Environ. Sci. Technol. 2009, 43, 3480–3486
Groundwater Pollution by Perfluorinated Surfactants in Tokyo M I C H I O M U R A K A M I , †,‡ KEISUKE KURODA,§ NOBUYUKI SATO,| TETSUO FUKUSHI,§ SATOSHI TAKIZAWA,§ AND H I D E S H I G E T A K A D A * ,† Laboratory of Organic Geochemistry (LOG), Institute of Symbiotic Science and Technology, Tokyo University of Agriculture and Technology, Fuchu, Tokyo 183-8509, Japan, “Wisdom of Water” (Suntory), Corporate Sponsored Research Program, Organization for Interdisciplinary Research Projects, The University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo 113-8656, Japan, Department of Urban Engineering, Graduate School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo, Tokyo, 113-8656, Japan, and IDEA Consultants Inc. 1334-5, Riemon, Ohigawa, Sida, Sizuoka, 421-0212, Japan
Received December 15, 2008. Revised manuscript received March 6, 2009. Accepted April 8, 2009.
Perfluorinated surfactants (PFSs) in groundwater were analyzed to reveal their distribution and sources. Sixteen groundwater and spring samples were collected from the Tokyo metropolitan area, and nine PFSs, including perfluorooctanesulfonate (PFOS) and perfluorooctanoate (PFOA), were analyzed by liquid chromatography-tandem mass spectrometry. A column test using artificial street runoff was also performed to study their behavior. PFSs were detected in all groundwater samples, some at concentrations comparable to those in wastewater and street runoff, suggesting widespread contamination of groundwater by PFSs. In particular, PFOS was more abundant in groundwater than in rivers, wastewater, and street runoff. This was attributed to its production from the degradation of its precursors, as supported by the column test. The occurrence of short-chain perfluorocarboxylates (PFCAs) in groundwater was also consistent with the results of the column test, showing that limited amounts of short-chain PFCAs were removed by soil, as the efficiency of removal increased with the chain length. We evaluated the contributions of PFCAs from wastewater and surface runoff to groundwater by using two indicators, the long/(short + long) ratio and the even/(even + odd) ratio. Both ratios showed good agreement in their calculated contributions in heavily contaminated groundwater where breakthroughs likely occurred. Wastewater and surface runoff contributed to 54-86% and 16-46% of PFCAs, respectively, in groundwater.
Introduction Perfluorinated surfactants (PFSs) such as perfluorooctanesulfonate (PFOS; C8F17SO3-) and perfluorooctanoate (PFOA; * Corresponding author phone: + 81-423-67-5825; fax + 81-42360-8264; e-mail:
[email protected]. † Tokyo University of Agriculture and Technology. ‡ Organization for Interdisciplinary Research Projects, The University of Tokyo. § Department of Urban Engineering, The University of Tokyo. | IDEA Consultants Inc. 3480
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 10, 2009
C7F15COO-) are used in fluoropolymer manufacture and firefighting foams. PFSs and their precursors are also included in carpeting, waxes, and textiles (1-3). PFSs were detected in surface waters (4-9), bays (7, 10, 11), aquatic organisms (10, 12-14), sediments (14, 15), aerosols (16), and street dust (17). More than half of the rivers in Japan recently exceeded the New Jersey (U.S.) preliminary health-based drinking water guidance for PFOA (40 ng/L) (9). PFSs are environmentally persistent and potentially harmful pollutants (18, 19). In addition, PFOS is bioaccumulative, and perfluorocarboxylates (PFCAs) with eight or more fluorinated carbons might be bioaccumulative (20). Because of their features, concern about them is increasing. However, there were limitations in investigations of groundwater contamination by PFSs except beneath specific pollution sites (i.e., fire-training facilities at military bases 21-23). Groundwater is contaminated by sewage through leakage from decrepit sewer pipes, the past practice of sewage infiltration to underground, and infiltration by contaminated river water (24, 25). This contamination often poses a major barrier to the use of groundwater. Fenz et al. used carbamazepine as a marker of sewage and indicated that the average exfiltration rate was 1% in Linz, Austria (24). Nakada et