Environmental Pollution 206 (2015) 104e112

Contents lists available at ScienceDirect

Environmental Pollution journal homepage: www.elsevier.com/locate/envpol

Characterizing direct emissions of perfluoroalkyl substances from ongoing fluoropolymer production sources: A spatial trend study of Xiaoqing River, China Yali Shi a, Robin Vestergren b, Lin Xu a, Xiaowei Song a, Xiameng Niu a, c, Chunhui Zhang c, Yaqi Cai a, * a

State Key Laboratory of Environmental Chemistry and Ecotoxicology, Research Center for Eco-Environmental Science, Chinese Academy of Sciences, Beijing 100085, China Norwegian Institute for Air Research (NILU), 9296 Tromsø, Norway c School of Chemical & Environment Engineering, China University of Mining and Technology, Beijing 10083, China b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 1 May 2015 Received in revised form 26 June 2015 Accepted 27 June 2015 Available online xxx

The spatial trend of perfluoroalkyl substances (PFASs) along Xiaoqing River and its tributaries was studied to characterize isomer profiles and quantify emissions from fluoropolymer (FP) manufacturers in P China. Substantially elevated PFAS concentrations downstream of tributary 4 demonstrated that the emissions from this FP manufacturer dominated total riverine discharges. Isomer profiles of perfluorooctanoic acid (PFOA) in water displayed a stepwise increase in percentage branched PFOA downstream of tributary 3 (14.0%) and 4 (22.7%) reflecting the importance of FP sources. Strong positive correlations between PFOA isomers in water downstream of tributary 4 indicated that isomer profiles were conserved from emission sources to the final reservoir. Riverine discharges of PFOA (23e67 t/yr) were in agreement with theoretical emission calculations from FP production (68 t/yr) whereas large discrepancies between the two methodologies were observed for perfluorobutanoic acid and perfluoropentanoic acid. Collectively, this study fills critical knowledge gaps for understanding ongoing global sources of PFASs. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Per- and polyfluoroalkyl substances (PFASs) PFOA isomers Source tracking Isomer fractionation

1. Introduction Per- and polyfluoroalkyl substances (PFASs) are a group of manmade chemicals which have been widely used in commercial products and industrial processes due to their unique amphiphilic properties and chemical stability (Posner, 2012). However, the same properties that make PFASs useful in commercial applications also make them problematic contaminants when released to the environment (Buck et al., 2011) and the global contamination of PFASs has attracted significant attention from scientists and regulators (Lindstrom et al., 2011). Perfluorooctane sulfonic acid (PFOS) and perfluorooctanoic acid (PFOA) have been recognized as two of the most problematic PFASs due to their persistence, ubiquitous presence in environmental samples, wildlife, and humans (Buck et al., 2011) and toxicity in animal models (Seacat et al., 2003;

* Corresponding author. E-mail address: [email protected] (Y. Cai). http://dx.doi.org/10.1016/j.envpol.2015.06.035 0269-7491/© 2015 Elsevier Ltd. All rights reserved.

Lau et al., 2007). The adverse environmental profile of PFASs has led to a series of regulatory actions and voluntary phase-out initiatives by the chemical industry. Between 2000 and 2002, the major global manufacturer of PFOS and PFOA phased out production of perfluorooctyl compounds (3M, 2000, 2003; Buck et al., 2011). In 2006, eight other leading global companies agreed to a stewardship program to reduce emissions and product content of PFOA and related chemicals by 95% by 2010 and to work towards their elimination by 2015 (US EPA, 2006). PFOS, its salts and perfluorooctylsulfonyl fluoride (POSF) were listed as Persistent Organic Pollutants (POPs) in Annex B of the Stockholm Convention in 2009 (UNEP, 2009), and PFOA and its ammonium salt (ammonium perfluorooctanoate, APFO) were added to the European chemicals regulation REACH candidate list as Substance of Very High Concern (SVHC) in 2013 (European Chemicals Agency, 2013). The initiatives to phase out perfluorooctyl compounds including PFOS and PFOA have been generally achieved by substituting to short-chain PFAS homologues. A parallel trend is the geographical shift in the

Y. Shi et al. / Environmental Pollution 206 (2015) 104e112

production and use of PFASs from Europe and North America to emerging economies including China (Wang et al., 2014a). This change of place of production has also led to a continued or increasing production and use of perfluorooctyl compounds which have largely been phased out in Europe and North America (Xie et al., 2013; Wang et al., 2014a). Although, numerous studies have reported highly elevated concentrations of both legacy and replacement PFASs in environmental samples and human blood serum from China (Bao et al., 2011; Zhou et al., 2013, 2014; Shan et al., 2014), a comprehensive understanding of the major emission sources is lacking. Global source inventories have demonstrated that the historical emissions of PFOA can be largely attributed to the use of APFO in fluoropolymer (FP) production (Prevedouros et al., 2006; Cousins et al., 2011). The relative importance of emission sources may, however, vary considerably for different PFCA homologues and over time as a consequence of phase out actions (Wang et al., 2014a). For example, direct emission from manufacture and use of APFO can account for 98e100% of the historical emissions of PFOA, whereas impurities of PFCAs in fluorotelomer products (Washburn et al., 2005) or degradation of fluorotelomer precursors (Ellis et al., 2004) are relatively more important for C4eC7 PFCAs (Wang et al., 2014a). Although theoretical cumulative emissions of PFOA and perfluorononaoic acid (PFNA) are broadly consistent with ocean water inventories based on measurement data (Armitage et al., 2006, 2009a, b; Prevedouros et al., 2006), large discrepancies between modeled and observed values are observed for short-chain PFCAs (CnF2nþ1COOH, n < 7) (Wang et al., 2014a). Further studies are therefore needed to evaluate the assumptions of global emission inventories which are sometimes derived from limited empirical data (Wang et al., 2014a). A different approach to investigate the emission sources of PFOA is the use of isomer profiles (Benskin et al., 2010a, b; 2012a). The principal idea behind this source apportionment tool is that the two major synthesis routes to produce PFOA result in distinct profiles of branched and linear isomers. Provided that ocean water is the final reservoir of PFOA, isomer profiles in water samples from industrialized or remote regions can be used for tracking production sources (Benskin et al., 2010b, 2012a; Yu et al., 2013; Fang et al., 2014a, 2014b). Electrochemical fluorination (ECF), which results in a mixture of branched and linear isomers, was used to produce the majority of APFO in Europe and North America between 1951 and 2000 (Prevedouros et al., 2006). Since 2003, the production of ECFproducts has rapidly increased in China to meet the demand of APFO as a polymerization aid in FP production (Wang et al., 2014a). Telomerization, which was developed in the 1970s, is distinct from ECF in that the composition of products is isomerically pure and typically >99% linear (Buck et al., 2011). However, production of isopropyl PFOA using the telomerization process has also been reported (Kissa, 2005; Benskin et al., 2010a). Although telomerization has been widely used to produce APFO in Europe and North America in the period 2003e2015, its use in China has been fairly limited (Ruan et al., 2010; Fuxin, 2013; Wang et al., 2014a). Fluorotelomer products, some of which can degrade to PFCAs in the environment (Wallington et al., 2006; Wang et al., 2009; Butt et al., 2010), are however used in numerous consumer product applications including textile and paper treatment (Wang et al., 2014a). A complicating factor of using isomer profiles for source elucidation is, however, that the differences in physicoechemical properties of structural isomers may lead to a fractionation in the environment (Benskin et al., 2010a; K€ arrman et al., 2011). Furthermore, it is not well established if the isomer profiles in Chinese APFO products are similar to historically used products in North America and Europe (Benskin et al., 2010a). The aim of this study was to advance our understanding of PFAS

