Dioxin stack emission concentrations are usually reported as
PCDD/PCDF (ng/dscm @ 7% O2) or TEQ (ng/dscm @ 7% O2) where
ng/dscm = nanograms per dry standard cubic meter @ 7% O2 refers
to a calculation performed to adjust the emisison concentration to a
standard oxygen level.
In performing dioxin stack testing, it is prudent to insure that
sufficient parameters are measured so that multiple comparisons can be
determined. It is helpful to be able to determine mass emission
rates, concentration emission rates, total PCDDs/PCDFs and toxic
equivalency factors (TEQs). Along with the gas flow rates in the
stack, oxygen should also be monitored during the stack testing, so
that concentrations corrected to a specific oxygen level can be
For example, total dioxins and TEQs are not the same, nor is there a
consistent ratio. According to the June 1994 draft USEPA dioxin
reassessment, typical dioxin concentrations in cement plant stack
emissions (14 kilns) range from 1.983E-09 to 1.998E-06 (0.000000001983
to 0.000001998). The toxicity equivalency (TEQ) for these same 14
kilns ranged from 1.750E-11 to 4.318E-08 (0.0000000000175 to
0.00000004318). While the lowest total PCDD/PCDF concentration
did equate to the lowest TEQ, the highest total PCDD/PCDF concentration
did not equate to the highest TEQ.
Process Sample Concentrations (solid matrix)
Dioxin concentrations are also determined in solid matrices such as
soils and cement kiln dust (CKD). As an example, in the USEPA
Report to Congress on Cement Kiln Dust (RTC), dioxin concentrations
were analyzed and reported as total concentrations (μg/Kg) and TCLP
concentrations (μg/L). PCDD/PCDF concentrations are often
reported with laboratory notations indicating something other
than detection at the value noted. Some of those notations are as
< = not detected, the associated value is the detection limit
N.A. = detection limits are not available
B = the constituent was detected in an associated blank
J = the concentration is an estimate
The dioxin/furan constituent of most concern is 2,3,7,8-TCDF.
Total 2,3,7,8-TCDF concentrations were reported for seven cement kilns
in the RTC. Results for three of those kilns were reported below
the analytical detection limit. Because of variations between
runs, detection limits varied for all three of those kilns. The
detection limits were 0.00065, 0.00087 and 0.00099 (μg/Kg) or
micrograms per kilogram. The four facilities that reportedly did
yield a measurable concentration ranged from 0.00039 to 0.038
μg/Kg. These are typical concentrations to target with
appropriate analytical procedures.
Often concentrations of dioxins are presented as toxicity equivalents
(TEQ). TEQs are determined by summing the products of multiplying
concentrations of individual dioxins times the corresponding TEF (See
Table 1-1) for that compound [Section 4 Code of U.S. Federal
Regulations (CFR) 266 Appendix IX].
Calculating TEQ's for PCDDs/PCDFs
Section 4.0 of Appendix IX, 40 CFR Part 266 does not state how to treat
"non-detect" or less than detection limit values. Section 4.0
does reference USEPA document number USEPA/625/3-89/016 which has
sample calculations. GCI obtained a copy of this document which
is imposingly entitled "Interim Procedures for Estimating Risks
Associated with Exposures to Mixtures of Chlorinated Dibenzo-p-dioxins
and Dibenzofurans (CDDs and CDFs) and 1989 Update." This document
was published/released in March of 1989.
The sample calculations in this document have a number of instances
where a value of zero is given for individual congener concentrations
in real waste or real stack emissions. The subsequent Toxic
Equivalent Quantity (TEQ) value that was calculated, based on the
Toxicity Equivalency Factor (TEF) supplied by the USEPA, was also
zero. There are a few instances where congener concentrations
were listed as "<" a certain value. In these cases the TEQ
value was listed as a zero or as a line, effectively a zero.
