Photodegradation of some brominated and phenolic micropollutants in raw hospital wastewater with CeO₂ and TiO₂ nanoparticles

Micropollutants in hospital wastewater are directly discharged into the sewage system without treatment since the conventional sewage/urban treatment plants can only remove macropollutants such as biochemical oxygen demand, chemical oxygen demand (COD), nitrogen and phosphorus in Turkey. The micropollutants decrease the biological treatment efficiency if they are not treated and then discharged into the receiving environment (Petrie et al. 2014).

Gemfibrozil (GFZ) is a blood lipid regulator and is used to treat hypertriglyceridemia (Grenni et al. 2013). It is frequently detected in surface water and wastewater (Krkošek et al. 2011). GFZ is toxic to aquatic microorganisms and may also cause endocrine disruption in fish (Chen et al. 2017). It has been found in wastewater treatment plants and also in freshwaters at a concentration ranging from 0.07 to 0.51 μg L⁻¹ (Fang et al. 2012). Triclosan (TCS) is a broad-spectrum antibacterial and antifungal agent (Xie et al. 2012) used in hospitals. It has been found in surface waters, wastewaters, sediments and sludges due its widespread usage. It has low acute and chronic toxicities (Tohidi & Cai 2017). The concentration of TCS has been found to range from 1.4 up to 133,000 ng L⁻¹ in surface waters, and from 20 to 133,000 μg kg⁻¹ dw in biosolids (Montaseri & Forbes 2016). 2,3,4,5,6-Pentabromoethyl benzene (PBEB) and 2,3,4,5,6-pentabromotoluene (PBT) are new brominated flame retardants (NBFRs) and have been used in order to prevent fires and in the formation of different types of polymers, resins and plastics (Cristale et al. 2012; Chen et al. 2012). NBFRs are of environmental concern because of their high lipophilicity, persistence and resistance to degradation (Barón et al. 2015). Their bioaccumulative and carcinogenic/mutagenic properties make them potentially dangerous for environmental health (Sun et al. 2015). PBEB was detected in samples of river sediments at concentrations ranging from 3.1 to 10.0 ng g⁻¹ dry weight (Cristale et al. 2013). In river water, PBT concentrations range from 0.0003 to 0.0210 μg L⁻¹ (Wu et al. 2010).

Nanoparticles with unique surface-active properties are beneficial in removing pharmaceuticals and micropollutants (Hu et al. 2012). Titanium (IV) oxide (TiO₂) as a photocatalyst is widely applied to wastewater treatment due to its photocatalytic oxidation ability, and stable and non-toxic characteristics (Chen et al. 2017). Cerium (IV) oxide (CeO₂) nanoparticles are widely used as a catalyst and as superconducting materials (Xu et al. 2017).

The treatability studies of GFZ, TCS, PBT and PBEB performed with nano-CeO₂ and nano-TiO₂ via photocatalytic processes are limited in recent literature. Chen et al. determined 61.8% GFZ removal with 0.05 g L⁻¹ nano-TiO₂ in a synthetic wastewater having a GFZ concentration of 2 mg L⁻¹ under 350 W UV light power at a pH of 7.00 through 8.0 min (Chen et al. 2017). Son et al. reported 82% TCS photodegradation rate within 20 min with 0.1 g L⁻¹ nano-TiO₂ at a pH of 7.00 under 1.37 × 10⁻⁴ Einstein L⁻¹ min⁻¹ UV (Son et al. 2009). Yu et al. found 95% photocatalytic degradation for 9 mg L⁻¹ TCS within 6 h using 100 mg L⁻¹ nano-TiO₂ at UV light power of 15 W (Yu et al. 2006). García et al. found 85% TCS photodegradation at 100 μg L⁻¹ TCS by 0.335 g L⁻¹ nano-TiO₂ under solar light irradiation after 120 min (García et al. 2012). The physicochemical properties (hydrophobicity, solubility, volatilization) of micropollutants used in this study are shown in Table 1.

The novelty of the study is to treat some brominated and phenolic micropollutants present in raw hospital wastewater for the first time in Turkey by photocatalytic processes under laboratory conditions. The aim of this study was to evaluate the photocatalytic removals of PBT, PBEB, TCS and GFZ in a raw hospital wastewater using CeO₂ and TiO₂. Effects of irradiation times (15; 30; 45; 60; 90 min), nanoparticle concentrations (0.25; 0.50; 1.00; 1.50 g L⁻¹), UV light powers (120; 210; 300 W) and raw hospital wastewater pH (4.00; 7.00; 8.50) on PBT, PBEB, TCS and GFZ photodegradation yields throughout photocatalysis at constant temperature (25 °C) were investigated using nano-CeO₂ and nano-TiO₂ generated under laboratory conditions. The recoveries of the nanoparticles and their costs were also investigated.

Raw hospital wastewater was taken from the Dokuz Eylul University Hospital (Izmir, Turkey) sewer channel, which flows through the sewer channel without any treatment. The influent COD concentration of the raw hospital wastewater was 1,900 ± 20 mg L⁻¹ while the influent PBT, PBEB, TCS and GFZ concentrations were 2.31 ± 0.003, 3.28 ± 0.004, 860.00 ± 0.05 and 53.30 ± 0.05 μg L⁻¹, respectively. The pH, the temperature, the total nitrogen and total phosphorus concentrations were 8.50 ± 0.02, 18 ± 2 °C, 0.40 ± 0.01 and 1.00 ± 0.01 mg L⁻¹, respectively.

