Multi-year atmospheric concentrations of per- and polyfluoroalkyl substances (PFASs) at a background site in central Europe

Abstract

A total of 74 high volume air samples were collected at a background site in Czech Republic from 2012 to 2014 in which the concentrations of 20 per- and polyfluoroalkyl substances (PFASs) were investigated. The total concentrations (gas + particle phase) ranged from 0.03 to 2.08 pg m⁻³ (average 0.52 pg m⁻³) for the sum of perfluoroalkyl carboxylic acids (ƩPFCAs), from 0.02 to 0.85 pg m⁻³ (average 0.28 pg m⁻³) for the sum of perfluoroalkyl sulfonates (ƩPFSAs) and from below detection to 0.18 pg.m⁻³ (average 0.05 pg m⁻³) for the sum of perfluorooctane sulfonamides and sulfonamidoethanols (ƩFOSA/Es). The gas phase concentrations of most PFASs were not controlled by temperature dependent sources but rather by long-range atmospheric transport. Air mass backward trajectory analysis showed that the highest concentrations of PFASs were mainly originating from continental areas. The average particle fractions (θ) of ƩPFCAs (θ = 0.74 ± 0.26) and ƩPFSAs (θ = 0.78 ± 0.22) were higher compared to ƩFOSA/Es (θ = 0.31 ± 0.35). However, they may be subject to sampling artefacts. This is the first study ever reporting PFASs concentrations in air samples collected over consecutive years. Significant decreases in 2012–2014 for PFOA, MeFOSE, EtFOSE and ƩPFCAs were observed with apparent half-lives of 1.01, 0.86, 0.92 and 1.94 years, respectively.

1. Introduction

Per- and polyfluoroalkyl substances (PFASs, C_nF_2n+1^—R) are a diverse class of industrial chemicals. From an environmental fate perspective, PFASs can be classified in two categories: i) the perfluoroalkyl acids (PFAAs) and ii) their precursors. The PFAAs include perfluoroalkyl carboxylic acids (PFCAs, C_n-1F_2n-1–COOH) and perfluoroalkyl sulfonates (PFSAs, C_nF_2n+1–SO₃H) while the PFAA precursors include compounds such as perfluorooctane sulfonamides (FOSAs) and perflurooctane sulfonamidoethanols (FOSEs) (Buck et al., 2011; Wang et al., 2017). Given their chemical properties, they have been widely used in industrial manufactured products such as inks, varnishes, waxes, fire-fighting foams, repellents for textiles, leather and paper products (Lai et al., 2016; Paul et al., 2009). It is estimated that the European emissions of C8-C13 PFASs were only 10–26% of the global budget in 1950–2010 which was dominated by North America and Asia (Armitage et al., 2009). Some PFASs have raised concerns because of their persistence, bioaccumulation and potential toxicity (Dreyer, 2010; Giesy and Kannan, 2002). These compounds, which are cycling pollutants (Armitage et al., 2009; Stemmler and Lammel, 2010; Thackray et al., 2020) have been globally detected in different environmental media such as water, air and biota as well as in human and in food (Giesy and Kannan, 2001; Kannan et al., 2004; Ostertag et al., 2009; Wang et al., 2015; Yamashita et al., 2005). Volatile PFASs precursors were also found in remote regions as a consequence of long-range atmospheric transport (LRAT) (Wang et al., 2015) while ionic PFAAs are transported via the aquatic system and in marine aerosols (Benskin et al., 2012; Karásková et al., 2018) as well as from air-sea exchange (Xie et al., 2013). Among the PFASs, only perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) and their salts so far have been listed under the Stockholm Convention on Persistent Organic Pollutants (POPs) (UNEP, 2019, 2017). On the other hand, other PFCAs, PFSAs and FOSA/Es have also been restricted under national or regional regulatory or voluntary frameworks (ECHA, 2019; EPA, 2006; OECD, 2015). However, it is unclear whether these regulatory frameworks are efficient in mitigating the harmful effects of PFASs on both human health and the environment.

