Technical Note
ORP/LV-76-9
SAMPLING AND DATA
REPORTING CONSIDERATIONS
FOR AIRBORNE PARTICULATE RADIOACTIVITY
DECEMBER 1976
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RADIATION PROGRAMS
LAS VEGAS FACILITY
LAS VEGAS, NEVADA 89114
-------
Technical Note
ORP/LV-76-9
SAMPLING AND DATA REPORTING CONSIDERATIONS
FOR AIRBORNE PARTICULATE RADIOACTIVITY
Gregory G. Eadie
David E. Bernhardt
December 1976
OFFICE OF RADIATION PROGRAMS - LAS VEGAS FACILITY
U.S. ENVIRONMENTAL PROTECTION AGENCY
LAS VEGAS, NEVADA 89114
-------
DISCLAIMER
This report has been reviewed by the Office of Radiation Programs -
Las Vegas Facility, U. S. Environmental Protection Agency, and approved
for publication. Mention of trade names or commercial products does not
constitute endorsement or recommendation for their use.
11
-------
PREFACE
The Office of Radiation Programs of the U.S. Environmental Protection
Agency carries out a national program designed to evaluate population exposure
to ionizing and nonionizing radiation, and to promote development of controls
necessary to protect the public health and safety. This report describes the
evaluation of selected air filters for use in environmental radiological air
quality monitoring studies conducted by the Las Vegas Facility. Readers of
this report are encouraged to inform the Office of Radiation Programs of any
omissions or errors. Comments or requests for further information are also
invited.
Donald W. Hendricks
Director, Office of Radiation
Programs, Las Vegas Facility
111
-------
CONTENTS
Page
PREFACE iii
LIST OF TABLES AND FIGURES vi
ACKNOWLEDGMENTS vii
SUMMARY 1
INTRODUCTION 2
AIR SAMPLING SYSTEM 3
1. System Description 3
2. Determination of the Air Volume Sampled 3
TYPES OF AIR FILTERS 7
1. Glass Fiber Filter 7
2. Microsorban Filter 7
3. Acropor (Membrane) Filter 7
4. Cellulose Fiber - Paper Filters 8
RADIOACTIVITY CONTENT OF AIR FILTERS 9
1. Blank Filter Analyses 9
2. Data Reporting Formats 11
AIRBORNE PARTICULATE COLLECTION EVALUATIONS 20
1. Dust Loading Determinations 20
2. Replicate Sampling Variability 20
3. Frequency of Filter Changing 22
4. Dust Loading Comparisons 22
S. Measurement of Sampled Volume 26
REFERENCES 29
APPENDIX A - METRIC CONVERSION TABLE 30
-------
LIST OF FIGURES
Number Page
1 Typical High-Volume Air Sampling Unit 4
2 Sampler Flow Rate Versus Sampling Time 27
LIST OF TABLES
Number
1 Air Pressure Correction Factor per 1000-foot
Elevation Increments 6
2 Glass Fiber Filter Radioactivity Content 10
3 Microsorban Filter Radioactivity Content 12
4 Acropor Filter Radioactivity Content 13
5 Whatman #541 Filter Radioactivity Content 14
6 Whatman #41 Filter Radioactivity Content 1S
7 Summary of Radioactivity Contents of Selected
Air Filters 16
8 Replicate Sampling Variability - Glass Fiber Filters 21
9 Effect of Frequency of Filter Changing - Glass Fiber Filters 23
10 Dust Loading Comparisons 24
11 Dust Loading Comparisons - Microsorban versus
Glass Fiber Filters 2S
VI
-------
ACKNOWLEDGMENTS
The authors would like to extend their grateful appreciation to the
laboratory personnel of the U.S. Environmental Protection Agency, Environ-
mental Monitoring and Support Laboratory (EMSL) - Las Vegas for their support
of this study. The use and calibration of equipment from the Environmental
Radiation Branch - EMSL is also gratefully acknowledged. The authors thank
Drs. Robert Kinnison (EMSL), Bernard Malamud (University of Nevada at Las
Vegas), Richard Gilbert (Battelle Northwest), Guy Merrill (Air Force McClellan
Central Laboratory), and Mr. Christopher Nelson (ORP) for their assistance in
determining the appropriate statistical techniques for error terms associated
with the data.
The authors, although recognizing the assistance of others, accept full
responsibility for the contents of this report.
vn
-------
SUNMARY
This report discusses the evaluation of selected air filters for their
suitability as collection media for the radiological analyses of airborne
particulate matter. Standard four-inch diameter filters were analyzed for
their natural radioactivity contents. Of the filters tested, glass fiber
filters had the highest radium-226 content (0.35 picocuries per filter) and
Microsorban filters contained roughly one-third of this activity. Microsorban
filters also have lower uranium and thorium contents than do the glass fiber
filters. For the analytical methods used in this study, all filter types
tested had undetectable polonium-210, lead-210, and radium-228 contents. Dust
loading characteristics of selected filters were also evaluated. The results
indicate that Microsorban filters have higher collection and dust retention
efficiencies (ranging from 6 to 26 percent greater) than do glass fiber
filters. As a result of these evaluations, Microsorban filters are being used
in the routine environmental radiological air quality monitoring networks
operated by the Office of Radiation Programs - Las Vegas Facility (ORP-LVF).
-------
INTRODUCTION
High-volume air sampling systems are routinely utilized to measure the
ambient airborne particulate concentrations of naturally-occurring radio-
nuclides such as radium-226 and -228, lead-210, polonium-210, isotopic uranium
and isotopic thorium. In many cases, it is necessary to distinguish the
contribution of normal background concentrations of these radionuclides from
quantities which may be contributed by some industrial activity. This report
discusses the comparisons of selected air filters for natural radioactivity
content and for airborne particulate collection properties. It also discusses
the air sampling procedures used by the Office of Radiation Programs - Las
Vegas Facility (ORP-LVF) to measure airborne levels of naturally-occurring
particulate radioactivity.
