ARTICLE IN PRESS

Water Research 39 (2005) 47974807 www.elsevier.com/locate/watres

Removal of selected pharmaceuticals, fragrances and endocrine disrupting compounds in a membrane bioreactor and conventional wastewater treatment plants
M. Claraa, B. Strenna, O. Gansb, E. Martinezb, N. Kreuzingera,, H. Kroissa
a

Institute for Water Quality and Waste Management, Vienna University of Technology, Karlsplatz 13, 1040 Vienna, Austria b Umweltbundesamt GmbH, Spittelauer Lande 5, 1090 Vienna, Austria Received 10 September 2004; received in revised form 1 September 2005; accepted 13 September 2005 Available online 19 October 2005

Abstract Eight pharmaceuticals, two polycyclic musk fragrances and nine endocrine disrupting chemicals were analysed in several waste water treatment plants (WWTPs). A membrane bioreactor in pilot scale was operated at different solid retention times (SRTs) and the results obtained are compared to conventional activated sludge plants (CASP) operated at different SRTs. The SRT is an important design parameter and its impact on achievable treatment efciencies was evaluated. Different behaviours were observed for the different investigated compounds. Some compounds as the antiepileptic drug carbamazepine were not removed in any of the sampled treatment facilities and efuent concentrations in the range of inuent concentrations were measured. Other compounds as bisphenol-A, the analgesic ibuprofen or the lipid regulator bezabrate were nearly completely removed (removal rates 490%). The operation of WWTPs with SRTs suitable for nitrogen removal (SRT410 days at 10 1C) also increases the removal potential regarding selected micropollutants. No differences in treatment efciencies were detected between the two treatment techniques. As in conventional WWTP also the removal potential of MBRs depends on the SRT. Ultraltration membranes do not allow any additional detention of the investigated substances due to size exclusion. However, MBRs achieve a high SRT within a compact reactor. Nonylphenolpolyehtoxylates were removed in higher extend in very lowloaded conventional WWTPs, due to variations of redox conditions, necessary for the degradation of those compounds. r 2005 Elsevier Ltd. All rights reserved.
Keywords: Wastewater treatment; Endocrine disrupting chemicals; Pharmaceuticals; Musk fragrances; Membrane bioreactor; Removal efciency

1. Introduction Several pharmaceuticals, ingredients of personal care products and so-called endocrine disrupting compounds
Corresponding author. Tel.: +43 1 58801 22622;

fax: +43 1 58801 22699. E-mail addresses: manfred.clara@umweltbundesamt.at (M. Clara), norbkreu@iwag.tuwien.ac.at (N. Kreuzinger).

(EDCs) (hormones and chemicals, which are suspected to have an impact on humans and wildlife hormone systems) are detected in surface waters in several European countries (Ternes, 1998; Fromme et al., 2002; Heberer, 2002a). Negative adverse health effects on aquatic organisms, which could be attributed to EDCs, are documented in various studies (Sonnenschein and Soto, 1998; Sumpter, 1998). Heberer (2002b) investigated the occurrence of pharmaceuticals in

0043-1354/$ - see front matter r 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2005.09.015

ARTICLE IN PRESS
4798 M. Clara et al. / Water Research 39 (2005) 47974807

drinking water and in wastewater and tracked some persistent compounds from municipal sewage to drinking water. As the concentrations measured in the environment are in the range of nanogram per litre (ng l1) or a few microgram per litre (mg l1) they were often referred to as micropollutants. Most of those compounds are of anthropogenic origin and wastewater treatment plant (WWTP) efuents are important point discharges for the presence of endocrine disrupting compounds and residuals of pharmaceuticals in rivers, streams and surface waters. Therefore the elimination of these substances within the WWTP or their retention is of elementary interest. In conventional activated sludge plants (CASP) the aeration tank and the nal clarier form one process unit. The separation of treated sewage and sludge occurs in the clarier via sedimentation. Therefore the ability to sediment is an important selection criterion. The biomass concentration in the mixed liquor is limited by the capacity of the clarier. In membrane bioreactors this parameter is of minor inuence, as separation is achieved via membrane ltration. Thus, the plant can be operated at higher biomass concentrations resulting in smaller plant sizes. The most important advantage of MBRs is the complete retention of suspended solids, thus reducing emissions to the dissolved fractions. Higher costs and higher requirements in operation and maintenance as well as power consumption compared to conventional systems are well-known disadvantages. MBRs are often highlighted regarding their potential to improve hazardous sub-

stances removal from sewage and their importance for the aquatic environment, respectively. One reason is linked to the complete retention of solids, as hydrophobic organic substances tend to accumulate onto the sludge. Furthermore, as all bacteria are held back, their adaptation to the more efcient mineralising of the respective compounds is possible (Ivashechkin et al., 2004). Another reason is that molecules larger than the molecular weight cut-off (MWCO) of the membrane cannot penetrate membrane pores due to size exclusion. However, only few comparative studies on the fate of organic micropollutants in MBRs and conventional systems are available. In wastewater treatment usually micro- and ultraltration membranes are used. During the last years the number of realized WWTPs equipped with membrane technology increased notably. Especially in regions with no suitable receiving waters or where treated wastewater is used for groundwater inltration MBRs are suitable alternatives to conventional systems, also due to their disinfection capabilities. The purpose of the present study is neither the optimisation of operational settings at the investigated MBR pilot plant nor the optimisation in respect to nutrient removal. The study investigates the treatment efciency of a MBR pilot plant equipped with an ultraltration membrane and operated at different solids retention times (SRTs) regarding the removal of selected micropollutants. The SRT describes the mean residence time of the biomass in the biological reactor. It is related to the growth rate of micro organisms, as only

