EPA/600/4-85/034
April 1985
EVALUATION OF PASSIVE SAMPLING DEVICES (PSD's)
by
Robert W. Coutant
BATTELLE
Columbus Laboratories
505 King Avenue
Columbus, Ohio 43201
Contract No. 68-02-3487
Project Officer
James D. Mulik
Methods Development and Analysis Division
Environmental Monitoring Systems Laboratory
Research Triangle Park, NC 27711
ENVIRONMENTAL MONITORING SYSTEMS LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27 711
-------
DISCLAIMER
The information in this document has been funded wholly or in part by
the United States Environmental Protection Agency under Contract NO. 68-02-
3487 to BatteLLe Columbus Laboratories. It has been subject to the Agency's
peer and administrative review, and it has been approved for publication as
an EPA document. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use.
i i
-------
FOREWORD
Measurement and monitoring research efforts are designed to anticipate
environmental problems, to support regulatory actions by developing an in-
depth understanding of the nature and processes that impact health and the
ecology, to provide innovative means of monitoring compliance with regula-
tions, and to evaluate the effectiveness of health and environmental protec-
tion efforts through the monitoring of long-term trends. The Environmental
Monitoring Systems Laboratory, Research Triangle Park, North Carolina, has
responsibility for! assessment of environmental monitoring technology and
systems for air, implementation of agency-wide quality assurance programs for
air pollution measurement systems, and supplying technical support to other
groups in the Agency including the Office of Air and Radiation, the Office of
Toxic Substances, and the Office of Solid Waste.
The determination of human exposure to toxic organic compounds is an
area of increasing importance to EPA. The passive personal monitors evalu-
ated in this study are possible candidate devices to be employed in future
exposure monitoring studies. Such devices presently are being employed in
industrial hygiene and have many characteristics which make them potentially
very useful in ambient air investigations. However, since the requirements
for ambient air monitoring are more stringent than those for industrial
hygiene, the present investigation was undertaken to obtain an independent
evaluation of a particular passive dosimeter developed for EPA by the Mon-
santo Research Corporation. A key goal of the program was to develop a sound
understanding of the sampling performance of this device to serve as a basis
for valid use of the passive monitors in ambient exposure monitoring.
Thomas R. Hauser, Ph.D.
Di rector
Environmental Monitoring Systems Laboratory
Research Triangle Park, North Carolina 27711
111
-------
ABSTRACT
The basic objectives of this study were to evaluate the performance of
Che EPA passive sampling device (PSD) for sampling of ambient level volatile
organic compounds (VOC's); to develop an understanding of the mechanics of
passive sampling using reversible adsorption; and to apply this understanding
to development of an improved PSD that is usable for sampling of VOC's over
periods of 8 to 24 hours. Laboratory and limited field evaluations of the
standard and modified PSD's were conducted and a model relating sorbent
properties and device design to sampling rates was developed. The results
show the standard PSD's to be useful for sampling of VOC's having large
retention volumes. Modified PSD's having greatly reduced sampling rates show
promise for sampling compounds having retention volumes as low as 5 to 10 L/g
over 8 to 24 hour sampling periods. The use of Spherocarb as an alternative
sorbent to Tenax® GC also was investigated as a means for improving the
performance of the PSD, This sorbent was found to be unsuitable because of
the high temperatures required for desorption. It is recommended that the
model which was developed be used for developing sampling plans for specific
applications, and that more extensive field evaluation of the reduced-rate
PSD's be conducted.
This report is being submitted in fulfillment of Contract No. 68-02-3487
by Battelle Columbus Laboratories under the sponsorship of the U.S.
Environmental Protection Agency. It covers a period from April 15, 1982 to
October 31, 1984, and work was completed as of October 31, 1984.
i v
-------
CONTENTS
Page
Disclaimer ii
Foreword iii
Abstract iv
Tables vi
Figures vii
1. Introduction . . . 1
2. Objective 3
3. Conclusions and Recommendations 4
4. Experimental Methods and Procedures 6
Approach 6
Passive Sampling Devices. 6
Tenax GC Traps 9
Trap and PSD Cleanup 9
Trap and Badge Analysis 9
Chamber Test Procedures 13
5. Experimental Results 18
Analytical System 18
Sampling Rates 19
Spherocarb-Fi1led PSD's 33
Reduced Rate PSD's 38
Shielded PSD's 43
Intercomparison Field Study 46
References 51
Appendices
A. Mechanics of Passive Dosimeter Sampling with Reversible
Adsorption 52
B. Chromatograph Inlet System 64
v
-------
TABLES
Number Page
1 Key to chemicals in Figure 3a 12
2 Badge analysis correction factors 19
3 Core data set for MRC badge performance 21
4 Sampling rateg and desorption efficiencies 22
5 Comparison of individual PSD's 23
6 Velocity effect . 24
7 Relative humidity effect 25
8 TWA sampling rates of PSD's . . 27
9 Retention volumes 33
10 Spherocarb retention volumes 35
11 Spherocarb desorption efficiencies 37
12 Percentage recoveries from spherocarb PSD's 39
13 A/L values for reduced rate PSD's 41
14 Summary of reduced rate PSD results 42
15 Reduced-rate PSD field samples 44
16 Comparison of shielded and unshielded PSD's 45
17 Summary of PSD/Tenax trap data: Field intercoraparison study. . . 48
A-l Calculated PSD sampling rates 58
v i
-------
FIGURES
Number Page
1 EPA thermally desorbable passive dosimeter 7
2 Device holders 8
3a Dual chromatogram of gas sample 11
3b Sample chromatogram of second series of chemicals 14
4 Exposure chamber 15
5 PSD loading and support device for chamber tests 16
6 Acrylonitrile 28
7 1,1-Dichloroethylene 29
8 Trichlorotrifluoroethane 30
9 1,2 Dichloroethane 31
10 Hexachlorobutadiene 32
11 PSD shield assembly 45
A-l Boundary layer over badge surface 55
A-2 Schematic representation of PSD diffusion barrier 57
A-3 Calculated effect of velocity on MRC badge sampling rate .... 59
A-4 Effect of time on sampling rate 61
B-l General purpose GC inlet 65
vii
-------
SECTION 1
INTRODUCTION
In recent years, there has been an increased awareness of the need for
monitoring individual or personal exposures to pollutants and toxic chemi-
cals. This awareness has prompted the development of a variety of personal
sampling devices including battery-driven pump systems, passive systems hav-
ing high specificity for individual compounds, and generalized passive sys-
tems intended for the collection of volatile organic compounds. Within this
latter category, the primary commercial emphasis has been on the use of
carbon-based sorbents for monitoring of the relatively high concentrations of
contaminants found in industrial workplaces. In two earlier programs for the
Environmental Monitoring Systems Laboratory, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina (RTP), Battelle's Columbus
Laboratories (BCL) explored the problems and limitations of using coioier-
cially available passive devices for monitoring ambient level organic chemi-
cals. In addition, a detailed evaluation was made of the performance of one
of these devices under simulated ambient conditions (1-2).
Results of these earlier studies have shown that the commercially avail-
able carbon-based devices are satisfactory for ambient monitoring of selected
volatile organic compounds under some conditions, but they are by no means
completely general in their applicability under realistic ambient conditions.
For example, their performance is impaired under conditions of high relative
humidity. The Environmental Monitoring Systems Laboratory therefore has
undertaken, under separate contract with the Monsanto Research Corporation
(MRC), the development of a passive sampler that is not subject to the same
restrictions as the commercially available devices. The basic concept
involved in the development of this new device has been to empLoy relatively
hydrophobic porous polymer sorbents in order to evolve a system that is
readily subject to thermal desorption for analysis. While much of the
initial work with this device has been conducted using Tenax® GC as the
sorbent, the fundamental applications concept is flexible to permit the use
of other porous polymer sorbents, or even activated carbon, as may be
required for specific applications.
This report addresses the findings of three Work Assignments conducted
consecutively at BCL. The primary objective was an independent evaluation of
the applicability of the MRC passive sampling device (PSD) with respect to
monitoring of volatile organic compounds (VOC's) under ambient conditions.
In the first Work Assignment, sampLing rates were determined for chloroform,
1,1,1-trichloroethane, carbon tetrachloride, trichloroethylene, tetrachlora-
ethylene, benzene, and chlorobenzene. A general.model of passive sampling
1
-------
using thermally reversible adsorption also was developed. The second Work
Assignment then was conducted to extend the list of chemicaLs to include
acrylonitrile, 1,1-dichloroethylene, trichlorotrifluoroethane, 1,2-dichloro-
ethane, a-chlorotoluene, and hexachlorobutadiene, and to test the general
applicability of the performance model. In the third Work Assignment, the
general precepts of the performance model were applied for the purpose of
modifying the PSD to enable long-term (8-24 h) sampling of VOC's. Results of
the first two work assignments and some of the developmental work performed
by MRC are summarized in two papers accepted for publication in the January
1985 issue of Analytical Chemistry (3,4).
2
-------
SECTION 2
OBJECTIVE
The basic objective of Work Assignment No. 13 was to provide an
independent assessment of the efficacy of the MRC passive dosimeter. This
assessment was to include:
1. Evaluation of dosimeter sampling rates and sampling precision at
ppbv concentration levels.
2. Evaluation of the effect of air veLocity on the sampLing rate.
3. Evaluation of the effect of humidity on the sampling rate.
The basic objective of Work Assignment No. 21 was to:
1. Extend the body of information on sampling rates of the MRC
dosimeter by consideration of ten additional compounds.
2. Evaluate the general applicability of the sampling performance
model developed under WA-13.
The basic objective of Work Assignment No. 33 was to improve the overall
performance of the PSD for sampling of volatile organic compounds by:
1. Replacement of :he Tenax® GC sorbent with Spherocarb and/or
2. Redesign of the physical structure of the device.
A specific goal of Work Assignment No. 33 was to achieve accurate
sampling of volatile organics for time periods of 8 to 24 hours.
3
-------
SECTION 3
CONCLUSIONS AND RECOMMENDATIONS
Based on the results of this work, it is concluded that the EPA PSD1s
offer some distinct advantages over other available passive sampling devices
for sampling ambient level volatile organic compounds. Thermal desorption of
collected samples provides more than adequate amounts of sample for use with
conventional GC and GC/MS analytical procedures. In addition the
independence of these devices from high relative humidities yields more
flexible field applicability than the commercially available devices using
activated carbon. Finally, the Tenax® GC can be replaced easily with other
sorbents for customized sampling applications. However, the PSD's should not
be utilized without strict attention to the mechanics of reversible adsorp-
tion and their implications with the respect to particular sampling
requirements. Specifically!
(1) The standard EPA PSD's can be used for sampling ambient levels of
VOC's, but careful attention must be paid to the retention volumes
of target compounds and the appropriate sampling period. In gen-
eral, the standard PSD's are useful for chemicals having large
retention volumes (>100 L/g), but can be used for'only short sam-
pling periods (a few hours or less) for compounds having small
retention volumes.
(2) Reduced rate PSD's having nominal sampling rates of the order of
2.5 cc/min show promise for applications requiring the sampling of
VOC's over extended periods of 8 to 24 hours. However, these
reduced rate devices should not be used for compounds having
retention volumes less than about 5-10 L/g. The current results
also indicate potential blank problems with the use of the reduced
rate devices over short sampling periods.
(3) In any case, the time-weighted average sampling rate for a parti-
cular sampling requirement should be estimated using the model
presented in this report and should be used as a guide in design-
ing the sampling plan for a particular application. In general,
rates significantly less than about 70-80 percent of the initial
rate (Ro) may indicate potential sampling error.
(4) The model of PSD performance presented in this report has been
shown to correctly represent the effects of retention volume,
sampling time, and air velocity on the effective time-weighted
average sampling rates of the EPA PSD's.
4
-------
(5) The use of Spherocarb in the EPA PSD's may offer some advantages
for sampling of a few selected VOC's, but general use of this
sorbent is not recommended because of problems associated with the
high temperatures needed for desorption. Contamination of the
sorbent by unpyrolyzed polymer presents some special difficulties
with cLeanup and preparation of Spherocarb for use in the PSD's.
(6) Because the EPA PSD's are not affected by high humidities, they
are not subject to the same limitations as the conur.ercially
available devices based on activated carbon.
(7) Protective shields developed for the EPA PSD's on this program can
provide protection against contamination during handling of the
devices in the field without significantly affecting their
sampling rates.
(8) Blank contamination of the EPA PSD's has generally not been a
problem, but a few instances of such have been observed. In
recognition of the fact that the PSD's may not always be handled
by trained laboratory personnel in the field, it is recommended
that a formalized procedure and containment system be developed
for cleanup and protection of the PSD's.
(9) It is recommended that further field testing of the reduced rate
PSD's be considered.
(10) Investigation of the application of the EPA PSD to sampling of
pollutants other than VOC's (e.g., NO2, volatile polar organics,
etc.) is recommended to take advantage of the capacity for use of
different sorbents with this device.
5
-------
SECTION U
EXPERIMENTAL METHODS AND PROCEDURES
APPROACH
The experimental approach involved procedures very similar to chose pre-
viously used (1-2) in evaluating commercially available passive dosimeters.
