United States
Environmental Protection
agency
Atmospheric Research and Exposure
Assessment Laboratory
Research Triangle Park, NC 27711
Research and Development
EPA/600/S3-90/009 May 1990
&EPA Project Summary
Laboratory and Field
Evaluations of Methodology
for Measuring Emissions of
Chlorinated Solvents from
Stationary Sources
Anna C. Carver, William G. DeWees, and Easter A. Coppedge
An evaluation of the use of EPA
Method 18 for sampling and analysis
of chlorinated solvents was needed
to support possible future regulations
by the U. S. Environmental Protection
Agency. The solvents specifically
addressed are: carbon tetrachloride
(CCI4), chloroform (CHCI3),
perchloroethylene (PERC), and
trichloroethylene (TCE). Laboratory
and field studies were performed to
evaluate sampling and analytical
procedures for measuring these
solvents from stationary sources.
Conclusions and recommendations
are made regarding the application of
Method 18 to organic solvent
sampling and analysis.
This Project Summary was
developed by EPA's Atmospheric
Research and Exposure Assessment
Laboratory, Research Triangle 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
Several chlorinated solvents are being
evaluated by the U. S. Environmental
Protection Agency (EPA) for possible
future regulations for emissions from
stationary sources. Entropy Environ-
mentalists, Inc. (Entropy) was
commissioned by the Quality Assurance
Division (QAD) of EPA's Atmospheric
Research and Exposure Assessment
Laboratory (AREAL) to evaluate methods
for sampling and analysis of some of the
high priority solvents. Those solvents
included in the evaluation were carbon
tetrachloride (CCI4) chloroform (CHCI3),
perchloroethylene (PERC), and
trichloroethylene (TCE). The work was
conducted in two phases: (1) a field
evaluation of sampling and analytical
techniques for measuring emissions of
PERC, arid (2) laboratory and field
evaluations of the collection of all four
solvents under high temperature/high
moisture conditions.
The use of EPA Method 181 was
previously evaluated as a sampling and
analysis method for PERC emissions in a
laboratory study sponsored by the EPA.2
Since perchloroethylene is widely
employed as a solvent in the chemical
degreasing industry, a degreasing facility
was selected for field evaluation of the
candidate method. Gaseous samples
collected in Tedlar bags were analyzed
directly using a gas chromatograph
equipped with a flame ionization detector
(GC/FID). Emission samples were also
adsorbed onto activated charcoal in
sorbent tubes and likewise analyzed by
GC/FID after desorption with an
appropriate solvent.
Laboratory evaluation of EPA Method
18 techniques modified to include an in-
-------
line gas conditioning system for high
temperature/high moisture sample
collection involved (1) the identification of
adsorption media that will effectively
collect organic solvents under these
conditions and (2) the effectiveness of the
gas conditioning system in permitting
sample collection in Tedlar bags. A field
evaluation of the sorbents chosen and the
gas conditioning system developed was
conducted at a gas-fired boiler where the
exhaust gas stream was spiked with the
chlorinated solvents. Samples collected
both in Tedlar bags and on the
adsorption tubes were analyzed using
GC/FID.
Experimental Procedures
All samples were analyzed using a
Hewlett-Packard 5890A series gas
chromatograph with dual flame ionization
detectors. One injection port was
equipped for liquid injections while the
other was equipped with a manually
operated gas sampling valve. Column
selection and operating procedures were
based on those followed in previous
evaluations.3 For all analyses, the
hydrogen and air pressures supplied to
the dual FID's were maintained at 22 and
20 psi, respectively. These values were
selected by optimizing the FID response
according to manufacturer's suggestions.
For the GC/FID analysis, calibration
standards were prepared, in bags with
nitrogen or in the appropriate desorbing
solution, using measured amounts of
each solvent. These were injected into
the GC to obtain a linear calibration curve
in the range of the samples to be
analyzed.
Dual quadruplicate-train sampling runs
were conducted at the inlet and outlet to
a carbon bed adsorber at the degreasing
facility according to EPA Method 18.
Gaseous samples were collected
simultaneously in four Tedlar bags and
four charcoal adsorption tubes. Double-
seamed Tedlar bags were fabricated for
the collection of gaseous samples as
specified, and each bag was leak-
checked before and after use in the field.