There is one instance of a "ND" being listed as a value for a
homologous group (A homologous group is not differentiated by isomer,
in this case all TCDDs.) The subsequent TEQ calculated was listed
as a short line, effectively zero. GCI, in examining these sample
calculations, was careful to discount any source data for which the TEF
Based on this referenced document any PCDD/PCDF values that are below
the analytical detection limit for the method should be listed as zeros
or non-detects before calculating the TEQ. GCI cautions that the
sampling and analytical method QA/QC requirements must be met to
demonstrate an adequate lower detection limit.
Sample PCDD and PCDFs Report
While report information and reporting format will vary between
analytical labs and stack testing firms, Table 1-2 provides an example
of what a PCDD/PCDF report might look like. Note that this
particular report lists specific runs across the top to the columns,
along with run start and stop times. Other information that is
presented are gas conditions, volumetric flow rates, TEQs and total
PCDFs & PCDDs. Averages of all the reported values are also
There is no standard reporting format for stack test data.
Consequently, a review of the report format with a testing firm
representative is recommended.
The following are things to look for when reviewing or comparing stack test data:
1. Are the
conditions "standard" as required by USEPA? That is at 25oC?
reports will indicate that 0 degrees C was used rather than 25 degrees
C. ("Standard" conditions for emission reports in Europe are
different than the U.S. For this reason, emission results and
regulatory limits cannot be directly compared.)
2. Is the
reported mass of the extract, in nanograms (ng), already divided by the
sample volume or is it the mass of the total sample?
Dioxin samples are collected onto resin and then extracted. The
testing company should report the dioxin that was present in the
extract as a mass without a volume divisor.
Example 1: The stack sample reveals 0.190 ng TEQ for 4.23 dry standard cubic meters of sample,
0.190 ng/4.23 m3, which can be written as 4.5E-02 ng/dscm.
In order to correct to 7% O2, use the following correction calculation:
where 21 is the percentage of oxygen in the
atmosphere, and 7% is the standard to which the value is to
(21 - XO2)
XO2 is the actual stack gas oxygen concentration demonstrated during the test being corrected.
(TEQ value in ng/dscm) x 14
(21 - XO2)
Example 2: the stack gas oxygen for example/sample 1 was 11.27%
4.5E-02 ng/dscm x 14
= 6.47E-02 ng/dscm TEQ @ 7% O2
(21 - XO2)
To convert concentration to a mass emission rate (e.g. grams per
second), the flue gas flow rate at standard conditions must be
known. Sometimes this is reported in dry standard cubic feet
(dscf) rather than in metric units such as dry standard cubic meters
If the report provides:
Dioxin concentration (ng/dscm) 6.47E-2
Volumetric flow rate standard conditions (dscfm) 55,802
Per the following example of this conversion first coverting the flow rate from dscfm to dscm/min.
x 1 dscm
= 1580.4 dscm/min
minute 35.31 dscf
6.47E-02 ng/dscm x [kiln stack flow rate/minute] x 1 minute/60 seconds
6.47E-02 ng/dscm x 1580.4 x 1
= 1.7 ng/sec = 1.7 x 10 -9 g/sec
As an added note, GCI would point out that the current USEPA "Guidance
on Structuring Trial Burns for Collection of Risk Assessment Data"
released as an "Internal Review Draft" in May of 1997 discusses
the use of the "full detection limit" value in calculation emissions of
PICs for use in risk assessments "If the permit writer is setting an
emission limit on the compound of concern..." Although this
statement has been limited to PICs and PICs are discussed separately
from PCDDs and PCDFs, a zealous permit writer may insert that the
Dioxin TEQs be calculated in this manner. The permit writer is
aided in being able to make this decision by the lack of a specified
method of TEQ calculation. For BIF units, a specified TEQ
calculation method is included in Appendix IX 40 CFR Part 266 which in
turn references other USEPA documents that document the use of "Zeros"
as the value for any non-detects (ND) in PCDD/PCDF emissions.
