United States
Environmental Protection
Agency
Atmospheric Research and Exposure
Assessment Laboratory
Research Triangle Park, NC 27711
Research and Development
EPA/600/S3-90/033 Aug. 1990
&EPA Project Summary
Audit Materials for Semivolatile
Organic Measurements During
Hazardous Waste Trial Burns
J. R. Albritton, R. S. Wright, W. C. Eaton, R. K. M. Jayanty, and Robert G
Fuerst
Two audit materials have been
developed to assess the accuracy
and precision of semivolatile organic
measurements using EPA Method
0010 SW 846. The first audit material
is prepared by spiking known
quantities of organic compounds
onto XAD-2 resin in glass cartridges.
It is used to evaluate the analytical
portion of the method. The second
audit material is prepared by spiking
known quantities of organic
compounds onto Carbotrap™ in
stainless steel cartridges. It is used
to evaluate the sampling and
analytical portions of the method.
Recovery efficiencies were found to
be between 88 and 98 percent for
XAD-2 audit cartridges and between
91 and 108 percent for Carbotrap ™
audit cartridges. In general, test
compounds were found to be stable
on XAD-2 audit cartridges under
refrigeration over an 8-month period
and on Carbotrap audit cartridges at
room temperature over a 2-month
period. An interlaboratory study was
used to assess the accuracy and
precision of the audit materials for
six test compounds: pyridine,
toluene, o-xylene, chlorobenzene,
1,1,2,2-tetrachloroethane, and
nitrobenzene. The study involving five
cooperating laboratories revealed an
overall mean bias percentage of -29
percent. The mean within-laboratory
variability was 22 percent. These
variabilities generally agree with
corresponding values obtained in an
interlaboratory comparison study of
methods for volatile and semivolatile
compounds on solid waste samples.
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
Operators of hazardous waste
incinerators are currently using Method
0010 to monitor stack emissions during
hazardous waste trial burns. The
sampling train consists of a series of
traps and impingers that collect
particulates and gaseous semivolatile
organics An adsorbent material, XAD-2
resin, is used to absorb organic vapors
with boiling points greater than 100°C.
The overall accuracy of the process of
trapping, desorbmg, and analyzing these
organic vapors by Method 0010 can be
assessed by Agency personnel using
audit materials.
Research Triangle Institute (RTI) has
investigated two audit materials for
delivery of test compounds that could be
used in performance audits during
hazardous waste trial burns. The first
audit material is XAD-2 resin contained in
glass cartridges that has been spiked
with known amounts of several
semivolatile organic compounds. This
audit material is sent to the laboratory
involved in the trial burn. The laboratory
recovers these compounds by Soxhlet
extraction, concentrates the extract by
Kuderna-Danish evaporation, and
analyzes the extract by gas
chromatography/mass spectroscopic
(GC/MS) detection. This audit material is
intended to assess only the analytical
portion of Method 0010. The second
audit material is an adsorbent, Carbotrap,
-------
contained in stainless steel cartridges
that have been spiked with known
amounts of semivolatile organic
compounds and delivered to the trial burn
sites. Once at the site, the compounds
are thermally desorbed from the audit
cartridge onto an operating Method 0010
sampling tram, which collects them on
XAD-2 resin for subsequent recovery and
analysis. This audit material is intended
to assess the combined sampling and
analytical portions of Method 0010.
Measurement errors can occur in either
the sampling or the analytical portions of
Method 0010. The use of two different,
but identically loaded, audit materials
allows one to begin to locate the source
of an observed measurement error. If the
measurement error were in the sampling
portion, one would expect to see
accuracy or precision problems for the
Carbotrap audit cartridges, but not for the
XAD-2 audit cartridges. If the
measurement error were in the analytical
portion, one would expect to see
accuracy or precision problems m both
sets of audit cartridges. Statistical tests
should be used to verify that any
apparent accuracy or precision problems
are significant. These audit materials
cannot detect measurement errors arising
from other sources such as the
volumetric sampling rate.
Experimental Procedures
RTI performed initial recovery and
stability studies for several test
compounds on the two audit materials. A
batch of identical audit cartridges was
prepared by injecting the same known
quantities of the test compounds onto the
sorbent material. These audit cartridges
were then analyzed by RTI or, in some
cases, by a cooperating laboratory. In the
recovery studies, the injected and
measured loadings for multiple cartridges
were compared to yield the means and
relative standard deviations of the
recovery efficiencies. In the stability
studies, the measured loadings for
multiple dates were compared to yield
data on the stability of the batch.
