EP A/600/A-97/069
Field Evaluation of EPA Proposed Method 0040
(Sampling and Analysis of Volatile Organic Compounds Using Tedlar® Bags)
James F. McGaughey, Joan T. Bursey, sod Raymond G. Merrill
Eastern Research Group, Inc.
Morrisville, North Carolina 27560
Robert G. Fuerst and Merrill D. Jackson
National Exposure Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
A field test has been performed to evaluate EPA Proposed Method 0040 (Sampling of Principal
Organic Hazardous Constituents from Combustion Sources Using Tedlar® Bags), a method designed for
the collection of volatile organic compounds present in a combustion source at concentrations above the
range of EPA Method 0030 (VOST). Proposed Method 0040 is based on the results of laboratory
studies to develop and refine a sampling train and methodology to collect and analyze volatile organic
compounds present in source emissions between 100 and 1000 ng/ra3, In the laboratory studies, a design
for the sampling train was developed, the train was evaluated using dynamic spiking techniques, and a
standard sampling method in SW-846 format was prepared. An analytical methodology using gas
chromatography/mass spectrometry (GC/MS) was also developed to provide qualitative identification of
specific compounds and quantitative results based on calibration for individual compounds.
The original design of the sampling train was modified by removing a Teflon® check valve at the
Tedlar® bag inlet to resolve discrepancies in volumes measured by the train. As a result, leak-check
procedures in the proposed method were simplified. A field test of the modified methodology was
performed to determine and document the systematic error (bias) and random error (precision) of the
method under stationary source sampling conditions. Four similar trains were operated simultaneously
using a quadruplicate sampling probe, with dynamic spiking of two of the four quad trains with specific
volatile organic compounds, while simultaneously sampling emissions from a coal combustion source.
Analytical data were statistically evaluated according to the procedures of EPA Method 301. Fifteen of
eighteen volatile organic test compounds met Method 301 acceptance criteria for performance while
using the Proposed Method 0040.
INTRODUCTION
A field test using the statistical experimental design outlined in EPA Method 3011 to evaluate a
method for sampling and analyzing high-level [Le., parts per million (ppm) in a stationary source] volatile
organic compounds from stationary source emissions using EPA Proposed Method 00402 has been
completed. Proposed Method 0040 was designed to collect volatile organic compounds (VOCs) at
concentrations that are above the concentration range that can be collected by EPA Method 003 03 (the
Volatile Organic Sampling Train, VOST).
The original sampling train used in this field study was developed by ERG staff under contract to
the U. S. Environmental Protection Agency. This method for collecting and analyzing VOCs at
concentrations between 100 and 1000 ng/m1 in stationary source emissions was developed and refined in
the laboratory. The laboratory development work involved the design of the sampling train, laboratory
evaluation of the train using gaseous dynamic spiking techniques, and preparation of a standard sampling
method in SW-846 format. The analytical portion of the method is based on gas chromatography/mass
spectrometry (GC/MS) and provides qualitative identification of VOCs for characterization of Products
of Incomplete Combustion (PICs) as well as quantitative emissions results based on calibration for
individual compounds. Additional laboratory studies were performed to address the question of accurate

