Grey Rutile TiO2 with Long-Term Photocatalytic Activity Synthesized Via Two-Step Calcination

Colored titanium oxides are usually unstable in the atmosphere. Herein, a gray rutile titanium dioxide is synthesized by two-step calcination successively in a high-temperature reduction atmosphere and in a lower-temperature air atmosphere. The as-synthesized gray rutile TiO2 exhibits higher photocatalytic activity than that of white rutile TiO2 and shows high chemical stability. This is attributed to interior oxygen vacancies, which can improve the separation and transmission efficiency of the photogenerated carriers. Most notably, a formed surface passivation layer will protect the interior oxygen vacancies and provide long-term photocatalytic activity.


Introduction
Among titanium oxides, TiO 2 is well investigated in the photocatalysis research field because of its high chemical stability, low cost, and nontoxicity [1]. However, it can only absorb ultraviolet light, resulting in low photocatalytic efficiency. To expand its light absorbance range and enhance the separation efficiency of the photogenerated carriers, many efforts, such as doping with other elements, sensitizing with dyes, and coupling with metal or nonmetal nanoparticles or different semiconductor materials, have been made to solve the aforementioned problems [2][3][4][5][6][7]. Very recently, TiO 2 nanotubes synthesized via the electrochemical anodization of titanium foil exhibited visible light response characteristics for the photodecomposition of formaldehyde [8].
It has been reported that when TiO 2 is partially reduced by H 2 or CO, or bombarded by high-energy particles (laser, electron, or Ar + ), the obtained colored TiO 2 powers show visible light photocatalytic activity. In 2010, a blue titanium dioxide with a mixture of anatase and rutile phase was synthesized via hydrolysis and the reduction of isopropyl titanium, showing a higher photocatalytic activity than that of commercial anatase TiO 2 . The higher photocatalytic activity was attributed to the presence of Ti 3+ in the interior of the titanium dioxide crystal [9]. In 2011, Giamello et al. used an isotope labeling method to study the existence of Ti 3+ in rutile titanium dioxide in detail [10]. In the same year, a black TiO 2 with a strong absorption of visible light was synthesized via a high-temperature hydrogenation reduction of P25 TiO 2 by Chen et al. The obtained higher photoactivity of the black TiO 2 was attributed to the reduced band gap of titanium dioxide caused by the generation of the surface disordered structure [11]. Another kind of simple method to produce colored TiO 2 is the addition of fluorine species during TiO 2 preparation [12,13]. In 2014, Xu et al. synthesized stable blue TiO 2 nanoparticles with a non-stoichiometric TiO 2−x core and stoichiometric TiO 2 shell structure for the photodecomposition of methylene blue (MB) dyes under visible light irradiation [12]. In addition to

Materials
Hexanoic acid (HA), tetrabutyl titanate (TBOT), methylene blue (MB) dye, and glucose were purchased from Sinopharm Chemical Reagent Company. All chemicals were of AR grade. The ultrapure water used in the experiment was obtained from a Mill-Q (electric resistivity 18.2 MΩ·cm) water purification system.

Synthesis of TiO 2 -GR and TiO 2 -WR
First, uniform spherical anatase TiO 2 particles (Figure 1a and Figure S1a) with a diameter of 200-300 nm were synthesized via a previous reported method [33]. In a typical process, hexanoic acid (0.46 g) dissolved in ethanol (230.0 mL), and TBOT (1.70 g, 10% ethanol solution) was mixed by stirring at room temperature. Then, 35.0 mL H 2 O was dropped into the mixture with vigorous stirring for 12 h at room temperature. The products were obtained after centrifugal separation and were then ready for use for the next two-step calcination procedure.
Second, the as-prepared TiO 2 nanosphere was firstly calcinated in a tubular high-temperature furnace with continuous argon flow at 900 • C for 3 h. Then, it was further calcined at 500 • C in air atmosphere for 10 h, and gray rutile TiO 2 particles with polyhedron morphology were obtained (TiO 2 -GR, Figure 1b and Figure S1b). A white rutile TiO 2 used as a reference sample (TiO 2 -WR, Figure 1c and Figure S1c) was prepared by the calcination of the as-prepared TiO 2 nanosphere at 900 • C for 3 h in an air atmosphere.
The as-prepared photocatalysts were stored in an air atmosphere at room temperature.