Table 1 Investigated compounds (CAS number, molecular weight MW (g mol1)) and respective consumption (kg year1) in Austria according to Sattelberger (1999) and Skutan (2003) Category Pharmaceutical Pharmaceutical Pharmaceutical Pharmaceutical Pharmaceutical Pharmaceutical Pharmaceutical Pharmaceutical Musk fragrance Musk fragrance EDC EDC EDC EDC EDC EDC EDC EDC EDC (analgesic) (analgesic) (lipid regulator) (antiepileptic) (contrast media) (tranquilizer) (antibiotic) (antibiotic) Substance Diclofenac (DCF) Ibuprofen (IBP) Bezabrate (BZF) Carbamaezpine (CBZ) Iopromide (IPM) Diazepam (DZP) Roxithromycin (ROX) Sulfamethoazole (SMX) Tonalide (AHTN) Galaxolide (HHCB) Bisphenol-A (BPA) Nonylphenol (NP) Nonylphenol monoethoxylate (NP1EO) Nonylphenol diethoxylate (NP2EO) Nonylphenoxyacetic acid (NP1EC) Nonylphenoxyethoxyacetic acid (NP2EC) Octylphenol (OP) Octylphenol monoethoxylate (OP1EO) Octylphenol diethoxylate (OP2EO) CAS 15307-86-5 15687-27-1 41859-67-0 298-46-4 73334-04-3 439-14-5 80214-83-1 723-46-6 1506-02-1 1222-05-5 80-05-7 25154-52-3 104-35-8 20427-84-3 3115-49-9 106807-78-7 140-66-9 2315-67-5 2315-61-9 Consumption 6143 6696 4474 6334 5386 125 Not available 963 Not available Not available $14,500,000 $7,000,000

Not available Not available Not available

ARTICLE IN PRESS
M. Clara et al. / Water Research 39 (2005) 47974807 4799

organisms can be detained and enriched that are able to reproduce themselves during this time. WWTPs operating at high SRTs allow the enrichment of slowly growing bacteria and consequently the establishment of a more diverse biocoenosis with broader physiological capabilities (e.g. nitrication). The results observed in the MBR are compared to several conventional WWTPs. Eight pharmaceuticals, two polycyclic musk fragrances and nine endocrine disrupting chemicals were studied (see Table 1).

2. Materials and methods 2.1. Analytical methods Different analytical methods were applied to determine the concentration levels of the analytes in the water samples. Analyses were performed by the Austrian Umweltbundesamt. Statistical evaluation of measurements and determination of limits of quantication (LOQ) were based according to DIN 32645 (1994). Indicated recoveries are mean recovery rates for the whole analytical method, including all steps of the analytical process. Alkylphenols and alkylphenol ethoxylates are present in plastics and detergents and glassware required special treatment prior to use. Therefore, case was taken to avoidas far as possibleplastics for sampling since traces of the surveyed industrial chemicals may leach the plastics into the sample. Layered sampling bottles (aluminium, 1.2 l; Burkle, Germany) to avoid contact of the sample with the plastic screw mount were provided by the Umweltbundesamt. The material can withstand freezing of the sampling bottle. In order to avoid contamination, bottles were regularly cleaned, rinsed with acetone and water and heated up to 110 1C over night. Afterwards, they were rinsed with isooctane in order to deactivate the surface and were dried until the remaining solvent evaporated. In the case of the selected pharmaceuticals and personal care products, also aluminium bottles, 1,2 l (Burkle, Germany) were used for the sampling and the samples were stabilized with sodium azide. For the analysis of the selected EDCs, 500 ml water samples were acidied with sulphuric acid to pHp3 and 25 ml of surrogate standard (BPA-d16 and NP1EC-d2) was added. The conditioning and loading steps were carried out automatically by the Auto Trace extraction Work Station (Zymark, Hopkinton, MA, US), using C18 phase cartridges (1 g, 6 ml). After drying with nitrogen for 20 min, the compounds were eluted by acetone and a mixture of methanol/methyl-tert.-butylether (1:9). The extracts were evaporated in a Turbo Vap II (Zymark, Hopkinton, MA, US) to 0.5 ml by nitrogen and lled up to a nal volume of 1 ml with

