2019.03.29_CCO.p11.a_PFASPlansFinal

March 29, 2019

Christel Compton
The Chemours Company FC, LLC
22828 NC Highway 87 W
Fayetteville, NC 28306-7332

RE: Review of Consent Order Paragraph 11a

Dear Ms. Compton:

This is to convey the final list of questions from our review of the documents related to the
Consent Order Paragraph 11a. Thank you for assisting with potential dates for the next few
weeks to discuss the items with our staff and EPA.

Geosyntec PFAS Characterization Sampling Plan
It appears that samples may either be sent to Eurofins or TestAmerica [both are referenced in
Tables 2 and 4]. Please submit the TestAmerica’s SOP(s) so we can understand how any of the
standard operating procedures may differ between labs.

Section 3.3.1 – Consider whether Clean Water Act approved methods [40 CFR Part 136] should
be used for the Field Parameter measurements taken during sampling. If so, those SOPs should
be included.

Section 3.3.2 – where are the boundaries for spatial and temporal composite sampling defined?
Please reference or provide detail for consistency in sampling events and documentation
requirements.

Section 3.3.2 – Do any of the markers (e.g., Sharpie and others) being used for sample labelling
contain PFAS?

Section 3.3.2 – Powderless nitrile gloves should be used and changed between samples. Gore-
Tex or other water-resistant, insect-resistant, or UV-protected synthetics or products should
not be worn during sampling.

2
Section 3.4 – Define storage temperature for archived samples since we do not know how long
archived samples may be stored. What type of QA/QC samples will be maintained (e.g., Field
Blank) should those archived samples be analyzed at a much later date?

Section 3.4 – The section states that sample coolers will be “taped shut” and signed across the
lid of the cooler. It is unclear what kind of tape is to be used. The use of custody seals is
recommended, and the bill of lading is to be retained with the sample records.

Section 3.5.1 – Recommend the use of Temperature Blanks to verify proper thermal
preservation during shipment.

Section 3.5.1 – The Field Blanks are required by the reference method and should be analyzed
regardless.

Section 3.5.1 – Field duplicates are used to assess the precision associated with sample
collection, handling and storage procedures as well as laboratory analytical processes. This
statement should be added to the first sentence under Field Duplicates.

Section 4.2 – If SPE is used, Method Blanks should be used at a frequency of one per extraction
batch rather than one per 20 samples when an extraction batch is less than 20 samples. This
will test the day-to-day variability between extraction batches.

Section 4.2 – Since this is a location- and time-limited project and since the bias these matrices
may attribute to these methods of analyses for these compounds is unknown, each sample
location should be characterized with Matrix Spikes rather than relying on random selection
during through ongoing QC requirements.

Section 4.2 – Indicate that CCVs are analyzed initially, after so many samples and at the end of
each analysis batch to bracket samples analyzed.

Section 5 – Provide all associated QC results and not just case narratives.

Table 2 – There are fairly significant differences in PQLs between the two contract laboratories,
TestAmerica and Eurofin Lancaster. Please describe how it is determined which lab is used.
It appears that Eurofins is not using the branched isomers of PFOS, PFHxS, NetFOSAA and
MeFOSAA in their calibration curves. What the rationale for that decision? Eurofins does
analyze a solution at the mid-point of the curve that contains the previous analytes and their
branched isomers. This is simply done for the analyst so they know how to fully integrate the
analytes in real samples. As such it seems that Eurofins is capturing the linear/branches
isomers when they analyze real samples, but they are not including them in the
calibration. The extraction/cleanup was left out as “proprietary”.

3
Table 4 – There is no mention of thermal preservation requirements [e.g., <10 degrees C
without evidence of freezing] for shipment and storage. Please clarify the requirements.

Table 4 – The significant difference in hold time [i.e., 40 vs 28 days unpreserved] used by the
two contract laboratories [TestAmerica and Eurofins] is concerning. Without stability validation
data in the matrices of interest, it is not known whether data may be impacted by the
significantly longer hold time especially if samples are not to be thermally preserved as
indicated in the Sampling Plan. The conditions for storage of extracts must also be addressed.
Table 5 – Footnote 2 may need adjustment based on previous comment regarding analysis of
Field Blanks and frequency of CCV based on previous comments.

