT FOR GROUP I*

Filter




























A
A
A
B
C
D
D
D

A
A
A
B
C
D
D
D

A
A
A
B
C
D
D
D
Magnification

6.000
15,000
30,000
30,000
30,000
30,000
6,000
15,000

6,000
15,000
30,000
30,000
30,000
30,000
6, 000
15,000

6,000
15,000
30,000
30,000
30,000
30,000
6,000
15,000
Instrument

SEM
SEM
TEM
TEM
TEM
TEM
SEM
SEM

SEM
SEM
TEM
TEM
TEM
TEM
SEM
SEM

SEM
SEM
TEM
TEM
TEM
TEM
SEM
SEM
Length (nm)
<1.0 1.0-2.5 >2. 5
Set
19. 4
18. 6
2. 3
14. 4
3. 5
4.2
12. 5
26. 6
Set
15. 4
7.4
2. 0
2.8
4. 3

7. 7
12. 6
Set
14. 1
55. 9
26.4
9. 3
7. 7
3. 0
22. 1
37. 5
J
23. 5
28.0
3. 3
3.0
1.4
4.4
26.3
28.0
2
17.9
16. 1
2.6
3. 0
7.0

14. 9
13. 6
3
59. 7
59.8
21. 5
43. 2
48. 0
12.8
32.9
51. 3

37. 2
40. 4
2. 1
0.9
1. 2
1. 8
36. 7
29. 5

38. 7
34. 9
1. 1
1. 1
3. 6

22. 1
22.9

35. 6
47. 5
7. 1
9.0
18. 5
5. 7
35.6
46. 6
Diameter (nm)
<0. 1

32.2
23.8
2. 7
0.3
0.2
3. 6
23. 5
9. 7

16.5
6. 7
0. 5
1. 7
3. 5

12. 5
10.8

27. 0
44. 3
20. 3
16.4
15.0
7. 8
22.6
24. 9
0. 1-1.0

46. 6
61.8
4.4
16.9
5. 1
6.4
51.8
73.8

54. 8
50. 4
4. 7
4.9
11. 1

32. 2
38.4

81.8
118. 4
34. 4
44. 6
57.9
13.2
66.9
110. 5
>1.0

1. 3
1.4
0.6
1. 1
0.8
0.4
0. 2
0. 7

0.9
1.3
0. 5
0.2
0. 5

0
0. 7

0. 5
0. 5
0.3
0.6
1.2
0. 5
1. 1
0
Total fiber
densities

80. 1
87.0
7. 7
18.4
6.0
10.4
75. 5
84. 1

72.0
58.4
5. 7
6.9
15. 1

44. 7
49. 1

109.4
163.2
55.0
61.5
74. 1
21. 5
90.6
135. 4
Mean
Sets
Sets
Sets
Set 3
Set 3
Set 3
1 and 2
1 and 2
1 and 2



6,000
15.000
30,000 (Ashed)
6,000
15,000
30,000
SEM
SEM
TEM
SEM
SEM
TEM
13.8
16. 3
4. 8
18. 1
46.7
11. 6
20. 7
21.4
3. 5
46.3
55.6
31. 4
33. 7
31. 9
1. 7
35.6
47. 1
10. 1
21.2
12.6
1.8
24.8
34.6
14. 9
46. 3
56. 3
7.6
74. 4
114. 5
37.5
0. 8
0. 9
0.6
0.8
0.3
0. 7
68. 2
69. 6
10. 0
100.0
149. 4
53.0

Fibers/mm  (in thousands)
                                            20

-------
                     TABLE 2.   FIBER  COUNT FOR  GROUP II"

Length (nm)
Filter

A
A
A
B
C
D
D
D

A
A
A
B
C
D
D
D

A
A
A
B
C
D
D
D
Magnification

6,
15,
30,
30,
30,
30,
6,
15,

6,
15,
30,
30,
30,
30,
6,
15.

6,
15,
30,
30,
30,
30,
6,
15,

000
000
000
000
000
000
000
000

000
000
000
000
000
000
000
000

000
000
000
000
000
000
000
000
Instrument

SEM
SEM
TEM
TEM
TEM
TEM
SEM
SEM

SEM
SEM
TEM
TEM
TEM
TEM
SEM
SEM

SEM
SEM
TEM
TEM
TEM
TEM
SEM
SEM
<1.

14.
31.
21.
15.
13.

17.
22.

18.
36.
19.
15.
24.
22.
24.
36.

15.
16.
21.
53.
11.
41.
31.
22.
0
Set
2
0
0
2
6

5
9
Set
0
8
1
4
5
0
8
4
Set
7
4
7
9
5
0
9
1
1. 0-2. 5
1
22.9
29.0
22.6
25. 1
18.4

19.0
19. 3
2
31.6
30. 9
18.8
20. 4
21. 3
20. 6
38. 5
25. 8
3
15. 0
18. 7
18.8
39.5
13. 6
33. 3
21. 5
28. 4
>2. 5

40. 4
24. 0
15. 3
20.2
14. 3

20. 5
10. 1

35. 3
26.6
Z2. 4
19.1
18. 1
25.6
38.0
21. 3

12. 3
15. 5
16. 7
19.8
14. 5
22. 2
17. 1
21. 8
Diameter (iun)
<0. 1

28.9
36.6
17.6
14.6
7.9

20. 7
14.5

22. 1
33.6
20. 5
16.0
6.5
18.9
32.9
19.3

9.0
10. 8
13.6
21.9
3. 2
14.8
22. 7
12.0
0. 1-1.0

48.5
47.4
40.8
44.9
37.9

36.3
37.8

62.4
60. 7
39. 6
38. 4
57.3
49.2
68.2
64.3

34. 0
39.9
43. 5
90.9
35. 7
81.5
47. 6
60.3
>1.0

0.2
0
0.5
0.9
0. 5

0
0

0.4
0
0. 3
0.5
0. 1
0. 1
0. 2
0

0
0
0
0. 3
0.8
0.3
0. 2
0
Total fiber
densities

77.5
84.0
58.9
60.4
46.2

57.0
52.3

84.9
94.3
60.4
55.0
63.9
68.2
101.3
83.5

43.0
50. 7
57.2
113. 1
39.6
96.5
70. 5
72.3
Mean



6,
15,
30,
000
000
000
SEM
SEM
TEM
20.
27.
23.
4
6
5
24.8
25. 4
22.9
27. 3
19.9
18.9
22. 7
20.6
14. 1
49. 5
52. 3
50.9
0. 2
0
0.4
72.4
72.9
65.4

Fibers/mm  (in thousands)
                                               21

-------
                   TABLE  3.   FIBER COUNT FOR GROUP Ilf

Length (nm)
Filter

A
A
A
B
C
D
D
D

A
A
A
B
C
D
D
D

A
A
A
B
C
D
D
D




Magnification

6,
15,
30,
30,
30,
30,
6,
15,

6,
15,
30,
30,
30,
30,
6,
15,

6,
15,
30,
30,
30,
30,
6,
15,

6,
15,
30,

000
000
000
000
000
000
000
000

000
000
000
000
000
000
000
000

000
000
000
000
000
000
000
000

000
000
000
Instrument

SEM
SEM
TEM
TEM
TEM
TEM
SEM
SEM

SEM
SEM
TEM
TEM
TEM
TEM
SEM
SEM

SEM
SEM
TEM
TEM
TEM
TEM
SEM
SEM

SEM
SEM
TEM
<1.

