NCD980602163_19930805_Warren County PCB Landfill_SERB C_Koppers Superfund Site - Demonstration of BCD Technology, Response to Quality Assurance Comments-OCR

t<tGt\VtU
PUG l '.3 1993
SUrERfONP SECTION
SITE DEMONSTRATION OF
THE BASE-CATALYZED DECOMPOSITION (BCD) TECHNOLOGY
AT THE
KOPPERS COMPANY, INC. SITE
MORRISVILLE, NORTH CAROLINA·
RESPONSE TO QUALITY ASSURANCE COMMENTS
Prepared for
U.S. EJ\iTVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Cincinnati, Ohio 45268
Work A5signment No.
Date Prepared
EPA Contract No.
PRC-EMI No .
PRC Project Manager
Telephone No.
EPA Project Manager
Telephone No.
0-11
August 5, 1993
68-C0-0047
047-1100
Robert Hutcheson
[phone]
Terry Lyons
[phone]
TABLE OF CONTENTS
Section
1.0 RESPONSE TO CONCERNS
2.0 RESPONSE TO MINOR ISSUES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.0 CHANGES RECOMMENDED BY SITE TEAM MEMBERS ....................... 16
3.1 CHANGES RECOMMENDED BY RADIAN CORPORATION ................. 16
3.2 CHANGES RECOMMENDED BY VERSAR, INC. ........................ 19
3.3 CHANGES RECOMMENDED BY INTERNATIONAL TECHNOLOGY
CORPORATION .............................................. 20
ATTACHMENTS
Attachment
A REVISED SAMPLE LOCATION MAP
;
B STANDARD OPERATING PROCEDURE FOR THE GAS CHROMATOGRAPHY/MASS
SPECTROMETRY ANALYSIS OF AIR TOXICS CANISTER SAMPLES USING METHOD T0-14
1.0 RESPONSE TO COl\.1MENTS
The Quality Assurance (QA) manager for the Risk Reduction Engineering Laboratory (RREL)
identified four concerns regarding the Quality Assurance Project Plan (QAPP) for the Base-Catalyzed
Decomposition Technology SITE demonstration at the Koppers Company, Inc. site in Morrisville,
North Carolina. The PRC SITE team's responses to concerns follow.
la. For sample points 2 and 3 (untreated and treated soil), Table 4-3 specifies 12 composite
samples per run, while Table 4-1 specifies four 2.5-hour composite samples. TI1e
associated text states that each of the 2.5-hour composites will consist of at least four
grab samples. Please clarify the number of samples that will be collected, and explain
the tintlng and procedures for collecting the subsamples.
Response: Four composite samples will be collected per test run at sample locations 2 and 3.
One composite sample will be collected from each sample location every 2 hours per 8-hour
test run. A composite sample will consist of four grab samples with one grab sample
collected every 30 minutes. Each grab sample will be contained within a compositing jar
which will be covered with a Teflon®-lined lid and placed in a cooler in between collection of
each grab sample. Originally, 12 composite samples were to be collected per test run from
each sample location (four samples every 2.5 hours for a total of 12 samples per sample
location).
lb. In Table 4-3 under sample location 2, there is an entry labeled "soclium carbonate
reagent" with zero samples and one blank. What is meant by this entry?
Response: One composite sample of the sodium bicarbonate reagent will be collected during
mixing of the contaminated soil with the reagent. The composite sample will consist of at
least 4 grab samples and will be analyzed for PCDDs/PCDFs, SVOCs, VOCs, and metals.
The QAPP will be revised to designate the sodium bicarbonate sample as sample location 11.
The original entry was meant to refer to sampling of a "reagent blank" to assure that the
reagent contains no contaminants of concern.
le. Table 4-3 provides for matrix spike samples for PCP for contantlnated soils, while the
teAi explains that these samples are too concentrated for matrix spiking. Consistency is
needed.
Response: Matrix spike samples will be collected for SVOCs analysis as outlined in Table 4-
3. The text of the QAPP will be revised for consistency.
ld. Table 4-1 and Section 4.4.1 list six composite samples for sample location 1 for the TCLP,
procedure, while Table 4-3 lists only three san1ples and one field duplicate. How many
samples are intended.
Response: The collection of TCLP samples at sample location 1 has been canceled. TCLP
samples (1 _per test run) will only be collected at sample location 3.
le. Table 4-3 lists oil and grease (O&G) for sample location 1 while O&G is not listed for
this stream in Table 4-1. Please clarify.
Response: Analyses for oil and grease (O&G) will be conducted on samples collected from
sample location 1. Table 4-1 will be revised to correspond with Table 4-3 .
lf. Table 4-3 lists pH for sample location 2 while pH is not listed for this stream in Table 4-
1. Please clarify.
Response: Samples will be collected for pH determination at sample location 2~ Table 4-1
will be revised to indicate the .collection of pH samples at sample location 2.
lg. Table 4-1 lists moisture for sample location 2 for the purpose of converting data to a
dry-weight basis to calculate the efficiency of the BCD process. This seems to be
inconsistent with the discussion concerning equation 1-1, which caJculates removal based
on the total mass of contaminant. Please clarify.
Response: The detem1ination of moisture content will be used for the purpose of converting
data to a dry-weight basis only. Table 4-1 of the QAPP will be revised to indicate the change
in rationale for collection of moisture content at sample location 2. Since data from both
locations 2 and 3 will be converted to a dry weight basis, the contaminant removal efficiency
of the BCD process will be determined by contaminant concentration as opposed to mass of
contaminant.
lh. Table 4-3 lists metals, O&G, and bulk density for sample location 3 while these
parameters are not listed for this stream in Table 4-1. Please clarify. Is bulk density
needed to caJculate the total mass of treated soil? If so, it should also be included for
untreated soil (Sample Point No. 1)).
Res ponse: The analysis of metals, O&G, and bulk density will be conducted for samples
collected at sample location 3. Also, bulk density will be detem1ined for soil at san1ple
location 1 as indicated in Table 4-3. Table 4-1 of the QAPP will be revised for consistency.
Bulk density is a non-critical parameter and will be used to determine soil property changes
resulting from the L TTR/MTTR. Soil mass will be determined by weighing drums of soil
entering the L TTR/MTTR units.
li. Table 4-3 lists three time-integrated samples per test run for each of the gaseous streams
while footnote "d'1 states one composite sample per run. Please clarify the number of
samples per run for the gaseous streams in Table 4-3.
Response: One time-integrated sample per test run will be collected for each gas stream
sample location. Table 4-3 will be revised accordingly.
::,
lj. Table 4-3 lists metals analysis for sample location 6 while metals analysis is not listed for
this stream in Table 4-1. Please clarify. Sodium hydroxide reagent is listed as a
parameter for sample location 6 in Table 4-3 with zero samples and 1 blank to be taken.
This entry is not understood. Please explain.
2
Response: Metals analysis will be conducted on samples collected from sample location 6 as
indicated in Table 4-3. Table 4-1 will be revised to correspond with Table 4-3.
Also, a separate sample of the sodium hydroxide reagent will be collected per test run of the
LTR. The sample will be analyzed for PCDDs/PCDFs, SVOCs, VOCs, and metals. The
QAPP will be revised to designate the sodium hydroxide reagent sample as sample location
12.
lk. Sample locations 6 and 7 are described as solid samples in Table 4-3. These are oil
samples. Please revise.
Response: Sample locations 6 and 7 are the L TR influent and effluent which consists
primarily of L W 110 oil from the LTTR/MTTR oil scrubber system. The QAPP will be
revised to indicate sample locations 6 and 7 as being oil samples.
11. Table 4-3 lists particulate loading and moisture content for sample location 8 while these
are omitted from Table 4-1; on the other hand, Table 4-1 lists O&G and pH for this
sample point, both of which are inappropriate for gas samples. Please clarify.
Response: Particulate loading and moisture content samples will be collected for analysis at
sample location 8 as indicated in Table 4-3 . Table 4-1 of the QAPP will be revised
accordingly. Also, oil and grease (O&G) samples and pH san1ples will be _removed from
sample location 8 in Table 4-1.
lm. Table 4-3 lists three time-integrated samples per run for sample location 8. Footnote "d"
states that there is 1 composite (integrated) sample per run. Please clarify the number of
samples per run for sample location 8 in Table 4-3.
Response: As stated in Response 1 i, one time-integrated sample per test run will be collected
from each gas stream location. Table 4-3 of the QAPP will be revised for consistency .