PFASs, as other semi-volatile organic compounds (SVOCs), partition in the air between the gaseous and the particle phases (Ahrens et al., 2012). This partitioning is crucial as it will influence their removal pathways, which are phase-specific, such as wet and dry deposition, photolysis or reaction with OH radicals (Ellis et al., 2003). Therefore, gas-particle partitioning of SVOCs will affect their atmospheric residence time and their potential for LRAT (Cai et al., 2012). Over the last decades, it has been shown that the gas-particle partitioning of SVOCs is influenced by the physico-chemical properties of the compounds (e.g. vapour pressure), the outdoor meteorological conditions (i.e. temperature and relative humidity) and the quantity and type of particulate matter (Ahrens, 2011; Lohmann and Lammel, 2004; Pankow, 1987). However, this general knowledge seems to not be applicable to PFASs. Indeed, because of their relative high vapor pressure, many PFASs are predicted to be mainly in the gaseous phase, while some of them were significantly found in the particle phase (Ahrens et al., 2012). Similarly, no influence of ambient temperature on their gas-particle partitioning has been reported (Ahrens et al., 2012). Therefore, more knowledge on PFASs gas-particle partitioning is needed.

The current knowledge on atmospheric PFASs is mainly restricted to coastal and oceanic areas (Cai et al., 2012; Dreyer, 2010; Shoeib et al., 2010; Wang et al., 2015, 2014) or to sites close to PFASs sources such as manufacturing facilities (Chen et al., 2018), while limited information is available for continental background/rural sites (Barber et al., 2007). However, such data are crucial to understand the environmental fate of PFASs. Similarly, most of the studies performed on atmospheric PFASs relied on short sampling campaigns (i.e. few weeks up to a year) whereas multi-year variations of PFASs have never been addressed. Given the fact that air is a medium reacting rapidly to differences in primary emissions (Harrad, 2015), such data would be useful to evaluate the efficiency of the various regulations applied to these compounds.

The main goal of this study is to provide novel data on the atmospheric levels of PFASs at a background site in Czech Republic. More specifically, their seasonal and multi-year variations as well as the influence of LRAT on their atmospheric concentrations will be investigated.

2. Methodology

2.1. Air sampling

Air was sampled from January 2012 to December 2014 at the National Atmospheric Observatory Košetice (49°34’24”N, 15°04’49”E) in Czech Republic. This site is part of the European Monitoring and Evaluation Programme (EMEP) network (Holoubek et al., 2007), besides others. This sampling site, located in an agricultural region in central Czech Republic with limited anthropogenic sources, as well as the sampling techniques have been already described in different studies investigating other POPs and SVOCs in air (Degrendele et al., 2020, 2018, 2016; Shahpoury et al., 2015). A high-volume air sampler (Digitel DH77 with PM₁₀ pre-separator) was used to collect air samples on a biweekly basis resulting in an average sampled volume of 5282 m³ corresponding to a flow rate of about 31.4 m³ h⁻¹ for 7 days sampling duration. Quartz fibre filters (QFFs, QM-A, 150 mm, Whatman, UK, pore size of 2.2 μm) and polyurethane foam plugs (PUFs, two in series, T3037, 110 × 50 mm, 0.030 g cm⁻³, Molitan a.s., Czech Republic) were used to collect the particle and gas phase, respectively. Prior sampling, PUFs were pre-cleaned with acetone and dichloromethane for 8 h each via Soxhlet extraction. After sampling, all QFFs and PUFs were wrapped in aluminum foil and stored in a freezer at −18 °C until analysis. Twenty-six samples were collected each year. However, four samples were discarded due to some failures (e.g. sudden change in the flow rate, electrical power shutdown) during sampling (Table S1 in the Supplement).