MDL and IDOC studies are generally performed in laboratory water matrix. How have each of
the proposed methods have been validated for analysis of PFAS in the matrices involved in this
project? Were any spike studies, etc. done? The method validation is indirectly referenced for
the non-targeted analyses by reference to following EPA’s Protocol for Review and Validation of
New Methods for Regulated Organic and Inorganic Analytes in Wastewater under EPA’s
Alternate Test Procedure Program (EPA, 2016), but is not mentioned for the targeted analyses.

Original questions submitted to Chemours on March 22, 2019……………………………………………………
EPA Analytical Services Branch - Science and Ecosystem Support Division (SESD)

EPA, Science and Ecosystem Support Division (SESD) conducted a limited review of the
following documents provided by Chemours to NC DEQ Division of Water as part of their
consent order:
• Geosyntec’s, PFAS Characterization Sampling Plan, Process and Non-Process
Wastewater and Stormwater – December 28, 2018
• Chemours, PFAS Non- Targeted Analysis and Methods Development Plan
• Chemours, Determination of Table 3 Plus Compounds by LC/MS/MS, Chemours Fluoro
products Analytical Method revision date 01/10/2019
• Eurofins, PFAS in Aqueous Samples by Method 537 version 1.1 using LC/MS/MS

SESD has provided the following comments:

PFAS Characterization Sampling Plan –
1. Page 4, 3.3.2 Sample Collection Procedures Common to All Locations: There are no
“procedures” listed in this section nor in Table 4, Sampling Containers, Preservation, and
Holding Times.

2. Page 5, first paragraph, first sentence: If the sample containers used for EPA Method
537 Mod contain Trizma® as a buffer, they should not be used to fill the flow-through
cell.

3. Page 5, first paragraph, third sentence: “For each sample type” should be “For each
sample” or “For each sample station”.
4

4. Page 5, second paragraph: For consistency between sampling events, more detail of
how the sampling will be conducted for each of the various media (i.e. stormwater,
process water, NCCW, etc.). When temporal composite samples are being collected, the
duration and interval of the aliquot collection should be specified. During storm events,
the “composite samplers” may need to be reprogrammed for each storm, but it is
necessary to detail the criteria for the field personnel and the regulators what
information is desired for each sampling event. Additionally, the protocol used to split
the composite sample and fill the nine sample containers needs to be stated.

5. Page 5, second paragraph: While it may be advantageous to collect composite samples
for variable solids loading in stormwater samples, it should be acknowledged that there
may be PFAS losses in the composite sampling process. From Section 3.1 Equipment and
Supplies of the Interstate Technology Regulatory Council's Site Characterization
Considerations, Sampling Precautions, and Laboratory Analytical Methods for Per- and
Polyfluoroalkyl Substances [PFAS]: “Not all PFAS are hydrophilic, and some are volatile.
As a result, these chemicals may sorb to sampling equipment and supplies or be lost
from samples during sample
collection. Preliminary data suggest that sorption may occur quickly. Additionally,
volatile losses have not yet been characterized. Until they are better quantified,
sampling efforts should consider whether these losses would affect project objectives
and adjust accordingly.”.

To evaluate the PFAS losses, it is recommended that a grab sample (or two) be collected
in addition to one of the proposed composite sample locations for each sampling event.

6. Page 5, second paragraph, seventh sentence: Provide details for the Hach SD900 and
the other composite samplers. Are they refrigerated or capable of storing samples on
ice? Is there one container for the aliquots that will comprise the of temporal composite
sample or are there several containers? What is the composition of the sampler’s tubing
and components?

7. Pages 5 & 6, Stormwater, Intake and Outfall, NCCW, Process Wastewater, and WWTP
Discharge Sampling: Due to the potential stratification of PFAS in solution, the sampling
location/depth in the water column needs to be specified for each sampling medium.
(See Section 3.6.2 Surface Water of the ITRC's Site Characterization Considerations,
Sampling Precautions, and Laboratory Analytical Methods for Per- and Polyfluoroalkyl
Substances [PFAS]).

