Hybrid visible-light responsive Al2O3 particles

Chemical Physics Letters 685 (2017) 416–421

Contents lists available at ScienceDirect

Chemical Physics Letters journal homepage: www.elsevier.com/locate/cplett

Research paper

Hybrid visible-light responsive Al2O3 particles Vesna Ðordevic´ a, Jasmina Dostanic´ b, Davor Loncˇarevic´ b, S. Phillip Ahrenkiel c, Dušan N. Sredojevic´ a,d, Nenad Švrakic´ d,e, Milivoj Belic´ d, Jovan M. Nedeljkovic´ a,⇑ a

Institute of Nuclear Sciences Vincˇa, University of Belgrade, P.O. Box 522, 11001 Belgrade, Serbia Institute of Chemistry, Technology and Metallurgy, University of Belgrade, Studentskitrg 12-16, 11000 Belgrade, Serbia c South Dakota School of Mines and Technology, 501 E. Saint Joseph Street, Rapid City, SD 57701, USA d Texas A&M University at Qatar, P.O. Box 23874, Doha, Qatar e Institute of Physics, University of Belgrade, Pregrevica 118, 11080 Belgrade, Serbia b

a r t i c l e

i n f o

Article history: Received 29 March 2017 In final form 7 August 2017 Available online 8 August 2017 Keywords: Inorganic/organic hybrids Charge transfer complex Visible light responsive material Al2O3 Photocatalysis

a b s t r a c t Detailed study of Al2O3, an insulator with the band gap of about 8.7 eV, and its different organic/inorganic charge transfer complexes with visible-light photo activity is presented. In particular, prepared Al2O3 particles of the size 0.1–0.3 lm are coated with several organic complexes – the specific details for catecholate- and salicylate-type of ligands are described below – and the light absorption properties and photocatalytic activity of such hybrids are scrutinized and compared with those of other organic/ inorganic hybrid materials previously studied. In addition, the obtained experimental results are supported with quantum chemical calculations based on density functional theory. Ó 2017 Elsevier B.V. All rights reserved.

1. Introduction Over the past decades, the metal oxides (TiO2, ZnO, SnO2 and CeO2), which are abundant in nature, biocompatible, and stable, have been broadly investigated for photocatalytic purposes [1–8]. The promising photocatalyst must have proper combination of electronic structure and light absorption properties, as well as efficient separation of photo-formed charge carriers. The efficient solar light utilization of oxide materials is hindered by their large band gap. For instance, the most thoroughly studied photocatalyst TiO2, with the band gap of Eg = 3.2 eV, at the high-frequency edge of the visible-light spectrum, absorbs less than 5% of the available solar light photons, allowing only UV photons (k < 380 nm) to produce electron-hole pairs and stimulate redox processes on the catalyst surface. There has been tremendous interest over the recent years to improve visible-light absorption of TiO2; the methods include dye sensitization for photo-excitation of TiO2 in the visible spectral region via photo-induced interfacial electron transfer [9– 11], doping with metal and nonmetal ions to promote less energetic excitation of electrons from mid-gap dopant levels to conduction band of TiO2 [12–17], and the use of plasmonic noble metal nanoparticles [18–20]. Another emerging approach to extend photo-absorption of TiO2 into a more practical range of solar spec⇑ Corresponding author. E-mail address: [email protected] (J.M. Nedeljkovic´). http://dx.doi.org/10.1016/j.cplett.2017.08.012 0009-2614/Ó 2017 Elsevier B.V. All rights reserved.