acetonitrile (ACN). 10 ml of internal standard (4-n-NP, BPA-d8) was added prior to analysis. LC-MS/MS applying electro spray ionisation (ESI) was used for the determination of the analytes. BPA, NP, OP and NP1,2ECs were detected in the negative ion mode, whereas the NPEOs and OPEOs were analysed in the positive ion mode. A detailed description of the analytical method for the determination of alkylphenols is given in Mar tnez et al. (2004). To analyse CBZ, IPM, DZP, SMX and ROX, two different solid phase extraction methods with CH (1 g, 6 ml) and ENV+ (200 mg, 6 ml) phases were employed. In the rst method 500 ml samples were acidied, whereas in the second method a neutral EDTA buffer was used. Prior to sample extraction a surrogate standard (josamycine, tylosine and dihydro-CBZ) was added to the samples. The conditioning, loading and elution steps were carried out automatically. The analytes were eluted with methanol, ethyl acetate and dichloromethane, in the case of the acidic sample preparation, and with ACN and a mixture of acetonitril/water/tri-ethyl-amine, in the case of neutral sample preparation. The extracts were evaporated to below 0.5 ml and then lled up to a nal volume of 1 ml with ACN/H2O 1:9. These compounds were analysed by LCMS/MS with ESI in the positive ion mode. To overcome problems due to ion suppression in the LC-MS method, recoveries of the surrogate standard and measurements of multiple dilutions of the extracts were performed. IBP, DCF, BZF, AHTN and HHCB were separated and analysed by GC-(ESI)MS in the multiple ion detection (MID) mode. Analytes were enriched by solid phase extraction on a RP-18 (3 g, 6 ml) column under acidic conditions. Prior to sample extraction a surrogate standard (meclofenamic acid and AHTN-D3) was added to the samples. After elution with ethyl acetate and dichloromethane a derivatization step with diazomethane was performed, followed by a cleaning up step with silica gel. The extracts were lled up to 1 ml after addition of 50 ml internal standard (decachlorbiphenyl). The results were corrected by the added surrogate standards, where it was possible. In the case where no appropriate surrogate standards were available the results were corrected by the recovery performed in the same serie. Mean recovery rates, the LOQ and the limits of detection (LOD) for the different investigated compounds are summarised in Table 2. 2.2. Membrane bioreactor The MBR is a pilot scale plant situated at the premises of a conventional WWTP (WWTP 1). From the grit chamber of the conventional plant the wastewater is pumped into a feed tank, from which the denitrication vessel is charged. In addition to the mechanic treatment of the full scale plant a 1 mm punched sieve is installed at

ARTICLE IN PRESS
4800 M. Clara et al. / Water Research 39 (2005) 47974807 Table 2 Investigated compounds, limits of detection (LOD) and quantication (LOQ) in ng l1 and mean recovery rates (%) Substance DCF IBP BZF CBZ IPM DZP ROX SMX AHTN HHCB BPA NP NP1EO NP2EO NP1EC NP2EC OP OP1EO OP2EO
a

Formula C14H11Cl2NO2 C13H18O2 C19H20ClNO4 C15H12N2O C18H24I3N3O8 C16H13ClN2O C41H76N2O15 C10H11N3O3S C18H26O C18H26O C15H16O2 C14H22O C17H28O2 C19H32O3 C17H26O3 C19H30O4 C15H24O C16H26O2 C18H30O3

MW 296.16 206.29 361.83 236.28 791.15 284.70 837.06 253.31 258.40 258.40 228.20 220.36 264.41 308.46 278.39 322 206.33 250.38 294

LOD (ng l1) 10 (50 ) 10 (50a) 10 (50a) 10 10 10 10 10 10 20 10 10 10 10 10 10 10 10 10

LOQ (ng l1) 20 (100 ) 20 (100a) 20 (100a) 20 20 20 20 20 20 40 20 20 20 20 20 20 20 20 20

Recovery (%) 98 97 99 75 98 88 65 84 83 88 80 79 72 66 90 90 82 89 84

Inow samples.

Table 3 Characterisation of the investigated MBR pilot plant and the sampled conventional activated sludge WWTPs T (1C) MBR 1st sampling 2nd sampling 3rd sampling WWTP1 1st sampling 2nd sampling 3rd sampling WWTP 2 WWTP 3 SRT (d) HRT (d) TSS (g l1) VSS TSS1 () F/M (g COD g TSS1 d1)

22.2 27.2 5.5 16.8 22.1 6.8 13.5 10.4

10 27 55 114 237 52 2 46

0.5 1.2 4.0 13.3 12.5 13.6 0.08 1.20

6.3 4.5 11.8 4.9 4.0 4.0 4.0 3.1

0.75 0.71 0.79 0.65 0.60 0.68 0.80 0.65

0.24 0.16 0.03 0.02 0.02 0.02 1.70 0.04

the extraction point, reducing the amount of particulate matter and bres entering the plant. From the nitrication tank the sludge is pumped to the external situated cross ow membrane module, where an ultraltration membrane is installed. The sludge is recycled to the nitrication vessel and the permeate is stored in a permeate tank. The membrane module is regularly backwashed with treated wastewater from the permeate tank. An internal sludge recycle connects nitrication and denitrication to avoid biomass accumulation in the nitrication tank and to obtain further nitrogen removal. The excess sludge is removed from this internal sludge recycle. Both nitrication and denitrication have

a volume of 2.5 m3. Whereas the nitrication tank is continuously aerated, the denitrication tank is mixed by an electric stirrer. No phosphorous precipitation occurred. The MBR was run with different operational settings. Different SRTs were installed during three investigation periods. Characteristic data on the MBR are summarised in Table 3. 2.3. Conventional wastewater treatment The conventional activated sludge WWTP, where the MBR pilot plant is installed, serves a rural community

