Highly stable visible‐light photocatalytic properties of black rutile TiO2 hydrogenated in ultrafast flow

The hydrogenation and introducing oxygen vacancies (VO) can lead to surface lattice disorder in TiO2, which is a new form of TiO2 named black TiO2, with excellent visible-light photocatalytic activity, but this TiO2 is easy to failure because oxidation makes the concentration of surface VO decrease rapidly in a short time. In this work, black TiO2 nanoparticles with VO almost concentrated inside nanoparticles were fabricated under ultrafast hydrogen flow. These bulk VO shortened the bandgaps of black TiO2, enhanced its visible light absorption, and meanwhile provided extremely strong stability. The location of VO in black TiO2 was evident from EPR, XPS with HRTEM investigation, and other characteristics of black TiO2 were obtained by XRD, UV-Vis, SEM, PL, and photocurrent techniques. The degradation experiments on Cr6+ or rhodamine B demonstrated the good visible-light photocatalytic performance of our material. After 18 months of natural aging treatment (in the air), our samples showed no discoloration and maintained 89.5% photocatalytic efficiency, and further study exhibited that this black TiO2 also contained excellent acid resistance and moderate alkaline resistance. This work could help design lattice disorder to obtain more stable and practical black TiO2.


Introduction
TiO 2 has been widely studied in dealing with environmental and energy problems since its photocatalytic property of decomposing water was discovered in 1972 [1][2][3]. However, naturally occurring TiO 2 can only be excited by ultraviolet to generate electron-hole pairs owing to its wide bandgap and weak visible-light absorbance [4,5]; this limitation makes TiO 2 a long distance to practical application. Doping [6,7], compounding with other semiconductor oxides [8], such methods have been tried for narrowing the bandgap of TiO 2 and enhancing its visible-light photocatalytic performance.
In 2011, Chen et al. reported a material called ''black TiO 2 '' with an amazing narrowed bandgap (1.54 eV) and significantly improved visible-light photocatalytic activities compared to white TiO 2 [9], quickly attracting public eyes. These TiO 2 nanoparticles changed from white to black after being reduced in a hydrogen atmosphere with high pressure, but their phase kept TiO 2 . This form of TiO 2 had significant absorption of visible light, and in subsequent reports, TiO 2 with a similar mechanism but a slightly lighter color was also called black TiO 2 [10,11]. Although many methods for the preparation of black TiO 2 have been developed [12][13][14], hydrogenation remains the most used method for the reason of high-efficiency and ease of industrial production. Extensive efforts had been done on the formation mechanism of black TiO 2 [13,[15][16][17], while the most generally accepted explanation was that oxygen vacancies (V O ) played a leading role in forming black TiO 2 . V O could change band structures and visible-light absorbance of black TiO 2 via introducing donor energy levels in conduction bands [9,[17][18][19]. The generation of V O was often considered to be accompanied by the appearance of Ti 3? [20,21], and they could both promote the separation of photogenerated electron-hole pairs. Other defects, for example, surface hydroxyl groups had also been reported to be in connection with the formation of black TiO 2 with excellent visible-light photocatalytic performance. Moreover, these crystal defects often accumulated on the material surface, resulting in surface disorder, and thus typical amorphous shells could be usually observed on the surface of black TiO 2 .
Although the ability of black TiO 2 to utilize visible light had been dramatically improved, the surface defects were easy to be oxidized, leading to a decrease in defect concentration, so black TiO 2 was unstable and its further application was seriously limited [22,23]. Some researchers began to concern and solve this problem, while the effect of long time treatment under various external risk factors (O 2 , H ? , OH -, etc.) on the photocatalytic performance of black TiO 2 was still not investigated. For example, Lan et al. adopted an in situ reduction method to form Ti 3? as much as possible inside the material to stabilize black TiO 2 [23]. At the same time, one study revealed that there are two types of V O and they may correspond to different locations [24].
Herein, black rutile TiO 2 nanoparticles with stable visible-light photocatalytic activity were prepared through hydrogen in an ultrafast flow. The conventional flow rate of H 2 was 0.01-0.1 L/min when fabricating black TiO 2 [25][26][27][28], but it reached 1.2 L/min in this work. A series of temperature gradient experiments were carried out to found an optimal parameter and the samples hydrogenated at 800°C contained the best photocatalytic properties. The experimental parameters of photocatalysis had great effects, and our samples could degrade most RhB (250ml, 4lM) in 120 min under visible light, and also had a favorable ability to reduce Cr 6? . Various characterization methods revealed good surface crystallization but a mass generation of V O in the body of black TiO 2 , and the high concentration of internal defects rather than surface defects was believed to an important cause of stabilizing black TiO 2 . The natural aging process of 18 months indicated that the shelf life of our samples was 3 times that of traditional black TiO 2 . Moreover, the high stability of black TiO 2 was further demonstrated by series acid-alkali resistance experiments: examining the residual photocatalytic activity of black TiO 2 after pretreated in different pH conditions.

