Removal of pharmaceuticals in WWTP effluents by ozone and hydrogen peroxide

Ozonation to achieve removal of pharmaceuticals from wastewater effluents, with pH values in the upper and lower regions of the typical range for Swedish wastewater, was investigated. The main aim was to study the effects of varying pH values (6.0 and 8.0), and if small additions of H 2 O 2 prior to ozone treatment could improve the removal and lower the reaction time. The effluents studied differed in their chemical characteristics, particularly in terms of alkalinity (65.3–427 mg∙ℓ −1 HCO 3- ), COD (18.2–41.8 mg∙ℓ −1 ), DOC (6.9–12.5 mg∙ℓ −1) , ammonium content (0.02–3.6 mg∙ℓ −1 ) and specific UV absorbance (1.78–2.76 ℓ∙mg −1 ∙m −1 ). As expected, lower ozone decomposition rates were observed in the effluents at pH 6.0 compared to pH 8.0. When pH 8.0 effluents were ozonated, a higher degree of pharmaceutical removal occurred in the effluent with low specific UV absorbance. For pH 6.0 effluents, the removal of pharmaceuticals was most efficient in the effluent with the lowest organic content. The addition of H 2 O 2 had no significant effect on the quantitative removal of pharmaceuticals but enhanced the ozone decomposition rate. Thus, H 2 O 2 addition increased the reaction rate. In practice, this will mean that the reactor volume needed for the ozonation of wastewater effluents can be reduced.


INTRODUCTION
A number of pharmaceuticals of differing therapeutic class, along with their metabolites, have been detected in aquatic environments (Ternes, 1998;Kolpin et al., 2002;Fent et al., 2006;Batt et al., 2006;Snyder, 2008;Verlicchi et al., 2012). The major source of these pharmaceuticals is considered to be the discharge of effluents by wastewater treatment plants (WWTPs) that are not designed for removing trace organic pollutants, in view of the recalcitrance of such pollutants to biodegradation and their limited biological activity, especially in cold climates. Accordingly, additional treatment following biological treatment is called for.
Ozonation is one of the most promising technologies for the removal of organic micropollutants contained in wastewater. The efficiency of ozone in removing pharmaceuticals and personal care products, both from water generally and from wastewater, has been tested in both laboratory-and pilot-scale experiments (Ternes et al., 2003;Huber et al., 2005;Buffle et al., 2006a,b;Bahr et al., 2007;Benner and Ternes, 2009;Hollender et al., 2009;Hansen et al., 2010;Zimmermann et al., 2011). Ozone-based oxidation can be more energy-efficient than UV-based oxidation, especially when used for treatment of waters high in UV absorbance (Rosenfeldt et al., 2006;Hansen and Andersen, 2012).
One of the benefits of using ozonation in aqueous solutions is that the hydroxyl (OH) radicals that are produced will react non-selectively with pharmaceuticals, which could be an advantage for those pharmaceuticals that are difficult to degrade by direct reaction with ozone (Lee and Von Gunten, 2010). The OH radicals are generated through the self-decomposition of ozone in water at pH levels above 7, where the hydroxide ions are acting as initiators (Hoigne and Bader, 1983). Laboratory experiments have shown that the addition of hydrogen peroxide (H 2 O 2 ) enhances the decomposition of ozone, promoting the production of OH radicals (Von Gunten, 2003). Furthermore, non-selective oxidation by highly reactive radicals usually enhances the reaction rates of ozone-resistant compounds, which will reduce the treatment time required (Zwiener and Frimmel, 2000;Huber et al., 2003). Balcioglu and Ötker (2003) reported that adding H 2 O 2 enhances both the UV absorbance (at 254 nm) removal and the decrease of COD in wastewater. The rapid reaction of OH radicals is preferable in practice since it reduces the reactor size needed for such treatment. The efficiency of ozone treatment for the removal of pharmaceuticals can also depend upon the reactivity of the wastewater matrix in general (Nöthe et al., 2009).
The present study aimed at investigating the impact of varying pH levels, within the natural interval in Sweden (pH 6-8), on the removal of pharmaceuticals from wastewater effluents by the addition of ozone. It also determined if the reactivity of ozone can be promoted by addition of small amounts of H 2 O 2 at low pH levels. Since the addition of H 2 O 2 can be expected to enhance the decomposition of ozone to OH radicals, it is of interest to investigate the effect this has in the case of effluents with a pH below 7, where the reaction rate can be expected to be lower and the pharmaceutical removal rate lower due to the lack of hydroxide ions that promote the decomposition of ozone.

