United States
Environmental Protection
Agency
Atmospheric Research and Exposure
Assessment Laboratory
Research Triangle Park NC 27711
Research and Development
EPA/600/S3-90/047 Aug. 1990
vvEPA Project Summary
Validation of Emission
Sampling and Analysis Test
Method for PCDDs and PCDFs
Jimmy C. Pau, Andres A. Romeu, Marilyn Whitacre, and John T. Coates, Jr.
The precision and accuracy of the
Modified Method 5 sampling and
analysis protocols for polychlorinated
dibenzo-p-dioxins and dibenzofurans
in municipal waste combustor stack
gas have been determined. This was
accomplished using a dynamic
spiking system designed to
continuously deliver stable isotopic
PCDD/PCDF congeners into the MM5
sampling train upstream of the
particulate filter during sampling of
MWC stack gas. The results from this
study indicate that the MM5 sampling
train provides quantitative and
reproducible measurements of
dioxins and furans under the
conditions used in this study.
Accuracy of the measurements as
determined using recovery of the
dynamically spiked compounds
ranged from 77.6% to 117%. Precision
of the measurements was high as
evidenced by low relative percent
differences between replicate trains.
The distribution of native dioxins and
furans within the sampling trains was
essentially the same as for the
dynamically spiked components. The
dynamic spiking system is a useful
tool in the determination of the
accuracy of the measurements. It
should be considered as a viable
alternative to determine method
accuracy during trial burns. It does
not affect the determination of native
dioxins and furans and does not
significantly impact on analytical
costs.
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
Municipal incineration of waste,
coupled with energy recovery, is a
practice that is gaining favor in current
efforts directed at finding economic,
effective and efficient alternatives to land
burial. The U.S. Environmental Protection
Agency has undertaken to regulate
municipal waste incinerators and to
characterize their emissions in order to
allay public concerns about their
emissions. Among pollutants of such
concern are the polychlorinated dibenzo-
p-dioxins (PCDDs) and polychlorinated
dibenzofurans (PCDFs).
A Modified EPA Method 5 (MM5)
sampling train, incorporating an ice-water
chilled condenser and XAD-2 resin
cartridge, has been used to collect
medium volatility organic compounds
from stack gases for a wide range of
combustion sources. The configuration
and operating procedures for this system
in determining PCDDs and PCDFs in
municipal waste combustor (MWC)
emissions were standardized by a
workshop sponsored by the American
Society of Mechanical Engineers, the
U.S. EPA, and the U.S. Department of
Energy held in September, 1984. The
product of the workshop was a draft
sampling and analysis procedures
document known as the ASME protocol.
Recently, the EPA Office of Solid Waste
has written a method for sampling and
-------
analysis of semivolatile organic
compounds from stationary source
emissions. This method is referenced as
SW-846 Method 0010.1
The ASME protocol, and more recently
SW-846 Method 0010. have become
industry standards for sampling stack
gases for PCDDs, PCDFs, PCBs, and
many other semivolatile organics.
However, low recoveries for specific
PCDD isomers spiked into MM5
sampling trains were observed during
pilot and field evaluation studies
conducted under contract to U.S. EPA's
Environmental Monitoring Support
Laboratory (EMSL). Since these
observations were based on limited
experiments, additional data were needed
to determine the precision and accuracy
of the MM5 system for collecting and
recovering PCDD and PCDF emissions.
The overall objective of the current
study was to characterize the precision
and accuracy of the MM5 sampling
system using stable isotopically labeled
PCDD and PCDF compounds spiked
during MWC flue gas sampling.
Procedures
The general experimental approach
was to measure the recoveries of specific
"C-iabeled PCDD and PCDF congeners
dynamically and statically spiked into
MM5 trains during incineration stack gas
sampling. Selected labeled congeners
representing tetra- through octachloro
PCDD and PCDF homologs were spiked
onto the XAD-2 resin prior to each test
(static spike) and continuously throughout
the test (dynamic spike). Following
completion of each test, the samples
were recovered from the trains and
analyzed for the spiked compounds and
for the native PCDDs and PCDFs.
