PB83-261651
Development of Improved Analytical
Techniques for Determination of
Asbestos  in Water  Samples
Ontario Research Foundation, Mississauga
Prepared  for

Environmental Research Lab., Athens, GA
 Sep 83
                     U.S. DEPARTMENT OF COMMERCE
                   National Technical Information Service

-------
                                         PB83-261651
                                        EPA-600/4-83-042
                                        September 1983
DEVELOPMENT OF IMPROVED ANALYTICAL TECHNIQUES

        FOR DETERMINATION  OF  ASBESTOS

              IN WATER SAMPLES
                     by
   E.J. Chatfield, M.J.  Dillon,  W.R.  Stott
         Electron Optical  Laboratory
        Department of  Applied Physics
         Ontario Research  Foundation
      Sheridan Park  Research  Community
    Mississauga, Ontario,  Canada L5K  1B3
             Contract 68-03-2717
               Project  Officer
              J. MacArthur Long
         Analytical  Chemistry Branch
      Environmental  Research  Laboratory
               Athens,  Georgia
                     30613
     ENVIRONMENTAL  RESEARCH LABORATORY ~
     OFFICE OF  RESEARCH  AND DEVELOPMENT
    U.S. ENVIRONMENTAL PROTECTION AGENCY
            ATHENS, GEORGIA 30613
    	  	 HPRODUCEinr
                NATIONAL TECHNICAL
               INFORMATION SERVICE
    -.--—--•      US. DEPARTMENT OF COMMERCE
    -_	 .      SPBIKGFIEIK. VA. 22161
                                                    U.S- ^"ft y[?||jtj Mill illII III II1UII Hlll|l|Il||ll II]

-------
                                  TECHNICAL REPORT DATA
                           (Please read Instructions on the reverse before completing)
  REPORT NO.
  EPA-600/4-83-042
4. TITLE AND SUBTITLE
   Development of Improved Analytical  Techniques
   for Determination of Asbestos  in Water Samples
             3. BECtflENT'S ACCESSIOi*NO.
              PB83-261651.
             5. REPORT DATE

              September  1983
             6. PERFORMING ORGANIZATION CODE
  AUTHOR(S)
   E.J.  Chatfield, M.J. Dillon and W.R.  Stott
                                                          8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
   Department of Applied Physics
   Ontario Research Foundation
   Sheridan Park Research  Community
   Mississauga, Ontario, Canada L5K  1B3
             10. PROGRAM ELEMENT NO.
                CBNC1A
             11. CONTRACT/GRANT NO.
                68-03-2717
12. SPONSORING AGENCY NAME AND ADDRESS
  Environmental Research Laboratory—Athens GA
  Office of Research and Development
  U.S. Environmental Protection  Agency
  Athens, Georgia  30613
             13. TYPE OF REPORT AND PERIOD COVERED
                Final, 10/78-9/81
             14. SPONSORING AGENCY CODE
                EPA/600/01
15. SUPPLEMENTARY NOTES
16. ABSTRACT           •                                  .
         Sample preparation techniques were-examined for the analysis  of asbestos
   fiber concentrations in water.  The carbon-coated Nuclepore technique using a
   polycarbonate filter proved  to  be superior, to either the "drop"  or  the collapsed
   membrane filter technique.   Compared with plasma ashing, ozone-ultraviolet light
   oxidation of water samples was  found to be a simpler and superior technique for
   removal of organic materials.   Experiments revealed that large proportions of the
   suspended asbestos fibers could become attached to the inside surfaces of sample
   containers.  This effect was caused by trace organic materials of bacterial origin.
   Ozone oxidation, carried out inside the collection container, released the attached
   fibers into the water again.-.  Initi&l experiments were carried out  to determine the
   effectiveness of the attachment phenomenon as a fiber separation technique.  Experi-
   ments into the nature of the scavenging effect of bacteria on container surfaces
   led to the development of stable reference dispersions of asbestos  fibers.  If
   bacteria and their products  were excluded initially, and if absolute sterility was
   maintained thereafter, suspensions of both chrysotile and crocidolite appeared to
   be stable for long periods of  time..  Tests of reference suspensions in sealed glass
   ampoules stored for almost two  years produced fiber concentration values statistical
   ly compatible with those obtained at the time of sample preparation.
17.
                               KEY WORDS AND DOCUMENT ANALYSIS
                  DESCRIPTORS
                                             b.IDENTIFIERS/OPEN ENDED TERMS
                          c.  COSATI Field/Group -i_
18. DISTRIBUTION STATEMENT
   RELEASE TO PUBLIC
                                              19. SECURITY CLASS (ThisReport)
                                                UNCLASSIFIED
                                                                        21. NO. OF PAGES
                              147
20. SECURITY CLASS (This page)
  UNCLASSIFIED
                          22. PRICE
EPA Form 2220-1 (9-73)

-------
                  NOTICE






THIS  DOCUMENT  HAS BEEN'REPRODUCED.



FROM THE  BEST  COPY FURNISHED US  BY




THE  SPONSORING AGENCY.  ALTHOUGH  IT



IS RECOGNIZED  THAT CERTAIN  PORTIONS




.ARE  ILLEGIBLE,  IT IS BEING RELEASED'



IN THE. INTEREST OF MAKING AVAILABLE




AS  MUCH INFORMATION AS  POSSIBLE.
                     l-o,

-------
                                 DISCLAIMER
The information in this document has been funded wholly or in part by the
United States Environmental Protection Agency under Contract No.  68-03-
2717 to Ontario' Research Foundation.  It has been subject to the  Agency's
peer and administrative review, and it has been approved for publication
as an EPA document.  Mention of trade names or commercial products does
not constitute endorsement or recommendation for use.
                                     T.T

-------
                                 FOREWORD

Nearly every phase of environmental  protection depends on a capability to
identify and measure specific pollutants in the environment.  As part of
this Laboratory's research on the occurrence, movement, transformation,
impact, and control of environmental contaminants, the Analytical  Chemistry
Branch develops and assesses new. techniques for identifying and measuring
chemical constituents of water and soil.

A 3-year study was conducted to develop improvements in the analytical
method for determination of asbestos fiber concentrations in water samples.
The research produced an improved sample preparation and analysis method-
ology, a rapid screening technique to reduce analysis cost, and a new
reference analytical method for asbestos in water.  The analytical method
for determining asbestos fibers in water is perceived as representing
current state-of-the-art.                             .

                                      William T. Donaldson
                                      Acting Director
                                      Environmental- Research Laboratory
                                      Athens, Georgia       •
                                   iii

-------
                                   PREFACE

The Preliminary Interim Method for Determining Asbestos in Water was issued
by the U.S. Environmental Protection Agency's Environmental Research
Laboratory in Athens, Georgia.  The method was based on filtration of the
water sample through a sub-micrometer pore size membrane filter, followed by
-preparation of the filter for direct examination and counting of the fibers
in a transmission electron microscope.  Two alternative techniques were
specified:  one in which a cellulose ester filter was prepared by dissolution
in a condensation washer; and another known as the carbon-coated NucleporeR
technique which used a polycarbonate filter.  In January 1980 the method was
revised (EPA-600/4-80-005) to eliminate the condensation washer approach, and
a suggested statistical treatment of the fiber count data was incorporated.

The research program described here is directed towards refinement of the
revised interim method.  It was recognized that clearly-defined methods for
fiber identification were required, in addition to further standardization of
the format for reporting of analytical results.  Major improvements were also
required in the method for removal of interfering organic materials.   The use
of ozone-ultraviolet light oxidation prior to filtration of the water sample
was investigated and found to be superior to any other technique.   This
oxidation technique also eliminated the sometimes serious scavenging of fibers
by sample container walls, an effect which had not been systematically investi-
gated.  Reference dispersions of asbestos fibers which are stable for long
periods of time were also developed.  These will now permit known analytical
quality assurance samples to be submitted for analysis along with field
samples.  The lack of standard samples was one of the major limitations of
asbestos fiber analyses.

The research program has allowed development of an analytical  method which
incorporates the best available technology.  This method should be used when
the most reliable analytical data are required.
                                     iv

-------
                                  ABSTRACT

A research program directed towards  improvement of the  analytical method  for
determination of asbestos fiber concentrations  in  water samples was  carried
out.  Sample preparation techniques  involving evaporation  of  small drops  on
transmission electron microscope (TEM)  grids were  found to yield  samples  with
uneven deposits of particulate, and  were thus unsuitable for  quantitative
determinations.  The collapsed membrane filter  technique was  found to  incur
significant size-dependent fiber losses, which  became more serious with
increasing"filter pore size.

Using naturally-occurring dispersions of both chrysotile and  amphibole fibers,
it was demonstrated that intra-laboratory analyses made by the same  operator
using the same instrument were statistically compatible at the 5% level.

Ozone-ultraviolet light oxidation of water samples was  found  to be a simpler
and superior technique for removal of organic materials than  plasma  ashing.
Experiments indicated that unknown and  sometimes very large proportions of
the suspended fibers may become attached to the inside  surfaces of the con-
tainer.  It was found that this effect  was a consequence of trace organic
materials of bacterial origin, and that ozone-UV oxidation, carried  out inside
the collection container, released the  attached fibers  into the water  again.
Initial experiments were carried out to determine  the effectiveness  of the
attachment phenomenon as a fiber separation technique.

The experiments into the nature of the  scavenging  of fibers by container  sur-
faces led to the development of stable  reference dispersions  of asbestos
fibers.  It was found that if bacteria  and their products  were excluded
initially, and if absolute sterility was maintained thereafter, suspensions
of both chrysotile and crocidolite appeared to  be  stable for  very long periods
of time.  Reference suspensions in sealed glass ampoules have now been stored
for almost 2 years, and they still yield, fiber  concentration  values  statisti-
cally compatible with those originally  obtained at the  time of preparation.

A fiber classification system was developed which  recognizes  instrumental
limitations, and if required, permits later re-evaluation  of  the  raw data
using different fiber identification criteria.   A computer program was written
which permits fiber identification on the basis of energy  dispersive X-ray
spectra and zone axis selected area  electron diffraction patterns.   The
identification procedure operates by selection  of minerals which  are con-
sistent with the measurements, using a  library  of data  from 226 minerals.  A
standardized format for reporting of fiber counting data was  also established,
and a computer program written to facilitate this.

A reference analytical method was written (published separately), which

-------
incorporates ozone-UV oxidation,  fiber identification  and  classification, and
a standardized reporting format.

A rapid screening technique was  developed which  can  be used  to eliminate from
further study those water samples which have  low fiber concentrations.  This
technique (published separately)  is  based on  the alignment properties of
asbestos fibers in magnetic fields.

This research program was performed  under contract 68-03-2717 under sponsor-
ship of the U.S. Environmental  Protection Agency.  This  report covers a
period from October 1978 to September 1981 and the work  was  completed as of
September 1981.
                                     vi

-------
                                  CONTENTS
Foreword		     111
Preface	      1 v
Abstract 			,		       v
Figures	      1x
Tables	      xi
Acknowledgment	    xi 11

   1.  INTRODUCTION	       1

   2.  CONCLUSIONS AND RECOMMENDATIONS.			       6

   3.  INVESTIGATION OF TEM SAMPLE PREPARATION TECHNIQUES	       9

       3.1  Drop Method	      10
       3.2  Collapsed Membrane Filter Method...	      11
       3.3  Summary of Alternative Sample Preparation Methods  	      17

   4.  REPROMCIBILITY OF ENVIRONMENTAL WATER.SAMPLE.'ANALYSES  ..........      18

       4.1  General 	      18
       4.2  Sample Collection	      20
       4.3  Analytical Reproducibility  	      21

   5.  OXIDATION OF ORGANIC MATERIALS  ;. 4	      23

       5.1  The Ozqne-UV Oxidation Method	      23
       5.2  High Pressure Oxidation	      26

   6.  EXPERIMENTS ON THE STABILITY  OF-  AQUEOUS FIBER DISPERSIONS  	      31

       6.1  Initial Observations	      31
       6.2  Effects of Biological Organisms on Suspension  Stability ..      37
       6.3  Preparation and Preservation of Reference
            Fiber Dispersions	      39
       6.4. Container and Storage Effects on  Fiber  Dispersions........      40
       6.5  Investigation of Fiber Scavenging Mechanism 	      48

   7.  FIBER  IDENTIFICATION PROCEDURES	      56
       7.1  General Considerations	      56
       7.2  Fiber Identification Techniques 	.-..	      57
            7.2.1  SAED Technique 		      57
            7.2.2  EDXA Technique	      58
            7.2.3  Optimum Fiber  Identification  Procedure  	      61
       7.3  Instrumental Limitations	      61

                                     vii

-------
      7.4   Specification of Adequate Instrumental Performance
            for EDXA Measurement 	     62
      7.5   Analysis of Fiber Identification Data 	     63
      7.6   Fiber Classification Categories 	     67
            7.6.1  Classification of Fibers With Tubular
                   Morphology, Suspected to be Chrysotile 	     73
            7.6.2  Classification of Fibers Without Tubular
                   Morphology, Suspected to be Amphibole	     76
            7.6.3  Reporting of Fiber Classifications	 •    80

 8.   DETERMINATION OF FIBROSITY 	     81

 9.   STATISTICS OF FIBER COUNTING 	     84

      9.1   Test for Uniformity of Fiber Deposit
            on Electron Microscope Grids 	     84
      9.2   Calculation of the Mean and Confidence
            Interval of the Fiber Concentration	     84

10.   DEVELOPMENT OF A REFERENCE METHOD FOR DETERMINATION
     OF ASBESTOS FIBER CONCENTRATIONS IN WATER 	     89

     10.1   Use of Ozone-Ultraviolet Light Treatment for
            Oxidation of Interfering Organic Materials  	     89
     10.2   Use of Ultrasonic Treatment	     90
     10.3   Use of the Condensation Washer for Removal  of
            Residual Undissolved Plastic From TEM Samples 	     90
     10.4   Introduction of Minimum Fiber Length to be  Reported ....     94
     10.5   Requirement for Quantitative Interpretation of
            Chrysoti 1 e SAED Patterns	 \..:.'.'.'	     94
     10.6   Requirement for Energy Dispersive X-Ray Analysis
            and Zone Axis SAED for Amphibole Identification 	     94
     10.7   Introduction of a Fiber Classification System 	     95
     10.8   Standardized Reporting Format	     95
     10.9   Introduction of Fibrosity Index 	     95
     10.10  Statistics of Fiber Counting 	     95
     10.11  Deletion of "Field of View" Fiber Counting	.".     96

     REFERENCES	     97

     APPENDIX A: COLLAPSED MEMBRANE STUDY:
                 DETAILED ANALYTICAL DATA  	    102

     APPENDIX B: REPLICATED ANALYSES OF ENVIRONMENTAL WATER SAMPLES:
                 RESULTS AND STATISTICAL ANALYSES	    110

     APPENDIX C: CONTAINER AND STORAGE STUDY:
                 FIBER LENGTH DISTRIBUTIONS	'......,	    127

     APPENDIX D: CONDENSATION WASHER STUDY:
                 DETAILED ANALYTICAL DATA 	    131
                                  viii

-------
                                   . FIGURES..
Number                 '             .         .                             Page

i   1.   Principal varieties of asbestos	       2

'  - 2  - Steps in the collapsed membrane  procedure
       of filter preparation	      11

;   3   Collapsed membrane filter method:
       Chrysotile transfer efficiencies  	      15

"•• 4" """Collapsed membrane filter "method:
       UICC.crocidolite transfer efficiencies	      16

   5   SEM micrograph of the deposit on  a Nuclepore filter
       produced, by- filtration of 20 ml  of lake water after
       the' sample was shaken	      19

   6   SEM micrograph of the deposit on  a Nuclepore filter
       produced by filtration of 20 ml  of. lake water after
       ultrasonic treatment at 2.2 W/L  for  30 minutes	      19

   7   Equipment used for the ozone-UV  oxidation  technique  	      24
                                       !
   8   Ozone-UV oxidation of water samples  in glass bottles	      25

   9   High pressure oxidation equipment	      29

   10   Union Carbide Calidria Chrysotile before high pressure
       oxidation treatment	      30

   11   Union Carbide. Calidria Chrysotile after high pressure
       oxidation treatment	      30

   12   Results  of inter-laboratory analyses using aqueous
       dispersion of Union Carbide Chrysotile  fibers
       (high concentration)	       33

   13   Results  of inter-laboratory analyses using aqueous
       dispersion of Union Carbide Chrysotile  fibers
       (low concentration)	....	       33
                                      ix

-------
14   Results of inter-laboratory analyses using a naturally-
     occurring aqueous dispersion of chrysotile fibers
     (Lloyd Lake, Ontario) 	     33

15   SEM micrographs showing asbestos fibers attached by
16   organic material to the inside surface of a glass
     contai ner	     34
17   Descriptive flowchart of computer program to
     match X-ray spectrum with stored mineral data	     65

18   Computer program output obtained from XMATCH program for
     input of quantitative EDXA elements from a fiber of
     ri ebecki te	     66

19   Two zone axis patterns obtained by tilting a single
     fiber about its axis 	...	     68

20   Solutions obtained from analysis of zone axis
     pattern 34 shown in Figure 19 ..	     69

21   Solutions obtained from analysis of zone axis.
     pattern 41 shown in Figure 19	     70

22   Only solutions remaining after permuting solutions from
     both patterns 34 and 41 and calculating inter-zone
     axis angles	'	     71

23   Classification chart for fiber with tubular morphology 	     74

24A  TEM micrograph of chrysotile fibril, showing morphology 	     75

24B  TEM micrograph of UICC Canadian chrysotile fiber after
     thermal degradation by electron beam irradiation	     75

25   SAED pattern of chrysotile fiber with diagnostic features
     1 abel 1 ed	,	     75

26   Amphibole SAED pattern (crocidolite) obtained from a
     fiber without precise orientation onto a zone axis 	     77

27   Classification chart for fiber without tubular morphology 	     78

28   Aspect ratio distribution for an aqueous dispersion
     of UICC amosite	     82

29   TEM specimen prepared from aqueous suspension of asbestos
     and other materials.  Polycarbonate filter dissolved using
     chloroform in Jaffe Washer only 	     92

30   TEM specimen shown in Figure 29, but after further treatment
     using chloroform in condensation washer for 180 minutes 	     93

-------
                                    TABLES





Number                                    -                              Page
1

2
3
4

5

6

7
8

9

10

11

12
13
14
15
16
Collapsed Membrane Study for Chrysotile:
Initial Measurements 	 	 	 	
Col 1 apsed Membrane Study for Chrysoti 1 e 	 :
Collapsed Membrane Study for Crocidolite 	 —
Chrysotile Fiber Concentrations in Municipal Drinking Water,
Sherbrooke , Quebec 	 	 	 	
Replicate Analyses of Environmental Water Samples:
Summary of Results 	 	
Results of Analyses of Chrysotile Dispersions
in 0.1% Starch Solution 	 	 .- 	
Stability of Very Dilute Chrysotile Fiber Dispersions 	 	
Shaking of Chrysotile Fiber Dispersions Under Selected
Ionic and pH Conditions 	 	
Stability and. Storage of Sterile Union Carbide
Chrysotile Fiber Dispersions 	 	 	 	
Stability and Storage of Sterile UICC Chrysotile
Fiber Di spersi ons 	 	 — 	
Stability and Storage of Sterile UICC Crocidolite
Fiber Di spersi ons — 	 	 	
Storage and Container Study: Union Carbide Chrysotile 	
Storage and Container Study: UICC Chrysotile 	
Storage and Container Study: UICC Crocidolite 	
Storage and Container Study: Sherbrooke Water (1st Series) .
Storage and Container Study: Sherbrooke Water (2nd Series) .

13
15
16

20

22

27
35

36

38

38

39
42
43
45
46
47
                                     xi

-------
                                           -  . .        .   ..        .     .  ,
17   Storage and Container Study:  Sherbrooke Water (3rd Series) .     49
18   Initial Fiber Separation Experiments Using
     Union Carbide Chrysotile	      51
19   Separation Experiments Using Union Carbide Chrysotile  	      53
20   Separation Experiments Using UICC Crocidolite 	.-.	      54
21   Separation Experiments Using UICC Amosite	      55
22   "K" Values	      60
23   Silicate Mineral Standards for Calibration of TEM-EDXA
     System 	      59
24   An Example of a Single Mineral Entry in the Mineral
     Library File 	      63
25   Classification of Fibers With Tubular Morphology	      72
26   Classification of Fibers Without Tubular Morphology 	      72
27   Levels of Analysis for Amphibole 	      79
28   Indices of Fibrosity for Selected Minerals ..•	      83
29   95% Confidence Limits for the Poisson Distribution 	      86
30   Condensation Washer Study:  Chrysotile Fiber Loss Analysis .      91
31   Condensation Washer Study:  Crocidolite Fiber
     Loss Analysis	      91
                                   xii

-------
                               ACKNOWLEDGMENTS


The authors wish to acknowledge their appreciation to Mr. L. Doehler and
Mrs. A. Liebert for their invaluable contributions to the project.  They also
appreciate the helpful technical discussions with Mr. O.K. Smith and Dr. C.H.
Anderson.  The authors also wish to thank Dr. G. Plant of the Geological
Survey of .Canada, and Drs. F.J. Wicks and R.I. Gait of the Royal Ontario
Museum for helpful discussio.ns relating to mineral identification and for
supply of the many mineral samples.

-------
                            	SECTION 1

                                INTRODUCTION


This research program had two primary objectives:

     (a)  to develop a rapid screening.technique which can be used to elimi-
          nate from further study those  .water samples which have low fiber
          concentrations; and

     (b)  to develop the analytical procedure for  determination of asbestos
          fiber concentrations in water to the point that a refereed method
  •  "     can be defined.     	"  "~ ""     ""  	~

The rapid screening technique development was based on the alignment property
of asbestos fibers in magnetic fields, and the results of these investigations
are reported separately.*

Analytical techniques for determination  of asbestos .fiber concentrations in
water samples have been generally based on transmission electron microscopy
(TEM).2  The existing procedure is essentially one in which each individual
fiber detected in a known fraction of the sample is identified and measured.•--
However, this existing procedure requires improvements in a number of areas.

     (a)  There must be a clearly-defined procedure for identification of
          fibers.

     (b)  It must incorporate a reliable means of  oxidation of organic
          materials which interfere with the analysis.  .

     (c)  A standardized fiber counting procedure  and reporting format must
          be defined.

     (d)  Reference suspensions must be incorporated so that quality assurance^
          samples of known values can be introduced "into"analytical programs."

     (e)  It must be capable of yielding reproducible analyses when these are
          conducted by different laboratories starting from actual liquid
          samples of both naturally-occurring and  laboratory-prepared fiber
          dispersions.                                                      .

     (f)  There must be a demonstration of the precision and accuracy attain-
          able, for both number and mass concentrations.

Since environmental water.samples contain many particles of elongated shape

-------
which are not asbestos  fibers,  during TEM analysis it is necessary to examine
each particle separately  in  order to discriminate the asbestos fibers.  Where
a fiber has limited electron scattering power its internal morphology may be
visible in the TEM.  This internal  morphology in some circumstances may be
sufficiently characteristic  to  be considered as an identification, but in
general other information is required before a fiber can be confidently iden-
tified.  In effect, the problem is  one of determining the precise crystal
structure and composition of an individual  particle which has crystallized in
a fibrous habit.   If the  analytical  method is to be used in a regulatory sense,
the fiber identification  procedure  must allow for no other interpretation to
be made.  Two fiber identification  techniques are available on a modern analy-
tical electron microscope:   selected area electron diffraction (SAED) and
energy dispersive  X-ray analysis (EDXA).   The SAED technique permits investi-
gation of the crystallography of an  individual fiber, and EDXA permits a
qualitative, or, in some  circumstances, a quantitative measurement of the
chemical composition of a fiber.

The inadequacy of  EDXA measurement  alone has been reported-.3  The problem
of fiber identification is illustrated in Figure 1, which shows the com-
positions of the more common fibrous silicate minerals which are collectively
thought of as asbestos.   It  will  be  noted that within one type there can be a
substantial variation in  composition.   In fact, the compositions overlap those
of many other non-fibrous and fibrous minerals.  Lee1* has demonstrated that
some specific amphiboles  and related minerals can be discriminated by quanti-
tative interpretation of  SAED patterns obtained by tilting the fiber so that
one of its principal crystallographic axes  (zone axes) is parallel to the
electron beam direction.   However,  demonstration that such patterns are com-
patible with the structure of a particular mineral does not exclude the possi-
bility that they may also be compatible with the structures of other minerals.
Only when the patterns have  been shown to be incompatible with all other
minerals in contention can it be stated that the fiber has been identified.
Precise and well-established methods for unequivocal identification of amphi-
bole asbestos in the TEM  have not been reported in the literature.
                                   . Asbestos
                                                    Amphiboles
             Serpentine
             Chrysotlle
           (White Asbestos)
           Mg3(S1205) (OH)4


Actln
I
AnthophylHte
(Mg. Fe)7(S18022) (OH)2
>Hte
Cs2(M9. Fe)5 (S18022) (OH)2


Amos


He
(Fe, Mg)7 (S1g022) (OH)2
                      TremoHte
                   Ca2Mg5 (S18022) (OH)2
                                                 CrocldoUte
                                                 (Blue Asbestos)
                                          Na2FeII3FeIII2 (S1g022) (OH)2
                  Figure  1.   Principal  varieties of asbestos.


                                  •'•''  2	  . .

-------
For the purposes of routine analyses of environmental  samples, nearly all
analysts have limited their identification procedure to observation of fiber
morphology and simple visual inspection of an SAED pattern obtained from the
fiber in whatever orientation it was found on the TEM specimen.   On the basis
of these observations, fibers were classified as chrysotile or amphibole,
using known chrysotile and amphibole fibers as reference standards.  Although
this procedure is probably adequate in many circumstances for identification
of chrysotile, it is-in-no way-specific for the classification of fibers as --
amphibole.5  Nevertheless, many analyses have been performed in which a layer-
type SAED pattern with a 0.53 nm spacing has been taken as evidence for the
presence of amphibole.  However, the more cautious analysts using this approach
have also used qualitative EDXA examination to supplement their observations,
even if these were not specifically called for in the published analytical
procedures.  Even with this more cautious approach, the precise identity of
fibers classified as amphibole by these routine methods is open to serious
question, and more sophisticated techniques are required for unequivocal
identification of fibers..                                                      .

Methods for preparation of water samples for examination in the TEM have been
the subject of considerable discussion... Of the. methods, that, have been used,
many have been discarded on account of particle losses during preparation.
One study reported an inter-laboratory disagreement by a factor of 300.6  The,
ideal TEM specimen for asbestos fiber counting has an absolutely uniform
deposit of particulate over the whole grid, and the density of the particulate
deposit is such that a significant degree of particle overlap does not occur.
The supporting substrate for the particulate should also be featureless, and
should not interfere with identification of the particles.  A further require-
ment is that a specific area of the TEM specimen should be related in an
accurate, quantitative manner to a known volume of the original  water sample.
Recently, the methods used for preparation.of TEM specimens have fallen into
several broad categories:

     (a)  Direct Transfer Methods

          These methods are based on filtration of the water sample through a  .
          membrane filter, after which the filtered particulate is transferred
          to a carbon-coated TEM grid by solvent dissolution of the filter.
          In general it is found that particulate is lost or moved during the
          dissolution of the filter.7

     (b)  Replicated Filter Surface Methods

          In the carbon-coated Nuclepore procedure,2 a 0.1 urn pore size
          capillary pore polycarbonate membrane filter is used to filter the
          water sample.  The active surface of the filter is carbon coated by
          vacuum evaporation, after which the filter is dissolved away in
          chloroform.  The particulate is retained in position by the carbon
          film during the dissolution.7  The conventional mixed-esters
          membrane filter has too much surface structure for this method to be
          directly applicable.  However, after filtration of the water sample
    ,      through such a filter, a method described by Ortiz and Isom8 can be
          used to prepare a TEM specimen.  The filter surface is exposed to


 	_ _  .._-.„.     '_. .;. 3.  	 __	   -..

-------
          acetone vapor, which causes the sponge structure to collapse into a
          continuous film.  The assumption is made that the particulate is
          still at the surface after this operation.   The procedure is then
          the same as for the carbon-coated Nuclepore preparation, except that
          acetone is used as the solvent.                                      :

     (c)  Drop Methods

          In these methods the TEM specimen is prepared by evaporation of a
          drop of liquid on a carbon-coated support grid.  There are several
          variations of this basic technique, but the common feature of all
          versions is the initial requirement for concentration of the fibers.
          Direct preparation from the water sample is not possible, since only
          5 microliters of the liquid can be conveniently mounted in a single
          drop on a 3 mm diameter carbon-coated electron microscope support
          grid.  If the concentration of asbestos in  the original water is
          106 fibers/liter, only 5 fibers would be transferred to the grid in
          this volume.  Even concentration by a factor of 100, e.g. filtration
          of 200 ml followed by redispersal of the fibers in a 2 ml volume,
          results in the transfer, of..only 500 fibers  unless multiple drops can
          be satisfactorily deposited on the grid. Moreover, much of the
          deposit is obscured by the grid bars which  may occupy as much as 60%
          of the total grid area.

          For the concentration step, ultra-centrifugation or filtration may
          be used.  Ultra-centrifugation of liter volumes of water, and redis-
          persal of the centrifugate in a smaller volume of clean water is a
          technique widely used in other fields, but  no work has so far been
          reported in which the asbestos fiber losses have been compared with
          known low-loss preparations.   Cross-contamination is undoubtedly a
          problem, but the technique does have the advantage that the use of
          filters is completely avoided along with their associated quality
          control deficiencies.  The alternative approach is filtration, and
          in order to redisperse the particulate in a small volume of liquid,
          either the filter must be ashed or the particulate must be ultra-
          sonically detached from it.  Ashing is undoubtedly the superior
          approach, since ultrasonic detachment leaves a doubt as to how much
          material remains on the filter.  Some results have been reported9
          which indicate that amphibole fiber losses  during low temperature
          ashing are insignificant.  No information is available for chryso-
          tile fiber losses, apart from that of Chatfield et al.7  The question
          then arises as to what proportion  of the droplet volume is repre-
          sented by the particulate deposit on one grid opening.   If the drop-
          let does not spread over the entire area of the grid before it
          evaporates, the actual area of particulate  deposit is not known and
          it is-also difficult to measure.  Moreover, it is not likely that
          the density of particulate deposits in areas near the edge of the
          evaporated droplet is the same as in those  areas close to the center.

As a first step in standardizing a satisfactory analytical method, a prelimi-
nary version of an interim procedure was published by the EPA in 1976.10 It
has since been shown that the carbon-coated Nuclepore technique included in

-------
this published procedure does not incur significant particle losses,7 and that
the uniformity of particle deposition is superior to that yielded by other
methods tested.  In contrast, this work7 also indicated that direct transfer
techniques involving reflux washing of unfixed particles on membrane filters
were at best critically dependent on the skill of the operator,  if not totally
irreproducible.  The non-reproducible behavior was  observed for  both the  Jaffe
Washer and the condensation washer.  In light of these results,  the membrane
filter-condensation washer approach was deleted from the revised version  of
the Interim Procedure published in I960.2

The carbon-coated Nuclepore procedure has been widely accepted as the most
satisfactory current preparation technique,  but no  complete statistical  studies
have been performed using samples from real  sources, as opposed  to laboratory-
prepared fiber dispersions.  Even in this case, when chrysotile  fiber disper-
sions in distilled water were analyzed by several laboratories,  the results
indicated that there was an unacceptable degree of  variability compared with
the results obtained when prepared filters were distributed.11  It has also
been found that there are some deficiencies  in the  Interim Procedure when
analyzing real water-samples, particularly in the case of organically contami-
nated waters.   Although low-temperature ashing is recommended in this proce-
dure and by others, 12>13»11* even the initial  filtration sometimes cannot  be
performed satisfactorily.   In more marginal  cases  the absence of replicated
filter detail  indicates that there was a coating of slime on the filter
surface.  Any fibers embedded in this slime  would not be transferred to the
TEM samples, and consequently the results from such samples are  of question-
able accuracy.  The author has previously reported  application of a technique
using ozone gas and ultraviolet light for oxidation of organic materials  in
liquid samples.15*16  The treatment is given prior  to the filtration, and in
most cases interfering organic.materials are oxidized to leave a clear liquid
which is readily filtered.  No other chemicals are  used.   Since  the oxidizing
agent is a gas, any amount of oxidizer can be applied to a sample without the
addition of foreign ions or any increase in  the total sample volume which
would result in dilution of the original fiber concentration.  Although the
effects of the technique have not been fully and systematically  investigated,
it allows the use of the low temperature asher and  high power ultrasonic
treatment to be avoided.  The technique does not remove large cellulose
fibers; other similar refractory organics may also  be incompletely oxidized
by the treatment.                      '

The research program directed towards improvement of the TEM analytical
technique is described in this report.  Sample collection and handling proce-
dures were investigated, and various techniques for preparation  of TEM
specimens from water samples were compared.   Particular attention was given
to the problem of precise fiber identification and  a standardized reporting
format was developed.  Computer programs were written to simplify the data
reduction for both fiber identification and  reporting.   This work forms the
background study supporting the issue of a new analytical technique for
determination of asbestos in water samples,  published separately.17

-------
                                  SECTION 2_i.

