United States
Environmental Protection
Agency
Atmospherfc Research and Exposure
Assessment Laboratory
Research Triangle Park, NC 27711
Research and Development
EPA/600/SR-94/080 July 1994
v*EPA Project Summary
A Test Method for the
Measurement of Methanol
Emissions from Stationary
Sources
B. A. Pate, M. R. Peterson, E. E. Rickman, and R. K. M. Jayanty
Methanol was designated under Title
III of the Clean Air Act Amendments
(CAAA) of 1990 as a pollutant to be
regulated. The U.S. Environmental Pro-
tection Agency (EPA), through a con-
tract with Research Triangle Institute
(RTI), has developed a test method for
the measurement of methanol emis-
sions from stationary sources. The
methanol sampling train (MST) consists
of a glass-lined heated probe, two con-
densate knockout traps, and three sor-
bent cartridges packed with Anasorb
747. The Anasorb samples were des-
orbed with a 1:1 mixture of carbon dis-
ulfide and N,N-dimethylformamide. All
samples were analyzed by gas chro-
matography with flame ionization de-
tection.
Following laboratory testing, field
tests of the MST and the National Coun-
cil of the Paper Industry for Air and
Stream Improvement (NCASI) sampling
method for methanol were conducted
at two paper and pulp mills. In accor-
dance with EPA Method 301, two pairs
of trains were run in parallel for six
runs, collecting a total of 24 samples
by each method. During each run, half
of the trains were spiked with a known
amount of methanol. The sampling lo-
cation at the first field test was an inlet
vent to a softwood bleach plant scrub-
ber where the methanol concentration
was about 30 ppm. The average per-
cent recovery of the spike was 108.3%
for the MST method and 81.6% for the
NCASI method. Although neither
method showed significant bias at the
95% confidence level, the biases of the
two methods were significantly differ-
ent. A second field test was conducted
at the vent of a black liquor oxidation
tank where the methanol concentration
was about 350 ppm. The average per-
cent recovery of the spike was 96.6%
for the MST method and 94.2% for the
NCASI method. The biases of the two
methods were not significantly differ-
ent for the second field test. The MST
had a practical quantitation limit (PQL)
of about 3 ppm for a 20-L sample.
Samples were shown to be stable for
at least 2 weeks after collection.
This Project Summary was developed
by EPA's Atmospheric Research and
Exposure Assessment Laboratory, Re-
search 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
Under Title III of the CAAA of 1990,189
toxic chemical species will be regulated
for stationary source emissions. Methanol
is one of the pollutants to be regulated.
EPA, through a contract with RTI, has
developed a test method for the measure-
ment of methanol emissions from station-
ary sources.
A literature search was conducted to
review the sampling and analysis meth-
ods currently used to measure methanol.
Based on the information obtained from
this literature search, the MST was devel-
oped and evaluated in the laboratory. The
MST consists of a glass-lined heated
probe, two condensate knockout traps in
an ice bath, and three sorbent cartridges
packed with Anasorb 747. A 1:1 mixture
Printed on Recycled Paper
-------
of caibon cfeulfideand N.hkfimethylformamide
is used to desorb the Anasoib samples. The
knockout traps, used to remove water va-
por, also collect a significant amount of
methanol if water is present. Condensate
and desorption samples are analyzed by
gas chromatography with flame ionization
detection.
A field evaluation of the MST and the
NCASI sampling method for methanol was
conducted at two pulp and paper mills in
accordance with EPA Method 301 proce-
dures. The first field test location was an
inlet vent to a softwood bleach plant scrub-
ber where the methanol concentration was
about 30 ppm. A second field test was
conducted at the vent of a black liquor
oxidation tank where the methanol con-
centration was about 350 ppm. Both meth-
ods met the criteria for precision and bias
as set forth in Method 301.
Experimental Approach
Laboratory Testing Apparatus
A dynamic dilution system was used for
mixing methanol and diluent humidified
nitrogen. The components were mixed in
a 1-L Kimak* dilution flask. Flow rates of
the test gas and the humidified nitrogen
were regulated with Tyian mass flow con-
trollers. The gas mixture was passed from
the dilution flask to a three-port manifold,
both of which were enclosed in an insu-
lated, temperature-regulated box. A cylin-
der containing 250 ppm methanol was
used as the test gas and was diluted to
the desired concentration using this dy-
namic dilution system.
