Environ. Sci. Technol. 2003, 37, 4442-4448
Occurrence and Removal of Estrogenic Short-Chain Ethoxy Nonylphenolic Compounds and Their Halogenated Derivatives during Drinking Water Production M I R A P E T R O V I C , * ,† A L F R E D O D I A Z , ‡ FRANCESC VENTURA,‡ AND D A M I AÄ B A R C E L O Ä † Department of Environmental Chemistry, IIQAB-CSIC, c/Jordi Girona 18-26, 08034 Barcelona, Spain, and AGBAR, Aigu ¨ es de Barcelona, P. Sant Joan 39, 08009, Barcelona, Spain
The elimination of nonylphenol (NP), nonylphenol monoand diethoxylates (NP1EO and NP2EO), nonylphenol carboxylates (NP1EC and NP2EC) and their brominated and chlorinated derivatives during drinking water treatment process in Sant Joan Despı´ waterworks in Barcelona was investigated utilizing a recently developed, highly sensitive LC-MS-MS method. The concentration of these potentially estrogenic compounds in raw water entering waterworks (taken from the Llobregat River, NE Spain) ranged from 8.3 to 22 µg/L, with NP2EC being the most abundant compound. Prechlorination reduced the concentration of short-chain ethoxy NPECs and NPEOs by about 2535% and of NP by almost 90%. However, this reduction of concentrations was partially due to their transformation to halogenated derivatives. After prechlorination, halogenated nonylphenolic compounds represented approximately 13% of the total metabolite pool, of which 97% were in the form of brominated acidic metabolites. The efficiency of further treatment steps to eliminate nonylphenolic compounds (calculated for the sum of all short-chain ethoxy metabolites including halogenated derivatives) was as follows: settling and flocculation followed by rapid sand filtration (7%), ozonation (87%), GAC filtration (73%), and final disinfection with chlorine (43%), resulting in overall elimination ranging from 96 to 99% (mean 98% for four sampling dates). A few of the nonylphenolic compounds (NP, NP1EC, and NP2EC) were also identified in drinking water; however, the residues detected were generally below 100 ng/L, with one exception for NP2EC in November 2001 when a concentration of 215 ng/L was detected.
Introduction Due to the scarcity of groundwater resources, Mediterranean drinking water treatment plants (DWTP) are frequently fed by surface waters. In most instances, drinking water is derived from river water, which contains a significant portion of wastewater originating from discharges of industrial and municipal effluents or agricultural activity in catchment * Corresponding author phone: (34)93 400 6172; fax: (34)93 204 59 04; e-mail:
[email protected]. † IIQAB-CSIC. ‡ Aigu ¨ es de Barcelona. 4442
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 19, 2003
zones. Thus, the occurrence of organic microcontaminants in raw water and their removal in the course of drinking water production and possible formation of disinfection byproducts are key issues in relation to the quality of drinking water supplies. The Llobregat River (Catalonia, NE Spain) is the most important source of drinking water for the Barcelona area with a population of 3.2 million people. The DWTP Sant Joan Despı´ with the daily production of 300 000 m3 of drinking water, which corresponds to 35% of water in the metropolitan supply network, is situated downstream of a densely industrialized area near the mouth of the river. Since the 1960s, the effluents from tannery, textile, pulp, and paper industries situated along the basin have been discharged into the river; consequently, a broad spectrum of organic chemicals has been found in river water entering DWTP, including hydrocarbons, pesticides, surfactants, and plasticizers (1, 2). One of the classes of compounds widely occurring in the Llobregat River is nonylphenol ethoxylates (NPEO) and their metabolites. Concern was raised regarding the environmental safety of this class of nonionic surfactants and particularly of their short-chain ethoxy metabolites: nonylphenol monoand diethoxylates (NP1EO and NP2EO), nonylphenol carboxylates and ethoxycarboxylates (NP1EC and NP2EC), and nonylphenol (NP). These compounds are substances perceived to possess estrogenic activity due to their ability to mimic endogenous hormone 17β-estradiol (3, 4). Several studies determined that alkylphenolic compounds accounted for a large portion of the estrogenicity of sewage effluents discharged into the Llobregat River or its tributaries (5-7). Thus, taking into consideration the high requirements on the quality of drinking water, it is important to study their behavior during treatment at DWTP. In early works by Ventura and co-workers (8, 9) a wide range of nonionic surfactants of polyethoxylate type and their neutral and acidic metabolites