Guidelines Cor Stac''. Testinq a
Waste Combustion Facilities
t Municip
inaJ
Midwest Researo
:h Inst., Kansas
City.
MO
Prepared for
Environmental Protection Agency
Research Tr xang
V.C* -A- - - ~
riangle Pa^,
Jun 38
U S. Department of Cwwwee
tetiwwi Technical b&Tnw6«* Service
' : U "
-------
TECHNICAL REPORT DATA
{rtecse retd (tua^catna on Ihe rtttne be.\*t comptetintf
1. REPORT NO- 3.
EPA/60J/S-S8/085
3. RECIPIENT'S ACCESSION NO.
4. TITL£ AND SUBTITLE
Guidelines for Stack Testing at Municipal Waste
Combustion Facilities
6. REPORT DATE
June 1988
t. PERFORMING ORGANIZATION CODE
7. AUTHORlSI , „ ' '
C. L. Haile (MR1) and J. C. L. rris (A. D. Little,
Inc.)*
t. PERFORMING ORGANIZATION REPORT NO.
8985-K (04)
». PERF OAMIKC OROANIZATICS NAMl AND ADDRESS
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 64110
10. PROGRAM ElEMEWT NO.
11. CONTRACT/GRANT NO
66-02-4395, Task 4; and
68-02-3999. Task 13
12. SPONSORING agency name ano address
EPA, Office of Research and Development
Air and Energy Engineering Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OP REPORT ANO PERIOO COVERED
Task Final; 10/87 - 3/88
14. SPONSORING AfiENCV CODE
EPA/600/13
is.sup*lementarv notes ^££j^l project officer is Sharon L. .Nolen, Mail Drop 61, 919/
541-7607. (*) A. D. Little, Inc., 15 Acorn Park, Cambridge, MA 02140.
«. abstract rep0r: gives guidance for stack testing at municipal waste combustion
(MWC) facilities. State and local environmental agencies have been required to de-
velop regulations for MWCs in response to a significant expansion in the number of
MWC facilities currently in operation or in some phase of planning and construc-
tion. As part of this regulatory process, permitting agencies need guidance on
identifying critical emission parameters and on methods that can be used to mea-
sure tho^e parameters, expecially toxic substances. The Northeast States for Coor-
dinated Air Use Management (NESCAU.M), with support from EPA's Control Tech-
nology Center, organized a work group of recognized experts to provide specific
guidance cn methods for measuring stack emissions ffom MWCs and reporting for-
mats to facilitate standardization of emissions tests. This methods guidelines re-
port is based on the discussions of the NESCAVM work groups Specific testing
recommendations, emissions, measurement methods, and recommended reporting
formats are presented. These recommendations are not intended to prescribe reg-
ulatory requirements, but to recommend emissions test programs responsive to
typical requirements. The measurement methods presented ape based primarily
on reference methods already in widespread use..
IT. KEY WOROS AND DOCUMENT ANALYSIS ,
» OESCR'PTORS
b.lOENTI*iERS/OPEN endeo terms
c. COSATi Field'GfOup
Pollution Measurement
Tests Licenses
Emission
Flue Gases
Wastes
Combustion
Pollution Control
Stationary Sources
Municipal Waste Com-
bustion
13B
14 B 05D
14G
21B
18. DlS TBI Bu<-' OS STATEMENT
Release to Public
is security class «iCE
CPA Form 2220-1
-------
PB89-2 3^d93
EPA/600/8-33/085
Jane 1988
GUIDELINES FOR STACK TESTING AT
MUNICIPAL WASTE COMBUSTION FACILITIES
by
Clarence L. Haile
Midwest Research Institute
425 Volker Boulevard
Kansas City, Missouri 64110
and
Judith C. Harris
Arthur D. Little, Inc.
15 Acorn Park
Cambridge, Massachusetts 02140
EPA Contracts No. 68-02-4395
Work Assignment No. 4;
and 63-02-3999,
Work Assignment No. 13
Project Officers
Sharon L. Nolen
Air and Energy Engineering Research Laoo-story
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
and
Nancy L. Seidman
Northeast States for Coordinated Air Use Management
85 Merrimac Street
Boston, Massachusetts 02114
AIR AKD ENERGY ENGINEERING RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NC 27711
-------
NOTICE
This document has been reviewed in accordance vtith
U.S. Environnental Protection Agency policy aw!
approved for publication. Mention of trade naaes
or commercial products does not constitute endorse-
Bi*n: or recoirsaendation for use.
-------
PREFACE
This report was prepared under EPA Contract Ho. 68-02-4395, Work Assign-
ment 4. The objective of this work assignment was to assist the U.S. Envi-
ronmental Protection Agency (EPA) Control Technology Center (CTC) and the
Northeast States for Coordinated Air Use Manageae'it (NESCAtiM) in developing a
methods guidance document for testing municipal waste combustors {MWCs).
A Work Group was organized by NESCAUM to address issues in MWC testing.
The overall goal of the Work Group Is to provide state and local agencies
permitting MWCs agencies with guidance on testing methods and reporting
formats. The specific objectives of the Work Group are as follows:
1. Develop standardized MWC stack test methods;
2. Develop standardized reporting formats for MWC test results;
3. Identify unresolved stack testing technical Issues; and
4. Promote exchange of technical Information on MWC stack testing.
This methods document is the product of the Work Group's efforts to
achieving these objectives.
(Revised 7/14/88)
-------
CONTENTS
PREFACE
ACKNOWLEDGMENTS*. 1
1.0 INTRODUCTION
2.0 TEST REC0M4EN0ATI0NS 2
2.1 COMPLIANCE TESTING 3
2.2 PERFORMANCE TESTING 4
2.3 RESEARCH AND DEVELOPMENT TESTING 5
2.4 SPECIAL PURPOSE TEST PARAMETERS 6
3.0 MEASUREMENT METHODS FOR COMPLIANCE/PERFORMANCE TESTING 6
3.1 POLYCHLORINATED DIBENZQ-o-DIOXINS AND DIBENZGFURANS... 6
3.1.1 Method Description.... 6
3.1.2 Critical Features 12
3.1.3 Quality Assurance and Quality Control Procedures. 16
3.1.4 Oata Quality Objectives 17
3.2 PARTICULATE MATTER 18
3.2.1 Method Description 18
3.2.2 Critical Features 13
3.2.3 Quality Assurance and Quality Control Procedures. 20
3.2.4 Oata Quality Objectives 21
3.3 METALS 21
3.3.1 Lead. Cadmium, and Total Chromium 21
3.3.2 Mercury 24
3.4 COMBUSTION GASES--CAPBCN MONOXIDE, CARBON DIOXIDE,
AND OXYGEN 25
3.4.1 Method Description 25
3.4.2 Critical Features 26
3.4.3 Quality Assurance and Quality Control Procedures. 26
3.4.4 Data Quality Objectives 27
3.5 TOTAL HYDROCARBONS (THC) , 27
3.5.1 Method Description..... 27
3.5.2 Critical Features 27
3.5.3 Quality Assurance and Quality Control Procedures. 28
3.5.4 Data Quality Objectives i 28
3.6 OTHER GASES 23
3.6.1 Hydrochloric Acid 28
3.6.2 Nitrogen Oxides 1 29
3.6.3 Sulfur Dioxide 30
Page
ill
viil
1
-------
CONTENTS (continued)
Page
4.0 RESULTS REPORTING * 32
4.1 OVERVIEW 32
4.2 RESULTS TO BE REPORTED . 32
4.2.1 Facility Operating Status 32
4.2.2 Combustion Process Data 33
4.2.3 Permit Conditions 33
4.2.4 Pollutant Emission Data 33
4.2.5 Conversion Factors 36
4.3 OATA TABULATION FORMATS 38
4.4 OATA QUALITY RESULTS REPORTING 38
4.5 REPORTING "NOT DETECTED" VALUES 38
5.0 REFERENCES 38
6.0 ADDITIONAL SOURCES OF INFORMATION 44
APPENOIX A ADDITIONAL MEASUREMENT METHODS FOR RESEARCH TESTING A-I
APPENDIX B SELECTED DETAILED PROCEDURES FOR MEASUREMENT OF PCDOs
AND PCDFs IN MWC STACK EMISSIONS B-l
vi
-------
FIGURES
Pace
3-1 Modified method 5 sampling train (from Method 0100,
SW-846) 8
3-2 Combined condenser - adsorbent cartridge unit, (from
Method 0010, SW-846) 9
3-3 Separate condenser and adsorbent trap units train.
(from Stanley, J. S., et al., EPA Report No. 560/5-82-014) 10
3-4 Example condenser and XA0 resin cartridge designed in
impinger form 14
3-5 Method 5 particulate train. (Code of Federal Regulations,
Title 40, Part 60, Appendix A) 19
A-l Schematic of the volatile organic sampling train (VCST) (from
Method 0030, SW-846) A-12
A-2 VOST system setup (from Method 0030, SW-846) A-13
A-3 VOST trap assembly (EPA 600/9-34-019), November 1984,
Proceedings: National Symposium on Recent Advances in Pollutant
Monitoring of Ambient Air and Stationary Sources, pp. 171-179,
"Volatile Organic Sampling Train (VOST) Development at MR I by
Fred Bergman" A-14
TABLES
Pace
3-1 List of analytes, method internal standards, surrogates,
and recovery internal standards for dioxin/furan analysis 13
3-2 Example target limits of detection 18
4-1 MWC facility operating data 34
4-2 Pro forma table for reporting permit conditions and test
results 35
4-3 Pro forma table for test conditions data 37
4-4 Pro forma table for conventional emission parameter 39
4-5 Example pro forma table for dioxins and furans emission data 40
4-6 Example pro forma table for metals emission data 41
4-7 Pro forma table for other organics emission data 42
4-8 Example pro forma table for data quality results 43
A-l Calibration compounds contained in PC8 standard solutions A-4
A-2 Example target limits of detection A-6
A-3 Example target limits of detection A-10
3-1 Example gas chromatographic column condition for
2,3,7,8-substituted PCDO and PCDF congeners 8-5
3-2 List of analytes, method internal standards, surrogates, and
recovery internal standards for dioxin/fu>-ar analysis 8-6
vli
-------
ACKNOWLEDGMENTS
This methods docurent is based on discussions by a Work Group organized
by the Northeast States for Coordinated A1r Use Management (NESCAUM). The
Work Group was created to provide State and Local Agencies responsible for
permitting municipal waste combustors (MWCs) with guidance on testing methods
and formats for reporting test results. Nancy L. Seidman, Special Projects
Director for NESCAUM, served as Project Manager. The Work Group- members were
drawn from State and Lccal Agencies, Federal Agencies in the United States and
Canada, firms operating MWC facilities, and liaisons from professional
organizations. The meibers of the Work Group are listed below.
American Society of Mechanical Engineers, Codes and Standards- Arlene
Cpadafino
American Society for Testing and Materials - Charles Velzy
California Air Resources Board - Peter Ouchida, Dean Simeroth
Connecticut Department of Environmental Protection - Len Bruckman, George
Miller
Environment Canada - P. K. Leung
Minnesota Pollution Control Agency - Lori Bartels, Carolina Schutt
New York Departaient of Environmental Conservation - A1 Columbus, Robert
Kerr
NESCAUM - Michael Bradley, Nancy Seidman
Ogden Projects, Ire. - Jeff Hahn, Henry Von Oem Fange
South Coast Air Q-'ality Management Oistrict - John Higuchi
U.S. Environmental Protection Agency
Control Technology Center - Sharon Nolen
Office of Air Quality Planning and Standards - George Walsh, Gary
McAlis'er
Environmental Monitoring Systems Laboratory - Larry Johnson
Wheelabrator Environmental Systems - Tim Porter
Clarence L. Haile of Midwest Research Institute and Judith C. Harris of
Arthur 0. Little, Inc., moderated and participated in the Work Group
discussions and prepared this document.
We gratefully acknowledge the time invested by the Work Group. Their
assistance and expertise :nade this project possible.
v i i-i
-------
GUIDELINES FOR STACK TESTING Of
MUNICIPAL WASTE COMBUSTION FACILITIES
SEMI ON 1.0
INTRODUCTION
Considerable interest and activity
rave recently been focused on assessing
emissions from municipal waste combust-
ors (MWCs). In particular, state and
local environmental agencies have been
required to develop regulations for
KvJCs in response to a significant
expansion in trie nuirber of MWC facili-
ties currently in operation or in some
phase of pfanning and construction. As
part of this regulatory process, per-
mitting agencies have need for guidance
cn identifying critical emission param-
eters and also for methods that can be
used to measure those parameters, espe-
cially toxic sjDstances.
The Northeast States for Coordinated
Air Use Management (NESCAUtf) with the
support of the U.S. Environmental Pro-
tection Agency (EPA) Control Technology
Center (CTC) organized a work group to
provide specific guidance on methods
for measuring stack emissions from MWCs
and reporting forr.ats to facilitate
standardization of emissions tests.
This methods guidelines report 1s based
cn trie discussions of the NESCAUM Work
Group. Specific testing recommenda-
tions, emissions, measurement methods,
and recommended results reporting
formats are presented in the following
sections.
The test recc^-eidations presented in
Section 2.0 are based on the Work
Group's knowledge of current require-
ments of EPA and state and local per-
mitting agencies and in anticipation of
requireverts likely to be imposed in
the near future. These recommendations
jre not intended to prescribe regula-
tory req jl rerrents, but to recoir.nr.enj
emissions test programs responsive to
typical requirements.
The measurement methods presented in
Section 3.0 are based primarily on
reference methods already in widespread
use, although few have been validated
specifically for MWC testing. Most
refer to EPA methods contained 1n the
Federal Register and/or 'PA methods
manual for solid wastes (Svi-^46). How-
ever, many method recommendations and
descriptions of critical method fea-
tures specific to MWC emissions testing
were drswn from the considerable knowl-
edge and broad experience of the Work
Group.
Section 4.0 of this document presents
emission te^t reporting formats devel-
oped by the Work Group to encourage
standardisation of data reporting to
both facilitate efficient emissions
assessment and improve the usefulness
of MWC te«t reports for comparing emis-
sions among facilities. Standardized
reporting can greatly facilitate use of
test results to meet the needs of
multiple users, e.g., risk assessment,
modeling, and MWC research. For
example, the Work Group recommends the
use of 7% 02 as the basis for
normalizing stack gas concentrations to
a common excess air level, altnough it
recognizes that 12% C0? is frequently
specified in permit limitations. There
are advantages and disadvantages to
both choices but the Group felt that
one approach should be selected to
increase consistency of reporting. In
those cases there the applicable permit
condition, emission limitation, or
regulation specifies reporting units
that are different from the NESCAUM
recommended format, the Work Group
recommends that results be reported
both ways.
1
{Revised 7/14/88)
-------
References for specific method
citations are provided in Section 5.O.,
Section 6.0 is a bibliography of useful
references for MWC testing.
Two appendices are Included with the
document. Appendix A provides addi-
tional measurement methods for MWC
research testing. Appendix B provides
selected detailed procedures for mea-
suring PCDDs and PCOFs.
SECTION 2.0
TEST RECOftODATIONS
Stack testing at MWCs may be under-
taken to address many different types
of needs. The specific objectives of
an MWC stack test can determine both
the process and emission parameters to
be measured and the methods appropriate
to conduct the measurements. It is
important at the outset of any testing
program that all parties involved
(facility owner/operator, regulatory
bcdies, organization performing the
test, interveners if any) have a common
agreement on the measurements to be
made and the use to be made of the
data. It is also important to recog-
nize that many potentially useful sam-
pling and analysis methods may not have
been validated for specific application
to MWC testing. In those cases, extra-
ordinary quality assurance and control
procedures may be required in order to
document the validity of the test
results.
Conceptually, it is useful to cate-
gorize MWC stack testing (measurements)
into four broad categories.
Regulatory compliance testing -
Testing to determine compliance with
specific emission limitations
imposed as permit conditions by fed-
eral, state, cr local regulations
(or guidelines), or by the terms of
the indivicual facility permit.
• Performance testing - Testing to
characterize eaissions of a specific
MWC unit or component, or to vt.-ify
performance in accordance with an
equipment vender's guarantee.
• Research and development testing -
Testing to develop data bases on MWC
emissions, to determine the influ-
ence of specific operating param-
eters on emissions, and/or to
characterize general classes of
MWCs.
• Continuous monitoring - Testing of
selected parameters on a continuous
basis using instrumental monitors,
to confirm that bulk combustion con-
ditions and/or air pollution control
device performance are not substan-
tially different from those observed
during initial compliance/perfor-
mance testing.
The NESCAUM Work Group decided that
this document should focus on the first
three of these categories: Compliance
testing, perT^mance testing, and
research and development testing. The
group recognized tnat there are also
important issues to be resolved in the
area or continuous emission monitors
(CEM), out felt that should be the sub-
ject cf a separate, dedicated effort.
The Work Group also determined that
procedures for sailing and analysis of
refuse, which have been the subject of
investigations by the National Bureau
of Standards, ASME, and others, were
beyond the scope of this project docu-
ment. A major objective of the NESCAUM
project was to scecify a set of "core"
measuressent procedures that cculd be
applied consistently fto MSW combustor
testing by the various federal, state,
and local jurisdictions. This will
facilitate comparisons among facilities
and lead to the generation of a more
useful data base on MWC emissions.
5
-------
2.1 COMPLIANCE TESTING
The objective of compliance testing
is to determine whether an MWC is meet-
ing all of the conditions regarding
emissions that ara imposed by its op-
erating permit. In this context, the
terrr. "compliance" is used to refer not
only to limits Imposed by regulations,
but to any other restrictions that may
be Imposed, on a case-by-case basis, in
the operating permit for a facility.
Typically, testing to determine compli-
ance with all permit conditions is re-
quired at least one time during the
first 5 months of facility operations,
after the shake-down phase is complete.
Some states are requiring follow-up
compliance testing at intervals of
3 months to 1 year during the first
years of operation, and annually or
semiannually thereafter. The follow-up
tasting may not require the full range
of measurements made during the initial
compliance testing. Depending on the
Initial test results, a set of param-
eters considered to be key indicators
of continuing compliance may be speci-
fied.
The parameters to be measured during
compliance testing will depend, of
course, on the emission limitations
written into each individual permit.
Based on their common understanding of
current permit requirements imposed by
Federal and, mere importantly, state
regulations, the NESCAUM Work Group de-
termined that the following parameters
are likely to be required in compliance
testing for most new MWCs. It should
be emphasized that this is not an
exclusive list of parameters for com-
pliance testing. Many state and local
governments impose additional require-
ments for compliance.
• Polychlorlnated dlbenzodioxins and
dibenzofurans including tetra-
through octachloro homologs ard/or
all individual isomers containing
chlorines at the 2,3,7, and 8-
positions;
• Total hydrocarbons (THC);
• Metals Including lead, mercury,
cadmium, and total chromium; .
• Combustion gases including carbon
monoxide, carbon dioxide, and oxy-
gen;
• Particulate matter and opacity;
• Other gases Including hydrochloric
acid (HC1), nitrogen ox^es (N0X),
and sulfur dioxide (S02) as an
indicator of S0X; and
• Process operating conditions,
including furnace and boiler data,
emissions control device data.
Measurement methods for the stack test
parameters are presented in Section 3.0
of this document.
It should be noted that the numerical
limits on emissions of the above param-
eters will vary on a casa-by-case
basis. Even for particulate matter,
covered by Federal EPA New Source Per-
formance Standards (NSPS) for MWCs of
0.08 gr/SCF at 12* C02, new facility
permits by the states are generally
limiting emissions to levels of 0.015
or 0.010 gr/SCF. Similarly, different
states have different criteria for set-
ting PCD0/PC0F emission limitations.
The critical numerical limits for
each measured parameter jhould be
established before compliance testing
plans are finalized. This will enable
thp measurement methods to be tailored,
if necessary, to ensure that the method
detection limits are appropriate for
the permit emission limit.
The criteria that will be used to
judge the acceptability of the compli-
ance test results must also be care-
fully defined- and documented.
Compliance testing typically requires a
minimum of three replicate tests under
3
-------
the same nominal operating conditions.
However, -different agencies may nave
different quality assurance criteria
regarding completeness. Some require
that at least three tests be success-
fully completed for all parameters of
concern. Others will accept the
results even if one sample run is lost
(e.g., laboratory error/accident) as
long as the two tests successfully
completed demonstrate compliance. The
standard of completeness and contin-
gency provisions, if any, should be
discussed with regulatory authorities
prior to testing. Requirements for
analysis and reporting of results for
QA audit samples, if any, should also
se specified in advance of emissions
testing.
The conditions tested should include
the measurement of stack emissions at
or near 1C036 of the design heat input
capacity expected during normal opera-
tions if this is substantially differ-
ent from the design capacity. If the
facility wishes its operations to be
permitted under lower refuse loads
(e.g., 803= of design capacity), addi-
tional tests should be made using these
coerating conditions. For compliance
testing, the furnace temperature,
excess air, pressure drop across the
boiler, air pollution control device
rarameters, etc., should be typical of
those expected in routine operation.
(Tests to determine the impact of these
parameters on facility emissions are
classified as research and development
testing.)
In facilities consisting of multiple
units of the same design (i.e., 750 TPD
facility with tliree 250 TPD units), it
is the consensus of tne NESCAUM Work
Group that bulk combustion parameters
(e.g., particulate, CO, CO,, 02) should
;e measured on each unit. The param-
eters the group describes as "fuel-
related," such as HC1, S02, trace
crganics and trace metals, need only be
reasured on one of the units, assuming
they are of the same design and oper-
ated under the same nominal conditions.
Mote: These are the NESCAUM Work Group
reccnmendations; seme regulatory
authorities may require additional
testing on each unit.
2.2 PERFORMANCE TESTING
The objective of emissions perfor-
mance testing is to determine whether a
facility is operating in accordance
with its design specifications, in con-
trast to compliance testing for com-
parison with emission limitations.
Thus, while compliance testing usually
involves sampling of the final stack
effluent, performance testing may
require sampling both upstream and
downstream of air pollution control
devices.
The parameters to be measured in per-
formance testing could include anv of
these listed under compliance testing.
In addition, there could be interest
in, for example, comufxison of control
device efficiencies with respect to
fire particulate matter (e.g., 10 or
2.5 mean cut diameter) or to
specific organic ar inorganic chemicals
such as those listed below under re-
search and development testing. The
NESCAUM Work Group is not aware cf any
parameters uniquely associated with
performance testirg. Therefore, there
are no methods in this document di-
rected specifically to performance
testing. The methods described under
compliance and research and development
testing can also be applied ror perfor-
mance tasting.
The test conditions f'cr performance
testing again would depend upon what
aspect of system performance w?s being
determined. Operating parameters such
as ES? electrical field, fabric filter
clcth-to-air ratio (net or gross), cr
dry scrubber injection rate are ex-
angles of variables that mignt be
controlled in performance testing. As
-------
in compliance testing, a minimum of
three runs at each condition of inter-
est should be completed successfully to
ensure that the results are representa-
tive of actual performance during the
test period.