Photocurrent Measurements
The photocurrent measurements were carried out on an electrochemical analyzer (CHI660D Instruments, Shanghai Chenhua Instrument Co., Ltd., Shanghai, China) using a standard three-electrode system. The as-prepared samples, a commercial Pt gauze electrode (Gaoss Union Technology Co., Ltd., Wuhan, China, 2 cm × 2 cm, 60 mesh), and saturated calomel electrode were used as working electrodes, counter electrode, and reference electrode, respectively. The working electrode was prepared as follows: 0.05 g of the sample was ground with 0.10 g terpinol for 10 min to make uniform slurry. Then, the slurry was evenly dripped onto a 4.0 cm × 1.0 cm indium tin oxide-coated glass (ITO glass) electrode masked by an adhesive tape with thickness of 0.5 mm and smoothed by a doctor's blade. Therefore, the formed film about had a thickness of 0.5 mm. Next, these electrodes were dried in an oven and were calcined at 350 • C for 30 min in an air atmosphere. The electrode was immersed in a 0.10 M NaClO 4 aqueous solution to measure the transient photocurrent under a 300 W Xe arc lamp irradiation with an incident light power density of 130 mW/cm 2 at 0.4 V vs. the saturated calomel electrode. exposed to 1000 W Xe lamp irradiation with or without the light cutoff filters (λ > 420 nm), under ambient conditions and magnetic stirring. At given time intervals, the reaction solution was sampled and analyzed by a UV-visible spectrophotometer (UV 2250, Shimadzu, SHIMADZU (CHINA) Co., Ltd., Shanghai, China).

Results and Discussion
A spherical anatase TiO2 (Figure 1a and Figure S1a) in a white color was fabricated as the raw material for the gray TiO2-GR via the hydrolysis of TBOT in the presence of alkyl chain carboxylic acids [33]. It was determined that alkylchain carboxylic acids remained on the surface of the TiO2 nanospheres, which were used as a reductant for the subsequent high-temperature reduction of titanium dioxide [4]. Both the gray rutile TiO2 and the reference sample (TiO2-WR) exhibited polyhedron morphology (TiO2-GR, Figure 1b and Figure S1b). After calcination at 900 °C in an Ar atmosphere, it can be seen from the High Resolution Transmission Electron Microscope (HRTEM) pattern that a surface layer composed of a large number of microcrystals surrounded by a disordered structure formed on the obtained gray TiO2 ( Figure S2a). The disordered structure is believed to be mainly caused by the presence of oxygen vacancies, which are response for the black color of TiO2 [14,30]. Then, after further calcination at 500 °C in an air

Photoactivity Measurements
The photocatalytic discoloration of MB dyes was performed on a reformative XPA-7 photocatalytic reaction instrument(Xujiang Electromechanical Plant, Nanjing, China). The incident light power was 162 mW/cm 2 , which was measured by a handheld Optical Power Meter (Newport 1916-R, Newport Corporation, California, CA, USA). The light exposure area of the quartz bottle was about 19.1 cm 2 . The discoloration effect was measured using the absorption spectroscopic technique. In the typical process, an aqueous solution of the MB dyes (10.0 mg/L and 30.0 mL) and 20.0 mg of the as-prepared photocatalysts were mixed in a 50 mL cylindrical quartz tube and left overnight in darkness to reach the adsorption equilibrium for the MB dyes. Then, the mixture was exposed to 1000 W Xe lamp irradiation with or without the light cutoff filters (λ > 420 nm), under ambient conditions and magnetic stirring. At given time intervals, the reaction solution was sampled and analyzed by a UV-visible spectrophotometer (UV 2250, Shimadzu, SHIMADZU (CHINA) Co., Ltd., Shanghai, China).

Results and Discussion
A spherical anatase TiO 2 (Figure 1a and Figure S1a) in a white color was fabricated as the raw material for the gray TiO 2 -GR via the hydrolysis of TBOT in the presence of alkyl chain carboxylic acids [33]. It was determined that alkylchain carboxylic acids remained on the surface of the TiO 2 nanospheres, which were used as a reductant for the subsequent high-temperature reduction of titanium dioxide [4]. Both the gray rutile TiO 2 and the reference sample (TiO 2 -WR) exhibited polyhedron morphology (TiO 2 -GR, Figure 1b and Figure S1b).