ARTICLE IN PRESS
M. Clara et al. / Water Research 39 (2005) 47974807 4801

in the South-East of Austria and is designed for 7000 population equivalents (pe). The design capacity is based on 100 g chemical oxygen demand (COD) pe1 day1. Actually approx. 3000 pe are connected to the sewer system. The treatment plant is charged with domestic sewage from a separated sewer system without industrial inuents and with strong seasonal uctuations due to local viniculture. These strong uctuations are illustrated by means of the COD in the inuent. While the mean COD concentration in the inuent in 2002 was 1080 mg l1, the maximum 2 weeks mean was 1900 mg l1 and peak values of more than 6000 mg l1 COD were measured in daily composite inow samples. The sewage plant is operated with simultaneous sludge stabilisation and simultaneous phosphorus precipitation with ferric chloride (FeCl3). Nitrogen removal is achieved by simultaneous and intermittent nitrication and denitrication. The WWTP consists of screen and grit chamber, two aeration tanks (1546 m3 each) and two secondary clarier for nal sedimentation. The excess sludge is removed very infrequently resulting in high SRTs. This conventional WWTP was sampled in parallel to the MBR. In order to compare the results observed for the different SRTs in the MBR, two CASP operating different SRTs were sampled. Twenty four hours composite samples were taken by automatic sampling and preserved by cooling. WWTP2 is a highly loaded plant, designed for carbon removal only. WWTP3 is a low-loaded plant designed for nutrient removal. Characteristic data as temperature (T), SRT, hydraulic retention time (HRT), total suspended solids concentrations (TSS), volatile suspended solids concentrations (VSS) and food to microorganism ratios (F/M) expressed in terms of COD are summarised in Table 3. 2.4. Mass balances A plausibility analysis of the measured values was made, using mass balances for phosphorus, nitrogen and COD according to Nowak et al. (1999). Based on that data, an evaluation of process characteristics (e.g. SRT, specic excess sludge production, etc.) was performed. The most important removal pathways an organic compound is subjected during wastewater treatment are stripping by aeration (volatilisation), adsorption to the sludge (excess sludge removal) and biotransformation/biodegradation. Considering octanolwater partitioning (log POW) and Henry coefcients (KH) of the investigated substances and the aeration systems in the different WWTPs, volatilisation is negligible for most of the selected substances (Struijs et al., 1991; Rogers, 1996). Only in WWTP2 operating supercial aeration systems for the two musk fragrances AHTN and HHCB volatilisation may be of importance.

Daily (24-h) composite samples (time-weighted) of inuent and efuent were taken by automated samplers preserved by cooling and ltered by uted lters. The samples were transported to the analysing institute and the dissolved concentrations were measured. The calculations of the mass balances are based on those measured dissolved concentrations. The absorbed fractions were considered by the application of equilibrium solid/liquid partition coefcients KD. From the investigated substances the EDCs and the musk fragrances are relevant for adsorption and have KD-values higher than 1000 l kg1, whereas for the investigated pharmaceuticals (KD-values below 100 l kg1) adsorption processes are negligible (Ahel et al., 1994; Clara et al., 2004b; Ternes et al., 2004). Removal from the liquid phase corresponds to the difference between inuent and efuent mass uxes. Biodegradation/biotransformation is the removal minus the mass ux removed from the system with the excess sludge. A mathematical description of the mass balances is summarised in Clara et al. (2005).

3. Results and discussion 3.1. Pharmaceuticals and personal care products (PPCPs) Mean dissolved concentrations of the PPCPs in inuent and efuent of the investigated WWTPs are summarised in Table 4. The tranquilizer diazepam was not detected in any of the analysed samples. The absence of this compound is explained by the low consumption in Austria, amounting to 125 kg year1 in 1997 (Sattelberger, 1999). The contrast media iopromide was not present in the inuent of WWTP1 and the MBR, what is explained with the absence of a hospital in the drainage area. In WWTP2 high concentrations of IPM were measured. Mean measured inuent concentrations amounted to 3840 ng l1, and mean measured efuent concentrations to 5060 ng l1. No reasonable justication for the increasing of IPM concentrations within the WWTP could be identied. The antibiotic SMX was only found during the 1st sampling in May. ROX was detected during all sampling campaigns, but notably higher concentrations were detected during the 3rd sampling in December compared to the samplings in May and July. Considering the relatively constant inuent ow rates during the three sampling campaigns (240726, 255725, 235725 m3 d1) this fact indicates a seasonal variation of antibiotics in the inuent and may be related to higher consumptions during the cold period of the year. This statement remains an assumption as no data on consumption rates for the relevant period are available. Comparable results

ARTICLE IN PRESS
4802 M. Clara et al. / Water Research 39 (2005) 47974807 Table 4 Mean dissolved concentrations (ng l1) of PPCPs in inow and efuent of the investigated WWTPs DCF 1st sampling Inow WWTP1 efuent MBR efuent 2nd sampling Inow WWTP1 efuent MBR efuent 3rd sampling Inow WWTP1 efuent MBR efuent WWTP2 Inow Efuent WWTP3 Inow Efuent IBP BZF CBZ IPM DZP ROX SMX AHTN HHCB

3250 1536 3464 4114 1533 2033 3190 1680 2140 1400 1300 905 780

1480 n.d. 22 2679 n.d. 22 2448 20 69 2300 2400 1200 24

1960 n.d. 103 2014 n.d. 73 6840 692 1550 7600 4800 1550 715

1850 1594 1619 1200 1337 1147 704 952 794 670 690 325 465

n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. 3840 5060 26 250

n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d. n.d.