-------
AIR SAMPLING SYSTEM
1. SYSTEM DESCRIPTION
Air particulate sampling is conducted using a heavy-duty air sampler*, as
shown in Figure 1, which may be run continuously for periods of about a year
without requiring any routine maintenance. This unit uses the English units
of measure and therefore, the following discussions and reported data will
utilize English units. (Appendix A - Metric Conversion Table has been
included for cross-referencing purposes.) This sampler has a carbon-vane pump
with a 10.5 cubic feet per minute (CFM) free flow capacity. The pump is
driven through a V-belt system by a 110-volt, 3/4-hp motor equipped with
thermal overload protection. Each unit has a built-in vacuum gauge, and is
calibrated to provide an air flow rate versus pressure drop calibration curve.
A running time meter, with readout to tenths of an hour, provides the total
sampling time and can be reset to zero for each new sampling period. The air
volume collected is determined from the calibration curve by averaging the
"on" and "off" air flow rates and multiplying this average by the total time
of sample collection. A quick change filter holder is mounted at one meter
above the ground surface and is secured such that the open face of the filter
(four-inch diameter) is toward the ground.
The average face velocity at the filter is 1.4 miles per hour (61 cm/sec).
Thus, sampling is sub-isokinetic other than under extremely low wind velocity
conditions. Normally, sampling sub-isokinetically results in an excess of
coarse material, versus fines, due to the inertia of these particles in the
air; but, the downward orientation of the filter holder reduces this bias.
The sampled air and associated particulate material is actually drawn from the
air stream; hence, there is discrimination against large particles due to
their inclination to follow their basic trajectory. This size discrimination
should be of limited significance for normal atmospheric aerosol particle size
distributions and for particle size distributions pertinent to inhalation by
man. For example, the settling velocity for a 10-micrometer equivalent
aerodynamic particle (settling velocity in air of an equivalent-sized particle
with a density of 1 g/on3) is 0.3 cm/sec (Silverman, et al. 1971). This is
over two orders of magnitude less than the face velocity of the air sampler.
2. DETERMINATION OF THE AIR VOLUME SAMPLED
A calibration curve showing the air flow rate (CFM) versus the pressure
drop, as measured by the built-in vacuum gauge, is obtained for each air
sampling unit. These curves are plotted from measurements made in the Las
Vegas, Nevada facility (elevation 2000 feet) and are normalized to a tempera-
ture of 21° C (about 70° F) and a pressure of 760 mm (29.92 inches) of mer-
cury, the standard pressure at sea level.
Heavy-Duty Air Sampler, Research Appliance Corporation, Allison Park, PA;
or Tempest Air Sampler, Gelman Instrument Corporation, Ann Arbor, MI.
-------
FIGURE 1. TYPICAL HIGH-VOLb~ME AIR SAMPLING UNIT
-------
In order to calculate the volume of air sampled under ambient conditions
at the specific air sampling site, an air density correction factor which
includes the effects of air pressure, temperature, and humidity, should be
considered. Such a correction factor has been derived based only on the air
pressure term for each 1000-foot increment in elevation and is shown in Table
1. The influence of humidity and temperature on the air density is not
considered to be significant compared to the air pressure term. For example,
a fluctuation of 20° F about a normal temperature of 70° F (i.e., a range of
40° F from 50° F to 90° F) represents less than a four percent variance in the
air density over this temperature range. For most ORP-LVF air sampling
locations, the annual mean temperature falls within 20° F of the normal 70° F;
therefore, the temperature correction term represents much less than a two
percent error in the air density determinations. The use of constant air flow
sampling units, which can be calibrated at the specific sampling site, would
overcome this difficulty in determining the true air volume sampled.
To obtain the ambient air volume sampled, the total air volume, as
calculated from the flow rate versus pressure drop calibration curve, is
multiplied by the respective correction factor (Table 1) to the nearest 1000-
foot increment of elevation for the specific sampling site. This air pressure
correction factor (Federal Register, 1971) represents approximately a 1.9
percent change in air volume sampled per 1000-foot increment of elevation.
Since most of the ORP-LVF air sampling sites are at elevations in the 4000 to
7000-foot range, this correction factor ranges from about 8 to 14 percent.
-------
TABLE 1. AIR PRESSURE CORRECTION FACTOR
PER 1000-FOOT ELEVATION INCREMENTS
Elevation
(feet)
Sea Level
(Zero feet)
1000
2000
3000
4000
5000
6000
7000
8000
* Taken from J.
Barometric Pressure*
(Inches of Mercury)
29.92
28.86
27.82
26.81
25.84
24.89
23.98
23.09
22.22
H. Perry and R. H. Perry, eds,
Air Pressure
Correction Factor**
1.000
1.018
1.037
1.056
1.076
1.096
1.117
1.138
1.160
1959, Engineering Manual,
Table 4-25, McGraw-Hill Book Co.
** Air Pressure Correction Factor = [standard Pressure at Sea Level]3
(Federal Register, 1971) L Pressure at Site Elevation J
-------
TYPES OF AIR FILTERS
1. GLASS FIBER FILTER
Glass fiber filters are routinely used in air sampling networks. They
are rather durable under various sampling conditions, easy to handle,
inexpensive and have very good particulate collection characteristics. They
are relatively non-hygroscopic, i.e., they maintain constant weight regardless
of the ambient humidity. Glass fiber filters are manufactured from micro-
sized filaments of pure glass. Type A filters have no special organic
binders, as used in the Type E filters, and are treated to remove any trace
amounts of organic fiber contaminants. Type A filters are extremely useful
for trace element air pollution monitoring purposes. Both types of glass
fiber filters have collection efficiencies of at least 99.7 percent for
particles larger than 0.3 micrometer (ym) at a collection face velocity of 100
feet per minute (fpm). For particles as small as 0.05 ym, collection effi-
ciencies may be as high as 98 percent (Air Sampling Instruments, 1972; Yaffe
et al., 1956). Particulates are retained on the filter's surface as well as
within the filter matrix; hence, the glass fiber filter is considered a
"depth" filter. Type E-glass fiber filters were used in this study.
2. MICROSORBAN FILTER - (Delbag)
The Microsorban filter consists of very fine thermoplastic filaments
(polystyrene) of a thickness of approximately one micrometer or less. These
filaments are of hydrophobic nature, and can be charged electrostatically.