Recoveries were determined for each
spiked compounds by comparing the
amount measured against the amount
spiked.
The incinerator selected for this study
was a mass burn waste combustor
burning municipal refuse. Electrostatic
precipitators were used to control
particulate emissions. Refuse was fed
into a reverse reciprocating stoke grate.
In addition to the motion of the grates,
the underfire grate assisted in further
agitation of the waste. The unit was
operated without auxiliary fuel. A total of
'Tost Methods for Evaluating Solid Wastes,
PhysJcafChomteal Methods (SW-846), 3rd Edition,
Olltco of Solid Waste and Emergency Response,
U.S, Environmental Protection Agency,
Washington, DC, 1986.
26 test runs (including two blank trains
sampling ambient air) were conducted
during six test days. Ports 1 and 2 were
the left- and right-hand liners in one dual
probe located at the inlet of the
electrostatic precipitator. Ports 3 and 4
were in the second dual probe located at
the outlet of the precipitator. Sampling
was conducted simultaneously at inlet
and outlet sampling ports. The two trains
in each probe had closely matched flow
rates, while the two pairs were sampled
at flow rates about 30% apart.
The MM5 sampling train consists of
four main sections (Figure 1): a
nozzle/probe assembly (front half), a
heated filter assembly with a cyclone for
trapping particulates, an ice water-chilled
condenser for trapping moisture (back
half), and an XAD-2 resin cartridge.
During collection of MWC flue gases,-the
trains were not traversed but rather set
up as single, average velocity points.
Each of the two average velocity points
sampled concurrently were considered to
replicate sampling velocity locations.
Each sampling train was leak-checked
prior to collection of the samples. The
stack gases were sampled for 240
minutes. The dynamic spiking system
injectors were started 30 minutes after
the start of the test and operated for a
total of 180 minutes. The MM5 trains
were disassembled, each section was
capped, and the components taken to a
clean area. Each appropriate section of
the sampling train was rinsed
sequentially with acetone and toluene
and the rinsates stored in amber glass
bottles with PTFE-lined screw caps.
Particulates (from the cyclone and filter)
were stored in bottles and wetted with
toluene. The XAD cartridge was sealed.
The field static spiking solution, 13C-
1,2,3,7,8-PeCDF was added to the XAD-2
resin prior to each test. The dynamic
spiking solution containing five 13C-
labeled PCDD/PCDFs (13C-1,2,3,4,7,8-
HxCDF, 13C-OCDF, 13C-1,2,3,4-TCDD,
i3C-1,2,3,7,8,9-HxCDD, and 13C-OCDD)
was continuously added throughout the
test using a dynamic spiking system
designed by MRI (Figure 2). The tests
were conducted using two spiking levels:
25 and 100 ng. Each level was spiked in
duplicate at each sampling location on
two of the test days and once on the
other test days. The test matrix is shown
in Table 1.
Each component of the MM5 sampling
train was individually spiked with a
Method Internal Standards mix (i.e.,
surrogates, i3C-2,3,7,8-TCDF, 13Q-
2,3,7,8-TCDD, i3C-1,2,3,7,8-PeCDD, and
i3C-1,2,3,4,6,7,8-HpCDD) prior to solvent
extraction. All solvent extractions were
conducted using toluene as the extraction
solvent. The solid samples (i.e., XAD-2
resin and combined particulates/filter)
were Soxhlet-extracted for 16 to 22 hr.
XAD-2 resin extracts with free water and
all front-half and back-half glassware
rinses were back-extracted using reagent
grade water. The sample extracts were
passed through sodium sulfate to remove
residual water. Each extract was
rotoevaporated to a final volume of 1-2
mL. A small volume (25 jiL) of tridecane
keeper was. added to each extract and
then further concentrated using a
nitrogen evaporator. Hexane was added
to the extracts to a final volume of 1 mL.
