Legacy and Emerging Airborne Per- and Polyfluoroalkyl Substances (PFAS) Collected on PM_2.5 Filters in Close Proximity to a Fluoropolymer Manufacturing Facility

Abstract

Large fluoropolymer manufacturing facilities are major known sources of per- and polyfluoroalkyl substances (PFAS), many of which accumulate in groundwater, surface water, crops, wildlife, and people. Prior studies have measured high PFAS concentrations in groundwater, drinking water, soil, as well as dry and wet deposition near fluoropolymer facilities; however, much less is known about near-source PFAS air concentrations. We measured airborne PFAS on PM_2.5 filters in close proximity to a major fluoropolymer manufacturing facility (Chemours’ Fayetteville Works) located near Fayetteville, North Carolina, USA. Weekly PM_2.5 filter samples collected over a six-month field campaign using high-volume air samplers at locations 3.7 km apart, north-northeast and south-southwest of the facility were analyzed for thirty-four targeted ionic PFAS species by liquid chromatography coupled to electrospray ionization tandem mass spectrometry. Twelve emerging and ten legacy PFAS compounds were detected. Thirteen PFAS were found at higher concentrations in these nearfield samples than at regional background sites, suggesting a local source for these compounds. Five emerging and five legacy PFAS compounds had maximum concentrations exceeding 1 pg/m³. PFBA, PFHxA, PFHxDA, PFOS, PMPA, NVHOS, PFO5DoA, and Nafion BP1 contributed the most to the total (legacy + emerging) PFAS concentration (86%). Six PFAS, specifically PFBA, PFOS, PFO5DoA, Nafion BP1, Nafion BP2, and Nafion BP4, provide a consistent representative profile of elevated species across the two sites (with detection frequency > 50%). To our knowledge, this is the first study to report both legacy and emerging ionic PFAS in air in close proximity to a U.S. fluoropolymer manufacturing facility.

Introduction

Per- and polyfluoroalkyl substances (PFAS) are a chemically diverse group of manufactured chemicals widely used to make consumer and industrial products because of their oil, heat, grease, and water repelling properties.¹ The global fluoropolymer market is growing rapidly; market value is projected to rise from $7.7 billion in 2019 to $11.7 billion by 2027.² In addition to increasing demand, there is also increasing concern about the fate and transport of PFAS products and byproducts in the environment and their impact on human health because of their persistence and toxicity.^3–8

Large fluoropolymer manufacturing facilities produce fluoropolymers using legacy and emerging PFAS as polymer processing aids. These facilities are important sources of PFAS contamination via direct air emissions, dispersion and deposition, as well as discharge of manufacturing waste and wastewater into local aquatic environments.^{9, 10} Previous studies have reported high PFAS concentrations in surface water, groundwater, soil, sediments, crops, vegetables, birds, and eggs, as well as other environmental media and biota in the vicinity of fluoropolymer manufacturing facilities.^11–17 Higher PFAS concentrations in serum and blood samples of residents in close proximity to manufacturing facilities compared to the wider population average have also been measured.^18–21 In addition, adverse health effects (e.g., thyroid dysfunction, decreased kidney function, and cancers) have been documented in populations living near fluoropolymer manufacturing facilities.^22–26

