Research Fate of Estrogens in a Municipal Sewage Treatment Plant HENRIK ANDERSEN,† HANSRUEDI SIEGRIST,‡ BENT HALLING-SØRENSEN,† AND T H O M A S A . T E R N E S * ,§ The Danish University of Pharmaceutical Sciences, Institute of Analytical Chemistry, Universitetsparken 2, DK-2100 Copenhagen, Denmark, EAWAG, Ueberlandstrasse 133, 8600 Duebendorf, Switzerland, Bundesanstalt fu ¨ r Gewa¨sserkunde (BfG), Kaiserin-Augusta-Anlagen 15-17, D-56068 Koblenz, Germany
The fate of the highly potent endocrine disrupters estrone (E1), 17β-estradiol (E2), and 17R-ethinylestradiol (EE2) was investigated in mechanical and biological sewage treatment as well as in sewage-sludge treatment at a municipal German sewage treatment plant (STP). The main outcome of the study was that a common municipal STP with an activated sludge system for nitrification and denitrification including sludge recirculation can appreciably eliminate natural and synthetic estrogens. As a consequence, the endocrine effects of biota in the receiving waters should be significantly reduced. All estrogen concentrations decreased gradually along the treatment train. In the STP effluent, the steroid estrogen concentrations were always below the quantification limit of 1 ng/L. The elimination efficiency of the natural estrogens (E1 and E2) exceeded 98%, and EE2 was reduced by more than 90%. The natural estrogens were largely degraded biologically in the denitrifying and aerated nitrifying tanks of the activated sludge system, whereas EE2 was only degraded in the nitrifying tank. Only about 5% of the estrogens are sorbed onto digested sewage sludge. It is very likely that conjugates (glucuronides and sulfates) of the estrogens were cleaved into the parent compounds mainly in the first denitrification tank.
Introduction Desbrow, Routledge, and their co-workers (1, 2) have reported that the estrogenic effect on fish caused by sewage treatment plant (STP) effluents may in some cases be attributed to the presence of the natural estrogens 17β-estradiol (E2) and estrone (E1) as well as the active ingredient of most birthcontrol drugs, 17R-ethinylestradiol (EE2). Recent publications suggest that steroids may be the main source of estrogenicity in many municipal STPs (3, 4). Since the sources of natural estrogens cannot be eliminated, specific treatment processes in STPs must be optimized. To date, it is known that municipal STPs reduce steroid estrogens to some extent, although frequently not to levels lower than the known effective * Corresponding author telephone: +49 261 13065560; fax: +49 261 1305363; e-mail:
[email protected]. † The Danish University of Pharmaceutical Sciences. ‡ EAWAG. § Bundesanstalt fu ¨ r Gewa¨sserkunde (BfG). 10.1021/es026192a CCC: $25.00 Published on Web 07/26/2003
2003 American Chemical Society
concentrations for fish (5-9). Ozonation, UV-radiation, membrane filtration, and activated carbon adsorption are potential treatments that might improve the effectiveness of estrogen removal in a STP (4, 5, 10-15). However, implementation of these techniques would increase the cost of wastewater treatment. Alternatively, understanding the fate of estrogens within the STPs might yield removal methods based on better management or minor modifications of current STPs. In batch experiments, Ternes et al. (16) investigated the basic aerobic microbial reactions of estrogens in contact with activated sludge taken from the old Wiesbaden plant, which removes BOD. Experiments were performed at 20 °C with a starting concentration of 1 µg/L and a total suspended solids (TSS) concentration of 0.52 g of TSS/L. Assuming a pseudofirst-order reaction (TSS ) constant): dCE2/dt ) -kE2‚TSS‚ CE2 with the pseudo-first-order reaction constant kE2 ≈ 150 L/(g of TSS‚d), E2 was found to have a T1/2 of about 0.2 h with nearly all the E2 being converted to E1. E1 was removed more slowly with a T1/2 of about 1.5 h at 20 °C in the same concentration range and with the same TSS concentration. The pseudo-first-order reaction constant was therefore kE1 ≈ 20 L/(g of TSS‚d). In this study, EE2 was not degraded significantly within 48 h. Layton et al. (17) conducted a similar study with estrogens using activated sludge from STPs operating in a warmer climate. E2 was mineralized to CO2 within a few hours. Thus, an accumulation of E1 could not be observed. As in the German study, EE2 was found to metabolize much more slowly, even though 40% was mineralized in 24 h. Activated sludge bacteria can therefore be expected to be more active at higher ambient temperatures. Estrogens from humans are largely excreted as conjugates, mainly glucuronides (18). In batch experiments, E2glucuronides were cleaved to E2 with pseudo-first-order reaction constants similar to those of E2 degradation, and the E2 produced by the reaction was then oxidized to E1 (16). In some studies, concentrations of E1 and E2 or estrogen activity