trum involves charge transfer (CT) from surface modifier into the conduction band of nanocrystalline TiO2 particles. The CT complex formation between surface Ti atoms and either catecholate- or salicylate-type of ligands, accompanied with red shift of absorption onset, has been primarily studied with colloidal TiO2 nanoparticles [21–28]. Although, the main purpose of extending the absorption spectrum of TiO2 towards the red spectral region is the usage of less energetic photons to drive photo-induced reactions, there is a lack of information regarding photocatalytic performance of surface-modified TiO2 particles. Recently, hydrogen evolution under visible light illumination of surface-modified TiO2 nanoparticles by catechol and its derivatives has been demonstrated [27,29], as well as degradation of organic dye crystal violet using commercial TiO2 powder (Degussa P25) modified with catechol [30], and degradation of methylene blue and crystal violet using surface modified TiO2 nanoparticles with ascorbic acid and dopamine, respectively [28,29]. In the present study we demonstrate that CT complex formation accompanied with the red shift of optical absorption peak is not exclusive of TiO2. Coordination of small colorless organic molecules with the surface of Al2O3 particles also leads to the formation of composites whose optical absorption is extended into the visible spectral region. The degradation of organic dye was used to test photocatalytic ability of surface-modified Al2O3 particles. Having in mind that pristine Al2O3 has the band gap of about 8.7 eV [31], and does not absorb solar photons, photocatalytic reactions, driven

V. Ðordevic´ et al. / Chemical Physics Letters 685 (2017) 416–421

using surface-modified Al2O3 particles, indicate that the CT complex formation is a promising way to enhance photocatalytic performance of metal oxides. The optical properties and photocatalytic abilities of surface-modified Al2O3 particles are discussed in terms of relative position of energy levels. To complement our experimental findings, corresponding quantum chemical calculations were performed using density functional theory (DFT), and the calculated values, obtained for a series of aluminum complex model systems analyzed here, compare well with experimentally determined band gap values. The details of the experimental results and theoretical model calculations are presented and summarized in figures and tables below.

2. Results and discussion The thorough microstructural characterization of Al2O3 particles, prepared by a sol-gel process via hydrolysis of aluminum isopropoxide, followed with calcination [32], was performed and the results are presented in Fig. 1. The XRD pattern of samples calcinated at 700 °C indicated the existence of crystalline c-Al2O3 (JCPDS Card No. 00-010-0425). The presence of other coexisting alumina phases or impurities was not detected. The calcinations at higher temperature (1100 °C) led to the formation of alumina with corundum crystal structure (JCPDS Card No. 00-056-0457); the XRD pattern is not shown since alumina particles with corundum phase are outside the scope of the present work. Highresolution TEM (Fig. 1B) revealed the presence of reasonably uniform spherical Al2O3 particles in the size range from 0.1 to 0.3 lm. Also, it can be seen that each submicron-size Al2O3 sphere is composed of many nanocrystal grains. This observation is consistent with the broadening of the XRD peaks and estimated size of constitutive building units of about 3.2 nm is determined by the use Scherrer’s equation. Nitrogen adsorption-desorption

417

isotherm of synthesized c-Al2O3 samples is shown in Fig. 1C, and specific surface area, calculated by BET method [33], was found to be 54.8 m2 g 1. It should be noted that specific surface area of synthesized Al2O3 powder is similar to commonly used commercial Degussa P25 TiO2 photocatalyst (50 m2 g 1) [34]. Pore size distribution of Al2O3 powder is shown in Fig. 1D. It is clear that the sample is mesoporous, and estimated pore radius was found to be in the size range 3–4 nm. The Kubelka-Munk transformations of diffuse reflection data for surface-modified Al2O3 particles with catechol (CAT) and 5-amino salicylic acid (5-ASA) are shown in Fig. 2. As can be seen in this Figure, the absorption spectra of surface-modified Al2O3 are redshifted due to formation of CT complexes between ligand molecules and surface Al atoms. The effective band gap energies of CAT/Al2O3 and 5-ASA/Al2O3, determined from the absorption onset, are found to be 1.26 and 1.66 eV, respectively. Our preliminary results (not detailed here) with a series of catecholate-type of ligands (catechol, 2,3-dihydroxy naphthalene, caffeic acid, dopamine, gallic acid) showed, as a general trend, that they induce larger red-shift than salicylate-type of ligands (salicylic acid, 5-amino salicylic acid) and ascorbic acid. Finally, the largest red-shift of optical absorption was observed upon coordination of CAT to the surface of Al2O3 particles (compare band gap value for bulk cAl2O3 (8.7 eV) [31] with absorption onset of CAT/Al2O3 at 1.26 eV). It transpires that the surface-modification, as described here, is a simple way to transform the insulator into a hybrid semiconductor-like material capable of harvesting the large portion of the solar spectrum. The small differences in the optical properties of surface-modified Al2O3 particles, i.e. fine tuning of the absorption properties, can be achieved by changing different electron donating/withdrawing functional side groups attached to the basic backbone structure of catechol or salicylic acid [27]. The ways the ligands bind to the Al2O3 surface was investigated using FTIR spectroscopy, and the FTIR spectra of free and adsorbed

Fig. 1. Microstructural characterization of synthesized Al2O3 samples: (A) XRD, (B) high-resolution TEM, (C) nitrogen adsorption-desorption isotherms, and (D) pore size distribution.