There is considerable evidence suggesting that air emissions from fluoropolymer manufacturing facilities contribute to both air and water contamination,^{8, 27–29} yet near-field measurements in air (gas and particle phases) are rare, particularly for emerging PFAS compounds. Stack and fugitive emissions from storage and production, as well as emissions from treatment of waste/process water and legacy site contamination could all contribute to air concentrations. Some previous studies have modeled air concentrations from known or estimated PFAS emissions,^{30, 31} or used surrogate substrates (e.g., vegetation) as indicators of atmospheric PFAS concentrations or deposition.^{32, 33} D’Ambro et al.³⁰ used 2017 PFAS emissions estimates from Chemours’ Fayetteville Works facility (prior to stack testing and facility emissions controls) and the Community Multiscale Air Quality (CMAQ) model to predict air concentrations up to 24.6 and 8500 ng m⁻³ for total hexafluoropropylene oxide-dimer acid (HFPO-DA, or “GenX”) and total PFAS (gas + particle), respectively, around Chemours’ Fayetteville Works facility. Roostaei et al. ³⁴ linked modeled atmospheric GenX deposition, which is highly dependent on wind direction, to high GenX concentrations in private wells near the same facility. Galloway et al.²⁷ and Schroeder et al.³⁵ tied fluoropolymer production facilities in West Virginia (WV), Vermont (VT), New York (NY) and North Carolina (NC) to downwind (but upstream) GenX and PFOA in well water, suggesting the contamination could be explained by atmospheric deposition. Measurements of PFAS in vegetation also suggest atmospheric concentrations of PFAS are elevated near PFAS manufacturing facilities.³² A few studies have measured near-source PFAS in air. Barton et al.⁹ measured PFOA along the fence line of the DuPont Washington Works facility located near Parkersburg, WV using the Occupational Safety and Health Administration Versatile Sampling (OVS) tubes capturing gases and particulate matter simultaneously and a high-volume cascade impactor over a 10-week sampling period. They reported 24-h averaged PFOA concentrations at fence-line of 0.12 to 0.9 µg/m³ (gas + particle). Chen et al. ³⁶ collected 8 air samples (gas + particle) using passive air samplers with sorbent-impregnated polyurethane foam (SIP) disks, with each sample integrated over 44 days in areas surrounding two fluorochemical manufacturing parks in Fuxin, China. The sum of measured legacy ionic PFAS species averaged 4900 ± 4200 pg/m³. No particle size information was provided. In summary, well water contamination, vegetation measurements, modeling, and limited air measurements all suggest that PFAS air concentrations near fluoropolymer manufacturing plants can be quite high.

Notably, prior air studies have primarily focused on measuring legacy PFAS compounds, which is important to documenting the impact of phasing out production of those compounds.^{16, 37} However, with the increasing production and usage of emerging PFAS compounds that serve as legacy PFAS replacements,^{38, 39} there is an urgent need to characterize the abundance and composition of emerging PFAS in close proximity to fluoropolymer manufacturing facilities. As a result, in this study we collected atmospheric PM_2.5 filter samples close to a fluoropolymer manufacturing facility near Fayetteville, North Carolina and measured the concentrations of emerging and legacy PFAS species to provide insights into near source PFAS concentration profiles. To our knowledge, this is the first study to report both legacy and emerging ionic PFAS in the air near a fluoropolymer manufacturing facility in the U.S.

Materials and Methods

Sampling Sites

D’Ambro et al³⁰ predicted the highest PFAS concentrations would be within 0.1 km of the facility. We identified two locations where we could obtain power and permission and that were: 1) in line with the prevailing wind directions and 2) between 1 and 2 km of the facility with the goal of characterizing the nearfield PFAS concentrations and composition profile. The two sampling sites were 3.7 km apart and on opposite sides of Chemours’ fluoropolymer manufacturing facility Fayetteville Works (near Fayetteville, North Carolina, USA) (Figure 1). The selected sites were south-southwest (SW) and north-northeast (NE) of the facility, in line with the prevailing wind directions for this area. Fayetteville Works is a major fluoropolymer manufacturing plant in the USA and worldwide. The facility began fluoropolymer production in 1980 under the Dupont Corporation. It has been owned by the Chemours Company since 2015. Multiple fluoropolymer products have been manufactured at various locations within Chemours’ Fayetteville Works, including Chemours Nafion^® Membrane (plastic film), Nafion^® Polymer Dispersions, hexafluoropropylene oxide (HFPO) monomer as well as vinyl ether monomers, fluorocarbon intermediates for Nafion^® membranes, as well as other fluorocarbon products and Fluoropolymer Processing Aids (PPA).⁴⁰ On-site wastewater treatment operations are in the southern part of the facility. Monomer ion exchange membrane and polymer processing aid areas are in the north. Direct emissions (stack and fugitive), as well as emissions from legacy contamination, could be a factor at this facility.

Figure 1.

Sampling locations (north-northeast (labeled NE) and south-southwest (labeled SW) of Chemours’ Fayetteville Works). Wind frequency and speed were averaged over the 6-month sampling time.

