United States
Environmental Protection
Agency
Environmental Sciences Research
Laboratory
Research Triangle Park NC 27711
Research and Development
EPA-600/S2-80-200 May 1981
Project Summary
Feasibility Study for an
Asbestos Aerosol Monitor
Catherine H. Skintik
This research project was initiated
to determine the feasibility of discrim-
inating and counting asbestos-fiber
aerosol particles by means of their
shapes, using a two-detector optical
aerosol counter. The asymmetry of
their optical diffraction patterns dis-
tinguishes fibers from other more
regular aerosol particles. A laboratory
prototype was designed, constructed,
tested, modified, and retested. The
feasibility of counting fibers by means
of their optical diffraction patterns,
using two detectors, was demon-
strated. In the latest version of the
device, the aerosal fibers were aligned
by a combination of fluid-velocity-
gradient and electric fields. A descrip-
tion of the latest version, including its
capabilities and its limitations, is pre-
sented in this summary. Testing of this
version under controlled conditions
had just begun at the end of the project
period.
This Project Summary was devel-
oped by EPA's Environmental Sciences
Research Laboratory, Research Tri-
angle Park, NC, to announce key
findings of the research project in a
separate report of the same title (see
Project Report ordering information at
back).
Introduction
Airborne, respirable asbestos fibers
are known to constitute a health hazard
to human beings. Therefore, it is neces-
sary to monitor the concentration of
asbestos-fiber aerosols in environments
where the potential for such a hazard
exists.
The fibrous aerosol monitor developed
in this study (known as the UVM for
University of Vermont) is characterized
by (1) a continuous sampling of the
ambient air; (2) passage of the sampled
air through a small, sensitive volume,
illuminated by an intense beam of light;
(3) alignment of fibrous aerosol particles
in the sensitive volume by a combina-
tion of flow-velocity-gradient and electric
fields; (4) simultaneous measurements
of the pulses of light scattered in dif-
ferent directions, when an aerosol
particle traverses the sensitive volume,
by two or more photomultiplier tubes;
(5) classification of each particle as a
fiber or nonfiber on the basis of the
ratios of intensities of light scattered in
different directions; and (6) display of
the current fiber concentration with
only a few seconds' delay.
The device described here detects
fibers with a reduced counting efficiency
if they are randomly aligned, and the
original proposal was based on the
assumption of random alignment. How-
ever, it was discovered late in the
project period that the combination of
flow and electric field alignment was
both effective and convenient; it offers
enormous advantages in interpreting
data produced by the instrument, as
well as greatly increased efficiency in
counting fibers.
The light-scattering fibrous aerosol
monitor counts all types of fibers in the
appropriate size range. Because it is
nonspecific, the monitor should be used
in situations where there is reason to
suspect the presence of a particular kind
of fiber. The fiber counter's nearly con-
-------
tinuous readings can be used as an
alarm to trigger more detailed analytical
procedures, which are too slow and
expensive for routine use in continuous
monitoring.
By its very nature, an optical fiber
counter can operate properly only within
a certain range of particle sizes and
concentrations. Particles that are too
small or too large may not be detected or
may overload the instrument. Low fiber
concentrations may take too long to
measure. High dust concentrations may
cause coincidence errors when two
particles are in the sensitive volume
simultaneously. Measurements of low
fiber concentrations in high background
concentrations of amorphous dust are
doubly difficult. This study has been
designed to extend the lower limit of
reliably detectable fiber sizes and to
detect small numbers of fibers among
much larger numbers of background
dust particles.
The UVM Fibrous Aerosol
Monitor
The two-detector, aerosol-fiber
counter depends on the optical principle
illustrated in Figure 1. In the figure, a
light beam is incident from the left on a
fiber oriented with its long dimension
perpendicular to the beam. Thef iber is a
fraction of a micron in diameter, com-
parable to the wavelength of light, and
many microns long. Some of the light is
scattered andforms a diffraction pattern
on a distant screen. The bright pattern is
wider in the equatorial plane of the fiber
and narrower in the plane that contains
the long dimension of the fiber and the
incident beam. If a photodetector were
placed at position 1 on the screen, it
would measure a higher light intensity
than one placed at location 2, even
'Light
Beam
Figure 1. Diffraction pattern of a
fibrous particle.
though both points are equally distant
from the center of the diffraction pattern.
