United States
Environmental Protection
Agency
Atmospheric Research and
Exposure Assessment Laboratory
Research Triangte Park NC 27711
Research and Development
EPA/600/S3-90/007 June 1990
&EPA Project Summary
Validation of Emission Test
Method for PCDDs and PCDFs
Jimmy C. Pau, John T. Coates, Jr., Clarence L. Haile, Andres A. Romeu, and
H. Michael Molloy
The precision and accuracy of the
Modified Method Five (MM5) sampling
and analysis scheme for poly-
chlorinated dibenzo-p-dioxins and
dibenzofurans (PCDD/ PCDF) in mun-
icipal waste combustor (MWC) stack
gas have been determined. This was
accomplished using a dynamic
spiking system designed to contin-
uously deliver stable isotopic PCDD/
PCDF congeners into the MM5
sampling train upstream of the partic-
ulate filter during sampling of
incinerator stack gas. Field validation
tests to measure the recovery of
statically and dynamically spiked
PCDD/PCDF during stack gas sam-
pling indicated that the resin effec-
tively retained the static spike during
operating conditions. Dynamic spike
recoveries were inadequate, partially
due to the use of dichloromethane as
the extraction solvent. Toluene was
subsequently shown to be more ap-
propriate than dichloromethane for
obtaining enhanced recoveries of
PCDD/PCDF.
Experiments were conducted to
determine if increasing filter box
temperatures provoked increased
migration of the PCDD/PCDF towards
the back components of the MM5.
Back-half glassware was coated with
Apiezon-L grease to enhance PCDD/
PCDF recovery. Recovery of the static
spikes was quantitative. Dynamic
spike recovery was mostly greater
than 60%, but reproducibility was
inconsistent between some of the
replicate trains, possibly due to in-
consistencies in Apiezon-L coatings
or the inability to effectively remove
the Apiezon during extract cleanup.
Increasing the temperature of the
filter box resulted in a redistribution
of dynamically spiked PCDD/PCDF
toward the back components of the
MM5 at all sampling locations. The
native dioxins and furans did not
undergo analogous redistributions.
This Project Summary was devel-
oped 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 cart-
ridge, 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 stan-
dardized 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.
The ASME protocol has become a de
facto industry standard for sampling
stack gases for PCDDs, PCDFs, RGBs,
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), now the Atmospheric
Research and Exposure Assessment
Laboratory (AREAL). Since these
observations were based on limited
exporiments, 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 study was conducted in several
phases. During Phase 1, a dynamic
spiking system for continuously
introducing 13C-labeled PCDD/PCDF
congeners into the MM5 train during
sampling was designed and tested. The
dynamic spiking system was then tested
using MWC flue gas background to
determine method reproducibility and
accuracy during Phase 2 of the study.
Based on the results obtained during
Phase 2, additional laboratory tests were
conducted (Phase 3) to investigate the
influence of extraction solvent and carbon
content on PCDD and PCDF extraction
from MWC ash. Phase 4 used the
information gleaned from the previous
phases to investigate alternative
PCDD/PCDF recovery enhancement
procedures from the MM5 sampling train.
Finally, Phase 5 used the recovery
enhancement procedures to determine
precision and accuracy of modified MM5
sampling and analysis techniques with
MWC flue gas background. Table 1
summarizes the phases of this study and
identifies the MM5 sampling runs
associated to each phase. Because of the
complexity of the study, the phase-
specific methods utilized will be dis-
cussed concurrently with the results
noted at each phase. The following
procedures were used throughout the
study.
Table 1. Summary of the Investiga-
tion with Associated MM5
Sampling Runs
Phase 1
Phase 2
Run 1
2
3
Run 4
5
6
7
8
9
10
11
Laboratory tests of
the dynamic spiking
system
Laboratory tests to
demonstrate recovery
of PCDD/PCDF at two
spiking levels using
MM5 sampling train
Field tests to
determine MM5
sampling and analysis
method reprodu-
cibility and accuracy
using MWC flue gas
Phase 3
Phase 4 Run 12
13
14
15
16
17
Laboratory tests to
investigate the
influence of extraction
solvents and
unburned carbon on
extractability of
PCDD/PCDF from ash
Laboratory tests to
investigate sample
recovery enhance-
ment procedures for
the MM5 sampling
train
Phase 5 Run 18 Field tests to deter-
19 mine the distribution
20 of PCDD/PCDF within
21 the MM5 sampling
train for high filter box
temperature and to
determine the precis-
ion and accuracy of
the MM5 protocol
using sample recov-
ery enhancement
procedures studied in
Phase 4
The MM5 sampling train consists of
four main sections (Figure 1): a
nozzle/probe assembly (front-half
glassware), a heated filter assembly with
a cyclone for trapping particulates, an ice
water-chilled condenser (back-half glass-
ware) for trapping moisture, and an XAD-
2 resin cartridge. Each sampling train
was assembled and leak-checked prior to
collection of the samples. During collec-
tion of actual 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 be
replicate sampling locations. The ambient
air or stack gases were sampled for a
minimum of three hours and the train
leak-checked after the sampling run.