In brief, the PSD's were exposed to mixtures of volatile organic compounds at
ppbv concentration levels under well-controlled conditions. Exposure levels
were monitored by two independent methods, direct GC analysis and active
(pumped) sampling using Tenax® GC traps. Inasmuch as validated reference
sampling rates were not available for the devices, apparent sampling rates
based on the amounts of each chemical adsorbed and exposure times were used
as a basis for comparison of PSD performance.
PASSIVE SAMPLING DEVICES
The EPA PSD is a cylindrical stainless steel (SS) sorbent container with
a double layer of fine mesh SS screen fitted into each end of the cylinder.
The volume between the screens is sufficient to permit filling with about
0.4 g of Tenax® GC. The screens serve as a diffusion barrier, and their
dimensions determine the effective sampling rate of the device. An expanded
view of the PSD is shown in Figure 1; critical dimensions of the device are
shown for one face in Figure A-2 in Appendix A. The device is designed so
that either one face or both faces can be exposed, as dictated by the sam-
pling rate requirements. In use, the device is mounted in a circular ring
clamp that is fitted with a spring clasp for attaching the device to the
subject's clothing. After exposure, the ring clamp can be removed and the
assembly can be stored in metal containers prior to analysis. For analysis,
the device is placed in a close fitting desorption oven for thermal
desorption and analysis by gas chromatography.
For this study, it was decided that prolonged storage was not necessary,
and the exposed devices were mounted directly into desorption h
holder consisted of two threaded SS halves with an internal 0-r
quick-disconnect fittings, as shown in Figure 2. Each dosimete
contained within its own holder at all times except during expo
O-ring materials including Teflon, Kelrez, and Viton were tried
program, with most of the work being conducted using Teflon O-r
cases, the O-rings were baked out prior to use.
olders. Each
ing seal and
r then was
sure. Several
during the
ings. In all
6
-------
STAINLESS STEEL
PERFORATED PLATE
SORBENT
CONTAINMENT
VOLUME
CAP
— STAINLESS STEEL
RETAINER RING
200 MESH
STAINLESS STEEL
DIFFUSION SCREEN
Figure 1. EPA thermally desorbabLe passive dosimeter,
7
-------
PSD holder
open
2d. High temperature holder
Figure 2. Device holders.
8
-------
The standard desorption holder described above could not be used for the
experiments conducted with Spherocarb because of the high temperatures (350-
450 C) needed for desorption of this sorbent. For those experiments, the
special high temperature desorption holder shown in Figure 2d was used. This
unit was constructed from a standard 1.5 in. stainless steel Swaglok cap and
plug assembly. An annular shim was added to the inside of the assembly to
provide a snug fit with the PSD, and the internal surfaces of the assembly
were plated with nickel. The gas inlet tube was coiled flush with the top of
the assembLy so that the inlet gas was well-heated before contacting the PSD.
Heating was provided by a commercially available ring heater of the same type
used with the standard desorption holders. Inlet and outlet tubes were
fitted with quick-disconnect fittings similar to those used with the standard
holders. The total assembly provided a leak proof desorption holder chat was
capable of operation at temperatures as high as 700 C.
TENAX® GC TRAPS
Exposure dosages for the PSD's were monitored both by direct GC analysis
of the test mixtures and by active sampling of the test gas using Tenax® GC
traps. The traps for this purpose consisted of 5-inch sections of 1/4 inch
O.D. SS tubing that were fitted on both ends with quick-disconnect fittings
(see Figure 2c). The tubes were filled with approximately 1/3 g 60/80 mesh
Tenax® GC, with glass wool plugs at the tube ends. During use, the traps
were "plugged in" to the test chamber and were pumped at rates of 10 to
20 cc/min, with the actual rates for each run being measured by a soap-bubble
flowmeter.. Traps were always run in duplicate, and, in some cases, backup
traps were used to test for breakthrough. In most cases, the total volume of
gas sampled per trap was less than 2000 cc, and no breakthrough was detected.
However, in some multiple hour exposure runs conducted later in the program,
the retention volumes of several chemicals were exceeded. In these cases,
primary emphasis was placed on the results of the direct analyses.
TRAP AND PSD CLEANUP
The routine procedure used for preparation of both Tenax3 GC traps and
the PSD's consisted of heating overnight at 200 C with zero-nitrogen (ZN2)
flowing through the devices. Spherocarb-filled PSD's were heated at 450 C.
Cleanup was always performed during the night directly preceding use of.the
devices. Blank contamination by compounds of interest was not detected on
the Tenax® GC devices but was a problem with Spherocarb.
TRAP AND BADGE ANALYSIS
From an analytical viewpoint, the PSD's differ from most commercially
available passive devices in that the entire amount of thermally desorbed
sample (as opposed to an aliquot of solvent desorbed sample) is available for
a single pass analysis. This means that the extreme detection sensitivity
required for analysis of ambient samples collected on conventional passive
9
-------
dosimeters is not needed or necessary for analysis of the PSD's. Conversely,
if a highly sensitive detection system, such as a series combination of elec-
tron capture (ECD) and photoionization (PID) detectors is used, the PSD sam-
pLe must be split in order to avoid detector saturation. Indeed, when the
PSD is used to sample ppbv levels of volatile organic chemicals, it is advis-
able that less sensitive detectors, such as the flame ionization detector
(FID), be operated at reduced sensitivity levels in order to avoid overly
complex chromatograms caused by the myriad of minor components chat are
always present in ambient air samples.
In the current work, both a split sample approach and a whole sample
analysis approach at reduced sensitivity were used with different groups of
samples. The split sample approach was used with experiments with the first
set of compounds, and the whole sample analysis was used for the second set
of chemicals. Both approaches have faults; the use of a splitter requires
special care in the characterization and/or elimination of selective split-
ting of the sample components, and the large amounts of sample available for
whole sample analysis can cause retention time shifts that confound component
identification. Although we obtained more consistent and reproducible
results with the whole sample approach for laboratory samples of known com-
position, some procedure involving subdivision of the sample may be necessary
for ambient field samples. The procedures used are discussed individually
below.
Split Sample Analysis
Traps and PSD's, in their holders were analyzed by identical procedures.
First, they were checked for leakage of the holders by slight pressurization
with GC carrier gas (ZN2). They were then heated to approximately 175 C and
were plugged into the injection port of the GC using an O-ring-free quick
disconnect fitting. Desorption was allowed to proceed for 5 minutes while
flushing the sample with carrier gas. The desorbed sample was passed through
a variable splitter and the sample was then cryofocused onto a fused silica
capillary column maintained at -60 C. In some cases, a separate detector was
used to monitor the unused splitter effluent. The output of that detector
indicated that desorption was essentially complete within two minutes.
j
A Varian Model 3700 gas chromatograph equipped with three detectors (ECD
and PID in series and FID in parallel) and coupled to three integrators was
used for analysis. In the early experiments using the split sample approach,
the FID was disconnected so chat the sample components passed through only
the ECD/PID pair. A pair of 50m fused silica capillary columns coated with
SE-30 was used for component separation. One of these columns was operated
in a splitless mode in conjunction with a lOcc gas sampling loop for analysis
of test gas samples. The other column was connected to the inlet splitter,
and both columns were common at the detector inlet. Split ratios of 150—
200:1 were commonly used for PSD and trap analyses. Dual chromatograms
obtained with the ECD/PID setup are shown in Figure 3a.
10
-------
CO
rv
ci
u">
10
ECD
wr>
KaJL
-------
TABLE 1. KEY TO CHEMICALS IN FIGURE 3a
Chemical
Retention
Time, min.
Decector(a)
ECD PID
Approximate^)
Concentration, ppbv
DL,
ppbv
Chloroform
15.41
X
-
130
0.08
1,1,1-Trichloro-
e thane
16.05
X
X
25
0.02
Carbon
tetrachloride
16.45
X
4
0.005
Trichloro-
ethylene
17.57
X
X
23
0.06
Tetrachloro-
ethylene
20.43
X
X
27
0.01
1,1,2,2-Tetra-
chloroethane
23.11
X
X
>58
(0.01)
1,1,1,2-Tetra-
chloroethane
24.50
X
X
(27)
(0.01)
p-dichloro-
benzene
24. 84
X
_
o-dichloro-
benzene
25.83
X
_
_
1,3,5-trichloro-
benzene
27.89
X
1,2,4-trichloro-
benzene
28.81
X
Benzene
16.48
-
X
3.6
0.1
Methylcyclo-
hexane
20.46
_
X
_
Toluene
19.39
-
X
(10)
(0.1)
Chlorobenzene
21.42
-
X
36
0.13
Ethylbenzene
21.82
-
X
(15)
(0.1)
m&p-xylene
22.01
-
X
(16)
(0.1)
p-ethyltoluene
24.25
-
X
(6)
(0.1)
1,2,A-trimethyl-
benzene
28.84
-
X
(37)
P.l)
(a) X = good response; x = fair response; - = poor response.
(b) Estimated concentrations and detection limits in parentheses; - indi-
cates no basis for estimate.
12
-------
Whole Sample Analysis
For whole sample analyses, desorption and column loading were conducted
in two steps. PSD's and traps were first desorbed into a cryogenic (LAr)
trap, which was then flash heated for loading onto a cooled (-60 C) column.
The same GC noted above was used, but a fixed splitter was used to divert
approximately 90 percent of the column effluent to the FID. In contrast to
results obtained with the inlet splitter (see Results Section) repeated tests
of this outlet splitter showed no evidence of selective splitting. With this'
arrangement, the FID output was used for component quantification, and the
ECD/PID outputs were used solely as an aid for component identification. An
annotated chromatogram obtained with this system is shown in Figure 3b.
CHAMBER TEST PROCEDURES
Test Chamber
The chamber used for device exposure studies is shown in Figure 4. This
chamber is a 200-L glass chamber constructed from two opposing bell jars that
are joined to a central anodized aluminum ring. This ring is provided with
numerous ports for loading, sampling, monitoring devices, etc. Stirring
within the chamber is achieved using a completely sealed internal fan that is
magnetically coupled to an external variable speed drive unit. For the cur-
rent experiments, a stainless steel shroud and swirl dampener was placed over
the fan in order to achieve well-directed flow within the chamber and a
uniform flow pattern in the vicinity of the devices.
Devices were loaded into the chamber using the sample holder shown in
Figure 5. This holder is a stainless steel cylinder with a central rod for
attachment of the passive devices. For any experiment, the devices could be
attached to the holder under a protective clean atmosphere; the cylinder was
closed and was flushed with zero nitrogen; and the cylinder was inserted inco
the chamber, with one end locked in place while the other end protruded
through a seal in the aluminum ring. The concentrations of test species
could then be adjusted in the chamber without exposure of the PSD's. At the
start of a chamber exposure, the outer shell of the cylinder was partially
withdrawn through the seal, leaving the PSD's suspended from the central rod.
At the completion of the run, this procedure was reversed, and the entire
assembly was removed to a glove bag for disassembly and preparation of the.
collectors for analysis.
Prior to each chamber run, the chamber was thoroughly flushed with zero
nitrogen and gas analyses were made by GC to confirm che initial condition of
the chamber. Addition of the test species was made by direct injection of an
appropriate liquid or gas mixture of the pure chemicals. Dilution then was
carried out in the chamber as needed to obtain the desired concentrations.
Once loaded, the chamber was operated in a dynamic mode with about 50 cc/min
of gas being removed for GC analysis and appropriate makeup gas being added
to maintain the chamber at about 0.25 cm H2O above ambient pressure.
13
-------
OQ
C
1
O"
n
LO
rr
0>
3
3
T3
M •
»—•
o
(t.
&>
1—»
O
(A
0*
•1
N
O
**1
3
t—i
&>
O
n
O
O
a>
1
rr
CU
(D
3
n
n
o
O
*-r>
1
w
•
fl>
n
o
p
ci-
n
H" •
(t.
0)
o
L
ACRYLONITRT.LE—5 . 638
+ 1, i-dichloroethylene— 5. 787
a
TRICHLOROTRTFLUOROETHANE—6.452
1,2-DICHLOROETHANE—11.08C
TRANS-1, 3-DICHl.OROPROPENE—15. 829,
1 , 2-DIBROMOE'i'HANE—16. 962
TOLUENE—16.202
_0- XYLENE—19.933
BENZYL CHLORIDE—22.675
1IEXACHLOROBUTADIENE—2 7. 519
RT: STOP RUN
-------
ur« chamb
-------
,, Iff I on O-rlnp, Se.il
Bndp.p on Support Rod
CyIi nde r
I'l usli I np 1'orl
In1n1oss
Cy1indcr
Figure 5. PSD loading and support device for chamber tests.