PBEB, PBT, TCS and GFZ were extracted the from raw hospital wastewater by solid-phase extraction method. Oasis HLB cartridges (200 mg) were used in the extraction experiments. PBEB and PBT extraction experiments were carried out according to Chokwe et al. (2015) while TCS and GFZ extraction experiments were carried out according to Diwan et al. (2010). Extraction recovery rates for PBEB, PBT, TCS and GFZ are shown in Table 2.

PBEB (CAS No. 85-22-3) and PBT (CAS No. 87-83-2) measurements were performed on a GC Agilent 7890A equipped with an Agilent 5975C inert mass selective detector. The gas chromatography–mass spectrometer (GC-MS) is equipped with a flame ionization detector and an HP-5MS capillary column with dimensions of 30.0 m (length), 0.250 mm (I.D.), and 0.25 μm (film thickness) from Agilent Technologies. The oven program was taken from Cristale et al. and developed as following: the oven was set at 40 °C (1 min), increased to 60 °C at 15 °C min⁻¹ (8 min), to 220 °C at 10 °C min⁻¹ and to 300 °C at 15 °C min⁻¹ (8 min). Helium (purity 99.999%) was employed as carrier gas with a constant flow of 1.5 mL min⁻¹. The standards were injected (2 μL) into the GC system in splitless mode, with a splitless time of 1.5 min. The injector, quadrupole and transfer line were set at 300, 150, and 280 °C respectively (Cristale et al. 2012). The retention times (RT), m/z transitions and method recovery rates for PBEB and PBT monitored by the GC-MS are shown in Table 2.

TCS (CAS No. 3380-34-5) and GFZ (CAS No. 25812-30-0) measurements were performed on an Agilent 1100 Series high-performance liquid chromatograph (HPLC) equipped with a diode array detector. A Thermo C18 column (5 μm, 250 mm × 4.6 mm, Thermo Scientific) was used and the injection volume of each sample was 20 μL. For TCS, the mobile phase consisted of pure acetonitrile and HPLC grade water (90:10 v:v) with a flow rate of 1.0 mL min⁻¹ for a column oven temperature of 40 °C and retention time of 3.606 min (Maaroof & Uysal 2014), while for GFZ, the mobile phase consisted of pure acetonitrile, pure methanol and 0.05 M (pH = 7.00) NH₄H₂PO₄ (60:30:10 v:v:v) with a flowrate of 0.8 mL min⁻¹ for a column oven temperature of 20 °C and retention time of 4.351 min (Elsherif et al. 2013). The RT and method recovery rates of TCS and GFZ monitored by the HPLC are shown in Table 2.

Conventional pollutants in hospital wastewater such as COD were measured according to Standard Methods (APHA/AWWA/WEF 2012). Total nitrogen and total phosphorus were measured with reagent kits in a Photometer Nova 60/Spectroquant. The pH and temperature were measured with WTW probes.

TiCl₄ (0.92 mL) was dissolved in 35 mL de-ionized water under an ice water bath, and then 0.8 g of P123 as a structure-directing agent was added into the solution under magnet stirring (Tang et al. 2016). When the temperature was raised to 50 °C, the mixture solution was aged for another 12 h. Finally, the solution obtained was dried via lyophilization. The obtained solids were calcined at 300 °C for 2 h, and the final product was named as P_0.5-TiO₂. One millilitre of po1yoxyethylene octylphenol ether was mixed with 1 mL 1-hexanol and 2.5 mL cyclohexane at a mass ratio of 1:1:2.5, then 0.5 mol L⁻¹ Ce(NO₃)₃·6H₂O was added to obtain an emulsion (Wang et al. 2015). This mixture was aged at the reaction temperature of 300 °C for 3 h before being centrifuged to separate the solid, which was subsequently rinsed and dried to obtain the nano-CeO_2.

The photocatalytic experiments were carried out in a covered stainless steel system consisting of quartz glass reactors with a volume of 1,000 mL and UV light lamps with a 254 nm wavelength (895.0 mm × 26.0 mm, 30.0 W, 0.36 A, G13 Model, Osram). The light power required was provided by increasing or decreasing the UV lamp numbers. The distance between the UV lamps and quartz glass reactors was 10 cm (Figure 1).

The control experiments were carried out under dark conditions for the study of the adsorption capacity of TiO₂ and CeO₂ nanoparticles. Raw hospital wastewater was first filtered through a 0.45 μm pore sized membrane to remove physical impurities. The nanoparticle was added to the filtered hospital wastewater in a quartz glass reactor and it was stirred at 25 °C. The control for photolysis was carried out under UV light without nano-TiO₂ and nano-CeO₂ to determine the photolysis capacity of the micropollutants. Raw hospital wastewater was filtered through a 0.45 μm pore sized membrane and then irradiated under UV in photocatalytic reactors without nanoparticles at 25 °C. All the experiments were performed in triplicate.

The effects of different nanoparticle concentrations (nano-TiO₂ and nano-CeO₂) (0.25; 0.50; 1.00; 1.50 g L⁻¹), irradiation times (15; 30; 45; 60; 90 min), UV light powers (120; 210; 300 W) and hospital wastewater pH (4.00; 7.00; 8.50) on the PBT, PBEB, TCS and GFZ photodegradation yields throughout photocatalysis at constant temperature (25 °C) were investigated.

After the first utilization of nano-TiO₂ and nano-CeO₂ in the photocatalytic experiments, the nanoparticles were separated by filtering them through 0.45 μm pore sized membranes. Then, the nanoparticles were regenerated using ethanol with 0.10 mol L⁻¹ HCl (Chowdhury & Balasubramanian 2014). The nanoparticles were dried under vacuum and they were reused for the second treatment process to treat TCS, GFZ, PBT and PBEB again. For each new treatment step the same procedure was applied to the nanoparticles. Six sequential treatment steps were investigated in order to detect the effect of reusability of the nanoparticles from the raw hospital wastewater after each treatment.