                       CONCLUSIONS AND RECOMMENDATIONS
The investigation of specimen preparation techniques for asbestos fiber count-
ing by transmission electron microscopy showed that the carbon-coated Nuclepore
method was superior to both the "drop" method and the collapsed membrane filter
method.  The "drop" method, in which a microliter volume of a concentrated
dispersion is evaporated on a carbon-coated TEM grid, was shown to produce
samples on which the fiber distribution was not sufficiently uniform to
warrant their use in quantitative determinations.  The collapsed membrane
filter method was investigated in detail, and wa^ shown to display strongly
size-dependent fiber losses relative to the Nuclepore preparation.  For chry-
sotile, the fiber losses increased with the pore size of the membrane filters
used.  For the 0.45 ym and 0.22 ym pore size filters, the losses were statisti-
cally significant at the 5% level; for the 0.1 ym pore size filters, fiber
losses were not. significant at the 5% level..  For fibers shorter than 1.0 urn,
the fiber losses using the 0.45 ym pore size filter were very high,  and only
24% of these fibers were transferred to the TEM samp-le.   The corresponding
value for the 0.22 ym pore size filter was between about 60% and 70%.   The
results were consistent with the postulate that the shorter fibers penetrated
the filter structure more deeply and were engulfed during the collapsing pro-
cedure.  The results for crocidolite were more difficult to interpret.   Al-
though the total fiber losses using the 0.22 ym and 0.1 ym pore size filters
were not significant at the 5% level, in some fiber size ranges significant
losses were observed.  On the basis of the results, the collapsed membrane
method was found to be unsatisfactory for quantitative analyses.    It is, how-
ever, recommended that if cellulose ester membrane filters must for some
reason be used, plasma etching of collapsed membranes should be investigated
as a means of increasing the transfer-efficiency of short fibers to  the
TEM specimens.

Samples from water sources contaminated by chrysotile fibers were collected
from Sherbrooke, Quebec, and it was shown that 10 replicate measurements from
each of these samples were statistically compatible.   The same conclusion was
drawn for samples contaminated by amphibole fibers collected from the Duluth
area of Minnesota.  This indicated that for a series  of sub-samples  filtered
at the same time, intra-laboratory measurements by a  single operator using the
same instrument were repeatable.

A method of oxidation of organic materials in water samples, based on the use
of ozone gas and short wavelength (254 nm) ultraviolet light.was  found  to be
successful.  This oxidation technique was found to remove those organic com-
ponents of drinking water samples which inhibit filtration, and was  an  effec-
tive and more convenient replacement for oxidation by two-step filtration and

-------
 low temperature  ashing.  When  the ozone-ultraviolet light technique was used,
.no changes  in. either the electron ...diffract! on__behavi or or .the chemical composi-
 tions of chrysotile  and amphiboje fibers were detected.  A second oxidation
 technique",  based on "thei use  of oxygen at pressures up to about  13.8 MPa and
 temperatures  of  up to 300°C, was found  to be effective" for removal of organic
 materials,  but some  degradation of  chrysotile fiber morphology was observed
 after treatment  at the most  extreme  conditions.   It was also found that con-
 tainers  made  of  polytetrafluoroethylene were required; since both glass and  •
 silica were attacked under the extreme  conditions used.  Because of the success
 of the ozone-UV  technique, no  further investigation of.the more involved high-
 pressure method  was  conducted.  However, it is recommended that for samples  "~~
 containing  large concentrations of  refractory organics, such as sewage or
 plant effluents, the technique warrants further consideration.

 Studies  of  the stability of  asbestos fiber dispersions yielded some surprising
 results.  Initial experiments  indicated that mechanical shaking of polyethylene
 bottles  containing chrysotile  fiber  dispersions in double-distilled water re-
 duced the suspended  fiber  concentrations to very  low values.  This effect did
 not occur if  the bottles were  exposed to continuous ultrasonic agitation for a
 similar  period of time.  The behavior was unaffected by either ionic or pH
 conditions.   The effect was  also observed for dispersions of crocidolite.
 It was  deduced  that  the  presence of trace organic materials of bacterial-:^ ,u
 origin  in  some  way promoted  the attachment of asbestos"fibers to the inside
 surfaces of the containers.   It was considered that this effect could seriously
 compromise the  results of  routine sample analyses.  Container and storage
 studies were conducted which  indicated that the effect was a consequence of an
 organic product of bacteria,  rather than a mechanism involving direct.interac-
_tipn with_the bacteriajthemsel.ves, and that.the organic material was probably	
 a variety  of polysaccharide.   It was found that the effect of this phenomenon
 on routine sample analyses could be eliminated by ozone-UV treatment carried
 out inside the  original  sample container.  This treatment was found to perform
 the double task of oxidation  of interfering organic materials and release of
 fibers  attached to the container.

 The observation of the interaction of asbestos fibers with the trace organic
 materials  had two other  consequences:, the development of stable reference
 fiber suspensions, and development of a separation technique which was at
 least partially specific for chrysotile.  Reference fiber suspensions have
 been required for some time  in order to facilitate analytical quality assur-
 ance programs,  but their stability has always been in question.  It was found
 that if the reference dispersions were prepared so as to exclude all bacteria
 and their  organic products,  they were then stable for long periods of time
 provided that absolute sterility was maintained.  It is recommended that a
 standards  agency maintain  a  supply of these reference dispersions, with
 appropriate certification  of their contents, so that analytical quality
 assurance  of future  sampling programs can be established by incorporation of
 control samples.

 Initial investigations of  a  separation technique based on fiber attachment to
 container  walls were carried out as support for development of the rapid
 screening  method.  The separation technique was found to be effective as a

-------
selective method for concentration of chrysotile asbestos  fibers  from water
samples; its selectivity for separation of amphibole fibers  was not tested.   .
It is recommended that this separation method should be developed further,  and
that the mechanism which gives rise to the attachment phenomenon  should be
investigated.  The observation of this strong interaction  between asbestos
fibers and organic materials of biological origin may have significance in
other fields unrelated to analytical  method development.

The fiber identification protocol based on zone axis SAED  and quantitative
EDXA is capable of more specificity than had previously been provided.by  TEM
analysis.  The identification procedure permits determination of  approximate
chemical composition, which is adequate for the general  classification of
amphibole fibers, but is not sufficiently precise for the  incorporation of
adjectival modifiers in the mineral description.   It is recommended that  the
identification procedure be reviewed on a regular basis, and that more precise
X-ray analytical procedures be developed and applied as they become available.

The improved reference analytical method for determination of asbestos  fibers
in water represents the best available technology.   A number of new features
were incorporated.  These include the introduction of ozone-UV oxidation  for
all samples, a fiber classification system, a minimum fiber  length for report-
ing, a standardized reporting format, quantitative interpretation of fiber
identification data, and a fibrosity index which appears to  permit discrimina-
tion of fibrous and non-fibrous species.  The changes and  additions  introduced
into the basic analytical method should eliminate the problems of poor inter-
laboratory reproducibility which have been observed in the past.

Conclusions and recommendations relating to the development  of a  rapid screen-
ing technique for measurement of asbestos concentration are  included in the
separate report on this aspect of the overall  program.

-------
. ••"	_	;	.;	SECTION s

              INVESTIGATION OF TEM SAMPLE PREPARATION TECHNIQUES


 Although application of the EPA Interim Method2 has apparently been success-
 ful, both before.and since its publication a number of other preparation
 ^techniques have;been used by some laboratories.  These techniques include
\several variations of the "drop" method and the collapsed membrane filter
 jmethod. _      ]	         i.	  _'	   _     ^_

 A serious problem associated with the carbon-coated Nuclepore method
 (recommended in the Interim Method) is the insolubility of some components of
 the filter 1n organic solvents.  This undesirable feature of the polycarbonate
 filter is associated with the polymer crystal!inity, and is actually both
 necessary and intentionally present so that the membrane has adequate strength.
 Unfortunately this property appears difficult to control during manufacture
 and is variable between individual batches of filters.  The replication of the
 filter surface and the pores has also been considered to be a minor incon-
 venience, since the surface detail creates a distrac.tion for the microscope
 operator during TEM examination.  However, when samples with very little
 deposit on the filter surface are examined, the replicated filter detail is
 useful in that it provides sample position reference for the operator.  The
 background detail in these circumstances is therefore an advantage.   The
 problems associated with limited filter solubility can be avoided by using
 either the "drop" technique or the collapsed membrane filter technique.  These
 techniques also yield TEM samples with reduced surface detail.

 The direct "drop" method consists of concentration of the fibers by filtration
 or centrifugation, followed by deposition of a few microliters of the concen-
 trated dispersion on a carbon or collodion-coated TEM grid.  There are a
 number of variations of this method,12*18 but the initial requirement is one
 of sample concentration.  Direct preparation from the water sample is usually
 not possible, since only 5 microliters of the liquid can be conveniently
 mounted in a single drop on a 3 mm diameter carbon-coated electron microscope
 Support grid.  The level of asbestos in most waters is too low to provide a
 sufficient number of fibers in the 5 yl drop.  If filtration is used to concen-
 trate the sample, the filtered particulate must be removed from the filter and
 redispersed in a small volume of liquid before the grid can be prepared.  The
 only routes currently available for this are ashing of the filter, or ultra-
 sonic detachment of the particulate.  Alternatively, centrifugation has some-
 times been used to concentrate the sample.  Although ult-ra-centrifugation of
 liter volumes of water followed by redispersal of the centrifugate in a
 smaller volume of clean water is a technique widely used in other fields, no
 work has so far been reported in which the asbestos fiber losses have been
 compared with known low-loss.preparations.  However,, this technique does avoid

-------
completely the use of filters and any associated quality control deficiencies.

When the concentrated suspension has been prepared, a few microliters are then
transferred to a coated grid.  The question of how uniform the deposit will be
after evaporation has been discussed extensively.  Mudroch and Kramer18 used a
carbonized collodion-coated grid and took no action to facilitate spreading of
the droplet, whereas Cunningham and Pontefract treated the carbon-coated grid
for a short period in a low temperature plasma asher to render its surface
hydrophilic.19- Moreover, the proportion of the drop represented by the deposit
on one grid opening must be accurately determined before the concentration of
fibers in the original sample can be calculated.

Another technique which has been used by some laboratories is the "collapsed
membrane" method.8  Essentially, the liquid is filtered through a conventional
mixed-esters sponge-type membrane filter, after which the filter structure is
collapsed into a continuous film by exposure to solvent vapor.   For TEM sample
preparation the filter surface is then carbon coated, after which the filter
is dissolved by a Jaffe Washer procedure using acetone as the solvent, leaving
an extraction replica which contains the particles which were originally on
the surface of the filter. -The technique requires that all of the fibers
remain on the surface of the filter after the collapsing process, so that they
are available for entrapment in the evaporated carbon film; any particles
which become covered during collapse of the structure will  not be transferred
to the TEM specimen.  Particle losses using this technique are likely to be
most serious for the larger pore-size filters, on which fibers  can penetrate
more deeply into the filter structure during sample filtration.

3.1  DROP METHOD   .-     .\

     A short study was made of the characteristics of TEM grids prepared by
     the "drop" technique.   The uniformity of the particulate deposit on the
     grid openings was investigated using several of the technique variations
     which have been published.18»19i20   The problem of uniformity of deposit  .
     was considered by Cunningham and Pontefract,19 who observed that the
     carbon film could be rendered hydrophllic by placing the grids in a
     plasma asher for a few minutes.  The Ontario Ministry of the Environ-
     ment20 attempted to solve this problem by adding detergent to the con-
     centrated fiber dispersion before mounting the drop on the grid.   The
     merits of inverting the grid during the evaporation of the sample drop
     have also been considered.

     Carbon-coated copper grids were prepared by the film cast  technique in
     which a drop of 1% collodion in n-pentyl  acetate is placed on water in
     a 10 cm diameter crystallizing dish.   After the solvent has evaporated,
     copper grids are dropped onto the floating collodion film.   A Whatman
     filter paper is then placed on the floating film,  and immediately removed
     and inverted.  The collodion film adheres to the filter paper, trapping
     the grids.  After the filter paper is dry, the collodion-coated grids  can
     be removed.  The grids are then carbon-coated by vacuum evaporation, and
     the collodion layer is dissolved away by reflux washing with acetone in
     a Jaffe Washer.21


                                     10

-------
     A standard dispersion of UICC chrysotile was  used as  the test dispersion
     for mounting on the grids.   Grids  were prepared from  the test dispersion
     as follows:

     (a)  untreated grids, evaporation  both upright and inverted;

     (b)  plasma-treated grids,  evaporation both upright and inverted;

     (c)  untreated grids, detergent added to dispersion,  evaporation both
          upright and inverted.

     Grid openings were selected randomly and the  number of fibers on each was
     recorded.   A chi-square test was used to test the uniformity  of the
     deposit on 10 grid openings of each grid.

     None of the grids displayed a uniform deposit of fibers at even the
     0.1% significance level.  Because  of this observation, further work on the
     "drop" methods of TEM grid preparation was not carried out.   Although
     these methods are convenient for mounting small  quantities of particulate
     material where quantitation is not required,  they are not suitable for
     quantitative fiber concentration measurements.

3.2  COLLAPSED MEMBRANE FILTER METHOD

     A significant feature of the carbon-coated Nuclepore  technique is  that
     during the filter dissolution step, all of the particles on the filter
     surface are held in position by the evaporated] carbon film,  resulting in
     a minimal  loss of particulate during the preparation.   The technique
     described by Ortiz and Isom8 apparently permits the same advantage to be
     obtained when a conventional mixed-esters membrane filter is  used  for the
     sample filtration.  In the technique of Ortiz and Isom, the sponge struc-
     ture of the filter is collapsed into a thin plastic film by exposing the
     filter to acetone vapor.  The steps in the procedure  are shown in
     Figure 2.   The exposure to acetone vapor must be monitored very carefully,
                                          1.   UNUSED MEMBRANE



                                          2.   AFTER FILTRATION


                                       :   3.   COLLAPSED MEMBRANE
                                          4.   CARBON  COATED AFTER COLLAPSE

                                          5.   CARBON  FILM AFTER DISSOLUTION
 Figure 2.  Steps in the collapsed membrane procedure of filter preparation.

                                     11

-------
 since under-exposure will  fail  to  collapse the surface.structure to a
 continuous film,  and over-exposure may cause  the  fibers  to  become engulfed
 in the plastic.   Although  chromium was incorporated with carbon  as the
 evaporated coating in the  published method, this  would  be unsuitable for
 application to  asbestos  measurement because it would yield  a  strong diff-
 raction pattern of its own during  examination of  fibers  by  SAED.   The
 coating material  is not  considered to  be  critical, (Ortiz,  L.W.,  personal
 communication)  and carbon  was selected.

 In an initial study,  a series of suitably loaded  filters was  prepared
 by filtration of  identical  volumes of  a dispersion of chrysotile.   The
 fiber size distribution  was selected to be similar to that  normally
 encountered in  water samples.   The filters used were:

 (a)   0.1 ym pore  size Nuclepore capillary pore polycarbonate;

 (b)   0.1 ym pore  size Millipore mixed  cellulose esters;

 (c)   0.22 ym pore size Millipore mixed cellulose  esters;  and

 (d)   0.45 ym pore size Millipore mixed cellulose  esters.

 The  filters were  prepared  by repeating the sequence (a)  to  (d) until
 5  filters of each type were prepared;  in  this  way any change  in  the fiber
 concentration of  the  dispersion during the preparation would  affect all
 of the filter types equally.  When using  the  collapsed membrane technique,
 a  50% success rate in preparing specimen  grids 'suitable  for TEM fiber
 counts was achieved.   Some filters disintegrated  immediately  upon  being
 placed in the Oaffe Washer.  In two cases  a film  of plastic remained
 which was sufficient to  interfere  with fiber  identification even after
 the  grids had been on the  Jaffe Washer for 17  hours.   There was always a
 significant degree of breakage  of  the  carbon  film during  the  filter dis-
 solution step;  and in some cases only  two  or  three grid openings remained
 intact on the grid.   So  far as  the removal of  filter structure detail  is
 concerned, it is  considered that the best results were obtained when  the
 sample was exposed to acetone vapor during the collapsing step until  the
 filter just cleared.

 The  results obtained during  the initial study  are shown  in Table 1.   The
 detailed fiber  counting  data are shown in  Table A-l of Appendix A.   In
 Table 1, comparison of the values  obtained from the membrane  filters with
 those from the  carbon-coated Nuclepore filters shows a progressively
 increasing degree of fiber loss as the membrane filter pore size
 increases.   At  the 5% significance level,  statistically valid numerical
 fiber losses of 20.4% and  56.5% are shown  for  the 0.22_um and 0.45  ym
"pore size Millipore filter preparations respectively."  '

 A  more detailed study was  made  on  the  performance of the 0.1 ym and
 0.22 ym pore size Millipore cellulose  esters membrane filters.  In  view
 of the initial  results,  no further work was performed on the 0.45 ym
 pore size membrane filters.  Using a standard  dispersion of refined Union
 Carbide Calidria  chrysotile, filters were  prepared as before, alternating

                                12

-------
TABLE 1.  COLLAPSED MEMBRANE STUDY FOR CHRYSOTILE:  INITIAL MEASUREMENTS
Preparation
Carbon-Coated Nuclepore
0.1 pm Pore Size
Polycarbonate
Collapsed Mlllipore
0. 1 pm Pore Size
Cellulose Esters
Collapsed Mlllipore
0.22 ym Pore Size
Cellulose Esters
Collapsed Mi Hi pore
0.45 vm Pore Size
Cellulose Esters
Number
of
Filters
Analyzed
3
3
5
3
Fiber Concentration
(106 Fibers/Liter)
Mean
26.9
26.0
21.4
11.7
95% Confidence
Interval
20.4 - 33.4
9.9 - 42.1
15.7 - 27.1
6.1 - 17..3
Fiber Loss
in
Preparation
(*)*
-
3.3
20.4
56.5
Is the Fiber
Loss
Detectable
at 5%
Significance?
-
No
Yes
Yes
              *Compared to the Carbon-Coated Nuclepore Preparation

-------
the types of filter throughout the series of filtrations.  Ten filters
each of the 0.1 ym pore size Nuclepore, and Mi Hi pore filters of 0.1 urn
and 0.22 ym pore size, were prepared with identical  fiber loadings.  TEM
specimen grids were prepared from the Nuclepore filters by the conven-
tional carbon coating procedure, and from the Mi Hi pore filters by the
collapsing technique.  The results are shown in Table 2, with the detailed
fiber counting data in Tables,A-2 to A-4 of Appendix A.  It can be seen in
Table 2 that the 0.1 ym Millipore filter preparations showed no detectable
fiber loss at 5% significance,  but that the 0.22 ym Millipore preparations
showed a mean loss of 23.4% relative to the Nuclepore preparations.

The fiber losses of the collapsed Millipore preparations might be expected
to be fiber size-dependent, since they are probably a consequence of
embedding of the deposit during the collapsing process.  If the carbon-
coated Nuclepore technique is assumed to incur negligible fiber losses,
the fiber length distributions  observed using the two techniques allow
the fiber transfer efficiency of the collapsed membrane technique to be
calculated as a function of fiber length.   These transfer efficiency
values are independent of the fiber length distribution of the dispersion
used for the comparison.  The transfer efficiency histograms for chryso-
tile, using 0.45 ym, 0.22 ym and 0.1 ym pore size membrane filters, are
shown in Figure 3.  It can be seen that for the 0.1 ym pore size membrane
filter, the transfer efficiency remained at about 100% for the entire
fiber length range investigated (0.2 ym to 10 ym).   In contrast, the re-
sults using the 0.45 ym pore size membrane filters were unsatisfactory,
for fiber lengths shorter than  1 ym, only 24% of the fibers were trans-
ferred to the TEM grids.  The results using the 0.22 urn pore size membrane
filters roughly paralleled those for the 0.45 ym pore size filters, but
they showed somewhat higher transfer efficiencies.   It is clear that a
significant proportion of the fibers, failed to transfer to the TEM grids
when the 0.22 ym and the 0.45 ym pore size filters were used,  and that
this effect was strongly fiber size-dependent.  Accordingly, the col-
lapsed membrane filter technique using the larger pore size filters
introduces a serious distortion into the observed fiber length
distribution.

To test the collapsed membrane  preparation technique using amphibole
fibers, a similar series of filter samples was prepared using a dispersion
of UICC crocidolite.  The results obtained are shown in Table 3, with the
detailed fiber counting data in Tables A-5 to  A-7 of Appendix A.   In
Table 3 it can be seen that statistically significant total  fiber losses
were not demonstrated for either the 0.1 ym or the 0.22 ym pore size
Millipore filter preparations.   Figure 4 shows the fiber transfer effi-
ciency histograms, which indicate that the behavior of crocidolite was
different from that of chrysotile.  In particular, for both the 0.1 ym
and the 0.22 ym pore size filters, the longer fibers were transferred
less efficiently to the TEM specimen grids.  The histograms show that
substantial distortions were introduced into the observed fiber length
distributions.
                                 14

-------
^tU^COUAPSED
                                          MEMBRANE STUDY FOR  CHRVSOTILE
                                           fiber Concentration        Fiber Loss  (Is the Fiber
                                           (1Q6 Fibers/Liter)            in

                              Filters   I - i - • - 1  Preparation
                              Analyzed    Mean     95% Confidence


                                                   Interval
                                                                  Loss

                                                               Detectable

                                                                 at 5%

                                                              Significance?
 Polycarbonate
 .   urn Pore Size

Ce"ulose Es
                                  15

-------
        TABLE 3.  COLLAPSED MEMBRANE  STUDY FOR  CROCIDOLITE
Preparation
Carbon-Coated Nuclepore
0. 1 ym Pore Size
Polycarbonate
Collapsed Mi Hi pore
0.1 ym Pore Size
Cellulose Esters
Collapsed Mi Hi pore
0.22 ym Pore Size
Cellulose Esters
Number
of
Filters
Analyzed
10
10
10
Fiber Concentration
(106 Fibers/Liter)
Mean
224
229
204
95% Confidence
Interval
158 - 290
200 - 258
183 - 225
Fiber Loss
in
Preparation
(*)*
-
0
8.9
.Is the Fiber
Loss
Detectable
at 52!
Significance?
-
No
No
                *Compared to the Carbon-Coated Nuclepore Preparation
I2O
z
UJ
£100
UJ
£ 80
UJ
u
u.
u.
w 60
at
UJ
u.
OT
I 40
0
E 20
UJ
3
n
T~~
161
p





"

-

-


t i i



1



O.ljjm PORE SIZE







1 1 1 t 1 1 1 II
IZO

100

80

1

60

40

20


LU-LjJ n



-


.



0.22 >im PORE SIZE 1 	 .

-

-


i i i i i 1 1 1 1 i i i i i 1 1 1 1
O.I
1.0
   10   O.I
FIBER LENGTH, jjm
..1.0
10
            Figure 4.   Collapsed membrane filter method:
                        UICC crocidolite transfer efficiencies.
                                  16

-------
TABLE 2.   COLLAPSED
MEMBRANE STUDV FOR CHRVSOTILE
 Number
   of
niters
Analyzed
                                               Fiber Loss    Is
                                                   in
                                               Preparation
                                        the Fiber
                                        Loss
                                      Detectable
                                       at 5%
                                    Significance?
               ""

-------
TABLE  3.   COLLAPSED MEMBRANE STUDY FOR  CROCIDOLITE
Preparation
Carbon-Coated Nuclepore
0.1 ym Pore Size
Polycarbonate
Collapsed Mi Hi pore
0.1 ym Pore Size
Cellulose Esters
Collapsed Mlllipore
0.22 urn Pore Size
Cellulose Esters
Number
of
Filters
Analyzed
10
10
10
Fiber Concentration
(106 Fibers/Liter)
Mean
224
229
204
95% Confidence
Interval
158 - 290
200 - 258
183 - 225
Fiber Loss
in
Preparation
W*
-
0
8.9
.Is the Fiber
Loss
Detectable
at 52S
Significance?
-
No
No
        *Compared to'the Carbon-Coated Nuclepore Preparation
120
h-
Z
UJ
£100
UJ
a.
V
z1 80
UJ
u
GI
u.
w 60
Jfn PORE SIZE
60


40


20




1 1 1 1 1 1 1 ll 1 1 1 1 t 1 1 1 1 n
-




_ ^™^^"*^^™




- •



0.22 jum PORE SIZE l_


-


-




i i i i i 1 1 il i i i i i 1 1 1 1
o.i i.o 10 "b.i .. i.o 10
                      FIBER LENGTH, >im

   Figure  4.   Collapsed membrane filter method:
               UICC crocidolite transfer efficiencies.
                         16

-------
3.3  SUMMARY.OF ALTERNATIVE SAMPLE PREPARATION METHODS^

     The purpose of the investigation  was  to  determine  if  any other TEM
     preparation technique would give  results superior  to  those of the
     carbon-coated Nuclepore method, or if a  simpler  technique of equivalent
     performance could be defined.  The "drop"  methods  were  found to  be quite
     unsatisfactory, since uniform deposits could  not be obtained reliably.
     Moreover,  because the area over which the  drop spread before it  evapo-
     rated is not easily measured,  it  is difficult to determine an accurate
     relationship between the fiber count  on  a  small  number  of grid openings
     and the fiber concentration of the suspension being investigated.
     Accordingly, the technique was not investigated  further.

     The collapsed membrane technique  appeared  initially to  be promising,  but
     it was found that only a small proportion  of  the grid openings had carbon
     film which was intact, and that the success rate for  satisfactory grid
     preparation was only about. 50%.   The  technique appeared to be equivalent
     in performance to the carbon-coated Nuclepore method  for chrysotile
     deposited on 0.1 ytn pore size filters.   For larger pore sizes, the
     method had significant size-dependent fiber losses.   For amphibole fibers,
     total numerical fiber losses were not demonstrated for  either 0.1 ym  or
     0.22 ym pore size Millipore filters,  but it appears that there was a
     trend towards lower transfer efficiencies  for the  longer fibers  on
     filters of both pore sizes.
                                     17

-------
                                  SECTION 4

           REPRODUCIBILITY OF ENVIRONMENTAL WATER SAMPLE ANALYSES
4.1  GENERAL
     Although satisfactory reproducibility has been demonstrated for analyses
     of a large number of Nuclepore filters which were prepared at one time
     from a single fiber dispersion,7'^2 this has not been the case.when a
     water sample was analyzed at different times after it was collected.
     Inter-laboratory analyses conducted by distribution of water samples,
     rather than filters, have largely been unsuccessful.11  These studies
     have displayed significant variability for both well-controlled fiber
     dispersions prepared in the laboratory and naturally-occurring fiber dis-
     persions.   Moreover, collection of separate samples from the  same source
     has sometimes yielded conflicting results.   A study was made  to determine
     the'intra-laboratory analytical reproducibility of several  naturally-
     occurring fiber dispersions.   The study was designed so that  any major
     effect of sample volume on the analytical reproducibility could also  be '
     determined.

     Preliminary work indicated that-fiber dispersions vary greatly-in their —
     ability to yield satisfactory TEM samples.   Natural  sources can often
     produce a very non-uniform, highly agglomerated deposit on  the TEM speci-
     men.   Inorganic particles tend to agglomerate with the gelatinous organic
     matter sometimes found in natural  waters.  This effect is illustrated in
     Figure 5,  which shows the deposit formed on a Nuclepore filter by the
     filtration of 20 mL of water from an Ontario lake which is  known to have
     high levels of chrysotile contamination.   The sample  was  shaken  vigorously
     for about 30 minutes before the aliquot was taken.   As can  be seen in the
     micrograph, the suspended matter has not been uniformly dispersed.  A
     TEM sample prepared in this way would not allow identification of most of
     the chrysotile which is present.   Figure 6  shows  the  deposit  from 20  mL
     of the same water sample, but the sample had been exposed to  ultrasonic
     treatment at a power of 2.2 W/L for a period of 30 minutes  before filtra-
     tion.   The deposit would still  not permit identification  of most fibers
     mixed with it.  This problem has been found to occur  frequently  in raw
     lake and river waters, particularly after storage of  the  samples.   In
     these cases, it is very difficult to produce a uniform, homogeneous sus-
     pension which will permit the preparation of useful specimens directly by
     the carbon-coated Nuclepore technique.   Prior'to  these studies,  the fac-
     tors which control the stability of an  asbestos fiber dispersion were not
     clearly understood.   Similar phenomena  had been observed  even for simple
     dispersions of chrysotile in distilled  water, although on some occasions
     such dispersions were found to be stable for long periods of  time.

                                     18

-------
                                                     2.0 jum
Figure 5.   SEM micrograph of the deposit on  a  Nuclepore  filter  produced  by
           filtration of 20 mL of lake  water after  the sample was  shaken.
                                                                                     -- ~
Figure 6.  SEM micrograph of the deposit on  a  Nuclepore  filter produced by
           filtration of 20 ml of lake water after ultrasonic treatment at
           2.2 W/L for 30 minutes.
                                    19

-------
     The volume of sample required in order to ensure that the sample is
     representative of the source concentration is probably related to the
     character of the source.  It can readily be demonstrated that samples
     collected sequentially from a faucet may yield statistically incompatible
     values for the fiber concentration, whereas the values obtained from
     replicate samples taken from the same bottle may be compatible.  Table 4
     shows the chrysotile concentrations found in a study of the Sherbrooke,
     Quebec water supply;23 and illustrates the limitation of "grab" samples.
     Although analyses of replicates from the same bottle were statistically
     compatible, the actual range of values obtained at the different points
     in the system indicates either a temporal variation or source
     inhomogeneity on the one liter scale.
    TABLE 4.
CHRYSOTILE FIBER CONCENTRATIONS IN MUNICIPAL DRINKING WATER,

                SHERBROOKE. QUEBEC
Water Type
Raw
Treated*
Distribution 1
Distribution 2
Fiber Concentration (105 Fibers/Liter)
Mean
72.5
26.1
222
80.3
95% Confidence Interval
59.1 - 85.9
16.3 - 35.9
177 - 267
30 - 129
           *Municipal water treatment, at the time of sampling,
            consisted only of the addition of chlorine gas and
            ozone to the raw water.
     A study was undertaken to evaluate the reproducibility of the  EPA Interim
     Method2 when applied to environmental  samples,  and to investigate any
     observed container or sample volume effects.  The analyses were  made  of
     samples collected from eight locations, four  of which were associated
     with an amphibole fiber source and four with  a  chrysotile source.

4.2  SAMPLE COLLECTION

     A sampling scheme was devised for the  collection of environmental  samples,
     both from tap sources and from large bodies of  water.   These samples  were
     used in the investigation of:                          .

     (a)   the reproducibility of analyses of replicates  from one collected   .
          sample;
                                      20

-------
      (b)  the reproducibillty of duplicate samples from a single source; and

      (c)  the determination of a suitable sample volume.

      The sampling scheme for collection of tap water samples was to fill one
      5 gallon polyethylene bottle, followed by alternate 1 liter and 250 ml
      bottles until thirty of each had been collected, and completing the series
  	-with another 5 gallon-bottle. • The sampling of a large body of water was
      subdivided into two sets.  Two sets of samples were then taken from .the
      body of water which serves as the raw water supply for the location in
      which the tap water samples were obtained.  In the first set a total of
      30 samples were collected, alternating between 1 liter and 250 ml sample
      bottles, followed by collection of a single 5 gallon sample.  The second
      set of samples was collected in a similar manner at a sampling point in
      a water treatment plant supplied from the raw water source.