418

V. Ðordevic´ et al. / Chemical Physics Letters 685 (2017) 416–421

Fig. 2. Kubelka–Munk transformations of UV–Vis-NIR diffuse reflection data of surface-modified Al2O3 powders with CAT (blue) and 5-ASA (red). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

ligands (CAT and 5-ASA) on Al2O3, as well as detailed analysis of FTIR spectra, are presented in Supporting Information. Based on FTIR measurements, we could not discriminate if the formation of bidentate mononuclear chelating or bidentate binuclear bridging complexes takes place. On the theoretical side, DFT calculations were performed with periodic boundary conditions (PBC) using PBE [35] and HSE06 [36] functional, in order to estimate the energy gaps of various complexes. Although, it is well-known that PBE functional underestimates bandgaps for insulators and semiconductors [37], some recent studies indicated that it may predict energy levels of catecholate-type interfacial charge transfer complexes quite accurately [27]. The rationale for the choice of HSE functional refers to their ability to calculate bandgaps for semiconductors more precisely comparing to the standard semi-local functional, such as PBE, with no significant increase of the computational cost [38,39]. On the other hand, it has been shown that standard LDA and GGA functional fail to reproduce the band gap of bulk c-Al2O3, however, a proper treatment has been achieved by using the modified semi-local Becke-Johnson exchange potential (mBJ) [40]. Four models of different complexity, based on the crystal structure of c-Al2O3, considering (1 0 0) and (1 0 1) surfaces, were constructed in attempt to calculate energy gaps of CAT/Al2O3 and 5ASA/Al2O3. The optimized structures for CAT molecules anchored to the surface of c-Al2O3 are presented in Fig. 3, while the data obtained by DFT calculations are summarized in Table 1. The results for 5-ASA/Al2O3 are presented in Supporting Information. The first two model systems refer to (1 0 0) plane, while the second two are connected with (1 0 1) plane of c-Al2O3. It is evident that both, PBE and HSE06 functional, highly overestimate energy gaps for the simplest model system (I); 3.36 and 4.13 eV, respectively (see Table 1). The more realistic description was achieved by the second model (II). The calculated gap using PBE functional moderately overestimates experimental value (1.62 vs. 1.26 eV), while usage of HSE06 functional yields energy gap that significantly deviate from experimental value (2.25 eV). The overestimated energy gap values were also obtained for both functional (1.63 and 2.00 eV for PBE and HSE06, respectively) when calculations were performed using the simpler model associated with (1 0 1) surface (model system III). Finally, the best agreement with experimentally determined band gap value was achieved using the most sophisticated model system (IV), with HSE06 value being overestimated by only 0.12 eV (1.38 vs. 1.26 eV).