2.2. Sample preparation and analysis

The techniques used in this study have been already published elsewhere (Karásková et al., 2018) and are only briefly summarized here. An automated extractor unit (B-811, Büchi, Switzerland) was used to extract all QFFs and PUFs using 5 mM ammonium acetate in methanol. Prior extraction, all samples were spiked with surrogate standards (M8PFOA, M8PFOS, Wellington Laboratories Inc., Canada). The concentrated extracts were filtered (nylon membrane, 13 mm diameter and 0.45 μm pore size) and transferred into polypropylene centrifuge tubes (Alpha Laboratories, UK) and they were concentrated to 0.5 mL and diluted with 0.5 mL (5 mM ammonium acetate in water). Internal standards (MPFBA, MPFHxA, MPFOA, MPFNA, MPFDA, MPFDoDA, MPFHxS, MPFOS, dMeFOSA, dMeFOSE, Wellington Laboratories Inc, Canada) were added to the samples prior the analysis. All samples were analyzed using high performance liquid chromatography (HPLC) (Agilent 1290, Agilent Technologies, Palo, Alto, California, USA) connected to a mass spectrometer (QTRAP 5500, AB Sciex, Foster City, California, USA), using isotope dilution method. Further information can be found elsewhere (Karásková et al., 2018). The target compounds were ten PFCAs (PFPeA, PFHxA, PFHpA, PFOA, PFNA, PFDA, PDUnDA, PFDoDA, PFTrDA and PFTeDA), five PFSAs (PFBS, PFHxS, PFHpS, PFOS and PFDS) and five FOSA/Es (FOSA, MeFOSA, EtFOSA, MeFOSE, EtFOSE) (see full names in Table S2).

2.3. Quality assurance and quality control (QA-QC)

A total of 11 field blanks and ten laboratory blanks were analyzed as per samples. Some of the target compounds were found in non-negligible quantities (i.e. > 20% of average mass in samples) in the field blanks which could be caused by the unavoidable use of Teflon parts in extraction apparatus. Similar blank issues contaminated were previously reported in atmospheric PFASs (Barber et al., 2007; Rauert et al., 2018). Limits of quantifications (LOQs) were identified as the maximum of the instrumental LOQs and the average concentrations in the field blanks plus three times their standard deviations. The reported concentrations have been corrected by subtracting the average of field blanks. The average recoveries of surrogate standards for all samples were 59% and 70% for M8PFOA and M8PFOS, respectively. Moreover, spike recovery tests were performed and the recoveries were ranging from 50.8% (PFTeDA) to 128% (PFPeA) on PUFs and from 101% (PFTrDA) to 128% (PFTeDA) on QFFs.

2.4. Meteorological data and air mass origin

Different meteorological parameters (i.e. temperature, precipitation, wind speed, wind direction and relative humidity (RH)) were provided by the Observatory. Additionally, the atmospheric boundary layer (ABL) height was retrieved from the Global Data Assimilation System (GDAS).

In order to evaluate the relationship between LRAT and PFASs atmospheric concentrations, an analysis of air masses using backward trajectories of selected samples was performed. The Lagrangian particle dispersion model FLEXPART (Stohl et al., 2010, 2005) was used to identify the origin of the air masses of selected samples. For this, meteorological data (0.5° and 3 h resolution, 91/137 vertical levels retrieved from ECMWF, http://www.ecmwf.int, last access: 20 January 2020) were used as inputs. In this study, for each case investigated, we simulated the release of 100,000 particles between 0 and 200 m a.g.l. which were tracked 5 days backward in time.