This would not be the case for a spherical
particle, and therefore, the principle can
be used to distinguish fibers from
spheres and near-spheres. Fibers of
known orientation are most easily
detected, but even randomly oriented
fibers can be detected with a certain
probability.
A simplified schematic of the airflow
system, which resembles that of many
optical particle counters, is shown in
Figure 2. The sensitive volume, about 1
mm3 in size, is at the intersection of the
inlet air stream and a 3 Mw, circularly
polarized He-Ne laser beam. The sample
air is surrounded by a sheath of clea n air
as it traverses the open space between
the inlet and collector tubes. An addi-
tional flow of filtered purge air is neces-
sary to clean the scattering chamber
after startup and/or small leaks of room
air. The inlet tube is calibrated so that
the maximum air sampling rate of 5
cmVsec, in a laminar flow, corresponds
to a vacuum of 2.5 cm (1 in) of water in
the scattering chamber, as measured by
the vacuum gauge. Flow meters and
associated valves, labeled FM in the
figure, measure and adjust the other
flow rates. Some filters have been
omitted from the diagram for clarity.
The geometry for the positioning of
the two photomultiplier tubes that
measure the scattered light is shown in
Figure 2. Simplified diagram of the
airflow system. The sen-
sitivity sample volume is
at the intersection of the
sample airstream and the
laser beam.
Figure 3. The line from the scattering
volume to detector #1 and the line of the
incident beam determine the "scattering
plane" of detector #1. Although both
detectors measure light scattered by the
same angle (0) from the incident beam
direction, detector #1, whose scattering
plane is the equatorial plane of the fiber,
is expected to be more brightly illumi-
nated than detector #2, whose scattering
plane includes the length of the fiber.
The photomultiplier tubes, corre-
sponding to detector #1 and detector #2,
are contained in two long cylinders
protruding from the left of the sampling
head. Detector #1 is at a scattering
angle, 0 = 20°, downward from the laser
beam, which travels horizontally from
right to left in the picture. The light trap,
a pipe with a 90° bend, slightly obscures
detector #2, which intercepts light
scattered by 20° in a horizontal plane.
The aerosol beam travels horizontally
outward from the page, and the hose in
the foreground carries the air from the
collector tube back to the pump. The
inlet is on the other side of the sampling
head. The fibers are aligned with their
long axes parallel to the airflow. The
sampling head is drilled to permit the
use of scattering angles of 40° and 90°,
as well as 20°. ,
When the light intensities are quite
low, a photon-counting technique is an
appropriate method of transferring the
intensity measurements to a digital
computer. Each electron liberated by
light from a photomultiplier's cathode is
amplified to a sensible electric pulse
Detector #1
Figure 3. Two light detectors mea-
sure the intensities of the
light scattered by the fiber
fat the left) at a common
scattering angle 0 from
tie incident beam, but at
different azimuthal angles,
$1 and 2 around thei
beam. {
-------
within the photomultiplier tube. A
photon counter for each tube counts
these pulses, representing single photo-
electrons. Every 100 miproseconds the
accumulated sum is transferred to the
microcomputer, and a new count is
initiated by a timer. The number of
photoelectrons is proportional to the
average light intensity during that 100
microsecond interval. When either
phototube exceeds a preset level of
intensity, the microcomputer determines
that a particle is present in the sensitive
volume. If the ratio of the peak intensity
(largest 100 microsecond count) from
detector #1 to the peak intensity from
detector #2 exceeds a preset threshold
value, the microcomputer decides that
the particle is a fiber. In that case, the
fiber count is incremented by one and
the size of the brighter intensity, I, is
recorded by a pulse-height-analyzer
subroutine as a measure of fiber size.