The MST, shown in Figure 1, consists
of a heated sampling probe, two conden-
sate traps in an ice bath, three Anasorb
747 (SKC) sorbent cartridges, and a pump.
(However, the results from the field tests
would later show that only two Anasorb
cartridges are necessary.) Each conden-
sate trap is an empty half-stem impinger.
The first sorbent cartridge contains 5 g of
Anasorb and the backups contain 1.5 g of
Anasorb each.
Laboratory Experiments
The Occupational Safety and Health
Administration (OSHA) recently promul-
gated a method for measuring methanol
in the workplace (OSHA Method 91) that
uses Anasorb 747 (SKC) as the collection
medium. Although OSHA Method 91 was
written for methanol as a non-stationary
'Mention of trade names or commercial products does
not constitute endorsement or recommendation (or
source, a literature review indicated that
Anasorb was the most promising sorbent
and the results outlined below show that
Anasorb was highly efficient in collecting
methanol.
The desorption efficiency of methanol
from Anasorb was determined at three
different loadings. A primary methanol
standard (94.8 pg/mL) was prepared by
diluting 3 mL methanol to 25 mL with
deionized water. Five grams of Anasorb
were placed in three separate vials and
were then spiked separately with 3 |iL, 6
H.L, and 9 \iL of the methanol standard.
The Anasorb in each vial was desorbed
with 30 mL of CS./DMF. Each Anasorb
sample was calibrated with a correspond-
ing secondary methanol standard, pre-
pared by spiking 3 (iL, 6 p.L and 9 (xL of
the primary standard into separate vials
containing 30 mL of solvent.
Several spiking tests were conducted to
determine the recovery efficiency of the
MST. A heated stainless steel tee, with a
septum in one arm, was placed between
a port on the dilution system and the MST.
The dilution system was set to a tempera-
ture just above 100°C, and the diluent
nitrogen contained 20% water by volume.
The sample flow rate was 1 L/min for all
runs. At the beginning of each run, ap-
proximately 5 H.L of methanol was injected
at the tee.
Field Testing and Method
Validation
Field testing of the MST was conducted
at three different sites. The primary objec-
tive of the field testing was to obtain an
estimate of the precision and accuracy of
the method under field conditions. A sec-
ondary objective was to compare the MST
to the NCASI method (shown in Figure 2)
for sampling methanol. In accordance with
EPA Method 301, a field test of a method
consisted of four trains run in parallel for
six runs (24 samples total). The four trains
were run as two pairs, with the two probes
of each pair taped together and inserted
perpendicular to the flow of gas in the
vent (or stack).
Field Site A: TMP Mill
The first field site was the stack from an
atmospheric cyclone at a thermomechanical
pulping mill. Source data provided by the
mill indicated that the methanol concen-
tration was around 100 ppm. The first run
of the presurvey was stopped after 30
seconds because condensing steam had
filled the condensate knockout traps and
the Anasorb cartridges, and water was
threatening the sampling pump. The tem-
perature of the stack gas was measured
at 212°F, indicating that the gas was nearly
all water. Because neither method was
suitable under the prevailing conditions, a
second field site was necessary.
Field Site B: Softwood Bleach
Plant Scrubber
The second field site was a 3-foot-di-
ameter inlet vent to a softwood bleach
plant scrubber at a pulp and paper mill.
The site was downstream from a fan that
created a strong positive pressure. The
vent gas temperature was just above
130°F. Sampling Trains 7 and 10 were
exposed to direct sunlight whereas Trains
11 and 14 were shaded. To prevent break-
through problems due to sunlight, the sor-
bent tubes for Trains 7 and 10 were
shielded from the sun with a moist rag.
The heated probes were placed about 18
in. into the vent and each sample was
collected at a rate of 1.0 L/min for 20 min.