TEQ Calculation Considerations
Generally the stack testing firms will report the results of the USEPA
Method 23 PCDD/PCDF analysis as total PCDD and PCDF concentration
values (or as subtotals and totals of families of congener, i.e. tetra
or penta) which are then converted to TEQ values and reported TEQ
totals. In such cases, the individual congener are not reported
and the owner/operator need not, and can not, calculate the TEQ
values. However, if the individual congeners are reported, the
worksheet (Appendix B, which may be copied) will aid in the calculation
of the TEQ values. Please remember that in the past there have
been a variety of toxic equivalency factors (TEFs). As an
example, the USEPA had a 1981 version as well as the currently accepted
1989 version. Ontario (Canada) and New York state have their own
versions as do the FDA, the Swiss and Britain. It is possible
that the USEPA will change the TEF values in the future based on
continuing scientific research of the health effects of these
compounds. Consequently, it is recommended that the analytical
report include the concentrations for each of the congeners as well as
the total PCDD/PCDF and total TEQ generally reported.
In the USEPA Combustion Emission Technical Resource Document (CETRED),
USEPA examined dioxin/furan emissions corrected to a stack gas oxygen
level of 7%. This is also consistent with continuous
emission monitoring required by the boiler and industrial furnace
regulations (BIF). It is important to note that European
guidelines require oxygen correction to 11% if you are comparing test
results from European facilities.
STACK SAMPLING METHOD
The sampling method required in the United States is USEPA Method
23. It can be found in Appendix C of this document. The
sample is withdrawn isokinetically and collected in the sample probe,
on a glass fiber filter, and on a packed column of absorbent
material. The PCDDs and PCDFs are extracted from the sample,
separated by high resolution gas chromatography, and measured by high
resolution mass spectrometry.
Prior to the actual sampling for dioxins certain preliminary
determinations must be made. These preliminary determinations are
common for several emissions stack sampling methods. The
procedures described partially in USEPA Method 5 (Appendix D) for
particulates are the same as for a number of emissions.
Determinations for some other emissions have their own USEPA method
number. They provide the basis for proper setup and selection of
such things as: sampling site and number of sample points, sampling
volume and time, size of nozzle and probe and the measuring of physical
conditions of the stack gas which enable the proper calculation of
emissions. USEPA Method 5 section 4.1.2 currently describes these
determinations and references USEPA Methods 1-4 for more specific
The proper sample cleanup procedure begins as soon as the probe is
removed from the stack at the end of the sampling period. The
sampling components must have their external surfaces carefully wiped
to prevent contamination of the final sample. Then the probes,
filters, impingers, etc. must carefully have their contents rinsed or
transferred to collection vessels for the analysis.
Detection Limits or Quantitation Limits
The detection limit for each isomer is determined by measuring the
amount of analytical instrument response from the injection of
calibration and internal standards. This determines the relative
response factor. If a blank were injected into the analytical
instrument, there would likely be some level of response, even though
no actual analyte was injected. This is considered to be
background noise. Using the response factor data and the
background noise in the analyzer, a signal to noise ratio is
calculated. The calculation includes ensuring that the ratio is
greater than 2.5, which is required by the method. A minimum
detection limit (MDL) is thus determined.
Determination of the amount of a given analyte would not be possible if
the amount detected was below the minimum quantitation limit. In
other words, quantitation would not be possible because the amount
detected would be below the analytical instrument quantitation limit
and consequently could not be determined accurately. Analyte
concentrations are often reported with laboratory notations reflecting
some analytical limitation. Some of these notations sometimes
used are as follows:
< = not detected, the associated value is the detection limit
N.A. = detection limits are not available
B = the constituent was detected in an associated blank
J = the concentration is an estimate, i.e. less than
the quantitation limit but greater than the detection limit.
(1) _______ Is the nozzle made of
nickel, nickel-plated stainless steel, quartz, or borosilicate glass ?
(2) _______ Are sample transfer lines heated?
(3) _______ Is the filter support Teflon or Teflon-coated wire?
