Water Research xxx (2015) 1e10

Contents lists available at ScienceDirect

Water Research journal homepage: www.elsevier.com/locate/watres

Substrate-immobilized electrospun TiO2 nanofibers for photocatalytic degradation of pharmaceuticals: The effects of pH and dissolved organic matter characteristics Sung Kyu Maeng a, Kangwoo Cho b, Boyoung Jeong b, Jaesang Lee c, Yunho Lee d, Changha Lee e, Kyoung Jin Choi f, Seok Won Hong b, * a

Department of Civil and Environmental Engineering, Sejong University, 98 Gunja-Dong, Gwangjin-Gu, Seoul 143-747, Republic of Korea Center for Water Resource Cycle Research, Korea Institute of Science and Technology (KIST), Hwarangno 14-gil 5, Seongbuk-gu, Seoul 136-791, Republic of Korea c Civil, Environmental, and Architectural Engineering, Korea University, 145, Anam-ro, Seongbuk-gu, Seoul 136-701, Republic of Korea d Department of Environmental Science and Engineering, Gwangju Institute of Science and Technology (GIST), Gwangju 500-712, Republic of Korea e School of Urban and Environmental Engineering, and KIST-UNIST-Ulsan Center for Convergent Materials (KUUC), Ulsan National Institute of Science and Technology, Ulsan 698-805, Republic of Korea f School of Materials Science and Engineering, and KIST-UNIST-Ulsan Center for Convergent Materials (KUUC), Ulsan National Institute of Science and Technology, Ulsan 698-805, Republic of Korea b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 3 March 2015 Received in revised form 15 May 2015 Accepted 17 May 2015 Available online xxx

A substrate-immobilized (SI) TiO2 nanofiber (NF) photocatalyst for multiple uses was prepared through electrospinning and hot pressing. The rate of furfuryl alcohol degradation under UV irradiation was found to be the highest when the anatase to rutile ratio was 70:30; the rate did not linearly increase as a function of the NF film thickness, mainly due to diffusion limitation. Even after eight repeated cycles, it showed only a marginal reduction in the photocatalytic activity for the degradation of cimetidine. The effects of pH and different organic matter characteristics on the photodegradation of cimetidine (CMT), propranolol (PRP), and carbamazepine (CBZ) were investigated. The pH-dependence of the photocatalytic degradation rates of PRP was explained by electrostatic interactions between the selected compounds and the surface of TiO2 NFs. The degradation rates of CMT showed the following order: deionized water > L-tyrosine > secondary wastewater effluent (effluent organic matter) > Suwannee River natural organic matter, demonstrating that the characteristics of the dissolved organic matter (DOM) can affect the photodegradation of CMT. Photodegradation of CBZ was affected by the presence of DOM, and no significant change was observed between different DOM characteristics. These findings suggest that the removal of CMT, PRP, and CBZ during photocatalytic oxidation using SI TiO2 NFs is affected by the presence of DOM and/or pH, which should be importantly considered for practical applications. © 2015 Elsevier Ltd. All rights reserved.

Keywords: Organic matter characteristics pH Pharmaceuticals Photocatalytic oxidation Substrate-immobilized TiO2 nanofibers

1. Introduction Because photo-driven reduction and oxidation reactions are initiated on a titanium dioxide (TiO2) surface, TiO2 photocatalysis offers various application opportunities, including photosynthesis, photovoltaic cells, electro- and photochromics, water splitting, and €tzel, 1995; Onozuka et al., sensing (Gr€ atzel, 2001; Hagfeld and Gra

* Corresponding author. E-mail address: [email protected] (S.W. Hong).

2006; Wang et al., 2011). The photocatalytic activity of TiO2 to produce an electronehole pair upon UV absorption leads to the
generation of hydroxyl radicals ( OH) as powerful and low-selective oxidants under ambient conditions, thus enabling rapid degradation and mineralization of a wide range of organic pollutants and natural organic matter (Fujishima et al., 2008; Lee et al., 2010; Ramasundaram et al., 2015; Uyguner-Demirel and Bekbolet, 2011). Most of the previous studies reported heterogeneous photocatalysis using fine particles of semiconductors such as TiO2 for removing contaminants from water; however, the use of heterogeneous catalysts requires an additional separation step to remove

http://dx.doi.org/10.1016/j.watres.2015.05.032 0043-1354/© 2015 Elsevier Ltd. All rights reserved.