were identified in raw water entering DWTP and also qualitatively detected in finished water. Moreover, the formation of halogenated derivatives (such as ring-brominated and chlorinated NPEOs, NPECs, and NPs during the chlorination process at DWTP) due to high concentration of bromide ions in raw water that arises from daily dumps of salt mining effluents into the upper course of the river was reported (10, 11). A recent study employing recombinant yeast assay (RYA) and MCF7 test for both estrogenic and anti-estrogenic activities and enzymelinked receptor assay (ELRA) to probe binding to the receptor indicate (12) confirmed that halogenated nonylphenolic derivatives retained a significant affinity for the estrogen receptors, suggesting that they may be able to disturb the hormone imbalance of exposed organisms. Several studies have reported levels of NPEOs and their degradation products in wastewater treatment plant (WWTP) effluents discharged into the Llobregat River or its tributaries (5, 7) and in river water entering DWTP (13, 14). Although levels of several tens of micrograms per liter were found in the Llobregat River upstream of DWTP, the behavior of estrogenic nonylphenolic compounds during processing of this contaminated water in waterworks and their possible occurrence in treated water has rarely been the scope of interest, and there are hardly any data available for drinking water. This study aimed to fill the gap and to contribute to the gathering of data that are an important prerequisite for a better ecotoxicological assessment of nonylphenolic compounds. Thus, the objectives of the work presented here were (i) to study the efficiency of different treatment steps to 10.1021/es034139w CCC: $25.00
2003 American Chemical Society Published on Web 09/05/2003
FIGURE 2. Scheme of drinking water treatment plant at the Llobregat River with sampling points (sites 1-6). Conditions for specific sampling dates: prechlorination 3-18 mg/L (break point); ozonation 0.22-0.40 mg/L; final chlorination 1.4-2.05 mg/L.
FIGURE 1. General structures and acronyms of the studied shortchain ethoxy NPEO metabolites and corresponding halogenated derivatives. Note: the exact structure of the alkyl chain is unknown (different branching is possible). remove nonylphenolic compounds during drinking water treatment at the water works Sant Joan Despı´ in Barcelona and (ii) to study the formation of chlorination byproducts (halogenated nonylphenolic derivatives). The study covers both neutral and acidic NPEO metabolites and their halogenated derivatives (structures and acronyms are shown in Figure 1).
Experimental Section Standards and Reagents. All solvents (water, acetonitrile, methanol) were HPLC grade and were purchased from Merck (Darmstadt, Germany). Technical grade 4-NP was obtained from Aldrich (Milwaukee, WI). NP1EO, NP2EO, NP1EC, and NP2EC were synthesized according to the method described elsewhere (14). The purity of all synthesized compounds was 95% or higher. BrNP was synthesized using elemental bromine according to the method described by Reinhard et al. (15). ClNP was prepared by chlorination of nonylphenol using sulfuryl chloride according to the method of Stokker et al. (16). BrNP1EC and ClNP1EC were synthesized by reacting brominated and chlorinated NP, respectively, with chloroacetic acid in the presence of sodium hydride and dimethylformamide as a solvent. These two synthesized compounds rendered BrNP1EO and ClNP1EO by reduction with lithium aluminum hydride in ether solution. BrNP2EO and ClNP2EO were synthesized by reacting BrNP and ClNP, respectively, with 2-(2-chloroethoxy)ethanol in the presence of NaOH in water. Finally, BrNP2EC and ClNP2EC were obtained from BrNP2-
EO and ClNP2EO, respectively, by oxidation with Jones reagent (15). The purity of all synthesized halogenated compounds was higher than 95%. Description of the DWTP and Sampling. A detailed scheme of installed treatment steps at the DWTP Sant Joan Despı´ (Barcelona, Spain), showing six sampling sites, is displayed in Figure 2. Water and sludge samples were taken in four occasions from July to November 2001. Raw water entering the DWTP (the Llobregat River) and water samples after each treatment step were collected as grab samples in Pyrex borosilicate amber glass containers, previously rinsed with Milli-Q water and acetone. To avoid further reaction with residual chlorine, samples were stabilized with 0.4% (v/v) ascorbic acid (0.1 M). Afterward, samples were stored at 4 °C and extracted as soon as possible (latest within 2 d). Sludge obtained from prechlorinated raw water after flocculation with aluminum sulfate and mixed in a minor proportion with sludge coming from