Improved Solar-Driven Photocatalytic Performance of Highly Crystalline Hydrogenated TiO2 Nanofibers with Core-Shell Structure

Hydrogenated titanium dioxide has attracted intensive research interests in pollutant removal applications due to its high photocatalytic activity. Herein, we demonstrate hydrogenated TiO2 nanofibers (H:TiO2 NFs) with a core-shell structure prepared by the hydrothermal synthesis and subsequent heat treatment in hydrogen flow. H:TiO2 NFs has excellent solar light absorption and photogenerated charge formation behavior as confirmed by optical absorbance, photo-Kelvin force probe microscopy and photoinduced charge carrier dynamics analyses. Photodegradation of various organic dyes such as methyl orange, rhodamine 6G and brilliant green is shown to take place with significantly higher rates on our novel catalyst than on pristine TiO2 nanofibers and commercial nanoparticle based photocatalytic materials, which is attributed to surface defects (oxygen vacancy and Ti3+ interstitial defect) on the hydrogen treated surface. We propose three properties/mechanisms responsible for the enhanced photocatalytic activity, which are: (1) improved absorbance allowing for increased exciton generation, (2) highly crystalline anatase TiO2 that promotes fast charge transport rate, and (3) decreased charge recombination caused by the nanoscopic Schottky junctions at the interface of pristine core and hydrogenated shell thus promoting long-life surface charges. The developed H:TiO2 NFs can be helpful for future high performance photocatalysts in environmental applications.

black TiO 2 nanoparticles by treating pristine TiO 2 nanoparticles (crystal-white) under 20 bar pure H 2 atmosphere at 200 °C for 5 days 30 . The authors also demonstrated an approach to enhance solar absorption by introducing disorder in the surface layers of nanoscale TiO 2 through hydrogenation 31 . The role of hydrogen in producing lattice disorder was presented in anatase TiO 2 nanoparticles, and the highly localized nature of the mid-gap states results in spatial separation of exciton in hydrogenated TiO 2 surface. It accounts for its high photocatalytic efficiency as verified by density functional theory 32,33 . Moreover, hydrogenated TiO 2 nanoparticles exhibit the characteristics of low bandgap, which matches well with visible light absorption [34][35][36][37] . Wang et al. reported the hydrogen treatment as a simple and effective strategy to improve the performance of photoelectrochemical water splitting using one dimensional hydrogenated TiO 2 material 32 . In practical applications, one dimensional material titanate materials are typically better than the corresponding nanoparticles. In addition, Liu et al. reported a facile synthesis of hydrogenated TiO 2 nanobelts. It shows an outstanding UV and visible photocatalytic decomposing of methyl orange and water splitting for hydrogen production 38 . An elongated one dimensional material is easier to achieve a percolated electrical network than with zero-dimensional materials. Bundling of one dimensional material contributes to mechanical strength in tangled networks and thus results in macroscopic films [39][40][41] . Furthermore, the hydrothermal synthesis has opened up new possibilities for large scale production of TiO 2 nanofibers by simply thermal treatment of the obtained titanate nanofibers in air 42 .
It is noted that hydrogenated TiO 2 may worsen the photocatalytic activity under simulated solar light as compared to the pristine material. High pressure hydrogenation can be counterproductive to improve the photocatalytic activity of TiO 2 due to the formation of bulk vacancy defects 29 . However, we suggest that the suitable staggered band alignments between highly-crystalline TiO 2 and disordered TiO 2 have the enhanced photocatalytic activity in hydrogenated TiO 2 , as it provides a driving force for the separation of photoexcited electron 43,44 . Hence, the hydrogenated process and its parameters play important role in whether the photocatalytic properties of the material improve of degrade.
In the present work, we demonstrate hydrogenated TiO 2 nanofibers (H:TiO 2 NFs) having highly crystalline one dimensional anatase TiO 2 core and highly defective surface with oxygen vacancies and Ti 3+ interstitial defects obtained by hydrothermal synthesis and subsequent heat treatment in H 2 of partial pressure in N 2 gas flow. An optimal calcination condition is proposed to fine tune the photocatalytic activities. The photo-induced charge carrier distribution and carrier dynamics are systematically investigated to understand the role of surface defects. Photo-induced decoloration of various organic dyes under solar light irradiation confirms the correlation between hydrogenation conditions and the photocatalytic activities.