The correlation (R) between data was performed using Excel Microsoft 2010 while the significance between parameters was performed in analysis of variance (ANOVA) tests using an α-value of 0.05 with P (probability) values. All of the differences were considered significant at p < 0.05. The Kruskal–Wallis one way test was applied to calculate the significant difference of the operational conditions on the photodegradation yields. Kruskal–Wallis ANOVA χ² values were calculated for P > 0.05(Quinn & Keough 2002).

The SEM image in Figure 2(a) shows that the TiO₂ nanoparticle samples exhibit a well-defined spherical structure after 300 °C calcination. The TEM figure shows that the as-prepared powder is completely crystalline, and consists entirely of a pure anatase phase (Figure 2(b)). The XRD analysis showed that the TiO₂ nanoparticle originated completely from the pure anatase phase with a good crystallinity (Figure 2(c)). From the XRD pattern, it is clear that the material formed has a tetragonal crystal structure with lattice parameters a = b = 3.785 Å and c = 9.513 Å. The crystallite size of the TiO₂ was in the range 10.5–11.5 nm and it had a unit cell volume of 136.27 Å (Table 3). The main peaks of anatase TiO₂ appeared at 25°, 28°, 39°, 48°, 54°, 55°, 63°, 69°, 70° and 76° (Figure 2(c)). These peaks of anatase in TiO₂ were in accordance with the peaks given by Hu et al. (2017) and Hassani et al. (2017) (at 25°, 28°, 38° and 48°, 54°, 55°, 63°, 69°, 70°, and 76°). The strongest and narrowest peaks were detected at 2θ = 25° for anatase. According to FT-IR analysis, the peaks of anatase TiO₂ appeared at 3415.95, 2360.83, 1630.35 and 1386.86 cm⁻¹ (Figure 2(d)). These peaks were similar to the findings of Hassani et al. (2017). In the FT-IR spectrum of the TiO₂ nanoparticle, the bands appeared at 3415.95–2360.83 cm⁻¹, which shows the bending vibration of OH molecules. A carbonyl peak appeared at around 1630.35 cm⁻¹. The peak at 1386.86 cm⁻¹ shows stretching vibrations of Ti-OH. The BET surface area of the synthesized TiO₂ nanoparticle sample shows less aggregation and a smaller particle size, which is in accordance with the SEM and TEM analysis. The BET surface area of the TiO₂ nanoparticle catalyst had an average pore diameter of 98.84 Å with a total pore volume of 0.402 cm³ g⁻¹ while the surface area was 205 m² g⁻¹ (Table 3). The absorption spectrum of the TiO₂ nanoparticle at a wavelength of 410 nm showed that it is a highly crystalline nanomaterial with a band gap value of 3.02 Ev due to the presence of a very fine anatase phase (Figure 2(e)).

SEM images showed that CeO₂ nanoparticles exhibited spherical morphologies (Figure 3(a)). Figure 3(b) shows the TEM image of the sample synthesized at 165 °C displaying a particle size of 1–3 nm and strong agglomeration. The XRD patterns reveal that the prepared CeO₂ nanoparticles had a cubic fluorite structure (Figure 3(c)). From the XRD peaks it is evident that the crystallites were oriented along 28°, 33°, 48°, 56°, 59° and 69°, corresponding to the cubic fluorite structure of CeO₂. Similar peaks at 28°, 33°, 48° and 56° were reported by Phuruangrat et al. (2017) and Channei et al. (2017). The crystallite sizes calculated were around 4.5 and 6.5 nm (Table 2). The BET surface area of the CeO₂ nanoparticle had an average pore diameter of 78.06 Å with a pore volume of 0.309 cm³ g⁻¹ while the surface area was 302 m² g⁻¹ (Table 2). The FT-IR spectrum was recorded in the range of 650 to 4,000 cm⁻¹ (Figure 3(d)). These FT-IR peaks are in accordance with the data obtained by Ravishankar et al. (2015) and Tambat et al. (2016) for CeO₂ nanoparticles). These peaks are due to (Ce-O) metal–oxygen bond vibrations. The band gap of nano-CeO₂ was higher than that of nano-TiO₂ (Table 3). The prepared nano-CeO₂ and nano-TiO₂ exhibited better properties than the commercial nanoparticles (Table 3). Nano-CeO₂ produced under laboratory conditions had a higher surface area than those of nano-TiO₂ while the pore volume, the average size and pore diameter of nano-CeO₂ was lower than nano-TiO_2. This increased the photocatalytic properties of nano-CeO_2. The UV–visible absorption spectral studies showed that the synthesized CeO₂ nanoparticles had a strong absorption band at low wavelength near 380 nm, corresponded to a band gap energy of 3.26 eV (Figure 3(e)). In comparison with the UV–visible absorption spectrum of CeO₂ nanoparticles reported in the literature (Farahmandjou et al. 2016), similar peaks in the spectrum located at around 400–700 nm were observed. The high band gap energy and low pore volume and pore diameter of nano-CeO₂ compared to nano-TiO₂ enhanced the UV light absorption of the CeO₂ nanoparticle.