Adsorption tubes were packed with
charcoal and checked for migration of
PERC from the primary to the backup
portion. The sampling flow rate for
collection of the Tedlar bag samples was
maintained using adjustable needle
valves at a rate of 55 - 60 cms/min, while
the charcoal tube samples were collected
at a flow rate of 0.08 L/min. also using
critical orifices. The gaseous Tedlar bag
samples were injected directly into the
GC/FID and quantified according to
Method 18. The adsorbed charcoal tube
samples were desorbed using a fixed
volume of methylene chloride to allow
GC/FID analysis using liquid injection of
the resulting solutions.
Prior to the second field evaluation,
seven commercially available
sorbents(Bio-Rad's Bio BeadsR SM-2 and
SM-4; Supelco's Amberlite XAD-8«
CarbotrapR and Carboxen; Water's C18-
OctadecylR and Union Carbide's TRI-X-
100R) were evaluated for collecting the
solvents of interest under simulated high
temperature/high moisture conditions. All
sorbents were pretreated according to
manufacturer's instructions. Two-gram
quantities of each were packed into
sampling tubes and subjected to
retention and breakthrough (time at which
sorbent is no longer retaining one or
more of the solvents) tests. Sorbent tubes
were challenged with a mixture of CCI4,
CHCI3, PERC, and TCE each at 100 ppm
from a Tedlar bag at an approximate flow
of 1.0 L/min, for a 1-hour period, using a
diaphragm pump. The gas exiting the
tubes was introduced directly into the
GC/FID.
High temperature/high moisture stack
gas containing the chlorinated solvents
was simulated by drawing the same
mixture at the same flow for the same
time period through a heated impinger
containing 400 ml of water, This gas was
then drawn through the candidate in-line
gas conditioning system condenser for
cooling and a primary sorbent tube,
followed by a knockout jar for condensate
collection and a second (backup) sorbent
tube. The pressure differential of the
system was monitored, as was the gas
flow rate. The resulting sample was
introduced directly into the GC/FID.
The two sorbents with acceptable
retention and breakthrough chara-
cteristics under the simulated high
temperature/high moisture conditions,
were examined for efficiency of solvent
removal for analysis (desorption
efficiency). The desorption efficiency was
determined for three different desorbing
solutions: acetonitrile, methanol, and
methylene chloride. Multiple tubes of
each sorbent were charged with 1.0 11!
each of the chlorinated solvents and set
aside for 1 h. Using a separate tube for
each sorbent/desorbing solution
combination, they were desorbed with 4
ml of the appropriate solution which was
then analyzed by GC/FID. The percent
desorption efficiency for each solvent
from each sorbent was calculated based
on the known charge of that solvent.
To apply Tedlar bag sampling to a high
temperature/high moisture source, an in-
line gas conditioning system was
designed. The system consisted of a
condenser to cool the gas stream,
followed by a knockout jar to collect the
moisture, and then the Tedlar bag. In a
laboratory check of this system design,
condensate collected in the knockout jars
was analyzed. High temperature/high
moisture stack gas containing the
solvents was simulated as previously
described and drawn through the
sampling system at approximately 1.0
L/min for 1 h. Aliquots of the condensate
collected were passed through a C18
cartridge, eluted with methanol, and
injected into the GC/FID, or placed into 4
ml of methylene chloride and injected
into the GC/FID.
An emission spiking system was
developed to introduce known quantities
of the solvents into a stack with relatively
clean emissions. A mixture of the
solvents equivalent to the desired stack
gas concentrations was passed through a
pressurized atomizer to create a fine mist
that would be vaporized in the high
temperature emissions. A SpeedaireR
paint sprayer was modified to include a
deVilbus nebulizing nozzle. The system
pressure was used to regulate the flow. A
Teflon flow meter was placed in-line for
fine adjustment of the flow rate.
A preliminary test run was conducted
to determine any need for modification in
the sampling system. Then field
evaluation testing using the refined
protocols was conducted according to
EPA Method 18 with modification of the
sampling systems to include in-line
conditioning for the high temperature/high
moisture gas stream. Each of five
sampling runs was conducted over a 1-h
time period. The spiking system was
started 5 minutes before sampling began,
and sampling continued 5 minutes after
all of the solvent mixture had been
introduced into the stack. The sampling
probe, which was constructed of Teflon
tubing, was inserted into a port with the
inlet at the center of the stack
approximately 18 feet downstream of the
point of injection of the solvents. A
diaphragm pump was used to pull the
sample gas into a heated sampling
manifold constructed of stainless steel
piping. Twenty sampling lines were
attached to the manifold, allowing
generation of up to 20 simultaneous
samples during a single sampling run
(see Figure 1 for a detailed schematic of
an eight-train setup). Eight of the sample
lines were equipped for Tedlar bag
sampling, while the other twelve were
equipped for sorbent tube sampling. A
sampling rate less than that specified by
Method 18 was used for 12 of the 16 bag
-------
samples collected. To validate the use of
this reduced flow rate, one set of four bag
samples was collected at the higher rate.