8. Page 6, Non-Contact Cooling Water: For the spatial composite samples, the length of
each ditch and the number aliquots collected needs to be stated so that data from
subsequent sampling events can be appropriately assessed.

5
9. Page 6, Process Wastewater: Process wastewaters have the potential for great
variability due to batch processing, process upsets, wash-downs, start-ups, etc., but grab
samples are being proposed for each production area. Grab samples are appropriate
where temporal variability over the course of one day is not expected at the sampling
location.

10. Page 7, 3.3.3 Decontamination Procedures: Tap water and de-ionized water should be
analyzed for PFAS and the one with the lower PFAS constituents should be used as the
final rinse in the decontamination procedure. If they both exceed the criteria to achieve
the DQOs for the project, PFAS-free water should be used as the final rinse.
Decontaminated equipment should be covered with polypropylene or other PFAS-free
plastic until it is ready for use.

11. Page 7, 3.4 Sample Shipping, Chain of Custody, and Holding Times, first sentence: To
prevent ice melt water from potentially contaminating the samples, it is recommended
that the samples be bagged in Ziplocs® or Whirl-Paks®.

12. Page 7, 3.4 Sample Shipping, Chain of Custody, and Holding Times, first sentence:
Composite sampling requiring ice preservation, should have composite aliquots stored
on ice or refrigerated during the sampling process.

13. Page 7, 3.5 Quality Assurance/Quality Control, second sentence: “Table 4” should be
“Table 5”.

14. Page 8, 3.5.1 Field QA/QC, second sentence: Insert at the beginning of the section
“Criteria for achieving...”. Also, “Table 3” should be “Table 2”.

15. Page 8, Equipment Blanks, first sentence: Equipment blanks should also be used to
demonstrate that sampling equipment is not contaminating the sample with analytes of
interests. Modify the first sentence to “Equipment blanks are used to evaluate
equipment and cleaning or decontamination procedures.”

The complete sampling system (containers, gloves, tubing, composite samplers, labels,
bags, etc.) should be evaluated to ensure all the equipment is PFAS-free and not
contributing to any PFAS detected in the samples. Initially this can be done by using the
composite sampler to collect a sample from PFAS-free water and labeling,
bagging, ice, and shipping to the laboratory for analyses. If PFAS are detected above the
PQLs, then further investigation will be required to determine the cause.

16. Table 2: Title should be “CRITERIA FOR ACHIEVING DATA QUALITY OBJECTIVES”.

17. Table 4: In the preservation column, “Ice” should be listed for each of the analytical
methods.

6
Chemours, PFAS Non- Targeted Analysis and Methods Development Plan –

1. The planning document overall, seems sound. There should be greater emphasis given
to synthesis of authentic standards by companies which maintain an arms-length
relationship with Chemours. Companies which specialize in the production of high-
purity analytical standards should be focused on for their expertise. This is especially
important, as new compounds are discovered and methods developed for their
quantification in real-world samples.

2. The use of high-field NMR would be of potential benefit in structural identification of
unknown compounds.

Chemours, Determination of Table 3 Plus Compounds by LC/MS/MS –

1. Details in this SOP are minimal, with limited procedural information available. A more
detailed version of the methodology is necessary for evaluation of the procedures.

2. Most of the target analytes are not available commercially. Suggest that funding be
directed toward synthesis of these compounds with non-related companies which
specialize in production of analytical laboratory standards so that primary and
secondary source standards are available. Doing so will give greater validity to the data.

3. Page 1, Sample Preparation. The note states that the 100 ppb mixed stock standard is
prepared in ultrapure water. Studies have shown that PFAS will bind to the sample
container in the absence of an organic solvent. Preparing stock or mixed standards in
water may cause low bias in quantitative results because of PFAS binding to the sample
container.

4. The 10-point calibration curve spans a concentration range from 10 – 50,000 ng/L. This
calibration range seems to span a range greater than possible with MS/MS detection.
Clarify how this is performed.

5. Page 2, Sample Preparation. The QC samples are prepared at a concentration of 5 ppb
(5,000 ng/L). Relative to the NC DHHS provisional health goal of 140 ng/L for HFPO-DA
and the US EPA health advisory for PFOA and PFOS of 70 ng/L, the QC spike
concentrations are quite high and not representative of protective health goals.

