Atmospheric Concentrations of Per- and Polyfluoroalkyl Substances in Michigan’s Ambient Air Using Passive Sampling

Abstract

As a part of Michigan’s efforts to identify and address per- and polyfluoroalkyl substances (PFAS) contamination statewide, more knowledge of atmospheric PFAS concentrations is needed to better understand their atmospheric transport and impact on other environmental media. This research aimed to measure atmospheric concentrations of PFAS in Michigan’s ambient air using low-cost and easy-to-use passive samplers and identify relationships with environmental factors. Passive samplers, consisting of polyurethane foam (PUF) discs and sorbent-filled polyethylene tubes (radiello-XAD samplers), were deployed for a month at 27 sites across Michigan and analyzed for both ionic and volatile, neutral PFAS. Short chain perfluorocarboxylic acids (PFCAs), specifically perfluoropropanoic acid (PFPrA) (d.f. 60 %) and perfluorobutanoic acid (PFBA) (d.f. 96 %), and perfluorooctane sulfonic acid (PFOS) (d.f. 63 %) were the most frequently detected compounds, with the ∑₃PFAS ranging from non-detect to 15.5 pg/m³ for the PUFs and from non-detect to 248 pg/m³ in the radiello-XAD samplers. 6:2 fluorotelomer alcohol was also frequently detected (d.f. 45 %) and had a positive correlation with both population density (r=.52, p<.05) and industrial sites’ density (r=.48, p<.05). Detection of specific compounds on each type of sampler provided insight into the preferential transport pathway of atmospheric PFAS. For example, PFOS was observed mostly in the gas-phase, predominantly captured by the radiello-XAD samplers, while PFBA was mostly in the particle-phase, predominantly captured by the PUFs. This study highlights the importance of developing detection tools for measuring atmospheric PFAS across a vast geographic area to identify contributing factors to ambient concentrations.

1. Introduction

Environmental exposure to per- and polyfluoroalkyl substances (PFAS) has become of increasing concern due to the growing body of research showing their notable persistence and associated human health risks.¹ Perfluoroalkyl acids (PFAAs), which are generally ionized in the environment, exhibit a strong affinity for the aqueous phase, and so a large focus has been on their presence in surface, ground, and ocean water.^2–4 However, research has emerged showing the global dispersion, long-range transport, and oxidation of their volatile polyfluorinated precursor compounds.^5–7 Despite this, we have minimal studies with high spatial resolution measurements, limited capabilities of detecting and analyzing the wide scope of compounds that are present, and a lack of standardized methods for PFAS air sampling. This is particularly the case for atmospheric PFAS, even with research highlighting inhalation as a possible significant exposure pathway, and PFAS having been detected even in the most remote parts of the world,^8,9 yet the understanding of their fate and transport is still quite limited.

The volatile polyfluorinated precursors, which includes compounds such as fluorotelomer alcohols (FTOHs) and perfluorooctane sulfonamides/ perfluorooctane-sulfonamido-ethanols (FOSA/FOSEs), are capable of transforming or degrading into ionic PFAAs, such as perfluorooctanoic acid (PFOA) and perfluorooctane sulfonic acid (PFOS), which have recently been classified as carcinogenic (PFOA) and possibly carcinogenic (PFOS).^10,11 Atmospheric sources of these precursors can come from landfills, industrial emissions, wastewater treatment plants (WWTPs), and households.^12–18 The low water solubility and high vapor pressure of the neutral PFAS precursors also means they can volatilize from water into air.¹⁹ Even lesser researched is the hypothesis that ionic PFAS can be directly atmospherically transported via gas phase or aerosol-mediated pathways.^20,21