Please cite this article in press as: Maeng, S.K., et al., Substrate-immobilized electrospun TiO2 nanofibers for photocatalytic degradation of pharmaceuticals: The effects of pH and dissolved organic matter characteristics, Water Research (2015), http://dx.doi.org/10.1016/ j.watres.2015.05.032

2

S.K. Maeng et al. / Water Research xxx (2015) 1e10

€ and Matilainen, the fine catalyst particles from water (Sillanp€ aa 2014). Membrane separation process such as ultrafiltration are used to separate TiO2 slurry from water (Lee et al., 2001), but the removal of fine TiO2 particles can cause membrane fouling. Despite the high catalytic longevity and reliability of TiO2, the separation and recovery processes to enable multiple-use scenarios have always been recognized as a challenge for the practical application of TiO2-mediated photocatalysis in water treatment processes. Moreover, recent findings that metal oxide nanoparticles cause adverse biological effects in aquatic environments suggest a need to post-process for minimal environmental release of TiO2 after the application (Barmo et al., 2013; Zhu et al., 2009, 2011). In order to resolve these drawbacks, various strategies to immobilize (nano- or micro-sized) TiO2 particulates on the substrate surface have been used such as the solegel method, thermal treatments, chemical vapor deposition and electrophoretic deposition. Among the various shapes of TiO2, e.g., nanoporous materials, nanowires, nanorods, nanoparticles (NPs), and nanofibers (NFs), the fiber-like one-dimensional nanostructure is more advantageous over other TiO2 shapes because of the enhanced catalytic activity resulting from the increase in the catalyst surface area per unit volume (Kim et al., 2009) and an easier separation from water, mainly attributable to its specific morphology (Zhang et al., 2010) together with nanotube arrays (Mor et al., 2005). Electrospinning, a technique for producing long fibers of nano-to micrometer diameter, has received great attention in the recent past because of its simplicity, versatility, low cost, and scalability for industrial scale production (Caruso et al., 2001; Zhang et al., 2002). Since the fabrication of TiO2 NFs by electrospinning was first reported in 2003 (Li and Xia, 2003), numerous types of TiO2 NFs have been developed; for example, polypyrrole-decorated AgeTiO2 NFs (Yang et al., 2013), iron phthalocyanine-TiO2 NF heterostructures (Guo et al., 2012), CdS nanoparticles sensitized electrospun TiO2 NFs (Wei et al., 2014), Au co-catalyzed TiO2 NFs (Nalbandian et al., 2015), and hollow mesoporous TiO2 NFs with an increased catalytic activity compared to that shown by TiO2 NPs (Zhang et al., 2012). Although all catalysts are expected to be reusable for multiple cycles, few attempts have been made to immobilize TiO2 NFs onto substrate. Heterogeneous photocatalysis is an effective method to remove organic micropollutants and dissolved organic matter (DOM) from €, 2010; Rivera-Utrilla et al., 2013; water (Matilainen and Sillanp€ aa Valencia et al., 2013). In the last few decades, heterogeneous photocatalysis using different semiconductors such as TiO2, ZnS, ZnO, SrTiO3, and WO3 has been extensively studied by the scientific €€ community (Sillanpa a and Matilainen, 2014). The removal of pharmaceuticals via heterogeneous photocatalysis is influenced by many factors such as catalyst concentration, light wavelength, water matrix, pH, and DOM concentration. Previous studies have reported that several factors (e.g., natural organic matter, pH, temperature, suspended solids, and alkalinity) influence the effi€nder, 2003). ciencies of advanced oxidation processes (Oppenla Choi et al. (2014) recently used heterogeneous photocatalysis for removing of pharmaceuticals from secondary wastewater effluent (SWE), giving more focus to the quantity of DOM. However, in spite of the above-mentioned studies, little is known about the effects of different characteristics of DOM on photocatalytic oxidation. Heterogeneous photocatalysis can attenuate organic micropollutants such as pharmaceuticals in wastewater before discharging effluents in aquatic environments or wastewaterimpacted rivers. Since the photocatalytic degradation of organic micropollutants is greatly affected by the background organic matter (Autin et al., 2013), it is important to monitor effluent organic matter (EfOM) in SWE. Choi et al. (2014) recently reported that the removal of pharmaceuticals from SWE is inversely related