The distribution of these intensities is
available at the end of a measurement
of many fibers. Depending upon the
amount of data analysis, the computa-
tion process requires 1 to5 milliseconds
per particle. This time interval intro-
duces no error in the measured particle
counts, because the computer automat-
ically extends the running time to
correct for it.
The fiber counter can operate with
randomly oriented fibers, but only the
favorably oriented fraction will be
counted. The counting efficiency is
greatly enhanced and a number of
statistical uncertainties are eliminated
by aligning the fibers to maximize the
ratio of light intensities in the two
detectors. The alignment is accom-
plished by the fortuitous reinforcement
of two effects. The gradient of the
airflow velocity in the sample stream
causes a partial alignment of the fibers
with their long axes parallel to the
streamlines. This alignment is reinforced
and stabilized by an electric field, in the
same direction as the flow, created by
making the inlet and sheath tubes carry
a positive charge and the collector tube
a negative charge.
Testing of the Fiber Counter
The time available for testing the fiber
counter was limited by the fact that the
successful design was not developed
until the final few months of the project
period. Efforts to reduce stray back-
ground light from the incandescent
source had been unsuccessful, and it
iwas finally replaced with a laser. The
laser source permitted detection of the
partial alignment of fibers in the sample
airstream and measurement of their
complete alignment within the final
month or two.
The test aerosols used with the fiber
counter were generated by two methods.
One was the nebulization of an aqueous
suspension of polystyrene latex spheres,
and the other was the mechanical
agitation of a dry sample of UICC Rhode-
sian chrysotile. A normal cycle of opera-
tion began by filling a sample chamber
with aerosol from the generator. Then
the chamber exhaust was shut down
and clean air was admitted at a rate just
adequate to replace the amount with-
drawn by the various sampling devices:
the fiber counter (5 cmvsec), a com-
mercial optical aerosol-particle counter
(usually 50 cmVsec), and occasionally a
sampling pump behind an open-face
filter in the chamber.
The flow of clean air through the
chamber may be increased to purge it of
contaminants. Even so, the limit on the
background aerosol count, in practice,
was the patience of the experimenter in
flushing the chamber, not the cleanli-
ness of the incoming air. In every case,
the levels of cleanliness obtained were
many orders of magnitude below the
count rates produced by the test aerosols.
During most of the tests, this large
(350 I) sample volume was filled with
rather clean air (10-100 particles/cm3
over 0.5 /urn in diameter) because the
fiber counter accumulated data rapidly
at these concentrations. Thus we avoided
any possibly confusing coincidence
effects and reduced the danger of a
large release into the laboratory in case
of an accident.
The photon-counting technique for
measuring light intensities employs an
actual count of the photoelectrons
leaving the cathode of the photomulti-
plier tube by amplifying each one suffi-
ciently to register in an electronic
counter. This technique reduces the
dependence of the light-intensity cali-
bration of the instrument upon the
current gain of the photomultiplier
tubes, as compared with a measure-
ment of the anode current. During initial
tests of the instrument with both the
white-light source (type 9824B photo-
multiplier tubes) and the laser source
(type 9798A photomultiplier tubes), it
was found that the photon-counting
system could be operated on a "plateau"
of the curve of count-rate versus high
voltage, where a change of 100 V in
operating potential, (representing a
factor of perhaps 2 in photomultiplier
gain) produced a change in photoelectron
count-rate of only about 10 percent.
This occurred at a high voltage of 1450
V for the 9798A tubes used with the
laser. Therefore, one can conclude that
the photon-counting technique makes
the intensity calibration relatively in-
sensitive to any changes in photomulti-
plier gain that might reasonably be
expected from high-voltage drift or
temperature effects.