Given that the atmosphere is now considered a key transport medium for PFAS,²² and the knowledge that air inhalation is an exposure pathway for humans,²³ a better understanding of the fate and transport of PFAS and the extent to which they are atmospherically present is needed. Active sampling and passive sampling are frequently utilized methods for measuring PFAS concentrations in air, often using materials such as polyethylene sheets and sorbent-impregnated polyurethane.^24–26 Active sampling, which actively pulls air through a sampling medium, can be useful when needing to sample a large volume of air, but relies on a power supply to do so. Often, active sampling spans a sampling period of a couple of hours, utilizing bulky and expensive equipment.²⁶ This makes data collection with high temporal and spatial resolution physically impractical or too expensive, especially in more remote areas. In these cases, passive sampling can provide a solution, typically measuring a (time-weighted) average concentration over a longer period of time using cheaper, less labor-intensive materials.²⁶ However, passive sampling methods require an aspect of calibration during development, to determine sampling rates, or equilibrium partitioning constants, which can be compound specific. Overall, depending on the environment and target compounds, different tools present various advantages or complications. For example, polyurethane foam (PUF) passive air sampling has been used in both indoor and outdoor studies targeting PFOA and PFOS.^26–29 However, PUF passive sampling is not an effective method for measuring more volatile compounds, like FTOHs, due to them reaching equilibrium within just a few hours.^30,31 Previous studies, such as Shoeib et al., 2008, reported PUF discs reaching equilibrium within 1 day, making them unsuitable to collect time-integrated concentrations for these compounds over a longer deployment period.³¹ On the other hand, other passive sampling techniques, such as sorbent-impregnated PUFs, polyethylene sheets, and XAD-2-passive air samplers, have shown to be suitable methods for measuring volatile PFAS, but few to none studies reporting concentrations for both volatile and ionic PFAS.²⁶

Across the USA, there are limited studies that have conducted local or regional measurements of atmospheric PFAS, and even fewer that measure a wide scope of both volatile and ionic PFAS. One example of such a study in the USA is the Integrated Atmospheric Deposition Network, in which atmospheric concentrations of both volatile and ionic PFAS are monitored across the Great Lakes.³² Sampling campaigns of a smaller scope have also been conducted in North Carolina to measure atmospheric PFAS near fluoropolymer manufacturing facilities, in New York to assess contamination pathways to urban lakes, and at 5 sites across North America as part of the Global Atmospheric Passive Sampling (GAPS) network.^33–35

As part of Michigan Department of Environment, Great Lakes, and Energy (MI EGLE), the Michigan PFAS Action Response Team (MPART) has been working to identify and address PFAS contamination across the state. Several studies have reported PFAS contamination in different environmental matrices throughout Michigan, mainly water and biota, with a large concern placed on the implications to human health.^36–39 Lesser studied is state-wide atmospheric concentrations of PFAS, and so the atmospheric transport, implications to other environmental matrices, and contributions to total exposure for humans are not yet fully understood.

The primary goal of this study was to measure atmospheric concentrations of PFAS in Michigan’s ambient air covering a vast geographical area with good spatial resolution, using low-cost and easy-to-use tools. To achieve this, the aims of this study were to (i) deploy a unique combination of two different passive samplers to simultaneously measure a wide scope of both volatile and ionic PFAS outdoors, (ii) identify spatial patterns or relationships with environmental factors and site characteristics, and (iii) assess the possible implications of the atmospheric concentrations on Michigan’s environment.

2. Methods

2.1. Chemicals and Reagents

Liquid chromatography-grade methanol (LC-MeOH) and water (LC-water) were purchased from Fisher Scientific (Pittsburg, PA, USA) along with ammonium hydroxide (NH₄OH), ammonium acetate (C₂H₇NO₂), ACS-grade ethanol (EtOH), ACS-grade ethyl acetate (EtOAc), and ACS-grade methanol (MeOH). Both native and mass-labelled analytical standards of all target analytes were purchased from Wellington Laboratories (Ontario, Canada) (SI, Table S2 and Table S3).

