United States
Environmental Protection
Agency
Environmental Monitoring
Systems Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S4-85/067 Jan. 1986
Project Summary
Comparison of Ambient Air
Sampling Techniques for
Volatile Organic Compounds
M. W. Holdren, D. L Smith, and R. N. Smith
The objective of this study was to
carry out a comparison of ambient air
sampling techniques. A series of four-
teen experimental sampling runs were
carried out at a field site adjacent to
Battelle's chemistry laboratory. Ambi-
ent air was drawn through a sampling
manifold and was continuously spiked
with a mixture of fifteen volatile or-
ganic compounds (VOCs) to give con-
centrations 1 to 3 ng/l above back-
ground air. These compounds were
chloroethene, 1,1-dichloroethene,
dichloromethane, 3-chloropropene,
1,1,2-trichloro-1,2,2-trifluoroethane,
trichloromethane, 1,2-dichloroethane,
1,1,1-trichloroethane, benzene, tetra-
chloromethane, trichloroethene,
toluene, tetrachloroethene, chloroben-
zene, and 1,2-dimethylbenzene.
During the sampling period passi-
vated stainless steel canisters were uti-
lized to collect whole air integrated
samples upstream and downstream of
the spiking region. An automated gas
chromatographic system was em-
ployed to analyze the contents of the
sample canisters using capillary
column separation and multiple detec-
tors for sample analysis (electron cap-
ture, flame ionization and mass selec-
tive detectors). Tenax GC adsorbent
samples were also collected down-
stream of the spiking region in parallel
with the integrated canister samples.
These samples were analyzed by stand-
ard gas chromatographic/mass spectra-
metric techniques.
In comparing analytical results,
whole air collection via canisters gave
better precision for the compounds of
interest in this study. An estimate of
precision (% standard deviation) for the
canister sampling method ranged from
4 to 10 percent (using the mass selec-
tive detector). For Tenax, precision val-
ues ranged from 8 to 16 percent. When
comparing measured recoveries with
expected concentrations, the canister
sampling approach yielded values from
89 to 120 percent. Recoveries, using
Tenax adsorbent, ranged from 8 to 104
percent.
This Project Summary was devel-
oped by EPA's Environmental Monitor-
ing Systems Laboratory, Research Tri-
angle Park, NC, to announce key
findings of the research project that is
fully documented in a separate report
of the same title (see Project Report
ordering information at back).
Introduction
The Methods Development Branch of
• the Environmental Monitoring Systems
Laboratory (EMSL) is responsible for
the development and evaluation of
state-of-the-art and emerging analytical
techniques for the determination of or-
ganic compounds in ambient air. Re-
cently, a priority listing of volatile or-
ganics has been established and the
EMSL is focusing on further develop-
ment of analytical methodology associ-
ated with the detection of these com-
pounds. Primary emphasis has been
placed on developing field-compatible
analytical systems.
During the past two years a joint ef-
fort by Battelle's Columbus Laborato-
ries and EPA has resulted in the devel-
opment and evaluation of a prototype
system for the analyses of sixteen
volatile organic compounds (VOCs).
-------
The system consists of a reduced-
temperature trap for condensing organ-
ics from ambient air, and a capillary
column gas chromatograph (GC) with
flame ionization (FID) and electron cap-
ture detectors (ECD) to quantitatively
monitor the VOCs. Software develop-
ment using the basic programming ca-
pability of the GC system permits auto-
matic sampling and analysis to be
achieved with minimal operator inter-
facing. The prototype system has been
tested with respect to sample drying
procedures (selective removal of water
vapor), co-collection of reactive ambi-
ent air species, and collection and re-
lease efficiency. The prototype design
has also been evaluated by comparing
two nominally identical systems using
calibration mixtures and ambient air.
Excellent results were obtained during
these laboratory tests and the following
recommendations were made:
• The automated GC system should
be field tested. During the field
tests the reduced temperature trap-
ping system should be compared
with other preconcentration tech-
niques such as solid adsorbents
and passive dosimeters.
• Many of the target compounds
tested in the laboratory also co-
elute with other ambient air spe-
cies. Although the combination of
capillary column and flame ioniza-
tion and electron capture detectors
alleviate some of the identification
and quantitation concerns, other
more selective detectors are
needed. Integrating a mass selec-
tive detector (MSD) into the auto-
mated gas chromatographic sys-
tem is recommended. The greater
specificity of this detector over
other detection systems will allow
better differentiation of co-eluting
GC peaks and thus improve quanti-
tative capability.
The full report focuses on the com-
parison of two ambient air sampling
techniques. Specifically, whole air col-
lection into specially treated canisters
was compared with Tenax GC adsor-
bent sampling techniques. Analysis of
the canisters' contents was accom-
plished with the automated GC system,
while Tenax adsorbed samples were
analyzed by standard gas chromato-
graphic mass spectrometric techniques.
