Degradation of Fluorotelomer-Based Polymers Contributes to the Global

Article

pubs.acs.org/est

Degradation of Fluorotelomer-Based Polymers Contributes to the Global Occurrence of Fluorotelomer Alcohol and Perﬂuoroalkyl Carboxylates: A Combined Dynamic Substance Flow and Environmental Fate Modeling Analysis † ‡ † † ‡ Li Li,*, , Jianguo Liu,*, Jianxin Hu, and Frank Wania † College of Environmental Sciences and Engineering, Peking University, 5 Yiheyuan Road, Beijing 100871, PR China ‡ Department of Physical and Environmental Sciences, University of Toronto Scarborough, 1095 Military Trail, Toronto, Ontario M1C 1A4, Canada

*S Supporting Information

ABSTRACT: Using coupled dynamic substance flow and environmental fate models, CiP-CAFE and BETR-Global, we investigated whether the degradation of side-chain fluorotelomer-based polymers (FTPs), mostly in waste stocks (i.e., landfills and dumps), serves as a long-term source of fluorotelomer alcohols (FTOHs) and perfluoroalkyl carboxylates (PFCAs) to the global environment. The modeling results indicate that, in the wake of the worldwide transition from long- chain to short-chain products, in-use stocks of C8 FTPs will peak and decline afterward, while the in-use stocks of C6 FTPs, and the waste stocks of both FTPs will generally grow. FTP degradation in waste stocks is making an increasing contribution to FTOH generation, the bulk of which readily migrates from waste stocks and degrades into PFCAs in the environment; the remaining part of the generated FTOHs degrade in waste stocks, which makes those stocks reservoirs that slowly release PFCAs into the environment over the long run because of the low leaching rate and extreme persistence of PFCAs. Short-chain FTPs have higher relative release rates of PFCAs from waste stocks than long-chain ones. Estimates of in-use and waste stocks of FTPs were more sensitive to the selected lifespan of finished products, while those of the emissions of FTOHs and PFCAs were more sensitive to the degradation half-life of FTPs in waste stocks. Our preliminary calculations highlight the need for environmentally sound management of obsolete FTP-containing products into the foreseeable future.

■ INTRODUCTION arrive at perfluoroalkyl carboxylates (PFCAs) as ultimate 5−7 Constituting almost 80% of the market of fluorotelomer-based degradation products. During the degradation, both the 1 fl intermediate and terminal degradation products may demon- substances worldwide, side-chain uorotelomer-based poly- ff 8−10 mers (FTPs) have been applied as durable water repellents strate adverse environmental e ects. (DWRs) on a wide range of finished textiles, fabrics, carpets, Despite the consensus that FTP degradation contributes to and garments2,3 and as oil and grease repellents in paper and the occurrence of FTOHs and PFCAs worldwide, the magnitude and temporal evolution of the problem have not packaging industries as well as other miscellaneous applica- 11 3 yet been well-elucidated and are thus still being contested. tions. The FTPs provide continuous water, oil, and stain 11−14 resistances for commercial finished products throughout the Several earlier modeling studies have attempted to product lifespans, during which FTPs might migrate into the evaluate the future worldwide releases of FTOHs and PFCAs from FTP degradation; however, these studies possessed environment because of abrasion and weathering. Afterward, a ff considerable amount of FTPs enters the waste stream and di erent scopes and relied on distinct assumptions and accumulates in waste stocks such as landfills and dumps, where methodologies. First, there is a lack of consistent, holistic consideration of mass flows of FTPs in both (i) in-use stocks aged FTPs undergo degradation on the time scale of decades or fi longer to generate various per- and polyfluoroalkyl substances during product service life and (ii) waste stocks (i.e., land lls (PFASs).4 Recent experimental studies have demonstrated that and dumps) during the waste-disposal phase. For instance, this process comprises a series of sequential step-wise transformations, which first form nonpolymeric fluorotelomer- Received: August 10, 2016 based substances like fluorotelomer alcohols (FTOHs) Revised: February 24, 2017 followed by a variety of immediate degradation products such Accepted: March 17, 2017 as saturated and unsaturated fluorotelomer carboxylates and Published: March 17, 2017

Figure 1. Transformation of FTPs, FTOHs, and PFCAs in the environment and waste stocks [R′ = H or methyl group; R = H or (meth)acrylic group]. Arrows in solid lines denote the substance flows quantified in this work: estimating stocks of FTPs, and annual releases of degFTOHs and deg-degPFCAs, are the main objectives (denoted as blue and red shadings); annual releases of resFTOHs, deg-resPFCAs, and impPFCAs are also calculated for comparison. Arrows in dashed lines represent substance flows not quantified. The box “other PFCA sources” includes all PFCAs sources in Wang et al.18 other than the three considered here.

Wang et al.11 assumed an immediate end of FTOH releases given to how and why those variables affect processes once in-use stocks of FTPs are depleted because they throughout the product lifecycle (e.g., in-use and waste stocks) postulated that all obsolete FTP-containing finished products and in the environment (e.g., FTP degradation in waste stocks will be “properly treated (i.e., safely landfilled or incinerated) versus after being released into the environment). In this and no longer available for degradation”.11 This assumption situation, process-oriented models can contribute to a better contrasts with the stated expectation that FTP degradation in mechanistic understanding of the complicated relationship landfills “potentially constitutes a large long-term environ- between key parameters and the estimated release of generated mental load”4 for these compounds. In fact, the potential for FTOHs and PFCAs. Such an approach is also helpful for degradation of FTPs in waste stocks has been confirmed identifying the most influential parameters requiring accurate experimentally,4,15 and considerable releases of FTOHs and quantification in future studies. fl PFCAs via landfill leachates and gases have been observed in a In this contribution, we combine a dynamic substance ow range of field studies.16,17 Coping with this situation analysis model, CiP-CAFE, and an environmental fate model, necessitates a better “understanding of the mass flow of side- BETR-Global, to mechanistically simulate the temporal chain fluorinated polymers during their whole life-cycle, evolution of in-use and waste stocks of FTPs, and the fi ” 11 environmental releases of FTOHs and PFCAs from the including in land lls , as stated by Wang et al. Second, fl there are substantial variability and uncertainty associated with degradation of FTPs. The in uence of variations in LSs and important input parameters. For example, (i) the lifespan of HLs on model predictions is explored with four scenarios. We finished products [hereafter “product lifespan”, (LS)], and (ii) seek to preliminarily characterize the current and future the degradation half-life of FTPs in the environment or waste contributions of FTP degradation, particularly during the stocks [hereafter “degradation half-life”, (HL)], have been waste-disposal phase, to the releases of FTOHs and PFCAs fi worldwide. Findings in this study may provide a long-term identi ed as two key parameters determining the contribution fl of FTP degradation.11 However, the LS adopted in earlier overview of the stocks and mass ows of FTPs throughout the 18 13 lifecycle as well as complement the current understanding of studies ranges from 10 years to 50 years, while the HL sources of FTOHs and PFCAs. derived from different degradation experiments spans 2 orders of magnitude from 10−17 years6 to 1200−1700 years.5 How different LS and HL values influence global annual FTOH ■ METHODS releases from FTP degradation has been studied previously Applications, Substances, and Terminology. Although using empirical emission factors.11 Limited consideration was fluorotelomer-based substances have found use as ingredients

4462 DOI: 10.1021/acs.est.6b04021 Environ. Sci. Technol. 2017, 51, 4461−4470 Environmental Science & Technology Article

Figure 2. Schematic of the modeling strategy in this study. Green boxes indicate model outputs, while blue boxes indicate data collected from the literature or calculated from model outputs. Red diamonds indicate a calculation using models.

