Black 3D-TiO2 Nanotube Arrays on Ti Meshes for Boosted Photoelectrochemical Water Splitting

Black 3D-TiO2 nanotube arrays are successfully fabricated on the Ti meshes through a facile electrochemical reduction method. The optimized black 3D-TiO2 nanotubes arrays yield a maximal photocurrent density of 1.6 mA/cm2 at 0.22 V vs. Ag/AgCl with Faradic efficiency of 100%, which is about four times larger than that of the pristine 3D-TiO2 NTAs (0.4 mA/cm2). Such boosted PEC water splitting activity primarily originates from the introduction of the oxygen vacancies, which results in the bandgap shrinkage of the 3D-TiO2 NTAs, boosting the utilization efficiency of visible light including the incident, reflected and/or refracted visible light captured by the 3D configuration. Moreover, the oxygen vacancies (Ti3+) can work as electron donors, which leads to the enhanced electronic conductivity and upward shift of the Fermi energy level, and thereby facilitating the transfer and separation of the photogenerated charge carrier at the semiconductor-electrolyte interface. This work offers a new opportunity to promote the PEC water splitting activity of TiO2-based photoelectrodes.


Photoelectrochemical Measurements
The PEC tests were conducted in a three-electrode configuration connected to a CHI 660E electrochemical workstation (CH Instrument, Chenhua Ltd., Shanghai, China), with the pristine and ECR-3D-TiO 2 NTAs with an exposed area of 1 cm 2 , Ag/AgCl (3 mol L −1 KCl-filled), and Pt mesh as the working, reference, and counter electrode, respectively. The supporting electrolyte was 1 M NaOH (pH = 13.6). The irradiation source was a 500 W Xe lamp (Solar 500, NBet Group Corp., Beijing, China) with calibrated intensity of 100 mWcm -2 . Moreover, a water filter was used between the lamp and electrochemical cell to remove solution heating from infrared light. An Ocean Optics oxygen sensor system equipped with a FOXY probe (NeoFox Phase Measurement System, Ocean optics, Orlando, FL, USA) was applied to determine the amount of evolved O 2 . The experiment was carried out together with the stability tests. Before the O 2 measurement, the headspace of the anodic compartment was purged with high purity N 2 (99.9995%) for 1 h under vigorous stirring. PEC water splitting with O 2 sensing continued for 180 min at 0.22 V vs. Ag/AgCl, and the O 2 yield was quantified to calculate the Faradic efficiency. Electrochemical impedance spectroscopy was carried out to understand the charge transfer process between photoelectrodes/electrolyte interfaces. All the measurements were performed under the open circuit condition with the frequency ranging from 0.01 Hz to 100 kHz. Mott-Schottky plots were derived from impedance potential tests conducted at a frequency of 1 kHz in dark conditions.

Morphological Characterization of the Pristine and ECR-3D-TiO 2 NTAs
The morphologies of the 3D-TiO 2 NTAs before and after electrochemical reduction were investigated by FE-SEM. The low-magnification overall FE-SEM image of the ECR-3D-TiO 2 NTAs−1.3 V displays that the diameter of a single Ti wires is about 0.12 mm and the percentage of the open area of Ti mesh is calculated to approximately 30%, suggesting the higher utilization efficiency of the Ti source ( Figure 1a). Figure 1c,d are the magnified FE-SEM images of the area marked by the red ellipse in Figure 1b, which clearly exhibits that TiO 2 NTAs are radially grown outward around the Ti wires, leading to the formation of 3D-TiO 2 NTAs. This highly ordered structure can be described by the 3D representation in Figure S1. The top and cross-sectional view FE-SEM images show such ECR-3D-TiO 2 NTAs with an average diameter of approximately 150 nm, a wall thickness of about 10 nm, and a similar length of 6 µm (Figure 1c,d and Figure S2), which are identical to those of the pristine 3D-TiO 2 NTAs. The effect of the electrochemical reduction on the morphologies and microstructures of 3D-TiO2 NTAs were further investigated by FE-TEM. From the low-magnification FE-SEM images, all the products possess a tightly packed tubular nanostructures with a mean external diameter of 150 mm, which is consistent with the FE-SEM results (Figures 1e and S3a,d,g). The selected electron diffraction patterns display very similar diffraction patterns, which demonstrate the polycrystalline structures of the 3D-TiO2 NTAs before and after electrochemical reduction (Figures 1f and S3b,e,h). In addition, the well-resolved lattice spacing of 0.305 nm are observed in all the products (Figures 1g and S3c,f,i), which corresponds to the {101} plane of anatase TiO2 [38,39]. The phase transition of the 3D-TiO2 NTAs induced by electrochemical reduction were analyzed by XRD. As shown in Figure  S4, all the diffraction peaks match well with crystal structure of the anatase TiO2 (JCPDS 21-1272) and metal Ti [38,39]. No other phase is detected, suggesting no change in the lattice structures after electrochemical reduction. The above FE-SEM, FE-TEM and XRD results imply that electrochemical reduction does not destroy the morphology, microstructures or phase of the 3D-TiO2 NTAs.