105

emissions related to ongoing FP production with particular emphasis on (i) evaluating and improving methods for emission estimation and (ii) evaluating the applicability of isomer profiles as a source apportionment tool for ongoing sources. To this end, we assessed the spatial trends, isomer profiles and riverine discharges of PFASs in Xiaoqing River which receives emissions from one of the major FP producers in China (Wang et al., 2014a, b). Although the focus of this paper is on identifying and quantifying sources of PFCAs, other PFASs are discussed with respect to their contrasting sources, transport and fate. 2. Materials and methods 2.1. Study area Xiaoqing River is located in Shandong province on the eastern edge of the North China plain stretching from the approximate coordinates 36 380 e37 160 north and 116 480 e118 520 east. The 216 km long Xiaoqing River has an SWeNE orientation and flows parallel to the Huang He River before emptying into Laizhou Bay of the Bohai Sea. Xiaoqing River has a catchment area of 13 000 km2 and five major tributaries feeding into the main river stream (Gao et al., 2014). The river basin is densely populated, flowing through the major cities of Jinan, Zibo, Binzhou, Dongying, and highly industrialized. The average annual precipitation is about 620 mm, with more than 60% of the rainfall during June, July and August. The average river water flux to the Bohai Sea has been estimated to 1.9  109 m3 per anumn (m3/yr) (Wang et al., 2014b). Previous monitoring studies performed by our research group (Table S3) and others (Wang et al., 2014b) have observed highly elevated concentrations of PFCAs in water samples from Xiaoqing River. Based on a recent emission inventory (Wang et al., 2014a) and detailed studies of Chinese industry patent literature (Tang et al., 2009; Cao et al., 2010; Wang et al., 2010; Xie et al., 2011; Xia and Wei, 2013), we identified two FP production facilities with suspected discharge to Xiaoqing River (Fig. 1) (Wang et al., 2014a, b). One FP production facility located in tributary 4 (T4) belongs to the Dongyue group and is currently one of the major facilities for polytetrfluoroethylene (PTFE) production in China with a reported production of 37 000 tonnes per anumn (t/yr). The same facility also produces approximately 500 t/yr of perfluorinated ethylene-propylene copolymers (FEP), 300 t/yr of polyvinylidene fluoride (PVDF) and 40 t/yr of APFO. The other FP production facility at tributary 3 belongs to the company 3F Jinan and has a substantially lower production capacity (100e1000 t PTFE/yr) (Wang et al., 2014b). 2.2. Sampling of water and sediment Based on suspected point sources, the sampling area was divided into four sections: urban and waste water effluents (Section 1), minor FP point source (Section 2), major FP source (Section 3) and marine estuary (Section 4). In total, 36 surface water (W1eW36) and 33 sediment samples (S1eS28, S32, S34eS36) were collected every 5e10 km from the main river stream and the Bohai Sea. One or two water (TW1, TW2, TW3-1, TW3-2, TW4-1, TW4-2, TW5) and sediment samples (TS1, TS2, TS3-1, TS3-2, TS41, TS4-2 and TS5) were also collected from each of the five main tributaries. All samples included in the spatial trend were collected on April 24e26th, 2014. Prior to the spatially resolved sampling campaign, two pilot studies on water samples from Xiaoqing River had been conducted. In the first pilot study, six water samples were collected on May 29th 2013 between site 22e24. In the second pilot study, five water samples were collected on August 28th 2013 from sampling sites 22e26.

106

Y. Shi et al. / Environmental Pollution 206 (2015) 104e112

Fig. 1. Spatial trends of PFASs in water and sediment samples from the main river stream, tributaries and estuary of Xiaoqing River. In order to illustrate the spatial variability in P PFAS concentrations, the area was divided into four different sections (note the shifts in y-axis values between sections). Water and sediment samples from marine water outside the Xiaoqing estuary are displayed in Section 4.

2.3. Sampling of technical APFO products Technical APFO mixtures were collected from Siyang Qingyun Fine Chemical Co (No. 1), Shandong Huaxia Shenzhou New Material Co., Ltd. (No. 2) and Shanghai Fluorine Fine Chemical Co., Ltd. (No.