A 1 x 10 cm tapered column was slur-
ry-packed in hexane with 1 g of silica gel
followed by 4 g of 40% (w/w) sulfuric acid
modified silica gel. A second column (1 x
30 cm) was also slurry-packed, but with 6
g of acid alumina followed by 1 g of
sodium sulfate, also in hexane. The two
columns were set such that the silica gel
column drained directly into the acid
alumina column. The sample extract was
applied to the silica gel column and
drawn into the packing together with
three rinses of the concentrated extract
container. The silica gel column was
eluted with 45 mL hexane into the
alumina column, which was then eluted
with 20 mL hexane. A solution of 20%
(v/v) dichloromethane/hexane was used
to elute the alumina column (20 mL).
Finally, 20 mL of a 50% (v/v) dichloro-
methane/hexane solution was used to
elute the column. The three fractions
were collected separately and archived.
The 20% dichloromethane/hexane
fraction was concentrated to 2-3 mL,
transferred to a Reaction vial, amended
with 25 nL of tridecane, and the volume
reduced to 25 or 100 pL, depending on
the field spiking levels. This cleaned up
extract was spiked with a Recovery
Internal Standard solution (130-1,2,3,
6,7,8-HxCDD) and analyzed for
PCDD/PCDF by HRGC/HRMS.
A Carlo Erba MFC-500 gas
chromatograph was fitted with a 60-m x
0.25 pm i.d. DBS fused silica capillary
column using helium carrier gas (20-40
cm/sec). A Grob-type injector in the
splitless mode at 270 °C was used to
inject a 1 u,L portion of the sample
extract. After 2 min. at 200°C, the
temperature in the GC was increased to
220°C (5°C/min), held for 16 min,
increased again to 235°C (5°C/min) and
held for 7 min, and finally increased to
330 °C (5°C/min). The capillary column
was threaded directly into the source
chamber of a Kratos MS-50 high
-------
(Optional)
Potentiometer \ Filter
Quartz/Glass Liner
Thermocouple
Wozz/e
nan
Check
Valve
Reverse - Type
Pilot Tube
toe Bath
TIC TIC Fine Control
Valve
Q Condenser with Ice Water Jacket
Qp XAD Resin Cartridge with Ice Water Jacket
© Greenburg-Smith, WOmL of Double Distilled in Glass H2O
QO Modified Greenburg-Smith, lOOmL of Double Distilled in Glass H2O
© Modified Greenburg-Smith, Dry
Qy Modified Greenburg-Smith, SiOz
Figure 1. MM5 Sampling train configuration used during the study.
resolution mass spectrometer (transfer
line temperature of 280-300 °C) scanning
from 202-472 amu in one sec or less.
The resolution on the mass spectrometer
was set to 3,000. All samples and
standards were analyzed using the
selected ion monitoring (SIM) mode. A
three-point calibration curve was
analyzed and relative response factors
calculated. Daily calibration checks were
performed. Table 2 presents the ions
monitored during data acquisition.
Results and Discussion
Static Spike Recovery
Results for the recovery of the static
spike of a 13C-labeled congener added to
the XAD-2 resin prior to sample collection
are presented in Table 3. The recoveries
of the compound statically spiked onto
the XAD-2 resin were somewhat
disappointing in the light of the
recoveries observed during previous
studies.2 It is believed that the low
recoveries observed, especially for Runs
28 and 29, were anomalous and, in light
of the results presented below, do not
represent an accurate view of the
2 Mid west Research Institute (MRI). 1989.
"Validation of Emission Test Methods for PCDDs
and PCDFs," Final Report on WA No. 23 prepared
for Jimmy C. Pau, Atmospheric Research and
Exposure Assessment Laboratory, U.S. EPA,
Office of Air Quality Planning and Standards.
sampling accuracy of the MM5 sampling
train.
Dynamic Spike Recovery (Inlet)
Table 4 presents the percent recovery
of the compounds dynamically spiked at
25 ng into the sampling trains at the inlet
sampling ports (Runs 28, 29, 30 and 31).
The average of all total recovery values
for the individual congeners spiked into
each of the sampling trains analyzed was
95.1% (n = 30, 28% RSD). This average
value was obtained from two sets of
replicate sampling trains and two sets of
individual, non-replicate sampling trains.