During the adsorption studies of micropollutants on the surface of nano-TiO₂ the maximum removal percentages of GFZ, TCS, PBT and PBEB via adsorption were 5.7%, 5.6%, 7.0% and 7.6%, respectively, at 0.50 g L⁻¹ nano-TiO₂ concentration after a 45 min stirring time at a pH of 8.50 and a temperature of 25 °C without UV (Figure 4(a)). The adsorption removal percentages of GFZ, TCS, PBT and PBEB were 9.0%, 6.0%, 13.4% and 17.1%, respectively, at 0.50 g L⁻¹ nano-CeO₂ concentration after a 45 min stirring time at the same operational conditions given above under dark experimental conditions (Figure 4(b)). It can be said that the adsorption of micropollutants on photocatalyst was not the primary mechanism for their removals. The adsorption yields were double for PBEB and PBT compared to GFZ and TCS. This can be explained by the physicochemical properties of PBEB (low solubility = 2.895 × 10⁻⁷ g L⁻¹ and high adsorption capacity log K_ow = 7.48) and PBT (low solubility = 9.351 × 10⁻⁷ g L⁻¹ and high adsorption capacity log K_ow = 6.99) compared to GFZ (low solubility = 4.964 × 10⁻³ g L⁻¹ and high adsorption capacity log K_ow = 4.77) and TCS (low solubility = 2.13 × 10⁻⁸ g L⁻¹ and high adsorption capacity log K_ow = 4.76) and by their structures (brominated groups in PBT and PBEB; phenolic groups in GFZ and TCS). ANOVA multiple regression analysis showed that there was a linear regression between adsorption yields and physicochemical properties of the studied compounds (R = 0.85) and this was not statistically significant (P = 0.21, α = 0.05). The Kruskal–Wallis one way test showed that the physicochemical properties had a significant effect on adsorption of all micropollutants (Kruskal–Wallis ANOVA, χ² = 4.29, P < 0.05).

The experiments were also performed using direct photolysis of micropollutants in the absence of nano-TiO₂ and nano-CeO₂ under 210 W UV at 254 nm. As a result, no obvious change in their concentrations within 45 min UV light irradiation was observed (9.0%, 10.0%, 14.2% and 15.9% removals for GFZ, TCS, PBT and PBEB, respectively) (Figure 4(c)). For sole UV exposure, it was found that micropollutants in hospital wastewater with phenol bonds (TCS and GFZ) or brominated bonds (PBT and PBEB) were not easily photolyzed by 254 nm at 210 W UV photons. During photolysis only the photons were generated during UV irradiation instead of generation of hydroxyl radicals (OH^•). The low removals of GFZ, TCS, PBT and PBEB in photolysis were expected to be achieved mostly by the low number of photons from UV irradiation (Son et al. 2009). Similar results were obtained for the direct UV photolysis of TCS and GFZ (Carlson et al. 2015). Stamatis et al. also found that the removal of 1.0 mg L⁻¹ TCS with adsorption and photolysis was low (1.8% and 4.9%, respectively) with 550 mg L⁻¹ TiO₂ at an irradiation intensity of 700 W m⁻² (Stamatis et al. 2014). The ANOVA multiple test statistic showed a linear regression between physicochemical properties and removals of micropollutants by photolysis (R = 0.84), although this regression was not significant (P = 0.15, α = 0.05). The Kruskal–Wallis one way test showed that photolysis had no significant effects on the removal of micropollutants (Kruskal–Wallis ANOVA, χ² = 16.14, P > 0.05).

Effect of irradiation time on the photodegradation of GFZ, TCS, PBT and PBEB. According to the previous studies given in the literature and from our preliminary studies it was decided to use a 0.25 g L⁻¹ nano-TiO₂ and nano-CeO₂ concentration at a 120 W UV power. The photocatalytic experiments performed with nano-TiO₂ showed that 45 min is the optimum irradiation time for the maximum photocatalytic removal of GFZ (80%), TCS (83%), PBT (72%) and PBEB (68%) at 25 °C at a pH of 7.00 (Figure 5(a)). Any further increase in irradiation time decreased all the micropollutants' photo-removals because of the deactivation of active sites by the deposition of by-products. Neves reported that the increasing of irradiation time from 1 h to 5 h positively affected the 0.01 g L⁻¹ TCS removal (from 68% to 78%) at a pH value of 7.00 with 0.50 g L⁻¹ TiO₂ with UV (Neves 2014). It is important to note that the TCS photodegradation yields in our study were higher than the data obtained by Neves although they were obtained with half of the nanoparticle dose and four times the irradiation time. Also, the photocatalytic experiments performed with nano-CeO₂ showed that 45 min was the optimum irradiation time for the maximum photocatalytic removals of GFZ (86%), TCS (88%), PBT (77%) and PBEB (75%) with 0.25 g L⁻¹ nano-CeO₂ concentration at 120 W power at 25 °C and at a pH of 7.00 (Figure 5(b)). It is widely accepted that UV excitation leads to the generation of electron (⁠⁠) – hole (⁠⁠) pairs on the nano-TiO₂ and nano-CeO₂ surface. These charges will react with water molecules and result in OH^•. The hydroxyl radicals oxidize the micropollutant molecules and cause the latter to decompose and mineralize to water and to carbon dioxide. ANOVA test statistics showed that a linear regression between irradiation time and micropollutant photo-removals was observed by increasing the irradiation time from 15 min to 60 min (R = 0.89) and this correlation was significant (P = 0.50). The Kruskal–Wallis one way test showed that the increased length of time up to 60 min had a significant effect on micropollutant removals (Kruskal–Wallis ANOVA, χ² = 1.12, p < 0.05). A significant regression between photodegradation yields of micropollutants and irradiation time for higher than 45 min (60 and 90 min) (R = 0.23) was not observed and this correlation was not significant (P = 12.07, α = 0.05) (Kruskal–Wallis ANOVA, χ² = 16.78, p < 0.05).