2.3 RESEARCH AND DEVELOPMENT TESTING
As in the case of performance test-
ing, the objectives of research and de-
velopment testing may be many and
varied. Among the possibilities dis-
cussed by the NESCAUM Work Group are:
• Development of improved data bases
through research and development
which will help determine if addi-
tional parameters should be sub-
jected to compliance testing or
dropped from tne current list cf
compliance pollutants,
• Expansion cf the data base cn
refuse-derived fuel (ROF) facilities
to determine whether apparent dif-
ferences in emissions between ir-ass
burn and ROF combustion or air-cnc
different RCF combustion systems are
characteristic (generic) differences
in the technologies or facility
(waste)-specific phenomena, or
• Tests, possibly including measu-e-
ments upstream of air pollution con-
trol devices, to assess the effects
of controllable combustion oaraj-
eters on emissions of specific pa'-
lutants,
• Tests of the hypothesis that
PCOF are formed primarily by pest-
combustion zone reactions,
• Evaluation of new ccmoustor tech-
nologies ,
• Evaluation of new or improved sam-
pling and analysis methodologies.
Among the parameters that the NESCAiM
Work Group identified as potentially cf
interest In MWC resea-ch and develop-
ment sampling are the following:
• Volatile organics on a compound-
specific basis, as contrasted with
the aggregate measurement repre-
sented by THC. The testing may be
directed toward target compounds,
e.g., formaldehyde, or to a broad
range gas chrcmatography/mass
spectrometry (GC/MS) screen,
• Polyrhlorinated biphenyls including
mono- through decachloro- homologs,
• Semi volatile organics possibly in-
cluding chlorinated benzenes, chlo-
rinated phenols, and polynuclear
aromatic hydrocarbons (e.g.,
benzo(a]pyrene),
• Metals possibly including arsenic,
beryllium, chromium (VI) versus
total chromium, nickel. Research
and development testing could also
include multiple metals screening
analysis.
Test conditions for research and
development testing are likely to be
even more variable than in the case of
performance testing. These might in-
clude, for example, deliberate opera-
tion of a facility outside of its
design specifications to define the
envelope of acceptable vs unacceptable
operation or to determine maximum emis-
sions under worst case operating condi-
tions. The comments made under compli-
ance and performance testing also
apply to research and development test-
ing. In the case of some research and
development testing, however, it nay be
necessary to perform six or more test:,
(vs the minimum of three specified
earlier) at each condition in order to
detect differences between emissions
under different conditions.
Procedures potentially suitable for
MWC research and develcement testing
are presented in Appendix A It should
5
-------
be noted that many of these procedures
were- initially developed fcr purposes
other than MWC testing, and may not
have been validated for this purpose.
It may be necessary to incorporate suf-
ficient quality assurance and control
measures in the test design to ensure
that the method performance is adequate
to meet the test objectives.
2.4 SPECIAL PURPOSE TEST PARAMETERS
Some MWC tests Include test param-
eters selected because of local envi-
ronmental concerns and special research
requirements. The Work Group discussed
several parameters in this category,
out decided not to include measurement
methods for them in this document.
for example, monochloro through
trichloro dioxin and furan homologs
were excluded primarily because homo-
logs with less than four chlorines are
not significant in determining risk
factors according to most commonly used
risk assessment protocols. The methods
described for tetrachloro- through
cctachloro-PCOOs and PCDFs in Sec-
tion 3.1 could be extended to the lower
homologs, provided that the QC proto-
cols are amended to include these cate-
gories.
Also, procedures for pollutants that
are of concern only in limited locali-
ties or with infrequently used tech-
nologies (i.e., ammonia emissions pos-
sibly associated with slippage in
de-NOx technologies) were also omitted.
SECTION 3.0
MEASUREMENT METHODS FOR COMPLIANCE/
PERFORMANCE TESTING
3.1 POLYCHLORINATED OIBENZO-2-OIOXINS
AND QIBENZQFURANS
The methods described in this sec-
tion are appropriate for measuring
tetrachloro through octachlora poly-
chlorinated dibenzo-g-dioxins (PCDOs)
and dibenzofurans (PCDFs) In MWC stack
emissions. The tetrachloro through
octachloro hcwologs can be determined
or the specific 2,3,7,8-sutstituted
congener can be determined simultane-
ously with the homologs. Alterna-
tively, 2,3,7,8-tetrachlorodibenzo-g-
dioxin (2,3,7,8-TCDD) may be determined
alone. Analysis of sample extracts for
2,3,7,8-lCDD or all 2,3,7,3-substituted
congeners requires significantly more
chromatographic separation than is
required for homolog measurements only.
Achieving the higher chromatographic
resolution requires a longer analysis
time, with proportionately higher
analysis costs.
3.1.1 Method Description
Stack emission samples are collected
for PCDD and PCDF measurement using an
EPA Method 5 train modified by the
addition of a condenser to cool the
gases and an adsorbent cartridge to
trap organics. This sampling train is
generally referred to as a Mcaified
Method 5 (MM5). Particulates are col-
lected in the filter and vaporous emis-
sions are retained by the cartridge.
The distribution of PCDOs and PCDFs
between the particulate catch ara the
cartridge are likely altered during
sampling. However, the sum of PCDOs
and PCDFs collected by the train repre-
sents the total concentrations in the
stack emissions. Following sampling,
the appropriate components of the train
are recovered and transferred to the
laboratory for analysis. The compo-
nents are extractedi the extracts are
cleaned to remove excessive levels cf
potential interferences, and the
cleaned extracts are analj^ed for PCODs
and PCDFs using combined <:as chronatog-
raphy/mass spectrometry (GC/MS).
6
-------
3.1.1.1 Sample Collection—
A detailed protocol for collecting
stack emissions using a MM5 sampling
train can be found 1n Method 0010 of
SW-846. This protocol was prepared spe-
cifically for measuring semivolatile
organics in hazardous waste incinerator
stack emissions to allow determination
of destruction and removal efficiencies
(OREs) of Principal Organic Hazardous
Components (POHCs) contained in the
waste feed. Method 0010 can be applied
nearly directly for measuring PCOD and
PCDF emissions since PCDOs and PCDFs
are categorized as semivolatiles. The
Method 0010 protocol is similar to the
draft protocols produced as a result cf
a workshop sponsored by the American
Society of Mechanical Engineers and
U.S. Environmental Protection Agency,
generally referred to as the ASME Draft
Protocol (1934).
Train configuration. The MM5 train
is assembled and operated in a manner
very similar to a Method 5 particulate
train with three exceptions:
a. Addition cf a condenser and water-
cooled aasorbent resin cartridge
immediately downstream of the filter
box.
b. T^e train precleaning to avoid
organic contamination.
c. Sample recovery procedures.
A schematic of the MM5 sampling train
from Method GO 10, SW-346 is shown in
Figure 3-1. The adsorbent cartridge 1s
filled with Amberlite5 XAO-2 resin spe-
cially precleaned according to the
protocol contained in Method 0010, Ap-
pendix 6, SW-846. The condenser and
adsoroent cartridge can be facricated
as a single unit (as shown in Fig-
ure 3-2) or as individual units (Fig-
ure 3-3). The condenser cools the gas
to 20"C or less and condenses part of
the TWisture. The cooled gases and
condensed moisture flow down through,
the XAD-2 resin which retains PCDDs and
PCDFs from the vaporous and aqueous
phases.
Train assembly and operation. All
train components from the first im-
plnger forward must be rigorously
cleaned before use to avoid organic
contamination. The*~train should be
assembled as shown 1n Figure 3-1. A
cyclone (also precleaned) may be
installed for sampling flue gas with
high particulate loading, e.g., inlet
of a particulate control device. Also,
an oversized first impinger or
additional iooingers may be used to
Increase the water collection capacity
for sampling high moisture gases, e.g.,
outlet of a wet scrubber.
The flue gas sample should be col-
lected by operating the train according
to the Method 5 specifications, includ-
ing leak checking, isokinetic sampling
rates, and stack or duct traversing.
However, additional leak checks should
be conducted after each port change.
The sample should be in the range of 3
to 6 cubic meters.
Sample recovery. At the completion
of sampling, the probe and umbilical
should be carefully removed from the
train, the ends of the probe and the
train should be capped with precleaned
aluminum foil or glass caps, and the
probe and train should be removed to a
clean field or permanent laboratory for
sample recovery. The sample recovered
consists of the following:
a. Front half rinse—rinse of the
prcbe, front half of the filter
holder, and rinse of connecting
lines between the probe and filter
holder (including a cyclone if
used).
b. Particulate catch—fiIter, including
cyclone eaten (if used.)
7
-------
TlieimiHnrlei
> fillm I (older
llrJlcil Airj
lcni|H>«4lii«t !>eiuof
Prolte
Reveiic Typ« Pilol Tulis
rn
Piti>l Mjnomcler
ncciir.ulalion Pump
L _ _
1
Imimignit
lly Pjm Vat*c
Dry (jj» M. ln Ai« lii|l't Punm
Figure 3-1.
Modified Method 5 sampling train,
(from Method 0100, SW-846)
-------
Flow Oii«i.ium
211/1? lull Jolul
(iUll Wf 211/12'Joi hcl Jtimi)
w
Figure 3-2. Combined condenser - adsorbent cartridge unit,
(from Method 0010, SW-646)
-------
CoilJtflUOf
ttu« Gut fluw
Sort>«i»l Trap
• 32 cm-
•25cm-
6 mm Clutt Coaling Coll f|
7
7
i 28/IS
WuUf Jucicl CuuIImj Cult
Glut* Wool (%y Will9t JimAcI fm-PuiUUJ Glau
XAl> - 2 Olik
( 75 G
-------
c. Back half rinse—rinses of back half
of the filter holder and all con-
necting lines from the filter holder
to the inlet of the resin cartridge.
d. Resin cartridge.
e. First Impinger—condensed sample
moisture.*
A detailed procedure for each train
component is presented in Appendix C.
3.1.1.2 Sample Analysis—
A detailed protocol for analyzing
PCODs and PCDFs in a wide range of sam-
ple types can be found in Method 8280
of SW-846. Although much of this pro-
tocol can be applied directly for
determining these compounds in samples
recovered from MM5 trains, sorre modifi-
cations are typically required to
achieve method detection limits suit-
able for MWC stack samples.
Sample preparation and extraction.
Solid sample components, i.e., the
filter ^nd cyclone (if used) catch and
the XAO-2 resin are Soxhlet extracted
using toluene or benzene. The aqueous
components and rinses are extracted by
simple partiticrirg with dichloro-
methane. A detailed procedure for
extracting each train component is pre-
sented in Appendix C.
Prior to extraction, the samples must
be spiked with solution of method
internal standard (MIS) comDOunds.
These are isotcuically labeled PCDO
and/or PCDF congeners (sometimes
referred to as surrogate compounds)
which are used to quantitate PCOOs and
PCDFs in the sample by an internal
standard method. If available, it
shou^ also contain a labeled congener
representing each homolog quantitated.
A typical MIS solution might contain
the following:
»3Cl2-2,3,7,3-TCD0 (tetrachloro-
dibenzo-jj-dioxin)
>3C,2-l,2,3,7,8-PeCD0 (pentachloro-
dibenzo-g-dioxtn)
l3Cj2-l,2,3,6,7,8-HxCDD (hexachloro-
dibenzo-p-dioxin)
i3CiZ-l,2,3,4,6,7,8-HpCD0 (hepta-
chlorodlbenzo-o-dloxin)
13Ci2-0CDD (octachlorodibenzo-D-
dioxin)
Extract cleanup. The extracts from
the individual train components should
be combined, concentrated, and cleaned
by column chromatography as described
in Method 8280 to remove coextracted
interferences. The final, cleared
extract should be prepared in a high
boiling solvent, such as tridecane or
hexadecane. Achieving very low detec-
tion limits typically requires a final
extract volume in the range of 15 to
100 yL. A detailed procedure for
extract cleanup is presented in
Appendix C.
SC/MS analysis. Sample extracts must
be analyzed by GC/MS as described in
Method 8280, SW-846 using a fused
silica capillary column and selected
ion monitoring (SIM) data acquisition.
The GC/MS system must provide suffi-
cient chromatographic separation and a
level of sensitivity sufficient to
acnieve the desired minimum detection
limits for all subject compounds.
The PCDDs and PCDFs are quantitated
using response factors normalized to
the MIS coiwjounds spiked prior to
sample extraction. A detailed proce-
dure for GC/MS analysis, including
guidance on instrument specifications
* The condensate ^ay be analyzed sepa-
rately to provide an indication of
gross breakthrcugn.
11
-------
and GC column selection, is presented
in Appendix C.
The GC/MS calibration standard solu-
tions should contain the following:
a. A native congener representative of
each PtfO or PCDF homo log to be
quantit&ted, ax "one of the three
calibration levels.
b. The HIS compounds at levels corres-
ponding to that spiked into the
samples, i.e., equivalent to 100X
recovery.
c. At least one recovery internal stan-
dard compound (RIS), to be used to
determine the recoveries of the MIS
compounds, at a level similar to the
MIS conrpounds.
An exajnple of the compounds included
in a calibration standard for deter-
nining 2,3,7,8-substituted PCDDs and
PCDFs is presented in Table 3-1.
3.1.2 Critical Features
3.1.2.1 Sample Collection--
Sampling train materials. All
sampling train components that ccme in
contacc -«ith the recovered sarr.pl e must
be comDosed of glass, Dolytetrafluoro-
ethylene (PTFE), or similar materials
to avoid potential sample contamination
or reactions with PCDOs and PCOFs.
Hence, the probe liner must be of boro-
silicate glass or quartz. Also, the
probe nozzle should be nickel or nickel
plated or (if available) constructed of
glass or quartz. If the location
and/or orientation of the sampling
ports requires, a shurt PTFE flexible
line may oe used to connect the probe
to the filter (or cyclcne, if used)
inlet. However, the internal diameter
cf the line must be the same as the
probe liner and tte line must be heated
to retard condensation of moisture.
Gaskets or other joint sells fabri-
cated from materials other than PTFE
should not be use*. This includes the
filter gasket. The use of silicone
sealant greases ts not permitted. The
use of silicone sealant greases and
other materials on train components not
contacting the sample is acceptable but
not advised.
Condenser/resin cartridge unit.
Alternate configurations for the
condenser and resin cartridge may be
acceptable. However, the units must be
configured so that flue gas and con-
densed moisture flows down through the
resin and the cartridge should be
charged with 20 to 80 g df XAD-2 resin
(40 g is typical). Figure 3-4 shows an
alternate configuration with the con-
denser ar.d resin cartridge designed in
the shape of iq^jir.gers to reduce the
height of the train and decrease the
potent!?! for breakage during handling.
Althoug.. the appearance of this con-
figuration is very different from those
shown in Fiyures 3-2 and 3-3, the unit
is acceptable oecause the condensed
moisture is required to flow down
through the resin ana the capacity cf
the cartridge is 65 g of resin. Ouring
sample recovery, any condensed moisture
collected in the bottom of the con-
denser unit is poured tnrough the XAO
resin.
Also, the quality of the resin must
be checked before packed cartridges are
used. At least one aliquot cf each
resin lot (manufacturers and cleanup
batcn) equivalent to, a cartridge charge
should be extracted' and analyzed as a
sample to confirm the background levels
of the conDOunds of interest or poten-
tial interferences are acceptable.
12
-------
TABLE 3-1. LIST OF ANAl.YTES, METHOD INTERNAL STANOARDS, SURROGATES, ANQ
RECOVERY INTERNAL STANDARDS FOR DIOXIN/FUSAN ANALYSIS
Analyte
Compounds in
calibration standard
Method
internal standard3
Recovery .
internal standard
Tetra-CDD
2,3,7,8-TCBD
l3C,2-2.3,7,8-TCDD
J'Clu-2,3,7,8-TC00 or
Tetra-COF
2,3,7,3-TCDF
13C12-2,3,7,8-TC0F
' 1CiZ-l,2,3,4-TCDD
Penta-CDO
1,2,3,7,8-PeCDD
»»C,2-1.2,3,7.8-PeCD0
Penta-COF
1,2,3,8,9-PeCOF
»3Cl2-l,2,3,8,9-PeCDF
Penta-CDF
2,3,4,7,3-PeCOF
13C12-1,2,3,7,8-PeCDO
Hexa-CDO
1,2,3,4 J,3-HxCDO
2,3,6,7,8-HxCOD
i»Cl2-l,2,3,6,7,8-HxCDD
Hexa-COO
1,2,3,6,7,3-HxCDD
13Ct2-l,?,3,6,7,8-HxCOO
Hexa-CDO
l,2,3,7,S,9-HxCDD
l3C l2-1,2,3,6,7,3-HxCDD
Hexa-CDF
1,2,3,4,7,6-Hx.CDF
1 JC,2-1,2,3,6,7,8-HxCOD
Hexa-CDF
1,2,3,6,7,3-HxCDF
13C i2-1,2,3,6,7,3-HxCOO
Hexi-COF
2,3,4,6,7,8-HxCQF
l3C,2-1,2,3,5,7,3-HxCDD
Hexa-CDF
1,2,3,£ ,3,9-HxCDF
l3C i2-1,2,3,6,7,3-HxCDD
Hepta-CDD
1,2,3,4,6,7,8-HpCDO
13C12-1,2,3,4,6,7,8-HpCDO
Hepta-COF
1,2,3,4,6,7,8-HpCDF
,3C l2-l,2,3,4,6,7,8-HpCDD
Hepta-CDF
1,2,3,4,7,3,9-HpCDF
13Cl2-l,2,3,4,6,7,8-HpCDD
Octa-COO
ocoo
t3Cl2-0C0D
Octa-CDF
OCOF
•3C,2-0C00
a Added to sample prior to extraction.
^ Added to sample at time of injection into GC/MS.
13
-------
Dumocowe'e
Wall
From
Hlrtr
XAO-2
Figure 3-4. Example condenser and XAO resin cartridge designed in
impinger form.
14
-------
Resin cartridge temperature. Care
roust be taken during sample collection
to ensure proper cooling of the con-
denser and resin cartridge unit. The
temperature at the inlet to the
cartridge should not be permitted to
exceed 20*£. Considerable quantities
of 1ce may be required for each sam-
pling run.
Use of sample for multiple param-
eters. §oiiie additional parameters can
be determined from a MM5 train used for
PCDD and PCDF measurements, with vary-
ing degrees of compromise. Solutions
charged Into the second and succeeding
impingers may be modified to facilitate
analysis of compounds such as HC1.
AHquots of the first itnpinger solution
can also be utilized. Also, the ex-
tracts) from train components prepared
for PCOO and PCDF analysis can be split
for use in determining other semivola-
tile organics. However, splitting of
the extract proportionately increases
the minimum detectable levels far PCDQs
and PCOFs as well as the other semi-
volatile parameters. The particulate
catch from the train should not be used
to measure particulate emissions. Dry-
ing the filter to achieve reliable
gravimetric analysis may result in sig-
nificant losses of PCDDs and PCOFs con-
tained on the filter. Similarly, the
filter should rot be allquoted to
facilitate other analyses, e.g., non-
volatile metals, since the distribution
of particulates on the filter may not
be uniform.
3.1.2.2 Sample Analysis—
Method internal standard fMIS>
spiking. The potential for reliable
quantitation of a PCDD of PCDF compound
may be significantly enhanced using an
MIS conpound with very similar chemical
and physical properties. Unfortu-
nately, the availability of appropri-
ately mass labeled PCDD and PCDF
congeners limits the number of MIS com-
pounds chat can be practicably used in
MWC stack measurements. Nonetheless,
at least two compounds, a labeled TCDD
and labeled OCQO or OCQF, should be
used to quantitate tetrachloro through
octachloro PCOO and PCOF homo logs. If
possible, a labeled representative of
each homolog class should be included
in the MIS compounds, e.g., as 1n
Table 3-1.
In the case of MM5 samples, the final
extract is a composite of extracts of
the several train components. Hence,
any one MIS compound should be spiked
into oniy ore train component prior to
extraction. However, different MIS
solutions can be spiked into different
components. The filter and XAD resin
components should be considered the
highest priority in developing MIS
spiking strategies.
Extract cleanup. The alumina and
carbon column cleanup procedures re-
quire considerable laboratory skill and
attention to detail to routinely
achieve acceptable PCDD and PCDF recov-
eries and a sufficient level of extract
cleanup. Laboratory staff conducting
these procedures should be familiar and
practiced with the adsorbents and with
handling trace quantities of PCDDs
PCOFs. Laboratories conducting PCOD
and PCDF analyses should have demon-
strated acceotable recoveries by pro-
cessing spiked blanks through the
cleanup procedures prior to analysis of
MWC samples.
Quality of standards. Analytical
standards for natives and labeled con-
geners are expensive. They are typi-
cally sold in smalil quantities of the
neat compound or in solution form.
Both the purity' and quantitative
accu-acy of standard solution should be
evaluated before luse. At the very
least, the origin 'and use of standards
solutions should be clearly traceable.
15
-------
Handling of am 11 extract volumes.
Achieving optimum detection limits for
PCDOs and PCOFs typically requires con-
centrating the final extract to con-
siderably less than 1 mL. It 1s very
Important that sma.11 volume extracts be
handled with extreme care. Whenever
possible, maintain the extracts at
larger volumes until just before GC/MS
analysis and concentrate to less than
500 ul only 1n the presence of a higher
boiling keeper solvent, e.g., tridecane
or hexadecane. Also, do not attempt to
split or aliquot extracts at volumes
less than 1.0 mL. Dilute the extract,
split or aliquot are required, then
reconcentrate as needed.
Selection of a GC/MS systea. The
principal considerations are the basic
sensitivity of the instruwent, the
effective SIM mass range during analy-
sis, and the mass, resolution. The
instrument level of detection required
should be calculated from the target
detective level in the flue gas sample.
The various spectrometer systems have
different capabilities and limitations.
Quadrupole focusing mass spectrometers
can provide good sensitivity for the
entire range of ions required to deter-
mine tetrachloro through cctachloro
PCDOs and PCDFs in a single GC/MS run.
Also, the set of ions monitored can be
changed under computer control during
the, run. The effective SIM range of
magnetic focusing mass spectrometers is
limited because the sensitivity in-
herently falls off at higher mass for a
fixed magnet setting. It is generally
not practical to change magnet settings
during a GC/MS run. Hence, multiple
GC/MS runs may be required to achieve
optimum sensitivity for determining
tetrachloro through octachioro PCDOs
and PCOFs, one for tetrachloro and
pentachloro homologs and a second for
hexachloro through octachioro hcmologs.