(see the terminology in Text S1) in a multitude of consumer statistics. Because technical PFASs are usually present and products, for simplification, three major applications (APs) can consumed in above applications as mixtures of homologues be roughly categorized on the basis of lifespan character- with different chain-lengths or derivatives with various istics:18,19 FTPs serve as DWRs on finished textiles, fabrics, functional groups, we defined “equivalents” to collectively carpets, and garments (AP1), while nonpolymeric fluoro- describe a series of similar PFASs with the same featured telomer-based derivatives serve as surfactants for treating moiety but without considering their differences in molecular consumer products (representing all uses with continuous weights, physicochemical properties, and degradation kinetics. releases throughout lifespan) (AP2) and additives in aqueous In this study, the following three categories of PFAS equivalents film forming foams (AFFFs, representing all uses with both were considered (Figure 1): accidental releases at accidents and intensive discharge at the (I) side-chain FTP equivalents (respectively, 4:2, 6:2, 8:2, end of shelf life) (AP3). Here, we did not consider FTP use in 10:2, and 12:2 homologues were considered in this paper and packaging industries due to absence of market study), which refer to a collection of side-chain

4463 DOI: 10.1021/acs.est.6b04021 Environ. Sci. Technol. 2017, 51, 4461−4470 Environmental Science & Technology Article

fluorotelomer-based acrylate, methacrylate, urethane, and processes termed “landfill (WD1)” and “dumping and simple other polymers. FTP-equivalents are used as ingredients landfill (WD3)” (Text S2.1). in DWRs in AP1 (denoted as FTPs) and have the To feed the CiP-CAFE model, we first calculated the annual potential to degrade into FTOHs and further PFCAs. production of FTPs on a homologue basis in individual CiP- (II) FTOH equivalents (respectively, 4:2, 6:2, 8:2, 10:2, and CAFE regions. The global annual production of total technical fl 18,19,27 12:2 homologues), which refer to a collection of (i) uorotelomer-based substances (Figure S1) were split fl into FTPs (for use in AP1) and nonpolymeric substances (for FTOHs and (ii) uorotelomer-derived nonpolymers 1 (e.g., fluorotelomer acrylates, FTA), the degradation of use in APs2 and 3) based on a reported ratio of 80% to 20%. 20 Furthermore, because technical fluorotelomer-based substances which generates FTOHs as intermediates. FTOH- ff equivalents can be released into the environment as are mixtures of di erent homologues of ingredients, residuals residuals in consumer products in the three APs and impurities, we divided these mixtures according to literature-reported homologue composition (for ingredients) (denoted as resFTOHs, Figure 1) and as degradation and contents (for residuals and impurities) in long-chain (C8- products from the degradation of FTPs (denoted as based) and short-chain (C6-based) products (Table S1). To degFTOHs, Figure 1) in both waste stocks and the reflect a worldwide transition from long-chain (C8-based) environment. products to their short-chain (C6-based) alternatives since (III) PFCA equivalents (respectively, C4−C12 homologues), fl 2006, we assumed that the fraction of long-chain products in which refer to both per uoroalkyl carboxylic acids and total fluorotelomer-based products decreased linearly from corresponding carboxylates. In this study, we consider 100% before 2005 to 0% after 2016. This simplified assumption releases of PFCA-equivalents into the environment as could underestimate the amounts of long-chain products in impurities in consumer products in the three APs developing countries (e.g., in China),27 as domestic production (denoted as impPFCAs, Figure 1) and as degradation and use of long-chain products are still ongoing in these products from both resFTOHs (denoted as deg- countries. Those in developed countries can be underestimated resPFCAs, Figure 1) and degFTOHs (denoted as deg- as well if they continued importing long-chain products from degPFCAs, Figure 1) in waste stocks and the environ- developing countries; however, the underestimation seems ment. quite moderate because, in most cases, the imports are The main objectives of this study are (i) to simulate the prohibited in developed countries in compliance with their temporal evolution of in-use and waste stocks of FTPs in AP1 regional or national trade regulations. For example, a sampling − “ and (ii) to investigate the contribution of FTP degradation to campaign in Norway during 2012 2013 indicated that the formation/release of degFTOHs and deg-degPFCAs (Figure emissions from consumer products imported from China account for 1.5[%] of the discharges of PFOA to wastewater 1). In addition, we also simulated the direct releases of fl ” 28 resFTOH residuals and impPFCA impurities, as well as the in uents and 0.3[%] of the emissions of 8:2 FTOH to air . formation of deg-resPFCAs, in all three APs for comparison. Our global production estimates on a homologue basis are We recognized that a small amount of FTOHs can also be believed to be reliable because, for example, the estimate for C8 FTPs for the period 1970−2007 (41.4−49.7 kt) compares generated from degradation of nonpolymeric FT-based 5 fl favorably with a previous report of 34.5 kt (32 kt of acrylate substances, e.g., poly uoroalkyl phosphate esters (PAPs) in 12 21,22 plus 2.5 kt of urethane polymers). While the absolute AP2 (Figure 1); we did not considered their contribution amounts of production volumes are admittedly uncertain, they to the total FTOH release because (i) inadequate available do not hinder us to arrive at meaningful conclusions in the market information renders estimating their contribution highly following because (i) we focus our attention on temporal uncertain and (ii) the release is much lower than degFTOHs trends in the relative importance of individual sources, and (ii) and resFTOHs, according to our preliminary calculation (Text 18 both CiP-CAFE and BETR-Global are linear models that S2). There are a number of other PFCA sources (Figure 1), enable scaling all absolute outputs to production volumes if such as the intentional uses of PFCAs as processing aid in updated production information is available in the future. Next, fluoropolymer production and the transformation of non- fl fl the global annual production of FTPs on a homologue basis polymeric uorotelomer-based derivatives (e.g., uorotelomer were attributed to the regions defined in CiP-CAFE: ∼5% in sulfonamido betaines and fluorotelomer thioamido sulfonates 29 30 23 Western Europe (RE5), 50% in North America (RE6), as ingredients in AFFFs, which normally do not transform to ∼ 27 24 15% in China (RE1, only after 2009), and the remainder in FTOHs ). Estimating PFCA releases from these sources is Japan and South Korea (RE2) (see the locations of beyond the scope of this work. Meanwhile, the lack of detailed manufacturers in Figure SF-1). information prevents consideration of the presence and releases CiP-CAFE accounted for geographical redistributions of the fl of other minor residuals, e.g., per uoroalkyl iodides and regional FTP production via international trade among the 25 fluorotelomer iodides. CiP-CAFE regions using formulated concentrated DWRs Overview of Modeling Strategy. Figure 2 summarizes (Table S2) as a surrogate for the FTPs subjected to textile the conceptual modeling strategy adopted in this study. In a and fabric finishing after “formulation (LC2)”, and clothing first step, the dynamic substance flow analysis model, CiP- (Table S2) as a surrogate for the FTPs in finished textile and CAFE, is used to calculate time-variant in-use and waste stocks fabric products reaching consumers after “processing (LC3)”. of FTPs worldwide for the period 1960 to 2040. The rationale Here, we assumed that FTPs contained in the concentrated for, and description of, the model have been detailed DWRs and clothing were imported to or exported from a previously26 and are briefly provided in Text S2.1. Here, region in the same proportion as the two respective surrogates substance accumulation in the process labeled “in-service because international statistics enabling discrimination between (LC5)” in CiP-CAFE is used to describe in-use stocks, while FTP-containing and FTP-free commercial products were not waste stocks are represented by substance accumulation in the available. To investigate the influence of different product

