Multiple pollutants in groundwater near an abandoned Chinese fluorine chemical park: concentrations, correlations and health risk assessments

Contamination and adverse effects from various pollutants often appear in abandoned industrial regions. Thus, nine groundwater samples were collected from the vicinity of the fluorochemical industry in Fuxin City, Liaoning Province, to determine concentrations of the ten heavy metals arsenic (As), chromium (Cr), cadmium (Cd), lead (Pb), nickel (Ni), copper (Cu), manganese (Mn), zinc (Zn), iron (Fe) and mercury(Hg), as well as those of fluorine (F−) and eighteen poly- and perfluorinated substances (PFASs), analyse correlation relationships, and assess the health risks for different age groups. The results showed that the levels of fluorine (F−) (0.92–4.42 mg·L−1), Mn (0.0005–4.91 mg·L−1) and Fe (1.45–5.61 mg·L−1) exceeded the standard limits for drinking water. Short chain perfluorobutanoic acid (PFBA) (4.14–2501.42 ng·L−1), perfluorobutane sulfonate (PFBS) (17.07–51,818.61 ng·L−1) and perfluorohexanoic acid (PFHxA) (0.47–936.32 ng·L−1) were the predominant substances from the PFASs group. No individual PFASs levels had significant relationships with F− or heavy metal contents. There was a positive relationship between short chain PFASs concentrations and water depth and a negative relationship between long chain PFASs concentration and water depth. The hazard quotient (HQ) decreased in the order F− > heavy metals > PFASs and also decreased for older age groups. In addition, As, Fe, Mn and perfluorooctanoic acid (PFOA) were the main sources of risk from the heavy metal and PFASs groups, respectively.

Sample collection and pretreatment. The study area and sampling points are listed in Fig. 1, and Google Earth Pro (V7.3.3.7721) software was applied to establish the latitudes and longitudes of each sampling point. Groundwater samples (sites G1-G9, n = 9) were collected from civic wells during August and September 2019. In the sampling process, each polyethylene bottle was rinsed 1-2 times with pure water and 2-3 times with the groundwater sample before sampling. Groundwater samples were collected at 3-5 m below the ground for storage in two cleaned 1 L PP bottles. One was used for heavy metal determination, and one was used for determining physicochemical properties and F − and PFASs levels. Then, 2 mL of 1% HNO 3 was added to each water sample for heavy metal determination to stabilize the forms of heavy metals. A pH < 2 was established with HNO 3 before storage at 4 °C.
The physicochemical properties of the water samples were determined in the laboratory. Ten heavy metals, As, Cr, Cd, Pb, Ni, Cu, Mn, Zn, Fe, Hg and F − were analysed as target pollutants in groundwater, and all experimental methods for determining physicochemical properties, heavy metals and F − from precise standards are given in Supplementary Table S1 online. All standard values for heavy metals and F − in drinking water are from GB/ T14848-2017 (Class III) 20 and were compared with the pollutant concentrations determined in the present study.
Next, PFASs (abbreviation for Supplementary Table S2 online) pretreatment methods were developed by Chen et al. 18 and Bao et al. 21 . Oasis WAX cartridges (500 mg, 6 mL, Waters, USA) were used to extract PFASs from the groundwater samples. The WAX SPE cartridges were conditioned with 5 mL of methanol containing 0.1% ammonium hydroxide, 5 mL of methanol, and 5 mL of Milli-Q water. The water samples were then loaded onto the cartridges at a rate of 1 drop/s. The cartridges were washed with 5 mL of 2.5 mM acetic acid/ammonium acetate (pH = 4) and then centrifuged for 10 min at 3000 rpm to remove residual water. Ionizable compounds were eluted with 5 mL of methanol and 5 mL of methanol containing 0.1% ammonium hydroxide. The eluent was concentrated under nitrogen and dissolved in 1 mL of acetonitrile. Then, the liquid was filtered by 0.45 μm membrane and stored in 1.5L vials for machine detection and analysis. All PFASs concentrations were analysed via high-performance liquid chromatography triple quadrupole mass spectrometry (HPLC-MS/MS), and measurement parameters are listed in Supplementary Tables S3 and S4 online. Health risk assessment. Noncarcinogenic methods were applied to assess potential health risks 22 .
The health risks of F − , heavy metals, and PFASs in groundwater were assessed for the different age groups 6-12 months, 6-11 years, 11-16 years, 16-18 years, 18-21 years, 21-65 years and > 65 years. Although local residents have various ways of contacting these pollutants, direct ingestion of a drinking water source is always thought to be the most harmful exposure method. The hazard quotient (HQ) is typically used in noncarcinogenic assessments, and the mixed hazard quotient (HQ mix ) is based on the sum of HQ values for single pollutants. If the HQ/HQ mix for an assessed pollutant exceeds 1, risks to human health are indicated. If the HQ/HQ mix for the assessed pollutant is below 1, the risk is within the acceptable range. The formulas for HQ/HQ mix calculation are given in Eqs. (1) to (3)