Results and Discussion
The hydrogen sodium titanate nanofibers were calcined at various temperatures in the mixture of H 2 /N 2 for 12 hrs to find the optimal calcination process that produces the most active photocatalyst. The crystal structure of various H:TiO 2 -X NFs was characterized by synchrotron X-ray diffraction (Fig. 1). (Note, the number in the name of the samples after H:TiO 2 denotes the calcination temperature.) The results show that H:TiO 2 NFs calcined below 600 °C comprises a major anatase TiO 2 phase along with a minor transition phase of monoclinic β -TiO 2 45,46 . The reflection intensity at 2θ = 16.8° increases with ascending calcination temperature. The higher calcination temperature improves the ordering of the anatase TiO 2 lattice. When the applied calcination temperature is above 650 °C, the crystal structure transforms to pure anatase TiO 2 phase. All diffraction peaks can be perfectly indexed as the body-centred tetragonal structure of anatase TiO 2 , with unit cell parameters a = b = 3.78 Å and c = 9.52 Å [COD ID:720675]. The reflection intensity at 2θ = 16.8° decreases at calcination temperatures above 700 °C indicating the formation of rutile TiO 2 phase from anatase 47 . Also, we synthesized a series of pristine TiO 2 NFs calcined at various temperatures under the air flow for 12 hrs in comparison with H:TiO 2 NFs (Fig. 1b) Fig. S1 of supplementary information. The XPS results suggest that the crystal surface has oxygen vacancy defects and Ti 3+ interstitial defects. The resolved Ti 2p orbital evidences the presence of Ti 3+ signals at around 457 eV ( Fig. S1 Table 2. The absorption and desorption isotherms and the pore diameter distribution curves of pristine TiO 2 -650 NFs and H:TiO 2 -650 NFs can be found in Fig. S2 of supplementary information. After hydrogenated process, the specific surface area and total pore volume of H:TiO 2 -650 NFs are larger than pristine TiO 2 -650 NFs. The reason could be due to the surface defect formation of H:TiO 2 -650 NFs for nitrogen gas adsorption/desorption, such as oxygen vacancy defects and Ti 3+ interstitial defects. As a result, the average pore diameter of H:TiO 2 -650 NFs should be decreased after the hydrogenated process due to the formation of small surface defects. Camera images and corresponding absorbance spectrum of pristine TiO 2 NFs and H:TiO 2 NFs are shown in the Fig. 3. H:TiO 2 NFs is having a greyish color with respect to the white pristine TiO 2 NFs. As compared to pristine TiO 2 NFs, the absorbance spectrum of H:TiO 2 NFs is enhanced in the visible region. The bandgaps of pristine TiO 2 NFs and H:TiO 2 NFs can be estimated to be approximately 3.17 and 3.14 eV respectively. The enhanced visible absorption behavior could be due to the surface defects, including the oxygen vacancy and the Ti 3+ interstitial defects. When the Ti 3+ interstitial defects which reduces Ti 4+ into Ti 3+ is on the surface, it introduces mid-gap state into TiO 2 crystal for enhanced optical absorption 33,48 . Computer simulation is used to examine the absorption behavior caused by the oxygen vacancy.  Tip-enhanced Raman spectroscopy (TERS) gives the information about the surface vibrational modes of the synthesized TiO 2 . Both pristine TiO 2 NFs and H:TiO 2 NFs were measured by two-laser TERS system to observe the phase transformation in certain depth profile. The information provided by 532 nm excitation probes more efficiently the outside surface structure, while the 633 nm scatters from the entire volume of the nanowires. As   Fig. 4(a)). The outside surface structure of H:TiO 2 NFs is similar to anatase phase, however, oxygen vacancy defect and the Ti 3+ interstitial defects (partial TiO 2 transformed to Ti 2 O 3 ) are included. It can be inferred that the formation