The comparison study was carried out
at a site adjacent to Battelle's chemistry
laboratory. To assure detectable con-
centrations, the sampling manifold was
spiked with a mixture of fifteen VOCs
and the comparison focused on these
compounds.
Procedure
Tenax and whole air sampling (using
passivated stainless steel canisters)
were operated in parallel employing the
sampling manifold and apparatus
shown in Figure 1. Fourteen test runs
were carried out at the field site. Eleven
sampling runs were made with spiked
ambient air while the remaining three
runs involved the injection of calibra-
tion mixtures. Each test run lasted two
hours.
Canister samples were analyzed with
an automated capillary column gas
chromatograph equipped with electron
capture, flame ionization and mass se-
lective detectors. This instrument was
also programmed to collect and analyze
"real time" integrated samples directly
from the manifold.
Five, ten (duplicate), and twenty liter
samples were collected using Tenax ad-
sorbent sampling techniques. These
samples were analyzed by gas chro-
matography/mass spectrometry.
Results and Discussion
Overview
The comparison study was carried
out at a field site adjacent to Battelle's
chemistry laboratory. The evaluation
study consisted of fourteen sampling
runs as listed in Table 1. Runs 4 and 11
were clean air experiments using dilu-
tion mixtures from an audit cylinder
supplied by EPA, while run 5 was a sim-
ilar run using Battelle's calibration cylin-
der. The remaining sampling runs were
made with spiked ambient air. The ana-
lytical precision of the three detectors
from the automated GC system was ob-
tained by replicate analysis of canisters
collected from the sampling manifold
during the tests. A comparison was also
made of concentrations of the six target
compounds that were detected by both
the electron capture and mass selective
detector. Whole air canister concentra-
tions determined by the cryogenic GC
system (using the mass selective detec-
tor response) were also compared with
those values obtained from Tenax GC
adsorbent devices.
Analytical Precision of MSD,
ECD, and FID Detectors
An estimate of the analytical preci-
sion of the cryogenic GC system was
determined from triplicate analyses of
spiked and background canisters col-
lected during each sample period. For
these analyses, twelve compounds
were detected with the FID; six com-
pounds were detected with the ECD.
Due to the limitation in software only six
compounds were monitored with the
MSD.
A summary of the analytical precision
of all three detectors is shown in
Table 2. In viewing the spiked canister
data set, an average analytical precision
of ±7.3 percent was found for the six
compounds analyzed with the MSD. Te-
trachloroethene had the lowest value
(±4.0%) while (Freon-113) displayed the
highest relative standard deviation
(RSD) value (±10.1%). A significantly
lower average RSD value of ±3.7 per-
cent was obtained for the same six com-
pounds analyzed with the ECD. The RSD
values ranged from ±1.3 percent (tetra-
chloromethane) to ±10.1 percent
(Freon-113). With the FID detector an av-
erage RSD value (±7.7%) similar to the
MSD was found. However when the six
additional compounds that were also
detected by the FID were included in the
calculations, the average RSD value
was lowered to ±5.8 percent. Toluene
exhibited the lowest RSD value (±1.8%)
while tetrachloromethane produced the
highest RSD value (±14.4%).
The background canister data, also
shown in Table 2, follow the same trend
as the spiked canister data. An average
RSD value of ±7.2 percent was found
for the six compounds analyzed with
the MSD, while an average RSD value of
±3.6 percent was obtained with the
ECD. With the FID detector, the average
RSD value was ±8.5 percent. When the
five additional FID detected compounds
were included in the FID calculations,
the average RSD value was lowered to
±7.0 percent (chloroethene was not
found in background air i.e. <0.1 ng/l).
Comparison of ECD and MSD
Responses
The six compounds that were
detected by both the electron cap-
ture and mass selective detectors
were 1,1,2-trich I o r o -1 ,2,2-
trifluoroethane, trichloromethane,
1,1,1-trich lo roetha ne, tetra-
chloromethane, trichloroethene and
tetrachloroethene. A comparison of
concentrations was carried out by de-
termining average percent relative dif-
ference values between the two detec-
tors i.e., [100 (ECD response-MSD
response)/average response]. Three
data sets were examined and include
the spiked canisters, the background
-------
Exhaust
\
Variable Speed
Pump and
Dry Test Meter
Solid
Adsorbent
Sampling
System
Calibration Cyl.
Audit Cyl.
Figure 1. Diagram of sampling manifold and instrumentation employed during field program.
Table 1. List of Sampling Runs
Run No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Sampling date
7/12/84
7/17/84
7/19/84
7/20/84
7/24/84
7/26/84
7/27/84
7/30/84
7/31/84
8/02/84
8/03/84
8/06/84
8/07/84
8/09/84
Description
Ambient air (trial run)
Ambient air
Ambient air
Audit cylinder
Calibration cylinder
Ambient air
Ambient air
Ambient air
Ambient air
Ambient air
Audit cylinder
Ambient air
Ambient air
Ambient air
canisters, and samples collected in real
time.