Overall experimental setup
Two effluents of relatively high pH (pH 8.0) were treated with ozone, whereas two other effluents, low in pH (pH 6.0), were treated with ozone in combination with H 2 O 2 . Treatment was carried out at these pH levels since they correspond to the upper and lower range of pH values typically found in Swedish WWTP effluents. The effluents selected were from plants with extended nitrogen and phosphorus removal. The difference in pH is due to the origin of the potable water (ground versus surface waters). Further, the effluents also differ with respect to other chemical parameters such as alkalinity and ammonium and organic matter content. The pharmaceuticals investigated represent different therapeutic classes commonly used in Sweden, most of them having been found to be present in WWTP effluents (Falås et al., 2012).
The production of OH radicals by ozone decomposition was followed indirectly through measuring the ozone concentration. The experiments were carried out initially in effluents with a pH range between 6 and 8, with the aim to determine the minimum amount of H 2 O 2 needed to increase the decomposition of ozone.

Chemicals
The H 2 O 2 solution (30%) employed was purchased from Sigma-Aldrich, the NaOH and H 2 SO 4 being purchased from Merck (Germany). The pharmaceutical reference standards were purchased from different suppliers as analytical grade (> 98%) solids (Appendix: Table S1). The stock solution of pharmaceuticals was prepared in methanol at a concentration of 100 mg•ℓ −1 . The ozone stock solution was prepared in a glass bottle containing purified water (Millipore-Billerica, MA) and provided with a diffuser to disperse the generated ozone from a 1.0 g•h −1 ozone generator (O3 Technology AB, Sweden) supplied with dry oxygen gas. The bottle was immersed in an ice bath to increase ozone solubility. Detailed description of the method is found in Antoniou and Andersen (2012).

WWTP effluents
The biologically-treated wastewater effluents investigated, differing in their characteristics and representing the typical variations in alkalinity, pH, and organic matter and ammonium content, were taken from 4 municipal WWTPs in Sweden: Öresundsverket (Effluent 1), Klagshamn (Effluent 2), Uppsala (Effluent 3) and Käppala (Effluent 4). The effluent samples differed from one another in pH on the day of collection and were adjusted at the start of the experiment by use of either NaOH or H 2 SO 4 , so as to be exactly pH 6.0 or pH 8.0. Table 1 shows the quality parameters of the effluents.

Analysis
COD and NH 4 + -N were determined by use of the Hach Lange test kits LCK 114 and LCK 304. To measure alkalinity, a 25 mℓ sample was titrated with 0.05 M HCl to a pH of 4.5, and then the alkalinity in mg•ℓ −1 HCO 3 was calculated. DOC was measured on the basis of wet chemical oxidation, using a Shimadzu TOC-Vwp analyser. The UV-absorbance at 254 nm was measured using a Varian CARY50 Bio UV-Vis spectrophotometer. The specific UV absorbance (SUVA), an indicator of the dissolved aromatic carbon that the wastewater contains, known to affect the reactivity of DOC to ozone, was determined by normalising UV absorbance at 254 nm to the DOC concentration (Weishaar et al., 2003). The O 3 doses delivered were analysed by the colorimetric method of indigo (λ = 600 nm) through preparing bottles of indigo trisulphonate solution in Milli-Q water in parallel with the treatment samples (Bader and Hoigne, 1981;Antoniou and Andersen, 2012).
For pharmaceutical analysis, 100 mℓ samples of the treated effluent were filtered through a 0.45 µm membrane filter (Millipore) and were acidified to pH 3 by use of sulphuric acid. After SPE extraction, LC/MS/MS analysis of the extracts was carried out, using a triple-stage quadrupole MS/MS TSQ Quantum Ultra EMR (Thermo Fisher Scientific, USA) coupled with an Accela LC pump (Thermo Fisher Scientific, USA) and a PAL HTC autosampler (CTC Analytics AG, Switzerland) having a Hypersil GOLD aQ TM column (50 mm x 2.1 mm ID x 5 µm particles, Thermo Fisher Scientific, USA). The method of analysing pharmaceuticals was used earlier by Hörsing et al.  Table S2.