2.2. Passive Sampler Assembly

Passive sampler set-ups consisted of two samplers: (a) a PUF disc encased in two stainless steel bowls, and (b) a polyethylene diffusive tube (radiello™) filled with a sorbent (XAD-4) (Figure 1). Polyethylene radial diffusive bodies, radiellos™, and Amberlite™ XAD-4 (Supelco) were purchased from Sigma Aldrich. PUF discs were purchased from Tisch Environmental (OH, USA). Other materials for sampler assembly and deployment such as metal rods, bowls, washers, nuts, and zip ties were purchased from McMaster-Carr (IL, USA). PUF discs were Soxhlet precleaned with LC-MeOH. All other sampler assembly materials, including radiellos™ and XAD sorbent, were pre-cleaned with a rinse of basic MeOH and LC-MeOH before assembly. The radiello™ bodies were filled with the XAD-4 sorbent and capped using a small amount of cotton wool (to prevent the XAD from falling into the adapter) and the radiello™ vertical adapter. Prior to deployment, all samplers were stored in muffled aluminum foil packets and plastic bags for chemical and physical protection.

Figure 1.

Figure showing passive sampler bowl set-up consisting of PUF and sorbent-filled tube sampler (radiello-XAD).

2.3. Field Deployment

Passive air samplers were deployed simultaneously at 27 locations across Michigan, USA for approximately a month-long period between September 2021 and October 2021 (Figure 2). Sites chosen were air monitoring sites regularly accessed by Michigan EGLE. Two of the chosen sites had a duplicate air passive sampler set-up and a third site had a total of 5 air passive sampler set-ups. Sampling location details can be found in the SI (SI Table S1).

Figure 2:

Site map showing the 27 locations where passive air samplers were deployed outdoors in Michigan, USA. Circles show a surrounding area with a 15 km radius.

2.4. Sample Extraction

Two subsequent fractions, a LC-fraction and GC-fraction, were taken from each radiello-XAD sampler using solid-liquid extraction (SLE) based on the method previously published.⁴⁰ All extraction materials were pre-cleaned with a basic MeOH and LC-MeOH rinse. The sorbent was removed from the diffusive tube, split into two, and each fraction spiked with the corresponding mass-labelled internal standards (listed in SI Tables S2 and S3) in a 15 mL Falcon tube. SLE was performed by vortexing and subsequent 30 minutes sonication using 4 mL of either EtOAc or LC-MeOH for the GC- and LC-fraction, respectively. Extracts sat for 24 h, after which the extract was collected in a pre-labelled, pre-cleaned Falcon tube. The SLE was repeated twice more, resulting in 12 mL total of extract. For the GC-fraction, the extract was concentrated to 100 µL in volume under a gentle stream of nitrogen and prepared for analysis via gas chromatography mass spectrometry (GC-MS). For the LC-fraction, the extract was concentrated to 0.5 mL, from which an aliquot of this final concentrate was diluted with 4 mM ammonium acetate in water to achieve 40/60 of methanol/water ratio, suitable for analysis via liquid chromatography tandem mass spectrometry (LC-MS/MS).

Extraction and analysis of PUF samples was based on the methods described elsewhere.^28,41 Briefly, PUF samples were spiked with mass-labelled internal standards (listed in SI Tables S2 and S3) and were extracted using SLE with 10 mL LC-MeOH. Extraction was repeated twice more and the combined extract evaporated down to 0.5 mL under a gentle stream of nitrogen. An aliquot of the concentrated extract was diluted with 4 mM ammonium acetate in water and prepared for analysis via LC-MS/MS as described above.

2.5. Instrumental Analysis

Samples were analyzed for neutral PFAS on an Agilent 7890B gas chromatograph coupled to an Agilent 5977A mass spectrometer (MS) device operating in positive chemical ionization mode using selected ion monitoring. Samples were analyzed for ionic PFAS on a liquid chromatograph SCIEX ExionLC AC UHPLC coupled to a SCIEX X500R quadrupole time-of-flight tandem mass spectrometer (QTOF MSMS), following the previously published method.⁴² For additional details and quality assurance, refer to the SI.