3e5), which represent the three major APFO producers in China. Sample No. 1e4 represent major technical products of APFO which altogether are produced in quantities up to 200 t/yr. In contrast, product No. 5 is reported to be sold in lower quantities due to its high production cost. All product samples were provided with an

Y. Shi et al. / Environmental Pollution 206 (2015) 104e112

approximate percentage of linear (L-) PFOA. For product samples from Shanghai Fluorine Fine Chemical Co., Ltd (No. 3e5), three technical products with different isomeric purity grades were available corresponding to 73%, 78%, and 90% L-PFOA. 2.4. Standards and reagents All samples were analyzed for 16 PFASs, including perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), PFOA, PFNA, perfluorodecanoic acid (PFDA), perfuoroundecanoic acid (PFUnDA), perfluorododecanoic acid (PFDoDA), perfluorotridecanoic acid (PFTrDA), perfluorotetradecanoic acid (PFTeDA), perfluorohexadecanoic acid (PFHxDA), perfluorooctetradecanoic acid (PFODA), perfluorobutane sulfonic acid (PFBS), perfluorohexane sulfonic acid (PFHxS) and PFOS. In addition, structural isomers of PFOA including L-PFOA and four brPFOAs (iso-, 5m-, 4m- and 3m-PFOA) (using the terminology of Benskin et al. (2012b)) were analyzed separately. All native and mass-labeled linear PFAS isomers were purchased as commercial mixtures from Wellington laboratories. For quantification of branched isomers the T-PFOA mixture (79% L-PFOA, 9% iso-PFOA, 3% 3m-PFOA, 4% 4m-PFOA, 4.5% 5m-PFOA) from Wellington standards was used. Detailed information about chemicals and reagents is given in the Supporting information. 2.5. Analysis and quantification All samples were treated according to previously developed methods (Zhou et al., 2013) and details about the extraction and clean-up procedures are given in the Supporting information. Instrumental separation and quantification of PFASs was performed using a high performance liquid chromatography (HPLC), including a dual pump and autosampler (Ultimate 3000, ThermoFisher Scientific Co.), coupled with electrospray ionization tandem mass spectrometry (ESI-MS/MS, API 4500, Applied Biosystems/MDS SCIEX, USA). The method used for linear PFASs quantification employed a regular C18 stationary phase (Zhou et al., 2013), whereas a similar methodology to that of Benskin et al. (2012b) was used for isomer-specific quantification of PFOA. Separation was achieved using an Acclaim 120 C18 column (5 mm, 4.6 mm  150 mm, ThermoFisher Scientific Co.) for linear PFASs and Ascentis Express F5 PFP Column (2.7 mm, 90 Å, 10 cm  2.1 mm, SigmaeAldrich) for PFOA isomers, respectively. Details of the mobile phase composition and HPLC gradient are provided in the Supporting information. Quantification was performed using the internal standard calibration curve consisting of a concentration gradient (0.05, 0.1, 0.5, 1.0, 5, 10, 20, 50 and 100 mg/L), spiked with 2 ng 13 C4PFBA,13C4PFOA, 13C2PFDoDA, and 13C4PFOS, using a 1/x2 weighted regression with a correlation coefficient greater than 0.99 for each analyte. Employing the methodology of Benskin et al. (2012b), specific product ions were used to individually quantify branched isomers of PFOA (see also Supporting information). The matrix-specific limits of detection (LOD) and limits of quantification (LOQ) were determined at the lowest concentration resulting in a signal-to-noise ratio (S/N)  3 and (S/N)  10, respectively. The total organic carbon (TOC) of sediment was determined by an O.I. Analytical Solids TOC Analyzer (O.I. Analytical, USA).

107

bottles filled with milli-Q water. However, all procedural and field blank samples were consistently below LOD. Solvent blank injections were performed to monitor sample carry over. Instrumental drift was monitored by injecting a calibration standard after every 10 injections and a new calibration curve was constructed if a deviation of ±20% from its initial value was observed. Spikerecovery tests of PFASs were performed for all matrices (n ¼ 4) to evaluate the accuracy and precision of reported data. The recoveries ranged from 70.4 ± 8.7% to 109.2 ± 4.1% in water and from 81.5 ± 2.1% to 103.3 ± 4.0% in sediment for all analytes except PFODA (Table S2). Briefly, 2 ng of individual linear PFAS homologues and 10 ng of T-PFOA for branched isomers of PFOA were added to low contaminated samples prior to the addition of internal standard. Specific measures were taken to ensure that all quantified concentrations were within the dynamic range. Most importantly, the amount of sample used for extraction was adjusted depending on the sampling location so that the analytes and their internal standards would be present at quantifiable concentrations in the final extract. For sample TW4-1, TW4-2 and W19 further dilution and re-quantification of the extracts was necessary due to the extremely high concentrations. The reproducibility of measurements in highly contaminated samples was evaluated by duplicate extraction and analysis of 11 water samples (Table S4). 2.7. Data analysis Statistical analysis of concentration data was performed using IBM PASW statistics 18.0 (SPSS Inc., 1993e2007) with a statistical significance threshold of p < 0.05. Spearman rank analysis was used to investigate correlations between PFASs concentrations in water and sediment samples. The KruskaleWallis H test was applied to examine differences in PFAS concentrations between river sections and differences in field-based distribution coefficients of PFOA isomers. In the statistical analysis, the concentrations of PFASs were treated as zero if the compounds were below LOD in samples. Field-based sediment/water, distribution coefficients (Kd, L/kg), organic carbon/water distribution coefficients (Koc, L/kg) were calculated according to the following equations

Kd ¼

Cs Cw

(1)

Kd foc

(2)

Koc ¼

where Cs and Cw are the concentrations of PFASs in sediment (ng/kg dw), water (ng/L) samples respectively and foc is the fraction of organic carbon in sediment samples (unitless). Although fieldbased distribution coefficients should not be confused with partitioning coefficients determined at equilibrium, Kd and Koc were estimated here to evaluate differences in distribution between PFAS homologues and PFOA isomers. Theoretical emission estimates from FP and APFO manufacture (EFP and EAPFO) were calculated and compared to empirically derived river discharges. The total emissions of PFCAs (Etot, t/yr) were calculated for the major FP and APFO manufacturing facility in T4 using the following equations

EFP ¼ PFP *URAPFO *EFAPFO *FW;FP *HD

(3)

2.6. Quality assurance

EAPFO ¼ PAPFO *EFAPFO *FW; APFO *HD

(4)

All laboratory consumables and solvents were routinely checked for contamination and one procedural blank sample was conducted in every batch of ten samples. Field blanks included poly-propylene