      For water with high levels of amphiboles, tap water samples were taken in
      Beaver Bay, Minnesota (in the Duluth area).   Here the water is taken
      from Lake Superior and pumped through a coarse sand filter directly into
      the distribution system.  The raw water samples were taken at the shore
      of Lake Superior off a rocky promontory at the Buchanan Town Marker, about
      11 miles northeast of Duluth.  The final series of samples was taken in
      the Duluth Pumping Station.  The raw water intake is stated to be about
      % mile from the shoreline and 75 ft. below the surface.

      Similar sets of samples were also collected in the Sherbrooke area of
      Quebec province where the river water and the municipal water were known
      to contain chrysotile fibers at concentrations adequate for this study.
..   ---The tap-water samples, were collected in Sherbrooke. .At the time. of.
      sampling, the municipal water was taken from the Magog River and treated
      with chlorine gas and ozone.  There was no filtration of the water before
      distribution.  The series of raw water samples was taken at the shore of
      the Magog River, approximately 1% miles upstream of the treatment plant
      intake.  Samples were also taken from the raw water intake at the
      Sherbrooke Water Treatment Plant.

4.3  ANALYTICAL REPRODUCIBILITY

      Ten replicate filters were prepared from each of the eight 5 gallon
      bottles.  These filters were analyzed to determine the analytical repro-
      ducibility of replicate samples prepared from natural waters known to
      contain either amphibole or chrysotile fibers.  In the case of the Duluth
      area samples, it was  found that the number of amphibole fibers found on
      10 grid openings of the TEM specimens was inadequate for satisfactory
      analysis of the data.  The level of total solids in the water did not
      permit filtration of  larger volumes.  Accordingly, the fiber counts were
      continued until 20 grid openings had been examined.  For the Sherbrooke
      area samples, the relative levels of chrysotile and other materials were
      such that fiber counts on 10 grid openings yielded sufficient numbers of
      chrysotile fibers for analysis of the data.
                                      21

-------
The detailed analytical data are shown in Tables B-l to B-8 of Appendix  B.
Statistical analyses have been made of the results, using the grid
opening fiber counts.  These analyses are also shown in Appendix B  in
Tables B-9 to B-16.  For each series of replicate analyses, the mean value
of the ten measurements has been calculated, along with an estimate of
variance for the mean.  Using the two-sided  ".f'-test, each individual
sample analysis result has been compared with the mean of ten measure-
ments, to determine if it is a replicate value of the mean.  In the
majority of cases, the results show that, at 5% significance, the
individual values are indistinguishable from the mean.  A summary of
the statistical results is shown in Table 5.  The study indicates that
from a single 5 gallon sample, with few exceptions it is possible to
obtain replicate results from 10 ml sub-samples, both for the amphibole
and the chrysotile dispersions.

As has been found on previous occasions, subsequent analyses of the
separately-collected 1 liter and 250 ml samples did not replicate the
values obtained from the 5 gallon samples.  The values obtained for the
first samples analyzed were all significantly lower than those which had
been obtained from the larger 5 gallon samples.  No further analyses
were performed on these samples, since it was considered more beneficial
to investigate the reasons for the instability of such suspensions.
     TABLE 5.  REPLICATE ANALYSES OF ENVIRONMENTAL VJATER SAMPLES:

                        SUMMARY OF RESULTS
Source
Beaver Bay Tap Water,
Initial Sample
Beaver Bay Tap Water,
Final Sample
Lake Superior Water
Duluth Raw Water
Sherbrooke Tap Water,
Initial Sample
Sherbrooke Tap Water,
Final Sample
Magog River, Sherbrooke
Sherbrooke Raw Water
Fiber Type
Amphibole
Amphibole
Amphibole
Amphibole
Chrysotile
Chrysotile
Chrysotile
Chrysotile
- Number of Analyses--
Statistically Compatible
with Mean*
9
10
10
9
9
7
9
10
Number of
Samples
Analyzed
10
10
10
10
10
10
10
10
                  •"Two-sided ".("-test at 53S significance
                                22

-------
 ".'	____	:	SECTIONS

                       OXIDATION OF ORGANIC MATERIALS


Organic materials, often present in water samples as slimes or algae, interfere
with TEM specimen preparation by the carbon-coated Nuclepore procedure.   A
coating of slime on the filter has two effects:  it retards or may even  prevent
filtration of the sample; and since only the top surface of the coating  is
replicated, any particles embedded in it are not transferred to the final
replica, which is examined in the TEM.  Furthermore, these organic materials
have been found to scavenge fibers from suspension.  A common method for re-
moval of the organic materials is the low temperature plasma ashing technique,
as described in the EPA Interim Method.2  However, in many cases the filter
becomes blocked before a useful volume can be filtered, and then it is not
possible to analyze the sample satisfactorily.

A new technique has been developed for removal  of organic materials by the use
of 1% ozone gas and exposure to short wave ultraviolet light.15'16  Although
primarily developed for, and applied to, the analysis of beverages such  as
wine and beer, the technique has been successfully applied to a number of
industrial effluent samples and urine.  The interfering organic materials in
most water samples can be oxidized by treatment for a period of about 4  hours.
However, the technique does not attack bulk cellulose rapidly, thus other
methods must be used if cellulose is considered an interference.  The advantage
of the ozone-UV technique is that no acids or chemicals are added to the sample,
and the equipment is both simple and inexpensive.

A different and more powerful oxidation technique was also investigated  for
application to those samples where large amounts of cellulose or other similar
refractory organics are present.  In this technique the water sample is  exposed
to oxygen gas at high pressure and at an elevated temperature.  In neutral or
alkaline conditions,.complete oxidation takes place at about 13.8 MPa and
300°C, without the requirement for addition of other reagents.  Suitable high
pressure research reactors are available commercially.  It is estimated  that
2 samples* per day could be processed in one reactor.

The advantages of the oxidation methods proposed are that no chemicals are
added which might carry contamination, and that the procedures are very  simple
to perform., Using the ozone-UV technique, it has become relatively simple to .
analyze many liquid samples which, because of slow filtration, could not
previously.be prepared satisfactorily.

5.1  THE:OZONE-UV OXIDATION METHOD



                                     23

-------
               GAS-LINE DRYING J
               TUBE 8 FILTER
                                        OXYGEN SUPPLY
                                         CYLINDER
Figure  7.   Equipment used for the ozone-UV oxidation technique.
                                  24

-------
use for simultaneous oxidation of two samples is shown  in  Figure  8.   An
air extract or fume hood is required to remove surplus  ozone.   Before
being ozone-UV treated, the polyethylene or glass bottle containing  the
water sample is placed in an ultrasonic bath for a period  of 15 minutes.
The level of the liquid in the sample bottle is marked  on  the  outside,
using a felt marker.  The quartz pipets (formed by drawing quartz tubing
to the required length and tip diameter), are thoroughly washed before
each use, and installed on the ozone supply, as indicated, so  that the
tip is close to the bottom of the sample bottle.  The UV lamp  is  also
thoroughly washed and then immersed in the sample and switched on.
      Figure 8.  Ozone-UV oxidation of water samples in glass
                 bottles.  The ozone supply line has been split
                 into two lines to permit simultaneous oxidation
                 of two samples.  A valve and a filter holder are
                 incorporated in each of the supply lines to the
                 samples.


At an ozone concentration of about 4% in oxygen, each sample is treated
with about 1 liter/minute of gas for approximately 3 hours.   At other
ozone concentrations, the oxidation time should be adjusted so that each
sample receives about 10 grams of ozone.  The gas flow rate should be
sufficient to  produce a mixing action in the liquid but should not splash
any of the sample out of the container.  It is not-easy to indicate when
oxidation has  been completed, but this treatment, as described, has been
found to be adequate for all water samples so far analyzed.   When oxida-
tion is complete, the UV lamp and quartz pi pet are removed after which
the bottle is  re-capped and placed in an ultrasonic bath for a period of

                                 25

-------
     fifteen minutes.  This allows particulate released from the oxidized /
     organic materials and the container surfaces to be uniformly dispersed
     throughout the sample.  During the oxidation procedure, evaporation may
     result in a reduction of the sample volume.   This loss of volume should
     be noted and, if significant, can be accounted for by either addition of
     distilled water or calculation.  The sample should be filtered immediately
     after it is removed from the ultrasonic bath.

     An initial evaluation of the ozone-UV oxidation technique was made using
     a dispersion of refined Union Carbide Calidria chrysotile.  The organic
     material chosen for the test was a starch solution.   After ozone-UV
     oxidation of a 0.1% starch solution, it was  possible to filter 100 ml of
     the liquid.  However, it was found that even a 0.001% solution of starch
     in double-distilled water could not be filtered through a 0.1 ym pore
     size Nuclepore filter.  A volume of 4 ml of the Union Carbide chrysotile
     dispersion, which had a concentration of about 200 x 106  fibers/liter,
     was added to 96 ml of 0.1% starch solution.   Four samples were prepared
     in this manner, and the standard,ozone-UV technique  was applied, followed
     by treatment in the ultrasonic bath for 15 minutes.   The  results obtained
     are shown in Table 6.  A statistical  analysis shows  that at 5% signifi-
     cance, the ozone-UV treated sample results do not differ from those of
     the controls.

     Further qualification of the ozone-UV technique was  demonstrated using
     both artifically-prepared and naturally-occurring chrysotile dispersions.
     These results are discussed in the overall context of Section 6, which
     deals with sample container effects.

5,2  HIGH PRESSURE OXIDATION

     The feasibility of oxidation of organic materials by high pressure oxygen
     was tested using a stirred reactor.   For this purpose,  several  refractory
     organic dispersions in water were used.   These were  selected to be repre-
     sentative of materials which may be present  in water supplies,  rivers or
     lakes.  The oxidations were carried out at a pressure of  4.8 MPa and a
     temperature of 250°C.  The total pressure of 4.8 MPa comprised  an oxygen
     partial pressure of 1.0 MPa and 3.8.MPa of steam.  The reactor  was
     initially charged with oxygen to a pressure  of 0.7 MPa, and  during the
     reaction time of 90 minutes the sample was stirred at 800 rpm.   The
     experiments described below demonstrated that the procedure  was very
     successful.

     (a)  Proteins

          A slurry of ground beef, including the  fats,  was oxidized  by the
          high pressure oxygen procedure,  leaving a clear, yellow filterable
          liquid.   Ozone-UV oxidation was  unsuccessful  for oxidation of this
          sample.

     (b)  Cellulose

          A dispersion of cellulose was prepared  by ultrasonic dispersal  of

                                     26

-------
TABLE 6.  RESULTS OF ANALYSES OF CHRYSOTILE DISPERSIONS IN 0.1% STARCH SOLUTION
Sample
CONTROL SAMPLES
Control 1
Control 2
Control 3
OZONE-UV SAMPLES
Ozone- UV 1
Ozone-UV 2
Ozone- UV 3
Ozone-UV 4
i
Fiber Concentration
Mean
(106 Fib/Liter)

165
148
251

199
140
122
117

95% Confidence
Interval
(106 Fib/Liter).

87 - 244
101 - 194
182 - 320

157 - 241
110 - 170
96.8 - 146
81.8 - 152

Estimated Mass
Concentration
(Nanograms/Li ter)

2400
1720
3150

1610
1250
1640
893

Concentration
Equivalent to
1 Fiber
Detected
(106 Fib/Liter)

1.62
1.51
2.28

1.84
1.44
1.11
1.22

No.
Fibers
Counted

102
98
110

108
97
110
96

No.
Grid
Squares

9
9
6

4
5
7
6

Grid Distribution
X2

35.1
14.4
6.3

1.4
2.3
.4.6
6.7

Significance
of
Uniformity
W

<0.1
5
25

50
50
50
10


-------
     two sheets of tissue.  After the high pressure oxygen procedure a
     clear, colorless liquid remained which was filterable.  The ozone-UV
     technique was not effective for cellulose oxidation.

(c)  Raw Sewage

     Sewage samples have been analyzed previously using the ozone-UV
   "  technique. "The high pressure oxygen procedure was much more effec-
     tive, and yielded a clear liquid with inorganic sediment.

(d)  Urine

     Oxidation of the organic components of urine was only partially
     achieved, yielding a filterable liquid.which required additional
     treatment.  The ozone-UV method has been found effective for oxida-
     tion of urine, but it frequently yields a precipitate of calcium
     oxalate.

(e)  Cellulose Esters (Millipore Filter)                      ;

     A Mi Hi pore filter was dissolved in acetone and the mixture dispersed
     ultrasonically in water to give an emulsion of cellulose esters.   The
     high pressure oxygen procedure was successful  in oxidation of this
     emulsion, yielding a clear filterable liquid.   The ozone-UV method
     was completely ineffective.

(f)  Lake Water                                '

     Both the  high pressure oxygen and ozone-UV methods were effective  in.
     oxidation of the organics present in lake water, although  some more
     refractory organics are known to remain after the ozone-UV treatment.

For evaluation of the high pressure oxygen technique, an oscillating high
pressure reactor which.did not require a mechanical  stirrer was selected,
thus eliminating one source of possible contamination.   The equipment is
shown in Figure 9.  Initial experiments were performed using the oscilla-
ing high pressure oxygen reactor at conditions up to 300°C and  13.8 MPa.
The immediate  conclusion of the first experiment was that glass inner
containers cannot be used at temperatures above 200°C, since they are
attacked by water under these conditions.  It is possible that  other
container materials may be more resistant, but the  choices are  somewhat
restricted in  view of the high pressure oxygen conditions.   After a
preliminary test using silica tubing, silica containers were rejected
since a precipitate of silica was found after the treatment.  A custom-
designed Teflon^ container was eventually fabricated, and this  was found.'
to be satisfactory.

At the highest pressure and temperature used, namely 13.8 MPa and 300°C,
it was found that morphological changes occurred in  chrysotile.   Although
no change in the Mg/Si ratio was found, non-fibrous  debris was  generated
by the procedure and some of the fibers showed evidence of unscrolling.
                                28

-------
           Figure 9.  High pressure oxidation equipment.
Figures 10 and 11 show the comparison of the fiber morphologies before
and after the treatment.  The non-fibrous debris was found to contain
magnesium and silicon, giving rise to speculation that the fibers had
possibly reacted with the water under the extreme conditions.  It is
possible that reduction of the temperature and pressure would permit
oxidation of the organics to be achieved without significant degradation
of the fibers.  Since the amphiboles are generally less reactive than
chrysotile, it is likely that these minerals would be unaffected.

The high pressure oxygen procedure may be useful for very difficult
samples such as effluents with high concentrations of refractory organic
materials.  However, it was concluded that the ozone-UV method was cer-
tainly the superior technique for routine drinking water analyses, par-
ticularly in view of later observations on scavenging by the sample con-
tainer surfaces.  Accordingly, no further investigation of the high
pressure oxygen technique was undertaken.
                                29

-------
                                                                    O.Zjum
                             0.2 jum
Figure 10.   Union Carbide Calidria
            Chrysotile before high
            pressure oxidation
            treatment.
Figure 11.   Union Carbide Calidria
            Chrysotile after high
            pressure oxidation
            treatment.
                                     30

-------
       		___	SECTION.6	.;,	

          EXPERIMENTS ON THE STABILITY OF AQUEOUS FIBER DISPERSIONS


6.1  INITIAL OBSERVATIONS

     It has been known for some time that aqueous dispersions  of asbestos
     fibers can yield lower fiber counts  when  they are  analyzed after storage
     for long periods of time.   The reasons for this  instability of the dis-
     persions were unknown.   The question arises as to  why fiber dispersions
     apparently remain stable in a large  water body,  when samples collected
     from it and stored in bottles eventually  become  unstable.   Reasons which
     have been put forward for the reduction in concentration  with time range
     from complete dissolution of the fibers to "plating  out"  of fibers onto
     the interior surfaces of the container.  Complete  aqueous dissolution of
     silicate mineral fibers under normal temperature and pressure conditions
    'was considered unlikely, and the mechanism of the  suggested scavenging
     action by the container surfaces was not  clear.

     The stability of asbestos fiber dispersions in water was  of interest  from
     two points of view:

     (a)  it is important to know that the fiber concentration measurement
          obtained from a bottle of water is the same as  that  which would  have
          been obtained at the time of collection; and

     (b)  in a reference analytical method, stable fiber  dispersions of known
          concentrations are required for the  purposes  of analytical quality
          assurance and qualification of  individual analytical  laboratories.

     Accordingly, investigations were made into the factors which control  the
     stability of asbestos fiber dispersions in water.

     It has been established that, within a single laboratory,  replicate
     filters prepared from the same dispersion of a single fiber type can
     yield replicate results when analyzed using the  carbon-coated Nuclepore
    • procedure.7  The results of Section  4 have shown that the same is also
     true for chrysotile and amphibole fiber dispersions  of natural origin.
     Inter-laboratory analyses of filters, all prepared by one laboratory  from
     a single fiber dispersion, have also yielded comparable results.22
     However, inter-laboratory analyses of liquid samples have not in general
     yielded satisfactory results.6'11
                                     31

-------
Figures 12  and  13 show the results obtained when glass ampoules containing
dispersions of purified Union Carbide Calidria chrysotile at two diffe-
rent concentrations in double-distilled water were distributed to a
number of  laboratories experienced in TEM determination of asbestos fiber
concentrations.  It is clear that in addition to an unacceptable degree
of inter-laboratory variability, there was also a significant trend to-
wards lower concentrations when there were longer time intervals between
issue of the liquid samples and preparation of the analytical filters.
The results from the Ontario Research Foundation are identified as
triangles, and show that after an extended period of storage repeat
analyses using techniques identical in all respects did not always yield
values comparable to the initial fiber counts.  Figure 14 shows similar
results for a naturally-occurring dispersion of chrysotile (Lloyd Lake,
Ontario),  and it can be seen that the results also show a trend towards
lower values with increasing periods of storage.

A large number of these ampoules had been prepared, and consequently
material was available for studies to_determine the reasons for their
failure as reference standards.   Several ampoules were opened and the
contents filtered directly without the use of any ultrasonic treatment.
No chrysotile fibers were detected on these filters, indicating that the
chrysotile was no longer in suspension.   The empty ampoules were then cut
open so that the interior surfaces could be gold coated and examined in
the SEM.  An interesting feature emerged from this examination:  many
chrysotile asbestos fibers were found attached, to the glass surface, and
aJUL 0& tiiw were associated with some organisms or organic material  of a
gelatinous appearance.  This effect is illustrated in Figures 15 and 16.
It was evident that the asbestos fibers  were cemented to the interior
surfaces of the glass ampoules by this organic material.   The results of
the inter-laboratory analyses indicate that ultrasonic treatment of the
ampoules"before they were opened  was only partially successful  in re-
dispersing the fibers.  The origin of the organic material  in these
samples was unclear at this point..

In an initial experiment, a chrysotile asbestos dispersion  was  divided
into nine equal volumes of about 800 mL  each in one-liter polyethylene
bottles.  Samples from three of these bottles were filtered immediately
for analysis.  The remaining six bottles were treated as follows:

(a)  two bottles were shaken continuously in a reciprocal  shaker;

(b)  two bottles were exposed continuously to ultrasonic agitation in a
     bath; and

(c)  two bottles were allowed to stand on the bench,  followed by
     15 minutes in an ultrasonic bath before sub-sampling  the liquid
     from them.

One bottle from each treatment was removed for analysis  after a period.of
24 hours, and the second one after seven days.   Sub-samples  from each
group of three bottles were then filtered and the filters were  prepared
for TEM fiber counting.   The results  obtained are  shown  in Table 7.   It

                                32

-------
                      I
                      I
                      I 40
                      "s
                          A

                          "A
                                                          • 190 doyi
                         0             SO            IOO
                          INTERVAL BETWEEN AMPOULE PREPARATION AND ANALYSIS, doyi
Figure  12.   Results of inter-laboratory analyses using  aqueous dispersion of
             Union Carbide  chrysotile fibers  (high concentration).
                       i.
rf*.
                                                         -190 don
                                                       0 • 190 doyt
                                                       •—190 day.
                                       90            100
                         . INTERVAL BETWEEN AMPOULE PREPARATION AND ANALYSIS , doyi
Figure  13.  Results of inter-laboratory analyses using  aqueous dispersion  of
             Union Carbide  chrysotile fibers  (low concentration).
                                                          • 170 Ooy»
                                                       »—• ITOdoyi
                                                          200 an*
                         0             SO            IOO
                          INTERVAL BETWEEN AMPOULE PREPARATION AND ANALYSIS, doyt
Figure  14.  Results of inter-laboratory analyses using  a  naturally-occurring
             aqueous dispersion of  chrysotile  fibers.   (Lloyd Lake,  Ontario).
                                          33

-------
                                                   1.0 jum
Figures 15 and 16.  SEM micrographs showing asbestos fibers attached by
                    organic material to the inside surface of a glass
                    container.
                                   34

-------
 is immediately evident that continuous shaking removed all  of the fibers
 from suspension within 24 hours,, whereas after seven days in an ultrasonic
 bath the results were essentially unchanged.   In static storage, followed
 by ultrasonic treatment for 15 minutes prior to filtration, the concentra-
 tion was reduced by about a factor of 2 after seven days.  The curious and
 unexpected results for continuous shaking were thought to be due to collec-
 tion of fibers by the inside surfaces of the containers, and it was con-
 sidered that re-suspension might be achieved by ultrasonic treatment.
 The bottles, which had been continuously shaken for 24 hours and seven
 days respectively, were placed in an ultrasonic bath for a period of 30
 minutes, and it was found that the re-suspended fiber concentrations were
 less than 10% of the initial value.  This result was very disturbing,  in
 view of the fact that the degree of agitation experienced by a water
 sample before it is analyzed cannot be controlled.   A second experiment
 was conducted in which the fiber dispersions were prepared in a 0.005
 Molar solution of sodium pyrophosphate (Na^Oy).  This material is a
 recommended dispersal agent for silicate mineral particles.21*  The results
 obtained were essentially the same as those obtained without the sodium
 pyrophosphate:  continuous shaking completely removed the fibers from
 suspension, whereas during static storage or continuous ultrasonic treat-
 ment the fibers mostly remained in suspension.

   TABLE 7.  STABILITY OF VERY DILUTE CHRYSOTILE FIBER DISPERSIONS

             (Fiber concentrations in 106 fibers/liter)
Time
Initial .
24 hours
7 days
Treatment
Continuous Shaking
107
<0.6
<0.6
Continuous Ultrasonic
100
Not Analyzed
78
Static Storage
with Routine
Preparation
117
75
51
Experiments were performed with polyethylene bottles  containing equal
volumes of the same dispersion, which contained approximately
3000 ng/liter of Union Carbide chrysotile.
 In a separate and parallel  study, experiments were  conducted  in order  to
 assess the effects of pH and polyvalent ions on  the stability of  aqueous
 dispersions of chrysotile.   Naumann and Dresher25 reported  that the
 stability of concentrated chrysotile dispersions is enhanced  by addition
                                  35

-------
of polyvalent ions to the dispersion at particular values of pH, and it
was thought that the same technique might be useful  at low fiber concent-
rations.  The following ionic and pH conditions were used:

(a)  aqueous dispersion adjusted to pH 4.0 using acetic acid;

(b)  0.005 molar ferric chloride adjusted to pH 4.0  using acetic
	acid; and	  *  	  	

(c)  0.005 molar zirconyl chloride adjusted to.pH 4.0 using acetic acid,

The results obtained are shown in Table 8, and were  similar to those
obtained previously:  continuous shaking for 7 days  removed most of the
fibers from suspension, whereas during static storage for the same period
most of the fibers remained in suspension.  These results indicated that
none of the wide range of conditions used appeared to affect significantly
the final result:  shaking of the suspension removed fibers almost
completely, presumably to the interior surfaces of the plastic container.
        TABLE 8.  SHAKING OF CHRYSOTILE FIBER DISPERSIONS

               UNDER SELECTED IONIC AND pH CONDITIONS

             (Fiber Concentrations in 106 fibers/liter)

lomc and
pH Conditions


Aqueous Dispersion,
Adjusted to pH 4.0
by addition of
acetic acid
0.005M FeCl3,
Adjusted to pH 4.0
by addition of
acetic acid
0.005M,ZrOCl2,
Adjusted to pH 4.0
by addition of
acetic acid

Time



Initial

7 days

Initial

7 days

Initial

7 days

Treatment


Continuous Shaking

50

0.6

41

10

33

2.6


Static Storage
with Routine
Preparation
50

37 .

41

33

33

20

                                36

-------
'EFFECTS "OF BIOLOGICAL  ORGANISMS  ON  SUSPENSION  STABILITY

Throughout the work  described  in Section  6.1,  no  particular effort  had
been made to  maintain  sterility  in  the  suspensions,  and  it was  at this
point that the very  important  part  which  bacteria and  their decomposition
products  play in the stability of dilute  dispersions of  chrysotile  asbes-
tos fibers was recognized.   The  importance  of  the electro-kinetic effects
of chrysotile in-removal  of polysaccharides is well  known in  the beverage
and pharmaceutical industry,26'" and this  lends  support to the view that
these organic materials may scavenge  chrysotile fibers from suspension
and subsequently adhere to  the container  walls.

New dispersions of Union  Carbide chrysotile, UICC Canadian chrysotile and
UICC crocidolite were  prepared,  using sterilized  glass containers and
double-distilled water taken directly from  the condenser outlet of  a
glass still.   The dispersions  were  immediately transferred to 50 mL glass
ampoules, and these,  were  flame sealed.  The ampoules were then  autoclaved
at 121°C  for  30 minutes in  order to sterilize  the contents.   All of the
preparation for each fiber  type  was completed  within one day.   Using this
procedure, few bacteria or  their decomposition products  could be present
in the dispersions,  and any organisms which may have accumulated during
handling  would be destroyed, so  that  no biological activity should  exist
in the final  ampoules. Table  9  shows the results obtained from analyses
of 10 mL  samples from  the ampoules  of Union Carbide  chrysotile  dispersion.
 It can be seen that  even  after storage  for  a period  of sixty  days,  the
fiber count remained unchanged,  and that  constant results were  obtained
without the use of ultrasonic  treatment.  The  dispersions also  showed no
change after  continuous shaking  for periods of up to seven days.  However,
.when an ampoule was  opened  and the  contents contaminated with unsterile
distilled water before being re-sealed, it  was found that very  variable
 fiber counting results could be  obtained  after shaking.  However, the
nature and concentration  of the  organisms added were not under  good
 control.   The results  of  similar work using UICC  chrysotile are shown in
Table 10.  The sterile suspensions  remained unchanged  for long  periods,
 but when  they were contaminated  with  water  containing  active  organisms
 the results were again unsatisfactory.  In  particular, the unsterile
 ampoule shaken for seven  days  yielded a very inhomogeneous filter,  and
 the chrysotile fibers  were  found to be  strongly aggregated and  attached
 to organic debris.   In this sample  a  large  mass of organic material  con-
 taining 20 fibers was  found on one  grid opening,  resulting in a very
 large confidence interval for  the calculated concentration value.

 Although  an extended study  using crocidolite was  not made, the  results
 shown in  Table 11 indicate  that  stable  dispersions of  UICC crocidolite
 can also  be made if  biological material is  excluded  from the  preparation.

 It has now become clear that much of  the  variability associated with inter-
 laboratory analyses  of aqueous dispersions  of  chrysotile fibers was  due
 to the presence of biological  organisms,  the decomposition products  of   .
which scavenged the  fibers  and subsequently adhered  to the surface  of the
 containers.  The fiber concentration  measurement  obtained was then  a
 function  of the extent to which  this  process had  already occurred and the

                                 37

-------
 TABLE 9.  STABILITY AND STORAGE OF STERILE UNION CARBIDE
               CHRYSOTILE FIBER DISPERSIONS
      (All concentration values in 106 fibers/liter)
Sample
Analysis at initial ampoule preparation
After continuous shaking for 24 hours
After continuous shaking for 7 days
After 60 days storage, no ultrasonic
treatment used
Unsterile ampoule shaken for 24 hours
Unsterile ampoule shaken for 7 days
Unsterile ampoule shaken for 4 days
Mean
82.6
73.5
69.5
76.8
80.0
29.4
16.7
62.4
95% Confidence
Interval
44.5 - 121
47.1 - 99.9
51.5 - 87.5
55.2 - 98.5
61.3 - 98.7
15.0 - 43.8
4.4 - 29.0
38.8 - 86.4
TABLE 10.  STABILITY AND STORAGE OF STERILE UICC CHRYSOTILE
                    FIBER DISPERSIONS
      (All concentration values in 106 fibers/liter)
Sample
Analysis at initial ampoule preparation
After continuous shaking for 24 hours
After continuous shaking for 7 days
After 52 days storage, no ultrasonic
treatment used
Unsterile ampoule shaken for 24 hours
Unsterile ampoule shaken for 7 days
Unsterile ampoule shaken for 4 days
Mean
19.1
27.9
25.1
24.8
22.4
29.4-
38.8
6.4
95% Confidence
Interval
11.1 - 27.0
18.2 - 37.6
15.9 - 34.3
12.5 - 37.0
15.9 - 28.9
21.0 - 37.8
0 - 77.7
3.3 - 9.6
                           38

-------
              TABLE 11.   STABILITY  AND  STORAGE  OF STERILE  UICC

                        CROCIDOLITE FIBER DISPERSIONS

               (All  concentration values  in  106 fibers/liter)
Sample
Analysis at initial ampoule
preparation
After 30 days storage, no ultrasonic
treatment used __
Mean
28.9
34.7
29.8
95% Confidence
Interval
15.4 - 42.4
26.9 - 42.5
23.2 - 36.5
     degree to which any ultrasonic treatment was  able  to  detach  and  redisperse
     the fibers.  "If biological  organisms  were excluded, and  absolute sterility
     was subsequently maintained,  standard dispersions  of  chrysotile  and
     crocidolite could be prepared which appeared  to  be stable  for  long periods
     of time.   These observations  on the effects of organic materials also
     indicate  that the. use of ultrasonic treatment on water samples prior to
     sub-sampling for analysis is  absolutely essential  if  reliable  and repro-
     ducible results are to be obtained.

6.3  PREPARATION AND PRESERVATION  OF REFERENCE FIBER  DISPERSIONS

     Asbestos  fiber dispersions  were prepared and  sealed in ampoules  so that
     stable dispersions of known fiber concentration  were  available .for the
     balance of the experimental program.   Both flame-sealed  borosilicate
     glass ampoules, and TeflonR-capped serum bottles  are satisfactory con-
     tainers in which sterility  can be maintained.  The procedure for prepara-
     tion of the dispersions is  described  in Steps  (a)  to  (e). .
          a)  The weight of asbestos  required  to yield the desired
              mass concentration  is calculated.      -

          b)  The asbestos  is weighed accurately and dispersed in a
              small  volume  of freshly-distilled water by grinding it
              gently in an  agate  mortar and  pestle.      _..  .  .-_  __
          c)  The slurry is then dispersed  in  a  large  volume of freshly-
              distilled water in a glass  container which has been sterilized
             .by ozone-ultraviolet light  treatment.  The container is then
              sealed using a plastic  stopper,  and it is treated in an ultra-
              sonic bath for a period of  about 30 minutes.

          d)  The suspension is shaken vigorously, and then the required
              volumes are pipetted into washed ampoules or serum bottles.
                                     39

-------
          e)  The ampoules or bottles  are  sealed,  labelled and autoclaved for
              30 minutes at 121°C to sterilize the contents.

     Several ampoules selected from the  series, including  the first and last
     prepared, should then be analyzed to  obtain the  precise  numerical  fiber
     concentration.   These analyses should be  conducted very  carefully, and a
     minimum of about 200 fibers  should  be counted in each so that the  size
     distribution can be defined  accurately.   It has  been  found in practice
     that for size distributions  characteristic-of water-borne asbestos,
     standards having particular  numerical fiber concentrations can be  pre-
     pared by assuming that UICC  chrysotile and UICC  crocidolite contain
     approximately 2 x 10llt fibers/g and 4 x 1012  fibers/g respectively.

6.4  CONTAINER AND STORAGE EFFECTS ON  FIBER DISPERSIONS

     The observation that laboratory-prepared  aqueous dispersions of chrysotile
     were de-stabilized by small  amounts of biological  material, raised ques-
     tions about the effects of container materials and storage on environ-
     mental samples collected in  the field.  Drinking water usually contains
     significant amounts of biological materials,, and. in view of the effects
     discussed in Section 6.3, it appears   that the type of containers  used
     and the storage conditions may have a profound effect on the fiber con-
     centration results obtained  from  field samples.