-------
Velocity Effect Experiments
Most of the runs in the chamber were performed with the air velocity set
at about 100 Epm in the vicinity of the PSD's, as determined by measurements
made with an Alnor hot-wire anemometer. In attempting to determine the
effect of air velocity on the apparent PSD sampling rates, initial experi-
ments were performed by simply reducing the fan speed. However, we found
that the air flow became poorly defined in the chamber at low fan speeds, and
an alternative procedure was adopted. In this procedure, a single PSD was
mounted in the loading cylinder, and the cylinder was withdrawn only a few
inches at the start of the exposure period. An external SS bellows pump was
then used to draw chamber air through the cylinder and over the PSD, with the
pumping rate being determined by a rotometer. The exhaust from the rotometer
was then recycled back into the chamber.
Relative Humidity Effects
Most of the experiments for this study were performed at low relative
humidities (in the range of 7 to 10 percent). However, for two runs the
chamber air was humidified to a level of approximately 90 percent RH.
17
-------
SECTION 5
EXPERIMENTAL RESULTS
ANALYTICAL SYSTEM
Initial work on this study consisted of checking the analytical system
with respect to the desorption and injection procedures used for the Tenax®
GC traps. Because of time delays in obtaining PSD's and the ring heaters
used for PSD desorption, PSD's were not included in this initial checkout,
and it was assumed that the performance of the system was similar for the
traps and PSD's because of the similarities in materials and construction of
the two types of devices. However, after making a number of chamber runs, it
was noted that although reasonably good precision was being obtained with the
PSD's, the absolute quantities of adsorbed test compounds apparent from the
PSD analyses were somewhat higher than expected. At that point, a series of
runs was performed in which both PSD's and traps were used in an active mode
to sample the chamber mixture. The results indicated that even though
identical procedures were being used and the trap data indicated essentially
complete recovery, the apparent recoveries of each of the test chemicals from
the PSD's were consistently greater than could be accounted for in terms of
normal analytical error. Indeed, the apparent recovered amounts of
chlorobenzene were approximately twice the anticipated amounts. Furthermore,
the apparent recoveries appeared to be subject to seemingly subtle changes in
the desorption and injection procedure. In the interest of utilization of
the early data and maintaining continuity, a set of "calibration" runs was
made using the PSD's in an active mode and employing the same procedures used
in the early runs. Chamber runs made after this point in time also were per-
formed using this standardized set of procedures for desorption and
injection. Although this approach is less than ideal from an analytical
viewpoint, it is believed that the correction factors (shown in Table 2)
derived from this approach are valid for the procedures used. In any case,
all raw data shown for the first set of compounds in this report are numbers
actually observed, and the correction factors have been applied only in data
analysis.
A similar set of active sampling experiments was conducted with the
second set of compounds, for which the analytical procedure was changed.
This second set of compounds contained both more volatile, aerylonitrile and
1,1-dichloroethylene, and less volatile, hexachlorobutadiene, chemicals.
However, the average "correction factor" found was 1.00 + 0.10, and we con-
clude that the necessity for a correction factor is due to an interaction in
the desorption and inlet splitter operation rather than some function unique
to the PSD's.
18
-------
TABLE 2. BADGE ANALYSIS CORRECTION FACTORS
(First Set of Chemicals)
Chemical Correction factor^3)
Chloroform
1.28
1,1,1-Trichloroethane
1.07
Carbon Tetrachloride
1.35
Trichloroethylene
1.98
Tetrachloroethylene
1.36
Benzene
1.74
Chiorobenzene
2.12
(a) To be divided into quantities of each chemical
recovered from PSD desorption.
SAMPLING RATES
The normal procedure for use of a passive sampler involves the use of
known sampling rates for each compound to relate the analytical results to
exposure concentrations. In principle, the effective sampling rates for each
compound can be calculated from knowledge of the physical dimensions of the
device, the diffusion coefficient for each compound, and an understanding of
the fundamental mechanics of the sampling system. For many of the commer-
cially available passive systems, the assumption of irreversible sorption and
application of Fick's first law suffice for reasonable estimates of the sam-
pling rates. However, in some cases, characterization of the effective dif-
fusional resistance of the physical structure is not perfectly straight-
forward and such estimates bear considerable uncertainty. For example, with
dosimeters employing porous membranes as the diffusion barrier, pore size
distributions and total porosity of the membranes are not easily measured
with sufficient accuracy, and it is more expeditious to determine effective
sampling rates experimentally. In most cases, estimates of sampling rates
neglect the effects of air velocity on mass transport to the surface of the
device, and the effects of interaction between the air velocity and the
physical structure of the device. In ,the case of reversible sorption, as is
the case with the EPA PSD's, there are, to the best of our knowledge, no
published procedures for estimating passive device sampling rates. Rates
estimated for the EPA PSD's are, therefore, based on the procedure advanced
in Appendix A of this report.
Because of the fact that sampling rates for some of the compounds consi-
dered in this study had not previously been determined for the EPA PSD's,
19
-------
this parameter was chosen as the basis for comparison of the experimental
data. The current data include three separate measures of the gas phase
concentrations in the test chamber: (1) direct GC gas phase analysis,
(2) active sampling via the Tenax® GC traps, and (3) passive sampling via the
EPA PSD's. These data can be ratioed in several different ways to yield
desorption efficiencies and apparent sampling rates:
. . QtV
desorption efficiency = DE =
QgvT
sampling rate = QgRj/Q-p = Rgj (2)
sampling rate = QgVgS/QgDE t s Rgg (3)
In equations (1), (2), and (3), the subscripts B, T, and g refer to PSD,
trap, and gas phase, respectively. Q represents the quantity of an
individual compound detected by the GC; S is the split ratio; V is the volume
of gas sampled; R is the sampling rate; and t is the sampling time. For the
core set of data shown in Table 3, Vg has a value of 10 cc and S = 167; Qg
values are time weighted average values based on 3-4 measurements made during
the course of a given run.
The values of sampling rates and desorption efficiencies derived from
the data in Table 3 using equations (1), (2), and (3) are shown in Table 4.
In making these calculations, values of Qg were adjusted by dividing the
observed values by the correction factors shown in Table 2. Also listed in
Table 4 are the one-hour dual-sided sampling rates estimated from the PSD
dimensions and published Tenax® GC retention volumes (3) using the procedures
outlined in Appendix A. Each derived value of a sampling rate given in Table
4 is a mean value, and is accompanied by the number of points used and the
standard deviation, expressed as a percentage of the mean. The numbers of
points used varies principally because of the fact that different numbers of
combinations were available for the different calculations. However, some of
the observed values were rejected as outliers. The criterion used for
rejection was that no point should differ from the mean by more than 50
percent or twice the standard deviation, whichever was greater. Although
this criterion is quite severe, it permits retention of most of the data.
It can be seen from Table 4 that Rg-p and Rgg values generally agree
reasonably well with each other, with Rg^ values probably being more repre-
sentative because of the larger number of points involved. In most cases,
Rgj values also agree with the estimated sampling rates, with values for
carbon tetrachloride and tetrachloroethylene being the exceptions. In the
case of carbon tetrachloride, difficulties with accurate sampling by both
passive and active samplers have been noted by ourselves (2) and other inves-
tigators (5), but this phenomenon has not yet been satisfactorily explained.
To the best of the author's knowledge, similar difficulties with sampling of
tetrachloroethylene have not been reported previously, and the low rates seen
in the current data may reflect analytical difficulties, or the effect may be
real.
20
-------
TABLE 3. CORE DATA SET FOR MRC BADGE PERFORMANCE
\ Run Huabet
m
763
"TO
?di
709
712
728
804
819
\ Tlat, hr
1/2
i
1
1
1
I
2
1-1/2
1
\tfjf tiir,
10
16.7
16.7
16.7
16.7
16.7
16.7
16.7
21.0/19.0
\xe/«to
Chaaical \
0.024/0.010
0.019/0.021
0.044/0.040
.032/.032
.269/.322
.051/.054
.221/.296
.140/.127
.832/.576
Chlorolora
0.063
0.0)5
0.069
0.052
0.543
0.103
0.164
0.103
1.11
0.200/0. 117/0
0.101/0.1)1/0
,23)/.212/.264
.136/.156/.190
1.48/1.49/1.62
.218/.316/.403
.844/.973/0
.442/.464/.575
1.995/2.06/2.28
0.108/0,102
0.086/0.125
.201/.15*
.128/.132
.591/.654
.198/.220
.386/.443
.218/.207
.652/.586
1,1.1-TtIthloroethart#
0.185
0,118
0 262
0.213
1.075
0.377
0.303
0.149
1.02
0.554/0.488/0
0.547/0.582/0
.794/.728/,753
.495/ .461/ .549
1.173/1.20/1.277
.723/.778/.838
1.105/1.013/0
.536/.535/,545
1.11/1.17/1.26
0.051/0.059
0,029/0.040
.080/.075
.065/.064
.188/.203
.095/.097
,100/,10?
.109/.103
.208/.184
Car boo (ttritMorU*
0.1 Of
0.074
0.133
0.109
0.318
0.170
0.110
0.118
0.331
0.132/0.110/0
0.109/0.125/0
.168/.158/.171
.132/.142/.154
.325/.328/.339
.177/.186/.203
.172/. 171/0
.17?/.183/.066
.316/.326/. 344
0.019/0.022
0.016/0.018
.057/.044
.038/.037
.219/.258
.047/.0*8
.074/.08?
.193/.169
1.00/0.800
Yrlchloro«thjri*Q«
0.051
0.031
0.082
0.065
0.458
0.093
0.062
0.142
1.39
0.252/0.162/0
0.158/0.163/0
.431/.435/.495
.268/.325/.401
1.50/1,58/1.66
.551/.684/.776
.703/.748/0
1.09/1.18/1.20
2.52/2.52/2.68
0.025/0.028
0.019/0.021
.062/.055
.040/.040
.221/.237
.085/.089
.255/.281
.112/.099
.302/.281
Tatrachloroathjrlaoa
0.0??
0.0)7
0.103
0.071
0.359
0.156
0.289
0.095
0.418
0,207/0.197/0
0.170/0.176/0
.2)7/.243/.255
.197/, 212/. 224
.508/.523/. 533
.275/.321/.363
.519/. 545/0
0/. 316/0
.608/.582/.60]
0.014/0,019
0.H2/0
.0)6/.025
.016/.016
.124/.142
.031/.032
.185/.219
.061/. 056
.302/.211
Beaten*
0.038
0.019
0.037
0.027
0.226
0.05?
0.152
0.048
0.372
0.184/0.162/0
0.113/0.144/0
.164/.165/.185
,104/.132/.144
.946/1.04/1,13
.271/.335/.179
1.05/1.14/0
,331/.421/.921
1.66/1.67/188
0.021/0.024
0.068/0.0/8
.052/.057
.034/.034
.214/.240
.130/.Ill
.316/,409
.051/.0*3
.311/.276
CMorobtatiQi
0.050
0.143
0.091
0.062
0.419
0.101
0.273
0,049
0,401
0,285/0.275/0
0.701/0.696/0
.)59/.380/.4 22
.232/.269/.299
1.81/2.00/2.18
. 367/.446/.534
2.24/2,59/0
.357/.418/.33?
2,29/2.06/2.20
Tabl. Ley; I. for «*cb chcaical daia group reprtNcnta
trap 1/irap 2
gas TWa
Badge 1/Badge 2/6«. |*tt - 10 cc
c. badge - badge rate ¦ tlw/npIU tmio
-------
TABLE 4. SAMPLINC RATESAND DESORPTION EFFICIENCIES
_(b) 0(c) (d) (e) _(f) (d) (e) __(g) (d) (e)
Chemical ResC R BT RSD % RSD N DE RSQ N
Chloroform 59.0 59.5 20 39 63.8 21 22 0.95 11 13
1,1,1-Tri-
chloroethane 46.8 45.1 25 40 47.4 32 20 1.02 14 13
Carbon-
tetrachloride 53.9 23.8 17 44 22.4 19 21 0.93 9.4 16
Trichloro-
ethylene 67.7 68.2 23 38 73.2 20 18 0.97 15 15
Tetrachloro-
ethylene 71.5 39.0 31 34 34.4 30 17 0.97 14 16
Benzene 72.2 70.1 22 41 77.0 16 22 1.03 12 13
Chloro-
benzene 66.0 61.3 19 45 61.3 13 22 0.98 12 14
All R's in unics of cc/min.
(b) Estimated using procedure of Appendix A.
Mean of rates calculated by Equation 2.
Standard deviation expressed as percentage of mean.
Number of points included in set.
(f) Mean of rates calculated by Equation 3.
-------
TABLE 5. COMPARISON OF INDIVIDUAL PSD's
Frequency of ranking
Relative^3 ^
Badge Number
Low
Medium
High
response
120
31
5
4
0.84 +_ .14
121
8
31
1
0.90 + .10
122
1
4
35
1.0
Average =
0.92 + 13%
Relative
to PSD number 122.
individual PSD's are reasonably consistent with the precision observed for
the traps, but the process of averaging over all three PSD's contributes
additional apparent imprecision. Inasmuch as the PSD's evaluated in this
program are prototype devices, it is likely that this source of variation in
PSD responses could be reduced significantly by refined manufacturing and
assembly techniques.
