Ozone Oxidation of Endocrine Disruptors and Pharmaceuticals in Surface Water and Wastewater

Transcription

1 Ozone: Science and Engineering, 28: Copyright # 2006 International Ozone Association ISSN: print / online DOI: / Ozone Oxidation of Endocrine Disruptors and Pharmaceuticals in Surface Water and Wastewater Shane A. Snyder, 1 Eric C. Wert, 1 David J. Rexing, 1 Ronald E. Zegers, 1 and Douglas D. Drury 2 1 Southern Nevada Water Authority, Henderson, Nevada, USA 2 Clark County Water Reclamation District, Las Vegas, Nevada, USA The oxidative removal of a diverse group of trace organic contaminants from surface water and wastewater was evaluated using ozone (O 3 ) and O 3 combined with hydrogen peroxide (O 3 /H 2 O 2 ). Target compounds included estrogenic and androgenic steroids, pharmaceuticals, pesticides, and industrial chemicals. Bench- and pilot- scale experiments were conducted with surface water spiked with the target compounds and wastewater effluent containing ambient concentrations of target compounds. Full-scale water treatment plants were sampled before and after ozonation to determine if bench- and pilot-scale results accurately predict full-scale removal. In both drinking water and wastewater experiments, the majority of target compounds were removed by greater than 90% at O 3 exposures commonly used for disinfection. Atrazine, iopromide, meprobamate, and tris-chloroethylphosphate (TCEP) were the most recalcitrant compounds to oxidize using O 3, with removals generally less than 50%. The addition of H 2 O 2 for advanced oxidation was of little benefit for contaminant removal as compared to O 3 alone. O 3 /H 2 O 2 provided a marginal increase in the removal of dilantin, diazepam, DEET, iopromide, and meprobamate, while decreasing the removal efficacy of pentoxifylline, caffeine, testosterone, progesterone, and androstenedione. In wastewater experiments, O 3 and O 3 /H 2 O 2 wereshowntoremoveinvitro estrogenicity. Collectively, these data provide evidence that O 3 is a highly effective oxidant for removing the majority of trace organic contaminants from water. Keywords Ozone, Endocrine Disruptors, Pharmaceuticals, Surface Water, Wastewater, Advanced Oxidation Process, Hydrogen Peroxide, Drinking Water, Wastewater, Water Reuse INTRODUCTION A diversity of organic contaminants has been shown to occur at ng L 1 (10 9 gl 1 ) concentrations in wastewater treatment plant (WWTP) effluents globally. Of particular Address correspondence to Shane A. Snyder, SNWA, 1350 Richard Bunker Ave., Henderson, NV 89015, USA. shane.snyder@snwa.com interest are trace pharmaceuticals and steroids that are not completely eliminated through conventional wastewater treatment and subsequently are released into the aquatic environment. While the earliest reports documenting trace pharmaceuticals and steroids were published in the 1960s and 1970s (Stumm-Zollinger and Fair, 1965; Tabak and Bunch, 1970; Garrison et al., 1975; Hignite and Azarnoff, 1977), more recent studies have established causality between the occurrence of trace steroids in WWTP effluents and reproductive impacts to aquatic wildlife (Desbrow et al., 1998; Jobling et al., 1998; Kramer et al., 1998; Snyder et al., 2001b; Folmar et al., 2002). Steroids are but one group of emerging water contaminants known as endocrine disrupting chemicals (EDCs), which are compounds that can mimic or block the action of endogenous hormones (Snyder et al., 2003). Several reports have been published that demonstrate the ubiquitous occurrence of pharmaceuticals and EDCs when analytical detection limits of ng L 1 or less are applied (Belfroid et al., 1999; Snyder et al., 1999; Ternes et al., 1999; Huang and Sedlak, 2001; Kolpin et al., 2002; Vanderford et al., 2003). Ozone (O 3 ) has been shown to be an effective disinfectant and powerful oxidizer (Hoigné and Bader, 1983a, 1983b; Haag and Yao, 1992; Rakness et al., 1993; Acero et al., 2000; Janex et al., 2000). O 3 reacts with organic contaminants through either the direct reaction with molecular O 3 or through the formation of free radicals, including the hydroxyl radical (OH). Molecular O 3 is a selective electrophile that reacts quickly with amines, phenols, and double bonds in aliphatic compounds. The OH reacts less selectively than molecular O 3 and with faster reaction rates. In most water treatment applications, the concentration of OH is generally low (in the order of M). Advanced oxidation processes (AOPs) using O 3 in combination with hydrogen peroxide (O 3 / H 2 O 2 ) can increase OH concentration for removal of more recalcitrant compounds (Hoigne, 1998; Acero et al., 2000; Acero and Von Gunten, 2001). The use of Ozone Oxidation of Endocrine Disruptors December