Posteriorly, the MM5 trains were dis-
assembled, each section capped, and the
components taken to a clean area. The
front-half and back-half glassware was
then rinsed with appropriate solvents and
the rinsates stored in bottles.
In the laboratory, only one of the
components from each of several sam-
pling trains was spiked with the Method
Internal Standards (MIS, i.e., surrogates)
prior to solvent extraction. This was done
for individual MM5 components which
were composited prior to analysis. For
those sampling trains which were
composited, a different component was
spiked with surrogates before extraction.
The solid samples (i.e., XAD-2 resin and
combined particulates/filter) were
Soxhlet-extracted for 16 to 22h. XAD-2
resin extracts and front-half and back-half
glassware rinses were back-extracted
using reagent grade water. The sample
extracts were passed through sodium
sulfate to effect extract drying. Each
extract was rotoevaporated to a final
volume of 1 to 2 mL. A small volume (25
nL) 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
slurry-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) methylene chloride/hexane was
used to elute the alumina column (20
mL). Finally, 20 mL of a 50% methylene
chloride/hexane solution was used to
elute the column and collected. The three
separate fractions were collected
separately and archived. The 20%
methylene chloride/ hexane fraction was
concentrated to 2 to 3 mL, transferred to
a reaction vial, amended with 25 yL of
tridecane and the volume reduced to 25
nL. This cleaned up extract was spiked
with recovery internal standards and
analyzed for PCDD/PCDF by GC/HRMS.
GC/MS conditions were as per Table
2. A three-point calibration curve was
analyzed and relative response factors
calculated. All sample and standards
-------
(Optional)
Potentiometer \ Filter
Quartz/Glass Liner
Thermocouple
Nozzle —
Reverse - Type
Pilot Tube
TIC
Check
Valve
TIC TIC Fine Control
Valve
u Condenser with Ice Water Jacket
.QQ XAD Resin Cartridge with Ice Water Jacket
(3) Greenburg-Smith, 100mL of Double Distilled in Glass H2O
QO Modified Greenburg-Smith, 700ml of Double Distilled in Glass H2O
Modified Greenburg-Smith, Dry
Modified Greenburg-Smilh, SiO2
Figure 1. MM5 sampling train configuration used during the study.
were analyzed in the selected ion
monitoring (SIM) mode. Table 3 presents
the ions monitored during data
acquisition.
Results and Discussion
Phase 1 - Development of a dynamic
spiking system — A dynamic spiking
system was designed to continuously
deliver the spike compounds at a
consistent and known rate in the vapor
phase during the 3- to 5-h duration of a
typical sampling run. Dual syringe
injection units were installed just ahead of
the MM5 filter box (Figure 2). The injec-
tion port was heated and located at an
angle to facilitate rapid vaporization and
mixing with influent sample gases.
Syringe pumps were used to inject
relatively small volumes of a con-
centrated dynamic spiking solution in
toluene. All system parts in contact with
dioxins or sample gas were in glass or
PTFE except for the stainless steel
syringe needle. The system was tested
under laboratory conditions to address
proper vaporization of the spiking
compounds and the use of toluene as the
solvent. A large glass preheater was
installed to maintain a stable temperature
at the injector. An MM5 train assembly
was connected to the dynamic injection
system and operated for 4h sampling
ambient air. A low spike (5ng) and a high
spike (500ng) were used. Each of the
PCDD/PCDFs dynamically spiked into
the system was extracted and analyzed
by GC/ECD.
Table 4 presents the results of the
initial laboratory tests. Results of the high
spike test indicate that recovery was
quantitative for each PCDD/PCDF spiked.
Furthermore, most of the dynamically
spiked compounds recovered were
impinged in the back-half glassware. The
low spike tests resulted in somewhat
lower recoveries of the dynamic spikes,
but the distribution of spiked compounds
also favored the back-half glassware.
However, some congeners were found in
notable amounts in the filter extracts. The
-------
Table 2. Operating Parameters for the
Capillary Column Gas Chro-
matographlc System for Analysis
of Tetra-Octa COD and CDF
Mass spectrometer
Gas chromatograph
Column
Liquid phase
Liquid phase
thickness
Carrier gas
Carrier gas velocity
Injector
Injector temperature
Injection volume
Initial column
temperature
Column temperature
program
Transfer line
temperature
Scan range
Scan time
Resolution
KRATOS MS-50
Carlo Erba MFC 500
60 n x 0.25 ID fused
silica
DB-5
0.25 ///n
Helium
20-40 c/n/s
"Grab" (splitless
mode)
270°C
200"C (2 min hold)
200-220 (5"C/min)
hold 16 min
220-235 (5"C/min)
hold 17 min
235-330 (5"C/min)
280-300°C
202-472 amu
0.5-1 s
3,000 or greater
data indicate that the development of the
dynamic spiking system was successful
and that it could be used as a field
spiking technique.