Velocity Effects
In principle, the performance of any passive sampler should be affected
by the velocity of air passing over the device. This is caused by the fact
that the sampling process tends to deplete the concentrations of sampled spe-
cies in the vicinity of the PSD, and these species must be replenished in
order to establish a steady-state sampling process. The nature of the effect
can be formalized in terms of boundary-layer theory as we have indicated
previously (2). The magnitude of the sampling rate is determined by the
dimensions and construction details of the individual passive device. How-
ever, inasmuch as most commercially available passive dosimeters, including
the EPA device, have similar gross physical structures and similar intrinsic
sampling rates, it is to be expected that the velocity effects would be
roughly the same for all of these devices. Qualitatively, only a minor
effect of air velocity on the sampling rate at velocities above about 30 to
50 fpm would be expected, but significant decrease in sampling rate should
result at velocities below 30 fpm (see Appendix A). In the current work,
most of the PSD evaluation experiments were performed using an air velocity
of about 100 fpm, i.e., a typical velocity for a person moving about or even
sitting in a well-ventilated area. However, additional experiments were per-
formed at reduced velocities in order to document the existence of the
velocity effect with the EPA PSD.
23
-------
Initial experiments to evaluate the velocity effect were conducted with
the PSD's suspended in the chamber in the normal manner and the fan speed
reduced. The results of these experiments showed poor reproducibility how-
ever, and a re-examination of flow patterns in the chamber revealed seemingly
random fluctuations in flow at the low fan speed. The procedure therefore
was changed to permit exposure of one PSD at a time under conditions of a
well-directed flow (see section on procedures). Three PSD's were exposed
under these conditions at a linear velocity of 10 fpm, with the results shown
in Table 6.
Based on the considerations given in Appendix A, it is estimated that
the effective rate at 10 fpm should be approximately 0.64 times the rate at
100 fpm—a value that is in good agreement with the data in Table 6.
Relative Humidity Effects
In previous studies of passive devices that employ activated carbon as
the sorbent, it was found that the apparent sampling rates of the devices
were significantly diminished at relative humidities greater than about 30
percent (2). This result was not unexpected because of the well-recognized
affinity of activated carbons for water. On the other hand, porous polymer
sorbents such as Tenax® GC are generally considered to be hydrophobic, and it
was expected that passive devices using Tenax® GC as the sorbent would be
relatively unaffected by high relative humidities. In the current work, two
exposure tests of the PSD's were made at relative humidities of 87 and
92 percent, with the results as summarized in Table 7. With only six data
points available for each chemical, median apparent sampling rates and mean
values are both cited in Table 7. Although the scatter is somewhat greater
for some of the chemicals than was observed for the core data set given in
TABLE 6.
VELOCITY
EFFECT
(v
= 10 fpm)
Apparent
sampling rate, cc/min
Chemical
1
2
3
R10
Rio/Rioo
Chloroform
A 7.7
45.6
52.4
48.6
0.82
1,1,1-Trichloroethane
23.9
23.6
23.5
23.6
0.52
Carbon tetrachloride
13.6
13.3
14.0
13.7
0.58
Trichloroethylene
38.3
39.9
37.7
38.6
0.57
Tetrachloroethylene
21.2
21.6
21.4
21.4
0.55
Benzene
49.3
51.8
54.3
51.8
0.74
Chlorobenzene
41.8
45.2
43.2
43.4
0.71
Mean
= 0.64 + I
Median
= 0.58
24
-------
TABLE 7. RELATIVE HUMIDITY EFFECT^
Apparent
sampling
rates,
cc/min
Chemical
Mean RSD, %
N
Median
Chloroform
95.3
15
6
96.1
1,1,1-Trichloro-
ethane
68.4
33
6
67.6
Carbon
tetrachloride
28.2
30
6
29.1
Trichloro-
ethylene
77.3
13
5
80.8
Tetrachloro-
ethylene
40.4
12
6
40.4
Benzene
66.1
37
6
62.6
Chlorobenzene
59.9
9.1
6
60.9
Rates obtained at RH = 87 and 92 percent
Table 3, the mean apparent rates (or the medians) compare favorably with
those obtained at low humidities in all cases except for chloroform and
1,1,1-trichloroethane. With the latter two materials, the apparent sampling
rates were significantly higher at the high humidities. A re-examination of
the raw data for 1,1,1-trichloroethane reveals that the results obtained at
92 percent relative humidity yield an average apparent sampling rate of
47.3 cc/min (shown in parentheses in Table 7) which agrees quite well with
the low humidity data. On the other hand, very high results were obtained in
the 87 percent humidity experiment. The latter results may be due to spuri-
ous contamination, but such cannot be supported within the framework of the
current results. However, in the case of chloroform, all of the results
indicated higher sampling rates than those obtained at low humidities. In
experiments conducted later in this program, it was noted that hexane was
sometimes a system contaminant that interfered with the chloroform GC peak.
Such contamination may have affected the results of these humidity
experiments.
Model Evaluation
Because of the apparent success in using the model presented in Appendix
A to predict the effects of retention volume and sampling time on the
25
-------
time-averaged sampling rates of the EPA PSD's, additional exposure
experiments were conducted to provide a more severe test of the model. In
these tests, ten more chemicals, including three very volatile compounds,
acrylonitrile, 1,1-dichloroethylene and trichlorotrifluoroethane were added
to the test set. At the same time, the desorption procedure was changed (see
the Procedures Section) to permit less ambiguous quantification of the
analytical results.
The initial plan of these additional experiments was to conduct PSD
exposures to all eighteen chemicals for various time periods, ranging from 15
minutes to 24 hours. However, after conducting several such experiments at
time periods up to 4 hours, two faults in this plan became apparent:
1. With several of the more volatile chemicals, the total volumes being
sampled with the traps exceeded the retention volumes for these
compounds even at the low sampling rates being employed (ca.
10 cc/min).
2. The large amounts of sample being collected by the PSD's caused
shifts in the GC retention times. These shifts along with the com-
plexity of having to identify 18 chemicals amidst an array of minor
system contaminants made peak identification uncertain or even
impossible in many cases.
For these reasons, the initial data were discarded and two changes were made
in the experimental plan:
1. Trap samples were still collected, but were used only as crosschecks
on the data for the less volatile compounds (those for which reten-
tion volumes were not exceeded in any given experiment). The direct
gas analyses then became the primary reference for determining
sampling rates.
2. The list of chemicals was limited to the ten new compounds to
facilitate visual comparisons of chromatogram patterns.
Exposures of the EPA devices, in triplicate, then were made using ppbv levels
of the ten chemicals for time periods of 1/4, 1/2, 1, 2, 4, 12, and 24 hours.
The results of these tests are given in Table 8. The median values of the
apparent sampling rates are used in this table because only three data points
were obtained for each sampling condition. In most cases, there were at
least two points close to the indicated median values.
Although there is some scatter in the data, the variation is not gener-
ally significant with respect to the expected experimental error, i.e., +10
percent. As predicted by the model given in Appendix A, the apparent
sampling rates of the very volatile compounds decline sharply as the sampling
time is increased, while those compounds having large retention volumes
display only slowly declining sampling rates. A better picture of the
correspondence between the model and the experimental data can be gained by
consideration of Figures 6 to 10. These curves depict the experimental
26
-------
TABLE 8. TWA SAMPLING RATES OF PSD's (Median Values, cc/min)
Averaging Period, hr
Chemical
0(a)
0.25
0.5
1
2
4
12
24
Acrylonitrile
94.7
72.3
50.8
26.9
16.2
9.3
2.4
1.1
1,1-Dichloroethylene
82.2
37.9
19.8
10.1
4.5
2.7
0.5
0.2
Trichlorotrifluoroethane
62.1
10.7
7.3
3.7
3.1
2.3
1.0
0.7
1,2-Dichloroethane
81.2
95.8
67.0
55.9
41.4
21.2
11.9
6.9
trans-1,3-Dichloropropene
71.0
86.2
71.0
66.7
53.6
74.8
54.3
32.9
Toluene
75.9
75.5
71.2
65.9
45.0
43.0
34.9
25.4
1,2-Dibromoethane
69.5
80.3
61.0
64.7
48.1
60.8
57.7
—
o-Xylene
65.1
74.8
70.3
57.2
62.5
75.2
44.1
28.5
a-Chlorotoluene
83.9
63.0
73.8
62.0
68.2
72.7
45.8
55.3
Hexachlorobutadiene
(42.0)
40.5
40.3
41.3
41.9
40.4
(27.1)
I
31.4
(a) Estimated based on device dimensions and chemical diffusion coefficients (see Appendix A).
(b) Estimated from current data.
-------
cc/min. 'JO
Figure 6
94.7 cc/min.
4.9 + 0.8 L/q
2.0 cc/min.
j 3 i i i i i i i—. i i i i _JL
11 12 13 14 16 16 17 18 19 20 21 22 23 24
t, hr
Acryloriitrlle.
-------
Ro = 82.2 cc/min
Vb ¦ I-5 +0.1 L/q
SDEV -2.8 cc/min.
cc/min
Figure 7. 1,1-Dichloroethylene,
-------
100
90
80
70
60
50
40
30
20
10
0
Ro = 62.1 cc/mln.
Vb = 0.5 + 0.08 L/q
SOEV +1.3 cc/mln.
\
Xx~. x
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
t, hr.
Figure 8.
Trichlorotrifluoroethane.
-------
100
90
80
70
60
50
40
30
20
10
0
Ro ® 81.2 cc/mln.
Vb = 18 + 4 L/q
SDEV = 3.3 cc/mln.
i
i.
x
j.
X
X
±
X
X
J.
X
X
X
X
A
X
_1_
X
X
X
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
t, hr.
Figure 9, 1,2 Dichloroethane.
-------
100
90
80
70
60
50
40
30
20
10
0
Ro = 42 cc/mln.
Vk =» 324 L/q
SDEV =1.3 cc/nln.
-xx * *
X
00
J 1 1 1 1 1 1 1 I I I I 1 I I I I I 1 I I I 1 1_
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
t, hr
Figure 10. Hexachlorobutadiene.
-------
points and curves calculated for several compounds using equation A-8'.
Retention volumes (V^) for these calculations were derived from the
experimental data by solving equation A-8' at each point followed by
averaging the resultant values of to obtain a best overall value. The
experimentally derived retention volumes for the more volatile compounds in
comparison with values cited in the literature are shown in Table 9.(5) In
principle, the same approach could be used to calculate the retention volumes
of the less volatile compounds, but, for those compounds, the experimental
fluctuations are of a magnitude equal to or greater than the expected
declines in sampling rates. Thus meaningful calculations cannot be made.
SPHEROCARB-FILLED PSD's
As indicated in the previous sections of this report, the Tenax® GC-
filled PSD's exhibit severely declining sampling rates when used for multiple
hour sampling of compounds having low retention volumes. One approach to
solution of this problem involves the use of a sorbent having larger reten-
tion volumes. Ideally, the sorbent selected would have a uniform retention
volume of say 500 liters/gram for all compounds. UnfortunateLy, such a
sorbent is fundamentally infeasible. A sorbent having increased retention
volumes for some compounds will necessarily have larger retention volumes for
all compounds, and there is the risk that in enhancing the ability to sample
certain compounds the ability to recover some of the less volatile compounds
in a practical manner may be lost.
Some preliminary work conducted with Spherocarb under a separate con-
tract with EPA (Contract No. 68-02-3745(WA7)) indicated that this sorbent
might be suitable for sampling of compounds as volatile as vinyl chloride and
acrylonitrile, and that compounds such as toluene and tetrachloroetbylene
still could be recovered at desorption temperatures below their decomposition
temperatures.
TABLE 9. RETENTION VOLUMES(a)
Chemical Vb(cal),L/g Vb(Lit),L/g
Acrylonitrile
A.9+0.8
0.3-7.0
1,1 Dichloroethylene
1.5+0.1
2.6
Trichlorotrifluoroethane
0.5+0.08
0.23-0.47
1,2-Dichloroethane
18+4
24.4
(a) Calculated from PSD data.
33
-------
Retention Volumes
Spherocarb is a relatively new sorbent material, and little was known
about the retention volumes for many of the compounds of interest. There-
fore, measurements of retention volumes were undertaken for the set of 12
organic compounds identified as test compounds for Task 33. These were
measured by the standard chromatographic procedure involving determination of
the retention times as a function of temperature and column flow rate. Also
included in this series of experiments was the determination of the retention
volume of water vapor as an indicator of the possible susceptibility of
Spherocarb to humidity. The results of these measurements are shown in
Table 10. It can be seen that retention volumes for all of the compounds in
Table 10 are significantly larger than those shown for Tenax® GC in Table
A-l, indicating potentially good sampling ability for these compounds. The
retention volume for water vapor is relatively low, suggesting that sampling
with Spherocarb should be relatively insensitive to humidity.
Preliminary Evaluation of Spherocarb
Initial tests of Spherocarb involved the use of a Spherocarb-filled PSD
as an active sampler to determine appropriate desorption conditions and effi-
ciencies. For these experiments, the PSD was mounted in the high-temperature
holder described above, and was kept sealed at all times. At this point,
several problems associated with the use of Spherocarb became obvious:
1. Apparent desorption efficiencies for the test compounds were not
readily reproducible, and were quite low for several of the chlorin-
ated hydrocarbons. In particular, the recovery efficiency of 1,1,1-
trichloroethane was very sensitive to temperature, with recoveries
steadily decreasing at temperatures above 300 C.