QA/QC.
To determine the accuracy of all pollutant analyses, all pretreatment and detection methods for heavy metals, F − and PFASs were taken from standard reference materials or prior studies [23][24][25] . Three analytical duplicates were included in the analysis of each water sample. A field blank was used to eliminate the influence from the sampling bottle on concentration, and a series of laboratory Milli-Q water samples was added into the sampling bottle to determine this free blank. Procedural blanks were identified for every batch of samples, and they were either not detected (nd) or lower than the limits of quantification (LOQs) to ensure data accuracy and method availability. LOQs were defined as the minimum injection volume needed for reproducible measurements of peak area within ± 20% for repeated injections. All water samples were membrane treated to eliminate impurities. The limits of detection (LODs) and LOQs for F − and heavy metals are shown in Supplementary  Table S1 online. The recovery rates for heavy metals ranged from 88.42-97.49%, and the relative standard deviations (RSD) ranged from 3.55 to 6.69%. The PFASs concentrations were identified with a concentration series containing 0, 0.01, 0.1, 1, 5, 10, 50, 100, and 500 μg·L −1 used to draw the standard curve (R 2 > 0.99). An internal solution (IS, see Supplementary information) was added to each sample to assess the accuracy for recovery of

Results and discussion
Characteristics of various pollutants in water. NO −

-N, NO − 3 -N, NH + 4 -N and F -.
Pollutant concentrations in groundwater from the FCP of Fuxin City are shown in Table 1. The NO − 2 -N and NO − 3 -N concentrations ranged from no detection (nd) to 0.063 mg·L −1 and nd to 10.77 mg·L −1 , which were below the standard values of 1.00 mg·L −1 and 20.00 mg·L −1 , respectively ( Table 1). The mean NH + 4 -N concentration (1.02 mg·L −1 ) was two times that of the standard value (0.5 mg·L −1 ) in GB/T14848-2017 20 . Three major sources of NH + 4 -N were identified: leather industry production, introduction of nitrogen chemicals, and human activities 26 .
A high F − background always exists in the local groundwater. Abundant fluorite (CaF 2 ) resources can dissolve and disperse along the water flow direction, which was identified as the main source of F − in groundwater. The highest F concentration appeared in G3 (4.42 mg·L −1 ), followed by G8 (3.43 mg·L −1 ) and G6 (1.50 mg·L −1 ) ( Table 1 and Fig. 1). These concentrations far exceeded the limit (1.0 mg·L −1 ) for Class III Chinese drinking water, although F concentration standards from different organizations and institutes are always controversial 20 . The U.S. Environmental Protection Agency (EPA) recommends fluoridated community water systems adjust fluoride to approximately 0.7 mg·L −127 . Additionally, other studies have indicated that only F − concentrations below 2 mg·L −1 preclude adverse effects on the human body from drinking water 28 . Dissanayake et al. suggested that if the F concentration in drinking water exceeds 1.5 mg·L −1 , the probability of dental fluorosis increases for children 29 . Regardless of which standard values are used, all F values in our study exceeded the screening or risk values, which indicated that F contamination in groundwater may pose a threat to the human body.
Heavy metals. As, Cr, Cu, Fe, Mn, Ni, Pb, and Zn were all detected, whereas Cd and Hg were assigned nd status. The predominant heavy metals were Mn (0.0005-4.91 mg·L −1 ), Fe (1.45-5.61 mg·L −1 ) and As (0.0025-0.01 mg·L −1 ) ( Table 1 and Fig. 1). In our previous studies, the levels of Ni (63.01 mg·kg −1 ), Cr (48.26 mg·kg −1 ), Cu (23.29 mg·kg −1 ), Pb (12.54 mg·kg −1 ) and As (9.80 mg·kg −1 ) in soil were higher than those in groundwater 19 . Surprisingly, Cd and Hg were detected in soil but not in groundwater. Heavy metals entering the soil have weak transport capacity and are easily adsorbed by soil colloids; moreover, the physical and chemical properties of soil affect the form adopted, especially for some metals that have difficulty migrating from soil to water 2,12 . The concentrations of the two heavy metals Fe and Mn were much higher than 0.3 mg·L −1 and 0.1 mg·L −1 , especially for Fe, which exceeded the standard value at all sampling points. Rich ore resources are the main sources of excess Fe and Mn in groundwater, and metal powder can enter groundwater during the extraction process.
PFASs. Eighteen PFASs groundwater concentrations from the FCP of Fuxin City are shown in Table 2. The ∑ 18 PFAS concentrations ranged from 72.49 ng·L −1 to 68,142.14 ng·L −1 , and the mean value was 11,108.42 ng·L −1 . Short chain PFBA (4.14-2501.42 ng·L −1 ), PFBS (17.07-51,818.61 ng·L −1 ) and PFHxA (0.47-936.32 ng·L −1 ) were the predominant substances (Table 2 and Fig. 1). The highest ∑ 18 PFAS concentrations were found at G5 (68,142.16 ng·L −1 ), G4 (26,464.56 ng·L −1 ) and G6 (3334.00 ng·L −1 ), which were very near the center of the FCP www.nature.com/scientificreports/ ( Fig. 1 and Table 2). Short chain PFASs, as alternatives to long chain substrates, have been widely used in the FCP of Fuxin City or in other countries 21 . The PFASs contamination in our study area was much higher than those in other regions, including Taiwan 30 , the Maozhou River basin 28 , a Tianjin suburb 31 and Hubei Province 32 . Short chain PFASs (PFBA, PFBS, PFHpA, PFHxA and PFHxS) and PFOA are commonly detected currently due to the heavy use of long chain substitutes and different industry types 17,32 . There is no current environmental standard for PFASs in China, whereas newly emerging pollutants have been of concern in Canada 33 , Swedish 13 , American 34 and so on. The U.S. Environmental Protection Agency (USEPA) limit for PFOA in drinking water is 400 ng·L −1 , and the PFOA concentration (35.34-11,305.65 ng·L −1 ) exceeds the standard value by several orders of magnitude 35 . The low concentration detected for long-chain PFASs (C ≥ 8) may result mainly from historical industry production and emissions. Short chain PFASs applied to improve hydrophilicity and fluidity tend to migrate easily in water bodies, whereas long chain PFASs are readily adsorbed by large particles or sediments 36 . The Xi River is the main river flowing through the FCP and receives many industrial pollutants throughout the year. Surface water recharge and infiltration moves PFASs into the groundwater system. Based on contamination in the Maozhou River basin, a potential link between the surface and groundwater was identified, and the PFASs concentration in surface water had a highly positive relationship with that in groundwater 37 . A PFOS substitute of 6:2 Cl-PFSEA was detected at low concentrations (0.01-0.17 ng·L −1 ), which indicates that the emerging PFASs substitute has been used in local industrial production 38,39 (Table 2). In addition, PFSAs concentrations were much higher than PFCAs concentrations. Different functional groups cause PFASs to exhibit varied environmental behaviour, especially short chain PFASs, which migrate easily in water bodies 40 .