4464 DOI: 10.1021/acs.est.6b04021 Environ. Sci. Technol. 2017, 51, 4461−4470 Environmental Science & Technology Article

Figure 3. Estimated ranges of global in-use stocks (a and e) and waste stocks (b and f) of FTPs, annual releases of degFTOHs (c and g), and deg- degPFCAs (d and h) for C8 (a−d) and C6 compounds (e−h) from 1960−2040 under the four simulation scenarios: (I) LS = 10 years and HL = 75 years, (II) LS = 10 years and HL = 1500 years, (III) LS = 50 years and HL = 75 years, and (IV) LS = 50 years and HL = 1500 years. The global annual releases of resFTOHs are presented in gray shading in panels c and g for comparison. lifespans (LSs) of finished textile and fabric products in AP1 parameters for AP1 were assumed the same as those used in and degradation half-lives (HLs) of FTPs in waste stocks on Step 1, and consumer products in APs 2 and 3 were assumed to our model results, we ran CiP-CAFE using four scenarios: (I) be domestically consumed with fixed product lifespans (Table ffi LS = 10 years and HL = 75 years, (II) LS = 10 years and HL = S3). Table S5 tabulates partitioning coe cients (KAW and KOW) 1500 years (Table S3), (III) LS = 50 years and HL = 75 years, for FTOHs and degradation half-lives of FTOHs in various Δ Δ Δ and (IV) LS = 50 years and HL = 1500 years. Emission and compartments. Internal energies ( UOA, UAW, and UOW) waste factors used in the CiP-CAFE calculation are tabulated in for adjusting the cited partitioning coefficients to different Table S4. Here, we assumed that FTPs are neither released nor temperature were calculated based on the method in ref 34. degraded during the product’s lifespan as there is substantial Activation energies for adjusting the degradation half-lives to difference in the observed reduction in oil and water repellency different temperature were default values in CiP-CAFE and or FTP weight among diverse (co)polymer compositions, BETR-Global (20 kJ mol−1 for water and wastewater treatment finishing treatments, and textile surface properties in a plants and 30 kJ mol−1 for soil, landfill, and simple landfill and − multitude of laundering or weathering studies.2,31 33 While dumping). Emission and waste factors used in the CiP-CAFE many of the studies reported negligible FTP losses (from not calculation are tabulated in Table S4. Fate and transformation − observed to <2%) in a limited test time,31 33 no information is of FTOHs in the waste stocks are calculated by the Model for available for textile and fabric use over a decadal time scale. At Organic Chemicals in Landfills (MOCLA) module contained in the end of their lifespan, obsolete FTP-containing finished CiP-CAFE. textile and fabric products in AP1 enter the waste stream, the To evaluate the CiP-CAFE results for the different scenarios distribution between different disposal approaches following and to calculate FTOH transformation in the environment, we the default time-variant regional waste-disposal ratios performed global environmental fate simulations for total (WDRs)26 supplied within CiP-CAFE. Degradation of FTPs FTOHs using the BETR-Global model (Step 3 in Figure 2). in the waste stocks was calculated using HLs in the four This model is described in detail in refs 35 and 36 and brieflyin assumed scenarios. Emissions of FTPs from the waste stocks Text S2.2. The CiP-CAFE-derived annual releases of FTOHs were assumed to be negligible because FTPs are neither volatile and the amounts of FTOHs formed during the degradation of nor soluble. FTPs in the environment were combined and allocated to the The second step is composed of another series of CiP-CAFE 15°×15° BETR-Global cells, as in ref 26. Time-variant runs to estimate the global annual releases of FTOHs from atmospheric concentrations of 8:2 and 6:2 FTOH calculated by waste stocks during the period 1960−2040 (Step 2 in Figure BETR-Global for different scenarios were then compared with 2). The formation of degFTOHs was calculated from the monitoring data reported in the literature. degradation of (i) FTPs in two waste stocks and (ii) FTPs Next, the annually degraded amounts of resFTOHs and released into the environment by assuming that FTOHs are degFTOHs in both waste stocks (calculated by CiP-CAFE) and formed from FTPs with the same perfluorinated chain length the environment (calculated by BETR-Global) were fractionally with a yield of 100%.13 The formed degFTOHs in waste stocks, converted into deg-resPFCAs and deg-degPFCAs, respectively, together with the annual production of resFTOHs in all three based on the median values of the estimated molar yields (mol APs, served as inputs for the CiP-CAFE calculation. In Step 2, %) for the transformation of n:2 FTOH to different PFCA