In. For sample location 9 in Table 4-3, seven samples are listed per test run. Footnote 11 r1
states that there is 1 sample for each thermal desorber run and LTR run. Please corred
the number of samples per run for sample location 9 in Table 4-3.
lo.
Response: One time-integrated gas sample will be collected from the stack gas sample
location (designated as sample location 5 in the revised QAPP) during six runs of the
LTTR/MTTR and two runs of the L TR. Table 4-3 of the QAPP will be revised to indicate
the changes.
Three samples per test run are listed for sample locations 10 and 11 while footnote "c"
lists four composites for each five runs. Please clarify the number of samples per run
for these streams.
Response: The condensate water will be stored in the condensate storage tank for use in the
LTTRJMTTR water scrubber system during the course of the demonstration. The condensate
water will be treated at the end of the demonstration, prior to demobilization. Three grab
samples will be collected from the...c.ondensate storage tank (designated as sample location 9 in
3
the revised QAPP) prior to treatment. The treated condensate (wastewater) will be placed
into a treated wastewater storage tank. Three grab samples (designated as sample location 10
in the revised QAPP) will be collected at the end of wastewater treatment. Tables 4-1 and 4-
3 in the QAPP will be revised to indicate the change in sample frequency . Also a sample of
the scrubber water/cooling water supply source will be collected to assure that no
contaminants of concern are detected with the water supply.
lp. Preliminary samples will be analyzed on site '"ith a portable gas chromatograph. Which
samples will be used for this purpose?
Response: Preliminary samples will be analyzed on site for all sample locations (primarily
for sample locations I and 3) except gas sample locations. Data will be used for process
control only and is considered non-critical.
lq. For most analyses at sample point 1, three composite samples will be collected over the
entire project. Section 3.3.1 explains the compositing procedure once subsamples have
been collected. However, there is no explanation regarding how the initial subsamples
will be collected. These should be timed to be representative of the entire mass of soil
that is treated.
Response: The text describes collection of grab samples in paragraph 3 of Section 4.4.1. A
determination of collection times for grab samples will be estimated during manual screening
of the soils based on the time necessary to screen approximately 0.5 ton. Screening of soils
for the demonstration will be completed prior to the start of the BCD process demonstration.
The text of the QAPP will be revised to indicate the method for determining the grab sample
collection time interval.
lr. According to page 4.0-3, samples of treated soil will be collected before it is cooled with
process water. This approach raises two concerns. First, this approach will overlook
any addition of contaminants from the process cooling water. Second, volatiles would
likely be lost during the sampling procedure. Is the description of this sampling point in
the text correct, or will the soils actually be sampled once they have cooled to room
temperature, as is suggested in Figure 4-1?
Response: Treated soils (sample location 3) will be collected after it is cooled with process
water. The text of the QAPP will be revised to indicate the point of collection for soils at
sample location 3.
The cooling water will consist of either tap water obtained from the local municipal water
supply system, or treated condensate water from the BCD process. Samples of the cooling
water will be collected to assure no contaminants of concern are detected in the cooling water
supply source. -';
ls. Approximately 10 percent of the samples in Table 4-3 are field duplicates. However,
there is no indication of how such duplicates will be collected. Please provide procedures
for collecting field duplicates. How '"ill the data from such duplicates be treated?
4
Response: Field duplicate samples will be generated by collecting twice the necessary volume
for a designated sample. Double volume samples will be composited and quartered prior to
containerization. The data for duplicate samples will be used to assess sample variability.
The duplicate results will be averaged for the sampling event that was duplicated. The text of
the QAPP will be revised to indicate the collection method and use of duplicate sample data.
lt. Volatile organics will collected with methanol exiraction procedure. Please note that this
approach will result in poorer detection limits for the VOCs than are listed in Table 3-4.
Please revise this table accordingly to be sure that all parties are in agreement regarding
this determination. Also note that this procedure will require that the mass of soil and
the volume of methanol be recorded to permit a calculation of the final concentrations in
the soil.
Response: Volatile organic samples will be collected as individual grab samples and analyzed
by Method 8240. The methanol extraction method will not be used. The text of the QAPP
will be revised to indicate the change in the volatile organic sample collection procedure.
2. How will the mass of treated and untreated soil be determined? This information is
needed to estimate treatment cost (objective 4) and for calculating removal efficiencies.
Will the soil be weighed in some manner, or will soil mass be estimated from volume
and bulk density measurements?
It is unclear whether soil mass should be considered as critical or noncritical. Section
1.5 discusses calculation of the removal efficiency based on the total amount of
contaminant rather than the contaminant concentration. These procedures ½ill require
that the mass of material used and recovered for each run be measured in a traceable
manner. However, the mass of feed material or product has not been designated as a
critical parameter, and no QA objectives been set for mass determinations. If the data
for this project will be analyzed according to equation 1-1, the mass determinations will
be critical and QA objectives must be set. Appropriate calibration and QC procedures
will also have to specified. Please make appropriate revisions.
Response: The soil mass is considered a critical measurement for detem1ining treated cost,
and will be determined by weighing feed soil entering the L TTR and MTTR. Drummed
contaminated soils will be emptied into the LTTRJMTTR feed hopper. The drums will be
weighed prior to and after emptying the contents of the drums . The mass of treated soil will
not be measured. As stated in Response lg, removal efficiency will now be determined by
concentration.
3. Section 1.3 lists 5 objectives for this demonstration. There are two concerns related to
these objectiYes.
3a. The second objective is to "Detennine the ability of the process to produce effiuents
(treated soil, oil, water, and gases) that meet applicable local, state, and federal
standards for disposal or discharge." Later in this section (page 24), the measurements
critical to this objective are listed as PCP, PCDDs, and PCDFs in gas, water, and oil, as
well as particulate matter in the stack gas. However, the numerical values for the local,
state, and federal standards are ~ote presented. Without these values, the
5
appropriateness of the sampling and analysis procedures cannot be evaluated. Please
supply the numerical values for these standards, if possible.
Response: The soil residual must be treated. For soil, PCP must be treated to less than 95
ppm and dioxins and furans must be treated to less than 7 ppb.The appropriate restriction for
disposal of residual wastewater and air are listed below.
Federal Effluent Requirements (40 CFR Part 268)
HXCDD -all Hexachloro-dibenzo-p-dioxins
HXCDF -all Hexachloro-dibenzofurans
PeCDD -all Pentachloro-dibenzo-p-dioxins
PeCDF -all Pentachloro-dibenzofurans
TCDD -all Tetrachloro-dibenzo-p-dioxins
TCDF -all Tetrachloro-dibenzofurans
2 ,4 ,5-Trichlorophenol
2 ,4,6-Trichlorophenol
2, 3 ,4,6-Tetrachlorophenol
Pentachlorophenol
Phenol
2 ,4-Dichlorophenol
2,6-Dichlorophenol
Hexachlorodibenzo-p-dioxin
Pentachl orod ibenzofurans
Pentachlorodibenzo-p-dioxin
Oil and grease
Phenols
pH
< 1 ppb
< 1 ppb
< 1 ppb
< 1 ppb
< 1 ppb
< 1 ppb
< 1 ppb
< 0.05 ppm
< 0.05 ppm
< 0.01 ppm
_.::;_ 0.039 mg/I
_.::;_ 0.044 mg/I
_.::;_ 0.44 mg/I
_.::;_ 0 . 000063 mg/I
_.::;_ 0.000063 mg/1
_.::;_ 0 . 000063 mg/I
1.5 lb/1000 cubic feet of wastewater
0.14 lb/1000 cubic feet of wastewater
8.0 to 9 .0 at all times
State Air Emmission Requirements (NCAC Part 20)
HXCDF
Pentachlorophenol
Phenol
Tetrachlorodibenzo-p-dioxin
Sodium dioxide
Nitrogen dioxide
Particulates
0.0051 lb/yr
0 .063 lb/day
0 .0064 lb/hr
0.24 lb/hr
0.00020 lb /day
2.3 lbs/million BTU input
0.6 lb/million BTU input
1. 83 lb/hr per 600 lb/hr process material
3b. The third objective is to "Evaluate potential effects of chemical and thermal reactions of
the BCD process on PCP, dioxins, and furans, such as forming other hazardous
compounds." The discussion on page 24 states that for the third objective "the critic.al
parameters are PCP, PCDDS, and PCDFs in the contaminated feed soil, treated sludge,
treated liquid effiuent, and gas __f2Citing the stack." The measurement of these parameters
6
is critical to the first objective; their applicability to the third objective is unclear. If one
objective of this project is actually to identify reaction products, one would expect to
employ analytical methods that are oriented towards the identification of such products.
