Environ. Sci. Technol. 2005, 39, 953-962
Occurrence of Some Organic UV Filters in Wastewater, in Surface Waters, and in Fish from Swiss Lakes
Downloaded via EASTERN KENTUCKY UNIV on January 24, 2019 at 08:31:57 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
MARIANNE E. BALMER,* HANS-RUDOLF BUSER, MARKUS D. MU ¨ LLER, AND THOMAS POIGER Agroscope FAW, Swiss Federal Research Station, CH-8820 Wa¨denswil, Switzerland
Organic UV filters are used in personal care products such as sunscreen products, and in cosmetics, beauty creams, skin lotions, lipsticks, hair sprays, hair dyes, shampoos, and so forth. The compounds enter the aquatic environment from showering, wash-off, washing (laundering), and so forth via wastewater treatment plants (WWTPs) (“indirect inputs”) and from recreational activities such as swimming and bathing in lakes and rivers (“direct inputs”). In this study, we investigated the occurrence of four important organic UV filter compounds (benzophenone-3, BP-3; 4-methylbenzylidene camphor, 4-MBC; ethylhexyl methoxy cinnamate, EHMC; octocrylene, OC) in wastewater, and in water and fish from various Swiss lakes, using gas chromatographic/mass spectrometric analyses. All four UV filters were present in untreated wastewater (WWTP influent) with a maximum concentration of 19 µg L-1 for EHMC. The data indicate a seasonal variation with influent loads higher in the warmer season (June 2002) than in the colder one (April 2002). The influent loads were in the order EHMC > 4-MBC ∼ BP-3 > OC. The concentrations in treated wastewater (WWTP effluent) were considerably lower, indicating substantial elimination in the plants. 4-MBC was usually the most prevalent compound (maximum concentration, 2.7 µg L-1), followed by BP-3, EHMC, and OC. UV filters were also detected in Swiss midland lakes and a river (Limmat) receiving inputs from WWTPs and recreational activities. However, all concentrations were low (99%; EHMC, >99% (E)isomer; OC, g98%; all compounds courtesy of Merck, Darmstadt, Germany. Different stable isotope labeled compounds were used as internal standards: 13C12-3,3′,4,4′tetrachlorobiphenyl (PCB#77), 13C3-caffeine, and 13C6metolachlor, all from Cambridge Isotope Laboratories, Cambridge, MA. The chemicals, reagents, and solvents were of the highest quality available (residue analysis grade, distilled in-glass), and precautions were taken to prevent contamination from personnel, equipment, and glassware. Prior to use, all glassware was cleaned twice with the solvent used in the different procedures, and polyethylene gloves were worn throughout all steps of sample processing. Description of Lakes and Rivers Investigated. The lakes selected for the study were Zu¨richsee, Greifensee, Pfa¨ffikersee, Hu ¨ ttnersee, Thunersee, and for background measurements, a remote mountain lake, Jo¨risee (Figure 2). The lakes were selected to cover inputs from WWTPs and recreational activities (swimming/bathing). In particular, the lakes differed with respect to their P/Q ratios which are considered as a measure for the anthropogenic burden from wastewater, increasing with the population in the watershed (number of persons, P) and decreasing with the throughflow of water (Q, m3 d-1). All lakes, except the mountain lake, are stratified during the warmer season (April-December) with development of an epilimnion and a hypolimnion. In winter (January-March), the lakes are mixed to considerable depth. Hydraulic data and further details on these lakes are given in refs 10 and 11. Zu ¨richsee, situated 406 m above sea level (asl), was chosen as the main study site because significant direct and indirect inputs of UV filters are expected. Approximately 330 000 people live in its watershed and the lake thus receives considerable inputs of anthropogenic compounds. Zu ¨ richsee is an important recreational area with, for example, more than 70 public swimming areas. The lake represents a typical situation with respect to WWTP-derived contamination from the resident population in its watershed (P/Q ) 0.043 persons m-3 d). 954
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 4, 2005
FIGURE 2. Lakes and rivers sampled and type of samples analyzed from the respective sites (f: fish; sw: surface water; spmd: semipermeable membrane devices; wwtp: wastewater treatment plant). Inset: SPMD sampling of Zu1 richsee (outflow), effluent of WWTP Werdho1 lzli, and river Limmat (at Dietikon) with sampling sites indicated by triangles.