The lower photodegradation yields in PBEB and PBT compared to GFZ and TCS can be attributed to the difficulties in separating the five brominated groups with methyl and ethyl during consecutive reductive debromination as reported by Lee & Kwon (2016). Their high log K_ow and low solubilities and their hydrophobicities compared to GFZ and TCS resulted in low removals. It was reported that debromination is a main photochemical reaction pathway for NBFRs (Covaci et al. 2011). During photooxidation, intramolecular elimination of HBr from PBEB and PBT can subsequently generate lower brominated compounds via consecutive debromination (Covaci et al. 2011). Higher photodegradation yields of TCS were observed (87%) to produce several intermediates including a dioxin congener (2,8-dichlorodibenzodioxin) under UV with TiO₂ via dechlorination as reported by Bianco et al. (2015). The separation of chlorinated and phenolic groups photodegraded the TCS to CO₂ and water. Wang et al. (2017) found that TCS was photodegraded with a yield of 78% under 130 W UV power while the photodegradation was enhanced to 89% with 1–5 kGy gamma irradiation, resulting in the mineralization of TCS. When the OH addition attacked the GFZ, transient states of the OH adduct radicals were formed, resulting in the formation of hydroxylated by-products (e.g., GFZ-Pa), followed by further radical oxidation to form stable by-products (GFZ-Pb and -Pd). When the OH addition reaction took place on the ipso aromatic C atom of GFZ-Pa, this ipso-substitution led to stabilized carbon-centered radicals. In addition, O-dealkylation reactions caused either side chain cleavage of the parent compound, giving rise to the generation of GFZ-Pc/k (Chen et al. 2017). The photo-removals of PBT, PBEB, TCS and GFZ found in this study were higher than in the literature cited above.

Photocatalytic removals of 84%, 88%, 78% and 75% were obtained for GFZ, TCS, PBT and PBEB with 0.50 g L⁻¹ nano-TiO₂ (Figure 5(c)). Increasing the amount of photocatalysts from 0.25 to 0.50 g L⁻¹ resulted in an increase in the photodegradation efficiencies of GFZ, TCS, PBT and PBEB. Among the 0.25, 0.50, 1.00 and 1.50 g L⁻¹ nanoparticle concentrations, 0.50 g L⁻¹ was determined as the optimum nanoparticle concentration at 120 W power at 25 °C and at a pH of 7.00 for maximum photodegradation yields of micropollutants. The photodegradation yields found for GFZ and TCS were higher than for PBT and PBEB due to their physicochemical properties. Concentrations of nano-TiO₂ and nano-CeO₂ higher than 0.50 g L⁻¹ slightly decreased the photodegradation yields of GFZ, TCS, PBT and PBEB. The increase in TiO₂ loading from 0.25 g L⁻¹ to 0.50 g L⁻¹ provided more binding sites for substrate molecules to adsorb to the TiO₂ surface. At low TiO₂ loading low photodegradation yields were obtained for all the micropollutants since the number of active sites on the surface of the nanoparticle was low and partially exposed to the pollutants. With excessive nano-TiO₂ and nano-CeO₂ loadings (1.00 and 1.50 g L⁻¹), it has been suggested that the TiO₂ particles in the centre of an irradiating vessel may be shielded from the incident UV by other nano-TiO₂ and nano-CeO₂ particles, and thus particles in the center do not contribute significantly to the production of electron (⁠⁠) – hole (h⁺_VB) pairs (Bethi et al. 2016). At high TiO₂ concentrations the contacting sites on the surface of the nanocomposite were not fully activated to bind to the micropollutants. Therefore, the photodegradation yields decreased. The optimum concentrations of nano-TiO₂ and nano-CeO₂ were useful both from the economic and the photodegradation efficiency standpoint, because an increase in opacity and light scattering of TiO₂ particles occurred at high concentrations, leading to the decreased passage of UV irradiation through the suspension (Bethi et al. 2016). By increasing the catalyst concentration to an optimum level, the surface area was increased, leading to an increase in the production of reactive species. However, increased nanoparticle concentrations would also induce greater aggregation of the catalyst and decrease the total active surface area, thereby leading to a reduction in the photocatalytic treatment efficiency. The Kruskal–Wallis one way test showed that significant effects of nanoparticle concentration on the removal of micropollutants were observed as the nanoparticle concentration was raised from 0.25 to 0.50 g L⁻¹ (Kruskal–Wallis ANOVA, χ² = 1.07, p < 0.05). Any further increase of nano-TiO₂ and nano-CeO₂ concentrations (from 0.5 to 1.5 g L⁻¹) did not have a significant effect on micropollutant removals (Kruskal–Wallis ANOVA, χ² = 17.31, p > 0.05).

Corena studied the effect of TiO₂ concentrations on the photocatalytic removal of GFZ at a pH of 5.00 for an initial GFZ concentration of 2 mg L⁻¹ at 100 W UV light power. The increase in TiO₂ concentration from 0.10 g L⁻¹ to 1.50 g L⁻¹ affected GFZ removal positively because greater surface area and more active sites became available for TiO₂. However, an increase in TiO₂ concentration from 1.50 g L⁻¹ to 2.50 g L⁻¹ decreased the ability of the UV radiation to reach the TiO₂ surface because a portion of the UV light was reduced and scattered by the TiO₂ nanoparticles (Corena 2015). A concentration of 0.50 g L⁻¹ nano-CeO₂ gave better results for the photocatalytic removal of GFZ (89%), TCS (92%), PBT (81%) and PBEB (79%) than nano-TiO₂ with 0.50 g L⁻¹ at 120 W power at 25 °C and at a pH of 7.00 (Figure 5(d)). This can be attributed to the higher specific surface area of nano-CeO₂ (302 m² g⁻¹) with smaller particle sizes compared to TiO₂ (205 m² g⁻¹). Balavi et al. showed that the photocatalytic activity characteristic of CeO₂ results in it being a proper alternative instead of TiO₂ and ZnO in photocatalytic removal of some colored wastewaters (Balavi et al. 2013).