The spatial distribution of crystal orbitals of unmodified c-Al2O3 and CAT/Al2O3 (model system IV), as well as orbitals of free CAT molecule are shown in Fig. 4. The data concerning spatial distributions of CAT/Al2O3 crystal orbitals for three other model systems, and data for 5-ASA/Al2O3 (model system IV) are presented in Supporting Information. Analyses of the electronic structure of c-Al2O3 revealed that the valence band maximum (VBM) and conduction band minimum (CBM) are positioned at 7.18 and 4.39 eV vs. AVS, giving the band gap of 2.79 eV. As expected, this functional drastically underestimates band gap energy of c-Al2O3. Furthermore, two types of donor levels are located within the predicted bandgap region of Al2O3. The first one consists of the p-orbitals from phenyl ring and bridging oxygen atoms (a-level; HOCO 2), and the second one includes similar orbitals hybridized with the lattice atoms (two b-levels; HOCO 1, HOCO). Thus, it can be concluded that the origin of interfacial charge-transfer transitions are related to the excitation from the donor level (a) to the CBM of the c-Al2O3. On the other hand, excitation from b donor level lies in the near IR region, and do not participate in the visible-light absorption. These findings are in accordance with previous results concerning CAT/TiO2 system [27]. In addition, the two highest occupied molecular orbitals of free CAT (HOMO-1, HOMO) are positioned inside the bandgap of c-Al2O3, while LUMO was found to be higher than CBM of c-Al2O3 (see Fig. 4 and Table 1). Similar to the CAT, bidentate coordination of 5-ASA to the surface of c-Al2O3 was found, and interfacial charge transfer transition may be ascribed as the excitation from the a-donor level to CBM (see Supporting Information). The calculated energy gap between these two levels using HSE06 functional was found to be 1.97 eV. This value is overestimated with respect to the experimental one by approximately 20% (1.97 vs. 1.66 eV). For comparison, calculated value obtained using the PBE functional is underestimated by the similar amount (1.31 vs. 1.66 eV). To explore whether the gross periodic calculations might be replaced with the modeling on computationally less demanding molecular systems, series of calculations were performed using three different cluster systems. The hexa-, nona-, and dodecaclusters have been constructed based on the (1 0 1) plane of cAl2O3 and adequacy of the clusters has been estimated with respect to CAT. The geometries of the cluster systems and the results of calculations are shown in Supporting Information. The estimated energy gaps for both functional increased with the increase of the cluster size and gradually get closer to the experimental gap value. However, this convergence is more rapid with HSE06 functional, and the best agreement was found for CAT/ (Al2O3)12 cluster (1.02 vs.1.26 eV). However, the better description of the optical properties of molecular systems can be achieved within TD-DFT formalism. For example, the accurate description of charge-transfer excitations in organic dyes for the application in DSSCs has been achieved with CAM-B3LYP method, which was proven to outperform several common functional types [41,42]. Hence, the oxidizing and the reducing power of photocatalyst is determined by the potentials of the valence and the conduction band, respectively, and efficient photocatalyst, besides being able to harvest large portion of solar light, must have band edges close to the H2O stability potentials. Namely, the distributions of HOCO and LUCO orbitals of CAT/Al2O3 and 5-ASA/Al2O3 complexes indicate that they are localized onto the organic part (CAT, 5-ASA) and inorganic part of composites, respectively. Photocatalytic degradation reactions of organic dye methylene blue (MB), under illumination condition that simulate sunlight, were carried out over 5-ASA/Al2O3 composite to establish whether or not synthesized hybrids might have any potential application in light-driven processes. The MB was suitable for this purpose since direct photolysis does not induce its degradation, and, in addition, degradation mechanism is well understood and described in liter-

V. Ðordevic´ et al. / Chemical Physics Letters 685 (2017) 416–421

419

Fig. 3. The optimized geometries of four different CAT/Al2O3 model systems used in periodic calculations.

Table 1 The estimated positions of donor level and conduction band minimum (CBM), as well as energy gaps of CAT/Al2O3 calculated by DFT using PBC models of different complexity. The data for the pure c-Al2O3 (model system IV) and the free catechol molecule are also included. All energies are given in eV. Structure

Level of theory

Model system

CAT/Al2O3

PBE/6-31G(d,p)

I II III IV I II III IV

4.92 4.16 5.19 6.57 5.45 4.38 5.46 6.39

HSE06/6-31G(d,p)

a b

Donor level

c-Al2O3

HSE06/6-31G(d,p)

IV

7.18 (VBM)a

CAT

HSE06/6-31G(d,p)

Catechol

6.33b ( 5.48)

CBM 1.56 2.54 3.56 5.40 1.32 2.13 3.46 5.01 4.39 0.02

Ecalc g 3.36 1.62 1.63 1.17 4.13 2.25 2.00 1.38 2.79 5.50

Valence band maximum. The data for the free molecules are associated with HOMO 1, (HOMO) and LUMO orbitals.