12.
15.
14.
28.
15.
1Z.
9.
16.

15.
17.
11.
15.
17.
22.
16.
11.

17.
14.
21.
18.
16.
16.
14.
11.

14.
14.
17.
0 1.
Set 1
5
3
4
1
6
1
6
7
Set 2
0
9
7
2
6
2
1
7
Set 3
9
9
3
6
7
7
0
3
Mean
2
6
5
0-2.5

14. 2
17.9
16.4
15. 7
15.0
10. 5
9.9
11.9

14. 0
17. 0
13.0
13. 2
11.8
13.8
14. 0
12. 5

14.8
13.3
16.9
12. 7
11.4
12.2
15.0
4. 3

13. 7
12.8
13. 6
>2. 5

11. 7
10. 8
16.4
17.6
14. 9
16.7
12.3
9. 7

14.8
12. 6
17. 0
16.7
15. 8
17. 1
12. 7
15. 3

17.6
16.5
17. 7
15.3
7.0
15. 4
17.2
13. 9

14.4
13. 1
15. 6
Diameter (urn)
<0. 1

14.9
6.4
10. 9
17. 7
9.0
11. 7
6.1
7.6

11.9
7. 7
13.4
14.9
12.0
17. 1
9. 3
10. 7

19. 4
11.3
18.8
15.6
9.8
17. 7
11.2
9.2

12.2
8.8
14. 1
0. 1-1. 0

23. 5
37.6
36.0
43.6
36. 2
27. 3
25. 7
30. 7

31.9
39.8
28. 3
30.1
33. 0
35.9
33. 5
28.8

30. 9
33.4
37.0
30. 9
24.9
26.5
35.0
20. 3

30.2
31.8
32.5
>1.

0
0
0.
0
0.
0.
0
0

0
0
0
0.
0.
0.
0
0

0
0
0.
0.
0.
0.
0
0

0
0
0.
0



2

2
3






2
2
2





2
1
2
2





2
Total fiber
densities

38.4
44. 0
47.2
61.4
45. 5
39.3
31.8
38.3

43.8
47. 5
41. 7
45.1
45.2
53. 1
42.8
39.5

50.3
44. 7
55.9
46.6
35. 1
44. 4
46.2
29.5

42.2
40. 6
46.7

Fibers/mm  (in thousands)
                                            2Z

-------
Q



Q
a
o
cc:
                                              SCALE
30



20




10




 0
                                            840 1020 1200 1380 1560 1740 1920 2100 2280


                                                 NUMBER OF  FIBERS COUNTED
                                                            A TEM

                                                            O SEM
  _A
                                                                                      A



                                                                                      A

                                                                             A    A

                                                                              AO  A   A
1 1
A ^
1 1 1 1 1 1 1 1 1 1 1
>%
\
A
|
A
A
|
^?v £&
AA £
I |
^/Q A^ "
>AZ$ A
r i i i T
-fs
£
60      120     180     240     300     360     420     480


                         NUMBER OF FIBERS COUNTED
540
                                                                        600.
660
                                                                                         720
     Figure 3.  Percent  standard deviation vs fibers counted per  sample.
                                            23

-------
normally distributed and that, except for the displacement due to uncertainty
in magnification, no systematic variation is present.

      For sample sizes larger than 200 fibers, the precision in the measured
fiber concentration  is reasonably invariant; for sample sizes less than 200
fibers, the precision in measured concentration is strongly affected by the
total fiber count.

      The distribution in the total fiber concentration for each group with
respect to the examining instrument and magnification is shown in Figure 4,
where Group I is divided into two parts because set 3 did not replicate sets  1
and  2.  The distributions of fiber length and fiber diameter for all fibers
counted are shown in Figures 5 and 6, respectively.  These distributions
show that the fibers counted are representative of airborne distributions;
most of the fibers are less than 1 \im in diameter,  and only a small number
are  longer than  5 urn.  These distributions were  replotted on a logarithm-
probability graph paper in Figure 7, where both length and diameter distri-
butions are shown to be log-normal; the mean fiber length is  1. 5  (im,  and the
mean fiber diameter 0. 17 (j.m.  This distribution shows that virtually all
fibers counted were biologically significant (< 3. 3 fim).   Only 16  percent  of
the fibers were  longer  than 5 fo.m,  and 1 percent were shorter than 0.  1 (j.m.
A discussion of  distributions, from the literature,  is presented in Appendix A.

      The distributions of aspect  ratio for  the magnifications used in the
study are plotted in Figure 8, where the aspect ratio distributions are shown
to be relatively  constant even after ashing.  Ashing, however, appears to
reduce the percentage of long fibers to  some degree. The data for this figure
were obtained from  reexamination of photographs and represent only a portion
of all fibers counted.

      The bar graphs in Figures 9 through 11 show fiber  length distributions
by set for Groups  I through III,  respectively.  Each of three length catego-
ries is represented by  a bar.  Three such graphs,  each representing the
distribution of a single set, are then superimposed to form a nine-element
graph.  The c.ata for instrument and magnification are represented by nine-
element  graphs, and three such graphs represert the group.  Figures 10 and
11 also show the average value for the three  sets by an additional bar that is
more prominently displayed than  that of the individual sets.

      The bar graphs in Figures 12 through 14 represent the  distribution of
fiber diameters for  each of the three groups.  Each magnification is shown
separately,  and the  range of diameters is divided into four parts.  The dis-
tributions for Group II  and III are nearly identical.  Group I distributions
are dissimilar,  and the reasons for this effect are discussed in Section 7
under Effect  of Fiber Concentration.
                                     24

-------
160
140
~£ 120
_E
5
^ 100
o
| 80
LU
O
O
0 60
LU
CQ
Ll_
40
20
0
r- GROUP \
S
—

—




GROUP \
SETS 1 AND 2

x
LU o
CO vO
X
S °
S ,rr
00 —1
X
8
o
5
n



X
S 8
UJ o
CO \o
ETS


x
S ^




x
._§
"°



GROUP

x
S X
LU §
CO so
X
o
LU irT
1
X
S °
LU O
GROUP III

x x
x S °
lll£-|s-
Figure 4.  Comparison of measured fiber densities vs magnification

           for each group.
   o
   o
    i
           12345.67    8    9    10    11   12


                             FIBER LENGTH, urn
   Figure 5.  Distribution of fiber lengths for all fibers counted.  Set 3


              of Group I is excluded.
                                 25

-------
70
§ 60
ID
0
o
en 50
ct:
LU
DQ
U_
-40
o
o 30
i—
LU
O
LU 20
o_

10



—

.




