United States
Environmental Protection
Agency
Environmental Research
Laboratory
Athens GA 30613
Research and Development
EPA-600/S4-83-044 Nov. 19
&ER& Project Summary
Asbestos Fiber Determination in
Water Samples: Preparation
Techniques, Improved Analytical
Method, and Rapid Screening
E. J. Chatfield, M. J. Dillon, P. Riis, and W. R. Stott
A three-year study was conducted to
improve the analytical method for
determination of asbestos fiber
concentrations in water samples. The
research produced an improved sample
preparation and analysis methodology
and an alternative method with the
potential for routinely screening drink-
ing water samples for asbestos.
Sample preparation techniques
investigated included the "drop" and
collapsed membrane filter techniques.
When these two techniques were
compared to the carbon-coated
Nuclepore technique using a polycar-
bonate filter, the carbon-coated
Nuclepore technique proved to be the
superior method of sample preparation.
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.Initial experiments were carried
out to determine the effectiveness of
the attachment phenomenon as a fiber
separation technique. Investigation of
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
crocidoIKe 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
statistically compatible with those
obtained at the time of sample
preparation.
An improved analytical method for
determination of asbestos fiber concen-
trations in water samples was
developed. In this method, the water
sample is initially treated with ozone
and UV light to oxidize suspended
organic materials. The water sample is
then filtered through a capillary-pore
polycarbonate filter (0.1 fjm pore size),
after which the filter is prepared by
carbon extraction replication for
examination in a transmission electron
microscope (TEM). Fibers are classified
using selected area electron diffraction
(SAED) and energy dispersive X-ray
analysis (EDXA). Measurement of char-
acteristic features on a recorded and
calibrated SAEO pattern is specified for
precise identification of chrysotile.
Quantitative determination of the
chemical composition and quantitative
interpretation of at least one calibrated
zone axis SAED pattern are specified
for precise identification of amphibole.
-------
Mineral identification and generation of
the standard reporting format specified
for the fiber count results are achieved
by using two computer programs that
are integral to the analytical method.
The high cost of asbestos fiber
counting using the TEM led to the
requirement for an inexpensive and
rapid method by which water samples
having low fiber concentrations can be
excluded from further analysis.
Alignment of asbestos fibers in
magnetic fields, combined with
measurements of the scattered light
from the aligned dispersions, was
investigated as a rapid analytical
technique. A rapid, fixed-fiber
alignment method and a dynamic
method of fiber measurement were
studied. The dynamic fiber method
proved to be the more sensitive
method. Detection limits of 0.5 million
fibers per liter (MFL) and 5 MFL were
achieved for crocidolite and chrysotile,
respectively. These detection limits
were achieved directly from the water
sample without any preconcentration
steps. The scattered light measurement
techniques were applied to the'
determination of the fiber concen-
trations in drinking water samples from
three sources, and the results were
consistent with those obtained inde-
pendently by transmission electron
microscopy.
Using the dynamic fiber method,
mineral species other than the asbestos
varieties were examined to determine
possible interferences. The results
indicated that nonfibrous material that
rotates with the magnetic field yields
broad scattered light maxima at about
45 degrees to the magnetic field
direction. Qualitative measurements
showed that many other fibrous mineral
species yielded alignment modes
similar to those obtained with the
asbestos varieties.
The analytical time required for a
single measurement using the current
instrumentation is less than 10
minutes. Labor requirements for a
sample preparation are variable
depending on the fiber concentration
steps incorporated, but these need not
exceed one man-hour per sample.
This Project Summary was developed
by EPA's Environmental Research Lab-
oratory. Athens, GA. to announce key
findings of the research project that is
fully documented in three separate
reports (see Project Report ordering
information at back).
Introduction
The Preliminary Interim Method for
Determining Asbestos in Water was
issued by the U.S. Environmental Protec-
tion Agency's Environmental Research
Laboratory in Athens, Georgia, in July,
1976. 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 TEM. 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
Nuclepore®* technique that used a poly-
carbonate 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 first part of the research program
described here was directed towards the
development of improved techniques for
the analysis of asbestos in water.
Although the revised interim method had
achieved substantial acceptance, other
techniques of specimen preparation
remained in use, including various
"drop" methods and the collapsed
membrane filter method. The most effec-
tive method was to be selected and used
to study its analytical reproducibility in
determining the levels of asbestos in
drinking water and drinking water
supplies. In addition, better techniques
were sought for the removal of interfering
organic materials because of the low
temperature plasma ashing procedure
had proved to be unsatisfactory in a
number of ways.