Unfortunately, however, the overall
calibration factor of the fiber counter for
scattered intensities cannot be said to
be stable. Tests revealed that the prin-
cipal problem is the alignment of the
laser beam, and its cause is an inade-
quate mounting system for the laser.
A measurement of the relative veloc-
ities of the aerosol particles in the
sample volume under different operating
conditions was available directly from
the calibration program in the micro-
computer. The minimum scattered in-
tensity could be set to some threshold
value, usually 25 photoelectrons per
110-microsecond sampling interval,
which defined a minimum particle. The
number of successive intervals during
which the intensity stayed above the
threshold value gave a measure of the
temporal width of the light pulse from
one particle. The calibration program
displayed a histogram of these widths.
The measured width-above-threshold
was a function of maximum scattered-
light intensity, as well as particle velocity,
so that the distribution of pulse widths is
most useful for monodisperse aerosols.
Alignment of fibers in the airflow
alone was noticed almost immediately
when the counter was first tested with a
pure asbestos aerosol of Rhodesian
chrysotile. Previously it had been found
that the average of the intensity
ratio R = IA/IB was essentially the same
whether measured with latex spheres
or room air (if not adequately spherical,
then presumably randomly oriented)
and was quite stable over a period of
days. The corresponding correction
factor, f = 1 /, printed by the calibra-
tion program and'intended to be applied
as a factor of R so that spheres would
give unity, had been stable at 1.06 ±
0.05 for many days of spherically sym-
metric calibration runs. Operation of the
calibration program with an asbestos
aerosol should have produced the same
average correction factor if the fibers
were randomly oriented, although indi-
-------
vidual tioers could be detected by large
intensity ratios if they happened to be
favorably oriented. The correction factor
printed by the calibration program
changed abruptly, however, to 1.28 as
soon as the pure asbestos aerosol was
introduced. This significant change in
the average intensities occurred despite
an orientation of the sample flow that
greatly reduced the asymmetry in scat-
tered light intensities from flow align-
ment. A more detailed measurement of
the distribution of intensity ratios for the
asbestos revealed that indeed there was
a marked tendency for one detector to
receive a higher light intensity, and
about 7 percent of the particles produced
a high enough intensity ratio to be
classified as fibers.
The sample flow was immediately
reoriented at 90° to the laser beam and
in the plane of the beam and one
detector to maximize the effect of any
flow alignment on the asymmetry of the
light scattered into the two detectors at
a scattering angle of 20°. The effect
should be to favor the detector whose
scattering plane was perpendicular to
the plane of the laser and aerosol
beams. The inlet and collector tubes
were also insulated so they could be
used to produce an electric field of up to
3,000 V/cm in the sample region.
A large number of measurements
were made of the alignment of chrysotile
asbestos fibers in various electric fields.
A simple, single parameter to measure
fiber alignment was chosen to be the
number of fibers (defined by an intensity
ratio exceeding 8) detected in a fixed
total count, essentially a fiber-counting
efficiency. This quantity was remarkably
constant for a given set of instrument
settings, while the particle and, there-
fore, fiber count rate per unittime varied
by a factor of at least five.
In addition to the total fiber count, the
fiber-counting program prints the dis-
tribution of intensity ratios of all particles
classified as fibers. It makes no distinc-
tion between which detector is brighter
in classifying a fiber, because it was
designed to operate with randomly
oriented particles, but one can tell from
the printout how many favored each
detector. Therefore, one can count the
number of fibers oriented normal to the
airstream, an unfavored orientation, as
a function of various parameters. With
an air flow of 5 cmVsec, directed at 90°
to the laser beam, at a scattering angle
of 20°, with 4,095 particles of Rhodesian
chrysotile passing through the sample
volume, and with no electric field, 824
fibers (intensity ratio exceeding 8) were
counted in one of the runs. Of these, 36
were oriented to scatter more light into
the detector in the plane of the laser and
aerosol beams, while 788 favored the
other detector. Some of the 4,095
particles were undoubtedly fibers,
oriented so that their scattered intensi-
ties happened to be symmetric in the
two detectors. It can be seen, however,
that there is already considerable align-
ment from the airflow alone. With a
potential difference of 2 kV across the
gap, 1,042 fibers were found in 4,095
particles in one measurement, and only
9 of these were oriented so as to favor
light scattering into the detector in the
plane of the laser beam and the aerosol
beam. At a potential difference of 4 kV,
1,440 fibers were found in one mea-
surement of 4,095 particles, only 1 of
which was in the unfavored orientation.