Correlation relationship analysis. Correlations among various pollutants and physicochemical proper-
ties (see Supplementary Table S6 online) are shown in Fig. 2. The relationship between F − concentration and physicochemical properties was weak in groundwater except for DO, and the result was the same as that of Loganathan et al. for soil 41 . The F − concentration had a strongly negative relationship (R 2 = − 0.717, p = 0.02) with DO, which indicates that DO content may influence F − migration or behaviour, but this also needs to be studied further. In the present study, the weak correlation between F − and heavy metals was due to frequent rain, which dilutes the pollutants in water. The F − in groundwater has many sources, including atmospheric deposition, agricultural activities, geographical factors, fertilizers and pesticides 42 . Although the F − concentration had no significant relationship with TH (R 2 = − 0.337, p = 0.3) in groundwater, F − concentrations are continuously enriched, even after the groundwater reaches equilibrium with respect to CaF 2 levels, due to removal of Ca by precipitation of calcite (CaCO 3 ) 42,43 . In addition, plant cover type and amount, rain and soil properties also affect the migration of F − from the surface to groundwater 40 . The F − concentration showed the only significant relationship with heavy metal levels (p < 0.1) in soils from Shifang County of Sichuan Province; similarly, this relationship and the solubility of F − in groundwater needs to be further studied 44   www.nature.com/scientificreports/ and penetration of water. Short-chain PFASs were more inclined to migrate in water environments than longchain PFASs because of their hydrophilicities. In addition, frequent rainfall may also accelerate flow migration and affect the PFAS concentration distribution in the groundwater system. Nevertheless, long chain PFDoA (R 2 = − 0.607, p = 0.07), PFTrDA (R 2 = 0.786, p = 0.01), PFTeDA(R 2 = − 0.807, p = 0.002) and PFHxDA(R 2 = − 0.670, p = 0.04) showed negative relationships with the water level. The strong hydrophobicities of long chain PFAS facilitate adsorption by particles in water and limit migration and mobility capabilities. The correlations indicate the fates of short-and long-chain PFASs in groundwater. The positive relationships between short chain PFASs or long chain PFASs indicates that carbon chain length plays a key role in migration. The solubility in water in the water body is inversely proportional to the carbon chain length 32 . A previous study from New Jersey, USA, also showed that long chain PFDA, PFUdA and PFDoA have higher affinities for organic material and preferentially partition into sediments 45 . Importantly, there were no relationships between PFASs and F − or heavy metal concentrations in our study. The reason may be that PFASs and heavy metals come from different sources or their environmental concentrations are low, but this requires more study.
Health risk assessments. Health risk assessments are widely used to evaluate harmful effects on the human body. The assessment method with the quotient is a simple model that effectively calculates the risk values for different pollutants in the groundwater system. However, the model only considers the risk of a single pollutant and does not apply complex mechanisms for different pollutants. The main detected pollutants, including heavy metals, individual PFASs, and F − , were assessed, and the results are shown in Fig. 3. Figure 3a shows that children aged 6-12 months were more vulnerable to toxic pollutants than residents in the other age groups. The mean HQ values for heavy metals and F − reached 4.82 and 5.13, respectively, which poses potential risk for the 6-12 month age group. The HQ value reflected a decreasing trend with increasing age. The mean HQ or HQ mix value for F or heavy metals exceeded 1 for all age groups, which indicates that residents may suffer the negative effects of drinking contaminated groundwater. In Fig. 3b, the potential sources of risk for adults aged 21-65 years came from the heavy metals As (45.0%), Mn (27.1%), and Fe (24.9%). Although the quotients for As, Mn, and Fe were not exceeded 1 for the 21-65 year group, the HQ values were 2.34, 1.41, and 1.29 in the 6-12 month group (Fig. 3c). In studies of the risk due to heavy metals, groups of infants were exposed to toxic pollutants [46][47][48][49] . Abundant CaF 2 resources were the main reason for the high F − concentrations in groundwater, and earlier industrial production was also identified as another source. The HQ value for F − exceeded 1 for all age groups (4.82 for 6-12 months, 1.75 for 6-11 years, 1.32 for 11-16 years, 1.38 for 16-18 years, 1.53 for 18-21 years, 1.47 for 21-65 years and 1.45 for > 65 years) (Fig. 3c). A previous study by Ozsvath et al. indicated that if a child ingests excess F − , a variety of adverse effects can occur in a healthy body, including dental fluorosis, www.nature.com/scientificreports/ skeletal fluorosis, increased rates of bone fractures and decreased birth rates 28 . Importantly, high F − levels in groundwater are also introduced to farmland or residential areas by irrigation, although the risk from soil is lower than that from groundwater 19 . A study of a phosphate industrial area from Sichuan Province found that the main exposure pathway for pollutants was ingestion or particulate inhalation by groups of children and adults, respectively, while the highest risk value was less than 3-4 times that in our study 22 . F − and PFASs were the predominant substances in the FCP, especially PFOA and PFOS, which were the common PFASs in the groundwater system. The toxicities of long-chain PFASs are much higher than those of short-chain PFASs, and they accumulate more easily in the human body 50,51 . In Fig. 3b, PFOA accounted for the largest proportion of risk among PFASs (89.4%), followed by PFOS (7.0%) and PFHxS (1.80%). PFASs in teenagers may influence the reproductive system and character development of residents living near a Chinese fluorine chemical plant, as shown previously 52 . Drinking water provides the major ingestion pathway for PFASs, and they have a long half-life in the human body.