Etot ¼ EFP þ EAPFO

(5)

108

Y. Shi et al. / Environmental Pollution 206 (2015) 104e112

where PFP is the production volume of FP (t/yr), URAPFO is the use rate of APFO per produced unit of FP on a mass basis (t APFO/t FP), EFAPFO is the emission factor (unitless), FW is the fraction of emissions to water for FP and APFO production respectively (unitless), HD is the homologue distribution of PFCAs in APFO technical solutions (g PFCA/g APFO), PAPFO is the production rate of APFO and EFAPFO is the emission factor from APFO production. PFP and PAPFO were estimated based on the reported production capacity of 37 000 t/yr of PTFE, 500 t/yr of FEP and 40 t/yr of APFO in 2012 (Dongyue Group Limited, 2012) and HD was derived from the measurements of impurities in technical APFO performed here (Section 3.2). URAPFO, EFAPFO, FW, and EFAPFO were parameterized according to the high, low and most plausible scenarios described by Wang et al. (2014a) (Table S5eS8). The annual riverine discharge of PFCAs from Xiaoqing River, Nriver (t/yr), was calculated according to

Nriver ¼

Criver *Qriver 1000

(6)

where Criver is the PFCA concentration (kg/m3) and Qriver is the river water flux (m3/yr). The denominator of 1000 was used to harmonize with the units of Etot estimates. Criver values were derived from measured concentrations of individual PFCA homologues in water samples close to the river mouth (W22eW26) whereas literature values for the annual average water flux of 60.3 m3/s (1.9∙109 m3/ yr) (Wang et al., 2014b) was used for Qriver. Three different scenarios were calculated for Nriver corresponding to the median, max and min concentrations in water samples collected at two different time points (August 2013, April 2014) to reflect the uncertainty range in riverine discharges of PFCAs (Table S8).

3.3. Spatial trends of PFASs in water and sediment The spatial variation of PFASs in water from Xiaoqing River is depicted in Fig. 1 and summary statistics of PFASs for river Sections P 1e4 are presented in Table S9. PFAS concentrations in Section 1 increased four-fold from the initial concentration of 36.5 ng/L at P PFAS concensample point W1 to W12. Two distinct peaks in trations were observed at T3 (1310 ng/L) and T4 (496 000 ng/L). The peak concentrations in tributary 4 were followed by an increase in P median PFAS concentrations of more than two orders of magnitude in Section 3 compared to Section 2. A marked decrease P in PFAS concentrations was observed at the first marine sampling station (W29) and concentrations decreased with increasing distance to the Xiaoqing River mouth. The large spatial variation in concentrations was primarily observed for PFOA and short-chain PFCAs which ranged 34 orders of magnitude. Differences in median concentrations between the river sections were found to be statistically significant (p < 0.05) for all analytes except PFUnDA using the Kruskal Wallis H-test. In contrast to PFCAs, the concentrations of PFSAs displayed small or non-significant differences along the main river stream. Spatial trends in sediment samples generally mirrored those in water with peak concentrations of P PFAS at T3 (244 ng/gdw) and T4 (4100 ng/g dw). Significantly P higher median PFAS concentrations were also observed in sediment samples from Section 3 compared to Sections 1 and 2 (P < 0.05). However compared to water samples, the concentraP tions of PFASs in sediment displayed a more irregular decrease with distance to peak concentrations at T4. A higher relative abundance of C9eC11 PFCAs and PFOS in sediment compared to water was also observed in Sections 1 and 2. 3.4. Isomer profiles of PFOA in environmental samples

3. Results 3.1. Concentrations of PFASs in environmental samples P Detection frequencies, PFASs concentrations and homologue profiles of PFASs in environmental media varied considerably for the different matrices and sampling sites (Table S9eS11). All analytes, except PFDoDA, PFTrDA, PFTeDA, PFHxDA and PFODA, were P detected in surface water with PFAS concentrations ranging from 36.5 to 496 000 ng/L. PFOA was consistently the dominant homologue in all water samples followed by PFHxA and PFHpA. In sediment, all PFASs except PFODA were detected although the detection frequency varied greatly for the different homologues P and sampling sites. PFASs in sediment ranged from 0.333 to 4100 mg/kg dw and PFOA was the main homologue (18.1e95% of P PFASs). 3.2. Isomer profiles and homologue composition of technical APFO products Isomer profiles of PFOA and homologue composition of PFCAs in five different T-APFO products are presented in Table S12. The relative abundance of branched isomers displayed a consistent pattern of iso-PFOA > 5m-PFOA > 4m-PFOA > 3m-PFOA in all product samples, although the total percentages of L-PFOA isomers varied from 73.9 to 90.4%. Overall, the measured percentage of LPFOA agreed well with those reported by producers (deviation of 0.4e2.9% between average measured value (n ¼ 3) and those reported by the producers). The composition of PFCA impurities in APFO products ranged from 0.4 to 3.7% with PFHpA being consistently the dominant homologue followed by PFHxA and PFNA. The percentage of PFBA and PFPeA varied between the products, but was consistently 5mPFOA  4m-PFOA > 3m-PFOA) in water samples from all sections. Strong positive correlations were also observed for individual isomers of PFOA for Sections 3 and 4 (r > 0.971; p < 0.001). Sediment samples displayed a consistently higher percentage of L-PFOA compared to water samples (Fig. 2). The percentage of LPFOA in sediment varied between 80.8 and 100%. Iso-PFOA was the dominant branched isomer followed by 5m-, 4m- and 3m-PFOA, accounting for 4.77e7.68%, 0.06e4.65%, 0e4.39% and 0.99e3.62% respectively in sediment samples downstream of tributary 4. The spatial trend in sediments resembled that of surface water with an increasing percentage of branched isomers from Sections 1e3. However, the lower detection frequency of branched isomers in Section 4 (outside of the dynamic range) hampered a quantitative comparison of isomer profiles between Sections 3 and 4. 3.5. Field-based distribution coefficients of PFOA isomers Field-based distribution coefficients for PFOA isomers are presented in Table S18 of the supporting information. Log Kd values (arithmetic mean ± standard error) for PFOA isomers in paired water-sediment samples downstream of tributary 4 were

Y. Shi et al. / Environmental Pollution 206 (2015) 104e112

109

Fig. 2. PFOA Isomer profiles in surface water and sediments from the main river stream and major tributaries (denoted TW and TS respectively) of Xiaoqing River. The numbers 1e4 denote the river section according to the map in Fig. 1. The five analyzed APFO products are displayed for comparison (for more details see Table S12).