to the concentration of organic carbon during heterogeneous photocatalytic degradation using TiO2/UV-A, TiO2/UV-C, and H2O2/ UV-C; however, detailed characteristics of EfOM in SWE were not provided. It is important to investigate EfOM characteristics in SWEs since wastewaters are discharged from different origins. Moreover, it is well known that EfOM in SWEs acts as a precursor of certain nitrogenous disinfection by-products (N-DBPs) during oxidative water treatment (Chon et al., 2013); this phenomenon becomes especially important where secondary and tertiary wastewater effluents are used for indirect potable reuse (e.g., groundwater replenishment systems). Toward these ends, we prepared TiO2 NFs by electrospinning the precursor solution, followed by hot pressing to enhance the adhesion of TiO2 NF films to a substrate for the purpose of increasing the reusability of photocatalysts without an additional separation process. The optimum preparation conditions of these photocatalysts were determined by evaluating the photocatalytic oxidation rate of furfuryl alcohol (FFA) under UV irradiation. The effect of pH on the degradation of three pharmaceuticalsdcimetidine (CMT), propranolol (PRP), and carbamazepine (CBZ)dduring photocatalytic oxidation was studied using substrate-immobilized (SI) TiO2 NFs. To the best of our knowledge, few attempts have been made to investigate the change in the DOM characteristics as well as the removal efficacy of pharmaceuticals using tailor-made photocatalysts such as SI TiO2 NFs. Since the properties of DOM are diverse, we have examined the photo-degradation rates of CMT, PRP, and CBZ in two different surrogates of DOM to understand the role of the background organic matter; we also compared with those rates in SWE. In this study, L-tyrosine and Suwannee River natural organic matter (SRNOM) are used as NOM surrogates; Ltyrosine is an amino used and is used as the starting material of humic-like compounds that form under natural conditions (Bosetto et al., 1997) and via a photo-humification process (Bianco et al., 2014). 2. Materials and methods 2.1. Chemicals CMT, PRP, and CBZ of analytical grade were purchased from SigmaeAldrich, Seoul, Korea. These target pharmaceuticals were chosen because of their frequent occurrence in the aquatic environments and physicochemical properties. Table S1 shows the physicochemical characteristics of these selected pharmaceuticals. Stock solution of pharmaceuticals, SRNOM, and L-tyrosine (SigmaeAldrich, Seoul, Korea) were prepared in deionized water (DIW) using a 0.45 mm PTFE membrane (Millipore, USA). 2.2. Preparation and characterization of substrate-immobilized TiO2 nanofibers TiO2 NFs were prepared on quartz substrates by electrospinning a mixture of acetic acid, titanium (IV) isopropoxide (TiP, SigmaeAldrich, Seoul, Korea), ethanol, and polyvinylpyrrolidone (PVP; Mw z 1,300,000, SigmaeAldrich, Seoul, Korea), followed by hightemperature annealing. The precursor gel was prepared by dissolving 2 g of TiP and 1 g of PVP in a solution of 14 mL of ethanol and 4 mL of acetic acid under vigorous stirring at 50  C for 30 min. The resulting solution was loaded into a plastic syringe connected to a spinneret (a stainless needle) having a diameter of 200 mm. Prior to electrospinning, a thin layer of Ti metal, ca. 100 nm thickness, was deposited on a ground quartz substrate by electron-beam evaporation in order to make it conductive. Then, the Ti-precursor/ polymer solution was electrospun onto a quartz substrate (2  2 cm2) under the following conditions: positive voltage, 20 kV;

Please cite this article in press as: Maeng, S.K., et al., Substrate-immobilized electrospun TiO2 nanofibers for photocatalytic degradation of pharmaceuticals: The effects of pH and dissolved organic matter characteristics, Water Research (2015), http://dx.doi.org/10.1016/ j.watres.2015.05.032