2.6. QAQC

The quantification of all targeted PFAS in samples and quality control (QC) samples was based on the isotope dilution method. The concentrations of the target compounds were recovery-corrected by a set of mass-labeled internal standards spiked to each sample prior to extraction (see SI, Table S2 and S3). Recoveries of the surrogate mass-labeled PFAS spiked into the real samples, blanks, and quality control samples were within QC criteria 60–140% (SI Table S4). Instrumental quality control was performed by running a mid-calibration standard with every batch of analyzed samples. Process blanks and instrumental blanks were included in each sample batch for laboratory quality control. Travel blanks and field blanks were also included to monitor the sample preparation method performance. Method detection limits (MDLs) were calculated as the field blank average plus 3 times the standard deviation, however, when a compound was not detected in the blanks, half the instrumental detection limits (IDLs) were used. IDLs represent the concentration of analytes giving the signal-to-noise ratio of 10. MDLs were not calculated for target compounds that were not detected in any blanks or samples. See Table S5 and S6 for MDLs. Only values above the MDLs were reported and concentrations were reported as the measured concentration minus the blank average.

2.7. Data Analysis

Volumetric concentrations (pg/m³) were calculated from the quantified amount of a compound (pg/sample) by dividing by the deployment length and sampling rate (m³/day). For both samplers, sampling rates for individual compounds have been previously determined and published for only a select number of compounds. Therefore, one sampling rate for each sampler was used to calculate air concentrations for all detected compounds. For the PUF samplers, the recommended sampling rate of 4 m³/day was used for all target analytes.²⁹ Previously published calibration studies reported sampling rates between 0.1 m³/day and 0.6 m³/day for XAD-passive samplers for a wide range of semi-volatile organic compounds (SVOCs).⁴³ Indoor calibration studies have been conducted specifically for the radiello-XAD samplers but sampling rates for only a few compounds were calculated due to limited detection.⁴⁰ Based on the previously conducted calibration studies specifically for the radiello-XAD samplers, a sampling rate of 0.1 m³/day was used for all target analytes.⁴⁰

Population density (people/km²) within a circular area with the radius of 5 km, 10 km, and 15 km surrounding the deployment site was calculated by taking population data from NASA Earthdata and dividing by the total area (see SI Table S7).⁴⁴ Average wind speed, wind direction, and temperature over the sampling period was calculated from meteorological data collected and shared by Michigan EGLE (see SI Table S1). Meteorological data was only collected at a portion of the sites. Industry density (number of industries potentially handling, using, or releasing PFAS within an area) was calculated using the EPA ECHO PFAS Analytical Tool to count the number of industries within a 15 km radius circular area surrounding the sampling site (see Table S8). Moran’s I test was used to explore spatial patterns in the data, however no significant spatial relationship was observed. For sampling sites with multiple samplers, average concentrations were used for data visualizations and statistical calculations. For statistical analysis, non-detects or values below the MDL were replaced with the MDL/$\sqrt{2}$. All data visualizations and statistical analysis were done with R (R version 4.2.2).

3. Results and Discussion

3.1. Site Characteristics

Sampling sites spanned a large area of Michigan covering urban and remote locations (Figure 2). Population density for the surrounding area (circles with radius of 5 km, 10 km, and 15 km) was calculated for each site (SI Table S7). Sites with the highest population density are those in Detroit, while sites with the lowest are the more northern sites (Seney, Houghton Lake, Harbor Beach, Benzonia). Population density ranged from 2 people/km² to 1421 people/km² for the surrounding area with a 15 km radius. Several sites are in close proximity to water, with 4 sites on the shore of Lake Michigan, 3 sites on the shore of Lake Huron, and 7 sites in Detroit, which is on the bank of the Detroit River connecting Lake St. Clair and Lake Erie (Figure 1). For sampling sites co-located with meteorological stations, the average wind speed and air temperature over the month sampling period ranged from 3 mph to 6 mph and 12 °C to 16 °C respectively (SI Table S1).

3.2. Occurrence of Atmospheric PFAS

Of the target ionic PFAS, 16 compounds were detected above MDLs in the air samples at one or more sampling locations (Figure 3). For the PUF samples, PFBA was the most frequently detected followed by PFPrA, with detection frequencies of 97 % and 58 % respectively. In general, across the different functional groups, short chain PFAS (C≤6) were more frequently detected in PUFs compared to long chain PFAAs (Figure 3). In the radiello-XAD samplers, PFOS was the most frequently detected, with a detection frequency of 63 %.

Figure 3:

Squares show when compounds were detected above MDL at sampling locations in either the PUF, the radiello-XAD samplers, or both. Only compounds detected in at least one site in one type of sample shown. *Neutral PFAS were only measured in radiello-XAD samplers. Compounds are listed in order of molecular weight.