2. Recoveries of hydrocarbons, notably benzene, on the other hand
exceeded 100 percent in all tests, and benzene was always found in
blank runs in spite of extensive bakeout periods at temperatures up
to 500 C.
These observations illustrate two separate problems with the use of
Spherocarb. A close examination of the sorbent, as received from the manu-
facturer, revealed the presence of unpyrolyzed polymer beads mixed in with
the Spherocarb. Examination of several different batches of the sorbent
showed some variability in the amount of polymer present, but all batches
examined contained at least some of this material.
Several attempts were made to clean the 1.5 grams of sorbent in one PSD
by pyrolyzing the material while flushing with UHP nitrogen. In these
experiments, the output gas from the desorption cell was connected directly
to a flame-ionization detector so that progress in cleaning the sorbent could
be monitored continuously. Large amounts of hydrocarbon (full-scale deflec-
tion on the least sensitive FID range) were observed in heating the PSD from
500 to 600 C. After several days at 600 C, the FID had still not returned to
baseline. After three additional 8-hour periods at 700 C, the output
34
-------
TABLE 10. SPHEROCARB RETENTION VOLUMES
Chemical
t ,C/Vb,L/g
Vb(298 K)(a>
Vb(311 K)(a'
Acrylonitrile
250/0.231
225/0.413
200/0.804
1700
720
1,1-Dichloroethylene
300/0.117
250/0.315
200/1.14
2400
1000
Methylene Chloride
250/0.137
225/0.230
200/0.421
415
190
Trichlorotrifluoro-
echane
300/0.294
275/0.494
250/0.900
14xl03
5600
Chloroform
275/0.324
250/0.533
225/1.01
4300
1800
ip2-Dichloroethane
30/0.266
275/0.443
250/0. 779
8500
3400
1,1,1-Trichloroethane
300/0.132
250/0.336
225/0.740
5000
2000
Benzene
350/0.342
325/0.521
300/1.07
5xl05
2xl05
Trichloroethylene
350/0.199
325/0.365
300/0.638
4xl05
lxiO5
trans-l,3-Dichloro-
propene
375/0.217
350/0.353
325/0.605
4xl05
lxlO5
Toluene
- 350/0.336
325/0.576
300/1.05
5xi05
2xl05
Te'rachloroethylene
375/0.463
350/0.799
6xl06
2xl06
Water
60/0.143
45/0.323
30/0.740
1.02
0.47
Estimated by extrapolation of high temperature results.
-------
appeared Co be clean, and subsequent blank desorption runs at temperatures up
to 450 C revealed no benzene or other hydrocarbons being emitted. However,
further tests of desorption efficiencies of the other compounds still showed
low and erratic recoveries.
It was suspected that a part of the problem in low recoveries of the
chlorinated hydrocarbons might be due to catalytic dehydrochlorination occur-
ring on the stainless steel surfaces of the PSD and desorption holder. This
hypothesis was tested by placing an empty PSD (no sorbent) in the holder, and
sampling gas from the chamber through the holder. During this process, the
temperature of the holder was increased stepwise from ambient to 400 C. At
temperatures up to about 250 C, the apparent gas composition was essentially
the same as measured directly in the chamber. However, at temperatures above
250 C, progressive degradation of the chlorinated hydrocarbons was noted,
with 1,1,1-trichloroethane appearing to be most affected. Indeed, at temper-
atures between 250 and 350 C, the gas analyses suggested the conversion of
1,1,1-trichloroethane to 1,1-dichloroethylene.
Catalytic dehydrochlorination is not an uncommon problem when dealing
with chlorinated hydrocarbons, and one common solution is to use nickel or
silica in place of stainless steel. We therefore proceeded to replace the
stainless screens of the PSD with nickel screen of the same size, and ail
other exposed parts of the PSD and the desorption holder were plated with
nickel. The experiment noted above was then repeated. Results indicated
better resistance to degradation and suggested that good recoveries might be
possible if the desorption temperatures could be held below 400 C.
Further tests of the desorption efficiencies then were conducted using
different temperature ramps in an attempt to gain the best compromise between
obtaining good recovery and accomplishing the desorption in a reasonable
amount of time. A comparison of desorption efficiencies obtained by three
different procedures is presented in Table 11. The first two of these
involved stepwise temperature programming, with isothermal operation at
selected temperatures during the process. The main difference between the
first two procedures lies in the higher final temperature for the first
method. The last two sets of data represent results of ballistic programming
of the desorption holder to 400 C. All of these data were obtained using the
PSD in an active mode, with the direction of gas flow during desorption being
opposite of that during sampling. These are not individually typical data,
but rather illustrate some of the extreme variability observed in these
experiments.
Passive Sampling with Spherocarb
Despite of the inconsistent results obtained with Spherocarb in the
preliminary experiments, conflicting information from other programs sug-
gested that this sorbent might still be useful for passive sampling of
volatile organics. Therefore a series of exposures of the nickel-plated
Spherocarb-fi1led PSD's in the 200 L chamber facility was conducted. Tests
were conducted for 0.5, 1, 8, and 24 hr. Mechanical problems related to
handling of the PSD's in the chamber were experienced in both the 0.5 and
36
-------
TABLE 11.
SPHEROCARB DESORPTION EFFICIENCIES
(active mode)
Chemical Run A Run B Run C Ran D
Acryloni trile
102
90
MA
NA
1,1-Dichloroethylene
9.6
35
22
18
Methylene Chloride
19
30
52
41
Trichlorotri 11 uoroe thane
9.4
33
26
17
Chloroform
49
(488)
14
10
1,2-Dichloroethane
93
98
20
15
1,1,1-Trichloroethane
28
41
• 18
14
Benzene
82
88
107
78
Trichloroethylene
74
77
27
23
trans-1,3-Dichloropropene
62
59
51
49
Toluene
82
78
49
42
Tetrachloroethylene
71
74
33
33
Run key: A. 5-minute hoLd at 150 C; 5-minute hold at
250 C; 10-minute hold at 350 C; 40-minute
total cycle.
B. Similar to A, but with additional 10 minutes
at 400 C; 60-minute total cycle.
C. Ballistic program to 400 C, with 10 minutes
at 400 C; 45-minute total cycle.
D. Same as C.
NA = aerylonitrile not analyzed because of interference.
Numbers in parentheses represent probably misidentified
peaks.
37
-------
1 hour runs, and the 1 hour run was repeated. With the second 1 hour run, we
experienced analytical problems. A review of the results from the 8 and
24 hour runs led us to the conclusion that additional attempts to complete
the short term exposure work would not be meaningful.
The results of the 8 and 24 hr runs are cited in Table 12. In this
table, the results are expressed as the apparent percentage recovery for each
chemical based on the known time-weighted average concentrations and the
sampling rates derived from the physical properties of the PSD's. With the
large retention volumes for all chemicals listed in Table 12, the 8 and 24 hr
sampling rates should be essentially the same as the Ro values listed for the
Tenax® GC-filled PSD's in Table A-l. Also listed in Table 12 are average
recoveries, standard deviations and number of points used for each chemical.
These data suggest that the Spherocarb PSD's may be satisfactory for a few
selected chemicals such as methylene chloride, 1,1-dichloroethylene, and
tetrachloroethylene, but for most of the other compounds, the devices are
clearly unsatisfactory. It also should be noted that the apparent recoveries
listed in Table 12 generally do not correspond well with those listed in
Table 11. This may be due—in part — to the fundamental difference in
desorption from an active sampler as opposed to desorption from a passive
sampler. (With a highly effective active sampler, most of the sample is
concentrated on the inlet edge of the sorbent bed, and desorption is effected
most easily by reversing the flow direction.) More generally, it is believed
that the lack of correspondence between the two tables reflects a persistent
inability to obtain good reproducibility with Spherocarb.
REDUCED RATE PSD's
Another possible solution to the problem of poor sam'pling o£ the very
volatile compounds by the standard PSD's is to reduce the sampling rates by
alteration of the diffusion barrier. This approach was taken as an
alternative to the use of Spherocarb in place of Tenax® GC.
In the current work, several different alternative diffusion barriers
were considered. These included: (1) filter paper between the screen
assemblies of the standard PSD's; (2) finer screen; (3) compound plate
assemblies with the holes in the first plate being offset from those in the
second plate; and (4) simple plates having a single small hole. Simple
calculations utilizing the precepts outlined in Appendix A indicated that in
order to meet the objective of 8-24 hr sampling of compounds having retention
volumes as low as say 5 L/g, the sampling rates of the PSD's would have to be
reduced to the order of a few cc/min. Such a reduction is not possible with
the first two approaches cited above, but can be achieved with either the
compound plate or single plate approaches.
Both approaches were tried, with the result that it was demonstrated
that rates of the order of 2-10 cc/min could easily be achieved. However, it
quickly became obvious that the reproducibility of sampling rates between
separate PSD's was not good. A careful examination of barrier plates and
PSD's revealed irregularities in the physical dimensions of the PSD's. For
38
-------
TABLE 12. PERCENTAGE RECOVERIES FROM SPHEROCARB PSD's
PSD No.
8 hr
24 hr
Ave/RSD/N^3)
Chemical
102
115
121
102
115
121
1,1-Dichloroethylene
150
78
60
750
39
250
82/48/4
Methylene Chloride
107
101
68
122
101
59
93/24/6
TrichLorotrifluoro-
ethane
14
5
1
46
14
22
13/6/5
Chloraform
53
33
23
(253
103
144)(b)
36/15/3
1,2-Dichloroethane
16
7
9
202
33
109
16/12/4
1,1,1-Trichloroethane
6
4
3
26
15
11
8/5/5
Benzene
412
230
360
(all
>1000)
700
Trichloroethylene
55
27
14
11
19
14
15/3/4
Toluene
24
17
13
(all
u
o
o
0
1 i
A
)
18/6/3
Tetrachloroethylene
91
81
32
300
53
99
88/8/4
Average value of percentage recovery; reLative standard deviation; and number
of points included in average.
(b) Chloroform believed to be obscured by interfering hexane peak.
(c) Possible blank or interference problem.
39
-------
example, most of the PSD's in hand have I.D.'s within 0.002 in. of the aver-
age values, but at least one PSD was found to vary in I.D. by 0.005 in. from
the average. Also, several of the PSD's were found to be slightly irregular
in shape. These variations are insignificant with respect to normal usage of
the standard PSD's but are very important when the barrier plates are used to
reduce the sampling rates.
For example, the hypothetical case where a barrier plate having a
central hole of 0.5 mm I.D. is used to replace the outer screen assembly of a
standard PSD may be considered. If a mismatch of 0.001 in. occurs between
the O.D. of the plate and the I.D. of the PSD because of an irregularity in
shape of the PSD, the area of the gap at the edge of the barrier plate (1.5
cm radius) would amount to 0.024 cm^ as compared with an area of 0.0079 cm^
for the center hole. That is, the sampling rate due to edge leakage would be
3 times as great as the rate through the center hole. It is obvious from
this example that either tolerances must be kept very close, or some
mechanism must be used to seal the edges of the barrier plate. Optional
approaches considered for sealing the edges included the use of thin copper
and aluminum gaskets and high temperature soldering. The latter approach
gave a good seal, but resulted in contamination of the PSD by the flux that
was necessary and presented potential difficulties in replacement of the
Tenax® GC bed. The gaskets appeared to give good seals initially, but these
failed after a few normal temperature cycles in the desorption unit.
The final approach used was to carefully select PSD's that were free of
shape irregularities and which had closely matched I.D.'s. Barrier plates
then were machined to O.D.'s that were very slightly greater than the I.D.,
of the PSD's, and were "sprung" into the PSD's in place of their standard
outer screen assemblies.
Three of the reduced rate devices were prepared by replacing the outer
screen assemblies of standard PSD's with 0.030 in thick stainless plates
having single 0.5mm I.D. holes to serve as the diffusion barriers. These
then were exposed for time periods of 0.5, 4, 8, and 24 hours to the twelve
challenge compounds at nominal concentrations of 10 ppbv each.
A cursory examination of the results obtained with these PSD's indicated
good precision, but poor correspondence with Ro values calculated by the
usual procedures. This was not unexpected because of the extreme difficulty
of providing a perfect seal at the edge of the device, while maintaining the
flexibility of exchangeable diffusers. The following approach was therefore
adopted:
1. It was assumed that the literature values of the diffusion
coefficients were correct, and that our model for reversible"
adsorption is applicable. (Appendix A)
2. The apparent value of A/L was calculated for each data point.
3. The raw data was screened to eliminate obviously incorrect points.
For example, all of the trichlorotrifluoroethane data was totally
40
-------
inconsistent with all previous experience with this chemical. Also,
there was an obvious problem with integration of some of the chloro-
form peaks, and these data were dropped from this analysis.
4. An iterative sorting procedure involving successive rejection of
points more than two sigma from a floating mean was used to identify
the most consistent value of A/L.