Phase 2 - Determination of dynamic
spiking method precision and accuracy -
Laboratory and field verification studies
were conducted to determine the
accuracy and precision of dynamic
spikes using the dynamic spiking system.
In the laboratory, replicate MM5 sampling
trains equipped with the dynamic spiking
system were assembled. The trains
sampled ambient air for 4h using the
dynamic spiking system at two spiking
levels: 5 and 500 ng (Runs 4 and 5). All
extracts were analyzed by GC/ECD and
in some cases, by GC/HRMS. Field tests
(Runs 6 to 11) were also conducted to
subject the dynamic spiking system to a
realistic sample matrix from a mass burn
waste combustpr burning municipal
refuse. In addition, a ISC-labeled
pentachlorinated dibenzofuran was
statically spiked onto the XAD-2 resin
prior to sample collection. This set of
tests used three static and dynamic
spiking levels: 5, 50, and 500 ng. All
sample extracts were analyzed by
HRGC/HRMS-SIM. The experimental
design is shown in Table 5. The design
provided for replication of one spike level
per each test day.
Analyses of the laboratory test runs of
the dynamic spiking system indicated
adequate (>60%) recovery of the spiked
compounds at both spiking levels (Table
6). The results also indicated that, in the
absence of particulate matter, gas phase
dioxins and furans were impinged in the
back half glassware of the MM5 sampling
train, specifically the condenser.
Field tests 'using realistic sample
matrices resulted in lower recoveries of
the dynamically spiked components
(average of 22% recovery) but high
recoveries of the static spike (Table 7).
The majority of the dynamic spike com-
pounds recovered using a realistic matrix
were found on the filter (Table 8).
Recoveries of method internal standards
(surrogates) indicated no analytical
problems during the analysis of the
samples.
Phase 3 - Laboratory tests to
investigate the influence of carbon
content and extraction solvent on PCDD
and PCDF extraction from MWC ash -
Two composite municipal refuse
incinerator fly ashes were augmented
with unburned carbon (carbon black)
such that the total carbon content was
approximately 10% and 20% more than
the original content. Duplicate extractions
and analyses were performed for each
ash (native and amended) using each of
three extraction solvents (dichloro-
methane, benzene and toluene). Prior to
extraction and during sample cleanup,
each ash aliquot was spiked with the
isotopically labeled PCDD and PCDF
congeners as indicated in Table 9. The
ashes were Soxhlet-extracted, cleaned
up, and analyzed for the spiked
compounds.
Average recoveries of 13C-
PCDD/PCDFs spiked directly into Ash
No.1 ranged from 13% to 84% using
dichloromethane, from 72% to 92% using
benzene, and from 69% to 100% using
toluene (Figure 3). Ash No.2 showed
average recoveries of 13C-PCDD/ PCDFs
spiked directly into the ash ranging from
5% to 78% using dichloromethane, 50%
to 83% using benzene, and 59% to 90%
using toluene (Figure 3). Since sample
preparation and analysis were completely
randomized across the three variables
under consideration (i.e., ash type,
percent carbon, and extraction solvent), it
appears that overall extractability of
dioxins and furans from ash is extraction
solvent and ash-dependent. For both
ashes, the labeled compounds added to
the cleanup columns were recovered
quantitatively, indicating no loss of I
material from cleanup to analysis.
Benzene and toluene were the morel
efficient extraction solvents, although the
addition of unburned carbon to the native
fly ashes resulted in an enhancement of
the recovery efficiency of spiked com-
pounds in the dichioromethane extracts.
A parallel effect was not noted for toluene
and benzene extractions. There was also
some specificity on the extraction effici-
ency based on the degree of chlorine |
substitution on the spiked compounds.
Phase 4 - Investigation of sample I
recovery enhancement procedures from
the MM5 sampling train - Since the
recovery of dynamically spiked ISC-
labeled dioxins and furans using
dichloromethane as the extraction solvent
during Phase 2 of the study were
unacceptably low, and because Phase 3
of the study indicated that dichlorome-
thane does not effectively extract dioxins
and furans from ashes, additional studies
were designed to test the hypothesis that
the use of an alternative solvent (i.e., I
toluene) would result in better recoveries
of spiked dioxins and furans. Further-
more, the results from Phase 1 of the
study indicated that, in the absence of
particulates, most of the spiked dioxins
and furans were impinged in the back-
half glassware. However, the results from
Phase 2 indicated that, in the presence of
particulates, most of the dioxins and
furans were impinged in the filter.
Therefore, in order to enhance dioxin and
furan recovery in the back-half glassware
of the sampling train, the back-half
glassware was coated with Apiezon-L |
chromatographic grease.
A total of six laboratory runs (Runs 121
to 17) were conducted by sampling
ambient air and exhaust gas from a I
kerosene combustor. Dynamic spiking
levels were 25 and 500 ng. The back-half
glassware of the MM5 sampling train was
coated with Apiezon-L for six of the MM5
sampling trains, and the other six trains)
used uncoated back- half glassware.