0.41 ± 0.37, 0.24 ± 0.39, 0.20 ± 0.41, 0.18 ± 0.35 and 0.13 ± 0.40 for L, 3m-, 4m-, iso- and 5m-PFOA respectively. Kd of L-PFOA was found to be significantly higher compared to monomethyl branched isomers (p < 0.05). However, no statistically significant differences in Kd values were observed for monomethyl branched isomers due to the variability in the sediment data. Normalizing Kd to fOC for the sediment samples significantly reduced the variability in field based sorption coefficients (Table S18) for PFOA isomers. However, no statistically significant correlations were observed between Kd and fOC. The lack of a correlation between Kd and fOC is somewhat surprising considering that organic carbon has been identified as the primary predictor of sediment sorption (Higgins and Luthy, 2006). The conflicting results between field-based observations and controlled batch experiments can probably be explained by the contribution of an inorganic fraction to the sorption of PFOA onto sediment (Johnson et al., 2007) and the fact that equilibrium of PFOA between surface water and sediments might not have been reached in this study. 3.6. Theoretical emission predictions and empirical riverine discharges Fig. 3 displays theoretical emission estimates and riverine discharges derived from measured water concentrations for C4eC8

PFCAs. For PFOA, the most plausible emission scenario (68 t/yr) was in good agreement with the range of riverine discharges (23e67 t/ yr). Theoretical emissions for C4eC7 PFCAs displayed a larger uncertainty compared to PFOA due to the wide range of short-chain PFCA impurities in technical APFO products (Table S12). Considering the range of plausible emission estimates, a reasonable overlap with riverine discharges was observed for PFHpA and PFHxA. However, for PFPeA and PFBA the theoretical emissions underestimated riverine discharges by 3e4 orders of magnitude. 4. Discussion 4.1. Identifying the sources of PFASs to Xiaoqing River The spatially resolved sampling of this study allowed us to assess the contribution of known point sources compared to diffuse emissions to the PFAS concentrations in Xiaoqing River. Overall, the peak PFAS concentrations at T3 and T4 coincided well with identified FP production plants. Furthermore, the significant increase in P PFAS concentrations from Sections 2e3 was consistent with the substantially higher production capacity of the FP facility at T4. Based on these results we conclude that the FP production facility at T4 was the dominant source of PFASs to Xiaoqing River. The good agreement between repeated sampling campaigns in river Section

110

Y. Shi et al. / Environmental Pollution 206 (2015) 104e112

Fig. 3. Comparison between theoretical (gray) and empirical (white) emission estimates for C4eC8 PFCAs. Solid bars represent the most plausible emission scenario (use rate of 0.3 wt% PFOA-based products as processing aids in FP production) and median water concentrations respectively. The error bars represent the range of plausible emissions and max/ min PFCA concentrations in water respectively.

3 (Table S3) and qualitatively similar spatial trends in sediment and water further indicate that these conclusions were not severely biased by fluctuations in emissions or hydrological conditions. Generally, the concentrations of PFASs in water from Section 3 of Xiaoqing River were among the highest ever reported (Lein et al., 2008; Bao et al., 2011; Shan et al., 2014; Wang et al., 2014b). Concentrations of PFASs in marine waters outside of the Xiaoqing estuary (W29eW36: 586e8120 ng/L) were also 2e3 orders of magnitude higher than concentrations previously reported for coastal seawater in China (Cai et al., 2012) and elsewhere (Zhao et al., 2012; Benskin et al., 2012a; Takemine et al., 2014). It P should, however, be noted that the spatial trend of PFASs was dominated by C4eC9 PFCAs which are primarily associated with FP production. The lack of clear spatial trends for PFSAs indicates that input of these substances are likely caused by other sources from the densely populated area including street run-off and waste water treatment discharges (Müller et al., 2011; Filipovic et al., 2013). 4.2. Isomer profiles as source tracking tool The theory of using isomer profiles for source elucidation relies heavily on the assumption that ECF-based APFO has got a consistent composition of 78% linear and 22% branched isomers (Benskin et al., 2010a). Although APFO from the major historical manufacturer displayed a remarkably consistent isomer profile in 18 production lots over 20 years (Benskin et al., 2010a), it is plausible that isomer profiles shift if production procedures change (Rayne et al., 2008). Our measurements demonstrate that the isomer composition in Chinese APFO products varies between 73.9 and 90.4% (Table S12). According to the limited information we received from APFO manufacturers, product sample No. 1e4 represent major technical products which are produced in large quantities. In contrast, product sample 5 (90.4% L-PFOA) is produced and sold in minor quantities. A reasonable assumption is, therefore, that the isomer composition of Chinese APFO products generally varies between 74 and 80% L-PFOA. Given this range in technical ECF products, the relatively low %br-PFOA in Section 1 (median 5%)

