Facile Synthesis of Defective TiO2−x Nanocrystals with High Surface Area and Tailoring Bandgap for Visible-light Photocatalysis

A facile hydrothermal approach has been developed to prepare defective TiO2−x nanocrystals using Ti(III)-salt as a precursor and L-ascorbic acid as reductant and structure direction agent. The prepared TiO2−x nanocrystals are composed of a highly crystallized TiO2 core and a disordered TiO2−x outer layer, possessing high surface area, controlled oxygen vacancy concentration and tunable bandgap via simply adjusting the amount of added L-ascorbic acid. The defective TiO2−x shows high photocatalytic efficiency in methylene blue and phenol degradation as well as in hydrogen evolution under visible light, underlining the significance of the present strategy for structural and bandgap manipulation in TiO2-based photocatalysis.

TiO 2 is one of extensively studied photocatalytic materials due to its excellent physicochemical properties as well as its earth abundance, nontoxicity and stability [1][2][3][4] . However, the large band gap (3.2 eV for anatase) limits its utilization of sunlight and thus its practical applications in many important fields such as photocatalytic hydrogen evolution 5 , environmental remediation 6 and solar energy conversion 7 . To overcome this barrier, great efforts have been devoted to engineer TiO 2 ′ s band gap from a variety of aspects in order to approach a high photoactivity under visible light irradiation. Introducing dopants is one of the most intensively investigated strategies to enhance its visible light utilization. Initially, metal ions were used as dopants to introduce states into the TiO 2 band gap [8][9][10] , but the subsequential problems such as thermal instability, increased carrier recombination centers 11,12 , and the need for an expensive ion-implantation facility significantly limit its applications. Nonmetal elements was also adopted 13,14 , if doped under the right conditions, can effectively narrow band gap of TiO 2 and improve its visible-light absorption. Compared to other elements, nitrogen doping is thought to be the most useful choice to enhance its visible-light photocatalytic activity 15 . But unfortunately, further studies find that the activity of N-TiO 2 for visible-light induced hydrogen evolution is quite low 16 .
Recently, the intrinsic defects in TiO 2 matrix such as oxygen vacancy (Vo) and Ti 3+ have been proved to trigger the visible-light activity of TiO 2 [17][18][19][20][21][22][23][24] . Chen et al. 17 have reported that hydrogen thermal treatment of TiO 2 nanoparticles can generate an amorphous layer near the surface to form defective black TiO 2−x nanoparticles. Such defective black TiO 2−x nanoparticles display excellent photoactivity and stability in photocatalytic hydrogen generation. Theoretical calculations demonstrate that high vacancy concentration could induce a vacancy band of electronic states just below the conduction band and Scientific RepoRts | 5:15804 | DOi: 10.1038/srep15804 narrow the band gap to 1.0 eV. Except its optical and electronic properties, the performance of defective TiO 2−x also depends largely on its morphological and structural properties [25][26][27] , like its surface area 28 , particle size and pore structure 29 . However, to the best of our knowledge, rare work has been reported on controlling optical and morphological structure properties simultaneously for defective TiO 2−x . Herein, we report a facile hydrothermal approach to produce defective TiO 2−x nanocrystals with high surface area and tailoring band gap using TiCl 3 as a precursor and L-ascorbic acid as reductant and structure direction agent.

Results
The obtained defective TiO 2−x samples display diffraction peaks at 25.4°, 37.9°, 48.1° and 53.1° in the XRD patterns, indicating that as synthesized TiO 2−x is in a pure anatase phase (Fig. 1a). It suggests that the presence of L-ascorbic acid does not influence the crystal structure of TiO 2−x . TEM analysis with the obtained defective TiO 2−x samples shows that the particle size of highly crystallized TiO 2 core gradually decreases from 50 nm to 10 nm (Figs. 1b-1d) as increasing the amount of L-ascorbic acid from 0, 0.3 to 0.7 g in 80 mL solution. The crystal lattices of TiO 2 core can be clearly identified in all the samples, indicating that the TiO 2 core is highly crystallized. Since the particles are overlapped, the disordered TiO 2−x outer layer cannot be distinguished in TEM, but its presence can be indirectly confirmed via further sintering of the defective black TiO 2−x at 500 °C in N 2 atmosphere considering that high concentration of vacancy could significantly reduce the melting temperature of TiO 2 25,26 . The sintered TiO 2−x exhibits a nanoscroll structure with several layers of TiO 2−x sheets and the interlayer spacing is 0.76 nm (see Fig.  S1a for details). EDX measurement (Fig. S1b) confirms that the TiO 2−x nanoscroll is only composed of Ti (56.4 at%) and O (28.8 at%) elements, implying the presence of defective TiO 2−x layer surrounding the TiO 2 core. It is worth to mention, the defective TiO 2−x nanosheet is a fascinating structure to apply in Li-ion battery and photocatalysis for its ultrahigh cycle rate and high efficiency 30,31 . This report provides a new and simple method for preparing the defective TiO 2−x nanosheets.