6. Page 4, QQQ Acquisition Parameters. If possible, add confirmatory ions where possible.
The qualifier ion ratio and retention time criteria for selection or exclusion of a
compound should be stated. This may require multiple MRM functions to obtain an
adequate number of scans across peaks.

7
7. Page 5, Data Processing, Calibration. No mention is made of the minimum number of
scans required across a chromatographic peak. A minimum of eight and preferably ten
scans are required to produce quantitative data.

8. Performing two calibration curves; one prior to sample analysis and one following the
sample sequence is not a technique typically seen. This may have the potential to
introduce bias to the determinations, especially if absolute instrument sensitivity has
been lost during analysis. If the practice is retained, precision and accuracy criteria
should be developed to judge acceptability of the closing calibration curves with respect
to the opening calibration.

9. Page 5, Data Processing, Blanks. As currently written, method blanks are optional (e.g.,
and/or). The requirement for a method blank is mandatory.

10. Page 5, Data Processing. Suggest adding a positive control spiked at a concentration
equal to the minimum reporting limit to verify method performance at the limit of
quantitation. A 5% frequency is reasonable (One per batch of 20 samples).

11. There is no mention of a method detection limit study. Suggest implementation of an
MDL study.

12. Page 5, Data Processing, Duplicates. Suggest eliminating the RPD 50% criterion and
using a single 25% criterion for all duplicates.

Eurofins, Polyfluorinated Alkyl Substances (PFAS) in Aqueous Samples by Method 537 Version
1.1 Modified Using LC/MS/MS -

1. The use of isotope dilution, where allowed, is to be commended. Isotope dilution and
extracted internal standards allow recoveries of target analytes to be corrected for
losses during sample processing.

2. The dependencies between internal standards (ISTDs) and the compounds they are used
to quantitate seems reasonable. Chemistry and chain length of ISTDs generally
corresponds to target compounds.

3. Branched isomers not in ICAL – are they included elsewhere? Integration of samples
should include both linear and branched chain isomers, where present. This is implied in
section D.12 (page 16) but needs clarification. If branched chain isomers are not
included during integration, the potential for underestimation of results exists.

4. No MRL confirmation performed. Recommend that a positive control sample spiked at
the MRL concentration be carried through the process to verify method performance.

8
5. With respect to the SOP title, and the use of “Modified” 537 1.1: The allowed changes to
Method 537 are limited and clearly stated. Once those allowed changes are exceeded,
the method is no longer considered Method 537. There have been major chemistry
changes made to this SOP when compared to Method 537. The title connotates
something that this method is not; this method can no longer be considered Method
537.

6. Redaction of certain SOP elements as proprietary make evaluation of the method in its
entirety impossible. The SPE and cleanup sections are critical to full evaluation of the
method.

7. There is no mention of HFPO-DA (aka, GenX) within the document. Presumed to not be
a target compound.

8. Page 5, Precaution to minimize method interference. This section is redacted as
proprietary content. No evaluation can be made.

9. Pages 6 – 7, Sample collection/preservation. Trizma is not added to samples unless they
originate from a chlorinated source. Method 537 uses Trizma as both a buffer and free-
chlorine scavenger. Its use in 537 increased recoveries for several compounds using
SDVB SPE (unpublished data). It is impossible to know what effect Trizma will have on
the samples analyzed by this method because of the (1) major differences in chemistry
and (2) the addition of compounds relative to Method 537.

10. Page 10, Initial Calibration (A.2). States there must be a detection of all analytes in the
MDL standard. If there is a signal-to-noise requirement for this criterion, state it.

11. Page 10, Initial Calibration (A.1/A.4). Section A.1 requires a minimum of five calibration
points. Section A.4 provides for the allowance of a second-order (quadratic) calibration
model. While not stated in Method 537, quadratic calibration requires a minimum of six
calibration points. Likewise, a first-order model requires a minimum of five calibration
points. These requirements should be stated in the SOP.

12. Page 11, Continuing Calibration (7.B.1.a) The requirement for use of the CS3 level
standard is stated. Method 537 §10.3 states “The beginning CCC of each analysis batch
must be at or below the MRL in order to verify instrument sensitivity prior to any
analyses.”