4465 DOI: 10.1021/acs.est.6b04021 Environ. Sci. Technol. 2017, 51, 4461−4470 Environmental Science & Technology Article homologues: the degradation of n:2FTOH yields 1 mol % for route varied over 3 orders of magnitude among the scenarios, each PFCA homologue with n − 1, n − 2, and n − 3 carbons with the calculated cumulative releases of 8:2 degFTOHs and 5 mol % for each PFCA homologue with n and n +1 ranging from 11.5−13.8 t under Scenario IV to 1700−2038 t carbons.18 The yields correspond to homologue-specific mass under Scenario I and those of 6:2 degFTOHs ranging from yields of 4.1%−6% on a FTOH basis, which agree well with − − − 0.7 0.9 t under Scenario IV to 126 152 t under Scenario I by those used in previous modeling studies.37 39 Admittedly, the 2015. In contrast, the releases of degFTOHs via the latter route PFCA yields could be somewhat conservative, because we were quite similar for the different scenarios, namely 22−37 t considered neither degradation intermediates of FTOHs (e.g., for 8:2 degFTOHs and 3−6 t for 6:2 degFTOHs by 2015. 38 fluorotelomer aldehydes) nor the PFCA formation from FTP Consequently, the relative importance of the two sources of degradation intermediates other than FTOHs (e.g., newly degFTOHs varied considerably between the four scenarios. fi identi ed 7:2 sFTOH [F(CF2)7CH(OH)CH3] in biodegrada- Furthermore, the atmosphere was predicted to receive the fl 20 ffi tion of 8:2 uorotelomer acrylate), due to insu cient dominant share (97% − 99%) of degFTOHs. Volatilization information on the multimedia partitioning behavior and with landfill gas is the predominant route by which degFTOHs degradation kinetics of these intermediates. Meanwhile, we enter the atmosphere. This finding echoes Washington et al.’s did not consider the PFCA formation in the environment assessment that “landfills are not airtight” and “commercial resulting from further transformation of the degradation FTPs potentially might be a source of fluorotelomers to the intermediates released from waste stocks. Finally, the converted environment even after disposal”.4 deg-resPFCAs and deg-degPFCAs in waste stocks, along with For both 6:2 and 8:2 degFTOHs, the annual releases in the annual production of impPFCAs in all three APs, served as scenarios with short HL (I and III) are almost 1 order of inputs for a third set of CiP-CAFE calculations (Step 4 in magnitude higher than in those using long HL (II and IV); that Figure 2) to calculate time-variant emissions of total ff fl − is, the estimated annual releases of degFTOHs are a ected to a uorotelomer-related PFCAs for the period 1960 2040. larger extent by a change in HL (which is in contrast to the Parameters for consumer products in three APs are the same sensitivity of stocks to LS) when both the HL and LS vary as those used in Step 2. Emission and waste factors used are within a realistic range. Annual releases of 8:2 degFTOH are given in Table S4; partitioning coefficients for neutral PFCAs anticipated to peak around 2022 in Scenario I and decline and degradation half-lives of PFCAs in various compartments afterward due to the rapid depletion of C8 FTPs waste stocks are tabulated in Table S6. in this scenario (Figure 3b), while in other scenarios, releases of ■ RESULTS AND DISCUSSION both 8:2 and 6:2 degFTOHs are projected to keep increasing throughout the simulation period. Temporal Evolution of Estimated Stocks and Re- The estimated annual multimedia releases of deg-degPFCAs leases. Figure 3 presents the calculated time-variant global in- in Figure 3d,h are the sum of deg-degPFCAs (i) liberated from use and waste stocks of FTPs, as well as annual releases of waste stocks and (ii) transformed from the degFTOHs released degFTOHs and deg-degPFCAs into the environment, for long- into the environment. The latter is currently dominant, with chain C8 compounds (Figure 3a−d) and short-chain C6 − − contributions to cumulative releases of 89% 95% for deg- compounds (Figure 3e h) under the four scenarios. degPFOA and 68% − 97% for deg-degPFHxA by 2015 under Within the given time frame from 1960 to 2040, for both C8 different scenarios. The dominance of the latter route is and C6 FTPs, scenarios based on long LS (III and IV) yield understandable as our calculations find that (i) 97% − 99% of larger in-use stocks (Figure 3a,e) but smaller waste stocks degFTOHs volatilize with landfill gas into the atmosphere, (Figure 3b,f) because longer use implies slower transfer to where 94% − 97% of them degrade into deg-degPFCAs within waste; in scenarios with long HL (II and IV), fewer FTPs are a year and, by contrast, (ii) less than 3% of the deg-degPFCAs degraded, resulting in larger waste stocks (Figure 3b,f). In general, both the in-use and waste stocks peak or plateau when generated from the degFTOHs remaining in waste stocks are fl fl annually released via leachate (Table S6). That leaching is not in ows and out ows are similar in magnitude. For example, in- ffi fi use stocks of C8 FTPs (Figure 3a) peaked at 21−25 kt in 2010 an e cient route for delivering a signi cant amount of deg- − degPFCAs into the environment in the short term occurs (Scenarios I and II) or 49 59 kt in 2014 (Scenarios III and IV) fi fl when increasing rates of discarding matched decreasing rates of mostly because the area-speci c leachate ow, which equates to ∼ −1 26 new use (the latter a result of the phase-out of C8 FTPs). the average annual precipitation in a region ( 1 m year ), is 16 ± Likewise, in-use stocks of C6 FTPs (Figure 3d) are expected to too small. Likewise, Yan et al. estimated that annually 1.8 3 fi level off at 59−70 kt after 2027 (Scenario I and II) when the t of PFOA were leached from land lls across China based on samples collected in 2013, which accounts for a mere ∼3% of increasing discard rates catch up with the annual rates of new 27 use of C6-based finished products (which is assumed constant the annual national emissions of PFOA (<60 t in 2012 )in ffi according to annual production data in ref 18). Furthermore, China. In fact, the ine ciency of leachate releases is also the 40 for both the in-use and waste stocks, the difference in stock size case for a wider range of soluble compounds such as phenol. due to a change in LS (i.e., Scenarios I and II versus Scenarios The above result implies that, due to such a “trickle”, the III and IV) is more notable than that due to a change in HL residence time of deg-degPFCAs generated in waste stocks can (i.e., Scenarios I and III versus Scenarios II and IV). This reach multiple decades if not centuries, i.e., waste stocks are a indicates that, when both are at possibly realistic levels, product reservoir slowly releasing deg-degPFCAs. Figure S2 presents an lifespan is more crucial than degradation half-life to an accurate additional long-term calculation to 2100 of annual releases of description of FTPs stocks. deg-degPFOA from the two routes under Scenario I. While the The estimated annual multimedia releases of degFTOHs in formation of deg-degPFCAs in the environment will moder- Figure 3c,g are the sum of degFTOHs (i) released from waste ately decline after 2020 due to the depletion of C8 FTP waste stocks and (ii) transformed from FTPs released in the stocks (Figure 3b), the releases of deg-degPFCAs from waste environment. The releases of degFTOHs via the former stocks will increase throughout the simulation period. After

4466 DOI: 10.1021/acs.est.6b04021 Environ. Sci. Technol. 2017, 51, 4461−4470 Environmental Science & Technology Article

Figure 4. Comparison between literature-reported measured and BETR-Global modeled atmospheric concentrations of 8:2 FTOH (a and b) and 6:2 FTOH (c and d) at diﬀerent sites (see Figure SF-1 for their BETR-Global cells). The BETR-Global simulations were based on the annual releases of resFTOHs alone (a and c) and combined resFTOHs and degFTOHs under the Scenario I (b and d). Diagonal lines represent perfect agreement (solid), and agreement within a factor of 100.5 (dashed line) and 10 (dotted line). Error bars indicate range of the concentrations (i.e., the diﬀerence between maximum and minimum).

