Reprinted from Journal of the Air & Wast* Manag
EPA/600/J-93/463
olatJon, VW. 42, No. 10, Oetobtr 1982
Measurements of VOCs from the TAMS Network
Gary F. Evans and Thomas A. Lumpkln
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina
Deborah L Smith
Battelle
Columbus, Ohio
Matthew C. Somervllle
ManTech Environmental Technology, Inc.
Research Triangle Park, North Carolina
Target volatile organic compounds (VOCs) were measured at • network of urban air
monitoring locations In Boston, Chicago, Houston, and the Seatte/Taeoma area.
Following a pilot-scale field eval
n of available techniques for determining concen-
nique based on evacuated stainless steel canisters
Jratlons of VOCs In ambient air
i/as selected to collect whole air samples. Twenty-four-hour Integrand samples were
collected every twelfth day at ten sites over a 2-year study period. Battelle Columbus
Laboratory (BCL) analyzed the samples for 25 target VOCs using cryogenic focusing,
gas chromatographto separation and mass selective detection with flame tonbatton
detection as backup. Duplicate canister samplers were operated each sampling period
at one of the ten sites In the Toxic Air Monitoring System (TAMS) network to estimate
overall method precision. In addition, every 10th analysis was repeated by BCL to obtain
a measure of analytical precision. -Finally, each sampling period a dean evacuated
canister was sent to Research Triangle Park to be filled with an audit gas mixture of
known concentrations. The audit canisters were analyzed along with the routine field
samples to estimate method accuracy. The target compounds found most ubiquitously
were benzene, toluene, xylene and ethylbenzene. These aromatic compounds were
highly correlated and proportionally related In a manner suggesting that the primary
contributors were mobile sources In an the urban locations studied. Annual median
concentrations for target compounds ranged from 0.1 to 4.0 part per bmion volume
(ppbv) range, while Individual 24-h concentrations occasionally reached as high as 20
ppbv.
In June 1984, a document entitled
"Strategy for Monitoring Ambient Air
Toxic Pollutants,"1 was issued by the
U.S. Environmental Protection Agency
(EPA) proposing a multi-year research
and development program in the area
of toxic air pollutant assessment. As
part of the overall strategy, the docu-
ment called for the development and
operation of a pilot-scale monitoring
network consisting of one air toxics
sampling site in each of three cities in
the first year, with subsequent expan-
sion to 10-15 sites possible in future
Implications
Title III of the 1990 Clean Air Act Amendment! require! that ambient monitoring
be conducted in urban areas for a broad range of hazardous air polhitantt,
including volatile organic compounds. This paper describes an approach to
sampling and analysis for VOCs which was used successfully by U.S. EPA In a
network application over a two-year period. The data quality achieved is
summarized, along with a discussion of observations from ten urban monitoring
locations. The monitoring system, based on whole air collection in a stainless steel
canister, generally performed well in the field and produced data of high quality.
October 1992
Volume 42, No. 10
REPRODUCED 8Y:
U.S.Dtptitnwnt of Comnwro*
National Ttchnteat Information Sarvte*
Springfield. Virginia 22161
years. EPA's Atmospheric Research
and Exposure Assessment Laboratory
in Research Triangle Park, North
Carolina (AREAL/RTP) was given the
responsibility for implementing this
pilot program, known as the Toxic Air
Monitoring System (TAMS) network.
The objectives established for this pro-
gram were to evaluate methods of
sample collection and analysis for toxic
air pollutants, begin to characterize
ambient concentrations in selected ur-
ban atmospheres, and transfer moni-
toring technology and results to re-
gional offices, state and local agencies.
Experimental Methods
The TAMS monitoring program was
initiated in 1985 with the establish-
ment of one air monitoring station
each in Boston, Chicago and Houston.
These three cities were selected for
this program by the Office of Air Qual-
ity Planning and Standards (OAQPS)
from a List of urban areas with popula-
tions greaterthan one million persons.
