SYNTHESIS O-g-C3N4/TiO2 RUTILE COMPOSITE MATERIAL FOR PHOTOCATALYTIC APPLICATION

The development of novel materials ensuring the use of solar radiation as an inexhaustible source of renewable and environmentally friendly energy is one of the actual problems of materials science. Scientific research towards of solving this important task showed the expediency of using photocatalytic processes with the participation of semiconductor systems. One of the most well-known catalyst titanium dioxide TiO2 has photoactivity only in the ultraviolet region of the spectrum that significantly restricts its use. The application of based on undoped graphitelike carbon nitride g-C3N4 or g-C3N4/TiO2 composite catalysts allows using only part of the visible spectrum of solar radiation (with a wavelength of less than 460 nm). It is found that the doping of carbon nitride by oxygen significantly improves its photocatalytic properties to enhancing solar energy utilization. Therefore, to improve the photocatalytic activity of semiconductor photocatalyst, the coupling O-doped g-C3N4 (O-g-C3N4) with rutile TiO2 is a good strategy. Novel composite material O-g-C3N4/TiO2 was synthesized by gas phase method of deposition of O-doped g-C3N4 on particles of rutile powder under the special reactionary conditions of the pyrolysis of melamine. Obtaining O-g-C3N4/TiO2 binary composite was confirmed through various analytical techniques including X-ray powder diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS) and ultraviolet-visible diffuse reflectance spectra (UV-Vis-DRS) methods. It is found that the absorption spectra of the O-g-C3N4/TiO2 powders show a bathochromic shift of the long-wavelength edge of the fundamental absorption band (to 600 nm) relative to the absorption band of g-C3N4/TiO2 (~ 460 nm). As a result, O-g-C3N4/TiO2 photosensitivity is observed in the significant part of the visible region and the band gap of synthesized product is determined to be less than 2.4 eV versus 2.7 eV for undoped g-C3N4 or g-C3N4/TiO2. One stage constructing heterojunction structure of O-g-C3N4/TiO2 composite may be used as a low-cost way to avoid the limitations of each component and realize a synergic effect in promoting the efficient generation and separation of charge carriers, thus boosting the photocatalytic activity to enhancing solar energy utilization.


INTRODUCTION
Using sunlight by semiconductors for the production of renewable and clean energy and purification of water and air from toxic organic pollutants and pathogenic microflora is considered as a promising approach to address the imperative energy and environment problems [1][2][3][4][5]. By harvesting solar light to drive a series of important chemical reactions, semiconductor photocatalysis has emerged as one of the most fascinating technologies for wastewater treatment and converting low-density solar energy into high-density chemical energy.
Titanium dioxide (TiO 2 ), as one of the most efficient metal oxide semiconductor photocatalyst, is widely used for many applications such as hydrogen evolution, CO 2 reduction and wastewater treatment because of its low cost, high chemical stability [1][2][3][4].
However, its photocatalytic activity is limited by rapid photogenerated electron-hole recombination and, most importantly, its wide band gap energy of ca. 3 eV (~ 3.2 and ~ 3.0 eV for anatase and rutile, respectively) which is activated only under ultraviolet (UV) light irradiation that accounts for about 5 % of solar light.