The actual flow rates for the small bag
samples were maintained (using critical
orifices) between 0.038 and 0.040 L/min,
while the higher flow rates were 0.440 to
0.470 L/min. The flow rates through the
primary and backup sorbent tubes
ranged between 0.230 and 0.240 L/min,
also maintained by critical orifices. The
resulting Tedlar bag samples were
analyzed within 4 hours of sampling, and
the sorbent tube samples were analyzed
within 5 days of sampling.
Results and Discussion
The optimum operating conditions
identified for analysis of PERC samples
collected in Tedlar bags during the first
field evaluation utilized a packed
chromatography column containing 20%
SP-2100/0.1% Carbowax 1500 on
100/120 mesh Supelcoport, operated
isothermally at 120°C. PERC solutions
obtained from the desorption of the
charcoal tubes using methylene chloride
were analyzed using an OV-101 column
and the following temperature
programming: 70 °C (3 minutes)-ramp of
30°C/min (1 minute) - 100°C (3 minutes).
For the PERC bag samples, the
estimated limit of detection was 1.6 ppm,
and the estimated quantifiable limit was
4.8 ppm. No upper detection limit was
apparent below the level at which room
temperature nitrogen is saturated with
perchloroethylene. For liquid injections of
4.0 ill, dissolved PERC in amounts as low
as 0.003 pi per 4 ml desorbing solution
resulted in discernable detector
response, while concentrations above
160 ul per 4 ml of desorbing solution
saturated the detector. Migration of
PERC from the primary to the backup
portion of the charcoal tubes was found
to be insignificant at charge levels from
1.3 to 36.0 nL
All bag samples from the first
evaluation were initially analyzed on site.
The average coefficient of variation
observed for the bag samples from the
controlled emissions of PERC was 9.8%.
A pooled standard deviation of 14 ppm
was calculated for charcoal tube samples
collected from the controlled emissions.
The average coefficient of variation for
these samples was 4.6%.
For the second field test, optimization
of instrument operating conditions for
3/8" Teflon Tubing
z
•3/8" Teflon Tubing
Figure 1. Detailed schematic of multiple bagisorbent tube sampling train system.
-------
analysis of bag and desorbed sorbent
tube samples resulted in analysis of both
types using a packed chromatography
column of 1% SP-1000 on 60/80
Carbopack-B Supelcoport, operated
isolhermally at 200°C. The carrier gas
How was maintained at 20 mL/min for
both the liquid and the gaseous sample
injections.
Of the seven sorbents examined for
chlorinated solvent collection under high
temperature/high moisture conditions,
two were selected for use in the field
evaluation test. Supelco's Carboxen-564,
a hydrophobia carbon molecular sieve
which is designed to replace charcoal for
the collection of airborne contaminants,
demonstrated excellent collection
efficiency for all the solvents. After 158
minutes of sampling the solvent mixture
through the Carboxen sorbent tube at a
flow rate of 0.460 L/min with 65%
moisture and a stack gas temperature of
173*F, there was no evidence of
breakthrough. Union Carbide's TRI-X-
100, which Is an inorganic highly
hydrophobia silica oxide generally used
for waste removal in industrial adsorption,
also showed acceptable collection
efficiency. Under similar sampling
conditions with 46% moisture, an average
stack gas temperature of 155°F, and a
sample flow rate of 0.590 L/min, it
showed 20% breakthrough of CCI4, while
the other three solvents were retained.
The other five sorbents did not retain the
solvents under similar conditions.
The desorption efficiencies of the
Carboxen and TRI-X-100 sorbents were
evaluated. Although each of the
chlorinated solvents could be analytically
separated from each of the desorbing
solutions, methylene chloride proved to
be the most efficient in removing the
solvents from both sorbents, with the
lowest recovery being 82%. Methylene
chloride was, therefore, chosen for
desorption of the sorbent tubes for the
remaining GC analyses.
GC analysis of the condensate
collected by the in-line gas conditioning
system for Tedlar bag sampling
confirmed that the solvents were not
being collected; none were detected.