However, the excellent sensitivity of a
high performance double focusing (mag-
netic md electrostatic focusing
sectors) mass spectrometer can be suf-
ficient to achieve acceptable detection
limits for tetrachloro through octa-
chioro PCDOs and PCOFs In a single
GC/MS run, typically significantly
lower than possible with a quadrupole
spectrometer. A double focusing (mag-
netic and electrostatic sectors)
spectrometer also offers the option of
providing addltionat-qualitative veri-
fication of PCDOs and PCDFs at higher
mass resolutions. However, high reso-
lution mass spectrometry is not typi-
cally required for MWC samples.
In many cases, lowest achievable
detection limits are requested for
PCDOs and PCOFs in MWC stack samples,
whether due to very low emission stan-
dards and/or very low actual emissions.
Hence, high performance magnet mass
spectrometers (single or double
focusing) are preferred for determining
PCDOs and PCOFs in MWC stack emission
samples.
3.1.3 Quality Assurance and Quality
Control Procedures
3.1.3.1 Calibration of Field Equip-
ment-
Method S provides specific procedures
for calibrating and leak-checking the
sampling equipment. However, addi-
tional leak checks should be conducted
following each port change.
3.1.3.2 Field Blank Saaples—
At least one blank train should be
assembled, recovered, and analyzed in
the same manner as a sample train for
every five tests at each facility. The
blank trains should be assembled, leak
checked, and the probe and filter box
maintained at normal operating tempera-
tures for the duration of a typical
test run. However, the sampling pump
should not be activated. If a facility
test consists of more than five sequen-
tial tests, at least cne blank train
should be run for every five tests.
16
-------
A complete set of other ffeld blanks
should also be taken and reserved for
diagnostic analysis if the background
contribution from any blank train is
unacceptable. These should include at
least one from each lot used of the
following: sample containers, filters,
resin cartridges, and aliquots of rinse
solvents.
3.1.3.3 Laboratory Blank Samples—
At least one complete method blank
should be analyzed for each lot of sam-
ples analyzed and fo; each extraction/
cleanup method combination used. Also,
a cqirplete set of laboratory solvents
and materials (at least, one from each
lot used) should be reserved for diag-
nostic analysis if the background
contribution from any method blank is
unacceptable.
3.1.3.4 Resin Prescreening—
Each lot of XAD-2 resin should be
screened for background contribution
prior to use for sample collection. A
resin aliquot equivalent to a cartridge
charge should be extracted and analyzed
in the same manner as a sample, with
the exception that extract cleanup can
be omitted. If the background contri-
bution is higher than the target quan-
titation limit for any analyte, that
lot of resin should be cleaned by
extraction with dichloromethane and a
second aliquot screened. Cleaned resin
stored for more than 2 weeks should be
stored under methanol. The methanol
should be removed by decanting and
resin dried as described in
Method 0010, Appendix B, SW-846. The
resin should not be heated.
3.1.3.5 Column Recovery Spikes—
Each lot of samples cleaned by column
chromatography should include at least
ore spiked blank to provide a check on
recovery.
3.1.3.6 Relative Response Factor Moni-
toring—
RRFs for each analyte and MIS com-
pound should be within 2054 of the mean
RRF for that contpound. The RRFs for
the daily standards should also fall
within 20% of the mean. Alternatively,
RRF control charts can be developed and
maintained for analyses conducted on a
routine basis.
3.1.3.7 Quality Control Check Samples—
A blind quality control check sample
should be analyzed with samples from
each facility test. Th<« concentrations
of native anaTytes should be in the
lower half of tie calibration range.
The concentrations of MIS and RIS com-
pounds should be the same as in sample
extracts. The accuracy results for
blind check standard should be within
60 to 120% of the spiked level.
3.1.4 Oata Quality Objectives
3.1.4.1 Limit of Detection—
Limits of detection objectives should
be based on recu'atory requirements.
Achieving detection limits signifi-
cantly lower than regulatory require-
ments can needlessly increase testing
costs. Nonetheless, it is anticipated
that very low limits of detection will
be required or cesired in most cases.
Table 3-2 presents target detection
limits for PCDGs and PCDFs that are
readily achievable using the methods
described above *ith a high performance
magnetic focusing GC/MS system. Sig-
nificantly lower detection limits have
been achieved for sooe MWC tests.
3.1.4.2 Accuracy—
The accuracy objective for blind
quality control check sample is 60 to
120* of the spike level. The accuracy
objective for recovery of MIS -compounds
should be 50 to ISO*.
17
-------
TABLE 3-2. EXAMPLE TARGET LIMITS OF
DETECTION
Congener
Target detection limit
(ng/Nm3)
TCOO
0.15
TCDF
0.15
PeCDO
0.15
PeCDF
0.L5
HxCDD
0.6
HxCDF
0.6
HpCDD
0.5
HpCDF
0.6
OCDO
1.5
OCDF
1.5
3.1.4.3 Precision-
Precision should be evaluated by
pooling all the accuracy values, i.e.,
recoveries, for each MIS compound. For
examcie, the recoveries for the MIS
compound 1JCti-2,3,7,8-TCDD should be
pooled to determine a single precision
value for 7CD0 determinations. The
target for precision is < 5C%.
3.1.4.4 Completeness—
The completeness objective for PCQOs
and PCQFs in MKC flue gases should be
at least 90S valid determinations.
3.2 PARTICULATE MATTER
3.2.J Method Description
The basic method for measurement of
particulate natter in stack gas to
determine compliance with emissions
limitations or to compare the perfor-
mance of alternative particulate con-
trol devices is that specified in EPA
Methcd 5 (Code of Federal Regulations,
Title 40, Part 60, Appendix A) and the
associated Methods 1-4. This method
operationally defines particulate
matter as any material that condenses
when stack gas is withdrawn isokineti-
cally through a temperature-controlled
glass-lined probe and high-efficiency
fiber filter.
The Method 5 sampling train is shown
schematically in Figure 3-5. The es-
sential elements arjs a glass- or
quartz-1ined temperature-control led
probe equipped with a Pi tot tube (for
measuring stack gas flow rate) and a
sharp-edged button-hook nozzle, a
glass- or quartz-fiber filter supported
in a glass filter holder inside an oven
immediately at the outlet of the probe,
an Impinger train or condenser to
remove water, and a pump and metering
system.
Procedures for selecting sampling
locations and for operation of the
train are specified in detail in
Method 5 and associated Methods 1 to
4. (Note: EPA has recently relaxed
the Hethod 1 cyclonic flew restrictions
for stack sampling. The NESCAUM Work
Group has determined that the relaxed
restrictions are probably not appropri-
ate for ttWC stack sampling and recom-
mends that the more stringent restric-
tions be maintained for this purpose.)
The procedure indicates a sampling rate
of 0.5 to 1.0 cubic feet per minute
(cfm) or 14 to 28 liters per minute
(Ipm) for a minimum of 1 h total sam-
pling time. The minimun total sample
size is thus 30 cubic feet (cf) or
1.1 ms.
3.2.2 Critical Features
3.2.2 1 Sampling Train Hardware-
Hardware suitable for use in EPA
Method 5 is commercially available from
a number of suppliers and there is no
reason to consider a nonstandard train
configuration for compliance/perfor-
mance measurement of particulate mat-
ter. Quart2 probe liners or water-
18
-------
SO
annr^
TEMPERATURE SENSOR
PROBE
IMPINGER TRAIN OPTIONAL. MAY BE REPLACE!!
BY AN EQUIVALENT CONDENSER
PITOT TUBE
TEMPERATURE
SENSOR
HEATED AREA
THERMOMETER
FILTER HOLDER
REVERSE-TYPE
PITOT TUBE
STACK
—"WALL
PR08E
THERMOMETER
/
PITOT MANOMETER
ORIFICE
IMPINGERS ICE SATH
BYPASS VALVE
/
THERMOMETERS
o
VACUUM
GAUGE
MAIN VALVE
DRY GAS METER
AIRTIGHT
PUMP
Figure 3-5. Method 5 particulate train. (Code of Federal
Regulations, Title 40, Part 60, Appendix A)
CHECK
VALVE
VACUUM
LINE
-------
cooled probes with borosillcate glass
liners should be used if the stack tem-
perature is expected to exceed 480*C
(900aF). At lower temperatures, boro-
sillcate glass can be used without
cooling. The metal probe option al-
lowed in Method 5 should not be used
for MWC testing due to the potential
for corrosion. If the only purpose for
which the sample is to be used is
determination of particulate matter
emissions, 1t does not matter whether
glass or quartz is chosen as the fiber
filter medium. Mote, however, that if
chemical analysis procedures for, e.g.,
metals, are to be applied to the col-
lected particulate material after its
mass has been gravlmetrlcally deter-
mined, the filter medium selected must
be shown to be free of analytical
interferences and/or background con-
tamination.
3.2.2.2 Sampling Collection Time—
Depending on the particulate material
emission standard with which a given
facility must comply, it may be neces-
sary to sample for longer than the 1 h
indicated in the Method 5 protocol. For
example, at an emission rate of
0.010 grains per dry standard cubic
feet (gr/dscf), the limit specified in
some recent permits, a 2 nJ or larger
sample of stack gas may be required to
ensure that the precision of the
compliance/performance determination is
adequate. The necessary stack gas
sample size (and sampling time) for a
given test must be calculated by divid-
ing the minimum weighable particulate
mass (taking into account the filter
tare weight) by . the particulate mass
emission rate specified in the permit.
3.2.2.3 Hot Zone Sampling-
Some MWC tests may require sampling
upstream of air pollution control de-
vices in order to, for example, quan-
tify the removal efficiency achieved by
the devlce(s). The basic method for
performing these wasurements is the
samie as described above. However, some
additional cons icferat ions and method
modifications are likely required. The
gas flow within the ductwork or breech-
ing from which samples are ..to be
acquired will often not meet the cri-
teria established in EPA Method 1 for
Method 5 use. Access to these sampling
locations may make traversing the stack
cumbersome, at best. If situations are
encountered in which exceptions must be
made to the Method 5 traversing pro-
cedure, it may be possible to perform a
velocity traverse and select a point of
average velocity, away from the walls
of the duct, as a single sampling loca-
tion. Sampling should be isokinetic at
this point. Also, the standard mate-
rials for probe construction may not
withstand the high temperatures and
high corrosivity often present at loca-
tions upstream of pollution control
devices. At temperatures above 900°C
(1650°F), a water-cooled probe jacket
is advised. Any exceptions to Method 5
should be approved by the regulatory
authority prior to testing.
3.2.3 Quality Assurance and Quality
Control Procedures
3.2.3.1 Calibration—
Reference MetNxJ 5 specifies in
detail the procedures that must be used
to calibrate and leak-check the sam-
pling equipment. THe analytical bal-
ance used for tne gravimetric deter-
minations must also be calibrated
regularly.
3.2.3.2 Blanks-
Method 5 also sctfcifies that a QC
blank sample, consisting of an aliquot
of the acetone solvent used for train
recovery,, be prepared. The Method does
not require a bla^k filter QC sample.
However, generation of a field blank
filter is standard cractice. To ensure
comparability witn ;t^e sample filters,
20
-------
this blank may be mounted In a filter
holder and held it 120* ± i4°C for a
time e3u.1l to the total sampling run
time.
3.2.3.3 Replicated-
It 1s standard practice to perform
Method 5 determinations 1n triplicate
(three separate stack sampling runs).
3.2.3.4 Spikes—
For the particulate matter measure-
ment, t>»e concept of "spiked samples"
is not applicable.
3.2.3.5 Calculations—
Method 5 presents the equations to be
used to calculate the percent iso-
k1neticity and the particulate matter
emission rate.
3.2.4 Data Quality Objectives
3.2.4.1 Isold net1city—
Methcc 5 requires that the isokine-
ticity ce in the range of 90 to 110X.
Some agencies require that this objec-
tive be ret for each sampling point, or
for specified fraction (percent) of the
sampling points, while others require
only that the average isokineticity
over eacn run meets the objective.
3.2.4.2 Accuracy-
Method 5 does not include data qual-
ity objectives for accuracy of the
particulate matter determination.
3.2.4.3 Precision—
Method 5 does not Include data qual-
ity objectives for precision of repli-
cate cstemri nations. It is usual
practice to comoute the mean of the
three separate ceterminations and
compare this mean value with the emis-
sion l:Tri:ation. Seme states, however.
require that each of the three measure-
ments fall within the emission limita-
tion.
3.3 METALS
This subsection describes measurement
methods for lead (Pb), cadmium (Cd),
total chromium (Cr), and mercury (Hg).
3.3.1 Lead, Cadmium,-and Total Chromium
3.3.1.1 Method Description-
It is the consensus of the NESCAUM
Work Group that EPA Method 12 (Code of
Federal Regulations, Title 40, Part 60,
Appendix A) is the method of chcice for
compliance testing of lead, cadmium,
and total chromium from municipal waste
conbustors. The basic sampling proce-
dure specified in Method 12 is an EPA
Method 5, with 0.1 N nitric acid re-
placing water in the first two imping-
ers of the train. (See Section 3.2 for
description of M5). After sample col-
lection and recovery, the probe wash
and iicpinger solutions are combined and
taken to dryness. The residue is com-
bined with the particulate matter on
the M5 filter and digested with 50%
nitric acid followed by hydrogen ier-
oxide. Lead analysis is by atomic
absorption spectroscopy (flame).
Cadmium and chromium are determined by
atomic absorption spectroscopy (AAS) or
inductively coupled plasma (ICP) spec-
troscopy.
3.3.1.2 Critical Features—
The features of the MS train dis-
cussed in Section 3.2., above, apply to
•M12 as well.
For determination of lead, cadmium,
and chromium it is important that the
glass fiber filters used for collection
of particulate matter have a lot assay
indicating a low background level of
the analytes.
21
-------
It 1s acceptable to use the same M5
runs for determination both of particu-
late matter and of cadmium, chromium,
and lead. When this option 1s used,
the probe liner should be rinsed with a
0.1 N nitric acid solution after the M5
acetone/water rinses have been per-
formed. This addle probe rinse is
kept separate from the acetone/water
rinses and is subsequently combined
with the 0.1 N nitric acid impinger
solutions, taken to dryness, digested,
and analyzed.
The detection limit cited for lead in
the method is 0.2 to 0.5 ug Pb/mL of
final digested solution. Since the M12
procedure results 1n a final volume of
250 mL and the standard M5 sample size
is 1.1 m3, the in-stack detection limit
for lead is on the order of 45 to
114 ug/m*. If a lower detection limit
is required to determine compliance
with permit conditions, three options
are readily available:
a. Collection of a larger stack gas
sample;
b. Evaporation of the 250 mL digested
sample to rear dryness and dissolu-
tion of the residue in a smaller
volume of water; or
c. Substitution of flameless AAS for
the flame procedure specified in the
method.
The first of these is practically
limited to a factor of 2 to 3 improve-
ment in sensitivity (2 to 3 h stack gas
san^le versus the standard 1 h). With
the second, an improvement of a factor
of 5 to 10 is probably achievable. A
volume smaller than 25 mt may not allow
^solubilization of the residue. This
alternative also introduces the possi-
bility of some losses through volatil-
ization. Flameiess AAS (e.g., SW-846
Method 7421) offers the greatest poten-
tial improvement in sensitivity, since
inherent detection limits for lead ty
this technique are about 1,000 times
lower than for flame AAS. (SW-846
cites a detection limit of O.t pg/L for
the furnace Method 7421, versus
0.1 mg/L for the flame AAS
Method 7420.) However, the presence of
interferences may limit the sensitivity
improvement achievable on actual sam-
ples to less than 1,000-fold.
The detection Halts for cadmium and
chromium in the solution analyzed by
the ICP method are about 4 yg/L and
7 ug/L, respectively. For a standard
1.1 ns M5 sample, and a final extract
volume of 250 ml (as specified in N12),
these correspond to 1n-stack detection
limits of 1 tig/m* for cadmium and
1.8 yg/m* for chromium. If flame AAS
is used for the determination, the de-
tection limits increase to 1.3 and
11 ug/m3.
If lower detection limits are
required ...for cadmium or chromium, the
options are essentially the same as
described under lead, above. The
simplest option is to reduce the volume
used in diluting the sample after
digestion. Depending on sample charac-
teristics, however, a final sample vol-
ume of less than 25 mL is probably not
practical. On the other hand, if
flaiusless (graphite furnace) AAS is
used as the analytical technique,
inherent detection limits are consider-
ably lower: 0.01 ug/m3 for cadmium and
0.1 ug/m* for chromium. Keep in mind
that interferences (e.g., iron) nay
raise these detection limits somewhat
for real samples.
3.3.1.3 Quality Assurance and Quality
Control Procedures
Calibration. The sampling equipment
must be calibrated as specified in M5.
Laboratory balances must be calibrated
according to a regular schedule. The
AAS system is calibrated for wavelength
accuracy according to the manufac-
turer's soecifications.
22
-------
M12 specifies that the quantification
be accomplished by reference to a six-
point external standard curve covering
the range of 0 to 20 ug/mL of lead.
Samples must be diluted as necessary to
fall within the calibration curve
range. The method specifies that the
curve 1s not to be force-fit through
zero. The same procedures should be
used 1n generating standard curves for
cadmium and chromlua.
It 1s acceptable to prepare a single
set of calibration standards containing
cadmium, chromium, and lead, depending
on the determinative technlque(s)
chosen for the analysis. An HC1 solu-
tion 1s required when the analysis Is
to be done by ICP or flame AAS, nitric
acid is used for furnace AAS.
Blanks. M12 specifies that a filter
blank be created using two filters from
each lot used 1n the sampling train.
The filter blank sample 1> digested in
the same fashion as the samples, how-
ever, the method specifies that the
blank be made to a final volume of only
100 mL.
The method also specifies that a
0.1 N nitric acid blank be prepared by
taking 200 mL of solution to dryness
and digesting the residue as if it were
a sample. Again the final volume is
made to IOC mL.
The states that the absorbance
readings for sanples are to be cor-
rected for both the filter blank and
the nitric acid "jlonk absorbances.
Replicates. Samples, filter blanks,
and nitric acids blanks are analyzed 1n
triplicate. The mean of the three
determinations is used to read the sam-
ple concentration frcm the standard
curve.
It is common practice to base compli-
ance determinations on the mean of the
lead emissions as measured in three
separate stack sampling runs.
Spikes. The method requires that at
least one sample from each source
tested shall be analyzed by the method
of- additions as a check on possible
matrix effects. One aliquot of the
saaple is analyzed unsplked and a
second aliquot is mixed with an equal
volume of standard solution. The con-
centration of the analyte in the sample
solution, Cs, is calculated as:
*s
cs ' cix*-rrs-where
C, * Analyte concentration in fie
standard solution
A. - Absorbance of the unsplked
sample solution
A* * Absorbance of the spiked sample
solution
Calculations. M12 presents the
equation for calculating the stack gas
concentration of lead in milligrams per
cubic meter (rag/**}. Analogous equa-
tions should be used to calculate stack
gas concentrations for cadmium and
chromium.
3.3.1.4 Data l^iality Objectives—
Calibration Curve. No numerical data
quality objectives are established for
acceptance of th* standard curve (e.g.
correlation coefficient). However, the
method implies that a y-intercept value
exceeding 0.003 absorbance units Indi-
cates a potential problem with standard
solution preparation or nonlinearity of
the standard curve.
Accuracy. The data quality objective
{DQS} is that Cs calculated as shown
above from the spiked and unspiked
saaisle absorbances be within 5% of the
value obtained by reading the sample
concentration directjy from the cali-
bration curve. Thii Is equivalent to
an accuracy goal of 95 to 105X recovery
for each spiked analyte. If DQOs are
not met, all samples must be analyzed
by the methoa of additions.
23
-------
Precision. No numerical data quality
objectives for precision of the repli-
cate measureaents on samples or blanks
is stated. However. M12 states that
calibration standard analyses shall be
repeated until two consecutive ab-
sorbance readings agree within 3*.
3.3.2 Mercury
3.3.2.1 Method Description—
It is the consensus of the NESCAUM
Work Group that EPA Method 101A (Code
of Federal Regulations, Title 40,
Part 61, Appendix A) is the method of
choice for compliance testing of mer-
cury emissions in MWC stack gas.
Method 101A collects mercury 1n
impingers containing acidic potassium
permanganate solution. The basic sajn-
pltng method is EPA Method 5. However,
the use of a filter 1s optional 1n the
train for the M101A determination. The
filter may be required when sampling
streams with high particulate loadings,
as in locations upstream of air pollu-
tion control devices. The probe wash
solution specified in the method is 250
to 4CO mL of fresh C% KMnO,, solution.
After sample collection and recovery,
the probe wash and impinger solutions
are combined. If a filter was used in
the train, it is added to the acidic
pemanganate solution. This 1s taken
to near dryness and then digested with
concentrated nitric acid. The digested
solution is combined with the Impinger
solutions and the filter discarded.
The combined solutions are diluted in
an acidic aqueous medium, the mercury
is reduced to elemental form, and
determined by cold vapor AAS.
3.3.2.2 Critical Features—
The critical features of the M5 train
discussed In Section 3.2, apply to MiOi
as well. For mercury determination, it
is important to ensure that the col-
lecting medium (acidic permanganate
solution) does not come into contact
with any greased surfaces. It is also
necessary to avoid the use of metal
probe liners. 8oros1licate glass or
quartz are acceptable.
A ¦inimum stack gas sampling period
of 2 h (versus 1 h for standard M5) is
recommended. However, the method notes
that high organic material concentra-
tions may cause the •«&!lection medium
to become exhausted (bleaching of
purple permanganate color) before the
2 h sampling is completed. If the
Impinger reagents have no perceptible
purple color at the completion of the
run, the sample should be considered
invalid and the run repeated. To avoid
the need for resampling, the sampling
crew is advised to monitor the reagent
color throughout the rur and to .replace
the reagents with fresh solution if
significant fading is observed. The
impinger reagents would then be com-
bined prior to analysis.
The detection limit for mercury
analysis by cold vapor techniques is
about 0.2 sg/L in the solution as
analy2ed. Using the standard dilutions
specified in M101A (probe wash and
impinger solutions made to 1 I, then
diluted 2 to 250 mL), this corresponds
to an in-stack detection limit of
11 wg/a) for the recommended 2 h
(2.2 »3) sample. If a lower detection
limit is required two options are
readily available:
a. Collection of a larger stack gas
saaple, or
b. Use of a smaller secondary dilution
factor.
The first option may be limited because
of the relatively low capacity of the
permanganate collection reagent, espe-
cially in sources emitting significant
quantities of organic material. The
second, option is probably capable of
achieving a factor of 10 improvement in
sensitivity.
24
-------
3.3.2.3 Quality Assurance and Quality
Control Procedures
Calibration. The sampling equipment
must be calibrated as specified in M5.
Laboratory balances must be calibrated
according to a regular schedule. The
AAS system 1s calibrated for wavelength
accuracy according to the manufac-
turer's specifications,
M10IA specifies that quantification
be accomplished by reference to a six-
point external calibration curve,
covering the range of 0 to 1,000 ng of
mercury per analysis. The calibration
curve is not force-fit to zero.