suggests that PFOA impurities in fluorotelomer products (Washburn et al., 2005) or degradation of fluorotelomer based precursors (Wallington et al., 2006; Wang et al., 2009; Butt et al., 2010) were the primary sources of PFOA upstream of the FP facilities. Contrastingly, the %br-PFOA in surface water from T3 and T4 (21.4e22.5%) were in very good agreement with isomer profiles in APFO products No. 1e4 (19.8e26.1%). This agreement in isomer profiles between APFO products and surface water samples demonstrates that the majority of PFOA downstream of FP production effluents in T4 was of ECF origin. Given the variability in %br-PFOA of APFO products No. 1e4, it is, however, difficult to quantify the exact proportion ECF emissions using the methodology of Benskin et al. (2010b). Nevertheless, the consistent relative abundance of isomers (iso-PFOA > 5m-PFOA > 4m-PFOA > 3m-PFOA) together with an approximate range of 20e26% br-PFOA would still be a useful qualitative signature that could be used to distinguish between ongoing sources of PFOA in China. Another important aspect of the isomer-profile methodology is to what extent differences in physicalechemical properties of branched and linear PFOA may lead to environmental fractionation. The strong correlations between all individual PFOA isomers (r > 0.885; p < 0.001) in surface water from the TW4-1 to W28 demonstrate that isomer profiles are well conserved in water P samples over a wide range of PFOA concentrations, organic carbon content and salinity. Although L-PFOA was significantly enriched in sediments, this fractionation process did not have a significant impact on the isomer profiles in water samples. The explanation for these seemingly divergent findings is likely that advective transport of PFOA with the bulk flow of water is a far more important transport process than sedimentation (Armitage et al., 2006; Higgins and Luthy, 2006). Analogously, the homologue patterns of C4eC8 PFCAs were very well conserved (r > 0.948; p < 0.001) in water samples downstream of tributary 4 which indicate that all of these homologues transported in a similar way with the bulk flow of water.

Y. Shi et al. / Environmental Pollution 206 (2015) 104e112

4.3. Implications for understanding the global sources of PFASs The use of APFO as a polymerization aid in fluoropolymer manufacture has been identified as the major historical emission source of PFCAs on a global scale (Armitage et al., 2006; Prevedouros et al., 2006; Cousins et al., 2011), but little empirical data has been available to evaluate estimated ongoing emissions (Wang et al., 2014a). The good agreement between theoretical emission estimates and riverine discharges for PFOA provides some confidence to the uncertain parameter values used by Wang et al. (2014a) to estimate emissions from FP production (e.g. the use rates of APFO). By comparing the riverine discharges calculated here with the total global emissions by Wang et al. (2014a) we further conclude that approximately 10e30% of the ongoing global PFOA emissions can be attributed to the FP production facility tributary 4 of Xiaoqing River. Theoretical emission calculations for C4eC7 PFCAs were associated with a much higher uncertainty due to the variable homologue distributions in APFO products (Table S12). Despite this uncertainty, the large discrepancy between theoretical and empirical emission estimates (especially for PFBA and PFPeA) indicates that emission processes for these homologues from FP production are not fully understood. One plausible explanation is that short-chain PFCAs are used as polymerization aids in combination with APFO. This practice has been reported in patents from FP producers in China (Cao et al., 2010; Xia and Wei, 2013) and elsewhere (Wang et al., 2014c), but has so far not been quantified in global emission inventories (Wang et al., 2014a). Interestingly, the riverine discharges of PFBA (1.3 t/yr) and PFPeA (1.1 t/yr) from Xiaoqing River are within the range of total global emissions of PFBA (0.4e17 t/yr) and PFPeA (0.4e23 t/yr) from all other quantified sources during the period 2003e2015 (Wang et al., 2014a). Thus, inclusion of hitherto overlooked emissions of short-chain PFCAs from FP production could significantly improve the accuracy of global emission estimates of short-chain PFCAs (Wang et al., 2014a). Acknowledgments This work was jointly supported by National Key Basic Research Program of China (2014CB114402), the National Natural Science Foundation of China (No. 21377145, 21321004) and the Strategic Priority Research Program of the Chinese Academy of Sciences (XDB14010201). The financial support from the Norwegian Research Council via research project PFC ChiNo (29666/E40) is also gratefully acknowledged. Dr. Zhanyun Wang is gratefully acknowledged for valuable discussions around global emission inventories of PFCAs and sharing of data. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.envpol.2015.06.035. References 3M Company, 2000. Voluntary Use and Exposure Information Profile for Perfluorooctanoic Acid and Salts. USEPA Administrative Record AR226e0595. www.regulations.gov. as document EPA-HQ-OPPT-2002-0051-0009. 3M Company, 2003. Health and Environmental Assessment of Perfluorooctane Sulfonic Acid and its Salts. U.S. Environmental Protection Agency, Washington, DC. Armitage, J.M., Cousins, I.T., Buck, R.C., Prevedouros, K., Russell, M.H., MacLeod, M., Korzeniowski, S.H., 2006. Modeling global-scale fate and transport of perfluorooctanoate emitted from direct sources. Environ. Sci. Technol. 40 (22), 6969e6975. Armitage, J.M., MacLeod, M., Cousins, I.T., 2009a. Modeling the global fate and