O








A



FIBER DIAMETERS
O <0. IfjLfn
A 0.1-0.5
• 0.5-1.0
A >1.0



• ^
             Figure 6.  Composite fiber diameters for all groups

                         and magnifications.
          10
          3.0
       3:
       1--
       Cl
          1.0
          0.3
          0.1
                                    I   I  I
                 I
                                                                           3.0
                                                                           1.0
                                                                              E
                                                                              a.
                                                                           0.3 2
                                                                              -
                                                                           0.1
           0.01   0.1      1      5  10  20  30 40  50 60 70  80  90  95

                               PERCENT LESS THAN STATED SIZE
99
       0.03
99.9   99.99
Figure 7.  Total fiber length and diameter distributions.   Set 3 of Group I

            is excluded from length data.
                                        26

-------
80





70





60
o 50
o
 .
UJ
oo 40
o
o:
             SEM

             6000 X
                                  TEM

                                  30,000 X
                                               TEM

                                               30,000 X (ashed)
                                                           £23 3:1-6:1

                                                           HI >6:1
30





20





10
Figure 8.  Distribution of aspect ratios for each magnification

            and preparation.
	 SET 1
cr«. 	 SET 2
SEM .._ SET 3
15'°°OX 02 Sum
SETS 1 AND 2 WERE
*_._....


•
ASHED PRIOR TO TEM
[

O
LAMINA
TEM
30, 000 X
A
TION
         Figure 9.  Group I fiber length distributions.
                                 27

-------
35
30
25
15
10
5
-

-


O
TEM
6000 X

A
, 	 1

•


	 1
O
SEM
15, 000 X

A

	
•




O
TEM
30, 000 X
	 SET 1
— SET 2
— SET 3
— AVERAGE VALUE
FOR ALL SETS
O <1.0 fj.m
A 1.0- 2. 5
• >2.5

A


	
•

    Figure 10.  Group II fiber length distributions.
UJ  0 —
             SEM
             6000X
     	SET 1
     — SET 2
     — SET 3
     — AVERAGE VALUE
 SEM     FOR ALL SETS      TEM
15.000 X  O<1.0/xm        30.000 X
18
15
12
9
6
3
"
-
	


0



A
	


•




O


A
£* 1. U
• >2


'
-t.'i
5/ilTl


0

A
	
•

   Figure 11.   Group III fiber length distributions.
                              28

-------
  80
  70
  50
o

540
o
0
  20
  10
              SETS 1 AND 2
 SEM
6000 X
                  SEM
                 15,000 X
                             12°
                             105
                                      90
                                             SEM
                                            6000 X
SET 3

SEM
5.000X
A 0.1-0.5
• 0.5-1.0
              Figure  12.   Group I fiber  diameters.
80


70
CM 60
E
E
O
1<0



SEM
15. 000 X








—
O







A











^
)
42
36


TEM 30









O







A

•^M

24



18


12
6
, —


~



—

—



	


—
^^i^^M







SEM
6000 X








•—
O







A











^1
o <0. l^,m
A 0. 1-0. 5
• 0.5-1.0
• >1.0



c
15









O
JEft
00







A
\

fl TEM
Dx 30, 000 X







: 	

^

O







A











i]
  Figure 13.  Group II fiber
               diameters.
                                Figure  14.   Group III fiber
                                              diameters.
                                  29

-------
                                SECTION 7

                     EFFECTS OF TEST PARAMETERS
FIBER DEPOSITION UNIFORMITY

      The first and last filters from each deposition run (filters A and D of
each set) were examined and compared in order to establish the consistency
of asbestos deposition on the sample filters.  Filters A and D from each set
of four filters were  initially examined in the SEM before any  other operation
was performed.  Because all four filters in a set were sequentially exposed
to the asbestos generator air stream,  without alteration of deposition
parameters, comparison of the first and last filter in each set provided a
measure of c.ny systematic variation in fiber deposition in each set.  Com-
parison of fiber concentration variations on filters A and D from all sets
provided a m.easure of both the  random and systematic variation in fiber
deposition or. all the filters.

      Fiber counts at both 6000  X and 15,000 X magnifications were made, and
fiber concentration variations were calculated from the total  areas observed.
The fiber concentrations  from all nine filter sets are given in Table  4.  The
difference  in fiber concentration between filters A and  D for each set was
calculated,  divided by the sum of the fiber concentration from both filters,
and expressed in percent.  The fiber concentration variation  among the nine
sets ranged from 1 to 24 percent of the  average fiber concentration for each set.

      The mean value for the variation in the percent fiber deposition from
all sets was 5 percent and represents the systematic variation from zero.
The standard deviation calculated from all sets at 12 percent represents the
random variation.  Because the systematic variation was small relative to
the random variation, it appears that there is no significant systematic
variation in fiber deposition among the filter sets.  The variation in fiber
deposition for filters in any group was not significantly different than that for
all groups;  it was concluded, therefore, that fiber deposition variation is not
a function of the actual fiber concentration.

      The experimental significance of the random variation in fiber concen-
tration on the total experimental program was further  evaluated by comparing
the data of this experiment with the total data base presented in Figure 3.
The number of fibers counted on filters A and D among all  sets varied
between 200 aad  600 fibers, from which the expected percent standard devia-
tion in accordance with Figure  3 would decrease  from  20 to 10 percent.
                                     30

-------
                    TABLE 4.  SEM COMPARISON OF FILTERS A AND D

Fiber concentration variation

Group I
Set
Set
Set
Group II
Set
Set
Set


1
2
3

1
2
3



4.
27.
18.

20.
-16.
-27.
6000 X
•A*
r

6/155.6
3/116.7
8/200.0

5/134. 5
4/186. 2
5/113. 5
15,OOOX
Percent

3
23
9

15
-9
-24


2.
9.
27.