2090, the annual releases from the two routes will be by Prevedouros et al.19 (∼100 t year−1 for 2002), Wania37 comparable. (100−200 t year−1 for 2000−2005) and Schenker et al.39 (60− Interestingly, the calculation by the MOCLA module in CiP- 155 t year−1 for 2000−2005); they are also within the range of CAFE suggests that short-chain PFCA homologues are almost ∼0.5 to ∼300 t year−1, which can be calculated from the data 2 orders of magnitude more likely than long-chain ones to presented in Wang et al.18 migrate from waste stocks to the hydrosphere (Table S6). This The relative contributions of degFTOHs and resFTOHs to is because short-chain PFCA homologues are more water- the total releases change over time. 8:2 resFTOH dominated soluble and adsorb less to landfill organic matter (i.e., lower log the annual total emissions of 8:2 FTOH prior to approximately KOW in Table S6). If we further take into account the migration 2010, its contribution exceeding that of 8:2 degFTOH by a of precursors, i.e., short-chain FTOH homologues volatilize factor of 2 (Scenario I) to 2 orders of magnitude (Scenario IV). more from waste stocks to the atmosphere due to their higher Since 2010, the relative importance of 8:2 degFTOH has been vapor pressure,41 we can definitely expect higher relative release increasing in the wake of the phase-out of long-chain products rates of short-chain deg-degPFCAs than long-chain deg- in most world regions. Such a transition from resFTOH- degPFCAs. For example, our calculations demonstrate that, dominant to degFTOH-dominant releases is also obvious for during the simulation period 1960−2040, the cumulative Scenario I of 6:2 FTOH (Figure 3g), in which the annual consumption of C8 FTPs (55−66 kt) is estimated to be 25- releases of 6:2 degFTOH would exceed that of 6:2 resFTOH fold higher than that of C4 FTPs (2.1−2.5 kt), yet the by approximately 2025. Our finding of the dominance of cumulative releases of deg-degPFOA (280−347 t) are only 6- resFTOHs is consistent with previous studies.11,13 For instance, fold higher than those of deg-degPFBA (data not shown in van Zelm et al.13 calculated that 8:2 resFTOH constituted up to Figure 3). That short-chain deg-degPFCAs are more readily 75% of the total 8:2 FTOH in European air, freshwater, and released implies that the use of short-chain FTPs may have a seawater prior to 2025, although their estimated annual releases more significant immediate influence on the environment, of 8:2 degFTOH are at least 1 order of magnitude higher than while long-chain FTPs are more likely to be a long-term threat. ours because the authors assumed that, before entering the use Analogous to the sensitivity of degFTOHs above, HL was phase, one-third of historically produced FTPs had been identified as the factor with the most influence on the annual liberated from industrial processes and subjected to degrada- releases of deg-degPFCAs, as those releases are considerably tion. higher in Scenarios I and III (Figure 3d,h). Scenario I led to the The dominance of resFTOHs before around 2010 also highest release of deg-degPFCAs. The annual releases of deg- provides justification for the good agreement between degPFOA and deg-degPFHxA are predicted to increase measured and modeled FTOH concentrations in a series of throughout the simulation period, except for Scenario I in earlier modeling studies,37,39,42 which were based on emission which releases of deg-degPFOA reach a plateau (Figure 3d). estimates of resFTOHs alone. Furthermore, we modeled the Rising Contributions of FTP Degradation to Global atmospheric concentrations of FTOHs using the BETR-Global FTOH Releases. In addition to the amount of FTOHs released model and our CiP-CAFE-derived multimedia emission as a result of FTP degradation (degFTOHs), the CiP-CAFE estimates of resFTOH and degFTOH under Scenario I. The model also estimated the amount of FTOHs released as modeled concentrations of 8:2 and 6:2 FTOHs were then residuals (resFTOHs, gray shading in Figure 3c,g). The annual plotted against measured concentrations (Figure 4), which had releases of 8:2 resFTOH are estimated to have substantially been reported from large-scale sampling cruises or on-land − increased since the 1990s with a peaked around 2007 (the year campaigns43 54 so as to encompass a wide span of after the worldwide transition from long-chain-based commer- representative BETR-Global cells (Figure SF-1). Cells covering cial products to their short-chain alternatives began); those of the territory of mainland China were excluded from the 6:2 resFTOH have increased steadily until reaching a plateau in comparison because the ongoing production and new uses of 2007. Our estimates agree with earlier studies. For example, our long-chain products in China could still be a cause for annual 8:2 resFTOH emissions (80−320 t year−1 for the period resFTOH releases (as elaborated in Methods section), thus 2000−2004 on average) are generally similar to those reported resulting in an underestimation in 8:2FTOH concentrations for