Blanks. Method 101A specifies that
blank determinations be performed on
the collection medium and on the filter
(1f used). In addition, a blank cor-
responding to the 0 mg/l calibration
standard should be rerun after every
five samples. The standard and sample
data are blank corrected.
Replicates. The method requires that
repiicate~aliquots of diluted samples
an£ cf calibration standards be ana-
lyzed until two consecutive peaks agree
within the data quality objective (see
below.) It is the mean of the two
values that is used to build the cali-
bration curve (standards) cr calculate
concentrations (samples). Method 101A
requires that compliance determinations
be based on the mean of the mercury
emissions as measured in three separate
stack sampling runs.
Spikes. M101A does not require the
analysis of spiked samples. However,
it does recommend that at least one
sample from each stack test be checked
by the method of standard additions.
The NESCAUM Work Group recommends this
procedure for MWC tests.
Calculations. M101A presents the
equations for calculating the total
mercury in the sample (ug) and the
emission rate in grams per -lay (g/day).
The NESCAUM Work Group recommends that
mercury results also be reported in
concentration units (ug/m3).
3.3.2.4 Data Quality Objectives—
Isokinetlcity. the Isokinetic vari-
ance and acceptance criteria are the
same as for M5.
Calibration curve. The method does
not specify a goodness-of-fit data
quality objective (e.g., correlation
coefficient) for the calibration curve.
It
-------
Method 10, a sample of stack gas is
withdrawn through a stainless steel or
sheathed Pyrex glass probe equipped
with a filter to remove particulate
matter, a condenser and/or silica gel
trap to remove excess moisture, and an
ascarlte trap to remove C02, then
introduced into the NOIR analyzer.
There is no requirement for traversing
the stack or for maintaining Isokinetic
sampling conditions; the sample Is
obtained from the midpoint of the stack
or from another convenient location at
least 3.3 ft from the stack wall.
For carbon dioxide, the equivalent
method is EPA Method 3A, using an HDIR
Instrument for the C0a determination.
The probe, filter, and condenser/silica
gel trap are as in EPA M10. However,
the ascarite tube is not Included in
the train. In this method, it is spe-
cified that a minimum of 8 to 12 tra-
verse points be sampled.
For oxygen, the recommended method Is
Instrumental monitoring using a para-
magnetic or pol.:rographic analyzer.
The reference method 1s 3A. As de-
scribed aoove for CO and C02 deter-
minations, a gas conditioning system
for removal of particulate matter and
excess noisture must be included up-
stream of the instrumental analyzer.
3.4.2 Critical Features
Housing the analyzers in a clean,
thermally stable, vibration-free envi-
ronment will minimize instrument drift.
When the CO concentration in the
stack or flue gas is low (I.e., less
than 50 p?m), some field investigators
have found that varying {%) levels of
C02 may cause interferences in the CO
determination. The use of an NOIR
instrument equipped with a gas filter
correlation system will minimize these
effects. As an alternative, the NOIR
Instrument can be calibrated with gas
¦Ixtures containing CO in CO; at the
levels expected in the gas stream.
If the MWC is not equipped with a
system for acid gas removal, there Is
potential for corrosion of optical com-
ponents within the HDIR instruments.
However, most HC1 will be removed from
the gas stream In the condenser.
3.4.3 Quality Assurance and Quality
Control Procedures
The methods specify the use of cali-
bration gases for performance testing
of the- NOIR Instruments. A minimum of
three gas mixtures - one high, one mid-
range, and one zero - are to 6e used.
For CO analysis the gas mixtures should
be CO in nitrogen. {Mote that in cal-
culating the CO concentration In the
stack gas, a correction must be made
for the volume of C0a removed by the
ascarlte gas conditioner.) For C0a,
¦ixtures in nitrogen, air, or 07/C0z/
S02 matrices may be used. For C02, the
method specifies that system bias be
checked by introducing the calibration
gases at the outlet of the probe, up-
stream of the sample conditioning sys-
tem (condenser and, for CO analysis,
ascarite removal tube). Certified gas
mixtures, traceable to the National
3ureau of Standards should be used for
bias checVs.
Since the instrumental methods recom-
mended here are accepted EPA proce-
dures, it is not necessary to perform
confirmatory testing by manual proce-
dures each time the measurements are
made. However, 1f instrument problems
are observed, or if the response to
calibration gas mixtures 1s nonrepro-
ducible, manual checks may be advis-
able. For COi and! 02 determinations,
Orsat measurement can be made in paral-
lel with the Instrumental determina-
tions. Method 3A cross references
Method 6C (Instrumental determination
of sulfur dioxide emissions) for speci-
fication of calibration error, sampling
26
-------
system bias, zero drift, calibration
drift, and Interference check deter-
minations.
3.4.4 Data Quality Objectives
Method 10 does not set numerical data
quality objectives for CO determina-
tion. However, it states that the pre-
cision of most NDIR analyzers 1s
approximately 2% of full span and the
accuracy about 5% of full scan.
For C02 deteminations, Method 3A, by
reference to Method 5C, sets data qual-
ity objectives as follows: analyzer
calibration error less than 2% of span
for zero, mid-range, and high calibra-
tion points; sampling system bias less
than 5* of span for zero, raid-range,
and high calibration mixtures; zero
drift less than 3% over the period of
each run; calibration drift less than
31 over the period of each run; and
interference check less than 7% of the
C0t concentration observed In each
run. Method 3A , also specifies that
differences between the Orsat and
instrumental results greater ihan 0.555
C02 should be investigated. Possible
sources of discrepancies are leaks in
t.*e sampling system, inadequate mois-
ture removal, or fouling of the optical
exponents in the instrument. Sampling
should not be continued until the
source of the problem has been found
and corrected.
The data quality objectives for
oxygen determination are the same cs
those specified above for C02 under
Method 3A. However, 1n the case of
oxygen there Is no numerical objective
for agreement between the Orsat and
instrumental measurements.
Equations for calculating the st?ck
cas concentrations are given the the
respective methods.
3.5 TOTAL HYDROCARBONS (THC)
3.5.1 Method Description
The NE5CAUM Work Group recommends the
use of EPA Method 25A for determination
of THC in MWC emissions. In this
method (Code of Federal Regulations,
Title 40, Part 60, Appendix A), a gas
sample is extracted through a stainless
steel rake type probe, heated gas
sample line, and glass fiber filter to
a flame ionization detector (FID).
3.5.2 Critical Features
The FID Is calibrated using propane
(or other a'tkane) in nitrogen. The
results may be reported as ppm carbon,
ppm as methane, ppm as propane, etc.
However, the basis of the reporting
method must be stated.
In Method 25A, the FID is operated as
if It were a total hydrocarbon ana-
lyzer. This is, of course, only an
approximation since different organic
compounds have different FID response
factors. Highly oxygenated or halo-
genated organics, for example, have
relatively weak FID responses. How-
ever, the FID is as close to being a
"universal" hydrocarbon detector for
field use at ppm levels as any device
currently available. If the parameter
of concern is nonmethane total hydro-
carbons (NMTHC), the measurement can be
made both with and without a carbon
filter 1n place upstream of the ana-
lyzer. The difference between the
total and the filtered gas stream mea-
surements will provide a measure of
NMTHC. An alternative approach is to
incorporate a gas chromatographic (GC)
separation prior to FIO detection.
This can provide at least tentative
identification of the types (boiling
point distribution and possibly spe-
cific compound assignments based on
retention times) of hydrocarbons pres-
ent in the gas stream.
27
-------
EPA Method 25 is « possible alterna-
tive to Method 25A. However, Work
Group members report that Method 25 is
subject to interferences from the high
levels of COi arid water present In MWC
combustor effluents, and may give er-
roneously high results. The Work Group
thus recommends Method 25A, although
some agencies still require that
Method 25 be used to determine compli-
ance.
Acceptability of the proposed THC
method should be confirmed with the
appropriate regulatory agency(1es)
prior to testing.
3.5.3 Quality Assurance and Quality
Control Procedures
Calibration. Method 25A requires
that the FID response range be defined
using zero gas and a span gas contain-
ing 1.5 to 2.5 times the expected THC
level. A three point calibration curve
is then constructed for quantification.
Replicates. The method provides no
specific guidance on the number of
replicate measurements or the averaging
time to be used in determining compli-
ance with permit emission limitations.
These must be established in discus-
sions with regulatory authorities for
each test.
3.5.4 Data Quality Objectives
The method specifies that 2ero drift
and span drift over the duration of a
series of measurements must ba no more
than 3». Also, the calibration error
shall be less than 5% of the calibra-
tion gas value.
3.6 OTHER GASES
The other gases the NESCAUM Work
Group members consider to be of poten-
tial concern in most MWC compliance
testing are the acid gases: NQX, S02
as an indicator of sulfur oxides, and
hydrochloric acid (HC1). Methods for
the first two of thesr acid gas param-
eters are well developed. Methods for
measurement of HC1 in MWC emissions
have been less widely used although
validation has not been completed.
This is particularly true of methods
for continuous wnitoring of HC1, which
are still in a method development
stage.
3.6.1 Hydrochloric Acid
3.6.1.1 Method Inscription—
The NESCAUM Work Group agreed that
HC1 should be collected in the impirger
portion of the H5 train. After sacple
collection and recovery according to M5
procedures, the ispinger solutions are
combined and made to a convenient vol-
ume (e.g., 1 L) For analysis. The pH
1s adjusted as necessary to accommodate
the analytical finish. The CI" ion
concentration in the resulting solution
is determined by ion chromatography,
specific ion electrode, or titration
against standard silver nitrate or
mercuric nitrate (e.g., SW-S46
Method 9252).
3.6.1.2 Critical Features—
TTiere is a lack of consensus on the
impinger reagent most appropriate for
collection of HC1 from MWC stack gas.
The use of 0.1 K NaOH 1s frequently
specified. Some Vtortc Group meacers
recommend use of a sodium carbonate
solution, since this reagent will not
absorb COj. There is evidence to
suggest that caustic reagent is ^ot
necessary, and that HC1 is efficiently
trapped 1n any aqueous medium. If this
is true, then distilled water {e.g.,
ASTM Type II reagent water) will be the
reagent of choice for collection of
HC1.
The features of the M5 train cis-
cussed in Section 3.2 above, apply to
HCl sampling as well. It is acceptable
23
-------
to perform the HC1 determination on the
back, half of the train used for partic-
ulate Batter determination. It has
also been suggested (EPA Environmental
Monitoring Systems Laboratory, Research
Triangle Park, North Carolina) that HC1
collection can be Incorporated Into a
multiple metals sampling train. How-
ever, a train containing both basic im-
pingers for HC1 and acidic impingers
for «etals has not yet been tested. If
such a train were used, it would be
necessary to ensure that all impinger
solutions are representatively tested
for all species of concern. If an
M5-type train 1s to be used only for
HC1 determinations, it is not necessary
to sasole isokinetically.
The detection limit for HC1 will
depend on the detection limit of the
analytical technique chosen. Using
SW-846 Method 9252, the detection limit
in the solution as analyzed 1s less
than 2.5 mg/t_. For a standard M5 stack
gas volume of l.l m*, assuming a final
Impimger solution volume of 1 I, this
corresponds to an in-stack detection
limit of about 2.3 mg/m*. The de-
tect^cn limits of the ion chromatog-
raphy and specific ion electrode
determination methods are typically
less than I mg/L. Many samples may
require dilution to fall within the
linear range of the CI" determination
method, especially for samples col-
lected upstream of the acid gas removal
systea (or at older facilities with no
acid gas removal system in place).
3.6.1.3 Quality Assurance.and Quality
Control—
The quality control procedures recom-
menced in SW-846 Method 9252 are appli-
caole to each of the methods described
above. These include:
Blanks. A minimum of one blank per
saiaoie oatch. The blank should consist
of a volume of 1 M NaOH equal to that
charged to the MS impingers.
Replicates. A minimum of one dupli-
cate sample for ewery-10 field samples.
Spikes. Spiked samples or standard
reference materials shciuld be analyzed
"periodically."
3.6.1.4 Data (jMlity Objectives-
Numerical data quality objectives
have not been established for the
determination of HC1 in MWC effluents.
It is expected that a precision cor-
responding to bettBufcjthan 3X CV and an
accuracy of better should be
achievable by tae specified analytical
¦ethods at tCl emission levels of
SO ppm or higher. The uncertainty
introduced by sampling variations
cannot be assessed at this time.
3.6.2 Nitrogen Oxides
3.6.2.1 Method Description—
The NESCAUM Work Group reconnends EPA
Hethod 7E as the basic procedure for
ccspliance testing of MSW combustors
for M0X emissions. In this method a
sasple is withdrawn continuously and
conveyed to an instrumental chemi-
luainescent analyzer for determination
of N0X concentration. The Work Group
also considers EPA Method 7A an accept-
able alternative. In this method, a
grab sample of stack gas is collected
in an evacuated flask containing a di-
luted sulfuric acid/hydrcgen peroxide
absorbing solution. The nitrogen oxide
emissions, except rlitrous oxide, are
oxidized to nitrate! and determined by
icn chromatography.
In both methods, the sample interface
consists of a tjorosilicate glass,
stainless steel, or Teflon probe
equipped with a filter (glass wool
plug) to remove 'particulate matter.
Selection of sampling locations is
cross referenced to EPA Method 7.
29
-------
3.6.2.2 Critical Features—
Method 7A Is specific for K0X species
other than nitrous oxide. The instru-
mental analyzer used in Method 7E 1s
t"#sed on detection of NQX as NO and
Incorporates a converter to reduce any
NO j present to NO prior to quan-
tification. It is critical that the
converter efficiency be checked prior
to each test. It is also critical that
an acceptable water removal mechanism
be incorporated Into the sample condi-
tioning system.
The grab sampling method, 7A, gives
only a snapshot indication of the N0X
emissions. A number of replicate
determinations (minimum three per test)
are neeced to ensure that the results
are representative of source emissions.
The nuaoer of replicates and the
required avera^inj time for NO emis-
sions should by confirmed with the
regulatory agency(ies) for each MWC
test.
Particulate plugging problems may be
expected when applying either of these
methods upstream of air pollution con-
trol devices.
3.6.2.3 Quality Assurance and Quality
Control Procedures—
Stack sampling shipment and labora-
tory instrumentation are to be cali-
brated as stated in the methods or
manufacturer's specifications.
If available, audit samples should be
analyzed. Audit gas samples are pro-
vided by the U.S. Environmental Protec-
tion Agency's Environmental Monitoring
Suoport Laboratory (Research Triangle
Park, North Carolina).
Methcd 7A requires the generation of
a five-point calibration curve for
quantification. Hezhod 7E requires the
use of t*ree calibration gases consist-
ing of HO In nitrogen at high, mid-
range, arc zero concentration levels.
Neither method explicitly requires a
blank determination and requirements
for replicate sample determinations are
not stated.
3.6.2.4 Data Quality Objectives—
Both Method 7A and Method 7C require
that the results of the QA audit sample
analysis be within of the actual
audit concentrations. Method 7A also
requires that the calibration curve be
sufficiently linear than none of the
five calibration points deviates from
the linear regression line by more than
7% of the concentration at that point.
Method 7E requires conformance to the
data quality objectives stated in
Method 6C for an;.!.zer calibration
error and sample system bias. Neither
method present numerical data quality
objectives for precision and accuracy
of sample measurements.
3.6.3 Sulfur Dioxide
3.6.3.1 Method Description—
The NESCAUM Work Group preferred
method for compliance testing measure-
ment of sulfur dioxide is EPA Method 6.
The basic sampling train specified in
this method consists of: a boro-
silicate glass or stainless steel probe
with a quartz or pyrex wool plug for
particulate removal; a midget bubbler
containing isopropyl alcohol for
absorption of S03; and midget impingers
containing hydrogen peroxide solution
for collection of S02; plus the usual
valves and metering system. The method
also allows the options of sampling S02
In an MS train, with hydrogen peroxide
replacing water in the impingers, or a
Method 8 train (like M5 but with an
unheated filter placed between the
first two impingers), with isopropyl
alcohol in the first impinger and
hydrogen peroxide in the back-up
Impingers. After sample collection and
recovery, the contents of the hydrogen
peroxide imoingers are combined and
30
-------
rnade to volume with distilled water.
The S02 is determined by as sulfate ion
oy titration with barium perchlorate to
a Thorin endpoint or by ior, chromatog-
raphy. An acceptable alternative is
Kethod 6C, based on an instrumental
analyzer.
3.6.3.2 Critical Features-
Free ammonia is & known interference
in Method 6. This could be a problem
wnen testing at an MWC that uses post-
ccmbustion NO, control techniques, such
as Selective Catalytic Reduction (SCR).
The midget bubbler containing the
isopropyl alcohol fcr S03 removal must
be packed with glass wool at the top to
prevent carry-over of sulfuric acid
mist.
When sampling in locations upstream
of particulate matter control devices,
it flay be necessary to use a high effi-
ciency glass fiber filter in place of
tf*e glass wool plug. High concentra-
tions of cations generated by collec-
tion of fine metal fumes in the hydrc-
cen peroxide impingers would interfere
ir tne analysis.
vethod 6 sets no specific criteria'
fcr selection of sampling sites. It is
reccwnended that sampling be performed
at a point of average velocity near the
center of the stack/duct, at least 3 ft
a>»ay from the stack wall. Isokinetic
Sonoling is not required.
The detection limit of Method 6 Has
been determined to be 3.4 mg/mi of
S02. The upper limit of the method is
afccut 93,000 mg/mJ for a 20 L sample of
stick gas.
"hen the instrumental analyzer
Metnod 6C is used, it is necessary to
crecx the instrument performance using
Method 5 as a reference method.
3.6.3.3 Quality Assurance and Quality
Control Procedures—
Calibration. Sampling equipment must
be calibrated and leak-checked as
described in the method. Laboratory
equisnert, including balances, must be
caliarrced on a regular schedule in
accordance with manufacturers' speci-
fications.
The collection efficiency of the sam-
pling train must be determined by per-
forming at least three tests with a
fourth hydrogen peroxide impinger in
the train. The contents of the fourth
impinger are analyzed separately.
The titrimetric quantitative analysis
technique is calibrated by titration of
sulfuric acid solutions that have been
standardized against 0.0100 N NaOH,
which has been standardized against
prf»»»-y standard grade potassium acid
phthalate.
Blanks. The method requires the
analysis of a blank with each series of
samples.
Reolicates. The method requires that
eacn sample be analyzed in duplicate.
Audit samples. If available, audit
gas sarsles should be analyzed. Audit
gas samples are provided by the
U.S. environmental Protection Agency's
Environmental Monitoring Support
Laboratory (Research Triangle Park,
Nortft Carolina).
3.E.3.4 Data Quality Objectives-
Precision. Reslicate titrations must
agree -itnin iX or 0.2 mL, whichever is
larger.
Accuracy. Numerical criteria for
equirrent calibration and leak checks
are given in the method.
31
-------
The results of analyses of the QA
audit safflples must agree to within 5%
of the known value.
The collection efficiency must be at
least 99V for a series of three tests.
This requires that the sulfate content
of the fourth impinger not exceed IX of
the total sulfate recovered from the
first three inpingers.
SECTION 4.0
RESULTS REPORTING
4.1 OVERVIEW
The NESCAUM Work Group agreed that
the reporting of results of MWC test-
ing, whether for compliance, perfor-
mance, or research and development,
should use a format that meets the
following criteria:
• Allows the reader to readiiy
determine what the test objectives
were and whether they were met,
• Expresses all data for compliance
testing in the units specified in
the permit and/or regulation,
• Provides an indication of the data
quality in terms of range, standard
deviation, or error bounds, and
compares the actual data quality
indicators with the data quality
objectives set for the tests,
• Cross-references summary data to
more ccoplete back-up information,
which say be in appendices, and
• To the extent possible, meets the
needs of multiple potential users:
permi tters /regu 1 a tors, 1 etj 1S1 ator s,
air suality modelers, risk
assessors/toxicolog-ists, data base
builders.
The Work Group felt that it would not
be burdensome to require that results
be reported in a consistent fcraat for
all tests. It was recognized that this
might require reporting some results in
two different forms—once in the com-
pliance units and once in the NESCAUM
cQsoon format.
The group developed a list of
auxiliary parameters^to be reported.
Along with basic process operating data
are included the information neeced to
convert the data from the NESCAUM com-
mon units to units appropriate for
other specific uses, e.g., allow easy
conversion from stack concentration in
iiO/b' to emission rate in gra-ns per
second (g/sec) or emission factor in
pounds per ton (lb/ton) of refuse. The
Work Group recognizes that so-ie of the
process operating data may net be
readily available for all tests. For
example, MWC facilities will usually
net have measurements of the heat con-
tent of the refuse in Btu/lb, although
they may have estimates based cn the
calorimetric performance of the com-
bustpr. To the extent possible, esti-
mated values of nonmeasured parameters
should be reported, along with an indi-
cation of the method used tc develop
the estimate. The Work Grace- recom-
mends that this list of auxiliary
parameters to be reported snauid be
used as a checklist by the fieic: test-
ing crew to ensure that as mucn as pos-
sible of the potentially useful infor-
mation is recorded for each unit
tested.
4.2 RESULTS TO 8E REPORTED
4.2.1 Facility Operating Status
The Work Group decided that the fol-
ding process data should be included
in the body of an MWC stack test
report:
* Feedrate of waste expressed as:
(1) the rate of steam production.
32
-------
(2) the percentage of the design
capacity steam production, (3) the
average ^asured or calculated mass
loading of waste (tons/hour (T/hj or
tons/day [T/dayl, and (4) the heat-
ing val-e of the waste in Btu/lb
(measured, estimated, or assumed for
purposes of calculating feedrate).
• All available data concerning the
composition and source of refuse
(e.g., i moisture, % ash, % resi-
dential, etc.).
• Combustion temperature(s) supported
by a sci^ematic drawing indicating
the measurement location(s), speci-
fication of the dev1ce(s) used for
temperature measurement, and an
indication of how the devices were
shielded, etc. (Note: Measurement
of combustion temperatures is still
an inexact science. Meaningful
comparisons between facilities can-
not ba srawn unless this supporting
information is reported.)
• Pressure arop across the boiler.
• Percent excess air.
• Air po'T-tion control device param-
eters (s-ch as ESP electrical data).
• If a scot blowing episode(s) was
includes in the sampling period, its
duration.
• Facility status prior to test, i.e.,
"cold" start or continuing opera-
tion.
•The Work. Group also agreed that most
other ancillary process data that were
recorded by the tester or by the facil-
ity durirg the testing period should be
included in appendices to the report.