31.
10.
-21.
f*

9/
3/


171. 1
107. 5
8/298.6

11
8/
6/

136.3
177.8
123.0
Percent

2
9
9

22
5
-18
Group III
Set
Set
Set
1
2
3
6.
1.
4.
6/70.2
0/86.6
1/96.5
9
1
4
5.
8.
15.
7/82.3
0/87.0
2/
74.2
7
9
20

f = A-D/A+D, where A = fiber concentration on filter A of a set,  and D = fiber concentration
   on filter D of the same set (See Tables 1 through 3. )  Mean =  5 percent, standard
   deviation = 12 percent.

-------
The measured standard deviation of 12 percent from all sets lies within this
range; therefore,  it may be concluded that the measured random variation
does, in fact, represent the true variation in fiber deposition on the filters,
and that no systematic variation in fiber deposition exists either within a set
or among the sets.  It may also be concluded that an accurate value of fiber
concentration from each set can be determined from a sample size that con-
sists  of 200 fibers or more.

EFFECT OF MAGNIFICATION

      The choice of magnification  involves a tradeoff between effective
resolving power and area of coverage.  Magnifications ranging from 6000
to 30, 000 X were used to establish reasonable guidelines for effective
measurement. At 6000 X on the SEM, even the smallest fibril could be
observed; however,  as the  concentration of the fibers on the filters  increased,
fiber  clusters were created, and the 6000 X magnification was not adequate
for resolution of the smallest fibrils when they were partially obscured by
other fibers.  With the SEM, 15, 000 x magnification was usually sufficient  for
resolution oi; fine  particles, and an increase to 30,000x did not  significantly
increase the instrument's ability to resolve fine fibrils partially obscured
by agglomeration.  Because of  its higher resolution, the TEM was better
able to  distinguish individual fibers in an agglomerate  at 15,000x magnifica-
tion;  an increase to 30, 000 X magnification was not required for further
improvement.

EFFECT OF SAMPLE PREPARATION PROCEDURES

      The fiber density on filters A through D was determined to be uniform
within each set.  When the  SEM fiber-count statistics for filters A and  D
were  compared with the TEM data for these  same  filters, it was possible to
evaluate the  effect of the TEM's extensive  preparation procedures on fiber
count and fiber distribution.  But, because both filters were previously exam-
ined in  the SEM, filter B was included in order to  eliminate possible effects
of the SEM examination. TEM data from filter B were then compared with
the TEM datci from filters A and D. Filter C, a fibrous filter, was  added to
the study to permit comparison between the two types of filters.  (Filters A,
B, and  D are membrane filters.)

      A comparison of fiber concentration and fiber distribution is presented
in Table 5.   There is  no discernible difference between SEM results from
filters A and D and TEM results from these  same  filters.  In addition,  these
results are generally  in good agreement with the results from filter B.  The
data from filter C also agree well with all other data.  However, results
from Group I filters with high fiber concentrations did not agree.

      It was  concluded that filter preparation techniques for  either the SEM
or TEM (without ashing) had no discernible effect on the  results.  Moreover,
type of  filter did not affect  the  TEM results,  and the data are identical to
the SEM results obtained with a membrane filter.  (Fibrous filters are not
easily used for SEM examination.  ) Thus,  neither  sample preparation nor
                                     32

-------
     TABLE 5.  COMPARISON OF FIBER STATISTICS FOR FILTERS
                 PREPARED FOR SEM AND TEM EXAMINATION

*
Group Set Filter
I 3 A + D
A + D
B
C
II 1 A + D
A + D
B
C
2 A + D
A + D
B
C
3 A + D
A + D
B
C
in i A + D
A + D
B
C
2 A + D
A + D
B
C
3 A + D
A + D
B
C
Instrument
SEM
TEM
TEM
TEM
SEM
TEM
TEM
TEM
SEM
TEM
TEM
TEM
SEM
TEM
TEM
TEM
SEM
TEM
TEM
TEM
SEM
TEM
TEM
TEM
SEM
TEM
TEM
TEM
Fiber
<1. 0 1
32.4
14. 7
9.3
7. 7
21.4
21.0
15. 2
13. 6
29. 0
20.6
15.4
24. 5
21.5
31.4
53.9
11.5
13.5
13.3
28. 1
15.6
15. 2
17.0
15. 2
17.6
14.5
19.0
18.6
16. 7
length
.0-2.5
50. 9
17.2
43. 2
48.0
22.6
22. 6
25. 1
18.4
31. 7
19. 7
20.4
21. 3
20.9
26. 1
39.5
13.6
13.5
13.5
15. 7
15.0
14.4
13.4
13.2
11. 8
11.9
14.6
12. 7
11.4
(Urn)
>2.5
41.3
6.4
9.0
18. 5
23.8
15.3
20. 2
14. 3
30. 3
24.0
19.1
18.1
16. 7
19.5
19.8
14. 5
11.1
16.6
17.6
14. 9
13.9
17. 1
16. 7
15.8
16.3
16.6
15. 3
7.0
Concentration
N/mm
124. 7
38. 3
61.5
74.1
67. 7
58.9
60.4
46.2
91.0
64.3
55.0
63. 9
59.1
76.9
113. 1
39. 6
38.1
30.0
61.4
45. 5
43.4
47.4
45.1
45.2
42. 5
50.2
46.6
35.1

*
 Filters A,  B,  and D were membrane filters; filter C was a fibrous filter.
                                   33

-------
filter is a variable in the measurement of asbestos fibers by either
microscope (exclusive of ashing for the TEM preparation).  However, the
membrane filter is more difficult to use for TEM and increases the proba-
bility of fiber loss.

      The effect of ashing was studied in a second  series of experiments on
sets 1 and 2 of Group I filters.  The data from this study (Table 6) show a
marked difference in both fiber concentration and fiber distribution as a
result of ashing, which was used in TEM sample preparation to remove
nonasbestost fibers from the filter.  The asbestos fiber concentrations from
the TEM examination are only 15  percent of those  obtained with the SEM.
It was concluded that ashing is responsible for a nominal 85-percent loss of
asbestos fibers.
      TABLE 6.  COMPARISON OF FIBER STATISTICS FOR FILTERS
                 PREPARED FOR SEM EXAMINATION AND BY
     	ASHING FOR TEM EXAMINATION	

                                      Fiber length (|J.m)     Concentration
                    *               	             2
Group Set Filter
I 1 A + D
A + D
B
C
2 A + D
A + D
B
C
Instrument
SEM
TEM
TEM
TEM
SEM
TEM
TEM
TEM
<1. 0
19.3
3.3
14.4
3. 5
10.8
2.0
2.8
4. 3
1.0-2.5
26.4
3.8
3.0
1.4
15.6
2. 6
3.0
7.0
>2. 5
36. 0
2. 0
0.9
1.2
29. 7
1. 1
1.1
3.6
N/mm
81. 7
5.8
18.4
6.0
56. 1
5.7
6.9
15.1

 Filters A, B, and D were membrane filters; filter C was a fibrous filter.