al. also revealed that groundwater in Tokyo was contaminated by pharmaceuticals and personal care products (PPCPs) and suggested that the composition of groundwater is ∼1% sewage on a metropolitan-wide average (25). PFSs are contained in wastewater (26-28) and surface runoff (8, 28). Soil infiltration column tests in our previous study revealed that PFOS and perfluorooctane sulfonamide (FOSA; C8F17SO2NH2) were not removed by soil during infiltration (29) and that PFOS concentrations increased, possibly through its production from the biodegradation of its precursors. Therefore, PFSs possibly reach groundwater and cause contamination. It is now required to investigate the extent of occurrence of PFSs in groundwater. Since information on sources is useful for controlling pollutants, source apportionment is also required. Past studies showed that PFSs were derived from direct discharge in relation to urban activities (7, 9, 11), and that PFSs in surface waters in urban areas were derived from sewage and street runoff (5, 8, 9, 28). However, to our knowledge, no reports of sources of PFSs in groundwater are available. This study had three aims. First, we investigated the extent of occurrence of PFSs in groundwater, comparing the results with observations from rivers, wastewater, and street runoff in Japan. Second, we analyzed the behavior of PFSs during soil infiltration in column tests to understand their transport in groundwater. This is the first report to reveal the behavior of perfluorocarboxylates (PFCAs) during soil infiltration, following our recent report of PFOS, FOSA, organic matter, nutrients, metals, PAHs, and toxicities as determined by bioassays (29). Third, we estimated the contributions from wastewater and surface runoff to PFSs in groundwater by using PFCA compositions, which are useful for distinguishing between wastewater and street runoff (28). In this study, we targeted nine PFS species: PFOS, FOSA, perfluoroheptanoate (PFHpA; C6F13COO-), PFOA, perfluorononanoate (PFNA; C8F17COO-), perfluorodecanoate (PFDA; C9F19COO-), perfluoroundecanoate (PFUA; C10F21COO-), perfluorododecanoate (PFDDA; C11F23COO-), and perfluorotridecanoate (PFTDA; C12F25COO-).
Experimental Section Groundwater Sample Collection. We collected 16 groundwater and spring samples from 0 to 33 m below the ground 10.1021/es803556w CCC: $40.75
2009 American Chemical Society
Published on Web 04/23/2009
in the Tokyo metropolitan area in September to November 2006 (Supporting Information (SI) Figure S1). A parallel study investigated PPCPs (25). SI Table S1 shows the characteristics of the groundwater samples, including type (unconfined, confined or spring), depth, pH, oxidation-reduction potential (ORP), and electrical conductivity (EC), which were measured by electrodes in the field. We preliminarily analyzed pH, ORP and EC in 121 groundwater samples in Tokyo. To confirm how well the 16 new samples represented the water quality of the Tokyo basin, we compared pH, ORP, and EC between them and the 121 previous samples. SI Figure S2a-c shows that the distributions were similar. Therefore, the conventionally measured water quality in the 16 samples could be regarded as being representative. Samples were filtered through prebaked glass fiber filters (GF/F, pore size: 0.7 µm; Whatman) and stored at 5 °C before analyses. Soil Infiltration Column Test. This test was performed with artificial street runoff equivalent to approximately 11-12 years of rainfall; details are described in Murakami et al. (29). Preparation of artificial street runoff allowed the setup of a reproducible, reliable, constraint-free experimental scheme. Briefly, soil samples were collected from a typical loamy layer in a park in the Kanto district, Tokyo, at depths of 90-120 cm from the surface. The Kanto loamy layer is a major soil in Tokyo (30). The samples were air-dried and sieved through 75- and 2000-µm mesh screens to obtain the 75-2000 µm fraction. The soil was then packed in one column to a depth of 20 cm and in two columns to 50 cm. Artificial street runoff was prepared from highway street dust and sediments and groundwater, which was collected from 120 m below the ground at Ohigawa, Shida, Shizuoka. The groundwater and solids were mixed at 25 L/kg and