An effort was made to include in the
study a diversity of emission and cli-
mate types, while not duplicating air
toxics research already underway by
other programs. The selection of spe-
cific monitoring locations within each
city was made in consultation with
OAQPS and the appropriate regional,
state and local authorities. Siting crite-
ria included population density, area
emission types and rates, site accessi-
bility, electric power availability, and
site security. In most cases, an existing
state or local agency monitoring sta-
tion was found to meet the require-
ments of the TAMS program. For each
site, a local person was recruit d, hired,
and trained specifically for tin opera-
Copgnithtl992-Alr*WuttM«oi«niwntA*ni
-------
tion of the TAMS air toxics sampling
equipment.
The initial sampling methodology
employed for the collection of VOCo at
the TAMS sites consisted of a distrib-
uted air volume (DAY) sampling sys-
tem (four air volumes and a field blank
per sampling period) with Tenax-GC
as the solid adsorbent. Twenty-four-
hour integrated samples were col-
lected every sixth day. Identification
and quantitation of target compounds
was accomplished at AREAL./RTP by
using thermal desorption of the sam-
ple followed by gas chromatographic
separation and mass spectrometric
(GC/MS) analysis for 96 target com-
pounds. Sample validation was based
on the degree of consistency in blank-
corrected concentration across the four
samples for each compound.
In August 1986, a methods compari-
son study was begun at a second site
established in the Houston area. Hous-
ton Site #2 is situated in suburban
Deer Park, about midway between the
Houston Ship Channel and the Bay-
port industrial complex. For a 10-
month study period, two polished stain-
less steel canister samples were
collected concurrently with the DAV-
Tenax samples and returned to RTF
for analysis. A comparative analysis of
the results2 demonstrated that while
the two methods yielded substantially
equivalent mean concentrations,
method precision was superior for the
canister collection technique. In light
of these results, it was decided in
August 1987 to replace the Tenax-
based samplers at existing TAMS sites
with canister-based systems and to
deploy the canister-based systems at
new TAMS sites called for in fiscal
year 1988 (FY-88) program plans. Also,
because of the projected increase in
sample load, a decision was made to
reduce the sampling frequency to ev-
ery twelfth day and to seek a contract
laboratory for provision of clean canis-
ters and analytical services to the ex-
panded network.
Reid Operations
During FY-88, the TAMS network
was expanded to include a total often
urban air monitoring stations. Three
sites were located in each of the origi-
nal cities: Boston, Chicago and Hous-
ton. In each city, the first two sites
were selected to represent the primary
industrial sectors of the urban area,
and the third site was located in a
more residential area for purposes of
contrast. Midway through the fiscal
year, a single station became opera-
tional in the Seattle/Tacoma metropol-
itan area. This site was located in an
industrial complex on land controlled
by the Port of Tacoma. Descriptive
information for each TAMS site ap-
pears in Table I.
The canister samplers deployed in
the network were similar in design to
the K-type samplers built by EPA for
use in the Kanawha Valley Study,3 two
of which were used in the TAMS meth-
ods comparison study. Prior to deploy-
ment, the samplers were cleaned and
certified according to procedures which
have since been codified in Method
TO-14 of EPA's Compendium of Meth-
ods.4 It was decided to operate the
TAMS samplers in a passive (i.e., vacu-
um-filling) mode to eliminate the pump
as a potential source of sample contam-
ination, mechanical failure, or air leaks.
Every twelfth day, a whole-air sam-
ple was collected at each TAMS site in
a clean, evacuated canister with initial
vacuum equal to -30 in. Hg. Sam-
pling commenced at 6 a.m. and contin-
ued for a 24-h period. Flow of air into
the canister was regulated by a mass
flow controller set to a rate of ~3.5
cc/min. A certified mass flow meter
was used to verify correct sample flow
before and after each sample collection
period. If properly followed, this proce-
dure resulted in a sample under a
slight vacuum (~2-3 in. Hg.) at the
conclusion of the 24-h sampling cycle.
Beginning and ending sampling times,
canister pressures, and flow rates were
recorded by the field operator.
Laboratory Operations
All canister samples collected at the
TAMS sites in FY-88 and FY-89 were
prepared and analyzed by Battelle Co-
lumbus Laboratory (BCD. The cryo-
genic GC/MSD system used to analyze
these samples was identical to the
system that had been used at AREAL/
Table I. Location of TAMS monitoring stations.
City
No.