Conclusion
Contaminant concentrations and correlations between F − , heavy metal, and PFAS levels were studied, and health risk assessments for different age groups living around an abandoned FCP were performed. F − , Fe and Mn concentrations exceeded the recommended limits of the drinking water standard, and the Cd, Cr, Cu, Ni, Pb and Zn concentrations were considerable. PFBA, PFBS and PFHxA were the predominant substances, whereas the PFOA concentration was much higher than the risk control value from the USEPA. The F − concentration showed weak correlations with various physicochemical properties, except for DO, and heavy metal concentrations. A strongly positive relationship existed between short chain PFASs and water depth, while long chain PFASs had a negative relationship with water depth. Carbon length plays an important role in the migration for PFASs. Health risk assessments for all age groups demonstrated that the risk according to source decreased in the order F − > heavy metals > PFASs, and HQ values decreased for older age groups. As, Fe and Mn in the heavy metal group and PFOA in the PFASs groups were identified as the main sources of risk, so these substances should be listed as priority pollutants in the local region. Different pollutants in groundwater originate from various sources, so measures should be adopted to control pollutants at different levels. Residents should also reduce the frequency with which they drink groundwater by using external water to decrease health risks. Our study clarified the characteristics of various pollutants in groundwater and analysed potential correlations to provide a reference for local pollution prevention and control.