Materials
White rutile TiO 2 (50 nm in diameter, purchased from Ansteel Research Institute of Vanadium and Titanium) was heated in nitrogen flow inside a furnace with a 10°C/min heating rate firstly. After the furnace temperature reached 150°C, nitrogen was replaced with ultrafast pure hydrogen flow (99.9 %, 1.2 L/min), while heating rates continued to be 10°C /min, and then the samples were held at different temperatures (700-900°C) for 1 h. Black TiO 2 was obtained after the samples cooled to room temperature. In the following, black TiO 2 hydrogenated at different temperatures was abbreviated to 700°C-H-TiO 2 , 750°C-H-TiO 2 , etc.

Photocatalytic test
4 lmol/L Rhodamine B (RhB, provided by ChengDu Chron Chemicals Co,.Ltd) aqueous solutions were the reaction substrate of most photocatalytic experiments. In each experiment, 10 mg TiO 2 nanoparticles were dispersed into a 250 mL RhB solution. After the mixture reached an adsorption-elution balance in the dark, visible-light photocatalytic experiments were carried out under a xenon lamp. This lamp was equipped with a 420 nm cutoff filter to obtain light with wavelengths longer than 420 nm. 5 mL mixture was taken for filtration and centrifugation each time, and then the absorbance of supernatants was measured to determine the concentration of RhB. The parameters of stability experiments under different environments would be described later in this article.

Characterization
The colors of black TiO 2 gradually deepened with hydrogenation temperature increasing, which matched with the ultraviolet-visible diffuse reflectance spectra (UV-Vis DRS, measured on Shimadzu UV-3600). The phase analysis was obtained by an X-ray diffraction spectrometer (XRD, Rigaku smartlab9) with Cu Ka X-ray. Electron paramagnetic resonance (EPR) was carried out at 77 K on a Bruker ELEXSYS-II E500 spectrometer. The information about surface hydroxyls and Ti states was obtained by an X-ray photoelectron spectrometer (XPS, Thermo Fisher Scientific K-Alpha). Microphotographs of materials were obtained by scanning electron microscopy (SEM, FEI Inspect F50) and transmission electron microscopy (TEM, FEI Tecnai G2 F20). Figure 1 illustrates the XRD patterns (normalization based on the intensity of diffraction peak (101)) of different TiO 2 . The main phase of samples hydrogenated at different temperatures all maintained rutile TiO 2 , and no obvious change was observed among 700°C-H-TiO 2 , 750°C-H-TiO 2 , 800°C-H-TiO 2 , and white TiO 2 . Higher hydrogenation temperatures led to the formation of suboxide titanium, reflecting as some impurity peak around the main diffraction peak (27.4°). The peaks at 24.32°and 28.06°were assigned to Ti 9 O 17 , while the peaks at 26.42°and 28.78°were assigned to Ti 6 O 11 , and a reconstruction process induced by titanium suboxides during reduction processes brought other effects which were detailed discussed in the supplementary information [29,30].