Location
Classification
Code
Boston
Chicago
Houston
Tacoma
1
2
3
!•
2
3
1
2
3
1
1 Summit Dr
115 Southampton St
Orchard & Common St
727 E. lllthSt
7800 W 65th St
O'Hare Internationl
1262 Mae Drive
1050 W Pasadena Dr
4510 Aldine Mail Rd
2301 Alexander Ave
Industrial
Industrial
Residential
Industrial
Industrial
Airport Property
Industrial
Industrial
Residential
Port/Industrial
BOSl
BOS2
BOSS
CHI1
CHI2
eras
HOU1
HOU2
HOU3
SEAT
RTP for processing earlier TAMS sam-
ples. Canisters were cleaned with a
vacuum and oven system4 and sent to
the field sites in time for each sam-
pling date. Returned canisters were
checked in and the final pressure was
recorded. Sample analysis was achieved
with a Hewlett-Packard (HP) model
5880 A (level 4) GC equipped with a
flame ionization and a mass selective
detection system (HP 5970). A 50-m
by 0.32-mm-i.d., HP-1 fused-silica col-
umn was used to resolve the target
compounds. The column exit flow was
split by a low dead-volume tee (Alltech,
Inc.). One-third of the flow was di-
rected to the mass selective detector
(MSD); the remaining flow passed
through the flame ionization detector
(FID).
A Perma Pure dryer (model MD-125-
48F) with a tubular hygroscopic ion-
exchange membrane (Nafion) was used
to remove water vapor selectively from
the sampled gas stream. Use of the
drier permitted collection of larger vol-
umes of sample air (350 cc), thereby
lowering the limit of detection. The
MSD was operated in the selective ion
monitoring (SIM) mode of operation
for canister analyses. In this mode, the
MSD monitored only preselected ions,
rather than scanning all masses contin-
uously between two mass limits. As a
result, increased sensitivity and im-
proved quantitative analysis were
achieved. For the canister samples,
two characteristic ions were moni-
tored for each target compound and
quantitation was based upon peak ar-
eas. The minimum quantitation limit
(MQL) used for all compounds was 0.1
part per billion by volume (ppbv).
Quality Assurance
In the field, duplicate canister sam-
plers were operated each sampling pe-
riod at Houston Site #2 to obtain
information for estimating the overall
precision of the measurement system.
In the laboratory, one canister was
reanalyzed each sample period to pro-
vide information on analytical preci-
sion. During the second year of the
program, a clean evacuated canister
was sent each sampling period to EPA
to be filled with a mixture of known
NlST-traceable concentrations of
VOCs from an audit gas cylinder. The
canister was then returned to BCL for
analysis with the routine field sam-
ples. These data were collected to ob-
tain estimates of analytical bias.
Results and Discussion
Results by TAMS site ranged from
73.9 to 97.5 percent data complete-
ness. These values were computed as
the percentage of scheduled sampling
periods for which valid aerometric re-
1320
J. Air Waste Manage. Assoc.
-------
suits were actually obtained. Reasons
for missing samples included mechani-
cal or personnel problems in the field,
missed shipment dates and laboratory
malfunction. Overall, the results were
judged to be acceptable in light of the
original data quality objective (DQO)
for data completeness (i.e., a 75 per-
cent).
In Figure 1, the percentage of TAMS
data reported as discriminant values
(i.e., >0.10 ppbv) is shown by com-
pound averaged across all sites. Al-
though the target compound list was
limited to VOCs which might be ex-
pected in urban air, many of the tar-
geted compounds were often not found
in measurable quantities at several of
the TAMS monitoring sites. Further
quantitative analysis was therefore
limited to compounds found at concen-
trations above the MQL at least 75
percent of the time. Thirteen of the 26
target compounds met this criterion,
ten having 95 percent or more discrim-
inant values (see Figure 1).
Data Quality
The replicate values provided by BCL
for every tenth sample were used to
calculate analytical precision estimates
for the 13 compounds of interest. Also,
the results from the duplicate sam-
plers operated at Houston site #2 were
used to estimate overall method preci-
sion for each compound. These esti-
mates were computed in each case as
the pooled standard deviation ex-
pressed as a percentage of the overall
mean for the paired data sets. Concen-
tration values below the MQL were set
to one-half that level for computation.
The resulting coefficients of variation
(%CVs) for replicates and duplicates
appear in Figure 2.