Experimental setup
For the ozone consumption experiments carried out, the biologically-treated municipal wastewater was ozonated at different pH levels and O 3 to H 2 O 2 ratios. Samples were taken at different reaction times for analysis of the O 3 content. For experiments involving pharmaceutical removal, the wastewater

Determination of ozone concentration profiles at different pH
As can been seen in Fig. 1, the ozone concentration in the wastewater effluents decreased rapidly after the first minute of ozone addition. Thereafter, the rate of ozone decomposition decreased gradually and stabilised. This relatively fast ozone consumption was to be expected due to the matrix components in the wastewater consuming the oxidant. In addition, the decomposition of ozone tended to proceed faster at pH 8 than at pH 6, in accordance with the results of other studies, such as Hoigne and Bader (1981), using drinking water spiked with organic compounds, and Elovitz et al. When H 2 O 2 was added to the effluents at H 2 O 2 /O 3 ratios ranging from 2 to 0.25 (Fig. 1), ozone was almost completely consumed during the first minute. As could be expected, the effluent of high pH (pH 8) exhibited the fastest ozone decomposition rate (< 1 min). An additional experiment was also carried out at pH 6.0, involving the use of lower doses of H 2 O 2 , resulting in significantly lower H 2 O 2 /O 3 ratios of 0.05-0.10. As can be seen (Fig. 1), the differences in ozone removal rate between the samples are most obvious in the first minute or so of treatment, where the decomposition of ozone increased with an increase in the H 2 O 2 /O 3 ratio for around 2 min, after which nearly all the ozone has been consumed.

O 3 concentration profiles in the WWTP effluents tested
On the basis of these findings, it is obvious that the addition of relatively small amounts of H 2 O 2 is able to change the ozone concentration profile appreciably. To investigate this further, a set of experiments was carried out using 4 different effluent wastewaters (Table 1), two with relatively high pH and two with relatively low pH. The effluents, after pH adjustments to 8.0 and 6.0, respectively, were treated with ozone and the decomposition was followed (Fig. 2). In the high-pH effluents ( Fig. 2A), about half of the ozone was already consumed during the first minute, especially in the case of Effluent 2. The differences observed can be attributed to the higher SUVA content in Effluent 2 than in Effluent 1 ( Table 1). The relatively high content of aromatic compounds, indicated by the relatively high SUVA level, could explain the increased ozone consumption in the early stages of treatment, due to fast reactivity of aromatic compounds, as also observed by Westerhoff et al. (1999). At pH 168 6.0, in contrast, Effluent 3 appears to have a much lower ozone demand than Effluent 4 (Fig. 2B), this probably being due to the lower organic content of Effluent 3, which is only about half that of Effluent 4. Similar to what can be seen above (Fig. 1), the addition of H 2 O 2 to the effluent led to an increased decomposition of ozone, measured as a decline in ozone concentration.
These findings show that it is important, when employing ozonation, to investigate the initial ozone demand of the wastewater due to matrix effects. The present findings also show that at low pH the combination of ozone and H 2 O 2 reduces the reaction time, indicating that it is possible to reduce the size of the reaction tank employed for treatment.