For the volatile, neutral PFAS, which were only measured in the radiello-XAD samplers, 5 target compounds were detected above the MDL at one or more sampling location (Figure 3). The FTOHs were the most frequently detected across all sites, with 6:2 FTOH detected at 45 % of sites, 8:2 FTOH at 24 % of sites, and 10:2 FTOH only at 1 site. N-methylperfluorooctane-sulfonamide (N-MeFOSA) and N-ethylperfluorooctanesulfonamide (N-EtFOSA) were also detected, but only at one site each (Figure 3). Air samplers detected the highest number of compounds in Dearborn, with 13 compounds measured above detection limits (Figure 3).

In only a few instances a compound was detected by both the PUFs and radiello-XAD samplers at one location (Figure 3). Without the use of PUFs, the sampling campaign would not have captured the frequent presence of short chain carboxylic acids (PFCAs) across Michigan. Similarly, without the inclusion of the radiello-XAD samplers, the study would not have included the measurement of any volatile PFAS. While previous studies have utilized sorbent-impregnated PUFs (SIP-PAS) to increase the uptake of volatile PFAS compared to PUF discs alone, various equilibration times have been observed by different studies using the SIP-PAS, adding uncertainty to calculated air concentrations.²⁶ Furthermore, in this study, the detection of a compound in one sampler but not in the other may provide insight into whether transport of the compound is primarily in the gas-phase or particulate-phase. While SIP-PAS has the capability to capture both particulate-phase and gas-phase PFAS, associating phases to detected compounds may be more difficult, providing less insight into the fate and transportation of atmospheric PFAS.²⁹

An assessment of gas-particle partitioning by Karásková et al., 2018, calculated the particle-associates fraction of several PFAAs in outdoor air.²⁸ They reported a statistically lower particle-associated fraction for perfluorosulfonic acids (PFSAs) compared to short-chain perfluorocarboxylic acids (PFCAs).²⁸ In our study, the radiello-XAD samplers, which are assumed to only collect gas-phase compounds, often detected PFOS when the PUF sample did not (Figure 3). This aligns with previously reported particle-associated fraction values and multiple studies’ observations of PFOS being predominantly in the gas-phase.^28,45–47 Short chain PFCAs, such as PFBA, were generally only captured by the PUF samples (Figure 3). Other studies’ reporting of PFBA in the particle-phase and the higher particle-associated fraction for short chain PFCAs is a possible indication that the PUFs were mainly capturing the particulate-phase PFAS in this study.^28,47,48

3.3. Concentrations of PFAS across Michigan

Across all sampling sites, PFPrA, PFBA, and PFOS were the only ionic PFAS frequently detected (detection frequency > 50 %). In PUFs, the sum of these frequently detected PFAS (∑₃PFAS) concentrations ranged from 4.8 pg/m³ to 15.5 pg/m³ when detected. PFBA had the highest concentrations ranging from 6.7 pg/m³ to 14.8 pg/m³ when detected. In radiello-XAD samplers, the ∑₃PFAS concentrations ranged from 3.2 pg/m³ to 248 pg/m³ when detected. Of the volatile PFAS measured in the radiello-XAD samplers, 6:2 FTOH had the highest concentrations, ranging from 80 pg/m³ to 500 pg/m³ when detected (See SI for concentrations of all detected compounds).

One sampling site, Dearborn, had a ∑₁₅PFAS (sum of all detected compounds) that was approximately 20 times larger than the median ∑₁₅PFAS of all sampled locations. Dearborn is one of seven sites within the Detroit area of Michigan (Figure 2), yet no other site in the Detroit area detected such range of compounds at similar levels (Figure 5). Notably, the Dearborn sampling location lies on the perimeter of a large car manufacturing facility (See SI, Figure S3). Vehicle manufacturing is known to rely on the use of PFAS for a wide variety of applications, including use in their fuel systems, brake systems, electronics, batteries, and interior treatments.^49–51 The next nearest sampling location to the car manufacturing facility is another 1 km away, which likely explains the observed decreased concentrations the other sampling locations within the Detroit area.