The 13 compounds of interest fall
into three classifications, as shown in
Figure 2: freons, aromatics, and chlori-
nated alkanes. Although the freons
exhibit relatively high %CVs for dupli-
cate samples, the %CVs for replicate
analyses are quite comparable with
those for other compounds. This sug-
gests that problems may have been
encountered in obtaining representa-
tive ambient air samples for the fre-
ons. For example, the duplicate sam-
plers were located on the roof of an
air-conditioned trailer from which fre-
ons may have been emitted. Also, freon-
113 may have been used as a solvent to
clean components in the mass flow
controllers. Because of the relatively
high %CV for duplicates, freon data
were excluded from further analyses.
The other compounds exhibited over-
all method precision in accordance with
the project DQOs (i.e., £ ±25 per-
cent). For carbon tetrachloride, the
%CV for replicate analyses exceeded
the %CV for duplicate samples. This
anomaly probably is due to the consis-
tently low concentrations of carbon
tetrachloride reported (i.e., just slightly
above the quantitation limit).
A comparison of the analytical re-
sults from BCL with the known or
spike values from EPA for audit gas
samples showed that analytical bias
was generally within ±25 percent for
the compounds in the mixture. An
explanation was found for each occa-
sion when excursions beyond these
DQO bounds did occur. Appropriate
corrective measures were then insti-
tuted, such as obtaining a calibration
gas of improved purity for one com-
pound and correcting the humidifica-
tion process for audit sample prepara-
tion at one point in the program.
Monitoring Data
The median and maximum 24-h
VOC concentration values for the two-
year study period are shown by site in
Table II. For the aromatic compounds,
concentration levels and their relation-
ships are very similar from site to site
and even across cities. The median
concentrations for chlorinated alkanes
are comparatively low (i.e., less than
1.0 ppbv in most cases). However, max-
imum 24-h values for the solvents
Compound
Benzene
Benzyl Chloride
Carbon Tetrachloride
CNorobenzene
3-Chtoropropene
12-Otehloroetriane
Dichloromethane
Ethyl Benzene
4-Ethyl Toluene
Freon-11
Freon-12
Freon-113
Methyl Chloride
Styrene
Tetrachtoroethene
Toluene
124-Trichlorobenzene
111-Trlchloroethane
TrJchloroethene
Trichloromethane
124-Trimethylbenzene
135-Trimethylbenzene
'Vinyl Chtorkte
rmp-Xylene
o-Xylene
Figure 1. Percent values above quantitation limit.
October 1992 Volume 42, No. 10
20
30 40 50 60 70 80
Percent Values > 0.10 ppbv
90 100 110
1321
-------
Compound
Freon-11
Freon-12
Freon-113
Benzene
Toluene
m+p-Xytene
o-Xylene
Ethylbenzene
124-Trimethylbenzene
Methyl Chloride
Dichloromethane
Carbon Tetrachtoride
111-Trichloroethane
20
30
%CV(REPs)
Co6ffictonts of vsrifltion toe rvpBcctes And dupOcAtu.
40 50 60
%CV(DUPs)
70
80
90 100
dichloromethane and 1,1,1-trichloro-
ethane did occasionally approach or
even exceed 20 ppbv at industrial sites
in Houston and Tacoma.
Review of data plots suggested that
some strong associations existed both
between and within sites. To explore
these relationships further, Pearson
product-moment correlations5 were
computed between sites within each
city. The highest inter-site correla-
tions were found to exist between Bos-
ton sites 1 and 2 and.Houston sites 1
and 3 for dichloromethane and the
aromatic compounds (r > 0.60 in all
cases). The lowest spatial correlations
(r < 0.33) were observed for carbon
tetrachloride. This compound exhib-
ited little concentration range at most
of the sites, remaining very close to the
global background level of 0.12 ppbv.
Friedman's test5 (a nonparametric pro-
cedure) was computed by city to deter-
mine whether consistent differences
existed between sites within each city.
Generally, in the cases where signifi-
cant differences were found, the more
industrialized areas (sites #1 and #2)
exhibited higher concentrations than
the corresponding residential area (site
#3).