On the other hand, graphite-like carbon nitride g-C 3 N 4 , characterized as a non-toxic and chemically highly resistant material, has been shown to have huge potential as a promising photocatalyst capable of absorbing visible light for renewable energy (in particular, hydrogen production by photocatalytic water splitting) and for broad range of environmental applications (decomposition of organic pollutants and the destruction of pathogenic microflora). However, the application of based on pristine g-C 3 N 4 catalysts allows us using only part of the visible spectrum of solar radiation (with a wavelength of less than 460 nm). Thus, in order to further enhance the light harvesting capability of g-C 3 N 4 , various bandgap engineering strategies, including doping by metal [6,7] or non-metal atoms, in particular, oxygen [5,[8][9][10][11][12], are used. A number of researches have shown that the doping g-C 3 N 4 by oxygen to form O-g-C 3 N 4 significantly improved its photocatalytic properties [5,[8][9][10][11][12][13][14][15][16][17]. As a result, the absorption spectra of the O-doped carbon nitride powders show a bathochromic shift of the longwavelength edge of the fundamental absorption band (to 498 nm or more) relative to the absorption band of pristine g-C 3 N 4 (~ 460 nm) [8,9,11,12]. It is found that O-modified g-C 3 N 4 presented intrinsic electronic/band structure modulation [9], resulting in extended light absorption range for more effective visible-light utilization, up-shifted conduction band for stronger reducibility and more effective separation of photogenerated charge carriers, which are beneficial for improving photocatalytic hydrogen evolution activity. For example, according to [9], when using thermally oxidized porous g-C 3 N 4 , the 1430.1 μmol·g −1 ·h −1 average photocatalytic hydrogen evolution is achieved in 8 h under visible-light irradiation, which was 4.3 times as high as that of the pristine g-C 3 N 4 sample (334.3 μmol·g −1 ·h −1 ). In report [10] is presented that O-doped material (O-g-C 3 N 4 ) shows 6.1 and 3.1 times higher hydrogen evolution reaction activity (with apparent quantum efficiency of 7.8 % at 420 nm) than bulk and even 3D porous g-C 3 N 4 . According to investigation [13], the photocatalytic hydrogen evolution rate of O-doped g-C 3 N 4 was 13.9 times higher than that of bulk g-C 3 N 4 .
Oxygen doped carbon nitride was successfully applied for production hydrogen peroxide (H 2 O 2 ) which is widely used in wastewater treatment, pulp bleaching, pharmaceuticals etc. [12]. According to [12], oxygen doping g-C 3 N 4 changes the catalyst properties, leading to the promoted photocatalytic H 2 O 2 production capability in the absence of hole scavengers. As-prepared oxygen doped g-C 3 N 4 displays the H 2 O 2 concentration of 3.8 mmol·L -1 , more than 7.6 times higher than that of neat g-C 3 N 4 [12].
Doping graphitic carbon nitride (g-C 3 N 4 ) with oxygen contributes to the effectiveness of the degradation of pollutants [14]. Thus, ozone treated carbon nitride increase photocatalytic rhodamine B (RhB) degradation constants by approximately 6 times [11], accelerates the photodegradation of methylene blue by a factor of 5 times and 2 times accelerates the generation of H 2 in comparison with untreated graphitic carbon nitride [4]. The 40 % metal-free oxygen doped porous graphitic carbon nitride (O-g-C 3 N 4 ) catalyst can degrade bisphenol A (15 mg·L −1 ) in 240 min with a mineralization rate as high as 56 % [15].
However, the known methods for oxygendoped carbon nitride (O-g-C 3 N 4 ) producing suggest a two-stage process, since the synthesis involves post-treatment of the pre-synthesized undoped g-C 3 N 4 with ozone, nitric acid or hydrogen peroxide [8,13,[15][16][17]. For example, long-term (10 h) heat treatment process (in a Teflon sealed autoclave at 140 °C) of the presynthesized g-C 3 N 4 by H 2 O 2 solution was used to obtain oxidized carbon nitride [8]. The pyrolysis method developed in Frantsevich Institute for Problems of Materials Science of NASU (IPM NASU) allows us to obtain oxygen-doped carbon nitride O-g-C 3 N 4 in one stage from melamine [18], urea [19][20][21] or cyanuric acid and urea mixture [22], and O-doped lownitrogen containing carbon nitride O-C 3 N from pyridine [23,24].
Besides the method of carbon nitride modification by oxygen, to increase the activity of the photocatalyst is also promising a different approach: creating g-C 3 N 4 based binary photocatalytic composite materials. For efficient light utilization, the fabrication and photocatalytic applications of g-C 3 N 4 /TiO 2 composite materials have attracted much attention [1][2][3][4]. The synergistic effect of the combination of g-C 3 N 4 and TiO 2 leads to a significant improvement in the absorption of visible light and the effective separation of photo-generated electron-hole pairs. This greatly enhances the photocatalytic activity of g-C 3 N 4 /TiO 2 composite material, for example, for inactivating against bacteria in water under visible light.