ature [43]. The photo-degradation kinetic data for different initial concentrations of MB are presented in Fig. 5. Based on kinetic measurements, some general features can be readily discerned. Firstly, the 5-ASA/Al2O3 hybrid composite photocatalytically performs, and complete decolorization of MB was observed after sufficiently long illumination time. As expected, unmodified c-Al2O3 powder does not display photocatalytic activity (see inset to Fig. 5). Secondly, the increase of the rate constant with the increase of the initial concentration of polluter molecules is a textbook example of Langmuir-Hinshelwood kinetics [44]. Thirdly, comparison of photocatalytic performance of 5-ASA/Al2O3 and the most studied commercial Degussa P25 TiO2 photocatalyst were performed either under illumination that simulate solar light or under visible light illumination conditions (inset to Fig. 5). Optical filter, based on NaNO2 solution, was used to reduce UV part of the emitted light from the light source. The obtained results clearly indicate that photocatalytic performance of 5-ASA/Al2O3 powder is almost the

same after reducing UV light. On the other hand, as expected, photocatalytic performance of Degussa P25 is significantly diminished when NaNO2 optical cut-off filter was used. Finally, photocatalytic performance of 5-ASA/Al2O3 was not diminished when samples were kept under standard laboratory conditions over a time period of one month. Of course, this is a preliminary study and determination of the level of enhancement of redox chemistry of surfacemodified Al2O3 powders and their possible applications will be the subject of our further research. 3. Conclusion In conclusion, we have synthesized a wide range of Al2O3 organic/inorganic complexes and thoroughly analyzed their visible light photo-activity by experimental and theoretical methods. Our initial photocatalytic results strongly support the conclusion that these novel low-cost hybrid materials exhibit enhanced photo-

420

V. Ðordevic´ et al. / Chemical Physics Letters 685 (2017) 416–421

Fig. 4. The frontier crystal orbitals drawn over a single repeated unit for CAT/Al2O3 (model system IV). The orbitals of free catechol molecule, as well as VBM and CBM of unmodified c-Al2O3 are also presented. HOCO and LUCO represent highest occupied and lowest unoccupied crystal orbitals.

Fig. 5. Photocatalytic degradation kinetics of MB (initial concentrations: (a) 1.0, (b) 2.0, and (c) 5.0 ppm) over 5-ASA/Al2O3 powder (0.5 mg/mL) under simulated sunlight. Inset: comparison of photocatalytic performances of unmodified c-Al2O3, 5-ASA/Al2O3 and TiO2 (Degussa P25) under identical experimental conditions (initial concentration of MB was 2 ppm and concentration of photocatalysts were 0.5 mg/mL) under simulated sunlight and with reduced UV light by optical filter based on NaNO2 solution.

redox chemistry, and are promising candidates for use in many light-induced processes.

4. Experimental and computational methods The c-Al2O3 was prepared similar to the method described elsewhere [32]. Briefly, the round bottom flask with 20 mol of distilled water was heated in oil bath up to 85 °C and 0.20 mol of aluminiumisopropoxide (Al[OCH(CH3)2]3, 98+%, Alfa Aesar) was added under vigorous stirring. After 15 min, 4.85 ml of nitric acid (HNO3, 69–70%, J.T. Baker) was added and the solution was stirred and kept uncovered for 2 h at 85 °C, which was sufficient time for isopropanol, formed during the hydrolysis, to evaporate. After