Rather, this project simply employs EPA method 8270 for standard analyte list. To
identify reaction products, one might concentrate on identifying likely products, such as
chlorinated phenols or partially oxygenated forms of pentachlorophenol. Perhaps
reaction products have been identified from laboratory studies, and these products could
be searched for during the current project. Please clarify this objective, and inclicate
which methods are critical to its attainment. Perhaps this objectiYe could be addressed
more easily with a small scale, laboratory study.
Response: Objective 3 is considered a secondary objective of the demonstration. Page 24 of
the QAPP will be revised.
The primary reason for choosing Method 8270 was the large number of compounds which can
be analyzed, simultaneously. Many of these compounds may be degradation or breakdown
products of the thermal decomposition of PCP. These compounds include many of the
chlorinated phenols, non-chlorinated phenols, ethers, and phenyl ethers. Another important
reason for choosing Method 8270 is in the qualitative, positive identification of target
compounds. The ability to positively identify compounds is a major advantage of this
method . Another advantage for the use of Method 8270 is that it is possible to identify
compounds which are not included in the target analyte list. This is performed through
tentatively identified compounds (TICs). The 20 TICs with the highest peaks will be
identified if possible. The mass spectrometer software program includes a 50,000 chemical
compound library of mass spectra data. This library can be invoked to help in the
identification and quantitation of non-identified chemical compounds found in samples. This
may be extremely helpful when searching for PCP degradation products.
4. It is questionable whether the accuracy objective of 50-150 percent recovery can be
achieved for pentachlorophenol (Tables 5-4 and 3-1) without preliminary effort. PCP is
notorious for poor recoveries. In fact, Table 6 in Method 8270 lists the range of
recoveries for PCP as 14-176 percent. The adclition of sodium carbonate may further
stabilize the PCP in the soil matrix and decrease recoveries. It is suggested that
adequate recoveries be demonstrated for PCP before this project is initiated. The
addition of a strong mineral acid to the soil may be necessary to release the PCP during
exiraction.
Response: Tables 3-1 and 5-4 will be changed to indicate that the accuracy objective for PCP
is 14-to 176-percent, as is specified in Table 6 of SW-846 Manual Method 8270. PRC
agrees that the recovery of PCP can be very poor when PCP analysis is performed with
standard EPA SW-846 Manual Method 8270. The poor recoveries of PCP may be
compounded by the addition of sodium carbonate during sample treatment.
To address the concern of low PCP recovery inherent in Method 8270, PRC will be collecting
a sample during the dry-run week (August 16) which has been run through the THERM-O-
DETOX treatment process, and will have had sodium carbonate added to it. This sample will
7
be submitted to Yersar before the start of the demonstration activities. Four aliquots of this
sample will then be spiked with PCP at each of four levels --10, JOO, 1,000, and 2,000 parts
per million, and analyzed using Method 8270. Versar will then perform a statistical analysis
of the matrix spike data including: mean percent recovery, standard deviation, and
determining the 95 percent confidence interval for percent recovery. Versar will transfer this
data to the PRC project manager before the demonstration activities begin. The 95 percent
confidence interval from this evaluation should fall within the range of 14-to 176-percent. If
the 95 percent confidence interval exceeds these ranges PRC will contact the EPA TPM to
determine corrective action.
Corrective actions which may be taken includes acidifying samples and analyzing the acidified
sample extracts by Method 8270. This course of action may reduce the number of
compounds which can be reported due to the possibility of acidic breakdown of some Method
8270 analytes. This action may also limit the TICs data derived from the sample analysis, for
the same reason. Another corrective action may include analysis of san1ples with Method
8270 and a separate analysis of samples with EPA SW-846 Manual Methods 8040 or 8 I 51.
These are gas chromatographic methods specifically designed to determine phenols and
chlorinated acids. The main advantage to the use of these methods is that they should provide
better recoveries for PCP and other chlorinated phenols. Another advantage is that the
quantification of PCP may be more accurate with these methods than with Method 8270.
Disadvantages of the use of these methods are that the number of analytes reported is much
smaller than with Method 8270, and the identification of specific compounds is not as exact as
with Method 8270. Method 8040 and 8151 require second column confirmation, or mass
spectrometer confirmation using Method 8270, for positive identification of compounds.
Corrective actions which will be used will be determined by the EPA TPM and the PRC
project manager.
8
2.0 RESPONSE TO l\UNOR ISSUES
RREL's QA Manager identified nine minor issues regarding the subject QAPP . The PRC SITE
team's response follow.
1. There are also some concerns regarding stack sampling procedures.
la. The estimated stack diameter and flow rate of the stack gas is needed to select the proper
sampling equipment. What are the approximate diameters and gas flow rates at the
various stack sampling locations for this process?
Response : Gas san1ples to be collected during the demonstration include L TTRJMTTR off-
gas, stack gas, and BCD liquid reactor off-gas (designated as sample locations 4, 5, and 8 in
the revised QAPP). The LTTR/MTTR off-gas will be collected after the gas strean1 has
passed through the oil and water scrubbers and condenser, prior to carbon polishing.
Originally, a separate off-gas sample was to be collected for the LTTR and MTTR
(designated as sample locations 4 and 5 in Draft QAPP; see Attachment A for revised sample
location map). The stack gas sample will be collected after the carbon polishing unit, prior to
release of the gas stream to the atmosphere. The BCD liquid reactor off-gas stream will be
sampled as the gas exits the reactor.
The off-gas flow rates of the L TTR/MTTR and the BCD I iquid reactor is estimated at 65
cubic feet per minute (cfm) and 6 cfm, respectively. The diameters of the off-gas/stack gas
pipes are estimated to range between 2 to 4 inches. Radian will finalize specific sampling
procedures during the week of the dry-run.
lb. It is unclear whether gas samples will be analyzed for PCDDs/PCDFs by Method 23
(HRGC/HRMS) or by Method 8280 (HRGC/LRMS). Samples ""ill be collected by
Method 23, and such san1ples are normally analyzed as well by Method 23. However,
Table 5-3 specifies Method 8280, although Table 3-2 gives detection limits from Method
8290, the SW-846 equivalent to Method 23. Clarification is requested. The primary
difference here is that Method 23, which is based on high resolution mass spectrometry,
giYes more definitiYe identification and frequently better detection limits. If Method
8280 is indeed intended for the air samples, the detection limits in Table 3-2 for
PCDD/PCDFs in air should be adjusted accordingly. The QAPP should also explain
what changes, if any, will be needed to analyze Method 23 samples by Method 8280.
Will labelled surrogates be added to the resin bed prior to sample collection?
In Section 4.4.4, page 19, the discussion of the Method 23 train states that separate
trains ~ill be run for PCB, PCDDs, PCDFs, and SVOCs. PCBs have not been _ . ,
previously identified as part of this project. Three sampling trains have been discussed,
one for PCDDs/PCDFs, a second for SVOCs including PCP, and a third for particulates.
Please clarify this discussion. Also, there is a typographical error in the description of
the resin in the same paragraph: "SAD" should be XAD-2.
Response: Meth od 8290 will be used for PCDDs/PCDFs analyses. Table 5-3 of the QAPP
will be changed for consistency wj_t_h Table 3-2 .
9
Also, the listing of a PCB gaseous sample train and the "SAD" designation for "XAD-2" in
Section 4.4.4, page 19, are typographical errors. The text of the QAPP will be revised
appropriately.
le. The minimum sample volume listed for the VOST train (VOCs in gas) is 1.8 liters.
Twenty liters (0.02 dscm) is the typical volume collected and is the volume used to
calculate the detection limits given in Table 3-4. Please clarify.
Response: Samples for VOC analyses will be collected using passivated Summa canisters .
The VOST sample train will not be used to collect a sample for VOC analysis . The text of
the QAPP will be revised to indicate the change in VOC sample collection methods .
Attachment A list the standard operating procedures for the gas chromatography/mass
spectrometry analysis of air toxics canister samples using Method T0-14.
ld. Section 4.7.1 discusses guidelines concerning construction and geometry of a 11standard
pitot tube." Will a standard pitot tube be used for this project; A type 11S11 pitot tube is
usually used for Method 5 sampling unless small ducts or stacks are involved. Please
clarify.