It is felt that this represents an im-
portant ccsortunity to document an
operatiora- baseline for the facility
during a c;~e when its emission status
was estaD*.;shed. These results would be
potentially useful to owner/operators
or regulators in trouble-shooting prob-
lems that might arise in the future.
Table 4-1 presents a list of facility
operating data that should be recorded
for each test run.
4.2.2 Combustion Process Oata
The report should include, in tabular
fcrnat, a summary of data relating to
the overall combustion and air pol-
lution control device performance.
This would include, for each test run:
4.2.3 Permit Conditions
The Work Group believes it would be
helpful to the MWC test report reader
to have a tabular summary of the emis-
sion limitations (if applicable) im-
posed on the facility as permit condi-
tions in the body of the report.
Table 4-2 is a pro fonr.a table for
listing permit conditions.
4.2.4 Pollutant Emission Data
For each pollutant tested, the rsoort
should show, in tabular form, the value
xeasured for each run as well as the
mean of the three (or more) replicate
stack tests. It was felt that a table
showing only the mean value would not
convey sufficient information concern-
ing the uncertainty of the results.
Tnis could result in misuse of the data
for comparative purposes.
The Work Group consensus is that
metric units should be used con-
sistently in the resorting of MWC
emission data. ?or organic and
inorganic chemical pollutants, the Work
Stack temperature
Stack gts % moisture
% Isokineticlty
Stack gas flow rate (mVsec)
Concentrations of combustion gases
and acid gases
Particulate loading
23
-------
TABLE 4-1. MWC FACILITY OPERATING DATA
1. Boiler/Furnace Data
Steam flow (Ib/h) set point and actual
Superheater outlet temperature (*F)
Superheater outlet pressure (psi)
Feedwater flow (lb/h)
Feedwater temperature (*F)
Combustion air temperature (°F)
Furnace draft (in. H20)
Total comtustion airflow (acfm)
Secondary air pressure(s) (in H20)
Furnace temperature(s) ('F)
Economizer exit gas temperature (°F)
Excess Air (%)
CO and 0? (ppmv and %)
Sootblowing (frequency and duration)
2. Electrostatic Precipitator Operating Data
Secondary field current (mA) for each field
Secondary field voltage (KV) for each field
Rappii.g cycle (frequency and duration)
3. Acid Gas Absorber Operating Oata
Inlet temperature (°F)
Outlet temperature (*F)
Wet: Slurry concentration {% or specific gravity)
Slurry feedrate (gpm)
Dilution* water flow rate (gpm)
Atomizing air pressure (psi)
Control mode - automatic/manual
Ory: Lime injection rate (Ib/h)
4. Baghouse Operating Oata
Inlet temperature (*F)
Outlet temperature (°F)
Pressure drop (in. HjG)
Cleaning cycle control - automatic/manual
Cleaning cycle - pressure drop versus timed cycle
Cleaning cycle - frequency and duration
34
-------
TABLE 4-2. PRO FORMA TABLE FOR REPORTING PERMIT COHOITIOHS AMD TEST RESULTS
Penait Averaging Test
Parameter Units limitation time results Cwnment/reference
Process paraieters
Furnace temperature
Refuse feedrate
Other (e.g., combustion efficiency:
Stack emissions
Particulates
CO
2,3,7,8-TCDD
equivalents
Lead
Mercury
Cadmium ¦
Chromium
HC1
NOx
SO,
Others:
a Citation for equivalents calculation.
35
-------
Group agreed that the appropriate unit
for reporting In the ccssci NESCAUM
format should be ug/NmJ or ng/NmJ.
Normal conditions are defiQ»«-=«^v dry,
0"C, and 1 atm corrected $7% 02y The
correction factor (CF) for/STl!, Is:
CF = —
Lr 20.9-f
where Y is the volumetric oxygen per-
centage measured in the stack.
The Work Group actively considered
both 12% C02 amd 7% 02 as tfee basis for
normalizing concentrations. Although
some permit conditions {e.g., particu-
late matter emission limitations! con-
tinue to be written on a 12% C32 basis,
the Work Group finally chose 7% 02 as
its recommended normalizing factor.
One reason cited for this selection was
that it obviates the need to specify
whether CO, produced by corabustion of
auxiliary fuel should "count" in the
correction factor. On the other hand,
the use of the 7% 02 basis is
complicated by possible in-leakage of
air between the combustion zone and the
measurement location. In the end, the
Work Group recognized thai this choice
was somewhat arbitrary and that neither
basis was strongly preferable to the
other on tecnnical grc^nas. The Group
did feel strongly that a single con-
sistent method of correcting the data
should be adopted.)
Seme state agencies also require that
uncorrected data be reported. If this
is a requirement, the Work Sroop recom-
mends that the raw (uncorrected) data
be included in an aacencix to the
report of test results.
Gaseous species (Oj, C02, CO, HC1,
S0X, and N0X) are exceptions to this
general rule and should be reported In
concentration units of percent or parts
per million volume (pprcv) corrected ta
standard conditions of: dry, 68JF,
1 atm, and 7% 0?.
In the NESCAUM consensus format, par-
ticulate matter should be reported in
milligrams per cubic meter (mg/Nm}) at
the same standard conditions. If the
emission limitation specified by the
regulation/permit is simply comparable
to the reported emission concentrations
for a particular pollutant, the group
felt that the emission limit should
also be' included in the summary table.
4.2.5 Conversion Factors
As noted above, the Work Group felt
that units reflecting mass of pollutant
{or volume for gases) per volume of
stack gas is the most generally useful
format for many purposes. However, it
was recognized that other reporting
bases are more convenient for many pur-
poses. Other types of units in common
use include:
• mass/time (e.g., g/sec)
• mass/mass c-f refuse (e.g., lb/ton)
• mass/HHV (e.g., lb/10* Btu)
• mass/mass of stack gas (e.g.,
lb/1,000 lb flue gas)
To accommodate users who wish to con-
vert the mass/volume concentration
units into one of these alternative
bases, the data shown in Table 4-3
should be included in the report.
Table 4-3 lists only the facility-
specific information that is required
for the conversions. Factors for
converting between English and metric
units or from nanogram to gram (ng to
g) can be included in an appendix to
the test results report. Table 4-3
also includes the heating value of the
refuse (Btu/ton) as one of the test
conditions to be reported. The
"comment/reference" column of the table
should indicate the basis for the
reported value (e.g., direct measure-
ment, calculated from steam production,
etc.).
36
-------
TABLE 4-3. PRO FORMA TABLE FOR TEST CONDITIONS DATA
Run
Parameter
Unit
1
2
3
Mean
RDP*
Comment/reference
u
Operating conditions
Refuse
Feedrate*
Heat content*
(as received)
Moisture* X
Total ash*
Steam flow rate
Heat absorbed by
boiler
Boiler efficiency
T/day
Btu/lb
%
lb/h
Btu/h
% of heat release
Slack gas parameters
Flow rate
Temperature
Moisture
NmVs
dscf/oitn
dscf/3tu
4b/h (dry)
Test time
24 h clock
* Indicates basis for estimate.
-------
4.3 DATA TABULATION FORMATS
Tables 4-4 through 4-7 present pro
forma tables for data reporting. The
NESCAUM Work Group feels strongly that
all data should be presented with an
appropriate Indication of Its reliabil-
ity. This 1s accomplished, In part, by
indicating the Individual measurements,
as well as the mean, 1n all data
tables. In addition, the group recom-
nends that any measurement not meeting
the relevant data quality objectives
(precisian, accuracy, recovery of
spiked material, etc.) should be
reported conditionally (e.g., asterisk
to an explaining footnote).
4.4 DATA QUALITY RESULTS REPORTING
Table 4-8 presents a suggested format
for reporting data quality results and
for comparing them with data quality
objectives. For purposes of Illustra-
tion, only the PCDD/PCDF data quality
results are shown 1n the pro forma
table. For an actual test, comparable
results should be reported for each
indicator of DQO attainment for each
parameter tested.
4.5 REPORTING "NOT DETECTED" VALUES
In an Ideal world, the issue of
reporting "not detected" (NO) results
would be a moot point, because the
sampling and analysis methods used
would have detection limits so far
below the level of possible concern
that an NO was functionally equivalent
to zero. Unfortunately, especially in
the case of PCDO/PCDF and some trace
metals, the analytical detection limit
and the lower level of concern (or
emission limitation) are essentially
the same (to within an order of
magnitude). This creates a temptation
to assign a concentration to NO values
equivalent to the detection limit or
some fraction of the detection limit.
When the emission limitation applies to
the sum of concentrations of a number
of related pollutants (e.g. PCOO/PCDF
or PCBs) the assignment of finite
numerical concentrations to a series of
NO results can easily give rise to
false positive results.
NO values should not be assigned
finite concentration values for sum-
mation purposes. The NO values should
be reported as such in the data summary
tables, with the detection lisit of the
method for that sample shown in paren-
thesis, e.g.,
2,3,7,8-PeCOO NO (7 no/fta^)
If other PCDOs above the detection
linit had been found 1n the saaole, the
total PCDD concentration would be
reported as the sum or the apparent
positive findings, but wcwld not
include a 7 ng/m3 contribution from
PsC00. If no PCDO had been found 1n
the sample, the total would be reported
as NO. However, in this case the
number in parentheses probably should
be the sum of the detection limits.
SECTION 5.0
CITED REFERENCES
Code of Federal Regulations, Title 40,
Part 60, Appendix A, July 1, 1985.
U.S. Environmental Protection Agency,
Test Methods for Evaluating Solid Waste,"
SW-846, Third Edition, November 1986.
American Society of Mechanical Engi-
neers, Sampling for the Deter-ritation of
Chlorinated Organic Compounds in Stack
Emissions and Analytical Proc edtxes to
Assay Stack" Effuent Samples and Residual
Combustion Products for Polycniorinated
Dilwiro-p-dioxim (PCDO) arxi Poly-
chlorinated Dibenzofurans (PCDF),
September 18, 1984.
Code of Federal Regulations, ?*'tle 40,
Part 61, Appendix A, 1987.
38
-------
TABLE 4-4 PRO FORMA TABLE FOR CONVENTIONAL EMISSION PARAMETER
Run
Parameter
Unit
1
i
3
Me on
RPD
Comment/reference
-¦ i i ¦ .n«
«o
Stack gas flow rate
Stack (jdi velocity
Stack temperature
Stack gas moisture
Isoklnetlclty
Particulate loading
Carbon monoxide
Carbon dioxide
Oxygen
1IC1
«ox
SO,
dscf/s
ft/s
m/s
%
%
mg/dscm
gr/dscf
Pf"»
X
%
ppmv
ppmv
ppmv
-------
1AML 4-5.! EXAMHlt PRO fORHA 1 ABU FOR PIOXINS AMD fURANS IMISSIOH OATA
Run
Compound or howto Vog
Metho.Ja
Units
i
2 j
Mean RPD Comment/reference
Total TCODs
2,3,7.8-TCDO
Total PeCODs
2,3,7,.8-PeCDDs
Total HxCDOs
2,3,7,8-HxCDDs
Total HpCDDs
2,3,7,8-HpCOOs
ocoo
ng/Nm3
ng/Nm3
ng/Nm3
ng/Nm3
ng/Nm3
ng/Nm3
ng/Nm3
ng/Nm3
ng/Nm3
Total TCOFs ng/Nm3
2,3,7,8-rCOF ng/NmJ
Total PeCDfs ng/Nm3
2,3,7,8-PeCDFs ng/Niii3
Total HxCDFs ng/Nm3
2,3,7,8-HxCDFs ng/NmJ
Total HpCDFs ng/Nm3
?»3,7,8-HpGt)Fs ng/Nm3
OCDF ng/Nm3
ToUl 2,3,7,R-TCDD equivalents0
a Method reference, e.y.,' MM5/SW-846 or MM5/GC/MS.
^ Dry normal at 7% 07.
c Citation for equivalents calculation.
-------
TABLE 4-6. EXAMPLE PRO FORMA TABLE FOR METALS EMISSION DATA
Run
Element
Method3
Units
T
2
3
Mean
RPO
Comment/reference
iig/Nm*
iig/Nm)
iig/Nin'
a EPA isetliud number (e.g., M12) or description (e.g., M5/SW-846 3050/ICP).
b Dry normal at 7% Oz.
Lead
Mercury
. Cadmium
¦
Chromium
Others:
(Hst)
-------
lAIiLE 4-7. PRO FORMA lAUlt
FOR 011ILR ORGAN ICS
EMISSION DATA
Run
Compound or homolog
Method3 Units
1 2 3
Mean RPO Comment/reference
n
-------
TABLE 4-8. EXAMPLE PRO FORMA TABLE FOR OATA QUALITY RESULTS
Parameter
Criterion
Data quality
objective
Data quality
results Comment/reference
2,3,7,8-TCOD
MH5 isokineticity
90-110*
MIS recovery
50-1502
Detection linrit
0.15 ng/Nm*
QC check saaple
60-120%
accuracy
MIS precision
< SOX RSO
Lead
MS isokineticity
90-110*
43
-------
SECTION 6.0
ADDITIONAL SOURCES OF INFORMATION
Ada»s* R. E., R. H. Janes, L. B. Farr,
M. M. Thomason, H. C. Miller, and L. D.
Johnson, "Analytical Methods for Deter-
mination of Selected Principal Organic
Hazardous Constituents in Combustion
Products," Environmental Science and
Technology, 20, 711 (1986).
Bursey, J., m. Hartman, J. Homolya,
B. McAllister, J. McGaughey, and
0. Wagoner, Laboratory and Field Evalu-
ation of the Semi-VOST Method, Vol- J,
EPA 600/4-S5-075a, NTIS PB86-1235d1/AS
and Vol. II, EPA 600/4-35-075b, NTIS
PB86-123569/AS, September 5, 1985.
"Determination of Total Gaseous Non-
methane Organic Emissions as Carbon,"
Method 25.1, South Coast A1r Quality
Management District (California), Draft
Report ,• Prepared by Technical Services
Division, January 1987.
"Determination cf Total Gaseous
Nonraethane Organic Emissions as Carbons
Using a Flajne Ionization Oetector,"
Method 25.2, South Coast Air Quality
Management District (California), Draft
Report, Prepared by Technical Services
Oivision, January 1987.
deVera, F. R., B. P. Simmons, B. 0.
Steshens, and 0. L. Stons, Samplers and
Sampling Procedures for Hazardous Waste
Streams, EPA 600/2-80-018, NTIS P880-
135353, January 1980.
Fuerst, R. G., T. J. Logan, M. R.
Midcett, and J. Probaska, "Validation
Studies of the Protocol for the
Volatile Organic Train,* Journal of the
Air Pollution Control Association, 37, 4,
388 (1987).
Gornan, P.. R. Hathaway, D. Wallace,
and A. Trenholm, Practical Cuide - Trial
Bums for Hazardous Waste Incinerators,
EPA-6CO/2-86-C50, NTIS PB86-190246/AS
(1986).
Guidelines and Specifications for Preparing
Quality Assurance Program Plans, Quality
Assurance Management Staff, Office of
Research Development, USEPA, Washing-
ton, DC, QAMS-004/80; EPA 600/8-83-024,
NTIS PB83-219667, September 1980.
Handbook for Sampling and Sample
Preservation of Water and Wastewater,
EPA 600/4-82-029, NTIS PB83-124503,
September 1982.
Hansen, E. M., Protocol for the Collection
and Analysis of Volatile PQHCs Using VOST,
EPA 600/8/84-007, NTIS PB84-170042,
March 1984.
Harris, J. C., D. J. Larsen, C. F.
Rechsteiner, and K. E. Thrun, Sampling
and Analysis Methods for Hazardous Waste
Combustion, EPA 600/8-84-002, NTIS
P834-155845, February 1984.
Interim Guidelines and Specifications for
Preparing Quality Assurance Project Plans,
Quaiity Assurance Management Staff,
Office of Research Development, USEPA,
Washington, CC, QAMS-005/30; EPA 600/4-
83-C04, NTIS PB83-170514, December
1980.
James, R. H., R. E. Adams, J. M.
Finkel, H. C. Miiler, and L. D.
Johnson, "Evaluation of Analytical
Methods for the Determination of POHC
in Cocoustion Products," J, Air Pollution
Control Association, 35, 9, 959 (1985).
Jayanty, R. K. M., S. W. Cooper, C. E.
Decker, and D. J. yon Lehmden, "Evalu-
ation of Parts-Per-8illion Organic
Cylinder Gases for |Jse as Audits During
Hazardous Waste Tr,ial Burn Tests," J.
Air Pollution Control Association, 35, 11,
1155 (1985).
44
-------
Oayanty, R. K. M., J. A. Sokash, W. F.
Gutknecht, C. E. Decker, and D. J.
von Lehoden, "Quality Assurance for
Principal Organic Hazardous Consti-
tuents (POHC) Measurements During
Hazardous Waste Trial Burn Tests,"
J. Air Pollution Control Association, 35, 2,
143 (1935).
Johnson, L. 0., "Detecting Waste Com-
bustion Emissions," Environmental Sci-
ence and Technology, 20, 223 (1986).
Johnson, L. 0., and R. G. Merrill,
"Stack Sampling for Organic Emissions,"
Toxicological and Environmental Chemistry,
6, 109 (1983).
Jungclaus, G. A., P. G. Gorman,
G. Vaughn, G. W. Scheil, F. J. Bergman,
I. 0. Johnson, and D. Friedman,
Development of a Volatile Organic Sampling
Train (VOST), presented at the Ninth
Annual Research Symposium on Land
Disposal, Incineration, and Treatment
of Hazardous Waste, Ft. Mltcneil, KY,
May 15&j. In Proceedings, EPA 600/9-
84-015, *TIS PBS'—234525, July 1984.
Lentzen, 0. E., D. E. Wagoner, E. D.
Ester, and W. F. Gutknect, /ERL-RTP
Procedures Manual: Level 1 Environmental
Assessment (Second Edition), EPA 600-7-78-
201, P2-293795, October 1973.
Maraescn, J. H., J. E. Knoll, M. R.
Midgect, D. E. Wagner, J. Rice, and
J. B. Hcmolya, "An Evaluation of the
Semi-VOST Method for Determining Emis-
sions from hazardous Waste Inciner-
ators,* Air Pollution Control
Association, 37, 9, 1067 (1587).
Menzies, K. T., and J. W. Adams,
Instrumental Monitoring of So nmethane
Hvdroccrbons at a Hazardous Waste
incinerator, EPA 600/2-86-077, NTIS PB87-
102331/AS, September 1986.
Methods for Chemical Analysis of Water and
Waste, EPA 600/4-79-020, NTIS
PB29-7*36, March 1979, revised 1983.
Nolen, S. L.f M. O. Jackson, and D. 6.
Harris, Mobile Laboratory for OnSit*
Monitoring of Hazardous &atte Incirteratort,
EPA 600/8-86-039, NTIS Pe87-140835,
November 1986.
Podlenski, J., E. Peduto, R. Mclnr«es,
F. Abell, and S. Gronberg, Feasibility
Study for Adapting Present Combustion
Source Continuous Monitoring Systems to
Hazardous Waste Incinerators, Volume I,
Adaptability Study ant Guidelines Docu-
ment, EPA 600/8-84-110a, NTIS PS84-
187814, March 1984.
Prohaska, J., T. J. Logan, R. G.
Fuerst, and M. P.. Midgett, Validation of
the Volatile Organic Sampling Train (VOST)
Protocol, Volume I, Labcraton; Phase, EPA
600/4-36/014a, NTIS P386-145547, U.S.
Environmental Protection Agency,
January 1986.
Prohaska, J., T- J. Logan, R. G.
fuerst, and M. R. Kidgett, Validation of
the Volatile Organic Sampling Train (VOST)
Protocol. Volume U, Field Validation Phase,
EPA 600/4-36-014b, NTIS P386-1455:4,
U.S. Environmental Protection Agercy,
January 1986.
Rappe, C.t "Analysis of ?oiychlorine*ed
Dioxins and rurars," Environ. Sci. Tee..t-
nol., 18, 13, 3, 78A (198i).
Schlickenrieder, L. M., J. W. Adisss*
and K. E. "i'hrun, Modified Method 5 Train
and Source .Assessment Sampling System
Operators Manual, EPA 6CG/2-35-003, *7IS
PB85-169873, February 1955.
Shaub, W. M., and W. Tsang, "Dioxfn
Formation in Incinerators," Environ. ZcL
Tecftnot., 1", 21, 721^1923).
Stern, D. A., 8. M. Hyatt, J. f.
Lachowski, and K. McGregor, Soect-
ation of Halogen and Hydrogen Halide Com-
pounds in Caseous Emissions, in ?-a-
ceedings of the Ninth Annual Rese;r'ch
Symposium on Incineration and Treatment
of Hazardous Wastes, EPA 600/9-84-CI5,
NTIS PSS4-234525, July 1934.
45
-------
APPENDIX A
ADDITIONAL MEASUREMENT METHODS FOR
RESEARCH TESTING
A-I
-------
1.0 P01YCHL0RINATED BIPHENYLS
The methods presented In this section
are appropriate for measuring poly-
chlorinated biphenyls (PCBs) in MWC
stack emissions. Monochloro through
aecachloro PCBs are determined by
hoaolog.
1.1 METHOD DESCRIPTION
Stack emission samples are collected
for PCB measurement using an MM5 train
Identical to that described in
Section 3.1 of this document for mea-
suring PCOOs and PCDFs. The sample
components recovered from the train are
extracted, the extracts are combined
and cleaned to remove excessive levels
of potential interferences, and the
cleaned extracts are analyzed for PCBs
by hoosolog using GC/MS.
1.1.1 Sample Collection
The MM5 train configuration, prepara-
tion, operation, and samDle recovery
described for determining PCDDs and
PCDFs in MWC stack gases (Section 3.1
of this document) can be directly
apolied to measuring PCBs.
1.1.2 Sample Analysis
A primary reference for determining
PCSs in MWC stack samples Dy hcmolog is
EPA Methoa 680. The GC/KS analysis
procedure was previously described by
Erickson eta!. Method 630 addresses
PC3 ineasurements in water and soil/
seciment by GC/MS. Although the sample
preparation, extraction, and extract
cleanup requirements for MWC stack sam-
ples differs somewhat from *ater and
soil/sediment samples, the SC/MS-SIM
analysis and quantitation protocol
presented in Method 680 are directly
anolicable.
1.1.2.1 Saaple Preparation and Ex-
traction-
Preparation and extraction of MM5
sample components described for PCDOs
and PCDFs (Section 3.1.1.2 of this
document) can be directly applied to
samples for PC8s with two exceptions.
Samples for PCB determination must be
spiked with a solution cf surrogate
recovery standards rather than the PCOO
and PCDF MIS solutfort. Also, MM5 resin
and particulate samples can be ex-
tracted for PCBs using dichloromethane
rather than toluene or benzene.
