Occurrence, partition and environmental risk assessment of per- and polyfluoroalkyl substances in water and sediment from the Baiyangdian Lake, China

This work examined the contamination of poly- and perfluorinated compounds (PFASs) in the water and sediment of the Baiyangdian Lake. The total concentration of PFASs in the surface water varied from 140.5 to 1828.5 ng/L, and the highest concentration of PFASs were observed near the entrance of Fuhe river. The topmost contaminant was sodium perfluorohexanesulfonate (PFHxS) and perfluorooctanoic acid (PFOA) in the north and south of the Baiyangdian Lake respectively, which indicated different contamination sources. The total concentration of PFASs in the sediment varied from 0.48 to 30 ng/g, and the distribution of PFASs in the sediment was similar with that in the surface water. The concentrations of polyfluoroalkyl phosphoric diesters (diPAPs) were three to four orders of magnitude lower than those of perfluorocarboxylates (PFCAs) and PFSAs. Although the pore water and the surface water had similar ΣPFASs, the concentration of perfluorodecanoic acid (PFDA) in pore water was 1.4 to 4.4 times higher than that in surface water, and the concentration of perfluoropentanoic acid (PFPeA) in pore water was 20–70% that in surface water. The results of ecological risk assessment showed that the PFASs were currently of no immediate risk to the aquatic life.

LC-MS grade methanol, acetonitrile, and methyl-tert-butyl ether (MTBE) were purchased from Fisher Scientific (Fair Lawn, NJ, USA). LC grade ammonium acetate waspurchased from Fisher Scientific (Fair Lawn, NJ, USA). Analytical grade sodium hydroxide (NaOH) was purchased from Sinopharm Chemical Reagent Beijing, Co., Ltd. Tetrabutyl ammonium hydrogen sulphate (TBAS) was purchased from J.T. Baker (Phillipsburg, NJ, USA). Oasis ® weak anion exchange solid phase extraction cartridges (WAX; 6 cc, 150 mg) were purchased from Waters (Milford, MA). Milli-Q water was used in all analytical experiments. Sample collection. Surface water and sediment samples were collected at Baiyangdian Lake in March 2016.
Surface water samples were collected with a stainless steel bucket and stored in 1 L polypropylene (PP) containers with a narrow mouth and a screw cap. A total of 15 surface water samples were collected. The corresponding sediment samples were collected with a bottom grab and stored in stainless steel containers.
Sample duplicates and field blanks were collected and analyzed along with laboratory and procedural blanks. The stainless steel bucket, stainless steel containers, and PP bottles were cleaned before use by rinsing sequentially with methanol, distilled water, and then water from the sampling site. All samples were kept in an ice bath during shipping, and all water samples were extracted immediately upon arrival at the laboratory. Figure 1 illustrates the sampling sites.
Sample preparation. Water samples. Both surface water samples and pore water samples were prepared.
Pore water samples were collected by centrifuging the sediment samples from S3, S4, S6-S11, S13, and S15 at 10000 rpm for 10 min. Water samples were prepared according to previously published methods 31 . Water samples were filtered with Whatman GF/F glass microfiber (0.7 μm pore size, 4.7 cm diameter) and then extracted with Oasis WAX solid phase cartridges (150 mg, 6cc, Waters, MA, USA). Before solid phase extraction, each mass-labeled PFSA (2 ng) was spiked into the water sample as internal standard. The WAX cartridge was conditioned with 1% NH 4 OH in methanol (6 mL), followed by methanol (6 mL) and Milli-Q Water (6 mL). Water samples were passed through the conditioned cartridge at 5 mL/min, and the loaded cartridge waswashed firstly with ammonium acetate buffer (pH = 4, 25 mmol/L, 6 mL) and then with MeOH (2 mL). The target compounds instrumental analysis. Target compounds were separated using Ultra-Pressure Liquid Chromatography (UPLC) coupled with a Micromass Xevo-TQD mass spectrometer (UPLC-Xevo-TQD, Waters, USA) operated in the negative electrospray ionization mode. The injection volume was 10 µL. The analytes were separate on a Waters Acquity BEH C18 column (50 mm × 2.1 mm i.d., 1.7 μm) using aqueous ammonium acetate and methanol as the mobile phase with agradient elution program similar to those reported elsewhere. Multiple reactions monitoring of target compounds and optimized mass spectrum parameters were also similar to those reported elsewhere 32 .
Gradient elution was used to separate different compounds by liquid chromatography.For the analysis of PFCAs and PFSAs, the mobile phase consisted of (A) 10 mmol/L ammonium acetate in HPLCgrade water and (B) 10 mmol/L ammonium acetate in 8:2 (v/v) methanol/acetonitrile. To analyze PFPiAs, diPAPs, and PFPAs, the mobile phase consisted of 0.1% NH 4 OH in HPLC grade water (A) and pure methanol (B). The flow rate was 300 μL/min and the injection volume was 10 μL. Multiple reaction monitoring (MRM) of target compounds and optimized mass spectrum parameters followed a reported set of conditions 33 .