26 41 n.d. 64 36 42 117 69 31 78 57 25 45

145 50 56 n.d. 18 n.d. n.d. n.d. n.d. 24 91 75 51

1106 145 163 989 170 92 1046 144 148 450 160 210 170

3060 451 455 4443 600 373 3360 652 536 1400 870 830 535

were obtained by McArdell et al. (2003) reporting two times higher inow loads during wintertime than during summer. Considering the low tendency of the antibiotics to accumulate onto sludge, removal rates between 50% and 60% were achieved. But especially in the case of antibiotics it has to be annotated, that metabolites should be included in the mass balances as a signicant amount enters WWTPs in metabolised form, e.g. SMX as N4-Acetyl-SMX (Gobel et al., 2004). The other PPCPs were detected in all investigated plants during all sampling campaigns. Except in WWTP2, lower efuent than inuent concentrations were measured for BZF, IBP, AHTN and HHCB indicating a notable removal potential during wastewater treatment (see Fig. 1). For these substances comparable efuent concentrations and removal rates were observed in the conventional activated sludge plant and the MBR. In WWTP2 no or only slight removal rates were achieved. Those low removal rates are attributed to the low SRT installed in the plant (Clara et al., 2005). The antiepileptic carbamazepine is not removed during wastewater treatment. In all investigated treatment facilities efuent concentrations in the range of the respective inuent concentrations were measured (Clara et al., 2004a). Contradictory results were observed for DCF. Whereas in WWTP1 removal rates of up to 70% were calculated, no removal was observed in the other conventional WWTPs. In the MBR no removal

occurred during the rst sampling campaign, when the plant was operated with a SRT of approx. 10 days. Dissolved efuent concentrations in the range of inuent concentrations were measured. From this observation can be concluded, that the membrane does not detain the substance by size exclusion. With increasing SRT also in the MBR a partial removal of DCF is observed. No reasonable explanation for these ndings could be found. The authors were not able to identify why only in WWTP1 remarkable removal of DCF occurred whereas in the other sampled WWTPs no similar behaviour was observed. But also in literature similar contradictory observations are reported. Whereas Zwiener and Frimmel (2003) and Heberer (2002a) report no signicant elimination of diclofenac, Ternes (1998) documented elimination rates of up to 70%. 3.2. Endocrine disrupting compounds (EDCs) Mean measured inuent and efuent concentrations of the EDCs are summarised in Table 5. BPA is nearly completely removed (495%) in WWTP1 as well as in the MBR and comparable dissolved efuent concentrations were measured. Considering the low efuent concentrations, biodegradation/biotransformation processes are the main removal pathway for BPA. In WWTP2 no removal was observed, indicating a dependency of degradation on the SRT as WWTP2 is a highly loaded plant, operating at SRTs between 1 and 2 days.

ARTICLE IN PRESS
M. Clara et al. / Water Research 39 (2005) 47974807 4803

Removal
120 100

Sorption to sludge

Biodegradation/biotransformation
120

IBP
relative amount [%]

BZF
100 80 60 40 20 0 -20

relative amount [%]

80 60 40 20 0 -20

2 TP W W 2 TP W W

1III

1III

1II

1II

1I

1I

II

III

II

BR

BR

BR

III

TP

TP

BR

BR

TP

TP

TP

TP

TP

TP

BR

120 100

AHTN
relative amount [%]

120 100 80 60 40 20 0 -20

relative amount [%]

80 60 40 20 0 -20

1III

1III

1II

1II

1I

II

III

1I

II

BR

BR

BR

III

TP

BR

TP

TP

TP

TP

TP

TP

BR

BR

Fig. 1. Removal of ibuprofen, bezabrate, tonalide and galaxolide by adsorption to the sludge and biodegradation/biotransformation in the investigated wastewater treatment facilities ( removal, sorption, & biodegradation/biotransformation).