The measurement campaign occurred at a time when several changes in emissions controls were underway, to abide by a consent order issued by the North Carolina Department of Environmental Quality (NC DEQ)⁴¹. Carbon beds were added, and scrubbers were enhanced to bring air emissions into compliance with a requirement for >90% control of GenX air emissions at the facility before fall 2019, which coincided with the start of our field campaign (September 20, 2019). By the middle of the campaign (December 27, 2019) a thermal oxidizer – four stage scrubber system was installed and fully operational,⁴² located north of the center of the facility. According to NC DEQ, Chemours’ Fayetteville Works facility typically conducts annual maintenance, testing and related operations during the month of October (personal communication).

Sample Collection

Six-day integrated PM_2.5 filter samples (Fri 16:00 to Thu 16:00 local time) were collected weekly on pre-baked (550 ℃) 102-mm quartz fiber filters (2500 QAT-UP, Pall Laboratory, Port Washington, NY, USA) using high-volume (HiVol) air samplers (~227 L/min) (Tisch Environmental, Inc., Cleves, OH, USA) at the two sampling locations (SW and NE sites) from September 20, 2019 to March 26, 2020. Trip blanks (filters loaded in a spare sampling head transported to and from both sites) were also collected weekly. PM_2.5 filter samples and field blanks were wrapped in pre-baked (550 ℃) aluminum foil inside Ziploc bags, transported and stored together in the dark at −20°C until analysis. A total of 26 and 20 quartz fiber filter samples (along with 25 and 22 field filter blanks) were collected at the SW and NE site, respectively. Note that quartz fiber filters have a large surface area and are known to adsorb some vapors as well as to efficiently collect PM_2.5.^43–46 While the gas phase is not collected efficiently on the filter, for some high volatility compounds, the filter may predominantly contain adsorbed vapor.

Real-time PM_2.5 mass concentrations were measured at the SW site for the entire sampling period and at the NE site from 01/03/20 to 03/26/20 using a personal dataRAM nephelometer (pDR-1500, Thermo Fisher Scientific, Waltham, MA, USA). Meteorological parameters (e.g., wind direction, wind speed, temperature, relative humidity [RH]) were measured using a meteorological station at a height of 3 meters (Precision Weather Station Vantage Pro2, Model 6162, Davis Instruments, Hayward, CA, USA) at the SW site.

Analyte Selection

Target analytes were ionic PFAS species selected based on measured or modeled species in Chemours stack emissions provided by NC Department of Environmental Quality (unpublished data), and preliminary measurements from the Cape Fear River, well water and drinking water (unpublished data). Additionally, we considered the availability of authentic standards.⁴⁷

Sample Preparation and Instrumental Analysis

Each filter was spiked with 0.5 ng of mass-labeled PFAS internal standards (Table S1) and extracted three times with 20 mL HPLC grade methanol (Thermo Fisher Scientific, Waltham, MA, USA) as described by Zhou et al.⁴⁷. Briefly, extracts were combined and evaporated to 5 ml under a gentle flow of nitrogen (Airgas, Radnor, PA, US). Each extract was filtered through a polypropylene-membrane syringe filter (Acrodisc GHP, 13 mm, 0.2 μm, Waters Corporation, Milford, MA, US) and further evaporated to ~25 μL under nitrogen. Milli-Q water was added to provide a final volume of ~100 µL with a 75:25 (v/v) solvent mixture of Milli-Q water and HPLC grade methanol to match the initial mobile phase composition. Thirty-four PFAS compounds, including 13 perfluoroalkyl carboxylic acids (PFCAs), 8 perfluorosulfonic acids (PFSAs), 9 perfluoroalkyl ether carboxylic acids (PFECAs), and 4 perfluoroalkyl ether sulfonic acids (PFESAs) were selected as target compounds in this study (Table S1). Analytical standards and mass-labeled internal standards were obtained from Wellington Laboratories (Guelph, Ontario, Canada) and the Chemours Company (Wilmington, DE, USA). Samples were analyzed by an AB SCIEX Triple Quad^™ 6500 ultra-high performance liquid chromatography-electrospray ionization tandem mass spectrometry (UHPLC/ESI-MS/MS) operated in negative ion mode with multiple reaction monitoring. Parts made of PEEK (polyether ether ketone) were used to replace Teflon^™ components in the system to minimize background PFAS contamination. UHPLC/ESI-MS/MS method parameters and operating details are given by Zhou et al.⁴⁷