Response: A standard pitot tube will be used for this project. The ducts are too small to a
Type S pi tot. \Vhere possible, ,we will install 4-inch PVC tubing (with no glued joints) with
ports at appropriate distances to comply with the 8 and 2 diameter rule of EPA Method 1.
Gas velocity and temperature will be measured with a standard pilot and thennocouple as
described in EPA Method lA and 2C. It may not be feasible to install a 4-inch diameter
sample line , in which case the smallest possible standard pitot will be used, although the EPA
methods cited do not apply to ducts less than 4-inches in diameter. If this approach is found
to be ineffective, alternate techniques, such as use of a hot wire anemometer or measurement
of total duct gas volume over time, may be attempted.
le. Will the thermal desorbers and the BCD liquid reactor be operated simultaneously? If
so, how can the gas emission at sample point 9 be apportioned between the various
sources?
Response: The them1al desorbers and the BCD liquid reactor will not be operated
simultaneously. The demonstration schedule presented in the QAPP will be revised to
indicate the days of operation for the respective units.
2. Scheduled QC and calibration
2a. In Section 3, page 5 of 14, it is stated that untreated soil (before and after the addition of
sodium bicarbonate) will not be spiked for PCP because the expected level of PCP is
greater than 1000 mg/kg. If matrLx spiking is not used, how will the accuracy of PCP in
this matrix be determined? Accuracy must be detemtined for each critical parameter in
each critical matrix. Please describe the procedure for detemtining the accuracy of PCP
in the feed soil. The next sentence in the same paragraph states that precision of PCP in
feed soil will be assessed by serial dilution. There is no explanation why tltis procedure
is being used over the typical laboratory duplicate. Please explain.
10
Response: Matrix spike samples and field duplicates will be collected for sample locations
and 2. The text of the QAPP will be revised accordingly.
2b. Table 5-4 specifies a 5% frequency for MS/MSDs. This appears to contradict Table 4-3,
which specifies a definite nwnber of MS/MSDs, typically at a higher frequency than 5%.
It is recommended that Table 5-4 be modified to be consistent with Table 4-3.
Response: Table 5-4 of the QAPP will be modified to be consistent with Table 4-3 .
2c. In Table 5-4 corrective action for MS/MSDs is to flag the data. It is recommended that,
at a minimum, the project manager be notified as soon as possible.
Response: The text of the QAPP will be modified to indicate notification of the PRC and
EPA project managers upon initiation of corrective actions for MS/MSDs.
2d. In Table 5-4 the acceptance criterion for initial calibration for dioxins is RSD < 30%. To
what measurement does this apply? The method requires that the RSD of each standard
injected in triplicate be < 15%. Is the intent of this entry to relax this criterion
somewhat? Or does this criterion apply to the linearity of the calibration curve?
Response: Table 5--4 will be changed to indicate that the acceptance criteria for initial
calibration for dioxins is ..::;_ 15 percent, as is specified in EPA SW-846 Manual Method 8280.
2e. In Table 5-4 there is no procedure listed to determine accuracy for total organic carbon
analysis. Section 3 mentions matrix spikes. Please resolve this inconsistency.
Response: Total organic carbon analytical accuracy is assessed through the use of matrix
spike samples. The matrix spike will be composed of potasium hydrogen phthalate.
Acceptable matrix spike recovery control limits generated by Versar are 80-to 120-percent
recovery. This matrix spike recovery acceptance criteria will be added to Table 5-4.
2f. According the Sections 9.1 and 9.2, accuracy for PCDDs/PCDFs will be determined by
MS/MSDs. This is probably incorrect because later in these subsections and in other
sections of the QAPP, other procedures are specified for accuracy and precision for
PCDD/PCDF analyses. Please resolve these inconsistencies.
Response : Sections 9-1 and 9-2 will be changed to indicate that accuracy for PCDDs/PCDFs
will be evaluated through the use of laboratory control standards . The accuracy criteria is 40-
to 120-percent recovery. Accuracy will also be estimated by determining internal standard
recoveries for each congener class.
2g. Page 3.0-4 states that precision for particulate matter will be determined by anal5'zmg ·'
duplicate aliquots of a field sample. A Method 5 sampling train is not appropriate for
aliquoting procedures. Precision of a Method 5 train is not routinely determined. Please
clarify.
Response: Precision for particulate matter will not be determined experimentally during the
demonstration. The reference to_r:Jarticulate matter in the first sentence on page 4 of Section 3
11
should be deleted. Note that footnote "f" to Table 3-1 (relating to paniculate matter) says,
("Precision and accuracy are based on EPA collaborative test." The following sentences will
be inserted on page 4 of Section 3 of the QAPP.
Precision and accuracy for measurements of particulate loading and moisture in gas
streams are not readily measured experimentally in a demonstration test. Adherence
to method protocols, which include performance-related activities such as sampling
equipment calibration, isokinetic sampling, balance calibration, desiccating filters to a
constant weight, etc. is the basis for achieving acceptable method accuracy and
precision.
2h. Footnote "d" to Table 3-1 indicates that the laboratory for gaseous PCDD/PCDF sample
analyses has not been determined. Thls is probably an oversight. Section 2 identifies
three subcontract laboratories; which one will do these analyses?
Response: Gaseous sample analyses for PCDDs/PCDFs will be performed by IT
Corporation. The text of the QAPP will be changed accordingly.
3. Project organization.
3a. According to Figure 2-1, the QA manager is "to be assigned. 11 This person should be
identified in the revised document to be sure that thls f uhction is covered.
Response: The IT QA Coordinator is Mary Tyler. The text of the QAPP will be revised
accordingly.
3b. There is some ambiguity regarding which laboratories will perform whlch analysis. Will
Radian perform all analyses on all gas samples, including dioxins and furans? Will all
dioxin and furan determinations be performed by IT? Clarification is needed to assure
that samples are sent to the correct laboratory.
Response: All gas sample analysis will be perfom1ed by Radian, except those to be analyzed
for PCDDs and PCDFs. IT Corporation will perform PCDD/PCDF analysis on all gas, soil,
water, and oil samples. Versar will perform all listed analysis for soil, water, and oil, except
for samples listed for PCDD/PCDF analysis. Table 3-1 of the QAPP will be revised to
indicate the appropriate lab for each listed analysis.
3c. Figure 2-1 suggests that the QA managers actually report to the project managers and
thus lack essential managerial independence. Please clarify.
Response: Figure 2-1 of the QAPP will be modified to indicate managerial independence of
the QA managers .
3d. Who is responsible for the physical measurements such as particle size distribution and
bulk density? Who is responsible for field measurements such as PCP?
Response: Versar is responsible for all physical measurements on soil samples including
particle size distribution and bulk d_~nsity . As stated in the 3b response, Table 3-1 of the
12
QAPP will be revised to indicate the appropriate lab for each listed analysis. PRC is
responsible for all field measurements, such as PCP in soil and water. Radian will measure
air flows and ETG will perform all process measurement.
4. Please describe the sample numbering system that ·will be employed during this project.
Entering this information in the QAPP helps the sampling team to properly identify
samples.
Response: Samples will be labeled according to sample location numbers (SL#), test run
number (TR#), and composite number (C#). For example, the third composite sample
collected during the fifth test run of the L TTRJMTTR at sample location 3 will be labeled
SL3-TR5-C3. The text of the QAPP will be revised to indicate the sample numbering system.
5. Table 4-4 should be modified to provide for addition of sulfuric acid to pH < 2 as a
preservative for water samples collected for oil and grease.
Response: Table 4-4 of the QAPP will be revised to indicate the addition of sulfuric acid to
pH < 2 for preservation of water samples collected for oil and grease analysis .
6. According to Figure 4-1, the BCD process employs activated carbon beds for cleaning
water and gas before discharge. How will the lifetime of this units be estimated for the
purpose of calculating operating costs? How will disposd.I costs for the these systems be
estimated, considering that they may be badly contaminated?
Response: Capacity and cost for granular activated carbon (GAC) units are readily available
from manufacturers literature. By measuring inputs to both water and air GAC during the
demonstration, an estimation of operating and disposal costs will be detem1ined.
7. There are a number of issues related to the analytical methods listed in Tables 5-1, 5-2,
and 5-3.