The XAD-2 audit cartridges were
loaded by volatilizing a solution of the
test compounds in methylene chloride
with a flash evaporation unit. Once
volatilized, the test compounds were
carried from the unit by a nitrogen flow,
were diluted with room air, and were
sorbed on the XAD-2 resin with a Nutech
Model 201 sample pump. The audit
cartridges were immediately sealed after
loading and then were stored under
refrigeration.
The XAD-2 audit cartridges to be
analyzed at RTI were Soxhlet extracted
for 16 hours with methylene chloride.
Internal standards having different boiling
points from those of the test compounds
were added to the extracts. The extracts
were concentrated by Kuderna-Danish
evaporation The concentrated extracts
were analyzed by gas chromatography
with flame lonization detection (GC/FID).
The same analytical techniques were
followed by the cooperating laboratory
with the exception that analysis was by
GC MS detection Deuterated versions of
the test compounds were added as
internal standards to the cooperating
laboratory XAD-2 resin.
The Carbo'rap audit cartridges were
loaded by volatilizing a solution of the
test compounds in methylene chloride
with a flash evaporation unit. Once
volatilized. Hie :est compounds were
carried from the unit by a helium flow and
were sorbed onto the Carbotrap. The
audit cartridges were immediately sealed
and then were stored at room
temperature.
The Carbotrap audit cartridges to be
analyzed at RTI were desorbed by two
different methods: direct thermal
desorption into a gas chromatograph and
thermal desorption onto XAD-2 resin. For
direct thermal desorption, the cartridges
were connected !o the sample inlet of a
gas chromatograph and were heated to
475°C. A helium flow carried the
volatilized test compounds from the
Carbotrap to the gas chromatograph's
column. For thermal desorption onto
XAD-2 resin, the cartridges were heated
and the volatilized test compounds were
carried from the cartridges by a helium
flow. The compounds were diluted with
room air and were sorbed onto the XAD-2
resin with a Nutoch Model 201 sample
pump The subsequent extraction,
concentration, and analysis were as
described above
The Carbotrap audit cartridges to be
analyzed by the cooperating laboratories
during the mterlaboratory study were
thermally desorbed onto XAD-2 resin.
The cartridges were heated to 400 °C and
were purged with a nitrogen flow. The
volatilized test compounds were diluted
with filtered ambient air and were sorbed
onto the XAD-2 resin with a Method 0010
sampling train.
The accuracy and precision of Method
0010 were measured by conducting an
interlaboratory study involving
cooperating laboratories. Audit samples
were prepared and delivered to
laboratory personnel who were
experienced in Method 0010. The
accuracy of the method was estimated
by comparing the theoretical amount of
test compounds loaded onto the audit
cartridges to the amount measured by
the cooperating laboratories. The average
bias percentage between the theoretical
loading and the multiple measured
loadings for each laboratory's
measurements of a compound was used
as an index of the accuracy of the
method. The between-laboratory
variability was estimated by comparing
the results of measurements at several
different laboratories. The relative
standard deviation of the average
measured loadings for multiple
laboratories' measurement of a
compound was used as an index of the
between-laboratory variability of the
method. The within-laboratory variability
was measured by providing several
cartridges loaded at, or near, the same
level. The relative standard deviation of
the multiple measured loadings for each
laboratory's measurements of a test
compound was used as an index of the
within-laboratory variability of the method.
Five laboratories agreed to participate
in the interlaboratory study. Each
laboratory received an audit kit containing
seven Carbotrap audit cartridges, seven
XAD-2 audit cartridges, and necessary
supplies for thermal desorption of test
compounds. The following test
compounds were spiked onto both types
of audit material: pyridine, toluene, o-
xylene, chlorobenzene, 1,1,2,2-
tetrachloroethane, and nitrobenzene. Of
the seven Carbotrap audit cartridges
supplied, three identical cartridges were
loaded at approximately 200 micrograms
(ng) of each compound per cartridge,
three identical cartridges were loaded at
approximately 2,000 ng of each
compound per cartridge, and one
cartridge was a blank The same loading
scheme was also used for the seven
XAD-2 audit cartridges.
The level at which each test compound
was loaded onto the audit material was
verified at RTI before the materials were
shipped to the cooperating laboratories.
Test compound names and a broad
loading range (i.e., 50 to 5,000 ng) were
provided to the laboratories, but the exact
level at which the compounds were
loaded was not disclosed.