A summary of the average percent
relative difference values from the three
data sets is shown in Table 3. For 1,1,2-
trichloro-1,2,2-trifluoroethane, negative
average percent relative difference val-
ues were observed for all three data sets
and indicated that the MSD recorded
concentrations were in general higher
than the corresponding ECD values. On
the other hand, trichloromethane, tetra-
chloromethane and trichloroethene
gave positive average percent relative
difference values for all three data sets.
The positive values denoted higher rela-
-------
Table 2. Analytical Precision Data From Replicate Analyses of Spiked and Background Canisters
Spiked canisters
detector precision <±%)
Background canisters
detector precision (±%>
Compound
Chloroethene
1, 1,2-Trichloro- 1,2,2-trifluoroethane
Trichloromethane
1,2-Dichloroethane
1, 1, 1-Trichloroethane
Benzene
Tetrachloromethane
Trichloroethene
Toluene
Tetrachloroethene
Chlorobenzene
1,2-Dimethylbenzene
MSD
NM(t»
10.1
7.3
NM
8.7
NM
6.4
7.3
NM
4.0
NM
NM
ECD
NM
10.1
4.3
NM
2.0
NM
1.3
2.8
NM
1.7
NM
NM
FID
7.4
10.0
3.3
3.6
3.6
1.6
14.4
11.9
1.8
2.7
2.9
6.7
MSD
NM
10.4
12.4
NM
4.5
NM
4.9
8.6
NM
2.3
NM
NM
ECD
NM
3.9
9.2
NM
2.0
NM
1.7
2.7
NM
1.8
NM
NM
FID
NM
18.7
1.6
1.1
2.9
3.6
15.1
7.2
1.7
5.2
15.2
4.9
NM - not detected.
Table 3. Average Percent Relative Difference Values Obtained By Comparing Electron Cap-
ture and Mass Selective Detector Responses During the Field Study1"1
Average relative difference values (%)
Compound
1, 1,2-Trichloro- 1,2,2-trifluoroethane
Trichloromethane
1, 1, 1-Trichloroethane
Tetrachloromethane
Trichloroethene
Tetrachloroethene
spiked
canisters
-11.8
9.2
3.2
23.2
19.6
-3.6
(ECD response -
background
canisters
-8.8
15.0
-6.0
0.5
36.9
5.7
MSDresponse\ ,„
real time
samples
-12.4
2.2
-1.9
23.4
24.1
-5.7
live ECD responses for these three com-
pounds. Tetrachloroethene and 1,1,1-
trichloroethane showed no consistent
bias. The spiked canister and real time
sampling data showed very similar per-
cent relative difference values as one
would expect since both types of sam-
pling were carried out downstream of
the sample spiking port. Although the
same general trend is also observed
with the background canister data, the
somewhat larger variation resulted be-
cause these samples were collected up-
stream of the spiking port and therefore
generally contained much lower
concentrations of the six target
compounds.
Tenax GC Adsorbent Versus
Whole Air Collection/Cryogenic
Preconcentration Comparison
Study for Ambient Air
In order to compare the data from the
two collection methods, the appropriate
Tenax GS concentration values were ra-
tioed to the corresponding concentra-
tions found by cryogenic trapping/GC
analysis of canister samples. The ratios
provided a direct measure of the rela-
4
average response /
tive recovery of the Tenax GC adsorbent
for each compound. The comparison
was done on this basis because the can-
ister/cryogenic trapping concentrations
generally agreed better with the diluted
concentration generated in the mani-
fold (see full report). Likewise literature
data has shown that several of the fif-
teen target compounds have low break-
through volumes on Tenax GC and are
therefore not efficiently collected.
In Table 4 a summary of the perform-
ance data for Tenax GC relative to canis-
ter/cryogenic trapping is given. Each
compound is listed along with its break-
through volume and the average Tenax
GC recovery relative to canister/cryo-
genic trapping obtained over the ten
sampling runs. As indicated in the table
for those compounds in which break-
through volumes were very low, the
nominally 5 liter Tenax GC adsorbent
sample was used in the comparisons;
the remaining concentration values
were taken from the 10 liter samples.
The values used for the canister sam-
ples were obtained using the mass se-
lective detector, because this detection
system was less subject to potential in-
terferences. However, the toluene and
1,2-dimethylbenzene values were ob-
tained using FID due to limitations on
the number of ions which could be
monitored using the mass selective de-
tector.