Figure 5:

Concentrations of PFAS measured in radiello-XAD samplers at each site shown in pg/m³.

While this study presents the first use of the combined PUF disc and radiello-XAD sampler to measure atmospheric PFAS, previous studies have utilized PUF discs alone to measure outdoor concentrations.^27,52–54 A study by Liu et al., 2015, which deployed PUFs on house roofs around Shenzhen, China, reported atmospheric concentrations of 3.4 pg/m³ to 34 pg/m³ for a sum of 11 PFAAs.⁵⁵ Across those sampling sites, PFOA was often detected at the highest concentrations with PFOS and short chain PFCAs also frequently detected.⁵⁵ While total concentrations detected in PUFs are comparable, the limited detection of PFOA in this study is likely a result of the phasing out of PFOA by 2015 in the USA.⁵⁶

In comparison to PFOA, PFOS was frequently detected by the radiello-XAD samplers across all sampling sites. While measured concentrations were relatively low compared to other detected compounds, frequent detection of PFOS was unexpected given the phase-out in the USA almost 20 years ago. A study by Zhou et al., 2021, observed similar unexpected detection of PFOS, reporting it dominated in ambient fine particulate matter samples collected across North Carolina.⁵⁷ Possible sources of PFOS suggested in Zhou et al., 2021, may also be applicable here. Detected PFOS may be due to imported products from countries still using PFOS, the use of long shelf-life products made in the USA prior to the phase-out (e.g. PFOS-based AFFF), atmospheric formation from precursors, resuspension from contaminated media, or long-range atmospheric transport from other countries.^57–61

A study by Xia et al., 2024, measured atmospheric concentrations of PFAS around the Great Lakes in 2021 using sorbent impregnated PUF disks (SIP disk).³² For a sum of 41 PFAS, Xia et al., 2024, reported concentrations of 179 pg/m³ to 1146 pg/m³ for urban sites and 38 pg/m³ to 442 pg/m³ for rural sites, similar to the sum concentrations measured in this study by the combined radiello-XAD and PUF samplers.³² The most abundant PFAS measured across all sites were FTOHs, predominantly 6:2 FTOH, and PFCAs, which is consistent with the observations of this study and those reported in previous studies.^32,35,45 A similar study by Yao et al., 2025, measured atmospheric concentrations of PFAS in New Jersey using active sampling.⁶² Reported concentrations measured in both the particle phase and gas phase were similar to those measured in PUFs and the radiello-XAD samplers deployed in Michigan, but unlike this study, long chain PFCAs (C>6) were the most frequently detected.⁶²

Both the study by Xia et al., 2024, and Yao et al., 2025, report higher concentrations measured in the gas phase versus the particulate phase.^32,62 In general, the concentrations measured by the radiello-XAD samplers were 10 to 100 times higher than those measured by the PUFs, again indicating that gaseous PFAS were mainly captured by the radiello-XAD sampler, whereas the PUF predominantly captured particle-phase PFAS.

3.4. Resolving Matrix Interference Using High Resolution Mass Spectrometry (HRMS)

Two short chain PFCAs- PFBA and perfluoropropanoic acid (PFPrA)- were frequently detected in the PUF samplers (Figure 3). It is known that PFBA, along with other short chain PFAS, has been falsely reported in air and biological samples due to analytical method interferences.^63–66 These matrix interferences are particularly challenging to resolve for these compounds with only one major MS/MS transition.^63,66 Confirmation of these compounds and their reported concentrations is achieved through utilization of high resolution mass spectrometry (HRMS). Matrix interferences were observed within 0.1 Da of the compounds’ exact mass (SI Figure S1 and Figure S2). The use of HRMS enabled us to mass-resolve matrix interferences which wouldn’t have otherwise been possible with unit resolution. Moreover, the frequent detection of PFPrA in the PUF samples lends additional confidence in the detection and quantification of PFBA.

3.5. Correlations with Environmental Factors

The observed meteorological conditions (wind speed and temperature) across the sampling locations did not result in meaningful correlations with measured concentrations. This is likely in part due to the lack of available data for the deployment locations, and the small variability in observed wind speed and temperature over the sampling period across the sampling locations.