Particulate collection efficiencies of almost 100 percent can be achieved for
sub-micron sized dust particles. Efficiencies exceeding 99.9 percent have
been measured for particles less than 0.3 ym (Air Sampling Instruments, 1972).
Particle penetration of the filter media does occur; hence, Microsorban is
also considered a "depth" filter. The dust storage capacity of the Micro-
sorban filter is extremely high due to the large surface area of the matted
polystyrene filaments.
3. ACROPOR (MEMBRANE) FILTER
Acropor is an acrylonitrile polyvinylchloride copolymer membrane,
reinforced with nylon. It is flexible, easy to handle, and will not chip or
break. The Acropor (AN-200) filter used for this study has a mean flow pore
size of 0.20 ym and corresponding particulate collection efficiency greater
than 99.9 percent (Air Sampling Instruments, 1972). Collection efficiencies
for particles smaller than this membrane pore size are approximately the same
due to the electrostatic forces resulting from the movement of air through the
filter. In contrast to the "depth"-type filters, particle deposition occurs
almost exclusively at the upper surface of the membrane filter.
-------
4. CELLULOSE FIBER - PAPER FILTERS
Whatman Type #41 and #541 cellulose fiber filters were also used in this
study. These filters are made of purified cellulose pulp and are relatively
inexpensive, easily obtainable, and have low air flow resistance. Although
there is considerable penetration of sub-micron sized particles through the
cellulose paper filters, their collection efficiencies are usually considered
adequate for most air sampling programs.
-------
RADIOACTIVITY CONTENT OF AIR FILTERS
1. BLANK FILTER ANALYSES
The Gelman Type E glass fiber filters which were evaluated were standard
four-inch diameter filters having an average mass with an associated two
standard deviations of 0.5251 ± 0.0172 grams (measured to a tenth of a milli-
gram accuracy). Two different sets of filters were analyzed for natural
radioactivity contents. The first set consisted of five single glass fiber
filters. The second set consisted of a series of five samples with each
sample representing an aliquot from a composite of four filters. Individual
radionuclide analysis was completed on separate aliquots of the total sample.
In general, one-fourth of the total sample was analyzed for uranium content,
one-fourth for thorium, one-fourth for radium, and the remainder for lead/
polonium. All radiological analyses were completed at the U.S. EPA - Environ-
mental Monitoring and Support Laboratory, Las Vegas, Nevada (EMSL-LV).
Standard radiochemical procedures (Johns, 1975) were employed. Table 2 shows
the individual radionuclide content and two-sigma counting error term for each
analysis. Usually, counting times ranged from 30 minutes for radium-226
analysis to 1000 minutes for the actinide analyses.
Also shown in Table 2, for each group of analyses, is the average radio-
nuclide content plus and minus the standard error of the mean at the 95
percent confidence level. A grand average of all results per radionuclide
analysis, and its standard error term, is also given.
The standard error term about the mean is calculated at the 95 percent
confidence level based on the "t-distribution" for the appropriate number of
degrees of freedom due to the small sample sizes (usually n<10). The stan-
dard error of the mean is defined as follows:
Ox = standard error of the mean
(7= standard deviation of the results
Xj= analytical result
X = mean value = -=^p
n = number of sample results
df = degrees of freedom = n-1
t » value of the t-distribution at the 95 percent confidence level for
the appropriate degrees of freedom (t values may be obtained from
Appendix M, Spurr and Bonini, 1973).
-------
TABLE 2. GLASS FIBER FILTER RADIOACTIVITY CONTENT*1^
Radium- 226
Single Filter
Analysis
0.53 ± .20
0.39 i .19
0.30 ± .17
0.24 ± .15
0.26 ± .17
Average * '
0.34 ± .15
Grand/,,
Average ( '
0.35
4-Filter
Composite
0.58 ± .15
0.43 ± .13
0.36 ± .12
0.18 ± .08
0.23 ± .09
0.36 ± .19
± .09
Uranium-238
Single Filter
Analysis
0.11 ± .06
0.07 ± .02
0.08 ± .03
0.08 ± .03
O.Q& ± -QZ
Average^3'
0.08 • .03
Grand,,,
Avarage {3>
n.nn
4-Filter
Compos i te
0.06 i .02
0.05 ± .02
0.05 ± .02
0.15 ± .03
O.OS ± .02
0.07 ± .06
± .02
Thorium-230
Single Filter
Analysis
0.19 ± .07
0.15 ± .14
0.17 ± .06
0.19 ± .08
0.18 ± .06
0.1S ± .03
0.20
4-Filter
Composite
0.52 ± .07
0.12 ± .04
0.14 ± .04
0.17 t .05
0.14 ± .05
0.22 i .21
± .08
Radium-228
Sinqle Filter
Analysis
4.78 ± 1.84
<1.52
<1.68
<1.68
.02
Polonium-210
Sinqle Filter
Analysis
<0.36
<0.04
<0.06
O.07
0.63 t .29
<0.23
<1.59 <0.
4-Filter
Comoosi te
0.04 ± .04
<0.08
<0.06
<0.06
17
(1) Average four-inch diameter glass fiber filter mass t two standard deviations of 0.5251
(2) U-235 calculated based on natural U-Z35 to U-238 activity ratio of 1:21.45 (or 0.0466)
(3) Average of all results with standard error about this mean based on the t-distribution
Uranium-234 Uranium-235'2'
Single Filter 4-Filter Single Filter
Analysis Composite Analysis
0.10 i .06 0.09 ± .02 0.0051 i .0028
0.13 * .03 0.07 i .02 0.0033 ± .0009
0.14 ± .04 0.06 » .02 0.0037 ± .0014
0.18 ± .04 0.08 ± .02 0.0037 t .0014
0.06 * .02 0.07 i .02 0.0028 ± .0009
0.12 » .06 0.07 i .01 0.0037 * .0011
0.10 i .03 0.0035
Lead-210
Sinqle Filter 4-Filter
Analysis Composite
<0.15 <0.28
0.17
O.47
<0.46
O.38
<0.33 <0.28
<0.32
± 0.0172 grains.
at the 95 percent confidence level.