4467 DOI: 10.1021/acs.est.6b04021 Environ. Sci. Technol. 2017, 51, 4461−4470 Environmental Science & Technology Article the most recent years. Figure 4a,b demonstrate good agreement end. This is in contrast to the release of deg-degPFOAs (Figure between measured and modeled 8:2 FTOH concentrations, 3d), which will last for decades and even centuries. Currently, whether the releases of 8:2 degFTOH from scenario I are the deg-degPFOA is estimated to account for 3% (Scenario II considered or not (Figure 3c). However, including releases of and IV) to 30% (Scenario I) of deg-resPFOA. Such a 8:2 degFTOH (Figure 4b) improves model agreement with the dominance of deg-resPFOA in the total FTP-related PFOA more recent measurements, most notably the latest available releases has also been reported by Russell et al.,5 who indicated measurements (at the bottom left), which were sampled that deg-degPFOA had contributed 18% to the PFOA between Oct. 2010 and Jan. 2011.44 Such an improvement historically released from FTP-related sources by 2007, while lends supports to our hypothesis of an increasing contribution deg-resPFOA had contributed 77%, based on the assumed of degFTOHs. Compared with Scenario I, the improvements in degradation half-life of 1000−2000 years. the other scenarios were less remarkable (data not shown) A number of other PFCA sources18,19 exist apart from the because the estimated releases of 8:2 degFTOH are much three (deg-degPFCAs, deg-resPFCAs, and impPFCAs) consid- smaller than those of 8:2 resFTOH (Figure 3c). It should be ered in this work (Figure 1). The most-recent reliable estimate noted that, while agreement with measurements is best using of the global total release of PFCAs (C4−C14) from these Scenario I (Figure 4b), we are reluctant to pronounce one sources is approximately 5600−13000 t between 1951 and scenario more realistic than another, because (i) the 2015 (under the plausible scenario),18 which is nearly 1 order contribution of degFTOHs is rather low at present (Figure of magnitude higher than the estimated cumulative release of 3c) and is thus easily drowned out by the considerable deg-degPFCAs (556−591 t, C4−C12) in our highest-release uncertainty or variability associated with the measurements Scenario I for the same period. Therefore, omitting the (error bars in Figure 4); and (ii) the number of recent contribution from the degradation of FTPs when estimating (especially since 2011) measured data is limited. In addition, historical PFCA emissions18,19,37 should not lead to a significant while we believe that the bulk of FTPs degrade during the underestimation. waste disposal phases, there can be continuous, but currently Nevertheless, the degradation of FTPs in waste stocks can be unquantifiable, losses or degradation of in-use FTPs due to a significant source of PFCAs in the future, particularly when washing, abrasion, or weathering.2 Figure S3 indicates that deliberate uses of (long-chain) PFCAs will terminate. Our assuming the degradation of an annual loss of 2%, 5%, and 40% estimates suggest that 1323−1535 t of deg-degPFCAs (C4− of in-use stocks of FTPs elevates the annual releases of C12) will be released during 2016−2040 in Scenario I (data not degFTOHs and deg-degPFCAs (under Scenario I) by a factor shown). This level is in the same order as the projected of approximately 3, 5, and 15, respectively. This would lead to cumulative release of 10−3830 t from intentional uses of an overestimation of the observed atmospheric occurrence of PFCAs (C4−C14) for 2015−2030 (worst case assuming no 18 total FTOHs, given the good agreements between our restrictions on PFCAs will be taken in developing countries). modeling results and measurements in Figure 4. Nevertheless, Meanwhile, our long-term calculation under Scenario I (Figure investigations on FTP loss during the use phase are required to S2) indicates that, between 2015 and 2100, another 7 times as allow quantification of this potential source. For 6:2 FTOH much will be released as the cumulative emissions of deg- (Figure 4c,d), while including 6:2 degFTOH releases improves degPFOA up to 2015. Moreover, by 2100, 1170−1380 t of deg- model agreement with the more-recent measurements as well, degPFOA is estimated to be present in waste stocks (data not the improvement is too minor to be readily perceived from the shown), which is available for leaching into the environment in figure because the estimated releases of 6:2 degFTOH are the following centuries. Our calculations highlight the need for nearly 2 orders of magnitude lower than those of 6:2 resFTOH. environmentally sound management of the waste stocks of From a regulatory perspective, our calculation implies that FTPs into the foreseeable future. On the one hand, instead of the increasing contribution of FTP degradation has the being deposited in landfills and dumps, destruction and potential to partially offset the reduction in FTOH residuals irreversible transformation techniques (e.g., incineration) in products achieved by the 2010/15 PFOA Stewardship should be the preferable disposal options for obsolete FTP- Program and other regulatory efforts. This could partially containing finished products; on the other hand, landfill gases explain the recent rebound of declining atmospheric concen- and leachates from historical and current landfills and dumps, trations of 8:2 FTOH; for instance, a multiannual trend analysis which are, respectively, the two major routes liberating indicated that 8:2 FTOH concentrations in samples from the degFTOHs and deg-degPFCAs into the environment, should Global Atmospheric Passive Sampling (GAPS) Network be appropriately treated or remediated. initially declined from 2006−2008 (33−46 samples each year) but then increased again from 2009−2011 (19−34 ■ ASSOCIATED CONTENT 55 samples each year). *S Supporting Information FTP Degradation in Waste Stocks as a Long-Term The Supporting Information is available free of charge on the Source of PFCAs. We compared the annual releases of deg- ACS Publications website at DOI: 10.1021/acs.est.6b04021. degPFCAs estimated above (Figure 3d,h) with other PFCAs sources, i.e., impPFCAs and deg-resPFCAs (both liberated Text describing terms and models used in this study and from waste stocks and formed in the environment) (Figure S4), the generation of FTOHs from degradation of which are also related to the uses of fluorotelomer-based polyfluoroalkyl phosphates; tables detailing homologue substances. Figure S4 shows that, for both PFOA and PFHxA, composition and contents, international trade and the annual releases of deg-resPFCAs are nearly 3 orders of lifespan distribution of consumer products, emission magnitude higher than those of impPFCAs. The releases of and waste factors, and physicochemical properties of impPFOA and deg-resPFOA are anticipated to cease within a modeled chemicals; and figures depicting global annual decade when the service life of nonpolymeric C8 fluorotelomer- production of technical fluorotelomer-based substances, based substances (e.g., AFFFs and surfactants) comes to the additional release estimates, and illustration of the