covering the flask, heating and stirring was maintained for additional 4 h. Then 12.0 g of water soluble polymer Polyethylene glycol 12,000 (H(OCH2CH2)nOH, M.W. range 11,000–13,000, Alfa Aesar) was added into cooled solution and gentle stirring was maintained overnight. Obtained sol was dried in air in open beaker heated to 70 °C until viscous white gel was formed. The gel was transferred into porcelain crucible and finally calcinated at 700 °C for 24 h in open-air furnace. Surface modification of c-Al2O3 powder was performed by dispersing 0.1 g of powder in 10 ml of water containing adequate amount (1.53  10 5 mol) of dissolved CAT (ex. pure, Fisher) or 5-ASA (99%, Acros Organic). After 2 days the powder was three times thoroughly washed with distilled water, and finally dried in a vacuum oven at 40 °C. The X-ray diffraction (XRD) powder patterns were recorded using Rigaku SmartLab instrument under the Cu Ka1,2 radiation. The intensity of diffraction was measured with continuous scanning at 2°/min. The data were collected at 0.02° intervals. Transmission electron microscopy (TEM) was performed using a JEOL JEM-2100 LaB6 instrument operated at 200 kV. TEM images were acquired with a Gatan Orius CCD camera at 2 binning. Nitrogen adsorption-desorption isotherms were determined on Sorptomatic 1990 Thermo Finnigan automatic system using nitrogen physisorption at 196 °C. Before measurement the samples were outgassed at 130 °C for 3 h. Specific surface area of the samples was calculated from the nitrogen adsorption-desorption isotherms according to the BET method [33]. Pore size, pore volume distribution, and porosity were determined by mercury intrusion porosimetry on Pascal 140/440, Thermo Scientific. Optical properties of surface-modified Al2O3 particles were studied in the wavelength range from 300 nm to 1.3 lm by diffuse reflectance measurements (Shimadzu UV–Visible UV-2600 spectrophotometer equipped with an integrated sphere ISR-2600 Plus). Infrared spectroscopy measurements were carried out using a Thermo Nicolet 6700 FTIR spectrometer at spectral resolution of 8 cm 1 in the region of 4000–400 cm 1. Photocatalytic ability of hybrid 5-ASA/Al2O3 particles was studied using the degradation of organic dye methylene blue (MB).

V. Ðordevic´ et al. / Chemical Physics Letters 685 (2017) 416–421

Photocatalytic decolorization of MB was carried out with 50 mg of the powdered photocatalyst suspended in 100 mL of MB solution. Initial concentrations of MB were in the range from 1.0 to 5.0 ppm. Prior to irradiation, the slurry was stirred in the dark for 30 min to obtain the equilibrium adsorption state. The light source, placed 50 cm above the top surface of the pollutant solution, was 300 W Osram Ultra Vitaluxlamp whose emission spectrum simulates solar radiation. The absorbance/concentration of MB was monitored using UV–Vis spectrophotometer (Thermo Electron Nicolet Evolution 500). For determination of photocatalytic degradation rate constants, initial concentration of MB was corrected for the amount of adsorbed organic dye onto surfacemodified Al2O3. The blank experiments, the degradation of organic dye under simulated sunlight in the absence of Al2O3, were also performed. When desired, the UV part of light source was reduced using optical filter based on NaNO2 solution. For comparison, photocatalytic experiments were also performed under identical experimental conditions with commercial TiO2 Degussa P25 powder. The DFT calculations for periodic boundary conditions (PBC) as well as for the molecular cluster systems were carried out with the Gaussian 09 suite of programs [45]. Both type of calculations were performed with the use of Perdew-Burke-Ernzerhof (PBE) functional [35], as the generalized gradient approximation (GGA) model, and Heyd-Scuseria-Ernzerhof screened hybrid functional (HSE06) [36]. The Pople 6-31G(d,p) valence double-zeta polarized basis set [46], have been used in all calculations. The ultrafine integration grid was also specified for all calculations. The unit cells that were used for the periodic calculations have been constructed based on the defect-free crystal structure of c-Al2O3 (7.887  7.887  7.887 Å). The Al2O3 (1 0 0) surface was modeled by using the two different slab models, which are both single layer Alterminated. First one contains two oxygen and four aluminum layers, while the second one comprises of three oxygen and six aluminum layers. The unit cells of both models are defined by the same lattice parameters (7.887  7.887 Å), with infinite vacuum thickness. To model Al2O3 (1 0 1) surface the two additional slab models were employed, which are partially and fully hydroxylated. This surface imposes the same number of Al- and O-layers, and we constructed four and five-layered unit cells, with the same lattice parameters (7.891  5.649 Å), the second lattice is capped with one oxygen atom on the bottom. These models also imply infinite vacuum space. The presence or absence of surface hydroxyl groups, and additional oxygen atom, were specified in order to keep the Al2O3 stoichiometry and charge neutrality of the unit cells. The optimization of the atomic coordinates was carried out within the unit cell, whereas the lattice parameters were fixed in all cases. For the calculations on the molecular systems, we used three different-sized clusters. These clusters were also built from the crystal structure of c-Al2O3, following the (1 0 1) surface plane. Hence, we constructed hexa-, nona-, and dodeca-clusters ((Al2O3)n (n = 6, 9, 12)), with 30, 45 and 60 atoms overall, respectively. The CAT and 5-ASA were anchored on the top of the clusters, and their atoms were relaxed during the optimization, whereas the atoms of the cluster have been frozen to preserve the crystal structure of cAl2O3. Acknowledgement This publication was made possible by NPRP grants #6-021-1005 and #5-674-1-114 from the Qatar National Research Fund (a member of Qatar Foundation). Work at the Vincˇa Institute of Nuclear Sciences and Institute of Physics, Belgrade is supported by the Ministry of Education, Science and Technological Develop-

