AQ_F_0900009_20180406_ST_RvwMemo (4)

Division of Air Quality
April 6, 2018
MEMORANDUM
To: Mike Abraczinskas, Director
Heather Carter, Fayetteville Regional Supervisor
From: Gary L.Saunders,Stationary Source Compliance Branch
Subject: The Chemours Company—Fayetteville Works
Fayetteville, Bladen County, North Carolina
Facility ID. No.0900009, Permit No.03735T43
Performance Testing for HFPO Dimer Acid (GenX)Conducted on January 22-25,2018 at
Vinyl Ethers North (Division Stack)and Polymer Processing Aid (PPA)Stacks by Weston
Solutions, Inc.
Tracking No. 2018-083ST
Summary of GenX Test Program
Sources Tested
During the week of January 22, 2018,emissions testing was conducted on two process area stacks.The
first series of test runs was conducted on the Division Stack located at the Vinyl Ethers North(VEN)
process area.The Division Stack is a common stack through which emissions from VEN Waste Gas
Scrubber(WGS)are combined with fugitive emissions collected from the enclosed areas of VEN
(sometimes referred to as"room air"). Gases from various reaction vessels and unit operations in the
hexafluoropropylene oxide (HFPO)and VEN processes are vented through the WGS,a caustic scrubber
which reduces emissions of GenX(HFPO Dimer Acid) and its precursors(HFPO Dimer Acid Fluoride). In
the HFPO process,oxygen is reacted with hexafluoropropylene (HFP)to produce HFPO,a key
intermediate in production of the perfluorinated vinyl ethers produced in the VEN process. During the
testing,the HFPO process was operating and the VEN process are was producing perfluoropropyl vinyl
ether(PPVE).
The second process area tested was the Polymer Processing Aid (PPA)area. HFPO dimer acid fluoride
(HFPO DAF) produced in the VEN process is used to produce HFPO dimer acid (also known as GenX).
There are a number of products that may be produced at PPA depending upon the raw materials and
the final product needs. Process gases pass through the PPA scrubber which is a pH controlled packed
bed scrubber. Fugitive emissions from enclosed areas of the PPA process(i.e. room air)are also vented
through the PPA stack though,similar to the Division Stack at VEN, it does not pass through the
scrubber.
Sampling Method
Testing was conducted using a modified EPA Method 0010 found in the SW-846 compendium of Test
Methods for Evaluating Solid Waste:Physical/Chemical Methods. This sampling train is a variation of the
EPA Reference Method 5 found in 40 CFR 60,Appendix A. The Method 0010 train extracts a sample
isokinetically from the gas stream, passes the sample through a temperature-controlled filter,through a

temperature-controlled condenser and into a series of XAD-2 resin "traps"and impingers to capture and
collect the materials that passed through the filter. The test method is designed to capture certain
particulate and condensable materials for later recovery and analysis.
This test method relies upon three actions to capture a,sample for analysis. Filterable particulate at
specified temperatures is captured on the initial filter.Cooled gases with an affinity for the resin are
absorbed and can later be extracted from the resin for quantification. Finally,the impingers capture the
remaining condensable components. These condensates are recovered for later quantification.
Several modifications to the sampling train were recommended and implemented. First,the filter
temperature was held at a lower temperature than the normally recommended temperature of 248°F, ±
25°F. To avoid possible thermal degradation of any materials entering the sampling train,the
temperature was held just high enough to prevent condensation in the probe or the filter. Second,two
resin traps were used instead of the single trap as recommended by the methodology. The initial trap
located immediately after the condenser was retained. A second trap was inserted between the
impingers and the silica gel dryer impinger. This second trap was added to provide an indication of
"breakthrough"of the train, if it occurred. Finally, it was recommended that both impingers utilize a
Smith-Greenberg style tip to increase contact between the gas passing through the train and the water
in the impingers.
Method 0010 recommends a three-hour sampling time per run with a nominal sample volume of one
dry standard cubic meter(dscm) per hour to obtain approximately three dscm per run. This run
time/volume is designed to provide adequate sample capture and recovery for low concentration gas
streams. As part of the sampling program, an analysis protocol was developed to determine the
emission rate of hexafluoropropylene oxide (HFPO)dimer acid (GenX). Emissions captured by the
sampling train may include HFPO DAF,GenX, and the ammonium salt HFPO dimer acid. The HFPO DAF is
used to form GenX. The combination of sample collection and processing cannot distinguish the original
form of these compounds entering the sampling train, only the final form as GenX. The results of testing
are reported as GenX.
The sample recovery and processing extracts GenX through a standardized process from the filter and
probe,the first XAD-2 resin trap,the impinger condensates,and the final XAD-2 "breakthrough" resin
trap. The sample processing is designed to concentrate the captured materials through a known series
of steps prior to analysis. During sample recovery,the analytes are spiked with a known quantity of an
isotopic version of the analyte under investigation to determined recovery percentage,to determine if
the recovery is within data quality objectives(DQOs),and to correct the indicated quantity of the
analyte by the appropriate recovery factor(s).
After sample recovery,the samples were sent to Chemours' contractor,Test America's laboratory in
Denver,Colorado. GenX was extracted from the resin traps. The DAQ required split samples after
extraction to be submitted to the DAQ for independent analysis. The split samples were sent to the EPA
laboratory in Athens,Georgia. This summary of results only addresses the results provided by Test
America for Chemours. The procedure for processing and analyzing the samples was shared between
labs. A known volume(aliquot)of each sample was passed through a liquid chromatography(LC)
column and then into a dual mass spectro mete r(MS). The analysis process is designated as LC/MS/MS.
The output of the mass spectrometer provides quantification of the analyte in question (GenX) and the
isotopic version of GenX.