The overall average indicates that the
sampling train quantitatively collected the
-------
Injector
Syringe Pump
5 mL Syringes
TFE Plunger
Luir Lock
316SS Needle
Sample
Box
Injector Ports on Top of
Bbow Angled Downward
TFE-faced 6 mm Cylindical
Septa Needle Tip at Bbow
Dual Probe
Single Pitot,
TIC
Vertical View
Sample]
Box
T
Console
Injector
(on bracket
above probe)
Bbow Injectors are
Pyrex Glass, Heat
Traced, Insulated,
Temperature
Controlled at 200°C
Dual Probe
Single Pitot, TIC
Injector (on bracket
above probe)
• Injector Ports on
Top of Bbow Angled
Downward TFE-faced
6 mm Cylindical
Septa Needle Tip at Bbow
Figure 2. Dynamic spiking system configuration relative to the MM5 sampling train.
dynamically spiked compounds added
during the sampling runs.
Runs 28 and 29 were conducted using
replicate sampling trains at both inlet
ports. Evaluation of the replicate data
from a qualitative point of view suggests
that for Run 28, reproducibility was good,
but for Run 29, the recoveries from Port
No. 1 were about one-half of that seen for
Port No. 2. It is suspected that during
Run 29, the syringe used for dynamic
spiking into Port No. 1 leaked at the Luer-
Lok fitting, thereby directing an
unascertained portion of the dynamic
spike into the sampling train. This could
not be confirmed from field observations,
but is a plausible explanation of the
results obtained. This conclusion is
supported by the results generated from
the analysis of native dioxins ancl furans.
-------
Table 1. Test Matrix for Field Sampling, Runs 28 to 33
Location
Inlet"
Outlet*
Test Day
1
2
3
4
5
6
Run No.
28
29
30
31
32
33
Port No. 1
/.c
L
L
H
H
H
Port No. 2
L
L
Hi
L
H
H
Port No. 3
L
L
L
HO
H
H
Port No. 4
L
L
H
L?
H
H
Blank
None
Outlet
None
None
Inlet
None
a The individual MM5 sampling trains at Ports No. 1 and 2 of the inlet sampling location
represent replicate trains sampling the flue gas simultaneously.
b The individual MM5 sampling trains at Ports No. 3 and 4 of the outlet sampling location
represent replicate trains sampling the flue gas simultaneously.
° L indicates that the dynamic and static spiking levels for this run were at the low level (i.e., 25
ng).
<*H indicates that the dynamic and static spiking levels for this run were at the high level (i.e.,
100 ng).
e These dynamic (but not the static) spiking levels were inadvertently switched by the field crew.
Thus, Port No. 3 was spiked at the low level, and Port No. 4 was spiked at the high level. The
effects of this overview were negligible.
Table 5 presents the results of the
recovery of the labeled compounds
dynamically spiked into the inlet
sampling trains at 100 ng (Runs 30, 31,
32, and 33). The average of all total
recovery values for the individual
congeners spiked into each sampling
train was 77.6% (n=30, 24% RSD). This
average value was obtained from two
sets of replicate sampling trains and two
sets of individual, non-replicate sampling
trains. The overall average for the
sampling trains dynamically spiked at
100 ng was approximately 20% lower
than those spiked at 25 ng.
Dynamic Spike Recovery
(Outlet)
The percent recovery results for
compounds dynamically spiked at 25 ng
into the MM5 trains in the outlet sampling
ports are presented in Table 6 (Runs 28,
29, 30 and 31). The average of all total
recovery values for the individual
congeners spiked into each of the
sampling trains analyzed was 112%
(n = 30, 12% RSD). This average value
was obtained from two sets of replicate
sampling trains and two individual, non-
replicate sampling trains. The overall
average indicates that the sampling trains
quantitatively collected the dynamically
spiked compounds added during the
sampling runs.
Table 7 shows the recoveries of the
labeled compounds dynamically spiked
into the outlet sampling trains at 100 ng.
The average of all sums of recovery
values for these sampling trains was
117% (n = 30, 13% RSD). This average
value was obtained from two sets of
replicate sampling trains and two
individual, non-replicate sampling trains.