Nonylphenol, nonylphenol ethoxylates and the nonylphenol carboxylates cannot be treated substance by substance. Due to the interactions between the different fractions an integrated evaluation of the behaviour of those compounds during wastewater treatment has to be performed. Biodegradation of nonylphenol polyethoxylates (NPnEOs) in wastewater treatment plants or in the environment is generally believed to start with a shortening of the ethoxylates (EO) chain, generating more persistent shorterchain NPEOs and NP (Ying et al., 2002). Partial NPnEO degradation seems to occur already in the sewer system as NP1EO is the dominant fraction in the inuent of the sampled WWTPs. Further transformation proceeds via oxidation of the EO chain, producing mainly nonylphenoxy ethoxy acetic acid (NP2EC) and nonylphenoxy acetic acid (NP1EC) (Ying et al., 2002). According to Brunner et al. (1988) NPnEOs are degraded more easily under aerobic, than under anaerobic, conditions, whereas a complete deethoxyla-

tion with formation of NP was observed under anaerobic conditions only (Ahel et al., 1994). As described in literature (Ying et al., 2002; Ahel et al., 1994), a change in the distribution of NP, NPEOs and NPECs was observed between inow and efuent. Whereas in the inuent NP and NP1EO were the main fractions, in the efuent NP1EC and NP2EC predominated (see Fig. 2). The following numbers (a b) indicate the proportionate fractions of the sum of NP, NPEO and NPEC and the respective standard deviations (n 19). In WWTP1 and the MBR NP contributed to approx. 32714% and NP1EO to 51713% to the inuent, whereas NP2EO (775%), NP1EC (674%) and NP2EC (372%) are of minor importance. In the efuent the distribution shifts to nonylphenol carboxylates. In the efuent of WWTP1 (MBR) the relative fractions amount to 1977% (1678%) for NP, to 373% (472%) for NP1EO, to 272% (271%) for NP2EO, to 5379% (58715%) for NP1EC and to

TP

TP

TP

HHCB

ARTICLE IN PRESS
4804 M. Clara et al. / Water Research 39 (2005) 47974807 Table 5 Mean dissolved concentrations (ng l1) of EDCs in inow and efuent of the investigated WWTPs BPA 1st sampling Inow WWTP1 efuent MBR efuent 2nd sampling Inow WWTP1 efuent MBR efuent 3rd sampling Inow WWTP1 efuent MBR efuent WWTP2 Inow Efuent WWTP3 Inow Efuent NP NP1EO NP2EO NP1EC NP2EC OP OP1EO OP2EO

2025 26 28 2376 35 16 2151 76 158 1710 1530 720 125

4031 487 371 2673 280 297 3129 326 482 1950 370 1280 285

7116 84 192 7299 n.d. 38 4450 64 93 4060 2580 6905 141

866 69 133 767 n.d. 44 835 78 60 600 1360 4645 78

724 905 2039 737 1117 4900 429 1544 1164 490 4730 2325 5235

362 196 589 107 290 1047 471 1648 874 840 8530 3635 7760

118 29 65 436 n.d. n.d. 215 15 74 680 90 146 106

213 20 19 552 n.d. n.d. 42 n.d. n.d. 660 470 281 n.d.

36 n.d. 15 55 n.d. n.d. n.d. n.d. n.d. n.d. 150 114 n.d.

120

120

cumulative frequency [%]

100 80 60 40 20 0

cumulative frequency [%]

WWTP 1

MBR

100 80 60 40 20 0

NP
200

NP1EO NP2EO NP1EC NP2EC

NP

NP1EO NP2EO NP1EC NP2EC

cumulative frequency [%]

cumulative frequency [%]

180 120 100 80 60 40 20 0

WWTP 2

WWTP 3
100 80 60 40 20 0

NP

NP1EO NP2EO NP1EC NP2EC

NP

NP1EO NP2EO NP1EC NP2EC

Fig. 2. Cumulative frequencies of nonylphenol and nonylphenolethoxylates in inow and efuent ( inow, & efuent) of the investigated treatment plants, with 100% equal to 44.5 nmol l1 (WWTP1), 44.5 nmol l1 (MBR), 30.5 nmol l1 (WWTP2) and 66.6 nmol l1 (WWTP3).

22714% (2078%) for NP2EC and no differences between the conventional activated sludge plant (WWTP1) and the MBR were detected. Also in WWTP2 and WWTP3 this shifting in the distribution between NP, NPEOs and NPECs can be observed.

Different removal rates were calculated for the alkylphenolic compounds in the different investigated WWTPs. The results of the mass balances are summarised in Fig. 3. The highest removal rates were observed in WWTP1, achieving more than 90%. The

ARTICLE IN PRESS
M. Clara et al. / Water Research 39 (2005) 47974807
Removal
120 100

4805

Sorption to sludge

Biodegradation/biotransformation
120 100

NP+NPEO+NPEC

OP+OPEO

relative amount [%]

80 60 40 20 0 -20
I M BR II M BR III W W TP 2 W W TP 3 1II 1I III TP BR 1TP TP M W

relative amount [%]

80 60 40 20 0 -20
1III M BR I M BR II M BR III W W TP 2 W W TP 3 1I TP W W W W TP W TP W 1II

W W

Fig. 3. Removal of alkylphenols and alkylphenol ethoxylates by adsorption to the sludge and biodegradation/biotransformation in the investigated wastewater treatment facilities ( removal, sorption, & biodegradation/biotransformation).