5. This value of A/L was used to calculate values of Ro, and the data
for each chemical were analyzed as a group, by reducing the data to
relative values of R/Rcal. (This approach increases the sizes of
the data sets by removing the time dependence of the data.)
The iterative sorting procedure yielded a value of A/L of 0.482
+ 9.8 percent with 87 points included in the analysis. Examination of data
for the individual devices gave values as shown in Table 13. It can be seen
that there is no significant difference between the three devices. Apparent
rates and values used in the analysis of the data are shown in Table 14.
Limited Field Study
The reduced rate PSD's were designed for use over extended sampling per-
iods of 8 to 24 hours. Inasmuch as the field intercomparison study permitted
sampling periods of only 2 hours, and results suggested a possible blank
problem with the use of the reduced rate devices for short sampling periods,
a limited field study was conducted to give preliminary evaluation of the
potential of these devices. In this field experiment, reduced-rate PSD's
were exposed in triplicate at three different sites, and active samples were
collected simultaneously using Tenax® GC traps. The three sites were:
1. At a BCL employee's home
2. In a hallway near the main BCL shop facility.
3. In a 5th floor BCL office complex, well removed from laboratory
activities.
TABLE 13. A/L VALUES FOR REDUCED RATE PSD's
Device No.
A/L, cm
RSD, Z
N
23
0.497
9.1
32
30
0.453
20
37
52
0.476
9.5
32
all
0.482
00
87
41
-------
TABLE 14. SUMMARY OF REDUCED RATE PSD RESULTS(a)
D(!i>
Vb( = '
Apparent
Race
S/Scal^)
Chemical
3. 5h
4h
8h
24h
Ro
8h
2ih
Acrylonitrile
6.35
4.9
2.34
2.26
2-45
8.79
5.64
7.38
3.55
3.31
3.27
1.46
2.04
1.70
3.06
1.15
-7-28!
n=9
0.75
0.46
1,i-Dichloroethylene
5.51
2
2.34
2. 12
2.34
2.02
1.65
1.90
1.81
1.62
1.61
1.23
1.13
1.20
2.66
1.17
+/-36Z
n=12
0.57
0.26
Methylene Chloride
(6.31)
5.5
1.54
1.42
1.77
2.65
2.07
1.90
2.02
1.69
1.71
1.53
1.57
1.81
3.04
0.89
-7-193:
n=9
0. 78
0.50
Trichlorotrifluoroethane
4.16
0.5
0. 72
1.50
1.45-
1.91
1.53
1.63
1 . 73
1.48
1.51
1.26
1.42
1.84
2.01
3.0
"7-250!
11
0.25
0.09
Chloroform
5.33
18.9
-------
Results of this limited study are shown in Table 15. Data for the run
near the BCL shop is restricted to two PSD samples; the third sample was lost
because of a power outage during the analysis. Trap analyses for the more
volatile components of the target set of compounds are considered question-
able because, in many cases, the retention volumes of these compounds were
approached or exceeded by the sample volumes.
For this series of runs, the reproducibility of Tenax0 GC trap analyses
was very poor, with reasonable duplication being obtained in only seven of
the pairs. However, the average precision of the PSD analyses is quite
reasonable and is in generally good agreement with precision observed in the
field intercomparison study and with earlier laboratory work. Because of the
tenuous nature of the Tenax® GC trap results, estimation of accuracy of the
PSD collection is arguable, and has been attempted only for the last 5 (least
volatile) compounds as shown in the last column of Table 15. These data sug-
gest that the reduced-rate PSD's have some promise for accurate sampling of
VOC's, but considerably more data will be required to validate their use. It
is particularly recommended that further evaluations of the reduced-rate
PSD's in the field include the use of a reference method that can be shown to
reliably sample the more volatile compounds. One possibility might be to use
pumped canister samples or Tedlar bag samples as references.
SHIELDED PSD's
While no special problems related to handling of the PSD's have been
observed in our laboratory evaluation, field tests conducted in conjunction
with other EPA programs have indicated the desirability of physical protec-
tion of the devices during exposure. In particular, it would be desirable to
have a light weight protective shield that would prevent direct handling and
contamination of the PSD during its field exposure. Several designs for such
a protective shield were considered, with the final design as shown in Fig-
ure 11. This simple shield consists of a thin aluminum shell that was
machined from 1.25 in. rigid aluminum conduit and end plates made from com-
mercially available chromium-plated sink strainers. The aluminum conduit was
bored out to accommodate the PSD, leaving a thin lip at one end to retain one
end plate. At the other end, the end plate is held in place by a snap ring.
The result is a rugged light-weight shield with end plates that present very
little restriction to air access to the PSD.
A total of five attempts were made to determine if the protective
shields influence the sampling rates of the normal PSD's. However, problems
were encountered in handling of these shielded devices because of the
increased overall dimensions of the devices — they are too large to fit
easily into our loading device for the chamber. In each of the first four
attempts, at least one shielded PSD was displaced from the suspension rod
because of rubbing against the loading device. In the fifth trial, we did
manage to recover the devices. This run included 2 shielded PSD's and one
unshielded PSD for comparison. Results for this run are shown in Table 16.
43
-------
TABLE 15. REDUCED-RATE PSD FIELD SAMPLES ^
Sampling Site,
time
PSD(b)
Chemical
Shop,
4h
Home,
8h
Office,
8h
Precision,
:
PSD/Trap
Acrylonitrile
21/5.5
ND/ND(d)
3/ND
29
51.4/45.1
1.98/1.63/1.95
2.0/.52/.95
1,1-Dichloroethylene
6.14/1.3
9.2/2.5
ND/ND
27
15.2/19.5
31.2/45.3/4.3
16/22/6.1
Methylene Chloride
5.55/7.20
11.9/49.2
ND/ND
12
17.9/14.1
103/122/36
1.3/8.8/7.9
TrichlorotrifluoroeChane
1.4/ND
ND/ND
ND/ND
13
10.1/10.1
7.2/10.2/8.7
ND/7.1/9.7
Chloroform
316/92
117/26
285/35
17
405/169
35/48/50
81/111/97
1,2-Dichloroethane
31.4/1.1
25.6/4.2
19.2/2.5
17
23.1/18.0
8.92/11.8/13.8
8.6/11.2/100
1,1,1-Trichloroechane
3700/720
11.9/ND
22.3/ND
11
0.64/0.13/7
3100/2600
ND/7.84/8.07
9.6/14.1/11.3
Benzene
10.0/8.2
13.5/7.3
12.9/2.4
23
1.26/0.25/5
13.1/105
4.0/11.5/9.5
2.8/5.4/3.8
Trichloroechylene
20.6/15.0
5.6/7.7
13.6/5.7
22
0.89/0.09/5
13.0/15.1
4.3/2.0/2.7
4.7/5.3/.65
Toluene
27.2/30.2
1.5/35
31.1/25.2
(26)
0.69/0.41/7
18.9/39.4
1.7/0.76/19
7.5/11.8/12.6
Tetrachloroethylene
15.2/14.5
15.5/2.8
10.1/8.4
25
0.69/0.34/6
18.3/13.7
0.96/1.08/10.6
2.3/4.8/6.3
Total
Average
22
0.81/0.23/23
_ _ Trap Concentrations
a. Ccr.centracions in ng/L; data group fornat = „ :
" or pSD Concentrations
b. Average of relaEive standard deviations within PSD group9
c. Relative concentrations/standard deviations/number of points included; Trap concentra-
tions for lighter chemicals considered inappropriate because of small retention volumes.
d. ND ¦ not detected.
44
-------
Figure 11. PSD shield assembly.
TABLE 16. COMPARISON OF SHIELDED AND
UNSHIELDED PSD's
Chemical Rsh/Runsh +/-
1,1, l.-Trichloroethane 1.04 0.03
Benzene 0.97 0.01
TrichLoroethylene 0.92 0.06
Toluene 0.98 0,20
Tetrachloroethylene 1.51 0.05
45
-------
With the exception of the case of tetrachioroethylene, these results indicate
little or no effect of the shields on the sampling rates of the PSD's. The
high results obtained for tetrachioroethylene are believed to be due to an
analysis anomaly.
INTERCOMPARISON FIELD STUDY
A brief intercomparison field study was conducted at BCL during July and
August, 1984, with the primary purpose of comparing cryogenic sampling
methodology with active sampling using several different sorbents. The bulk
of this study was conducted under Work Assignment 37 of this contract, and
the general results of the study were included in the Final Report for that
Work Assignment. However, a limited field study of the performance of Tenax®
GC-filled PSD's also was conducted in conjunction with this intercomparison
study.
Briefly, a sampling station was set up in the BCL parking lot, with
instrumentation for cryogenic sampling, multiple active sampling, and collec-
tion of pressurized air samples. The air was collected through a common
manifold with provision for fortification of the air with standard mixtures
of the VOC's of interest. This permitted fortification of selected samples
as a part of the quality assurance program. A cylinder of air containing
selected components of interest was prepared as a blind sample by EPA/EMSL as
a further check on sampling and analytical quality. Analyses of the cryo-
genic samples and pressurized air samples were conducted within the sampling
station by gas chromatography using flame ionization, electron capture, and
mass selective detectors. Active samples collected as a part of the main
body of this study were returned to the laboratory and were analyzed by
GC/MS.
It was inappropriate to collect the passive samples within the station's
manifold system. Therefore, PSD's were exposed in triplicate on the roof of
the station at a point close to the manifold inlet. Separate active samples
were collected at the same point using two Tenax® GC trap samplers. Standard
high-sampling rate PSD's were used in all runs, and additional low-rate PSD's
were exposed in selected runs. PSD and Tenax® GC traps were returned to the
laboratory and were analyzed by thermal desorption/gas chromatography, using
a flame ionization detector, A pressurized gas sample collected within the
station also was analyzed using the same GC system for purposes of comparison
with results obtained at the sampling station. All exposure periods were of
2 hours duration. Additional data collected were air temperature, wind
velocity, relative humidity, and barometric pressure. The sampling rates for
the PSD's were corrected for temperature and wind velocity, and concentra-
tions were corrected to 25 C, 760 torr, and 0 percent relative humidity for
purposes of comparison of the data with results of the main body of the
study. The temperature correction for the PSD sampling rates included
corrections of the diffusion coefficients and the Tenax® GC retention
volumes.
46
-------
Calibrations for the three separate analytical facilities used on the
study were referenced to N.B.S. standards for benzene and tetrachloro-
ethylene, and to a 3CL prepared standard mixture of all components of
interest. Cross checks of analyses on the pressurized gas samples and on the
EPA audit cylinder showed generally good agreement between the different
analytical facilities.
A summary of the results of the PSD study is shown in Table 17. Data in
this table are grouped in the following format:
Canister Concentration
Tenaxa GC Trap Concentrations
PSD Concentrations
where the canister concentration refers to the pressurized gas sample taken
within the sampling station. Only six chemicals were considered; a number of
other chemicals were included in the overall intercomparison study, but these
included very volatile chemicals for which (a) the retention volumes were
exceeded with the external active samplers, and (b) the standard PSD's are
known to be inaccurate because of the low retention volumes.
In the last 5 columns of Table 17 the apparent precision and accuracy of
the PSD's are summarized. Precision here is defined as the reproducibility
within each group of 3 PSD's, and is calculated separately for the standard
and low rate devices for each chemical. Accuracy is defined relative to the
canister analyses, with the relative accuracy of the Tenax® GC trap samples
being shown also.
A cursory examination of the data in Table 17 reveals several apparent
inconsistencies. Except for the cases of 1,2-dichloroethane and toluene,
data for the low rate PSD's are generally much higher than corresponding data
for the standard PSD's, Tenax® GC traps and canister samples. This may be
due to the relatively short duration of the sampling period — the Low-rate
devices are designed to sample over periods of 8 to 24 hours. Because of
their low sampling rates, the accuracy of blank corrections is very critical
when they are used for short-term sampling. Indeed, for several of the
samples in Table 17 where "not-detected" is indicated, the chemicals were
actually detected, but at levels below the blank correction.
A second problem is evident in consideration of the Tenax® GC trap
results. The reproducibility of the traps was less than desirable in a num-
ber of cases. Finally, there appears to have been a consistent problem with
the GC analyses for tetrachloroethylene on the standard PSD's, with appreci-
ably more tetrachloroethylene being recovered than was present in the
canister and trap samples.
In general, the Tenax® GC trap results are of the order of 20-25 percent
lower than corresponding canister analyses. This is consistent with the
observations made on active sample and canister analyses made within the sam-
pling station. Results for the standard PSD's are, on the average, inter-
mediate between the trap and canister results. The standard deviations for
47
-------
TABLE 17. SUMMARY OF PSD/TENAX TRAP DATA: FIELD INTERCOMPARISON STUDY(a)
Concentration, ng/l.
Run Number
Chomi r.n 1
731
731L^<"
719
806
8061.