4-Filter
Compos i te
0.0028 ± .0009
0.0023 * .0009
0.0023 ± .0009
0.0070 * .0014
0.0023 + .0009
0.0033 • .0025
> .0010
-------
Then:
OK-ffH ^'"n; I Equation (1)
A \n |_ n(n—1) J
Therefore, the average and grand average are reported plus and minus the
95 percent confidence level standard error of the mean term: (t (J^ ). All
results are reported in units of picocuries per four-inch diameter filter
(pCi/filter). The Ra-228, Po-210, and Pb-210 contents per filter are essen-
tially at or below the minimum detectable activity (MDA) levels for the
routine analytical procedures used in this study. However, appreciable quan-
tities of other naturally-occurring radionuclides (i.e., Ra-226 and isotopic
uranium and thorium) are present in the glass fiber filter blanks. No deter-
mination of the specific source of radioactivity was made; hence, the radio-
activity content reported here represents all possible sources, such as filter
media, reagents, laboratory equipment, and errors introduced by analytical
methods.
The results of the H4SL-LV radiochemical analyses of Microsorban filters
are shown in Table 3. Fluorophotometric analysis of the natural uranium
content of Microsorban filters by Argonne National Laboratory (Golchert, 1975)
indicated less than 0.008 pCi/filter. Isotopic thorium analyses using alpha
spectrometry indicated less than 0.00065, 0.00081, and 0.00032 pCi/filter of
thorium-228, -230, and -232 respectively. These results indicate a lower
level of blank analytical sensitivity for uranium and thorium (from a factor
of five for uranium to 30-80 for thorium) than that achieved at the EMSL-LV
laboratory (Table 3). This further indicates that individual laboratories
involved in the determinations of environmental levels of natural radio-
nuclides must exercise extreme caution to minimize analytical errors and that
a quality assurance program to periodically evaluate the radioactivity content
of blank filter media should be instituted.
The results of radioactivity analyses for selected radionuclides in
Acropor and Whatman #541 and #41 filter media are shown in Tables 4, 5, and 6,
respectively. Table 7 presents a summary of the radioactivity content of the
various air filters analyzed. Glass fiber and cellulose-type filters contain
the most radium-226 activity, 0.35 and about 0.27 pCi per filter, respectively.
Microsorban and Acropor filters contain roughly one-third of this amount of
radium-226 activity. Compared to glass fiber filters, Microsorban filters
have appreciably lower contents of thorium and uranium, about one-sixth and
one-fourth the activity, respectively. The concentrations of Po-210, Pb-210,
and Ra-228 in the various filters were below the analytical sensitivities for
the radiochemical procedures used for these analyses. In general, the analy-
tical sensitivities for Po-210, Pb-210, and Ra-228 were at least twice the
sensitivity (or MDA) for Ra-226, thorium, or uranium analyses.
2. DATA REPORTING FORMATS
The following discussions concerning the statistical treatment of the
data are based on the assumption that the data are normally distributed.
11
-------
Radium-226 Thorium-230
TABLE 3. MICROSORBAN FILTER RADIOACTIVITY CONTENT
{pCi/filter ± two sigma counting error terms)
Thorium-232 Uranium-234 Uraniun-235^2) Uranium-238
(1)
Radium-228 Polonium-210
Lead-210
0.18 ± .07
0.15 ± .07
0.06 ± .05
0.05 ± .05
0.19 ± .08
Grand , .
Average IJ)
0.13 ± .08
<0.015
<0.030
<0.032
<0.022
<0.021
<0.024
<0.012
<0.027
0.047 ± .038
<0.016
-------
TABLE 4. ACROPOR FILTER RADIOACTIVITY CONTENT
{pCi/filter ± two sigma counting error terms}
Radium-226
0.08 ± .04
0.16 ± .05
0.06 ± .04
<0.04
0.21 ± .06
Grand /«\
Average v '
0.11 ± .09
Radium-228
<0.43
0.46 ± .41
<0.40
-
<0.43
<0.43
Polonium-210
0.04 ± .03
<0.05
<0.06
<0.05
<0.04
<0.05
Lead-210
<0.54
<0.09
<0.34
<0.29
<0.27
<0.31
(1). Average four-inch diameter Acropor filter mass ± two standard
deviations of 0.0495 ± 0.0072 grams; single filter analyzed.
(2). Grand Average of all results with two standard error terms about
this mean based on the t-distribution at the 95 percent
confidence level.
13
-------
TABLE 5. WHATMAN #541 FILTER RADIOACTIVITY CONTENT^1^
(pCi/filter ± two sigma counting error terms}
Radium- 226
Single Filter
Four Filter
Composite
Grand Average ^ '
0.40
0.32
0.18
0.26
1.01
<0
0.20
0.42
0.13
0.11
0.31
± .17
± .18
± .13
± .15
± .27
.08
± .09
± .13
± .09
± .06
± .19
Radium-228
<2.14
2.14 ± 1.70
2.52 ± 1.95
<1.83
<1.76
<0.95
-
<1.01
<0.88
<1.01
<1.58
Polonium-210
0.25 ± .08
<0.07
<0.04
<0.06
<0.08
<0.18
<0.25
-------
TABLE 6. WHATMAN #41 FILTER RADIOACTIVITY CONTENT
{pCi/filter ± two si'gma counting error terms}
(1)
Radium-226 Radium-228
0.15
0.11
0.20
0.45
0.23
Grand
0.23
± .09
± .07
± .09
± .13
± .09
Average^ '
± .16
<0.98
<0.61
<0.80
<0.92
<0.86
<0.83
Polonium-210
-
<0.20
<0.86
<0.43
<0.17
<0.42
Lead-210
-
<1.41
<1.84
<0.92
<1.29
<1.37
(1). Average four-inch diameter Whatman #41 filter mass ± two
standard deviations of 0.6119 ± 0.0324 grams, four filter
composite analyzed.
(2). Grand Average of all results with standard error about
this mean based on the t-distribution at the 95 percent
confidence level.