111

transport of perfluorooctanoic acid (PFOA) and perfluorooctanoate (PFO) emitted from direct sources using a multispecies mass balance model. Environ. Sci. Technol. 43, 1134e1140. Armitage, J.M., MacLeod, M., Cousins, I.T., 2009b. Comparative assessment of the global fate and transport pathways of long-chain perfluorocarboxylic acids (PFCAs) and perfluorocarboxylates (PFCs) emitted from direct sources. Environ. Sci. Technol. 43, 5830e5836. Bao, J., Liu, W., Liu, L., Jin, Y., Dai, J., Ran, X., Zhang, Z., Tsuda, S., 2011. Perfluorinated compounds in the environment and the blood of residents living near fluorochemical plants in Fuxin, China. Environ. Sci. Technol. 45 (19), 8075e8080. Benskin, J.P., De Silva, A.O., Martin, J.W., 2010a. Isomer profiling of perfluorinated substances as a tool for source tracking: a review of early findings and future applications. Rev. Environ. Contam. Toxicol. 208, 111e160. Benskin, J.P., Yeung, L.W.Y., Yamashita, N., Taniyasu, S., Lam, P.K.S., Martin, J.W., 2010b. Perfluorinated acid isomer profiling in water and quantitative assessment of manufacturing source. Environ. Sci. Technol. 44 (23), 9049e9054. Benskin, J.P., Ahrens, L., Muir, D.C.G., Scott, B.F., Spencer, C., Rosenberg, B., Tomy, G., Kylin, H., Lohmann, R., Martin, J.W., 2012a. Manufacturing origin of perfluorooctanoate (PFOA) in Atlantic and Canadian Arctic seawater. Environ. Sci. Technol. 46 (2), 677e685. Benskin, J.P., Ikonomou, M.G., Woudneh, M.B., Cosgrove, J.R., 2012b. Rapid characterization of perfluoralkyl carboxylate, sulfonate, and sulfonamide isomers by high-performance liquid chromatography-tandem mass spectrometry. J. Chromatogr. A 1247, 165e170. Buck, R.C., Franklin, J., Berger, U., Conder, J.M., Cousins, I.T., De Voogt, P., Jensen, A.A., Kannan, K., Mabury, S.A., Van Leeuwen, S.P.J., 2011. Perfluoroalkyl and polyfluoroalkyl substances in the environment: terminology, classification and origins. Integr. Environ. Assess. Manag. 7 (4), 513e541. Butt, C.M., Muir, D.C., Mabury, S.A., 2010. Biotransformation of the 8:2 fluorotelomer acrylate in rainbow trout. 1. In vivo dietary exposure. Environ. Toxicol. Chem. 29, 2726e2735. Cai, M., Zha, Z., Yang, H., Yin, Z., Hong, Q., Sturm, R., Eblinghaus, R., Ahrens, L., Cai, M., He, J., Xie, Z., 2012. Spatial distribution of per- and polyfluoroalkyl compounds in coastal waters from the East to South China Sea. Environ. Pollut. 161, 162e169. Cao, S., Wang, H., Zhou, H., Yu, H., Wu, H., 2010. Patent: Method for Preparing Microemulsion of Fluorinated Ethylene Propylene Powder. CN 100560619. (in Chinese). Cousins, I.T., Kong, D., Vestergren, R., 2011. Reconciling measurement and modeling studies of the sources and fate of perfluorinated carboxylates. Environ. Chem. 8, 339e354. Dongyue Group Limited, 2012. 2012 Annual Report (March 10 2014). http://www. dongyuechem.com/UploadFiles/2013-3/News/20133201940820043.pdf. Ellis, D.A., Martin, J.W., De Silva, A.O., Mabury, S.A., Hurley, M.D., Andersen, M.P.S., Wallington, T.J., 2004. Degradation of fluorotelomer alcohols: a likely atmospheric source of perfluorinated carboxylic acids. Environ. Sci. Technol. 38, 3316e3321. European Chemicals Agency, 2013. Candidate List of Substances of Very High Concern for Authorisation. http://echa.europa.eu/web/guest/candidate-listtable. Fang, S., Chen, X., Zhao, S., Zhang, Y., Jiang, W., Yang, L., Zhu, L., 2014a. Trophic magnification and isomer fractionation of perfluoroalkyl substances in the food web of Taihu Lake, China. Environ. Sci. Technol. 48 (4), 2173e2182. Fang, S., Zhao, S., Zhang, Y., Zhong, W., Zhu, L., 2014b. Distribution of perfluoroalkyl substances (PFASs) with isomer analysis among the tissues of aquatic organisms in Taihu Lake, China. Environ. Pollut. 193, 224e232. Filipovic, M., Berger, U., McLachlan, M.S., 2013. Mass balance of perfluoroalkyl acids in the Baltic Sea. Environ. Sci. Technol. 47, 4088e4095. Fuxin, 2013. Product Catalogue: Fluorinated Surfactants. Fuxin Hengtong Fluorine Chemicals Co. Ltd (in Chinese). http://www.htfluo.com/FluorinatedSurfactant/ 10351464.shtml. Gao, M., Qiu, J., Li, C., Wang, L., Li, H., Gao, C., 2014. Modeling nitrogen loading from a watershed consisting of cropland and livestock farms in China using ManureDNDC, Agriculture. Ecosyst. Environ. 185, 88e98. Higgins, C.P., Luthy, R.G., 2006. Sorption of perfluorinated surfactants on sediments. Environ. Sci. Technol. 40 (23), 7251e7256. Johnson, R.L., Anschutz, A.J., Smolen, J.M., Simcik, M.F., Penn, R.L., 2007. The adsorption of perfluorooctanesulfonate onto sand, clay, and iron oxide surfaces. J. Chem. Eng. Data 52, 1165e1170. €rrman, A., Elgh-Dalgren, K., Lafossas, C., Møskeland, T., 2011. Environmental levels Ka and distribution of structural isomers of perfluoroalkyl acids after aqueous firefighting foam (AFFF) contamination. Environ. Chem. 8, 372e380. Kissa, E., 2005. Fluorinated Surfactants and Repellents, second ed. Marcel Dekker, New York. Lau, C., Anitole, K., Hodes, C., Lai, D., Pfahles-Hutchens, A., Seed, J., 2007. Perfluoroalkyl acids: a review of monitoring and toxicological findings. Toxicol. Sci. 99 (2), 366e394. Lein, N.P.H., Fujii, S., Tanaka, S., Nozoe, M., Tanaka, H., 2008. Contamination of perfluorooctane sulfonate (PFOS) and perfluorooctanoate (PFOA) in surface water of the Yodo River basin (Japan). Desalination 226 (1e3), 338e347. Lindstrom, A.B., Strynar, M.J., Libelo, E.L., Field, J.A., 2011. Guest comment: perfluoroalkyl acid focus issue. Environ. Sci. Technol. 45 (19), 7951e7953. Müller, C.E., Gerecke, A.C., Alder, A.C., Scheringer, M., Hungerbühler, K., 2011. Identification of perfluoroalkyl acid sources in Swiss surface waters with the help of the artificial sweetener acesulfame. Environ. Pollut. 159, 1419e1426.