of Ti 3+ interstitial defects in anatase results in the red shift of E g phonons (144 and 200 cm −1 ) caused by the multi-phonon B 1g of the Ti 2 O 3 . The third E g phonon at 640 cm −1 is blue shift affected by the A 1g phonon in Ti 2 O 3 . It is also noted that the mixed phase of anatase TiO 2 , oxygen vacancy defect and the Ti 3+ interstitial defects in the outside surface of H:TiO 2 NFs has broaden peaks with respect to pristine TiO 2 NFs as shown in Fig. 4(b) 50,51 TEM microstructure analysis and the TERS reveals that H:TiO 2 NFs contains a highly crystalline anatase TiO 2 core and a hydrogenated TiO 2 shell. The photo-assisted Kelvin probe force microscopy (photo-KPFM) is a useful technique to predict the photocatalytic capability of materials in the development of high performance photocatalysts 52,53 . Here, it was applied to elucidate the carrier distribution on pristine TiO 2 NFs and H:TiO 2 NFs. Topographic images and surface potential mappings of pristine TiO 2 NFs and H:TiO 2 NFs in Fig. 5. Figure 5   pairs can be generated under UV-B light leading to an upward shift of Fermi energy from E f to E fn and the resulting detected negative shift of surface potential. We assume that the electrons of the H:TiO 2 NFs excited by pulse laser will be transferred to the H:TiO 2 NFs surface. Figure 6(a) shows the PL spectra of pristine TiO 2 NFs and H:TiO 2 NFs excited by 375 nm picosecond pulsed laser. Intense PL at the position approximately 500 nm from the pristine TiO 2 NFs is surprising at first glance 54 . Even though pristine TiO 2 NFs has defect density in the structure so as to give strong PL response around 500 nm, we expected that the H:TiO 2 NFs would provide higher carrier transport based on the results of the TERS and photo-KPFM in highly crystalline of anatase TiO 2 55 . In order to address the behaviour of intrinsic PL, the results of micro time-resolved photoluminescence (μ-TRPL) was obtained by keeping the wavelength at 425 nm for understanding the carrier transport ( Fig. 6(b)). The transient PL decay plots were fitted by bi-exponential kinetics function 56 : where A 1 and A 2 are the corresponding amplitudes. τ 1 and τ 2 are fast decay time and slow decay time. The average lifetime was calculated using the following equation 57 : The transient PL decay fitting curve of pristine TiO 2 NFs and H:TiO 2 NFs depicts that the hydrogenated process could influence the charge transport efficiency. Table 3 is the summary of the measured fast decay time (τ 1 ), slow decay time (τ 2 ), and PL average lifetime (τ avg ) for pristine TiO 2 NFs and H:TiO 2 NFs. For the pristine TiO 2 NFs, the fast decay lifetime is 0.50 ns, the slow decay lifetime is 1.45 ns and their corresponding amplitudes are 54.4% and 45.6% respectively. Surprisingly, the fast decay lifetime of H:TiO 2 NFs significantly decreases to 0.34 ns  and the amplitudes increases to 94.3%. It suggests the improvement of the efficiency of electron transfer to surface and reduces the electron-hole recombination. It is reasonable to know that the average lifetime of the pristine TiO 2 NFs is 0.93 ns and that of the H:TiO 2 NFs is 0.40 ns. For the inner structure of the H:TiO 2 NFs, the highly crystalline anatase TiO 2 phase could deliver electrons effectively (Fig. 1(c)) It is believed that the outside surface structure of the H:TiO 2 NFs has large amount of surface defects (including the oxygen vacancy and Ti 3+ interstitial defect). The hetero-phase junction delivers electron to the surface defect on the outside structure of H:TiO 2 NFs. The excited electron irradiated by ultraviolet light is located in the surface defects, and it is easily trapped in the mid-state of conduction band which consists with the large negative surface potential at photo-KPFM studies.