The analysis of samples collected
during the preliminary sampling yielded
an average relative standard deviation of
(RSD) 14.1% and an average accuracy of
±9.6% for the solvents collected on
TRI-X-100. The average RSD for the
solvents collected on the Carboxen-564
was 10.0%, with an average accuracy of
±8.8%. For the Tedlar bag sample
analysis, the precision, in terms of the
average RSD, was 6.1%, with an average
accuracy of ±9.8%. All adsorption tubes
were desorbed with 4 ml of methylene
chloride. One problem noted with the
sampling system during the preliminary
test run was that the sample conditioning
devices connected to the end of the
manifold farthest from the stack gas flow
were not collecting as much condensate
as the others. As a result, the sorbent
tubes were not equally exposed to the
high moisture conditions and system
modification was necessary to ensure
that the gas sampled by all the trains was
equivalent. Therefore, the manifold was
wrapped with a heating coil to prevent
loss of heat at the end of the manifold. It
was also discovered that some of the
critical orifices were clogging, which was
most likely caused by fine particles of the
sorbent exiting the tubes. This was
remedied by including filters in-line
behind the tubes.
The refined sampling system was used
to collect the organic solvents spiked in
the high temperature/high moisture gas-
fired boiler emissions. The overall
average RSD for the small bag samples
was 13.5%, and the overall average RSD
for the large bag samples was 23%, with
an average accuracy of -14.0% and
+ 8.7%, respectively. For the sorbent
tube analysis, over 90% of each solvent
was collected on the primary Carboxen
tube, while as much as 50% of each
solvent was found on the backup TRI-X-
100 tube when the sorbent was exposed
to high moisture conditions. A pooled
standard deviation was calculated for
both the TRI-X-100 tubes and the
Carboxen tubes. The pooled standard
deviations for TRI-X-100 tubes were:
4.28% for CCI4, 12.95% for CHCI3,
3.13% for PERC, and 3.06% for TCE.
The pooled standard deviations for the
Carboxen tubes were: 6.09% for CCI4,
5.92% for CHCI3, 5.50% for PERC, and
3.18% for TCE. The overall average
standard deviation for solvent collection
on TRI-X-100 was 5.85%, while the
overall average standard deviation for the
Carboxen samples was 5.17% using a
combination of the primary and backup
tube values. The average accuracies
calculated for the Carboxen and the TRI-
X-100 tubes were +8.8% and +9.8%,
respectively.
Conclusion
Method 18 sampling procedures and
gas chromatography using flame
ionization detection were found to be
acceptable for the measurement of PERC
at degreasing facilities.
Method 18 procedures modified for
sample collection under high
temperature/high moisture conditions
were found to be acceptable for sampling
and analysis of emissions of CCI4,
CHCI3, PERC, and TCE.
The Supelco sorbent, Carboxen-564, is
effective for sample collection of CCI4,
CHCI3, PERC, and TCE emissions in high
temperature and high moisture
environments. TRI-X-100 appears to be
effective for the collection of these
solvents under high temperature
conditions; however, its collection
efficiency is decreased in the presence of
high moisture.
An in-line gas conditioning system
employing a condenser for cooling the
sample gas stream and a knock-out jar
for eliminating moisture collection is
appropriate for Tedlar bag sampling of
the above solvents under high
temperature and-high moisture con-
ditions.
References
1."Method 18: Measurement of
Gaseous Organic Compound
Emissions by Gas Chromatography,"
40 Code of Federal Regulations 60,
July 1, 1987.
2. Knoll, J. E., M. A. Smith, and M. R.
Midgett, "Evaluation of Emission
Test Methods for Halogenated
Hydrocarbons" - Vol. 1, CCI4,
C2H4Cl2 CI2CI4, Publication No. EPA-
600/4-79-02, U. S. Environmental
Protection Agency, Research Tri-
angle Park, North Carolina.
3."Method 18: Measurement of
Gaseous Organic Compound
Emissions by Gas Chroma-
tography," Quality Assurance Hand-
book for Air Pollution Measurement
Systems, Volume III, Publication No.
EPA-600/4-77-027b, August 1977.
-------
-------
A.C. Carver and W.G. DeWees are with Entropy Environmentalists, Research
Triangle Park, NC; £ A Coppedge (also the Project Officer) is with the
Atmospheric Research and Exposure Assessment Laboratory, Research
Triangle Park, NC. 27711.
The complete report, entitled "Laboratory and Field Evaluations of Methodology
for Measuring Emissions of Chlorinated Solvents from Stationary
Sources," (Order No. PB90-155 565/AS; Cost: $17.00, subject to change)
wili 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:
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S3-90/009
------- |