The results of the dynamically spiked I
labeled compounds are presented in
Table 10. Overall, the results indicate that
Apiezon-L effectively increased the
recovery of the ISC-labeled PCDD/PCDF|
dynamically spiked congeners.
Phase 5 - Determination of precision I
and accuracy using modified recovery \
procedures with a MWC flue gas I
background - The sample recovery
modifications investigated during Phase 4
and the dynamic spiking system were
used during a field test while sampling
incinerator flue gas laden with high levels
of particulate matter. Replicate runs were
-------
Table 3.
Characteristic Ions for Tetra-Octa COD/CDF Analysis
Ion (Relative Intensity)
Compound
Primary [M+2] Secondary [M] Ratio Range
Stable Isotopes (Field Spiked Analytes)
13C-1,2,3,7',8-PeCDFa
13C-1,2,3,4,7',8-HxCDFa
13C-OCDF<>
13C-1,2,3,4-TCDD>>
13C-1,2,3,7,8,9-HxCDD<>
13C-PCDD*
Native (Calibration Standard)
2,3,7,8-TCDF
2,3,7,8-TCD
1,2,3,7,8-PeCDF
1,2,3,7,8-PeCDD
2,3,4,6,7,8-HxCDF
1,2,3,7,8,9-HxCDD
1,2,3,4,6,7,8-HpCDF
1,2,3,4,6,7,8-HpCDD
OCDF
OCDD
Stable Isotopes (Method Internal Standard,)
13C-2,3,7,a-TCDF
13C-2,3,7,8-TCDD
13C-1,2,3,4,6,7,8-HpCDD
Recovery Internal Standard
13C-1,2,3,7,8,9-HxCDD
351.9005 (100)
385.8615 (100)
455.7806 (WO)"
333.9344 (100)
401.8564 (100)
471.7755 (100)°
305.8987 (100)
321.8936(100)
339.8597 (100)
355.8546 (100)
373.8207 (100)b
389.8156 (100)b
407.7817 (100)*>
423.7767 (100)b
443.7398 (100)
459.7347 (100)
317.9395 (100)
333.9343 (100)
435,8175 (100)>>
349.9035 Q61)
387.8586 (87)"
453.7836 (87)<*
331.9373 (76)
403.8535 (82)°
469.7785(87)^
303.9016 (76)
319.8965 (76)
337.8627 (61)
353.8576 (61)
375.8178 (82f
391.8127 (82f
409.7787 (98 )c
425.7737 (99)c
441.7428 (87)
457.7377 (87)
315.9424 (76)
331.9373 (76)
437.8145 (99)c
0.49-0.73
0.66-0.98
0.70-1.04
0.61-0.91
0.66-0.98
0.70-1.04
0.61-0.91
0.71-0.91
0.49-0.73
0.49-0.73
0.66-0.98
0.66-0.98
0.78-1.18
0,79-1.19
0.70-1.04
0.70-1.04
0.61-0.91
0.61-0.91
0.79-1.19
401.8564 (100)t> ' 403.8535 (82)° 0.66-0.98
3 = Quantitation was on the primary ion + secondary ion.
b = [M]* ^
d = [m+4]*
conducted at both inlet and outlet sam-
pling ports (relative to the electrostatic
precipitator), and each of the field ana-
lytes was dynamically spiked at two
levels (25 and 500 ng). In addition, the
filter box temperature was run at 250 °F
and at 420 °F for each spiking level at the
inlet and outlet locations to determine if
the filter box temperature affected the
distribution and impingement of spiked
and native PCDDs and PCDFs within the
MM5 sampling train. The incinerator was
the same as that used during Phase 2 of
this study. A total of 16 MM5 runs (Runs
18 to 21) were collected at both inlet and
outlet location in duplicate. The test
matrix is shown in Table 11. Toluene was
used as the solvent for rinsing all
glassware components (front-half and
back-half) after the sampling runs.
Static spike recoveries are presented
in Table 12. In most cases, the spike
recoveries were quantitative. Inlet static
spike recoveries were somewhat lower
when the filter box temperature was set
at 420 °F as compared to when the lower
• temperature was used. Overall, outlet
static spike recoveries were higher than
inlet spike recoveries, especially at the
higher temperature.
The results for the recovery of the
dynamic spikes collected at the inlet are
presented in Table 13. The average of
the sum of the dynamic spike recoveries
ranged from 49.5% to 70.6%. The 49.5%
average recovery was from .the high
spike, high filter box temperature
sampling train, in which one of the
replicate trains was lost due to a jeak.