However, both pristine g-C 3 N 4 and g-C 3 N 4 /TiO 2 composite exhibits photoactivity in the visible spectrum with limited utilization of solar energy (in the part of the visible diapason with wavelength below 460 nm only). While for modified by oxygen graphitic carbon nitride samples (O-g-C 3 N 4 ) synthesized in IPM [18][19][20][21][22] photosensitivity extends to a much larger region of the visible spectrum. It is found [8-11, 13, 15] that the doping of carbon nitride by oxygen significantly improves its photocatalytic properties to enhancing solar energy utilization. Thereby, to improve the photocatalytic activity of semiconductor photocatalyst, the coupling O-doped graphitic carbon nitride (O-g-C 3 N 4 ) with rutile TiO 2 is a good strategy. Therefore, the aim of the present was to obtain a nanocomposite based on O-doped carbon nitride and rutile phase of titanium dioxide (binary composite O-g-C 3 N 4 /TiO 2 rutile phase).

EXPERIMENTAL
New composite material -O-g-C 3 N 4 /TiO 2 (rutile phase) was synthesized in accordance with the one-step method developed in IPM for the synthesis of oxygen-doped carbon nitride (O-g-C 3 N 4 ) by gas phase method under the special reactionary conditions of the pyrolysis [25] of melamine [18], urea [19][20][21] or cyanuric acid and urea mixture [22]. Formation of O-g-C 3 N 4 /TiO 2 composite was carried out under ambient pressure and pyrolysis-generated selfsupporting atmosphere at the presence of a fixed volume of air. Nanoparticles of O-doped carbon nitride are formed in an vapor-gas reactionary space and are located by means of the deposition and the condensation on particles of rutile powder in more lower temperature (concerning the most highly temperature reaction zone) zones of reactionary space, far from a place of melamine precursor localization. The light yellow colored O-g-C 3 N 4 /TiO 2 (rutile) was obtained by melamine heating with the rate not exceeding 10 deg•min -1 and heat treatment at 550-580 °C for 1-1.5 h.
Deposition of O-g-C 3 N 4 (~ 6 % O) on the particles of rutile powder is confirmed through various analytical techniques including X-ray powder diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS) and ultraviolet-visible diffuse reflectance spectra (UV-Vis-DRS) methods.
The powder X-ray diffraction (XRD) measurements were obtained from 5 to 80 ° of 2θ by a DRON-UM diffractometer with CuK α -radiation (λ = 0.154 nm) and nickel filter.
Fourier transform infrared spectroscopy (FTIR) in the reflectance mode was determined between 450-4000 cm -1 with the spectral resolution of 8 cm −1 using a Nexus Nicolet FTIR spectrometer (Thermo Scientific) equipped with a Smart Collector reflectance accessory. Samples under investigation were prepared by pressing with KBr in 1:100 ratio.
X-ray photoelectronic spectra (XPS) of the samples were measured with a photoelectron spectrometer manufactured by SPECS Surface Nano Analysis Company (Germany). The system is equipped with a PHOIBOS 150 hemispherical analyzer. A base pressure of a sublimation ion-pumped chamber of the system was less than 5×10 -10 mbar during the present experiments.
The Mg K α radiation (E = 1253.6 eV) was used as a source of XPS spectra excitation. The XPS spectra were measured at the constant pass energy of 25 eV. The energy scale of the spectrometer was calibrated by setting the measured Au4f 7/2 and Cu2p 3/2 binding energies to 84.00±0.05 and 932.66±0.05 eV, respectively, with regard to E F . For investigated sample, all the spectral features are attributed to the constituent element corelevels or Auger lines.
Diffuse reflectance spectra (DRS) of the powders were recorded in spectral diapason 200-1000 nm using a Perkin-Elmer Lambda Bio 35 UV-Vis with an integrating sphere Labsphere RSA-PR-20 with BaSO 4 as a scattering standard and converted into ultraviolet-visible (UV-Vis) absorption spectra by the Kubelka-Munk function [26].