During photodegradation of GFZ, TCS, PBT and PBEB, semiconductor nano-TiO₂ and nano-CeO₂ were excited by photons (hv) and produced pairs of electrons (⁠⁠) and holes (⁠⁠) in the conduction and valance bands, respectively. The photogenerated electrons (⁠⁠) and holes (⁠⁠) were able to react with O₂ and OH^- to generate OH^•. OH^• tended to attack the carbon atoms with the highest electron density (Yu et al. 2006). The steps given below removed the micropollutants from the raw hospital wastewater (Equations (3)–(24)).

• (a)
  Generation of the electron (⁠⁠) – hole (⁠⁠) pairs:

  

  (3)
• (b)
  Keeping the holes at the valance band:

  

  (4)

  

  (5)
• (c)
  Electron (⁠⁠) transfer to the conduction band:

  

  (6)

  

  (7)

  

  (8)

  

  (9)

• (d)
  Oxidation of TCS: The phenol-type intermediates (benzoic acid and phenol) of TCS occurred during photocatalysis due to the formation of OH^• (Son et al. 2009). Photodegradation of TCS to CO₂ and H₂O occurred via OH^• according to the following equations (Equations (10)–(13)):

  

  (10)

  

  (11)

  

  (12)

  

  (13)
• (e)
  Oxidation of GFZ: Photodegradation of GFZ to CO₂ and H₂O occurred via OH^• and ^•O₂⁻ according to the following equations (Equations (14)–(18)) (Chen et al. 2017):

  

  (14)

  

  (15)

  

  (16)

  

  (17)

  

  (18)
• (f)
  Mineralization of PBT and PBEB to CO₂ and H₂O via OH^• and ^•O₂⁻ occurred via exciting the bromines according to the following equations, respectively (Equations (19)–(22)) (Chen et al. 2012):

  

  (19)

  

  (20)

  

  (21)

  

  (22)
• (g)
  Conjunction of the charge conductors (electron (⁠⁠) – hole (⁠⁠) pairs get together and the reaction ends):

  

  (23)

  

  (24)

The photocatalytic reactor was operated with a 0.50 g L⁻¹ nanoparticle (nano-TiO₂ and nano-CeO₂) concentration for 45 min UV irradiation at 25 °C and at a pH of 7.00 at increasing UV light powers (120, 210 and 300 W) in order to determine the optimum UV light power for the maximum micropollutant photodegradation. The micropollutant photodegradations were obtained as 84%, 88%, 78% and 75%, respectively, for 120 W UV light power after 45 min irradiation time at 0.50 g L⁻¹ nano-TiO₂ (Figure 5(e)). When the UV light power was increased to 210 W, the micropollutant photodegradation removals reached 90%, 92%, 88% and 83%, respectively, at the same irradiation time (Figure 5(e)). As the UV light power was increased to 300 W the photodegradation yields for GFZ, TCS, PBT and PBEB decreased slightly (Figure 5(e)). In this case, more hydroxyl radicals were available to attack the aromatic rings and the rate of removal increased. For the photocatalytic experiments carried out with nano-CeO₂, the micropollutant photodegradation yields were obtained as 89%, 92%, 81% and 79%, respectively, at 120 W UV light power (Figure 5(f)). The micropollutant photodegradation yields increased significantly when the UV light power was increased to 210 W (Figure 5(f)). Finally when the UV light power was increased to 300 W the photodegradation yields for GFZ, TCS, PBT and PBEB decreased slightly (Figure 5(f)). The optimum UV light power for maximum micropollutant photo-removals was found to be 210 W at 0.50 g L⁻¹ nano-TiO₂ and nano-CeO₂ concentrations after 45 min irradiation time at 25 °C at pH = 7.00. Kruskal–Wallis test statistics showed that the UV power significantly increased micropollutant removals, up to 210 W UV power (ANOVA, χ² = 2.31, p < 0.05). UV power higher than 210 W decreased micropollutant removal ((ANOVA, χ² = 14.88, p>0.05).

It should be pointed out that the photodegradation rates of micropollutants mentioned above increased when the UV light irradiation power was increased from 120 W up to 210 W. After 210 W UV power had been reached for a certain period of time, the margin of increase dropped. At this UV power, formation of long-lived intermediate by-products did not accumulate in the solution before their reaction with hydroxyl radical and photo-transformation into carbon dioxide and water (Bai & Acharya 2016). This is the reason for the high photodegradation yields of micropollutants found in this study. The removal yields of the micropollutants mentioned above during the photocatalysis process increased linearly when the UV light power was increased from 120 W to 210 W. The multiple regression analysis showed that a significant linear correlation between GFZ, TCS, PBT, and PBEB photodegradation yields and UV power up to 300 W (R = 0.86) and this regression was significant (P = 0.53, α = 0.05). As the UV power was increased to 300 W, the electron (⁠⁠) – hole (⁠⁠) recombination causes lower effects on the photo-removal yields since some metabolites accumulated and this decreased effective OH radical production, as reported by Caudillo-Flores et al. (2016). Ma et al. (2016) and Chen et al. (2017) found similar results for GFZ photodegradation under 230 W and 298 W powers. Chen et al. (2016a, 2016b) mentioned that for a maximum photodegradation kinetic constant an optimum UV power (267 W) was necessary. Covaci et al. (2011) and Ezechiás et al. (2014) mentioned that optimum UV powers (180–190 W) were necessary for the cleavage of brominated rings in PBEB and PBT from NBFRs since these organics have low volatility, low water solubility, and high log K_ow levels.