2 O 3 /H 2 O 2 for emerging contaminant removal has been shown previously (Acero et al., 2000; Zwiener and Frimmel, 2000; Huber et al., 2003; Ternes et al., 2003; Westerhoff et al., 2005). The oxidation of various EDCs and pharmaceuticals in water using O 3 also has been demonstrated (Zwiener and Frimmel, 2000; Adams et al., 2002, 2005; Huber et al., 2003; Ternes et al., 2003; Deborde et al., 2005; Kamiya et al., 2005; McDowell et al., 2005; Westerhoff et al., 2005). Huber et al. (2005) pilot tested O 3 for oxidation of 11 pharmaceuticals spiked into municipal wastewater effluents. This report showed that 10 of the pharmaceuticals were readily oxidized at O 3 doses >2 mg L 1 ; however, the X-ray contrast agent iopromide was not well oxidized at an O 3 dose of 5 mg L 1 (Huber et al., 2005). Ternes et al. (2002) studied the removal of 5 pharmaceuticals using O 3 at both laboratory and full-scale drinking water treatment conditions (Ternes et al., 2002). Only 3 of the 5 pharmaceuticals were detected at the full-scale drinking water treatment plant investigated, and all 3 were well removed by O 3. Ternes et al. (2003) also reported on the efficacy of O 3 and O 3 /H 2 O 2 at pilot scale for the removal of several pharmaceuticals, one steroid, and two musk fragrances. This investigation used O 3 doses of 5, 10, and 15 mg L 1 and a single O 3 / H 2 O 2 condition of 10 mg L 1 O 3 with 10 mg L 1 H 2 O 2. Once again, iodinated contrast media were found to be the most refractory contaminants investigated with less than 50% removal at an O 3 dose of 5 mg L 1. In this case, O 3 /H 2 O 2 was found to provide only a meager increase in oxidation as compared to O 3 alone. In the current study, O 3 was evaluated at bench-, pilot-, and full-scale in both surface water and wastewater for the oxidation of 36 structurally diverse contaminants. All bench- and pilot-scale experiments were conducted with Colorado River water (CRW) collected from drinking water intakes in Lake Mead, Nevada USA. O 3 doses from 1 3 mg L 1 were evaluated with and without H 2 O 2 addition. Tertiary wastewater (prior to disinfection) was collected from the Clark County Water Reclamation District (CCWRD) treatment plant located in Clark County, Nevada, USA, which discharges to Lake Mead. CCWRD wastewater was used to evaluate O 3 and O 3 / H 2 O 2 at pilot scale for the removal of emerging contaminants. Wastewater O 3 doses from 2 9 mg L 1 were evaluated. Full-scale O 3 oxidation was evaluated at 4 drinking water plants and 1 wastewater treatment plant. MATERIALS AND METHODS Analytical and Bioassay Methods Methods used to identify and quantify target compounds have been described previously (Vanderford et al., 2003; Trenholm et al., 2006; Vanderford and Snyder, 2006). Briefly, 1-Liter water samples were preserved using 1 gram of sodium azide to prevent microbial degradation and to quench any dissolved O 3 residual. Samples were extracted using automated solid-phase extraction with 500 mg hydrophilic-lipophilic balance (HLB) cartridges from Waters (Milford, MA). The resulting extract was analyzed by both gas chromatography with tandem mass spectrometry (GC-MS/MS) and liquid chromatography with tandem mass spectrometry (LC- MS/MS). Method reporting limits (MRLs) ranged from approximately 1 to 10 ng L 1. Table 1 provides the list of target compounds along with the analytical technique, MRL, and molecular weight (MW). Log K ow,pka,and water solubility for these compounds has been provided previously (Snyder et al., 2006). A human breast carcinoma in vitro bioassay was used to measure estrogenicity in O 3 and O 3 /H 2 O 2 wastewater experiments. The ability of cellular bioassays in the screening of wastewater effluents has been shown previously (Zacharewski, 1997; Desbrow et al., 1998; Snyder et al., 2001; Onda et al., 2002; Drewes et al., 2005). Water samples (1-L) were extracted similarly to samples for instrumental analyses, except surrogate compounds were not added. The cellular bioassay method employed measures cell proliferation and has been described in detail previously (Drewes et al., 2005). Bioassay data are reported as estradiol equivalents (EEq s). Bench-Scale Experiments Methods used for bench-scale O 3 testing were described previously (Westerhoff et al., 2005). Benchscale testing was conducted using prefiltered (0.7 mm) CRW spiked with target compounds to achieve nominal concentrations between 100 and 300 ng L 1. Average water quality data for CRW collected from the drinking water intake at Lake Mead are shown in Table 2. O 3 demand and decay tests were performed to determine the O 3 dose required to achieve a dissolved O 3 residual of 0.2 to 0.3 mg L 1 after 3 min contact time and zero residual within 10 min. This O 3 residual and contact time yield a USEPA regulatory-based concentration-time (CT) value of approximately 0.8 min-mg L 1, which is appropriate for primary disinfection of Giardia and virus inactivation. Dissolved O 3 concentrations were measured using the indigo method (Standard Methods 4500-O 3 ). The impact of OH promotion was tested by adding H 2 O 2 at mg per mg O 3. Pilot-Scale Experiments Pilot-scale O 3 testing was conducted using two dynamic pilot testing systems with flow rates of 1.0 L min 1 and 23 L min 1. A bench-top pilot plant (BTPP) with a flow rate of 1.0 L min 1 was used to conduct multiple O 3 and O 3 /H 2 O 2 contaminant oxidation experiments using both CRW and WWTP effluent collected post-filtration from CCWRD. BTPP also was used to evaluate O 3 and O 3 /H 2 O 2 disinfection and by-product 446 S. A. Snyder et al. December 2006

3 TABLE 1. Target Compounds, Analytical Method, and Physical Properties Compound Class Method MRL [ng L 1 ] MW Acetaminophen Pharmaceutical LC Androstenedione Steroid LC Atrazine Pesticide LC Benzo[a]pyrene PAH GC Caffeine PCP LC Carbamazepine Pharmaceutical LC DDT Pesticide GC DEET PCP LC Diazepam Pharmaceutical LC Diclofenac Pharmaceutical LC Dilantin Pharmaceutical LC Erythromycin Antimicrobial LC Estradiol Steroid LC Estriol Steroid LC Estrone Steroid LC Ethinylestradiol Steroid LC Fluorene PAH GC Fluoxetine Pharmaceutical LC Galaxolide Fragrance GC Gemfibrozil Pharmaceutical LC Hydrocodone Pharmaceutical LC Ibuprofen Pharmaceutical LC Iopromide Pharmaceutical LC Lindane Pesticide GC Meprobamate Pharmaceutical LC Metolachlor Pesticide GC Musk Ketone Fragrance GC Naproxen Pharmaceutical LC Oxybenzone PCP LC Pentoxifylline Pharmaceutical LC Progesterone Steroid LC Sulfamethoxazole Antimicrobial LC TCEP PCP LC Testosterone Steroid LC Triclosan Antimicrobial LC Trimethoprim Antimicrobial LC MRL = Method Reporting Limit; PCP = Personal Care Product; PAH = polycyclic aromatic hydrocarbon. formation using CCWRD (Wert et al., 2006). The larger pilot plant, with a flow rate of 23 L min 1, was used to evaluate O 3 oxidation using only CRW. Bench-Top Pilot Plant. The 1.0 L min 1 BTPP consisted of a continuous-flow O 3 contactor constructed using inert materials such as glass, fluorocarbon polymers, and stainless steel that do not leach or adsorb target compounds. A 208 L stainless steel drum was filled with 170 L of water and a peristaltic pump was used to control flow rate. Prior to entering the O 3 contactor, chemical feed ports allowed the injection of H 2 O 2 into the process stream followed by 2 static mixers. The O 3 contactor consisted of 12 glass chambers, each providing 2 min of contact time. Each glass chamber was equipped with Teflon sample ports. O 3 feed gas was produced from oxygen using a laboratory-scale O 3 generator (model LAB2B, Ozonia North America Inc., Elmwood Park, NJ, USA). O 3 was added in the first contactor chamber with counter-current flow through a fritted glass diffuser with a nominal bubble size of 0.1 mm. An O 3 feed gas flow rotameter and feed gas concentration analyzer were used to calculate the O 3 dosage. Off-gas was collected from the top of each cell into one central manifold and sent to an O 3 destruction unit containing manganese dioxide destruct catalyst. Transfer efficiency was calculated from the concentrations of O 3 feed gas, dissolved Ozone Oxidation of Endocrine Disruptors December