13. Page 11, Procedure, Sample Preparation (A.1) State how many significant figures are
necessary during weighing. Also applies to page 12, section A.3.j, and elsewhere.

14. Page 11, Procedure, Sample Preparation (A.3) The option for processing of an aliquot is
described. Numerous workers have shown that PFAS sorption to sample containers is a
source of sample loss and non-quantitative transfer. If aliquots are taken from the
9
original sample container, the sample must be qualified explaining the possibility for low
bias in reported values.

15. Pages 12 - 13, Procedure, Solid Phase Extraction (B.1 - 16) Most (16 of 21) procedural
steps of this section, which are critical to successful sample analysis have been redacted
as proprietary. As such, no judgement of the extraction can be made. NOTE:
EXTRACTION IS A CRITICAL STEP IN THE ANALYSIS.

16. Pages 13 - 14, Procedure, Solid Phase Extraction (B.18 -19) Steps 18 (tare bottle) and 19
(weigh bottle) appear to conflict with each other. Clarify.

17. Page 14, Procedure, Solid Phase Extraction (B.20) The SOP states that following addition
of labeled internal standards, the sample is ready for instrumental analysis. However,
section C (Extract Cleanup) follows this stated sample processing endpoint. Clarify the
flow of the sample extract and state if Section C Extract Cleanup is optional.
In section B.20, state what equipment is allowed or disallowed (pipette, syringe, etc.) for
sample transfer.

18. Page 14, Procedure, Extract Cleanup (C) All procedural steps of this section, which are
critical to successful sample analysis have been redacted as proprietary. As such, no
judgement of the cleanup procedure can be made. NOTE: CLEANUP STEPS ARE CRITICAL
TO THE ANALYSIS.

19. Page 14, LC/MS/MS Analysis, Mass Calibration and Tuning (D.1.b) The SOP provides a
criterion for mass axis calibration to be within ± 0.5 Daltons of the true value. There is
no statement indicating how this is performed. Note: Failure to maintain mass
calibration will impact instrument sensitivity.

20. Page 15, LC/MS/MS Analysis, Mass Calibration and Tuning (D.3) Acquisition Method,
Attachment 3 is redacted indicating it to be proprietary. Therefore, no judgement of the
LC/MS/MS acquisition procedure can be made. NOTE: ACQUISITION PARAMETERS ARE
CRITICAL TO THE ANALYSIS.

21. Page 15, LC/MS/MS Analysis, (D.5) The acronym L/B standard is used but undefined;
assume this to be Liner/Branched-chain. Clarify.

22. Page 16, LC/MS/MS Analysis, (D.11) The use of surrogates is mentioned; however, their
identity is not given. Clarify.

23. Page 16, LC/MS/MS Analysis, (D.12) The use of an MDL standard by which to judge data
quality is referenced. Its use is also mentioned in section A.2 (page 10). Clarify the origin
of the MDL standard.

10
24. Page 18, Mass Transitions AB Sciex 4500, Attachment 1. The stated transitions appear to
be nominal values. Mass transitions should be optimized on an instrument-by-
instrument basis. During optimization, the analyst should determine the precursor and
product ion masses to one decimal place (0.1 Daltons).

25. The precursor and product transitions for PFOS (2) appears to be incorrect. The table
entry m/z 413 > 169 should probably be 499 > 99.

EPA Office of Research and Development
Summary:
1) Groundwater analysis is not included. Comments on groundwater analysis may be
provided separately.
2) In section 2 of the data quality objectives PFAS are split into 2 groups: 1) Emanating
from Chemours; and 2) Not emanating from Chemours. A third group is needed: 3)
contributed to by Chemours. PFAS are in the Cape Fear River, however there are legacy
PFAS present contributed to by Chemours.
3) The PQLs listed in Table 2 for most of the analytes is too high at 50-120 ng/L. PQLs from
laboratories have been 5-10 ng/L.
4) The plan does not incorporate any spiked field sample or spiked trip samples to assess
recovery and accuracy of the intended analytes. Table 5 does indicate the plan for
matrix spikes which seem to be prepared on the day of analysis. EPA and DEQ have
implemented trip spikes in our project. Spiked field sample or spiked trip samples are
essential. If not done it is hard to assess what happens to samples during holding times,
while shipped on ice, frozen and archived and then extracted.