4468 DOI: 10.1021/acs.est.6b04021 Environ. Sci. Technol. 2017, 51, 4461−4470 Environmental Science & Technology Article

influence from considering assumed FTP loss during the (13) van Zelm, R.; Huijbregts, M. A.; Russell, M. H.; Jager, T.; van de use phase. (PDF) Meent, D. Modeling the environmental fate of perfluorooctanoate and its precursors from global fluorotelomer acrylate polymer use. Environ. Toxicol. Chem. 2008, 27 (11), 2216−2223. ■ AUTHOR INFORMATION (14) Washington, J. W.; Jenkins, T. M. Abiotic hydrolysis of fluorotelomer-based polymers as a source of perfluorocarboxylates at Corresponding Authors − * the global scale. Environ. Sci. Technol. 2015, 49 (24), 14129 14135. Phone: 86 10 62753746; e-mail: [email protected]. (15) Lang, J. R.; Allred, B. M.; Peaslee, G. F.; Field, J. A.; Barlaz, M. *Phone: 86 10 62759075; e-mail: [email protected]. A. Release of per- and polyfluoroalkyl substances (PFASs) from carpet ORCID and clothing in model anaerobic landfill reactors. Environ. Sci. Technol. 2016, 50 (10), 5024−5032. 0000-0002-5157-7366 Li Li: (16) Yan, H.; Cousins, I. T.; Zhang, C.; Zhou, Q. Perfluoroalkyl acids Frank Wania: 0000-0003-3836-0901 in municipal landfill leachates from China: Occurrence, fate during Notes leachate treatment and potential impact on groundwater. Sci. Total − − The authors declare no competing financial interest. Environ. 2015, 524 525,23 31. (17) Ahrens, L.; Shoeib, M.; Harner, T.; Lee, S. C.; Guo, R.; Reiner, ■ ACKNOWLEDGMENTS E. J. Wastewater treatment plant and landfills as sources of polyfluoroalkyl compounds to the atmosphere. Environ. Sci. Technol. The authors thank Roland Weber for insightful discussion on 2011, 45 (19), 8098−8105. PFAS leaching from waste stocks. This study was financially (18) Wang, Z.; Cousins, I. T.; Scheringer, M.; Buck, R. C.; supported by the National Natural Science Foundation of Hungerbühler, K. Global emission inventories for C4−C14 perfluor- China (grant no. 21577002). L.L. acknowledges scholarships oalkyl carboxylic acid (PFCA) homologues from 1951 to 2030, Part I: from the China Scholarship Council and the Shanghai Tongji production and emissions from quantifiable sources. Environ. Int. 2014, 70,62−75. Gao Tingyao Environmental Science and Technology Develop- (19) Prevedouros, K.; Cousins, I. T.; Buck, R. C.; Korzeniowski, S. H. ment Foundation. Sources, fate and transport of perfluorocarboxylates. Environ. Sci. Technol. 2006, 40 (1), 32−44. ■ REFERENCES (20) Royer, L. A.; Lee, L. S.; Russell, M. H.; Nies, L. F.; Turco, R. F. (1) USEPA Telomer Research Program Update. Presentation to USEPA- Microbial transformation of 8:2 fluorotelomer acrylate and meth- − acrylate in aerobic soils. Chemosphere 2015, 129,54−61. OPPT, November 25, 2002; U.S. Public Docket AR226 1141; 2002. ’ (2) Holmquist, H.; Schellenberger, S.; van der Veen, I.; Peters, G.; (21) Lee, H.; D eon, J.; Mabury, S. A. Biodegradation of Leonards, P.; Cousins, I. Properties, performance and associated polyfluoroalkyl phosphates as a source of perfluorinated acids to the − hazards of state-of-the-art durable water repellent (DWR) chemistry environment. Environ. Sci. Technol. 2010, 44 (9), 3305 3310. for textile finishing. Environ. Int. 2016, 91, 251−264. (22) Liu, C.; Liu, J. Aerobic biotransformation of polyfluoroalkyl − (3) OECD. OECD/UNEP Global PFC Group, Synthesis Paper on Per- phosphate esters (PAPs) in soil. Environ. Pollut. 2016, 212, 230 237. and Polyfluorinated Chemicals (PFCs); OECD: Paris, France, 2013. (23) Backe, W. J.; Day, T. C.; Field, J. A. Zwitterionic, cationic, and (4) Washington, J. W.; Jenkins, T. M.; Rankin, K.; Naile, J. E. anionic fluorinated chemicals in aqueous film forming foam Decades-scale degradation of commercial, side-chain, fluorotelomer- formulations and groundwater from US military bases by nonaqueous based polymers in soils and water. Environ. Sci. Technol. 2015, 49 (2), large-volume injection HPLC-MS/MS. Environ. Sci. Technol. 2013, 47 − 915−923. (10), 5226 5234. (5) Russell, M. H.; Berti, W. R.; Szostek, B.; Buck, R. C. Investigation (24) Butt, C. M.; Muir, D. C. G.; Mabury, S. A. Biotransformation of the biodegradation potential of a fluoroacrylate polymer product in pathways of fluorotelomer compounds. Environ. Toxicol. Chem. 2014, − aerobic soils. Environ. Sci. Technol. 2008, 42 (3), 800−807. 33 (2), 243 267. (6) Washington, J. W.; Ellington, J. J.; Jenkins, T. M.; Evans, J. J.; (25) Ruan, T.; Wang, Y.; Wang, T.; Zhang, Q.; Ding, L.; Liu, J.; Yoo, H.; Hafner, S. C. Degradability of an acrylate-linked, Wang, C.; Qu, G.; Jiang, G. Presence and partitioning behavior of fluorotelomer polymer in soil. Environ. Sci. Technol. 2009, 43 (17), polyfluorinated iodine alkanes in environmental matrices around a 6617−6623. fluorochemical manufacturing plant: Another possible source for (7) Ellis, D. A.; Martin, J. W.; De Silva, A. O.; Mabury, S. A.; Hurley, perfluorinated carboxylic acids? Environ. Sci. Technol. 2010, 44 (15), M. D.; Sulbaek Andersen, M. P.; Wallington, T. J. Degradation of 5755−5761. fluorotelomer alcohols: A likely atmospheric source of perfluorinated (26) Li, L.; Wania, F. Tracking chemicals in products around the carboxylic acids. Environ. Sci. Technol. 2004, 38 (12), 3316−3321. world: introduction of a dynamic substance flow analysis model and (8) Phillips, M. M.; Dinglasan-Panlilio, M. J. A.; Mabury, S. A.; application to PCBs. Environ. Int. 2016, 94, 674−686. Solomon, K. R.; Sibley, P. K. Fluorotelomer acids are more toxic than (27) Li, L.; Zhai, Z.; Liu, J.; Hu, J. Estimating industrial and domestic perfluorinated acids. Environ. Sci. Technol. 2007, 41 (20), 7159−7163. environmental releases of perfluorooctanoic acid and its salt in China. (9) Giesy, J. P.; Naile, J. E.; Khim, J. S.; Jones, P. D.; Newsted, J. L., Chemosphere 2015, 129, 100−109. Aquatic toxicology of perfluorinated chemicals. In Reviews of (28) Vestergren, R.; Herzke, D.; Wang, T.; Cousins, I. T. Are Environmental Contamination and Toxicology, Whitacre, D. M., Ed.; imported consumer products an important diffuse source of PFASs to Springer: New York, 2010; Vol. 202,pp1−52. the Norwegian environment? Environ. Pollut. 2015, 198, 223−230. (10) Ahrens, L.; Bundschuh, M. Fate and effects of poly- and (29) ECHA. ANNEX XV Restriction Report Proposal for A Restriction. perfluoroalkyl substances in the aquatic environment: A review. Perfluorooctanoic Acid (PFOA), PFOA Salts and PFOA-Related Environ. Toxicol. Chem. 2014, 33 (9), 1921−1929. Substances; European Chemicals Agency: Helsinki, Finland, 2014. (11) Wang, Z.; Cousins, I. T.; Scheringer, M.; Buck, R. C.; (30) USEPA. Long-Chain Perfluorinated Chemicals (PFCs) Action Hungerbühler, K. Global emission inventories for C4−C14 perfluoro- Plan; U.S. Environmental Protection Agency: Washington D.C., 2009. alkyl carboxylic acid (PFCA) homologues from 1951 to 2030, part II: (31) Castelvetro, V.; Aglietto, M.; Ciardelli, F.; Chiantore, O.; The remaining pieces of the puzzle. Environ. Int. 2014, 69, 166−176. Lazzari, M.; Toniolo, L. Structure control, coating properties, and (12) Russell, M. H.; Berti, W. R.; Szostek, B.; Wang, N.; Buck, R. C. durability of fluorinated acrylic-based polymers. J. Coat. Technol. 2002, Evaluation of PFO formation from the biodegradation of a 74 (928), 57−66. fluorotelomer-based urethane polymer product in aerobic soils. (32) DuPont, E. I. Oil- and Water-Repellent Copolymers. EU Patent Polym. Degrad. Stab. 2010, 95 (1), 79−85. 87300551.6, September 2, 1987.