On the first day of sampling, seven
NCASI runs were performed. Each NCASI
train consisted of two impingers (one dry
and one filled with 20 mL of deionized
water) and a silica gel tube. The methanol
sampling train was tested on the second
day of sampling. Each pair of trains was
alternately spiked with 6 (J.L of a methanol
standard. Thus, each spike contained
about 569 pg methanol, approximately the
same amount collected during each run.
The spikes were injected with a 10-|iL
syringe immediately after the start of a
run.
Field Site C: Black Liquor
Oxidation Tank
The third field site was a 3-foot-diam-
eter exhaust vent from a black liquor oxi-
dation tank at a pulp and paper mill. The
vent gas temperature was about 170°F.
Six NCASI runs were performed on the
first day of sampling. Each NCASI train
consisted of a single impinger (filled with
20 mL deionized water) and a silica gel
tube. Because the presurvey indicated a
high methanol concentration (~350 ppm),
all runs were conducted at a flow rate of
0.5 L/min for 20 min. Each pair of trains
was alternately spiked with 4 |iL (3,170
ng) of neat methanol, which was equiva-
lent to a methanol spike concentration of
244 ppm for a 10-L sample volume. The
spikes were injected with a 10-jj.L syringe
6 to 8 min. after the start of a run. The
four probes were placed in a single 3-in.-
I.D. port because there was no second
port that was perpendicular to the first
port. The methanol sampling train was
tested on the second day of sampling.
Each methanol sampling train consisted
of two condensate traps and three sor-
-------
Heated
Sampling
Probe
Condensate
Knockout
Traps
Anasorb 747 Tubes
Back
Ice Bath
Figure 1. Methanol sampling train.
Unheated
Teflon
Line
Water-Filled
Impinger
(20 mL)
Silica Gel
Tube
(520/260 mg)
I
\
1
Sampling
Pump
Ice Bath
Figure 2. NCASI sampling train for methanol.
-------
bent tubes containing 5,1.5, and 1.5 g of
Anasorb, in that order.
Results and Discussions
Laboratory Evaluation
The average overall recovery (desorp-
tion efficiency) of methanol from Anasorb
for the three samples was 98.1 ± 1.7%.
The three vials that were spiked with 3
III, 6 JJ.L, and 9 jiL of methanol had re-
spective overall recoveries of 95.8, 98.5
and 100.0%.
The average overall recovery efficiency
(Table 1) of the methanol spike from the
front (3 g), middle (1.5 g), and back (1.5
g) Anasorb cartridges was 97.3 ± 2.6%. A
comparison of Runs 1 and 3 with Run 2 of
the spiking tests indicates there was sub-
stantial breakthrough on the front Anasorb
cartridge when 60 L of sample was col-
lected. When the sample volume was re-
duced to 30 L, there was little break-
through. In Run 4, breakthrough was elimi-
nated by increasing the amount of Anasorb
in the front tube from 3 to 5 g.
Field Testing and Method
Evaluation
Field Site B: Softwood Bleach
Plant Scrubber
The presurvey results at the softwood
bleach plant scrubber vent gave concen-
trations that were very similar. A major
concern with the presurvey data was
breakthrough of methanol on the sorbent
tubes. The breakthrough may have been
due to elevated temperatures at the site
because the ambient temperature was
97°F and the MST was exposed to sun-
light, resulting in a dry gas meter tem-
perature of 114°F.
The paired trains for NCASI and MST
methods showed very good precision. The
average spike recovery for the MST was
Tabla 1. Percent Recovery of Methanol Spike
considerably higher than for the NCASI
method: 108.3% compared to 81.6%. The
methods had nearly identical average
unspiked concentrations, but the average
spiked concentration of the MST was
higher than the average spiked concen-
tration of the NCASI method. The reason
for this difference is unknown. Both meth-
ods had similar, and rather high, standard
deviations of spike recoveries. The high
standard deviation of the spike recoveries
was due to a bias that existed between
the two pairs of trains. Trains 11 and 14
gave results that were consistently about
6% higher than the results from Trains 7
and 10. As a result, when pair 11 and 14
was spiked, there was a high spike recov-
ery, and when pair 7 and 10 was spiked,
there was a low spike recovery. The spike
recovery for pair 11 and 14 averaged about
25% higher than the spike recovery for
pair 7 and 10. This discrepancy was found
for both methods and the cause is un-
known. Because of the precision between
the paired trains, it is unlikely that there
was a leak problem.