Greifensee (435 m asl) is situated 10 km east of Zu ¨ rich. The sizable population (about 100 000 persons live in its watershed) and the relatively low water throughflow lead to one of the highest P/Q ratios of all sizable lakes in Switzerland (P/Q ) 0.303 persons m-3 d). Greifensee also has recreational activities with several swimming sites and therefore some direct inputs of UV filters are expected. Pfa¨ffikersee (537 m asl) is a small lake located 20 km northeast of Zu ¨ richsee, close to Greifensee with some recreational activities. Direct and indirect inputs of UV filters are expected (P/Q ) 0.173 persons m-3 d). Hu ¨ttnersee, situated at 658 m asl near Zu ¨ richsee, is a small lake with a short water residence time of ∼120 d. There is no WWTP located in its watershed and therefore no indirect inputs of UV filters are expected (P/Q ) 0 persons m-3 d). However, there is public access to the lake and during summer there are recreational activities at the lake. The number of bathers relative to the lake volume and water throughflow is high, so that this lake may represent a worst-case situation with respect to possible contamination with UV filters from direct inputs (6). Thunersee is a pre-alpine lake, situated at 558 m asl. Although there is a sizable population living in its watershed (94 300), the rapid flushing with water from alpine areas leads to generally good water quality (P/Q ratio, 0.010 persons m-3 d). The relatively high mean altitude of its watershed (1790 m above seal level) accounts for generally low water temperatures, and hence expectedly, lower direct inputs of UV filters from recreational activities. Jo¨risee is a small mountain lake located at an altitude of 2450 m, remote from human activities, receiving inputs only from rain, snow, ice, and dry deposition. There are no recreational activities at the lake and no inputs of UV filters
are expected. Therefore, this lake was selected for background measurements. The river Limmat starts at Zu ¨richsee (outflow) and consists initially of the water of that lake. The Limmat passes the city of Zu ¨ rich and it was sampled at Dietikon, a location approximately 15 km downstream of the lake, 6.5 km downstream of the discharge of WWTP Zu ¨ rich (370 000 persons serviced). At that site, there is some additional contribution from the joining river Sihl (∼30 000 persons in its watershed, inset Figure 2). Surface Water Sampling. Surface water samples were taken between July and September 2002 from Greifensee, Zu ¨ richsee, Hu ¨ ttnersee, and Jo¨risee. Samples were generally taken at 1-2.5 m depth. From Zu ¨ richsee, additional samples from the hypolimnion (20 m depth) were analyzed. Samples from Zu ¨ richsee and Greifensee were taken above the deepest point using a 10-L Niskin bottle; samples from Hu ¨ ttnersee and Jo¨risee were grab samples taken by boat in the middle of the lake and from the shore, respectively. All water samples were filled on-site into methanol-rinsed, 1-L glass bottles, protected from light and refrigerated at 4 °C upon arrival at the laboratory. Selection and Sampling of WWTPs. Wastewater samples were obtained from eight municipal WWTPs located in the region of Zu ¨ rich (Table 1), six of them discharging into Zu ¨ richsee (not discharging into Zu ¨ richsee are Wetzikon and Kloten-Opfikon). The installations serve populations of about 10 000-30 000 persons and are modern, three- or four-stage installations that include mechanical, biological, and chemical treatment, and in most cases subsequent sand filtration. Untreated wastewater (influent) was taken after the primary sedimentation basins, and treated wastewater (effluent) was taken after sand filtration. With few exceptions, samples were flow-proportionally collected during 24 h. In addition, the effluent of a further, larger four-stage WWTP (Werdho¨lzli, Zu ¨ rich, 370 000 persons) was sampled using SPMDs in April/ May and in August/September 2002, respectively. The wastewater samples were from three sampling campaigns carried out in April 2002, June 2002, and August/ September 2003, respectively, whereby some of the samples were collected initially for another study on caffeine in wastewater (12). In April 2002, samples were taken during a longer period of cool weather with maximum