Results
The XAD-2 recovery study
demonstrated that test compounds could
be loaded onto and quantitatively
recovered from XAD-2 audit cartridges.
For RTI's analyses, the uncorrected
mean recovery percentages ranged from
-------
73 to 103 percent, and the corrected
mean recovery percentages ranged from
79 to 140 percent. The uncorrected data
indicate generally good recoveries for the
test compounds. The corrected data were
obtained by dividing the uncorrected
mean recovery percentages for the test
compounds by the mean recovery
percentages of the internal standards. In
general, the corrected percentages
improved the recovery percentages, but
in some cases the recovery percentages
declined after correction. The corrected
data suggest that it is difficult to correct
for evaporative losses in test compounds
with low boiling points. This difficulty is
due to the use of internal standards with
boiling points and recovery percentages
that are different from those of the test
compounds.
For the analyses by the cooperating
laboratory, the corrected mean recovery
percentages of the XAD-2 audit
cartridges ranged from 88 to 98 percent.
The improvement relative to RTI's
corrected values is probably due to the
cooperating laboratory's use of internal
standards that are deuterated versions of
the test compounds.
The Carbotrap recovery study
demonstrated that test compounds could
be loaded onto and quantitatively
recovered from Carbotrap audit
cartridges. For these cartridges that were
desorbed onto XAD-2 resin, the mean
recovery percentages ranged from 80 to
108 percent.
The XAD-2 stability study
demonstrated that test compounds were
stable on XAD-2 audit cartridges under
refrigeration for periods of at least 269
days. The mean recovery percentages
ranged from 98 to 106 percent after 4
days, from 99 to 106 percent after 151
days, and from 97 to 109 percent after
269 days. The greatest change in the
measured loadings during the 269-day
period was -6 percent. RTI did not use
internal standards for these analyses. The
excellent recovery percentages during
the XAD-2 stability study are due to
improvements that were made in the
Kuderna-Danish evaporator after the
completion of the XAD-2 recovery study.
The Carbotrap stability study
demonstrated that most of the test
compounds were stable on Carbotrap
audit cartridges at room temperature for
periods of at least 60 days. The
nitrobenzene loading on the cartridges
declined by 23 percent during this
period. For the remaining test
compounds, the mean recovery
percentages ranged from 91 to 99
percent after 1 day, from 90 to 100
percent after 30 days, and from 95 to 95
percent after 60 days. The greatest
change in the measured loadings
(excluding nitrobenzene) during the 60-
day period was -6 percent. RTI directly
desorbed the test compounds from these
Carbotrap audit cartridges into a gas
chromatograph
After the recovery and stability studies
had been completed, XAD-2 and
Carbotrap audit cartridges were prepared
for the mterlaboratory study. A portion of
each batch of cartridges was analyzed by
RTI. The mean recovery percentages for
the lowloading (i.e., -200 ng of each
compound) Carbotrap audit cartridges
ranged from 97 to 105 percent. The
mean recovery for the high-loading (i.e.,
~2,000 ng of each test: compound)
Carbotrap audit cartridges ranged from
101 to 102 percent. The mean recovery
percentages for the low-loading XAD-2
audit cartridges ranged from 104 to 110
percent. The mean recovery percentages
for the high-loading XAD-2 audit
cartridges ranged from 94 to 100 percent.
The relative standard deviations of the
recovery percentages ranged from 1 to
23 percent with a mean value of 6
percent. These results indicate that RTI
had accurately and reproducibly loaded
the XAD-2 and Carbotrap audit
cartridges
The results of the mterlaboratory study
are summarized in Tables 1 and 2. Table
1 gives the test compounds loaded onto
the audit cartridges, the expected
loadings of each compound on the audit
cartridges, and the average bias
percentages for audit cartridges that were
loaded by RTI and analyzed by the
cooperating laboratories. The latter
values are the averages for
measurements of three identical
cartridges that were given to the
cooperating laboratories. They compare
the cooperating laboratories' measured
loadings with RTI's expected loadings.
The overall average bias percentage
for both audit materials in the
mterlaboratory study is -27 percent. The
unusually high value for the Laboratory
E/pyridine/XAD-2/low-loading
combination was found to be an outlier at
the 99-percent confidence level, relative
to the other measured pyridine loadings.