EFFECT OF FIBER CONCENTRATION

      The degree to which fiber-count accuracy is affected by fiber concen-
trations on filters can be seen in the length distributions plotted in Figures 9
through 11.  Group I results (Figure 9) were highly erratic,  complicated by
an accidental departure from the test plan.  Group II results (Figure 10)
Indicate reasonably good agreement between the distributions obtained from
1:he TEM analyses at 30, 000 X  and those obtained from the SEM analyses at
15,OOOX.  The SEM analyses at  6000 x show a lower proportion of short
fibers and a higher proportion of long fibers than the other two analyses.
Close examination of the SEM photographs revealed that the  difference
                                    34

-------
between the 6000 x and 15,000x magnifications was a result of the smaller
fibers being obscured by larger ones at the lower magnification.  At the
higher magnification,  it is possible to separate some of the fibers that would
otherwise appear as an agglomerated mass at lower magnifications.  The
average value of fiber concentration for all Group II filters was 70,300
fibers/mm^, which is within 12 percent of the intended value.  The standard
deviation from the  mean concentration for individual samples was  25 percent.

      Group III data (Figure 11) show good agreement among all three dis-
tributions, which indicates that, at this concentration,  there is no discernible
effect from either instrumentation or magnification.  The average fiber con-
centration from all Group El filters was 43, 200 fibers/mm^, which is within
8 percent of the intended value.  The standard deviation from the  mean
concentration for individual samples was 15 percent.

      In both Group II and Group III, the individual distributions of each set
are similar and are well represented by the average.  Moreover,  agreement
among sets provides a measure of their accuracy and reliability.   Group I
data, however,  did not show the agreement seen  in Groups II and  III.

      Group I distribution analyses (Figure 9) can be divided into  two parts.
The  first, composed of sets 1 and 2, had average fiber concentrations  of
68, 900 fibers/mm^, 54 percent from the desired value, and a 23-percent
standard deviation from the mean concentration for individual samples.  The
second part,  composed of set 3, had an average fiber concentration of
125, 700 fibers/mm^,  17 percent from the desired value, and a 25-
percent standard deviation from the mean concentration for individual sam-
ples.  The particle-size distribution from set 3,  with the highest  fiber con-
centrations, produced the most erratic  results.  The three distributions
from the same filters  differ in total concentration, and the individual fiber
length distributions also differ.  Figure 9 shows  that the SEM analyses at
15, 000 X magnification had the highest total concentration,  140, 400 fibers/
mm^, and a nearly uniform size distribution.  The SEM analyses at 6000 x
had an average concentration, 100,000 fibers/mm  , and were noticeably
deficient in the proportion of fibers less than 1 Jim in length.  The obscura-
tion  of the finer fibers by the larger ones, as discussed for Group II,
appears to be the primary reason for this observed distribution.  Moreover,
because of the high concentration of fibers on the filter, even larger fibers
were obscured at the lower magnification and were not  individually resolved;
thus, a lower total concentration is observed.  The average concentration
from the TEM analyses was only 53,000 fibers/mm^ and was deficient in all
size categories.  The  TEM results were caused by fiber pileup.   When the
carbon coating is applied to a heavily laden filter, the uppermost  fibers
shield underlying fibers from the coating.  Subsequently, when the filter is
dissolved, the previously protected fibers are flushed away and are not
available to the  TEM for  observation.
                                                                 ^j
      Although sets 1  and 2 of Group I had concentrations nearly equivalent
to those of Group II, the distributions of fiber lengths were sufficiently
different to justify giving individual consideration to the data from the two
                                    35

-------
groups.  In SEM analyses, there was good agreement between the fiber
length distributions and concentrations from sets 1 and 2 of Group I at both
magnifications.  Individual sets of Group II at a similar concentration also
exhibited good agreement.  With respect to fiber length distribution,  sets  1
and 2 of Group I were low in short fibers and high  in long fibers relative to
Group II.  Although fiber obscuration is likely at the lower magnification,  as
observed in Group II,  the appearance of the same distribution at higher
magnification indicates that the fiber length distributions were different than
Group II distributions because of deposition,  not obscuration.

      It was concluded that fiber concentration on filters affects both observed
concentration and distribution.  At concentrations  of 40,000 fibers/mm^,
reliable,  accurate measurements can be made with either the SEM or TEM
and at magnifications  of from 6000 X to 30, 000 x.  At concentrations of
80,000 fibers/mm^, the SEM analyses at  6000 x showed a loss of fine fibers,
although bot.a the SEM analyses at  15, 000 X and the TEM analyses at 30, 000 X
provided accurate results.  At concentrations greater than 80,000 fibers/
     ,  no reliable data were obtained.
EFFECT OF INSTRUMENT PERFORMANCE

      Samples prepared for and examined with the TEM were reexamined
with the  SE.M to determine if the SEM data differed from the TEM data,
exclusive of sample preparation techniques.  From photographs taken of
identical areas with both microscopes at 15,OOOX and 30, 000 X,  every iden-
tifiable fiber was individually noted,  as  shown in Figures 15 and 16.  Three
types of  discrepancies between matching pairs of photographs were consid-
ered: agglomeration, confusion, and absence.  These data are given in Table?.

      An agglomeration is a cluster of fibers lying so close together that
some cannot be distinguished.  Confusion arises when the microscope cannot
distinguish asbestos fibers from the background.  When a fiber  is not visible
in one instrument but can be seen with the other,  the term  "absence" is used.

      The overall performance of both instruments  was comparable in that
each microscope failed to observe 26 percent of the fibers  observed by the
other microscope.   Of the fibers observed by the TEM but  not by the SEM,
84 percent were not observed because of agglomeration.  Of the fibers
observed by the SEM but not by the TEM, 97 percent were  not observed
because  of confusion.  The TEM was better  able to  resolve individual fibers
in agglomerates because of its superior resolution; however, because the
TEM also views both sides of the carbon film that supports the asbestos,
fibers become confused -with background.  This is shown in Figures  15 and 16
for both  membrane and fibrous filters.