(4) _______ Do the number of
sampling points add up to the minimum requirements of the method being
(5) _______ Were the sample trains assembled without the use of any sealing greases?
(6) _______ Were all leak checks performed as required?
(7) _______ Was a clean
contaminant-free enclosed location provided for sample train breakdown
and sample recovery operations?
(8) _______ Do field check sheets
provide notes on observations as well as a record of all temperature
and leak checks?
(9) _______ Do lab reports contain
a narrative of any QA/QC observations during sample
transportation, sample preparation, analysis, data reduction, or
(10) _______ Does the lab report
contain all required maintenance and calibration documentation?
(11) _______ If an audit sample
was required, was it run and was it an approved USEPA audit sample?
Sampling Train Collection Efficiency Check. An aliquot of the
specified surrogate standard must be added to the sample train
cartridges before collecting the field samples.
After the stack sample has been collected, USEPA SW846 analytical
method 8290, section 6.0 Sample Collection, Preservation, and Handling,
calls for stack samples to be extracted within 30 days and completely
analyzed within 45 days of extraction.
Typical QA/QC Problems
1. Breakage of sample train collection vessel. Recommend
taking an extra run to cover breakage of at least one vessel.
2. Failure of sample train to pass all leak check procedures.
3. Plugging of the resin trap causing a build up of vacuum in the sampling train.
SAMPLE TESTING (ANALYSIS)
The complexity of USEPA Method 23 (Appendix C) is such that, in order
to obtain reliable results, analysts should be trained and experienced
with the analytical procedures. Normally, a stack testing
contractor would have an experienced analytical lab which already does
their work. If you have previously used an analytical lab which
has given you good, reliable, defensible data, then it would be prudent
to consider using them. Section numbers references are for USEPA
Method 23 as published in 40 CFR Part 60, Appendix A (July 1, 1996
While method accuracy is the responsibility of the stack testing crew
and the analytical laboratory, it may be helpful to have a basic
understanding of the concerns involved. The accuracy of this method
depends upon proper preparation of the sampling system and associated
apparatus (Section 4.1.1). Also critical are the preliminary
determinations (Section 4.1.2) and the sample recovery procedures
(Section 4.1.3). The first phase of analysis is the extractions
performed on the containers from the sample recovery (Section
4.2). These extractions are then cleaned up and fractionated
(Section 5.2). From this point the next step is to inject an
aliquot of the prepared extracts into a gas chromatograph/mass
spectrophotometer (GC/MS), however, just prior to injection into the
GC/MS, a known spike of a recovery solution is added to the prepared
extract (Section 5.3). This is used with the quality control for
Every analytical method has a detection limit. The detection
limit is dependent on a wide variety of factors including the
analytical instrument, in this case a GC/MS. While normal method
quality assurance/quality control (QA/QC), such as initial instrument
calibration (Section 6.1.1) and daily performance checks (Section
6.1.2), helps to qualify method detection limits, it is an
important issue that is sometimes confusing. For instance, in
some cases, analytical results will indicate that there is a high
likelihood that a substance has been detected but because of the
limitations of the method/analytical instrument, it may not be possible
to quantify how much of the substance was present. A particularly
confusing issue can occasionally arise when a regulatory limit is lower
than the detection limit of any known method or analytical
instrumentation. Matrix specific interferences in any given
run/sample exasorbate the problem. All of these issues should be
addressed to some degree in the final analytical laboratory report.