Optical Absorption Properties of the Pristine and ECR-3D TiO2 NTAs
We have investigated the UV-vis reflectance spectra of the ECR-3D-TiO2 NTAs as a function of external bias applied in the electrochemical reduction and then compared with that of the pristine 3D TiO2 NTAs. Clearly, the pronounced absorption can be clearly observed in the UV region (<390 nm) of all the products, which can be attributed to the intrinsic band-to-band absorption of TiO2 [38,39,44]. Compared with the pristine 3D TiO2 NTAs, the visible light absorption (400-800 nm) is significantly enhanced after electrochemical reduction. As the applied bias changes from −1.2 to −1.4 V, the visible light The effect of the electrochemical reduction on the morphologies and microstructures of 3D-TiO 2 NTAs were further investigated by FE-TEM. From the low-magnification FE-SEM images, all the products possess a tightly packed tubular nanostructures with a mean external diameter of 150 mm, which is consistent with the FE-SEM results ( Figure 1e and Figure S3a,d,g). The selected electron diffraction patterns display very similar diffraction patterns, which demonstrate the polycrystalline structures of the 3D-TiO 2 NTAs before and after electrochemical reduction ( Figure 1f and Figure S3b,e,h). In addition, the well-resolved lattice spacing of 0.305 nm are observed in all the products (Figure 1g and Figure S3c,f,i), which corresponds to the {101} plane of anatase TiO 2 [38,39]. The phase transition of the 3D-TiO 2 NTAs induced by electrochemical reduction were analyzed by XRD. As shown in Figure S4, all the diffraction peaks match well with crystal structure of the anatase TiO 2 (JCPDS 21-1272) and metal Ti [38,39]. No other phase is detected, suggesting no change in the lattice structures after electrochemical reduction. The above FE-SEM, FE-TEM and XRD results imply that electrochemical reduction does not destroy the morphology, microstructures or phase of the 3D-TiO 2 NTAs.

Optical Absorption Properties of the Pristine and ECR-3D-TiO 2 NTAs
We have investigated the UV-vis reflectance spectra of the ECR-3D-TiO 2 NTAs as a function of external bias applied in the electrochemical reduction and then compared with that of the pristine 3D-TiO 2 NTAs. Clearly, the pronounced absorption can be clearly observed in the UV region (<390 nm) of all the products, which can be attributed to the intrinsic band-to-band absorption of TiO 2 [38,39,44]. Compared with the pristine 3D-TiO 2 NTAs, the visible light absorption (400-800 nm) is significantly enhanced after electrochemical reduction. As the applied bias changes from −1.2 to −1.4 V, the visible light absorption increases gradually, which are further verified by the color variation of the ECR-3D-TiO 2 NTAs. This implies that the ECR-3D-TiO 2 NTAs may respond to the visible light region (Figure 2a). Moreover, the bandgaps of the pristine 3D-TiO 2 NTAs and ECR-3D-TiO 2 NTAs−1.2, −1.3 and −1.4 V, estimated from the intercept of the tangents to the curves of (αhυ) 2 vs. photon energy by assuming TiO 2 as a direct semiconductor, are about 3.09, 2.95, 2.65 and 2.63, respectively ( Figure 2b). These results suggest that the electrochemical reduction not only promote the visible light absorption, but also reduce the bandgap of the 3D-TiO 2 NTAs, which can be ascribed to the presence of the defect state in the bandgap of TiO 2 created by the O-vacancies. The boosted visible light absorption and bandgap shrinkage means that visible light trapped by the 3D configuration can excite electron-hole pairs and thus effectively improve the PEC water splitting activity of the 3D-TiO 2 NTAs.