The total organic carbon content (TOC) of water and sediment samples was analyzed using a multi N/C 2100 S system (AnalytikJena, Germany) with a procedure similar to those described elsewhere 32 . Quality assurance and quality control. Quality assurance and control measures included field blank, travel blank, procedural blank, calibration curve, spike recoveries (both blank and matrix), and limit of quantification (LOQ). Field blanks were prepared by filling precleaned 1 L collection bottles with laboratory Milli-Q water that was previously determined to be free of PFSAs. Procedural blanks were analyzed with every batch of samples. Procedural and travel blanks were below the corresponding LOQs. Analyte recoveries were checked to determine the accuracy of the methods. Matrix spike recovery tests were performed for both water and sediment. To reduce instrumental background contamination from HPLC or solvents, an isolate trap column was connected between the solvent mixing cell and the six-way valve. Teflon-coated lab ware and glassware were avoided during all steps of sampling, pretreatment, and analysis to minimize contamination. The limits of quantification (LOQs) were defined as the smallest mass of injected compound that could afford a reproducible measurement of peak area within ±20% of the duplicate injection. The LOQ and recoveries for each compound were shown in Table 1. The PFASs concentrations were quantified using external calibration curves consisting of a concentration Because toxicological data of PFASs in sediment are lacking, the sediment PNEC was calculated based on equilibrium distribution. Thus, according to the technical guidance document of the European Union for the risk assessment of chemical substances (TGD) 34 .

Results and Discussions
PFASs in surface water. Among the 15 analyzed PFCAs and PFSAs, ten were detected in the water samples.
The concentrations of PFUnA, PFDoA, PFTrA, PFTeA and PFDS were lower than the LOQs. There must be different contamination sources since PFOA was dominant at S4-S7 however, PFHxS was dominant at other sites. The sites S4-S6 are near the entrance of the Xiaoyi River and the Zhulong River. As one of nine rivers entering the Baiyangdian Lake, the Zhulong River carries abundant wastewater from textile and fur plants and had a high PFOA concentration up to 8397.23 ng/L 29 . The highest concentrations of total PFASs were detected at S14, S15, and S1, which are near the entrance of the Fuhe River. The Fuhe River passes through the Baoding city and carries untreated urban sewage as it flows into the Baiyangdian Lake, and it is thus a principal source of pollution 35 . A photographic film production plant from one of the largest Chinese manufacturers is by the Fuhe River, and previous studies showed that the Fuhe River contained abundant PFHxS (>1000 ng/L). www.nature.com/scientificreports www.nature.com/scientificreports/ The pollutants in the samples from sites other than S4-S7 showed a similar distribution pattern (Fig. 2), which hinted on a common contamination source. The study of the Baiyangdian Lake in 2016 showed that PFOA (up to 8397 ng·L − 1) and PFHxS (up to 1478 ng·L − 1) were the predominant PFASs detected in the surface water 29 , which indicated the common contamination source of S1-S3, S8-S15. The PFSAs concentrations of the surface water collected from the Baiyangdian Lake in October 2010 ranged in 14.8-95.6 ng/L, and the lowest concentration detected in the current study was even higher than the highest concentration previously reported 30 . The comparison indicated that contamination of PFSAs in the Baiyangdian Lake deteriorated since 2008.
All surface water samples were free of PFPAs and PFPiAs, and 6:2 diPAP and 8:2 diPAP were detected in 60% and 73% of the surface water samples, respectively. The concentration ranged in nd-0.134 ng/L and nd-1.43 ng/L for 6:2 diPAP and 8:2 diPAP, respectively,three or four orders of magnitude lower than the concentrations of PFCAs and PFSAs (Fig. 3). The concentration of 8:2 diPAP was clearly much higher than that of 6:2 diPAP. The diPAPs must have come from a contamination source different from that of PFCAs and PFSAs because of their distribution characteristics and the higher concentration of 8:2 diPAP. Since PAPs are primarily used in paper products for food packaging, the diPAPs probably came from domestic sewage and household garbage 36 . Since the degradation of diPAPs to PFCAs can occur in wastewater treatment plants, diPAPs must be both a precursor of PFCAs and a potential fluorinated contaminant of their own 37,38 . To the best of our knowledge, this work is the first report that determined diPAPs in the Baiyangdian Lake. S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 S13 S14 S15  www.nature.com/scientificreports www.nature.com/scientificreports/ The regression analyses of the contaminants in the surface water of the research area indicated that significant correlations (p < 0.05) were found between various compounds. Especially, the correlations between PFOS and PFHxS, PFOS and PFBS, as well as PFHxS and PFBS were all more than 0.93, i.e., these three compounds might share similar sources and transport routes.