Quality Assurance/Quality Control

Targeted PFAS compounds were identified and quantified with mass-labelled internal standards and a 5-point calibration curve, respectively. Reported concentrations were corrected for extraction efficiency (~70–120%)⁴⁷ using internal standards spiked on each filter before sample extraction. Analytical detection limits, determined via U.S. Environmental Protection Agency (U.S. EPA) protocol,¹ were 0.02–0.5 ng/mL of extract (0.0011–0.028 pg/m³ of air; Table S2). Reported concentrations were trip blank subtracted if the mean blank was significantly greater than zero (p<0.05, one sample t-test). Field measurement detection limits were taken to be the larger of the analytical detection limit or 3 times the standard deviation (3σ) of the trip blank values (Table S3). Analytical precision was better than 15% for most compounds (<5 – 30%), estimated as the pooled coefficient of variation of duplicate sample analyses.

Results

PFAS Concentrations and Profiles in PM_2.5 Filter Samples.

Twenty two of the thirty-four targeted filter-collected PFAS compounds were detected at concentrations from <0.0018 to 7.02 pg/m³ (Figure 2, Table S3). Ten PFAS had weekly-averaged concentrations >1 pg/m³. Among the 22 detected PFAS compounds, 10 were legacy PFAS compounds (i.e., 7 PFCAs and 3 PFSAs) while 12 were emerging PFAS compounds (i.e., 8 PFECAs and 4 PFESAs). Six were detected at both sites with detection frequency > 50% (Table S4: perfluorobutanoic acid (PFBA); perfluorooctane sulfonic acid (PFOS); perfluoro 3,5,7,9,11-pentaoxadecanoic acid (PFO5DoA); 2-[1-[difluoro[(1,2,2-trifluoroethenyl)oxy]methyl]-1,2,2,2-tetrafluoroethoxy]-1,1,2,2-tetrafluoroethanesulfonic acid (Nafion BP1); 1,1,2,2-tetrafluoro-2-[1,1,1,2,3,3-hexafluoro-3-(1,2,2,2-tetrafluoroethoxy)propan-2-yl]oxyethanesulfonic acid (Nafion BP2); and 2,2,3,3,4,5,5,5–4-(1,1,2,2-tetrafluoro-2-sulfoethoxy) pentanoate (Nafion BP4). Three were detected at high frequency (88.5%, 96.2% and 65.4%, respectively) at the SW site but only at low frequency (25%, 25%, and 5%, respectively) at the NE site: perfluorohexanoic acid (PFHxA); perfluoro-2-methoxypropanoic acid (PMPA); and 2,2,3,3-tetrafluoro-3-((1,1,1,2,3,3-hexafluoro-3-(1,2,2,2-tetrafluoroethoxy) propan-2-yl)oxy) propanoic acid (Hydro Eve). Three PFAS were only detected at the SW site and with low detection frequency: perfluorododecanoic acid (PFDoA, 1 out of 26), perfluoro-2-ethoxypropanoic acid (PEPA, 2 out of 26), and perfluoro (3,5,7,9-tetraoxadecanoic) acid (PFO4DA, 3 out of 26). Five PFAS were only detected at the NE site and with low detection frequency: perfluoroheptanoic acid (PFHpA, 1 out of 20 samples), perfluorononanoic acid (PFNA, 1 out of 20 samples), perfluorooctadecanoic acid (PFODA, 3 out of 20 samples), perfluorohexane sulfonic acid (PFHxS, 3 out of 20 samples), and perfluoro (3,5-dioxahexanoic) acid (PFO2HxA, 6 out of 20 samples). Perfluorooctanoic acid (PFOA) was among the PFAS not detected.

Figure 2.

Time series of legacy and emerging PFAS profiles at the SW and NE sites. 6d PCP is the 6-day integrated precipitation. No sample was collected from 01/31/20 to 02/13/20 for the SW site and from 01/10/20 to 02/13/20 for the NE site.

