THE INFLUENCE OF TITANIUM DIOXIDE MODIFICATION BY SULFUR AND CARBON ON PHYSICO-CHEMICAL AND PHOTOCATALYTIC PROPERTIES

The nanocomposites based on TiO 2 doped with sulfur (S/TiO 2 ), carbon (C/TiO 2 ), carbon and sulfur (S/C/TiO 2 ) have been obtained. The powders were characterized by XRD, XPS, BET, SEM, EDX, TEM and UV-VIS spectroscopy. EDX and XPS spectroscopies prove that titanium dioxide powder includes only Ti and O elements, composites C/TiO 2 include the elements Ti, O, C, composites S/TiO 2 - Ti, O, S and composites C/S/TiO 2 - Ti, O, C, and S. XRD analysis revealed phase of anatase in all composites, rutile phase appeared with increasing of sulfur amount in sulfur-containing powders. It has been found that composites consist of roundish agglomerates in the range of 5–30 mm. Sulfur additives decrease grain growth of titanium dioxide particles from 14 to 9–10 nm in S/TiO 2 composites, carbon leads to increase in particle size from 14 to 19 nm, simultaneous modification of titanium dioxide by carbon and sulfur leads to the formation of particles with sizes of 7–8 nm. Analysis of nitrogen sorption–desorption isotherms for all synthesized samples has shown the presence of a hysteresis loop which is the evidence for mesoporous structure of the powders. The isotherms correspond to type IV of IUPAC classification for mesoporous materials with H1 type for C/TiO 2 and H 2 type for S/TiO 2 , and C/S/TiO 2 of hysteresis loop. The modification of TiO 2 by carbon and sulfur leads to increase of specific surface area (of about 1.8 times in the case of C/TiO 2 , about 3.3 times for S/TiO 2 and about 4.7 times for C/S/TiO 2 ), average pore volume and decrease of radius pore volume compared with TiO 2 . Absorption spectra of nanocomposites showed a bathochromic shift as compared with the absorption band of pure TiO 2 . It has been found that modification leads to band gap narrowing. Nanocomposite samples showed higher photocatalytic activity in the destruction of safranine T under UV and visible irradiation compared to pure TiO 2 . It may be related to the participation of dopants in the inhibition of electron-hole recombination, prolongation of charges lifetime, increasing efficiency of interfacial charge separation and formation of doping electronic states.


INTRODUCTION
Titanium dioxide is widely used in photocatalysis, in particular, for destruction of organic and inorganic pollutants. However, it has several serious disadvantages: an insufficiently high quantum yield of the reaction, wide band gap (3.2 eV), high rate of electron-hole recombination, and peculiarity of light adsorption by TiO 2 resulting in its photochemical activity only in the UV region of spectrum. So, an urgent problem in photocatalysis is a search for photocatalysts active under visible light irradiation. Researchers are trying in many ways to extend the photoactivation TiO 2 wavelength in the visible region of the spectrum, which will increase the use of solar energy and prevent recombination of the electron-hole pairs and thus allow more charge carriers to successfully diffuse to the surface. One of the ways is modification of titanium dioxide with different additives [1][2][3][4][5][6].
Currently researchers are concentrated on non-metals as dopants due to their advantages.
These authors have reported that such doping leads to an appearance the photocatalytic activity under visible light due to incorporated S or C atoms into the lattice of titanium dioxide and thus causes the generation of intermediate energy level in TiO 2 , located above valence band, narrowing the energy gaps of TiO 2 and shifting the optical absorption from UV region to visible one. So, the modification of titanium dioxide with non-metals can be an effective way to improve the photocatalytic activity under visible irradiation.
Considering the above, the aim of our work was to determine and compare the effect of titanium dioxide modification with different amount of sulfur and carbon on the physicochemical and photocatalytic properties of these composites.
Characterization of photocatalysts. Phase composition of the samples was determined by X-ray powder diffraction. A computerized Bruker D8 Advance diffractometer was equipped with CuK α (λ = 0.15406 nm) X-ray source. All XRD peaks were checked and assigned to known crystalline phases. Average crystallite size was determined using broadening of the most intensive reflex using the Debye-Scherrer equation. Interplanar distance (d, nm) was calculated using Wulff-Bragg's equation.
Diffuse reflectance spectra (DRS) of the powders were measured using a Perkin-Elmer Lambda Bio 35 spectrophotometer in the range between 200 and 1000 nm which allows converting data of corresponding spectra using the Kubelka-Munk equation. The value of E g was estimated by the method proposed by Wood and Tauc by the extrapolation of the linear part of the plot (hν*α(hν))1/n versus hν toward energy axis at α(hν) = 0 (n = ½ for direct allowed transitions). The absolute and relative errors were ±0.01 eV and ±0.3 %, respectively.