3. Results and discussion

3.1. PFASs total concentrations

In this study, individual PFCAs, PFSAs and FOSA/Es were detected in 32–76%, 47–93% and 35–58% of all samplesanalyzed (Table S3). The total concentrations of ΣPFCAs ranged from 0.03 to 2.08 pg m⁻³ with an average value of 0.52 pg m⁻³ (Table S3, Figure 1). These concentrations were slightly lower than those observed at a suburbanbackground site located about 100 km away, reporting concentrations ranging from 4.79 to 13.78 pg m⁻³ with an average of 8.42 pg m⁻³ (Karásková et al., 2018). ΣPFCAs concentrations were dominated by PFHxA, PFPeA and PFDA which accounted on average for 32%, 20% and 17%, respectively (Figure 2a). PFOA, which is included in the Stockholm Convention was found in only 32% of the samples with an average concentration of 0.14 pg m⁻³. This is lower than what was found in other studies in Europe and elsewhere (Ahrens et al., 2012; Barber et al., 2007; Liu et al., 2015).

Figure 1:

Total concentrations (in pg m⁻³) of ΣPFCAs, ΣPFSAs and ΣFOSA/Es

Figure 2:

Contribution of individual compounds to gaseous, particle and total concentrations of ΣPFCAs (a), ΣPFSAs (b) and ΣFOSA/Es (c)

The total concentrations of ΣPFSAs ranged from 0.02 to 0.85 pg m⁻³ with an average value of 0.28 pg m⁻³ (Table S3, Figure 1). These concentrations are similar with those measured at a semi urban site in Toronto, Canada (Ahrens et al., 2012). PFSAs were dominated by PFOS and PFBS, which contributed on average for 45% and 30%, respectively (Figure 2b). This contribution profile has been reported in many other studies (Ahrens et al., 2012; Karásková et al., 2018; Rauert et al., 2018) and is consistent with the fact that PFBS is one of the main replacement compound of PFOS (Wang et al., 2009).

Regarding ΣFOSA/Es, their total concentrations ranged from ND to 0.18 pg.m⁻³ with an average value of 0.05 pg.m⁻³ (Table S3, Figure 1), which is about one order of magnitude lower than those found for ΣPFSAs, as previously reported elsewhere (Rauert et al., 2018). The ΣFOSA/Es concentrations reported here are generally lower than those reported in other studies (Table S4). ΣFOSA/Es were slightly dominated by EtFOSE contributing on average for 34% (Figure 2c).

3.2. Factors affecting the inter-sample variations

The short temporal variations of SVOCs atmospheric concentrations can be influenced by the meteorological conditions along transport (e.g. temperature, precipitation) but also by the advection of air masses or the efficiency of removal processes (degradation and deposition) (Degrendele et al., 2020, 2018).

The influence of meteorological parameters (i.e. precipitation amount, ambient temperature, wind direction, wind speed, ABL height and RH) on PFASs concentrations was evaluated using simple correlation and multiple regression analysis. In general, no strong correlations were observed (Table S5). However, it is worth noting that the coefficient related to the ABL height in the multiple regression analysis was statistically significant (p<0.05) for PFHxA, PFOA, PFOS, MeFOSE, ΣPFCAs and ΣPFSAs (Table S6). This is consistent with the overall knowledge that ABL height is a crucial meteorological parameter affecting the levels of SVOCs including PFASs (Li et al., 2017).

Similarly, the influence of ambient temperature on PFASs concentrations was assessed using the Clausius–Clapeyron equation (see Supplement for details, Table S7). For all PFASs, except for PFPeA, PFHpA and FOSA, no significant (p > 0.01) correlations between the natural logarithm of partial pressure of PFASs and the inverse of ambient temperature were found. This support the hypothesis that the gas phase concentrations of most PFASs were not controlled by sources affected by the temperature (e.g. revolatilisation from surfaces or degradation), but rather by advection/LRAT (Hoff et al., 1998), which is consistent with findings from previous studies (Dreyer, 2010; Wang et al., 2015, 2014).