Thirteen PFAS were found at higher concentrations in these nearfield samples than at regional background sites (Table 1), suggesting a local source of these compounds. In 2019 statewide North Carolina monitoring of regional PFAS air concentrations, Zhou et al.⁴⁷ analyzed quarterly-composited PM_2.5 filter samples for the same 34 PFAS and only found detectable concentrations of PFOA, PFOS and PFHpA. In the current study, PFOA was not detected and PFHpA was only detection in 1 out of 46 samples. To compare the nearfield PFAS concentrations reported herein with concentrations of PFAS in regional-background NC air, weekly-averaged PFAS concentrations measured during this study are presented as quarterly-averaged PFAS concentrations for October-December 2019, the time period when both were measured. During these months, 4 legacy and 9 emerging PFAS were elevated in the PM_2.5 filter samples collected in close proximity to this facility (nearfield samples, this study; N=25), in comparison to the background sites (Zhou et al.⁴⁷; N=5). Most notably, quarterly-averaged PFHxA and PMPA were 0.3 – 1.2 pg/m³ in nearfield samples and <0.005 pg/m³ at all regional background sites. PFBA, 1,1,2,2-tetrafluoro-2-(1,2,2,2-tetrafluoro-ethoxy) ethane sulfonate (NVHOS), PFO5DoA and Nafion BP1 were 0.1 – 0.7 pg/m³ in quarterly-averaged nearfield samples and <0.02 pg/m³ at all regional background sites. Quarterly-averaged PFOS concentrations calculated from this study were 0.89 and 1.3 pg/m³ for the SW and NE sites, respectively, whereas the quarterly concentrations at the background sites collected concurrently were 0.16–0.72 pg/m³. The elevated nearfield concentrations measured for the 4 legacy and 9 emerging PFAS compounds, relative to the regional background, suggests a local source for these compounds.

Table 1.

Quarterly-averaged PFAS concentrations in PM_2.5 filter samples collected at the nearfield sites and in the regional background NC air from Oct.-Dec. 2019 (pg/m³).

	Legacy PFAS			Emerging PFAS
	PFBA	PFHxA	PFOS	PMPA	PFO2HxA	NVHOS	GenX	PFO5DoA	Nafion BP1	Nafion BP2	Nafion BP4	Hydro Eve
SW (N=13)	0.41	1.15	0.88	1.20	<0.021	0.65	<0.048	0.38	0.10	0.039	0.007	0.034
NE (N=12)	0.40	0.29	1.33	0.37	0.059	0.73	0.10	0.43	0.29	0.029	0.014	<0.015
Regional sites (N=5) a	<0.010	<0.004	0.16–0.72	<0.003	<0.018	<0.017	<0.026	<0.007	<0.004	<0.004	<0.005	<0.004

^a: data were derived from Zhou et al. 2021.⁴⁷ Among the five regional sites representing background PFAS in NC air, Fayetteville site is located closest to the facility (~20 miles northeast), perpendicular to the typical wind direction.

The time series of legacy and emerging PFAS concentrations are shown in Figure 2. NE and SW site PFAS species profiles for all PFAS species with detection frequencies > 10% are provided in Figure 3. Together, PFBA, PFHxA, PFHxDA, PFOS, PMPA, NVHOS, PFO5DoA, and Nafion BP1 contributed the most to the sum of measured (legacy + emerging) PFAS concentrations (86%). With detection frequencies > 50% for both sites, PFBA, PFOS, PFO5DoA, Nafion BP1, Nafion BP2, and Nafion BP4 provide a consistent representative profile of elevated species across the two sites (Figure 3, Table S4).

Figure 3.

Box-plot of weekly-averaged PFAS species concentrations collected on PM_2.5 filters (pg/m³). Shown are all PFAS species with detection frequencies > 10%. The top, central, and bottom of each box represent the third quartile, median, and the first quartile, respectively. Red squares represent the mean value. See also Figure S3.

Spearman’s rank correlation analysis was performed between the paired SW and NE site measurements (Table S5), and among PFAS species (Table S6) with detection frequency higher than 20%. The between-site comparison is influenced by wind direction but may provide insights regarding the ability of a single site to represent near source composition. The latter may provide insights into PFAS species that are co-emitted, used in the same process, or formed through the same mechanism. Significant (p < 0.01) positive between-species Spearman correlation coefficients (ρ) exist between several PFAS species (Table S6). PFBA was significantly positively associated with GenX, Hydro Eve, PFO5DoA and Nafion BP4. In contrast, PFHxA was significantly positively associated with PMPA, NVHOS, and Nafion BP2. PFBA and PFHxA were negatively correlated (ρ = −0.406.) Nafion BP1 was significantly positively associated with Nafion BP4 and PF05DoA. Correlations between species may reflect their use in common processes, their use at common locations within the facility, or emission/formation via a common mechanism. Significant (p<0.01) positive between-site ρ values were observed for PFBA, PFOS, and NVHOS (ρ = 0.618, 0.600, and 0.975, respectively). In contrast, Nafion BP1 was always detected but the between-site ρ concentrations were negatively correlated (ρ = −0.595).

