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 LaboratoryAthens 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
------- |