4 TABLE 2. Water Quality for Colorado River Water and Tertiary Effluent CRW Tertiary Effluent Parameter Units Ave 2005 June 2005 January 2006 Dissolved Oxygen mg L Total Organic Carbon mg L ph units Temperature C Total Alkalinity mg L Ammonia mg L 1 as N <0.08 <0.08 <0.08 Nitrate mg L 1 as N Bromide mg L Bromate mg L 1 <1.0 <1.0 <1.0 Assimilable Organic Carbon mg L 1 72 NM 330 Aldehydes mg L 1 7 NM 26 Carboxylic Acids mg L 1 40 NM 126 NM = Not Measured. O 3, and O 3 off-gas and ranged from 40 70% depending on water flow rate Transferred ozone dose was used in the evaluation of contaminant destruction. The BTPP testing was performed using two 170 L batches of CRW spiked with target compounds, with two experiments performed on each batch. O 3 dosages of 1.25 and 2.50 mg L 1 were selected to compare findings from the bench-scale study. During O 3 /H 2 O 2 experiments, H 2 O 2 was added approximately 0.5 min prior to the O 3 contactor. Initial O 3 /H 2 O 2 experiments were designed to model bench-scale conditions using a mg L 1 H 2 O 2 dose. A second set of O 3 /H 2 O 2 experiments were performed using a 0.2 H 2 O 2 to O 3 mass ratio. Once the operating conditions were established, the system was operated for 1 h to assure steady-state conditions were achieved before sampling. Samples were collected after 2, 6, 14, and 24 min of contact time to determine relative reaction rates. BTPP testing also was performed in June 2005 and January 2006 using 170 L batches of non-disinfected tertiary treated wastewater. Water was collected at CCWRD and immediately transported to the pilot facility (<1 hr transport time). Table 2 provides water quality data from CRW and CCWRD. Since wastewater entering Lake Mead was previously shown to contain endocrine disruptors and pharmaceuticals, (Snyder et al. 1999, 2001; Snyder, Villeneuve et al., 2001; Boyd and Furlong, 2002), spiking of the wastewater was not performed. Dissolved O 3 residual was measured at 2, 6, 10, 14, and 18 min contact times (Table 3). During January 2006 testing, TABLE 3. Operational Parameters for BTPP Experiments Water Source Date O 3 Dose [mg L 1 ] H 2 O 2 Dose [mg L 1 ] Dissolved Ozone Residual [mg L 1 ] 2 min 6 min 10 min 14 min 18 min CT [mg-min L 1 ] CRW Oct CRW Oct CRW Oct CRW Oct CRW Oct CRW Oct CCWRD Jun CCWRD Jun CCWRD Jun CCWRD Jan CCWRD Jan CCWRD Jan CCWRD Jan CCWRD Jan CCWRD Jan S. A. Snyder et al. December 2006

5 O 3 and O 3 /H 2 O 2 were evaluated using 2 batches of filtered tertiary wastewater. After contaminant removal testing was complete, OH production was investigated through duplicate O 3 and O 3 /H 2 O 2 experiments using the remaining 70 L of wastewater spiked with the probe compound para-chlorobenzoic acid (pcba). Due to the addition of pcba, duplicate experiments were required to avoid scavenging of OH during the contaminant testing, which would have distorted removal results. Pilot-Plant Experiments The 23 L min 1 pilot plant uses raw water from Lake Mead and includes ozonation, coagulation, flocculation, and filtration (Wert et al., 2005). A syringe pump was used to introduce the target compounds into the process stream. Two static mixers followed the contaminant spike to provide homogenization. The O 3 contactor consisted of 12 cells to provide approximately 24 min of contact at the design flow rate of 23 L min 1. Each contactor cell had sample ports to allow sampling after various contact times. Ambient air supplied the O 3 generator (model SGC21, Pacific Ozone Technology, Benicia, CA) to produce O 3 feed gas. The O 3 feed gas concentration was measured (Model HI-X, IN USA Inc., Needham, MA), controlled by a gas rotameter, and injected countercurrently through a porous stone diffuser mounted horizontally near the bottom of the first cell. Contactor cells 3, 5, and 9 were equipped with dissolved O 3 monitors used to calculate disinfection levels and other O 3 parameters such as half-life, CT, demand, and decay. Dissolved O 3 residual monitors were calibrated using the indigo method (Standard methods 4500-O 3 ). Off-gas from each contactor cell was collected into a central manifold and measured using an O 3 monitor. The feed-gas concentration, off-gas concentration, and dissolved O 3 measurements were all interfaced into computer software that calculates critical operating parameters, such as transferred O 3 dose and O 3 demand. O 3 demand was evaluated with and without the solvent carrier and contaminant spike. Full-Scale Evaluations Water samples were collected before and after O 3 disinfection at 4 drinking water facilities and 1 wastewater facility. While all target compounds were analyzed, only a limited number were detectable in raw drinking water. O 3 residual was quenched using ascorbic acid and samples express shipped to the laboratory for analysis (Trenholm et al., 2006). RESULTS AND DISCUSSION Bench-Scale Experiments Of the 36 target compounds evaluated, 22 were removed to below detection using an O 3 dose of 2.5 mg L 1. Removal of the remaining 14 target compounds is presented in Figure 1. Most target compounds investigated exhibited >50% removal except atrazine, (%) Removal Percent i e DDT Atraz n FIGURE 1. CRW. O3=2.5 mg/l O3=2.5 mg/l, H2O2= mg/l DEET Diazepam Dilantin Galaxolide Ibu ro p fen Iopromide Lindane Meprobamate Metolachlor Musk Ket e iopromide, lindane, musk ketone, and TCEP. H 2 O 2 addition ( mg L 1 ) at the 2.5 mg L 1 O 3 dose resulted in a 5 10% increase in target compound removal. Bench-Top Pilot Plant Results O 3 dose, H 2 O 2 dose, and dissolved O 3 residual information for both CRW and CCWRD effluent during BTPP experiments are provided in Table 3. Details on the experimental conditions and subsequent disinfection by-product (DBP) formation using CCWRD effluent have been shown previously (Wert et al., 2006). The wastewater effluent had nearly twice the total organic carbon (TOC) as CRW (Table 2), which contributes to the faster O 3 decay rate observed. Colorado River Experiments During BTPP O 3 experiments, 13 of the 36 target compounds were removed by greater than 90% within the first 2 min of contact time at an applied O 3 dose of 1.25 mg L 1 (Table 4). The addition of H 2 O 2 at 0.25 mg L 1 generally resulted in a minor increase in target compound removal. For certain compounds, (i.e., androstenedione, pentoxifylline, testosterone, progesterone, metolachlor, fluorine, benzo(a)pyrene, and caffeine) the overall removal was 15% less using O 3 /H 2 O 2 than with O 3 alone. Conversely, other target compounds (i.e., ibuprofen, dilantin, DEET, iopromide and meprobamate) showed a removal increase of 10% during O 3 /H 2 O 2 when compared to O 3 alone, indicating that removal was enhanced by the non-selective OH. For compounds with decreased overall removal, percent reduction was consistently greater at 2 min for O 3 /H 2 O 2 as compared to O 3 alone (Tables 4 and 5). Example reaction rate data for relatively slow-reacting compounds are plotted in Figures 2 and 3. This is the result of increased reaction kinetics with OH as compared to molecular O 3. As expected, increased O 3 doses resulted in superior compound removal; however, the percent removal is more on T E C P Triclosan Bench-scale removal of select target compounds in Ozone Oxidation of Endocrine Disruptors December