The relative percent difference (RPD)
between each replicate sampling train
provides an indication of sampling
precision. An RPD value of 35% or less
was chosen arbitrarily to indicate
reproducible sampling of the stack mate-
rial. Overall, the majority of the replicate
results had less than 35% RPD (39 of 60
replicate measurements). For the low
temperature, low spike inlet samples, the
OCDF had poor RPD for each portion of
the sample train analyzed. At the high
temperature, low spike samples, poor
precision was accounted mostly by the
back-half and XAD subsamples. This
may reflect some difficulty in consistently
mobilizing the compounds trapped in the
cyclone/ filter by increasing the tempera-
ture of the filter box. The largest variabi-
lity between the replicates at the low
temperature, high spike samples was ob-
served on the cyclone/filter subsamples,
and of these, it was observed on the
-------
HxCDF, TCDD, and HxCDD. This may be
indicative of variable trapping of these
congeners on the cyclone/filter. No
evaluation of precision of the field
samples for the high temperature, high
spike sample trains can be made
because one of the replicate trains
leaked, invalidating its results.
Table 14 presents the results of the
recoveries of the dynamic spikes at the
outlet location. The average of the sums
of recovery for each sampling train
ranged from 62.9% to 107%.
Precision between the field replicates
of the outlet samples was lower than the
inlet samples. Fifty-eight of 80 replicate
values had RPDs exceeding 35%. For
Injector
Console
' Syringe Pump
5 mL Syringes
TFE Plunger
Luir Lock
316 SS Needle
§"• 'I Probe
lie Pilot,
Vertical View
Sample
Box
Injector Ports on Top of
Elbow Angled Downward
TFE-faced 6mm Cylindrical
Septa Needle Tip at Elbow
ITr^
Elbow Injectors are
Pyrex Glass, Heat
Traced, Insulated,
Temperature
Controlled at 200°C
Injector
(on bracket
above probe)
Dual Probe
Single Pilot, TIC
Injector (on bracket
above probe)
Injector Ports on
Top of Elbow Angled
Downward TFE-faced
6mm Cylindrical
Septa Needle Tip at Elbow
Figure 2. Dynamic spiking system configuration relative to the MM5 sampling train.
-------
Table 4. Recovery of 13C-PCDD/PCDFs from Preliminary Tests of
Dynamic Spiking System
Percent Recovery
Spike
Run Level
2 500
ng
3 5 ng
Spike
Compound
13C-TCDD
13C-HxCDD
13C-HxCDF
13C-OCDF
13C-TCDD
13C-HxCDD
13C-HxCDF
13C-OCDF
FH Rinse
NO
ND
<1
<1
ND
ND
ND
Filter
8
1
<1
5
31
13
BH Rinse
87
108
113
113
56
51
76
XAD
4
4
4
4
*
8
3
Sum of
Recoveries
99
113
117
121
64
82
92
* Chromatographic interference.
Table 5. Experimental Design for
Initial Field Evaluation
Testing
Field Spiking Level* for Test Run
Stack Number:
Location 6'
8
10 ' 11
1 L Lb Mb Mb H H
2 L° M Mb H H L
3 M Lb H L M &
H H
M
a Field static and dynamic spiking levels:
L = low (5 ng); M = medium (50 ng);
and H = high (500 ng).
b The MM5 train components of these test
runs were analyzed individually.
o The XAD spike (static) was 50 ng
instead of 5 ng.
d The XAD spike (static) ' was 25 ng
instead of 500 ng.
the low spike, low filter box temperature
run, the relative percent difference
between the replicate recoveries of
almost every congener collected in the
front half, back half, and XAD exceeded
35%. All of the samples from the first
replicate train of this run had high
amounts recovered on the back-half and
low on the XAD, whereas the second
replicate train samples exhibited the
opposite, namely higher recoveries in the
XAD than in the back half. This may be
the result of differential coating of the
Apiezon in both trains or of the inability to
totally remove the Apiezon from the
extracts during cleanup. For the sampling
trains from the low spike, high filter box
temperature run, a large degree of
variability can be observed in the rep-
licate front component recoveries. How-
ever, the actual sample concentrations
were close to the detection limits, so that
random variability can be important. For
the high spike, low filter box temperature
run, significant variability was encoun-
tered for all congeners from the front
components.
The average recoveries for each
subsample in the train were normalized
to the sum of these recoveries and
plotted as stacked bar plots in Figure 4
for the dynamic spikes'at the inlet and
outlet. This plot aids in determining the
distribution of the spiked compounds in
the MM5 sampling train components. For
each set of congeners, consecutive
stacked bars represent low and high filter
box temperatures, respectively.
For the dynamically spiked com-
pounds collected at the inlet, there was a
change in the overall distribution of each
one favoring the back part (back half plus
XAD) of the sampling train at the higher
filter box temperature. This was observed
for both the low and high spikes. This
suggests that increasing the filter box
temperature produced less adsorptivity of
spiked dioxins and furans in the cyclone/
filter assembly. The dynamic spike
recoveries for the outlet location show
that, as was the case for the inlet
sampling locations, the higher filter box
temperature increased the relative
percentage. of material being trapped in
the back-half train components, of the
sampling train. Note that at low spike
levels, the high filter box temperature
decreased the amount of ISC-labeled
dioxin and furan congeners being trapped
on the XAD. relative to that collected in
the back-half glassware, while at high
spike level, a relatively greater proportion
of the compounds were trapped in the
XAD with high filter box temperatures. A
potential explanation for this was
described above, related to differential
Apiezon-L coating. on the trains or
residual Apiezon in the sample extracts,
and this is not reflected in the bars.