The overall average indicates that the
sampling trains quantitatively collected
the dynamically spiked compounds
added during the sampling runs.
Native Dioxins and Furans
Figures 3 and 4 present the distribution
of the dynamically spiked components
and native dioxins and furans collected
during all of the runs for this study. The
bars represent the normalized overall
average of all compounds collected in all
sampling trains for each separate
component of the MM5 trains. The
distribution of spiked components and for
the native dioxins and furans for the inlet
sampling locations. are essentially the
same, with the distribution heavily
favoring the filter assembly. The
distribution in the outlet sampling trains
for the spiked compounds and for the
natives are also very similar, but in this
case, the distribution favored the filter
assembly and the XAD-2 trap almost
equally. This is expected, since
paniculate loading at the outlet sampling
port is decreased, thereby allowing
compounds in the vapor phase to break
through to the back part of the sampling
train. However, based on the High
percentage of compounds captured in
the filter assembly, it appears that
particulate loading at the outlet was
relatively high for this set of sampling
runs. Particulate material in the flue
adsorbs the gas phase dioxins and furans
(native or spiked) when it impinges on the
filter assembly, thereby effectively
trapping these compounds in the filter
assembly of the MM5 sampling train.
Statistical Analysis
A number of analysis of variance
models were run on the data generated
during this study. The variables used to
conduct these tests were run number
(demonstrative of day-to-day variations),
dynamic spike level, sampling location,
replicate train recoveries, and specific
dynamically spiked compounds. Several
analysis of variance models were
considered and, after each model, the
factors or combination of factors which
did not have a significant effect on the
result were removed and an additional
model fitted.
To test the null hypothesis that the
MM5 sampling trains did not produce
reproducible (precise) results, the
dynamic spike recovery was used as the
dependent variable. Differences between
-------
Table 2. Specific PCDDs and PCDFs Analyzed and Corresponding Ions Monitored During
HRGCIHRMS Analysis
Ion (relative intensity)
Compound
Statically spiked compound
13C-1,2,3,7,8-PeCDF
Primary
351.9005(100)
Secondary
349.9035(61)
Ratio Range
0.49-0.73
Dynamically spiked compounds
'3C-t,2.3,4,7,8-HxCDF 385.8615(100) 387.8586(87) 0.66-0.98
13C-OCOF 455.7806(100) 453.7386(87) 0.70-1.04
'3C-T,2.3,4-rCDD 333.9344(100) 331.9373(76) 0.61-0.91
'3C-r,2,3,7,8,9-HxCDD 401.8564(100) 403.8535(82) 0.66-0.98
"C-OCD0 471.7755(100) 469.7785(87) 0.70-1.04
N$tiV9 PCDDIPCDFs
2,3,7,8-TCDF 305.8987(100) 303.9016(76) 0.61-0.91
2.3,7,8-TCDD 321.8936(100) 319.8965(76) 0.71-0.91
1,2,3,7,8-PeCDF 339.8597(100) 337.8627(61) 0..49-0.73
1,2,3,7,8-PeCDD 355.8546(100) 353.8576(61) 0.49-0.73
2,3,4,6,7.8-HxCDF 373.8207(100) 375.8178(82) 0.66-0.98
1,2,3,7,B.9-HxCDD 389.8156(100) 391.8127(82) 0.66-0.98
1,2,3,4,6,7,3-HpCDF 407.7817(100) 409.7787(98) 0.78-1.18
1,2,3,4,6,7,8-HpCDD 423.7767(100) 425.7737(99) 0.79-1.19
OCDF 443.7398(100) 441.7428(87) 0.70-1.04
OCDD 459.7347(100) 457.7377(87-) 0.70-1.04
Surrogates (method Internal standards)
'3C-2,3,7,8-rCDF 317.9395(100) 315.9424(76) 0.61-0.91
'3C-2,3,7,8-rCDD 333.9343(100) 331.9373(76) 0.61-0.91
'3C-T,2,3,7,8-PeCDD 367.8954(100) 365.8984(61) 0.47-0.73
'3C-J.2,3,4,6,7,8-HpCDD 435.8175(100) 437.8145(99) 0.79-1.19
Internal standard (recovery internal standard)
'3C-r.2.3,6,7,8-HxCDD 401.8564(100) 403.8535(82) 0.66-0.98
the aggregate of all replicate sampling
trains were found to be nonsignificant
(P=0.103). This was also true for all but
one of the individual replicate runs (Run
29). The factor that was found to have the
most influence on sampling precision was
sampling location (i.e., inlet or outlet), and
day-to-day variability, the specific
dynamically spiked compound, and the
interaction between run and sampling
location had lower but approximately
equal influence. These variables
accounted for 74% of the variability
observed. These observations indicate
that for any given day, a pair of MM5
sampling trains sampling a particular
portion of a stack effluent for any given
compound will produce results of
acceptable precision.