Moderate (0.3 ≤ r ˂ 0.6) positive correlations were found between ∑volatile PFAS concentrations and both population density (r =.46, p<.05) and industry density (number of industries potentially handling, using, or releasing PFAS within an area) (r=.39, p<.05). Slightly stronger correlations were observed between 6:2 FTOH concentrations and both population density (r=.52, p<.05) and industry density (r=.48, p<.05), possibly a result of 6:2 FTOH having the highest detection frequency of the volatile compounds. Industry density is strongly positively correlated with population density (r=.93, p<.05). This is likely attributed to a large number of the counted industries being WWTPs, which are built as a direct response to population. Population growth in areas is also linked to the availability to work in industrial facilities, and so a strong correlation between population and industries is expected.

Positive correlations between industry/population density and volatile PFAS concentrations are expected as many urban sources, such as WWTPs or landfills, industrial facilities, and homes are known to emit volatile PFAS into the environment.^13–15,18 Previous studies have also reported higher concentrations at urban sites.^32,67 For example, the study within the Great Lakes region by Xia et al., 2024, noted that the median ∑₄₁PFAS concentrations measured in air were significantly higher at their urban sampling sites compared to their remote sampling sites.³² Furthermore, an extensive study by Zhan et al., 2024, investigated the spatial patterns of organic contaminants using hierarchical cluster analysis with data from passive air sampler networks.⁶⁸ Their work found that elevated concentrations of FTOHs had a “population” signature (that is they were highly correlated with population density, more so than point sources), which is consistent with the correlations observed in this study.⁶⁸

Measured concentrations of ionic PFAS, whether in the PUFs or radiello-XAD samplers, did not result in meaningful correlations with population density or industry density. For example, PFBA concentrations measured in the PUF samples did not show significant correlations with population density (r = 0.1) or industry density (r = 0.15). A portion of ionic PFAS measured in the atmosphere can likely be attributed to the degradation of neutral PFAS.^5,6,67 Several studies have shown how photodegradation of an n:2 FTOH results in a range of PFCAs, such as PFNA and PFOA in the case of 8:2 FTOH or perfluoroheptanoic acid (PFHpA) and perfluorohexanoic acid (PFHxA) in the case of 6:2 FTOH.^6,18,69 However, studies have also noted that the efficiency of these reactions is only a small percentage, with the yield of PFOA from its precursors being in the range of 1% to 10% for example.^5,7,69 The low efficiency of these reactions coupled with the relatively long atmospheric half-lives of fluorotelomer alcohols (10–80 days) can be a likely explanation for a lack of significant correlations observed between precursors and PFAAs.^12,70 In the case of this study however, many PFAAs had lower detection frequencies, making it difficult to draw conclusions on correlations between compounds.

The higher detection frequency of PFPrA and PFBA may be a result of multiple possible sources. In the case of these shorter chain and ultra-short chain PFCAs, several precursors are known to have C3 and C4 degradation products. For example, atmospheric degradation of longer chain FTOHs (6:2 FTOH, 8:2 FTOH, 10:2 FTOH), and, while not target analytes in this study, 4:2 FTOH and 4:2 fluorotelomer acrylate can all result in both PFBA and PFPrA as products.^6,71,72 Another indirect source of PFPrA and PFBA are fluorinated gases such as hydrofluorocarbons (HFCs) and hydrochlorofluorocarbons (HCFCs), which are commonly used for refrigeration and air conditioning, and perfluoroethers, which were introduced as replacements for PFOA and PFOS.^73–77 HFCs, HCFCs, and perfluoroethers are known to atmospherically degrade and be a source of short chain PFAAs.^78,79 Alternatively, direct emissions of short chain PFCAs into the environment may also cause higher detection compared to other PFCAs, as their increased use as alternatives to the long chain compounds that have been phased out, such as PFOA.^80–83

4. Limitations and Implications

Future directions, particularly for Michigan’s understanding of ambient air concentrations of PFAS, may include more deployments at different times of the year to capture any seasonal variations that cannot be observed from this study alone. Seasonal variations in atmospheric PFAS concentrations are expected, mainly due to meteorological changes, and have been observed by previous studies conducted in other locations.^{29,35,84–87} However, seasonality has been reported to have different impacts on remote versus urban areas, and so a repeat study at the same sampling sites in a different season would be beneficial.⁸⁸ Alternatively, while spatial relationships were not found in this study, future deployments at different sites, reducing data gaps, could help identify if a spatial relationship is present.