In photocatalysis, pH can affect the reaction of OH radical production from H₂O/OH⁻ and nano-TiO₂. The photodegradation rates of GFZ, TCS, PBT and PBEB were researched under acidic (pH = 4.00), neutral (pH = 7.00) and alkaline pH (pH = 8.50) conditions after 45 min irradiation time at 210 W UV light power with a nanoparticle concentration of 0.50 g L⁻¹ nano-TiO₂ at 25 °C. Under acidic pH (4.00) GFZ, TCS, PBT and PBEB photodegradation yields were 47%, 50%, 40% and 35%, respectively (Figure 5(g)). The photocatalytic yields at neutral pH (7.00) increased (Figure 5(g)). Under alkaline conditions (8.50), GFZ, TCS, PBT and PBEB photodegradation yields reached the maximum (93%, 94%, 90% and 87%, respectively) under the same experimental conditions with the nano-TiO₂ (Figure 5(g)).

The surface of the photocatalyst may become positively or negatively charged depending on the ambient pH. TiO₂ has a point of zero charge at around pH = 6 (Son et al. 2009). The surface of the photocatalyst will be positively charged in solutions below this pH and negatively charged above. Therefore, the pH will impact the adsorption of micropollutants on the TiO₂ surface depending on their chemical structure. In this study, it was found that the most efficient pH was 8.50 for this nanoparticle whereas more extreme pH conditions negatively affected the photodegradation yields. This is due to the chemical structure of GFZ and TCS at pH = 8.50. Both the phenoxy and chloro groups on chloro-phenol moieties in GFZ (dissociation constant pK_a = 8.6) and dimethyl and phenoxy moieties (pK_a = 6.1) in TCS are deprotonated with the overall charge being negative (Son et al. 2009). Therefore, because of negative surface charge of TiO₂, maximal photodegradation occurs. Similarly, at low pH, both these functional groups are protonated, and hence the overall positive charge of GFZ and TCS results in low photodegradation yields as reported by Martínez-Zapata et al. (2013). Molinari et al. studied the effect of pH on the photocatalytic degradation of GFZ with UV/TiO_2. They observed that the maximum GFZ photodegradation yield was at alkaline pH values since the TiO₂ surface was charged negatively and the carboxylic group of GFZ was in the anionic form causing a repulsion phenomenon with the TiO₂ nanoparticle (Molinari et al. 2008). The Kruskal–Wallis test showed that the effect of pH on micropollutant photodegradation yields was significant (ANOVA, χ² = 2.03, p < 0.05).

The GFZ, TCS, PBT and PBEB photodegradation yields were lower (48%, 52%, 43% and 37%, respectively) in acidic pH (4.00) compared to neutral pH (7.00) conditions at 0.50 g L⁻¹ CeO₂ concentrations for the same operational conditions given for TiO₂ (Figure 5(h)). The photocatalytic yields of GFZ, TCS, PBT and PBEB increased to 95%, 97%, 92% and 90%, respectively, at a pH of 8.50 (Figure 5(h)). The reason for the increase of the photodegradation yields of the four micropollutants at pH = 8.50 can be explained by the hydroxyl radicals being more readily produced at higher pH values. As a result, the larger amount of hydroxyl radicals produced increased the photocatalytic degradation removals, as reported by Chow et al. (2012). Son et al. reported that the decreasing of photocatalytic removals under acidic conditions is due to inhibition of hydroxyl radical generation (Son et al. 2009). Since the pH of the raw hospital wastewater was 8.50, no pH adjustment was needed to achieve maximum GFZ, TCS, PBT and PBEB photodegradation yields with nano-TiO₂ and nano-CeO₂. This would decrease the cost of chemicals used to adjust the pH of hospital wastewater.

In order to decrease the catalyst cost, recoveries of nano-TiO₂ and nano-CeO₂ were researched in this study. Six sequential treatment steps were investigated in order to determine possible reusability of the TiO₂ and CeO₂ nanoparticles. The TiO₂ and CeO₂ nanoparticles were reused at the optimum experimental conditions (45 min irradiation time, 0.50 g L⁻¹ nanoparticle concentration, 210 W UV light power, pH = 8.50 at 25 °C). TCS, GFZ, PBT and PBEB removal efficiencies were taken into consideration to detect the reusability of nano-TiO₂ and nano-CeO₂ throughout the sequential six treatments. After the first treatment step with nano-TiO₂, the maximum removal efficiencies of TCS, GFZ, PBT and PBEB were determined as 94%, 93%, 90% and 87%, respectively (Table 4). After the second treatment step, the removal efficiencies of TCS, GFZ, PBT and PBEB were determined as 92%, 91%, 90% and 86%, respectively (Table 4). The removal efficiencies of TCS, GFZ, PBT and PBEB were obtained as 88%, 88%, 85% and 81%, respectively, after the fifth treatment step. In the sixth treatment step, the removal efficiencies decreased to 80%, 78%, 75% and 74% for TCS, GFZ, PBT and PBEB, respectively (Table 4).