7a. First, Table 5-1 should distinguish between aqueous samples and oil samples. The term
liquid includes both oil and water although different preparation procedures are typically
required. Modified Method 3020 is listed for the preparation of liquid samples for
arsenic, lead, and selenium analyses. Methods 7060 and 7740 have preparation
procedures for aqueous samples contained in them. Why is a modified method being
proposed? Method 3020 is acceptable for preparing aqueous samples for lead analysis.
Why is a modified method being proposed? What preparation method will be used for
oil samples?
Response: Table 5-1 will be changed to indicate three types of matrices: (1) water; (2) solid,
sludge, and oil; and (3) gas. Oil samples are treated differently than water samples, an-d the ';
preparation techniques employed for oil samples are usually by solvent dilution for organics,
and by solvent digestion for metals . Oils typically do not have the same density as water (1.0
gram per milliliter). For example, a degreasing oil may contain greater than 1,000,000
milligrams per liter of tetracholorethylene. To obtain more realistic and accurate data, oil
samples will be reported on a weight-to-weight basis.
13
Table 5-1 will be changed lo indicate that water sample preparation methods for arsenic will
be EPA SW-846 Manual Method 7060, and for selenium will be EPA SW-846 Manual
Method 7740 . The aqueous sample preparation method used for the analysis of lead will be
EPA SW-846 Manual Method 3020.
Oil samples will be prepared by EPA SW-846 Manual Method 3040 for metals analysis .
7b. Method 3520 is listed in Table 5-2 for preparing solid samples for SVOC analysis. Was
Method 3550 intended?
Response : The preparation methods used for soil and sludge sample preparation will include
both EPA SW-846 Manual Methods 3540 and 3550. The analytical laboratory performing
sample analysis will be given the option of choosing the preparation method, and will be
responsible for verifying the methods applicability to sample matrices encountered during this
demonstration.
7c. Modified Method 3050 is listed for arsenic, lead, and seleniwn. Footnote "d" indicates
that the modification involves the addition of hydrogen peroxide. ivf ethod 3050 already
includes the use of peroxide. Please explain.
Response: Table 5-2 will be changed to read Method 3050 and footnote "d" will be removed.
7d. Methods 413.2 and 9252 are listed for preparing solid samples in Tahle 5-2 for oil and
grease and chloride analyses. These methods are for water samples. Please supply
appropriate methods for preparing solid samples.
Response: The method used for the analysis of oil and grease in soil, sludge, and oil samples
will follow the same procedures that are listed in EPA SW-846 Manual Method 9071. This
method is applicable to sludge samples, and Versar has indicated that they follow the sample
preparation and analysis procedures for soils, sludge, and oil. Table 5-2 will be changed to
show that Method 9071 will be followed for soil, sludge, and oil analysis of oil and grease.
The method used for the analysis of chloride in soil, sludge, and oil will involve the
separation and transfer of chloride ions from the soil, sludge, and oil into water. Once the
chloride ions have been exchanged to water the remainder of the analysis steps are identical to
those listed in EPA SW-846 Manual Method 9252 . PRC and Versar have not been able to
identify a specific EPA-approved method performing soil, sludge, and oil sample analysis for
chloride. The method which Versar will use will be referred to as Modified Method 9252.
The standard operating procedure for Modified Method 9252 will be provided to EPA by
Versar. Table 5-2 will be changed to indicate that Modified Method 9252 will be used for
preparation and analysis of chloride samples . . ,
8. According to Table 1-9 (schedule) only one week is allowed for shakedown and process
optimization. Will this be enough time for optimization, including completion and
interpretation of analytical results?
14
Response: The schedule of activities to be presented in the revised QAPP is as follows:
Week of August 9
Week of August 16
Week of August 23
Week of August 30
Week of September 6
Equipment mobilization and soil
delineation , excavation, and
preparation
Dry run of BCD process equipment
Three test runs using the
LTTR/MTTR, one test run using the
L TR, and the visitors day (tentative)
Three test runs using the
L TTR/MTTR, and one test run using
the LTR
Equipment demobilization
9. Data reduction and reporting. Who will store raw data and samples, and for how long?
Providing a checklist of forms that will be required in the analytical report for each
analysis will expedite the report preparation procedure. Will the investigators require
data on magnetic media?
Response: Samples will be stored by the respective analytical labs for at least 4 weeks after
the submittal of data analysis reports. The data analysis reports will be stored for no less than
1 year.
15
3.0 CHANGES RECOMMEl\'DED BY SITE TEAM lY1EMBERS
The SITE team QA managers identified changes to the QAPP that are listed below.
3.1 CHANGES RECO1\1MENDED BY RADIAN CORPORATION
1 a. After the first sentence in the second paragraph of Section 3 .1, insert:
For gas samples, duplicate media spiked samples (spiked blank sorbent resins), rather than
matrix spiked samples, will be used for semivolatile and volatile organics. Matrix spikes of
impinger solution aliquots will be used for metals and HCI in gas samples .
lb. To the end of the last paragraph on page 1 of Section 3.1, add:
Precision for PCOD/PCOF analysis in gas samples will be estimated as the relative standard
deviation for recoveries of the pre-sampling surrogate spike in the replicate test runs.
le. Change the first sentence of the last paragraph on page 4 of Section 3 .1 to read:
Accuracy for SVOCs, VOCs, metals, chloride, HCl/chlorine, and TOC will be ~stimated as
percent recovery of the true analyte from matrix spike samples (media spike samples for
SVOC and VOCs in gas samples.) [Note: the change is shown in parenthesis.]
ld. To the second paragraph on page five of Section 3.1, add:
Accuracy for PCOO/PCOF in gas samples will be estimated from pre-sampling surrogate
spike recoveries .
2. Change precision objective for met3.ls in gas samples listed in Table 3-1 to 20 RPD and
recovery objective of 75-125 % . Also, change precision for HCl/chlorine in gas samples to 20
RPO.
3. Change the precision objective for metals trains in Table 3-6 to 20 RPO.
4. Insert the following to replace Section 4.4.5.
The LTTR/MTTR and LTR will be run separately; thus, the resulting stack gas from each of
these two processes will be sampled separately. The reactor off-gases will be sampled at he
carbon bed stack for PCDDs/PCDFS, particulates, moisture, SVOCs, VOCS, and HCI/Cl2.
These samples will be collected through ports installed in the carbon bed exit line. The exit
line from the carbon bed will be modified to accept 4-inch PVC tubing. The tubing will theri;
be routed to facilitate sampling . Ports will be installed at appropriate distances to comply
with the 8 and 2 diameter rule of EPA Method 1. Gas velocity and temperature will be
measured with a standard pitot and thermocouple as specified in EPA Method lA. The gas
samples will be collected by either standard U .S. EPA stack sampling methods as published in
40 CFR Part 60 (Standards for Performance for New Stationary Sources, Appendix A,
16
Reference Methods) or methods outlined in the Methods Manual for Compliance with the BIF
Regulations. The sampling trains to be used during the test are briefly described below.
Samples for SVOCs will be collected by a modified Method 5 (MM5) sampling train, Method
0010 in SW-846. Samples for PCDDs/PCDFs will be collected by M23. The MM5 and
M23 train are shown schematically in Figure . The sampling approach and requirements
for both MM5 and M23 are identical. Differences exist in the preparation and recovery of
the trains . Method 23 has more stringent requirements for preparing the filter and XAD resin
than does Method 0010.
• · A heated, three-foot borosilicate glass-lined sampling probe will be inserted
into the duct to withdraw the sample . The probe will be equipped with a
stainless steel (316) nozzle, sized to withdraw the sample isokinetically.
• Gas will pass from the probe to a borosilicate filter holder containing a 90
mm Reeves-Angel 934 A-H glass fiber filter (or equivalent). The filter
assembly will be enclosed in a heated chamber capable of maintaining a
temperature of 120 °C.
• The filter will be followed by an impinger train consisting of a spiral-type, ice
water-cooled condenser; an ice water-jacked sorbent module containing
approximately 20 grams of 30/60 mesh XAD-2 resin (pre-extracted); a
temperature sensor; a 1-liter condensate trap; two standard Greenburg-Smith
impingers; each containing approximately 100 mls of deioni zed water and a
final modified Greenburg-Smith impinger containing silica gel plus a
thermocouple to detect sample gas exit temperatures. The impinger train is
contained within an ice bath.