The concentrations of each native
dioxin and furan congener in each portion
of the sampling train were normalized to
their sum and plotted in stacked bar
graphs similar to those discussed above.
These are shown in Figure 5. Increasing
the filter box temperature had little or no
mobilization effect on the native dioxins
and furans collected at the inlet and
outlet locations. This is in stark contrast
to the ISC-labeled dynamic spikes,
whereby significantly lower adsorption of
these compounds in the front parts of the
train was achieved by raising the
temperature of the filter box. Furans were
more easily mobilized by increasing the
filter box temperature than dioxins. This
would suggest that the native dioxins and
furans are differentially associated with
the particles. The difference between the
dioxins and furans likely reflects
differences in their chemical structure
and ability to sorb to particles.
All the evidence points to a previous
association between flue particles and the
native compounds that is dissimilar to
that for the spiked compounds with the
same particles. The residence time of the
natives in the .incinerator system (from
incineration to sampling port) is in the
order of 10s or less, whereas the resi-
dence time for the dynamic spikes is
about 2s. Kinetic considerations indicate
that this is not sufficient time to establish
the associations ' between the particles
and the native dioxins and furans. The
implication is therefore that the associa-
tion was established prior to incineration,
between the precursors of the solid ash
particles and the precursors of the native
dioxins and furans.. This suggestion is
clearly speculative and would require
experiments to test the hypothesis.
Conclusions and .
Recommendatiqns
• Results from this investigation have
demonstrated that the dynamic spiking
system developed is appropriate for
field use. Given the proven suitability
of the .technique for PCDDs/PCDFs,
. consideration should be given to
expanding its application to other
semivolatile organic compounds of
interest.
-------
too
&• 60
8 40
CD
C
20
0
(a) Ash No. 1, 6.5% carbon
OCDF
7CDD
HxCDD
OCDD
90
80
70
o 50
8 40
£ 30
20
70
(a) Ash No. 2, 13% carbon
ZL I
TCDD OCDD
OCDF HxCDD
enzene
KJj:}JM!t!l
Toulene
(b) Ash No. 1, 76.5% carbon
OCDF HxCDD
OCDD
« 50
(b) Ash No. 2, 23% carbon
HxCDF
TCDD
OCDF HxCDD
OCDD
(c) Ash No. 1 26.5% carbon
720
HxCDD
OCDD
100
&• 60
8 40
20
0"
(c) Ash No. 2, 33% carbon
HxC
1
s
i
§
!
'DFo
"
%
I
1
m
'•>
f
'rcbo
VDF HxC
m
i
K
I
T
__
I
DCM
Benzene
Toulene
' OCDD '•
:DD ;
Figure 3. Recovery of l3C-labeled dioxins and furans spiked into Ash No. 1 and No. 2: (a) Ash No. 1 unamended; (b) Ash No. 1 amended
with 70% carbon; (c) Ash No. 1 amended with 20% carbon; (d) Ash No. 2, unamended; (e) Ash No. 2, amended with 10%
carbon; (f) Ash No. 3, amended with 20% carbon..
10
-------
Table 10. Recovery of 130-Labeled PCDDs and PCDFs Dynamically Spiked into MM5 Trains Apiezon-L
Coated and Uncoated Back-Half Glassware
Runs 12 and 13s
Runs 14 and 15>>
Runs 16 and It7b
Analyte
Coated Train0
13C-1,2,3,4,7&HxCDF
13C-OCDF
13C-1,2,3,4-TCDD
13C-1, 2,3,7, 8,9-HxCDD
Uncoated Train
13C-1, 2,3,4,7, 8-HxCDF
13C-OCDF
13C-1,2,3,4^TCDD
13C-1,2,3,7,8,9-HxCDD
25-ng
Back
Half
129
37.2
68.2
82.6
18.9
• 16.4
8.1
12.1
Spike
XAD
:NA
NA
NA
NA
30.3
15.5
NA
20.4
500-ng
Back
Half
133
90:6
66J8
89:5
35.3
26.5
15.6
21.6
Spike
XAD
NA
:NA
:NA
NA
21.2
12.3
NA
14.5
25-ng
Back
Half
113
21.7
86.8
74.2
14.0
6.1
7.3
10.0
Spike
XAD
:2.8
03
NA
.2.8
57.7
31.9
NA
49.3
500-ng
•Back
Half
212.1
58.6
118
156
NA
NA
NA
NA
Spike
XAD
86.1
0.9
NA
1'.2
37.4
19.8
NA
29.4
25-ng
•Back
Half
NA
37.2
NA
74.2
20.6
21.3
NA
20.7
Spike
XAD
61(8
NA
NA
NA
34.3
22.8
NA
30.2
500-ing
Back
.Half
NA
47.2
NA
61.5
22.5
13.6
NA
17.0
Spike
XAD
NA
NA
NA
25 9
13.6
NA
20.3
NA = not analyzed.
a Flue gas from lab air.
b Flue gas from kerosene combustor.