      Fiber absences in each microscope accounted for 3 percent  of the total
discrepancy between the instruments. Absences may be caused by extreme
agglomeration or confusion, by a loss (or gain) of fibers in handling between
the instruments,  or by insufficient resolving power.
                                    36

-------
                                                      (b)
  Figure  15.  TEM photomicrograph (a) and SEM photomicrograph (b)
              from Nuclepore filters.  Note confusion due to filter
              texture and ease of agglomerate separation in (a).
                (a)
(b)
Figure 16.  TEM photomicrograph (a) and SEM photomicrograph (b)
            from Millipore filters.   Again note confusion due to
            filter texture in (a).
                                  37

-------
                              TABLE 7.  SAMPLE AS OBSERVED BY SEM AND TEM
00


Total number of fibers seen
by either instrument
Total fibers seen by TEM
but not by SEM
Total iibers seen by SEM but
not by TEM
Agglomerations separated by
TEM but not by SEM
Agglomerations separated by
SEM but not by TEM
Fibers seen well in TEM but
confusing in SEM
Fibers seen well in SEM but
confusing in TEM
Fibers seen in TEM but not
seen in SEM
Fibers seen in SEM but not
Total
fibers
837

215
217
181
2
27

210

7

7

I
35

13
8
13
1
0

7

0

0
Group
II
349

95
85
72
0
16

82

7

3

m
454

107
124
96
1
11

119

0

4
Percent of
total fibers
100

26
26
22
0
3

25

1

1
Percent of
fibers lost
(each instrument)


100
100
84
1
12

97

3

3
          seen in TEM

-------
      As shown in Table 7, the distribution of fiber loss among the three
groups was proportional to the number of fibers counted in each group.
It is concluded, therefore,  that the concentration of fibers on the filter does
not contribute to fiber loss by any of the  three identified discrepancies; how-
ever, at high fiber concentrations, large loss occurs in TEM preparations.
No difference could be noted in comparisons  of the  relative loss at I5,000x
and 30, 000 X from both microscopes.  At both magnifications, fine fibers at
high filter  concentrations were affected equally.
                                    39

-------
                              REFERENCES

 1.    Thomson,  J.  G. ,  R. O. C. Kaschula,  and R.  R.  McDonald.  Asbestosis
      as a Modern Urban Hazard.  S. African Med.  J. ,  37:  77,  1963.

 2.    Thomson,  J.  G.   Asbestos and the Urban Dweller.  Ann. N.  Y.  Acad.
      Sci. , 132:   196, 1965.

 3.    Sullivan, R. J. , and Y. C. Athanassiadis.   Preliminary Air Pollution
      Survey of Asbestos. NAPCA No. APTD 62-27,  Public Health Service,  1969.

 4.    Selikoff, I. J. , J. Phurg, and E. C. Hammond.  The Occurrence of
      Asbe»tosis Among Insulation Workers  in the United States.  Ann. N.  Y.
      Acad.  Sci. , 132:  139,  1965.

 5.    Dree»sen,  W. C. , J. M. Dallavalle, T.  I. Edwards, J.  W. Miller,
      and R.  R.  Sayers.  A Study of Asbestosis in the Asbestos Textile
      Industry.  Public  Health Bulletin No. 241, Washington, D. C. , 1938.

 6.    Merewether,  E. R.  A.  , and C. W.  Price.   Report on  Effects of
      Asbestos Dust on  the Lungs and Dust Suppression in the Asbestos
      Industry.  H. M.  S. O. ,  London, 1930.

 7.    Houri.hane, D. O.   The Pathology of Mesotheliomata and an Analysis
      of Their Association With Asbestos  Exposure.   Thorax, 19: 268-278,
      1964.

 8.    Owen, W.  G. Mesothelial Tumors  and Exposure to Asbestos Dust.
      Ann.  N. Y. Acad. Sci., 132:   674,  1965.

 9.    Newhouse,  M. L.  , and H. Thompson.  Mesothelioma of Pleura and
      Peritoneum Following Exposure  to Asbestos in the  London Area.
      Brit. J. Ind.  Med., 22:  261,  1965.

10.    Lynch, K,  M. , and  W.  A. Smith.  Pulmonary Asbestosis:  Carcinoma
      of Lung in  Asbesto-Silicosis.   Am.  J.  Cancer, 24: 56, 1935.

11.    Knoxr  J. F. , R.  S.  Doll,  and  I.  D. Hill.  Cohort Analysis of Changes
      in Incidence of Bronchial Carcinoma in a Textile Factory.  Ann.  N. Y.
      Acad,  Sci., 132: 526, 1965.

12.    Isselbacher, K. J. , H. Klaus, and  H.  L. Hardy.  Asbestosis and
      Bronchogenic Carcinoma. J.  Am.  Med. Assoc. ,  15:  721, 1953.
                                   40

-------
13.    Noro, L.  Occupational and Non-Occupational Asbestosis in Finland.
      Am.  Ind. Hyg. Assoc.  J. , 29:  195-201, 1968.

14.    Wagner, J. C.  Current Opinions on the Asbestos Cancer Problem.
      Ann.  Occup. Hyg.,  15:  61-64, 1972.

15.    Selikoff, I. J. ,  E. C. Hammond, and J.  Churg. Asbestos Exposure,
      Smoking and Neoplasia. J. Am. Med. Assoc. , 204:  106,  1968.

16.    Harington,  J. S. , and F. J.  C. Roe.  Studies of Carcinogenesis of
      Asbestos Fibers and Their Natural Oils.  Ann.  N.  Y. Acad.  Sci. ,
      132:  439-450, 1965.

17.    Holt, P. F. ,  and D.  K. Young.  Asbestos Fibers in the Air of Towns.
      Atm. Environ. ,  7:  48, 1973.

18.    Scholl, E. L. Present  Threshold Limit Value in the U.  S.  A. for
      Asbestos Dust:  A Critique.  Ann.  N. Y. Acad. Sci., 132:  316, 1965.

19.    Merliss, R. R. Talc-Treated  Rice and Japanese Stomach Cancer.
      Sci. , 174:  1141,  1971.

20.    News release* U. S.  Environmental Protection Agency,  Washington,
      D.  C.,  1974.

21.    Nicholson,  W. J. , C. J.  Maggiore,  and I. J. Selikoff.  Asbestos
      Contamination of Parenteral Drugs.  Sci.,  177:  171, 1972.

22.    Hills, D. W.  Economics of Dust Control.  Ann. N.  Y.  Acad. Sci. ,
      132:  322-337, 1965.

23.    Whitby, K. T. , R. E.  Chalson,  W. E. Wilson, and R.  K.  Stevens.
      The Size of Suspended Particle Matter in Air.  Sci. ,  183:  1098,  1974.

24.    Selikoff, I. J.  Medical Aspects  of Asbestos  and Asbestos-Related
      Diseases.   Symposium: Am. Ind.  Hyg.  Assoc.  , Anaheim, California,
      1972.