IS
-------
TABLE 7. SUMMARY OF RADIOACTIVITY CONTENTS OF SELECTED AIR FILTERS
fn
Type Of
Filter
Glass Fiber
Microsorban
Whatman 141
Whatman #541
Acropor
Ci /filter ± s
Radium-226
0.35 ± .09
0.13 ± .08
0.23 ± .16
0.31 ± .19
0.11 ± .09
(1). Grand Averaae of all
andard error
Radiura-228
<1.59
<0.66
<0.83
<1.58
-------
a. Net Result Calculation
Since glass fiber air filters have been routinely used in ORP-LVF
air sampling systems, a program has been established to adjust the measured
gross radioactivity results to account for the natural radioactivity content
of the blank filter. This blank filter subtraction calculation consists of
subtracting the appropriate grand blank filter content (Table 2) from the
'gross analytical result to obtain a corrected net result per composite sam-
pling period. The number of sample filters associated with a given composite
sampling period must, therefore, be used to determine the appropriate quantity
of total activity contained in the blank filters. Since continuous air
samples are usually obtained, with individual filter changes about once per
week, the composite sampling period is normally four weekly samples combined
into one monthly period. Then, for example, the appropriate blank activity
subtraction value would be four times the grand average blank content for each
radionuclide of interest. No blank subtractions are made for the three radio-
nuclides (Ra-228, Po-210, and Pb-210) which are at the analytical MDA levels.
Let G = gross analytical result
B = blank value
N = net result
Then N = G - (Number of Filters per Composite x B) Equation (2)
b. Standard Deviation of the Net Result
The standard deviation of the net result is the square root of the
sum of the variances of the gross result and of the blank value.
LetO"G= one- sigma counting error term of the gross result
O^s one- sigma error term of the blank value
(7N= one standard deviation of the net result
Than << + tu8* x
-------
For any gross result which is equal to the two-sigma counting error term
(due to sample and equipment background counts) , the analytical result is
reported as a less than (LT or the symbol <) value. This is also the EMSL-LV
definition of the minimum detectable activity (MDA) (Bernhardt , 1976). For
calculational purposes, all "less than" values are considered as real numbers
with a two-sigma counting error term of equal value.
d. Volume Weighted Monthly Average Result
Data are usually reported as a monthly composited sampling period.
Individual filters are composited based on sample "OFF" dates within the same
calendar month. Any sampling period of less than a month's duration (e.g.,
weekly or daily samples) are statistically averaged, using a sampled volume
weighting procedure.
Let X = volume weighted monthly average
Xi= individual net result
Vi = individual volume
n = number of samples
Wi= sampled volume weighting factor
Wi=
Equation (4)
Then X* Equation (5)
A sampled volume versus collection time weighting factor was chosen
because of the operational variability between air sampling units. That is,
two identical units may have operated for exactly the same collection period,
but the volume of air sampled will probably not be the same for both units.
This is due to different sampling rates caused by filter loading and typical
operational variances between the units.
e. Standard Error Term of the Volume Weighted Monthly Average Result
The standard error term of the volume weighted monthly average is
obtained by selecting the larger error term of either the variance between
individual net results and the average, or the variance within the individual
net results.
Let SD2 between = variance calculated from the volume weighted individual
net results and the average value, divided by n
SD2 within = variance within individual net results
SE = standard error term of the volume weighted monthly
average result
18
-------
Where:
SD2 between = 'SW|' ^~ Equation (6)
SD2 within = - - — bL Equation (7)
n2
Then:
SE = the larger of either SD between or SD within Equation (8)
For repetitive counting of the same sample, the between error and within
error are identical. The between error for an average of a number of samples
includes the within error and the error from the uniqueness of the samples as
well as any analytical error (Jaffey, 1960). Thus, for most cases the between
error is larger than the within error.
The use of the concept of less than values, for values below the detec-
tion limit, can often result in a group of samples being reported as identical
less than results. The calculated between error for such a group of identical
less than values would be zero. Such a result would be the figment of the
"less than" reporting technique, and not a valid error estimate. Thus, when
less than values are reported, the within error should be calculated by
setting the individual counting errors equal to one -half of the less than
value (the less than value is the two-sigma error) . Both the between and the
within error terms should be calculated and compared, and the larger of the
two error terms, between or within, should then be used as the standard error
term of the volume weighted monthly average result.
f. The t -Distribution
Since the number of samples being averaged is usually small
(i.e., n < 20) the "t-distribution" should be used, instead of the normal
distribution, to obtain the standard error term of the monthly average at the
95 percent confidence level. Then the reporting format becomes: 3( ± t x SE.
Values of t may be obtained for the appropriate degrees of freedom at the
95 percent confidence level from Appendix M, Spurr and Bonini, 1973. For
example, for an average determination using five sample results, the degrees
of freedom would be four and the t value would be 2.78 instead of the 1.96
value for the normal distribution at the 95 percent confidence level.
g. Grand Average
A volume weighted grand average result is calculated for selected
time periods (usually an annual average) using all available individual data
and volume weighting these results. The standard error of this grand average
is also calculated. These calculations are preformed in the same manner as
shown in Equations 5 through 8 and the appropriate t-distribution value for
the 95 percent confidence level is applied to the calculated standard error
term.
19
-------
AIRBORNE PARTICULATE COLLECTION EVALUATIONS
1. DUST LOADING DETERMINATIONS
The dust loading of each filter type was determined by measuring the mass
of each filter prior to and at the end of a sampling period. These mass
measurements were made under controlled laboratory conditions, at 21° C room
temperature. In order to establish the degree of hygroscopicity of each
filter type, relative humidity measurements were periodically made using a
sling psychrometer. Mass measurements were made at an average relative
humidity of 35 percent, ranging from 15 to 55 percent. Glass fiber filters
showed a relative error of 2.3 percent of the filter mass due to filter
hygroscopicity. Similarily, Microsorban and Whatman #41 media showed
1.1 percent and 1.5 percent mass variations with changes in relative humidity,
respectively.
Dust loading determinations were also attempted using a "dry ashing"
procedure (i.e., igniting the filter media at 800° C). This method proved
inadequate for the glass fiber media which leaves a silicon residue upon
ashing. The Microsorban and Whatman #41 media were essentially ashless;
therefore, dust loading capacities could be determined from the mass of the
remaining residue if the organic and volatile components (e.g., Polonium-210)
of the airborne particulates are unimportant. Dust loads are reported in
units of microgram per volume of air sampled (pg/ft3).