The N 2 adsorption-desorption isotherms (Fig. 2a) of defective TiO 2−x samples exhibit a typical type-IV isotherm with a distinct hysteretic loop, indicating mesoporous features. The average pore size (Fig. 2b) decreases with increasing the amount of added L-ascorbic acid, while the corresponding BET surface area (Tab. S1) dramatically increases from 64.56 m 2 g −1 (white TiO 2−x ), 188.75 m 2 g −1 (brown TiO 2−x ) to 263.95 m 2 g −1 (black TiO 2−x ). Therefore, the presence of L-ascorbic acid molecules significantly affects the pore structure and surface area of the obtained TiO 2−x samples. The porous structure makes the defective TiO 2−x materials suitable for photocatalysis application because of their abundant porous channels.
Electron paramagnetic resonance (EPR) measurements were conducted at room temperature to verify the presence of high concentration Vo. As shown in Fig. 3a, all the defective TiO 2−x samples show a very strong EPR signal at g-value of 2.003 which indicates the significant presence of Vo. As discussed previously, the EPR signal appearing at g-value of 2.003 is caused by electrons trapped on surface Vo 32,33 . However, the representative signal of Ti 3+ usually appearing at g ≈ 1.94 34,35 is not shown here, which suggests the absence of rhombic Ti 3+ in defective TiO 2−x samples. Furthermore, the EPR signals of brown  and black TiO 2−x show a great enhancement in the intensity, which indicates their substantially increased Vo concentrations due to the increase in the amount of added L-ascorbic acid. Therefore, the defects in the colorful TiO 2−x are present as Vo instead of Ti 3+ . X-ray photoelectron spectroscopy (XPS) characterization of the white, brown and black TiO 2−x samples (Fig. 3b) shows only Ti2p 1/2 and Ti2p 3/2 peaks with slightly difference but no Ti 3+ peaks appeared, which further confirms that Ti 3+ does not exist in the defective TiO 2−x nanocrystals. Structural properties of the obtained TiO 2−x samples were further examined by measuring Raman scattering. For comparison, P25 Degussa was also analyzed. As shown in Fig. 3c, the six (3Eg + 2B1g + A1g) Raman-active modes of anatase phase with frequencies at 144, 197, 399, 515, 519 (superimposed with the 515 cm -1 band), and 639 cm -1 were detected in all investigated samples. Compared with P25, the defective TiO 2−x nanocrystals display a varying degree of blue-shift in Raman bands (Eg from 139 to 144, 153 and 164 cm −1 , respectively), which indicates that the original symmetry of TiO 2 lattice is broken down due to the disordered TiO 2−x layer formed by introducing of L-ascorbic acid 36 . In addition, the peak shift of defective TiO 2−x gradually increases along with the darkening color, suggesting the increase in the Vo concentration. This result further supports that the Vo concentration could be controlled by the amount of L-ascorbic acid directly. The optical property of all the defective TiO 2−x powder samples was characterized by UV-visible diffuse reflectance spectra (DRS), as shown in Fig. 3d. The white TiO 2−x exhibits a strong absorption in the UV range, while the brown and black TiO 2−x samples display a broad absorption over the entire UV-vis wavelength range investigated. Furthermore, the black TiO 2−x sample shows even higher absorption of UV-vis light than the brown TiO 2−x sample which further confirms the assumption that the high concentration of Vo is capable of generating a new vacancy band locating just below the conduction band edge of pure TiO 2 .
The photocatalytic degradation of MB (20 mg/L) and phenol (10 mg/L) was performed using 0.5 g/L of the as-synthesized white, brown and black TiO 2−x powders under a 300 W Xenon lamp with UV cut-off filter (λ > 420 nm). As shown in Figs. 4a and 4b, the black TiO 2−x exhibits the highest efficiency in both MB and phenol degradation. The kinetic reaction rate of MB degradation at black TiO 2−x is 1.98 × 10 −2 min −1 , while it is 1.18 × 10 −2 min −1 at the brown TiO 2−x , and 0.76 × 10 −2 min −1 at the white TiO 2−x . Similar improvement was also observed for phenol degradation, the kinetic rate at black TiO 2−x is 2.95 × 10 −2 min −1 , which is about 25.41 times greater than that at white TiO 2−x . The phenol molecules could be totally decomposed at black TiO 2−x nanocrystals under visible light irradiation in about 80 min.