The PCB surrogate compounds are used
only to assess PCB recoveries rather
than as the method internal standards
used for PCDOs and PCDFs. PCB quan-
titation does employ an internal stan-
dard spiked just prior to GC/MS
analysis. The PCB surrogate solution
should contain at least four re-
labeled congeners representing differ-
ent hcmologs. The spike level should
be in the range of 5 to 10 times the
anticipated detection liait. Example
labeled congeners which may be used as
recovery surrogates include the
following:
»3C6-4-monochlorobipheny1
13C, 2-3»31,5,5',6,6'-cctachoro-
biphenyl
*>Cia-2»2',3,3',5,5',6,6'-octa-
chlorobiphenyl
lJCl2-decachlorobiphenyl
Since the sample preparation and
extraction of MM5 samples for PCBs is
the same as for PCDDs an
-------
Unequal SDlittinc, can be employed,
depending on target detection limits.
1.1.2.2 Extract Cleanup—
The extracts from each MM5 train
should be combined, dried by passage
through a short column of anhydrous
sodium sulfate, and concentrated to
5 ml using Kuderna-Danish evaporators.
A 15-tnL aliquot of hexane should be
added and the extract concentrated to
2 raL using Kuderna-Danish evaporators
and nitrogen blowdown. The concen-
trated extract should then be cleared
by chrcmatograpny prior to tw'MS
analysis. A detailed protocol for
cleanup using florisil is presented in
Method 3520, SW-856. A silica gel
cleanuo procedure, as described in
Metnou 3630^-^-846 can also be used.
The. craned extract should be concen-
trated to 500 ul using Kuderna-Danish
evaporators and nitrogen blowdown.
1.1.2.3 GC/MS Analysis-
Sample extracts rcust be analyzed by
GC/MS using a fuseo silica capillary
column and selectee ion monitoring
(SIM) aata acquisition. A detailed
protocol is presented in Method coO.
The recommended capillary column is a
30-m DB-5 or SE-54.
PC3s and recovery surrogate ccnscunds
are quantitated in the extracts using
response factors romalized to inte*-na<
standard compounds saike into tne
cleaned extract just prior to analysis.
The relative response factors (;*F$)
are determined by analyzing at least
three levels of calibration solutions
prior to sample analysis. The calibra-
tion solutions should contain the
native PCB congeners, retention time
calibration compouncs, and internal
standard compounds shown in Table A-i.
The levels of t^e native congeners
snould cover a 100-fold concentration
range and the lowest concentration
should be 5 tines the practicao'e
detection limit. The retention time
calibration and internal standard com-
pounds should be present at the same
level in all solutions. The calibra-
tion solutions should also contain the
recovery surrogate compounds at con-
centrations corresponding to the spike
level in samples.
The RRFs for the native congeners and
recovery surrogates should agree with a
relative standard deviation (RSD) of
20X for all compounds. Once acceptable
calibration has been achieved, 1t is
verified each day of analysis by
analyzing the mid-level standard at the
beginning and end of the day. RRFs for
the daily standards must be within 2GS
of the calibration RRFs, otherwise
recalibrate is necessary.
The concentrations Of mono through
decachlorobiphenyl homologs are deter-
mined using their RRFs and are reported
as mass of homolog per sample volume,
e.g., nanogram of tetrachlorobiphenyl
per normal cubic meter (ng/Nm*).
Detection limits should be determined
for each homolog not detected using a
response of 2.5 times the background
noise signal at the appropriate reten-
tion time. The recoveries of the sur-
rogate compounds are also quantitated
using RRFs. The surrogate recoveries
should be within the range of 50 to
15C*. Recoveries outside this range
may indicate problens in the analysis.
1.2 CRITICAL FEATURES
1.2.1 Sample Collection
The critical features of NW5 sample
collection for PCBs are essentially the
same as those for PCDDs and PCOFs, as
described in Section 3.1.2.1 of this
document.
1.2.2 Sample Analysis
The critical features of analysis of
MWC stack samples for PCBs are somewhat
A-3
-------
TABLE A-l. CALIBRATION COMPOUNOS CONTAINED IN PCB STAHOARO SOLUTIONS
PCB homo log
Congener
number4
Chlorine substitution
Concentration Calibration Standard
Monochlorobiphenyl
1
2
Olchlorobiphenyl
5
2.3
Trlchlorobiphenyl
29
2.4.5
Tetrachlorobiphenyl
50
2.2',4.6
Pentachlorob i phenyl
87
2.2',3.4,5'
Hexachlorobiphenyl
154
2,2',4,4,\5,6'
Heptachlorobiphenyl
188
2.2',3,4',5,6,6'
Octach1c rob i p iieny1
201
2,2',3*3',4,5',6,6'
Nonach1orob i pheny1
207
2.2',3,3",4,4',5,6,6'
Oecachlorobiphenyl
209
2,2',3.3',4.4',5,5',6,6'
Retention Tirre Calibration Standards
Tetrachlorcbiohenyl
77
3.." ,4,4"
Pentachlorobicnenyl
104
2,2',4.6,6'
Nonachlorobipnenyl
208
2.2',3,3',4,5.5',6,7'
Internal Standards
dl0-anthrac2ne
dj2-chrysene
d6-3,3',4,4',-tetrachlorobiphenyl
4 Numbered according to system of 3allschmiter and Zell (Ballschmiter, X.,
and M. Zell, Fresenius Z. Anal. C^.em., 302, 20, 1980.
-4
-------
typical for analyses of trace contami-
nants in environmentally-related sam-
ples in general. The principal
considerations are the potential for
sample contamination and analyte
losses. For example, significant
losses of PCB*» (or even PCOOs and
PCOFs) can re_jlt from allowing an
extract to evaporate to dryness.
1.3 QUALITY ASSURANCE ABM QUALITY
CONTROL PROCEDURES
The primary quality assurance and
quality control procedures that should
be employed for PCB determinations in
MWC stack samples are essentially the
same as described fo" PCOOs and PCOFs
in Section 3.1.3 of this document.
Procedures for calibration of field
equipment, field lab blank samples,
laboratory blank samples, resin pre-
screening, and quality control check
samples are Identical. RRFs should be
monitored in a similar manner.
1.4 DATA QUALITY OBJECTIVES
1.4.1 Limit of Detection
As in the case of PCDO and PCDF mea-
surements (Section 3.1.4.1 of this
document) objectives for limits of
detection should be based on regulatory
requirements. Table A-2 presents
target limits of detection for PCB
homolcgs that are readily achievable
using the methods described above with
a high performance quadrupole GC/MS
system. Significantly lower detection
limits have been achieved for some MWC
tests.
1.4.2 Accuracy
The accuracy objective for blind
quality control check samples is 60 to
12035 of the spike level. The accuracy
objective for recovery of surrogate
compounds should be SO to 15CS.
1.4.3 Precision
Precision should be evaluated by
pooling all the accuracy values, i.e.,
recoveries for each surrogate compound.
For example, the recoveries for l3Cl2-
3,3',4,4'-tetrachlorobiphenyl should be
pooled to determine a single precision
value. The target for precision should
be s 50X RSO.
1.4.4 Completeness
The completeness objective for PCBs
in MWC flue $ases should be at least
90* valid deterainations.
2.0 OTHER SEMIVOLATILE ORGAN ICS
These methods are appropriate for
determining chlorinated benzenes (CBs),
chlorinated phenols (CPs), and poly-
cyclic aromatic hydrocarbons (PAHs) in
MWC stack gases.
2.1 METHOD DESCRIPTION
Stack emissions are saspled for CB,
CP, and/or PAH measurement using an MM5
train identical to that described in
Section 3.1 of this document for deter-
mining PCDOs and PCDFs. The sample
components are combined and cleaned to
remove excessive levels of potential
interferences (if required), and the
extracts are analyzed using GC/MS.
2.1.1 Sample Collection
The W5 train configuration, prepara-
tion, operatic--, and sanole recovery
described for determining PCDDs and
PCOFs in MWC stack gases (Section 3.1
of this document) can be directly
applied to measuring CSs, CPs, and/ar
PAHs.
a-5
-------
TABLE A-2. EXAMPLE TARGET LIMITS OF DETECTION
Congener
Target detection limit
(ng/Nm3)
Monoch1orobipheny1
5
Dichlorobipnenyl
5
Trichlcrobiphenyl
5
Tetrachlorobiphenyl
S
Pentachlorobiphenyl
10
Hexach1oroo1pheny1
10
Heptachloroblphenyl
10
Octach1crab ipheny1
15
Nonach1crob1pheny1
IS
Decachlorobipnenyl
15
-------
2.1.2 Sample Analysis
A .primary reference for determining
CBs, CPs, and/or PAHs in MWC stack
samples is Method 8270, SW-846.
2.1.2.1 Sample Preparation and Ex-
traction—
Preparation and extraction of MM5
Sample components described for PCDDs
and PCDFs (Section 3.1.1.2 of this
document) can be applied nearly
directly to samples for C8s, CPs, and
PAHs. The only exceptions are the
selection of extraction solvent, pH
adjustment of aqueous samples prior to
extraction, and surrogate recovery
spiking. Extraction of the MM5 resin
and particulate catch with toluene or
benzene can cause significant losses of
monochloro and dichlorobenzenes during
subsequent extract concentration.
Hence, extraction with dichloromethane
1s preferred if quantitation of these
compounds is required. The aqueous
samples must be adjusted to pH < 2 with
505£ aqueous sulfuric acid prior to
extraction to ensure quantitative
extraction of the more acidic CPs.
Prior to extraction, the sample com-
ponents should be spiked with recovery
surrogate compounds to facilitate
assessment of analyte recoveries.
GC/MS quantitation, as described in
Method 8270. SW-846, employs internal
standard compounds spiked just prior to
GC/MS analysis. The surrogate solution
should contain i3C or deuterium-labeled
representatives of each analyte class,
as available. The spike level should
be in the range of 5 to 10 times the
anticipated detection limit. Several
deuterium-labeled PAH compounds are
available for use as recovery surro-
gates or internal standards for GC/MS
quantitation. These include the
following.
d#-Naphthalene
d10-Anthracene
d ^-Phenanthrene
dt0-Acenaphthene
du-Chrysene
dl2-Perylene
Example labeled compounds which may
be used as recovery surrogates for CBs
and CPs ^re shown below.
1 6-1,2,4,5-Tetrach1orobenzene
* 3C,-2,4,5-Tri ch1oropheno1
6-Pentachloropheno1
Since the sample preparation and
extraction of MM5 samples for CBs, CPs,
and PAHs is the same a? for PCBs,
PCCQs, and PCOFs, the combined extract
fro* a single train (resin and particu-
lates must be extracted with toluene or
benzene) can be split to accommodate
multiple analyses. However, splitting
the extract increases the potential
detection limits for both sets of ana-
lytes. Also, using a single sample for
all analytes compromises quantitation
of monochloro and dichlorobenzenes
since PCDD and PCDF determination
requires extraction of the resin and
particulate catch with toluene or
benzene. Unequal splitting can be
employed, depending on target detection
limits.
2.1.2.2 Extract Cleanup—
Extract cleanup is typically required
to achieve low detection limits for MWC
stack samples. The extracts from each
MM5 train should be combined, dried by
passage through a short column of
anhydrous sodium sulfate, and concen-
trated to 5 ml using Kuderna-Danish
evaporators. A 15-mL aliquot of hexane
should be added and the extract concen-
trated to 2 ml uising Kuderna-Oanish
evaporators and nitjrogen blowdown. If
cleanup is net required, the extract
should be further concentrated to
500 vL for GC/MS analysis.
A detailed protocol for cleanup of
extracts for CBs using florisil is
-------
presented irr: Method 3620, SW-846. A
silica gel procedure for PAHs and
derivatized CPs 1s described in
Method 3630, SW-846. Either method can
be used for CBs, CPs (underlvatized),
and PAHs. The cleaned extract should
be concentrated to 500 uL using
Kuderna-0an1sh evaporators and nitrogen
blowdown.
2.1.2.3 GC/MS Analysis-
Sample extracts must be analyzed by
GC/MS using a fused silica capillary
column. A detailed protocol 1s pre-
sented In Method 8270, SW-846. The
recommended capillary column Is a 30-m
D8-5. Method 8270, SW-846 specifies
full scan data acquisition. Although
full scan data facilitates compound
Identification - confirmation by review
of complete mass spectra, the levels of
detection achievable are frequently
insufficient for MWC stack samples.
Hence, SIM data acquisition is recom-
mended. The same characteristic ions
and procedures described in
Method 8270, SW-846 can be used
directly, but data' should be acquired
in the SIM mode.
CBs, CPs, and PAHs and the recovery
surrogate compounds are quantitated in
the extracts using response factors
normalized to internal standard com-
pounds spike into the cleaned extract
just prior to analysis. The relative
response factors (RRFs) are determined
by analyzing at least three-levels of
calibration solutions prior to sairaie
analysis. The calibration solutions
should contain.: the native analytes,
recovery surrogates, and internal stan-
dard compounds. The internal standards
should include 1-3 deuterium-labeled
PAH compounds other than those used as
recovery surrogates. The levels of the
native analytes should cover a 100-fold
concentration range and the lowest con-
centration should be 5 times the
pract'caDle detection limit; The cali-
bration solutions should also contain
the recovery surrogate compounds at
:oncentrat1ons corresponding to the
spike level in samples.
The RRFs for the native analytes and
recovery surrogates should agree with a
relative standard deviation (RSO) of
20% for all compounds. Once acceptable
calibration has been achieved, 1t 1s
verified each day of analysis by ana-
lyzing the mid-level standard at the
beginning and end of .the day. RRFs for
the dally standards must be within 20*
of the calibration RRFs, otherwise
recalibratlon is necessary.
The concentrations for each compound
are determined using their RRFs and are
reported for each '^dividual compound
as mass per sample volume, e.g., nano-
gram of 1,2,4,5-tetrachlorobenzene/
dscm. Detection limits should be de-
termined for -each compound using a
response of 2.5 times the background
noise signal at the appropriate reten-
tion time. The recoveries of the
surrogate compounds are also quanti-
tated using RRFs. The surrogate
recoveries should be within the range
of 50 to 150%. Recoveries outside this
range may indicate problems in trie
analysis.
2.2 CRITICAL FEATURES
2.2.1 Sample Collection
The critical features of MM5 sample
collection for C3s, CPs, and PAHs are
essentially the same as those for
PCDOs, and PCDFs, as described in Sec-
tion 3.1.2.1 of this document.
2.2.2 Sample Analysis
The critical features of analysis of
MWC stack samples for CBs, CPs, and
PAHs are somewhat typical for analyses
of trace contaminants in environmen-
tally-related samples in general. The
principal considerations are the po-
tential for saoole contamination and
A-8
-------
analyte losses. For examole, signifi-
cant losses can result from Slewing an
extract to evaporate *.o dryness. Also,
PAH compounds *re susceptibly to photo-
degradation and should be protected
from exposure to lignt.
2.3 QUALITY ASSURANCE AND QUALITY
CONTROL PROCEDURES
The primary quality assurance and
quality control procedures that should
be employed for C3, CP, and PAH deter-
minations in MWC stack samples are
essentially the saroe as described for
PCDDs ' and PCDFs in Section 3.1.3 of
this document. Procedures for calibra-
tion of field equipment, field lab
blank staples, laboratory blank sam-
ples, resin prescreening, and quality
control check samples are identical.
RRFs should be monitored in a similar
manner.
2.4 DATA QUALITY OBJECTIVES
2.4.1 Linit of Detection
As in the case cf PCCQ and PCDF mea-
surements (Section 3.1.4.1 of this
document) objectives for limits of
detection should be based on regulatory
requirements. Table A-3 presents
target limits of detection for specific
compounds that ara readily achievable
using the methods described above with
a high performance quadrupole GC/MS
system. Significantly lower detection
limits have been achieved for some MWC
tests.
2.4.2 Accuracy
The accuracy objective for blind
quality control check samples is 60 to
120* of the spike level. The accuracy
objective for recovery of surrogate
compounds should be 50 to 150%.
2.4.3 Precision
Precision should be evaluated by
pooling all the accuracy values, i.e.,
recoveries for each surrogate compound.
For example, the recoveries for uCs-
2,4,5-trichlorophenol should be pooled
to determine a single precision value.
The target for precision should be
S SOX RSD.
2.4.4 Completeness
The completeness objective for PCBs
1n MWC flue gases should be at least
90* valid determinations.
3.0 VOLATILE ORGANICS
These methods are generally appropri-
ate for measuring organics with boiling
points in the range of 30"to 100'C in
MWC stack emissions. Compounds with
boiling points as as high as 145'C way
also be determined. Some compounds
with boiling points lower than 30°C can
also be determined, although signifi-
cant breakthrough may occur causing
underestimation of concentrations.
3.1 METHOO DESCRIPTION
The method involves passing the stack
gases through adsorbent cartridges.
The volatile organics retained by the
cartridges are transferred to the lab-
oratory where the analytes a«
thermally desorbed from the cartridges
and determined by GC/MS.
3.1.1 Sanple Collection
Samples are collected using the vola-
tile organics sampling train (VOST). A
detailed protocol for VOST sailing can
be found in Method 0030 of SW-346.
A-9
-------
TABLE A-3. EXAMPLE TARGET LIMITS OF QETECTIOM
Target detection limit
Compound
(ng/MmJ)
Chlcrcbenzenes
1,3-D1chlorobenzene
10
1,4-Dichlorobenzene
10
1,2-0'fchloroberizer.e
10
1,3,5-Trtchlorobenzene
10
1,2,4-Trlchlorobenzene
10
1,2,3-Tr1ch1orobenzene
10
1,2,3,4-Tetrich1orobenzene
10
1,2,3,5-Tetrachlorobenzena and
10
1,2,4,5-Tetrach1orobenzene
Pentachlorobenzene
10
Hexich]orobenzene
10
Chloroohenols
2-Chlorophenol
10
3-Chlorophenol and
10
4-Chlorcphenol
2,3-01chlorophenol
10
2,6-Dichlorophenol
10
3,5-OlcMorophenol
10
3,4-Olchlorophenol
10
2,4-DicMorophenol and
10
2,5-01ch1oropheno1
2,3,5-Trichloropienol
10
2,4,6-Trfch'orophtno?
10
2,4,5-TricMorophenol
10
2,3,4-Trichloropherol
10
2,3,6-Trlchlorophenol
10
2,3,5,6-Tetrachlorophenol
15
2,3,4,5-Tetrachlorophenol
15
Pent achl crropheno 1
50
Polycyclic Aromatic Hydrocarbons
Nathpnatene
20
Acenaphthylene
5
jScenaphthene
15
Fluorene
15
Phenanthrene
20
Anthracene
20
Huoranthene
20
Gyrene
20
Benz|a!anthracene
20 .
Chrysere
20
Senzo(b1fluoranthene
30
3enzo[kIf11 uoranthene
30
Benzofajpyrene
30
IndenoT1,2,3-cd1pyrene
30
Oibenzja,njanthracene
30
3enzo[ sTh .i.! pery1ene
30
A-10
-------
3.1.1.1 Train Configuration—
The primary components of the VQST
system are the probe, condenser,
condensate trap, a second condenser,
and a backup resin trap. A schematic
of the VOST system 1s shown in
Figure A-l. Figure A-2 shows a drawing
of the train. The first cartridge 1s
packed with approximately 1.6 g of
Tenax-GC resin. The second cartridge
is packed with Tenax-GC and petroleum-
based charcoal (1 g of each, approxi-
mately 3:1 by volume) with the charcoal
on the outlet end. The first trap
retains most of the higher boiling ana-
lytes. lower boiling analytes and the
portion of the higher boiling analytes
that break through the first cartridge
are retained on the second trap. Ana-
lytes that collect in the condensate
trap are purged into the second con-
denser and trap units.
Design details of the traps may
differ as appropriate for the specific
fittings used for tne train and for the
soecific design of the thermal desorp-
tion unit used to analyze the concents
of the traps. An example trap design
is shown in Figure A-3.
3.1.1.2 Adsorbent Cartridge Condi-
tioning
Prior to use, both new and previously
used adsorbent cartridges must be con-
ditioned and checked for background
contamination. This involves heating
the traps to 190"C and passing organic-
free nitrogen through the traps (in the
reverse of the sampling direction) at
> AO ml/min for L2-28 h. The condi-
tioned traps are then analyzed for
background contamination by thermal
desorption and GC/FID detection. This
procedure is described in Metnod 0030,
SW-846. A convenient alternative is to
continuously measure the trap back-
ground by analyzing the gas exiting the
traps during conditioning. The exit
gases from a trap can be sampled using
a gas sampling valve and introduced
onto a "dutmy* column (3-m x 1/16-in 00
SS tubing). The FIO response should be
equivalent to < 5 ppb propane.
3.1.1.3 Train Operation—
The sampling flow rate should be
approximately 1 L/min for 20 min for a
total of 20 L. The traps should be
removed, a new pair installed, and sam-
pling continued at the same rate. A
total of six pairs of traps should be
used. Alternatively, the sampling rate
can be reduced to as low as 0.25 L/min
to extend the sampling period. The
total sampling tine should not be less
than 1 h and at least 3 pairs of traps
should be used. The total volume sam-
pled by a pair of traps should not ex-
ceed 20 L. All traps should be tightly
sealed and stored at ice temperature
until analysis.
3.1.2 Sample Analysis
A detailed protocol for analysis of
VOST traps can be found in Method 5040,
SW-846. The volatile organic contents
of the traps are thermally desorbed and
analyzed by GC/NS. Method 3240,
SW-846, describes the GC/MS system and
procedures.
The design of the thermal desorption
units may differ for various adsorbent
trap designs. However, the desorption
unit must be leak-free and direct the
gas flow through the cartridge in the
reverse of the samole collection flew.
Prior to desorption and analysis,
each trap is soiked with an internal
standard solution (in methanol). The
internal standard compounds should be
stable isotope-labeled analogs of the
compounds of interest or similar com-
pounds. Typical internal standards
include d6-benzere, dl0-ethylbenzene,
and d„-dichioroethane.
A-ll
-------
\c
BE
Glass or Teflon Probe
Sample Flush
&
Purge
I
Condenser
Charcoal Filter
Teflon Tubing
/
Tenax Trap
Micro Valve
(T2)
Silica Gel
(DT)
Vacuum
Gauge ' 1
Vacuum
Gauge '2
9
/ Shuf Off Valve
Rotameter
Dry Cm Meter
rgon Tubing f0'" *1/ Uok Fr« PumP
57 Tubing
Tencx-CKarcoal
Trcp
¦Submersible Pump
•Viton O-Ringed Nickel Plated Fittings
Figure A-l. Schematic of the volatile organic sampling (train (VOST)
(from Method 0030, SW-84SJ. ¦'
A-12
-------
Teflon Plug Valve with Socket Joint
Silica Gel Holder
Condensers
t!
Vacuum Gages
Tubing
Tenax Trap
m
Flow Meter
Tenax/Charcoai
Trap
Submersible
Pump
Ice Water Bath
Case
Figure A-Z. 7G$T system setup (from Method 0030, SW-846).