S.K. Maeng et al. / Water Research xxx (2015) 1e10

tip-to-collector distance, 10 cm; and solution flow rate, 3 mL/h. Hot pressing was employed in order to enhance the adhesion of TiO2 NF films to the quartz substrate under a pressure of 10e15 MPa at 170  C for 10 min. Finally, the conversion of TiP to crystalline TiO2 and the complete removal of PVP from the as-electrospun NFs were carried out by high-temperature calcination at 500e900  C for 1 h in air. The morphologies of both Ti-precursor/PVP NFs and calcined SI TiO2 NFs were studied by field-emission scanning electron microscopy (FE-SEM, NOVA NanoSEM-200, FEI Company, USA). The crystallinity of TiO2 NFs calcined at different temperatures was investigated by X-ray diffraction (XRD) analysis. The XRD patterns were recorded on a Bruker D8 Advance X-ray diffractometer (Bruker AXS GmbH, Germany) using a Cu Ka radiation source (20 kV and 20 mA). 2.3. Photocatalytic experiments All the photocatalytic experiments were carried out in the batch mode under air-equilibrated conditions at room temperature (23 ± 2  C). A 60-mL cylindrical quartz reactor was placed in a black acrylic box equipped with a six 4-W commercial black light blue lamp (BLB lamp, Philips TL4W). The emission wavelength region ranged from 350 to 400 nm and the light intensity of all the six lamps was determined to be 3.33  104 E min1L1 using ferrioxalate actinometry. Two pieces of 2  2 cm2 quartz substrates loaded with TiO2 NFs were vertically immersed into the quartz reactor; each side of the substrates loaded with TiO2 NFs faced the light source. CMT, PRP, and CBZ as well as FFA were introduced into a magnetically-stirred reactor before initiating the experiments. Since FFA is known to be weakly adsorbed onto the surface of TiO2, it is used as a free OH radical probe compound (Cunningham and k, 1994). The reaction solution containing 0.1 mM FFA was Sedla buffered at pH 7.5 by using 10 mM phosphate. A certain amount of the stock solution was spiked to bring the initial concentration to 10 mM of the selected pharmaceuticals in DIW, SRNOM, and Ltyrosine solution. Sample aliquots of 1 mL were withdrawn from the photo-reactor at several time intervals and filtered through a 0.45-mm PTFE membrane (Millipore). Then, the filtered solution was directly transferred to a 2-mL amber vial for HPLC analysis. The SWE samples collected from a local municipal wastewater treatment plant (Jungnang, Seoul, Korea) were used, and the selected pharmaceuticals were spiked to obtain an initial concentration of 10 mM. The DOC concentrations of L-tyrosine, SRNOM, and SWE were adjusted to approximately 5.0 mg/L in order to investigate the effects of DOM characteristics. 2.4. Sample analysis Quantitative analyses of the initial and residual concentrations of the target compounds were performed by HPLC (Shimazdu, LC20AD) equipped with a C-18 column (ZORBAX Eclipse XDB-18) and a UVevis detector (SPD-20AV). The mobile phase was a binary mixture of 0.1% (v/v) aqueous phosphoric acid solution and methanol or acetonitrile (typically 70: 30 v/v). The analysis of FFA and carbamazepine was performed using the mobile phase of a 40% aqueous methanol and a mixture of 60% acetonitrile and 40% water, respectively. For HPLC analyses, the detection wavelength was set at 230, 270, or 290 nm based on the compound used. The DOM characteristics were determined by using dissolved organic carbon (DOC), specific UV absorbance (SUVA), liquid chromatography organic carbon detection (LC-OCD), and fluorescence excitation-emission matrix (EEM). The concentration of DOM was determined by using a total organic carbon analyzer (Shimadzu