The data obtained by both methods
were homogenous and a linear regres-
sion of train pair differences versus con-
centration showed there was no depen-
dence on concentration. A statistical analy-
sis following EPA Method 301 procedures
showed that the precision of paired trains
was not statistically different for the two
methods tested and that no significant bias
was found with the MST and the NCASI
methods ayhe 95% confidence interval.
A small amount (0.1%) of methanol
broke through the front Anasorb tube in
spiked field test samples and a negligible
amount broke through in unspiked field
test samples. This result was significant
because of the breakthrough problems that
occurred during the presurvey. Apparently
,the extreme heat at the presurvey did
cause breakthrough, but the corrective
measures (he., shielding the tubes from
Run
1
2
3
4
Volume
Sampled
(Q
60
SO
60
60
Condensate Traps (%)
Front Back Total
45.2
44.0
51.0
44.5
3.9
0.4
5.0
1.6
49.1
44.4
56.0
46.1
Anasorb Cartridaes (%)
Front
28.6
45.2
17.8
51.7
Middle
4.6
19,0
3.6
Back
18.8
0.0
4.4
0.0
Total
47.4
49.8
41.2
55.3
Average recovery
Standard deviation
Total
Recovery
96.5
94.2
97.2
101.4
97.3
2.6
the sun with a wet cloth) taken during the
field test were adequate to solve the prob-
lem.
Field Site C: Black Liquor
Oxidation Tank
The results for the presurvey at the
black liquor oxidation tank indicated a
methanol concentration between 300 and
400 ppm. The vent gas temperature
ranged between 170°F and 190°F for the
three runs. Because the ambient tempera-
ture was measured at 107°F, the Anasorb
tubes were protected from the sun and
heat by a wet cloth. Several unidentified
compounds were found in the Anasorb
samples, indicating the presence of com-
pounds other than methanol in the vent
gas matrix.
The paired trains for each method
showed good precision, although there was
substantial bias between Trains 3 and 9
for the NCASI method. Train 9 collected
an average of 35 ppm more methanol
than Train 3. The average spike recovery
for the MST was slightly higher than for
the NCASI method, 96.6% to 94.2%. Both
methods had a similar standard deviation
of the spike recovery, 8.7% for the NCASI
method and 9.8% for the MST method.
Concentration data for both methods
were homogenous at the 95% confidence
level, and a linear regression of train pair
differences versus concentration showed
there was no dependence on concentra-
tion. A statistical analysis conducted ac-
cording to EPA Method 301 procedures
showed that the precision of paired trains
was not significantly different for the two
methods at the 95% confidence level and
both methods were found to be accurate
because neither had a statistically signifi-
cant correction factor for its bias. No
methanol broke through the first Anasorb
tube even though the methanol level in
the spiked samples was almost 600 ppm.
This field test showed that a third Anasorb
tube is probably unnecessary for the MST
method if the sorbent tubes are properly
shielded from excessive heat.
Analytical Instrumentation and
Performance Evaluation
Analytical Systems
The majority of the samples from the
laboratory evaluation were analyzed on a
Hewlett-Packard 5890 Series I gas chro-
matograph (GC) equipped, with a flame
ionization detector (FID). A 30-m DB-Wax
megabore column was used for the analy-
sis. The temperature program started at
50°C and increased by 10°C/min to a final
temperature of 140°C. The temperature
-------
program was shortened when water analy-
ses were performed.
The field test samples were analyzed
on a Hewlett-Packard 5890 Series II GC
equipped with an FID and a Hewlett-
Packard 7673 autosampler. A 30-m DB-
624 megabore column was used to ana-
lyze all samples for the field test at Site B.
For the field test at Site C, a 60-m
Supelcowax-10 column was used to ana-
lyze Anasorb samples and a 3Q-m DB-
624 megabore column was used to ana-
lyze the water samples.