Results and discussion
The information about appearance, absorbance, and bandgaps of different TiO 2 was given in Fig. 2. According to Fig. 2a, the colors of black TiO 2 were gradually darkened as hydrogenation temperatures went up. The deepening of colors brought photocatalysis marked significance that visible light could not be absorbed by white TiO 2 but black TiO 2 broke through this limitation. The UV-Vis DRS showed in Fig. 2b exhibited a similar result that the visible-light absorbance of white TiO 2 was almost 0, while black TiO 2 had significantly stronger absorbance proportional to hydrogenation temperatures. The values of bandgaps shown in Fig. 2c were obtained by mathematical processing Fig. 2b and the following equation: Here, a, t, and E g represented absorption coefficient, frequency, and bandgaps, respectively. This formula was the main calculation method of the bandgaps of black TiO 2 [31,32]. Herein, direct application of this formula was inaccurate due to the high concentration of defects and mid-gap states in black TiO 2 , so an improved method, baseline approach, was adopted to improved accuracy [9,33,34]. The bandgap of white TiO 2 (rutile) in our experiments was 3.02 eV which was closed to experienced data 3.0 eV, and the bandgaps of black TiO 2 were nearly 2.95 eV. Although these hydrogenated samples exhibited a small difference in bandgap compared with raw material, their visible-light absorbance dramatically increased, which possibly resulted from V O localized states below the conduction band. In general, TiO 2 nanoparticles were successfully transformed to black TiO 2 . Figure 3 presents the micromorphology information of black TiO 2 hydrogenated under different temperatures. White TiO 2 nanoparticles without treatment were in the shape of spindles with a diameter of nearly 50 nm, while black TiO 2 nanoparticles had ellipsoidal shapes with longer diameters due to the thermal effect in annealing processes. A summary of statistics on the size of different TiO 2 nanoparticles is shown in Fig. 4, and the size of TiO 2 nanoparticles slightly increased with annealing temperature raising. In particular, size distributions of the samples 800°C-H-TiO 2 , 850°C-H-TiO 2 , and 900°C-H-TiO 2 exhibited only a little difference, so the role of particle size could be ignored when the changes of photocatalytic activity were discussed later. The lattice-disordered shell was a significant feature of traditional black TiO 2 , which was interestingly not observed in our hydrogenated materials. Lattice stripes at the edge of black TiO 2 nanoparticles were still clear, indicating quite low surface disorder and defect concentration. White TiO 2 had an interplanar spacing of 0.319 nm in (110) face, while this parameter in black TiO 2 expanded to 0.342 nm. This 7.2 % expansion was due to the existence of V O enriched in the body of black TiO 2 , and subsequent analysis of crystal defects further verified this conclusion.
The photocatalytic activity of different TiO 2 was determined by the time-depend residual of RhB, as shown in Fig. 5. RhB itself did not degrade under visible light, so its concentration change was caused by the photocatalytic effect of black TiO 2 . When the different mixtures of TiO 2 powders and RhB water solution had just been exposed to light, their absorption-dissociation balance reached in the dark would be disturbed, causing an abnormal upward change in the C/C 0 curve of RhB in the first 20 min, and this phenomenon disappeared with time growth. As expected, white TiO 2 had no visible-light photocatalytic activity due to its poor ability to utilize visible light, while black TiO 2 had obvious enhanced photocatalytic activity driven by visible light, further suggesting that black rutile TiO 2 were successfully prepared. RhB degradation rates of black TiO 2 increased with hydrogenation temperatures and visible-light absorbance before 800°C, but an abnormal phenomenon of sharply decreased photocatalytic activity was observed in the experiments of 850°C-H-TiO 2 and 900°C-H-TiO 2 , although they had darker colors and narrower bandgaps. This anomaly could be attributed to the changes in crystal planes of different black TiO 2 and the difference in photocatalysis induced by crystal planes [35,36]. This would be detailly discussed in the supplementary information. In general, 800°C-H-TiO 2 had the best performance that it could degrade more than 70 % RhB in 2 h, and Table 1 exhibited a comparison on the photocatalytic performance between the present study with previously reported works. Besides, we also chose another substrate Cr 6? that was more difficult to decompose to further illustrate the excellent photocatalytic activity of our material, which is revealed in Fig. S1. 800°C-H-TiO 2 was selected for more tests and a series of stability experiments because of its optimal performance, and we adopted EPR measurement to reveal the changes from white TiO 2 to black TiO 2 from the angle of lattice defects. As shown in Fig. 6a, for 800°C-H-TiO 2 , a small peak at g = 2.005 indicated the generation of a small number of V O [24,47], while the spectrum of white TiO 2 with few crystal defects appeared horizontal. This result explained the variation in absorbance from white TiO 2 to 800°C-H-TiO 2 . V O produced donor levels under the conduction band [17] and then increased the visible-light absorption of black TiO 2 . It should be noted that different V O peaks corresponded to diverse positions of existence, and black TiO 2 reported previously usually had the peaks at both nearly g = 1.98 (in the surface, coexisted with Ti 3? ) and nearly g = 2.005 (in the body) [24], while the peak at g = 1.98 in 800°C-H-TiO 2 in this paper seemed not obvious. This analysis meant that in 800°C-H-TiO 2 V O was concentrated mainly inside nanoparticles. According to Fig. 6b, the double peaks at 458.7 eV and 464.5 eV in white TiO 2 could be all assigned to Ti 4? [14]; the spectrum of 800°C -H-TiO 2 showed a similar result that the total peak was almost identical to the Ti 4? , also indicating that defects Ti 3? barely existed in the surface of black TiO 2 nanoparticles. Fig. S2 further illustrates that only a few surface defects existed in our material, and those characteristics may contribute to the excellent stability of 800°C-H-TiO 2 .