3

TOC-VCPN, Japan) and was represented as DOC. Specific UV absorbance (SUVA), the DOC to UV absorbance at 254 nm, was used to determine the relative aromaticity of the DOM during photocatalytic degradation. The LC-OCD (DOC-LABOR, Germany) separates DOM in four fractions based on molecular size and weight: biopolymers (BP), humic substances (HS), building blocks (BB), and low-molecular-weight (LMW) neutrals. The LC-OCD can be used to effectively monitor the polar components of NOM with a lower SUVA (Huber et al., 2011). For EEM, all samples were adjusted to the DOC concentration of 1 mg/L by diluting the samples before measuring by using the LS50B spectrofluorometer (Perkin Elmer, USA). Fluorescence intensities were detected at emission wavelengths of 280e600 nm in 1 nm intervals and at different excitation wavelengths ranging from 200 to 400 nm in 10 nm intervals. Previous studies have defined four peak regions to characterize the DOM: tryptophan protein-like peak A (ex/em ¼ 270e280/ 320e350 nm), aromatic protein-like peak B (ex/em ¼ 220e240/ 320e350 nm), humic-like peak C (ex/em ¼ 330e350/ 420e480 nm), and fulvic acid-like peak D (ex/em ¼ 250e260/ , 2003; 380e480 nm) (Chen et al., 2003; Leenheer and Croue Yamashita and Tanoue, 2003). 3. Results and discussion 3.1. Morphology of TiO2 nanofibers and their crystalline structures Fig. S1(a) shows an SEM image of as-electrospun TiP-PVP composite NFs on the substrate. Most of the NFs were observed to possess structural uniformity with a diameter ranging from 200 to 500 nm. This observation suggests that a continuous injection of TiO2 sol dispersed in the polymer matrix has successfully occurred during electrospinning under the given condition. Fig. S1(b) shows the morphology of TiO2 NFs calcined at 600  C without hot pressing. After calcination, the diameter reduced to 80e150 nm, mainly due to the burn-off of the PVP polymer owing to the transformation of the precursor to crystalline TiO2. The morphology of TiO2 NFs present on the quartz substrate after hot pressing at 15 MPa and 170  C, followed by calcination at 600  C, is presented in Fig. S1(c). The hot-pressing step did not result in significant changes in the fiber diameter although a slight bending at the fiberefiber junctions was observed. Further increasing of the applied pressure up to 20 MPa led to too much increase in the compression, thereby enlarging the NFs size (Fig. S1(d)); this increase in the NF size probably decreased the surface-area-to-volume ratio of TiO2 NFs. The phase evolution and crystal structure of the electrospun TiO2 NFs were evaluated using XRD. No evidence of TiO2 nucleation was observed in the XRD pattern after hot pressing at 170  C (data not shown). Fig. S2(a) shows the XRD patterns revealing the phase evolution of TiO2 NFs annealed at different temperatures. As reported by Kumar et al. (2007), TiO2 NFs began to crystallize into the anatase structure at 500  C. The crystalline structure changed to the rutile phase at temperatures above 600  C, and to the monocrystalline rutile phase at 900  C. Fig. S2(b)e(d) shows highresolution SEM images of TiO2 NFs annealed at 500, 700, and 900  C, respectively. It can be clearly seen that the morphological evolution was accompanied with an increase in the annealing temperature. Although the samples annealed at 500  C have a relatively smooth surface, thermal annealing temperatures higher than 700  C resulted in a significant growth of discrete grains. The average grain size increased to ~50 and 150 nm by annealing at 700 and 900  C, respectively. From the XRD patterns, the ratio of rutile phase was calculated using the Spurr equation (Spurr, 1957). The fractions of the rutile content normalized to the sum of anatase and rutile phases at various calcination temperatures are summarized in Table 1.

Please cite this article in press as: Maeng, S.K., et al., Substrate-immobilized electrospun TiO2 nanofibers for photocatalytic degradation of pharmaceuticals: The effects of pH and dissolved organic matter characteristics, Water Research (2015), http://dx.doi.org/10.1016/ j.watres.2015.05.032

4

S.K. Maeng et al. / Water Research xxx (2015) 1e10

Table 1 Rutile fractions in electrospun TiO2 NFs as a function of annealing temperature and the corresponding pseudo-first-order rate constants for the photocatalytic degradation of FFA. Annealing temp. ( C)

500

600

700

800

900

Rutile fraction (%) Pseudo-first-order rate constants (min1)

0 0.0014

30 0.0026

75 0.0020

93 0.00098

100 0.0017

3.2. Photocatalytic activities of substrate-immobilized TiO2 nanofibers In Table 1, the oxidative power of SI TiO2 NFs annealed at different temperatures after the immobilization was evaluated as a rate of photocatalytic FFA oxidation. Either sorption in the dark or direct UV photolysis led to a negligible reduction in the FFA concentration (data not shown). Compared with phase-pure anatase and rutile TiO2 NFs, anatase-rutile mixed phase that effectively retards the recombination of photogenerated electronehole pairs (Riegal and Bolton, 1995) caused a more rapid decomposition of FFA. Furthermore, photocatalytic degradation showed the maximum efficacy with a 30% rutile fraction. The result is compatible with the earlier findings that an anatase to rutile ratio of 70:30 (approximately similar to that of commercial Degussa P25) is optimal for photocatalysis (Doh et al., 2008; Li et al., 2011). Lower activity of pure anatase TiO2 NFs than pure rutile is possibly attributable to poor crystallinity of the former associated with low processing temperatures. To investigate the effect of the film thickness, the photocatalytic activity of the SI TiO2 NFs (annealed at 600  C) was monitored as a function of the feeding volume of the Ti-precursor/polymer solution (Fig. 1). The inset shows a cross-sectional SEM image of the TiO2 NF film with a feeding volume of 2 mL. Based on the experimental results, the volume-to-thickness conversion factor was calculated to be 2.1 mm/mL. The photocatalytic activity linearly increased up to the feeding volume of 1 mL and then slowly increased afterward, which mainly resulted from the limited diffusion of FFA and/or limited light penetration through the TiO2 NF film on the quartz substrate. These results imply that the SI TiO2 NF photocatalyst has an optimal film thickness beyond which the