If this value is discarded, the overall bias
percentage is -29 percent. The overall
average bias percentage for the
Carbotrap audit cartridges alone is
approximately equal to the corresponding
value for the XAD-2 audit cartridges alone
(i.e., -28 and -32 percent, respectively,
and excluding the anomalous Laboratory
E value).
The overall average bias percentage
for the low-loading audit cartridges
(excluding the outlier) is -23 percent. This
value is less than the corresponding
value of -35 percent for the high-loading
audit cartridges.
The overall average bias percentages
for each laboratory are given below:
Laboratory
A
B
C
D
E
-60
-39
-5
-39
-1ia
aExcluding the outlier.
The results from the mterlaboratory
study show large variations in bias
percentages between laboratories and
within individual laboratories. For
example, the average bias percentage for
the six test compounds at the two
loadings for both types of audit material
for Laboratory A is -60 percent (range of -
88 to -27 percent) compared to
Laboratory C's average bias percentage
of -5 percent (range of -33 to +21
percent).
The large range in the average bias
percentages for test compounds within a
particular laboratory may be attributed to
the group of test compounds selected for
the interlaboratory study. The individual
compounds were selected with more
emphasis placed on the class of
compounds that they represent rather
than the boiling point of the particular
compound Many cooperating
laboratories indicated the need to modify
their existing Method 0010 analytical
procedures to accommodate low-boiling
test compounds such as toluene
(B.P. = 110°C), chlorobenzene (B.P. =
132°C), and tetrachloroethane (B.P. =
146°C).
The average measured loadings that
were reported by the cooperating
laboratories were used to calculate the
between-laboratory variability. This value
is defined as the relative standard
deviation of the average measured
loading among the five cooperating
laboratories for each test compound/audit
material'loading combination The
between-laboratory variabilities for the
interlaboratory study are given in Table 2.
These values range from 10 to 68
percent (excluding the outlier) with an
overall average value of 42 percent. In
general, the between-laboratones
variabilities are consistent across the
various audit material/loading
-------
Table 1 Mean Bias Percentage for Measured Loadings Inter/nboratory Study
Cooperating laboratories and average bias percentage
Compound
Pyridine
Toluene
o-Xylene
1,1,2,2-Tetra-
chloroethane
Chlorobenzene
Nitrobenzene
Low
197
174
176
319
221
241
High
1,970
1,740
1,760
3,190
2,210
2,410
Low
-64
-48
-58
-49
-76
-27
High
-68
-88
-70
-69
-82
-46
Low
-19
-27
-19
-30
-13
-17
High
-31
-56
-39
-43
-59
-27
Low
+ 5
+ 7
+ 1
+ 1
+ 7
-17
High
+ 12
+ 13
+ 21
-4
+ 10
+ 4
Low3
-63
+ 76
-45
-56
-51
-14
High
-42
-59
-42
-27
-56
+ 32
Lowb
4
-22
-23
-28
+ 13
-34
High
-34
-29
-32
-46
-29
+ 7
XAD-2 resin audit cartridge
Compound
Expected loading
(fig/cartridge)
Low
Cooperating laboratories and average bias percentage
Low
High
Low
Low
Low
High
Lowc
High
Pyridine
Toluene
o-Xylene
1,1,2,2-Tetra-
chloroethane
Chlorobenzene
Nitrobenzene
197
174
176
319
221
241
1,970
1,740
1,760
3,190
2,210
2,410
-52
-57
-53
-52
-71
-37
-60
-83
-61
-63
-73
-44
-20
-51
-30
-33
-53
-18
-39
-66
-46
-46
-67
-35
+ 20
-16
-13
-30
-15
-6
+ 7
-27
-24
-33
-25
-20
-42
-53
-44
-27
-52
-10
-64
-71
-56
-48
-57
-2
+ 306d
+ 15
+ 19
+ 13
+ 36
+ 20
+ 8
-23
-24
-33
-2
-25
aOne of the three Kuderna-Dan/sh evaporators went dry: data for this sample are not included in the calculations.
bOne of three loaded cartridges was reported as "non detected": data for this sample are not included in the calculations.
cTwo of three loaded cartridges were reported as "sample lost": the average percent bias is for a single sample.
dFound to be an outlier at the 99-percent confidence level.