     Previous work28 on the effects of containers  and storage conditions  on  '
     laboratory-prepared fiber dispersions in  lake waters  demonstrated  that:

          a)  there was a large increase in the amount of  skeletal
              diatomaceous material during storage in glass bottles
              in illuminated conditions;   .      ...

          b)  samples stored in polyethylene bottles  did not  show a
              large increase in diatomaceous material;

          c) - after ultrasonic treatment of the bottle before sub-
             .sampling, and using the  analytical  techniques of the  .
              time, no significant reduction in the suspended fiber
              concentrations was   detected after storage of the
              samples for 8 months.

     The ability to collect water samples  from a source,and to transport  them
     without significant changes  to a  laboratory for  analysis,  is fundamental
     to all existing analytical techniques for the. determination of asbestos
     fiber concentrations.  Therefore, a study was made  in which both
     laboratory-prepared and naturally-occurring fiber dispersions  were
     placed in different types of containers and transported  under realistic
     conditions to the laboratory.

     Concentrated sterile dispersions  of Union Carbide  Calidria  chrysotile, .
     UICC Canadian chrysotile, and UICC  crocidolite were prepared in sealed
     glass ampoules so that they  could be  transported and  used  to prepare
     samples at the location of the naturally-occurring  dispersion.  The


                                     40

-------
 naturally-occurring fiber dispersion  selected for this  study was  the
 drinking.water of Sherbrooke,  Quebec, which  is known  to contain a high
 concentration of chrysotile fibers.   Four types of one-liter container
 were investigated; flint glass,  borosilicate glass, polyethylene  and
 polypropylene.

 At the sampling site the ampoules of  the sterile dispersions were opened,
 and diluted dispersions were prepared in the four types of container  by
 adding 20  ml of the concentrate  to 400 ml of double-distilled water.
 Nuclepore  filter controls were prepared at the same time from samples of
 the sterile dispersions.  Samples of  the natural dispersion were  collected
 directly from the faucet into the four types of container.   Since it  was
 thought that the fiber scavenging effect of the containers may be a
 function of fiber concentration, additional  samples of the natural dis-	
"pension were "also dlVuteeQ:3"injlpuble-di_stilled water before they were
 pia'ced"in  the containers.  Care  was taken to ensure "that changes  in" the
 fiber concentration of the source during sampling could not significantly
 affect the results of the study.  Each sample was collected directly
 into the particular container, and a  small volume withdrawn immediately
 •for filtration through- a- 0.1 ym  pore  size Nuclepore filter.  The
 Nuclepore  filter controls were analyzed later to obtain the fiber concen-
 trations of the dispersions before any transport or storage effects had
 occurred.

 The results obtained for the Union Carbide chrysotile dispersion  are
 shown in Table 12.  It can be seen that after transport and storage for
 3 days, followed by the usual treatment of fifteen minutes in an  ultra-
 sonic bath.before sub-sampling for analysis, the results obtained were
 statistically identical with those obtained by analysis_of. the control
 filters made at the time of preparation of the samples.  All of  the
 types of container yielded the same satisfactory results.

 The samples were then stored for another 36 days, after which they were
 shaken on a laboratory shaker for 2 days.  Sub-samples were taken for
 analysis, and it was found that  the suspended fiber concentrations in
 the plastic containers had been  reduced to the detection limit of the
 measurements.  This is a result consistent with the earlier work.  The
 fiber concentrations remaining in suspension in the glass containers  had
 been reduced to about 30/6 of the original values, indicating that the
 fiber scavenging mechanism of glass surfaces was less effective.   It  was
 decided that if, as suspected, the fibers were attached by organic
 material to the inside surfaces  of the containers, ozone-UV treatment
 of  the sample in the storage container, for about 4 hours, could  oxidize  .
 the organic material and release the fibers into suspension.  In  Table 12
 it  can be seen that this was the case, and the most satisfactory  result
 was obtained for the polyethylene container.  Although the flint  glass
 and polypropylene containers yielded results showing  a similar pattern,
 the final  results obtained were significantly lower than the fiber con-
 centration of the original dispersion.

 The results of the study using UICC chrysotile are shown in Table 13.
 The results obtained were similar to those using the  Union Carbide

                                 .41

-------
    TABLE 12.   STORAGE  AND CONTAINER STUDY:   UNION CARBIDE  CHRYSOTILE

              (All  concentration values  in 106  fibers/liter)*

Sample Treatment


Water filtered at time
of preparation of
diluted samples in bottles
Storage for 3 days
followed by 15 minute
treatment in ultrasonic bath
Additional storage for 36 days
followed by 48 hours on
laboratory shaker. No .
further ultrasonic treatment
i
Additional storage for 83 days
followed by ozone-ultraviolet
treatment and 15 minutes
ultrasonic agitation in bath
Container Material
Flint Glass
Mean
"

76.8


83.3


30.7



49.7


95* Confidence
Interval

55.2 - 98.5


61.1 - 106


17.9 - 43.3



27.6 - 71.7


Polypropylene
Mean


-


80.4


0.6



40.7


955S Confidence
Interval

Not Analyzed


63.8 - 97.0


0 - 1.94 •



25.5 - 55.8


Borosilicate Glass
(Ground Glass Stopper)
Mean


80.0


85.3


26.0



-


95% Confidence
Interval

61.3 - 98.7


66.2 - 104


19.8 - 32.1



Not Analyzed


Polyethylene
Mean


-


78.5


<0.5



65.0


95* Confidence
Interval

Not Analyzed


54.0 - 103


-



55.0 - 74.5


*A11  concentration values -refer to the fiber concentration in  the original undiluted suspension.
 Analysis of the undiluted suspension at the time of the ampoule preparation yielded a
 concentration  value of 82.6 x 106 fibers/liter, with a 95% confidence interval  of
 (44.5 - 121) x 106 fibers/liter.

-------
CO  I
    I
                           TABLE  13.   STORAGE AND CONTAINER STUDY:   UICC CHRYSOTILE

                                 (All  concentration values  in 106 fibers/liter)*
Sample Treatment
Mater filtered at time
of preparation of
diluted samples In bottles
Storage for 3 days
followed by 15 minute
treatment In ultrasonic bath
Additional storage for 56 days
followed by 48 hours on
laboratory shaker. No
further ultrasonic treatment
Additional storage for 89 days
followed by ozone-ultraviolet
treatment and 15 minutes
ultrasonic agitation 'in bath
Container Material
Flint Glass
Mean
24.8
27.3
7.7
31.3
95% Confidence
Interval
12.5 - 37.0
16.7 - 37.8
2.9 - 12.5
21.6 - 41.1
Polypropylene
Mean
-
23.9
4.5

952 Confidence
Interval
Not Analyzed
18.8 - 29.2
- 0.6 - 8.3
Not Analyzed
Boroslllcate Glass
(Ground Glass Stopper)
Mean
22.4
25.2
5.0
37.5
95% Confidence
Interval
15.9 - 28.9
18.4 - 32.0
0.1 - 9.9
21.7 - 53.3
Polyethylene
Mean
-
22.3
<0.6
36.6
95% Confidence
Interval
Not Analyzed
17.1 - 27.6
-
26.9 - 46.1
                     *A11 concentration values refer to the fiber concentration 1n the original undiluted suspension.
                      Analysis of the undiluted suspension at the time of the ampoule preparation yielded a concentration
                      value of 19.1 x 106 fibers/liter, with a 95X confidence interval of (11.2 - 27.0) x 106  fibers/liter.

-------
chrysotile.  Storage of the samples for 3 days, followed by ultrasonic
treatment, yielded values which were statistically compatible with the
original value for the dispersion.  Long term storage, followed by shaking
for 2 days, reduced the suspended fiber concentrations significantly.  In
the case of the polyethylene container the concentration was reduced to
the detection limits of the measurement.  In the flint glass, borosilicate
glass and polyethylene containers, it was found that the ozone-UV treat-
ment restored the suspended fiber concentrations to values which were
statistically compatible with the original value for the dispersion'.

It was originally thought that the fiber scavenging effect by the con-
tainer surfaces would be specific to chrysotile, since fibers of this
mineral carry a surface charge which is of an opposite sign from that
carried by particles of most other mineral species.  If this were the
case the same behavior would not be expected for crocidolite.  The results
of the container study using UICC crocidolite are shown in Table 14, and
it can be seen that the pattern is very similar to that obtained for
chrysotile.  The fiber scavenging effect of the container surfaces was
very pronounced for the plastic containers, and the ozone-UV treatment
was effective .in restoring the fibers to suspension.

The fiber scavenging effect of the container surfaces was also demonstrated
for Sherbrooke, Quebec, drinking water.  Table 15 shows the results
obtained for the first series of samples.  The three Nuclepore filters
prepared at the time of sample collection showed very consistent results
of 64.6 x 106, 63.4 x 106, and 66.4 x 106 fibers/liter.  After storage
for 40 days the containers were shaken to re-di'sperse settled particulate,
and a sample from each was analyzed.  The results obtained were signifi-
cantly lower than those from the control filters, and even the use of
ultrasonic bath treatment in the normal preparation technique did not
yield fiber counts higher than about 50% of the original values.  Subse-
quent continuous  shaking for 48 hours produced some additional increase
in fiber concentration for the glass container, but reductions occurred
for the two plastic containers.  After storage for 15 more days the
samples were given ozone-UV light treatment for 3 hours without removing
them from the containers, followed by mild ultrasonic agitation.in a bath
for 15 minutes.  After this treatment the fiber concentrations were res-
tored to values statistically indistinguishable from those originally
observed at the time of collection.  These results were parallel to those
obtained using laboratory-prepared dispersions in that shaking appeared
to enhance the scavenging of fibers by the walls of the plastic containers.
The fibers scavenged by the container walls could then be returned to
aqueous suspension by ozone-UV treatment.

Results from the second series of samples of Sherbrooke drinking water
are shown in Table 16.  In this case the Sherbrooke water was diluted 1:3
with double-distilled water before placing in the containers for trans-
port and storage.  The pattern of results was similar to that of the first
series, except that the scavenging effect was much more effective.   The.
conclusion could be drawn that either the magnitude of the effect was
dependent on fiber concentration, or that some agent in the double-
distilled water enhanced it.

                                 44

-------
                          TABLE 14.   STORAGE  AND CONTAINER STUDY:   UICC CROCIDOLITE

                                (All  concentration  values in  106  fibers/liter)*

Sample Treatment


Mater filtered at time
of preparation of
diluted samples in bottles
Storage for 3 days
followed by 15 minute
treatment in ultrasonic bath
Additional storage for 49 days
followed by 48 hours on
laboratory shaker. No
further ultrasonic treatment
i
Additional storage for 96 days
followed by ozone-ultraviolet
treatment and 15 minutes
ultrasonic agitation in bath
Container Material
Flint Glass
Mean


34.7


31.7


27.8


36.5



95% Confidence
Interval

26.9 - 42.5


24.2 - 39.3


18.6 - 37.1


26.7 - 46.3



Polypropylene
Mean


-


30.9


<1.1


' -



95% Confidence
Interval

Not Analyzed


22.7 - 39.1


•


Not Analyzed



Borosilicate Glass
(Ground Glass Stopper)
Mean


29.8


15.4


4.4


30.6



95% Confidence
Interval

23.2 - 36.5


10.7 - 20.2


0.2 - 8.6


19.7 - 41.5



Polyethylene
Mean


-


35.3


0.7


25.2



95% Confidence
Interval

Not Analyzed


27.3 - 43.3


0 - 2.3


13.4 - 37.0



01
                      *A11 concentration values.refer to the fiber concentration in the original undiluted suspension.
                       Analysis of the undiluted suspension at the time of  the ampoule preparation yielded a concentration
                       value of 28.9 x 106  fibers/liter, with a 95% confidence interval of (15.4 - 42.4} x 106 fibers/liter.

-------
   TABLE 15.   STORAGE  AND  CONTAINER  STUDY:   SHERBROOKE UATER (1st  SERIES)1

             (All  concentration  values  1n  106  fibers/liter)*
Sample Treatment
Water filtered on site
Storage for 40 days,
hand shaken before
sub-sample removed
Additional 15 minute
treatment in ultrasonic bath
Additional 48 hours on
laboratory shaker. No
further ultrasonic treatment
Additional storage for 15 days,
followed by ozone-ultraviolet
treatment and agitation in
ultrasonic bath for
15 minutes
Container Material
Flint Glass
Mean
64.6
15.3
26.9
35.5
73.2
95% Confidence
Interval
38.7 - 90.5
11.3 - 19.3
20.7 - 33.0
31.1 - 40.0
57.9 - 88.4
Polypropylene
Mean
63.4
10.4
20.7
15.2
73.0
95% Confidence
Interval
51.6 - 75.2
8.4 - 12.5
16.2 - 25.2
12.2 - 18.3
62.0 - 84.0
Polyethylene
Mean
66.4
22.5
37.3
26.9
62.9
95% Confidence
Interval
48.3 - 84.5
15.0 - 29.9
31.0 - 43.6
20.8 - 33.0
36.6 - 89.1
*A11 concentration values in 106 fibers/liter refer to the fiber concentration
 in the original undiluted suspension.

-------
  TABLE 16.  STORAGE AND CONTAINER STUDY;  SHERBROQKE MATER  (2nd SERIES)
                               I
             (All concentration values in 106 fibers/liter)*
Sample Treatment
Water filtered on site
Storage for 104 days,
hand shaken before :
sub-sample removed
Additional 15 minute
treatment in
ultrasonic bath
Additional 48 hours on
laboratory shaker. No
further ultrasonic treatment
Additional storage for 3 days,
followed by ozone- ultra violet
treatment and agitation in
ultrasonic bath for
15 minutes
Container Material
Flint Glass
Mean
65.0
10.4
25.3
30.5
74.5
95% Confidence
Interval
47.?'- 82.3
7.2 - 13.5
21.0 - 29.7
16.8 - 44.1
61. 0'- 88.5
Polypropylene
Mean
42.4
5.9
21.3
7.4
43.7
95% Confidence
Interval
31.9 - 52.8
3.7 - 8.1
16.8 - 25.7
6.0 - 8.8
35.5 - 52
Polyethylene
Mean
49.1
7.8
18.1
4.0
53.0
95% Confidence
Interval
33.9 - 64.3
4.3 - 11.3
14.1 - 22.1
2.0 - 6.1
33.4 - 72.5
*A11 concentration values in 106 fibers/liter refer to the fiber concentration
 in the original undiluted suspension.


-------
     In the third series  of Sherbrooke water samples  no  dilutions  were made.
     The tests  made on  this series  did not incorporate the  extended  mechanical
     shaking step, a procedure which  would not normally  be  a  component of
     routine sampling and analysis.   The  results  are  shown  in Table  17.   It
     can be seen  that after storage for only 40 days, very  low fiber concen-
     tration values were  obtained when the samples were  simply hand-shaken
     before sub-samples were removed.  Treatment  of the  containers for 15
     minutes in an ultrasonic bath  elevated the fiber concentrations to about
     30 - 50% of  the value obtained at the time of collection.   After storage
     for a further period of 41 days,  the samples were treated with  ozone-UV
     light and  agitated for 15 minutes in an ultrasonic  bath.   The fiber con-
     centrations  were then found to be statistically  indistinguishable from
     that obtained from the Nuclepore control  filter  prepared at the time of
     sample collection  81 days earlier.

     The fiber  length distributions at the various stages of  the Sherbrooke
     water studies were found to be unchanged  from those of the initial  fiber
     dispersions  determined from the  Nuclepore filter controls.  The tabulated
     fiber length distributions are shown in Appendix C.

     The overall  conclusions concerning container and storage  effects  were  as
     follows.

          a) Plastic containers scavenged more fibers than glass
             containers.                                               4.:.

          b) Samples stored for more than a few  day's yielded  very low
             fiber concentration values.

          c) Ultrasonic  treatment  by itself was  not  able to  re-disperse
             all of the  fibers scavenged by the  container  surfaces.

          d) The ozone-UV technique,  combined with ultrasonic treatment,
             was able  to re-disperse the fibers  scavenged  by  the  container
             surfaces  such that the  suspended fiber  concentration was res-
             tored to  its original value at the  time of collection.

          e.) Fibers from laboratory-prepared  dispersions were  scavenged
             more completely by the  container surfaces  than  those from
             naturally-occurring dispersions.

          f) The fiber scavenging  effect did  not appear to be  size-
             selective.

6.5  INVESTIGATION OF FIBER SCAVENGING MECHANISM

     Limited investigations were carried  out in order to determine the agents
     responsible  for the  fiber scavenging by container surfaces.   The  effect
     was considered to  be an important one, since it  could  potentially be
     developed  into a separation technique specific for  chrysotile asbestos
     fibers. Moreover, if the consequences of the effect are  to be  controlled
     in routine sample  handling, a  full understanding of the  mechanism is re-
     quired.
                                     48        	

-------
TABLE 17.  STORAGE AND CONTAINER STUDY:  SHERBROOKE HATER (3rd  SERIES)

          (All concentration values in 106 fibers/liter)*
Sample Treatment
Water filtered on site
Storage for 40 days,
no shaking before
sampling by pi pet
Additional 15 minute
treatment in
ultrasonic bath
Additional storage for 41 days,
folldwed by ozone- ultra violet
treatment and agitation in
ultrasonic bath for
15 minutes
Container Material
. , Flint Glass
Mean
73.5
17.2
39.9
68.6
95% Confidence
Interval
58.2 - 88.9
13.0 - 21.4
30.7 - 49.1
38.0 - 99.2
Polypropylene
Mean
7
5.1
15.8
94.3
95% Confidence
Interval
Not Analyzed
3.5 - 6.6
12.3 - 19.3
53.4 - 135
Polyethylene
Mean
-
15.0
25.7
62.8
95% Confidence
Interval
Not Analyzed
12.2 - 17.8
17.2 - 34.2
60.8 - 64.8
 *A11  concentration  values  in  106  fibers/liter refer  to  the fiber concentration
  in  the  original  undiluted suspension.

-------
The earlier Observations on the stability of fiber dispersions indicated
that the container itself did not appear to be responsible for the fiber
scavenging effect, and that the presence of some biological  agent or
organic material was required.  Moreover, it was found that plastic sur-
faces scavenged more fibers than glass surfaces.  These observations are
consistent with work by Fletcher2^ on bacterial  attachment,  in which it
was found that the number of bacteria which became attached to surfaces
immersed in water  depends on-the nature of the surfaces;  - In particular,
it was found that much lower numbers of bacteria attached  to surfaces such
as glass or mica, than was the case for plastic surfaces such as poly-
ethylene and polystyrene.  It was also shown that RF plasma treatment of
the plastic surfaces reduced the numbers of bacteria which attached to
them.  A possible mechanism for the fiber scavenging effect by sample
container surfaces could be that the fibers become initially attached to
the organic component or bacteria, which subsequently adheres to the con-
tainer surface.  Alternatively, it could be that the organic component
coats the interior of the container such that, fibers which contact the
surface then adhere to it.  In either case, mechanical agitation would
assist the process by making contact with the container walls a more
frequent event.  Further experiments were conducted to elucidate the
mechanism further.

The initial approach taken was to attempt identification of the organisms
present in the non-sterile double-distilled water used for the container
studies and in the stored Sherbrooke water samples.   Standard plate counts
were made to determine the total number of viable organisms  present in
the water samples.  The double-distilled water was found to  contain
2.1 x 107 organisms/liter and the Sherbrooke water 8.6 x 106 organisms/
liter.  Swabs taken from the inside walls of the containers  also showed
the presence of some fungi.  The predominant viable organism in both
water samples was Pseudomonas spp..  Typical colonies of the  micro-
organisms were isolated and cultured in the laboratory.  A washed sus-
pension of the cultured micro-organisms was then employed  in further fiber
separation experiments.

The first series of experiments was  designed to determine whether the
cultured micro-organisms were effective in attachment of chrysotile
fibers to the inside surfaces of polyethylene containers.  About 800 ml
of each water sample was placed in a one-liter polyethylene  container, to
which was added 50 ml of a sterile suspension of Union Carbide chrysotile.
The container was then mechanically shaken for 60 hours  at a frequency
of about 2.5 Hz.  The water was then transferred to an identical  new
polyethylene bottle, and given ozone-UV treatment in the usual manner,
followed by exposure to ultrasonic agitation in  a bath for 15 minutes.
A sample was then filtered and analyzed.  The original empty polyethylene
bottle was re-filled with 850 mL of double-distilled water,  and ozone-UV
treatment was given to this sample, followed by  15 minutes in an ultra-
sonic bath.  A sample was then filtered and analyzed.  The results of
the 6 experiments are shown in Table 18.  It can be seen that the un-
filtered, unsterile, double-distilled water (Experiment 1) yielded
attachment of 84.5% of the chrysotile fibers to  the container, a high
value which is consistent with the earlier work.  A second sample of the

                                50

-------
TABLE 18.  INITIAL FIBER SEPARATION EXPERIMENTS
        USING UNION CARBIDE CHRYSOTILE
Experiment
1. UnfUtered,
un'sterile
double-distilled
water
2. Unsterile
double-distilled
water, filtered
through 0.2 ym
Nuclepore filter
3. Unfiltered
Mississauga
tap water
4. Unfiltered
Mississauga
tap water with
addition of
107 viable
organisms/ liter of
Pseudomonas spp.
5. Filtered
Mississauga
tap water
6. Filtered
Mississauga
tap water with
addition of
107 viable
organisms/liter of
Pseudomonas spp.
Fiber Concentration
Remaining Suspended
(106 fibers/liter)
20.8
17.6
100
61.7
0.8
55.3
Separated Fiber
Concentration
(106 fibers/ liter)
113
134
39
45.5
173
116
<
Proportion of
Fibers
Separated, %
84.5
88.4
28.1
42.4
99.5
67.7
                      51

-------
unsterile, double-distilled water was filtered through a 0.2 vm pore size
Nuclepore filter before the experiment was performed, (Experiment 2).
In this experiment 88.4% of the chrysotile .fibers became attached to the
polyethylene container, indicating that the active agent was capable of
penetrating a 0.2 urn pore size filter.  The experiment using unfiltered
Mississauga tap water (Experiment 3), yielded an attachment of only 28.1%,
which was not greatly increased in Experiment 4 by the initial  addition
of 107 viable micro-organisms/liter.of the Pseudomonas spp culture.
Filtered Mississauga tap water by itself (Experiment 5), yielded an
almost complete attachment of 99.5% of the fibers, which appeared to be
suppressed to 67.7% by initial addition of the cultured micro-organisms,
(Experiment 6).  The results, therefore, indicated that the active agent
was present in Mississauga tap water, that it was capable of passing
through a 0.2 um pore size filter, and that the cultured micro-organisms
appeared to be ineffective.

A second series of experiments was performed using the same procedure,
firstly as a partial repeat of the first study, and secondly to extend  it
to UICC crocidolite and UICC amosite dispersions.  The results  for Union
Carbide chrysotile are shown in Table 19.   In Experiment 1, a high attach-
ment of >99.6% was demonstrated using unsterile, double-distilled water.
In Experiment 2, sterile conditions were maintained as efficiently as
possible, using water directly from the output of the still.   In this case
only 5.9% of the fibers were observed to attach.to the container walls.
In Experiment 3, the sterile, double-distilled water was used again, but
with the addition of 107 viable micro-organisms/liter of the culture.   No
significant attachment of fibers was observed as a consequence  of the
addition.  In Experiment 4, addition of the cultured micro-organisms was
made to Mississauga tap water, and the attachment of- 75.5% agreed with
that observed in the initial experiment (67.7%).   Although the  actual
proportions, of attached fibers were generally lower, the overall patterns
of the results for crocidolite and amosite were similar to that observed
for chrysotile.  These results are shown in Tables 20 and 21.

In summary, the attachment of asbestos fibers to polyethylene surfaces
appears to be promoted by an organic material produced by bacteria, rather
than by the bacteria themselves. .  The material either coats the inside
surface of the container, which then retains any fiber which contacts the
surface, or it attaches to the fibers themselves which subsequently contact
the container surface and remain there.  The precise nature of the organic
material is not clear, but it is likely to consist of some varieties of
polysaccharide of bacterial origin.30  These are generally water-soluble,
protein free, relatively heat-resistant complexes with molecular weights
up to 10s.  These materials are known to be present in drinking water,31
and it has been shown that they are strongly adsorbed by chrysotile
asbestos as a consequence of its very high positive zeta potential.32
The action of these organic materials on the amphibole asbestos varieties
is not currently understood.  Further investigations were not made into
the mechanism, but it is probable that the effect could be developed into
a reliable,, and at least partially-specific, separation technique for
chrysotile asbestos.  Application of the technique to the amphiboles is


                                 52                                .

-------
TABLE 19.  SEPARATION EXPERIMENTS USING
        UNION CARBIDE CHRYSOTILE

Experiment
1. Unfiltered,
unsterile
double-distilled
water
2. Sterile
double-distilled
water
3. Sterile
double-distilled
water with
addition of
107 viable
organisms/liter of
Pseudomonas spp.
4.. Unfiltered
Mississauga
tap water with
addition of
107 viable
organisms/liter of
Pseudomonas spp.
Fiber Concentration
Remaining Suspended
(106 fibers/liter)

<0.5


136



119





24.6




Separated Fiber
Concentration
(106 fibers/liter)

129


8.50



6.66





75.9




Proportion of
Fibers
Separated, %

>99.6


5.9



5.3





75.5




                  53

-------
TABLE 20.   SEPARATION EXPERIMENTS USING
           UICC CROCIDOLITE
Experiment
1. Unfiltered,
unsterile
double-distilled
water
2. Sterile
double-distilled
water
3. Sterile
double-distilled
water with
addition of
107 viable
organisms/liter of
Pseudomonas spp.
4. Unfiltered
Mississauga
tap water with
addition of
107 viable
organisms/liter of
Pseudomonas spp.
Fiber Concentration
Remaining Suspended
(10s fibers/ liter)
46.9
189
136
58.4
Separated Fiber
Concentration
(106 fibers/ liter)
118
1.85
. 8.63
79.1
Proportion of
Fibers
Separated, %
71.6
1.0
6.0
57.5
                  54

-------
    	:.	^  	~   -	-	 - ...   ....
    less certain,  and further understanding  of the mechanism is required so
    that a.controlled procedure can be developed.
                   TABLE 21.   SEPARATION E5PERIMENTS USING
                                  UICC AMD!
                   ITE
     Experiment
-Fiber-Concentratiqn\J. _Separated Fiber
Remaining SuspendedU    CdncentrStiw~
  (106 fibers/liter)! (10s fibers/liter)
              Proportion of
             	Fibers---
               Separated,  %
1.  Unfiltered,
    unsterile
    double-distilled
    water
       2.01
12.1
85.8
2.  Sterile
    double-distilled
    water
      55.6
 1.73
 3.0
3.  Sterile
    double-distilled
    water with
    addition of
    107  viable
    organisms/liter of
    Pseudomonas spp.
      46.6
 4.37
 8.6
4.  Unfiltered
    Mississauga
    tap water with
    addition of
    107 viable
    organisms/liter of
    Pseudomonas spp.
      25.1
10.5
29.5
                                      55

-------
                                 SECTION 7

                      FIBER IDENTIFICATION PROCEDURES
7.1  GENERAL CONSIDERATIONS

     Precise methods for identification of asbestos  fibers  have-never been  spe-
     cified as part of the published analytical  procedures.2*33  Consequently,
     the electron microscopist has been free to  decide what-combination  of
     measurements or observations constitutes an adequate identification1 for a^
  ""particular fiber. " Chrysotile was  originally thought "to  be relatively
     simple to identify, since similar  tubular morphology occurs  in  only a  few
     other minerals and its selected area electron diffraction  (SAED) pattern
     has some characteristic features.   However, it  has been  found that  other
     species such as vermiculite can yield particles which  have a scrolled
     structure, and these can be mistaken for chrysotile if the SAED patterns
     are not interpreted quantitatively.31*  It has always been  considered
     difficult to identify the precise  mineral species of a single sub-
     micrometer fiber of amphibole.  The layer type  .SAED patterns which  are
     usually obtained have a 0.53 nm layer separation.  Comparison of this  type
     of pattern with those obtained from known amphibole asbestos fibers has
     often been accepted as identification for an~amphibole.fiber, but in fact
     many other minerals can yield similar or identical patterns. Accordingly,
     the more cautious analyst has obtained energy dispersive X-ray  analyses
     (EDXA) of at least a few of the fibers to provide additional confidence
     in the identification, although these analyses  were not  called  for  in  the
     specified procedure.  Even this approach is subject to error, since other
     minerals exist which have elemental  compositions similar to  those of the
     amphiboles.  Therefore, as part of the analytical procedures for deter-
     mination of asbestos fiber concentrations in water, a  logical and prac-
     tical protocol for fiber identification must be defined.

     When the mineralogical species of  the fibers has been  correctly identified
     the question still remains as to whether the fibers are  actually bto>t>.
     Although chrysotile presents little difficulty  in this regard,  it is not
     routinely possible in the TEM to classify an individual  small amphibole
     fiber as either asbestos or as a non-asbestos cleavage fragment, since
     crystal habit is the only basis on which these  can be  discriminated.
     Previous work35 has shown that popu&ttuwA  of fibers which originate from
     the fibrous and non-fibrous minerals can be discriminated  on the basis of
     the distributions of their fiber aspect ratios.   A~demonstrated and
     reliable means of discrimination is necessary,  since many  minerals  other
     than the asbestoses, but compositionally similar to them,  yield fragments
     of relatively high aspect ratios.   This aspect  of fiber  identification is
     discussed more fully in Section 8.     	

                                     56   .

-------
     Before it is  incorporated  into a  fiber count, each particle with an
     aspect ratio  of 3:1  or  greater, and not of obviously biological origin,
     must be Identified according  to precisely defined criteria.  Fiber
     internal  morphology,  chemical composition and crystal structure are the
     properties on which  these  criteria must be based.  Fiber internal mor-
     phology allows amphiboles  and other crystalline mineral fibers to be dis-
     criminated from chrysotile and a  few other minerals which display a
  	tubular appearance in the  TEM.  Further analysis of each fiber must then
     be conducted  using SAED and EDXA  methods.

     The crystal structure of some mineral fibers, such as chrysotile, is
     easily degraded by the  high current densities required for EDXA examina-
     tion.  Therefore, SAED  investigation of these sensitive fibers must be
     completed before attempts  are made to obtain EDXA spectra.  When examining
     more stable fibers,  such as the amphiboles, the order of work is unimpor-
     tant.

7.2  FIBER IDENTIFICATION TECHNIQUES

   - -7.2.1 -SAED Technique -  -         	

            The SAED technique  can be  either qualitative or quantitative.  For
            quantitative  work,  a thin  film of gold should be evaporated on the
            underside of  the specimen  grid as an internal calibration of
            camera length.  Qualitative SAED consists of visual examination of
            the pattern obtained on the microscope screen from a randomly
            oriented fiber.   For non-cylindrical fibers, quantitative (zone
            axis)  SAED requires alignment of the fiber so that a principal
            crystal!ographic axis  is parallel to the electron beam.  The
            pattern is then  recorded and its consistency with zone axis
            patterns from known mineral structures can be examined.  The SAED
           . pattern obtained from  one  zone axis may not be sufficiently speci-
            fic to identify  the mineral fiber, but it is often possible to
            tilt the fiber to another  angle and to record a different zone
            axis pattern. The  angle between the two axes can also be checked
            for consistency  with the structure of a suspected mineral.  Inter-
            pretation of  the zone  axis patterns, and computation of the inter-
            zone axial angles of known mineral structures are complex and are
            best performed by a computer program.  SAED patterns obtained from
            fibers with cylindrical symmetry, such as chrysotile, are an excep-
            tion since they  are not sensitive to axial tilt, and patterns from
            randomly oriented fibers can be interpreted quantitatively.

            For visual examination of  the SAED pattern, the camera length of
            the TEM should be set  to a low value and the SAED pattern then
           .should be viewed through the binoculars.  This procedure minimizes
            the irradiation  and possible degradation of the fiber.  However,.
            the pattern is distorted by the tilt angle-of the viewing screen.
            For recording purposes, a  camera length of at least 2 meters must
            be used if accurate measurement of the pattern is to be possible.
            It is  of extreme importance that, when obtaining an SAED pattern
            for either recording or visual evaluation, the sample height should

                                    57

-------
       be properly adjusted to the eucentric point and  the  image should
       be focussed in the plane of the selected area  aperture.  .If this
       is not done there may be some components of the  SAED pattern which
       do not originate from the selected area.  It will  be found in
       general  that the smallest SAED aperture will be  necessary.