Fig. 1. Photocatalytic degradation of FFA as a function of the thickness of TiO2eNF films ([FFA]0 ¼ 10 mM). The inset shows the cross-sectional SEM image of the TiO2eNF film with a feeding volume of 2 mL (n ¼ 3). Each data point represents the mean of three replicates; error bars indicate the standard deviation of the mean.

photocatalytic activity shows no appreciable increase. The potential capability of the SI TiO2 NFs, annealed at 600  C with a feeding volume of 2 mL for the following experiments, as the photocatalytic system was explored based on its multiple reuses for the photocatalytic degradation of CMT. As shown in Fig. 2, the repetition test shows a marginal reduction in the photocatalytic activity, with k(CMT) ¼ 0.0123 min1 for the 1st cycle versus k(CMT) ¼ 0.0111 min1 for the 8th cycle. 3.3. Changes in the degradation of the selected pharmaceuticals at different pH values Changes in the pH values can affect the photocatalytic degradation rates of pharmaceuticals mainly by altering the electrostatic interactions of charged compounds with the amphoteric TiO2 NF surface (Kim et al., 2004). Fig. 3 shows the effects of pH 2, 5.8e6.7, and 10 on the degradation of (a) CMT, (b) PRP, and (c) CBZ using the SI TiO2 NFs in DIW under UV irradiation. When the pH is increased from 2 to 10, the pseudo-first-order rate constants for PRP degradation also increased from 0.0079 to 0.0226 min1. The degradation rate of PRP showed the highest enhancement, followed by CMT (slight enhancement, 0.0101 to 0.0164 min1) and CBZ (negligible change, 0.0158 to 0.0166 min1). The observed behaviors could be well explained by the pH-dependent electrostatic interaction between the TiO2 NF surface and the pharmaceuticals. The zeta potential measurement (Fig. S3) shows that the point of zero charge of the electrospun TiO2 NF was ~5. The pKa of the secondary amine moiety of PRP was 9.4, indicating that PRP existed predominantly as a positively charged species below pH 9.4. Similarly, the pKa of the imidazole moiety of CMT is 6.8; it becomes positively charged below pH 6.8 and neutral above pH 6.8. CBZ occurs as a neutral species in the entire pH range. The electrostatic repulsion of the positively charged PRP or CMT species from the positively charged TiO2 NF surface was responsible for the slow degradation of these compounds at acidic pH (pH ¼ 2). On the other hand, the

Fig. 2. Repetition tests of substrate-immobilized TiO2 NFs for the degradation of cimetidine under UV irradiation ([cimetidine]0 ¼ 10 mM).

Please cite this article in press as: Maeng, S.K., et al., Substrate-immobilized electrospun TiO2 nanofibers for photocatalytic degradation of pharmaceuticals: The effects of pH and dissolved organic matter characteristics, Water Research (2015), http://dx.doi.org/10.1016/ j.watres.2015.05.032

S.K. Maeng et al. / Water Research xxx (2015) 1e10

5

Fig. 3. Effect of pH on the degradation of cimetidine (a), propranolol (b), and carbamazepine (c) under UV irradiation ([pharmaceutical]0 ¼ 10 mM, n ¼ 3).

electrostatic attraction between the positively charged PRP and the increasing negative charge of the TiO2 NF surface appears to significantly enhance the degradation rate of PRP with increasing pH. The lower pKa value of CMT than that of PRP reduces the attraction between CMT and the TiO2 NF surface, which may explain less enhancement in the degradation rate of CMT with pH. The degradation rate of CBZ as a neutral species did not change with pH variation. These observations confirm that pH is a prominent factor in the water treatment for removing ionic pharmaceuticals using SI TiO2 NFs. 3.4. Effects of organic matter characteristics on the degradation of the selected pharmaceuticals Photocatalytic degradation of the selected pharmaceuticals by the SI TiO2 NFs was conducted both in the absence and presence of L-tyrosine and SRNOM (Fig. 4). When the SI TiO2 NFs were applied without photo-irradiation, the concentrations of the pharmaceuticals did not decrease at all in the aqueous solutions containing L-