The maximum removal efficiencies of TCS, GFZ, PBT and PBEB were determined as 97%, 95%, 92% and 90%, respectively, after the first treatment step with nano-CeO₂ (Table 4). After the second treatment step, removal efficiencies of TCS, GFZ, PBT and PBEB were obtained as 96%, 93%, 91% and 90%, respectively (Table 4). After the fifth treatment step, the removal efficiencies of TCS, GFZ, PBT and PBEB were determined as 92%, 89%, 86% and 85%, respectively. The removal efficiencies decreased to 86%, 83%, 80% and 79% for TCS, GFZ, PBT and PBEB, respectively after the sixth treatment step (Table 4). The results showed that the nano-TiO₂ and nano-CeO₂ can be used effectively for five sequential runs (the lowest removal efficiency was 81% with nano-TiO₂ and it was 85% with nano-CeO₂ for PBEB at the end of the fifth treatment step) for the photocatalytic treatment of TCS, GFZ, PBT and PBEB from raw hospital wastewater to decrease the cost of treatment.

A cost analysis was carried out for the photocatalytic treatment of TCS, GFZ, PBT and PBEB from 1 m³ raw hospital wastewater at optimum experimental conditions (45 min irradiation time, 500 g L⁻¹ nanoparticle concentration, 210 W UV light power, pH = 8.50 at 25 °C) with nano-TiO₂ and nano-CeO₂. The UV power required was provided using 20 UV lamps. Also, electricity consumption and nanoparticle synthesis costs were considered. The total cost of the photocatalytic treatment of TCS, GFZ, PBT and PBEB from 1 m³ raw hospital wastewater was found as 8.70 € and 2.28 € at the optimum experimental conditions with nano-TiO₂ and nano-CeO₂, respectively (Table 5). For both nanoparticles, 4 × 10⁻³ € was spent on 20 UV lamps and 9.5 × 10⁻² € was determined as the investment cost. The chemical cost of synthesizing the TiO₂ nanoparticle and electricity consumption cost were only 8.60 €, while the chemical cost to synthesize the CeO₂ nanoparticle and electricity consumption cost were only 2.18 €. The cost analysis for the photodegradation of TCS, GFZ, PBT and PBEB from raw hospital wastewater showed that the photocatalytic removal with nano-CeO₂ is cheaper compared to nano-TiO₂. In this study, the main part of the cost consisted of the chemicals (8.57 € for nano-TiO₂ synthesis and 2.15 € for nano-CeO₂ synthesis). Conducting the studies in sunlight significantly decreased the cost of the study. Since these studies were performed during January and February in the laboratory, the power of the sun was low. Furthermore, a UV light simulator was not used. Preliminary studies showed that photocatalytic experiments' performed in sunlight decreased the total cost by 76%.

No recent literature was found concerning the photocatalytic cost of the micropollutants studied. However, some costs were compared with our data. Zhang et al. (2011) calculated that 99 € was needed to treat 500 m³ of wastewater containing 34 mg L^-1 methyl orange dye using only 1.9 g L⁻¹ TiO₂. Fenoll et al. (2012) and Wandre et al. (2016) found that the cost of photodegrading 1 m³ of wastewater containing 12 mg L⁻¹ TCS and 20 mg L⁻¹ methyl orange was 102 € and 127 €, respectively, in the presence of 2 g L⁻¹ nano-TiO₂ and CeO₂. These costs are higher than our cost of treating the raw wastewater containing the four micropollutants.

In this study the nano-TiO₂ and nano-CeO₂ were produced under laboratory conditions to photodegrade TCS, GFZ, PBT and PBEB from raw hospital wastewater for the first time in Turkey. XRD analysis showed that the TiO₂ nanoparticle completely originated from the pure anatase phase with a tetragonal crystal structure. The XRD patterns of CeO₂ showed that they had a cubic fluorite structure and exhibited strong agglomeration. The FT-IR spectrum of the TiO₂ nanoparticle showed the bending vibration of OH molecules, the carbonyl peak and stretching vibrations of Ti-OH. The FT-IR spectrum peaks of CeO₂ nanoparticles originated from the (Ce-O) metal–oxygen bond vibrations.

Under optimized operational conditions (45 min irradiation time, 0.50 g L⁻¹ nanoparticle concentration, 210 W UV light power, pH = 8.50 at 25 °C) the nano-CeO₂ had a higher photocatalytic activity (97% TCS, 95% GFZ, 92% PBT and 90% PBEB) than nano-TiO₂ (94% TCS, 95% GFZ, 90% PBT and 87% PBEB) to remove these micropollutants from raw hospital wastewater since its pore volume, average size and pore diameter were low with high band gap energy. The slightly lower alkaline pH (8.50) maximized the photo-removals of micropollutants from the raw hospital wastewater. TiO₂-photocatalytic efficiency decreased in acidic conditions by inhibiting the generation of OH radicals. The nano-CeO₂ and nano-TiO₂ used in this study were recovered effectively to be used again with photodegradation yields as high as 85–94% and 83–90%, respectively. The cost of treating the TCS, GFZ, PBT and PBEB from 1 m³ raw hospital wastewater was 2.28 € with CeO₂ nanoparticles. The results of this study showed that TCS, GFZ, PBT and PBEB in raw hospital wastewaters can be effectively removed with photodegradation by the utilization of nano-CeO₂ and nano-TiO₂ generated under laboratory conditions for the first time in Turkey. This process could be used effectively in other pharmaceutical and hospital wastewater studies, both as pilot projects and as real treatment in the field.

The authors would like to thank TUBITAK for the PhD dissertation grant given to Gokce Guney on the Priority Section Subjects (Grant number 2211-C Prior Sections; Water; Prior, Specific and Micro Pollutants).