-------
and reproducible measurement of the volume collected in the Tedlar® bag. In the original laboratory
experiments, the volume of air collected by evacuating a nominal 20 liters of air from the rigid container
was determined before and after the removal of the Teflon® check valve assembly located just prior to
the bag inlet valve. Significant deviation from the theoretical 20 liters collected was encountered with the
check valve in place. Removal of the check valve resulted in actual volumes ranging from -1.0 to 3.5
percent relative deviation from the theoretical amount. Therefore, Method 0040 was modified prior to the
field test. The Teflon® check valve was replaced with a glass/Teflon® stopcock. Additional
thermocouples were also added inside the rigid container to monitor the temperature of the sampled gas.
The field evaluation of the modified method was performed to determine and document the
systematic error (bias) and random error (precision) of the method under stationary source sampling
conditions.
EXPERIMENTAL APPROACH
Proposed Method 0040 was evaluated for sampling and analysis of the compounds listed in Table
1. EPA Method 301 was used for the statistical design of the sampling strategy and for the statistical
evaluation of the results obtained from dynamic spiking of two of four collocated trains. The field
evaluation was conducted at a coal-fired power plant. An 800-megawatt unit with an electrostatic
precipitator but no caustic scrubber was selected for testing. Since the coal contained 1% sulfur, S02
was present in the emissions, as well as particulate matter and NOx. The stack temperature was
approximately 132°C (270°F).
Unit 2 at this test site had been characterized previously for VOST sampling. Several non-
halogenated VOCs had been detected in the VOST samples, but at levels typically found in ambient air
and 1% or less of the proposed dynamic spiking levels for Proposed Method 0040.
Field Sampling
The sampling train used in the field evaluation study is shown in Figure 1. A quad probe was
used with four similar sampling trains. The quad probe consists of four similar heated sampling probes
that can be inserted in the stack as ok unit. This configuration allows the simultaneous collection of
stack gas in four similar trains, with gaseous dynamic spiking in two of the trains.
Prior to shipment to the field test site, all glass components of the sampling train were cleaned
according to the procedures described in Proposed Method 0040, wrapped in aluminum foil, and
segregated to prevent contamination. Tedlar® bags were cleaned and blanked according to the
procedures of Proposed Method 0040; 10% of the bags were analyzed by GC/MS to verify that
appropriate cleaning criteria had been met.
All Tedlar® bags and rigid containers were leak-checked prior to transport to the test site. The
rigid containers were evacuated to a vacuum of approximately 25" Hg and allowed to stand for
30 minutes while a vacuum gauge was monitored. All containers were leak-free. Each container and the
corresponding Ed were uniquely numbered so these two components were always used as a unit. In
addition, each Tedlar® bag was filled almost to capacity with clean nitrogen, sealed, and allowed to stand
overnight. Each bag was visually inspected for leaks. The nitrogen in 10% of these bags was analyzed to
demonstrate the absence of VOCs that were to be dynamically spiked. After Irak-checks, each bag was
uniquely numbered, attached to the Quick-Connect® fitting on a rigid container lid, re-evacuated, and
sealed in the corresponding rigid container. Containers and bags were then transported to the test site.
Dry gas meters were leak-checked and calibrated. Other preparations included leak-checking the
sampling umbilicals, temperature readout calibration, and checking the operation of other associated
sampling equipment.
Sampling trains were assembled in the field laboratory without Tedlar® bags and associated rigid
containers. The bags in containers were transported separately to the sampling location and installed in
the sampling train after the probe was positioned in a single stack port.