After calcination at 900 • C in an Ar atmosphere, it can be seen from the High Resolution Transmission Electron Microscope (HRTEM) pattern that a surface layer composed of a large number of microcrystals surrounded by a disordered structure formed on the obtained gray TiO 2 ( Figure S2a). The disordered structure is believed to be mainly caused by the presence of oxygen vacancies, which are response for the black color of TiO 2 [14,30]. Then, after further calcination at 500 • C in an air atmosphere, a dense layer with ordered lattice was formed by the refilling of oxygen atoms into the oxygen vacancies on the outmost layer of TiO 2 particles ( Figure S2b). The formed dense layer with the size of 2-5 nm is on the outermost layer of the TiO 2 -GR particle, which would act as a surface passivation layer to hinder the further diffusion and infiltration of oxygen molecules into the interior oxygen vacancies. As a result, the interior lattice disordered structure would be retained. The lattice width of the surface passivation layer is 0.21 nm, which is ascribed to the (210) crystal faces of the rutile TiO 2 (JCPDS 21-1276; Figure S2b). However, no such surface layer structures were observed on the surface of the TiO 2 -WR particles ( Figure S2c). The lattice widths are 0.32 nm and 0.25 nm, which belong to the (110) and (101)  . The Rietveld analysis (TOPAS V 6.0) of the XRD patterns shows that TiO 2 -GR has an average particle size of 48.5 nm. This result is different from that intuitively observed from the HRTEM patterns, which is attributed to the different detection areas between XRD and HRTEM. Herein, the HRTEM patterns are mainly afforded the surface layer crystal structure of TiO 2 particles. Therefore, it can be inferred that the as-synthesized TiO 2 -GR nanoparticles are mainly composed of microcrystals with an average size of about 48.5 nm, while their surface layers are composed of smaller microcrystals. Additionally, the analysis results indicate that the broadening of the diffraction peak is mainly due to grain refinement, and there is no existing microstrain. However, compared with the cell parameters of the standard rutile TiO 2 (PDF# 87-0710), both TiO 2 -GR and TiO 2 -WR show a lattice expansion, with average lattice distortions of 0.11% and 0.13%, respectively. It is proposed that this is mainly caused by the different treatments during the high-temperature calcination process. No peaks centered at 2θ = 25.9 • , ascribed to carbon (JCPDS 26-1079), were observed in the XRD patterns of gray TiO 2 [34]. atmosphere, a dense layer with ordered lattice was formed by the refilling of oxygen atoms into the oxygen vacancies on the outmost layer of TiO2 particles ( Figure S2b). The formed dense layer with the size of 2-5 nm is on the outermost layer of the TiO2-GR particle, which would act as a surface passivation layer to hinder the further diffusion and infiltration of oxygen molecules into the interior oxygen vacancies. As a result, the interior lattice disordered structure would be retained. The lattice width of the surface passivation layer is 0.21 nm, which is ascribed to the (210) crystal faces of the rutile TiO2 (JCPDS 21-1276; Figure S2b). However, no such surface layer structures were observed on the surface of the TiO2-WR particles ( Figure S2c). The lattice widths are 0.32 nm and 0.25 nm, which belong to the (110) and (101)  . The Rietveld analysis (TOPAS V 6.0) of the XRD patterns shows that TiO2-GR has an average particle size of 48.5 nm. This result is different from that intuitively observed from the HRTEM patterns, which is attributed to the different detection areas between XRD and HRTEM. Herein, the HRTEM patterns are mainly afforded the surface layer crystal structure of TiO2 particles. Therefore, it