2. REPLICATE SAMPLING VARIABILITY
Air particulate collection effectiveness and the dust loading were
evaluated using selected air filters and standard air sampling procedures.
Since adequate facilities were not available to provide constant airborne
particulate concentrations of known particle sizes, tests were run out-of-
doors at the Las Vegas facility under ambient field conditions. These tests
were conducted using the air sampling units as described above. The units
were arranged so that the air sampling inlets were in very close proximity to
each other in order to assure sampling of equivalent airborne particulate
concentrations.
In order to establish the sampling variability of the air sampling unit,
two identical units were operated side-by-side during the same time period.
Glass fiber filters were employed as the particulate collection medium and the
results of these tests are shown in Table 8. Assuming that each sampling unit
was capable of sampling an equivalent airborne particulate concentration, the
relative sampling variability between the two units ranged from zero percent
to 12.64 percent, with an average of 6.79 ± 4.72 (one-sigma) percent. Using
the t-distribution for six degrees of freedom at the 95 percent confidence
level, the t value would be 2.447 and the error would then be 11.55 percent.
20
-------
TABLE 8. REPLICATE SAMPLING VARIABILITY* - GLASS FIBER FILTERS
On
Flow
CFM
10.3
9.2
10.4
9.2
10.4
9.2
9.6
9.3
9.4
9.7
9.2
9.7
9.9
10.4
* Replicate
Off
Flow
CFM
10.1
9.2
9.9
9.1
9.5
9.0
8.9
9.1
9.2
9.2
9.1
9.4
8.7
8.7
Sampling
Total
Time
Hours
45.4
45.4
69.8
69.7
53.2
53.4
47.7
47.7
71.7
71.7
50.4
50.5
70.9
71.0
Variability
Total
Volume
ft3
27784
25060
42718
38474
31920
29156
26617
26330
40009
40869
27821
29088
39562
40896
= Maximum
Total Dust
Load
Grams
0.0245
0.0217
0.0552
0.0464
0.0778
0.0634
0.0916
0.0994
0.0512
0.0569
0.0442
0.0530
0.1454
0.1505
Dust Load - Minimum
Dust Load
Capacity
uq/ft3
0.88
0.87
1.29
1.21
2.44
2.18
3.44
3.78
1.28
1.39
1.59
1.82
3.68
3.68
Dust Load
Replicate Sampling
Variability
Percent
1.14
6.20
10.66
8.99
7.91
12.64
0
-------
The resultant sampling variability is, therefore, 6.79 ± 11.55 percent, with a
maximum value of 18-.34 percent.
3. FREQUENCY OF FILTER CHANGING
The evaluation of weekly versus daily and every-other-day air filter
changing was conducted using three air sampling units positioned side-by-side.
Glass fiber filters were used as the particulate collection medium and the
dust loading was determined for each filter. The results of these tests are
shown in Table 9.
The greatest dust loading (0.2383 grams, corresponding to a concentration
of 2.89 yg/ft3) was obtained for the weekly sampling period using only one air
filter. This weekly sampling period result is in excess of the expected maxi-
mum 18.34 percent replicate sampling variability from the results of the other
two shorter sampling periods results (i.e., 2.38 and 2.37 yg/ft3). Similar
conclusions are obtained using the collection time weighted results (Table 9}.
A possible explanation is that frequent glass fiber filter changing results in
the loss of particulate matter from the filter's surface due to normal hand-
ling during the filter changing procedure. For routine air sampling using
glass fiber filters, sample collection periods of about one-week duration per
filter appears to be adequate. It should be noted that this conclusion is
based on only one trial result.
4. DUST LOADING COMPARISONS
In order to evaluate selected air filters for suitability as air par-
ticulate collection media, field tests were conducted to permit relative
comparisons of the dust loading of each filter type. Three air sampling units
were operated side-by-side using the same sampling periods, but each unit
utilized a different filter medium. The results of these tests are shown in
Table 10.
Glass fiber appears to have a dust loading capacity about twice that of
Whatman #541 paper filters and the Acropor membrane filters. The dust loading
capacity of glass fiber filters ranges from roughly two to ten times the
values obtained using Whatman #41 paper filters.
Microsorban versus glass fiber filter comparisons are shown in Table 11.
For the two tests using one-week collection periods, Microsorban filters had
the greater dust loading of 1.28 versus 1.07 and 2.17 versus 1.62 yg/ft3, a
range of 16 to 25 percent greater than the glass fiber filter results.
The results from Microsorban filters changed weekly versus about every
two-days show that frequent Microsorban filter changes yield greater total
dust loading results. For the first series of tests, the weekly dust load was
1.28 yg/ft versus the average dust load for three filter changes of 1.73
yg/ft , a 26 percent greater dust load than the weekly value. For the second
series of tests, the weekly value was 2.17 yg/ft3 versus the every two-day
change cycle value of 2.31 yg/ft3, a six percent increase. Similar conclu-
sions are obtained using the collection time weighted dust loading results.