Surface Oxidation State of the Pristine and ECR-3D TiO2 NTAs
To solidify the presence of O-vacancies in the ECR-3D TiO2 NTAs, the chemical composition and surface oxidation states of the pristine 3D-TiO2 NTAs and ECR-3D-TiO2 NTAs were further examined by XPS. Only Ti, O and C signals are observed in the survey spectra of all the products, which reveals that electrochemical reduction does not introduce other impurities ( Figure S5a). For the pristine 3D TiO2 NTAs, the Ti 2p core level spectrum has two peaks centered at 458.3 and 464.1 eV, which are typical for the Ti 2p3/2 and 2p1/2 peaks of Ti 4+ in TiO2 (Figure 3a) [39,43,50]. After the electrochemical reduction, the Ti 2p3/2 and 2p1/2 peaks shift to the low binding energy of 457.9 and 463.7 eV, illustrating the different bonding environment of the Ti atom. By subtracting the normalized Ti 2p spectra of the ECR-3D TiO2 NTAs−1.3 V with that of the pristine 3D TiO2 NTAs, two extra peaks at 457.7 and 463.3 eV were observed, which were indexed to the Ti 2p3/2 and 2p1/2 peaks of Ti 3+ [39,43,50]. This indicates that O-vacancies are introduced in the ECR-3D

Surface Oxidation State of the Pristine and ECR-3D-TiO 2 NTAs
To solidify the presence of O-vacancies in the ECR-3D-TiO 2 NTAs, the chemical composition and surface oxidation states of the pristine 3D-TiO 2 NTAs and ECR-3D-TiO 2 NTAs were further examined by XPS. Only Ti, O and C signals are observed in the survey spectra of all the products, which reveals that electrochemical reduction does not introduce other impurities ( Figure S5a). For the pristine 3D-TiO 2 NTAs, the Ti 2p core level spectrum has two peaks centered at 458.3 and 464.1 eV, which are typical for the Ti 2p 3/2 and 2p 1/2 peaks of Ti 4+ in TiO 2 (Figure 3a) [39,43,50]. After the electrochemical reduction, the Ti 2p 3/2 and 2p 1/2 peaks shift to the low binding energy of 457.9 and 463.7 eV, illustrating the different bonding environment of the Ti atom. By subtracting the normalized Ti 2p spectra of the ECR-3D-TiO 2 NTAs−1.3 V with that of the pristine 3D-TiO 2 NTAs, two extra peaks at 457.7 and 463.3 eV were observed, which were indexed to the Ti 2p 3/2 and 2p 1/2 peaks of Ti 3+ [39,43,50]. This indicates that O-vacancies are introduced in the ECR-3D-TiO 2 NTAs−1.3 V. In addition, the O1s spectra of the ECR-3D-TiO 2 NTAs−1.3 V Nanomaterials 2022, 12, 1447 6 of 12 was also different from that of that of the pristine 3D TiO 2 NTAs. In the O1s spectra, the main peak located 529.7 eV is the characteristic peak reported for lattice oxygen of TiO 2 , while other peaks centered at 531.4 eV can be associated with oxygen species absorbed at O-vacancies [39]. As displayed in Figure 3b and Figure S5b, the peaks of area of 531.4 eV of ECR-3D-TiO 2 NTAs increase gradually with electrochemical reduction bias reducing from −1.2 V to −1.4 V, which suggests that the amount of the O-vacancies increases with the deceasing electrochemical reduction bias. This is why the visible light absorption increases gradually with the electrochemical reduction bias reducing from −1.2 to −1.4 V.
Nanomaterials 2022, 12, x FOR PEER REVIEW 6 of 12 TiO2 NTAs−1.3 V. In addition, the O1s spectra of the ECR-3D TiO2 NTAs−1.3 V was also different from that of that of the pristine 3D TiO2 NTAs. In the O1s spectra, the main peak located 529.7 eV is the characteristic peak reported for lattice oxygen of TiO2, while other peaks centered at 531.4 eV can be associated with oxygen species absorbed at O-vacancies [39]. As displayed in Figures 3b and S5b, the peaks of area of 531.4 eV of ECR-3D-TiO2 NTAs increase gradually with electrochemical reduction bias reducing from −1.2 V to −1.4 V, which suggests that the amount of the O-vacancies increases with the deceasing electrochemical reduction bias. This is why the visible light absorption increases gradually with the electrochemical reduction bias reducing from −1.2 to −1.4 V.