air temperatures of 17 °C and expectedly little outdoor activities and less use of sunscreen products. In contrast, the samples in June 2002 were taken during a very warm and sunny period with afternoon temperatures up to 31 °C and expectedly more outdoor activities and use of sunscreen products. The last sampling campaign (August/September 2003) represented a typical late summer/early fall situation. The first and the last sampling campaigns thus reflect more usual situations, and the June 2002 campaign likely a “worstcase” situation with respect to UV filters from the use of sunscreens. Solid-Phase Extraction (SPE) and Cleanup of WWTP and Surface Water Samples. Extraction of UV filters from wastewater and surface water samples was done by solidphase extraction (SPE) using reusable glass columns containing approximately 10 mL of a macroporous polystyrenedivinylbenzene polymer absorbent (Bio-Beads SM-2, 20-50 mesh, Bio-Rad Laboratories, Hercules, CA). Separate columns were used for WWTP samples and surface water. Most WWTP influent samples (∼300 mL) were centrifuged prior to extraction using a Sorvall ultracentrifuge (samples from the third campaign were not centrifuged); WWTP effluent and surface water samples were extracted directly. For analysis, WWTP samples (100-200 mL) were fortified with 13C3-caffeine as internal standard (influent, 1300-2600 ng; effluent, 130260 ng). Surface water samples (1 L) were fortified with 13C6metolachlor (50 ng), as routinely done for the analysis of
neutral pesticides (13). All samples were passed through the SPE columns at about 10 mL min-1. The analytes were recovered from the SPE columns using methanol/dichloromethane, and the extracts were cleaned up on silica minicolumns using ethyl acetate/methanol 95:5 (WWTP samples) or ethyl acetate (surface waters) as previously described (12). The purified extracts were then carefully concentrated to 100-1500 µL with a gentle nitrogen stream. Fossil groundwater was processed as procedural blank and fortified at concentrations of 0.02, 0.05, 0.1, and 2 µg L-1 of each UV filter for the determination of recoveries. Recoveries were also determined in effluent water from WWTP Wetzikon at a concentration of 20 µg L-1 of each UV filter. All four UV filters were recovered acceptably (78-129%). The limits of detection (LODs) were 2 (surface water) and 10 ng L-1 (wastewater). Procedural blanks showed the presence of small amounts of BP-3 and EHMC (2 and 3 ng L-1), and the data were corrected accordingly. The data reported, however, are not corrected for recoveries. Semipermeable Membrane Devices (SPMDs). SPMDs serve for integrative in situ concentration of lipophilic contaminants and measure time-weighted average concentrations of the dissolved (bioavailable) compounds. Therefore, SPMD sampling is an appropriate sampling method for compounds with varying concentrations in low concentration ranges, such as UV filters (14, 15). Standard SPMDs (Exposmeter AB, Tavelsjo¨, Sweden) were mounted in perforated stainless steel containers and exposed for ∼3 weeks at a depth of 1-2 m in Zu ¨ richsee (outflow), Greifensee (outflow), Hu ¨ ttnersee (midlake), the river Limmat, and for background measurements, in the remote mountain lake, Jo¨risee (∼6week exposure, near shore). In addition, SPMDs were exposed directly in the effluent of the WWTP Zu ¨ rich at the same time as in Zu ¨ richsee and in the river Limmat. All SPMDs were dialyzed with cyclopentane/dichloromethane 95:5 as reported in refs 16 and 17. GPC cleanup of dialyzates was done in the same way as described for fish extracts (see below). For control purposes, nonexposed SPMDs were processed as procedural blanks, and dialyzates were fortified with 50, 100, and 200 ng of each UV-filter for recovery determinations. Recoveries were 51% for BP-3, 72% for 4-MBC, 110% for EHMC, and 73% for OC. Data from duplicate exposures showed reasonable agreement. The limits of detection (LODs) were 1 ng SPMD-1 for methyl triclosan and 5 ng SPMD-1 for 4-MBC and OC. The LODs for BP-3 and EHMC were higher (25 and 100 ng SPMD-1, respectively) because of the presence of these compounds in storage blanks. The data were corrected for these blank contributions, but not for recoveries, and concentrations were not quantified when