• A vacuum line will connect the outlet of the impinger train 10 a control
module consisting of a vacuum pump, a calibrated dry gas meter, a calibrated
orifice and an inclined manometer.
For each test run, the following material will be recovered from the MM5 and the M23
sampling trains as described in Method 0010 and M23; rinsate from the sample probe and the
front half of the filter holder, the filter, the XAD-2 resin sorbent module, liquids collected in
the I-liter condensate trap, and rinsate from components between the back half of the filter
holder and the condensate trap. M23 also includes a final toluene rinse . The recovered
material will be extracted and analyzed for SYOCs, or dioxins and furans. Each sampling
train will generate several samples , however, the extracts of each sample will be combined to
produce one final sample per train.
The BIF Method 0050 sampling train will be used to measure particulate HCl/Cl:1 and
moisture content of the gas stream. The BIF Method 0050 train is shown schematicall y on
Figure _. This sampling is similar to the MM5 /M23 train described above, with the
following exceptions:
• The water cooled condenser and sorbent module are removed .
17
• The impinger train will consist of five impingers. The first is empty. The
second and third will contain 100 ml of 0 .1 N sulfuric acid. The fourth
impinger will contain 100 ml of 0.1 N sodium hydroxide. The last impinger
will contain silica gel as a final water trap.
Volatile organic compounds will be collected using passivated Summa canisters for analysis
by GC-MS following the general guidelines of U.S. EPA Compendium Method TO-14.
5. Insert the following changes in Table 4-4.
Dioxins and Furans, Gas Sample Volume -3 dscm not 10 dscm
HCI/Cl2, Gas Sample Volume -1 dscm, not 30 dscm
Semivolatile Organics, Gas Sample Volume -3 dscm, not 10 dscm
Metals, Gas Sample Volume -3 dscm, not 30 dscm
Moisture, Gas Sample Volume -1 dscm, not 20 dscm
Particulate, Gas Sample Volume -1 dscm, not 30 dscm
6a. Modify the first sentence of paragraph one in Section 4.7.3 as follows:
Incline manometers or magnehelic gauges will be used during this project to measure
differential and static pressures.
6b. Delete the second sentence of the paragraph.
7. IntoSection4.7.6:
• Delete "(NBS Class S traceable)".
• Last senrence of section, change ''± 0.1 gram" to + 0.5 gram.
8. Insert the following changes to Table 5-3 :
• For PCDDs/PCDFs, change "8280" to "8290".
• For Volatile Organics, change Analvsis Method from "5040" to "8240".
• For Metals Analyses, Preparation Methods are as follows:
Arsenic SW-846 3020
Barium SW-846 3005
Cadmium SW-846 3005
Chromium SW-846 3005
Copper SW-846 3005
Lead SW-846 7421
Mercury SW-846 7470
Nickel SW-846 3005
Selenium SW-846 3020
Silver SW-846 3005
18
. ,
Sodium
Zinc
SW-846 3005
SW-846-3005
• Preparation and analysis for Chloride (HCl/Cl:J in stack gas should be given as BIF
0050.
3.2 CHANGES RECOl\1MEl\TDED BY VERSAR, INC.
1. Samples for the following parameters may be collected together in the same sample container.
The volume listed is the total amount required for all the parameters. Table 4-4 will be
modified to reflect the combined samples .
• Solids:
•
Total metals + TCLP metals
Chloride + moisture + Oil & Grease + pH
PSD + density
Liquids:
Chloride + TSS + pH
• Oils:
If oily or sludge-like, treat as solid:
Chloride + Oil & Grease + Moisture Content + pH
If low-viscosity liquid, especially if aqueous, treat as a liquid sample:
Chloride + pH
Oil & Grease separate
2. The following sample volume requirements should be added to Table 4-4 .
19
500 ml
1000 ml
1000 ml
500 ml
1000 ml
500 ml
1000 ml
Minimum Sample
Volume
Parameter Media
TCLP (Metals) Solid 500 g
for total & TCLP
metals, collected
together
TCLP (VOCs) 500 g
TCLP (SVOCs) 500 g
Container
Wide Mouth G
Teflon-lined cap
Teflon-lined cap
Maximum
Preservative Holding Time
Cool ~ 4 °C Analyze mercury in
28 days; for other
metals: 6 mos. to
TCLP extraction, 6
mo . to analysis
For VOCs: 14 days
to TCLP extraction;
14 days to analysis
For SVOCs: 7 days
to TCLP extraction;
7 days to
preparation
extraction; 40 days
to analysis
3. The analytical method for bulk density listed in Table 5-4 will be changed to ASTM 2974.
3.3 CHANGES RECOM!\1ENDED BY INTERNATIONAL TECHNOLOGY
CORPORATION
1. In Table 7-2, the SAL uses 13C-1,2,3,6,7,8-HxCDD as an internal standard instead of 13C-
l,2,3,4,7,8-HxCDD as stated in Table 7-2. Also, 13C-OCDF is not used as an internal
standard.
2 . The retention time window check solution is analyzed at the beginning and end of each 12
hour shift (Table 5-14).
3. Under acceptance criteria in Table 5-4 for the calibration standard, the term "for native
compounds" will be added .
4. In Table 3-2, the target detection limits for gas samples will be changed to the following:
Compound
2,3,7,8-TCDD
Total TCDD
Total PeCDD
Total HxCDD
Total HpCDD
OCDD
Detection Limit (ng/dscm)
0.003
0.003
0.016
0.016
0.016
0.030
20
2,3,7,8-TCDF 0.003
Total TCDF 0.003
Total PeCDF 0 .016
Total HxCDF 0.016
Total HpCDF 0.016
OCDF 0 .030
5. Method SW-846 8290 will be added to Table 5-4 . The following changes are listed on the
following page ..
21
SUMMARY OF CALIBRATION AND INTERNAL QUALITY REQUIREMENTS Parameter Analytical Calibration Frequency Acceptance Criteria Corrective Action Method PCCDS and PCDFs SW-846-82 90 Initial calibration Before sample Relative standard 1) Repeat initial calibration multipoint, 5 levels analysis begins and deviation (RSD) Native 2) If still unacceptable, make when calibration compounds -20% necessary adjustments check standard Labeled compounds-30% 3) Repeat initial calibration TCDD Beginning of 12 25 % valley between Adjust GC conditions or chromatography hour shift 2,3,7,8-TCDD and replace column check closest eluting TCDD isomer PCDD/PCDF Beginning of 12 MID switch points Rerun retention time hour shift properly located standard, adjust selected ion monitoring windows Calibration check Beginning and end Agreement with 20 % of 1) Rerun calibration check of 12 hour shift predicted value from 2) Repeat initial calibration initial calibration for native compounds, 30% of predicted value from initial calibration for labeled compounds Method blank 5 % of l per batch <I= PQL 1) Reanalysis if blank if problem still exists flag data associated with blank Surrogate spike Every sample 40-135% 1) Evaluate system recovery 2) Flag data 22 . '
ATTACHMENT A
REVISED SAMPLE LOCATION MAP
CONT .W IHA TU)
SOIL
SCR£Di
CD
SCRITNED
COlfTAMIHA TU)
SOIL
IJ(X[R
0
SOIL
STOC><Pll.£
VAPOR OtSCHARGES
VAPOR O!SCW,RC€S
LOW TOI P CR;.. TUR£
Tl-l ER\.W.. 0£50fiPTION
fITD(R
OISO-WIO(
~
\:::/ \./
9CO SOUOS RV,CTOR
l,l[[)t\Jl,/ n:w. T}jER\.W._
OCSOR!7110N
COOUHC SCRDI'
~
WATER SPRAY
0
ON-SITT: f3,.l,C)(flU
OR ------l
OfT -sm: DISPOSAL
LEGEND
DECO!fT>,J.JiHATED SOI\.
CONT ).JN[R OR
STOCKP\L(
@-Sample Location Number
TR£ATU)
WATER
TO
l)(SCK.\.RC£
SUPPLDJ (NT
0
COt<OOl30\
BCD
UOUIO
R(>.CTOR
OCCHLORJM TlOt-1
R(AC{J,'T,
/
(j)
ruo.. ------, Tl<(ATU)
ClU./K >J-10 DISPOSAL
KOPPE[1S COMPANY, INC.
MORRISVILLE, NORTH CAROLINA
FIGURE 4-1
BCD PROCESS FLOW DIAGRAM
SAMPLE LOCATIONS
PIU: ENVIRONMENTAL MANA GEMENT, IN C.