The analytical equipment was calibrated with a series of known concentrations of GenX to produce a
calibration curve for the instrumentation. Ideally,the sample recovery and concentration techniques
place the analyte concentration in the sample aliquot in the middle of the instrument calibration curve
and allow the instrumentation to measure both the analyte of concern and the spiked analyte so that a
determination of concentration and recovery can be discerned from the analysis. In the calibration
process, minimum detection level and quantification levels were also set.
The concentration of the analyte exceeded the calibration curve concentration levels. The sample was
"diluted" by a known amount and followed specific procedures to bring the analytical results back into
the calibration range of the instrumentation. Using this procedure,a maximum dilution level of 50 was
specified.
Preliminary "Shakedown"Tests
Since this was the first time that GenX had ever been measured from a Chemours stack,the facility
proposed a series of preliminary or"shakedown"tests to determine if the testing and analytical
procedures worked acceptably and to make modifications to test procedures to address any issues
before actual testing began on all identified sources of GenX and HFPO DAF.
GenX preliminary testing was conducted during the week of January 9, 2018. The testing revealed that
several modifications were needed with regard to stack access on the Division Stack and sampling and
analysis procedures for quantification of the GenX at both VEN and PPA. In addition,the tests at PPA
gave an indication of room air fugitive emission rates since the stack configuration allows room air to be
tested separately from process emissions.
During analysis of the sample from the preliminary test, it was found that the concentration of GenX in
the recovered sample was higher than initially anticipated. The higher concentrations required higher
dilution rates on some samples to bring the target analytes into the proper calibration range of the
instrumentation. To address the high concentration issue, a smaller sampling volume would need to be
captured to reduce the amount of sample dilution necessary. A smaller sample volume could be
obtained by using a smaller diameter nozzle for isokinetic sampling or by reducing the 3-hour run time.
A sampling time of 90-minutes per run was set as a way to reduce the total sample while still assuring
that certain batch cycle characteristics were sampled in each process area. The test results discussed in
this review reflect the 90-minute sampling run time.
Test Results
The reported GenX test results reflect corrected emission rates accounting for dilution and spike
recovery values. Although presented on a per run basis,average emission rates are presented as the
average of three runs per method procedures. Run to run emissions indicate the need for additional
testing and source emissions characterization.
The shorter sampling times and smaller sample nozzle size for each run appear to have addressed the
analytical issues associated with high dilution ratio requirements that were identified during the
preliminary testing and analytics.

Division Stack/Vinyl Ethers North Test Results
GenX emissions testing of the Division Stack was conducted on January 22 (Run 1) and 23 (Runs 2 and 3),
2018 while producing PPVE. Production operations were considered normal and representative during
the sampling runs. Each sampling run was 90 minutes in length. The emission rate is the combination
of the process gases through the scrubber and the room air emissions because of the current stack
configuration. The sample analysis data indicated the sampling train captured most of GenX before the
second XAD-2 trap. The per run emission rate and average for all three runs is displayed in the table
below.
Table 1. Summary of Division Stack Test Results
GenX Emission Rate
Test Method Run Number
Ib/hr g/sec
1 0.246 0.0310
2 0.404 0.0509
Modified Method 0010
3 0.238 0.0300
Average 0.296 0.0373
PPA Area Test Results
GenX emissions testing of the PPA Area Stack was conducted on January 24(Runs 1 and 2) and 25(Run
3),2018 while the PPA process was producing GenX. Each sampling run was 90 minutes in length. The
emission rate is the combination of the process gases through the scrubber and the room air emissions.
An additional run was conducted to measure GenX from the room air portion of the effluent. The
sample analysis data indicated the sampling train captured most of GenX before the second XAD-2 trap.
The per run emission rate and average for all three runs is displayed in the table below.
Table 2. Summary of PPA Area Test Results
GenX Emission Rate
Test Method Run Number
Ib/hr g/sec
1 0.0258 0.00325
2 0.715 0.0900
Modified Method 0010 3 0.0252 0.00317
Average 0.2553 0.03214
Room Air Test Run 0.0058 0.00073

As shown in Table 2,the reported emissions from Run 2,are higher than the emissions from Runs 1 and
3. The test report from Chemours indicates that Run 2 was conducted under different process.
However, NC DAQ's assessment is that the operations and emissions captured by Run 2 represents
routine and normal operations for this process and there were no upsets or malfunction conditions that
should void consideration of the run. Rather,the operations that were measured during the test are
characteristic of normal unit operations that include batch operations. NC DAQ required that
emissions testing be conducted such that emissions from both continuous and batch operations were
captured and characterized and believes that this test(including all three runs) met this requirement.
Therefore,the emission rate is the average from all 3 runs.
Summary and Conclusions
Division of Air Quality staff were on site during each day that source testing occurred. DAQ staff
observed the source test teams,the sample recovery and the process operations. DAQ required
Chemours to develop test and analysis methodology within less than a four-month window. The test
methodology is still under review for accuracy and appropriateness at all Chemours sources. Based
upon the onsite observation of the testing and review of the test report, NC DAQ concludes that the
testing was conducted in accordance to the modified testing protocol submitted by Chemours and that
the analytical results appear representative of the stack conditions and process operations during the
testing. The test results indicate that additional testing of the PPA area stack may be needed to further
characterize the emissions profile during normal operations of the process. Additional PPA tests were
conducted during the week of February 26,2018 and will be reviewed separately.
Cc: Central Files—Bladen County
[BEAM Documents—0900009