Figure 4. Degradation of MB (a) and phenol (b), hydrogen evolution (c) and cycle test (d) of the defective TiO 2−x nanocrystals under visible light (>420 nm) .
Moreover, the photocatalytic activity of the defective TiO 2−x was also tested for hydrogen generation under visible light (Fig. 4c). All the catalysts were loaded with 0.6 wt% Pt, and methanol was used as sacrificial agent. The black TiO 2−x nanocrystals show a much higher photocatalytic activity with a hydrogen evolution rate of 116.7 μ mol g −1 h −1 compared to the white TiO 2 (20.6 μ mol g −1 h −1 ) and brown TiO 2−x (38.9 μ mol g −1 h −1 ) under visible light irradiation in the presence of 20 mg/L photocatalyst. The cycling test results of the visible-light driven photocatalytic activity of black TiO 2−x nanocrystals for hydrogen evolution as a function of time during a 25-hour testing period are shown in Fig. 4d. No noticeable decrease in H 2 production rate for black TiO 2−x in 5 cycle tests were observed within the test period, indicating good stability of the black TiO 2−x nanocrystals in the photocatalytic production of hydrogen from water under visible light. The photocatalysis results confirm that the presence of defective TiO 2−x outer layer with well-developed porosity is the key factor leading to improved photoactivity. The high Vo concentration in the defective TiO 2−x nanocrystals enhances the absorption of visible light and thereby the generation of charge carriers, which are further transformed into abundant active species of ·O 2 − and ·OH to degrade the pollutants and split water to produce H 2 under the visible light irradiation. The high surface area and rich pore structure increase the collision possibility of the pollutant molecules with catalyst surface and the adsorbed active radicals. These factors are responsible for the enhancement of photocatalytic activity of defective TiO 2−x nanocrystals for pollutants degradation and hydrogen evolution. This is further confirmed by the measurement of photocurrent densities with the defective TiO 2−x nanocrystals photoanodes at a constant potential of 0 V (vs Ag/AgCl) under visible light (Figs. 5 and 6). The photocurrent density of black TiO 2−x is the highest in all the samples, which is almost 10 times of that of white TiO 2−x . Thus, the high concentration of Vo defects gives rise to high visible-light induced photo-electron transformation and results in high efficiency in photocatalytic activity.

Discussion
Due to its unique optical property and superior visible-light-driven photocatalytic activity, defective TiO 2−x , has attracted plenty of attentions recently. But most of the reports focused on tailoring the photochemical properties, rare study is reported on its structure and morphology. To the best of our knowledge, the highest surface area of defective TiO 2−x reported was near 90 m 2 /g 37 . In this report, we presented a novel method for fabricating defective TiO 2−x nanocrystals with tunable bandgap and high surface area using Ti(III)-salt as a precursor and L-ascorbic acid as reductant and structure direction agent. The formation of defective TiO 2−x nanocrystals is schematically shown in Fig. 5. During hydrolysis, the -TiOH 2+ was formed first (Eq. 1) and oxidized by the dissolved oxygen to the Ti(IV)-oxo species (Eq. 2) 38 , and L-ascorbic acid molecules were adsorbed on the initial particle surface through  Ti-O-C bond, while the excessive Ti 3+ was diffused in the interspace of L-ascorbic acid molecules. The Ti(IV)-oxo species is assumed to be an intermediate between TiO 2+ and TiO 2 , consisting of partially dehydrated polymeric Ti(IV)hydroxide 39 . Following hydrothermal process, Ti(IV)-oxo was transferred to highly crystallized TiO 2 (Eq. 3). If without the presence of L-ascorbic acid molecules, the excessive Ti 3+ would react with the dissolved oxygen molecules in the solution to grow into TiO 2 crystals. But the chemical adsorption of L-ascorbic acid inhibited the diffusion of dissolved oxygen molecules to the TiO 2 core surface, and hindered the growth of TiO 2 crystals 40,41 . Since the oxidation process suffered from the insufficient supply of oxygen, after removing the adsorbed L-ascorbic acid, Vo was produced outside of the TiO 2 core 20 . Therefore, a defective, nonstoichiometric TiO 2−x layer was formed surround the TiO 2 core with rich oxygen vacancies. In addition, abundant pore structures were formed in the defective TiO 2−x layer after removing the adsorbed L-ascorbic acid, evidencing its critic role as a structure direction agent. In summary, a facile hydrothermal approach has been developed for preparing defective TiO 2−x nanocrystals with TiCl 3 as a precursor. L-ascorbic acid plays critical role in controlling the morphology and bandgap structure towards engineering the prepared defective TiO 2 . The defects in the defective TiO 2−x nanocrystals proved to be of Vo, while the Vo concentration and band gap of the defective TiO 2−x nanocrystals could be easily tailored by varying the introduced amount of L-ascorbic acid. Comparing with the white and brown defective TiO 2−x , the black TiO 2−x shows much higher surface area and efficiency in degradation of organic pollutants (MB and phenol) and hydrogen evolution under visible light irradiation. The present work provides an alternative approach for fabricating defective TiO 2−x nanocrystal photocatalysts with controllable band gap and morphological structure for environmental remediation and solar fuel generation.