0 Back-half glassware was coated with Apiezon-L.
Table 11. Test Matrix for Second
Field Validation Test for
Recovery of Polychlorin-
ated Dibenzo-p-Dioxins and
Dibenzofurans
Location
7esf
Day
1
2
3
4
Port
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
Inlet
Spike/F.B.
Temp.a
25/250
25/250
25/420
25/420
500/420
500/420
500/250
500/250
Outlet
Spike/F.B.
Temp.a
25/250
25/250
25/420
25/420
500/250
500/250
500/420
500/420
Blank
Train
outlet
inlet
Table 12. Percent Recovery of Static
Spikea on XAD-2 Resin
. During Second Field
Validation Test
Percent Recovery on
XAD Resin
Sp*e Filter Box Filter Box
Sampling Level Temp. Temp.
Location (ng) 250°F 420°F
Inlet
Outlet
25
500
25
500
99.4
107
89.0
77.7
101
84.3
.86.6
93.4
78.6
81.5
71.9
97.5
92.4
44.3
92.6
a T3C-1,2,3,7,8-pentachlorodibenzofuran.
1 Spike designates dynamic spike level
(ng), and F.B. designates filter box.
11
-------
Table 13. Percent Recovery of Dynamic Spikes from the Inlet Sampling Location MM5 Sampling Trains During Second Field
Validation Test
SpJte 73C- Filter Box Temperature = 250° F Filter Box Temperature = 420" F
Level Labeled
(ng) Analyte
25 HxCDF
OCDF
TCDD
HxCDD
OCDD
500 HxCDF
OCDF
TCDD
HxCDD
OCDD
Cyclone/
Filter
Average"
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
47.1
28.9
44.9
73.1
37.6
12.0
42.5
24.9
56.1
32.4
53.0
60.0
55.9
12.7
44.5
61.6
50.7
66.5
67.5
28.6
Front
Half
19.4
6.72
20.5
39.6
18.4
11.4
17.7
6.23
24.6
20.0
4.95
11.9
9.77
29.2
4.18
31.9
4.84
4.13
10.0
25.0
Back
Half
1.94
20.6
0.345
82.7
4.13
25.9
2.26
6.62
0.576
85.1
7.98
44.9
3.22
24.8
11.6
21.6
4.47
45.0
2.65
1.89
Sum of Cyclone/
XAD Recov. Filter
3.57
9.52
0.933
46.4
7.34
18.5
2.69
5.95
0.822
7.30
x =
S =
RSD =
1.58
18.4
0.557
33.9
2.09
. 6.24
1.04
23.6
0.421
45.3
x =
S =
RSD =
72.0
66.6
67.4
65.1
82.0
70.6
13.5
19.1%
67.5
69.4
62.3
61.0
80.5
68.1
15.1
22.2%
32.4
17.3
48.4
9.9
26.0
39.2
36.4
5.78
55.0
7.83
25.4
NA"
45.8
NA
20.7
NA
24.9
NA
49.4
NA
Front
Half
2.94
6.80
3.96
4.04
2.74
32.8
3.34
0.90
4.37
2.52
2.61
-. NA
4.79
NA
2.38
NA
2.41
NA
4.90
NA
Back
Half
28.3
88.3
12.3
59.1
24.2-
24.0
18.9
56.8
8.69
69.3
11.9
NA
6.35
NA
21.4
NA
8.08
NA
5.38
NA
XAD
3.73
135
2.97
159
5.19
80.5
4.81
87.3
1.71
200
~x
S
'RSD
2.73
NA
1.79
NA
3.31
NA
1.92
NA
1.53
NA
x
S
RSD
Sum of
Recov.
67.4
67.6
58.1
63.3
69.7
= 65.2
= 8.44
= 72.9%
42.6
58.7
47.8
37.3
'NA
61.2
= 49.5
= 10.3
= 20.7%
a Average recovery of two replicate sampling points.
b Replicate MM5 train leaked during sampling and was therefore not analyzed.