Correlations were also computed be-
tween compounds measured at each
TAMS site to estimate the degree of
linear relationship between pollutants
at the same location. This procedure
revealed that o-xylene, m+p-xylene,
and ethylbenzene were highly (r >
0.90) and mutually correlated at every
TAMS site. Also, at most sites, ben-
zene and toluene were positively corre-
• lated with one another and with o-xy-
Table n. 24-H VOC median and maximum by rite (ppbv).
Compound
Benzene
Toluene
m + p-Xylene
o-Xylene
Ethylbenzene
Methyl Chloride
Dichloromethane
Carbon Tetrachloride
1 1 1-Trichloroethane
124-Tri-Methy! Benzene
Stat
Med
Max
Med
Max
Med
Max
Med
Max
Med
Max
Med
Max
Med
Max
Med
Max
Med
Max
Med
Max
BOS1
0.87
2.73
2.07
5.60
0.92
3.03
0.37
0.93
0.31
0.81
0.54
1.18
0.29
3.10
0.12
0.16
0.58
3.34
0.35
1.02
BOS2
1.15
1.95
3.01
5.48
1.42
2.96
0.53
1.12
0.44
0.84
057
1.01
0.39
3.89
0.13
0.17
0.71
2.72
0.59
1.43
BOS3
0.70
3.18
1.85
11.8
0.73
3.15
0.31
1.13
0.23
0.92
0.58
1.02
0.42
1.80
0.13
0.21
0.68
2.61
0.35
1.33
cmi
1.03
4.25
3.36
15.2
1.36
7.79
0.43
1.64
0.49
2.71
0.57
0.82
0.42
3.97
0.12
0.44
0.59
3.89
0.41
1.32
CH12
1.17
7.28
2.60
12.8
1.05
4.70
0.35
1.62
0.35
1.49
0.56
4.76
0.37
3.21
0.12
0.15
0.68
4.50
0.37
1.86
CHI3
0.86
2.02
1.83
9.49
0.81
2.41
0.40
1.16
0.26
0.81
0.58
1.34
0.38
3.40
0.13
0.19
0.66
1.84
0.36
0.87
HOU1
1.82
8.65
3.00
18.6
1.51
7.83
0.54
2.78
0.50
2.41
0.79
2.09
0.34
20.6
0.16
0.88
0.53
17.4
0.63
3.20
HOU2
1.09
6.51
1.77
9.86
0.87
3.24
0.34
1.01
0.29
1.06
0.72
2.42
0.31
5.71
0.13
1.14
0.41
7.08
0.37
1.19
HOU3
1.34
8.96
3.16
19.4
1.44
9.00
0.55
3.31
0.47
2.46
0.73
2.23
0.31
3.29
0.12
0.34
0.79
16.9
0.62
4.48
SEAT
0.95
8.21
2.22
16.3
1.16
8.33
0.44
3.00
0.37
2.30
0.56
2.99
0.35
14.1
0.12
0.17
0.53
13.7
0.46
2.82
1322
J. Air Waste Manage. Assoc.
-------
Table HI. Ratios of toluene: benzene: xylene: ethylbenzene.
City
Site
Boston Chicago Houston Tacoma
1
2
3
7:3:4:1
7:3:4:1
8:3:5:1
7:2:4:1
7:3:4:1
7:3:5:1
6:4:4:1
6:4:4:1
7:3:4:1
6:3:4:1
lene, m+p-xylene, and ethylbenzene.
These strong linear as relations sug-
gest that a common source category is
predominantly responsible for the con-
centrations of aromatic compounds ob-
served at all the TAMS monitoring
locations. The ratios of the median
concentrations for aromatic com-
pounds at the TAMS sites are shown
in Table III. These ratios, expressed as
toluene: benzene: total xylene: ethyl-
benzene, are quite similar from site to
site and closely approximate published
ratios among these compounds for mo-
tor vehicle emissions (7:3:5:1).6 Clearly,
mobile sources are most likely respon-
sible for the major portion of aromatic
VOC concentrations observed at these
sites. Comparison of the TAMS data-
base with other results reported and
summarized in the literature''"10 show
that the VOC concentrations observed
at the TAMS locations are typical of
the atmospheric levels of these com-
pounds generally found in U.S. urban
areas.