       If a zone axis SAED analysis is to be attempted  on a fiber,  the
    ..... sample must be placed' in an appropriate holder.  The most con-
    .   venient holder allows complete rotation of the sample and single
       axis tilting.   The sample should be rotated until  the fiber  image
       indicates that the fiber is oriented with its  length coincident
       with the tilt axis of the goniometer.   The sample  height  should
       then be adjusted until  the fiber is at the eucentric position.
       The fiber is tilted until a pattern appears which  is a symmetrical,
       two-dimensional array of spots.  The recognition of  zone  axis  .
       alignment conditions requires some experience  on the part of the
       operator.  Not all zone axis patterns which can  be obtained  are
       useful or definitive.  Only those which have closely-spaced  reflec-
       tions corresponding to low indices in at least one direction should
       be recorded.  Patterns in which all  d-spacings are less than about
       0.3 nm are not specific and are usually very wasteful  in  computer
       analysis time.  A useful guideline is  that the lowest angle  reflec-
       tions should be within the radius of the first gold  diffraction
       ring (111), and that patterns with smaller distances between the
       reflections are usually the most definitive.

7.2.2   .EDXA Technique

       Correct identification of individual._mineral fibers  requires quan-
       titative data on their compositions.   In addition, the quantitative
       analysis "procedure "should be transferable between  instruments.

       The technique described by Cliff and Lorimer36 offers  a convenient
       method by which relatively accurate (^10%)  quantitative analyses
       can be obtained.  The X-rays generated in a thin specimen by an .
       incident electron beam have a low probability  of interacting with
       the specimen.   Thus the mass absorption and fluorescence  effects
       are negligible.  In a specimen composed of elements  I  and J,  the
       following relationship can be used to  perform  quantitative analyses
       in the TEM.
       where Aj and Aj are the elemental  integrated peak areas measured
       on the  TEM, Cj and Cj are the weight or atomic fractions of the
       two elements, and k is a constant.  To  incorporate correction for
       the particle size effect on peak area ratios,37 the Cliff and
       Lorimer technique has  been extended by  obtaining separate values
       of k for different ranges of fiber diameter.
                               58

-------
Calibration of the TEM-EDXA combination was achieved using
reference silicate minerals.  Great care was exercised in the
selection of suitable calibration standards, so that they were as
homogeneous as possible.   Since many of the fiber analyses will
involve the commercial  asbestos varieties,  some of the standards
were selected to have compositions close to these.  Each mineral
standard was examined under a binocular microscope, and a single
crystal' was extracted using forceps.  The crystal was cleaved,
and one fragment was embedded in methacrylate, after which it was
polished for microprobe analysis.  The other fragment was ground
using a mortar and pestle.  The crushed mineral was dispersed in
water, which was then filtered using a 0.1 um pore size Nuclepore
filter.  TEM grids were prepared from the filter using the carbon-
coated Nuclepore procedure.  EDXA spectra were obtained from about
20 fibers in each of four fiber diameter ranges.  The spectra were
corrected for background, and the peak area for each element was
expressed as a ratio to the silicon peak area.  Quantitative
microprobe analysis was performed on the polished mineral fragment,
and the composition of the mineral was determined using a conven-
tional ZAP procedure.  The mean k values obtained for a range of
minerals are shown in Table 22 and these form the basis of the
quantitative TEM-EDXA calculation procedure.  If quantitative
determinations are required for minerals containing elements other
than those shown in Table 22, suitable standards can be selected
and calibration information obtained by following the procedure
described above.

The quantitative EDXA technique can readily be transferred to
another system with only a minimum of calibration.  Standards of
some selected minerals in Table 22 have been prepared on TEM grids
for this purpose, and the only requirement is to obtain EDXA
spectra from about 20 particles of each mineral.  The mineral
standards selected for this purpose and the elements for which
they are used are shown in -Table 23.

        TABLE 23.  SILICATE MINERAL STANDARDS
         FOR CALIBRATION OF TEM-EDXA SYSTEM
Mineral Standards
Riebeckite
Chrysotile
Halloysite
Phlogopite
Wollastonite
Bustamite
Elements
Na, Fe, Si
Mg. Si
Al, Si
K, Si
Ca, Si
Mn, Si
                          59

-------
          TABLE 22.   "K" VALUES
(Atomic Ratio/Peak  Area Ratio, Relative to Silicon)






Mg








Fe





Ca



Na
K
AT
Mn




Actinolite
Cummingtonite
Grunerite
Hornblende
Talc
Tremolite (A)
Tremolite (B)
Chrysotile
Actinolite
Crocidolite
Cummingtonite
Grunerite
Hornblende
Phlogopite
Riebeckite
Tremolite (A)
Actinolite
Hornblende
Tremolite (A)
Tremolite (B)
Wollastonite
Crocidolite
Riebeckite
Phlogopite
Halloysite
Bustamite
Rhodoni te
Fiber Width Range, urn

<0.25
1.783
1.673
1.894
1.741
1.902
1.619
1.681
1.678
0.995
1.123
1.160
1.166
1.043
N.A.
- 1.225
0.984 .
0.975
1.001
0.870
0.801
1.080
2.979
3.344
N.A.
1.219

0.25 - 0.5
1.743
1.651
2.006
1.687
1.892
1.709
1.672
1.855
1.181
1.059
1.189
1.101
1.021
1.128
1.194
1.096
0.956
0.971
0.853
0.901
1.050
3.185
3.177
1.026
1.235
< 	 1.198 	 >
< 	 1.086 	 >

0.5 - 1.0
1.724
1.712
2.021
1.770
1.807
1.679
1.714
1.641
1.050
1.038
1.136
1.068
0.956
1.024
1.112
1.026
0.928
0.935
0.836
0.824
1.036
3.324
3.631
0.925
1.235

>1.0
1.808
1.746
2.167
1.943
1.784
1.745
1.818
1.591
0.945
1.042
1.046
0.992
0.885
0.990
1.079
0.910
0.894
0.867
0.814
0.777
0.993
3.475
3.890
0.916
1.235
< — i 1.181 	 >
N.A.
N.A.
                      60

-------
     7.2.3  Optimum Fiber Identification  Procedure

            Since  heat-sensitive  fibers such as chrysotile are degraded by
            electron beam irradiation, it is recommended that attempts to
            obtain SAED patterns  be made  first.  The process of accumulating
            an EDXA spectrum requires  intense  irradiation of the fiber, and
            sensitive fibers will  not  always retain their crystallinity after-
            wards.  However, interpretation of the fiber identification data
            is normally achieved  using a  computer program.  It has been found
            that the most expedient identification procedure is first to
            select, on the basis  of chemical composition, those minerals con-
            sistent with the EDXA spectrum of  the unknown fiber.  The more
            labor-intensive technique  of  quantitative  interpretation of zone
            axis SAED patterns can then be used for comparison with a limited
            number of possible minerals,  rather than with the entire mineral
            vocabulary.

7.3  INSTRUMENTAL LIMITATIONS

    -A modem analytical electron microscope (AEM) has a resolution of the
     order of 0.2 nm, which is adequate for imaging of the smallest fibers of
     interest.  However, in using the  instrument an appreciation of the analy-
    . tical limitations is required.

     The smallest area of the image from  which an SAED pattern can be obtained
     is obviously a critical factor in the analyses, since it may be impossible
     in a heavily-loaded specimen to completely isolate a particular fiber
     under investigation.  The smallest area which can be analyzed without
     interference from surrounding particles is given  by:

                          A = J (§+ 2000 Cse3)2

     where:

               A  = Effective SAED area in urn2

             .  D  = Diameter of SAED aperture  in ym

               M  = Magnification of objective lens

               Cs - Objective lens spherical aberration coefficient in mm

               9  = Maximum required Bragg.angle in radians

     Although almost all instruments of current manufacture meet these
     requirements, many older instruments which are still in service do not.
     It. is obviously not possible to reduce the area of analysis indefinitely
     by use of .apertures smaller in diameter than those specified by the manu-
     facturer, since there is a fundamental limitation imposed by the spheri-
     cal aberration coefficient of the objective lens.  It is extremely impor-
     tant that before SAED operation is attempted, the-image should be
     properly focussed, since the analyzed area expands rapidly with the
     degree of de-focus.
                                    61

-------
     For EDXA measurements,  the AEM should have an  illumination  and condenser
     lens system which is capable of forming electron  probes  100 nm or less  in
     diameter.  The minimum probe diameter possible is limited by the spherical
     aberration of the condenser lens system, and in some instruments the
     minimum beam diameter may not have adequate current density for the X-rays
     generated from very thin fibers to be detected.   In some instruments,  this
     problem is overcome by performing all EDXA measurements  in  scanning
     transmission electron microscopy mode. (STEM).

     It is not always possible to collect SAED and  EDXA data  from a particular
     fiber.   It may be impossible to obtain a satisfactory SAED  pattern if  the
     fiber is too close to a bar of the support grid.   When the  sample is
     tilted to locate a zone axis, the grid bar may obscure the  fiber.   The
     fiber may also be inappropriately tilted about an axis perpendicular to
     the goniometer tilt axis so that no principal  zone axes  are encountered
     within the range of the goniometer.   A double  tilt goniometer allows
     precise alignment of a fiber zone axis, but in this type of goniometer  it
     is not possible to align the fiber axis with one  of the  tilt axes.   If
     the fiber is large, extensive twinning in the  structure  may prevent the
     observation of a simple zone axis pattern.   In the case  of  EDXA, the fiber
     may be too close to a bar of the support grid, giving rise  to a high
     background signal unrelated to that from the fiber.   In  other cases, the
   1  grid bars or participate on. the sample may shield the detector and no
     spectrum from the fiber will be obtained.

7.4  SPECIFICATION OF ADEQUATE INSTRUMENTAL PERFORMANCE FOR EDXA MEASUREMENT

     The microscope specifications required for SAED performance have already
     been stated in 7.3; the EDXA performance is a  function of the microscope,
     the EDXA crystal itself, and a number of geometrical  factors.  The minimum
     acceptable performance for X-ray analysis is therefore rather more diffi-
     cult to specify.  Since suitable standard samples are not available,
    . evaluation of the combined TEM-EDXA system must be based on examination
     of the spectrum obtained from a small diameter reference fiber which is
     excited by an electron beam of known diameter. It was concluded that  the
     specification should be stated in terms of the relevant  X-ray line which
     is most difficult to measure.  X-ray detectors are generally least sensi-
     tive in the low energy region, and so measurement of sodium in UICC
     crocidolite was selected as the performance criterion.

     Investigations were made into the performance  of  the  Philips EM400T and
     EDAX PV9760/04 30 mm2 detector, which is considered to be a state-of-the-
     art instrumental combination for this type of  analysis.   An electron beam
     diameter of 100 nm, and an accelerating potential  of 100 kV were selected.
     The beam was allowed to impinge on a UICC crocidolite fiber of 50 nm dia-
     meter, and it was found that a net NaKa peak area count  rate of 5 cps was
     obtained, with a peak/background ratio of 5.0. Using an earlier model
     Philips EM301S and a Kevex 3214-301-V 10 mm2 detector, a corresponding
     NaKa peak area count rate of 1 cps was obtained with  a peak/background
     ratio of 1.0.  It was considered that the net  peak area  should exceed
     100 counts in order for it to be considered statistically valid.   Accor-
     dingly, under these conditions using the EM30.1S,  a counting period of

                                      62

-------
     100 seconds would be required in order to accumulate a satisfactory
     spectrum.   It has not been possible to survey all of the instrumental
     combinations in.use, but the results of the EM301S - Kevex 3214-301-V
     combination represents a reasonable minimum performance which is accep-
     table.   Relaxation of this performance requirement leads to long counting
     times and perhaps even an inability to detect the presence of sodium in
     the smaller fibers normally found in water samples.  Inability to detect
     sodium would seriously compromise the specificity of analysis, and in
     particular would.not allow discrimination of crocidolite from some other
     mineral species.  If the instrumental combination yields the specified
     performance for the NaKa line from typical UICC crocidolite fibers, the
     performance for the higher energy X-ray lines will normally exceed the
     requirements.

     In some circumstances where detection of sodium is not important, for
     example in chrysotile analyses, the instrumental requirements could be
     relaxed and replaced by a similar criterion based on detection of mag-
     nesium in reference chrysotile fibrils.

7.5  ANALYSIS OF FIBER IDENTIFICATION DATA	-.	

     Because the fiber.identification procedure is involved and time-consuming,
     a Fortran computer program has been developed which permits the work to
     be performed by individuals without extensive training in mineralogy.
     The program is listed in the publication, Analytical Method for Determina-
     tion of Asbestos Fibers in Water:17  This program permits the EDXA and
     zone axis SAED measurements to be compared against a library of composi-
     tional  and structural data for 226 minerals.  The mineral library includes
     fibrous species which have been listed by several authors,38"43 together
     with other minerals.which are known to be similar to amphibole or chry-
     sotile in either their compositions or some aspects of their crystallo-
     graphy.  Amphibole compositional ranges and terminology were adopted in
     accordance with the most recent mineralogical classification by the
     International Mineralogical Association.1*4  An example of a single mineral
     entry in the file is shown in Table 24.


               TABLE 24.  AN EXAMPLE OF A SINGLE MINERAL ENTRY
                         IN THE MINERAL LIBRARY FILE

NAME:      CUMMINGTONITE
FORMULA:   (FE,MG)7 SIS 022 (OH)2                                                  ;
MANDATORY:    14  8.00  8.00  12  2.10  4.90   26   2.10  4.90   0   0.00  0.00
OPTIONAL:  \   20  0.00  1.34  11  0.00  1.34    0   0.00  0.00   0   0.00  0.00
A.B.C.ALPHA.BETA.GAMMA.SYM:      9.600  18.300     5.300  90.000 101.833   90.000    5
                                      63

-------
The nominal composition for the mineral is stated, along with the per-
mitted ranges of the number of atoms for four mandatory elements.  In
the case of cummingtonite, the published ranges for the mandatory ele-
ments are Si (8.00), Mg (2.10 - 4.90), and Fe (2.10 - 4.90).   Elements
which may also be present, but which are not mandatory in this mineral,
are Ca (0 - 1.34) and Na (0 - 1.34).  The crystallographic data are also
listed together with a numerical code for the symmetry group.  Additional
minerals may be added to the library if they are thought to be of concern
in particular situations.  The addition of new minerals to the data file
does not require any further calibration of the EDXA system.

It is important to recognize that a demonstration that the measurements
on an unknown fiber are consistent with the data for a particular com-
parison mineral does not uniquely identify the unknown, since the
possibility exists that data from other minerals may also be  consistent.
It is, however, very unlikely that a mineral of another structural class
could yield data consistent with that from an amphibole fiber identified
uniquely by a quantitative chemical analysis and indexed SAED patterns
from two zone axes.

The computer program classifies fibers initially on the basis of chemical
composition.  Either qualitative or quantitative EDXA information may be
used.  The procedure using qualitative EDXA consists of entering the list
of elements which originate from the particle.  For quantitative EDXA,
the list of elements and the areas under the corresponding X-ray emission
peaks, after background correction, form the input data for the computer
program.   The calculated elemental composition of the unknown is compared
with each mineral in the stored mineral library.  The program selects
from the file a list of minerals which are consistent in composition with
that measured for the unknown fiber.  The published compositional ranges-
specified in the data file are increased by ± 20% to accommodate experi-
mental error.  In the case of sodium, the ranges are increased by + 50%,
reflecting the increased difficulty of quantifying this element.  For a
mineral to be selected as, consistent in composition, the mandatory
elements must be present in the unknown and, if the input data is quan-
titative,, then the mandatory elements must be present within these  '"'"'"
bounds.  In addition, the remaining elements entered for ;the unknown  '
fiber are compared either qualitatively or quantitatively with the   .. ......
optional  elements in the comparison mineral.                           .

Figure 17 shows a complete flow diagram for the matching procedure
based on chemical composition.  Figure 18 shows an example of the output
obtained for a fiber of riebeckite.  It can be seen that the original
list of 226 minerals has been reduced to four compositionally-consistent
varieties.

The computer program is linked to..an electron diffraction pattern
analysis  routine which is an extension of that originally developed
by Rhoades45.  The program requires measurements of 5 reflections from
the zone-axis SAED pattern, which are then tested for consistency with
the crystallographic data of the minerals previously selected as com-
positionally-consistent with the unknown fiber.  If more than a single

                                64

-------
                                                X M A T C H   COMPUTER   PROGRAM

                             A program to match X-ray spectrum with minerals tabulated 1n the computer ffle  "OIFOAT"
- Read 1ri particle Identifier
- Read In X-ray spectrum-elements and background subtracted peak Intensities


                             • If silicon present with  an X-ray peak intensity, go to quantitative analysis section
                               N0«-
                     QUALITATIVE ANALYSIS
                                                                                                           'YES
                                                                                                   QUANTITATIVE ANALYSIS
     -  Read data on each mineral from file  "DIFDAT"

     -  Included In mineral data are up to eight elemental
       possibilities with upper and lower bounds on
       atomic concentration
     -  Extract from mineral data up to four mandatory elements
       from first four elemental possibilities listed

NO   -  Hatch mandatory mineral-elements with spectrum-elements

     -  All mandatory mineral-elements must be matched with  the
       exception of a mineral-element with zero lower atomic
       concentration.  (Only one matching failure Is allowed.
       This exception allows for mineral end members).
     - Extract from mineral data up to four optional elements
       from  last four elemental possibilities listed

    '- Compare unused spectrun-'elenents'w't'tX optionaT'mineraT-
       elements

     - Up to two spectrum-elements are allowed to remain
       unmatched

     - A sum total of three unused spectrum-elements and
       optional mineral-elements are allowed
                                 YES
           WRITE THE MATCHED MINERAL IN  FILE "MATH1N"
                                                                             NO
                                                                                  - Read in particle width
 - Read data on each mineral from file "DIFDAT"

 - Included 1n mineral  data are up to eight elemental
   possibilities with upper and lower bounds on atonic
   concentration •
 - Extract from mineral data up to four mandatory elements
   from first four elemental possibilities  listed

 - From the spectral  peak areas, the ratio  of atomic
   concentrations to  silicon are calculated, and then
   using the average  number of silicon  atoms In mineral,
   the atomic concentrations are calculated

 - Hatch up the spectrum-elements with  mandatory
   mineral-elements

 - Atomic concentrations must be within range of 80% of
   lower mineral atomic concentrations  and  120% of upper
—mineral atomic concentrations.(sodium has 505 - 1505 .
   limits)

   All mandatory mineral-elements must  be matched with
   the exception of a mineral-element with  zero lower
   atomic concentration. (Only one matching failure 1s
   allowed.  This exception allows for  mineral end
   members)
 -Extract from mineral data up to four optional elements
   from  last four elemental possibilities  listed

 - Compare unused spectrum-elements with optional
   mineral-elements

 - To be matched the atomic concentration  must be less
   than  120% of upper mineral atomic concentration
   (ISO! for sodium)

 -- If concentration is betoeen 1201 of upper atomic
   concentration and upper limit plus 12.51 of average
   number of silicon atoms 1n mineral,  then this 1s a
   conditional failure of which a maximum  of Z are
   allowed
 - If concentration 1s greater than upper  linrit plus
   12.51 of average number of silicon atoms in mineral,
   then  a natch between mineral and spectrum Is rejected
   outright.  (If the spectrun-element Is  not included
   1n the optional mineral-elements, then  the upper linrit
   It defined as zero).

 - A sun total of three unused spectrum-elements and
   optional limral-elements are allowed
                                                                                                             YES
                                                                                        WRITE THE HATCHED MINERAL IN FILE "MATHIN"
                 Figure117.    Descriptive  flowchart  of  computer program  to
                                      match  X-ray  spectrum  with  stored  mineral   data
                                                                  65

-------
                      PARTICLE IDENTIFICATION
PARTICLE: UNKNOWN SAMPLE X
                                                           DATE: 08-JUN-82
    WIDTH OF PARTICLE:
0.700  micrometers
    X-RAY SPECTRUM:   ELEMENT    PEAK AREA

                         SI      5325.00
                         NA       597.00
                     ELEMENT    PEAK AREA

                        FE      2157.00
     CALCULATED
   ATOMIC RATIOS:
SMENT
SI
NA
RATIO
1.000
0.430
ELEMENT
FE
RATIO
0.425
   MINERALS WITH COMPOSITIONS CONSISTENT WITH X-RAY SPECTRUM
AEGIRINE
CROSSITE
FE-RICHTERITE
RIEBECKITE
     NA FE SI2 06
     NA2 (MG,FE)3 (FE,AL)2 SIS 022 (OH)2
     NA CA NA FES SIS 022 (OH)2
     NA2 FE3 FE2 SI8 022 (OH)2
           Figure  18.  Computer program output obtained from
                      XMATCH program for input of quantitative
                      EDXA elements from a fiber of riebeckite
                                  . 66

-------
 : ,  mineral is still reported, measurements from an additional zone axis SAED
     pattern can be entered, together with the tilt angle observed between the
     positions at which the two patterns were obtained.  For the riebeckite
     example, two zone axis patterns which were obtained by tilting a single
     fiber about its axis are shown in Figure. 19. .The angle between
     the positions at which these patterns were obtained was 21°.  Figures 20
     and 21 show the solutions obtained from these two patterns.  Figure 22
  	shows the only consistent solutions which remain after permuting the solu-
     tions from the two patterns and calculating the inter-zone axis angles.
     The cell constants for the three minerals are very similar, and discrimina-
     tion- between them on the basis of SAED methods alone would be very diffi-
     cult..      :         .           .      .

:     Even using,the computer program, the identification procedure is time-
     consuming; however, it is likely that the SAED data could be transferred
     directly from the TEM to the computer without photographing the patterns.
     Thus,, an "on-line" particle identification system can be envisaged which
     incorporates both the EDXA and SAED measurements.  For the majority of
     fibers in a sample, it will not.be economic to attempt the complete zone
     axis-identification: procedure,-and. a lower -level of identification will
     have to be accepted in.most samples.  In general, the complete procedure
     will be used only to establish the presence of a particular mineral
     species, and other fibers will, then be identified on the basis of mor-
     phological, crystallographic, or chemical similarity. .Once a particular
     EDXA spectrum has been associated with a completely identified mineral
     species in a sample, similar spectra from other particles in the same
     sample need not be processed by the computer program in order to qualify
•     as quantitative analyses.         r           •             .

 7.6  FIBER CLASSIFICATION CATEGORIES

•:     It is not always possible to proceed to a, definitive identification of a
     fiber; this may be due to instrumental limitations, obstruction by sample
     support grid bars, or the actual nature of the fiber.  In many analyses a
     definitive identification of each fiber may not actually be necessary if
     there is other knowledge available about the sample, or if the concentra-
     tion is below a level of interest.  The analytical procedure must there-
     fore take account of both instrumental limitations and varied analytical
     requirements.  Accordingly, a system of fiber classification has been
     devised to permit accurate recording of data.

     In this identification protocol the general principle is to establish the
     most specific fiber classification (target classification) which is to be
     attempted, and then to record for each fiber the classification which is
     actually achieved.  Depending on the intended use of the results, criteria
     for acceptance of fibers as "identified" can then be established at any
     time after completion of the analysis.

     The classifications as shown'in fables 25 and 26 are directed towards
     identification of chrysotile and amphibole fibers respectively.  In an
     unknown sample chrysotile can be regarded as confirmed only if a recorded,
     calibrated SAED pattern from a representative fiber in the CD category

                                     67

-------
Electron Diffraction Pattern 34
Electron Diffraction Pattern 41
        Figure 19.   Two zone axis patterns obtained by tilting a
                    single fiber about its axis.
                                      68

-------
                                                            DATE:  08-JUN-82
                       PARTICLE IDENTIFICATION



 PARTICLE: UNKNOWN SAMPLE X


ELECTRON DIFFRACTION PATTERN: FIBER 2   PATTERN 34


CAMERA CONSTANT"   83.030  mm*A

DISTANCES OF DIFFRACTION SPOTS (mm)   •

       4.580  15.520  30.700  15.520   4.580

ANGLES BETWEEN SPOTS (degrees)

       80.70   89.80   97.50  180.00
   COMPLETE-ELECTRON DIFFRACTION ANALYSES MAY BE FOUND IN.FILE "XINDEX"
                      RESULTS OF ZONE AXIS 'ANALYSIS
          MINERAL
 CROSSITE
    -2  0 -1     2  0  1

 FE-RICHTERITE
     20 >L    -2  0 -1

 RIEBECKITE
     201    -2.  6 -1
        B      C    ALPHA   BETA  GAMMA.
9.65  17.91   5.32  90.00 103.60  90.00
9.82  17.96   5.27  90.00 104.33  90.00
9.75  18.00   5.30  90.00 103.00  90.00
                Figure 20.  Solutions obtained from analysis  of zone
                           axis pattern 34 shown in Figure  19.

                                     69

-------
                                                             DATE: 08-JUN-82

                        PARTICLE IDENTIFICATION
  PARTICLE: UNKNOWN SAMPLE X      '.''.'..'


 ELECTRON DIFFRACTION PATTERN: FIBER 2   PATTERN 41


 CAMERA CQNSTANT=   .81.480  mm*A        •.'.•.

 DISTANCES OF DIFFRACTION SPOTS (mm)    ^

       .12.120    9.070   15.420  .10.520   12.110

 ANGLES BETWEEN  SPOTS (degrees)'      .   '        ""

        57.50    98.50   134.00  179.90
    COMPLETE ELECTRON DIFFRACTION ANALYSES MAY BE FOUND  IN FILE  "XINDEX1
          '':..        RESULTS OF ZONE AXIS ANALYSIS               . .
                ..""''•  -"  ' '     - l   .      •     • •    ' .        .  .    '     • '.


           MINERAL '        .       .    A      B      C    ALPHA .  BETA  GAMMA
.  CROSSITE            -  '            9.65  17.91   5.32  90.00103.60  90.00
      5-1  2    -5 -1-2    -10  1     10-1                  .

,  FE-RICHTERITE                      9.82  17.96   5.27  90.00104.33  90.00
>      5 -1  2.    -5 -1 -2    -1  0 .1-    1  0 -1

  RIEBECKITE                         9.75  18.00   5.30  90.00 103.00  90.00
      5 -1  2     1  0 -1    -1 1 0:  1    -5 -1 -2      ~

             Figure  21.   Solutions obtained from analysis  of zone
                         axis pattern 41 shown in Figure 19.


                                     70

-------
                                                            DATE:  08-JUN-82

                       PARTICLE IDENTIFICATION
 PARTICLE:  UNKNOWN SAMPLE X
    ELECTRON DIFFRACTION PATTERNS:

       I   #1:.  FIBER 2   PATTERN 34
          #2:  FIBER 2   PATTERN 41
    MEASURED INTER-ZONE AXIS ANGLE=  21.00  +/-   6.00   degrees
COMPLETE INTER-ZONE AXIS ANGLE ANALYSIS MAY BE FOUND IN FILE "PHIDAT
                                                             I'TJUTTIAT'1
               RESULTS OF INTER-ZONE AXIS ANGLE ANALYSIS

           MINERAL                    ZONE AXIS OF #1  ZONE AXIS OF.#2 ANGLE
CROSSITE                    ••            -20-1         -5 -1-2      21.14
CROSSITE                            <      201        .5-12      21.14
FE-RICHTERITE                      -      201          5-12      20.90
FE-RICHTERITE                         :   -2  0 -1         -5-1-2      20.90
RIEBECKITE                           .     201      -5-12"  20.99
RIEBECKITE               ,.              -2  0-1         -5 -1 -2      20.99
           Figure 22.  Only solutions remaining after permuting
                       solutions from both patterns 34 and 41
                       and calculating inter-zone axis angles.
                                      71

-------
    TABLE 25.  CLASSIFICATION OF FIBERS WITH TUBULAR MORPHOLOGY
TM

CM
CD
CQ
CMQ

CDQ

NAM
Tubular morphology not sufficiently characteristic
for classification as chrysotile
Characteristic chrysotile morphology
Chrysotile SAED pattern
Chrysotile composition by quantitative EDXA
Chrysotile morphology and composition by
quantitative EDXA
Chrysotile SAED pattern and composition by
quantitative EDXA
Non-asbestos mineral
    TABLE 26.- CLASSIFICATION OF FIBERS WITHOUT TUBULAR MORPHOLOGY
UF
AD

AX

ADX

AQ
AZ
ADQ

AZQ

AZZ,

AZZQ

NAM
Unidentified fiber
Amphibole by random orientation SAED (shows layer.
pattern of 0.53 nm spacing)
Amphibole by qualitative EDXA.  Spectrum has elemental
components consistent with amphibole
Amphibole by random orientation SAED and
.qualitative EDXA
Amphibole by quantitative EDXA
Amphibole by one zone axis SAED
Amphibole by random orientation SAED and
quantitative EDXA
Amphibole by one zone axis SAED pattern and
quantitative EDXA
Amphibole by two zone axis SAED patterns with
consistent inter-axial angle
Amphibole by two zone axis SAED patterns, consistent
inter-axial angle and quantitative EDXA
Non-asbestos mineral                        .
                                72

-------
is obtained.  Amphibole can be regarded as confirmed only by obtaining
recorded data which yield. exdlua-iveXf/ amphibole. &oluutionjt> for fibers
classified in the AZQ, AZZ or AZZQ categories.

7.6.1  Classification of Fibers with Tubular Morphology,
       Suspected to be Chrysotile

       Many fibers are encountered which have tubular morphology similar
       to that of chrysotile, but which defy further attempts at charac-
       terization by either SAED or EDXA.   They may be non-crystalline,
       in which case SAED techniques are not useful, or-they may be in a
       position on the grid which does not permit an EDXA spectrum to be
       obtained.  Alternatively, the fiber may be of organic origin, but
       not sufficiently definitive that it can be disregarded.  Classi-
       fication attempts for individual fibers of the same mineral will
       meet with various degrees of success.  Figure 23 shows the classi-
       fication procedure used for fibers  which display any tubular mor-
       phology.  The chart is self-explanatory, and essentially every
       fiber is eventually rejected as a non-asbestos mineral (NAM), or
       classified in some way which could  still contribute to the
       chrysotile fiber count.

       Morphology is the first consideration, and if this is not similar
       to that usually seen in chrysotile  standard samples, the initial
       classification is TM.  Regardless of the doubtful morphology, the
       fiber is still examined by SAED and EDXA methods according to
       Figure 23.  It may be possible to classify the fiber as having
       chrysotile morphology (CM) if it possesses the following morpholo-
      .gical characteristics.

            a)  The individual fibrils should have high aspect ratios
                exceeding 10:1 and be about 40 nm in diameter.

            b)  The electron scattering power of the fiber at 60 to
                100 kV accelerating potential should be sufficiently
                low for internal structure to be visible.

            c)  There should be some evidence of internal structure
                suggesting a tubular appearance similar to that shown
                in Figure 24A, which may degrade in the electron beam
                to the appearance shown in Figure 24B.

       Every fiber having these morphological characteristics should be
       examined by the SAED technique, and only those which give diffrac-
       tion patterns with the precise characteristics of Figure 25 should
       be classified as chrysotile by SAED (CD).  The relevant features
       in this pattern for identification  of chrysotile are indicated in
       Figure 25.  The (002) reflections should be examined to determine
       that they correspond approximately  to a spacing of 0.73 nm, and
       the layer line repeat distance should correspond to 0.53 nm.   There
       should also be "streaking" of the (110)  and (130) reflections.
       Using millimeter calibrations on the microscope viewing screen,

                                 73

-------
                           FIBER WITH TUBULAR MORPHOLOGY
                             Is fiber morphology characteristic
                             of that displayed by reference chrysotile?
                                NO
                    TM
               Examine by SAED
      Pattern not
      chrysotile
   NAM
Chrysotile
pattern
         CD
             Pattern not present
               or Indistinct
        Examine by quantitative EDXA
    Composition not
    that of chrysotile
   Chrysotile
   composition
                 No Spectrum
                    TM
             CQ
                           YES
                                        CM
                                    Examine by SAED
    Chrysotile
    pattern
  CD
   Pattern not
   chrysotile
                                   Pattern not present
                                     or indistinct
                                                                   CM
                             Examine by quantitative EDXA
Chrysotile
composition
Composition not
that of chrysotile
                                     No Spectrum
                                    NAM
                                Examine by quantitative EDXA
                           Composition not
                           that of chrysotile
                          Chrysotile
                          composition
                                        No Spectrum
                       NAM
                                   MM^jfcM

                                   Fool
Figure 23.   Classification chart  for  fiber with tubular morphology.
                                         :  74  :

-------
Figure 24A.
TEM micrograph of
chrysotile fibril,
showing morphology.
Figure 24B.   TEM micrograph of UICC
             Canadian chrysotile fiber
             after thermal  degradation
             by electron beam irradia-
             tion.
            Figure 25.  SAED pattern of chrysotile fiber with
                        diagnostic features labelled.   Necessary
                        criteria are the presence of 0.73 nm
                        spacing for the 002 reflections, 0.53 nm
                        spacing for the layer line repeat and
                        characteristic streaking of the 110 and
                        130 reflections.
                                     75

-------
       these observations can readily be made at the instrument.   A TEM
       micrograph of at least one representative fiber should be  recorded,
       and its SAED pattern should also be recorded on a separate film or
       plate.  This plate must carry calibration rings from a known poly-
       crystalline substance such as gold.  This calibrated pattern is
       the o\i£y documentary proof that the particular fiber is chrysotile
       and not some other tubular or scrolled species such as halloysite,
       palygorskite, talc or vermiculite.