Performance Evaluation
The single sample method was used to
estimate the analytical limit of detection
(LOD) and limit of quantitation (LOQ) for
water and CS^DMF samples. For each
solvent, a methanol solution at a concen-
tration about 1.5 times the estimated LOQ
was analyzed eight or nine times to deter-
mine a value for s, the approximate stan-
dard deviation at the lowest level of mea-
surement. From this determination the LOD
was calculated as 3s and the LOQ as
10s. Both methanol standards were pre-
pared at a concentration of 0.95 ng/mL.
The CS./DMF samples had an analytical
LOD of 0.20 ng and a LOQ of 0.69 ng
while the water samples had an analytical
LOD of 0.15 ng and a LOQ of 0.50 ng.
The analysis was performed with the ana-
lytical system used for the analysis of field
test Site B samples.
The linearity of the analytical system
used for the field test samples was evalu-
ated with CS/DMF and water standards.
A primary standard was prepared by dilut-
ing 3 mL methanol to 25 mL with deion-
ized water, resulting in a methanol con-
centration of 94.8 ng/mL. Six standards of
each solvent were prepared by spiking
various amounts of solvent with the pri-
mary standard. These solutions ranged in
methanol concentration from 4 to 35 (ig/
mL. Each solution was analyzed three
times and the FID response was shown to
be linear over a methanol concentration
range of 5 to 85 ng/L for CS./DMF samples
and 4 to 76 ng/mL for water samples.
Quality Control
Sample Stability
Sample stability tests were performed
for Anasorb, silica gel, and water. Samples
were analyzed on the day they were spiked
(Day 0) and on Days 1, 3, 7, and 14
following spiking. Three spiked samples
and one blank of each medium were ana-
lyzed on each day. The recovery for each
day represented the average of three tubes
analyzed and blank corrected. Samples in
water and silica gel were shown to be
stable for at least 2 weeks while the Anasorb
samples were shown to be stable for at
least 1 week. Other Anasorb samples were
found to have a recovery of about 90%
after 2 weeks. The recovery of methanol
from Anasorb was 107% after 1 week.
The recovery of methanol from water was
96% after 2 weeks and 111 % of methanol
was recovered from silica gel after 2
weeks.
Conclusions and
Recommendations
A laboratory evaluation of the MST
. showed that Anasorb had a desorption
efficiency for methanol of 98.1 ± 1.7%.
Samples taken where methanol was col-
lected on the Anasorb or in the conden-
sate knockout trap were found to be stable
for up to 2 weeks. The analytical system
used to measure methanol had a limit of
quantitation of 0.5 ppm, and the MST
method had a quantitation limit of 3 ppm.
The MST and NCASI methods both met
the guidelines specified by EPA Method
301 in both field tests. Both methods
showed very good precision, which was
not significantly different at the 95% confi-
dence level. Although neither method
showed significant bias, the biases of the
methods were significantly different at the
95% confidence level for the first field
test. The disparity in the biases of the
methods is evident from the spike recov^
ery values. Only 81.6% of the spike was
recovered using the NCASI method, while
108.3% of the spike was recovered with
the MST method. The reason for this dif-
ference is unknown; however, the heated
glass-lined probe and injection port used
in the MST may have been more efficient
at transporting methanol than the unheated
Teflon sampling line and injection port rec-
ommended by the NCASI method.
The bias between methods that was
seen in the field test at Site B was not
found in the field test at Site C. The aver-
age percent recovery of the spike was
96.6% for the MST method and 94.2% for
the NCASI method. The precision of paired
trains was not significantly different for the
two methods at the 95% confidence level.
An advantage of the MST is that the
composition of the emission source can
be determined by using a mass spectro-
metric detector to analyze the desorbed
Anasorb samples. Emissions at Field Site
B were considered clean because no com-
pounds other than methanol could be de-
tected. Field Site C had a more complex
emission matrix and provided important
information about the capabilities of the
MST method. Field Site C also had a high
methanol concentration, which helped de-
termine the capacity of the MST method.
The information in this document was
funded wholly or in part by the U.S. Envi-
ronmental Protection Agency under Con-
tract No. 68-D1-0009 to Research Triangle
Institute, Research Triangle Park, NC.
•&U.S. GOVERNMENT PRINTING OFFICE: 1994 - 550-047/8028!
-------
------- |