-------
Each heated train component was heated to 130°C, and each sampling train was leak-checked in
the field according to the simplified leak-check procedures. Samples were collected by evacuating the
rigid container of the Tedlar® bag, withdrawing stack gas through the quad probe. The front end of the
quad probe was positioned in the center of the stack and remained in that location during each day of
testing. Traversing was not required for this method evaluation study since the true concentration of the
components of the stack gas was of no interest to this program. The sampling flow rate always exceeded
the dynamic spiking flow rate by at least a factor of 2 to ensure that all of the spiked VOCs were carried
totally into the sampling trains.
The sampling flow rate was set to a nominal 0.33 Lpm and a sample was collected for one hour
simultaneously in each of the sampling trains. At the end of one hour, the rigid container was isolated
from the rest of the train and the trains were leak-checked. Each rigid container was opened to visually
mspect the Tedlar® bag to determine if it had been filled to approximately 80% of capacity, indicating the
sample had been collected. Since no condensate was observed in any of the trains during any of the
sampling runs, no condensate was collected. Samples were shipped daily by ground transport to the
laboratory to ensure that the Tedlar® bag samples remained above 0°C and that the samples could be
analyzed within 72 hours of sampling. Two quad runs were scheduled and performed each day of
sampling to ensure that laboratory analytical capacity would not be exceeded. On three of the four days
of sampling, the rigid containers were heated to maintain the temperature above 0°C.
In addition to the eleven quad sampling runs performed for method evaluation, four field blanks
(one for each of the sampling trains) were taken. The field blanks were transported and analyzed with the
stack gas samples. Two Tedlar® tegs, filled with high purity nitrogen and labeled as laboratory blanks,
were left in the laboratory under the same storage conditions as the field samples.
Dynamic Spiking Procedures
The compounds dynamically spiked in the field (Table 1) were contained in a commercially
prepared and certified compressed gas cylinder. The concentration of each VOC in the cylinder was
verified by GC/MS before field use. During each quad sampling ran, spiking gas was continuously
introduced into two of the four sampling trains through two fine metering valves. The flow rate of the
spiking gas into the sampling trains was nominally 80 mL/min to introduce approximately 20 ppm of each
VOC over the sampling period of one hour. Each gas metering system was equilibrated for
approximately 30 minutes before the start of sampling. The gaseous spike was introduced into each train
at a point immediately after the probe and before the filter and condenser. The regulator and tubing
leading to each sampling train were maintained at a temperature of 130-140°C.
ANALYSIS
Gaseous samples in Tedlar® bags were analyzed by GC/MS, using an injection loop to inject a
constant volume of sample into the GC. Analytes were cryofocused and then introduced onto the head of
a fused-silica DB-1 (0.32 mm ID, 60 m length, 1 |i film thickness) capillary column.
Quantitative calculations were based on the injection of a 5 mL sample from the Tedlar® bag, at a
nominal concentration of200-600 ng on column on the basis of field-spiked values. Appropriate
dilutions from the spiking cylinder were used to prepare calibration standards. All standards were
prepared in Tedlar® bags and stored at ambient temperatures. A 5-mL gaseous aliquot of the internal
standards (bromochloromethane, l-bromo-4-fluorobenzene, chlorobenzene-dj, and 1,4-difluorobenzene)
was cryofocused along with each sample prior to introduction onto the capillary column. The response
factors were verified daily. Tedlar® bag field samples were stored at laboratory ambient temperature and
analyzed within 72 hours of sampling to meet the method-specified hold time. All field samples were
analyzed and the results were used in the statistical calculations.
Repeated analysis of several samples verified the stability of the VOCs over time in a source
matrix. Percent change from the original analysis was determined by comparing area counts of the
subsequent analyses to the original analysis. The percent difference for every compound on each of the

-------
subsequent analysis days is the same within experimental error, indicating that the compounds were stable
in the Tedlar®) bags for at least 10 days following the initial analysis.
RESULTS AND DISCUSSION
The theoretical concentration of each analyte in the spiked trains was calculated by determining
the amount of dilution of the volume of gas spiked into each Tedlar® bag. The volume of spiked gas was
calculated by multiplying the average of the spiking gas flow rate (values from pre- and post-sample
collection) in mL/min by the length of the spiking period (nominally 60 min). This value was then divided
into the sum of the dry gas meter volume and the spiking volume to determine a dilution factor. This
dilution factor (nominally a value of 5) was then divided into the concentration of each analyte contained
in the spiking gas cylinder (nominalfy 100 ppm).
The percent recovery for each VOC was calculated by dividing the analyzed value by the
theoretical value and multiplying by 100. Recoveries for each quad run are shown in Table 1, with mean
and standard deviation. Using a criterion of acceptable recoveries from 50-150%, all compounds except
bromomethane fell in the acceptable range. Using a precision criterion of percent relative standard
deviation less than 50, all compounds except 1,3-butadiene and dichlorodifluoromethane performed
successfully.
The results of the Method 301 statistical calculations are shown in Table 2. Using the acceptance
criteria of EPA Method 301, all the compounds met acceptance criteria except bromomethane,
1,3-butadiene, and dichlorodifluoromethane.
CONCLUSIONS
The Proposed Method 0040 sampling train performed successfully in its modified configuration.
Modification of the method leak-check procedures was successful. These modifications should therefore
be made a part of the Method. The total volume of sample collected in the Tedlar® bags of the
dynamically-spiked sampling trains was determined by summing the volume read by the dry gas meter and
the volume of spike gas added. This measurement approach was confirmed by calculating the percent
recovery of deuterated analogs of several of the VOCs that were injected as a liquid spike into the bags
during the collection of the samples.
REFERENCES
1.	EPA Method 301. Protocol for the Field Validation of Emission Concentrations from Stationary Sources.
U. S. Environmental Protection Agency. EPA 450/4-90-0015. April, 1991.
2.	Proposed Third Update to SW-846 Manual, Published for Public Comment, Federal Register July,
1995.
3.	"Test Methods for Evaluating Solid Waste, Physical/Chemical Methods, SW-846 Manual, 3rd
ed." Document No. 955-001-0000001. Available from Superintendent of Documents, U. S.
Government Printing Office, Washington, D.C. November, 1986.
DISCLAIMER
The information to this document has been funded wholly by the United States Environmental
Protection Agency under EPA Contract Number 68-D4-0022 to Eastern Research Group, Inc. It has
been subjected to Agency review and approved for publication. Mention of trade names or commercial
products does not constitute endorsement or recommendation for use.