increased after primary settling as compared to the raw influent (19-21), while Holbrook et al. (13) found no intermediate increase in steroid estrogen concentrations at five U.S. STPs. It was suggested (19) that the intermediate increase of these concentrations could be explained by cleavage of conjugated steroid estrogens in the primary clarifier. This assumption is consistent with the findings of Adler et al. (22) that on average 58% of total E1, 50% of total E2, and 26% of total EE2 were conjugated in raw sewage from different parts of Germany. Matsui et al. (20) recently performed a detailed profile of estrogen removal in a Japanese STP using an immunoassay for E2 in combination with the yeast estrogen screening (YES) assay for measuring estrogen activity. The estrogenicity measured by YES tended to decline during the treatment train, but the major reduction was found in the denitrification step, which was also the first step in the active sludge treatment. It was further observed that the E2 concentrations and estrogen activity of the dewatering liquid from the sludge treatment were even more than twice as high as the inflow to the plant. The aim of the current paper was to investigate the fate of E1, E2, and EE2 due to sorption and degradation in each treatment step of a municipal STP with nitrification, denitrification, and both biological and chemical phosphate removal. VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
4021
FIGURE 1. Flow scheme of the Municipal STP Wiesbaden with sampling locations
Experimental Section Estrogens in Water. Estrogens were analyzed in water according to the method reported by Ternes et al. (19). The analytes were extracted from filtered water samples onto RP-C18 cartridges. Extracts were cleaned with a deactivated silica gel column. The estrogens were trimethylsilylized before analyzing with GC-ion trap-MS. As against the original method, however, the brand of the RP-C18 solid-phase material was changed to RP-C18 Bulk Sorbent from Separtis GmbH (Grenzach-Wyhlen, Germany). Furthermore, an additional cleanup step based on gel permeation chromatography (GPC) was introduced to remove matrix interference from raw sewage extracts. Details of the GPC method and evaluation of the GPC procedure are reported by Ternes et al. (23). A 10-point calibration was performed over the whole procedure after spiking groundwater with the respective estrogens at concentrations between 0.25 and 100 ng/L. In each analysis series, a blank sample of deep groundwater was run in parallel. According to the evaluation method originally described (19), the recovery of the steroid estrogens was g93%, and the relative standard deviation (RSD) of the method was e14%. A further quality parameter designed to ensure that matrix effects did not influence the extraction or cleanup, not mentioned in the original description (19), was the ratio between the surrogate standard, 17β-estradiol-17acetate, and the instrumental standard, Mirex (CAS Registry No. 2385-85-5). The average of this ratio was determined as a mean of all calibration samples. The quantitative result of a real sample was only considered valid if this ratio deviated less than 30% from the average value of the calibrations. The newly included GPC cleanup step exhibited a systematic loss of 11%. This loss was effectively corrected by the surrogate standard, 17β-estradiol-17-acetate. Estrogens in Sludge. The method used for measuring estrogens in sludge was recently described in Ternes et al. (23). Sludge was freeze-dried, and aliquots of 0.5 g were successively extracted twice with methanol and subsequently twice with acetone. For each extraction step, the slurry of the sample in solvent was ultrasonicated for 10 min. The four solvent fractions were combined, and a cleanup was carried out with GPC and silica gel. The extracts were then derivatized and analyzed by GC-ion trap-MS. Standards for the calibration curve and a blank for each analysis series were made by spiking the combined solvent mixture used for extraction followed by the full cleanup procedure described for native samples. The 9-point calibration curve was performed with the respective estrogens at concentrations between 1 and 200 ng/g. The recoveries of the analytes were g83% with an RSD of e19%. As in the water analysis described above, the ratio of the surrogate standard 17β-estradiol-17β4022
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 18, 2003