6 TABLE 4. BTPP Removal Using O 3 and O 3 /H 2 O 2 in CRW Low Dose O 3 Dose [mg L 1 ] H 2 O 2 Dose [mg L 1 Reaction Time [min] Spiked CRW Target Compound (ng L 1 )(n=9) Percent Removal Percent Removal Acetaminophen >99 > >98 Androstenedione Atrazine Benzo(a)pyrene <1 < Caffeine Carbamazepine 373 >99 >99 >99 >99 >99 >99 >99 >99 DDT <1 <1 24 <1 <1 <1 14 DEET Diazepam Diclofenac 313 >99 >99 >99 >99 >99 >99 >99 >99 Dilantin Erythromycin 12 >89 >89 >89 >89 >92 >92 >92 >92 Estradiol >99 >99 >99 >99 99 >99 >99 Estriol 361 >98 >98 >98 >98 >99 >99 >99 >99 Estrone > >99 Ethinylestradiol 348 >99 >99 >99 >99 >99 >99 >99 >99 Fluorene <1 < Fluoxetine >99 >99 >99 99 >99 99 >99 Galaxolide Gemfibrozil 357 >99 >99 >99 >99 >99 >99 >99 >99 Hydrocodone 303 >99 >99 >99 >99 >99 >99 >99 >99 Ibuprofen Iopromide Lindane 280 <1 <1 <1 <1 <1 <1 <1 <1 Meprobamate Metolachlor < Musk Ketone 260 <1 <1 <1 <1 1.6 < Naproxen 298 >99 >99 >99 >99 >99 >99 >99 >99 Oxybenzone 72 >99 >99 >99 >99 >98 >98 >98 >98 Pentoxifylline Progesterone Sulfamethoxazole >99 >99 >99 >99 >99 >99 >99 TCEP 341 <1 <1 <1 2 <1 <1 <1 9.3 Testosterone Triclosan Trimethoprim 312 >99 >99 >99 >99 >99 >99 >99 >99 dramatic for those compounds that are more resistant to oxidation (i.e., iopromide, meprobamate, atrazine). TCEP, lindane, and the synthetic fragrance musk ketone proved to be the most challenging compounds to oxidize using O 3, with percent removal generally less than 20%. Wastewater Effluent Experiments June Dissolved O 3 residual had decayed after 12 min of contact time at all O 3 doses (Table 3). Analysis showed that 17 of the 36 target compounds were detectable in the tertiary effluent (Table 6). Of the 17 contaminants detected, seven target compounds were removed to less than detection using the lowest O 3 dose evaluated (4.9 mg L 1 ). When compounds were removed to less than the MRL, percent removals are shown as greater than the value calculated from the MRL. The concentration of the fire-retardant TCEP was essentially unchanged at all O 3 doses. The synthetic fragrance musk ketone also was difficult to oxidize, with 38 and 68% concentration 450 S. A. Snyder et al. December 2006

7 TABLE 5. BTTP Removal using O 3 and O 3 /H 2 O 2 in CRW High Dose O 3 Dose [mg L 1 ] H 2 O 2 Dose [mg L 1 ] Reaction Time [min] Target Compound Spiked CRW (ng L 1 ) (n = 9) Percent Removal Percent Removal Acetaminophen 117 >99 >99 >99 >99 >99 >99 99 >99 Androstenedione > Atrazine Benzo(a)pyrene 113 NA NA NA 41 NA NA NA 24 Caffeine >97 >97 > Carbamazepine 373 >99 >99 >99 >99 >99 >99 >99 >99 DDT 83 NA NA NA 41 NA NA NA <1 DEET Diazepam Diclofenac 313 >99 >99 >99 >99 >99 >99 >99 >99 Dilantin Erythromycin 12 >92 >92 >92 >92 >92 >92 >92 >92 Estradiol 347 >99 >99 >99 >99 >99 >99 >99 >99 Estriol 361 >99 >99 >99 >99 >98 >98 >98 >98 Estrone >99 >99 >99 98 >99 >99 >99 Ethinylestradiol 348 >99 >99 >99 >99 >99 >99 >99 >99 Fluorene 349 NA NA NA 72 NA NA NA 82 Fluoxetine 208 >99 >99 >99 >99 99 >99 >99 >99 Galaxolide 121 NA NA NA 57 NA NA NA 52 Gemfibrozil 357 >99 >99 >99 >99 >99 >99 >99 >99 Hydrocodone 303 >99 >99 >99 >99 >99 >99 >99 >99 Ibuprofen Iopromide Lindane 280 NA NA NA 9 NA NA NA <1 Meprobamate Metolachlor 454 NA NA NA 68 NA NA NA 83 Musk Ketone 260 NA NA NA 17 NA NA NA 14 Naproxen 298 >99 >99 >99 >99 >99 >99 >99 >99 Oxybenzone 72 >99 >99 >99 >99 >98 >98 >98 >98 Pentoxifylline >99 >99 > Progesterone > Sulfamethoxazole 327 >99 >99 >99 >99 >99 >99 >99 >99 TCEP < <1 Testosterone > Triclosan Trimethoprim 312 >99 >99 >99 >99 >99 >99 >99 >99 NA = Not Analyzed. reduction at the low and high O 3 doses, respectively. Oxybenzone was detectable in post-o 3 samples at the low and high doses; however, this compound is used in a variety of skin care products, and detection at 1 8 ng L 1 is likely due to contamination during sample handling since bench-scale results suggested rapid oxidation at even small O 3 doses. Significant estrogenicity was detected in both raw sewage and tertiary effluent (Table 6). Estrogenicity was reduced to less than detection at all O 3 doses applied, which suggests that oxidation by-products formed were not estrogenic at the MRL of bioassay (0.06 ng L 1 EEq). January Tertiary effluent was collected on 2 sequential days in January 2006 for evaluation of O 3 and O 3 /H 2 O 2 for contaminant destruction. As shown in Table 2, water quality between June and January events was remarkably similar with the exception of temperature. O 3 demand and decay were lower in January as compared to June (Table 3). GC-MS/MS target compounds were not analyzed during this experiment; therefore no data exist Ozone Oxidation of Endocrine Disruptors December

