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

-------