809
809L
814
814L
31.9
36.3
12.4
20.2
12.4
1 ,2-Dir.hloroe thane
2.76/25.3
34
.3/24.5/29.6
32.1/32.1
8.21/1.5
13.5/15.0/16.8
11.0/1.8
14/20/30
2.89/0.45
8.34/12.4/10.1
8.37/3.76
18.3/20.7/21.6
5.76/4.59/2.67
9.2/9.7/8.5
1.56/0.72/0.64
32.6
88.4
15.5
22.5
8.0
1,1,l-Trichloroethanc
2. 57/24 . 1
ND/ND/ND
78/81
9.02/1.7
ND/ND/ND
13.6/2.6
ND/19.3/7.6
2.53/0.32
17.2/24.6/12.4
26.9/14.3
124/134/13/
14.7/18.1/13
29.6/33.2/27.5
16/11/6.3
4.8
375
9.1
8
8.4
Bonr.rnr
4.39/6.05
49/55/45
376/394
5.93/2.92
22.5/13.7/30
6.3/4.3
15.5/30/165
1.69/0.53
5.3/6.9/5.4
2.65/ND
388/386/391
4.94/4.34/3.72
6.9/6.6/12.3
0.87/0.81/1.27
2.0
30
11.4
5.0
3.4
Trichloro^tliylcne
Nn/2.90
0.
72/0.14/0.14
15.6/14.7
10.2/5.96
14.2/16.4/13.5
2.1/0.3
ND/5.4/1. 7
0.91/0.74
4.6/1.0/1.0
ND/ND
26/22.6/26.2
7.7/ND/6.0
2.1/2.9/1.6
ND/ND/ND
14.5
22
20.4
16.6
10.3
Toluene
13.8/15.1
14
.4/14.1/14.7
20.6/20.9
14.3/14.8
14.2/24.1/22.6
13.5/14.6
23.9/22/20
9.45/5.51
9.1/11.9/12.7
19.3/17.7
23.9/24.4/26.8
24.6/ND/24.4
24/23/22
11.1/10.7/11.0
10.4
12
3.9
7.1
3. 73
Tetrachlorocthylene
3.10/5.32
23.4/ND/ND
12.2/12.2
1.2/1.8
NT)/ 1 J . 2/ND
5.56/4.65
7.1/ND/ND
3.90/2.43
11.4/14.3/9.8
13.3/12.5
17.5/4.4/36.7
22.8/17.2
19.4/19.5/22
13.8/25.5/19.9
-------
TABLE 17. (Continued)
Prec i 0ion.
, X<*>
Accuracy ^c ^
Standard Reduced Rate
Trap/
PSU/
LPSD/
Chemical
PSD's
PSD' a
Canister
Can ister
Cani ster
1,2-Dichlorocthane
27
14
.75/.15/5
.46/
. 10/9
.92/.18/10
1,1,1-Trichloroethane
21
(34 )
.74/.15/5
1.20/
.29/12
(1.6/1.1/5)
Benzene
20
(21 )
.82/.20/6
-75/
.26/9
(-72/.12/3)
Trieh loroeLhylene
20
(15)
. 72/.40/6
.67/
.17/7
(.72/.53/11)
Toluene
4
15
.85/.15/10
1 . 19/
.12/12
1.08/.17/10
Tetrachloroethylene
(19)
(19)
.80/.23/7
1-±L
.96/8
(3.0/.62/5)
Total
.77/.23/39
.91/.
33/48
(a) Format of data group ® Cani3ter/Trapa/PSD'a; Canister and trap results not repeated for reduced
rare PSD data group*.
(b) Average reproducibility with PSl) triplets (numbers in parentheaes indicate limited data group).
(c) Average ratioft of traps, PSU's and reduced-rate PSD's to canister results formatted a9:
average/standard deviation/number of points included.
(d) Reduced-rate groups denoted by "L".
-------
the PSD/canister ratios are somewhat larger than the precision calculated fo
the PSD's, but this is to be expected because of the additional analytical
uncertainty associated with the canister concentrations. The apparent preci
sion far both the standard and low-rate PSD's are consistent with those
observed in the laboratory evaluations of these devices (see Tables 4 and
14).
50
-------
REFERENCES
R. W. Coutant and D. R. Scott, "Applicability of Passive Dosimeters for
Ambient Air Monitoring of Toxic Organic Compounds", Environ. Sci.
Technol. 16 410-413 (1982).
R. W. Coutant, final report on Evaluation of Passive Monitors for
Volatile Organics to Environmental Monitoring Systems Laboratory,
U.S. Environmental Protection Agency, R.T.P., November, 1982
(EPA 600/4-83-014).
R. W. Coutant, R. G. Lewis, and J. D. Mulik, "Passive Sampling Devices
with Reversible Adsorption", Anal. Chem. 57^ 219-223, (1985).
R. G. Lewis, J. D. Mulik, R, W. Coutant, G. Wooten, and C. McMillin,
"A New Thennally-Desorbable Passive Sampling Device", Anal. Chem. 57,
214-219, (1985).
K. J. Krost, E. D. Pellizari, S. G. Walburn, and S. A. Hubbard, "Collec
tion and Analysis of Hazardous Organic Emissions", Anal. Chem., 54^, 810
817 (1982).
R. B. Bird, W. E. Stewart, and E. N. Lightfoot, Transport Phenomena,
John Wiley & Sons, Inc., New York, 1960, pp 145 and 607.
51
-------
APPENDIX A
MECHANICS OF PASSIVE DOSIMETER SAMPLING
WITH REVERSIBLE ADSORPTION
MECHANICS OF PASSIVE DOSIMETER SAMPLING
A general model of the mechanics of passive sampling with reversible
adsorption is given in reference 3. The following discussion summarizes the
simplified model that is appropriate for the EPA PSD.
Passive dosimeters in general rely on diffusion-limited sampling to
achieve their characteristic, compound selective "pumping" rates# It is nor-
mally assumed that the function of these devices can be expressed in terms of
Ficks first law, i.e.,
mi = Di A (H1-) (A-l)
where mj and are the mass flow rate and diffusion coefficients of the i-th
species and A is the sampling area, i.e., the face area, A0, times the
porosity, £.
(i1)- /* U-2>
s c
where C. and C. are the concentrations at the surface of the device and at
1 1
the surface of the collection medium, respectively. 2, is then the distance
from the surface of the device to the surface of the collection medium. C.
is effectively the equilibrium concentration of the i-th species over the
collection medium under a given condition of loading of the collection
medium.
When the collector is an activated carbon, the value of Cc is usually
taken to be zero because of the strong chemisorption that occurs with that
substrate. With a polymeric collector, however, Cc may not be effectively
zero (Cc £ 0.1 Cs). This is especially the case when the sorbent is chosen
to permit thermal desorption. In such cases, it is desirable that only
physical adsorption be involved in the collection process. With physical
adsorption, however, the adsorbed material is energetically not much
different from a condensed liquid phase and the equilibrium value of Cc may
52
-------
have a significant dependence on the amount of adsorbed material. If the
adsorption isotherm of a given species is represented by a plot of the weight
of adsorbate per gram of sorbent, Ws, versus the equilibrium gas phase
concentration, Cc, the slope of the curve at any point has the units of
volume. Walling, et al. (J. F. Walling, R. E. Berkley, D. H. Swanson, and
F. J. Toth, Sampling Air for Gaseous Chemicals Using Solid Adsorbents,
Application to Tenax, EPA 600/5-4-82-059, 1982) show that this volume
parameter is equivalent to the familiar GC retention volume, V^. That is,
Further, Walling notes that V^, can be considered constant at low surface
coverages and in the absence of interfering species.
The significance of Equation (A-3) is that we can expect Cc to increase
as the surface becomes loaded and that the rate of increase is inversely pro-
portional to the retention volume. In terms of passive monitor operation,
this means that the effective sampling rate will decrease at a rate that is
inversely proportional to the retention volume. For species that have large
retention volumes, the rate of sampling will be relatively insensitive to
loading, but we can expect the sampling rates of weakly bound species to
change with time. An estimate of the magnitude of this effect can be gained
from the following considerations.
In accord with Equation (A-l), the mass flow per unit weight of sorbent
into the device can be expressed as
where RQ is the intrinsic diffusion limited rate under irreversible condi-
tions, DA/5,. Part of the mass flow to the device will result in an increase
in surface loading and part of the mass will contribute to an increase in Cc,
1- • £ • I
dWs = V5dCc
(A-3)
(A-4)
£(dm) = dW5 ~ §£(dCc) .
(A-5)
Combining Equations (A-3), (A-4), and (A-5), we obtain
dCc
Rq/M
(A-6)
(Cs - Cc) Vb + &A/2W '
53
-------
For a sorbent/device combination that is of any practical significance, >>
&A/2W, we can simply neglect the latter term. Making the substitution k =
^o/^b, we can integrate Equation (A-6) between the Limits of Cc = 0 at t =
0, and Cc at t = t to obtain
= 1 = e_l<-c . (A-7)
Cs
Now the quantity 1 - Cc/Cs represents the effective relative sampling rate,
R/R0 at any instant in time. Inasmuch as the passive sampler is an integrat-
ing device, it is more appropriate to know the time weighted average relative
sampling rate, Rt.
" _ /0e"kcdt _ 1 - e"kc ,. „v
t " ft ux 1 (A"8>
/
dt
o
We see then that for small k's, i.e., for large retention volumes, the sam-
pling rate is inversely proportional to R0 and directly proportionaL to V^,
i.e., a reLativeLy constant sampling rate is favored by the combination of a
large retention volume, a property of the sorbent/sorbate combination, and a
relatively low intrinsic sampling rate, a property determined by the physical
design of the device.
For the sake of completeness, it is necessary to include one more effect
that is usually ignored in considerations of passive dosimeter mechanics —
the veLocity effect. Obviously, when a passive dosimeter removes molecules
of some species from the air space surrounding the device, the concentrations
of those species must be replenished by bulk movement of the air in order to
achieve a steady-state sampling condition. Whenever there is bulk flow over
a surface, there is drag between the gas stream and the stationary surface,
with the result that a relatively stagnant boundary layer develops near the
surface. The thickness of this boundary layer depends on the velocity and
the distance from the leading edge of the surface. (Surface shape and rough-
ness also affect the boundary layer development, but are ignored in this ele-
mentary treatment.) The significance of this boundary layer is that it does
present a small diffusional resistance that is additional to the intrinsic
resistance that is determined by the physical structure of the dosimeter
(noted above in terms of R0).
The veLocity effect for dosimeters previously has been treated employing
irreversible adsorption (2), and use a similar approach for the reversible
case:
54
-------
A-l.
Consider the simplified representation of a dosimeter given in Figure
Sorberit
J-, J.
Figure A-l. Boundary layer over badge surface,
Here the boundary layer thickness is represented by 5, and x represents the
position of the surface of the dosimeter. Under steady-state conditions, the
mass fluxes and J2 must be equal. Therefore,
DA0(C«i - Cx)/6 = DAQe(Cx - C0)/JL . (A-9)
Defining the quantity e5/S, as z, we then find that
Cx = (Coo + zC0)/(l+z) . (A-10 )
Inasmuch as Cx = Cs and C0 = Cc, as used in Equation A-4,
= Rn/W(l+z)
(cs - Cc) (A 6 >
where Equation 6' is now the equivalent of Equation A-6, but including the
contribution of the boundary layer. We can see then that this effect merely
amounts to reduction of R0 by a factor of l/(l+z), or
R/ = R0/(l + y~) . (A-11)
55
-------
Equation A-8 can then be corrected for the velocity effect,
R' = 1 ~ e~k>t. (A-8' )
t k't
In order to apply Equation A-8', the following relationship given by Bird,
Stewart and Lightfoot (6) is used:
5 = 4.64 (Hi) 1/2/i.026 sj/3 = 3.94 (H*)l/2 for air (A-12)
vp vp
where U is the viscosity, x is the mean distance from the leading edge, v is
velocity, p is the gas density, and Sc is the Schmidt number. (For an
average organic molecule in air at 25 C, Sc - 1.5).
The EPA Device
The diffusion barrier of the EPA device is a multilayer structure and in
determining the effective £ / &, the several layers must be treated as a series
of diffusional resistors. Thus,
(e/S,)"1 = I (ei/Jli)-1 . (A-13)
i
The structure and parameters for the EPA device are shown in Figure A-2.
(One face shown, normal use employs both faces.)
Referring to the dimensions in Figure A-2,
(e/jM-l = 2(—-—)~1 + 2( -379X.395 x-i 2( '-?5-)~l
.127 V .0107 ' . 0655
= 0.729
or £/1 = 1.37
and
A0£/5, = 9.70 cm for single sided sampling
or 19.40 cm for two sided sampling.
It should be noted that the open recess at the face of the device is included
in these calculations. If this recess is not included, A0e/1 = 11.75 cm for
single sided sampling.
56
-------
1_
Sorbent
Face
Fine Screens; *-i - 3.01C7 c
M ¦ 0.373
Open
Recess
Coarse Perforated Plates;
£
Figure A-2. Schematic representation of PSD diffusion barrier.