In general, optimum dust loading was obtained for Microsorban filters with
sample collection periods of about two days; but weekly filter changes also
22
-------
TABLE 9. EFFECT OF FREQUENCY OF FILTER CHANGING - GLASS FIBER FILTERS
to
C/4
Sampling Period
Weekly
2 days
3 days
2_days
7 days TOTAL
1 day
1 day
3 days
1 day
1 day
7 davs TOTAL
* Weighted Dust
On
Flow
CFM
9.2
9.7
9.9
9.7
10.5
10.6
10.4
10.4
10.4
Load = z(Dust
Off
Flow
CFM
7.4
9.2
8.7
9.6
10.3
10.3
8.7
10.4
10.2
Load)
Total
Time
Hours
165.6
46.6
70.9
47.9
165.4
22.5
24.0
71.0
23.2
24.5
165.2
(Collection
Total
Volume
ft3
82469
26562
39562
27878
94002
14040
15120
40896
14477
15141
99461
Time)
Total
Dust Load
Grams
0.2383
0.0478
0.1454
0.0305
0.2237
0.0267
0.0264
0.1505
0.0138
0.0184
0.2358
Dust Load
ug/ft3
2.89
1.80
3.68
1.09
2.38
1.90
1.75
3.68
0.95
1.22
2.37
Weighted
Dust Load*
iiq/ft3
2.89
-
2.40
-
2.41
Total Time
-------
JCABLE JO. DUST LOADING COMPARISONS
Filter
Type
Glass Fiber
Whatman #541
Acropor
Glass Fiber
Whatman #541
Acropor
Glass Fiber
Whatman #541
Acropor
Glass Fiber
Whatman #41
Glass Fiber
Whatman #41
On
Flow
CFM
9.6
9.9
6.3
10.5
9.5
4.5
9.5
9.7
4.5
9.9
8.9
9.5
4.6
Off
Flow
CFM
6.0
3.6
1.9
6.7
3.8
2.6
6.5
3.8
1.5
6.9
8.2
6.0
3.6
Total
Time
Hours
95.1
95.1
95.1
164.9
165.1
164.9
262.7
261.7
262.7
211.7
211.7
601.3
601.2
Total
Volume
ft3
44506
38801
23395
85088
66370
35618
126096
106774
47286
106697
109237
281408
147895
Total
Dust
Grams
0.4756
0.2025
0.0987
0.2599
0.1116
0.0587
0.4751
0.1694
0.1254
0.2260
0.0230
0.5159
0.1314
Oust Load
u q/f 1 3
10.69
5.22
4.22
3.05
1.68
1.65
3.77
1.59
2.65
2.12
0.21
1.83
0.89
-------
Filter
Type
Microsorban
TOTALS
Microsorban
Glass Fiber
Microsorban
TOTALS
Microsorban
Glass Fiber
* Weighted Dust
On
Flow
CFM
9.7
9.6
9.9
.
10.8
9.4
9.7
9.7
9.8
9.2
10.8
Load =
Off
Flow
CFM
9.5
9.3
9.5
.
9.2
8.7
9.2
9.6
9.7
8.7
9.8
s(Dust Load)
Tota
Total
Time
Hours
46.3
65.6
52.5
164.4
164.6
164.6
48.0
64.9
55.9
168.8
169.0
169.0
(Collection
1 Time
Total
Volume
ft3
26669
37392
30555
94616
98760
89872
27360
37772
32869
98001
91260
104442
Time)
Total
Dust
Grams
0.0489
0.0404
0.0472
0.1635
0.1259
0.0961
0.1176
0.0569
0.0520
0.2265
0.1977
0.1696
Dust Load
yg/ft3
1.83
1.08
2.43
1.73
1.28
1.07
4.30
1.51
1.58
2.31
2.17
1.62
Weighted
Dust Load*
ug/ft3
-
1.72
1.28
1.07
-
2.33
2.17
1.62
-------
appear to be adequate for routine sample collection. Of all the air filters
tested, Microsorban .filters had the greatest dust loading capacity.
5. MEASUREMENT OF SAMPLED VOLUME
The technique for estimating sampled volume was described above in the
"Air Sampling System" section of this report. There are several potential
sources of error or inaccuracy associated with the estimation of sampled
volume. These include:
1. Leakage of air around the filter and the gasket of the sample
filter holder.
2. Inaccuracies in reading the vacuum gauge (the vacuum readings are
translated to flow rate). The gauge can only be read to two signi-
ficant figures and operators are prone to read them to only one
significant figure.
3. The sampled volume is based on the average sampling rate as deter-
mined from the initial and final flow rates. This assumes a linear
decrease in sampling rate with time.
The above items are difficult to quantitate, but some information was
gathered concerning Item 3.
Figure 2 presents time versus flow-rate profiles indicative of those
obtained for several types of filters. The plots are flow rate, in cubic feet
per minute, versus sampling time. The data points (measured line) represent
actual vacuum measurements and associated flow rates and, thus, the best
estimate of the actual sampled volume. The dashed line represents the flow
rate estimated from the initial and final flow rates and is indicative of the
flow rate actually used to calculate the concentration of the various radio-
nuclides in the sampled air. The indicated percent error is the difference
between the area under the dashed line and the measured line divided by the
area under the measured line. The net grams of airborne particulate collected
on each filter is also indicated.
Figures 2a and 2b indicate that the normal estimate of sample volume is
in error by up to 70 percent for the indicated samples using Acropor or
Whatman #541 filters. These plots relate to sampling resulting in a mass
loading of 0.1 to 0.2 grams of particulate matter on the filter. This is
relatable to an ambient dust concentration of 50 ug/m3 (1.4 ug/ft3) and a
sampling rate of 8 ft3/min for one week, which gives a dust loading of 0.1
gram.
There is probably an additional error as a result of the decreased
entrapment of airborne particulates due to the reduced face velocity resulting
from the increased air resistance of the filter as dust is accumulated. Both
the error in the estimated volume and the reduced entrapment of airborne
particulates may account for some of the discrepencies in the mass loading
indicated in the previous section.
26
-------
ALL FIGURES:
Y = SAMPLER FLOW RATE
(ftVmin)
X= SAMPLING TIME
(hours)
%ERROR = % THAT THE
AVERAGE VOLUME
IS LESS THAN THE
TRUE INTEGRATED
VOLUME
g DUST = GRAMS OF
DUST COLLECTED
ON FILTER
FIGURE 2(a).