ATTACHMENT B
ST AND ARD OPERA TING PROCEDURE FOR THE
GAS CHROMATOGRAPY/MASS SPECTROMETRY ANALYSIS
OF AIR TOXICS CAl'HSTER SA.ivfPLES USING
METHOD T0-14
The following is excerpted from Radian SOP 327-MS-038 which ia e. RadilUl Corporation Confidential Reaearch and
Engineering Document.
1.0
STANDARD OPERATING PROCEDURE FOR THE GAS CHROMATOGRAPHY/MASS
SPECTROMETRY ANALYSIS OF AIR TOXICS CANISTER SAMPLES
USING METHOD TO-14
PURPOSE
This standard operating procedure provides a method for the analysis of canister air samples by gas
chromatography, using the method and analyte list from EPA Compendium Method TO-14 with a
Finnigan 4500 GC/MS/DS in either multiple ion detection (MID) mode or the full scan mode in
conjunction with the sample interface system built by Radian Corporation. This analytical method
can be used to quantify many volatile organic compounds with boiling points less than 200°C.
Analytical properties of polar compounds in canisters are not well defined currently. The GC/MS
analysis of gas samples from canisters in restricted to use by, or under the supervision of, analysts
experienced in the use of GC/MS systems, and skilled in the interpretation of mass spectra and their
use as quantitative tools.
2.0 SCOPE AND APPLICABILITY
This method may be applied to the analysis of air toxics samples which are introduced to the GC/MS
using the Radian sample interface system with the method and analyte list from EPA Compendium
Method TO-14. The method is applicable to ambient air, indoor air, to landfill gas, and to any air
samples with volatile organic analytes present below hundreds of ppbv. The method is not applicable
directly to samples from combustion processes. A modification of the method to allow direct gaseous
injection may be performed to analyze air samples which contain high levels of organic compounds.
3.0 METHOD SUMMARY
Field samples are received at the Radian laboratory in canisters. These samples are logged into the
Radian Sample and Analysis Management (SAM) system and scheduled for analysis in the GC/MS
laboratory. An analytical mode, either full scan for maximum nexibility or MID for maximum
sensiti vity, is selected and appropriate gaseous calibration standards are prepared. The instrument
is tuned and the range of calibration standards is analyzed. A calibration curve is obtained by linear
regression, with a goal of a correlation coefficient of 0.995 for each analyte. After the calibration
samples have been analyzed, a zero air sample containing pre-purified humidified air is analyzed.
Then, the samples are analyzed. Quantitie·s of the analytes in the sample are calculated from the
linear regression and reported through the SAM system. If qualitative/semi-quantitative analysis of
non-target analytes is desired, enhanced mass spectra obtained in the full scan mode are searched
against the 42,222-compound NIH/EPA reference library, and the data are interpreted by an
experienced mass spectroscopist. Semi-quantitative calculations are performed by comparison of the
signal level for a tentatively identified compound (TIC) to a compound in the sample with a
concentration which is accurately known.
4.0 INTERFERENCES
This analytical method is based on the identification of mass spectra using the appearance of the
spectra, comparison to reference spectra, and (for the target analytes) comparison to accurately
known retention time. Any elements of the gas matrix which might interfere with the ability to
identify the mass spectra, obtain accurate peak areas, or obtain an accurate retention time which can
be compared to a reference standard will affect the performance of the analysis. If intense
interfering peaks are encountered, the chromatography will be distorted. The only possible solution
to the problem of excessive compound loading is dilution of the sample or analysis of a smaller
volume. If a coeluting compound is encountered, the mass spectrum generally can be interpreted
unless the coeluting compound is an isomer of the compound of interest and the masses are the same
or approximately the same. In this case, within the limits of the present method, the problem cannot
be resolved. The analysis of blanks will demonstrate that the analytical system is free from
interferences and a field blank can serve as a check on the occurrence on contamination in the
sampling and handling procedures. The laboratory where volatile analysis occurs is completely free
of solvents.
5.0 MATERIALS AND APPARATUS
5.1 Calibration Standards: supplied in a canister ready for use on the
M • : ;
analytical system. Calibration standards are prepared in the Radian
Volatile Organic Compound (VOC) laboratory. Standard canister
numbers, analyte concentrations, source of supply, name of preparer,
date prepared, and date received are supplied to the GC/MS laboratory
with the standards.
2
5.2
5.3
5.4
5.4.1
5.4.2
5.4.3
5.4.4
5.4 .5
Radian Sample Interface Svstem: designed and built by Radian to
interface between the sample contained in the canister and the
analytical system.
Bubble Flow Meter
Gas Chromatograph/Mass Spectrometer System
Gas Chromatograph: an analytical system complete with
temperature-programmable gas chromatograph with sub-
ambient capabilities and all required accessories, including
gases, analytical columns, and separator.
Chromatographic Column: DB-624 megabore fused silica
capillary or equivalent, 30 m in length.
Mass Spectrometer: capable of scanning from 15-350 amu
every I sec or less, using 70 volts (nominal) electron energy in
the electron ionization mode and producing a mass spectrum
that meets all criteria for the manufacturer's specifications for
perCTuorotributylamine (PITBA) tuning.
Glass Jet Separator: to accept a flow rate of approximately 15
mL/min from the mega bore column and 15 mL/min of makeup
gas while allowing the mass spectrometer to operate at a
vacuum of l O.s torr or better.
Data System: to allow for the continuous acquisition and
storage on machine-readable media of mass spectra or MID
data obtained throughout the duration of the chromatographic
program, interfaced to the MS. The computer must have
software that allows searching any GC/MS data fit~ for id~s
of a specified mass and performing quantitative calculations
against the linear regression generated b y the calibration of
the instrument. The software must allow integration of the ion
abundances at a specified mass between a selected time or scan
number limits. The data system also must include the most
3
6.0
recent version of the EPA/NIH mass spectral reference library
for identification of mass spectra.
CHEMICALS AND REAGENTS
Analytical standards are prepared in the VOC laboratory in canisters at the approximate dilution, so
no chemicals outside the canisters are handled in the MS laboratory.
7.0 SAFETY
Standard safety procedures for the mass spectrometry laboratory require that safety glasses with side
shields be worn at all times. The use of equipment, the canisters, and the interface requires attention
to electrical hazards, as well as sharp, hot surfaces. No analyst is allowed to operate the
instrumentation without appropriate training and an SOP for the use of the Radian interface is
available (Radian SOP 262-A T-001 ).
4
8.0 PROCEDURE
8.1 Preparation of the Gas Chromatograph
8.2
8.1.1 Install the chromatographic column: DB-624 megabore (0 .53
mm) fused silica capillary column or equivalent, 30 min length.
8.1.2
8.1.3
8.1.4
8.1.5
8.1.6
Set the column helium carrier gas flow to 15 mL/min.
Set the flow of the helium make-up gas into the glass jet
separator at 15 mL/min.
Set the temperature of the separator oven to 220°C.
Set the temperature of the injector port to 105°C.
Set the program of the GC to hold at 10°C for five min, then
program to 200°C at 6°C/min, and hold at 200°C until peak
elution ceases.
Preparation of the Interface
Connect the interface exit fitting directly into the carrier gas inlet to
the chromatographic column after verifying helium flow through the
interface with a bubble flow meter.
5
8.3 Preparation of the Mass Spectrometer
8.3.l Full Scan Mode -If the full scan mode is selected for analysis,
the optimum sensitivity which can be obtained with MID is
sacrificed but the full potential of the MS as a universal
detector is realized.
8.3.2
8.3.l.1 Zero the MS according to Radian SOP 327-MS-038.
8.3.J.2 Tune the MS according to Radian SOP 327-MS-038.
8.3.l.3 Set scan parameters according to Radian SOP 327-MS-038.
8.3.l.4 Cool the MS according to Radian SOP 327-MS-038.
8.3.l.5 Introduce the sample to the MS according to Radian SOP 327-
MS-038.
8.3 .l.6 Initiate scan cycle and GC program according to Radian
SOP 327-MS-038.
Multiple Ion Detection (MID) Mode
In the Multiple Ion Detection (MID) mode, the mass
spectrometer is restricted to monitor only previously selected
masses rather than scanning a mass range. Sensitivity is
enhanced over the full scan mode, since no time is spent
monitoring masses which are not specifically of interest.