MBR showed comparable results. Signicantly lower removal of alkylphenolic compounds occurred in WWTP3 and no removal was observed in the high loaded WWTP2. A complete degradation of alkylphenol ethoxylates requires an anaerobic treatment step in order to obtain a de-ethoxylation (Ahel et al., 1994). WWTP2 is continuously aerated and in the efuent aerobic degradation metabolites as NP1EO and NP1EC are accumulated. WWTP3 operates an anaerobic cascade for biological phosphorous removal, but as in WWTP2 also in WWTP3 an accumulation of nonylphenol carboxylates is observed. Due to the anaerobic cascade and the low loading of the plant higher removal rates would be expected to occur. No reason for those comparably low removal rates could be found. WWTP1 is a very low-loaded plant achieving very high SRTs. Mean daily aeration times amount to approx. 5 h and especially during summer the formation of anaerobic zones in the biological reactor is realistic. Due to this and the high retention times a nearly complete removal occurs. Comparable results are reported by Ahel et al. (1994). The authors observed the highest removal rates for NPEOs in low-loaded conventional municipal WWTPs. In the MBR anaerobic zones are expected to appear in the denitrication tank due to high organic loading. In the MBR the sludge recirculation is important in order to avoid sludge concentration in one compartment of the plant. Oxygen transfer from the aerated compartments to the denitrication due to this sludge recirculation may inuence the removal of alkylphenolic compounds and be the reason for the slightly lower removals in the MBR compared to WWTP1. Summarising it can be stated that only negligible differences in the removal potential for the investigated substances could be detected between conventional

WWTPs and the sampled MBR. This observation leads to the conclusion, that the operated ultraltration membrane does not allow any further retention of the investigated substances and degradation is dependent on the operated SRT (Clara et al., 2005). Comparable results for BPA and NP are reported also by Ivashechkin et al. (2004) and comparable observations for pharmaceuticals are reported by Joss et al. (2005). Considering the molecular weight of the investigated compounds (see Table 2), an additional detention due to pure ltration effects by the applied ultraltration membrane based on size exclusion is not to be expected (Stephenson et al., 2000).

4. Conclusions Comparable efuent concentrations and removal rates for the different investigated substances were observed in the conventional WWTP and the membrane bioreactor. Some substances are not removed during wastewater treatment (e.g. carbamazepine). They pass the investigated plants without any reduction and efuent concentrations in the range of inuent concentrations were measured. Other substances as BPA, IBP, BZF, etc are degraded during wastewater treatment, achieving notable removal rates of more than 90% and resulting in low-efuent concentrations. For these compounds only slight differences in the efuent concentrations could be detected between conventional activated sludge wastewater treatment and membrane bioreactors. This observation leads to the conclusion, that the operated ultraltration membrane does not allow any further retention of the investigated substances due to size exclusion. Dense membranes as usually applied in reverse osmosis and/or nanoltration

ARTICLE IN PRESS
4806 M. Clara et al. / Water Research 39 (2005) 47974807 Clara, M., Kreuzinger, N., Strenn, B., Gans, O., Kroiss, H., 2005. The solids retention timea suitable design parameter to evaluate the capacity of wastewater treatment plants to remove micropollutants. Water Res. 39 (1), 97106. DIN 32645, 1994. Chemische Analytik; Nachweis-, Erfassungsund Bestimmungsgrenze; Ermittlung unter Wiederholbedingungen; Begriffe, Verfahren, Auswertung. Beuth Verlag GmbH, Berlin-Wien-Zurich. Fromme, H., Kuchler, T., Otto, T., Pilz, K., Muller, J., Wenzel, A., 2002. Occurrence of phthalates and bisphenol A and F in the environment. Water Res. 36 (6), 14291438. Gobel, A., McArdell, C.S., Suter, M.J.F., Giger, W., 2004. Trace determination of macrolide and sulfonamide antimicrobials, a human sulfonamide metabolite, and trimethoprim in wastewater using liquid chromatography coupled to electrospray tandem mass spectrometry. Anal. Chem. 76 (16), 47564764. Heberer, T., 2002a. Occurrence, fate and removal of pharmaceutical residues in the aquatic environment: a review of recent research data. Toxicol. Lett. 131 (12), 517. Heberer, T., 2002b. Tracking persistent pharmaceutical residues from municipal sewage to drinking water. J. Hydrol. 266 (23), 175189. Ivashechkin, P., Corvini, P., Fahrbach, M., Hollender, J., Konietzko, M., Meesters, R., Schroder, H.F., Dohmann, M., 2004. Comparison of the elimination of endocrine disrupters in conventional wastewater treatment plants and membrane bioreactors. In: Conference Proceedings of the Second IWA Leading-Edge Conference on Water and Wastewater Treatment Technologies, 14 June 2004, Prague, Czech Republic. Joss, A., Keller, E., Alder, A.C., Gobel, A., McArdell, C.S., Ternes, T., Siegrist, H.R., 2005. Removal of pharmaceuticals and fragrances in biological wastewater treatment. Water Res. 39 (14), 31393152. Mart nez, E., Gans, O., Weber, H., Scharf, S., 2004. Analysis of nonylphenol polyethoxylates and their metabolites in water samples by high-performance liquid chromatography with electrospray mass spectrometry detection. Water Sci. Technol. 50 (5), 157163. McArdell, C.S., Molnar, E., Suter, M.J.F., Giger, W., 2003. Occurrence and fate of macrolide antibiotics in wastewater treatment plants and in the Glatt Valley watershed, Switzerland. Environ. Sci. Technol. 37 (24), 54795486. Nowak, O., Franz, A., Svardal, K., Muller, V., Kuhn, V., 1999. Parameter estimation for activated sludge models with the help of mass balances. Water Sci. Technol. 39 (4), 113120. Rogers, H.R., 1996. Sources, behaviour and fate of organic contaminants during sewage treatment and in sewage sludges. Sci. Total Environ. 185 (13), 326. Sattelberger, R., 1999. Arzneimittelruckstande in der Umwelt Bestandsaufnahme und Problemdarstellung. Reports R162, Umweltbundesamt GmbH. Vienna, Austria, ISBN 385457-510-6. Skutan, S., 2003. Risikomanagement: Modulteil Stoffstromanalysen. In: Umweltbundesamt GmbH (Ed.), Hormonwirk same Stoffe in Osterreichs GewassernEin Risiko? Final Report, July 2003, Vienna, Austria, IV-14IV-45, ISBN 3-85457-695-1. Available from http://www.arcem.at/ endbericht.pdf