Jan. 24, to Feb. 4. 1975
10j
Y
7
I
5-
Y
l
GLASS FIBER 15% ERROR
K^~r.. 048gDUST
^^^^*" """ **" •"-
100 260 300
-_ ACROPOR 45% ERROR
L "~"~^^^ 0.13 g DUST
>V^~"^-^.^^
^**"**"****-^">^ii^_
100 200 300
10-
IU
Y
5-
>. WHATMAN 541 70% ERROR
V^ 0 17 g DUST
X^v
Ny ^\
^•*"»««> *- ^*
100 200 300
FIGURE 2(b)
Jan. 20, to Jan. 24, 1975
10-
7l \ WHATMAN 541
( ACROPOR V^NS 30% ERROR
V^l 30°/0 ERROR \ N 0.20 g DUST
Y5' Tvv 0.099 g DUST y \ N
1 .\. * L..\
6 50 100 0 50 100
X X
FIGURE 2(c)
April 9, to April 16, 1975
. GLASS FIBER <1% ERROR
1 , >,_ 0.096 g DUST
FIGURE 2. SAMPLER FLOW RATE (ftVmin) VERSUS SAMPLING TIME (hours)
27
-------
The plots in Figures 2a and 2b note errors of 30 to 70 percent for
Acropor and Whatman #541 for mass loadings of 0.1 to 0.2 grams. Figure 2a
notes an error of only 15 percent for a mass loading of 0.48 grams on glass
fiber. The plots in Figure 2c show a negligible error in the volume estimate
for mass loadings of 0.1 to 0.2 grams on Microsorban or 0.1 gram on glass
fiber filters.
The plots illustrated in Figure 2 are generally indicative of the flow-
rate versus time profiles of the other similar samples of this study. Thus,
it is concluded that glass fiber and Microsorban filters are preferable to
Whatman #541 and Acropor for mass loading in the range of the stated experi-
mental values (generally 0.1 to 0.5 gram per 4-inch diameter filter). There
is the additional concern that as a result of the decreased flow rates towards
the end of the sampling period, that Whatman #541 (also #41) and Acropor
filters will not sample the airborne activity present at a representative
sampling rate.
In summary, since Microsorban filters have minimal natural radioactivity
content (Table 3), compared to glass fiber filters (Table 2), the need to
perform a blank filter radioactivity content subtraction from the ambient
gross results, as is the case with the glass fiber filter, is usually not
necessary unless background determinations are being made. Compared to the
other filter types tested, the Microsorban filter also appears to have the
highest dust loading and retention capacity. As a result of these compari-
sons, Microsorban filters have replaced glass fiber filters for routine
airborne particulate sampling conducted by ORP-LVF.
28
-------
REFERENCES
Air Sampling Instruments, 4th Edition, 1972, American Conference of
Governmental Industrial Hygienists.
Bernhardt, David E., 1976, Evaluation of Sample Collection and Analysis
Techniques for Environmental Plutonium, Technical Note ORP/LV-76-5, Office of
Radiation Programs-Las Vegas Facility.
Federal Register, 1971, Vol. 36, No. 84, Appendix B.
Golchert, Norbert, W., 1975, Private communication to G. Eadie.
Jaffey, A. H., 1960, Statistical Tests for Counting, Nucleonics.
November, p 180.
Johns, Fredrick B., ed., 1975, Handbook of Radiochemical Analytical Methods,
EPA-680/4-75-001, Las Vegas, Nevada.
Perry, John H. and R. H. Perry, eds., 1959, Engineering Manual, McGraw-Hill
Book Co., New York, New York.
Silverman, L., C. E. Billings, M. W. First, 1971, Particle-Size Analysis
in Industrial Hygiene, Academic Press.
Spurr, William A., and C. P. Bonini, 1973, Statistical Analysis for Business
Decisions, Richard D. Irwin, Inc., Homewood, Illinois.
Yaffe, Charles D., D. H. Byers and A. D. Hosey, eds., 1956, Encyclopedia
of Instrumentation for Industrial Hygiene. University of Michigan, Ann Arbor,
Michigan.
29
-------
APPENDIX A
Multiply
Cubic feet (ft3)
Cubic feet per minute
(GEM)
Feet (ft)
Inches (in)
Inches (in)
Inches (in)
Micrograms per cubic
foot Gig/ft?)
Temperature
(degrees F - 32)
METRIC CONVERSION TABLE
By_
0.02832
472
0.3048
25.4
2.54
0.0254
35.31
0.555
To Obtain
Cubic meters (m3)
Cubic centimeters per
second (cm3/sec)
Meters (m)
Millimeters (mm)
Centimeters (cm)
Meters (m)
Micrograms per cubic
meter (yg/m )
Temperature
(degrees C)
30
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
ORP-LV-76-9
2.
4. TITLE AND SUBTITLE
Sampling and Data Reporting Considerations for
Airborne Particulate Radioactivity
7. AUTHORIS)
Gregory G. Eadie
David E. Bernhardt
9. PERFORMING ORGANIZATION NAME AND ADDRESS
U.S. Environmental Protection Agency
Office of Radiation Programs, Las Vegas Facility
P. 0. Box 15027
Las Vegas, NV 89114
12. SPONSORING AGENCY NAME AND ADDRESS
Same as above
3. RECIPIENT'S ACCESSION'NO.
5. REPORT DATE
December 1976
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NO.
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report discusses the evaluation of selected air filters for their
suitability as collection media for the radiological analyses of airborne
particulate matter. Standard four- inch diameter filters were analyzed for
their natural radioactivity contents. Of the filters tested, glass fiber
filters had the highest radium- 226 content (0.35 picocuries per filter) and
Microsorban filters contained roughly one- third of this activity. Microsorban
filters also have lower uranium and thorium contents than do the glass fiber
filters. For the analytical methods used in this study, all filter types
tested had undetectable polonium- 210, lead- 210, and radium- 228 contents. Dust
loading characteristics of selected filters were also evaluated. The results
indicate that Microsorban filters have higher collection and dust retention
efficiencies (ranging from 6 to 26 percent greater) than do glass fiber filters.
As a result of these evaluations, Microsorban filters are being used in the
routine environmental radiological air quality monitoring networks operated by
the Office of Radiation Programs - Las Vegas Facility (ORP-LVF) .
17.
a. DESCRIPTORS
Natural radioactivity
Air filters
Air sampling
Airborne particulate
Data reporting
18. DISTRIBUTION STATEMENT
Release to Public
KEY WORDS AND DOCUMENT ANALYSIS
b.lDENTIFIERS/OPEN ENDED TERMS
Air pollution
Radiochemi s try
19. SECURITY CLASS (This Report)
Unclassified
20. SECURITY CLASS (This page)
Unclassified
c. COS AT I Field/Group
1808
0705
21. NO, OF PAGES
31
22. PRICE
EPA Form 2220-1 (9-73)
------- |