To analyze sample composition (elemental analysis) and its morphology a scanning electron microscope (SEM JSM 6490 LV, JEOL, Japan) with an integrated system for electron microprobe analysis INCA. Energy based on energy-dispersive and wavelength-dispersive spectrometers (EDS + WDS, OXFORD, United Kingdom) with HKL Channel system was used.
Transmission electron microscopy (TEM) for received materials was carried out on a transmission electron microscope JEM-1200 ЕX (JEOL, Japan).
The values of the specific surface area (S sp ) of the samples as well as pore size distribution were determined using a Quantachrom NovaWin 2 device. The specific surface area of the samples was obtained from isotherms of nitrogen adsorption-desorption using the Brunauer-Emmet-Teller (BET) approach. The pore radius (R) and the pore volume (V tot ) were calculated from the desorption branches of the isotherms using the Barret-Joiner-Halenda method.
Presence of chemical elements and chemical bonds features in the samples were analyzed using X-ray photoelectron spectroscopy (XPS) on an UHV-Analysis-System equipment produced by SPECS Surface Nano Analysis Company (Berlin, Germany). The instrument was equipped with a semi-spherical analyzer PHOIBOS 150. XPS spectra of core-level and valence electrons were analyzed in an UHV-Analysis-System chamber under residual pressure not higher than 710 -8 Pa. XPS-spectra were activated by X-ray MgK  -irradiation (E = 1253.6 eV) and recorded at a constant pass energy of 30 eV. The energy scale of the device was graded by method [15] using reference metals Au and Cu. Surface charge of the samples was taken into account in reference to the energy of the C1s-line from hydrocarbon adsorbates which was set to 284.6 eV as recommended for transition metal oxides [16,17].
Photocatalytic experiment. Photocatalytic activity of the samples was evaluated by the change of the concentration of safranine T (ST) (c = 0.03 g L -1 ). Before irradiation, the catalyst suspension (2 g L -1 ) in an aqueous solution of ST was kept in dark up to achieve adsorption equilibrium. The time to reach the adsorption equilibrium in the system powder -ST does not exceed 2 h for all the samples.
Irradiation of aqueous solutions (pH-6.5) of dye was performed at room temperature in a quartz reactor in the presence of air. The light source was a high-intensity Na discharge lamp GE Lucalox with power of 70 W, the latter emitting in the visible region with maxima at 568, 590 and 600 nm and an UV lamp BUV-30 with the power of 30 W and radiation maximum at 254 nm.
Concentrations of the ST were measured spectrophotometrically using a Shimadzu UV-2450 spectrophotometer at λ = 520 nm for ST.
Photocatalytic rate constants for the model compound were calculated using the pseudofirst order kinetic equation.

RESULTS AND DISCUSSION
Investigation of the obtained powders by energy-dispersive spectroscopy based on energydispersive technique ( Analysis of SEM-images of the samples shows randomly distributed agglomerates in the range of 5-30 m (Fig. 1). The crystallite size in the agglomerates of the samples was calculated through Debye-Scherrer equation. The agglomerates of titanium dioxide consist of the particles of 14 nm in size. Modification with carbon leads to an increase in particle size: as the amount of carbon increases from 1 to 21 wt. %, the particle size increases from 14 to 19 nm. Sulfur additives inhibit grain growth of TiO 2 and form the particles of 9-10 nm in S/TiO 2 composites. Simultaneous modification of TiO 2 with carbon and sulfur leads to the formation of particles with a size of 7-8 nm. The change of TiO 2 particle sizes due to the doping with carbon and sulfur was confirmed by the TEM study (Fig. 2).  (Fig. 2 a). All composites also show intensive peaks corresponding to the anatase phase (Fig. 2). No signal of brookite was detected. The peaks are quite broad, indicating small crystallite sizes. In contradiction to [2], additives of carbon into the binary composites do not induce new crystalline forms different from anatase ( Fig. 2 b), which corresponds to the papers [5,7,8,12]. In the case of S/TiO 2 (Fig. 2 c), XRD patterns show the reflexes of anatase ((101), (004), (200), (105), (211), (213)) and rutile ((110), (101)) phases regardless of sulfur concentration.