25.    Edwards,  G.  H. , and J. R.  Lynch.  The Method Used by the U.S.
      Public  Health Service for Enumeration of Asbestos  Dust on Membrane
      Filters.  Am.  Occup. Hyg., 2:  1, 1968; 11: 1, 1968.
                   %
26.    Zumwalde, R. , S. Bayer, and N.  Leidel.  Sampling and Evaluation
      of Airborne Asbestos Dust.  NIOSH No. 84,  National Institute for
      Occupational Safety and Health,  Cincinnati,  Ohio, 1974.

27.    Asbestos — The Need for and Feasibility of Air  Pollution Controls.
      National Academy of Sciences, Washington, D. C. ,  1971.

28.    Wagner, J. C. ,  and J. W. Skidmore.  Asbestos Dust Deposition and
      Retention in Rats. Ann. N.  Y. Acad. Sci. ,  132:  77, 1965.
                                    41

-------
29.   Roach, S. A. , Measurement of Airborne Asbestos Dust by
      Instruments Measuring Different Parameters.  Ann.  N. Y. Acad.
      Sci. ,  132:  306,  1965.

30.   Addingley,  C. G.  Dust Measurement and Monitoring in the Asbestos
      Industry.  Ann. N.  Y.  Acad. Sci., 132:  298, 1965.

31.   Timbrell, V.  The Inhalation of Fibrous Dusts.   Ann.  N.  Y. Acad.
      Sci. ,  132:  255,  1965.

32.   Holt,  P.  F. , J.  Mills, and D. K.  Young, Experimental Asbestosis
      With Four Types of Fibers:  Importance  of Small Particles.  Ann.
      N. Y. Acad.  Sci., 132:  87,  1965.

33.   Thompson, R. J. , and G. B. Morgan.  Determination of  Asbestos
      in Ambient Air.  In: Proceedings of International Symposium on
      Identification and Measurement of Environmental Pollutants, Ottawa,
      Canada,  1971.

34.   Pattnaik, A. , and J. D.  Meakin.  Development of an Instrumental
      Monitoring Method for Measurement of Asbestos Concentrations in or
      Near Sources.  EPA-6501 2-73-016, U.S.  Environmental Protection
      Agency, Washington, D. C. ,  1973. 40pp.

35.   Craig,  D. K. , A. P. Wehner, and W.  G. Morrow.  The Generation
      and Characterization of a Respirable Aerosol of Chrysotile Asbestos
      for Chronic Inhalation Studies.  Am. Ind. Hyg. Assoc.  J. , 33:  283,
      1972.

36.   Lynch, J. R. , H. E. Ayer,  and D. L. Johnson.  The Interrelationships
      of Selected Asbestos Exposure Indices.   Am. Ind. Hyg. Assoc.  J. ,
      31:  598,  1970.

37.   Hamilton, R. J. , and W. A.  Walton.  The Selective Sampling of
      Respirable Dust.  In:  Inhaled Particles and Vapours,
      C. N. Davies, ed. , Pergamon Press,  London, 1961. pp.   465-475.

38.   Lippmann, M. , and W. B. Harris. Size-Selective Samplers for
      Estimating "Respirable" Dust Concentrations, Health Phys., 8:  155,
      1962.
                                   42

-------
                               APPENDIX A

                SIZE DISTRIBUTIONS OF ASBESTOS DUSTS
      The size distribution of asbestos dusts has been measured by several
investigators.  Although similarities exist among the distributions, each is
strongly dependent on the method of measurement.   In many cases, an opti-
cal microscope was used;  consequently, the distribution did not include the
finer fibers not resolved with this instrument.  Other instruments  used were
also insensitive to the finer sizes, and accurate correlation between studies
could not be made.

      A distribution of the lengths of airborne asbestos fibers of chrysotile,
amosite, and crocidolite was measured by Wagner and Skidmore with a
thermal precipitator (28).   For each of the minerals, 90 percent of the fibers
had particle lengths less than 7 (o.m.  At greater fiber lengths, the  proportion
of long fibers was greater for chrysotile than for the amphibole minerals.
Less than 0. 1 percent of the amphibole fibers were longer than 30  |j.m, and
about 0. 01 percent were longer than 75 |im.  For chrysotile fibers, 0. 1 per-
cent were longer than 50 |am, and about 0. 01 percent were longer than 100
      Roach (29) used a thermal precipitator and oil immersion objective
lens to measure the length and diameter distributions of amosite fibers in the
vicinity of a bagging operation.   He found that 85 percent of the visible par-
ticles were shorter than 1 p.m and only 5 percent were longer than 5 (J.m.
When the  same particles were sized by diameter, 94 percent had  diameters
less than  0. 5 (im; only 0. 6 percent were thicker than 1 fim.  The results of
Wagner and Skidmore and Roach are essentially the same, inasmuch as
Roach may have measured a slightly finer portion of the airborne
distribution.

      Addingley (30) measured chrysotile fibers in the workrooms of weaving,
doubling,  and carding operations, using a Royco  particle counter  at equivalent
sphere diameters between 0. 3 and 10 |j.m in 15 separate  size ranges.  The
distributions for the weaving and doubling operations were nearly identical.
About 50 percent of the fibers were in the 0. 5- to 0. 7-(j.m range; less than
2 percent were greater than 1 |j,m.  Below 0. 5 |J.m, a decrease in  frequency
is observed that may correctly define the distribution or may instead be a
consequence of a loss of instrument sensitivity at the finer sizes.  Because
the Royco particle counter measures an equivalent fiber diameter based on
the cross section of a fiber, direct correlation is not possible.
                                    43

-------
      A maximum in the particle  size distribution of chrysotile fibers was
observed by Craig et al.  (35), who used an Anderson Cascade Impactor and
mass calculations, and by Lynch  et al. (36), who worked with a transmis-
sion electron microscope and a sample collected on a membrane filter.
Craig found that the aerodynamic mass mean diameter  lies between 1 and
3 (am and that 5 percent of the fibers had equivalent diameters  greater than
7 (Jim for three samples measured.  Lynch measured asbestos  fibers in the
textile, friction,  and pipe industry.  Median fiber lengths varied from
1. 4 fim for the carding operation  to 0. 7 [im for the pipe finishing operation.
The  percentage of fibers  longer than 5 |o,m ranged from 4 percent in the
carding operation to  1 percent in  the pipe finishing operation.  For all other
operations monitored, the median fiber length was 0. 9  |J.rn, with 2  percent
of the fibers Longer than 5 (Jim.

      If the size distributions of the studies are correlated  (giving  recognition
to the many qualification  required), a log-normal distribution of asbestos
(chrysotile) fiber size might be defined:  a maximum number of fibers in the
distribution a.re between 0. 5  and  1 (am in length; less than 5 percent are
more than 5 |im in length; and,  if a reasonably symmetric distribution is
assumed, about 5 percent are less than 0. 1 |j.m in length.   The works of
Addingley,  Craig et al. and Lynch et al are in reasonable agreement with
this  correlation,  as are those of Roach and Wagner and Skidmore if account
is made  of the  fine  fibers not visible by optical microscopy methods.