Conclusions
EPA canister method TO-14 proved
fieldworthy in an urban network appli-
cation over a two-year period of opera-
tion. Coupled with GC/MSD analysis,
it produced data of acceptable quality
in terms of completeness (90 percent
overall), precision (within ±25%), and
accuracy (within ± 25%) for most VOCs
of interest. The 24-h ambient levels of
targeted VOCs ranged from less than
0.1 ppbv to slightly greater than 20
ppbv. Measured concentrations were
comparable in all urban areas moni-
tored, although the more heavily indus-
trialized areas tended to exhibit higher
concentrations than corresponding res-
idential areas. Toluene was the com-
pound generally found at highest con-
centrations, with individual 24-h
concentrations occasionally approach-
ing 20 ppbv. Mobile sources were most
likely the dominant contributors of
the aromatic compounds (i.e., ben-
zene, toluene, xylene, and ethylben-
zene) observed at all sites in each of
the four cities. Annual median concen-
trations for these compounds ranged
from 0.2 to 4.0 ppbv. Annual median
concentrations of chlorinated aliphatic
compounds were typically less than 1.0
ppbv, although individual 24-h concen-
trations ranged to 20 ppbv suggesting
the need for finer time-scale resolution
for applications where peak VOC con-
centrations are of interest.
Disclaimer
This paper has been reviewed in ac-
cordance with the U. S. Environmental
Protection Agency's peer and adminis-
trative review policies and approved
for presentation and publication. Men-
tion of trade names or commercial
products does not constitute endorse-
ment or recommendation for use.
References
1. Strategy far Monitoring Ambient Air
Toxic Pollutants, U.S. Environmental
Protection Agency, Office of Air and
Radiation, Research Triangle Park, NC
(1984).
2. "FY-87 Annual Report on the Opera-
tion and Findings of TAMS," U.S.
Environmental Protection Agency, Of-
fice of Research and Development, Re-
search Triangle Park, NC (1987).
3. "Report on Air Monitoring in the
Kanawha Valley, West Virginia," U.S.
Environmental Protection Agency, Of-
fice of Research and Development, Re-
search Triangle Park, NC (1987).
4. "The Determination of Volatile Or-
ganic Compounds (VOCs) in Ambient
Air Using Summa* Passivated Canis-
ter Sampung and Gas Chromatographic
Analysis," in Compendium MethodTO-
14, U.S. Environmental Protection
Agency Office of Research and Develop-
ment, Research Triangle Park, NC
(1988).
5. Cpnover, W. J. Practical Nonparamet-
ric Statistics, John Wiley and Sons,
New York, NY. (1980) p. 251 and pp.
299-308.
6. Scheff, P. A.; Wadden, R. A.; Bates, B.
A.; Aronian, P. F. "Source fingerprints
for receptor modeling," J. Air Waste
Manage. Assoc. 39:469 (1989).
7. Shah, J. J.; Singh, H. B. "Distribution
of volatile organic chemicals in outdoor
and indoor air," Environ Sci Technol.
22:1381 (1988).
8. National Volatile Organic Compounds
Database Update, U.S. Environmental
Protection Agency, Office of Air and
Radiation, Research Triangle Park, NC
(1988).
9. Edgerton, S. A; Holdren, M. W., Smith,
D. L; Shah, J.J. "Inter-urban compar-
ison of ambient volatile organic concen-
trations in U.S. cities," J. Air Waste
Assoc. 39:729(1989).
10. Michael, L. C.; Pellazari, E. D.; Perritt,
R. L.; Hartwell, T. D.; Westerdahl, D.;
Nelson, W. C., "Comparison of indoor,
backyard, and centralized air monitor-
ing strategies for assessing personal
exposure to volatile organic com-
pounds," Environ. Sci. Technol. 24:
996, (1990).
G. F. Evans and T. A. Lumpkin
are senior environmental engineers
with the U.S. EPA's Atmospheric
Research and Exposure Assessment
Laboratory in Research Triangle
Park, NC 27711. D. Smith is a re-
searcher with Battelle Memorial In-
stitute in Columbus, OH 43201. M.
C. Somerville is a senior scientist
with ManTech Environmental Tech-
nology, Inc. in Research Triangle
Park, NC 27709. This manuscript
was submitted for peer r- -new on
December 23, 1991. The revised
manuscript was received on May 28,
1992.
October 1992
Volume 42, No. 10
1323
------- |