3.3.2 Sample Collection Procedures Common to All Locations
Page 5, 5th line: “… archive for potential future analyses.”
How are these samples going to be managed? Frozen? Extracted and archived?

3.5.1 Field QA/QC
The plan does not include spiked field samples. Chemours has all of these standards for
assessment. Add: 1) spiked blank water samples and/or 2) spikes of field collected
sample.
These samples should go out into the field (or are spiked in the field) and are cooled,
transported
and extracted as unknown samples are managed. This will determine if there is good
recovery for the chemicals analyzed, which determines accuracy. The assessment of
blank contamination and duplicate precision is in the proposal but NOT accuracy. Also,
this is NOT the same as the Matrix Spikes in section 4.2 below as the samples here are in
the lab on the day of analysis.

Table 2: TestAmerica and Eurofins Lancaster Current PQL’s are higher than expected.

Table 5: Spike sample must be included for Field samples.
11

PFAS Non-Targeted Analysis and Methods Development Plan
Process and Non-Process Wastewater and Stormwater
1 Introduction - What if any are the differences in how groundwater analysis is going to be
conducted?

1.1. Non-Targeted Analytical Background – what will be done in the absence of an authentic
standard?

1.2 Scope and Rationale – how is the “…prioritized set of the highest abundance additional
PFAS,” set? What is the cutoff for the highest?

2.5 Quality Assurance/Quality Control – how are the archived duplicates going to be
managed? Frozen? Extracted and archived?
3.1 Sample Preparation
How is it determined what gets SPE? If after a direct injection is done and no peaks are seen or
many large peaks are seen, how does the process proceed? Results can indicate many PFAS of
widely varying concentrations.

3.3 Compound Identification
Direct Injection is the least sensitive approach. If no peaks are detected via direct injection how
do you proceed?

3.3.2 Enhanced Assessment
Again, how is the prioritization conducted?

4 Reporting
Include that the QTOFMS datafiles will be provided to DEQ.

Determination of Table3
Sample Preparation – dilutions listed for calibration curve do not include values low enough to
support 5-10 ppt PQL.

Note on mixed stock standard - has the stock stability on water been assessed, and then
assessed as stored in a refrigerator?

“A 2x dilution will be appropriate for most groundwater analyses (trace levels expected).” This
is not the case for Chemours groundwater nor many of the surfacewater and effluent samples.

QC sample: “… stock standard solution separate from the one used for calibration…” Is this a
secondary standard prepared from the same chemical or from a secondary source?

12

Analytical Method
LC Operating Conditions: Column A – given a 50:50 methanol:water is injected into a 85:15
water:acetonitrile, you would expect early eluting peaks to split and not chromatograph well.

QQQ Acquisition Parameter – major concern in monitoring only one MRM for all of the
compounds. Modern QQQs such as your Agilent 6470 has no problems with two MRMs for all if
available, which makes ion ratios possible to evaluate. If not monitored, this is impossible. One
is listed for HFPO-DA but for no others. No stable isotope labled standards are listed. Most may
not be currently available but Wellington Labs does have 13C3 HFPO-DA.

Data Processing – reference indicates “The curve should be linear,…” Dr. Strynar indicates
quadratic generally works better even if the data is linear. If the data is non-linear, a linear fit is
not good.

Blanks – is the target analyte concentrations in all blanks defined as 0.01 ppb?

Duplicates – how will you work with data when one sample is above the LOQ and one is below
the LOQ?

Appendix A – please add molecular weight as another column of information in the table.
DFSA, Byproduct 4 and Byproduct 5 – are diprotic, what about M-2H?
MMF – this compound is diprotic and low molecular weight. The MRM is for the M-H. It
will be missed if it is M-2H as that MRM is not monitored for. The structure given is
difluoromalonate. What about monofluoromalonate as a possible contaminant?

Cover letter dated January 30, 2019 from Brian Long “The lab standards are in the form of a
0.1% (by weight) solution in water for each compound, and Chemours can ship these standards
solutions to DEQ promptly upon request.” - have stability tests for standards in this solution
been performed?