Data at 6 kV show no fibers in the
unfavored orientation, and the change
in fiber-counting efficiency between 4
and 6 kV appears to be quite small.
Time did not permit the correction of a
defect in the fiber counter that became
obvious in studying the data from the
Rhodesian chrysotile. Because of this
defect, the roughly 35 percent maximum
fiber-counting efficiency must be con-
sidered a lower limit on the fraction of
aligned fibers. The data showed no
intensities larger than 2,000 photo-
electrons in a sampling interval, but
many particles near the maximum
intensity. The maximum observed count
rate of 2 x 107 photoelectrons/sec is in
fact about the limit of the electronics.
Fibers producing larger intensities in
the brighter detector saturate the elec-
tronics and produce too small an inten-
sity ratio. Measurements at reduced
laser intensities to catch these largest
fibers were not made owing to lack of
time in the project period.
Except for one unsuccessful attempt
to produce aerosols of flat plates from
talc, there was too little time left in the
project period to clean the aerosol
chamber and study other asbestos
aerosols and nonasbestos aerosols
properly. A few spot checks were made,
however, with the sample inlet open to
room air. These can be summarized
briefly. The "fiber count" in the labora-
tory air hovered at about 0.02/cm3,
despite the fact that readings with truly
clean air or latex spheres were consid-
erably less. No significant increase from
this value was obtained from various
clouds of talcum powder, chalk dust, or
smoke wafted before the inlet. However,
a piece of asbestos-cloth heating tape,
"bumped" in front of the inlet, produced
a very high fiber count. Also, it appeared
that calcium sulfate crystals from the
drying agent could produce a fiber count
when "dusted" in front of the fiber
counter. It was noted that, although
cigarette smoke did not produce any
fiber counts, it could paralyze the in-
strument by essentially filling the sample
inlet with a continuous haze that raised
the background count, between large
particles, above threshold.
Conclusions
The operation of the fiber counter
developed in this study demonstrated
the feasibility of using simultaneous
measurements of the light scattered in
two different directions by an aerosol
particle to differentiate between fibers
and more symmetrically shaped respi-
rable particles. Specific findings of the
study include the following:
• The device samples the ambient air
at a rate of 5 cm3/sec (0.01 cf m). At
this rate, a fiber counter can mea-
sure the OSHA threshold limit
value (TLV) of 2 fibers/cm3 to 10
percent accuracy in 10 seconds. In |
the UVM device, the sensitive
volume, within which a fiber is
counted, is about 1 mm3. Therefore,
the background dust concentration
can be as high as 100 total particles/
cm3 before coincidence errors reach
10 percent.
• The minimum detectable fiber size
for the. UVM instrument is limited
by signal-to-noise considerations,
related to background scattered
light when there are no aerosol
particles. In the proposal it was
estimated that one could achieve a
minimum detectable fiber volume
of 0.02 yum3. Despite a long battle
against background light, it has
been possible so far to achieve
reliable fiber discrimination only
down to a volume of 0.06 yum3.
• The present version of the UVM
fiber counter is limited by its elec-
tronics to a maximum intensity of
scattered light. The resulting limi-
tation on maximum fiber size is not
fixed, but rather one of the range of
sizes (dynamic range) measurable at
one time with the present photo-
detector-microcomputer interface
circuits. The present design served
well to measure small fibers be-j
-------
cause it is so flexible. However, we
have only recently begun to appre-
ciate the practical implications of
the fact that the scattered light
intensity at the forward scattering
angles used in this instrument
varies much more widely with
particle volume for fibers than it
does for spheres.