Clearly defined, unequivocal methods
for fiber identification, particularly of
amphibole asbestos, were not
incorporated in the Interim Method. It
was recognized that adequacy of fiber
identification procedures was a major
issue when results were discussed,
particularly if analyses were the subject
of litigation. In other situations, where a
large number of analyses were required,
complete identification of each fiber was
not economically possible. A substantial
component of the research program was
devoted to development of a systematic
multi-level fiber classification and
identification system. A standardized
form of data reporting was also required.
•Mention of trade names or commercial products
does not constitute endorsement or recommenda-
tion for use by the U.S. Environmental Protection
Agency.
The TEM fiber identification and count-
ing procedure in the approved analytical i
method is labor-intensive, and the '
resulting high analytical costs have
limited the extent to which water supplies
can be monitored routinely for asbestos
fibers. The cost factor has led to a
requirement for a rapid method that
would allow samples containing less than
a pre-defined concentration of asbestos
fibers to be eliminated from further
analysis, allowing the TEM characteriza-
tion to be confined to those samples that
require a more detailed analysis. A rapid
method is also required for routine
monitoring of fiber concentrations in
water sources where asbestos fibers are
known to be present at concentrations of
concern. To meet these needs, the
research program investigated a rapid
technique based on the measurement of
the light scattered by magnetically
aligned asbestos fibers.
The research project is described in
three reports: Development of Improved
Analytical Techniques for Determination
of Asbestos in Water Samples, Analytical
Method for Determination of Asbestos
Fibers in Water, and Rapid Screening
Technique for Detection of Asbestos
Fibers in Water Samples. While supplies
last, copies of Analytical Method for
Determination of Asbestos Fibers in
Water are available from ORD
Publications, Center for Environmental
Research Information, USEPA,
Cincinnati, OH 45268.
Conclusions and
Recommendations
Improved Sample Preparation
The investigation of specimen prepara-
tion techniques for asbestos fiber count-
ing by TEM 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 dis-
tribution was not sufficiently uniform to
warrant its use in quantitative
determinations. The collapsed
membrane filter method was shown to
display strongly size-dependent fiber
losses relative to the Nuclepore
preparation.
For chrysotile, the fiber losses
increased with the pore size of the
membrane filters used. For the 0.45 /urn
and 0.22 /um pore size filters, the losses
were statistically significant at the 5%
-------
level; for the 0.1 yarn pore size filters, fiber
losses were not significant at the 5%
level. For fibers shorter than 1.0 /jm, the
fiber losses using the 0.45 //m pore size
filter were very high, and only 24% of
these fibers were transferred to the TEM
sample. The corresponding value for the
0.22 //m 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 procedure.
The results for crocidolite were more
difficult to interpret. Although the total
fiber losses using the 0.22pm and 0.1 /JTT\
pore size filters were not significant at the
5% level, significant losses were
observed in some fiber size ranges.
On the basis of the results, the
collapsed membrane method was found
to be unsatisfactory for quantitative
analyses. It is recommended, however,
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
prepared using the carbon-coated
Nuclepore technique. It was shown that
10 replicate measurements from each of
these samples were statistically compat-
ible. The same conclusion was drawn for
samples contaminated by amphibole
fibers collected near Duluth, 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 and short wavelength (254
nm) uv light was found to be successful.
This oxidation technique was found to
remove those organic components of
drinking water samples that inhibit
filtration and to be an effective and more
convenient replacement for oxidation by
two-step filtration and low-temperature
ashing. When the ozone-uv light
technique was used, no changes in either
the electron diffraction behavior or the
chemical compositions of chrysotile and
amphibole fibers were detected. A
second oxidation technique, based on the
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 morphol-
ogy was observed after treatment at the
most extreme conditions. It was also
found that containers made of polytetra-
fluoroethylene were required, because
both glass and silica were attacked under
the extreme conditions used. Because of
the success of the ozone technique, no
further investigation of the more involved
high-pressure method was conducted.
For samples containing large concentra-
tions of refractory organics, such as
sewage or plant effluents, the high-
pressure 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
reduced 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.