The precision of the dynamic spike
sampling and analysis for each
compound was estimated and is
expressed as the pooled standard
deviation of the difference in recovery
between each set of replicates:
isc-HxCDF
13C-OCDF
13C-TCDD
130-HxCDD
13C-OCDD
12.3%
7.01%
8.27%
7.18%
5.77%
Thus, for 13C-HxCDF, the difference
between replicate sampling trains is
estimated to have a standard deviation of
12.3% recovery about two-thirds of the
time.
The absolute difference between the
dynamic spike recoveries and 100% was
used as the dependent variable for fitting
additional analysis of variance models.
Run number, sampling location,
replication, and the interaction between
run and sampling location significantly
affected the results in 40% of the
measurements. The variability in the
balance of the measurements may be
attributed to random fluctuations,
unknown variables, or to a skewed
distribution of the measurements. If no
variability had occurred, the predicted
difference between 100% and the actual
dynamic spike recovery was estimated to
be 18% (i.e., 82 to 118% recovery). This
range increased to somewhat less than
30% when all other factors influencing
variability were considered.
Conclusions and
Recommendations
The results generated during this study
indicate that the MM5 sampling train
provides quantitative and reproducible
measurements of dioxins and furans
under the conditions used in this study.
Accuracy of the measurements as
determined using dynamic spike
recoveries ranged from 77.6% to 117%.
Precision of the measurements was high
as evidenced by low relative percent
differences between replicate sampling
trains (with one exception) during the
measurement of dynamically spiked
dioxins and furans. The distribution of the
native and dynamically spiked dioxins
and furans within the sampling trains was
essentially the same.
-------
Table 3. Percent Recovery of Statically Spiked 13C-PeCDF on XAD Resin Component of MM5
Sampling Trains.
Percent
Sampling Recovery on
Run No. Spike Level (ng) Location Sampling Port XAD Resin
28 25
29 25
30 25
100
25
100
31 100
25
TOO3
25^
32 100
_
33 100
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
Inlet
Outlet
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
55.1
34.8
57.3
38.8
51.4
48.8
57.0
63.3
52.2
78.5
66.1
78.3
73.8
70.9
85.8
75.8
76.7
86.3
74.5
67.1
75.8
77.3
80.3
80.8
a The dynamic spiking levels on these samples were inadvertently switched during the test run,
but the static spike levels were as planned.
The dynamic spiking system is a useful
tool in the determination of the accuracy
of the measurements. It should be
considered as a viable alternative to
determine sampling accuracy and
precision during trial burns. Some
additional validation may be required to
determine if the dynamic spiking system
causes the spontaneous in situ formation
of additional PCDDs and PCDFs by
comparing replicate trains, one of which
has the dynamic spiking system.
It is recommended that during trial
burns, selected MM5 trains be equipped
with the dynamic spiking system in order
to assess the accuracy of the
measurements. This modification should
have minimal impact on sampling and
analysis costs.
-------
Table 4. Percent Recovery of Dynamically Spiked PCDDs and PCDFs Spiked at 25 ng at the
Inlet Sampling Locations
Average Total Relative Percent
Run No. 13C-Labeled Analyte Recovery (%) Difference
28*
29*
3Q<>
3-lb
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF
OCDF
TCDD
HxCDD
OCDD
125
108
126
92.6
111
80.1
69.5
72.7
64.2
71.8
83
109
102
74.4
95.3
137
122
94.9
92.0
126
16
2.0
19
11
4.0
76
64
50
64
73
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
'Duplicate sampling trains were collected and analyzed for Runs 28 and 29.
h Single sampling trains were collected and analyzed for Runs 30 and 31.