Some differences were evident between the two sampling sites (SW vs NE), even after accounting for the dominant wind direction (Figure 4). This might reflect the fact that different types of production and processing have taken place at different locations within the facility, both now and in the past. It should be noted that each sample was collected over 6 days, and while a dominant wind direction could be identified for several samples, wind direction was variable over the sampling periods. Finer time resolution will be needed to conduct an upwind-downwind analysis. Sum of measured PFAS concentrations were typically higher at the SW site (Figure S1). Several PFAS compounds (PMPA, PFHxA and Hydro Eve) were higher at the SW site regardless of the dominant wind direction. This can be seen in Figure 4, which shows 10 pairs of samples with a dominant NE or SW wind direction (Figure S2, Table S7), as well as a Wilcoxon Signed Rank test (α=0.05, Table S5). Although not statistically significant, perfluorohexadecanoic acid (PFHxDA) was also higher at the SW site (Figure 4). PFOS, PFODA, GenX, Nafion BP1 and Nafion BP2 were higher at the NE site (Figure 4), although the difference was only statistically significant for PFOS and Nafion BP2 (Table S5). No clear seasonal variation between fall and winter was observed in our study.

Figure 4.

Average PM_2.5 filter-collected PFAS species concentrations for 10 pairs of samples with a SW or NE prevailing wind direction (see Figure S2, Table S7) measured in upwind and downwind sites.

We observed a change in the PFAS species detected after, compared to before, the first week of November; this change can be most clearly seen in Table S3 (detected species shown in bold). It is also seen in Figure 2. PMPA (at NE site) and NVHOS (at both sites) were detected only before the first week of November but not after, whereas PFBA (at NE site) and PFO2HxA (at NE site) were detected only after that time. GenX (at both sites), PFO5DoA (at both sites) and Nafion BP4 (at NE site) were also more frequently detected after the first week of November. Before November 7th, PMPA, NVHOS and PFO5DoA contributed the most to the total emerging PFAS profile (average contribution: 94.5% for the NE site). From November 7^th, 2019 to the end of the field campaign, PFO5DoA, GenX and Nafion BP1 contributed the most to the total emerging PFAS profile at the NE site (average contribution: 88.8%). Figure 2 shows two legacy PFAS concentration spikes at the NE site: one in the first week of December, when PFHxDA was dominant, and one in the last week of February when perfluorododecane sulfonic acid (PFDoS) was elevated. Two spikes of GenX with concentrations > 1 pg/m³ were observed in January 2020.

Notably, the average weekly sum of emerging PFAS concentrations from both sites were lower after November 7^th compared to before, with a 26% reduction at the SW site and a 61% reduction at the NE site. These changes during the first week of November were predominantly among emerging PFAS; trends in legacy PFAS concentrations and composition were difficult to discern. The changes in emerging PFAS might be a result of changes in activities at the site, manufacturing processes/products synthesized at this facility, changes in control technology (e.g., installation of thermal oxidizer/4-stage scrubber for most process emissions), or changes in the storage of precursors or products. Although changes in local conditions/meteorology/chemistry could contribute, meteorological changes are an unlikely explanation, given that a comparable change in legacy PFAS concentrations did not occur. Since the thermal oxidizer was required to be fully operational by December 2019,^{48, 42} it is certainly possible that changes in air concentrations observed in early November 2019 are associated, at least in part, with changes in facility stack emissions controls. We speculate that the more dramatic shifts in emerging PFAS composition occur because of changes in active operations and controls (stack emissions and fugitive emissions from storage and waste); while secondary emission from legacy site contamination is a reasonable explanation for the more consistent compositional pattern observed for legacy PFAS.

Discussion

Comparison with other studies.