       The proportion of fibers which can be successfully identified as
       chrysotile by SAED is variable, and to some extent dependent on
       both the instrument and the procedures of the operator. The fibers
       that fail  to yield an identifiable SAED pattern therefore  remain in
       the TM or CM categories unless they are examined by EDXA.

       In the EDXA analysis of. chrysotile, there are only two elements
       which are relevant.  For fiber classification, the EDXA analysis
       should be quantitative.  If the spectrum displays prominent peaks
       from magnesium and silicon, «tt£fi XkoJui. OAZOA in tiin. apptoptuate.
       tiaJUio, and with only minor peaks from other elements,  the  fiber
       should be classified as chrysotile by quantitative EDXA, in the
       categories CQ, CMQ or CDQ, as appropriate.

       For chrysotile analyses there are essentially three possible levels
       of analysis which are shown below:

            1.  morphological and SAED discrimination only
                (target classification CD);

            2.  in addition, EDXA of only those fibers which
                remained.unclassified by SAED (target classi-
                fication CD);

            3.  EDXA in addition to SAED of all  fibers (target
                classification CDQ).

7.6.2  Classification of Fibers Without Tubular Morphology,
       Suspected to be Amphibole

       Every particle without tubular morphology and which is not
       obviously of biological origin, with an aspect ratio of 3  to 1 or
       greater, and having parallel or stepped sides, should  be considered
       as a suspected amphibole fiber.  Further examination of the fiber
       by SAED and EDXA techniques will meet with  a variable  degree of
       success, depending on the nature of the fiber and on a number of
       instrumental limitations.  It will  not be possible to  identify
       every fiber completely, even if time and cost are of no concern.
       Moreover, confirmation of the presence of amphibole can be achieved
       onty by quantitative interpretation of zone axis SAED  patterns,  a
       very time-consuming procedure.'  For routine samples from unknown
       sources, zone axis SAED work should be performed on at least one
       fiber typical of each compositional class reported.  When  a higher

                                 76

-------
degree of certainty is required, it may be necessary to identify
more fibers by the zone axis technique.  When analyzing samples
from well-characterized sources, the cost of identification by
zone axis methods may not be justified.

The 0.53 nm layer spacing of the random orientation SAED pattern
is not by itself diagnostic for amphibole.  However, the presence
of c-axis twinning in many fibers leads to contributions to the
layers in the patterns by several individual parallel  crystals of
different axial  orientations.   This apparently random positioning
of the spots along the layer lines, as shown in Figure 26, if also
associated with a high fiber aspect ratio, is a characteristic of
some types of amphibole asbestos, and thus has some limited diag-
nostic value.  If a pattern of this type is not obtained, the
identity of the fiber is still ambiguous, since the absence of a
recognizable pattern may be a consequence of an unsuitable orien-
tation relative to the electron beam, or the fiber may be of some
other mineral species.
  Figure 26.  Amphibole SAED pattern (crocidolite)
              obtained from a fiber without precise
              orientation onto a zone axis.
                                    *

As for chrysotile, the procedure is first to define the target
level of classification to be attempted, and then to record the
degree of success achieved for each fiber.   Figure 27 shows the
fiber classification chart for suspected amphibole fibers.   This
chart shows all the classification paths possible in the analysis
of a suspected amphibole fiber, when examined systematically by
SAED and EDXA.  Initially, two routes are possible, depending on
whether an attempt to obtain an EDXA spectrum or a random orien-
tation SAED pattern is made first.  The normal  procedure for
analysis of a sample of unknown origin is to examine the fiber by
random orientation SAED, qualitative EDXA,  quantitative EDXA, and


                          77

-------
                                FIBER  WITHOUT  TUBULAR MORPHOLOGY
                    Does fiber EDXA spectrum
                    snow elements consistent
                    with araohibole?
                            I
                                        Examine by random
                                        orientation SAED
                      YES
ro spectrum

 ^
         Does quantitative EDXA
         give fiber com>osit1on
         com Is tent with
  Is 1st zone axis SAED
  pattern consistent
  with amphlbole?
    YES


     No Pattern


       |AQ|
                                                                          Layer pattern wlt
                                                                          0.53 no spacing
                                                                  Pattern definitely
                                                                  not araphibole type
                                                   -E
                                               Does EDXA spectrum thaw
                                               elements consistent Mitt)
                                               aapnibole?
                                                         Does EDXA spectrun show
                                                         elements consistent with
   Is 1st tone axis SAED
   pattern consistent
   with anphlbole?
                                                                                Is 1st zone axis SAED
                                                                                pattern consistent »Uh
Does quantUatlve EDXA   Does quantitative EDXA
give fiber composition   give fiber composition
consistent with        consistent with
amphlbolt?            a
                    Are 2nd zone axis
                    SAED pattern and
                    Inter-axial angle
                    consistent with
                    anohlbole?
                                                             Are 2nd zone axis SAED pattern
                                                             and Inter-exial angle consistent
                                                             with anpMboleT i
                                                                               Amphlbole
                                                                               Solution
Figure  27.    Classification  chart  for  fiber  without  tubular  morphology.

                                                                78                                   .          -                -

-------
zone axis SAED, in this sequence.   The final  fiber classification
assigned will  be defined either by successful  analysis  at the
target level  or by the instrumental  limitations.   The maximum
classification achieved for each fiber should be  recorded.   The
various classification categories can then be combined  later in
any desired way for calculation of the fiber  concentration,  and a
complete record of the results from each fiber is maintained for
re-assessment of the data,  if necessary.	

Depending on the particular analytical requirements,  four levels
of analysis can be defined in this analytical  procedure,  and these
levels are shown in Table 27.

In the routine unknown sample, a level 3 analysis is  required if
the presence of amphibole is to be confirmed.   For this level of
analysis, attempts are made to raise the classification of every
fiber to the ADQ category.   In addition, at least one fiber from
each type of suspected amphibole found must be examined by zone
axis SAED methods to confirm the identification..
     TABLE 27.  LEVELS OF ANALYSIS FOR AMPHIBOLE
Level of
Analysis
1
2
3
4
Application
Routine monitoring of
known and well -charact-
erized sources for one
mineral fiber type.
Routine monitoring of
known and well -charact-
erized sources where
discrimination between
two or more amphibole
fiber types 1s required
Routine samples from
uncharacterized sources
in which presence or
absence of amphibole
is to be confirmed.
Samples where precise
identification of all
amphibole fibers is
an important issue.
Target
Classification
for all Fibers
ADX
ADQ
ADQ
AZQ
Required Classification
for Confirmation of
Amphibole in a Proportion
of the Fibers
Not Applicable
Not Applicable
. AZZ, AZQ or AZZQ -
Solutions must
include only
amphiboles.
AZZQ - Solutions
must Include only
amphiboles.
                          79

-------
7.6.3  Reporting of Fiber Classifications

       Before the fiber count data can be processed to give concentration
       values, a decision must be made as to which fiber classifications
       are to be considered adequate as identification of the fiber
       species in question.  This decision will  depend on how much is
       known about the particular source from which the sample was
       collected'.'        '  "  " "  •        '"'  ".	

       If the sample was collected from a completely uncharacterized
       source, it is recommended that the classified fibers be grouped
       as below.

            a)  Confirmed Chrysotile:       CDQ  + CD

            b)  Chrysotile Best Estimate:   CDQ  + CD + CMQ + CQ

            c)  Suspected Chrysotile:       CM

            d)  Confirmed Amphibole:        AZZQ + AZQ + AZZ
                                            (Solutions must include
                                            only amphiboles)

            e)  Amphibole Best Estimate:    AZZQ + AZQ + AZZ + AZ +
                                            ADQ  + AQ

            f)  Suspected Amphibole:        ADX  + AX + AD

       The Be/s;t Es£ono£e should be reported only if some fibers are also
       reported in the confirmed category, otherwise all  fiber classifi-
       cations can only be considered as .auipec-ted amphibole or Chrysotile.

The classification procedure defined has been found to be practical, and
accommodates both the inherent instrumental limitations and the limited
mineralogical training of most TEM operators. Using this procedure, the
decision as to what combination of measurements  constitutes an adequate
identification of a fiber can be made after the  fiber count has been made,
and upgrading of the fiber classifications can be made by subsequent
review or additional measurements, if necessary.  Moreover, TEM operators
have appreciated the transfer of the onus for fiber identification to the
user of the data, and interpretation of data from different operators or
from different samples is considerably easier, more complete,  and more
reliable.
                                80

-------
                                 _ SECTION  8

                           DETERMINATION  OF FIBROSITY
 The data available  on  adverse  health  effects  due  to  exposure  to  asbestos
 fibers are largely  confined to considerations of  exposure  to  those  minerals
 which are exploited commercially.   Consequently,  regulations  regarding maximum
 permissible levels  in  workplace atmospheres usually  refer  to  "asbestos" fibers,
 where "asbestos"  is defined as one  of the commercial  varieties.  When  fiber    :
 concentrations  are  measured in industries where the  commercial varieties  are  •
 not in use, or  in the  general  environment, the question  arises as to whether a
 particular fiber  is "asbestos".   The  amphibole asbestoses  all have_non-fibrpus_
'counterparts which  display  very prominent cleavage parallel to the  c-axis, andj.
 crushing of'these "minerals  yields a "large number  of  elongated "cleavage "frag"- ,;
 ments.   These fragments  are chemically identical  to  asbestos  fibers which
 originate from  the  fibrous  version  of the mineral.   Thus,  EDXA measurements
 are incapable of  discriminating between them.  The short range crystallography
 of the cleavage fragment is also identical to that of the  fibrous version, •
 although in the latter there is often substantial twinning parallel to the
 c-axis which occasionally may  allow some discrimination  in the TEM  between the
 two types.5 However,  for routine sample analysis there  appears  to  be  no  simple
 way to assign a single fiber of these materials as either  "asbestos" or
 "cleavage fragment".1*6

 It has been suggested  that  this ambiguity of  identification,  and the possible
 confusion of minerals  such  as  cummingtonite and minnesotaite, could be re-
 solved by increasing the aspect ratio for definition  of  a  fiber.47  The value'
 of 10:1 has been  discussed, but the suggestion has received little  support.
 The aspect ratio-cumulative number  distribution for  an aqueous dispersion of
 UICC amosite, shown in Figure  28,   indicates  that use of the  10:1 aspect  ratio
 definition would  involve rejection  of about 80% of the fibers counted  under
 the current 3:1 aspect ratio definition.   While rejection  of  such a large pro-
 portion of the  particles is considered unacceptable  by most administrations, .
 it is thought necessary  to  be  able  to discriminate asbestos fibers  from
 cleavage fragments.

 It is clear that  discrimination between hand  samples  of  the fibrous and non-
 fibrous minerals  presents no difficulty.   It  is only  when  the maximum  fiber
 length is small that the ambiguity  appears.   At low magnifications  on  the
 optical microscope, it is still  relatively easy to determine whether amphibole
 particles are  fibrous  or whether they are cleavage fragments.  However, when
 the size distribution  of the fibers is such that  the  TEM is required for  the
 observation, there  is  usually  a large proportion  of  low  aspect ratio fibers,
 regardless of whether  they  originate  from the fibrous or the  non-fibrous
 mineral.                    	   	    	

                  •                     81.

-------
                lOOr
                80 :

                60-
                20 •
                 0.01   O.I     I  2  • 5 • 10  20  '40  60   80 90 9S  98 99
                   PERCENTAGE NUMBER OF FIBERS BELOW STATED ASPECT RATIO
99.9
         Figure  28.   Aspect ratio -distribution for an aqueous dispersion
                  _iof UICC amosite.       i         ______

 The concept of an "index of fibrpsity" has been discussed by Wylie35 and it
 was decided to investigate this approach further.  The description of a
 material  as "fibrous"  requires that some proportion of the fibers have high
 aspect ratios,  and  conversely, a high median aspect ratio for the population
 of fibers indicates a  high degree of fibrosity.  If there is a large propor-
 tion of low aspect  ratio fibers, as is the case for the amosite distribution
 in Figure 28,   the  median aspect ratio will  be relatively low, but the range
 of aspect ratio above  the median may be very large and the material will still
 be considered  as fibrous.  Accordingly, the description of fibrosity will
 involve the median  aspect ratio and the standard deviation of the aspect ratio
 distribution above  the median.  To evaluate this concept, aqueous dispersions'""
 of a number of fibrous and non-fibrous minerals were prepared.  The minerals
 were first ground using a mortar and pestle, -after which they were dispersed
 in double-distilled water.  TEM. samples were then prepared from the dispersions
 by the carbon-coated Nuclepore procedure.   Fiber aspect ratio distributions
 were obtained  by measuringjthe J[engths_and widths qf_about. 250. .fibers ...( a? p_ect._ .
^ratios exceeding 3:1)  from each sample.  ThVaspect "ratio versus cumulative
 number distributions were plotted in the same way as for Figure 28.  It was
 observed that  although the aspect ratio distributions for most non-fibrous
 minerals were  approximately logarithmic-normal, those of the fibrous minerals
 frequently deviated from this approximation as illustrated in Figure 28.

 To describe the fibrosity of a population of particles, the properties of the
 empirical function:
 were investigated, where A is the median aspect ratio, and g is the geometric
 standard deviation in the region above the median.   The value of g was
 obtained by extrapolation of the region of the graph between the 84.13% and
 97.73% points,  corresponding to the region between  one and two standard
 deviations above the median.
                                       82

-------
 Table  28  shows  values of the median aspect  ratio, geometric standard deviation
 between the  84.13% and 97.73% points, and the calculated fibrosity  index  F  for
 a  range of non-fibrous and  fibrous minerals.  It can be seen  that the  non-
 fibrous minerals  have values below about 50 and the values for  the  known
 asbestos  varieties are very high.  This fibrosity index can be  derived from
 the measurement of only 50  fibers in  some cases, but about 150  - 350 fibers
"were measured to  obtain the values in Table 28.  It is important to recognize
 that this fibrosity  index is-simply an-empirical number which describes the-•
 aspect ratio distribution Jun. the. pot&.co&w. 4omp£a, and that  the value can  be
 lowered by selective removal of the high .aspect ratio fibers.   Low  values are
 also obtained if  the number of fibers measured is too low to  define adequately
 the high  aspect ratio side  of the distribution.  With these reservations, the
 fibrosity index derived from TEM fiber counts can be used in most cases to
 discriminate between particle populations of fibrous and non-fibrous mineral
 species.       .

 Anthophyllite seems  to be an exception to this rule, and material from three
 sources was  examined.  The"  aspect ratio distribution of UlCC  anthophyllite  ;
 yielded the  lowest fibrosity index encountered", and inspection  of the  data  "
 showed that  there were no .large aspect ratio fibers on the TEM  samples.
 Anthophyllite from two other sources was examined and although  the  values of
 the fibrosity indices correlated with the appearances of the  bulk mineral.
 samples,  the aspect  ratio distributions.in  the TEM size range were  not similar
 to those  from the other commercially-exploited asbestos varieties.


           TABLE  28.  INDICES OF FIBROSITY  FOR SELECTED MINERALS
Mineral Fiber
Grunerite
Cummingtonite
Tremolite
Hornblende
Actinolite
Riebeckite
Anthophyllite (UICC)
Anthophyllite (Quebec)
Anthophyllite (Salt Mt.)
UICC Amos He
Union Carbide Chrysotile
UICC Crpcidolite
Beaver Bay Water (Amphibole)
Taconite Tailings
Sherbrooke. Drinking Water (Chrysotile)
Tremolite in Vermiculite Sample
Number
of
Fibers
Counted
330
370
349
329
308
308
315
391
303
140
141
199
181
673
197
330
Median
Aspect
Ratio
5.1
5.1
5.6
5.75
5.70
4.9
4.95
6.35
5.34
6.3
21.0
7.8
6.4
6.7
13.0
12.0
Geometric.
Standard
Deviation
above
84.13? Point
1.76
1.82
1.75
1.88
1.87
2.38
1.69
1.91
2.68
3.48
2.29
2.81
J.33
2.89
2.23
'2.45
Fibrosity
Index
F
17.7
19.5
20,4
26.7
25.8
43.9
14.9
33.9
89.4
604.9
1066.3
321.2
483.7
241.8
304.9
441.0
                                     83

-------
                       .  ______________ SECTION 9

                       STATISTICS OF FIBER COUNTING


9.1  TEST FOR UNIFORMITY OF FIBER DEPOSIT ON ELECTRON MICROSCOPE  GRIDS

     A check should be made using the chi-square test, to determine whether
     the fibers  found on individual grid openings are randomly  and uniformly
     distributed among the grid openings.  If the total  number  of fibers
     found in k  grid openings is n, and the areas of the k individual grid
     openings are designated AI to A^, then the total area examined is:
   -...!>
                            *-..
                                  i = 1
     The fraction  of the  total area examined which is represented  by  the
     individual  grid opening area, p-j, is given by A'i/A.   If the fibers are
     randomly and  uniformly dispersed over the k grid openings  counted, the
     expected number of fibers falling in one grid opening with area  AT is
           If the  observed number found on that grid opening is n-j, then:
                                  i = k
                            i .
                           x
2  =  \  •    (n1  -  np.)2
     This value  is  compared with significance points of the  x2 distribution,
     having (k - 1) degrees of freedom.  Significance levels lower than 0.1%
     are caused  for the sample analysis to be rejected, since this corresponds
     to a very inhomogeneous deposit.  If this occurs, a new filter should be
     prepared, paying more attention to both uniform dispersal of the sus-
     pension and the filtration procedure.

9.2  CALCULATION OF THE MEAN AND CONFIDENCE INTERVAL
     OF THE FIBER CONCENTRATION

     In a fiber  count a maximum of 20 grid openings is usually-sampled from .-
     a population of grid openings, and it is required to determine the mean
     grid opening fiber count for the population on the basis of this sampling.
     The interval about the sample mean, which, with 95% confidence, contains
     the population mean, is also required.  In previous work7*23 Gaussian


                                     84

-------
 statistics'were"~used "for' this  calculation because" it had been noted  that
 the  fiber counts were  frequently not  Poissonian.  However, at low fiber
 counts,  this  assumption  leads  to lower 95% confidence limits which are
 negative.  The  distribution  for low fiber counts  is certainly skewed if
 not  accurately  Poissonian.   Assumption of the Poisson distribution,  when
 the  distribution is demonstrably not  Poissonian,  is not correct either.
 Because  some  interpretation  of low fiber counts is required, the fiber
 distribution  statistics-of a-large number of samples-were reviewed,  and
 a logical  procedure was  developed.

 The  distribution of fibers on  the grid .openings should theoretically
 approximate to  a Poisson distribution.  Because of fiber aggregation,
 filter defects  and size-dependent identification  effects, the actual  data
 often  does not  conform to the  Poisson distribution, particularly at  high
 fiber  counts.   Simple  assumption of a Poisson distribution may therefore
 lead to  confidence intervals narrower than are justified by the data.1*8
 Moreover,  if  a  Poisson distribution is assumed, the variance is fixed in
 relation to the total  number of fibers counted.   Thus a particular fiber
 count  conducted on one grid  opening is considered to have the same con-
 fidence  .interval as that ..for. the same.number of. fibers found on many grid
 openings.  However, the  area of sample actually counted is very small in
 relation to the total  area of  the filter, and for this reason fibers .
 must be  counted on a minimum of 4 grid openings taken from different areas
 of the filter in order to ensure representative evaluation of the deposit.

 At high  fiber counts,  where  there are adequate numbers of fibers per grid
 opening  to allow a sample estimate of the variance to be made, the dis-  .
 tribution can be approximated  to Gaussian, with independent values for
.the.mean: and  variance..._Where.jthe sample_estimate.of variance exceeds  ...  .
 that implicit in the Poissonian assumption,, use of Gaussian statistics
 with the variance defined by the actual data is the most conservative
 approach to calculation  of confidence intervals.

 At low fiber  counts it is not  possible to obtain  a reliable sample esti-
 mate of the variance,  and the  distribution also becomes asymmetric,  but
 not necessarily Poissonian.  For 30 fibers and below, the distribution
 becomes  sufficiently asymmetric that  the Gaussian fit is no longer a
 reasonable one, and sample variance estimates are unreliable.  Accor-
 dingly,  for fiber counts below 31 fibers, the assumption of a Poisson
 distribution  must be made for  calculation of the  confidence intervals.
 Table  29 shows  the 95% confidence limits for selected numbers of fibers,
 assuming the  Poisson distribution.  For total fiber counts less than 4,
 the lower 95% confidence value corresponds to less than one fiber, and
 in addition,  the upper 95% confidence value corresponding to a fiber
 count  of zero is 3.69  fibers.  Therefore, it is not meaningful to quote
 lower  confidence interval points for  fiber counts of less than 4,.and
 the result should be specified as "less than" the corresponding Poisson
 upper  95% confidence value.  For fiber counts higher than 30, the sample
 estimate of variance can be  calculated, and the larger of the two con-
 fidence intervals may  be selected.  For calculation of Poisson 95% con-
 fidence intervals, Table 40  of the reference by E.S. Pearson and
 H.O. Hartley1*9  should  be used, with an-extension  to an expectation of 100.

                                 85

-------
    TABLE 29.  95% CONFIDENCE LIMITS FOR THE POISSON DISTRIBUTION

Observed Number of
Fibers
0
1
2
3
4
5
10
20
30
40
50
100
200
Lower 95% Confidence
Limit
0.00
0.025
0.24
0.62
1.09
• 1.62
4^80
12.22
20.24
28.58
37.11
81.4
174
Upper 95% Confidence
Limit
3.69
5.57
7.22
8.77
10.24
11.67
18.39
30.89
42.83
54.47
65.92
121.6
230
For more than 100 fibers, the Poisson distribution can be accurately
approximated by a Gaussian distribution, still  using the Poisson variance
estimate.  For counts of more than 30 fibers the 95% confidence interval,
based on a sample estimate of variance, is calculated using the Student's
"t" distribution.  For the two-sided Student's  "t" calculation, k values
of grid opening fiber count are compared with the expected values for the
areas of the grid openings concerned.

In summary, fiber counting data should be reported as follows.

     No fibers  detected

     The value  should be  reported  as  less  than  369%  of the  concentration
     equivalent to one fiber.

     1 to 4 fibers

     When 1 to  4 fibers are  counted,  the result should be reported as
     less than  the corresponding upper 95% confidence  limit (Poisson).
                                86

-------
      5  to  30  fibers

      Mean  and 95% confidence  intervals  should be  reported on  the basis
      of the Poisson  assumption.

      More  than 30 fibers                                 " '

	When- more: than- 30  fibers are- counted, both the Gaussian  95% confi-
      dence interval  and the Poisson 95% confidence interval can be  calcu-
      lated.   The  larger of these  two  intervals should  be selected for
      data  reporting.  When the Gaussian 95%  confidence interval is
      selected for data  reporting,  the Poisson interval  should also  be
      noted.

 Fiber counts  performed  on less than 4 grid openings yield very wide 95%
 confidence intervals when using Gaussian  statistics.   This  is because  the
 value of Student's "t"  is very large  for  1 and 2  degrees of freedom.
 Accordingly,  fiber counts should  not  be made on less than 4 grid openings.

 The. sample estimate, of- variance-S2-is -first  calculated:   	  	-

                                i  = k                     .
                         52
            ;•  .                   :   (k  -  1)
 where:  ••'".'- •'•   \ •                              -

       ni    =  number of fibers  on  the  i'th grid opening

       n    =  total  number of fibers found in  k grid openings

       Pi    =  fraction of the total area examined represented
               by the i'th grid  opening
       k    =  number of grid openings

 For the 95% confidence  interval,  the value of to. 975 is obtained  from
 tables  for (k -  1) degrees of freedom.  If the  mean value of  fiber count
 is calculated to  be n,  the upper and lower values of the 95%  confidence
 interval  are given by:
 where;-

       ny   -   upper 95% confidence  limit

       (\l .  =   lower 95% confidence  limit

                                  87

-------
n    =   mean number of fibers  per grid opening
s    =   standard deviation  (square root of sample
         estimate of variance)
k    =   number of grid openings
                           88

-------
                                SECTION 10

            DEVELOPMENT OF A REFERENCE METHOD FOR DETERMINATION

                 OF ASBESTOS FIBER CONCENTRATIONS IN WATER
A reference method for determination of asbestos fiber concentrations in
water was written, which incorporates many of the results of the research
program.  Some of the changes or additions made to the EPA Interim Method2
were major ones, although the fundamental  carbon-coated Nuclepore specimen
preparation method remained the same.

10.1  USE OF OZONE-ULTRAVIOLET LIGHT TREATMENT FOR
      OXIDATION OF INTERFERING ORGANIC MATERIALS

      Ozone-UV treatment was specified for oJUL 4omp£e4.  This involves a
      3-4 hour period during which 1% ozone is bubbled through the liquid
      Jbi tke. collection conjunct., while it is also irradiated with 254 nm UV
      light.  This treatment is necessary for the following reasons.

           a)  A small amount of high molecular weight organic materials,
               probably polysaccharides, is normally present in drinking
               water.  Particularly if the container is agitated during
               transportation, these materials,, along with the fibers, are
               collected on the inside surfaces of the container.  This
               scavenging effect is capable of almost complete removal of ;
               suspended asbestos fibers, both chrysotile and amphibole.
               The process is largely irreversible without chemical treat-
               ment to oxidize the organics.  The proposed treatment has
               been shown to restore the scavenged fibers to suspension,
               and to yield reliable recoveries not otherwise possible.

           b)  Filtration rates through the Nuclepore filters are always
               improved due to the removal of interfering organics.

           c)  Removal of the polysaccharide material virtually eliminates
               the occurrence of uncountable aggregates of fibers.

           d)  Use of the. ozone-UV method virtually eliminates the require-
               ment for low temperature ashing.  Low temperature.ashing
               appears to be unreliable and it also concentrates any con-
               tamination associated with the filters.
                                     89

-------
10.2  USE OF ULTRASONIC TREATMENT

      Ultrasonic treatment of water samples in a bath-type unit,  both before
      and after ozone-UV treatment, is specified.   Use of mild ultrasonic
      treatment has been found to be the only way of achieving consistent
      results from identical  samples.   No evidence has been found that use of
      ultrasonic treatment at these low powers for the specified  times results
      in measurable fiber disintegration.	

10.3  USE OF THE CONDENSATION WASHER FOR REMOVAL OF
      RESIDUAL UNDISSOLVED PLASTIC FROM TEM SAMPLES

      Recent batches of Nuclepore polycarbonate filters are very  difficult to
      dissolve in chloroform in a Jaffe Washer, even after treatment extending
      over several days.  It is found that after the Jaffe dissolution step,
      residual undissolved plastic remains  on the sample such  that during the
      TEM examination a significant proportion of the fibers may  not be seen.
      After consultation with Nuclepore Corporation, it seems  unlikely that
      this situation is going to improve.  Use of methylene chloride, or
      N-methyl-2-pyrrolidone as alternative solvents was suggested.   Methylene
      chloride was found to evaporate very  rapidly and required the Jaffe
      Washer to be tightly sealed.   Although it yielded TEM samples which were
      more satisfactory, it did not completely remove the plastic.   When
      N-methyl-2-pyrrolidone was used, it was found to leave an oily film on
      the samples, which could only be removed by treatment with  another sol-
      vent.  Moreover, the rapid initial  solvent attack on the plastic caused
      a significant degree of breakage of the carbon film.

      The most effective technique was found to be an initial  Jaffe Washer
    .  dissolution, followed by a short treatment in a condensation washer.
      Use of the condensation washer as a single dissolution step is not
      recommended, since this requires that initially chloroform  be pipetted
      directly onto the TEM sample to avoid curling of the filter in the
      chloroform vapor.  Using the original design of Jaffe Washer,  the piece
      of paper under the samples can easily be moved to the cold  finger of a
      condensation washer.  After treatment in the condensation washer for
      30 to 60 minutes, the TEM samples.are then found to be satisfactory.
      Use of the condensation washer in this way should not incur fiber losses,
      since all particles are thought to be held in position by the carbon
      film.  However, additional experiments were conducted to demonstrate
      that this was the case.

      The Nuclepore control filters used originally for the collapsed membrane
      filter study were used to make measurements  of any fiber losses incurred
      by the additional condensation washing step.   The control filter results
      for chrysotile and crocidolite are shown in  Tables A-2 and  A-5 of
      Appendix A.  Four filters from each series were analyzed again after pre-
      paration, using the condensation washing step.  The detailed results are
      shown in Tables D-l and D-2 of Appendix D.   The fiber loss  analyses  for
      chrysotile and crocidolite are shown  in Tables 30 and 31 respectively...
                                    90

-------
TABLE 30.  CONDENSATION WASHER  STUDY:   CHRYSOTILE FIBER LOSS ANALYSIS
Sample Preparation
Control Samples
Prepared by Jaffe Washer only
Samples from Same Series
prepared by Successive Jaffe
and Condensation Washing
Number
of
Filters
Analyzed
10
4
Fiber Concentration
(106 fibers/liter)
Mean
,264.8
258.0
95% Confidence
Interval
236.6 - 293.6
208.1 - 307.9
Fiber Loss
in
Preparation
- '
2.6
Is the Fiber
Loss
Detectable
at 5%
Significance?
-
No
                *Compared to the Carbon-Coated Nuclepore Preparation
TABLE 31.  CONDENSATION WASHER  STUDY:   CROCIDOLITE FIBER LOSS ANALYSIS

Sample Preparation


Control Samples
Prepared by Jaffe Washer only
Samples from Same Series
prepared by Successive Jaffe
and Condensation Washing
•Number
of
Filters
Analyzed

10

4


Fiber Concentration
(10s fibers/liter)

Mean

224

243



95% Confidence
Interval
158 - 290

198 - 288


Fiber Loss
in
Preparation
tt\*
\kl
_

0


Is the Fiber
Loss
Detectable
at 5%
Significance?
_

No


                *Compared to the Carbon-Coated Nuclepore Preparation
  It can be seen that no fiber  losses  could be detected at the 5% signi-
  ficance level.

  Figures 29 and 30 show the deposit from an aqueous  dispersion containing
  chrysotile and other materials.  These  micrographs  show the same area of
  a TEM sample both before and  after condensation  washing for a period of
  180 minutes.  It can be seen  that no particle losses  or movement have
  occurred as a result of re-washing of the TEM grids in the condensation
  washer.  In the area of the sample illustrated,  the residual plastic was
  almost.completely dissolved by  the Jaffe Washer  treatment.  The remaining
  plastic was not, in this case,  removed  by the condensation washer, since
  the initial observation in the  TEM caused cross-linking of the plastic
  and increased its insolubility.

  It was concluded that the condensation  washing step did not compromise the
  samples in any way, and that  the operating conditions of the washer were
                                  91

-------
  «- ** •
    '   '              '
 '             UK.-'
            **.     5*
       ^•••-•^r.j.-'/>%^
  '^*4-"-*:               •*'.&
                    j&
             "•^.  V '»'- V •»»

                   ;^  Al   *!S^;*S•••*•&$-•+•   TOf  ?,
          ,   v    -•  ^  •?^^^»!lA^:-»!fc*^ibi?ift •
     l .
   •-
Figure  29.  TEM specimen prepared from aqueous suspension of asbestos and
           other materials.  Polycarbonate filter dissolved using chloroform
           in Jaffe Hasher only.
                                   92

-------
                                                                  2-0 jam
Figure 30.  TEM specimen shown in Figure 29, but after further treatment
            using chloroform in condensation washer for 180 minutes.
                                                                                    -;.— .-*
                                     93

-------
      not critical.   Accordingly,  it was  incorporated  into  the  analytical
      method as an optional  step in  the .routine  procedure.