      Equally important in the evaluation  of measuring  devices is the  relation-
ship of asbestos size to its deposition in the respiratory system.   The deposi-
tion  of inhaled airborne dust  of different  sizes has been evaluated by several
investigators (29,  37, 38). The remarkable agreement  in the results indicates
that  the maximum aerodynamical equivalent diameter (AED) is about 10 (j.m.
Thus, if the factor of 1/3 is used to define the diameter of asbestos fibers
relative to the  AED,  3. 5 (j.m appears to be the expected upper limit of long
asbestos fibers that can reach pulmonary air spaces (31).  The coarser par-
ticles are deposited in the ciliated portion of the respiratory tract —down to the
terminal bronchioles, removed by ciliary action,  and eventually either swal-
lowed or spat out (29).  Observations on the histological distribution of the
dusts in the lungs of  rats  have shown that dusts tend to  accumulate in the alveoli
arising from the respiratory  bronchioles  (28).  An evaluation of fiber deposi-
tion  by Timbrell (31) suggests that sedimentation and inertial precipitation
operate throughout the respiratory system to deposit particles in the nose, at
the bifurcation of the respiratory tract, and on the walls  of the bronchi and
bronchioles.  The distribution of  particles that proceed to the pulmonary air
spaces is composed primarily of  the finer particles. However, many of these
finer particles are  eliminated from the respiratory system without being de-
posited,  presumably because of their ability to follow the laminar air stream
within the respiratory air passages. The deposition and  retention  study con-
ducted on rats  by Wagner and Skidmore (28) revealed that the elimination rate
of chrysotile is three times greater than that of amosite and crocidolite, which
suggests that the reduced fibrigenicity of  chrysotile is the consequence of the
greater elimination of this mineral.  The  reason for the difference in elimina-
tion  is not understood, but it  is proposed here that the finer size distribution
of chrysotile fibers relative to amphibole  fibers is responsible for this behavior.
                                    44

-------
                               APPENDIX B

                           DATA CORRECTIONS
      The distributions of fiber lengths used here have been corrected for
finite frame size.  When photographs are taken of a sample, some fibers
will inevitably extend beyond the edge of the frame.  The lengths of fibers
only partially in the field are not known, and the  problem becomes more
severe as the magnification is increased.  If the  magnification were so great
that any fiber would overlap an edge, count statistics per frame would be
very poor, and no information about fiber lengths would be obtainable.  In
the other extreme, if the magnification were so low that the longest fibers
were very much smaller than the frame  size, no corrections would be
needed, although resolving power would  be adversely affected.  In order to
have both sufficient resolving power and meaningful length distribution data,
the data have  been corrected to eliminate the effects of finite frame sizes.
This was accomplished by first assuming that, on the  average,  each fiber
extending beyond the edge of the frame was halfway  in and halfway beyond the
frame.  Statistics were  then collected on the number and length of fibers per
frame entirely within the field and on the number and length of fibers only
partially within the frame.   The distributions corrected by the above assump-
tions were combined with the distribution for totally observed fibers; in this
way, distributions  for all fibers, even those partially  observed, were
obtained.  An effective area of the photographic frame was then corrected to
include all fibers,  including those that extended beyond the frame.  In calcu-
lation of the extended area,  it was assumed that the fiber concentration for
totally observed fibers using less than the total frame area was identical to
the fiber concentration for all fibers.  These corrections converted data at
all magnifications to a common base.  (Methods of counting only totally
observed fibers do not provide correction for different magnifications. )

      When the above corrections of data were made,  it was determined that
the actual magnification of any individual photograph differed by as much as
10 percent from the nominal magnification of that series.  Thus,  super-
imposed on the data is an uncertainty in  magnification that increases the.
probable error of each individual measurement.  Unquestionably,  this uncer-
tainty is  responsible for the displacement of the chi-squared plot in Figure 3
and increases the standard deviation calculated from results of individual
photographs.
                                     45

-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-77-059
2.
4. TITLE AND SUBTITLE
Evaluation of Electron Microscopy for Process
Control in the Asbestos Industry
7. AUTHOR(S)
R. M. Gerber and R. C. Rossi
9. PERFORMING ORGANIZATION NAME AIN
The Aerospace Corporation
P.O. Box 92957
Los Angeles , California 90
ID ADDRESS
009
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
3. RECIPIENT'S ACCESSION-NO.
5. REPORT DATE
February 1977
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
ATR-77(7552)-l
10. PROGRAM ELEMENT NO.
1AB015; ROAP 21AFA-011
11. CONTRACT/GRANT NO.
R802394
13. TYPE OF REPORT AND PERIOD COVERED
Final; 1/74-12/76
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES j.ERL-RTP Project Officer for this report is D. B.
Drop 62, 919/549-8411, Ext 2557.
Harris, Mail
16. ABSTRACT rp^ report gives results of an evaluation of the transmission electron micro-
scope (TEM) and the scanning electron microscope (SEM) as potential tools for fine
particle asbestos fiber counting for process control in the asbestos industry. The
study defined the capabilities and limitations of the instruments in applications where
asbestos specificity is not necessarily required, and where analysis cost must be
minimal. The study showed that the microscopes are equally capable of counting all
fibers in the full particle size distribution; but, for reasons of agglomeration and
confusion with the filter texture, each microscope can observe only 75% of the distri-
bution. In contrast, present standard light microscopy methods observe only the
coarser 10% of the distribution, without resolving the fine fibers. Optimum asbestos
fiber counting was done at 15,000 times magnification and at fiber concentrations on
the filter between 40,000 and 80,000 fibers per sq mm. The minimum number of
fibers counted to obtain high statistical confidence was 200 fibers per datum point.
Standard techniques for filter sample preparation were found to have no effect for
either instrument. Ashing of filters to remove non-asbestos fibers was responsible
for 85% asbestos fiber loss.
17.
a. DESCRIPTORS
KEY WORDS AND DOCUMENT ANALYSIS
b.lDENTIFIERS/OPEN ENDED TERMS
Air Pollution Microscopes Air Pollution Control
Industrial Processes Electron Micro- Stationary Sources
Asbestos scopes Fine Particulate
Fibers Transmission Standard Light
Measurement Scanning
Particles
18. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
20. SECURITY CLASS (This page)
Unclassified

c. COSATI Field/Croup
13 B
13H
HE
14B
21. NO. OF PAGES
55
22. PRICE
EPA Form 2220-1 (9-73)
                                                             47

-------