For the compounds, chloroethene
and 1,1-dichloroethene, no meaningful
data was obtained with the Tenax GC
adsorbent. With the exception of the
anomalous behavior of 1,1,2-trichloro-
1,2,2-trifluoroethane, the compounds
dichloromethane through trichloro-
ethene gave very reasonable relative
recoveries ranging from 83 to 130 per-
cent. However a dramatic fall off in
recovery is observed for the less vola-
tile compounds, toluene, tetra-
chloroethene, chlorobenzene and 1,2-
dimethylbenzene. The reason for the
relatively low recoveries for these
higher boiling compounds is unclear.
Conclusions
The significant findings from this
study are presented below:
(1) Replicate analysis of canisters
collected upstream and down-
stream of the spiking region with
the automated GC system pro-
vided precision data for each de-
tector. For concentrations varying
from 1 to 70 ng/l an average BSD
of ±3.7 percent was found for the
six compounds detected by the
electron capture detector (1,1,2-
trichloro-1,2,2-trifluoroethane,
trichlo rometha ne, 1,1,1-
trichloroethane, tetra-
chloromethane, trichloroethene,
and tetrachloroethene). An RSD
of ±7.3 percent was obtained for
the same six compounds ana-
lyzed with the mass selective de-
tector. A similar value of ±7.7 per-
-------
Table 4. Performance Data for Tenax Relative to Cryogenic Trappin'g/GC Analysis of Canis-
ter Samples
AverageTenax GC
recovery relative to
Compound volume'31, liters/cartridge cryogenic trapping, %
Tenax GC breakthrough
volume1"1, liters/cartridge
Chloroethene
1, 1-Dichloroethene
Dichloromethane
3-Chloropropenelcl
1,1,2-Trichloro-1,2,
2-trifluoroethane
Trichloromethane<0>
1,2-Dichloroethanelc>
1, 1, 1-Trichloroethane
Benzene
Tetrachloromethanelc>
Trichloroethene
Toluene
Tetrachloroethene
Chlorobenzene
1,2-Dimethylbenzene
0.8
Not Given
4
6
Not Given
13
18
9
27
13
28
122
106
249
334
(b)
Ibl
83 (21)
87 (35)
39 (25)
WO (36)
100 (15)
130 (42)
100 (18)
110(37)
1 12 (26)
70 (19)
88 (27)
78 (35)
55 (21)
Data from reference 5 in the full report at 90°F.
Low volume (nominally 5 liters) Tenax GC value used for these compounds. Medium volume
(nominally 10 liters) Tenax GC value used for all other compounds.
Value in parentheses is standard deviation for all sampling runs (excluding audit and calibra-
tion cylinder sampling runs).
cent was found with the flame
ionization detector.
(2) The six compounds that were de-
tected by both the electron cap-
ture detector and mass selective
detector were compared using
percent relative difference values
[100 (ECD response-MSD re-
sponsel/average response]. Rela-
tive difference values ranged
from -11.8 percent (1,1,2-
trichloro-1,2,2-trifluoroethane) to
36.9 percent (trichloroethene).
(3) Tenax GC adsorbent samples also
were collected in parallel with the
integrated canister samples and
then were analyzed by standard
gas chromatographic mass spec-
trometric techniques. Analysis of
duplicate Tenax samples (10 L) re-
sulted in RSD values that ranged
from 8 to 16 percent. Recovery rel-
ative to canister sampling was as
follows: chloroethene and 1,1-
dichloroethene (no meaningful
data), dichloromethane (83%), 3-
chloropropene (87%), 1,1,2-
trichloro-1,2,2-trifluoroethane
(39%), trichloromethane (100%),
1,2-dichloroethane (100%), 1,1,1-
trichloroethane (130%), benzene
(110%), tetrachloromethane
(110%), trichloroethene (112%),
toluene (70%), tetrachloroethene
(88%), chlorobenzene (78%) and
1,2-dimethylbenzene (55%).
Recommendations
Additional studies should be under-
taken under controlled conditions with
analytical uncertainties minimized as
much as possible by employing the
same analysis procedure for all col-
lected samples. Techniques to be com-
pared should include distributive air
volume sampling with Tenax adsor-
bent, passive sampling using personal
exposure devices, and whole air collec-
tion in canisters.
M. W. Holdren, D. L. Smith, and R. N. Smith are with Battelle's Columbus
Laboratories. Columbus, OH 43201.
W. M. McClenny is the EPA Project Officer (see below).
The complete report, entitled "Comparison of Ambient Air Sampling Techniques
for Volatile Organic Compounds," (Order No. PB 86-120 953/AS; Cost: $11.95,
subject to change) will be available only from:
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
Environmental Monitoring Systems Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
U. S. GOVERNMENT PRINTING OFFICE: 1986/646-116/20759
-------
United States
Environmental Protection
Agency
Official Business
Penalty for Private Use $300
EPA/600/S4-85/067
Center for Environmental Research
Information
Cincinnati OH 45268
0 i.
0000329
U S
PS
PROTECTION AGENCY
------- |