VB-XPS spectra shown in Fig. 6c  -H-TiO 2 was possibly through introducing localized states below CBM [17], so their VB and CB did not significantly extend. Figure 7 gave the appearance and photocatalytic activity of 800°C-H-TiO 2 after natural aging treatment for 18 months in the air atmosphere. As shown in Fig. 7a, 800°C-H-TiO 2 after 18 months of aging treatment still retained a typical appearance of black TiO 2 , with no significant difference from initial samples. Figure 7b shows the comparison of photocatalytic performance of 800°C-H-TiO 2 before and after aging treatment. 800°C-H-TiO 2 after aging treatment could degrade 65.3% RhB in 2 h, and its photocatalytic activity was only 10.5% lower than that of the samples before aging treatment, indicating that our black TiO 2 contained great stability and were only slightly oxidized after long-term storage in air. Corresponding to our works, black TiO 2 mainly depending on surface V O experienced a similar degree of photocatalytic performance attenuation in only 6 months [22]. The circulation experiment result of 800°C-H-TiO 2 after aging treatment is shown in Fig. 7c. Residual efficiency was calculated by the following equation: Here, C 0x or C x represented the concentration of RhB at the beginning or end in photocatalytic reactions of untreated black TiO 2 , while C 0y or C y represented the concentration of RhB at the beginning or end in photocatalytic reactions of black TiO 2 which had been recycled or other treated. The suspension (include the part which was used to measure absorbance) during catalytic experiments was all collected, after that, black TiO 2 powder was filtered from the suspension, cleaned in ultrasonic, and dried in argon. RhB concentration was correspondingly reduced in each subsequent cycle experiment due to the slight quality loss of catalyst, while other experimental parameters of photocatalysis were the same as those of Fig. 5.
The photocatalytic activity of 800°C-H-TiO 2 during 5 cycles of experiments did not change significantly, which further demonstrated its good storability.
To highlight the effect of pH on the material itself, the black TiO 2 powders were pretreated with acidbase instead of directly being put into reaction solutions with different pH as commonly used. Aqueous solutions with different pH were regulated by HCl or NaOH, and then the 800°C-H-TiO 2 powder after 18 months of aging treatment was dispersed in different solutions for 180 h, respectively. The parameters of subsequent processes were the same as above, and   [46] relevant results are shown in Fig. 8. Long-term immersion in acid solution did not decrease the photocatalytic activity of black TiO 2 , and the photocatalytic activity of black TiO 2 was even improved due to changes in surface states driven by protonation [48]. A large number of H ? interacted with black TiO 2 for a long time, which might enhance the absorption of black TiO 2 to RhB and thus increased their photocatalytic activity. The experiments of alkali groups gave results of slightly reduced photocatalytic properties of black TiO 2 after being placed in alkali solutions for a long time. In the group of  Fig. 8c shows that the photocatalytic reactions of black TiO 2 before and after treatment were all first-order reactions. In conclusion, our black TiO 2 had excellent acid resistance, while when they were long-term used in alkali water, a little risk of performance decrease may be concerned. Figure 9 reveals the charge transfer mechanism in the photocatalysis process of 800°C-H-TiO 2 . The electrons transition from VB to CB after light irradiation, leaving holes in their original positions. The high concentration of V O in 800°C-H-TiO 2 brought the formation of localized states below CB, which promotes the transition and transfer of photoelectrons driven by visible light. These photogenerated charges then reacted with O 2 or OHto form free radicals to degrade RhB into non-toxic H 2 O and CO 2 .

Conclusions
In summary, black rutile TiO 2 with extremely stable visible-light photocatalytic performance was effectively prepared in this work. 800°C-H-TiO 2 had the best photocatalytic activity and the high concentration of bulk V O may be responsible for its enhanced visible-light photocatalytic activity and stability. The stability experiment for 800°C-H-TiO 2 illustrated that our black TiO 2 could maintain photocatalytic activity for 18 months, and also could be stably used in acidic water.  Fig. 9 The charge transfer mechanism in photocatalysis process of 800°C-H-TiO 2

Data Availability
The source of the chemical, the process of the experiment, the handling of the data, etc., are transparent.

Code availability
The software application used in this work was also available on other computers.

Declaration
Conflict of interest There is no conflict of interest or competing interests to declare that are relevant to the content of this article.