The photodegradation of several organic dyes, including methyl orange, rhodamine 6G and brilliant green, under simulated solar light irradiation were performed by AEROXIDE ® TiO 2 P25, pristine TiO 2 NFs and H:TiO 2 NFs. The absorption spectra of methyl orange, rhodamine 6G and brilliant green, as a function of irradiation time were recorded in Fig. 7. The λ max in the measured absorbance spectrum is used to calculate the various organic dye concentration using a calibration curve. The λ max of methyl orange, rhodamine 6G and brilliant green are 464.0, 527.5 and 624.5 nm. The colour of suspension changed from the initial colour to colourless and showed good agreement with first-order kinetics i.e. ln(C/C o ) = − kt; where C is the concentration of the dye at time t, C 0 is the initial concentration, and k is the apparent reaction rate constant 58 . For the catalyzed photodegradation of various organic dyes, the H:TiO 2 NFs is superior to pristine TiO 2 NFs and the commercial AEROXIDE ® TiO 2 P25 59,60 . Based on our results thus three mechanisms may be assumed for the high photocatalytic activity of H:TiO 2 NFs: (1) highly crystalline anatase TiO 2 exhibit the high charge transport rate ( Fig. 1), (2) the hydrogenated process promotes the visible absorption behaviour to increase exciton generation (Fig. 3), and (3) surface charge can photo-induce the electron to decrease charge recombination (Fig. 6). The photocatalytic degradation mechanism of organic dye over H:TiO 2 NFs is described in equations (3)~(9) 61 . First, when TiO 2 is irradiated by a light that energy is greater or equal to its bandgap, the photon will excite the valence electron (e − ) to the conduction band and electron-hole pair will be generated. After that, the electron reacts with the oxygen (O 2 ) to form superoxide ions ( • − O 2 ). The superoxide ions possess a significant reducing ability, hence it will react with proton (H + ) and reduce to hydroperoxyl radical (HO 2 •). Whenever the organic molecules adsorbed on the photocatalyst surface, the hole (h + ) would react with the hydroxide ions (OH − ) or water molecules to form hydroxyl radicals (OH• ) and H + . Most of these free radicals behaves excellent oxidation ability, among which OH• and HO 2 • have the strongest oxidation potential. They will quickly adsorb any organics on the surface of TiO 2 and undergo oxidation-reduction reactions leading to the production of low molecular weight intermediates. It finally oxidizes these intermediates into environmentally harmless products such as water or carbohydrate. Many studies have focused on the factors that affect the OH• formation such as irradiation time, pH and phase structures. Under acidic environment, low pH will promote the formation of OH• because of the lower redox potential for hole at valance-band. Also, the phase structures of TiO 2 affects the formation rate of OH• significantly. Amorphous TiO 2 possesses lots of defect that induces the recombination of electron-hole pairs and suppress the OH• formation. Thus, the proper crystalline phase structure design facilitates the photocatalytic phenomenon 62,63 . The high-performance photocatalyst should maintain the activities after repeated irradiations. To further evaluate the stability and reusability of the pristine TiO 2 NFs and H:TiO 2 NFs, the recycled photocatalytic activities were measured by executing repeated degradation reaction of methyl orange over pristine TiO 2 NFs and H:TiO 2 NFs for five recycling runs under UV-B light irradiation. The photostability testing of pristine TiO 2 NFs and H:TiO 2 NFs were shown in Fig. S5   When H:TiO 2 NFs absorbs solar light with energy larger than its bandgap, excitons are generated. The electrons generated in H:TiO 2 NFs are effectively transferred to the surface defect, and it can capture the photogenerated electrons effectively thus reducing the rate of electron-hole recombination. In order to confirm the position of valence band and conduction band, the pristine TiO 2 NFs and H:TiO 2 NFs were measured by Ultraviolet Photoelectron Spectroscopy (UPS). The UPS spectra of pristine TiO 2 NFs and H:TiO 2 NFs are shown in Fig. S6 Fig. 8(a). The holes generated in H:TiO 2 NFs could stay on the area without surface defects due to the VBM of H:TiO 2 NFs is lower than that of pristine TiO 2 NFs. If the holes are not directly recombined with electrons in H:TiO 2 NFs, they are able to be further transferred to react with the organic dyes. During the charge separation and migration processes, some of the excited charges may recombine. If the electrons generated in H:TiO 2 NFs are effectively transferred to the oxygen vacancy and Ti 3+ interstitial defects, it captures the photogenerated electrons effectively and to reduce the rate of electron-hole recombination. In addition, the electronic structure and optical properties were also calculated to confirm the result with UPS study by the first-principles calculations based on density functional theory. The density of states of pristine TiO 2 and H:TiO 2 is also calculated, and the detail data of structure is shown in Fig. S7 of the supplementary information. Each electronic structure was analyzed in order to obtain the origin of the band discontinuity. In comparison of density of state, the conduction band and valence band of both the pristine TiO 2 and H:TiO 2 is attributed to the Ti 3d and O 2p orbital, respectively. The results show that a mid-state of H:TiO 2 can be considered as an extension of conduction band. As a consequence, the mid-state of H:TiO 2 could narrow the band gap and lead to the excited electrons richly transported from conduction band to new mid-state (Fig. S7 of the supplementary information). Figure 8(b) illustrates the outside material of H:TiO 2 NFs is consisted of the surface defects, including the oxygen vacancy and Ti 3+ interstitial, and highly crystalline anatase TiO 2 . The photocatalytic activity depends on the amount of working electrons and holes on the surface of the photocatalyst. The H:TiO 2 NFs with a core-shell structure prepared by the hydrothermal synthesis and subsequent heat treatment at low H 2 partial pressure in the N 2 gas flow can be helpful for searching the high-performance visible-light-active photocatalyst in the field of degradation of pollutants with solar light.