Apparently, the presence of the trace
organic materials of bacterial origin in
some way promoted the attachment of
asbestos fibers to the inside surfaces of
the containers. This effect could
seriously compromise the results of
routine sample analyses. Container and
storage studies indicated that the effect
was a consequence of an organic product
of bacteria, rather than a mechanism
involving direct interaction with the
^bacteria themselves, and that the organic
material was probably a variety of
polysaccharide. The effect of this
phenomenon on routine sample analyses
could be eliminated by ozone 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 the development of a
separation technique that was at least
partially specific for chrysolite. Reference
fiber suspensions have been required for
some time in order to facilitate analytical
suspensions have been required for some
time in order to facilitate analytical
quality assurance programs, but their
stability has always been in question. 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 certifi-
cation of their contents, so that analytical
quality assurance of future sampling
programs can be established by incorpo-
rations of control samples. Also, the
separation method should be developed
further and the mechanism that 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 a significance in other
fields unrelated to analytical method
development.
Improved Analytical Method
The improved analytical method devel-
oped in the study represents the best
available technology for determination of
asbestos fibers in water. 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 reporting, a standardized
reporting format, quantitative interpreta-
tion of fiber identification data, and a
fibrosity index that appears to permit
discrimination of fibrous and nonfibrous
species. The fiber classification system
that was developed recognizes
instrumental limitations, and if required,
permits later reevaluation of the raw data
using different fiber identification criteria.
A computer program was written that
permits fiber identification on the basis of
EDXA and zone axis SAED patterns. The
identification procedure operates by
selection of minerals that are consistent
with the measurements, using a library of
data from 226 minerals. A computer
program for reporting of fiber counting
data in a standardized format was also
established. The fiber identification
protocol based on zone axis SAED and
quantitative EOXA is capable of more
specificity than had previously been
provided by TEM analysis. The
identification procedure permits deter-
mination of approximate chemical com-
position, 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. The study
recommends 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 changes and
additions introduced into the basic
analytical method should eliminate the
problems of poor interlaboratory repro-
ducibility that have been observed in the
past.
Rapid Screening
The measurement of scattered light
from magnetically aligned asbestos fibers
has been demonstrated as a suitable
method for detection of asbestos fibers in
water samples. Two techniques were
investigated.
A rapid, fixed fiber alignment method
was studied in which an aqueous fiber
dispersion is filtered through a
membrane filter located in a strong
magnetic field. This results in a filter on
which the asbestos fibers are perma-
nently aligned in preferred directions.
When the filter is exposed to solvent
vapor, the structure collapses and the
filter becomes transparent. Rotation of
the filter in a collimated beam of light
yields maxima in the intensity of the
scattered light; the positions of these
maxima are related to the alignment
direction of the fibers.
A dynamic method of fiber
measurement also was investigated in
which the behavior of aqueous asbestos
fiber dispersions in a rotating magnetic
field is observed. A spectrophotometer
cell containing the fiber dispersion is
placed between the poles of a rotating
magnet and illuminated by a collimated
beam of light. The fibers rotate in syn-
chronism with the magnetic field, and
maxima in the intensity of the forward
scattered light are observed. Because
light is scattered more strongly in direc-
tions normal to the lengths of fibers, a
maximum in intensity of the scattered
light occurs for every 180 degrees of fiber
rotation. When the scattered light is
monitored and the intensity displayed as
a function of magnet position, the areas
under the peaks are proportional to the
fiber concentration.
For the fixed fiber alignment method,
the lowest detection level was limited by
the residual structure in the collapsed
membrane filter. For the UICC crocidolite
and UICC amosite, the detection level
was about 0.1 ng/mm2, and for chrysotile
about 1 ng/mm2. If 25-mm-diameter
filters were used, these detection levels
correspond to filter loadings of about 20
ng and 200 ng. respectively. In order to
detect a concentration of 0.2 MFL of
chrysotile, filtration of about 100L of
water through an active filter area of 200
mm2 would be required. Signal
enhancement techniques such as RF
plasma etching of the filter and
shadowing of the paniculate by evapo-
rated metal films failed to improve the
detection limits significantly. The
dynamic fiber method achieved much
lower detection limits of 0.5 MFL (180
ng/L) and 5 MFL (30 ng/L) for crocidolite
and chrysotile, respectively. These
detection limits apply to the 5mL volume
of aqueous fiber dispersion in the
spectrophotometer cell and correspond to
detection of 0.9 ng of crocidolite and 0.15
ng of chrysotile. The required detection
limits of 0.2 MFL or 1 ng/L can be
achieved with the incorporation of a
selective fiber concentration technique.
A limited study was made of the high
gradient magnetic separation (HGMS)
technique for amphibole fibers. A new
method was also devised for separation
of chrysotile fibers by selective adhesion
to organic materials.