12
-------
Table 14. Percent Recovery of Dynamic Spikes from the Outlet Sampling Location MM5 Sampling Trains During Second
Field Validation Test
Spike 13C- Filter Box Temperature = 250°F Filter Box Temperature = 420°F
Level Labeled
(ng) Analyte
25 HxCDF
OCDF
TCDD
HxCDD
OCDD
500 HxCDF
OCDF
TCDD
HxCDD
OCDD
Average3
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
Average
RPD
Cyclone/
Filter
1.29
0.00
3.42
7.91
0.827
17.3
2.22
13.1
3.96
23.2
6.63
49.5
19.3
69.1
5.42
41.3
7.41
57.2
21.9
74.9
Front
Half
0.358
57.5
0.707
117
0.124
102
1.19
11.8
0.791
121
0.929
54.2
2.81
86.8
0.677
46.1
0.983
56.5
3.01
80.4
Back
Half
46.0
72.0
37.4
66.3
38.0
70.4
37.3
65.8
34.9
70.3
56.7
37.0
57.4
33.1
59.5
32.3
51.0
41.4
53.9
40.1
XAD
28
73.8
18.2
126
34.0
108
22.9
73.1
18.7
108
X
S
RSD
10.5
58.0
8.14
3.19
12.6
42.7
8.72
50.1
8.32
7.45
X
S
RSD
Sum of Cyclone/
Recov. Filter
75.5
59.7
72.9
63.5
58.3
= 66.0
= 8.43
= 12.8%
74.7
87.6
78.2
68.1
87.1
= 79.2
= 9.19
= 11.6%
1.85
83.2
3.06
90.2
2.03
77.5
2.46
57.7
4.01
93.8
3.00
35.1
3.94
23.9
2.71
29.5
3.14
42.4
3.76
68.4
Front
Half
0.382
200
0.498
200
0.505
200
1.06
61.0
0.615
200
0.128
134
0.267
152
0.0940
104
0.202
96.0
0.275
154
Back
Half
65.3
47.2
97.3
28.3
58.1
43.4
63.7
40.2
95.8
21.3
46.5
0.22
45.0
63.1
44.7
6.94
43.8
0.91
44.4
67.4
XAD
24.8
24.2
33.2
3.92
23.9
2.94
24.8
rs.2
32.85
4.58
~
S
RSD
17.3
125
8.75
74.4
20.5
110
16.9
90.7
9.57
13.2
X
S
RSD
Sum of
Recov.
92.3
134.0
84.5
92.0
133.2
= 107
= 25.6
= 23.8%
66.8
58.0
67.9
64.0
57.9
= 62.9
= 73.3
= 20.9%
a Average recovery of two replicate sampling points.
13
-------
(a) Inlet Recovery of 13C Dynamic Spike
High Spike (Low/High Temp)
HxCDF OCDF TCDD HxCDD OCDD
(b) Inlet Recovery of 13C Dynamic Spike
Low Spike (Low/High Temp)
b1-2
2 1
o
CD 0.8
o:
•o 0.6
I 0.4
E
0.2
Cycl/Fil
Front Half
¥>.'*-f;Hl%\
Back Half
XAD
HxCDF OCDF TCDD HxCDD OCDD
C\J — CO (O •<)• (V O
»- ci c> o c>
A/GAOoey paziieuuoN
(o) Outlet Recovery of 13C Dynamic Spike
High Spike (LowiHigh Temp)
-
—
—
m
—
-
—
-
—
-
mm
HxCDF OCDF TCDD HxCDD OCDD
Cycl/Fil
Front Half
i >
Back Half
XAD
Normalized Recovery
p p p p -*
o Ko ^ en co -* KJ
(d) Outlet Recovery of 13C Dynamic Spike
Low Spike (Low/High Temp)
-.
—
__
—
„
<
—
r~
"™
—
—
*< '_
HxCDF OCDF TCDD HxCDD OCDD
Cycl/Fil
Front Half
Back Half
XAD
Figure 4. Normalized recoveries of ISC-labeled dynamically spiked compounds: (a) High spike (500 mg), inlet sampling location; (b) Low
spike (25 mg), inlet sampling location; (c) High Spike (500 mg), outlet sampling location; (d) Low spike (25 mg), outlet location.
Alternating bars represent low and high filter box temperature recoveries, respectively.
14
-------
(a) Inlet Native Furans
(Low/High Temperature)
75 0.6
12 0.4
o
o 0.2
I -
2,3,7,8 PENT A HEPTA
TETRA HEXA OCTA
Furan
(b) Inlet Native Dioxins
(Low/High Temperature)
',3,7,8 PENTA HEPTA
TETRA HEXA OCTA
Dioxin
(c) Outlet Native Furans
(Low/High Temperature)
(d) Outlet Native Dioxins
(Low/High Temperature)
2,3,7,8 PENTA HEPTA
TETRA HEXA OCTA
Furan
2,3,7,8 PENTA HEPTA
TETRA HEXA OCTA
Dioxin
Figure 5. Normalized recoveries of native dioxins and furans: (a) inlet furans; (b) inlet dioxins; (c) outlet furans; (d) outlet-dioxins. Low
and high dynamic spike run results were combined; alternating bars represent low and high filter box temperature recoveries,
respectively.
15
-------
Jimmy C. Pau is with Atmospheric Research and Exposure Assessment
Laboratory, Research Triangle Park, NC 27711. John T. Coates, Jr.,
Clarence L. Haile, Andres A. Romeu, and H. Michael Molloy are with
Midwest Research Institute, Kansas City, MO 64110.
Jimmy C. Pau is the EPA Project Officer (see below).
The complete report, entitled 'Validation of Emission Test Method For PCDDs
and PCDFs," (Order No. PB 90-187 246/AS; Cost: $53.00, 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
Environmental Protection
Agency
Center for Environmental Research
Information
Cincinnati OH 45268
Official Business
Penalty for Private Use $300
EPA/600/S3-90/007
------- |