the washing of sand filters was collected in precleaned amber glass bottles. The suspension (concentration of dry matter 3.5-5 g/L) was centrifuged at 4500 rpm, and the solid matter was separated and frozen at -20 °C before being freeze-dried. Chemical Analysis. NPEOs and their halogenated derivatives were analyzed using LC-MS as described by Petrovic et al. (2), whereas NP, NPECs, and their halogenated derivatives were analyzed using the recently developed LCMS-MS method (17). Briefly, target compounds were isolated from water samples using solid-phase extraction (SPE) on C18-silica (Accubond, J&W Scientific, Folsom, CA). Freeze-dried sludge was extracted with acetone-methanol (1:1, v/v) using a Dionex ASE 200 (Dionex, Idstein, Germany) as described elsewhere (18). Sludge extracts were then purified by SPE using LiChrolute C18 cartridges (Merck, Darmstadt, Germany). LC-MS-MS analyses were performed on a Waters 2690 series Alliance HPLC (Waters, Milford, MA) with a quaternary pump equipped with a 120-vial capacity sample management system. The analytes were separated on a narrow-bore 3-µm, 55 × 2 mm i.d. C18 reversed-phase column Purospher STAR RP-18 end-capped (Merck, Darmstadt, Germany). The sample injection volume was set at 10 µL. A binary mobile phase gradient with methanol (A) and water (B) was used for analyte separation at a flow rate of 200 µL/min. The elution gradient VOL. 37, NO. 19, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4443
TABLE 1. List of MRM Channels and Ions Monitored in SIM Mode for the Analysis of Nonylphenolic Compounds and Their Halogenated Derivatives ionization compound mode NP1EO NP2EO ClNP1EO ClNP2EO BrNP1EO BrNP2EO NP1EC NP2EC NP ClNP1EC ClNP2EC BrNP1EC BrNP2EC ClNP BrNP
PI PI PI PI PI PI NI NI NI NI NI NI NI NI NI
LC-MS-MS MRM1
LC-MS MRM2
SIM 297 331 321/323 365/367 365/367 409/411
277 f 219 321 f 219 219 f 133 311/313 f 253/255 355/357 f 253/255 355/357 f 297/299 399/401 f 297/299 253/255 f 167/169 297/299 f 79/81
219 f 133 219 f 133 219 f 147 253/255 f 167/169 253/255 f 167/169 355/357 f 79/81 297/299 f 79/81 253/255 f 181/183 297/299 f 211/213
was linearly increased from 30% A to 85% A in 10 min, then increased to 95% A in 10 min, and kept isocratic for 5 min. A benchtop triple quadrupole mass spectrometer Quattro LC from Micromass (Manchester, UK) equipped with a pneumatically assisted electrospray probe and a Z-spray interface was used for this study. Capillary voltage was set at -2.8 kV, extractor lens at 7 V, and RF lens at 0.6 V. The source and desolvation temperatures were 150 and 350 °C, respectively. The nitrogen (99.999% purity) flows were optimized at 50 L/h for the cone gas and at 540 L/h for desolvation gas. For MS-MS experiments, the argon collision gas was maintained at a pressure of 5.8 × 10-3 mbar. Quantitative LC-MS-MS analysis of compounds detected under negative ionization (NI) conditions (NP, NPECs, and corresponding chlorinated and brominated analogues) was carried out in multiple reactions monitoring (MRM) mode (see Table 1), while compounds detected under positive ionization (PI) conditions (NPEOs and corresponding chlorinated and brominated analogues) gave only [M + Na]+ adduct ions and produced no fragmentation. As a result, these compounds were analyzed using a single-stage MS in selected ion monitoring (SIM) mode. Under the experimental conditions used (short LC column), ClNPEO and BrNPEO coeluted, which was an obstacle for a separate quantification of isobaric compounds (e.g., ClNPn+1EO and BrNPnEO). The concentration of these compounds was estimated using a standard mixture of available compounds (ClNP1EO, ClNP2EO, BrNP1EO, and BrNP2EO, respectively); all present in the same amount. The obtained signals at m/z 321 corresponded to ClNP1EO, at m/z 365 corresponded to BrNP1EO + ClNP2EO, and at m/z 409 corresponded to the sum of BrNP2EO and ClNP3EO. The final results were expressed as a sum of all obtained concentrations, which therefore corresponded to the sum of ClNPEO (nEO ) 1-3) and BrNPEO (nEO ) 1-2). Further details on the methodology used, including detailed description of the fragmentation pathway and specific issues
related with the optimization of MS detection parameters and reduction of ion suppression effects, are given elsewhere (2, 17, 18). The method yielded detection limits of 2 ng/L for compounds detected by LC-MS-MS and 20 and 10 ng/L for mono- and diethoxylates, respectively, in water and 1-2 ng/g (LC-MS-MS), 10 ng/g (monoethoxylates), and 5 ng/g (diethoxylates) in sludge. Recoveries from water ranged from 72 to 98%, with standard deviations below 7%. Extraction of sludge yielded recoveries higher than 60%, with standard deviations below 14%.