Ionic PFAS concentrations measured in our current study (PM_2.5 filter-collected PFAS only) were lower than total (gas + particle) ionic PFAS concentrations measured near a fluoropolymer manufacturing facility in China,³⁶ and lower than those modeled by D’Ambro et al.³⁰ using Chemours’ Fayetteville Works 2017 self-reported emission information (before controls were implemented) and the Community Multiscale Air Quality Modeling System (CMAQ). D’Ambro et al.³⁰ predicted maximum downwind air concentrations of total (gas + particle) GenX and sum of speciated PFAS of 24.6 and 8500 ng/m³ at 0.1 km downwind, respectively, and concentrations on the order of ∼0.1 and ∼10 ng/m³ at 35 km downwind, respectively. For the purpose of modeling, they assumed that all PFAS were emitted in the gas phase. The lower PFAS concentrations in the current study are explained, in part, by the fact that we measured only filter-collected PFAS and the fact that the current study was conducted after (and during) the installation of significant controls on stack emissions in 2018 (and 2019). Removal of 99.99% of GenX in the thermal oxidizer would bring the D’Ambro gas+particle GenX concentrations at 0.1 km to approximately 2 pg/m³. Given the likely large uncertainties in emissions and gas-particle partitioning, this is in reasonable agreement with our filter-collected GenX measurements within 2 km (SW site < 3.2 pg/m³ with 88% below 0.06 pg/m³; NE site < 0.45 pg/m³ with 55% below 0.06 pg/m³).

Notably, several compounds measured in this study (e.g., GenX and Hydro Eve) were on the list of emissions reported by Chemours’ Fayetteville Works in 2017.³⁰ Most legacy PFAS measured in this study were also measured near fluoropolymer manufacturing plants in China (e.g., PFBA, PFHxA, and PFOS).³⁶ Barton et al. ⁹ measured PFOA in air in a six-event, 24-hr monitoring campaign during November 5, 2003 through January 16, 2004, along the fence line of the Washington Works facility in Parkersburg, WV, USA, and reported the particle phase PFOA concentration ranged from 0.12 to 0.9 µg/m³. In contrast, PFOA was below the field measurement detection limit (<0.63 pg/m³) in our study. This finding is consistent with expectations following US EPA’s PFOA Stewardship Program⁴⁹ launched in 2006, which aimed to eliminate PFOA from emissions and products no later than 2015.

There are some similarities between the compounds observed in this study and those observed in other types of samples that may be influenced by the Fayetteville Works facility. A study by Kotlarz et al.⁵⁰ collected blood and serum samples from communities living downstream on the Cape Fear River. Since the Chemours’ Fayetteville Works plant is a major source of PFAS to the Cape Fear River, we might see some similarities. In their study, a 99% detection frequency of Nafion BP2 in serum from the participants was reported, similar to our 93% detection in near-field PM_2.5 samples. GenX and PMPA were not detectable in their blood and serum samples. GenX was not detected at high frequency (26%) in our PM_2.5 samples. However, it has been measured in Fayetteville Works emissions and consistently in wet and dry deposition near the plant.^{51, 52} It is possible that this discrepancy is caused by detection limit differences. Unlike the serum study, we had a high detection frequency for PMPA. It should be noted that PMPA is a C4 short-chain PFAS, and compared to the long-chain Nafion BP1, Nafion BP2, and Nafion BP4, PMPA might have lower bioaccumulation potential,⁵³ and thus, leading to lower detection frequency than Nafion BP2 in the human body. The median PFOS concentration in serum was almost twice that of the median PFOA concentration in serum. Likewise, PFOS air concentrations were much higher than PFOA air concentrations in our study. In a separate study, Nafion BP2, PFO4DA, and PFO5DoA were detected in most blood samples (detection rate ≥88%) from volunteers living near this facility.^{26, 50} Nafion BP2, PFO4DA, and PFO5DoA were also detected in the current study, although not at the same detection frequency (94%, 61%, and 11%, respectively). While it should be recognized that there are also indirect sources of exposure to PFAS that undoubtedly influence blood and serum samples (e.g., from PFAS-containing consumer product use),⁵⁴ the presence of Chemours’ signature compounds (e.g., Nafion byproducts) is consistent with the influence of Chemours’s Fayetteville Works facility on PFAS exposure through inhalation, dermal and/or ingestion of contaminated air, water and/or food.