8 % Removal Dilantin AOP Diazepam AOP DEET AOP Dilantin O3 Diazepam O3 DEET O Reaction Time [min] FIGURE 2. BTPP reaction kinetics with O 3 (2.5mg/L) and AOP (O 3 =2.5mg/L, H 2 O 2 =0.5mg/L) Part I. % Removal Ibuprofen O3 40 Iopromide O3 Meprobamate O3 Ibuprofen AOP 20 Iopromide AOP Meprobamate AOP Reaction Time [min] FIGURE 3. BTPP reaction kinetics with O 3 (2.5mg/L) and AOP (O 3 =2.5mg/L, H 2 O 2 =0.5mg/L) Part II. for galaxolide and musk ketone as shown in the June 2005 experiment. Raw sewage also was not evaluated in the January investigation. Concentrations of target compounds were remarkably similar in June and January, except estriol, estrone, iopromide, meprobamate, and trimethoprim, which showed significant increases in concentration (Table 7). The concentration of TCEP decreased by nearly 50% in the January sampling, which was the only compound to show this degree of reduction between the June and January events. Percent removal was quite consistent between the two events. In June, an O 3 dose of 7.3 mg L 1 removed all detectable LC-MS/MS compounds except meprobamate and TCEP (87% and 10% reduction, respectively),andinjanuarya7.0mgl 1 O 3 dose removed meprobamate and TCEP by 83% and <1%, respectively. Iopromide was removed by 91% in June using a 7.3 mg L 1 O 3 dose, while in January removal was 81% at 7.1 mg L 1 O 3. Estrogenicity in the tertiary effluent was greater in January, which is expected considering the detection of estrone and estriol. As determined by the bioassay, estrone and estriol are far weaker estrogens than the primary endogenous estrogen, 17bestradiol (Drewes et al., 2005). It is interesting that estrogenicity was 3-fold greater on the second day of effluent sampling in January, which is likely due to the nearly 4- fold increase in estrone concentration. O 3 provided only a minor reduction in estrogenicity at the 2.1 mg L 1 dose, while the reduction was over 90% at the 3.1 and 7.0 mg L 1 O 3 doses. The small amount of remaining estrogenicity at TABLE 6. BTPP Removal of WWTP Contaminants June 2005 Source Water Raw Tertiary Ozonated Tertiary Ozone [mg/l] Compound ng L 1 ng L 1 %Rem %Rem %Rem Caffeine >80 >80 >80 Carbamazepine >99 >99 >99 DEET Diclofenac >98 >98 >98 Dilantin >99 Erythromycin >99 >99 >99 Galaxolide >99 >99 Hydrocodone >99 >99 >99 Ibuprofen >94 >94 >94 Iopromide >95 Meprobamate Musk Ketone Naproxen >92 >92 >92 Oxybenzone <1 >83 >83 Sulfamethoxazole >99 >99 >99 TCEP < Trimethoprim >97 >97 >97 EEq* >90 >90 >90 *EEq = Estradiol equivalent units; %Rem = %Removed. 452 S. A. Snyder et al. December 2006

9 TABLE 7. BTPP Removal of WWTP Contaminants January 2006 O 3 Dose [mg L H 2 O 2 Dose [mg L 1 ] Target Compound Ave Day 1 (ng L 1 ) Ave Day 2 (ng L 1 ) %Removal %Removal %Removal Androstenedione >39 22 >39 38 >58 >58 Caffeine >53 65 >68 >68 Carbamazepine >99 98 >99 >99 >99 >99 DEET Diclofenac >99 98 >99 >98 >98 >98 Dilantin >99 >99 Erythromycin >99 >99 >99 >99 Estriol 5.7 <5 <1 NA <1 NA >94 NA Estrone <1 44 >81 >94 >91 >94 Fluoxetine >93 81 >93 >91 >99 >91 Gemfibrozil >94 74 >94 >99 >99 >99 Hydrocodone >99 >99 >93 >99 Ibuprofen <1 <1 >82 <1 >24 >93 Iopromide < >24 Meprobamate >98 91 Naproxen >96 96 >96 >98 >66 >98 Oxybenzone <1 3.0 NA 67 NA >66 NA >66 Sulfamethoxazole >99 97 >99 >99 TCEP < Testosterone 1.8 <1 >44 NA >44 NA >98 NA Triclosan >99 >98 Trimethoprim >99 95 >99 >99 97 >99 EEq* *EEq = Estradiol equivalent units; NA = Not Applicable. higher O 3 doses was indistinguishable from 1 of the 3 travel blanks associated with these samples. Advanced oxidation using O 3 /H 2 O 2 also was evaluated during the January 2006 investigation (Table 7). Another 170 L sample of tertiary wastewater was collected the day following O 3 experiments previously described. Target compound concentrations were generally the same or slightly greater in samples collected on the second day (Table 7). Estrone, ibuprofen, naproxen, and EEq increased by 3 4 fold, while gemfibrozil had a remarkable 35-fold increase from 16 to 567 ng L 1. Iopromide showed a 3-fold decrease in concentration from day 1 to day 2, with a concentration decrease from139to45ngl 1. These results show that target compound concentrations fluctuate diurnally. Target compound net removal from wastewater was generally within 10% using either O 3 or O 3 /H 2 O 2. However, results indicate that using O 3 /H 2 O 2 versus O 3 could minimize the required contact time. Disinfection capability of O 3 /H 2 O 2 via OH exposure is poor when compared to O 3 and additional biodegradable DBPs are formed (Wert, Rosario-Ortiz et al. 2006). Removal of estrogenicity appears to be greater using O 3 /H 2 O 2 as compared to O 3 alone; however, the initial estrogenicity was 3-fold greater during O 3 /H 2 O 2 experiments, which may bias percent removal comparisons. Figure 4 shows pcba decomposition during O 3 and O 3 /H 2 O 2 in the January experiment. As expected, the pcba concentration decreased with increasing O 3 exposures as more OH was formed. Greater initial loss of pcba occurred during O 3 /H 2 O 2 versus O 3 when pcba(c/co) FIGURE 4. O3=2.1mg/L O3=3.6 mg/l O3=3.6 mg/l, H2O2=2.5 mg/l O3=7.1 mg/l O3=7.1 mg/l, H2O2=3.5 mg/l Time (min) pcba decay using O 3 and O 3 /H 2 O 2 in wastewater. Ozone Oxidation of Endocrine Disruptors December