PFSAs in pore water. The distribution characteristics of PFCAs and PFSAs in pore water were shown in Fig. 4. Since the content of dissolved organic matter (DOM) in pore water was usually more than one order of magnitude than that in surface water, it was excepted the total concentration of PFASs in pore water would higher than those in surface water. However, the results displayed in Fig. 4 didn't support the expectation. As shown in Fig. 4, total concentration of PFASs in pore water of S3, S4, S6, S7, S8, S13 and S15 were higher than those in correspond surface water, while in sampling sites S9, S10 and S11, the results was in the opposite.
water and the pore water of the Baiyangdian Lake, SW stands for surface water, PW stands for pore water Figure 5 presents the distribution of each PFSA congener in the surface water and the pore water. The concentration of PFPeA was much higher in surface water than in pore water, whereas PFDA was more enriched in pore water. Interestingly, although PFBA has a shorter carbon chain and presumably higher solubility in water, it was not enriched in surface water. The distribution of PFHxA, PFHpA, PFOA, PFBS, PFHxS, and PFOS showed a varied preference between surface water and pore water. The 6:2 diPAP and 8:2 diPAP were enriched in pore water in most cases.
Comparison with literature showed that the observed PFASs concentration at the Baiyangdian Lake was much higher than the ΣPFASs of the Songhua River (0.143-1.41 ng/g), the Peal River estuary (nd-2.41 ng/g), the Nansi Lake (0.47-1.81 ng/g), and the Daliao River (0.13-0.49 ng/g) [39][40][41][42] . In whole worldwide, the ΣPFASs in sediment varied greatly. The ΣPFASs level in sediment of target area was similar with those in Lake Superior (nd-10.5 ng/g), Lake Huron (nd-26.0 ng/g) and Lake Michigan (0. Sediment-water diffusion. The partition of organic pollutants between water and sediment affects their environmental behavior and fate, and the partition is governed by their physical and chemical properties as well as sediment characteristics such as organic carbon content, pH, ionic strength, salinity, etc. 45 . In current study, we calculated the distribution coefficient (K d ) of PFOA, PFOS and PFHxS between sediment and surface water as well as between sediment and pore water. We only calculated K d for PFOA, PFOS and PFHxS because their detection rate was > 90%. Significant correlations (p < 0.01) were found between sediment TOC and the sediment-pore water K d of PFHxS, PFOA, and PFOS (Tables 2-4).
There was no significant correlation (p > 0.01) between sediment TOC and the sediment-surface water K d of the tested compounds. Hence, after equilibrium was established for the distribution of the compoundsbetween pore water and sediment, the sediment TOC would become the dominant factor affecting the distributionof the pollutant between pore water and sediment.
The partitioning of PFASs is commonly evaluated by K oc as follows: www.nature.com/scientificreports www.nature.com/scientificreports/ where C s is the PFASs concentration of the entire sediment based on dry weight (ng/g); C w is the PFASs concentration of surface or pore water at equilibrium (ng/mL); and f oc is the organic carbon fraction of the sediment (%  www.nature.com/scientificreports www.nature.com/scientificreports/ Environmental risk of PFASs: an assessment. The RQ of PFASs were calculated to evaluate their environmental risk.
The PNEC water of PFOA, PFOS, PFNA, PFHxA, and PFDA were 100, 25, 100, 97, and 11 µg/L, respectively 47 . The PNEC sediment of PFOA and PFOS were 2060 and 67 µg/kg, respectively. The RQ water of PFOA, PFOS, PFNA, PFHxA, and PFDA were all < 0.01, which indicated very low environmental risk. The RQ sediment of PFOA was <0.01 but the RQ sediment of PFOS ranged in 0.002-0.13. Therefore, the environmental risk level of PFOS varied from very low to intermediate.
Since current toxicity data of PFHxS and PFBS indicate that they are less toxic than PFOS and PFOA, it was expected that they should have higher PNEC than PFOS [48][49][50][51][52] . The measured concentration of PFHxS and PFBuS in water and sediment were even lower than the PNEC of PFOS. Hence, PFHxS and PFBS in the Baiyangdian Lake did not have immediate environmental impact on the aquatic life.
The results showed that the PFASs in the Baiyangdian Lake would not generate immediate environmental impacts on the aquatic life. Nevertheless, uncertainty might arise in assessing the ecological risk of PFASs and their impact on aquatic organisms due to the scarcity of toxicity information and toxicity data of pollutants   Table 2. Correlations of PFOA. **Significant correlation at p < 0.01 (1-tailed).  Table 3. Correlations of PFOS. **Significant correlation at p < 0.01 (1-tailed).  Table 4. Correlations of PFHxS. **Significant correlation at p < 0.01 (1-tailed).