For the circular MRC dosimeter, x = ^ (diameter) = 0.96 cm,
and 6 = 1.55/vl (v in cm/sec). Therefore the veLocity correction factor is
f =1 + 1.37(1.55/vi/z)
or, @100 fpm, = 1 >1q.3o = °*76 9
and the effective sampling rate for two sided sampling is
R.' = D: 14.92 cm^/min. (A-14)
l 1
The rates calculated for some volatile organic compounds using Equa-
tion (A-14) are shown in Table A-l. Relative velocity effects, expressed as
Rv/Riqo are shown in Figure A-3.
57
-------
TABLE A-l. CALCULATED PSD SAMPLING RATES
(a)
(b)
(c)
(d)
Compound n
ui
, cm^/min
cc/min
Vb, L/g
, cc/min
t
Chloroform
5.33
79.5
18.9
59.0
1,1,1-Trichloroethane
4.76
71.0
11.8
46.8
Carbon Tetrachloride
4.97
74.2
16.5
53.9
Trichloroethylene
5.25
78.3
39.3
67.7
Tetrachloroethylene
4.97
74.2
154.0
71.5
Benzene
5.59
83.4
42.5
72.2
Chlorobenzene
4.48
66.8
372 .0
66.0
Acrylonitrile
6.35
94.7
0.3-7
23.0
1,2-Dicnloroethylene
5.51
B2.2
2-6
25.0
Trichlorotri fluoroethane
(4.16)
62.1
0.23-0.47
2.0
1,2-Dichloroethane
5.44
81.2
24.4
63.9
trans-1,3-Dichloropropene
4.76
71.0
335.0
69.9
Toluene
5.09
75.9
193
73.7
1,2-Dibromethane
(4.66)
69.5
183.0
67.6
o-Xylene
4.36
65.1
2800.0
64.9
a-Chlorotoluene
4.28
63.9
1984.0
63.7
Hexachlorobutadi ene
(3.3)
49.2
324.0
48.7
(a) Diffusion coefficients taken from Lugg (AnaL. Chem., 40, 1072 (1968))
except for estimated values given in parentheses.
(b) Calculated sampling rates at zero time and face velocity of 100 fpm.
(c) Retention volumes of Tenax GC taken from Pellizari, et al., (Anal. Chem.
54 810-817 (1982)) except where range of values is indicated.
(d) Time-weighted-average sampling rates (cm-Vmin) calculated for one hour
sampling times;
R = R0 [1kt 6 kt] > where k = V0.4Vb
58
-------
4
3
2
9
8
7
6
5
3
0
90 100
150
200
Velocity, fpm
Figure A-3. Calculated Effect of Velocity on MRC Badge Sampling Rate (e/J. = 1.37).
-------
Ic should be noted that the rates given in Table A-l are minimum rates
for the MRC device, i.e., they are based on the assumption that the open
recess at the face of the badge is part of the diffusion barrier. An upper
limit estimate of these rates would yield values about 20 percent higher.
The calculations shown in Figure A-3 imply that the sampLing rate
approaches zero as the velocity goes to zero. This is somewhat misleading
because the calculated rate represents only the steady-state rate. Under
perfectly quiescent sampling conditions, the diffusion layer would
continuously expand to infinity, taking an infinite time to do so. During
this time, the device would continue to sample in a transient mode at an ever
decreasing rate.
Normal air velocities are likely to range from about 10-20 fpm in a
passive indoor situation to 440 fpm (5 mph) in outdoor applications. The
expected variation in rates is thus 0.65 R^qq-1.15 Rioo- Under normal indoor
working conditions the expected rate is Rioo t 5 percent.
The effect of sampling time on the time weighted average sampling rate
can be seen in Figure A-4. In this figure Rc is plotted vs t/t, where
This approach allows the performance of the reversible sorption passive
dosimeter for different chemicals to be compared on the same graph once the
characteristic values of t are calculated for chemicals of interest. Also
indicated in Figure A-4 are the values of T calculated for single sided
sampling of the chemicals used in this program on the EPA PSD.
In principle, Figure A-4 could be used to correct sampling data taken
over various time periods to a common basis. If this type of correction is
deemed undesirable, however, it is likely that Rt values less than say 0.8
would be acceptable. Arbitrary selection of such a limit would determine the
alLowed sampling time for each chemical. For example, choice of Rc * 0.8
would allow a 39 hour sampling time for chlorobenzene but only 1 hour
sampling of 1,1,1-trichloroethane .
The most useful aspect of Figure A-4 may lie in its use for device
design. For any chemical having a known characteristic value of Vj,, Fig-
ure A-4 specifies the ratio of W/R0 that is needed to obtain relatively time
insensitive sampling.
Effect of Fluctuating Concentration
Although Equation A-8' correctly applies to sampling of relatively con-
stant concentrations of volatile organics (Cs constant), it does not apply
when Cs varies significantly during the sampling period. For example, if the
species being sampled is initially at say Cs for the first haLf of the sam-
pling period, and then drops to Cs=0 during the second half of the sampling
60
-------
R
c o.u
0.6
0.4
0.2
1 -e
t/x
ft/tT
u v,
* = k
(U=0.4q)
Compound
t, hr (single side sampling)
Chloroform 3.17
1,1,1-Trichloroethane 2.22
Carbon Tetrachloride 2.96
Trichlorethylene 6.69
Tetrachloroethylene 27.7
Benzene 6.79
Chlorobenzene 74.5
Acrylonitrile 0.7
1.1-D1chloroethylene 0.2
Trichlorotrifluoroethane 0.1
1.2-Dichloroethane 2.96
[rans-1,3-Dichloropropene 62.9
Toluene 33.9
1,2-Dibromoethane 35.1
o-Xylene 573
a-Chlorotoluene 414
Hexachlorobutadiene 102
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
t/,
2.0
2.2
2.4
Figure A-4. Effect of Time on Sampling Rate.
-------
time (e.g., when Che user moves from one building to anocher) some of the
chemical adsorbed during the first exposure condition will be lost because of
the reversible nature of the adsorption process. The rate of loss will not be
as great as the initial rate of adsorption however because of the fact that
the concentration gradient will be smaller ((Cc-0)/Jl) during the second
period. Indeed, the error induced by fluctuating concentrations can be
either positive or negative, depending on the way in which the exposure con-
dition changes. In principle, we could express Cs as a function of time and
numerically integrate Equation A-6, but this would require prior knowledge of
Cs for each sampling condition. A more reasonable approach is to consider
some specific examples and use the results to provide further guidelines to
use the PSD. Consider the following simple cases:
Case 1: Cs = 1 ppbv for t = 0-1 hr; Cs - 0 for' t = 1-2 hr
R0 = 81.2 cc/min; k = 0.676 hr"* (two sided sampling for 1,2-
dichloroethane).
During the first hour, the time-weighted average sampling rate is cor-
rectly given by Equation A-8' and is equal to 59.0 cc/min. The amount
adsorbed is therefore 60 m'in x 59 cc/min x 1 ppbv = 3.54 x 10"^ cc.
During the second hour, Equation A-6 reduces to
Cc/CC = exp-kt
o r
so that 50.9 percent of the adsorbed chemical will be lost during the
second hour. That is, 1.74 x 10"^ cc will remain at the end of the 2-hr
sampling period. If the user does not know that the concentration
change has occurred, he will assume that Che correct sampling rate is
Che 2-hr cime-weighted race (TWR) or 44.5 cc/min and he would calculaCe
the TWR concencracion as
1.74 x 10-6/(120 x 44.5) = 0.326 ppbv .
The Crue TWR for this sampling condition is 0.5 ppbv; a negative error
of 34.8 percent has occurred.
Case 2: Same concentration profile as Case 1; R2hr = ®.8 R0 = 70 cc/min
In this case, the Newton-Raphson method can be used Co solve Equa-
tion A-8* for k with the result that k = 0.232 hr~^. The amount sampled
during the first hour is Chen 3.75 x 10"^ cc. The amount lefc afcer Che
second hour is 2.97 x 10-^ cc. The user will assume rhaC Che badge sam-
pled ac a TWR of 56 cc/min and will calcuLate the TWR as 0.442 ppbv.
The true TWR is again 0.5 ppbv, so that a negative error of 11.6 percent
occurs.
62
-------
Case 3: Concentration profile reversed from Case 2, i.e., Cs = 0 for t = 0-1
hr; Cs = 1 ppbv for 1-2 hr; R0 and k remain as in Case 2
This time there is no material lost, but the device does not begin to
sample until the second hour, during which time it samples at the 1-hr
TWR. The amount adsorbed is therefore 3.75 x 10"^ cc. Again, the user
assumes that the 2-hr TWR is appropriate, and he calculates a TWR of
0.558 ppbv, or a positive error of 11.6 percent.
These three cases probably represent extreme exposure conditions; more
gradual changes with less total change in Cs would be more realistic, and we
would expect that the error would therefore generally be less than found in
these examples. However, the results serve to illustrate the effect of a
large value of k on the sample integration accuracy. The results also sug-
gest that the restriction of the TWR to 0.8 R0 may be a reasonable choice.
63
-------
APPENDIX B
CHROMATOGRAPH INLET SYSTEM
CHROMATOGRAPH INLET SYSTEM
As noted in the text of this report, analytical procedures used for the
PSD's were upgraded several times during the course of the 3 work assignments
covered by this report. One of the more significant changes involved the
development of a muLti-purpose chromatograph inlet system that incorporates
capabilities for: (1) PSD desorption, (2) direct sampling of external gas
streams, and (3) cryogenic preconcentration of gas samples. A schematic of
this inlet system is shown in Figure B-l. Briefly, this unit consists of the
following components:
1. A quick-disconnect inlet for PSD connection.
2. A gas sample inlet.
3. An outlet to a vacuum or pumping station.
4. An outlet to a capillary column.
5. A fixed volume loop for capture of gas samples.
6. A resistance wound cryogenic loop for trapping and release of PSD or
gas samples.
7. Two 6-port Carle valves for switching of flow directions and
injection of samples into the chromatograph.
The complete assembly, except for the cryogenic loop is contained within an
insulated temperature controlled box. When using the cryogenic sampling cap-
ability, the pumping outlet is connected to an evacuated fixed volume system
that is fitted with a Wallace and Tiernan gage for accurate determination of
sample volume.
64
-------
PSD
Input
Gas
Inlet
Nafion
Dryer
Carr
Puilo
Vent
Dry N2
SL
Thermos tat
Strip Heater
cryogenic
TraD
Figure B-l. General purpose GC inlet.
65
-------
TECHNICAL REPORT DATA
(Please read Instruction! on the reverse be/ore completing)
1 "epT/oT)?]/4-85/034 2
3 RECIPIENT S ACCESSION NO.
PB8 5 5L9bU8/iS
4. TITLE AND SUBTITLE
Evaluation of Passive Sampling Devices (PSD's)
6. REPORT DATE
April 1985
6. PERFORMING ORGANIZATION CODE
7. AUTHORtS)
Robert W. Coutant
B. PERFORMING ORGANIZATION REPORT NO.
S. PERFORMING ORGANIZATION NAME AND ADDRESS
Battelle-Columbus Laboratories
505 Ring Avenue
Columbus, Ohio 43201
10 PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO
68-02-3487
12. SPONSORING AGENCY NAME AND ADDRESS
US Environmental Protection Agency
Environmental Monitoring Systems Laboratory
Research Triangle Park, NC 27711
13. TYPE DF REPORT AND PERIOD COVERED
14 SPONSORING AGENCY CODE
EPA/600/08
15. SUPPLEMENTARY NOTES
16 ABSTRACT
The basic objectives of this study were to evaluate the performance of the EPA passive
sampling device (PSD) for sampling of ambient level volatile organic compounds
(VOC's); to develop an understanding of the mechanics of passive sampling using
reversible adsorption; and to apply this understanding to development of an improved
PSD that, is usable for sampling of VOC's over periods of 8 to 24 hours. Laboratory
and limited field evaluations of the standard and modified PSD's were conducted and
a model relating sorbent properties and device design to sampling rates was
developed. The results show the standard PSD'e to be useful for sampling of VOC's
having large retention volumes. Modified PSD's having greatly reduced sampling rates
show promise for sampling compounds having retention volumes as low as 5 to 10 L/g
over 8 to 24 hour sampling periods. The use of Spherocarb as an alternative sorbent
to Tenax GC also was investigated as a means for improving the performance of the PSD.
This sorbent was found to be unsuitable because of the high temperatures required for
desorption. It is recommended that the nodel which was developed be used for
developing sampling plans for specific applications, and that more extensive field
evaluation of the reduced-rate PSD's be conducted. This report is being submitted in
fulfillment of Contract No. 68—02—3487 by Battelle Columbus Laboratories under the
sponsorship of the US Environmental Protection Agency. It covers a period of April
15, 19B2 to October 31, 1984, and work was completed as of October 31, 1984.
17. KEY WORDS AND DOCUMENT ANALYSIS
a. DESCRIPTORS
b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Fit Id,'Group
18. DISTRIBUTION STATEMENT
RELEASE TO PUBLIC
IB SECURITY CLASS (This Report)
UNCLASSIFIED
21. NO. OF PAGES
74
20. SECURITY CLASS (This page)
UNCLASSIFIED
22. PRICE
CPA Form 2220-1 (Rav. 4-77) Pntvioua idition iiouoliti ^
------- |