Results and Discussion Raw Water Quality. The Llobregat River (154 km) receives, directly or via its tributaries, effluents from more than 30 WWTP. The whole river basin is heavily industrialized, and some of these WWTP receive significant percentage of wastewaters of industrial origin, mainly from textile industry and tanneries. The portion of treated water in the lower course of the Llobregat River shows pronounced seasonal variations, and during summer period, when the overall river flow is low, it may increase significantly. This has a significant impact on the quality of raw water entering DWTP and results in high levels of some organic and inorganic pollutants. Another characteristic of water from the Llobregat River is elevated concentrations of Cl- and Br- ions, which are the consequence of runoff waters from salt mines situated in the upper course of the river. Selected data characterizing the raw water entering DWTP are listed in Table 2, whereas Figure 3 shows the concentration of alkylphenolic compounds entering the DWTP. The daily variability of the concentration of alkylphenolic compounds was estimated to be low, and samples, although taken as grab samples, were considered to represent a daily average concentration. In the raw water short-chain ethoxy, metabolites were detected in the low micrograms per liter range, with NP2EC being the most abundant compound. Concentrations of acidic metabolites (NP1EC and NP2EC) ranged from 2.0 to 4.1 µg/L and from 2.8 to 15 µg/L, respectively, whereas less polar neutral metabolites were present in concentrations of 1.1-2.2 µg/L NP, 0.7-1.8 µg/L NP1EO, and 1.2-2.4 µg/L NP2EO for the studied sampling dates (from July to November). The variability in detected concentrations during the sampling period was probably influenced by the hydrological regime of the river. Low flow of the river during dry periods may result in a worsening of the quality of the raw water due to a higher proportion of treated wastewater. Thus, the highest concentrations of nonylphenolic compounds were detected in October 2001 when the river flow was only 3.2 m3/s, which represented the lowest value of the annual range river flow. However, the concentration of nonylphenolic compounds in raw water also depends on other factors, such as concentration of dissolved oxygen and water temperature. These parameters strongly influence biotic and abiotic transformation reactions and are closely related to the selfpurification capability of the water. The levels found are generally higher than those detected in surface waters in other studies. The concentrations in 30
TABLE 2. River Flow and Physicochemical Analysis of Raw Water sampling date
river flow (m3/s)
temperature (°C)
pH
DOC (mg of C/L)
conductivity (µS/cm)
NH4+ (mg/L)
Cl(mg/L)
Br(mg/L)
07/24/2001 09/17/2001 10/16/2001 11/13/2001
4.60 4.30 3.20 3.70
24.0-27.5a 21.8-23.0 18.7-20.6 10.5-13.1
7.96-8.47a 7.99-8.41 7.74-8.12 8.09-8.23
3.5-4.4a 4.3-4.9 4.3-5.9 5.4-7.5
1235-1280a 1150-1270 1200-1270 1390-1920
0.09-0.28a 0.03-0.28 0.03-0.12 0.30-1.90
234-260a 213-238 234-269 298-389
0.61-0.76b 0.42-0.66 0.54-0.67 0.55-0.76
a
Daily range.
4444
9
b
Monthly range.
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 19, 2003
FIGURE 3. Concentration of NPEC (a), NP (b), and NPEO (c) in waters of different treatment steps. Samples 1-4 correspond to the sampling dates July 24, 2001; September 17, 2001; October 16, 2001; and November 13, 2001, respectively. U.S. streams ranged from