The influence of LRAT was assessed using the analysis of air masses, as described in section 2.4, in order to identify potential sources. When considering the 20 samples with the lowest PFASs concentrations (i.e. < 0.45 pg m⁻³), 14 of them were mainly originating from the West (i.e. air that has passed through the Atlantic Ocean or the North Sea, Figures 3a and 3b) while the remaining were mainly originating from the East (i.e. air that has passed through Eastern continental regions). On the other hand, 12 out of the 20 samples with the highest PFASs concentrations (i.e. > 1.10 pg m⁻³) originated mainly from continental areas (Figures 3c and 3d) while the remaining had a maritime origin (i.e. passed through the Mediterranean Sea, Atlantic Ocean, North Sea or Baltic Sea). The highest PFASs concentrations mainly observed from different directions with the highest residence time in continental areas suggests a homogeneous continental emission source of PFASs, as previously found for other POPs (Degrendele et al., 2018). It is consistent with the fact that PFASs are used in the industry of flame retardants and can be emitted during the incineration of products or wastes containing PFASs (Paul et al., 2009; Prevedouros et al., 2006). Therefore, besides the fact that maritime sources are actually possible for PFASs as found in other studies (Cai et al., 2012; Shoeib et al., 2010), this source seems to be negligible compared to continental sources, as it was already found in China (Lai et al., 2016). However, we should keep in mind that given the low PFASs levels found in this study, European air is relatively clean in terms of PFASs compared to other regions of the world, as previously reported (Wang et al., 2018).

Figure 3:

Selected examples of 5 days backward trajectories of the 20 samples with the lowest (a and b, corresponding to samples collected on 4–11 December 2013 and 7–14 May 2014, respectively) and highest (c and d, corresponding to samples collected on 13–20 February 2013 and 9–16 October 2013, respectively) ΣPFASs concentrations

A recent study suggested that the ratio of long-chain to short-chain PFCAs could improve the knowledge on precursor chemistry (Thackray et al., 2020). Indeed, a strong gradient in the long-chain to short-chain ratio moving from more-polluted to less-polluted air masses could be the signature of atmospheric formation of PFCAs via the degradation of precursors (Thackray et al., 2020). In this study, the long-chain to short-chain ratio for the 20 samples with the lowest PFASs concentrations were 0–27.8 (average of 5.44 ± 8.43) while those for the highest PFASs concentrations were 0.45–57.6 (average of 14.3 ± 18.6). However, the differences were not statistically significant (p > 0.05), which do not confirm this hipothesis. Further data are needed to understand the influence of air masses on PFCAs profile.

3.3. Gas-particle partitioning in air samples

Among the different sub-classes of PFASs investigated, FOSA/Es showed the lowest particle fraction (θ) followed by short-chains PFCAs, and then long-chains PFCAs and PFSAs (Figures 4 and S1).

Figure 4 :

Average measured particle fraction (θ_measured) of PFASs

PFCAs were generally consistently found both in the particle and gas phases (Figure 4). PFOA had lower particle fraction (average 0 = 0.14 ± 0.33) compared to other PFCAs (average 0 ranged from 0.30 ± 0.46 to 0.65 ± 0.48, Figure 4). This repartition between both phases have also been reported in other studies (Ahrens et al., 2012; Karásková et al., 2018). Regarding the long chains, the particle fraction generally increased with the length of the carbon chain (Figure 4). Similar to the PFCAs, PFSAs were detected consistently in both phases. However, an important variation of the particle fraction was found among the individual compounds (Figure 4). Indeed, PFDS was principally found in the gas phase (θ = 0.12 ± 0.29), PFOS in the particle phase (θ = 0.80 ± 0.29) while the remaining compounds were found distributed across both phases (average θ ranged from 0.38 ± 0.44 to 0.65 ± 0.48). This large variation of θ was also reported by Ahrens et al. (2012) who found PFOS and PFDS only in the particle phase (θ = 1.00) and PFHxS only in the gas phase (θ = 0). Regarding ƩFOSA/Es, these were predominantly detected in the gas phase. The particle fraction of individual FOSA/Es ranged from 0.03 ± 0.09 (FOSA) to 0.22 ± 0.38 (EtFOSE) (Figure 4). This predominance in the gas phase had also been highlighted in other studies (Ahrens et al., 2012; Barber et al., 2007; Karásková et al., 2018; Wang et al., 2014). Moreover, no significant differences between the particle fraction of FOSAs and FOSEs were found, unlike what has been reported by Ahrens et al. (2012).