processes would be required, to detain the substances investigated in the present study; this would add further to the increased power consumptions of MBRs. Nevertheless the MBR offers advantages compared to conventional systems. The membrane allows the detention of particulate matter leading to an efuent free of suspended solids. Slightly lower total emissions can be achieved in MBRs compared to conventional WWTPs for strongly adsorbing substances. MBRs achieve high SRTs within compact reactor volumes and as degradation is a function of the operated SRT, this fact represents another advantage of MBRs in comparison to conventional systems. Especially in regions with no suitable receiving waters or where a reuse of the treated wastewater is planned, MBRs represent an attractive solution due to the mentioned advantages.

Acknowledgments The results presented in this work were obtained within the EU-funded project POSEIDON (EVK1-CT2000-00047), the ARCEM project (Austrian Research Cooperation on Endocrine Modulators) funded by the Austrian Federal Ministry of Agriculture, Forestry, Environment and Water Management, Kommunalkredit Austria AG and the nine Austrian Federal States. The results for the membrane bioreactor were derived from the VALIUM project (Verhalten von bestimmten ArzneimitteLruckstanden, Industrie- und Umweltche mikalien in Membranbioreaktoren), funded by the Government of Burgenland, Department Gewasserauf sicht. The authors acknowledge the VA-TECH WABAG for installation and operation of the membrane pilot plant.

References
Ahel, M., Giger, W., Koch, M., 1994. Behaviour of alkylphenol polyethoxylate surfactants in the aquatic environmentI. Occurrence and transformation in sewage treatment. Water Res. 28 (5), 11311142. Brunner, P.H., Capri, S., Marcomini, A., Giger, W., 1988. Occurrence and behaviour of linear alkylbenenesulphonates, nonylphenol, nonylphenol mono- and nonylphenoldiethoxylats in sewage and sewage sludge treatment. Water Res. 22 (12), 14651472. Clara, M., Strenn, B., Kreuzinger, N., 2004a. Carbamazepine as a possible anthropogenic marker in the aquatic environment: investigations on the behaviour of Carbamazepine in wastewater treatment and during groundwater inltration. Water Res. 38 (4), 947954. Clara, M., Strenn, B., Saracevic, E., Kreuzinger, N., 2004b. Adsorption of bisphenol-A, 17b-estradiole and 17a-ethinylestradiole to sewage sludge. Chemosphere 56 (9), 843851.

ARTICLE IN PRESS
M. Clara et al. / Water Research 39 (2005) 47974807 Sonnenschein, C., Soto, A.M., 1998. An updated review of environmental estrogen and androgen mimics and antagonists. J. Steroid Biochem. 65 (16), 143150. Stephenson, T., Judd, S., Jefferson, B., Brindle, K., 2000. Membrane Bioreactors for Wastewater Treatment. IWA Publishing, London, UK. Struijs, J., Stoltenkamp, J., Meent, D.V.D., 1991. A spreadsheet-based box model to predict the fate of xenobiotics in a municipal treatment plant. Water Res. 25 (7), 891900. Sumpter, J.P., 1998. Xenoendocrine disruptersenvironmental impacts. Toxicol. Lett. 102103, 337342. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Res. 32 (11), 32453260. 4807 Ternes, T., Hermann, N., Bonerz, M., Knacker, T., Siegrist, H.R., Joss, A., 2004. A rapid method to measure the solidwater distribution coefcient (Kd) for pharmaceuticals and musk fragrances in sewage sludge. Water Res. 38 (19), 40754084. Ying, G.-G., Williams, B., Kookana, R., 2002. Environmental fate of alkylphenols and alkylphenol ethoxylatesa review. Environ. Int. 28, 215226. Zwiener, C., Frimmel, F.H., 2003. Short-term tests with a pilot sewage plant and biolm reactors for the biological degradation of the pharmaceutical compounds clobric acid, ibuprofen, and diclofenac. Sci. Total Environ. 309 (13), 201211.