RESULTS AND DISCUSSION
At the heat treatment of melamine at 570 °C for 1.5 h in the most highly temperature reaction zone, in the placed in lower temperature zone the ceramic crucible with rutile powder, the light yellow colored powdery product are formed. The product obtained at 570 °C for 1 h has an even lighter shade. The obtained samples were denoted as O-g-C 3 N 4 /TiO 2 -1.5h and O-g-C 3 N 4 /TiO 2 -1h. To confirm the accordance of the obtained light yellow powder with the O-g-C 3 N 4 /TiO 2 composition, in the experiments the synthesized (IPM) O-doped carbon nitride (O-g-C 3 N 4 ) [18], pristine g-C 3 N 4 and TiO 2 (rutile phase) powdered samples also used. To understand the formation of O-g-C 3 N 4 /TiO 2 , the crystal structure of O-g-C 3 N 4 , pristine g-C 3 N 4 , rutile phase TiO 2 and presumably obtained O-g-C 3 N 4 /TiO 2 composite were firstly Synthesis O-g-C3N4/TiO2 rutile composite material for photocatalytic application ______________________________________________________________________________________________ investigated. Diffractograms of products O-g-C 3 N 4 /TiO 2 -1.5h and O-g-C 3 N 4 /TiO 2 -1h are practically identical. In the diffractograms of the both rutile titanium dioxide (Fig. 1, curve 4) and obtained light yellow powdered product O-g-C 3 N 4 /TiO 2 -1.5h (Fig. 1, curve 3) there are strong peaks characteristic for the rutile phase (JCPDS No. 21-1276) with the most intense reflexes at 2θ = 27.4 ° (110), 36.2 ° (101) and 54.4 ° (211). As is known, there are two characteristic reflexes at 2θ = 27.49 ° (002) and 12.40 ° (100) in XRD patterns of pristine g-C 3 N 4 ( Fig. 1, curve 1). These reflexes are caused by an interlayer stacking of aromatic heteroatomic rings with an interplane distance (0.324 nm) and the periodicity of stacking of heptazine fragments in one layer (0.714 nm). However, for O-g-C 3 N 4 /TiO 2 rutile composite sample (Fig. 1, curve 3), one typical peak of both-g-C 3 N 4 and O-g-C 3 N 4 (002) is not clearly observed due to the coverage by that of rutile TiO 2 (110), while the absence of another peak at 12.4 ° of O-g-C 3 N 4 (Fig. 1, curve 2) is owing to its high dispersion on the surface of the composites. Occurrence of an additional reflex (wide galo) at 2θ = 21.45 ° (d = 0.414 nm) in the diffraction pattern of O-doped carbon nitride (O-g-C 3 N 4 ) (Fig. 1, curve 2) is caused, as it was offered in [18,19], with partial distortion of a planarity of its polymeric network ((C 6 N 7 )−N) n because of a dearomatization of some heterocycles at an oxidation. It is important to note that in the diffractogram of the O-g-C 3 N 4 /TiO 2 binary composite a wide halo at 21.45 ° (characteristic for O-doped carbon nitride) is also clearly observed (Fig. 1, curve 3). 10 20 To stronger evidence the existence of O-g-C 3 N 4 and TiO 2 in the as-obtained sample, the surface chemistry of O-g-C 3 N 4 /TiO 2 -1.5h was analyzed by XPS analysis. Four elements (carbon, nitrogen, oxygen and titanium) can be identified in the O-g-C 3 N 4 /TiO 2 composite sample in the survey scan XPS spectra (Fig. 2 a).
Under CVD conditions of the melamine precursor excess, the surface of the particles of rutile powder is almost completely covered with O-modified carbon nitride (according to XPS, the content of TiO 2 on the surface of the binary material is less than 10 %). This fact explains the weak intensity of the Ti peak on the survey spectrum. As one can see from the Fig. 2 b, the XPS Ti2p core-level spectrum recorded for this sample corresponds to that of titanium dioxide. As shown in the high-resolution Ti2p spectrum (Fig. 2 b), peaks at 458.8 and 464.5 eV are assigned to Ti2p 3/2 , and Ti2p 1/2 of TiO 2 .