The presence of PFAS in the atmosphere can have implications on other environmental media, such as surface water, largely through atmospheric deposition.⁷⁸ Previous studies have explored how different atmospheric deposition pathways, such as rain or snow, may play a significant role in contamination of water sources.^32,33 Beyond inhalation, atmospheric PFAS can therefore also impact human exposure through drinking water contamination. Simultaneous measurements of air and water samples, particularly at sites in close proximity to known point sources such as Dearborn, could show the impact of air concentrations on surface water, or vice versa. Further exploration could be conducted to investigate the implications of atmospheric PFAS on other environmental compartments. For example, through dry or wet deposition, atmospheric PFAS can impact soil contamination both locally and far away.^89–91 Previous studies have reported environmental contamination from industrial atmospheric emissions, such as Schroeder et al., 2021, who observed soil and groundwater contamination in conserved forest lands significant distances from emission sources.⁹¹ In Michigan, atmospheric emission sources such as those possibly in Dearborn, may have an impact on both local and distance soils, vegetation, and water bodies. Furthermore, the atmosphere is now considered a key transport medium for PFAS, and so PFAS detected in Michigan can have a wider impact on environments beyond the state and to more remote areas.^5,6,92–94

Overall, this study provides an expansive investigation into the ambient concentrations and distribution of atmospheric PFAS across Michigan. The ambient air concentrations were similar to those reported in other outdoor studies, all of which are generally lower than the concentrations measured in indoor environments. This study also reported higher detection frequencies and relatively higher concentrations of 6:2 FTOH, PFBA, PFPrA, and PFOS. Concentrations of the volatile compounds, specifically 6:2 FTOH, were significantly correlated to population and industry density.

This research demonstrated the novel combined passive sampler as an effective, easy-to-use, and affordable detection tool, highlighting possible future utilization for carrying out expansive studies. However, further calibration studies for the radiello-XAD samplers under different environmental conditions and the addition of reliable depuration compounds to the PUFs could reduce uncertainty in sampling rates for individual compounds. Despite this, the observed detection of specific compounds on each type of sampler, along with our current understanding of how the samplers function, adds to our understanding of in what state different PFAS may undergo atmospheric transport. In particular, PFOS was observed in the gas-phase, as captured predominantly by the radiello-XAD samplers, and PFBA in the particle-phase, as it was captured mostly by the PUFs. Furthermore, the analytical methods used, as in the case of PFPrA and PFBA, highlight the importance of deeper exploration with multiple approaches to increase confidence in reported concentrations, particularly for shorter chain compounds that can have a lot of matrix interference. The high detection frequency of short-chain PFCAs suggests that including measurements of ultrashort-chain PFAS in future atmospheric studies would be worthwhile and would offer a more comprehensive view of their presence in the environment. In summary, future studies such as this can help identify potential sources of environmental contamination, improve our understanding of background levels present even in remote areas, and contribute to our broader knowledge of the fate and transport of atmospheric PFAS.

Supplementary Material

Supplementary Information

Additional details on analytical methods, standards, quality control and assurance, instrument conditions, site information, and individual air concentrations are available in the attached excel sheet (tables) or document (figures and details).

Figure 4:

Concentrations of PFAS measured in PUF samples at each site shown in pg/m³.

Figure 6:

Heat maps showing industry density (number of industries potentially handling, using, or releasing PFAS within area surrounding the sampling site) (top left), population density (per km²) (top right), and ∑volatile PFAS (6:2 FTOH, 8:2 FTOH, 10:2 FTOH, MeFOSA, EtFOSA) in measured in the radiello-XAD samplers (pg/m³) (bottom).