10.4  INTRODUCTION OF MINIMUM FIBER  LENGTH TO BE REPORTED

      A minimum fiber length of 0.5  ym has been  specified for the  following
      reasons.

           a)  Collection  of fibers  shorter than 0.5 ym  by  a  0.1 ym nominal
               pore size Nuclepore filter becomes less efficient,  and  may  be
               variable depending  on both the batch number  and  the filtration
               rate.

           b)  Identification of fibers less  than about  0.05  ym in diameter
               becomes very  unreliable.

           c)  Nuclepore filters contain  many artifacts  which can  be mistaken
               by inexperienced operators for fibers below  0.5  ym  in length.

           d)-  Operators vary in their interpretation when  asked to count  all
               fibers, particularly  when  large numbers of fibers are present.
         		_.	 .    ...	  _..  _ _    .. *•  	•   .          « .   V
      Counting  fibers with.lengths less than  0.5 ym is acceptable,  however, if
    •  they can  be identified  as asbestos  fibers.

10.5  REQUIREMENT FOR QUANTITATIVE INTERPRETATION OF
      CHRYSOTILE SAED PATTERNS    ...

      A gold-ring-calibrated SAED pattern from typical fibers has  been speci-
      fied as part of the  report,  from which  the 0.53  nm layer  line spacing
      and the 0.73 nm (002)  spacing  can be calculated.   These and  other rele-
      vant features for identification of chrysotile by  SAED  are also  speci-
      fied.  This limited  quantitative interpretation  is required  in order to
      discriminate chrysotile from halloysite, palygorskite and scrolled
      vermiculite.

10.6  REQUIREMENT FOR ENERGY DISPERSIVE X-RAY ANALYSIS AND
      ZONE AXIS SAED FOR AMPHIBOLE IDENTIFICATION

      Unequivocal analysis for amphiboles is  extremely difficult and can be
      achieved onty by ui/tng -the mo&t &opklt>£icat
-------
10.7  INTRODUCTION OF A FIBER CLASSIFICATION SYSTEM

      A fiber classification system has been incorporated which requires that
      the operator specify the identification action taken and its result for
      each fiber.  The system removes the onus of -tdentt^ccotuw from the
      microscope operator, who is instructed to attempt a specified degree of
      classification by morphology, SAED and'EDXA.  The classifications are
  	 then reviewed by!the user of the data, who makes "the decision as to
      what constitutes adequate identification for the particular analysis.
      This concept permits different levels of analysis to be defined.  From
      these levels of analysis, the analyst can select the appropriate level
      according to. both economics and data requirements.

10.8  STANDARDIZED. REPORTING FORMAL   '..      \

      A standardized reporting, format and statistical evaluation of data have
'.      been defined/These incorporate a minimum standard of fiber count
      reporting,, and. require that the operator record the classification of
      each fiber.  A computer program is provided which is compatible with the
'*•• •new-data reporting technique.'  For reporting, the user defines the	
      classification categories- to be combined in the final result, but the
      raw data is also presented for re-evaluation of the individual classifi-
      cations,, if necessary.           .              .
             '•  '  '•'••'.- :"",•••.-:•"• •	v •:-.-'•    •".-''•'              •
10.9  INTRODUCTION OF FIBROSITY: INDEX        .   .  ;. .   ;            .    .       /

      To resolve the cleavage fragment versus fiber, controversy in TEM analy-
      ses,, a "fibrosity index" is defined.  On a routine basis it is not
      possible to determined a single fiber of a particular aspect ratio is
      a cleavage fragment or an asbestos fiber.  It is possible to distinguish
      between populations of the two forms on the basis of their aspect ratio
      distributions.  The fibrosity index is a function of the median aspect
      ratio, and the distribution of aspect ratios higher than the median.
;      A minimum number of about 50 fibers is required in order to define
      whether or not the population arises from a fibrous:mineral.

      The question of re-definition of the fiber to one of aspect ratio 10:1
      has resolved itself.  Even in the case of UICC crocidolite and amosite,
      80% of the waterborne fibers have aspect ratios £ei* than 10:1; thus,
      use of the 10:1 definition would be completely unjustified.

10.10 STATISTICS OF FIBER COUNTING

      The statistics used to calculate the confidence intervals of measurements
;      have been modified to incorporate logical interpretation of low fiber
      counts.  The 95% confidence interval is calculated on the basis of
      Poissonian statistics.  Where the number of fibers is adequate for
      calculation of an independent standard deviation on the basis of
      Gaussian statistics, the corresponding 95% confidence interval is derived
      and the larger Of the two confidence intervals is the one used for repor-
      ting.  In this way, the more conservative estimate of the analytical
'      precision is used..	  	,	_	_.•		

                        „   ..     '••'.•  95 ;  '  	  '	-	"...

-------
10.11 DELETION OF "FIELD OF VIEW"  FIBER COUNTING

      The alternative "field of view"  counting  technique  for examination of
      heavily-loaded TEM samples has been  deleted.  Where this situation exists,
      the sample should be re-prepared to  have  a  lighter  loading.  The
      rationale for this modification  was  that  operators  have been found to
      produce lower fiber counts from  heavily-loaded TEM  samples, relative to
      ideally-loaded TEM samples prepared  from  the same water sample.
                                      96

-------
                                 REFERENCES
1.  Chatfield, E.J. and P.  Riis (1982).   A Rapid Screening Technique for
    Detection of Asbestos Fibers in Water Samples.   Contract No.  68-03-2717
    S.E. Environmental Research Laboratory, U.S. Environmental  Protection
    Agency, Athens, Georgia, 30613.

2.  Anderson, C.H.  and J.M. Long (1980).   Interim Method for Determining
    Asbestos in Water, Report EPA-600/4-80-005.  U.S.  Environmental  Protection
    Agency, Athens, Georgia.  Available  through the National Technical
    Information Service, Springfield, Virginia, 22161.

3.  Ruud, G.O., C.S. Barrett, P.A.  Russell and R.L. Clark (1976).  Selected
    Area Electron Diffraction and Energy Dispersive X-ray Analysis for the
    Identification of Asbestos Fibres, a Comparison.   Micron, 7_,  115-132.

4.  Lee, R.J. (1978).  Basic Concepts of Electron Diffraction and Asbestos
    Identification using SAD.  Part I:  Current Methods of Asbestos
    Identification using SAD; Part II:  Single Crystal  and SAD.   Scanning
    Electron Microscopy/1978/1 (ed. 0. Johari) SEM Inc., AMF O'Hare, Chicago,
    Illinois, 677-686 and 687-694.

5.  Whittaker, E.O.W. (1979).  Mineralogy, Chemistry and Crystallography of
    Amphibole Asbestos.  In:  Short Course in Mineralogical  Techniques of
    Asbestos Determination (ed. R.L. Ledoux), Mineralogical  Association of
    Canada,  /o Department of Mineralogy, Royal Ontario Museum,  100 Queen's
    Park, Toronto, Ontario, Canada, 1-34.

6.  An Inter-Laboratory Study of Asbestiform Mineral  Fibre Levels in the
    Water Supply of Thunder Bay, Ontario (1975).  Ontario Ministry of the
    Environment, P.O. Box 213, Rexdale,  Ontario, Canada, M9W 5L1.

7.  Chatfield, E.J., R.W. Glass and M.J.  Dillon (1978).  Preparation of
    Water Samples for Asbestos Fiber Counting by Electron Microscopy, Report
    EPA-600/4-78-011.  U.S. Environmental Protection Agency, Athens, Georgia.
    Available through the National  Technical Information Service, Springfield,
    Virginia, 22161.

8.  Ortiz, L.W. and B.L. Isom (1974).  Transfer Technique for Electron
    Microscopy of Membrane Filter Samples.  American Industrial  Hygiene
    Assoc. Journal, Vol 35 (No. 7), 423-425.
                                     97

-------
 9.  Bishop, K., S. Ring, R. Suchanek and D. Gray (1978).  Preparation Losses
     and Size Alterations for Fibrous Mineral Samples.  Scanning Electron
     Microscopy/1978/1 (ed. 0.. Johari) SEM Inc., AMF O'Hare, Chicago, Illinois,
     207-212.

10.  Anderson, C.H. and J.M. Long (1976).  Preliminary Interim Method for
     Determining Asbestos in Water.   U.S. Environmental Protection Agency,
     Athens, Georgia-30613.•-'-		—			

11.  Chopra, K.S., D.  Beaman, P.  Cook (1980).  Interlaboratory Measurements
     of the Chrysotile Asbestos Fiber and Mass Concentrations in Water
     Samples.  N.B.S.  Special Publication 619, Proceedings of the NBS/EPA
     Asbestos Standards Workshop, Published 1982.   U.S. Department of
     Commerce, National Bureau of Standards, Washington, D.C. 20234, 121-131.

12.  Cunningham, H.M.  and R. Pontefract (1971).   Asbestos Fibres in Beverages
     and Drinking Water.   Nature, Vol., 232, 332-333.

13.  Biles, B. and T.R. Emerson (1968).   Examination  of Fibres in Beer.
     Nature, Vol.  219, 93-94.-	 —     	-    .

14.  Chatfield, E.J.  and R.W. Glass  (1976).  Analysis of Water Samples for
     Asbestos:  Sample Storage and Technique Development Studies.  Symposium
     on Electron Microscopy of Microfibers, Published 1977,  F.D.A.,. 123-137.

15.  Chatfield, E.J.  and M.J. Dillon  (1978).  Some Aspects of Specimen
     Preparation and Limitations  of Precision in Particulate Matter Analysis
     by SEM and TEM.   Scanning Electron Microscopy/1978/1 (ed.  0. Johari).
     SEM Inc., AMF O'Hare, Chicago,  Illinois 60666, 487-496,  468.

16.  Chatfield, E.J.  (1978).  A New Technique for Preparation of Beverage
     Samples for Asbestos Measurement by Electron Microscopy..  Proceedings of
     the Ninth International Congress of Electron Microscopy, Toronto, 1978,
     Vol. II, 102-103.

17.  Chatfield, E.J.  and M.J. Dillon  (1982).  Analytical Method for.
     Determination of Asbestos Fibers in Water.   Contract No. 68-03-2717
     S.E. Environmental Research  Laboratory, U.S.  Environmental Protection
     Agency, Athens,  Georgia, 30613.

18.  Mudroch, 0.  and J.R. Kramer  (1976).  Enumeration of Asbestos Fibres.
     Proceedings of the Microscopical Society of Canada, Third Annual  Meeting.
     20-23.           .      -

19.  Cunningham,  H.M.  and R.D, Pontefract. (1976).   Measurement of Asbestos in
     Air, Water,  Foods, Tissues and Excreta.  Symposium on Electron Microscopy
     of Microfibers,  Published 1977,  F.D.A.  162-165.
                                    98

-------
20.  Ontario Ministry of the Environment (1976).  An Interim Method for the
     Determination of Asbestiform Mineral Fibre Levels in Air and Water.
     Ontario Ministry of the Environment, Laboratory Services Branch,
     P.O. Box 213,. Rexdale, Ontario, Canada, M9W 5L1.

21.. Jaffe, M.S. (1948).  Handling and Washing Fragile Replicas.  Proceedings
     of EMSA, Toronto.  Journal Applied Physics 19_, 1187.

22.  Chopra, K.S. (1978).  Inter!aboratory Measurements of Amphibole and
     ChrysotiTe Fiber Concentrations in Water.  Journal of Testing and
     Evaluation, Vol.. 6, No. 4, 241-247.

23.  Chatfield, E.J. and M.J, Dillon (1979).  A National Survey for Asbestos
     Fibres in Canadian Drinking Water Supplies.  Health and Welfare Canada
     Report 79-EHD-34.  Information Directorate, Department of National Health
     and Welfare, Brooke Claxton Building, Ottawa, Canada K1A OK9.           ;

24.  Herdan, G., (1960).  Small Particle Statistics, Second Revised Edition.
     Butter-worths, London,_ 350. _           „•_•'_     __

25.  Naumann, A.W. and W.H. Dresher (1968).  Colloidal Suspensions of
     Chrysotile Asbestos:  Surface Charge Enhancement.  Journal of Colloid
     and Interface Science, Vol. 27, No. 1, May 1968, 133-140.

26.  Wilke, H. and H.E. Voss (1954).  The Removal of Pyrogenic Substances
   '••" from Injection Solutions" by'Hltratjon..'_" Arzneim-Forsch. ,4, 8-14. ..,

27.  Kienholz, M. (1964).  Experiments with a Plant for Sterilization,
	--Demineralization and De-pyrogenization of Water. • Arzneim-Forsch., 14,
     1259-1262.             .    .         .

28.  Chatfield, E.J. and R.W. Glass (1976).  Improved Methodology for
     Determination of Asbestos as a Water Pollutant.  Ontario Research
    - Foundation Report, Mississauga, Ontario, Canada.

29.  Fletcher, M.. (1979).  The Attachment of Bacteria to Surfaces in Aquatic
     Environments.  In: _Adhesion pf_ Microorganisms to Surfaces.  Academic
    rPress7New York",. N7Y". (edV^DVC.'EViwood,7j.T M^                     87-108.

30.  Luderitz, 0. (1959).  Fiebererzeugende Stoffe mit besonderer Berucksichti-
     gung bakterieller Pyrogene.  Schriftenreihe Wander, 1-29.

31.  Lechat, P. (1961).  Etat actuel de quelques problernes poses par les
     pyrogenes bacteriens.  Produits pharm. 16, 509-523.

32.  Wilke; H. (1976).  New Rules Issued by the FDA About the Use of Asbestos-
     Containing Filter Sheets and Their Consequences in Pharmaceutical
     Technology.  In:  Drugs Made In Germany, 19, (Editio Cantor/D-7960
     Aulendorf, W.. Germany) 18-36.
                                    99

-------
33.  Samudra, A.V., C.F. Harwood and J.D. Stockham, (1978).  Electron
     Microscope Measurement of Airborne Asbestos Concentrations:  A
     Provisional Methodology Manual.  Report EPA-600/2-77-178, U.S. Environ-
     mental Protection Agency, Research Triangle Park, North Carolina, 22711.
     Available through National Technical Information Service, Springfield,
     Virginia, 22161.

34.  Chatfield, E.J. and G.M. Lewis (1980).   Development and Application of an
     Analytical Technique for Measurement of Asbestos Fibers in Vermiculite.
     Scanning Electron Microscopy/1980/1.(ed. 0. Johari) SEM Inc., AMF O'Hare,
     Chicago, Illinois, 329-340, 328.

35.  Wylie, A.G. (1979).  Fiber Length and Aspect Ratio of Some Selected
     Asbestos Samples, In:  Health Hazards of Asbestos Exposure,
     (I.J. Selikoff and E.C. Hammond,  ed).  Annals of the New York Academy
     of Sciences, 1979; 330: 605-610.

36.  Cliff, G. and G.W. Lorimer (1975).  The Quantitative Analysis of Thin
     Specimens.  Journal of Microscopy, 103, 203-207.

37.  Small, J.A., K.F.J. Heinrich, D.E. Newbury and R.L. Myklebust (1979).
     Progress in the Development of the Peak-to-Background Method for the
     Quantitative Analysis of Single Particles with the Electron Probe.
     Scanning Electron Microscopy/1979/1I (ed. 0. Johari) SEM Inc., AMF 0'Hares
     Chicago, Illinois, 807-816.

38.  Deer, W.A. ,-R.A. Howie and J. Zussman (1966). ' Rock-Forming Minerals,
     Part 2, Chain Silicates.  Longmans, London, England, 99-192.

39.  Mason, B. and L.G. Berry (1968).   Elements of Mineralogy.   W.H.  Freeman
     and Company, San Francisco, Calif.

40.  Bates, R.L. and J.A, Jackson (ed.) (1980).  Glossary of Geology, Second
     Edition.  American Geological Institute, Falls Church, Virginia, 22041.

41.  Roberts, W.L., G.R. Rapp and J. Weber (1974).  Encyclopedia of Minerals.
     Van Nostrand Reinhold Company, New York, N.Y.

42.  Spumy, K.R., W. Stober, H. Opiela and G. Weiss (1979).   On the
     Evaluation of Fibrous Particles in Remote Ambient Air.  The Science of
     the Total Environment 11, 1-40.

43.  Zumwalde, R.D. and J.M. Dement (1976).   Review and Evaluation of
     Analytical Methods for Environmental Studies of Fibrous Particulate
     Exposures.  Symposium on Electron Microscopy of Microfibers.
    .Published  1977," F.D. A. ,"139-150....

44.  International Mineralogical Association (1978).  Nomenclature of
     Amphiboles.  (compiled by B.E. Leake),  The Can. Mineralogist 1978;
     16:  501-520.
                                    100

-------
45.   Rhoades, B.L. (1976).   XIDENT - A Computer Technique for the Direct  -../
     Indexing of Electron Diffraction Spot Patterns, Research Report 70/76. '
     Department of Mechanical  Engineering, Univ. of Canterbury, Christchurch,
     New Zealand.

46.   Champness, P.E., 6.  Cliff and G.W. Lorimer (1978).  The Identification
     of Asbestos Using Electron Optical Techniques.  In:  Electron Microscopy
     and X-ray Applications to Environmental and~Occupational'Health
     Analysis (ed, P.A.  Russell and A.E.  Hutchings).  Ann Arbor Science,
     Ann Arbor, Michigan, 149-168.

47.   Campbell, W.J., R.L. Blake, L.L. Brown, E.E.  Gather and J.J. Sjoberg,
     (1977).  Selected Silicate Minerals  and Their Asbestiform Varieties,
     Mineralogical Definitions and Identification - Characterization,
     Information Circular 8751,. Bureau of Mines, Avondale Research Center,
     4900 LaSalle Road,  Avondale, Md. 20782..

48.   Chatfield, E.J. (1979).  Preparation and Analysis of Particulate Samples
     by Electron Microscopy, with Special Reference to Asbestos.  Scanning
     Electron Microscopy/1979/1 (ed.. 0. Johari), SEM Inc., AMF O'Hare,
     Chicago, II. 563-578,  486.

49.   Pearson, E.S. and H.O. Hartley (1958).  Biometrika Tables for
     Statisticians, Volume  I.   Cambridge  University Press, 32 East 57th
     Street, New York 22, N.Y.
                                    101

-------
        APPENDIX A

   TABLES A-l TO A-7
COLLAPSED MEMBRANE STUDY:
DETAILED ANALYTICAL DATA
           102

-------
                      TABLE A-l.  COLLAPSED MEMBRANE STUDY FOR CHRYSOTILE:   INITIAL  STUDIES
Sample
Nuclepore ,
Controls «
3
Collapsed 1
0.1 urn 2
'ore Size 3
lillipore
Collapsed 1
).22 \an 2
5ore Size 3
tfllipore 4
i 5
Collapsed 1
).45 pin 2
'ore Size 3
lilllpore
Fiber Concentration
Mean
(106 Fib/LHer)
24.1
27.4
29.2
33.5
22.5
22.0

28.7
20.1
18.5
17.1
22.5
13.1
16.5
5.5

95X Confidence
Interval
(106 Fib/Liter)
19.2 - 29.0
21.8 - 33.0
24.5 - 33.9
22.4 - 44.6
15.6 - 29.4
15.7 - 28.3

21.7 - 35.7
13.2 - 27,0
10.1 - 26.9
9.0 - 25.2
14.2 - 30.8
8.0 - 18.2
10.1 - 22.9
1.7 - 9.3

Estimated Mass
Concentration
(Nanograms/Liter)
165
477
355
392
192
205

218
. 176
194
230
190
142
238
51 .

Concentration
Equivalent to
1 Fiber
Detected
(106 Fib/Liter)
0.290
0.280
0.292
0.266
0.296
0.271

0.279
0.279
0.289
0.276
0.296
0.272
0.275
0.290

No.
Fibers
Counted
83
98
100
126
76
81

103
72
64
62
76
48
60
19

No.
Grid
Openings
10
10
10
10
10
10

10
10
10
10
10
10
10
10

Grid Distribution
X2
6.28
8.53
7.60
28.4
13.74
12.46

10.11
16.32
19.75
22.52
15.81
12.8-
15.67
15.21

Significance
of
Uniformity
m
50
25
50
<0.1
10
10

25
5
1
0.5
5
. 10
5
5

. o
-OJ

-------
TABLE A-2.   COLLAPSED MEMBRANE STUDY FOR CHRYSOTILE
   0.1 um Pore Size Nuclepore Filter Controls
Sample
1
2
3
4
5
6
7
8
9
i
10
Fiber Concentration
Mean
(106 F1b/L1ter)
240
250
220
230
270
230
350
260
270

310
95% Confidence
Interval
(106 F1b/L1ter)
160 - 330
180 - 320
160 - 280
160 - 310
180 - 370
160 - 300
310 - 390
120 - 410
97 - 450

160 - 470
Estimated Mass
Concentration
(Nanograms/Uter)
1600
1200
1300
2000
1300
1400
2800
1300
1600

2100
Concentration
Equivalent to
1 Fiber
Detected
(106 Fib/L1ter)
1.94
2.45
2.17
2.08
2.23
2.14
3.03
2.52
2.44

3.05
No.
Fibers
Counted
124
102
103
112
123
106
115
105
112

103
No.
Grid
Openings
7
5
6
6
6
. 6
4
5
5

4
Grid Distribution
X2
15.53
4.27
5.86
8.40
10.78
7.99
0.41
17.23
23.99

7.17
Significance
of
Uniformity
1
25
25
10
5
10
90
0.1 .
<0.1

5

-------
                                 TABLE A-3.  COLLAPSED MEMBRANE  STUDY FOR CHRYSOTILE

                                        0.1 ym Pore Size Mil\_1pgre Filters
Sample
. 1
2
3
4
5
6
7
8
9
i
10
Fiber Concentration
Mean
(106 Fib/Liter)
290
350
290
340
300
270
340
270
230

170
952 Confidence
Interval
(106 Fib/Liter)
170 - 410
250 - 450
190 - 390
.230 - 450
150 - 450
150 - 390
190 - 490
230 - 310
190 - 280

150 - 190
Estimated Mass
Concentration
t
(Nanograms/ Liter)
1400
2300
1200
1600
1300
1400
1700
1400
1100

980
Concentration
Equivalent to
1 Fiber
Detected
(106 Fib/Liter)
2.64
3.22
2.37
3.30
2.84
2.65
3.05
2.14
2.10

1.59
No.
Fibers
Counted
111
108
123
102
105
102
111
126
111

107
No.
Grid
Openings
5
4
6
4
5
•5.
4
6
6

8
Grid Distribution
X2
9.50
2.76
10.46
3.28
13.84
10.36
6.33
2.15
3.33

1.55
Significance
of
Uniformity
m
2.5
25
5
25
0.5
2.5
5
75
50

97.5
t
I
j !-•
' o
 en

-------
                TABLE A-4.   COLLAPSED MEMBRANE STUDY  FOR CHRYSOTILE
                         0.22 urn Pore Size  Mi Hi pore  Filters
Sample

1
2
3
4*
5
6
7
8
9
i
10
Fiber Concentration
Mean
(106 Fib/Liter)

240
180
220
170
190
260
200
180
210

180
955K Confidence
Interval
(106 F1b/L1ter)

180 - 300
140 - 220
140 - 310
0-530
130 - 260
190 - 340
120 - 280
130 - 240
120 - 300

120 - 230
Estimated Mass
Concentration
i
i
(Nanograms/Liter)
1
1300
iooo
1500
1300
1800
1700
1000
j
1600
j970
|
fOOO
i
Concentration
Equivalent to
1 Fiber
Detected
(106 Fib/Liter)

2.25
1.82
1.91
6.35
1.90
2.52
1.92
1.63
1.94

1.58
No.
Fibers
Counted

106
100
117
26
101
104
104
113
106

112
No.
Grid
Openings

6
7
7
2
7
5
7
8
7

8
Grid Distribution
X2

5.42
4.91
16.01
0.77
12.16
4.68
15.37
12.39
20.19

13.77
Significance
of
Uniformity
1%)

25
50
1
25
5-
25 .
1
5
0.1

5
*Unbroken carbon film present on only 2 openings of the 3 specimen grids  prepared.

-------
TABLE A-5.  COLLAPSED MEMBRANE STUDY FOR CROCIDOLITE
       0.1 urn Pore Size Nuclepore Filter Controls
Sample
1
2
3
4
5
6
7
8
9 i
1°
Fiber Concentration
Mean
(106 Fib/Liter)
211
259
222
261
' 226
210
155
256
165
270
95% Confidence
Interval
(106 F1b/L1ter)
168 - 253
166 - 352
152 - 293
238 - 284
187 - 265
163 - 257
110 - 200
181 - 331
131 - 199
216 T 323
Estimated Mass
Concentration
(Micrograms/L1ter)
63
140
35
43
31
29 \
129
34 -
38
45
Concentration
Equivalent to
1 Fiber
Detected
(106 F1b/L1ter)
2.17
2.35
2.25
2.64
2.24
2.21
1.28
2.49
1.28
2.59
No.
Fibers
Counted
97
110
99
99
101
95
121
103
129
104
No.
Grid
Openings
5
5
5
4
5
5
8
4
8
4
Grid Distribution
X2
2.07
7.46
5.35
0.23
1.55.
2.46
12.8
2.62
7.22
1.28
Significance
of
Uniformity
m
50
10
25
95
75
50
5
25
25
50

-------
                             TABLE  A-6.   COLLAPSED MEMBRANE STUDY  FOR CROCIPOLITE


                                    0.1  ym Pore  Size Millipore  Filters
Sample
1
2

3
4
5
6
7
8
9 '
10
Fiber Concentration
Mean
(106 Fib/Liter)
187
211

185
224
218
192
272
262
229
310
95% Confidence
Interval
(106 Flb/Uter)
110 - 265
180 - 242

134 - 237
167 - 281
180 - 255
134 - 251
187 - 357
203 - 322
128 - 329
146 - 475
Estimated Mass
Concentration
(M1crograms/L1ter)
43
22
1
28
26
23
70
107
49 . '
72
149
Concentration
Equivalent to
1 Fiber
Detected
(106 F1b/L1ter)
1.95
2.17

1.76
2.20
2.20
1.81
2.70
2.10
2.33
3.07
No.
Fibers
Counted
96
97

105
102
99
106
101
125
98
101
Ho.
Grid
Openings
6
5

6
5
5
6
4
5
7
4
Grid Distribution
X2
12.8
1.07

6.03
3.40
1.57
7.48
2.87
3.37
L8.0
8.29
Significance
of
Uniformity
(%)
2.5
75

25
25
75
10
25
25
0.5
2.5
o
co

-------
                            TABLE A-7.  COLLAPSED MEMBRANE STUDY FOR CROCIPOLITE

                                   0.22 pro Pore Size Millipore Filters
Sample
1
2
3
4
5
6
7
8
9 '
10 -
Fiber Concentration
Mean
(106 F1b/L1ter)
180
196
227
181
159
223
228
182
208
253
95% Confidence
Interval
(106 F1b/L1ter)
147 - 213
118 - 275
107 - 346
139 - 224
97.6.- 219
168 - 279
157 - 298
128 - 237
170 - 246
211 - 295
Estimated Mass
Concentration
(Mi crograms/Li ter )
32
24
39
46
•18
33
isq
,M
82
115
Concentration
Equivalent to
1 Fiber
Detected
(106 F1b/L1ter)
1.78
2.18
2.27
1.93
1.74
2.09
1.98
1.42
2.12
2.66
No.
Fibers
Counted
101
90
100
94
91
107
115
128
. 98
95
No.
Grid
Openings
6
5
5
6
6
5
5
7
6
5
Grid Distribution
X2
2.61
7.71
14.41
3.60
10.1
3.44
5.81
11.6
2.69
1.38
Significance
of
Uniformity
m
75
10
0.5
50
5
25
' 10
5
50
75
10
: 10 I

-------
                     APPENDIX B





               TABLES B-l TO B-16
REPLICATE ANALYSES OF ENVIRONMENTAL WATER SAMPLES:



        RESULTS AND STATISTICAL ANALYSES
                       110

-------
TABLE B-l.  BEAVER BAY. TAP WATER. INITIAL SAMPLE
Sub-
Sample
Number
1
2
3
4
5
6
7
8
9
10
"" AMPHIBOLE - SAED
Mean
Concentration
(106 fib/liter)
12.1
9.05
7.75
11.4
7.85
8.01.
5.84
7.89
9.50
9.79

ANALYSES OF
SUB-SAMPLES
i
lean Fiber
Concentration
(106 fib/liter)
lean Mass
Concentration
(H9/ liter)
95} Confidence
Interval
(106 flb/Hter)
9.8 -
6.81 -
5.95 -
8.7 -
5.91 -
5.82 -
4.21 -
6.55 -
6.35 -
7.05 -
14.4
11.3
9.55
14.1
9.79
10.2
7.47
9.23
12.6
12.5
Estimated Mass
Concentration
(pq/L)
20.7
6.77
4.87
8.82
5.50
2.59
;7.26
7.98
7.74
9.50
PROBABLE AMPHIBOLE
Mean
Concentration
(106 fib/liter]
8.05
9.22
8.20
7.02
8.64
7.13
7.06
6.95
8.22
5.89
AMPHIBOLE - SAED
Mean
(x)
8.92
8.17
Standard
Deviation
(S)
1.86
4.85
Variance
(S2)
3.47
23.6
95% Confidence
Interval
(106 flb/Hter)
5.36 -
6.98 -
6.27 -
5.08 -
6.54
4.77
4.99
4.97
6.50
4.58
• 10.7
• 11.5
• 10.1
• 8.96
- 10.7
- 9.49
- 9.13
• 8.93
- 9.94
- 7.20
Estimated Mas;
Concentration
(uq/L)
1.05
6.67
2.29
7.68
3.07
3.89
3.84
2.58
4.32
6.00
TOTAL FIBERS
Mean
Concentration
(106 fib/Uteri
20.2
18.3
16.0
18.4
16.5
15.1
12.9
14.8
17.7
15.7
PROBABLE AMPHIBOLE
Mean
(x)
7.64
2.70
Standard
Deviation
(S)
0.99
2.94
Variance
(S2)
0.98
8.64
istlma ted Mass
Concentration
(uq/L)
21.8
13.4
7.16
16.5
8.57
6.48
11.1
10.6
12.1
16.5
GRID DISTRIBUTION
(TOTAL FIBERS)
X*
29.54
23.54
19.12
38.34
29.17
33.99
23.67
13.93
34.62
29.58
Significance
Level of
Uniformity
(I)
5
10
10
0.5
5
1
10
10
1
5
TOTAL FIBERS
Mean Standard
Deviation
(x) (S)
16.6 2.11
12.4 4.77
Variance
(S2)
4.46
22.7

-------
                               TABLE B-2.   BEAVER BAY.  TAP WATER.  FINAL  SAMPLE
Sub-
Sample
Number
1
2
3
4
5
6
7
8
9
10
AHPHIBOLE - SAED
Mean
Concentration
(106 fib/liter)
5.79
7.08
5.31
4.88
5.77
5.30
4.94
5.22
3.59
4.07

ANALYSES OF
SUB- SAMPLES
I
lean Fiber
Concentration
(106 fib/liter)
lean Mass
Concentration
(ug/llter)
95% Confidence
Interval
(106 fib/liter)
4.52
5.31
3.89
3.74
3.89
3.66
3.77
3.39
2.17
2.64
7.06
• 8.85
• 6.73
• 6.02
- 7.65
- 6.94
- 6.11
• 7.05
- 5.01
- 5.50
Estimated Mass
Concentration
(ug/L)
11.80
2.96
8. 52
3.08
4.44
3.03
2.09
4.46
8.24
2.28
PROBABLE AMPHIBQLE
Mean
Concentration
(106 f1b/11ter]
5.70
3.63
3.60
5.32
4.74
5.21
4.39
3.48
5.34
3.89
AHPHIBOLE - SAED
Mean
(x)
5.20
5.09
Standard
Deviation
(S)
0.96
3.29
Variance
(S2)
0.92
10.8
95% Confidence
Interval
(106 fib/liter)
4.19
2.22
2.35
3.89
3.30
3.95
3.07
1.79
3.87
2.86
- 7.21
- 5.04
- 4.85
- 6.75
- 6.18
- 6.47
- 5.71
- 5.17
- 6.81
- 4.92
Estimated Mass
Concentration
(pq/U
0.63
1.23
1.56
10.40
4.74
2.14
2.55
1.53
1.33
0.97
TOTAL FIBERS
Mean
Concentration
(106 fib/Uteri
11.5
10.7
8.91
10.2
10.5
10.5
9.33
8.70
8.93
7.96
PROBABLE AMPHIBOLE
Mean
(x)
4.53
2.71
Standard
Deviation
(S)
0.84
2.94
Variance
(S2)
0.71
8.64
Estimated Mass
Concentration
d-fl/L)
12
4.
10
13
9.
5.
4
19
1
5
18
17
4.64
5.99
9.
57
• 3.25.
GRID DISTRIBUTION
(TOTAL FIBERS) -
X2
18.71
20.80
19.90
12.69
20.37
16.96
24.62
27.66
22.91
22.38
Significance
Level of
Uniformity
(*)
10
10
10
10
10
10
10
5
10
10
TOTAL FIBERS
Mean
(if)
9.73
7.80
Standard
Deviation
(S)
1.11
3.62
Variance
(S2)
1.24
13.1
ro

-------
TABLE B-3.  LAKE SUPERIOR WATER
Sub-
Sample
Number

1
2
3
4
5
6
7
8
9
10
AMPHIBOLE - SAED
Mean
Concentration
(106 fib/liter)
1.62
2.07
1.88
1.22 .
1.81
2.04
1.28 >
1.88
1 1.98 ' ;
3.27 V.