A-13
-------
CT7T
2 Layers-
S. S. Screen
C±E
1/2 UltTo-
Torr Fitting
"C" Clip
Tenax TA
1/2 Ultra-
Torr Fitting
Figure A-3. VOST trap assembly (EPA 600/9-84-019, November 1984,
Proceedings: National Symposiun on Recent Advances
in Pollutant Monitoring of Amoient Air and Stationary
Sources, pp. 171-179, "Volatile Organic Sampling Jrain
(VOST) Development at MRI by Fred Bergman).
A-14
-------
Following desorption of the sample
from the VCST trap onto the analytical
trap, the contents of the analytical
trap are thermally desorbed onto the
GC/MS systen. The GC/MS data are
acquired us'og the scanning mode. SIM
data acquisition can be used to de-
crease limits of detection. Analytes
are quantitated using relative response
factors {relative to the internal stan-
dards) determined by analyzing standard
solutions of the analytes (in methanol)
spiked onto a blank VOST trap. Detec-
tion limits should be determined for
each compoisxl not detected using a re-
sponse of 2.5 times the background
noise signal at the appropriate reten-
tion time.
The relative response factors (RRFs)
should be determined from at least
three levels of standard solutions over
a 100-fold rarrge. The lowest standard
level should be approximately 5 times
the detection limit. The RRFs should
agree within a relative standard devi-
ation of 2C*. Once an acceptable cali-
bration is achieved, it is verified
each day cf sample analysis by analyz-
ing the ale-level standard at the
beginning 2TaJ end of the day. RRFs for
the daily staroards must be within 20»
of the c*>'bration RRFs; otherwise,
recalibrat;cn is necessary. Alterna-
tively, cntroi charts can be used to
monitor RRFs.
VOST tr;a samples and field blanks
should be irilyzed within 1-1 d of sam-
pling. inalysis results from traps
held longer tnan 14 d may be considered
suspect for scne compounds.
3.2 CRITICAL FEATURES
Avoiding contamination of samoles and
standards is crucial to successful VQST
samcnng arid volatiles analysis.
Adsorbent cartridges and analytical
standards srsuld be prepared in a clean
laborat:**y fat is practically solvent-
free. -cerate cleaning of the traps
prior to use and avoiding contamination
during field heixtling, shipment to the
analysis laboratory, and storage prior
to analysis are especially critical.
Each trap should be screened to ensure
that, the background is acceptable im-
mediately following precleaning. Clean
traps should be stored in sealed metal
cans with an activated charcoal packet
in the can. Immediately following sam-
pling, the used traos should also be
stored in sealed cans with activated
charcoal. However, clean and used
traps usust not be stored in the same
can,
3.3 QUALITY ASSURANCE AND QUALITY
CONTROL PROCEDURES
3.3.1 Field Blank Savples
At least one pair of blank traps
should be analyzed for each test run.
The blank traps should be taken to the
sampling site and opened, i.e., end
caps removed, for a time period typi-
cally required for changing traps in
the VCST train. The caps should be
replaced and the blank traps shipped to
the analysis laboratory for analysis.
Also, a pair of trip blank traps should
be analyzed for each MWC facility test.
Trip blanks are handled similar to
other field blan* samples, althouc'' the
end caps are not rescved.
3.3.2 Laboratory Blank Samples
A laboratory blank sample, i.e.,
blank trap, must be analyzed at the be-
ginnlrg of eac* cay standards or sam-
ples are analyzed to check the back-
ground attributable to the VOST trap
desorption and surge-trap-GC/MS sys-
tems. The laboratory blank should be
spiked with the internal standard
solution.
A-IS
-------
3.3.3 Relative Response Factor Moni-
toring
Relative response factors must be
monitored for each analyte. The RRFs
for the daily standards should fa71
within 20* of the mean from the cali-
bration RRFs. Alternatively, RRF
control charts can be maintained. A
mid-2evel standard should also be
ana'yzed in triplicate to determine
precision. The relative response
factors should agree within ±20X.
3.3.4 Quality Control Check Standards
A blind quality control check stan-
dard should be analysed with samples
fro® each facility test. The check
standard should be spiked onto a VOST
trap. The analyte concentrations
should be within the lower half of the
calibration range. The accuracy
resets for blind check standards
sheila be within 60-12CX of the spiked
level.
I- available, audit gas samples may
also ae analyzed. iuiit gas samples
are provided by the EPA's Environmental
Mcnitcring Support Laboratory,
Research Triangle Park, North Carolina
(it tne request of the permitting
autr-crity) for VOST performance audits
re'ated to hazardous «aste incinerator
burns. The aud"
-------
Each lot of glass fiber filters used
1n the sampling train should be checked
prior to field testing to confirm that
background levels are low. Polytetra-
fluoroethylene (PTFE) coated glass
fiber filters mty be used.
When sampling at source locations
containing high HC1 levels (especially
at facilities without acid gas removal
or at points upstream of the acid gas
removal system), use of a stainless
steel buttonhook nozzle may lead to
unacceptably high nickel contamination.
Use of quartz nozzle 1s recommended for
all nickel determinations.
The detection limit for arsenic by
cold vapor AAS 1s on the order of
1-2 ug/L 1n the solution as analyzed.
Assuming a standard sample of l.l m*,
and a final digested sample volume of
250 mL each for the probe, filter, and
Implnger contents of the train (as in
Method 108), this corresponds to an in-
stack detection limit of C.iS to
0.3 ug/m1. If lower detection limits
are required to meet the test objec-
tives, the most promising alternative
is to collect a larger stack gas sam-
ple. For arsenic, neither furnace AAS
nor ICP offer a substantially lower
detection limit than that obtainable "by
hydride generation.
The detection limit for nickel deter-
mination by flame AAS is about 40 ug/l.
Assuming a standard M5 stack gas sampie
volume of 1.1 and a final digested
sample volume of 250 mL (as specified
in M12), this corresponds to an in-
stack detection limit of 9 ug/mJ. If
lower detection limits are required to
meet the test objectives, notions
include:
• Collection of a larger stack gas
sample,
• Use of a smaller volume for the
digested sample dilution, or
Substitution of ICP for the AAS
analysis.
The sensitivity enhancement possible by
using these options are, respectively,
a factor of 2-3, a factor of 2-5, and a
factor of 3.
4.1.3 Quality Assurance and Quality
Control Procedures
4.1.3.1 Calibration—
The sampling equipment must be cali-
brated as specified In M5. Laooratory
balances must be calibrated according
to a regular schedule. The AAS systen
is calibrated for wavelength accuracy
according to the manufacturer's speci-
fications.
Quantification should be accomplished
by reference to an exterral standard
carve based on at least five standards
covering the range from 0 to 25 wg
As/L. Samples must be diluted as
necessary to fall within tfte calibra-
tion curve range. The calibration
ccrve should not be force-fit througn
rero.
4.1.3.2 Blanks—
* filter blank should be created
using two filters frcm each Tot used in-
sampling train. The filter blank
"S digested in the same *ay as the sam-
ples. If the impinger solutions are
included in the analysis, trie blar*
swjuld include a volume of 0.1 H nitric
acid equal to that charged to the is»-
pingers. Blanks of the iicpinger water
and the solutions') used for rinsing
tne sampling "rain should be analyzed.
Replicates. Samples, standards, and
blanks should be jinalyred at least in
ctalicate. It is common practice to
Sase compliance determinations on the
nean of the arsenic a-vj/cr nickel emis-
sions as measured in three separate
stack sampling runs.
-------
4.1.3.3 Spikes—
As with other metal determinations.
It is not standard practice to analyze
true "spikes", in which a known quan-
tity of the metal is added to an ali-
quot of the sample prior to digestion.
However, 1t is considered good practice
to analyze at least one sample from
each source by the method of standard
additions.
4.1.3.4 Audit Samples-
It is recommended that quality con-
trol procedures include the analysis of
at least two unknown arsenic audit saw-
pies from the EPA Quality Assurance
Office. These should be analyzed con-
currently with the field samples. As a
preliminary check on laboratory perfor-
mance, audit samples of known concen-
tration are available through the EPA
Environmental Systems Laboratory at
Research Triangle Park.
4.1.3.5 Calculations—
TI.e concentration of arsenic and/or
nickel in the stack gas, Cs, is cal-
culated as follows:
Ca (h/L) x D x Va (L)
Cs (u/m3 * vs (ma) *
where
C_ » concentration in tne solution as
analyzed, from the calibration
curve,
0,« dilution factor(s), if any,
tf * volume of digested samDle, and
V* » volume of stack gas sample.
4.1.4 Oata Quality Objectives
no numerical data quality objectives
jjiave been established for the aeter-
illnation of arsenic or nickel in NWC
emissions.
It is expected that the precision of
replicate analyses of the same digested
sample should not exceed 5X CV. The
accuracy of the determination, as esti-
mated by the method of standard addi-
tions, should be similar to that de-
scribed for lead (Section 3.3 of this
document).
4.2 BERYLLIUM
4.2.1 Method Description
The basic method for beryllium deter-
mination is EPA Method 104. This uses
the same sampling train configuration
as M5. However, the method specifies
the use of a cellulose acetate membrane
filter (e.g., Millipore AA), backed-up
be a glass fiber filter if necessary to
prevent tearing, for collection of
beryllium. The sample 1s collected
isokinetically. A minimum sampling
time of 2 h is recommended.
Following sample collection and
recovery, the filter 1s digested with
concentrated nitric add, followed by a
sulfuric add/perchloric acid diges-
tion. The probe wash is combined with
the contents of the first three im-
pingers (water), evaporated to dryness,
and cigested in the same way as the
filters. After digestion, the residues
are combined and diluted with 252 HC1.
The solution is analyzed toy flame AAS.
4.2.2 Critical Features
The cellulose acetate membrane filter
specified in M1C4 may not withstand the
temperature and corrosion conditions in
MWC effluent. A glass fiber filter may
be used provided that the analysis of a
representative number of filters from
each lot shows acceptable background
levels of Be. Mote that the digestion
procedure specified for Be
(HNOj/HjSO^/HClOi,) is not the same as
that in the other metals methods
(HNQj/HjOj). it is acceptable to take <
a portion of the total sample from.
A-13
-------
e.g.* the lead digestion, and subject
It to a subsequent Be digestion accord-
ing to M104.
The critical features discussed under
M5 (Section 3.2.2 of tMs document) for
particulate apply to beryllium saspling
as well.
The detection limit for Be using
flame AAS 1s on the order of 5 tg/L in
the solution as analyzed. Assuming the
recomnended 2-h sampling time (M5 sam-
ple volume of 2.2 uJ and a digested
sample volume of 10 mL, as specified in
MI04, this corresponds to an in-stack
detection Unit of 0.025 ug/ra». How-
ever, these limits apoly to clean
(water) samples. The Be analysis Is
subject to interference from aluminum,
magnesium, and silicon, all of which
nay be present In MWC stack particulate
matter. The actual detection limits
observed for MWC stack gas samples may
therefore be considerably higher than
0.025 wg/m».
If even lower detection Units were
required to meet the test objectives,
either furwce AAS or ICP could be
substituted as the analytical finish.
The detection limits oted in SW-846
are 0.2 ug/t for furnace AAS and
0.5 ug/L for ICP.
4.2.3 Quality Assurance and Quality
Control Procedures
Method 104 scecifies tne use of a
single calibration standard of i ug/L
it concentration. However, fcr mea-
surement of Be in Mv«C effluents, the
pie of a minimum 5-:oint calibration
curve is advised.
Method 104 provides guiea*ce on
analysis of blanks, "epHcates, or
spikes. For MWC effluent characteriza-
tion. it is recommences that flanks,
spikes (analysis be rretrcd cf standard
additions), ard repHeatss be '-.eluded
In the QAPP. Tne procedures srecified
for other metals (e.g. lead, chromium)
can be used as guidelines for frequency
and types of QC samples.
M104 specifies that the mean of three
tests runs be used for the determina-
tion of corcoiiai in
MWC effluents is the draft method
developed by EPA's Environmental .^ni-
torlng Support Laboratory at Research
Triangle Park, North Carolina. In this
method, particulate chromium is col-
lected 1n the probe and filter cf a
standard MS train, sampling isoklreti-
cally. Glass fiber filters cr ®?FE-
coated glass fiber filters may be
used. After sample collection and
recovery according to standard MS
procedures, the probe rinse and filter
samples are combined, evaporated to
dryness, and digested with a solution
of 2% NaOH and 3% Na,C0, in deionized
distilled water.
A-19
-------
The digested solution Is brought to
volume and tT>e Cr(VI) determined
colorlmetrically as Its dlphenyl-
carbizone derivative. This analytical
procedure 1s described 1n SW-846 Method
7196.
4.3.2 Critical Features
All of the critical features dis-
cussed under M5 (Section 3.2.2 of this
document) sampling apply to the Cr(VI)
determination.
For determination of chromlum(VI), 1t
1s especially critical that replicate
filters from each lot of filter medium
be analyzed prior to testing to confirm
that background levels of Cr(VI) are
acceptably low.
In sampling MWC effluents containing
high levels of HC1, especially when
there Is no acid gas removal or the
seunpling point 1s upstream of the acid
gas removal system, stainless steel
probe nozzles should not be used.
Quartz is a recommended substitute.
The'detection limit of the method is
stated to be 50 ug/100 ml, with 100 mL
as the standard volume for the digested
sample. For a standard M5 sample of
1.1 mJ, this corresponds to an in-stack
detection limit of 45 jig/m3* However,
some NESCAUM Work Group members have
observed that interferences have caused
precipitate formation and/or cloudiness
of the solution. These difficulties
may increase the /practical limit of
quantitation.
4.3.3 Quality Assurance and Quality
Control Procedures
4.3.3.1 Calibratlon-
The sampling equipment is to be cali-
brated as described in M5. Laboratory
balances should be calibrated according
to a regular schedule. The spectro-
photometer used For the colorimetric
wlyslV: is calibrated according to the
•anufacturer's recommendations. The
EPA draft method specifies that a five-
point calibration curve covering the
range from 0 to 2,500 ug/L. This
should be adequate for MWC emission
Measurement.
4.3.3.2 Blanks—
The draft aethod does not specify
procedures for blanks other than a
zero, I.e., 0.0 ug/L, calibration
standard. For MWC testing, It 1s
recommended that a blank consisting of
2 filters firm each lot used in the
sampling train' be analyzed along With
tfie samples.
4.3.3.3 Replicates—
the opportunity for replicate analy-
sis 1s limited because the method
specifies that 50 ml from the 100 ml
total d 1 gested samp le vol une be used
for each analysis. For characterizing
the Cr(VI) concentration in MWC emis-
sions, the mean of three replicate
sampling and analysis runs should be
used.
4.3.3.4 Spikes—
As for other EPA metals analysis pro-
cedures, the method specifies that at
•east one samole from each source be
analyzed by the method of standard
additions. The procedure 1s the same
as that specified in M12 for lead.
4.3.4 Data Quality Objectives
Data Cfquality objectives for 1 so-
Tcfneticity, leak check. etc., are as
specified in K5;
No numerical da1j[a "quality objective
for precision hai been established.
However, EPA tests at a ferrochrome
shelter, a chemical plant, and 'a
refractory brick plant gave Relative
standard deviations of 4.4, 3.3, and
13.3%.
-------
The numerical data quality objective
for accuracy 1s stated 1n terras of the
method, of standard additions. The
value of the Cr(VI) concentration cal-
culated by that method must be within
Si of the value obtained by routine
spectrophotometry analysis. If not,
all samples must be analyzed by the
method of standard additions.
4.4 MULTIPLE METALS SCREEN
4.4.1 Method Description
It Is obvious that the sampling and
sample preparation methods are the same
or similar for many of the metals of
potential interest with regard to MWC
research and development. Recognizing
this, the Environmental Monitoring
Systems Laboratory at Research Triangle
Park, North Cirolina, has been devel-
oping a train for multiple metals sam-
pling. The recommended configuration
is an M5 train equipped with a nozzle,
glass fiber filter, and five impingers
as follows: (1) empty, (2 and 3) 5X
HNOj and 10X H2O2 (will be reduced to
0.1 N HNO3 if research shows it to be
adequate), (4) 1.5% KMnO„ and 10*
H2SQ„, and (5) silica gel. After
sample collection and recovery, a
nitric acid/hydrogen peroxide digestion
as in M12 or SW-846 Method 3050 will
convert most metals to soluble form.
(Exceptions are: mercury, which would
be lost t.irough volatilization and is
determined by the cold vapor AAS tech-
nique; beryllium, which requires a
nitric/sulfuric/perchloric acid method
for digestion; and chromium (VI), which
requires a basic extraction method.)
For MWC research and develcpnent, the
extract obtained from the nitric acid/
hydrogen peroxide digestion can be
screened by I CP to accomplish the
determination of 25 different metals.
These are:
Aluminum
Antimony
Arsenic
Barium
Beryllium
Boron
Cadmium
Calcium
Chromium
Cobalt
Copper
Iron
Lead
Manganese
Molybdenum
Nickel
Potassium
Selenium
Silicon
Silver
Sodium
Thallium
Vanadium
Zinc
The analytical procedure 1s presented
in SW-846 Met&cd 6010.
4.4.2 Critical Features
The features of the MS train dis-
cussed in Section 3.2 of this document
apply to the sampling of metals for ICP
screening, note that if silicon is to
be determined, it is Important that no
silicone grease be used in sealing the
train.
If chromiua. iron and/or nickel are
to be determined, the use of stainless
steel nozzles is not recommended.
Quartz should be substituted.
The detection limits for ICP analysis
as given in Method SW-S46 6010 range
from 0.3 ug/L (Be) to 75 Pg/L (Si).
Assuming a standard M5 sample volume of
1.1 mJ and a final digested samole
volume of 255 aL. these correspond to
in-stack detection limits of 0.07 wg/mi
(3e) to 17 s»g/sr* (Si). For MWC that
incorporate Use scrubbing for acid gas
removal, levels of calcium in the stack
particulate matter may be high. This
could cause interference in the deter-
mination of other metals. It could
also require serial dilution of the
digested sample prior to calcium deter-
mination in order to get the concen-
tration of calcium within the linear
range of the IC?.
A-21
-------
Note that while SW-846 lists arsenic
and beryllium as elements to which the
ICP screen applies, alternatives to the
screening method described in this sec-
tion are preferred for these elements.
For arsenic, the AAS method is pre-
ferred because of the lower detection
limit. For beryllium, an alternative
digestion procedure is recommended.
4.4.3 Quality Assurance and Quality
Control Procedures
It is recommended that calibration,
blanks, replicates, spikes, and cal-
culations for the ICP screen follow the
guidelines given in Section 3.3 of this
document for lead.
4.4.4 Data Quality Objectives
There are no established numerical
data quality objectives for ICP screen-
ing of MWC emissions. These should be
specified in the QA?P for each research
and development program.
5.0 OTHER INORGANICS
5.1 FLUORIDE
5.1.1 Method Description
The Work Group recommends EPA
Method 138 as the procedure for mea-
surement of fluoride in MWC emissions.
The sa-T.oling train in this nethod is
essentially the sase as in Method 5,
except that the filter may be inserted
at a different location with respect to
the impingers and that a glass fiber
filter is not used. Total fluoride
collected in the train is determined by
analysis using specific ion electrode.
(Ion chrcnatography is an alternative
method cf determination.)
5.1.2 Critical Features
This method determines total
fluoride, not just HF. Therefore, if
permit limitations are expressed as HF
concentrations, the appropriate regula-
tory authorities should be consulted to
determine whether a modified or alter-
native procedure should be used.
The critical features discussed under
M5 for particulate matter determina-
tions apply to fluoride concentration
eeasuretftent as well. For fluoride
determinations, distilled ' deionized
water 1s specified as the probe rinse
solution for seunple recovery. Prior to
sample recovery, it is important that
all M5 train surfaces potentially 1n
contact with the sample be wiped free
of grease, as residual grease could
result in loss of fluoride.
The filter material selected should
be paper or an organic membrane. Glass
fiber filters will irreversibly react
with HF and give erroneously low
recoveries. For this reason, the Work
Group does not recommend that fluoride
determinations be performed on samples
collected in the HC1 train described
earlier in this report. Three filters
from each lot used for sampling should
be analyzed prior to the test to con-
firm that the fluoride blank level is
acceptably low.
The method notes that samples and
standards should be at the same tem-
perature (±2"C) to ensure accurate
ceterminations.
The detection limit for fluoride
cited in the method is 0.02 u/ml
(0.02 mg/l). However, concentrations
lower than 0.1 (ug/mL require "extra
care." Assuming then that 0.1 ug/mL
represents a practical lower level of
quantification, a 250 mL final sample
volume and a 1:1 dilution with buffer
as specified in Method 13B, the in-
stack detection limit for a standard
-------
1*1 mi M5 stack gas sample is about
11 ng/Nm» (0.011 mg/NmJ).
5*1.3 Quality Assurance and Quality
Control Procedures
5.1.3.1 Calibration--
The sampling equipment must be cali-
brated as specified in Method 5. Lab-
oratory balances and pH meters roust be
calibrated according to a regular
schedule.
The specific ion electrode calibra-
tion curve 1s prepared by analysis of
10"2, 10"J, lO"11, and 10"* H standard
solutions of sodium fluoride and
plotting electrode response (nv)
against the logarithm of fluoride
concentration. If the curve 1s
nonlinear 1n the 10"* to 10~s M range,
additional calibration standards at
intermediate concentrations are added
to the calibration curve.
5.1.3.2 Blanks!—
As noted above under critical
features, at least three filters from
each lot should be analyzed as blanks:
the acceptance level is scecifted as
15 ug {0.015 mg) F/cm* of filter area
(Method 13A).
5.1.3.3 Replicates—
The method providti no specific
guidance concerning field or laboratory
replicate requirements. It is recom-
mended that three sampling runs be
performed and the average calculated
for comparison with permit limitations,
unless the regulatory authority speci-
fies otherwise. The sample preparation
procedure in the method provides a suf-
ficient volume of final solution to
allow replicate analytical measurements
on each sample if required by the regu-
latory authority.
5.L.3.4 Spikes—
The method does not requ i re the
analysis of spikes. However, it 1s
recommended that at least one sample
from each source be analyzed by the
method of standard addition.
5.1.3.5 Audit Samples—
If available, audit samples should be
analyzed and the results reported along
with sample data.
5.1.3.6 Calculations—
The equations for calculating the
total fluoride content (mg/Nni*) in the
sample are provided in Methods 13A and
138.
5.1.4 Data Quality Objectives
Method 138 provides guidance that a
standard deviation of ±0.056 mg F/Nra'
can be expected at fluoride concentra-
tions of 0.1 to 1.4 mg/HmJ with five
degrees of freedom (i.e., six samples).