4469 DOI: 10.1021/acs.est.6b04021 Environ. Sci. Technol. 2017, 51, 4461−4470 Environmental Science & Technology Article

(33) Watkins, M. H.; Covelli, C. A.; Seals, L. R. Fabric Treated with (51) Cai, M.; Xie, Z.; Möller, A.; Yin, Z.; Huang, P.; Cai, M.; Yang, Durable Stain Repel and Stain Release Finish and Method of Industrial H.; Sturm, R.; He, J.; Ebinghaus, R. Polyfluorinated compounds in the Laundering to Maintain Durability of Finish. U.S. Patent 2006/ atmosphere along a cruise pathway from the Japan Sea to the Arctic 0228964 A1. October 12, 2006. Ocean. Chemosphere 2012, 87 (9), 989−997. (34) MacLeod, M.; Scheringer, M.; Hungerbühler, K. Estimating (52) Ahrens, L.; Shoeib, M.; Del Vento, S.; Codling, G.; Halsall, C. enthalpy of vaporization from vapor pressure using Trouton’s rule. Polyfluoroalkyl compounds in the Canadian Arctic atmosphere. Environ. Sci. Technol. 2007, 41 (8), 2827−2832. Environ. Chem. 2013, 8 (4), 399−406. (35) MacLeod, M.; von Waldow, H.; Tay, P.; Armitage, J. M.; (53) Xie, Z.; Wang, Z.; Mi, W.; Möller, A.; Wolschke, H.; Ebinghaus, Wöhrnschimmel, H.; Riley, W. J.; McKone, T. E.; Hungerbühler, K. R. Neutral poly-/perfluoroalkyl substances in air and snow from the BETR Global−A geographically-explicit global-scale multimedia Arctic. Sci. Rep. 2015, 5, 8912. contaminant fate model. Environ. Pollut. 2011, 159 (5), 1442−1445. (54) Vento, S. D.; Halsall, C.; Gioia, R.; Jones, K.; Dachs, J. Volatile (36) MacLeod, M.; Riley, W. J.; McKone, T. E. Assessing the per- and polyfluoroalkyl compounds in the remote atmosphere of the influence of climate variability on atmospheric concentrations of western Antarctic Peninsula: an indirect source of perfluoroalkyl acids − polychlorinated biphenyls using a global-scale mass balance model to Antarctic waters? Atmos. Pollut. Res. 2012, 3 (4), 450 455. (BETR-Global). Environ. Sci. Technol. 2005, 39 (17), 6749−6756. (55) Gawor, A.; Shunthirasingham, C.; Hayward, S.; Lei, Y.; Gouin, (37) Wania, F. A global mass balance analysis of the source of T.; Mmereki, B.; Masamba, W.; Ruepert, C.; Castillo, L.; Shoeib, M.; Lee, S. C.; Harner, T.; Wania, F. Neutral polyfluoroalkyl substances in perfluorocarboxylic acids in the Arctic Ocean. Environ. Sci. Technol. − 2007, 41 (13), 4529−4535. the global atmosphere. Env. Sci. Process. Impact 2014, 16 (3), 404 413. (38) Wallington, T.; Hurley, M.; Xia, J.; Wuebbles, D.; Sillman, S.; Ito, A.; Penner, J.; Ellis, D.; Martin, J.; Mabury, S.; et al. Formation of C7F15COOH (PFOA) and other perfluorocarboxylic acids during the atmospheric oxidation of 8:2 fluorotelomer alcohol. Environ. Sci. Technol. 2006, 40 (3), 924−930. (39) Schenker, U.; Scheringer, M.; Macleod, M.; Martin, J. W.; Cousins, I. T.; Hungerbühler, K. Contribution of volatile precursor substances to the flux of perfluorooctanoate to the Arctic. Environ. Sci. Technol. 2008, 42 (10), 3710−3716. (40) Kjeldsen, P.; Christensen, T. H. A simple model for the distribution and fate of organic chemicals in a landfill: MOCLA. Waste Manage. Res. 2001, 19 (3), 201−216. (41) Kim, M.; Li, L. Y.; Grace, J. R.; Yue, C. Selecting reliable physicochemical properties of perfluoroalkyl and polyfluoroalkyl substances (PFASs) based on molecular descriptors. Environ. Pollut. 2015, 196, 462−472. (42) Yarwood, G.; Kemball-Cook, S.; Keinath, M.; Waterland, R. L.; Korzeniowski, S. H.; Buck, R. C.; Russell, M. H.; Washburn, S. T. High-resolution atmospheric modeling of fluorotelomer alcohols and perfluorocarboxylic acids in the North American troposphere. Environ. Sci. Technol. 2007, 41 (16), 5756−5762. (43) Li, J.; Del Vento, S.; Schuster, J.; Zhang, G.; Chakraborty, P.; Kobara, Y.; Jones, K. C. Perfluorinated compounds in the Asian atmosphere. Environ. Sci. Technol. 2011, 45 (17), 7241−7248. (44) Wang, Z.; Xie, Z.; Mi, W.; Möller, A.; Wolschke, H.; Ebinghaus, R. Neutral poly/per-fluoroalkyl substances in air from the Atlantic to the Southern Ocean and in Antarctic snow. Environ. Sci. Technol. 2015, 49 (13), 7770−7775. (45) Xie, Z.; Zhao, Z.; Möller, A.; Wolschke, H.; Ahrens, L.; Sturm, R.; Ebinghaus, R. Neutral poly-and perfluoroalkyl substances in air and seawater of the North Sea. Environ. Sci. Pollut. Res. 2013, 20 (11), 7988−8000. (46) Shoeib, M.; Vlahos, P.; Harner, T.; Peters, A.; Graustein, M.; Narayan, J. Survey of polyfluorinated chemicals (PFCs) in the atmosphere over the northeast Atlantic Ocean. Atmos. Environ. 2010, 44 (24), 2887−2893. (47) Shoeib, M.; Harner, T.; Vlahos, P. Perfluorinated chemicals in the Arctic atmosphere. Environ. Sci. Technol. 2006, 40 (24), 7577− 7583. (48) Oono, S.; Harada, K. H.; Mahmoud, M. A.; Inoue, K.; Koizumi, A. Current levels of airborne polyfluorinated telomers in Japan. Chemosphere 2008, 73 (6), 932−937. (49) Barber, J. L.; Berger, U.; Chaemfa, C.; Huber, S.; Jahnke, A.; Temme, C.; Jones, K. C. Analysis of per-and polyfluorinated alkyl substances in air samples from Northwest Europe. J. Environ. Monit. 2007, 9 (6), 530−541. (50) Dreyer, A.; Weinberg, I.; Temme, C.; Ebinghaus, R. Polyfluorinated compounds in the atmosphere of the Atlantic and Southern Oceans: evidence for a global distribution. Environ. Sci. Technol. 2009, 43 (17), 6507−6514.

4470 DOI: 10.1021/acs.est.6b04021 Environ. Sci. Technol. 2017, 51, 4461−4470