The types of PFAS compounds observed in this study also show similarities to water and wastewater samples that are expected to be influenced by the Fayetteville Works facility. Eight PFAS compounds were measured only near and downstream of Chemours’ Fayetteville Works facility in the Cape Fear River and also in groundwater samples collected from 5 tributaries of the Cape Fear River near the Fayetteville Works facility.^55,56 Similar to the groundwater samples, PMPA, Nafion BP2, and Nafion BP4 were detected (at high frequency, 65%, 94%, and 65%, respectively) in the current air study. In contrast, PFOS was detected at low detection frequency and low concentration in the groundwater samples,⁵⁵ but it was detected at high frequencies (>95%) at both sites in the current study. PFO2HxA, PFO3OA, and PFO4DA have also been found in the Cape Fear River,^{17, 57} although they were detected at modest frequencies in the air samples in our study. Additionally, Geosyntec, Inc. measured PFAS concentrations in the process and non-process wastewater generated by the Chemours Company - Fayetteville Works facility and reported many of the same PFAS measured in the current study.^{58, 59} Legacy PFAS (e,g., PFBA, PFHxA, and PFHxDA) and emerging PFAS (e.g., PFO2HxA, PFO3OA, PFO4DA, PFO5DA, PMPA, PEPA, NVHOS, and Hydro Eve) were measured in both the wastewater and air. Wastewater treatment has been reported to be a source of PFAS to air.⁶⁰ Thus, treatment of Chemours wastewater and process water could be a source of these PFAS species to near-field air concentrations. For example, it could help to explain the higher PFHxA and Hydro Eve concentrations at the SW site which is nearer to the wastewater treatment facility, compared to the NE. Fugitive emissions from storage tanks and emissions from stacks are also possible sources.

Regardless of the specific air emission pathway, the presence of these compounds in the process water clearly demonstrates that they are being used at the site. The measurement of similar PFAS species across multiple environmental media, linked to Chemours in different ways, adds confidence that the principal source of the elevated nearfield air PFAS concentrations is the Chemours’ Fayetteville Works facility. Additional discussion of the potential sources of individual PFAS can be found in the supplementary information.

Insights relevant to fate and transport.

Chain length has an important impact on the properties and fate of PFAS compounds in the environment.^{61, 62} In our study, twelve of the 34 measured PFAS were short-chain PFAS (n < 6 for PFSAs and PFESAs, as well as n < 7 for PFCAs and PFECAs),^{63, 64} with PFBA, PFHxA and PMPA most frequently detected (> 80% detection at ≥1 site). In comparison, ten of the 34 total PFAS measured were long-chain PFAS, with PFOS, Nafion BP1 and Nafion BP2 most frequently detected (>80% detection at ≥1 site). Typically, short-chain PFAS species are considered to be less bioaccumulative and toxic, but have higher long-range transport potential and aquatic mobility.^{39, 63} Long-chain PFAS are typically more carcinogenic and bioaccumulative.^{65, 66} Several short-chain PFAS have been used and manufactured as alternatives to long-chain PFAS;^{67, 68} for example, GenX has been used to replace PFOA. However, recent studies have shown that short-chain PFAS are persistent in the environment and some of them are more toxic (e.g., GenX vs. PFOA).^{69, 70} Thus, the potential environmental and public health effects of short-chain PFAS should not be overlooked.

Limitations.

One limitation of our study is that PFAS species were collected on quartz fiber filters that are highly efficient at collecting particles but also collect a portion of the gas phase by adsorption. We did not attempt to measure or correct for these artifacts. Quartz fiber filters have a large surface area but a small pressure drop. They are known to adsorb some organic gases, resulting in a net positive artifact, for example, when measuring total particulate organic carbon or particulate ionic PFAS.^43–46 Thus, concentrations reported include a contribution from the gas phase. Second, gas phase species are not collected efficiently on filters. Low carbon number PFAS species are likely present at substantially higher concentrations in the gas phase than in the particle phase, and PFAS can be transported longer distances⁷¹ in the gas phase than in the particle phase. Thus, gas-phase PFAS measurements are also needed to fully assess the impact of PFAS air emissions from fluoropolymer plants.^72–74 Third, we did not measure coarse particles (>2.5 μm in diameter). Resuspension of contaminated soil may result in PFAS in coarse particulate matter. Fourth, we did not collect samples during summertime. Contaminated soils and surface waters surrounding Chemours facility could potentially cause additional PFAS burden due to volatilization and recondensation of semi-volatile PFAS. Finally, we collected 6-day integrated samples to overcome detection limits. Development and use of real-time methods for PFAS detection (or sample collection with higher time resolution), could yield substantial new insights.⁷⁵