tyrosine or SRNOM, which confirmed negligible pharmaceutical removal by adsorption. Under direct UV photolysis, CMT, PRP, and CBZ showed negligible decomposition, regardless of the presence of organic matter. In the absence of DOM, the photocatalytic oxidation of the three pharmaceuticals by SI TiO2 NFs proceeded at similar rates (k(CMT) ¼ 0.0125 min1; k(PRP) ¼ 0.0136 min1; k(CBZ) ¼ 0.0124 min1), which appears to be plausible because the reactivity of the OH radical is not sensitively dependent on the substrate. Fig. 4 shows that the addition of L-tyrosine and SRNOM as protein- and humic-like substances, respectively, kinetically retarded the photocatalytic degradation rates of the pharmaceuti
cals. This can be ascribed to the dual roles of organics: OH scavenging and competitive mass transport to the reactive surface [24]. The inhibitory effects of the model organics on pharmaceutical degradation were roughly in the order of CMT < PRP < CBZ. As discussed earlier, CMT and PRP are cationic species while CBZ is a neutral species under the bulk pH condition of these experiments. Therefore, the mass transport of CBZ would be governed by

Please cite this article in press as: Maeng, S.K., et al., Substrate-immobilized electrospun TiO2 nanofibers for photocatalytic degradation of pharmaceuticals: The effects of pH and dissolved organic matter characteristics, Water Research (2015), http://dx.doi.org/10.1016/ j.watres.2015.05.032

6

S.K. Maeng et al. / Water Research xxx (2015) 1e10

Fig. 4. Effect of organic matter characteristics on the degradation of cimetidine (a), propranolol (b), and carbamazepine (c) under UV irradiation ([pharmaceutical]0 ¼ 10 mM, [SRNOM]0 ¼ 5.3 mg/L, [L-tyrosine]0 ¼ 5.1 mg/L, n ¼ 3).

hydrophobic interactions rather than by electrostatic attraction. Considering the log D values shown in Table S1, it is expected that the hydrophobic interaction would play more important roles for CBZ (log D ¼ 2.28 at pH 5.5) rather than for CMT (log D ¼ 1.15 at pH 5.5). The overall degradation rates are controlled by mass transport, and the inhibition magnitude could be explained by the fact that L-tyrosine and SRNOM primarily impede the Langmuir kinetics associated with hydrophobic interactions. On the other hand, pharmaceutical degradations were affected more by SRNOM than by L-tyrosine, especially for CMT. The ratio of the pseudo-first-order rate constant for pharmaceutical degradation in the presence of L-tyrosine (kty) to that in the presence of SRNOM (ksr), (kty/ksr) was 2.39 for CMT, 1.28 for PRP, and 1.55 for CBZ. The degree of inhibition by L-tyrosine or SRNOM would be a combinational consequence of physicochemical characteristics such as reactivity with hydroxyl radical, hydrophobicity, and specific UV absorption. To this end, SUVA254 illustrates the degree of UV-absorbing aromatic structure in DOM which was even correlated with the reactivity of hydroxyl radicals (Sakkas et al., 2007).

SUVA254 of SRNOM (SUVA254 ¼ 4.09 L/mg m) was approximately 10 times as high as that of L-tyrosine (SUVA254 ¼ 0.41 L/mg m), reflecting a stronger quenching of OH radicals and hydrophobicity. Because of the presence of a large number of aromatic carbons (humic) in SRNOM, more loss of UV radiations could occur, thus reducing the degradation of the selected pharmaceuticals. The EEM of SRNOM and L-tyrosine clearly exhibited that SRNOM has fluorescence regions (region peaks C and D) for humic- and fulvic-like substances, whereas L-tyrosine shows no characteristic fluorescence response (Fig. S4). LC-OCD also illustrated that humic substance fractions in SRNOM was dominant (71%), and no humic-like substances were detected in the L-tyrosine solution (Fig. S5). 3.5. Changes in effluent organic matter characteristics of the secondary wastewater effluent As in the case with model compounds, photocatalytic degradation of pharmaceuticals was also kinetically retarded in SWE (Fig. 5) in terms of pseudo-first-order rate constants shown in

Please cite this article in press as: Maeng, S.K., et al., Substrate-immobilized electrospun TiO2 nanofibers for photocatalytic degradation of pharmaceuticals: The effects of pH and dissolved organic matter characteristics, Water Research (2015), http://dx.doi.org/10.1016/ j.watres.2015.05.032

S.K. Maeng et al. / Water Research xxx (2015) 1e10

7

Fig. 5. Effect of effluent organic matter characteristics on the degradation of cimetidine (a), propranolol (b), and carbamazepine (c) under UV irradiation ([pharmaceutical]0 ¼ 10 mM, n ¼ 3).