In Fig. 3 a, the C1s spectra of O-g-C 3 N 4 /TiO 2 -1.5h could be deconvoluted into two main peaks cored at 284.6 and 288.0 eV. The major C1s peak at ~ 288 eV is identified as the tertiary carbon C−(N) 3 in the g-C 3 N 4 lattice. The weaker C1s peak at ~ 284.6 eV (which always present in the spectra of carbon nitride) is assigned to adventitious carbon species in good agreement with literature data [8,16,27]. Furthermore, the third weak peak at higher binding energy of ~290 eV was detected in O-g-C 3 N 4 /TiO 2 -1.5h sample, which was accredited to C=O or C-O [8,16,17], suggesting the partial oxidation and formation of oxygen functional groups.
The deconvolution of the N1s spectrum (Fig. 3 b) of the O-g-C 3 N 4 /TiO 2 composite showed the presence of two main peaks centered at 398.5 and 399.8 eV, which were attributed to sp 2 -hybridized nitrogen (C=N−C) and tertiary sp 3 -hybridized nitrogen (N−(C) 3 ) atoms in the carbon nitride network, respectively. The deconvolution of the O1s spectrum of the O-g-C 3 N 4 /TiO 2 composite showed the presence of two main maxima (Fig. 4). These peaks at binding energies about 531. Further demonstration of the formation of O-g-C 3 N 4 on rutile TiO 2 nanopowder surface is given by FTIR spectra (Fig. 5). IR spectra of all the samples (Fig. 5) contain, first of all, the peak near 810 cm -1 , which could be ascribed to the characteristic out-of-plane bending mode of tri-s-triazine (heptazine) units in both g-C 3 N 4 and O-g-C 3 N 4 [18,19]. Besides, in the IR spectra of Synthesis O-g-C3N4/TiO2 rutile composite material for photocatalytic application ______________________________________________________________________________________________ all samples there are several distinct intensive absorption bands in the 1200-1650 cm -1 region, which also are characteristic for g-C 3 N 4 and O-g-C 3 N 4 [18,19] and correspond to the stretching vibrations of aromatic CN bonds in condensed nitrogen-carbon heterocycles. It should be noted that the IR spectrum of a sample O-g-C 3 N 4 /TiO 2 -1h (Fig. 5, curve 2), which obtained with a shorter heat treatment time (1 h), is weakly structured. As one can see from the Fig. 5, curve 3, in the IR spectrum of obtained with a longer thermal treatment (for 1.5 h) sample O-g-C 3 N 4 /TiO 2 -1.5h (which contains more than 5 % oxygen) there are absorption bands characteristic of heptazine fragments of the g-C 3 N 4 structure, as well as bands of the -OH, >C=O and -COOH groups, which are characteristic only for oxidized carbon nitride. The peak at 1089 cm -1 is attributed to the stretching vibration of C-O, together with the N-H and O-H signals around 3000-3600 cm -1 generally related to the presence of hydroxyl groups. Most intensive signal becomes to emerge at ~ 1750 cm -1 , which suggests the formation of carbonyl (carboxyl) groups. The presence of carboxyl groups is an evidence of a stretching band at ~ 2700 cm -1 , characteristic for bond -O-Н of carboxyl group. Note that the presence oxygen-containing functional groups which confirms the oxygen is doped into O-g-C 3 N 4 lattice, is in good agreement with the literature data [8,12,14,16,28]. However, some researchers believe (and provides convincing evidence also) that O doping preferentially occurs at two-coordinated N position [10,13,15]. So, in [10] by experiment and DFT computation were identified that oxygen atoms preferentially replaced nitrogen atoms at twocoordinated N position. According to [10], after Ar + etching to remove surface layer, O-g-C 3 N 4 sample still shows this lattice oxygen, confirming that O atoms are uniformly doped in the matrix. In attempt to obtain more evidence relative to the structure alternation due to the O doping, in [15] EPR was conducted to detect the spin state of unpaired electrons. The bulk g-C 3 N 4 exhibits one single Lorentzian line with a g factor of 2.003 due to the unpaired electron of the π-bonded aromatic rings [15]. It has been found [15] that in the EPR spectrum of O-doped carbon nitride a shoulder peak appears, probably because substituting the sp 2 N atom in the heptazine unit with an O atom creates one unpaired electron. Thus, oxygen position in O-modified carbon nitride network currently still under discussion. Probably in the structure of the O-g-C 3 N 4 lattice, both the addition of oxygen in the form of functional groups and the substitution of two-coordinated nitrogen with oxygen is possible. So, according to [13], the FTIR spectrum and XPS analysis of O1s and C1s verifies that doping O atoms either take the Synthesis O-g-C3N4/TiO2 rutile composite material for photocatalytic application ______________________________________________________________________________________________ place of two-coordinated nitrogen atoms or exist as hydroxide radicals linked to C atoms in the C=N-C ring.