can be inferred that the as-synthesized TiO2-GR nanoparticles are mainly composed of microcrystals with an average size of about 48.5 nm, while their surface layers are composed of smaller microcrystals. Additionally, the analysis results indicate that the broadening of the diffraction peak is mainly due to grain refinement, and there is no existing microstrain. However, compared with the cell parameters of the standard rutile TiO2 (PDF# 87-0710), both TiO2-GR and TiO2-WR show a lattice expansion, with average lattice distortions of 0.11% and 0.13%, respectively. It is proposed that this is mainly caused by the different treatments during the high-temperature calcination process. No peaks centered at 2θ = 25.9°, ascribed to carbon (JCPDS 26-1079), were observed in the XRD patterns of gray TiO2 [34]. The chemical state of the surface species of TiO2-GR and TiO2-WR was determined by X-ray photoelectron spectroscopy (XPS), which was further analyzed by an XPS peak-fitting program (version 4.0, Hong Kong, China). The C1s XPS peaks of TiO2-GR centered at 284.6 eV (FWHM = 4.55 eV) and 282.9 eV (FWHM = 1.96 eV), which is similar to that of TiO2-WR, were ascribed to the *C and (*CO)Ti species caused by the carbon contaminant (Figure 3a,b) [35,36]. It was reported that if a carbon atom was doped in the crystal lattice of TiO2, a bonding energy peak ascribed to C* or Ti*-C emerged at 281.6 eV or 454.90 eV, respectively [37,38]. However, no such carbon bonding energy peaks were observed for either TiO2-GR or TiO2-WR. This indicates that there were no carbon atoms doped in the crystal lattice of the prepared gray TiO2, which means that the gray color did not originate from the carbon residues. As shown in Figure 3c,d, the Ti2p XPS peaks of TiO2-GR were The chemical state of the surface species of TiO 2 -GR and TiO 2 -WR was determined by X-ray photoelectron spectroscopy (XPS), which was further analyzed by an XPS peak-fitting program (version 4.0, Hong Kong, China). The C1s XPS peaks of TiO 2 -GR centered at 284.6 eV (FWHM = 4.55 eV) and 282.9 eV (FWHM = 1.96 eV), which is similar to that of TiO 2 -WR, were ascribed to the *C and (*CO)Ti species caused by the carbon contaminant (Figure 3a,b) [35,36]. It was reported that if a carbon atom was doped in the crystal lattice of TiO 2 , a bonding energy peak ascribed to C* or Ti*-C emerged at 281.6 eV or 454.90 eV, respectively [37,38]. However, no such carbon bonding energy peaks were observed for either TiO 2 -GR or TiO 2 -WR. This indicates that there were no carbon atoms doped Nanomaterials 2020, 10, 920 5 of 11 in the crystal lattice of the prepared gray TiO 2 , which means that the gray color did not originate from the carbon residues. As shown in Figure 3c,d, the Ti2p XPS peaks of TiO 2 -GR were centered at 464.53 eV (FWHM = 2.84 eV) and 458.20 eV (FWHM = 3.82 eV), which are ascribed to Ti 4+ 2p 1/2 and Ti 4+ 2P 3/2 of TiO 2 , respectively [39], which is similar to that of TiO 2 -WR. Two reduced titanium ion XPS peaks centered at 462.43 eV (FWHM = 3.91 eV) and 455.92 eV (FWHM = 2.63 eV) were observed for TiO 2 -GR, which could be ascribed to the low valence state titanium of nonstoichiometric TiO 2−x (0 < X < 2), mainly including Ti 3+ 2p 1/2 of Ti 2 O 3 and Ti 2+ 2p 3/2 of TiO [40,41], which is consistent with the O 1s XPS peak results. However, the TiO 2 -WR showed no reduced titanium ion XPS peaks. The O 1s XPS peaks mainly consisted of three components (Figure 3e,f). The two peaks centered at 529.47 eV (FWHM = 3.30 eV) and 531.58 eV (FWHM = 4.48 eV) were ascribed to the lattice oxygen of the stoichiometric TiO 2 [42] and nonstoichiometric TiO 2−x (0 < X < 2) [43,44], respectively, and the latter may also include some hydroxyl oxygen species [45]. The small O 1s peak centered at 527.51 