10 equivalent dosages of O 3 were applied. However, the net amount of pcba decomposition was similar during O 3 and O 3 /H 2 O 2. The pcba data shows that similar OH exposure can be expected in high TOC wastewater using either O 3 or O 3 /H 2 O 2. These results agree with Acero and von Gunten (2001), who showed OH formation was not greatly enhanced in high TOC waters by O 3 /H 2 O 2 compared to O 3 due to the promotion of O 3 decomposition by NOM (Acero and Von Gunten, 2001). Pilot-Plant Results Bench-scale testing showed that CRW spiked with target compounds exerted greater O 3 demand than unspiked CRW due to the solvents contained in the spiking solution. Therefore, pilot-plant testing was performed to determine the additional O 3 demand from the spiking solvent and the spiking solvent with target compounds. A solvent mixture without target compounds was infused into the pilot plant at the same rate calculated for the spiking solution. The solvent mixture was designed to match composition and quantity of solvents found in the target analyte spiking solution. Once the disinfection goal was achieved, the solvent infusion was stopped and raw water was passed through the O 3 contactor at the same O 3 dose. Dissolved O 3 residual and corresponding disinfection level increased as steady state was achieved, illustrating the demand from the solvent solution. The spiking solution containing target compounds was then infused, and O 3 dose increased to meet the combined demand. The change in O 3 operating conditions is summarized in Table 8. The solvents and target compounds clearly exert an O 3 demand, as demonstrated by the decrease in O 3 residual and CT. Raw water samples were collected every 10 min throughout the 90-min test to assure steady-state spiked concentrations of the target compounds. Measured target compound spiked concentrations ranged from 39 to 182 ng L 1. Samples were collected at 2, 6, and 24 min with residual O 3 quenched using ascorbic acid. Percent removal of target compounds during pilot-plant testing was remarkably comparable to removal shown using the BTPP and spiked CRW at an O 3 dose of 2.5 mg L 1 (Table 9). For instance, at 2, 6, and 24 min of contact time, pilot-plant removal of dilantin was 42%, 72%, and 83%, respectively, while the BTTP removal was 48%, 71%, and 86%, respectively. In all cases there was good agreement between pilot, BTPP, and bench-scale testing at similar O 3 doses (Figure 5), even though GC-MS/MS compounds showed greater analytical variability. Full-Scale Results Four full-scale drinking water treatment plants utilizing ozonation were investigated for contaminant removal. All 4 plants operate using ozone doses between 1 and 3 mg L 1 to achieve disinfection. Of the target compounds analyzed, 15 were detectable at one or more full-scale drinking water treatment facilities. Utility 4 was screened after a change in the target analyte list (Vanderford and Snyder, 2006); therefore, DEET, erythromycin, ibuprofen, and iopromide were not analyzed. The percent removal of target compounds was calculated based on concentrations detected before and after ozonation (Table 10). Despite variability in concentrations of individual compounds among these utilities, removal of compounds was consistent. DEET, dilantin, meprobamate, and iopromide were moderately removed (50% or less concentration reduction), while all other analytes were reduced to less than the MRL. Table 10 also compares full-scale removal to BTPP removal using spiked CRW at an O 3 dose of 1.25 mg L 1 and 24 min contact time. BTPP results are shown as ranges of performance. Excellent agreement was observed between BTPP and full-scale treatment removal despite differences in water quality, O 3 dose, and contaminant concentration. These data demonstrate that ozonation of drinking water for disinfection will result in predictable organic contaminant oxidation. One full-scale WWTP using ozonation was evaluated for contaminant removal. This facility uses an advanced treatment train, which includes ultrafiltration, pre-oxidation (using a small ozone dose), biological activated carbon (BAC), and ozone disinfection. An alternative analytical method using isotope-dilution LC-MS/MS was used for evaluation of target contaminants at this facility (Vanderford TABLE 8. Impact of Solvent Carriers on O 3 Operating Parameters During Pilot-Scale Experiments with CRW Raw Water Raw + Solvent Raw + Spike Transferred Ozone Dose (mg L 1 ) Ozone Demand (mg L 1 ) Half Life (min) Ozone Decay Rate (min 1 ) Ozone 6 min (mg L 1 ) Ozone 10 min (mg L 1 ) Ozone 18 min (mg L 1 ) CT (mg-min L 1 ) S. A. Snyder et al. December 2006

11 TABLE 9. Pilot-Plant Removal using O 3 in CRW Colorado River Water Source Water O 3 Dose [mg L 1 ] 2.4 Reaction Time [min] Target Compound Ave Raw (n = 9) (ng L 1 ) Percent Removal Acetaminophen >97 >97 Androstenedione >99 Atrazine Benzo(a)pyrene Caffeine >99 >99 Carbamazepine 122 >99 >99 >99 DDT DEET Diazepam Diclofenac 111 >99 >99 >99 Dilantin Erythromycin 111 >99 >99 >99 Estradiol 173 >99 >99 >99 Estriol 164 >99 99 >99 Estrone 182 >99 >99 >99 Ethinylestradiol 169 >99 >99 >99 Fluorene Fluoxetine 82 >99 >99 >99 Galaxolide Gemfibrozil 139 >99 >99 >99 Hydrocodone 81 >99 >99 >99 Ibuprofen Iopromide Lindane Meprobamate Metolachlor Musk Ketone Naproxen 58 >98 >98 >98 Oxybenzone 77 >99 >99 >99 Pentoxifylline >99 >99 Progesterone >99 Sulfamethoxazole 57 >98 >98 >98 TCEP 88 <1 4.9 <1 Testosterone >99 Triclosan 102 >99 >99 >99 Trimethoprim 110 >99 >99 >99 and Snyder, 2006). As a result, some compounds shown previously were not evaluated while other compounds were added, such as the blood pressure regulating pharmaceutical atenolol, the plasticizer bisphenol A, and the fluoxetine (Prozac) metabolite norfluoxetine. Removal of detectable target compounds through this treatment train is shown in Table 11. The pre-oxidation step had little impact on contaminant removal, while O 3 disinfection showed removal similar to results using CCWRD effluent. Most compounds were removed by >90%, except atrazine, dilantin, and meprobamate, which also were difficult to oxidize during bench- and pilot-scale tests. In general, contaminant removals from ozone disinfection observed at this WWTP are in good agreement with removal from both CRW and CCWRD at disinfection doses. While not the focus of this study, it is interesting to note that UF had little impact on contaminant removal, while BAC was effective for some contaminants and ineffective for others. The UF and BAC observations are consistent with previously published results (Snyder et al., 2006). Ozone Oxidation of Endocrine Disruptors December