Significant (p < 0.05) seasonal variations of the particle fraction, with higher particle fractions in winter (i.e. December, January and February) compared to summer (i.e. June, July and August), was observed only for individual PFCAs (Figure 4) such as PFPeA and PFHpA. Moreover, PFTrDA and MeFOSA also showed seasonal variations with higher particle fractions in summer (Figure 4). It is unclear why some PFASs behave differently. Higher particle fractions of FOSA/Es in winter had been previously reported (Jahnke et al., 2007).

However, we should keep in mind that the reported particle fractions of PFASs from this study may be affected by the sampling design. Indeed, sampling artifacts such as blow-on, blow-off, breakthrough, and degradation may occur for SVOCs when using conventional high volume sampling techniques (Ahrens et al., 2011; Arp and Goss, 2008; Melymuk et al., 2014; Wang et al., 2014). Moreover, given the large volumes sampled in this study (V = 5282 m³ on average), breakthrough may have occurred for the most volatile PFASs as it was previously observed (Barber et al., 2007). Ahrens et al. (2012) have done an interesting study comparing the particle fraction of PFASs using high volume air sampler equipped with and without a diffusion denuder. They reported significantly higher particle fractions of most PFASs investigated, except FOSA/Es, when using a high volume air sampler which is not equipped with a denuder, as in our study. However, in terms of concentrations, the PFASs air concentrations for both the gas and particle phases agreed within a factor of four between the two sampling techniques (Ahrens et al., 2011). Finally, the adsorbent used in this study (i.e. PUF) may also affect the reported particle fractions as it is known that for polar compounds such as PFASs, XAD has a stronger sorbing capacity (Dobson et al., 2006).

3.4. Seasonal and inter-annual variations

To the best of our knowledge, this is the first study reporting PFASs concentrations in several consecutive years. This type of data are crucial to evaluate the effectiveness of international regulations. Following what has been previously applied for other POPs and SVOCs (Degrendele et al., 2020, 2018), the harmonic regression presented in Equation (1) was used (Venier et al., 2012).

$$\ln {\mathrm{C}}_{\mathrm{i}}={\mathrm{a}}_0+{\mathrm{a}}_1\sin (\mathrm{zt})+{\mathrm{a}}_2\cos (\mathrm{zt})+{\mathrm{a}}_3\mathrm{t}$$ (1)

where C_i is the total concentration of individual PFAS, t is the time of sample collection, z = 2π/365.25 is a coefficient that fixes the periodicity to one year, a₀ is an intercept that rectifies the units, a₁ and a₂ are the harmonic regression coefficients that describe the seasonal variations and a3 is a first-order rate constant (in days⁻¹) that describes the multi-year variations. The apparent halving/doubling time (τ_1/2) represents the time needed to reduce or increase the PFAS concentrations to half or twice their initial values. Here, it was calculated from a₃ as:

$${\mathrm{\tau }}_{1/2}=\frac{\ln 2}{a_3}$$ (2)

For most of the PFASs, the coefficients a₁ and a₂, describing the seasonal variations, were not statistically significant (Table S8), which is consistent with the large variations observed in the winter-to-summer ratios of their concentrations (Table S9). Similarly, no influence of ambient temperature on PFASs concentrations was previously reported in Northwest Europe (Barber et al., 2007) suggesting that the influence of LRAT on the PFASs concentrations is more pronounced than temperature-dependent processes such as air-surface exchange.