• The present version of the UVM
instrument was tested with aer-
osols made from a sample of UICC
Rhodesian chrysotile. The concen-
tration of particles ranged from 0 to
100/cm3. About 35 percent of the
total particles over a given threshold
size in this aerosol were classified
as "fibers" by the device. It is not
known just what fraction of these
chrysotile particles actually de-
served to be classified as fibers, but
the fraction is probably somewhat
larger than 35 percent. The dis-
crepancy is attributed largely to the
limited dynamic range of the pres-
ent device; many fibers were too
large to be classified properly at the
high sensitivity setting used in
those measurements.
• The present instrument seems to
discriminate adequately among
latex spheres, common room dust,
chalk dust, and cigarette smoke,
butan inadequate number of quan-
titative tests have been completed
to permit a definitive statement. It
is especially sensitive to asbestos
fibers. It does, however, also display
a sensitivity to any needlelike parti-
cles, including some commonly
found in the laboratory.
• The alignment of the aerosol fibers
within the sensitive volume by a
combination of flow and electric
fields has added a new dimension
to the measuring capabilities of the
instrument. It now appears possible
to obtain a distribution of fiber sizes
from the instrument, as well as a
simple fiber count.
Several problems arose because of
aspects of the UVM instrument that
were not adequately foreseen in the
proposal. These included the following:
• The scattering geometry of this
method of fiber discrimination
makes the instrument awkwardly
sensitive to background light in the
scattering chamber. As a result,
the instrumental design was
changed in the last 6 months of the
project to use a laser rather than an
incandescent light source.
• As soon as the device was working
properly with the new light source,
it became obvious that the fibers
were exhibiting a strong preferen-
tial alignment in the sampling
volume owing to the airflow alone.
The addition of an electric field
strengthened this preferential align-
ment to essentially 100 percent
along the direction of flow.
Recommendations
Based on the information and prob-
lems generated during this study, the
following recommendations for future
work are suggested:
• The UVM fiber counter should be
tested by monitoring hazardous
fibrous aerosol particles in real
time, in a background of other dust.
• Efforts should be made to extend
the dynamic range of the fiber
counter to count not only the long,
thin fibers currently believed to be
most active pathogenically, but
also as many as feasible of the
smaller fibers, which can be pres-
ent in even larger numbers and
may also be hazardous. In addition,
the problems associated with light
scattering from longer, curly chrys-
otile fibers should be dealt with.
• Quantitative tests are needed to
establish the ability of the fiber
counter to discriminate against
false fiber counts from nonfibrous
background dusts. Two cases are
especially likely to lead to this prob-
lem: high dust concentration, which
will be particularly troublesome
when most of the particles are too
small to be detected individually,
and asymmetric particles (such as
oblate spheroids) that are not fibers.
• Investigations should be made into
the possibility of engineering the
UVM into a portable package that
could give a simple fiber count.
• Further theoretical calculations
need to be made on the light scat-
tering from long, thin fibers.
-------
This Project Summary was authored by Catherine H. Skintik WAPORA, Inc..
Cincinnati, OH 45233.
Jack Wagman is the EPA Project Officer (see below).
The complete report, entitled "Feasibility Study for an Asbestos Aerosol Moni-
tor," was authored by Robert W. Detenbeck, who is with the University of
Vermont, Burlington. VT 05405.
The above report (Order No. PB 81-120 800; Cost: $ 12.50. subject to change) will
be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
t> U.S. GOVERNMENT WBNTINO OFFICE: 1M1 -757-OU/7U4
-------
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Postage and
Fees Paid
Environmental
Protection
Agency
EPA 335
Official Business
Penalty for Private Use $300
PS 0000329
U S fcJNVIR PROTECTION
REGION 5 LIBRARY
230 S DEARBORN STREET
CHICAGO It, 606U4
------- |