stirred for 6 h at 250 rpm at 20 °C in the dark. The supernatant obtained after more than 15 h of settling was regarded as the artificial street runoff. In a previous study (31), we found that a prewashing of soils is necessary to leach water-soluble pollutants out of the columns. Before the soil column test using the artificial street runoff, the groundwater collected from 120 m below the ground at Ohigawa, Shida, Shizuoka, was fed into the columns at 5 mm/h (2.6 mL/min) for 14-15 days. The artificial street runoff was then fed into the columns at 10 mm/h (1680 mm/ week) under either continuous or intermittent flow. It was fed continuously into the 20-cm column and one of the 50cm soil columns for a total of 79 days. It was also fed intermittently into the other 50-cm column every alternate day for a total of 157 days. Effluent samples from the columns were collected after 2, 16, 30, 44, 58, 72, and 79 days under continuous flow and 3, 31, 59, 87, 115, 143, and 157 days under intermittent flow. The effluent was analyzed 3 times (at 2, 72, and 143 days). The test was conducted at a constant 20 °C in the dark. Column influent and effluent samples were filtered through prebaked glass fiber filters as described above and stored at 5 °C before analyses. Analyses. Details of chemicals and analyses are given in Murakami et al. (9). Briefly, PFSs in the filtrate (500-1000 mL) were concentrated by solid-phase extraction. 13C-labeled PFOA was spiked into aliquots of samples, and the samples were passed through Sep-Pak Plus tC18 cartridges (Waters) preconditioned with 20 mL methanol and then 10 mL distilled water. A flow rate of less than 10 mL/min was maintained. The cartridges were washed with 7 mL 30% (v/v) methanol in distilled water, and then with 7 mL 55% (v/v) methanol in distilled water that was acidified with 4 M HCl to pH 2.0-2.5. The target compounds were eluted with 20 mL methanol. The eluate was concentrated to 0.5-1 mL, and PFSs were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS; Agilent 1100 and TSQ Quantum) with a Zorbax Rx-C8 column (4.6 × 150 mm, 5 µm, Agilent) in electrospray negative ionization mode. A gradient mobile
FIGURE 1. PFS concentrations in groundwater in Tokyo. phase of 0.5 mM ammonium acetate in distilled water and 0.5 mM ammonium acetate in acetonitrile was used. At a flow rate of 0.3 mL/min, the mobile phase gradient was ramped from 20 to 100% 0.5 mM ammonium acetate in acetonitrile in 5 min, maintained at 100% for 11 min, and then ramped down again to 20%. The capillary temperature was maintained at 350 °C. Ions were monitored in selected reaction monitoring mode. Details of parent and product ions are listed in SI Table S2. The reproducibility and recovery rates were confirmed by using secondary effluent from a wastewater treatment plant. Relative standard deviation (n ) 4) was less than 15% for all PFSs. A 150-ng aliquot of each PFS standard was spiked into 1 L of each wastewater sample (n ) 4) to confirm the recovery rates. The recovery rates ranged from 76 to 109%. An operational blank was run with every batch using HPLCgrade distilled water. The detection limit was set at 3 times the detected amount in the operational blank, and ranged from 0.1 to from 0.6 ng/L for all PFSs. The recovery rates of the label standard in environmental samples were high enough, and there were no large differences between them: 67 ( 35% (arithmetic mean ( standard deviation) in groundwater, 97 ( 23% in river water (9), 88 ( 28% in wastewater influent (28), 71 ( 27% in wastewater effluent (9, 28), and 65 ( 35% in street runoff (28). Thus, matrix effects on PFSs were not a major concern. The recovery rates of the label standard in column influent and effluent were 26 ( 11% and 35 ( 11%, respectively, suggesting ionization suppression of PFSs. However, the difference was not significant (t-test; P > 0.15), indicating that the matrix effects on PFSs were similar. The matrix effects did not significantly affect the efficiencies of removal of PFSs by soils, which were estimated from the concentrations in the column influent and effluent. Also, the behavior and removal of PFSs in the column test with and without label-recovery correction were similar, supporting the reliability of the findings (SI).