The collaborative study cited in the
method did not find any bias in the
analytical method. Therefore, the re-
sults of a method of standard additions
check would be expected to give concen-
trations within 5 to I0?f of those
determined by direct measurement.
5.2 SULFUR TRIOXIDE/SULFURIC ACID MU-T
5.2.1 Method Description
The Work Group did not reach a con-
sensus on a method for measurement of
sulfur trioxide/sulfurlc acid mist in
MWC emissions. The Work Group con-
cluded that there has been insufficient
method validation on this source cate-
gory to allow a method to be recom-
mended with confidence. Methods that
were discussed as potentially appli-
cable include EPA Method 8 (Method 5
approach with isopropaool in the first
impinger, followed by a filter,
A-23
-------
followed by hydroger peroxide 1»-
pingers), EPA Method 6 (a midget bub-
bler version of Method 8), controlled
condensation, and a buffered foraalde-
hyde impinger procedure provided to
Ogden Projects, Inc., by Exxon Research
and Engineering Co.
Until more information concerning the
potential applicability of one or
several, of these methods to MWC testing
has been developed, it is the judgment
of the Work Group that methods for
sulfur trioxide/sulfuric acid mist sam-
pling and analysis must be selected on
a case-oy-case basis during discussions
with the regulatory agency(ies)
requesting these data.
5.2.2 Critical Features
Because it is not aware of satisfac-
tory method validation data for mea-
surement of sulfur trioxide/sulfuric
acid mist in MWC emissions, the Work
Group recommends that especially care-
ful consideration be given to the
design of a QA/QC protocol for any such
testing cn a case-by-case basis.
6.0 FORKAIDEHYDE
6-1 METHOO (DESCRIPTION
There are no standard EPA methods
that are directly applicable to the
determination of formaldehyde *n MWC
effluents. The procedure recommended
by the NESCAUM Work Group for formalde-
hyde measurement is an MM5 train using
a special XAD-2 resin that has been
coatea «ith 2,4-dinitrophenyhydrazine
(DNPH) hydrochloride. During samoling,
the 'crtaldehyde is converted to its
2,4-Cnph derivative. The derivative is
recovered by solvent desorption and
analyre-J by high performance liquid
Chromatography (HPLC).
An alternative method 1s use of the
MS train, with ONPH reagent in the
impingers.
6.1.1 Sanple Collection
Procedures for preparing the ONPH-
codted XAD-2 are presented in an NTIS
report, EPA Nunber 600/9-84-015, Pro-
ceedings of the Ninth. Annual Research
Symposium on Incineration and Treatment of
Hazardous iv'ajte. The relevant publica-
tion Is 6eltis et aL, Stack Sampling end
Analysis of Formaldehyde.
The MM5 train is operated as de-
scribed in Section 3.1 of this docu-
ment. It is critical that the tempera-
ture of the gas stream entering the
sorbent cartridge not exceed 20°C to
prevent decomposition of the coated
resin. Other critical features of the
sampling are as described in Sec-
tion 3.1.2 of this document. It is
recommended that the formaldehyde train
be run as a dedicated sample for form-
aldehyde determination.
6.1.2 Sample Analysis
The sorbent cartridge is desorbed
using acetonitrile solvent in a Soxhlet
apparatus. The resulting extract is
concentrated to a volume of 10 mL using
a Kuderna-Danish evaporative concen-
trator.
The HPLC analysis procedure is
described in Kuwata etal., "Deter-
mination of Aliphatic and Aromatic
Aldehydes in Polluted Airs as Their
2,4-Dinitriphenylhydrazones by High
Performance Liquid Chromatography," J.
Chrcm. Set., 17, 264, May 1979. This
procedure can also be applied to form-
aldehyde collected using M5 with ONPH
Impinger reagent. In that case, the
aqueous reagent is extracted with
chloroform, which is back-washed with
2N HC1 and distilled water, con-
centrated to dryness, and recissolved
in acetonitrile.
A-24
-------
6.2 CRITICAL FEATURES
These procedures have been applied to
only a limited number of full-scale
combustion sources and not* to the Vlork
Group's knowledge, an MVC tests. Form-
aldehyde sampling and analysis should,
therefore^ be regarded as a research
and development activity. The ratio-
nale for reconsitervding this relatively
unvalidated procedure is that it is -one
of the few methods for formaldehyde
that is compound-specific, rather than
being a total aldenyde/ketone procedure
(the M8TH and chroma tropic acid proce-
dures are for total carbonyls}.
Many lots of XAD-2 resin {and many
chemical laboratories) have relatively
high background levels of formaldehyde
contamination, which can result In
positive interferences. Hence, the
analysis of blahk samples is critically
important. (It Is possible that alter-
native substrates to XAD-2, e.g.,
Tenax, may give lower background
levels. If so, they should be substi-
tuted for the XAO-2).
6.3 QUALITY ASSURANCE ANQ QUALITY
COHTROL PROCEDURES
6.3.1 Calibration
A minimum of five calibration solu-
tions consisting or formaldehyde
dinitriphenylhydrazone in acetom'trile
and spanning the concentration range
expected in the sample srtust b« ana-
lyzed. The curve must not be force-fit
through zero. A minimum of two cali-
bration standards must be analyzed each
day that samples are run.
6.3.2 Blanks
A method blank consisting of an
unused, coated KAD-2 cartridge (or vol-
ume of DNPH impinger reagent) must be
analyzed with each batch of samples. In
addition, each lot of sorbent (or
impinger reagent) must be checked for
an acceotable background level prior to
field sampling.
6.3.3 Replicate
A minimum of three stack gas sampling
runs should be performed. Replicate
determinations (reinjections of the
sample extract into the HPLC system)
should be performed on at least I03t of
the samples or at least once on eacn
batch of sasples.
6.3.4 Spikes
With each batch of samples, at least
one spiked saaple must be analysed*
The spike should consist of a blank
sorbent cartridge (or volume of
Impinger reagent) to which formaldehyde
has been added at a level corresponding
to the emission limitation specified in
the permit {assuaing a i.i ro* stack gas
sample).
6.4 DATA QUALITY OBJECTIVES
6.4.1 U«it of Detection
The detection limit for formaldehyde
in the cited procedure is about O.S ng
formaldehyde/20 «L of acetonitrile.
Assuming an M5 or MM5 train stack sam-
ple volume of 1.1 m* and a final sample
extract volume of 10 mL acetonitrile,
this corresponds to an in-stack detec-
tion limit of about 0.25 pg/m*. Baseo
on data In the Report to Congress
(EPA/530-SW-S7-C2lb), the expected
formaldehyde concentration in MWC stacic
gas is on the order of 100 u9/mJ or
higher; detection limits should, there-
fore, be adequate for MWC testing.
6.4.2 Accuracy
The recovery of formaldehyde from the
spiked sample should be in the range cf
60 to 1003C. ^esu'ts should not be cor-
rected for recovery, but the percent
recovery must Se stated in the report.
i-25
-------
6.4.3 Precision
The correlation coefficient for the
calibration curve should be 0.990 or
higher. Replicate HPLC analyses of the
sample extracts should agree within
±10%. There is Insufficient history of
application of this method to MWC test-
ing to allow a OQQ for agreement among
replicate stack gas samples to be
established.
7.0 CITED REFERENCES
A1 ford-Stevens. A., T. A. Bellar, J. V.
Eichelberger, and W. L. Buddo.
Method 680. Determinantm of Pesticides
and PCBs in Water and Soil/Sediment by Ca*
Chromatography/Mass Spectrometry. U.S.
Environmental Protection Agency, Envi-
ronmental Monitoring and Support Lab-
oratory, Cincinnati, OH. November
1985.
8eltis et al. Stack Samptihg and Analysis
of Formaldehyde. In Proceedings of the
Ninth Annual Research Symposium on
Incineration and Treatment of Hazardous
Waste. EPA- 500/9-34-015, NHS P384-
234525, 1934.
Code of Federal Regulations, Title 40,
Part 61, Appendix B, 1987.
Erickson, M. 0. Anclyticcl Method: The
Analysis of By-Product Chlorinated 3iphenyls
in Commercial Products and Product Wastes,
Revision 2, U.S. Environmental Protec-
tion Agency, Office of Toxic Sub-
stances, Washington, DC. EPA 560/5-85-
010, NTIS PB86-109089 (1984a).
Erickson, M. O. Analytical Method: The
Analysis of By-product Chlorinated Biphenyls
in Air, Revision 2. U.S. Environmental
Protection Agency. Office of Toxic Sub-
stances, Washington, DC. EPA 560/5-85-
011, NTIS PB86-109097 (1984b).
Erickson, M. 0. Analytical Method: the
Analysis of By-Product Chlorinated Biphenyls
in Water, Revision 2. U.S. Environmental
Protection Agency, Office of Toxic Sub-
stances, Washington, DC. EPA 560/5-85-
012, NTIS PB86-109105 (1984c).
Kuwata et at., "Determination of Ali-
phatic and Aromatic Aldehydes 1n Pol-
luted Airs as Their"2,4-Olnltrophenyl-
hydrazones by" High Performance Liquid
Chromatography," J. Chrom. Sci., 17:264,
Mav 1979.
If.S. Environmental Protection Agency,
Test Methods for Evaluating Solid Waste,
SW-846. Third Edition. November 1986.
Exxon, Procedure for the Titrimetric
Determination of Sulfur Trioxide and Total
Sulfur Oxides in Flue Stack Emissions.
Personal communication from 8. E.
Hurst, Exxon Research and Engineering
Company, P.O. Box 101, Florham Park,
New Jersey 0793Z to H. Von Oem Fange,
Cccper Engineers, 1301 Canal Blvd.,
Richmond, California 94804-3555,
October 3, 1985.
-------
APPENDIX B
SELECTED DETAILED PROCEDURES FOR
MEASUREMENT OF PCOOs~ANO PCDFs
IN HWC STACK EMISSIONS
-------
The detailed procedure descriptions
presented 1n this appendix are Intended
to provide advisory guidance for
selected sampling and analysis opera-
tions presented in Section 3.1.1 of the
body of this document.
1.0 FIELD SAMPLE RECOVERY
At the completion of sampling, the
probe and umbilical should be carefully
removed from the train, the ends of the
probe and the train should be capped
with precleaned aluminum foil or glass
caps, and the probe and train should be
removed to a clean field or permanent
laboratory for sample recovery. The
recovery procedure for each train com-
ponent is described below.
1.1 PARTICULATE CATCH
The filter should be carefully
removed and folded to avoid loss of
collected particulates and placed In a
precleaned sample jar. The contents of
the cyclone (if used) should also be
transferred to the same jar.
1.2 FRONT HALF RINSE
The srooe should be thoroughly
brushed using a natural bristle brush
with steel rod to recovery particu-
lates. During brushing, the probe
should be inclined and rinsed three
times with methanol followed by three
rinses with dichlorcmethane into a
precleaned sample bottle. The front
half of the fiUer holder, cyclone (if
used), and connecting lines between the
filter holder and probe should be simi-
larly rinsed sequentially with methanol
and dicnlorofliethane into the same or a
separate precleaned sample bottle.
1.3 8ACK HALF RINSE
The back half of the filter holder,
condenser, and connecting lines between
the filter holder and the resin
cartridge should also be rinsed threei
times with metnanol following three
rinses with dichloromethane Into a pre-
cleaned samole bottle.
1.4 RESIN CARTRIDGE
The resin cartridge should be removed
from the train, the ends capped with
precleaned aluminum foil or glass caps,
and the cartridge returned to the
analysis laboratory intact.
1.5 FIRST IHPINGER
The water contained in the imp1nc;er
should be aeasured gravimetrically or
by pouring it into a precleaned gradu-
ated cylinder. Following this deter-
mination, the water should be poured
into a precleaned sample bottle. The
impinger should be rinsed three times
with methanol followed by three
dichloromethane rinses and the rinses
added to tne sample bottle. If the
water was measured using a graduated
cylinder, the cylinder should be
similarly rinsed into the sample
bottle.
Following recovery, all sample con-
tainers and the resin cartridge should
be stored at J'C or at wet ice tempera-
tures and srotected from light unt^l
analyzed. The remainder of the
impingers should be recovered as speci-
fied in Method' 5 for determination of
the moisture content of the sample.
2.0 SAMPLE EXTRACTION ANO CLEANUP
2.1 SAMPLE EXTRACTION
2.1.1 Solid Samples
The XAD-2 resin and particulates
catch shculj be extracted with toluene
or benzene in a single Soxhlet ex-
tractor. Pour the resin Into a
B-2
-------
pre-extracted ceilulose thimble in the
sample chamber of the extractor and
rinse the resin cartridge with methanol
followed by the extraction solvent
(toluene or benzene) Into the thimble.
Add the filter to the thimble on top of
the resin. If a cyclone was used, pour
the cyclone catch onto the filter and
resin and rinse the sample container as
required, also into the sample thimble.
If the prooe rinse contains a notice-
able anount of particulates, the rinse
can be filtered through 1 pre-rinsed
cellulose filter and the filter added
to the extraction thimble. The MIS
solution can be spiked before or after
adding the MM5 filter to the thimble.
The combined solids samole should be
extracted for approximately 16 h at a
rate of about 3 cycles/hr. Following
extraction, the extract should be con-
centrated to 5 mL using a rotary
evaporator. Kuderna-Qanish evaporative
concentration can be used for benzene
extracts.
2.1.2 Aqueous Components and Rinses
Aqueous and rinse components should
be partitioned between water and
dichlcromethare. Pour the entire
contents of the sample into an appro-
priate size separatory funnel. Spike
the sample with the MIS solution unless
the extract is to be combined witn
another soiked extract. If tne samp 1e
is composed primarily cf the train
rinse solvents, add a sufficient amount
of reagent water so that the dichloro-
methane is no longer miscible with the
methanol and water, then separate the
layers. Extract the aqueous layer with
three portions of aichiorcmetftane and
compine the extracts. The combined ex-
tracts snould be concentrated using
Kuderna-Oanish equipment.
2.2 EXTRACT CLEANUP
The exacts from the solids (resin
and oarticu'ate catch) and rfnses
should be combined and concentrated to
approximately I at using nitrogen blow-
down. At this paint, approximately
15 uL of tridecane, hexadecane, or
similar keeper should be adceo and the
extract further evaporated to near dry-
ness. The extract should be redis-
solved in 15 kL of hexane in prepara-
tion for cleanup.
The extract s&ould be cleaned as
described in Method 5280. Tr*e crude ex-
tract is initial>y cleaned ay parti-
tioning with 2QS aqueous KOH, 5X
aqueous NaCl, concentrates sulfuric
acid, and 5* aqueous HaCl. Tt?e extract
is then dried by passage through a
cclumn of anhyaraus sodiui sulfate and
concentrated to J by Kcsama-Oanish
and nitrogen blc<sis.
Just prior to iC/MS analysis, an ali-
quot of the cleared extract should be
further concentrates to 2S in tri-
decane or hexacecane. T*e final con-
centration show be cosxueted as
follows.
a. Dispense C5.Q yL of tridecane cr
hexadecane into a 1.0-wl conical
Seacti-vial ana aark tfte level of
solvent on the vial;
b. Add 0.5 to 2.75 mL of tne cleaned
extract to the vial and coventrate to
25 to 50 uL;
-------
c. Repeat the addition and concen-
tration to quantitatively transfer t!»e
entire extract into the Reacti-vial
(including rinses as requires);
d. Concentrate to the nark or
slightly below and make up to the mark
with the keeper solvent; and
e. Spike the extract with the recov-
ery internal standard (RIS) solution.
3.0 SAMPLE EXTRACT AHALYSIS
Sample extracts must be analyzed fcy
GC/MS as described in Method 82S3,
SW-846 using a fused silica capillary
column and selected ion monitoring
(SIM) data acquisition.
3.1 6C/HS INSTRUMENTATION
The GC/MS system must provide suf-
ficient chromatographic separation and
a level of sensitivity sufficient to
achieve the desirad minicaura detects
limits for all subject PCC3s and PCOFs
in the flue gas sample.
3.1.1 Gas Chromatograph
A high performance solitless injec-
tion system and fused silica capillary
column are required. Co'^jot select*^
must consider chromatcgracr^c resolu-
tion of the analytes, but should a'ss
consider tne run time required for ac-
ceptable resolution and t.^e durabilH;«
of the column. Several columns hav*
been successfully used. Method SC?;
recommends D3-5 (30 m), CP-Si 1-53
(50 m), and SP-2250 (30 mj columns.
example column and temperature prcgris
that have been used for ceternunin§
tetrachloro through octacr^ro PCOD arc
PCOF ^cmoiogs is presented b^low.
Example GC Column and Conditions for
PCOD and PCOF Homotogs
Column: 06-5, 30 m
Initial Temperature: 200*C, 4 min
Temperature frcqram: S'C/nir to
33Q*C, 5-min hold
Approximate Analysis Time: 35 nsin
Determination of 2,3,7,8-TCOO and/or
the 2,3.7,8-substituted PCOO and PCCF
congeners requires demonstration of
resolution, of 2,3,7,8-TCOD frco
1,2,3,4-TCDD with < 25X valley. Hence,
longer columns and slower temcerature
programs are typically required. Two
example columns and condition? that
have been used for determining the
2,3,7,3-substituted PCOO and PCF
congeners are presented in Table B-l.
3.1.2 GC/MS Interface
The gas chromatographic column should
be interfaced with the mass spectroa-
eter by threading the -exit end of the
column directly into the ion source of
the spectrometer. This avoias dead
voluste in connecting to a transfer
line. If the column cannot be threaded
into the source, a fused silica trans-
fer line can be used.
3.1.3 Mass Spectrometer
The mass spectrometer must be capable
of selected ion monitoring with suffi-
cient sensitivity to achitve the target
detection limits for PCDOs and fCOFs in
the flue gas sample. The spectrometer
must also be interfaced with a mass
spectrometry data system to provide
continuous data acquisition and storage
during the GC/MS rur) and permanent data
storage in machine-readable form. The
required instrumental detection limits
should be calculated from the target
detection limits, the flue gas samcle
volume, and the final extract vo'ure.
If the extracts are spit:, the fraction
of the total extract avail ao'.e for
analysis must also be considered.
-------
TABLE 8-1. EXAMPLE GAS
CHROMATOGRAPHIC COIUMM CONDITIONS
FOR 2,3,7,8-SUBSTJTUTEO PCDO
AKO PCOF CONGENERS
Example 1
Column: 03-5, 60 m
Initial Temperature: 200°C, 2 min
Temperature Program 1: 5°C/min to
220"C. 16-min hold
Temperature Program 2: 5*C/min to
235*C, 7-«in hold
Temperature Program 3: 53C/min to
330*C, 5-»in hold
Approxiaate Analysis Time: 60 min
Example 2
Column: DB-5, 60 m
Initial Temperature, 190"C, I m1n
Temperature Program 1: 5°C/min to
220*C» 16-min hold
Temperature Program 2: 5aC/min to
235'C, 7-ann hold
Temoeratjre Program 3: S^C/min to
250'C, l3-=nin hold
Temperature Program
-------
TABLE B-2. LIST OF ANALYTES, METHOD INTERNAL.STANDARDS, SURROGATES, AND
RECOVERY INTERNAL STANDARDS FOR DIOXIN/FURAN ANALYSIS
Analyte
Compounds In
calibration standard
Method
internal standard4
Recovery .
internal standard
Tetra-CQO
2.3,7,8-TCDD
1;j-2,3,7,8-TCOO
3?Cll,-2,3,7,8-TCD0 or
Tetra-CDF
2,3,7,3-TCCF
iiC,,-2,3.7,3-TC0F
l3Ci2-l,2,3,4-TCDD
Penta-CQQ
1,2,3,7,3-PeCDO
11Cl2-l,2t3,7,3-PeC00
Penta-CDF
1,2,3,3,9-PeCDF
nCljri,2,3,8,9-?eCDF
Psnta-CDF
2,3,4,7,3-PeCOF
l,Cn-l,2t3,7,3-PeC0D
Nexa-CDD
1,2.3,4,7,3-HxCDO
1JC»1,2,3,6,7,3-HxCDO
i3C,j-l,2.3,6,7,3-H*COO
Hexa-COD
1,2,3,5,7,8-HxCDD
*»Cll-l.2.3i6,?.8-HxCD0
Hexa-C30
1,2,3,7,8,9-HxCDO
l3Cll-l,2,3,6,?,S-HxC0D
Hexa^CDF
1,2,3,4,7,8-HxCDF
i>Cix-*..2»3.6,?,3-HxCD0
Hexa-CDF
1,2,3,5,7,8-HxCDF
iiCtl-I,2,3,o.7,3-HxC00
Hexa-CDF
2,3,4,6,7,3-HxCOF
lKi2-l,2,3,6,?,3-HxCD0
He-xa-CDF
1,2,3,4,3,9-HxCDF
^C,2-I,2,3,6,7,3-HxCDO
Heo*a-CDD
1,2,3,4,6,7,8-HpCDD
i JC, 2-l,2,3,4,5,7,3-HpCOO
•Jeo:a-CDF
1,2,3,4,5,7,8-HpCDF
iiC,j-l,2,3,4,5,7,3-HpCD0
Heota-CDF
1,2,3,4,7,3,9-HpCDF
"Cn-1,2,3,4,5,7,3-HpCGQ
Octa-CDQ
OCDO
»3C,j-CC00
Octa-COF
OCDF
>JC,2-OCOO
d Added to sample prior to extraction.
b -dded to sairple at time of inject{cn into SC/MS.
-------
Once the calibration has been
achieved, ft is verified each day of
saoole analysis by analyzing the aid-
level standard at the beginning and end
of the day. The RRFs for the daily
standards oust be within 20* of the
calibration RRFs, otherwise the recall-
bratlon Is necessary. If 2,3,7,8-TCDD
or all the 2,3,7,8-substltuted con-
geners are to be determined, a column
performance solution must also be ana-
lyzed each day to verify chromato-
graphic separation between 2,3,7,8-TCOO
and 1,2,3,4-TCDO. Separation must be
achieved with s 2524 valley.
3.3 RESULTS CALCULATION
The concentrations of the native
PCOQs and PCOFs are determined using
their response factors relative to the
HIS compounds. ' Results should be
reported as mass of homolog or congener
per staple tfohae, e.g.,.ng TCOO/dscra.
Since an .Internal standard method 1s
used, the. quantities of native PCDOs
and JPCOFs de.*emfped are corrected for
losses during sample extraction,
extract cleanup* extract concentration,
and other extract handling. The recov-
eries of the MS compounds spiked into
the samples prior to extraction are
determined using RRfs relative to the
RIS compound. The recoveries of the
MIS compounds should be within the
range of 50 to 1503C. Recoveries, out-
side this range may indicate problems
in the analysis. Detection limits
should be determined for each specific
congener or homolog not detected using
a response of 2.5 times the background
noise signal at the appropriate reten-
tion time.
I
------- |