The addition of carbon (1.5 wt. %) and sulfur (from 0.6 to 9 wt. %) into ternary composites practically does not influence on the TiO 2 XRD pattern view (all reflexes of anatase are present) (Fig. 2 d, curve 1), which corresponds to the data obtained in preparation of nanocomposite by other methods [12,18,19]. But an increase in sulfur amount (12 and 15 wt. %) in the samples leads to appearance of new reflexes at 2 = 27.4 and 36.2 ° (Fig. 2 d, curve 2). It corresponds to the (110) and (101) planes of rutile. Thus, an increase of the sulfur amount in the ternary samples leads to the formation of rutile structure.
All composites show the presence of a hysteresis loop which is the evidence for mesoporous structure of the powders (Fig. 3). The isotherms correspond to type IV of IUPAC classification for mesoporous materials with H1 type for C/TiO 2 (Fig. 3 a) and H2 type for S/TiO 2 (Fig. 3 b) and C/S/TiO 2 (Fig. 3 c) of a hysteresis loop.
Predominance of pores up to 3.5 nm is characteristic of pure titanium dioxide, whereas the modification leads to a decrease in an average pore radius up to 2.5 nm for C/TiO 2 and to 1.7-2.4 nm for S/TiO 2 . The ternary composites also have narrow pore size distribution -from 1.5 to 2.5 nm for samples with 0.6 to 9 wt. % of sulfur. Higher sulfur content (12 and 15 wt. %) in C/S/TiO 2 powders results in a larger pore size and broader pore size distribution (up to 3.5 nm). Similar results were obtained in [12]. It has been found that the surface area of composites depends on the dopant amount. The value of the specific surface area is increased of about 1.8 times in the case of C/TiO 2 (35-53 m 2 /g), about 3.3 times for S/TiO 2 (60-95 m 2 /g) and about 4.7 times for C/S/TiO 2 (80-133 m 2 /g) as compared with that of titanium dioxide (28 m 2 /g). These results are in accordance with the XRD, because the modification of titanium dioxide with carbon and sulfur leads to a decrease in the crystallite size and to an increase in the specific surface area of the samples.
Modification of titanium dioxide with carbon or sulfur in the case of binary and ternary composites leads to an increase in pore volume (for C/TiO 2 to 0.08, for S/TiO 2 -0.21, for C/S/TiO 2 -0.17) compared to TiO 2 (0.05 cm³/g) and the radius decrease (for C/TiO 2 to 3.0, for S/TiO 2 -2.1, for C/S/TiO 2 -2.03 nm) compared to that of TiO 2 (3.8 nm).
The XPS spectra also prove that the composites include Ti, O, C and S (Fig. 4). The features of the XPS spectra in all the samples can be attributed to the core-level electronic levels of titanium, oxygen and carbon atoms, and the presence of sulfur is detected for S/TiO 2 and C/S/TiO 2 composites. Modification of titanium dioxide with sulfur and carbon does not lead to the appearance of some fine-structure features or to change in the energy position of the XPS spectra of core-level electrons (Table).
It has been previously reported [20] that a sulfur-containing material shows a binding energy value of around 170 eV. In our samples the presence of sulfur was confirmed by a peak at 168.4 eV (Fig. 4, 5 a). This peak is due to the presence of the SO 4 2anion [21,22]. In all S/TiO 2 and C/S/TiO 2 composites a peak at 168.4 eV was visible (Fig. 4, 5 a). Binding energy of S2p-electrons in composites S/TiO 2 and C/S/TiO 2 (Table, Fig. 5 a) corresponds to that for sulfur atoms in the compounds Na 2 SO 4 and FeSO 4 [23].  Energy positions of the XPS spectra of corelevel Ti2p 3/2 -and O1s-electrons (Table, Fig. 5 b, c) in the investigated samples are in a good agreement with the literature data regarding the values of binding energies of Ti 2p 3/2 -and O1s-electrons in TiO 2 (458.5-459.0 eV for Ti2p 3/2 -electrons and about 530 еV for O1selectrons [23]).
It can be seen from Fig. 5 d that C 1s spectra at 295 to 282 eV can be observed. These species can be assigned to adventitious carbon species on the catalyst surface [24,25]. The peak (about 285 eV) is thought to signal the presence of adventitious elemental carbon and the peak (289-290 eV) indicates the presence of C-O bonds [24][25][26] which are in agreement with our previous results on the C/TiO 2 composite FT-IR spectrum [27]. These peaks are more intensive for C/TiO 2 composite. There is a peak at about 282 eV for C/TiO 2 powder and it can be ascribed to carbon substituting for oxygen atom in the lattice of TiO 2 , which resulted in formation of O-Ti-C bonds [24,28]. The formation of carbonate species could cause an obvious longtail absorption in the visible region [28]. Indeed, doping of titanium dioxide with carbon and sulfur leads to an appearance of a bathochromic shift as compared to the absorption band of pure TiO 2 , the composite samples extended the absorption to the visible range (around 430 nm) (Fig. 6 a, c, e). Binary and ternary composites have absorption in the entire visible region in the form of a broad, unstructured band without expressed absorption maxima. This absorption can be divided into two parts: shoulder at about 400-430 nm and tail from 430 nm and further.