Results and Discussion Occurrence of PFSs in Groundwater. Figure 1 and SI Table S3 show PFS concentrations in groundwater in Tokyo. PFSs were detected in all groundwater and spring samples, suggesting diffuse contamination. PFOS, PFHpA, PFOA, and PFNA were major species in groundwater, consistent with the results in rivers in Japan (9). The respective concentrations in the groundwater samples were 0.28-133, 50% difference in the concentrations in the column influent between quantifications based on the two monitoring ions. PFDDA and PFTDA were also omitted owing to low concentrations ( 0.30) in indicators between wastewater influents and effluents (i.e., long/(long + short): influents 0.03 ( 0.02 (arithmetic mean ( SE), effluents 0.06 ( 0.02; even/(even + odd): influents 0.45 ( 0.05, effluents 0.48 ( 0.04), we used indicators of wastewater influents to estimate the contributions from wastewater and street runoff to PFSs in groundwater. We considered street runoff to be representative of surface runoff. This decision was supported by the finding that PFS concentrations in street runoff were more than 1 order of magnitude higher than those in rainfall (28). The contributions from wastewater and street runoff were estimated as follows: RGW ) a × RWW + b × RSR where RGW is the long/(short + long) ratio or the even/(even + odd) ratio in the groundwater samples, RWW is the ratio in the wastewater influent samples, RSR is the ratio in the street runoff samples, a is the contribution from wastewater influent (%), and b is the contribution from street runoff (%) (a + b ) 100%). Figure 6 compares the two indicators among groundwater, wastewater influents, and street runoff (the latter 2 from Murakami et al. (28)). Both ratios were clearly higher (t-test; P < 0.005) in street runoff (long/(short + long), 0.26 ( 0.03; even/(even + odd), 0.72 ( 0.04) than in wastewater influents (0.03 ( 0.02; 0.45 ( 0.05). The values in groundwater ranged from 0 to 0.28 (long/(short + long)) and from 0.31 to 0.97 (even/(even + odd)), and the even/(even + odd) ratios in some groundwater samples exceeded those in street runoff. This excess could be explained by differences in behavior among PFCAs. Figure 4f, g and SI Figure S3e, f show the variations in both ratios during soil infiltration. Long/(short 3484
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 10, 2009
FIGURE 7. Contributions from wastewater and street runoff to four PFCAs in groundwater. (a) Estimation from long/(short + long) ratios; (b) estimation from even/(even + odd) ratios. + long) ratios in effluent were lower than those in influent because long-chain PFCAs were preferentially partitioned to the soil. These variations indicate that the contribution from street runoff to PFCAs in groundwater would be underestimated if long/(short + long) ratios were used as indicators. Conversely, even/(even + odd) ratios in effluent were higher than those in influent, although the differences were smaller in the late stage owing to smaller differences in hydrophobicity between even- and odd-chain PFCAs. Therefore, the use of even/(even + odd) ratios would overestimate the contribution from street runoff to PFCAs in groundwater. Accordingly, when PFCAs were not partitioned to the soils, the estimated contribution from wastewater and street runoff should be same between the uses of long/(short + long) and even/(even + odd); however, partitioning underestimated the contribution of street runoff in the use of long/(short + long) and overestimated it in the use of even/(even + odd). Figure 7 shows contributions from wastewater and street runoff to the 4 PFCAs in groundwater estimated by the two indicators. Although there were differences in estimated contributions between the two ratios in less contaminated groundwater samples (i.e., GW-1304 ∼ 301), consistent agreement was observed in heavily contaminated samples (i.e., GW-1601, 2401, 1101, and 302). This is probably because breakthroughs occurred in the heavily contaminated samples, whereas the partitioning was the governing factor in less contaminated samples. GW-1601 and GW-302, in which both indicators suggested large contributions from wastewater, contained 7-12 ng/L of crotamiton, a persistent marker of sewage, whereas GW-1101, in which they suggested large contributions from street runoff, contained less than the average limit of quantification (