12 Removal Percent DDT DEET Diazepam Dilantin Galaxolide bu 20 0 CONCLUSIONS Bench BTPP Pilot I profen Iopromide Lindane Meprobamate Metoachlor Musk Ketone TCEP Triclosan FIGURE 5. Scale comparison of percent removal of select target compounds at 2.5 mg/l O 3. O 3 and O 3 /H 2 O 2 processes treatment are effective for removal of most organic contaminants in water. Of the 36 compounds considered in this study, 22 were well removed in surface water by O 3 doses of 1.25 mg L 1 or greater (Table 12). Only 6 of the 36 compounds investigated had removals that were generally less than 50%. Musk ketone, lindane, and TCEP were the most challenging compounds to oxidize, with removals less than 20%. Trace contaminant removal in CRW was remarkably similar to removal demonstrated in CCWRD wastewater. At an O 3 dose of 2.1 mg L 1, removal of target compounds was slightly less than removal observed in CRW at 1.25 mg L 1 ; however, at O 3 doses of 3.6, or greater, mg L 1 in CCWRD water, removal was comparable or even superior to CRW. The critical parameter in selecting an O 3 dose for contaminant destruction is clearly O 3 exposure (CT value). From the experiments shown here, an O 3 residual of 0.5 mg L 1 after 2 min of contact time is sufficient to remove the majority of organic contaminants from both surface water and wastewater effluent. The addition of H 2 O 2 will increase the rate of contaminant decomposition, but will not likely increase overall removal given sufficient contact time. In fact, the use of O 3 /H 2 O 2 may result in a net decrease in contaminant removal. Surface water oxidation studies shown here are in generally good agreement with previous studies. Zwiener and Frimmel investigated the removal of 6 acidic pharmaceuticals from surface water using O 3 and O 3 /H 2 O 2 (Zwiener and Frimmel, 2000). The river water used had similar water quality to CRW (DOC = 3.7 mg L 1 and alkalinity = 122 mg L 1 ). Diclofenac was well degraded at O 3 doses of 1 mg L 1 and greater, while ibuprofen required O 3 /H 2 O 2 at 5/1.8 mg L 1 for efficient removal. These results compare favorably to CRW results (Tables 4 6 and 8). Adams et al. (2002) evaluated ozone for the removal of 7 antibiotics in surface water. Trimethoprim showed rapid degradation, with over 95% removal in less than 1.5 min with an O 3 residual of less than 0.05 mg L 1, which agrees with the >99% removal observed in CRW (Table 4). Boyd et al. (2003) showed that naproxen was reduced from 63 ng L 1 to less than 0.4 ng L 1 in a surface water treatment plant using ozone and chlorine oxidation, which was confirmed in Utility 3 in our study (Table 10). Ternes et al. (2002) determined the removal of 5 pharmaceuticals during drinking water treatment. Carbamazepine and diclofenac were removed by >97% using an O 3 dose TABLE 10. Full-Scale Drinking Water Treatment-Concentrations Before O 3, % Removed by O 3, and Removal Observed during Spiked BTPP BTTP Utitilty # Compound ng L 1 % Rem. ng L 1 % Rem. ng L 1 % Rem. ng L 1 % Rem. % Rem. Atrazine 1.9 > >28 <1.0 NA Caffeine <10 NA <10 NA 36 > >80 Carbamazepine 4.8 > >71 16 > >70 >80 DEET 7.9 > NM NA Dilantin Erythromycin <1.0 NA <1.0 NA 2.5 >60 NM NA >80 Estrone <1.0 NA 1.4 >28 <1.0 NA <1.0 NA >80 Galaxolide <10 NA <10 NA <10 NA >80 Gemfibrozil 2.0 >50 <1.0 NA 4.8 > >58 >80 Ibuprofen 1.7 >41 <1.0 NA NM NA Iopromide <1.0 NA NM NA Meprobamate Naproxen <1.0 NA <1.0 NA 12 > NA >80 Sulfamethoxazole 12 >91 11 >90 10 >90 12 >91 >80 Triclosan <1.0 NA <1.0 NA 3.2 > >66 >80 NA = Not Applicable; NM = Not Measured; BTTP = Bench-Top Pilot Plant 1.25 mg/l O3 24 min contact time. 456 S. A. Snyder et al. December 2006

13 TABLE 11. Full-Scale WWTP Removal Ultrafiltration Pre-Oxidation BAC Ozone Disinfection Secondary Effluent Compound ng L 1 ng L 1 % Rem ng L 1 % Rem ng L 1 % Rem ng L 1 % Rem Atenolol < Atrazine < Bisphenol A < Carbamazepine <0.50 >99 Diazepam <0.25 >38 Diclofenac < <0.25 >75 Dilantin Fluoxetine < <0.50 >98 <0.50 NA Gemfibrozil <0.25 >93 Meprobamate < Naproxen <0.50 >97 <0.50 NA Norfluoxetine < <0.50 >64 <0.50 NA Sulfamethoxazole Triclosan < <1.0 >98 <1.0 NA Trimethoprim < >94 NA = Not Applicable; % Rem = % Removal; BAC = Biologically Active Carbon. TABLE 12. Ozone Removal Summary >80% Removal 80 50% Removal 50 20% Removal <20% Removal Acetaminophen Benzo(a)pyrene Atrazine TCEP Androstenedione DDT Iopromide Lindane Caffeine DEET Meprobamate Musk Ketone Carbamazepine Diazepam Diclofenac Dilantin Erythromycin Fluorene Estradiol Ibuprofen Estriol Metolachlor Estrone Ethinylestradiol Fluoxetine Galaxolide Gemfibrozil Hydrocodone Naproxen Oxybenzone Pentoxifylline Progesterone Sulfamethoxazole Testosterone Triclosan Trimethoprim 0.5 mg L 1 and a contact time of 20 min in a controlled experiment. These results were confirmed in CRW at an O 3 dose 0.5 mg L 1 (Table 4). Removal of both pharmaceuticals to less than detection also was demonstrated at a full-scale drinking water plant using an O 3 dose of 1.2 mg L 1 and a contact time of 10 min with raw water containing 2.4 mg L 1 DOC. In our study, carbamazepine was removed to less than 1 ng L 1 in all 4 full-scale drinking water treatment plants investigated, while diclofenac was not detected at all (Table 10). Huber et al (2003). demonstrated the effectiveness of O 3 and O 3 /H 2 O 2 for the removal of 9 Ozone Oxidation of Endocrine Disruptors December