Table 2
Relative Standard Deviations for the Average Measured Loadings Between the Cooperating
Laboratories in the Interlaboratory Study
XAD-2 resin
Carbotrap
audit cartridge
audit cartridge
Compound
Pyridine
Toluene
o-Xylene
1 ,1 ,2,2-Tetrachloroethane
Chlorobenzene
Nitrobenzene
Low
47
49
33
32
51
10
High
42
68
48
40
61
32
Low
43a
36
38
33
63
23
High
49
59
30
22
56
21
aThe single measured loading from Laboratory E is excluded as an outlier.
-------
combinations and across the various test
compounds. The mean relative standard
deviations across the various audit
material/loading combinations are given
below:
Audit material 'loading
comb/nation
Mean RSD
Carbotrap/low loading
Carbotrap/high loading
XAD-2/low loading
XAD-2/high loading
37
48
40
aExcluding the outlier.
The mean relative standard deviations
across the various test compounds are
given below:
Mean RSD
Test Compound (°'°)
Pyndme
Toluene
o-Xylene
1 ,1 ,2,2-Tetrachloroethane
Chlorobenzene
Nitrobenzene
45a
56
37
32
58
22
aExcluding the outl/er.
The relative standard deviations for the
three measured loadings within each
laboratory are given in Table 3. These
values are used to calculate the within-
laboratory variabilities, which are defined
as the mean relative standard deviation
for each laboratory. These values ranged
from 0 to 94 percent with an overall mean
value of 22 percent. The minimum,
maximum, and mean relative standard
deviations for each laboratory are given
below:
Relative standard deviation (%)
Lab.
Min.
Max.
Mean
A
B
C
D
E
9
2
0
7
0
94
20
31
54
82
49
8
11
23
19
In general, the relative standard
deviations are consistent within each
laboratory, but tend to differ between
laboratories. Also, note that within-
laboratory variabilities are less than the
between-laboratories variability. The
mean within-laboratory (i.e., 22 percent)
is approximately one-half of the mean
between-laboratories variability (i.e., 42
percent).
In 1984, EPA-Las Vegas published the
results of an inter laboratory comparison
study of methods for volatile and
semivolatile compounds.(1 ) Nine
laboratories participated in the study.
Semivolatile compounds with boiling
points up to 500°C were spiked onto five
different waste samples. Each laboratory
was requested to analyze three replicates
of each waste sample. The semivolatiles
were e
-------
Table 3 Relative Standard Deviations for Measured Loadings Within Each Cooperating Laboratory in the Interlaboratory Study
Carbotrap audit cartridge
Cooperating laboratories and percent relative standard deviations
Compound
Pyndine
Toluene
o-Xylene
1,1,2,2-Tetra-
chloroethane
Chlorobenzene
Nitrobenzene
Low
49
62
73
31
66
24
High
36
47
46
32
28
19
Low
1 1
12
8
9
13
9
H*.
M
1 1
;'0
3
10
9
Low
31
20
21
24
20
9
High
16
8
3
7
6
0
Lowa
38
36
26
25
29
8
High
7
25
14
29
27
15
Lowb
82
37
37
18
28
0
High
13
5
0
17
21
13
XAD-2 res/n audit cartridge
Cooperating laboratories and percent relative standard deviations
Compound
Pyridine
Toluene
o-Xylene
1,1,2,2-Tetra-
chloroethane
Chlorobenzene
Nitrobenzene
Low
92
73
94
89
69
68
High
22
9
17
17
21
13
Low
4
2
5
5
6
a
High
8
9
8
3
6
4
Low
6
10
8
7
8
11
High
8
5
9
10
9
11
Low
7
16
16
28
13
38
High
14
30
24
13
54
21
Low1- High
11
17
17
1 1
13
6
aOne of the three Kuderna-Danish evaporators went dry, data for this sample are not included in the calculations.
t>One of the three loaded cartridges was reported as "none detected": data for this sample are not included in the calculations
cTwo of three loaded cartridges were reported as "sample lost"; no relative standard deviation can be calculated.
-------
-------
J. R. Albritton, R. S. Wright, W. C. Eaton, and R. K. M. Jayanty are with Research
Triangle Institute, Research Triangle Park. NC 27709. The EPA author,
Robert G. Fuerst (also the EPA Project Officer, see below) is with the
Atmospheric Research and Exposure Assessment Laboratory, Research
Triangle Park, NC 27711.
The complete report, entitled "Audit Materials for Semivolatile Organic
Measurements During Hazardous Waste Trial Burns," (Order No. PB 90-
239 971/AS; Cost: $15.00, 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:
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
United States Center for Environmental Research
Environmental Protection Information
Agency Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S3-90/033
------- |