Method
Materials synthesis. For the preparation of reduced TiO 2 nanocrystals, different amounts of L-ascorbic acid (0, 0.3 g and 0.7 g) were added to 70 mL DI water and stirred for 10 min at RT. Subsequently, 3.1 mL of TiCl 3 was added and a purple solution was formed. Then, NaOH solution (1 mol/L) was added to raise the pH to 4. After stirring for another 30 min at RT, the mixture was transferred to a 100 mL Teflonlined stainless steel autoclave and heated at 180 °C for 12 h. The obtained precipitates were collected by centrifugation, rinsed with water and ethanol for several times. After drying at 80 °C for overnight, the defective TiO 2−x samples were labeled according to its color as white, brown and black TiO 2−x .
Characterization. X-ray diffraction (XRD) patterns of the samples were collected on Bruker D8 Advance powder diffractometer over scattering angles from 20° to 80° using Cu Kα radiation. Transmission electron microscopy (TEM) characterization was performed on a JEOL-JEM 2100 electron microscope. Optical property was examined by UV− Visible diffuse reflectance spectrophotometer (DRS) (Shimadzu SolidSpec-3700DUV). The electron paramagnetic resonance (EPR) spectra were characterized with Bruker E500 Spectrophotometer. X-ray photoelectron spectra (XPS) of the samples were measured using a Kratos Analytical AMICUS XPS instrument.
Methylene blue (MB) and phenol degradation. 40 mL MB (20 mg/L) solution or phenol solution (10 mg/L) was placed in a 50 mL quartz photoreactor. The photocatalyst (0.5 g/L) was dispersed into the solution at neutral pH. In order to attain adsorption-desorption equilibrium, the solution was stirred in dark for 40 min. The solution was then irradiated by a 300 W Xenon lamp with UV cut-off filter (λ > 420 nm) at RT. Samples were taken at given time interval to test the concentration of MB and phenol. The concentration of MB was measured by UV-visible spectrophotometer (UV-1800, Shimadzu). The phenol concentration was determined by Thermo Fisher Ultra 3000 HPLC equipped with a 25 cm × 4.6 mm Cosmosil C18 column.
Photocatalytic hydrogen evolution. The photocatalytic reactions of H 2 evolution were carried out in a closed gas circulation system with an external-irradiation type of a glass reactor. The light source was a 300 W Xenon lamp with UV cut-off filter (λ > 420 nm). The co-catalyst Pt was loaded by an in-situ photodeposition method. The 0.6 wt% of Pt-loaded catalyst (25 mg) was dispersed with a magnetic stirrer in a methanol aqueous solution (10 mL of CH 3 OH and 90 mL of H 2 O). The evolved gas including H 2 was analyzed using an online gas chromatograph (7890A, Agilent) equipped with a thermal conductivity detector (TCD).
Scientific RepoRts | 5:15804 | DOi: 10.1038/srep15804 Photocurrent measurement. The photocurrent was performed with an electrochemical instrument CHI660E using a three-electrode system. The samples (0.1 g) were loaded on conductive surface of ITO glass and 0.5 M Na 2 SO 4 solution was used as electrolyte. 300 W Xenon lamp equipped with UV cut-off filter (λ > 420 nm) was used as light source, and standard calomel electrode (SCE) as reference electrode, Pt slice as counter electrode.