421

ment of the Republic of Serbia, under the projects 45020 and 171017, respectively. Appendix A. Supplementary materials Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cplett.2017.08. 012. References [1] A. Fujishima, K. Honda, Nature 238 (1972) 37. [2] E. Pelizzetti, C. Minero, Comments Inorg. Chem. 15 (1994) 297. [3] M.R. Hoffmann, S.T. Martin, W. Choi, D.W. Bahnemann, Chem. Rev. 95 (1995) 69. [4] A. Fujishima, X. Zhang, D.A. Tryk, Surf. Sci. Rep. 63 (2008) 515. [5] B. Ohtani, J. Photochem. Photobiol. C Photochem. Rev. 11 (2010) 157. [6] A.B. Djurišic´, Y.H. Leung, A.M. Ching Ng, Mater. Horizons 1 (2014) 400. [7] J. Schneider, M. Matsuoka, M. Takeuchi, J. Zhang, Y. Horiuchi, M. Anpo, D.W. Bahnemann, Chem. Rev. 114 (2014) 9919. [8] H. Wang, L. Zhang, Z. Chen, J. Hu, S. Li, Z. Wang, J. Liu, X. Wang, Chem. Soc. Rev. 43 (2014) 5234. [9] B. O’Regan, M. Grätzel, Nature 353 (1991) 737. [10] D. Kuang, J. Brillet, P. Chen, M. Takata, S. Uchida, H. Miura, K. Sumioka, S.M. Zakeeruddin, M. Grätzel, ACS Nano 2 (2008) 1113. [11] M. Zhang, C. Chen, W. Ma, J. Zhao, Angew. Chemie Int. Ed. 47 (2008) 9730. [12] R. Asahi, T. Morikawa, T. Ohwaki, K. Aoki, Y. Taga, Science 293 (2001) 269. [13] S. Sakthivel, H. Kisch, Angew. Chemie Int. Ed. 42 (2003) 4908. [14] Y. Cong, B. Tian, J. Zhang, Appl. Catal. B Environ. 101 (2011) 376. [15] R. Long, N.J. English, Phys. Chem. Chem. Phys. 13 (2011) 13698. [16] R. Long, N.J. English, Chem. Phys. Lett. 513 (2011) 218. [17] K.A. McDonnell, N.J. English, M. Rahman, D.P. Dowling, Phys. Rev. B 86 (2012) 115306. [18] Y. Tian, T. Tatsuma, J. Am. Chem. Soc. 127 (2005) 7632. [19] C. Gomes Silva, R. Juárez, T. Marino, R. Molinari, H. García, J. Am. Chem. Soc. 133 (2011) 595. [20] S. Naya, K. Kimura, H. Tada, ACS Catal. 3 (2013) 10. [21] T. Rajh, J.M. Nedeljkovic, L.X. Chen, O. Poluektov, M.C. Thurnauer, J. Phys. Chem. B 103 (1999) 3515. ˇ omor, J.M. Nedeljkovic´, J. Phys. Chem. C 113 [22] I.A. Jankovic´, Z.V. Šaponjic´, M.I. C (2009) 12645. [23] T.D. Savic´, I.A. Jankovic´, Z.V. Šaponjic´, M.I. Cˇomor, D.Zˇ. Veljkovic´, S.D. Zaric´, J.M. Nedeljkovic´, Nanoscale 4 (2012) 1612. ˇ omor, J.M. Nedeljkovic´, M.D. Dramic´anin, M.G. [24] T.D. Savic´, Z.V. Šaponjic´, M.I. C