eV (FWHM = 2.21 eV) could be attributed to the attached ionic oxygen of CO or O 2 [46]. , which are ascribed to Ti 4+ 2p1/2 and Ti 4+ 2P3/2 of TiO2, respectively [39], which is similar to that of TiO2-WR . Two reduced titanium ion XPS peaks centered at 462.43 eV (FWHM = 3.91 eV) and 455.92 eV (FWHM = 2.63 eV) were observed for TiO2-GR, which could be ascribed to the low valence state titanium of nonstoichiometric TiO2−x (0 < X < 2), mainly including Ti 3+ 2p1/2 of Ti2O3 and Ti 2+ 2p3/2 of TiO [40,41], which is consistent with the O 1s XPS peak results. However, the TiO2-WR showed no reduced titanium ion XPS peaks. The O 1s XPS peaks mainly consisted of three components (Figure 3e,f). The two peaks centered at 529.47 eV (FWHM = 3.30 eV) and 531.58 eV (FWHM = 4.48 eV) were ascribed to the lattice oxygen of the stoichiometric TiO2 [42] and nonstoichiometric TiO2−x (0 < X < 2) [43,44], respectively, and the latter may also include some hydroxyl oxygen species [45]. The small O 1s peak centered at 527.51 eV (FWHM = 2.21 eV) could be attributed to the attached ionic oxygen of CO or O2 [46]. It has been reported that the produced Ti 3+ originated from the oxygen vacancies on the surface of the gray TiO2. The removed oxygen atoms left behind two excess electrons per oxygen vacancy, which could be harvested by the neighboring Ti atoms, and induce the formation of Ti 3+ ions showing EPR signals [47]. Therefore, Electron Paramagnetic Resonance (EPR) is one powerful method for identifying the presence of oxygen vacancies in solid materials. A low-field signal with a g-value close to the free-electron value (g = 2.0023) is generally attributed to an unpaired electron trapped on an oxygen vacancy site [11]. Herein, as shown in Figure 4, an EPR signal with a g-value of 1.997 is attributed to the Ti 3+ centers in the rutile phase environment. As a comparison, there were no EPR peaks at the same position observed from the TiO2-WR EPR signals. It is believed that the surface Ti 3+ would tend to adsorb atmospheric O2, which would be reduced to O2 − , and shows an EPR signal at g ≈ 2.02 [11]. The absence of such a peak in the TiO2-GR EPR signals indicates that after long calcination in an air atmosphere, the surface oxygen vacancies are refilled by oxygen atoms and the Ti 3+ is mainly present under the formed surface passivation layer, which is proposed as a key factor for the observed excellent stability of TiO2-GR. It has been reported that the produced Ti 3+ originated from the oxygen vacancies on the surface of the gray TiO 2 . The removed oxygen atoms left behind two excess electrons per oxygen vacancy, which could be harvested by the neighboring Ti atoms, and induce the formation of Ti 3+ ions showing EPR signals [47]. Therefore, Electron Paramagnetic Resonance (EPR) is one powerful method for identifying the presence of oxygen vacancies in solid materials. A low-field signal with a g-value close to the free-electron value (g = 2.0023) is generally attributed to an unpaired electron trapped on an oxygen vacancy site [11]. Herein, as shown in Figure 4, an EPR signal with a g-value of 1.997 is attributed to the Ti 3+ centers in the rutile phase environment. As a comparison, there were no EPR peaks at the same position observed from the TiO 2 -WR EPR signals. It is believed that the surface Ti 3+ would tend to adsorb atmospheric O 2 , which would be reduced to O 2 − , and shows an EPR signal at g ≈ 2.02 [11]. The absence of such a peak in the TiO 2 -GR EPR signals indicates that after long calcination in an air atmosphere, the surface oxygen vacancies are refilled by oxygen atoms and the Ti 3+ is mainly present under the formed surface passivation layer, which is proposed as a key factor for the observed excellent stability of TiO 2 -GR.