112

Y. Shi et al. / Environmental Pollution 206 (2015) 104e112

Posner, S., 2012. Perfluorinated compounds: occurrence and uses in products. In: Polyfluorinated chemicals and transformation products, pp. 25e39. Prevedouros, K., Cousins, I.T., Buck, R.C., Korzeniowski, S.H., 2006. Sources, fate and transport of perfluorocarboxylates. Environ. Sci. Technol. 40, 32e44. Rayne, S., Forest, K., Friesen, K.J., 2008. Relative gas-phase free energies for the C3 through C8 linear and branched perfluorinated sulfonic acids: Implications for kinetic versus thermodynamic control during synthesis of technical mixtures and predicting congener profile inputs to environmental systems. J. Mol. Struct. THEOCHEM 869, 81e82. Ruan, T., Wang, Y., Wang, T., Zhang, Q., Ding, L., Liu, J., Wang, C., Qu, G., Jiang, G., 2010. Presence and partitioning behavior of polyfluorinated iodine alkanes in environmental matrices around a fluorochemical manufacturing plant: another possible source for perfluorinated carboxylic acids? Environ. Sci. Technol. 44, 5755e5761. Seacat, A.M., Thomford, P.J., Hansen, K.J., Clemen, L.A., Eldridge, S.R., Elcombe, C.R., Butenhoff, J.L., 2003. Sub-chronic dietary toxicity of potassium perfluorooctanesulfonate in rats. Toxicology 183, 117e131. Shan, G., Wei, M., Zhu, L., Liu, Z., Zhang, Y., 2014. Concentration profiles and spatial distribution of perfluoroalkyl substances in an industrial center with condensed fluorochemical facilities. Sci. Tot. Environ. 490, 351e359. Takemine, S., Matsumura, C., Yamamoto, K., Suzuki, M., Tsurukawa, M., Imaishi, H., Nakano, T., Kondo, A., 2014. Discharge of perfluorinated compounds from rivers and their influence on the coastal seas of Hyogo prefecture. Jpn. Environ. Pollut. 184, 397e404. Tang, Y., Xu, P., Zhou, X., 2009. Recovery and application of perfluorooctanoic acid (PFOA). Organo-Fluorine Ind. 4, 43e47 (in Chinese). United Nations Environment Programme, 2009. Governments Unite to Step-up Reduction on Global DDT Reliance and Add Nine New Chemicals under International Treaty. http://chm.pops.int/Convention/Pressrelease COP4 Geneva 8 May 2009/tabid/542/language/en-US/Default.aspx. US Environmental Protection Agency, 2006. 2010/2015 PFOA Stewardship Program. http://www.epa.gov/opptintr/pfoa/pubs/stewardship/. Wallington, T.J., Hurley, M.D., Xia, J., Wuebbles, D.J., Sillman, S., Ito, A., Penner, J.E., Ellis, D.A., Martin, J., Mabury, S.A., Nielsen, O.J., Sulbaek Andersen, M.P., 2006. Formation of C7F15COOH (PFOA) and other perfluorocarboxylic acids during the atmospheric oxidation of 8:2 fluorotelomer alcohol. Environ. Sci. Technol. 40, 924e930. Wang, N., Szostek, B., Buck, R.C., Folsom, P.W., Sulecki, L.M., Gannon, J.T., 2009. 82 fluorotelomer alcohol aerobic soil biodegradation: pathways, metabolites, and metabolite yields. Chemosphere 75, 1089e1096.

Wang, H., Fang, J., Li, W., Li, X., Hao, H., 2010. Patent: Fluorine-containing Emulsifier for Fluoropolymer Emulsion Polymerization and Preparation Thereof. CN 101745338. (in Chinese). Wang, Z., Cousins, I.T., Scheringer, M., Buck, R.C., Hungerbuehler, K., 2014a. Global emission inventories for C4-C14 perfluoroalkyl carboxylic acid (PFCA) homologues from 1951 to 2030, part I: production and emissions from quantifiable sources. Environ. Int. 70, 62e75. Wang, P., Lu, Y., Wang, T., Fu, Y., Zhu, Z., Liu, S., Xie, S., Xiao, Y., Giesy, J.P., 2014b. Occurrence and transport of 17 perfluoroalkyl acids in 12 coastal rivers in south Bohai coastal region of China with concentrated fluoropolymer facilities. Environ. Pollut. 190, 115e122. Wang, Z., Cousins, I.T., Scheringer, M., Buck, R.C., Hungerbuehler, K., 2014c. Global emission inventories for C4-C14 perfluoroalkyl carboxylic acid (PFCA) homologues from 1951 to 2030, part II: the remaining pieces of the puzzle. Environ. Int. 69, 166e176. Washburn, S.T., Bingman, T.S., Braithwaite, S.K., Buck, R.C., Buxton, L.W., Clewell, H.J., Haroun, L.A., Kester, J.E., Rickard, R.W., Shipp, A.M., 2005. Exposure assessment and risk characterization for perfluorooctanoate in selected consumer articles. Environ. Sci. Technol. 39, 3904e3910. Xia, J., Wei, W., 2013. Patent: Method for Preparing Fluorine-containing Polymer by Using Mixed Fluorine-containing Surfactant. CN 103204961. (in Chinese). Xie, X., Qu, J., Bai, R., Zhang, J., Yang, X., 2011. Patent: Peroxidic Fluoropolyether and its Use in Emulsion Polymerization of Fluorin-containing Monomer. US 2011/ 0207880 A1. Xie, S., Wang, T., Liu, S., Jones, K.C., Sweetman, A.J., Lu, Y., 2013. Industrial source identification and emission estimation of perfluorooctane sulfonate in China. Environ. Int. 52, 1e8. Yu, N., Shi, W., Zhang, B., Su, G., Feng, J., Zhang, X., Wei, S., Yu, H., 2013. Occurrence of perfluoroalkyl acids including perfluorooctane sulfonate isomers in Huai River Basin and Taihu Lake in Jiangsu Province, China. Environ. Sci. Technol. 47 (2), 710e717. € llera, A., Sturm, R., Tang, J., Zhang, G., Ebinghaus, R., 2012. DisZhao, Z., Xie, Z., Mo tribution and long-range transport of polyfluoroalkyl substances in the Arctic, Atlantic Ocean and Antarctic coast. Environ. Pollut. 170, 71e77. Zhou, Z., Liang, Y., Shi, Y., Xu, L., Cai, Y., 2013. Occurrence and transport of perfluoroalkyl acids (PFAAs), including short-chain PFAAs in Tangxun Lake, China. Environ. Sci. Technol. 47 (16), 9249e9257. Zhou, Z., Shi, Y., Vestergren, R., Wang, T., Liang, Y., Xu, L., Cai, Y., 2014. Highly elevated serum concentrations of perfluoroalkyl substances in fishery employees from Tangxun Lake, China. Environ. Sci. Technol. 48, 3864e3874.