acetate and the instrument standard Mirex was used as the quality parameter for the analysis. Furthermore, as sewage sludges vary considerably between different STPs, the suitability of the method was confirmed for each type of sludge (at each sampling location) by spiking 100 ng/g of the estrogens in an individual recovery experiment. The main quality criterion was that the recovery was within the 95% confidence interval range found for the different types of sludge used in the original evaluation method. Flow Scheme of Plant and Wastewater Characterization. The schematic of the municipal STP Wiesbaden is shown in Figure 1. The mechanical treatment consists of a screen, an aerated grit-removal tank, and a primary clarifier. The primary sludge collected in the primary clarifier is pre-thickened and then pumped into the mesophilic digester (33 °C, 20-d retention time). The primary effluent is directed to the activated sludge system for biological and chemical phosphate removal, denitrification, and nitrification. Fe(II)Cl2 is added in the first denitrification tank for efficient mixing in the water before oxidation to Fe(III) and subsequent precipitation with phosphate in the aerated nitrification tanks. After settling in the secondary clarifier, the activated sludge is returned to the inlet of the first denitrification tank. Internal recirculation supplies additional nitrate to the inlet of the first tank to improve denitrification. The secondary effluent is released to the river Rhine after its suspended solids had been reduced in a rotary sieve (not indicated). The activated sludge system is operated with a solids retention time of 11-13 d, which is typical for a nitrifying plant with predenitrification. The excess sludge is dewatered to about 5% total solids concentration in a rotary sieve before being pumped into the digester. The digested sludge is thickened and then treated thermophilically at 200 °C (not indicated) for sludge reduction before being dewatered in a filter press. The sludge liquid produced after thickening and dewatering is sent to the primary clarifier. Several parameters (e.g., BOD5, COD, ammonium, nitrate, phosphate, and T) were measured regularly according to German DIN standards at selected points in the treatment process. The nitrogen levels are of special interest for comparison with other STPs, as the current study suggests a link between steroid estrogen elimination, nitrification, and denitrification. On November 13, the nitrogen concentrations in the secondary effluent were 1.0 mg of NH4-N/L and 7.3 mg of NO3-N/L. The ranges during the preceding 2 weeks were 0.6-1.3 mg of NH4-N/L and 6.3-7.6 mg of NO3N/L, which illustrates well-functioning nitrification and denitrification. On the sampling days, the water temperatures in the biological treatment units were 16-17 °C. The average for the 2-week period preceding the sampling campaign was
TABLE 1. Measured Concentration and Mass Flux of Estrone (E1), 17β-Estradiol (E2), and 17β-Ethinylestradiol (EE2) in the STP Wiesbadene
VOL. 37, NO. 18, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
a Total concentration including fraction sorbed on suspended solids. b Average of nitrification and excess sludge. c Average of inlet and primary sludge. d Assumed to be similar as in digestion. e Boldface type indicates measured data. The range of minimum to maximum is indicated in parentheses. Italic type indicates estimated data. Regular type indicates calculated data.
9
4023
FIGURE 2. Measured mass flux of estrone (E1) and 17β-estradiol (E2) in g/d. Adler et al. (22) estimated the conjugated (glucuronide, sulfate) estrogen quantities for a municipal STP inlet to be about 130% of the nonconjugated compounds. The estimated inlet load of 14 g/d corresponds to about 45 µg/(p d). 17.7 °C with extremes of 14.8 and 19.9 °C. Furthermore, the inlet, return sludge, internal recirculation, and sludge streams as well as the sludge-liquid recycle streams were recorded and used for calculating the loads of estrogens and the dilution effects in the first denitrification tank. The water inflow to the STP was 66 000 m3 on November 13, which is within the range seen in the preceding 2 weeks of 40 000110 000 m3. On November 13, the return sludge flow and internal recirculation were about 50% and 200% of the inlet flow, respectively. Sampling. The sampling locations are shown in Figure 1. Sampling was generally carried out at the outlet of each treatment step. The sewage was not accessible for sampling before the grit removal tank. Water samples were taken as 24-h flow-proportional composite samples (cooled at 4 °C) from the outlet of the grit-removal tank, the primary and secondary sedimentations being collected from midnight to midnight. All other sludge-liquid and sludge samples were taken randomly between 9 a.m. and 11 a.m. The steps in the water treatment process were sampled on November 13 and 14, 2001, while the sludge treatment process was only sampled on November 13. The retention time of the sludge in the digester is about 20 d. The samples taken from the digester do not therefore correspond directly to the analyzed inlet sludge but rather to the average load of the previous 20 d. In general, the total concentrations of suspended solids in the 24-h composite samples of the main water flow path were low (