Conclusion
In summary, H:TiO 2 NFs was prepared by a safe and easy process, and its characteristics were studied to understand the correlation between the hydrogenated process and the solar-light-assisted photocatalytic performance. The high absorption in solar light is due to the oxygen vacancy and Ti 3+ interstitial defects on the surface of the H:TiO 2 NFs. The photo-KPFM analysis and μ -TRPL confirms the lower recombination rate and higher charge transport in H:TiO 2 NFs compared with pristine TiO 2 NFs. For the photodegradation of various organic dyes, including methyl orange, rhodamine 6G and brilliant green, H:TiO 2 NFs gave the fastest decoloration phenomenon under solar light irradiation than TiO 2 P25 and pristine TiO 2 NFs. Our study indicates that the significant photodegradation activity is obtained by adding the surface defect (the oxygen vacancy and Ti 3+ interstitial defect) into TiO 2 NFs surface. Three mechanisms were elucidated: (1) enhancement in absorbance to increase exciton generation, (2) highly crystalline anatase TiO 2 to increase the charge transport rate, and (3) decreased charge recombination to increase surface charge. The result illustrates a soft controlling of the hetero-phase junction and highly crystalline anatase TiO 2 . It may strongly change the ability of the materials in photodegradation of pollutants.

Methods
Preparation of H:TiO 2 NFs. For the preparation of H:TiO 2 NFs, we suspend 2.50 g TiO 2 anatase powder (Aldrich, 98%) in 62.5 mL of 10.0 M NaOH aqueous solution, followed by a treatment in a Teflon-lined autoclave at 150 °C for 24 hrs, applying revolving around its short axis. Then, sodium titanate NFs was then washed in 0.10 M HCl to exchange sodium ions for protons. The neutralized product was washed with deionized water and finally filtered and dried in the air at 70 °C to obtain the hydrogen sodium titanate NFs. The hydrogen sodium titanate NFs were calcined at in 15% H 2 (in N 2 buffer) flow for 12 hrs to obtain the various H:TiO 2 -xxx NFs.
Characterization of TiO 2 NFs. The crystal structure of pristine TiO 2 -650 NFs and H:TiO 2 -650 NFs were determined by synchrotron X-ray spectroscopy (l~1.025 Å) on beam line 13A1 of the National Synchrotron Radiation Research Center (NSRRC) in Taiwan. Spherical-aberration corrected field emission transmission electron microscope (JEOL, JEM-ARM200FTH, Japan) was used to observe the microstructures of pristine TiO 2 -650 NFs and H:TiO 2 -650 NFs. In addition, UV-vis absorption spectra of various synthesized TiO 2 samples were measured by absorption spectrophotometer (Jasco Analytical Instruments, V-650, Japan) in the 200-800 nm wavelength range. XPS (X-ray photoelectron spectroscopy) spectra were recorded with a PHI 5000 Versa Probe system (ULVAC-PHI, Chigasaki) using a micro focused (100 μ m, 25 W) Al X-ray beam. BET surface area, BJH cumulative volume of pores and BJH average pore width of pristine TiO 2 -650 NFs and H:TiO 2 -650 NFs were measured by Accelerated Surface Area and Porosimetry System (ASAP 2020, Micromeritics). The system of micro time-resolved photoluminescence (μ -TRPL) with one lasers as Picosecond diode laser driver with 375 nm Laser head (with integrated collimator and TE cooler for temperature stabilization) was integrated by UniNanoTech Co., Ltd. Andor iDus CCD with 1024 × 128 pixels was used to take the PL signal and the Pico Quant PMT Detector head with 200-820 nm, < 250 ps IRF was integrated to take the μ -TRPL signal. In a particular measurement, tip-enhanced Raman spectroscopy was performed using a UniRAM system (UniNanoTech) combined with MV4000 (Nanonics) scanning probe stage at excitation wavelengths of 532 nm (10 mW) and 633 nm (13 mW). The signal collection was detected by a CCD panel having 1024 × 256 pixels. The work function and HOMO (Highest Occupied Molecular Orbital) of pristine TiO 2 -650 NFs and H:TiO 2 -650 NFs were measured by ultraviolet photoelectron spectroscopy (UPS, ULVAC-PHI, Chigasaki) using ultraviolet light source of He I emission (21.2 eV, B50 W) and take-off angle of 90°. Low energy secondary electrons were collected by applying 10 V dc to specimens.