Table 2. In control experiments, the concentrations of the selected pharmaceuticals remained unchanged in the absence of UV irradiation. EfOM and particles in SWE would reduce the photocatalytic activity of the SI TiO2 NF and/or decrease UV light penetration. As shown in Table 2, the degradation rates of CMT followed the order of deionized water > L-tyrosine > SWE (EfOM) > SRNOM, and were affected by DOM characteristics. EfOM can contain both L-tyrosine

(soluble microbial products from biological wastewater treatment) and NOM (from drinking water sources) based on their origins. The highest photodegradation rate of PRP (0.0062 min1) was resulted from the electrostatic interactions between the positively charged PRP (pKa: 9.4) and the negatively charged TiO2 NF surface at pH 7.2 of SWE. The interferences in the SWE matrix were more distinct for CBZ, indicating that the hydrophobicity of EfOM is the primary

Table 2 Pseudo-first-order rate constants for the photodegradation of three pharmaceuticals, cimetidine, propranolol, and carbamazepine, in different solution matrices using substrate-immobilized TiO2 NFs under UV irradiation. Compounds

Cimetidine Propranolol Carbamazepine

Pseudo-first-order rate constants (k values, min1) Deionized water matrix

L-tyrosine

Secondary WW effluent matrix

Suwannee river NOM

0.0125 0.0136 0.0124

0.0074 0.0037 0.0017

0.0051 0.0062 0.0012

0.0031 0.0029 0.0011

Please cite this article in press as: Maeng, S.K., et al., Substrate-immobilized electrospun TiO2 nanofibers for photocatalytic degradation of pharmaceuticals: The effects of pH and dissolved organic matter characteristics, Water Research (2015), http://dx.doi.org/10.1016/ j.watres.2015.05.032

8

S.K. Maeng et al. / Water Research xxx (2015) 1e10

reason for the inhibition. Pseudo-first-order rate constants for CBZ were not significantly changed depending on DOM characteristics (SRNOM: 0.0017 min1, L-tyrosine: 0.0012 min1, and EfOM: 0.0011 min1); in fact, they were all retarded by the presence of DOM. Choi et al. (2014) reported that the rate of photocatalytic degradation of CBZ in SWE was more affected by the concentration of DOM compared to the rate affected by light sources or the initial CBZ concentrations. In order to keep the photocatalytic activity of the SI TiO2 NFs in the SWE, pretreatments such as ion exchange would be required to remove DOM, though it is generally impractical because of the high capital and operating costs. The interaction of background organic matter with the SI TiO2 NFs was further probed by using the EEM contour maps for SWE (Fig. 6). Photocatalytic treatment of SWE resulted in a significant decrease in the fluorescence intensities of peaks C and D, which is associated with the substantial loss of humic- and fulvic-like substances (fluorescent properties) during the photocatalytic oxidation by the SI TiO2 NFs. Such preferential removal of humic- and fulvic-like substances was similarly observed in the photocatalytic transformation of NOM present in water bodies (Hoffmann et al., 1995), which indicates a strong adhesion of these compounds to the SI TiO2 NFs via hydrophobic interactions. In addition, oxidation of SWE by the SI TiO2 NFs was accompanied by the maximum intensity shifts to shorter wavelengths in peak regions C and D. The

Fig. 7. Bulk organic matter characteristics determined by liquid chromatographyorganic carbon detection (biopolymers, humics, building blocks, and low-molecularweight neutrals) before and after the photo-treatment of secondary wastewater effluent with substrate-immobilized TiO2 NFs under UV irradiation (n ¼ 3).

blue shift indicates the fragmentation of larger molecules into smaller ones and the decomposition of condensed aromatic moieties, which increased the aliphatic hydrocarbon fraction (Hoffmann et al., 1995). Nevertheless, EEM measurements were not carried out to investigate the changes in the organic matter characteristics of SWE, SRNOM, and L-tyrosine with CMT, PRP, or CBZ. The detection methods used for the EEM are interfered by CMT, PRP, and CBZ because of their fluorescence characteristics. Fig. 7 compares the DOM concentrations of the four fractions (see Section 2.4) in the intact and photocatalytically treated wastewater effluents using LC-OCD. The concentrations of total organic carbon before and after the photocatalytic treatment confirm that no significant mineralization of EfOM occurred (