14 pharmaceuticals (including 7 common to this study) from surface water. At O 3 doses of >2 mg L 1, carbamazepine, diclofenac, ethinylestradiol, and sulfamethoxazole were readily removed by >95%, while diazepam, ibuprofen, and iopromide were removed by less than 80%, which is in good agreement with the data shown in the current study (Table 12). The removal of ibuprofen was significantly increased (by 28 43%) with the addition of 0.7 mg L 1 H 2 O 2 at the 2 mg L 1 O 3 dose, similar to the current study where an increase of 18% was demonstrated in CRW with the addition of 0.5 mg L 1 at a constant O 3 dose of 2.5 mg L 1 (Table 5). The removal of carbamazepine, diazepam, caffeine, and pentoxifylline was shown in Rhine River water (DOC of 1 mg L 1 ) using O 3 doses from 1 2 mg L 1 with a contact time of 20 min (McDowell et al., 2005). Similarly to the results shown here (Table 5), carbamazepine, caffeine, and pentoxifylline were oxidized at >95% with an O 3 dose of 0.3 mg L 1, while diazepam showed approximately 60% removal at an O 3 dose of more than 2mgL 1. Results from O 3 and O 3 /H 2 O 2 wastewater experiments also were in good agreement with studies previously published. Ternes et al. (2003) investigated the removal of a variety of emerging contaminants in wastewater using O 3 and O 3 /H 2 O 2 at pilot scale with a contact time of 18 min. Eleven of the analytes investigated by Ternes et al. also were target compounds in the current study (Table 1). Nine of the 11 common compounds were removed to less than quantitation using an O 3 dose of 5 mg L 1. Caffeine and iopromide exhibited only moderate removal (50%) at the lowest O 3 dose investigated (5 mg L 1 ). Advanced oxidation using 10 mg L 1 O 3 combined with 10 mg L 1 H 2 O 2 increased the removal of iopromide to 89%. These results were confirmed in CCWRD effluent at similar doses (Tables 6 and 7). Huber et al. (2005) found similar results when investigating an ozonation pilot for contaminant removal in 3 types of wastewater effluents. At an O 3 dose of 2 mg L 1 with approximately 8 min of contact time, only iodinated contrast media exhibited less than 40% removal. Iopromide also showed poor removal in CCWRD effluent (Tables 6 and 7). The reduction of estrogenicity of wastewater spiked with estradiol using ozone was shown by Kamiya et al. (2005). When 0.1 mg L 1 of ozone was consumed, estrogenicity decreased below the detection limit of the bioassay. Onda et al. (2002) reported that in vitro estrogenicity in wastewater was reduced from >1 to 0.1 EEq by both O 3 and an AOP, which is in excellent agreement with results shown here (Tables 11 and 12). Data from both drinking water and wastewater applications clearly demonstrate the efficacy of ozonation for the reduction of a wide variety of organic contaminants. Challenging contaminants, such as iopromide and ibuprofen will require greater oxidant doses, which may not be economically feasible in most water treatment applications. In general, most target compounds were readily oxidized using either O 3 or O 3 /H 2 O 2. The addition of H 2 O 2 at a constant O 3 dose provided only a small increase in target analyte removal. The chlorophosphate flame retardant, TCEP, was consistently the most recalcitrant compound to oxidize, with removal generally less than 20%. Estrogenicity, as measured by an in vitro bioassay, was readily removed from wastewater using O 3 and O 3 /H 2 O 2 at the dosages investigated. Data presented here for both surface water and wastewater compare well to previously published results. Based upon occurrence and removal data, carbamazepine, sulfamethoxazole, and trimethoprim are excellent indicators for O 3 /H 2 O 2 performance, where near complete removal is expected at even small oxidants doses. Detection of significant concentrations of these compounds would suggest inefficient oxidation and a possible malfunction of the oxidation process. Conversely, TCEP, musk ketone, and meprobamate occur at significant concentrations and are not well removed by most O 3 and O 3 /H 2 O 2 processes at common oxidant doses. Thus, these compounds would likely be detectable in O 3 and O 3 /H 2 O 2 process effluent. No single treatment process will eliminate all trace organics to less than the detection limits of modern analytical instrumentation. Toxicological relevance of exposure to trace concentrations of these contaminants is required in order to establish appropriate water treatment goals. Ozone is a viable process for the oxidation of a great diversity of organic contaminants; however, at economically feasible doses ozone will not result in mineralization (i.e., complete oxidation to carbon dioxide and water). Therefore, treatment by-products will be formed during ozone oxidation, ozone advanced oxidation, and UV advanced oxidation processes. If dissolved organic carbon is not significantly reduced while the majority of contaminants are oxidized, by-products should be expected. In the case of estrogenic endocrine disruptors shown here, the by-products of ozone and ozoneadvanced oxidation were no longer estrogenic as determined by the cellular bioassay. This is expected, as ozone and hydroxyl radicals will attack the phenolic functional group associated with the vast majority of highly estrogenic contaminants. With regards to other forms of toxicity, since natural organic matter (NOM) occurs in drinking water and wastewater at mg L 1 levels as compared to ng L 1 levels of trace contaminants, it is more logical to prioritize investigations regarding by-product toxicity from NOM rather than on trace contaminants. ACKNOWLEDGMENTS This project was funded in part by American Water Works Association Research Foundation (AwwaRF) project 2758, Evaluation of Conventional and Advanced Treatment Processes for the Removal of Endocrine Disruptors and Pharmaceuticals and by AwwaRF project S. A. Snyder et al. December 2006