. ANALYSES OF
SUB-SAMPLES
lean Fiber
Concentration
(106 fib/liter)
lean Mass
Concentration
lug/liter)
951 Confidence
Interval
(106 fib/liter)
0.59 -
1.05 -
0.64 -
0.44 -
0.63 -
0.99 -
iO.51 -
0.94 -
0.84,-
1.23 -
2.65
3.09
3.12
2.00
2.99
3.09
2.05
2.82
3.12
.5.31
Estimated Mass
Concentration
(ng/L)
0.67
13.5
0.81
4.22
1.45
2.16
2.20
7.67
0.77 ..
3.20
PROBABLE AMPHIBOLE
Mean .
Concentration
(106 fib/ liter]
3.51- . •-.
4.13 !• ,
2.95
. 3.66 '-
3.62,
.3.40 .
2.71
2.55
4.10 ,
2.86
AMPHIBOLE -.SAB)
Mean • ; '
(0 '"
1.90
3.66
Standard
Deviation
',. 0.564
4.06
Variance
(S2)
0.318
16.5
95* Confidence
Interval
(106 fib/ liter)
2.15 -
2.07 -
1.25 -
2.04 -
2.31 -
1.89 -
0.88 -
0.93 -
2.33 -
1.64 -
4.87
6.19
4.65
5.28
4.93
4.91
4.54
4.17
5.87
4.08
Estimated Mas;
Concentration
(v.g/0
1.32
2.00
4.78
7.52
4.76
1.66
0.40
2.86
2.77
0.81
TOTAL FIBERS
Mean
Concentration
(106 fib/liter)
5.13
6.20
4.83
4.88
5.43
5.44
3.99
4.43
6.08
6.13 .
PROBABLE AMPHIBOLE
Mean
(x)
3.35 .
2.89
Standard
Deviation
(S)
0.56
2.21
Variance
(S2)
0.31
4.88
LStimated Mass
Concentration
(uq/L)
1.97
15.5
5.59
11.
7
6.21
3.82
2.60
10.
5
3.54 .
4.01
GRID DISTRIBUTION
(TOTAL FIBERS)
X2

15.38
23.32
43.05
18.85
18.26
18.73
43.05
23.81
25.53
23.54
Significance
Level of
Uniformity
(*)
10
10
0.1
10
10
IP •
0.1
10 .
10
10
TOTAL FIBERS
Mean
5.25
6.55
Standard
Deviation
(S)
0.75
4.51
Variance
0.56
20.4

-------
TABLE B-4.   DULUTH, RAW WATER
Sub-
Sample
Number

1
2
3
4
5
6
7 >
8
9
10
AHPHIBOLE - SAED
Hean
Concentration
(106 fib/liter)
1.65
1.41
. 1.77
1.64
0.68
1.08
1.63
1.57
1.88
1.79

ANALYSES .OF
SUB- SAMPLES
tean Fiber
Concentration
(106 fib/liter)
tean Mass
Concentration
tug/liter)
95t Confidence
Interval''
CIO6 flb/Uter)
0.95 -
0.87 -
1.02 -
0.82 -
0.34 -
0.42 -
0.80 -
1.06 -
1.20 -
1.17 -
AMPIIIB
Hean
(5)
' 1.51
1.27
2.35
1.95
<2.52
'2.46
1.02
1.74
2.46
2.08
2.56
2.41
Estimated Mass
Concentration
(1-9/0
1.86
0.75
0.84
1.39
0.40
0.77
2.04
0.77
0.83
3.00
PROBABLE AHPHIBOLE
Mean
Concentration
(10s fib/liter}
2.55
1.69
1.98
1.78
1.16
1.58 .
1.91
1.71
••' 2.44 •
- 2.34
aE - SAED
Standard
• Deviation
(S)
0.37
0.81
Variance
0.13 , .
0.65
95% Confidence
Interval
(106 fib/Uter)
1.60 -
0.90 -
1.28 -
0.98, -
0.57 -
0.92 -
1.17 -
0.91 -
1.72 -
1,47 -
3.50
-2.48'
2.68
2.58
1.75
2.24
2.65
.2.51
3.16
3.21
Estimated Mass
Concentration
(UQ/U
2.03
0.41
0.52 .
0.38
1.46
0.60
0.23
1.87 '*
0.68
1.01
TOTAL FIBERS
Hean
Concentration
(106 fib/liter)
4.20
3.10
3.75
3.42-
1.84
2.66
•'" 3.54
3.2& . .-
4.32 . .
4.13
PROBABLE AHPHIBOLE
Hean
(x) •
1.91
0.92
Standard
Deviation '
0.43
0.64
'Variance'^ '
.V.4S.2)...,
"..*•--•
0.41
Estimated Mass
Concentration
(vg/L)
3.89
1.16
1.36
1.77
1.86'
1-37. , ;
2.27
~ 2.64
. 1.51
4.01
GRID DISTRIBUTION
(TOTAL FIBERS)
X2

25.23
20.66
18.62
32.72
16.56
19.69
28.29
16.14
15.60
19.87
Significance
Level of
Uniformity
(*)
10-
10
10
2.5
10 '
10
5
10
10
10
TOTAL FIBERS
Hean '
,.: (x) |
- 3:42
2.18
Standard
Deviation •
(S)
0.76
1.03
Variance ~
.'"^6.58
1.06

-------
TABLE B-5.   SHERBROQKE,  TAP WATER.  INITIAL  SAMPLE
Sub-
Sample
Number
1
2
3
4
5
6
7
8
9
10
Mean
Concentration
(10e fiber/ liter)
34.2
51.3
50.9,
47.9
50.1
44.9
34.0
41.9
47.0
40.9
95X Confidence
Interval
(105 fiber/ liter)
22.6 - 45.8
40.5 - 62.1
37.7 - 64.1
38.5 - 57.3
36.7 - 63.5
32.4 - 57.4
26.1 - 41.9
34.3 - 49.5
32.6 - 61.4
32.5 - 49.3 .
Estimated Mass
Concentration
(n9/L)
0.22
0.25
0.31
0.28
0.53
0.27
0.14
0.18
. 0.32
0. 17
Grid Distribution
X2
24.68
14.85
11.77
6.98
13.44
12.59
11.84
5.26
16.04
6.54
Significance
Level of
Uniformity (20.
0.1.
. 5
10 .
10
' 5 .
. 5
10 .
10
1.0
10
OVERALL
VALUES
Mean Fiber
Concentration
(106 fib/liter)
. Mean Mass
Concentration
(ug/liter)
Mean
(x)
44.31
0.267
Standard
Deviation
(S)
6.43
0. 11
Variance
(S2)
41.36
1.2 x 10" 2
                   •  i 115 .

-------
TABLE B-6.  SHERBROOKE, TAP WATER. FINAL SAMPLE
Sub-
Sample
Number
i
2
3
4
5
6
7
8
9
10
Mean
Concentration
(106 fiber/liter)
45.3
44.7
41.0'
44.4
32.3
51.0
40.1
34.7
36.3
27.1
95% Confidence
Interval
(106 fiber/liter)
40.4 - 50.2
38.3 - 51.1
34.3 - 47.7
34.9 - 53.9
24.8 - 39.8
35.8 - 66.2
33.2 - 47.0
28.8 - 40.6
27.1 - 45.5
20.2 - 34.0
Estimated Mass
Concentration
(yg/L)
0.26
0.22
0.21
0.27
0.14
0.38
0.20
0.17
0.44
0.22
Grid Distribution
X2
1.71
5.68
6.87
7.58
10.81
17.16
4.32
3.80
8.17
11.12
Significance
Level of
Uniformity (%)
10
10
10
10
10
1
10
10
10
10
OVERALL
VALUES
Mean Fiber
Concentration
(106 fib/ liter)
Mean Mass
Concentration
(ug/1iter)
Mean
(x)
39.69
0.25
Standard
Deviation
(S)
7.13
9.3 x 10~2
Variance
(S2)
50.90
8.65 x 10" 3
                     116

-------
TABLE B-7.  MAGOG RIVER, SHERBROOKE
Sub-
Sample
Number
1
2
3
4
5
6
7
8
9
10
Mean
Concentration
(106 fiber/liter)
16.8
17.3
14.2
18.6
18.6
14.3
15.5
16.5
12.4
12.4
95% Confidence
Interval
(106 fiber/liter)
11.5 - 22.1
7.2 - 27.4
9.8 - 18.6
13.3 - 23.9
10.5 - 26.7
• 10.0 - 18.6
10.1 - 20.9
10.1 - 22.9 .
7.4-17.4 .
6.8 - 18.0
Estimated Mass
Concentration
(u9/D
0.13
0.12
0.045
0.13
1.09
0.38
0.13
,0.085 '
0.070
0.15
Grid Distribution
x2
5.79
18.56
4.04
4.85
11.74
4.12
6.20
8.29
6.31
7.99
Significance
Level of
Uniformity (%)
10
2.5
10
10
10
10
10
10
10
10
OVERALL
VALUES
Mean Fiber
Concentration
\ (106 fib/liter)
Mean Mass
Concentration
(ug/liter)
Mean
(x)
15.66
0.233
Standard
Deviation
(S)
2.29
0.315
Variance
(S2)
5.25.
9.9 x lO'2
                117

-------
TABLE B-8.  SHERBROOKE. RAW WATER
Sub-
Sample
Number
i
2
3
4
5
6
7
8
9
10
Mean
Concentration
(106 fiber/liter}
29.9
30.1
24.9
23.5
22.6
23.2
27.6
24.6
22.8
19.7
95% Confidence
Interval
(106 fiber/liter)
19.5 - 40.3
23.6 - 36.6
20.1 - 29.7
19.0 - 28.0
15.7 - 29.5
16.4 - 30.0
21.0 - 34.2
17.4 - 31.8
13.9 - 31.7 .
14.7 - 24.7
Estimated Mass
Concentration
(ua/L)
0.22
0.25
0.14
0.11
0.18
0.23
0.13 .
0.13 .
0.14
0.15
Grid Distribution
:x2
24.68
9.10
6.18
5.43
14.76
12.72
10.65
13.30
21.39
8.55
Significance
Level, of
Uniformity (%)
0.1
10
10
10
5
10
10
10
1.0
10
OVERALL
VALUES
Mean Fiber
Concentration
(106 fib/ liter)
Mean Mass
Concentration
(ug/ liter)
Mean
(x)
24.89
0.17
Standard
Deviation
(S)
3.35
4.9 x 10"2
Variance
(S*)
11.22
2.4 x 10"3
               118

-------
                            TABLE B-9.  STATISTICAL ANALYSIS OF REPLICATE SUB-SAMPLING
                               BULK SAMPLE;  BEAVER BAY. TAP WATER. INITIAL SAMPLE
Sub-Sample
Number
1
2
3
4
5
6
7,
8
9
10
Number of Fibers per Grid Opening
Overall Sample
Mean
Total Fibers*
X




•9.45





S




• 3.84





S^




•14.71





Sub-Samples
Total Fibers*
X
11.52
10.36
8.85
10.82
9.28
8.64
7.41
8.67
10.28
8.65
4
4'.16
, i
3;. 58
3.03
4. 73
3.83
3.90
3.04
2.54
4.33
3.62
i
s;5
17.32
12.85
9.19
22.40
14.64
15.21
9.24
6.45
18.75
13.10
Two- Sided
-t-test
x - x
2.07
0.91
0.60
1.37
0.17
0.81
2.04
0.78
0.83
0.80
u
2.01
1.75
1.50
2.27
1.86
1.86
1.50
1.29
2.09
1.76
Is Sub-Sample
Replicate of Mean
(at 5% Significance)
Marginal
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
VO
                            *Amphibole Random SAED plus Probable Amphibole

-------
                            TABLE B-10.  STATISTICAL ANALYSIS OF REPLICATE SUB-SAMPLING

                                BULK SAMPLE:  BEAVER BAY, TAP WATER. FINAL SAMPLE
Sub-Sample
Number
1
2
3
4
5
6
7'
8
9
10
Number of Fibers per Grid Opening
Overall Sample
Mean
Total Fibers*
X
-.



• 5.48





S




-2.47





S*




«6.09




.
Sub-Samples
Total Fibers*
X
6.46
6.08
4.93
5.87
6.09
5.94
5.13
4.88
4.85
4.41
Sn
2.50
2.56
2.28
1.99
2.56
2.34
2.57
2.59
2.43
2.36
£
6.26
6.56
5.20
3.97
6.54
.5.46
6.62
6.73
5.88
5.56
Two-Sided
-t-test
x - x
0.98
0.60
0.55
0.39
0.61
0.46
0.35
0.60
0.63
1.07
u
1.21
1.24
1.11
0.98
1.24
1.14
1.24
1.25
1.17
1.15
Is Sub-Sample
Replicate of Mean
(at 5% Significance)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
ro   i
o
                             *Amphibole Random SAED plus Probable Amphibole

-------
TABLE B-ll.   STATISTICAL ANALYSIS OF REPLICATE SUB-SAMPLING
            BULK SAMPLE:  LAKE SUPERIOR WATER
Sub-Sample
Number
1
2
3
4 •••••:
5
6
7.
8
9
10
Number of Fibers per Grid Opening
Overall Sample
Mean
Total Fibers*
X
-.



• 1.91




•
S
1



- 1.60





S2




»2.55





Sub-Samples
Total Fibers*
X
1.89
2.21
1.75
1.81
1.96
1.97
1.32
1.67
2.28
2.23
Sn
1.25
1.68
1.98
1.34
.1.37
1.38
1.76
1.54
1.74
1.87
$
1.56
2.81
3.91
1.78
1.89
; 1.90
3.09
2.37
3.04
3.50
Two-Sided
-t-test
x - x
0.02
0.30
0.16
0.10
- 0.05
0.06
0.59
0.24
0.37
0.32
u
0.62
0.81
0.95
0.66
0.67
o;e8
0.84
0.75
0.84
0.90
Is Sub-Sample
Replicate of Mean
(at 5% Significance)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
  *Amphibole Random SAED plus Probable Amphibole

-------
                           TABLE  B-12.  STATISTICAL  ANALYSIS OF  REPLICATE  SUB-SAMPLING

                                        BULK SAMPLE:   DULUTH.  RAW WATER
Sub-Sample
Number
1
2
3
4
5
6
7 ,
8
9
10
Number of Fibers per Grid Opening
Overall Sample
Mean
Total Fibers*
X




•2.45





S




•1.69





S2




•2.88




\
Sub-Samples
Total Fibers*
X
3.04
2.21
2.70
2.46
1.38
1.86
2.48
2.31
3.11
3.01
Sn
i
1^95
1.54
1.63
1.96
1.11
1.38
1.93
1.41
1.59
1.78
*i!
3.80
2.37
2.66
3.85
1.22
.1.90
3.73
2.00
2.53
3.17
Two-Sided
-t-test
x - x
0.59
0.24
0.25
0.01
1.07
0.59
0.03
0.14
0.66
0.56
u
0.94
0.75
0.79
0.94
0.56
0.68
0.93
0.70
0.77
0.86
Is Sub-Sample
Replicate of Mean
(at 5% Significance)
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
Yes
ro
ro
                             *Amphibole Random SAED plus Probable Amphibole

-------
                            TABLE  B-13.   STATISTICAL  ANALYSIS OF REPLICATE SUB-SAMPLING
                               BULK SAMPLE:   SHERBROOKE. TAP WATER.  INITIAL SAMPLE
Sub-Sample
Number
1
2
3
4
5
6
7.
8
9
10
Number of Fibers per Grid Opening
Overall Sample
Mean
Total Fibers*
X
-,



•16.21





S




- 5.42





S2




•29.37





Sub-Samples
Total Fibers*
X
12.50
18.80
18.25
17.25
18.88
16.63
12.60
16.00
17.75
14.88
sn
5^93
5.53
5.65
4.04
6.05
5,54
4.11
3.48
6.52
3.65
«S
35.16
30.58
31.92
16.32
36.60
30.69
16.89
12,11
42.51
13.32
Two- Sided
i-test
X - X
3.71
2.59
2.04
1.04
2.67
0.42
3.61
0.21
1.54
1.33
u
4.32
4.02
4.70
3.44
5.11
4.62
3.06
2.99
5.48
3.12
Is Sub-Sample
Replicate of Mean
(at 5% Significance)
Yes
Yes
Yes
Yes
Yes
Yes
No
Yes
Yes
Yes
ro
                            *Chrysotile SAED plus Chrysotile Morphology Only

-------
                           TABLE B-14.  STATISTICAL ANALYSIS OF REPLICATE SUB-SAMPLING
                                BULK SAMPLE:  SHERBROOKE. TAP WATER. FINAL SAMPLE
Sub-Sample
Number
•v
1
2
3
4
5
6
7 '
8
9
10
Number of Fibers per Grid Opening
Overall Sample
Mean
Total Fibers*
X




»14.45





S




- 4.62





S2
-.



'21.34




•
Sub-Samples
Total Fibers*
X
17.38
16,80
14.50
16.63
11.00
19.00
14.75
13.13
13.13
9.50
Sn
2.23
3.37
3.33
4.25
3.56
6.79
3.02
2.66
4.00
3.40
s^
4.97
11.36
11.09
18.06
12.67
46.10
9.12
7.08
16.00
11.56
Two- Sided
t-test
X - X
2.93
2.35
0.05
2.18
3.45
4.55
0.30
1.32
1.32
4.95
u
1.98
2.52
2.50
3.58
2.65
5.65
2.59
2.31
3.39
2.54
Is Sub-Sample
Replicate of Mean
(at 5% Significance)
No
Yes
Yes
Yes
No
Yes
Yes
Yes
Yes
No
ro
                              ^Chrysotile SAED plus Chrysotile Morphology Only

-------
                            TABLE B-15.   STATISTICAL ANALYSIS OF REPLICATE SUB-SAMPLING


                                      BULK SAMPLE:  MAGOG RIVER. SHERBROOKE
Sub-Sample
Number
1
2
3
4
5
6
7'
8
9
10
Number of Fibers per Grid Opening
Overall Sample
Mean
Total Fibers*
X




•2.88





S
-.



• 1.58





S2




'2.48





Sub-Samples
Total Fibers*
X
3.20
3.10
2.60
3.40
3.40 |
2.60
2.90
3.10
2.30
2.20
sn
1.42
2.54
1.13
1.34
2.08
1.09
1.42
1.69
1.29
1.39
si;
2.02
6.45
1.28
1.80
4.33
.1.19
2.02
2.86
1.66
1.93
Two-Si ded
*-test
X - X
0.32
0.22
0.28
0.52
0.52
0.28
0.02
0.22
0.58
0.68
u
1.04
1.82
0.33
0.99
1.51
0.81
1.04
1.23
0.95
1.02
Is Sub-Sample
Replicate of Mean
(at 5% Significance)
Yes
Yes
Marginal
Yes
Yes
Yes
Yes
Yes
Yes
Yes
ro  j
en  •
                            *Chrysotile  SAED plus Clirysotile Morphology Only

-------
                            TABLE B-16.   STATISTICAL ANALYSIS OF REPLICATE SUB-SAMPLING


                                        BULK SAMPLE:  SHERBROOKE. RAM WATER
Sub-Sample
Number
1
2
3
4
5
6
7 i
8
9
10
Number of Fibers per Grid Opening
Overall Sample
Mean
Total Fibers*
X
-,



»9.29




•
S
•j



•3.70





S2




•13.70


...


Sub-Samples
Total Fibers*
X
11.50
11.10
9.40
8.30
8.70
8.50
10.50
8.80
8.30
7.60
Sn
5.57
3.36
2.53
2.24
3.74
3.51
3.52
3.60
4.54
2.72
si
31.02
11.29
6.40
5.02
13.99
12.32
12.39
12.96
20.61
7.40
Two-Si ded
*-test
x - x
2.21
1.81
0.11
0.99
0.59
0.79
1.21
0.49
0.99
1.69
u
4.00
2.45
1.89
1.69
2.73
2.55
2.55
2.61
3.26
2.02
Is Sub-Sample
Replicate of Mean
(at 5% Significance)
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
ro
CT»
                             *Chrysotile SAED plus  Chrysotile Morphology Only

-------
         APPENDIX C





    TABLES C-r TO C-3
CONTAINER AND STORAGE STUDY:



 FIBER LENGTH DISTRIBUTIONS
             127

-------
                            TABLE C-l.   FIBER LENGTH DISTRIBUTION:  (1st  SERIES)
                                       (CUMULATIVE PERCENTAGE NUMBER)
Particle
Size
Range,
yltl
0.10 - 0.15
0.15 - 0.22
0.22 - 0.32
0.32 - 0.47
0.47 - 0.69
0.69 - 1.00
1.00 - 1.50
1.50 - 2.20
2.20 - 3.20
3.20 - 4.70
i
4.70 - 6.90
6.90 - 10.00
10.00 - 15.00
15.00 - 22.00
22.00 - 32.00
Container Material
Flint Glass
On
Site
0.00
0.00
1.83
11.01
41.28
69.72
89.91
96.33
98.17
99.08
00.00
.00.00
00.00
100.00
.00.00
Hand
Shake
0.00
0.00
0.00
13.33
47.62
75.24
89.52
98.10
98.10
98.10
98.10
100.00
100.00
100.00
100.00
Routine
Prep.
0.00
0.00
0.92
7.34
47.71
73.39
89.91
93.58
97.25
99.08
100.00
100.00
100.00
100.00
100.00
48 hr.
Shake
0.00
0.00
0.00
9.71
46.60
72.82
85.44
97.09
98.06
99.03
100.00
100.00
100.00
100.00
100.00
Ozone
UV
O.OC
O.OC
O.OC
8.4<
47.17
72.64
92.45
98.11
100. OC
100. OG
100. OC
100. OC
100. OC
100. OC
100.00
Polypropylene
On
Site
0.00
0.00
0.00
17.76
60.75
81.31
91.59
98.13
98.13
99.07
LOO. 00
100.00
100.00
100.00
100.00
Hand
Shake
0.00
0.00
0.00
12.50
41.67
76.39
83,33
91.67
98.61
98.61
98.61
98.61
100.00
100.00
100.00
Routine
Prep.
0.00
0.00
0.00
10.89
55.45
73.27
91.09
95.05
97.03
98.02
100.00
100.00
100.00
100.00
100.00
48 hr.
Shake
0.00
0.00
0.00
13.86
55.45
78.22
89.11
98.02
99.01
99.01
100.00
100.00
100.00
100.00
100.00
Ozone
UV
0.00
0.00
0.00
11.01
57,80
82.57
89.91
96.33
98.17
100.00
100.00
100.00
100.00
100.00
100.00
Polyethylene
On
Site
0.00
0.00
0.00
18.02
65.77
82.88
93.69
97.30
99.10
99.10
100.00
100. OC
100.00
100.0C
100. OC
Hand
Shake
0.00
0.00
1.00
17.00
54.00
74.00
90.00
97.00
99.00
100.00
100.00
100.00
100. OC
100. OC
100. OC
Routine
Prep.
0.00
0.00
0.95
12.38
57.14
81.90
95.24
97.14
98.10
99.05
99.05
100.00
100.00
100.00
100.00
48 hr.
Shake
0.00
0.00
0.00
7.14
44.90
74.49
89.80
98.98
98.98
100.00
100.00
100.00
100.00
100.00
100.00
Ozone
UV
0.00
0.00
0.00
11.71
55.86
84.68
90.99
97.30
99.10
100.00
100.00
100.00
100.00
100.00
100.00
00

-------
TABLE C-2.   FIBER LENGTH DISTRIBUTION:  (2nd SERIES)
         (CUMULATIVE PERCENTAGE NUMBER)
Particle
Size
Range,
iim
0.10 - 0.15
0.15 - 0.22
0.22 - 0.32
0.32 - 0.47
0.47 - 0.69
0.69 - 1.00
1.00 - 1.50
1.50 - 2.20
2.20 - 3.20
3.20 - 4.70
i
4.70 - 6.90
6.90 - 10.00
10.00 - 15.00
15.00 - 22.00
22.00 - 32.00
Container Material
Flint Glass
On
Site
o.oc
o.oc
o.oc
le.s;
55.45
83.17
93.07
98.02
99.01
99.01
99.01
100.00
.00.00
LOO. 00
LOO. 00
Hand
Shake
O.OC
O.OC
O.OG
8.4S
59.15
81.69
92.96
98.59
100.00
100.00
100.00
100.00
100.00
100.00
100.00
Routine
Prep.
0.00
0.00
1.08
9.14
54.84
78.49
90.86
97.31
98.39
100.00
100.00
100.00
100.00
100.00
100.00
48 hr.
Shake
0.00
0.00
0.00
15.04
58.41
78.76
96.46
99.12
100.00
100.00
100.00
100.00
100.00
100.00
100.00
Ozone
UV
0.00
0.00
0.89
22.32
68.30
87.05
94.64
96.43
97.77
99.11
99.55
99.55
lOO.OO
,00.00
.00.00
Polypropylene
On
Site
0.00
0.00
0.00
10.67
54.67
78,67
93.33
97.33
97.33
98.67
98.67
LOO.OO
100.00
100.00
LOO.OO
i
Hand
Shake
0.00
0.00
2.50
37.50
52.50
72.50
85.00
92.50
95.00
95.00
100.00
LOO.OO
LOO.OO
100.00
100.00
Routine
Prep.
0.00
0.00
0.00
18.35
61.47
85. 32
95.41
99.08
99.08
99.08
99.08
100.00
100.00
100.00
100.00
48 hr.
Shake
0.00
0.00
0.00
3.77
43.40
79.25
94.34
96.23
98.11
100.00
100.00
100.00
100.00
100.00
100.00
Ozone
UV
0,00
0.00
0.00
15.13
68.07
88.24
95.80
98,32
99.16
100.00
100.00
100.00
100.00
100.00
100.00

Polyethylene
On
Site
0.00
0.00
1.11
18.89
57.78
80.00
90.00
95.50
100.0(
LOO.OO
100.00
LOO.OO
100.00
100.00
100.00
Hand
Shake
0.00
0.00
1.75
21.05
64.91
84.21
94.74
98.25
98.25
98.25
100.00
100.00
100.00
100.00
100.00
Routine
Prep.
0.00
0.00
0.00
20.37
66.67
81.48
93.52
98.15
98.15
99.07
100.00
100.00
100.00
100. 00
100.00
48 hr.
Shake
0.00
0.00
3.45
6.90
51.72
86.21
89.66
100.00
100.00
100.00
100.00
100.00
100.00
100.00
100.00
Ozone
UV
0.00
0.00
0.00
18.47
67.52
86.62
94.90
97.45
98.73
99.36
99.36
100.00
100.00
100.00
100.00

-------
                               TABLE C-3.   FIBER LENGTH DISTRIBUTION:  (3rd SERIES)

                                           (CUMULATIVE  PERCENTAGE NUMBER)
Particle
Size
Range,
urn
0.10 - 0.15.
0.15 - 0.22
0.22 - 0.32
0.32 - 0.47
0.47 - 0.69
0.69 - 1.00
1.00 - 1.50
1.50 - 2.20
2.20 - 3.20
3.20 - 4.70
4.70 - 6.90
6.90 - 10.00
10.00 - 15.00
15.00 - 22.00
22.00 - 32.00

Flint Glass
On
Site
0.00
0.00
1.87
24.30
66.36
89.72
95.33
98.13
98.13
100.00
LOO. 00
LOO. 00
100.00
100.00
100.00
Salnpl ing
Pipet*
0.00
0.00
1.02
11.22
64.29
87.76
95.92
95.92
96.94
97.96
100.00
IOO.OO
100.00
100.00
100.00
Routine
Prep.
0.00
0.00
0.00
9.80
51.96
77.45
87.25
96.08
99.02
100.00
100.00 ,
100.00
IOO.OO
100.00
100.00
48 hr.
Shake






0)
ID
O
•r-
O.
5-
4J
O
z




r.
Ozone
UV
0.00
0.00
0.83
14.17
58. 33
83.33
93.33
98.33
99.17
99.17
99.17
100.00
100.00
100.00
100. OC
Container Material
Polypropylene
On
Site






•o
Ol
N
CO
C
o





Sampling
Plfeti
0.00
0.00
0,00
8.57
60.00
77.14
82.86
88.57
97.14
IOO.OO
100.00
100.00
100.00
IOO.OO,
'100 ;00
Routine
Prep.
0.00
0.00
0.00
.15:24
54.29
.76.19
87.62
92.38
96.19
99.05
99'. 05
100.00"
100.00..
100.00
100.00
48 hr.
Shake






0)' .".
ID
O
O.
5-
0 .
z •
o,-
- <


- —
Ozone
UV
0.00
0.00
0.00
11.72
54.69
79.69
92'. 19
98.44
98.44
ioo.oo
100.00
100700
100 .'00
IOO.OO
mo. p.o

Polyethylene
On
Site






TJ
C
O




.... .
Sampling
by ^
0.00
'o.oo
0.00
9.00
'55.00
84.00
92.00
: 97.00
99.00
99.00
99.00.
99.00
100.00
100.00
100.00
Routi ne
Prep.
0.00
0.00
0.00
13.51
52.25
.80.18
92.79
99.10
100.00
100.00
100.00.
100.00
100.00
100.00
100.00
48 hr.
Shake






to
O
tQ.
CL
1




•
Ozone
UV
0.00
0.00
0.00
8.00
52.00
75.20
92.00
96.80
98.40
99.20
100.00
100.00
100.00
100.00
100.00
10
o
                         '*Sub-'sample taken from bottle before any shaking.

-------
         APPENDIX D





   TABLES D-l AND D-2
CONDENSATION WASHER STUDY:



 DETAILED ANALYTICAL DATA
           ,131 :

-------
                            TABLE D-l.  CONDENSATION WASHER STUDY USING CHRYSOTILE

Sample

6
7
9
10
Fiber Concentration >
Mean
(106 Fib/Liter)
298
235
268
231
951 Confidence!
Interval j
(10? Fib/Liter) '
199 - 397 . !.
189 - 281 :
151 - 384 ',
176 - 287 ;
Estimated Mass
Concentration
(Nanograms/Llter)
2750
1180
1200
1120
Concentration
Equivalent to
1 Fiber
Detected
(106 Fib/Liter)
2.38
2.20
2.52
2.18 "
No.
Fibers
Counted

125
107
106
106
Ho.
Grid
Squares

5
6
5
6
Grid Distribution
X2

7.18
3.16
10.1
4.56
Significance
of
Uniformity
W
10
50
2.5
25
: CO
• ro

TABLE D-2.

•• ' .•"' : • " • ' i; •; .' ' . V .
. , t • j •
CONDENSATION WASHER STUDY USING' CROCI OOLITE '
1- , '
i
• i •
i

Sample
i
2

3
4
5
Fiber Concentration .
Mean •:
(106 Fib/Liter)
226

236
285
226
952 Confidence
Interval
(106 Fib/Liter)
163 - 289
i
168 - 304
216 - 353 I
128 - 325
Estimated Mass
Concentration
Nicrograms/Liter)
137

152
20.5
22.2
Concentration
Equivalent to
1 Fiber
Detected
(106 Fib/Liter)
2.31

2.20
2.66
2.41
No.
Fibers
Counted

98

107
107
94
No.
Grid
Squares

5

5
5
5
Grid Distribution
X2

3.94

4.55
3.21
9.60
Significance
of
Uniformity
(*)
25

25
50
2.5

-------