However, only information specifically requested is obtained
when the mass spectrometer is operated in the MID mode, so
use of the MS as a universal detector is sacrificed for the
additional sensitivity.
8.3.2.l to 8.3.2.8 of Radian SOP 327-MS-038 detail the set up
procedures for MID mode and should be followed as detailed
there.
6
8.4 Initial Calibration
8.4.1 An initial calibration curve, either three-or five-point, must be
prepared from canister standards provided by the VOC laboratory.
The analyte list currently used in the T0-14 analysis currently includes
the following compounds:
1,3-butadiene
chloromethane
bromomethane
methylene chloride
I, 1-dichloroethane
bromochloromethane
I, I, I -trichloroethane
1,2-dichloroethane
trichloroethene
bromodichloromethane
toluene
cis-l ,3-d ichloropropene
tetrach loroethene
chlorobenzene
m-xylene
a-xylene
l, 1,2,2-tetrachloroethane
p-dichlorobenzene
I, l -dichloroethene
cis-1,2-dichloroethene
7
vinyl chloride
propylene
chloroethane
I rans-1,2-dichloroethene
chloroprene
chloroform
carbon tetrachloride
benzene
1,2-dichloropropane
/rans-1,3-dichloropropene
n-octane
I, 1,2-trichloroethane
dibromochloromethane
ethyl benzene
p-xylene
bromoform
m-d ich lorobenzene
o-d ichlorobenzene
l ,2-dibromomethane
Halocarbon 11
Halocarbon 113
Halocarbon 12
acrylonitrile
ethyl toluene
1,2,4-trimethylbenzene
hexachloro-1,3-butadiene
Halocarbon I I 4
acetonitrile
chloromethylbenzene
1,2,4-trichlorobenzene
1,3,5-trimethyl benzene
8.4.2
8.4.3
8.5 Daily Analysis
The data for the calibration samples are
entered into a response factor database, and a
linear regression is performed upon the
calibration data points.
A correlation coefficient of at least 0.995 is the
goal for every compound in the database. If
the correlation coefficient for a given
compound is below this value, additional
samples may be analyzed until the desired
correlation coefficient is obtained.
The following steps shall be taken in accordance with Radian SOP 327-MS-038 where the
appropriate section of the SOP is indicated in parentheses: daily calibration check (8 .5. 1 ), linear
regression (8.5.2), calculate percent recoveries (8.5.3), analyze zero air (8.5.4), quantify analytical
sample (8.5.5), optional sample dilution (8.5.6), analyte identification verification (8.5.7). A flow
chart for these steps is given in the cited SOP as Figure 2.
9.0 DAT A HANDLING AND CALCULATIONS
9 .1 Preparation of the Project Data File
All data associated with instrument tuning, calibration, and analyses are compiled into a project
file. The project file included the following information:
2)
Canister Information: The chain of custody sheets for the canister and any
other notes or comments relating to the canisters becomes a part of the MS
lab data file.
Program Instructions: A copy of the project instructions or QAPjP wh_ich
• ~ • I
will include any non-routine aspects of the analysis such as analytes which
are not on the method target list and/or use of qualitative or semi-
quantitative analysis for non-target analytes will be a portion of the program
instructions. The program instructions also will describe special analytical
procedures and deliverables.
8
3) Library and Standards Information: For every analytical procedure, the
GC/MS analyst must create a library. A check of the library against the
target list for the assay will verify that all targets are included in the
automated procedures. Analytical standards are supplied in canisters
prepared by the VOC laboratory. A data sheet is supplied with each canister
that lists the compounds which are components of the standard, the original
concentrations of all these compounds, and the concentrations of all the
components in the diluted sample, calculated both in the units of ppbv and
ug/m3•
4) Tuning Data: The instrument is required to demonstrate that tuning criteria
have been met prior to the initiation of any analysis. The tuning data for the
MS include a mass spectrum of PITBA, the tuning standard, or any other
compound specified as a tuning standard, and a list of the tabulated masses
of this compound.
5)
6)
Calibration Data: For each of the analyses of the three-or five-point
calibration, a chromatogram and the output of the data system listing area of
peaks for each of the target analytes are included. The inclusion of the raw
data for the calibration will allow a reviewer or another analyst to reconstruct
the linear regression. An additional component of the calibration data is a
table of the correlation coefficients obtained for each of the analytes.
Zero Data: For each of the purified, humidified air samples analyzed, a
chromatogram and the output of the data system are included in the project
file. If any analytes are identified in these zero air samples, a mass spectrum
and a calculated quantity are supplied.
7) Sample Data: For each of the canister samples, an enhanced mass spectrum
is included in the data folder for each analyte identified. A chromatogram
for each sample is also included, and the hardcopy from the data system
which includes quantitation peaks for the analytes identified, a peak area for
the major ion, and a retention time.
8) Calculation Sheet: Quantitative calculations for the target analytes identified
are performed by the computer using the linear regression generated in the
analysis of the calibration samples. If any additional calculations are
necessary (e .g., unit conversion), a calculation sheet is included to
9
9.2
demonstrate how the calculations are performed and the results written
directly onto the computer data sheets.
Preparation of the Project Archive
All data are archived on streamer tapes in the laboratory for at lea.st five years. The tape numbers
are recorded in the analysis logbook so that the correct streamer tape can be retrieved from the
analysis data and the sample data acquisition file number. The actual files listed on the streamer
tape are recorded in the tape logbook which is maintained for each instrument in the GC/MS
laboratory. An archivial copy of a data package is retained for at least five years in the laboratory
data archive.
10
9.3 Analysis Log Book
Each analysis performed on each instrument in the GC/MS laboratory is logged into an analysis
logbook for each instrument which contains the following information: file name, date, GC
parameters, GC column, sample identifier, SAM number, method, and analyst. The specifics of the
information to be recorded are listed in Section IO of Radian SOP 327-MS-038.
10.0 QUALITY CONTROL
I 0.1 Detection Limits
Detection limits for air toxics canister analysis have been experimentally determined using the EPA
method of seven replicates at 99% confidence level. The detection limits, determined according to
Radian SOP 327-MS-027, are listed in Section 11 of Radian SOP 327-MS-038.
Method precision, as defined by the reproducibility of replicate measurements has been determined.
Accura"cy as determined by the analysis of external audit samples has been determined by the
accumulation of a historical database for audit samples.
10.2 Method Quality Control Checks
The quality control checks listed below are used to assure the production of data of known quality.
I 0.2.1
10.2.2
10.2.3
Calibration Criteria: A three-or five-point calibration is performed
initially for the method and a linear regression is performed on the
calibration data points. A correlation coefficient of 0.995 or better
must be obtained for each of the target analytes. If the correlation
coefficient is less than 0.995 for the analytes, additional calibration
samples are analyzed until the criterion is met.
Tuning Criteria: The instrument is tuned according to the
manufacturer's criteria for perfluorotributylamine, with masses f3'1
and 2 I 9 at 30% of mass 69, the base peak. If these criteria are not
met, the instrument is tuned until the criteria are met.
Analvsis of Zero Air Samples: No target analytes should be observed
in the zero air samples at levels above the method detection limits. If
I I
10.2.4
target analytes are observed, a new zero air sample is analyzed until
a clean blank is obtained.
Daily Calibration Check: A daily calibration check at the mid-point
of the calibration range is performed prior to the initiation of any
quantitative analysis. The amounts for the calibration check are
generated in the data system by linear regression. Percent recoveries
are calculated for the calibration curve. For the linear regression
obtained from the original calibration data to be valid, the percent
recoveries must be within 30 percent of the initial curve. If this
criterion cannot be met, an additional daily calibration check sample
is analyzed. If the criterion still is not met, a new multi-point
calibration is generated.
12
DOCUMENTATION
The analyst will report the following information as specified in Section 11 of Radian SOP 327-
MS-038: canister information, program instructions, library and standards information, sample
dilution, tuning data, calibration data, zero air sample data, sample data, calculation sheets, .
reference spectra, appropriate sections of the instrument logbook, and a tabulation of instrument
conditions.
The SAM work order number is used as the data file reference number if the sample is logged into
SAM. All data a re recorded on hardcopy from a computer or in black ink. Corrections should be
made by crossing through the original entries with a single line so the original entry is not obscured,
and initialling the change.
All instrument maintenance is documented in a maintenance log kept next to the instrument. Blank
sample canister numbers are documented in the analysis log and on all chromatograms and
analytical data.
I 3