Photo-KPFM analysis. The surface potential mapping was measured using a photo-KPFM (Kelvin probe force microscope, Digital Instruments, Nanoscopes III). Pristine TiO 2 -650 NFs and H:TiO 2 -650 NFs were dispersed in ethanol and spin-coated on a gold coated (thickness of 100 nm) silicon wafer and then dried before analyses. The experimental setup of photo-KPFM was conducted using UV-B light (λ max ~ 312 nm, 8 W) exposure. The surface potential maps of samples were taken with and without illumination at room temperature. N-type silicon cantilever (Nanosensors, average force constant of 2.8 N/m) is coated with chromium as a buffering layer. A platinum-iridium5 alloy was used as a conductive layer. With this method, the height variation and contact potential by electrostatic force between the conductive tip and the surface of the samples are measured simultaneously. A line is scanned using AFM in tapping mode to acquire the topographic information of the material, then the same line is rescanned with the tip lifted to a height of 20 nm. During the second scan, V DC is applied at the tip to nullify the electrostatic oscillations, position by position, and the contact potential difference is observed and measured. The surface potential distributions of pristine TiO 2 -650 NFs and H:TiO 2 -650 NFs were mapped in the dark or under the illumination of a UV-B lamp (Sankyo Denki, G8T5E, 8 W). In addition, the function of the cross-section analysis was used to get detailed information on the topographic height and the surface potential across the selected line. The surface potential was obtained in the dark or under UV-B illumination. The surface potential difference is then denoted as the photo surface potential shift.  21 . In this measurement of photostability testing, 20.0 mg of pristine TiO 2 -650 NFs and H:TiO 2 -650 NFs were sonicated for 10 min in 300 mL of 10.0 ppm methyl orange aqueous solution, respectively. The temperature of the stirred dispersion was kept near room temperature. The distance between the 4 pieces of UV-B lamp (Sankyo Denki, G8T5E, 8 W)) and reactor was about 10.0 cm. After the reaction of first run testing under UV-B light irradiation, the suspensions were centrifuged to obtain the photocatalyst, which was washed with ethanol and deionized water carefully and then dried at 105 °C for 24 hr. The fresh 10.0 ppm methyl orange aqueous solution was mixed with the used photocatalyst to perform the 2 nd run photoactivity testing. Similarly, the recycled 3 rd , 4 th and 5 th tests were also performed.
Computational simulation. Computational simulation used in this paper is based on density functional theory with a GGA-PBE (Generalized Gradient Approximation Perdew-Burke-Ernzerhof) functional implemented in CASTEP which uses a plane wave basis set to expand the election wave function. As for the pseudopotential, two setups are adopted depending on the characteristic we are simulating. This is because that pseudopotential will have better accuracy in predicting the properties they suit. In the simulations of absorption spectrum and density of states (DOS), TiO 2 is modelled by a (3 × 3 × 1) supercell with/without oxygen vacancy (Fig. S2 of supplementary information). In this case, a norm-conserving pseudopotential is used due to its accuracy in predicting optical properties, and the calculations are conducted with an energy cutoff of 450.0 eV and a k-point set of 1 × 1 × 2.