-------
Table 1. Summary of spiked compound recoveries.
Compound
Mean"
%RSD
1,1,1 -Trichloroethane
92.9
23.9
1,1,2-Trichloroethane
94.5
21.4
1,1-Dichloroethane
93.7
21.8
1,1 -Dichloroethene
92.8
24.1
2,2,4-Trimethylpentane
105
22.8
Allyl chloride
82.0
25.6
Benzene
98.0
24.9
Bromo methane
168
31.6
1,3-Butadiene
52.9
56.9
Carbon tetrachloride
101
21.6
Chloromethane
123
22.9
Dichlorodifluoromethane
51.1
60.9
n-Hexane
94.0
20.5
Methylene chloride
93.4
25.8
Toluene
84.7
29.8
Trichlorofluoromethane
121
24.4
Vinyl chloride
109
25.3
Vinyl bromide
112
26.4
* Average of 22 samples

-------
Table 2. Summary of Method 301 statistical calculations.
¦ ii ii aBgaMaaagagaaaaeaag
Compound
Bias Significant
•
Correction
Factor
Precision,
%RSD
1,1,1 -Trichloroethane
No

15.0
1,1,2-Trichloroethane
No

16.6
1,1-Dichloroethane
Yes
1.12
15.0
1,1 -Dichloroethene
Yes
1.14
17.1
2,2,4-T rimethylpentane
No
1
13.2
Allyl chloride
Yes
1.29
18.0
Benzene
No
a
14.9
Bromomethaoe
Yes
0.66"
17.1
1,3-Butadiene
Yes
1.81*
41.5
Carbon tetrachloride
No
—#
14.6
Chloromethane
Yes
0.83
23.3
Dichlorodifluoromethane
Yes
2.14b
70.8C
n-Hexane
Yes
1.11
13.9
Methylene chloride
Yes
1.12
23.2
Toluene
Yes
1.22
29.8
Trichlorofluoromethane
Yes
0.84
14.0
Vinyl chloride
No
ai i
17.6
Vinyl bromide
No
		a
20.6
•Calculation of a Correction Factor is not required if bias is not significant,
bCorrection factor outside of Method 301 criteria for an acceptable method.
°%RSD outside of Method 301 criteria for an acceptable method.

-------
.SficaGd
ProbetootaBcnVfrfce .
ClwooaTap
0*
FtarHoMar
Ntt>FIter)
<>=®
Pttd Tim
Tata? Union
• -1
AirTchl Conlainar*^
ChvcodTytp
Quick Corrector
/^=oo^
To
WOST Control
Conaoto
Out*
Comectore
Miction Port
toSefc
Figure 1. Proposed Method 0040 train configuration.

-------
TECHNICAL REPORT DATA
1. REPORT HO. 2.
EPA/600/A-97/069
3.RI
4. TITLE AND SUBTITLE
Field Evaluation of EPA Proposed Method 0040
(Sampling and Analysis of Volatile Organic Compounds
Using Tedlar® Bags)
5.REPORT DATE
6.PERFORMING ORGANIZATION CODE
7. A0THOR
-------