PEC Water Splitting Activity of the Pristine and ECR-3D TiO2 NTAs
The influence of the electrochemical reduction bias on the PEC water splitting activity of 3D-TiO2 NTAs were also studied, and the results are shown in Figure 4. All the 3D TiO2 NTAs-based photoelectrodes display negligible dark currents in comparison with their respective photocurrents, suggesting no occurrence of the electrocatalytic water splitting. Under irradiation, the photocurrent densities of the ECR-3D-TiO2 NTAs increase steeply and are distinctly larger than that of the pristine 3D-TiO2 NTAs in the whole potential window from −0.9 to 0.6 V vs. Ag/AgCl, which reveals that the electrochemical reduction can significantly promote the PEC performance of the 3D-TiO2 NTAs. Figure 4b compares the transient photocurrent responses of the pristine and ECR-3D-TiO2 NTAs measured at 0.22 V vs. Ag/AgCl. It can be seen that all the 3D TiO2 NTAs-based photoelectrodes show excellent sensitivity to the light irradiation. There is a steep rise in current density from almost zero in dark conditions to a stable value upon illumination. In addition, the ECR-3D TiO2 NTAs−1.3 V generate a maximal photocurrent density of 1.6

PEC Water Splitting Activity of the Pristine and ECR-3D-TiO 2 NTAs
The influence of the electrochemical reduction bias on the PEC water splitting activity of 3D-TiO 2 NTAs were also studied, and the results are shown in Figure 4. All the 3D-TiO 2 NTAs-based photoelectrodes display negligible dark currents in comparison with their respective photocurrents, suggesting no occurrence of the electrocatalytic water splitting. Under irradiation, the photocurrent densities of the ECR-3D-TiO 2 NTAs increase steeply and are distinctly larger than that of the pristine 3D-TiO 2 NTAs in the whole potential window from −0.9 to 0.6 V vs. Ag/AgCl, which reveals that the electrochemical reduction can significantly promote the PEC performance of the 3D-TiO 2 NTAs. Figure 4b compares the transient photocurrent responses of the pristine and ECR-3D-TiO 2 NTAs measured at 0.22 V vs. Ag/AgCl. It can be seen that all the 3D-TiO 2 NTAs-based photoelectrodes show excellent sensitivity to the light irradiation. There is a steep rise in current density from almost zero in dark conditions to a stable value upon illumination. In addition, the ECR-3D-TiO 2 NTAs−1.3 V generate a maximal photocurrent density of 1.6 mA/cm 2 , which is about four times larger than that of the pristine 3D-TiO 2 NTAs (0.4 mA/cm 2 ). This photocurrent density value is superior or comparable to the previously reported values on self-doping TiO 2 NTAs formed on Ti foil (Table S1) [39,43,[46][47][48]51]. This means that the optimal electrochemical reduction bias is −1.3 V, which can be attributed to the two-faced effect of the O-vacancies on the PEC water splitting performance, and will be discussed thoroughly in the following text. mA/cm 2 , which is about four times larger than that of the pristine 3D-TiO2 NTAs (0.4 mA/cm 2 ). This photocurrent density value is superior or comparable to the previously reported values on self-doping TiO2 NTAs formed on Ti foil (Table S1) [39,43,[46][47][48]51]. This means that the optimal electrochemical reduction bias is −1.3 V, which can be attributed to the two-faced effect of the O-vacancies on the PEC water splitting performance, and will be discussed thoroughly in the following text.  The structural and chemical stability is a critical parameter for a photoelectrode during the PEC water splitting. To assess this property, the photocurrent density vs. time (J-t) curves of the pristine and ECR-3D-TiO 2 NTAs−1.3 V are obtained at 0.22 V vs. Ag/AgCl under continuous illumination (Figure 4c). No sign of decrease in photocurrent densities for the pristine and ECR-3D-TiO 2 NTAs-1.3 V are detected during the entirely measured 180 min. To further identify whether the observed photocurrents derive from the water splitting reaction, the amount of oxygen evolved from the ECR-3D-TiO 2 NTAs−1.3 V was determined by a fluorescence sensor. The amount of evolved oxygen increases linearly with test time with unity Faradic efficiency. Figure S6 presents the FE-SEM image and XRD pattern of the ECR-3D-TiO 2 NTAs−1.3 V after continuous PEC water splitting for 180 min, which prove that the surface morphology and crystal phase of the ECR-3D-TiO 2 NTAs-1.3 V remains intact. These results sufficiently confirm that excellent stability of the ECR-3D-TiO 2 NTAs−1.3 V, which is suitable for the potential long-term PEC water splitting application.
To investigate the effect of the electrochemical reduction on the electronic characteristics of 3D-TiO 2 NTAs, electrochemical impedance spectra (EIS) measurements were performed and the Nyquist plots are shown in Figure 5a, where the scatter points are the original experimental data, and the solid lines are the fitted curves utilizing the equivalent circuit mode in the inset of Figure 5a. It can be clearly seen that the equivalent circuit model fits well with the two samples. In this equivalent circuit model, R s corresponds to the overall series resistance of the circuit, and R ct represents the charge transfer resistance [47,52]. As depicted in Figure 5a, the ECR-3D-TiO 2 NTAs−1.3 V has a smaller semicircle diameter than the pristine 3D-TiO 2 NTAs under illumination, suggesting the smaller charge transfer resistance of the ECR-3D-TiO 2 NTAs−1.3 V. The charge transfer resistance can be obtained by fitting the Nyquist plots with the equivalent circuit model. As expected, the charge transfer resistance R ct of the ECR-3D-TiO 2 NTAs−1.3 V is reduced from 440.45 to 133.08 Ω, which indicates a more effective separation of the photogenerated electron and hole and/or a faster interfacial charge transfer of the ECR-3D-TiO 2 NTAs−1.3 V. Moreover, the electrochemical active surface areas of the pristine 3D-TiO 2 NTAs and ECR-3D-TiO 2 NTAs−1.3 V are estimated from the capacitive region of cyclic voltammograms (CV). The data shown in Figure S7 reveal that the electrochemically active area of the ECR-3D-TiO 2 NTAs−1.3 V is only 1.05 times than that of the pristine -3D-TiO 2 NTAs, indicating that both samples have comparable electrochemically active areas. In addition, the slope of the Mott-Schottky plot collected from the ECR-3D TiO 2 NTAs−1.3 V is much smaller than that of the pristine 3D-TiO 2 NTAs, which suggest an improvement of donor densities (Figure 5b). The donor densities were estimated from the slopes of Mott-Schottky plots using the following equation:

Discussion
Based on the above experimental results, the boosted photoelectrochemical water splitting performance of ECR-3D-TiO2 NTAs can be ascribed to the introduction of the Ovacancies. Firstly, PEC water splitting performance of the photoelectrode largely depend on its capability of effectively absorbing visible light. In the present case, the presence of O-vacancies results in the generation of a new defect energy level near the conduction band, which lead to the bandgap shrinkage, hence being favorable for the visible light harvesting. More importantly, the incident, reflected and/or refracted visible light captured by the 3D configuration is also absorbed by defect energy level near CB created by oxygen vacancy. Secondly, the introduction of the O-vacancies (Ti 3+ ) in ECR-3D-TiO2 NTAs generally work as electron donors, which leads to the enhanced electronic conduc-

Discussion
Based on the above experimental results, the boosted photoelectrochemical water splitting performance of ECR-3D-TiO 2 NTAs can be ascribed to the introduction of the O-vacancies. Firstly, PEC water splitting performance of the photoelectrode largely depend on its capability of effectively absorbing visible light. In the present case, the presence of O-vacancies results in the generation of a new defect energy level near the conduction band, which lead to the bandgap shrinkage, hence being favorable for the visible light harvesting. More importantly, the incident, reflected and/or refracted visible light captured by the 3D configuration is also absorbed by defect energy level near CB created by oxygen vacancy. Secondly, the introduction of the O-vacancies (Ti 3+ ) in ECR-3D-TiO 2 NTAs generally work as electron donors, which leads to the enhanced electronic conductivity and upward shift of the Fermi energy level, thereby facilitating the transfer and separation of photogenerated charge carrier at the semiconductor-electrolyte interface. Nevertheless, the excess Ovacancies may be the recombination centers for photogenerated carriers, hence limiting the generation of photocurrent [53,54]. Therefore, the optimized amount of the O-vacancies is essential to the PEC water splitting performance. The XPS result illustrates that the amount of the O-vacancies increases with deceasing electrochemical reduction bias (Figure 3 and Figure S5). Consequently, it can be included that the ECR-3D-TiO 2 NTAs−1.4 V may possess excess amount of the O-vacancies (Ti 3+ ), which lead to the recombination of photogenerated carriers before reaching the TiO 2 /electrolyte interface. Accordingly, the optimal electrochemical reduction bias is −1.3 V from the perspective of PEC water splitting activity.

Conclusions
In conclusion, black 3D-TiO 2 NTAs have been successfully fabricated via an electrochemical reduction and employed as a photoanode for PEC water splitting. The introduction of the O-vacancies results in bandgap shrinkage, which can effectively boost the utilization efficiency of visible light including the incident, reflected and/or refracted visible light captured by the 3D configuration. Moreover, the O-vacancies (Ti 3+ ) can work as electron donors, which leads to the enhanced electronic conductivity and upward shift of the Fermi energy level, thereby facilitating the transfer and separation of photogenerated charge carrier at the semiconductor-electrolyte interface. Benefiting from the oxygen vacancy, the optimized photocurrent density of ECR-3D-TiO 2 NTAs under white light illumination generated the photocurrent density of 1.6 mA/cm 2 at 0.22 V vs. Ag/AgCl, which is superior or comparable to the previously reported values on self-doping TiO 2 NTAs formed on Ti foil.