Regarding the annual variations, the coefficient a₃ of PFHpA, PFOA, PFTrDA, MeFOSA, EtFOSA, EtFOSE, MeFOSE and ƩPFCAs were significant (p < 0.05, Figure 5, Table S9). Among these compounds, PFOA, MeFOSE, EtFOSE and ƩPFCAs showed a decreasing trend with a halving time of 1.01, 0.86, 0.92 and 1.94 years, respectively. On the other hand, an increase was observed and doubling times of 0.48, 2.55, 1.18, 0.63 and 0.43 years were found for PFHpA, PFTrDA, PFDS, MeFOSA and EtFOSA, respectively (Figure S2, Table S9).

Figure 5:

Multi-year of PFASs for which statically significant decreases in 2012–2014 were found

The significant decreases of PFOA, MeFOSE, EtFOSE and ƩPFCAs could reflect a lower use of these compounds at the global scale. This is particularly true for PFOA which has recently been added in the Stockholm Convention (UNEP, 2019). However, no significant (p > 0.05) variations was observed for PFOS, which was included in the Stockholm Convention in 2009, prior to the one of PFOA (UNEP, 2017). This unexpected result could suggest a stagnation of the degrease in primary emissions during the time period investigated due to the many specific exemptions of production and use of PFOS listed in the Stockholm Convention (UNEP, 2017) and reported for specific countries such as Brazil (Rauert et al., 2018).

In the case of PFHpA and PFDS, results from a previous study analyzing PFASs in different building materials from Czech Republic can explain why an increasing trend was found. Indeed, high levels of PFHpA was previously found in wooden building materials from the Czech Republic, exceeding in several samples the levels found for PFOA (Bečanová et al., 2016). Similarly, PFDS was found in all electronic waste samples analyzed suggesting its use either as a side product during the PFOS production either as an alternative to PFOS (Bečanová et al., 2016). Therefore, the increasing concentrations of PFHpA and PFDS measured at our background site may reflect a shift from the usage of C8-compounds to shorter chains, more pronounced for PFCAs than for PFSAs. In the case of PFTrDA, MeFOSA and EtFOSA, it is unclear why an increasing trend was found and longer atmospheric monitoring is needed to further confirm this increasing trend.

4. Conclusions

A total of 20 PFASs were investigated in air samples at a background site in Czech Republic in 2012–2014. This study suggested that the atmospheric concentrations of most PFASs were not controlled by meteorological parameters but rather by LRAT. Indeed, the LRAT analysis performed with FLEXPART highlighted that the highest PFAS concentrations were generally associated with continental areas, which suggests a rather homogeneous continental emission source. Regarding the gas-particle partitioning, PFCAs and PFSAs were consistently found in both phases while FOSA/Es were mainly present in the gas phase. However, these findings may be subject to sampling artefacts. In this study, which is the first ever reporting multi-year atmospheric data of PFASs, significant decreases of PFOA, MeFOSE and EtFOSE were found with apparent half-lives of 1.01, 0.86 and 0.92 years, respectively, highlighting their long-term disappearance from the atmosphere. These decreases reflect the efforts made to reduce primary emissions of these compounds at the global scale and therefore highlight the effectiveness of international treaties such as the Stockholm Convention. However, the absence of a significant trend or even apparent positive concentration trends observed for the other PFASs suggest that the effectiveness is at least incomplete and such efforts should pursue. This linkage highlights the need of multi-year data of atmospheric PFASs in Europe but also for other regions worldwide. Finally, we should keep in mind that about 200 other PFASs are currently on the global market for intentional uses. However, their environmental fate and possible toxic effects have not yet been assessed. In order to protect humans and the environment from these possibly harmful chemicals, future research involving analytical techniques able to target a larger group of analytes is needed.

• PFASs concentrations were governed by long-range atmospheric transport

• Significant decreases were found for PFOA, MeFOSE, EtFOSE

• Significant increases were found for PFHpA, PFTrDA, PFDS, MeFOSA and EtFOSA