Sample Matrix Issues
It is important to choose the appropriate analytical method. Just
because a specific analytical method is designated does not necessarily
mean that it is the best method for what you are trying to
accomplish. For instance, USEPA initially used SW846 8280
(Appendix E) to analyze CKD samples for the CKD report to
Congress. This method was debatably inappropriate based upon the
sample matrix. The appropriateness of analytical methods was so noted
in Section D of the Regulatory Determination on Cement Kiln Dust
published in the 2-7-95 Federal Register. Make sure that the chosen
analytical method is the most appropriate for what you are trying
USEPA SW-846 Method 8290
SW846 8290 (Appendix F) provides procedures for the detection and
quantitative measurement of PCDDs and PCDFs in a variety of
environmental matrices at part-per-trillion (ppt) to
part-per-quadrillion (ppq) concentrations. The sensitivity of
this method is dependent upon the level of interferences within the
This procedure uses matrix specific extraction, analyte specific
cleanup, and GC/MS analysis techniques. The method also provides
selected cleanup procedures to help eliminate encountered
interferences. Quantitation of individual PCDD/PCDF congeners,
total PCDDs and total PCDFs is achieved in conjunction with the
establishment of a five point calibration curve for each homologue,
during which each calibration solution is analyzed once.
Stack samples must be stored at 4E C in the dark, extracted within 30
days and completely analyzed within 45 days of extraction.
Internal standards are used in this procedure and specific calibration
procedures must be followed.
A gas chromatograph (GC) column performance check is only required at
the beginning of each 12 hour period during which samples are
analyzed. A method blank run is required between a calibration
run and the first sample run. The same method blank extract may
thus be analyzed more than once if the number of samples within a batch
requires more than 12 hours of analyses.
Chromatography time for PCDDs and PCDFs exceeds the long term mass
stability of the mass spectrometer (MS). Because the instrument
is operated in the high resolution mode, mass drifts of a few ppm can
have serious adverse effects on instrument performance.
Therefore, a mass drift correction is mandatory, as described in the
A field blank is required for each batch of samples to be
analyzed. Many times, a rinsate that was used to rinse the
sampling equipment, is also included. Analysis of the rinsate
helps to insure that the samples were not contaminated by the sampling
equipment. Duplicate analyses of some samples is also required as
are matrix spikes and matrix spike duplicates.
Method 8290X is a modification of USEPA SW846 8290 used by Triangle
Laboratories of Research Triangle Park in South Carolina. While
the methods are very similar, Method 8290X is designed to provide a
greater analytical sensitivity, more comprehensive PCDD/PCDF
quantitation and stronger defensibility of the analytical
results. A comparison between SW846 8290 and the modification
8290X is provided in Appendix G. Other laboratories have made
similar modifications to the method.
Typical Checklist (Partial) for Sample/Analysis/Data Receiver
(1) _______ Was all
glassware cleaned and made free of silicone grease, especially on glass
(2) _______ Is the nozzle
made of nickel, nickel-plated stainless steel, quartz or borosilicate
(3) _______ Do sample transfer lines need to be heat traced?
(4) _______ Is the sample transfer line as short as possible?
(5) _______ Is the filter support made of Teflon or Teflon-coated wire?
(6) _______ Does the condenser conform to Figure 23.2?
(7) _______ Were the Section 3.0 reagent procedures followed?
(8) _______ Were traps loaded in a clean area, i.e., not in the field?
(9) _______ Were all samples sealed with aluminum foil or Teflon tape?
(10) _______ Were gas entry temperatures monitored and kept below 20EC for the XAD-2 resin?
(11) _______ Were all leak check procedures performed?
(12) _______ Was the proper
cleanup procedure begun as soon as the probe was removed from the stack
at the end of the sampling period?
(13) _______ Were all openings to
sample trains, probes, etc. capped except when inserted or when
sampling was underway?
(1) _______ Were appropriate sample extraction procedures followed?
2) _______ Were appropriate
sample cleanup and fractionation procedures followed?
Quality Control (7.0):
(1) _______ Was the PCDD/PCDF internal standard added to every sample before extraction?
(2) _______ Were the surrogate
compounds added to the resin absorbent sampling cartridge before the
sample was collected?
(3) _______ Was the toluene QA rinse reported separately from the total sample catch?
Quality Assurance (8.0):
(1) _______ Were audit samples run along with the stack samples?
(2) _______ Were the surrogate recoveries between 70 and 130 percent?
(3) _______ Were all samples extracted within 30 days of collection?
(4) _______ Were all samples analyzed within 45 days of collection?