In the investigation of light scattering
from liquid suspensions, it was found that
particles of random shape that rotate with
the magnetic field produce a broad
maximum of scattered light intensity
corresponding to alignment at an angle of
45 degrees to the magnetic field
direction. This effect was observed, for
example, with borosilicate glass frag-
ments, and is in contrast with the simple
increase in constant scattering obtained
from particulate that is unaffected by the
magnetic field. In general, however, the
presence of other particulate degrades
the detection limit, and therefore specific
fiber separation techniques were investi-
gated. HGMS was successful in extracting
UICC crocidolite and amosite, having a
95% numerical collection efficiency for
numerical collection efficiency for
dispersions of amosite. Because they are
not strongly magnetic, fibers of chrysotile
were not retained by the magnetic
separator. Therefore HGMS is a useful
technique for separating chrysotile
asbestos from amphibole asbestoses that
contain high concentrations of iron.
A new separation technique based on
scavenging of fibers by organic materials
was successful for specific separation of
chrysotile. The same technique also
appears to allow concentration of croci-
dolite and amosite fibers, but separation
specificity has not yet been established.
The recovery of separated chrysotile was
between 87% and 100% for standard
dispersion, falling to about 45% in the
case of drinking water samples.
Three municipal water supplies were
analyzed directly by the rotating fiber
method. The particulate of random
shapes yielded prominent, broad peaks at
45 degrees and 225 degrees and it was
necessary to perform profile subtractions
in order to extract the signal originating
from the fibers present. The residual
peaks after this procedure agreed with
the known asbestos fiber levels. For a
water sample from Beaver Bay,
Minnesota, the 45 degree component was
subtracted and this resulted in residual
peaks at 0 degrees and 180 degrees and
at 90 degrees and 270 degrees. This
agrees with the known asbestos content
of the water: cummingtonite is known to
align parallel and grunerite normal to the
magnetic field direction. It was possible to
measure directly the chrysotile fiber
concentration in a municipal water that
had a concentration of 40 MFL. Applica-
tion of the fiber separation technique to
the same sample yielded a concentrated
suspension for analysis that contained
chrysotile.
Development of computer profile
subtraction techniques will permit the
separation of the components corres-
ponding to mineral fibers from the total
scattered light profile. This refinement
will reduce the amount of sample
preparation required for separation and
pre-concentration. Variation of the rota-
tion rate and strength of the magnetic
field may provide additional information
by which particle species may be
differentiated. In routine use, it is
estimated that water samples could be
analyzed directly in five to ten minutes,
whereas samples requiring separation or
pre-concentration would require less
than one man-hour for preparation and
analysis.
The alignment modes of a number of
fibrous mineral species in a magnetic
field were investigated qualitatively.
Some yielded broad scattered light
profiles similar to those from chrysotile;
others displayed sharper peaks from
fibers aligned in directions parallel or
normal to the magnetic field. If the
primary purpose is the detection of
"asbestos" then there is some potential
for interference by fibrous species other
than those normally considered to be
asbestos. Assuming that the purpose of
the technique is to determine whether
any fibrous mineral species are present,
then it is highly successful, extremely
sensitive, and allows for some discrimi-
nation between mineralogical species.
Assuming some pre-concentration of
the sample, the magnetic alignment
technique has the required detection
level and sensitivity for measurement of
-------
fiber concentrations in water. It is capable
of significant further development,
particularly for the determination of fiber
dimensions. More extensive fiber charac-
terization could also be achieved on the
basis of iron content and alignment
mode. Additional research is also
required to optimize the specific fiber
separation techniques.
E. J. Chatfield, M. J. Dillon, P. Riis. and W. R. Stott are with the Ontario Research
Foundation, Mississauga, Ontario, Canada L5K 1B3.
J. M. Long is the EPA Project Officer (see below).
This Project Summary is based on the three reports listed below:
"Rapid Screening Technique for Detection of Asbestos Fibers in Water
Samples,"(Order No. PB83-262 915; Cost: $11.50)
"Development of Impro vedA nalytical Techniques for Determination of A sbestos
in Water Samples," (Order No. PB 83-261 651; Cost: $ 14.50)
"Analytical Method for Determination of Asbestos Fibers in Water," (Order No.
PB 83-260 471; Cost: $23.50)
The above reports are available only from: (costs subject to change)
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Research Laboratory
U.S. Environmental Protection Agency
College Station Road
Athens, GA 30613
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
BULK RATE
U.S. POSTAC
PAID
Cincinnati, O
Permit No. G
Official Business
Penalty for Private Use $300
U.S. GOVERNMENT PRINTING OFFICE: 1983-759-102/Oi
------- |