Nanomaterials 2020, 10, 920 6 of 11 Nanomaterials 2020, 10, x FOR PEER REVIEW 6 of 11 The photocatalytic activity of the as-prepared TiO2-GR was determined by the photocatalytic discoloration of the MB dyes. As shown in Figure 5a,b, the gray TiO2 showed a higher photoactivity than that of TiO2-WR under visible light or full-spectrum light irradiation. Herein, different from most of the previously reported colored TiO2 materials, the as-synthesized gray TiO2 showed a longlife photocatalytic activity (Figure 5c,d). The TiO2-GR samples retained their color and properties even after six months of storage, and did not exhibit any reduction in their photocatalytic activity after six photocatalysis cycles. This indicates that the gray TiO2 has an excellent chemical stability, which is ascribed to the special nanostructure caused by the two-step calcination treatment. As shown in Figure S2b, the formed surface passivation layer will protect the interior oxygen vacancies. As a result, the T 3+ on the most superficial layer will disappear, which has been confirmed by the aforementioned EPR results. The thermal stability of TiO2-GR was further studied by Thermogravimetric Analyzer (TGA) in open air, as shown in Figure S3. The sample was thermally stable up to 650 °C in open air, with negligible weight variation. The slight weight gain and loss wave before 125 °C are ascribed to the adsorption and desorption of O2, CO2, or H2O on the surface of TiO2-GR in the air. A distinguishable weight loss from 380 °C is ascribed to the dissociation of the surface -OH. Above 650 °C, the obvious weight increase is ascribed to the refilled interior oxygen vacancies, indicating that the surface passivation layer would be destroyed at this temperature. This shows that the as-prepared TiO2-GR has high thermal stability. The photocatalytic activity of the as-prepared TiO 2 -GR was determined by the photocatalytic discoloration of the MB dyes. As shown in Figure 5a,b, the gray TiO 2 showed a higher photoactivity than that of TiO 2 -WR under visible light or full-spectrum light irradiation. Herein, different from most of the previously reported colored TiO 2 materials, the as-synthesized gray TiO 2 showed a long-life photocatalytic activity (Figure 5c,d). The TiO 2 -GR samples retained their color and properties even after six months of storage, and did not exhibit any reduction in their photocatalytic activity after six photocatalysis cycles. This indicates that the gray TiO 2 has an excellent chemical stability, which is ascribed to the special nanostructure caused by the two-step calcination treatment. As shown in Figure S2b, the formed surface passivation layer will protect the interior oxygen vacancies. As a result, the T 3+ on the most superficial layer will disappear, which has been confirmed by the aforementioned EPR results. The thermal stability of TiO 2 -GR was further studied by Thermogravimetric Analyzer (TGA) in open air, as shown in Figure S3. The sample was thermally stable up to 650 • C in open air, with negligible weight variation. The slight weight gain and loss wave before 125 • C are ascribed to the adsorption and desorption of O 2 , CO 2 , or H 2 O on the surface of TiO 2 -GR in the air. A distinguishable weight loss from 380 • C is ascribed to the dissociation of the surface -OH. Above 650 • C, the obvious weight increase is ascribed to the refilled interior oxygen vacancies, indicating that the surface passivation layer would be destroyed at this temperature. This shows that the as-prepared TiO 2 -GR has high thermal stability.
A higher photocatalytic activity is attributed to the presence of oxygen vacancies that can create a higher light absorbance and improve the separation and transmission efficiency of photogenerated carriers, which is preliminarily confirmed by the UV-vis spectra, photocurrent, and photoluminescence spectra. The suggested photocatalysis mechanism is shown in Figure 6a. It is proposed that the photocatalysis of TiO 2 -GR may undergo two different photogenerated carrier transfer pathways when pumped by UV light and visible light separately. It can be seen, in both pathways, that the oxygen vacancies all play a vital role. Compared with TiO 2 -WR, the as-prepared TiO 2 -GR exhibits a broad spectral absorption in the visible light region (Figure 6b). This can be attributed to the transitions from the TiO 2 valence band to the oxygen vacancy levels, or from the oxygen vacancies to the TiO 2 conduction band pumped by visible light [30], which is responsible for the distinguishable higher photoactivity of TiO 2 -GR than that of TiO 2 -WR. These results are consistent with the photocurrent density results under visible light irradiation (Figure 6c). This indicates that, in this case, the visible light absorption of TiO 2 -GR does lead to charge carrier generation and contributes directly to the photocurrent. However, as shown in Figure 6c,d, the photocurrent density under a full-spectrum light condition is about 100 times that under a visible light condition, which indicates that the contribution of visible light to the improvement of the photocatalytic activity is very limited. This is consistent with previously reported results [48]. Therefore, it is proposed that the main factor for the higher photocatalytic activity of TiO 2 -GR is the photogenerated carrier transfer path pumped by UV light. [49] In this process, the Vo is still proposed to be a key factor for the improvement of the separation efficiency of the photogenerated carriers. First, the Vo can act as a trap site for the temporary storage of electrons, which can be further pumped to the conduction band to react with the substrates, resulting in suppressed recombination