For the O-g-C 3 N 4 /TiO 2 samples, despite the strong light scattering of TiO 2 at low wavenumbers as well as the fact that he surface of the particles of rutile powder is almost completely covered with O-doped carbon nitride (according to XPS), a weak wide signal near 800 cm -1 was observed (Fig. 5, curves 2, 3).
UV-vis diffuse reflectance spectra (DRS) were used to study the light absorption properties of pure O-g-C 3 N 4 and O-g-C 3 N 4 /TiO 2 rutile composite compared to pristine rutile phase TiO 2 and undoped g-C 3 N 4 (Fig. 6 a). The optical absorption spectra reveal that the pure undoped g-C 3 N 4 and rutile phase TiO 2 samples feature an intrinsic semiconductor-like absorption in the blue region of the visible spectrum and ultraviolet region respectively. There is a bathochromic shift of the absorption edge which expands to the significant part of the visible light region for all synthesized powders doped with oxygen (from 460 nm for undoped g-C 3 N 4 to 600 nm with absorption trailing extended to more than 700 nm for O-g-C 3 N 4 in g-C 3 N 4 → O-g-C 3 N 4 /TiO 2 -1h → O-g-C 3 N 4 /TiO 2 -1.5h → O-g-C 3 N 4 range). It is assumed that in comparison with bulk g-C 3 N 4 , the absorption band edge of O-g-C 3 N 4 is red shifted due to the presence of O-doping, and samples with more O-dopants show more obvious red shift [10]. The absorption band edge of O-g-C 3 N 4 /TiO 2 composite is located between pure TiO 2 and oxygen modified O-C 3 N 4 samples, which further confirmed the electronic coupling of these two components in the O-g-C 3 N 4 /TiO 2 heterojunction. Overall, the absorption intensity of all oxygen-doped samples (pure O-g-C 3 N 4 , O-g-C 3 N 4 /TiO 2 -1h and especially O-g-C 3 N 4 /TiO 2 -1.5h) is significant higher than that of pristine g-C 3 N 4 in the full spectrum and the absorption edges of oxygenated products are red shifted. It is assumed [13], that the red shift in the absorption wavelength indicates that the introduction of oxygen results in the absorption of more light energy to produce more photogenerated electron-hole pairs, which contribute to the improvement in the photoactivity of the catalysts. The band gap energies of the materials were estimated by the Kubelka-Munk equation by transforming the spectra into (αhν) 1/2 versus hν (for g-C 3 N 4 , the n is ½ for the indirect band gap semiconductor). The band gap energy (E g ) estimated from the intercept of the tangents to the plots of (αhν) 1/2 vs photon energy decreases in the range O-g-C 3 N 4 < O-g-C 3 N 4 /TiO 2 -1.5h < O-g-C 3 N 4 /TiO 2 -1h < g-C 3 N 4 < TiO 2 (rutile phase) and has values of 2.2, 2.3, 2.4, 2.6, and 3.0 eV, respectively, as shown in the Fig. 6 b. The obtained data confirm that the construction of heterojunction structure can be responsible for the overall improvement of light adsorption intensity and show the possibility of practical use of oxygen-doped O-g-C 3 N 4 /TiO 2 composite in photocatalysis.