Removal of pharmaceuticals by O 3 and H 2 O 2
In the ozonation of pH 8.0 effluents, a significant reduction in the different pharmaceuticals was found, especially in the case of Effluent 1 (Fig. 3A), even at relatively low doses of ozone. At the lowest dose (1.5 mg•ℓ −1 O 3 ), 9 of the 40 pharmaceuticals (clomipramine, sulfamethoxazole, repaglinide, ethinyl estradiol, fexofenadine, codeine, naproxen, diltiazem and eprosartan) already showed removal rates of 90-100%; only 8 of the pharmaceuticals (bupropion, oxazepam, levonorgestrel, memantine, fluconazole, flutamide, ketoprofen and ibuprofen) exhibited < 50% removal. As the ozone dose was increased, most of the pharmaceuticals, including the less reactive ones, were degraded.
On the other hand, in Effluent 2 (Fig. 3B), the pharmaceuticals were poorly removed, even when the O 3 dose was increased. This can be attributed to the high SUVA level (2.76 as compared with 1.78) of this effluent. The high ozone reactivity of the aromatic components of the DOC may have contributed to the decrease in pharmaceutical removal from the effluent. Also, as can be observed in Fig. 3, some of the pharmaceuticals in Effluent 1 exhibited a high level of removal in response to the lowest ozone dose but did not follow the same pattern of removal in Effluent 2. For example, both clomipramine and repaglinide showed a high degree of removal at the lowest ozone dose, yet when treated with the same O 3 dose in Effluent 2 it was only clomipramine for which the degree of removal was significant (~50%). This shows clearly that both the level of removal and the reactivity of pharmaceuticals can vary depending upon the composition of the wastewater matrix.
The efficiency of ozone in removing pharmaceuticals from pH 6.0 effluents (Fig. 4A) showed that ozone alone could remove > 90% of half of the pharmaceuticals present in Effluent 3 at the lowest ozone dose (1.8 mg•ℓ −1 ). When the dose was increased to 4.4 mg•ℓ −1 , all pharmaceuticals except fluconazole were degraded by over 90%. A still further increase in the ozone dose resulted in over 99% removal of all pharmaceuticals, except fluconazole (93%) and ibuprofen (96%). In contrast, the ozonation of Effluent 4 resulted in > 90% degradation of the pharmaceuticals when rather high doses of ozone (> 5 mg•ℓ −1 ) were employed (Fig. 4B)  169 (Effluent 1), the observed high removal of pharmaceuticals in the pH 6.0 effluent (Effluent 3) is likely due to the very low organic content of this effluent (Table 1), which resulted in more ozone being available (Fig. 2B) to react with the pharmaceuticals. Figure 5 illustrates the contribution of H 2 O 2 addition to the removal of those pharmaceuticals that have been shown to have the lowest reactivity towards ozone. For fluconazole (Effluent 3), as can be seen, there was only a slight increase in removal after the addition of H 2 O 2 , whereas for ibuprofen no improvement in its removal occurred. Thus, the addition of H 2 O 2 (at an H 2 O 2 / O 3 ratio of 0.08-0.13) to an initial ozone dose of 10 mg•ℓ −1   170 could not be expected to have any appreciable impact on the removal of pharmaceuticals in this effluent. On the other hand, for Effluent 4, the addition of H 2 O 2 was found to enhance the removal of ibuprofen, fluconazole, levonorgestrel, sulfamethoxazole and ketoprofen by only 4-16% and of naproxenl by ~15%.
The overall findings of this study show that reaction time can be reduced when ozone is combined with small amounts of H 2 O 2 , which will be advantageous when practical implementation of the technology takes place. As for most pharmaceuticals, this addition has no impact on removal efficiency, i.e., neither increasing nor decreasing its removal.
The majority of pharmaceuticals included in this study contained acidic and/or basic groups, having different and pH-dependent charges (positive, neutral or negative) and, as a result, may also differ in their tertiary chemical structure, as a function of pH. The pharmaceuticals that are acidic can be protolysed at pH 6, with no further changes occurring then when the pH is increased to 8. In contrast, those pharmaceuticals that have basic group(s) and low pK b values go from being unprotolysed at pH 6 to being protolysed at pH 8, the charge thus changing from positive to neutral, which can result in a change in the tertiary structure. Those pharmaceuticals having both acidic and basic groups may also undergo changes in the charge and in their tertiary structure. This can be expected to have an impact on the oxidation rate. It is not possible, however, on the basis of the experiments carried out here, to draw any final conclusions regarding this. Table 2 provides an overview of the findings regarding removal efficiencies for the pharmaceuticals that were investigated. It can readily be seen that an ozone dose of around 5 mg•ℓ −1 is sufficient to remove over half of the target pharmaceuticals, except in the case of Effluent 2, in which a much higher ozone dose may be required for removing a large fraction of the pharmaceuticals, this most likely being due to the higher SUVA level of Effluent 2. At the same time, it appears that, in the case of wastewaters such as Effluent 3 that are low in pH and organic content, a reasonable dose of ozone to remove over 90% of the pharmaceuticals is around 5 mg•ℓ −1 .
The oxidation of pharmaceuticals may lead to the production of by-and transformation products. Since these can be toxic to varying degrees as compared with the mother compound, toxicity evaluation of a given technology should be performed before it is considered for implementation.

CONCLUSIONS
The following conclusions can be drawn on the basis of the results of the study: • Ozonation can be employed as an additional treatment step to enable trace pharmaceuticals to be removed effectively from wastewater effluents. • The amount of ozone required for the removal of pharmaceuticals is dependent upon the chemical composition of the wastewater and upon the target compounds, with the content of organic matter in general and its aromaticity being of considerable importance. • Ozone decomposition can be stimulated by adding hydrogen peroxide at low pH. The addition of hydrogen peroxide has only a limited impact on the quantitative removal of pharmaceuticals. However, it reduces the treatment time and, accordingly, the reaction volume needed, which will be advantageous in practice. Amitriptyline