of photogenerated carriers. [50]. Herein, the suppressed recombination of photogenerated carriers is preliminarily confirmed by the photoluminescence spectra. If the recombination of carriers was suppressed, the photoluminescence of semiconductor materials would be quenched to some degree [51,52]. As shown in Figure 6e, compared with the reference TiO 2 -WR sample, the TiO 2 -GR samples show a much lower photoluminescence intensity. This indicates that the as-prepared gray TiO 2 -GR has a much higher photoinduced charge separation efficiency than that of the white TiO 2 -GR materials. In addition, because of the presence of free electrons bound loosely to the titanium atom in the oxygen vacancies [47], the surface electric conductivity of TiO 2 -GR will be improved, as a result of improving the carriers' transmission efficiency [26], which is also helpful for improving the photocatalytic activity. A higher photocatalytic activity is attributed to the presence of oxygen vacancies that can create a higher light absorbance and improve the separation and transmission efficiency of photogenerated carriers, which is preliminarily confirmed by the UV-vis spectra, photocurrent, and photoluminescence spectra. The suggested photocatalysis mechanism is shown in Figure 6a. It is proposed that the photocatalysis of TiO2-GR may undergo two different photogenerated carrier transfer pathways when pumped by UV light and visible light separately. It can be seen, in both pathways, that the oxygen vacancies all play a vital role. Compared with TiO2-WR, the as-prepared TiO2-GR exhibits a broad spectral absorption in the visible light region (Figure 6b). This can be attributed to the transitions from the TiO2 valence band to the oxygen vacancy levels, or from the oxygen vacancies to the TiO2 conduction band pumped by visible light [30], which is responsible for the distinguishable higher photoactivity of TiO2-GR than that of TiO2-WR. These results are consistent with the photocurrent density results under visible light irradiation (Figure 6c). This indicates that, in this case, the visible light absorption of TiO2-GR does lead to charge carrier generation and contributes directly to the photocurrent. However, as shown in Figure 6c,d, the photocurrent density compared with the reference TiO2-WR sample, the TiO2-GR samples show a much lower photoluminescence intensity. This indicates that the as-prepared gray TiO2-GR has a much higher photoinduced charge separation efficiency than that of the white TiO2-GR materials. In addition, because of the presence of free electrons bound loosely to the titanium atom in the oxygen vacancies [47], the surface electric conductivity of TiO2-GR will be improved, as a result of improving the carriers' transmission efficiency [26], which is also helpful for improving the photocatalytic activity.

Conclusion
In summary, gray rutile titanium dioxide was synthesized via two-step calcination, performed successively in a high-temperature reduction atmosphere and in a lower-temperature air atmosphere. The results indicate that, compared with the white rutile titanium dioxide, the asprepared gray titanium dioxide exhibits the typical characteristics of black-or blue-color TiO2, such as the presence of Ti ions in a low valence state, surface disorder structure, and oxygen vacancies, which are caused by the loss of oxygen atoms under reduction reaction conditions. According to previous reports [14], it is proposed that the presence of Ti 3+ or a surface disorder structure is mainly induced by oxygen vacancies. The as-synthesized gray titanium dioxide exhibits a higher photocatalytic activity than does white rutile TiO2. This is attributed to the interior vacancies, which can create a higher light absorbance and improve the separation and transmission efficiency of photogenerated carriers. Most notably, it is proposed that the two-step calcination can produce a surface passivation layer on the surface of gray titanium dioxide particles, as a result of protecting the interior oxygen vacancies, which provides long-term photocatalytic activity. This study provides a considerable reference for the design and synthesis of other semiconductor photocatalysts rich in oxygen vacancies, with high activity and high stability.

Conclusions
In summary, gray rutile titanium dioxide was synthesized via two-step calcination, performed successively in a high-temperature reduction atmosphere and in a lower-temperature air atmosphere. The results indicate that, compared with the white rutile titanium dioxide, the as-prepared gray titanium dioxide exhibits the typical characteristics of black-or blue-color TiO 2 , such as the presence of Ti ions in a low valence state, surface disorder structure, and oxygen vacancies, which are caused by the loss of oxygen atoms under reduction reaction conditions. According to previous reports [14], it is proposed that the presence of Ti 3+ or a surface disorder structure is mainly induced by oxygen vacancies. The as-synthesized gray titanium dioxide exhibits a higher photocatalytic activity than does white rutile TiO 2 . This is attributed to the interior vacancies, which can create a higher light absorbance and improve the separation and transmission efficiency of photogenerated carriers. Most notably, it is proposed that the two-step calcination can produce a surface passivation layer on the surface of gray titanium dioxide particles, as a result of protecting the interior oxygen vacancies, which provides long-term photocatalytic activity. This study provides a considerable reference for the design and synthesis of other semiconductor photocatalysts rich in oxygen vacancies, with high activity and high stability.