Nikolic´, D.Zˇ. Veljkovic´, S.D. Zaric´, I.A. Jankovic´, Nanoscale 5 (2013) 7601. ˇ omor, J.M. Nedeljkovic´, D.Zˇ. Veljkovic´, S.D. Zaric´, V.M. Rakic´, I. [25] T.D. Savic´, M.I. C A. Jankovic´, Phys. Chem. Chem. Phys. 16 (2014) 20796. ˇ omor, N.D. Abazovic´, Z.V. Šaponjic´, M.T. Marinovic´-Cincovic´, D. [26] T.D. Savic´, M.I. C Zˇ. Veljkovic´, S.D. Zaric´, I.A. Jankovic´, J. Alloys Compd. 630 (2015) 226. [27] S. Higashimoto, T. Nishi, M. Yasukawa, M. Azuma, Y. Sakata, H. Kobayashi, J. Catal. 329 (2015) 286. [28] B. Moongraksathum, P.-T. Hsu, Y.-W. Chen, J. Sol-Gel Sci. Technol. 78 (2016) 647. [29] I. Vukoje, T. Kovacˇ, J. Dzˇunuzovic´, E. Dzˇunuzovic´, D. Loncˇarevic´, S.P. Ahrenkiel, J.M. Nedeljkovic´, J. Phys. Chem. C 120 (2016) 18560. [30] B. Milic´evic´, V. Ðordevic´, D. Loncˇarevic´, S.P. Ahrenkiel, M.D. Dramic´anin, J.M. Nedeljkovic´, Microporous Mesoporous Mater. 217 (2015) 184. [31] B. Ealet, M.H. Elyakhloufi, E. Gillet, M. Ricci, Thin Solid Films 250 (1994) 92. [32] W. Wu, Z. Wan, W. Chen, M. Zhu, D. Zhang, Microporous Mesoporous Mater. 217 (2015) 12. [33] S. Brunauer, P.H. Emmett, E. Teller, J. Am. Chem. Soc. 60 (1938) 309. [34] C.B. Almquist, P. Biswas, J. Catal. 212 (2002) 145. [35] J.P. Perdew, K. Burke, M. Ernzerhof, Phys. Rev. Lett. 77 (1996) 3865. [36] J. Heyd, G.E. Scuseria, M. Ernzerhof, J. Chem. Phys. 118 (2003) 8207. [37] J.M. Crowley, J. Tahir-Kheli, W.A. Goddard, J. Phys. Chem. Lett. 7 (7) (2016) 1198. [38] J.E. Moussa, P.A. Schultz, J.R. Chelikowsky, J. Chem. Phys. 136 (2012) 204117. [39] E.N. Brothers, A.F. Izmaylov, J.O. Normand, V. Barone, G.E. Scuseria, J. Chem. Phys. 129 (2008) 011102. [40] M. Yazdanmehr, S.J. Asadabadi, A. Nourmohammadi, M. Ghasemzadeh, M. Rezvanian, Nanoscale Res. Lett. 7 (2012) 488. [41] P. Dev, S. Agrawal, N.J. English, J. Chem. Phys. 136 (2012) 224301. [42] P. Dev, S. Agrawal, N.J. English, J. Phys. Chem. A 117 (2013) 2114. [43] A. Houas, H. Lachheb, M. Ksibi, E. Elaloui, C. Guillard, J.-M. Herrmann, Appl. Catal. B Environ. 31 (2001) 145. [44] K.V. Kumar, K. Porkodi, F. Rocha, Catal. Commun. 9 (2008) 82. [45] M.J. Frisch, et al., Gaussian 09, Revision B.01, Gaussian, Inc., Wallingford CT., 2009. [46] W.J. Hehre, R. Ditchfield, J.A. Pople, J. Chem. Phys. 56 (1972) 2257.