Also, modification of titanium dioxide with carbon leads to the narrowing of the band gap of the composites from 3.39±0.01 to 3.26±0.01 eV, doping of sulfur -from 3.25±0.01 to 3.04±0.01 eV, additives of carbon and sulfurfrom 3.25±0.01 to 3.14±0.01 eV which contributes to reduce the energy required for photoactivation (Fig. 6 b, d, f). The bathochromic shift of the long-wavelength edge of the absorption band for the samples and the narrowing of the band gap occur because a part of the oxygen (lattice) atoms forming the valence band of TiO 2 are replaced by carbon atoms (Fig. 5 d). The appearance of absorption in the range of 3.12-3.14 eV for C/TiO 2 was observed and related to the formation of impurity (defect) levels in the band gap of TiO 2 upon carbon doping; that leads to sensitizing of C/TiO 2 composites to irradiation in the visible region of spectrum [10,29].
The photocatalytic activity of all the materials was tested for destruction of safranine T in aqueous solution under UV and visible irradiation. Prior to irradiation, photocatalyst-dye systems were kept in the dark to achieve a sorption equilibrium. The adsorption-desorption equilibrium was achieved within 120 min. The sorption value of the dye on composites C/TiO 2 was 55-75 %, S/TiO 2 -25-35 %, and S/C/TiO 2 -60-80 %.
It is known that the dye molecules are capable to absorb UV light following their chemical transformations. The irradiation of safranine T solution without a photocatalyst shows that a negligible amount of dye molecules undergo photolysis under UV light (Fig. 7 a). The process of degradation of the dye is accelerated in the presence of powders (Fig. 7 a).
During the irradiation of safranine T with visible light (in the absence and in the presence of titanium dioxide), the destruction of the dye did not occur (Fig. 7 b). When the dye water solutions were irradiated with visible light in the presence of the composites, a decrease of ST concentrations was observed. The rate of process under UV and visible irradiation was dependent on catalyst composition and structure.
The composite samples showed higher photocatalytic activity in the decomposition of SF than pure titanium dioxide under UV and visible light (Fig. 7).
Experimental results have revealed that the rate of photocatalytic destruction of dye under UV and visible irradiation depends on dopants content. So, for С/ТіО 2 sample the highest activity was detected with 1.5 wt. % of carbon. For ternary composites the rate is increased with an increase of sulfur content in the range of 0.6 to 1.3 wt. % and for S/TiO 2 of 2 to 17 wt. %. The sample S/C/TiO 2 with 1.3 wt. % of sulfur and the sample S/TiO 2 with 17 wt. % of sulfur possess the highest photocatalytic activity both under UV and visible irradiation (Fig. 7). Besides, these samples also showed the highest adsorption activity regarding safranine T. This indicates that, first of all, adsorbed molecules undergo photodestruction.
The enhanced UV and visible photocatalytic activity of nanocomposites is related to the participation of carbon in the inhibition of the recombination of photogenerated electrons and holes due to efficient phase separation of charge [5]. The high photocatalytic activity of the composites upon irradiation by visible light may be attributed to the formation of additional energy levels in the band gap of TiO 2 , which also leads to sensitization to visible light [12,30]. In S-containing samples S(VI) ions can substitute Ti(IV) and form the Ti-O-S bonds, which allows us to modify the electronic structure of TiO 2 by displacing the electronic sites from O to S [31,32]. The oxygen atom becomes a center that hinders the recombination of the electron-hole pairs under irradiation [33].
With the increasing of dopants amount the process rate is decreased due to high concentration of dopant acting as electron-hole recombination sites [12,34] and hence photocatalytic activity is decreased.
To investigate the photocatalytic capability of the most active photocatalyst in repeated applications, five repetition cycles were performed. High activity of the photocatalyst in all the cases is observed up to the third cycle. After the fifth use, deactivation is noted that can be explained by blocking the active sites with the reaction products, which prevents photons from accessing active sites and reduces the number of radicals participating in photodestruction and by losses of the photocatalyst during the experiment. A similar effect was recorded in the paper [35].