Table 5. Percent Recovery of Dynamically Spiked PCDDs and PCDFs Spiked at 100 ng at the
Met Sampling Locations
Average Total Relative Percent
Run No. 13C-Labeled Analyte Recovery (%) Difference
30«
37«
32"
33*»
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF
OCDF
TCDD
HxCDD
OCDD
102
80.4
91.9
83.2
95.1
114
103
99.7
97.7
120
70.9
52.5
65.5
58.1
60.9
80.7
61.6
84.7
63.8
71.8
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
30
6.0
11
23
11
2.0
1.0
5.0
2.0
1.0
"Single sampling trains were collected and analyzed for Runs 30 and 31.
b Duplicate sampling trains were collected and analyzed for Runs 32 and 33.
-------
Table 6. Percent Recovery of Dynamically Spiked PCDDs and PCDFs Spiked at 25 ng at the
Outlet Sampling Locations
Average Total Relative Percent
Run No. 13C-Labeled Analyte Recovery (%) Difference
28*
29*
30»
31"
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF --
OCDF
TCDD
HxCDD
OCDD
124
102
121
105
105
140
96.7
124
113
106
115
104
124
102
120
127
153
139
118
165
16
2.0
7.0
17
3.0
4.0
9.0
1.0
3.0
6.0
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
a Duplicate sampling trains were collected and analyzed for Runs 28 and 29.
h Single sampling trains were collected and analyzed for Runs 3C and 31.
Table 7. Percent Recovery of Dynamically Spiked PCDDs and PCDFs Spiked at 100 ng at the
Outlet Sampling Locations
Average Total Relative Percent
Run No. i3C-Labeled Analyte Recovery (%) Difference
30"
31"
32",
33"
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF
OCDF
TCDD
HxCDD
OCDD
HxCDF
OCDF
TCDD
HxCDD
OCDD
124
102
113
105
118
127
87.2
114
98.8
102
119
109
105
107
127
118
108
114
103
118
NA
NA
NA
NA
NA
NA
NA
NA
NA
NA
24
22
14
10
15
1.0
2.0
6.0
2.0
0.0
aSingle sampling trains were collected and analyzed for Runs 30 and 31.
>> Duplicate sampling trains were collected and analyzed for Runs 32 and 33.
-------
80%
Sampling Location
{Filter ggg Front Half Ui Back Half \ \XAD
Figure 3. Distribution of dynamically spiked compounds in the W/W5 sampling train.
60%
60%
I
1
30%
20%
ro%
n
OUTLET
Sampling Location
Filter gZ2 Front Half
Back Half
Figure 4. Distribution of native PCDDs and PCDFs in the MM5 sampling train.
10
-------
The EPA author, Jimmy C. Pau (also the EPA Project Officer, see below), is with
Atmospheric Research and Exposure Assessment Laboratory, Research
Triangle Park, NC 27711; and Andres A. Romeu, Marilyn Whitacre, and John T.
Coates, Jr.. are with Midwest Research Institute, Kansas City, MO 64110.
The complete report, entitled 'Validation of Emission Sampling and Analysis Test
Method for PCDDs and PCDFs," (Order No. PB90-235 847/AS; Cost: $17.00,
subject to change) will be available only from:
National Technibal 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
-------
?!
0) O
> m i
o o
•< D
Is
CD
(O
n>
CO
8
3
S
I
CD (S
g!
S Jjj
IS
3 ^
O 3
^. w
CD O
0) X
f §
w m
2S
i/fl
III
i 03 01
3^ o w
Q} D* (0
0.5 3
ake all nece
copy, and i
corner.
2 g
c S
S °
o to
Q. O
^ 3
W 3"
— . O
3 Q)
ff
•n ~
It
es
i
------- |